Provision of a Redox Donor to Oxy-Ferrous PrnB is Required for Pyrrolnitrin Synthesis

Abstract

PrnB, a longstanding member of the recently reclassified heme dependent aromatic oxygenase (HDAO) superfamily of histidine-ligated heme enzymes, catalyzes the conversion of 7-chloro-L-tryptophan (7-Cl-Trp) to monodechloroaminopyrrolnitrin (MCAP). This unique ring-rearrangement is an essential step in the biosynthesis of the broad-spectrum antifungal pyrrolnitrin. The conversion of 7-Cl-Trp to MCAP by PrnB differs from other HDAOs which typically affect the mono- or di-oxygenation of aromatic substrates. However, the molecular basis for this transformation has remained enigmatic due to the inability to reconstitute enzymatic activity in vitro. Transient kinetic approaches reveal that the sequential binding of 7-Cl-Trp and dioxygen results in the formation of a long-lived oxy-ferrous complex. Despite similarity to intermediates found in other HDAOs, the PrnB-oxy ternary species does not react with bound 7-Cl-Trp. However, the provision of a surrogate redox delivery system supports turnover and single-turnover studies reveal that activation of the oxy-complex is required. Together, these studies reveal the molecular basis for functional expansion of the HDAO structural framework through the alteration of the strategy of oxygen-activation.

Graphical Abstract

INTRODUCTION

The microbial natural product pyrrolnitrin (Prn) was initially isolated over half a century ago,¹ and has been widely recognized for its broad antifungal activity. The Prn biosynthetic pathway was originally identified in Pseudomonas pyrrocinia and has been subsequently characterized in several other bacteria.^{2, 3} The biosynthetic gene cluster (BGC) consists of four enzymes: two flavin-dependent halogenases (PrnA and PrnC),^{4, 5} the heme enzyme (PrnB) that has been purported to convert 7-Cl-L-tryptophan to monodechloroaminopyrrolnitrin (MCAP), and PrnD, a Rieske and Fe²⁺ dependent oxygenase involved in nitro-group installation (Figure 1A).⁶

Figure 1.

(A) Biosynthetic logic for pyrrolnitrin synthesis. Chlorination of L-tryptophan is followed by decarboxylation and ring-rearrangement by PrnB to form monodechloroaminopyrrolnitrin (MCAP). Subsequent halogenation and amine-oxygenation furnishes the final pyrrolnitrin product. (B) The prototypical dioxygenase reaction performed by the heme-dependent aromatic oxygenases (HDAOs) tