Elucidating the Role of O2 Uncoupling for the Adaptation of Bacterial Biodegradation Reactions Catalyzed by Rieske Oxygenases

Oxygenation of aromatic and aliphatic hydrocarbons by Rieske oxygenases is the initial step of various biodegradation pathways for environmental organic contaminants. Microorganisms carrying Rieske oxygenases are able to quickly adapt their substrate spectra to alternative carbon and energy sources that are structurally related to the original target substrate, yet the molecular events responsible for this rapid adaptation are not well understood. Here, we evaluated the hypothesis that reactive oxygen species (ROS) generated by unproductive activation of O2, the so-called O2 uncoupling, in the presence of the alternative substrate exert a selective pressure on the bacterium for increasing the oxygenation efficiency of Rieske oxygenases. To that end, we studied wild-type 2-nitrotoluene dioxygenase from Acidovorax sp. strain JS42 and five enzyme variants that have evolved from adaptive laboratory evolution experiments with 3- and 4-nitrotoluene as alternative growth substrates. The enzyme variants showed a substantially increased oxygenation efficiency toward the new target substrates concomitant with a reduction of ROS production, while mechanisms and kinetics of enzymatic O2 activation remained unchanged. Structural analyses and docking studies suggest that amino acid substitutions in enzyme variants occurred at residues lining both substrate and O2 transport tunnels, enabling tighter binding of the target substrates in the active site. Increased oxygenation efficiencies measured in vitro for the various enzyme (variant)-substrate combinations correlated linearly with in vivo changes in growth rates for evolved Acidovorax strains expressing the variants. Our data suggest that the selective pressure from oxidative stress toward more efficient oxygenation by Rieske oxygenases was most notable when O2 uncoupling exceeded 60%.


■ INTRODUCTION
Understanding biodegradation processes of anthropogenic organic contaminants in natural and engineered environments including soils, sediments, as well as (waste)water treatment plants is a critical step for designing measures that maintain ecosystem services and sustainable access to food and water under increasing human impact. 1−5 Contaminant biodegradation routes often involve redox reactions with molecular O 2 as the oxidant.−13 However, predicting the occurrence of oxidative biodegradation based on genomic inferences of oxygenase abundances in the environment is particularly difficult.−25 On the other hand, the substrate specificity is highly variable, even for structurally related enzymes, 24−28 and subject to evolutionary pressure.−45 Reactions with ROS can, in fact, inactivate proteins.−49 However, it is so far unclear whether alterations of protein sequences and structures emerge as a consequence of ROS formation in O 2 -activating enzymes and whether this process ultimately results in a reduced O 2 uncoupling as microorganisms adapt to new substrates.−54 Consistent with this hole hopping process and the notion that cellular reductants would scavenge these oxidation equivalents, reconfiguration of metabolic fluxes has been observed upon ROS release from the oxygenase concomitant with faster turnover of reduction equivalents. 55Thus, even in the presence of defense mechanisms at the enzyme level, modifications of contaminant-degrading oxygenases that allow for reduced extents of ROS formation would appear (at least) metabolically beneficial to the oxygenase-expressing microorganism.These oxygenases enable access to an additional, potentially broad substrate spectrum of environmental contaminants.Several contaminant-degrading oxygenases are considered promiscuous 56−59 and their alteration to accept alternative substrates constitutes a competitive advantage for the respective microorganisms. 11However, it is not known whether a modification of contaminant-degrading oxygenases as a consequence of exposure to xenobiotic substrates is indeed accompanied by reduced O 2 uncoupling and concomitant decrease of ROS formation.
In this study, we aimed to establish the relationship between microbial adaptation to alternative substrates, modification of enzyme sequence and structure, and the oxygenation efficiency of contaminant-degrading oxygenases.To that end, we studied Rieske non-heme ferrous iron oxygenases, 20,60−65 an important class of contaminant-degrading enzymes.−86 Despite the well-known role of Rieske oxygenases in biocatalysis, the factors that lead to successful substrate oxygenation are still largely elusive.−94 Yet, it remains unclear whether any of these structural and electronic factors are optimized in Rieske oxygenases as microorganisms adapt to alternative contaminants as primary substrates.
Previous laboratory evolution experiments with nitroarene dioxygenases 95−99 offer a promising avenue to examine the above questions.These works showed that Acidovorax sp.strain JS42 expressing the Rieske oxygenase 2-nitrotoluene dioxygenase (2NTDO) altered its substrate specificity from 2nitrotoluene to 3-and 4-nitrotoluene as growth-supporting substrates if exposed to one of these alternative substrates over weeks and months, respectively.This adaptation process was accompanied by one and two amino acid substitutions in the enzyme variants compared to the wild-type (wt) 2NTDO.Note, however, that selective sequencing of the ntdAcAd genes encoding this oxygenase precluded the identification of additional mutations related to ROS defense mechanisms.We recently quantified in vitro oxygenation efficiencies of Rieske oxygenases as the fraction of consumed O 2 used for substrate hydroxylation.Our studies of oxygenation efficiencies of wt 2NTDO revealed that the share of productively activated O 2 by 2NTDO with 3-and 4-nitrotoluene amounts to less than 16 and 6%, respectively.Conversely, 98% of O 2 was recovered in hydroxylation products when 2-nitrotoluene was used as the substrate. 24ere, we revisit the work of Parales et al. 95,96 to examine two hypotheses regarding the adaptation of 2NTDO toward oxygenation of 3-and 4-nitrotoluene.(i) Enzyme evolution in such biodegradation processes correlates with increasingly efficient substrate hydroxylation and thus reduction of ROS generation.(ii) This process occurs as substitutions at residues that are located in structural "hotspots" for catalytic performance. 64,70,91,100To that end, we carried out the following three tasks.(1) We quantified the substrate oxygenation efficiencies for four nitroaromatic substrates with wt 2NTDO as well as five adapted variants exhibiting single and double amino acid substitutions 95,96 and studied their relevance for enabling growth of Acidovorax sp.strain JS42.(2) We examined whether the kinetic mechanism associated with the catalytic cycle of Rieske oxygenases 25,101,102 was conserved in enzyme variants by evaluating the kinetics of O 2 activation and substrate hydroxylation.(3) We analyzed whether the amino acid substitutions are localized at or near the active site or in substrate transport tunnels and whether these structural modifications modulate substrate binding and/or intramolecular substrate transport 103,104 in ways that could affect the extent of O 2 uncoupling.Finally, we assessed our results in an environmental context by examining the relevance of the variants identified by Parales et al. 95,96 through the bioinformatic analysis of naturally occurring Rieske oxygenase sequences.

METHODS
All chemicals and materials used are reported in Section S1 of the Supporting Information.Enzyme purification, experimental, and analytical procedures are identical to methods described before by Bopp et al. 24

Bacterial Strains and Site-Directed Mutagenesis
Escherichia coli DH5α expressing recombinant dioxygenases from plasmids pKSJ90 (2NTDO M248I) and pKSJ92 (2NTDO L238V-M248I) were obtained from Rebecca E. Parales. 95Plasmids corresponding to pKMM32, pKMM33, and pKMM35 96 carrying the ntdAcAd genes for the I204A, I204T, and I204V variants of 2NTDO, respectively, were obtained from site-directed mutagenesis using pDTG800 99 carrying the ntdAcAd genes of wt 2NTDO as template.We performed site-directed mutagenesis using KLD mix (New England BioLabs) with forward primers including the exchanged codon and nonoverlapping reverse primers that are listed in Table S1.Incubation and purification procedures followed the methods described for NBDO and 2NTDO. 24Activities, purities, and yields were comparable to those for 2NTDO.A comparison of the two amino acid sequences of the catalytically active α subunits of NBDO and 2NTDO is shown in Figure S1.

Enzyme Assays
NADH-limited enzyme assays were used to determine the turnover of nitroaromatic substrates to organic and inorganic reaction products, O 2 disappearance, and H 2 O 2 formation, as well as changes in 13 C/ 12 C ratios of the organic substrate and 18 24,25 900 μL was withdrawn from each NADH-limited enzyme assay and mixed with 100 μL of an HRP assay in MES buffer to final concentrations of 10 mg L −1 HRP and 500 μM 4methoxyaniline.Kinetics of O 2 consumption and substrate oxygenation were performed in modified assays as described before 24,25 and in Supporting Information Sections S2.2 and S2.3.

Chemical and Isotopic Analyses
Organic Substrate and Product Concentrations.Organic substrates, namely, nitrobenzene and the three nitrotoluene isomers, as well as nitrobenzyl alcohols and catecholic products were quantified by high-performance liquid chromatography, whereas NO 2 − was measured photometrically.All procedures have been described previously. 24,25table Isotope Analyses.The 12 mL vials were prepared for analysis of 18 O/ 16 O ratios in O 2 by creating a 3 mL headspace filled with N 2 . 24,105,106After partitioning of dissolved O 2 into the headspace, 1000 μL of gaseous sample was withdrawn and injected into a GC/IRMS (Thermo Fisher Scientific) equipped with two sequentially connected PLOT columns (Restek from BGB Analytik; 30 m × 0.32 mm ID, 30 μm film thickness).Instrument parameters and calibrations followed methods described in Bopp et al. 107 Carbon isotope ratios ( 13 C/ 12 C) of organic substrates were analyzed on a GC/IRMS after solid phase microextraction as described in earlier works. 106,108ta Evaluation Reaction Stoichiometries.Substrate transformation and product generation were quantified in terms of reaction stoichiometries, which were normalized to the amount of consumed NADH as shown in eq 1. Stoichiometric coefficients of species j, |υ j |, were calculated through linear regressions of eq 1 for the different concentrations of the nitroaromatic substrate, dissolved O 2 , hydroxylated aromatic product, and NO 2 − obtained from assays with different amounts of added NADH after completion of turnover.
where |υ j | is the stoichiometric coefficient of species j, [j] is the measured molar concentration of the species after complete consumption of NADH, and [NADH] is the nominal concentration of NADH with q as the y-intercept.Uncertainties of |υ j | reflect errors arising from standard deviations in the measurements and linear regression analysis and are reported as 95% confidence intervals.
The extent of O 2 uncoupling, f O uc , was determined by linear regression of eq 2.
where [NO [ ] and [O 2 ] are the initial and final O 2 concentrations, respectively.We accounted for the possible constant O 2 background consumption independent of NADH concentration according to procedures described in Bopp et al. 24 The corresponding data are shown in Section S2.5 together with a discussion of the sensitivity of f O uc 2 to the accuracy of reaction product quantifications at small substrate turnover.
Isotope Effects.Kinetic isotope effects pertinent to the hydroxylation of aromatic carbon, 13 C-KIE, were determined through non-linear regressions of the observable changes in 13 C/ 12 C ratios versus the fractional amount of residual substrate resulting in C isotope signatures, δ 13 C, and C isotope enrichment factors, ε C , according to eqs 3 and 4.
where δ 13 C and δ 13 C 0 are the C isotope signatures of the residual substrate after the reaction and its original value, respectively.[S] and [S] 0 are the final and initial substrate concentrations, respectively, and are documented in Section S2.6.Parameter n C is the number of carbon atoms in the substrate to account for the isotopic dilution of the isotope effect assuming an asynchronous hydroxylation mechanism. 101,106Non-linear regression fits including the standard deviation of triplicate measurements were carried out in Igor Pro (WaveMetric Inc.).In a similar procedure, kinetic isotope effects of O 2 activation, 18 O-KIE, were derived as average for both O atoms in O 2 according to eq 5.
where [O 2 ] and [O 2 ] 0 are the final and initial dissolved O 2 concentrations, respectively.

Computational Analyses of the Protein Structure
Generation of Structural Models for 2NTDO and Its Variants.Homology models of wild-type and variants of 2NTDO were generated by submitting the corresponding sequences as a α 3 β 3 heterohexamer to the AlphaFold-Multimer 109 while keeping the number of multimer predictions per model to 1.The model structures obtained were aligned to the crystal structure of NDBO (PDB ID: 2BMQ) using PyMOL (v 2.5.2,Schrodinger Inc.).After including hetero atoms, modeled structures were refined applying the Rosetta FastRelax protocol 110,111 to resolve potential clashes.The model structures of wt 2NTDO are shown exemplarily in Figure S2.All homology model structures were predicted with high confidence and overall template modeling scores (ipTM and pTM scores >0.94). 109he homology models of variants closely resemble the wt 2NTDO with all-atom root-mean-square deviations (RMSD) between 0.35 and 0.95 Å (Table S6).
Substrate Tunnel Identification and Evaluation of Substrate Transport.CaverDock analyses 112 were carried out with the structural models obtained using the CaverDock Web Interface 113 by selecting the catalytic iron from the protein chain C (see Figure S2) as the starting point for tunnel identification and using a minimum probing radius of 0.6 Å while keeping default parameters otherwise (i.e., shell depth of 4, shell radius of 3, clustering threshold of 3.5, a maximal distance of 3, and desired radius of 5).Tunnels for nitroaromatic substrate transport analyses were selected as the highest-priority tunnels according to the likelihood ranking of CaverDock.For O 2 transport analyses, tunnels were selected based on close proximity to the substrate tunnel and a high CaverDock priority score.Ligand transport analyses were carried out with CaverDock 112 by applying default parameters, namely, discretization delta (Å) of 0.3 and calculating lower-bound trajectories only.
Substrate Docking Studies.Molecular docking was carried out using the homology models described above.The ADFRsuite 1.0 was used to prepare the receptors and ligands with fixed torsion as pdbqt files.AutoDock Vina 1.2.3 114 was subsequently employed for the molecular docking with a fixed seed of 42, exhaustiveness of 64, box size of 15 × 15 × 15 Å 3 , and the catalytic iron of chain C in the αsubunit as center coordinates (see Figure S2).All other parameters were set to default values, namely, a maximum number of binding modes of 9, a minimum RMSD between output poses of 1, a maximum energy difference between the best and worst binding mode of 3 kcal/mol, and a grid spacing of 0.375 Å. Docking poses were ranked according to the shortest Asn258-NO 2 bonding distance.Specifically, poses exhibiting the shortest distance were chosen as the most plausible ones based on previous knowledge of H-bonding between nitroaromatic substrates and Asn258. 26,115

Bioinformatic Analysis
The National Center for Biotechnology Information protein databases were accessed on 13th of January, 2023.We used BLASTP with default parameters and 2-nitrotoluene dioxygenase (AAB40383.1)from Acidovorax sp.strain JS42 95 as a query sequence against two databases: non-redundant protein sequences (nr) and proteins from WGS metagenomic projects (env_nr).Both databases were updated on 12th of January, 2023.For each, the top 5000 hits were pulled and filtered for significant hits with a bitscore above 250.Working with each data set separately, we aligned protein sequences using Clustal Omega 116 with default parameters and extracted alignment positions corresponding to variants of interest (M248I, L238V-M248I, I204T, I204A, and I204V).To test for phylogenetic patterns in mutations in each data set, an approximate maximum-likelihood phylogeny was estimated using FastTree v.2.1.11with default parameters. 117

Efficiency of Substrate Oxygenation by Wild-Type 2NTDO and Its Variants
2NTDO and its variants catalyzed the transformation of nitrobenzene and nitrotoluenes to dioxygenated (methyl)catecholic products, monooxygenated nitrobenzylalcohols, and NO 2 − (Figure 1a).To assess the efficiency of oxygenation vs the unproductive activation of O 2 , we quantified all substrate and product concentrations in enzyme assays at different extents of substrate turnover.In the following, we compare the dioxygenation efficiency of the wt 2NTDO with a variant enzyme arising from a single amino acid substitution in laboratory evolution experiments with 4-nitrotoluene as the substrate (referred to as 4NT + experiments) 95 for a general illustration of our experimental approach.
Figure 1b shows the consumption of O 2 and 4-nitrotoluene and concomitant formation of NO 2 − by wt 2NTDO.The 4methylcatechol product was detected but at concentrations below its quantification limit due to the low substrate turnover.While most of the NADH is used for O 2 activation (υ Od 2 = 0.85 ± 0.01 mol/mol NADH, Table 1, entry 4b), only minor − in enzyme assays of wt 2NTDO after complete consumption of different amounts of NADH. 244-Nitrobenzylalcohol was not detected in any of the assays, whereas 4-methylcatechol was below the limit of quantification.The black lines and shaded areas represent linear fits with 95% confidence intervals with slopes shown in Table S7 amounts of 4-nitrotoluene (υ S = 0.10 ± 0.03 mol/mol NADH, Table 1, entry 4a) are transformed to NO 2 − (υ NOd 2 − = 0.05 ± 0.01 mol/mol NADH, Table S7).The complete compilation of stoichiometric coefficients for 4NT + experiments obtained with eq 1 is shown in Table S7.The pronounced discrepancy between O 2 activation and formation of the dioxygenation product was reflected in an almost completely inefficient O 2 activation by 2NTDO and a fraction of O 2 uncoupling, f O uc 2 , of 0.94 ± 0.01 (eq 2, Table 1, entry 4).Thus, only 6% of the activated O 2 was used efficiently in the oxygenation of the substrate, whereas the remaining 94% were assigned to inefficient O 2 activation.This observation concurs with previous ones of low activities of 2NTDO with 4-nitrotoluene. 95he M248I variant of 2NTDO, in contrast, developed the ability to dioxygenate 4-nitrotoluene with much greater efficiency (Figure 1c).While the stoichiometric coefficient of O 2 consumption, υ Od 2 , remained similar to the wild-type (0.S7), the oxygenation by the M248I variant was 8-fold higher compared to the wt 2NTDO.Thus, the extent of O 2 uncoupling, f O uc 2 , dropped from 94 to 53% for wild-type and the variant, respectively.With increasing NADH concentrations, however, we observed a decrease in the sum of the nitroaromatic substrate and NO 2 − concentrations indicated by mass balance calculations.Similar effects were observed for experiments with 2NTDO and its 4NT + variants, catalyzing 3-and 4nitrotoluene oxygenation.This observation points to an additional but minor unknown reaction product that did not become evident in NO 2 − , methylcatechol, or nitrobenzylalcohol formation and impeded quantification of oxygenation efficiencies by eq 2.An alternative method to determine  S4).In the following discussion, we will stick to the more conservative estimates based on eq 2 and refer to Section S2.5 for our reasoning.Regardless of this uncertainty, the efficiency of the O 2 activation by the M248I variant significantly increased compared to that of wt 2NTDO by at least 40%.The comparison of 4-nitrotoluene turnover by wt 2NTDO and the M248I variant illustrates the drastic changes in efficiency of oxygenation caused by a single amino acid substitution.Ju and Parales 95 showed that an additional substitution (L238V-M248I) evolved directly from the M248I variant.Indeed and as shown in Figure 1d 108 We carried out identical types of experiments for the quantification of oxygenation efficiencies and O 2 uncoupling for the 3-nitrotoluene adapted variants I204A, I204T, and I204V denoted henceforth as 3-NT + experiments (Figure S3).These enzymes originate from laboratory evolution experiments with Acidovorax sp.strain JS42 adapted on 3nitrotoluene over one month. 96In analogy to 4-NT + variants M248I and L238V-M248I, we found that each of these substitutions improved the efficiency of O  S8, entries 15, 18 and 21), again reflecting an adaptation to the new nitrotoluene substrate isomer.

Evaluation of Oxygenation Efficiency and O 2 Uncoupling as a Measure for the Adaptation of Substrate Specificity
We studied the consequences of amino acid substitutions in 2NTDO on the substrate spectra of the enzymes.To that end, we compared oxygenation efficiencies of 2NTDO and five variants from the 4NT + and 3NT + experiments with the four substrates nitrobenzene and the three nitrotoluene isomers.
Figure 2a shows the fraction of inefficiently activated O 2 , The three amino acid substitutions at position I204 of 2NTDO allowed growth on 3-nitrotoluene. 96Figures 2b and  S3 show the adaptation trends according to the same visual aids used for the 4NT + experiment in Figure 2a.Each variant improved in oxygenation efficiency with 3-nitrotoluene significantly to 0.37 (I204A), 0.41 (I204T), and 0.39 (I204V) compared to the wild-type (0.16).Contrary to the 4NT + experiments, however, the mutations in the 3NT + experiments gave rise to a substantial decrease in oxygenation efficiency for the remainder of the substrates.The f O uc 2 -values of nitrotoluene and 2-nitrotoluene increased between 2-and 30-fold to 0.6 and 0.7, respectively, indicating a substantial decline of substrate specificity.4-Nitrotoluene was a poor substrate for I204 variants and the substrate turnover in these enzyme assays was mostly insufficient to allow reliable quantification of f O uc 2 close to unity.We note that the serial enrichment cultures in the 4NT + experiments ran for 6 months 95 as compared to only one month in the 3NT + experiments. 96Whether this 6-fold shorter adaptation period in the 3NT + experiments is responsible for the more exclusive adaptation toward this one substrate is unclear.

Structural Analysis of Point Mutations
We explored whether structural changes in the enzymes upon amino acid substitutions relate to the observable changes in oxygenation efficiency from two perspectives.First, we studied if the observed substitutions affected substrate and O 2 transport from the protein surface to the active site.In addition, we evaluated changes in the binding of nitroaromatic substrates in the active site through molecular docking by constraining the docking modes based on the interaction between Asn258 and the nitro group via hydrogen bonding.The H-bonding interaction with Asn258 is not only a known prerequisite for catalytic activity of nitroarene dioxygenases 26 but could also modulate the ability of the enzyme to hold the substrate in place for efficient hydroxylation by reactive Feoxygen species. 42ollowing procedures used in previous studies to identify small-molecule tunnels in non-heme Fe enzymes, 104,118−121 we characterized possible substrate and O 2 transport tunnels in wt 2NTDO and the five variants using CaverDock 112 with minimum probing radii of 0.9 and 0.6 Å for substrates and O 2 , respectively.In all homology models, tunnels from the catalytic non-heme Fe site to the surface were localized and could potentially be assigned as nitroaromatic substrate tunnels.Tunnel dimensions and throughput derived with CaverDock are compiled in Table S10 and show lengths between 20 and 23 Å, as well as mean and bottleneck radii of 1.6−1.9 and 1.1−1.5 Å, respectively.Figures 3 and S7 show examples of substrate tunnels for 2NTDO, as well as variants L238V-M248I and I204A from the 4NT + and 3NT + experiments, respectively.As evident from this structural analysis (Figure 3b), I204 is located at the substrate tunnel, whereas residues L238 and M248 are not.The complete list of amino acid residues in the various tunnels is compiled in Table S11.Evidence of I204 acting as a bottleneck residue was obtained from ligand transport analyses.Substrate binding energies calculated with CaverDock along the substrate tunnels are shown in Figure S5.These data indeed revealed lower substrate binding energies at approximately 12 Å distance from the non-heme Fe site through substitutions at the I204 residue.
The same type of CaverDock analysis with a smaller probing radius enabled tentative assignment of O 2 transport tunnels (Figure S8).While several tunnels can typically be identified which could act as O 2 transport tunnels (see above and refs 122−124 ), a common tunnel structure formed by multiple  hydrophobic residues close to the substrate tunnel was identified in all variants.The architecture of the identified tunnels (Table S10) as defined by the mean and bottleneck radii and throughput numbers by CaverDock agrees well with similar analyses of O 2 tunnel structures in lipoxygenases, another class of non-heme ferrous iron enzymes. 119e found that residues L238 and M248 are part of O 2 transport tunnels identified for wt 2NTDO.However, these residues no longer appear as part of the tunnels that were identified as most likely in the evolved enzyme variants.By contrast, we observed a slightly increased O 2 tunnel radius and a reduced tunnel curvature for the most likely tunnel in variants M248I and L238V-M248I in comparison to the wildtype.O 2 binding energies calculated along these O 2 transport tunnels (Figure S6) are smaller than those for the substrates in substrate transport tunnels (Figure S5).Moreover, differences in O 2 binding energy are not as pronounced among the variants carrying different amino acid substitutions.In contrast to substrate transport tunnels (Figure S5), bottlenecks for transport of O 2 are located at a greater distance from the active site and thus in closer proximity to L238 and M248 residues.
This preliminary and, in the absence of molecular dynamics simulations of nitroarene dioxygenases, 115 purely static evaluation of structural features of O 2 transport tunnels in 2NTDO and its variants underscores the well-known relevance of tunnels.−127 While our analysis illustrates that amino acid substitutions in the 4NT + and 3NT + experiments occurred at residues that likely affect substrate and O 2 transport, it is currently unknown how mutations in positions L238, M248, and I204 resulted in a reduction of O 2 uncoupling.The different locations of substitutions in 4NT + and 3NT + variants appear to point to structurally different modes of adaptation for the two isomers.Changes in the O 2 transport tunnel in 4NT + variants largely maintained the oxygenation efficiency of other substrates (Figure 2), whereas alterations to the substrate tunnel following adaptation to 3nitrotoluene decreased oxygenation efficiencies of other substrates.Evidence from engineering O 2 channels in lipoxygenase 118 showed that 2-fold larger bottleneck radii increased oxygenation efficiencies and specific protein activities.Here, we do not observe that changes in kinetic parameters associated with O 2 consumption (e.g., k cat /K mvalues in Table S12) of wt 2NTDO, variants, and substrates correlate with their O 2 uncoupling behavior.
We further studied the consequences of the three point mutations on substrate binding in the active site by evaluating the H-bonding interactions between the aromatic NO 2 group and residue Asn258 in the various docking poses for each enzyme/variant-substrate combination.Here, we deliberately focus on poses with the shortest Ar-NO 2 -Asn258 binding distance.The comparison of all substrate binding affinities with the extent of O 2 uncoupling, f O uc 2 , is shown in Figure S9.-values of this substrate.We found that binding affinities of 4nitrotoluene and 3-nitrotoluene in wt 2NTDO increase (i.e., become more negative) as a consequence of mutations M248, L238-M248, and I204 (Figure 4b,d).Moreover, the increase of binding affinities correlates approximately linearly with f O uc 2 and supports the conclusion that tighter substrate binding might be advantageous to hold the substrate in place for efficient hydroxylation as invoked for variations of O 2 uncoupling observed for α-ketoglutarate dependent nonheme ferrous iron oxygenase. 42,43However, such correlations were not observed for each enzyme/variant-substrate combination (Figure S9).While selected enzyme variants stemming from the 4NT + and 3NT + experiments show tighter substrate binding compared to wt 2NTDO for some of the tested substrates, systematic trends were observed only for substrates 3-and 4-nitrotoluene (Figure 4).According to common views of enzyme catalysis, one expects tuning of transition state stabilization to occur with a concomitant optimization of the enzymes' specific activity (i.e., k cat /K m ) 129 according to metabolic needs. 29A preferred optimization of substrate binding alone, that is lowering of K m , under the conditions of the adaptation experiments where substrate concentrations exceed K m might not necessarily improve catalysis. 130odulation of enzymatic activity through tighter binding of the substrate as a means to avoid O 2 uncoupling and concomitant production of ROS, by contrast, would appear as a quite meaningful adaptation strategy.

Catalytic Cycle of 2NTDO Variants
We examined the catalytic cycle of 2NTDO variants shown in Figure 5 and probed for changes in rate-limiting steps of O 2 activation and substrate hydroxylation upon amino acid substitution by evaluating nitroaromatic substrate and O 2 kinetic isotope effects (KIEs).Competitive 18 O and 13 C kinetic isotope effects of dissolved O 2 and the nitroaromatic substrate, respectively, were quantified following the methodology of our previous works 24,25,106 and the data are compiled in Tables 1  and S8. Figure 5b shows the 18 O-KIEs of O 2 activation in the presence of nitrobenzene and different nitrotoluene isomers for enzyme variants. 18O-KIEs of most variant-substrate combinations ranged from 1.012 to 1.020 with an average value of 1.017 ± 0.003.This outcome is indicative of a kinetic mechanism in which the rate-limiting step of O 2 activation involves formation of Fe III -(hydro)peroxo species 131 (2 in Figure 5a).This interpretation is corroborated by the detection of H 2 O 2 from O 2 uncoupling in assays of variants (Table S2).Almost identical observations in terms of 18 O-KIE were made with wt 2NTDO (1.016 ± 0.002) 24 and another nitroarene dioxygenase 25 with an even larger set of substrates implying that the mechanism of O 2 activation remained identical in 2NTDO variants.We do, however, also find three 18 O-KIE values between 1.022 and 1.026 for oxygenation of nitrobenzene and nitrotoluene isomers by I204 variants which exceed the values of the other 18 O-KIE in the rest of our data.These values approach theoretical 18 O equilibrium isotope effects ( 18 O-EIEs) that are considered representative of the rate-limiting formation of Fe IV �O species ( 18 O-EIE of 1.0287). 131This mechanism of O 2 activation is also consistent with reactions of other Rieske oxygenases. 132,133Whether this finding hints at changes in O 2 activation mechanisms in I204 variants is currently unclear.The higher 18 O-KIE values make up 30% of the data for I204 variants but less than 6% of the approximately 50 18 O-KIE values that we have determined so far for nitroarene dioxygenases. 24,25 13  Kinetic isotope effects ( 13 C-KIEs) of nitroaromatic substrate hydroxylation were used as probes for changes at later stages of the catalytic cycle in the variants.Specifically, we characterized the timing of substrate hydroxylation vs O 2 uncoupling (green arrows in Figure 5a). Th 13 C-KIE values for substrate hydroxylation by the different 2NTDO variants were, again, similar to what was observed previously for wt 2NTDO (0.998−1.011, Figure 5c, Tables 1 and S8).24 These data indicate a lack of reversibility of species 3 formation, from which hydroxylation occurs with an intrinsic 13 C-KIE of up to 1.039.108 As a consequence, the unreacted substrate released upon O 2 uncoupling was not subject to any isotope fractionation from reaction 3 → 4 (Figure 5a).24 The two exceptions to this interpretation with large 13 C-KIE values (I204T and I204A of 1.013 ± 0.002 and 1.022 ± 0.008, respectively, Figure 5c) also come from reactions catalyzed by I204 variants.In analogy to the interpretation of 18 O-KIE data above, these data could be seen as first and thus preliminary evidence for an alteration of the hydroxylation mechanism in enzyme variants.Larger 13 C-KIE values, by contrast, are not uncommon.Such values are indicative of the slightly different catalytic scheme as observed previously for another nitroarene dioxygenase where the magnitude of C isotope fractionation in the substrate was shown to correlate with the extent of O 2 uncoupling.25

Comparison of In Vivo Evidence for Adaptation with Oxygenation Efficiency Quantified In Vitro
We hypothesize that O 2 uncoupling and the ensuing formation of ROS limits the fitness of Acidovorax sp.strain JS42 in a substrate-dependent manner, which would also correlate with the adaptation of mutant strains to alternative substrates.To examine O 2 uncoupling as a measure for the characterization of microbial adaptation processes, we compared the oxygenation efficiencies obtained in vitro (in enzyme assays) with in vivo parameters for microbial activity (as NO 2 − formation rates) and growth (from inverse doubling times).The in vivo parameters were taken from previous experiments with evolved Acidovorax strains 95,96 and are referred to here collectively as NT + strains.
The comparison of specific in vivo NO  6a) became especially evident in the doubly substituted oxygenase variant L238V-M248I but not so much in M248I.This trend implies that the increase in dioxygenation activity and thus NO 2 − formation rate in vivo could have been the consequence of the decrease of O 2 uncoupling that we quantified in vitro.Deviations from the linear correlation illustrate that additional factors other than the oxygenation efficiency of the variants are relevant for microbial activity, a finding that points to the inherent limitations of assessing substrate specificity exclusively from the activity of the dioxygenase.
Microbial growth quantified here as inverse doubling times 95,96 represents a more holistic measure of adaptation in vivo because Acidovorax sp.strain JS42 utilizes nitroaromatic substrates as the sole source of carbon and energy.Activity of the dioxygenase and its efficiency in oxygenating the substrate are thus expected to be reflected in microbial growth rates.In fact, while Acidovorax sp.strain JS42 expressing wt 2NTDO could not grow on 3-and 4-nitrotoluene, prolonged exposure resulted in mutations in the codons for residues M248, L238, and I204 that supported growth of 4NT + and 3NT + strains on the respective substrate (Table S13).We note that some of these mutations can be related to G → T transversions typically hypothesized for ROS-induced modifications of nucleic acids 134 (Section S3.6).In Figure 6b, growth-related observations are shown as changes of O 2 uncoupling, f O uc 2 , and of inverse doubling times, Δ(1/doubling time), where no growth is assigned a value of 0 h −1 .Increased growth rates of 3-NT + and 4-NT + strains with their target substrate 3-and 4nitrotoluene are correlated linearly with their oxygenation efficiency gain (top left quadrant in Figure 6b).Conversely, the loss of efficiency is accompanied by decreased growth rates on non-target substrates (bottom right quadrant of Figure 6b).The most notable exceptions of these trends stem from 3NT + experiments where growth of the evolved strains is hardly affected even if O 2 uncoupling increases to almost 60% (I204T, I204V with 2-nitrotoluene).These data points suggest that not every increase in O 2 uncoupling is directly translated into reduced growth in vivo probably due to other adaptations not assessed in the original adaptation studies.
A comparison of absolute inverse doubling times vs

■ CONCLUSIONS
Environmental microorganisms capable of oxidative biodegradation have been shown to evolve their enzymatic machinery upon exposure to xenobiotic compounds.Their biodegradation makes anthropogenic contaminants available as alternative sources of carbon and energy. 11,38,39,95,96To that end, adaptation processes have been postulated that would be initiated by unproductive O 2 activation concomitant with production of ROS during the first chemical steps of contaminant transformation.Many biochemical processes are indeed necessary to lead from O 2 uncoupling, ROS release, and ROS scavenging to beneficial mutations that ultimately allow for the expression of oxygenases with improved function.Yet, our work suggests that the changes of the efficiency of substrate oxygenation could hint at the occurrence of this complex adaptation process.Point mutations at the terminal oxygenase of 2NTDO acquired from two independent laboratory evolution experiments both led to variants that enabled evolved strains to grow on alternative substrates concomitant with a substantial efficiency improvement of the oxygenases toward these alternative substrates.Analysis of sequence data also confirms the relevance of these variants in an environmental context through the presence of all variants in environmental sequences with relatively high natural abundances and taxonomic distributions among 2NTDO homologues (Figure S12).The observation that these 2NTDO point mutations could be associated with substrate and O 2 transport tunnels furthermore supports the notion that an enzymatic reaction can become more efficient while maintaining the kinetic mechanism of enzyme catalysis.The same structural elements have, in fact, been identified as hotspots for engineering Rieske oxygenases not only toward altering C−H hydroxylation selectivity 92−94 but also for transformation of alternative substrates. 66The observation that evolving biodegradation potential and rational engineering can be related to amino acid substitution at the same structural entities offers a promising avenue to assign change in the Rieske oxygenase structure to its function.This kind of insight will be critical for making more causal connections between structural features of this class of enzymes and the biodegradability of organic soil and water contaminants.We note, however, that it remains to be elucidated how modifications in substrate and O 2 transport tunnels would modulate substrate binding affinities in the active site and the oxygenation efficiencies of target substrates only.

Figure 1 .
Figure 1.(a) Oxygenation reactions of nitrobenzene and nitrotoluene isomers catalyzed by 2NTDO and its variants.(b) Concentrations of 4nitrotoluene, dissolved O 2 , organic products, and NO 2− in enzyme assays of wt 2NTDO after complete consumption of different amounts of NADH.24 4-Nitrobenzylalcohol was not detected in any of the assays, whereas 4-methylcatechol was below the limit of quantification.The black lines and shaded areas represent linear fits with 95% confidence intervals with slopes shown in TableS7.The mass balance includes the concentrations of 4-nitrotoluene, NO 2 − , and 4-nitrobenzylalcohol. (c) Experiment with the M248I variant adapted to 4-nitrotoluene.Note that the concentrations of nitrite and 4-nitrocatechol are almost identical whereas 4-nitrobenzylalcohol remained undetected.(d) Efficiency of O 2 activation as the sum of oxygenation products NO 2 − and nitrobenzylalcohols vs O 2 consumed, ΔO 2 .The performance of the wild-type oxygenase (2NTDO, green) with 4-nitrotoluene as substrate is compared with variants M248I and L238V-M248I.Lines represent linear fits, their slopes correspond to the oxygenation efficiency f 1 Figure 1.(a) Oxygenation reactions of nitrobenzene and nitrotoluene isomers catalyzed by 2NTDO and its variants.(b) Concentrations of 4nitrotoluene, dissolved O 2 , organic products, and NO 2− in enzyme assays of wt 2NTDO after complete consumption of different amounts of NADH.24 4-Nitrobenzylalcohol was not detected in any of the assays, whereas 4-methylcatechol was below the limit of quantification.The black lines and shaded areas represent linear fits with 95% confidence intervals with slopes shown in TableS7.The mass balance includes the concentrations of 4-nitrotoluene, NO 2 − , and 4-nitrobenzylalcohol. (c) Experiment with the M248I variant adapted to 4-nitrotoluene.Note that the concentrations of nitrite and 4-nitrocatechol are almost identical whereas 4-nitrobenzylalcohol remained undetected.(d) Efficiency of O 2 activation as the sum of oxygenation products NO 2 − and nitrobenzylalcohols vs O 2 consumed, ΔO 2 .The performance of the wild-type oxygenase (2NTDO, green) with 4-nitrotoluene as substrate is compared with variants M248I and L238V-M248I.Lines represent linear fits, their slopes correspond to the oxygenation efficiency f 1 O uc 2 as in eq 2. Shaded areas indicate 95% confidence intervals of linear regressions.

2 ( 4 - 2 .
NT), for wt 2NTDO (green), as well as the M248I (blue) and L238V-M248I (pink) variants.The increasing oxygenation efficiency of 4-nitrotoluene in the order 2NTDO, M248I, and L238V-M248I corroborates the observation made by Ju and Parales95 who found that the single amino acid substitution at M248I was prerequisite for further adaptation to 4-nitrotoluene through the additional amino acid substitution L238V.The negative 1:1 line in Figure2astands for the gradual improvement of the oxygenation efficiency of enzyme variants toward 4-nitrotoluene from 6% with wt 2NTDO to 47 and 57% with M248I and L238V-M248I, respectively.The colored bars designating changes in oxygenation efficiency of the 4NT + experiment in Figure2aallow one to assess the relatively minor consequence of this adaptation for the other nitroaromatic substrates.While 2nitrotoluene and nitrobenzene were oxygenated slightly more inefficiently with increases of f O uc 2 by up to 0.10 (dashed lines in Figure2a), oxygenation of 3-nitrotoluene improved by a similar amount albeit at a substantially higher f O uc

Figure 3 .
Figure 3. CaverDock analysis of 2NTDO homology models reveals potential tunnels for the transport of the nitroaromatic substrates.Orange spheres indicate the non-heme Fe active site of (a) 2NTDO, (b) I204A, and (c) L238V-M248I.

Figure 5 .
Figure 5. (a) Catalytic cycle of wt 2NTDO and enzyme variants.Upon substrate binding to the 6-coordinate non-heme Fe II resting state 1, O 2 is bound concomitant with electron transfer from the reduced Rieske cluster (Fe II −Fe III ).O 2 uncoupling and release of unreacted substrate and ROS (H 2 O 2 ) (dashed arrows) occur from the Fe III -oxo-peroxo species 2. Substrate oxygenation can be catalyzed by species 2 and high-valent Feoxo(hydroxo) species 3. 18 O kinetic isotope effects ( 18 O-KIE) probe for processes involved in O 2 activation and are indicated with blue arrows.13 C-KIEs of nitroaromatic substrate hydroxylation steps are shown with green arrows.(b) 18 O-KIE values of O 2 activation in the various enzyme/ variant-substrate combinations vs O 2 uncoupling.Marker shapes stand for substrate type; colors denote enzyme or variant.(c) Corresponding data for 13 C-KIEs of substrate hydroxylation.

Figure 4
Figure 4 shows the interaction network of docked 4nitrotoluene into the homology model of variant L238V-M248I and the correlation of binding affinities with f O uc 2 1616O ratios of dissolved O 2 .

Table 1 .
Stoichiometric Coefficients, υ j , for O 2 Activation and Dioxygenation of wt 2NTDO and 4NT + Variants with Nitrobenzene and Nitrotoluenes as Well as the 13 C-KIEs and 18 O-KIEs of the Organic Substrates and Dissolved O 2 , , the doublesubstituted variant L238V-M248I formed more oxygenation products per O 2 consumed (ΔO 2 ) than both the wild-type and t h e s i n g l y s u b s t i t u t e d v a r i a n t M 248 I ( 2 = ± , Table 1, entry 12). a The corresponding data for 3NT + experiments are shown in Table S8.b NADH-normalized stoichiometry of (co)substrate consumption based on eq 1. c O 2 uncoupling based on eq 2. d Reproduced from Bopp et al. 24 e Reproduced from Pati et al.
Figure (a) Rates of nitrite formation by Acidovorax sp.J42 and evolved 4NT + strains (per mg of protein) normalized to value obtained for 3nitrotoluene by wt 2NTDO vs f O uc 2 from enzyme assays with wt 2NTDO and variants from 4NT + experiments for 4 substrates.(b)Changesofinversedoubling times, Δ(1/doubling time), of evolved 4NT + and 3NT + strains relative to inverse doubling times determined with Acidovorax sp.J42 strain and the 4 substrates vs changes of O 2 uncoupling, f O uc uncoupling even further.Indeed, our observations hint at adaptations that help handle oxidative stress, as growth rates of complemented Acidovorax strains carrying no mutations other than the ntdAcAd genes encoding 2NDTO (FigureS11) were generally lower.Our observations show that O 2 uncoupling above the 63% threshold was indeed limiting the adaptation of Acidovorax to new substrates.Both the inefficient use of energy for O 2 activation and the cost of dealing with elevated ROS levels are detrimental to cell growth.Narrowly confined f f O uc 2 for all strains carrying wt 2NTDO and variants as well as any of the four substrates in Figure 6c hints at approximately 60% of O 2 uncoupling as important threshold.Regardless of the enzyme (variant)-substrate combination, Acidovorax sp.strain JS42 and NT + strains were only able to grow with substrates that resulted in f O uc 2 < 0.63.All tested amino acid 2 , of variants relative to f

■ ASSOCIATED CONTENT * sı Supporting Information The
Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsenvironau.4c00016.Biological and chemical materials used, descriptions of experimental, analytical, and computational procedures for protein purification, enzyme assays, chemical, and isotopic analyses and protein structure evaluation, additional results on substrate oxygenation efficiency, tunnel identification, substrate docking, and enzyme kinetics; sequence alignment and similarities scores for nitrobenzene dioxygenase and wt 2NTDO; and comparison of oxygenation efficiencies with fitness parameters and analysis of taxonomic distribution of 2NTDO homologues (PDF)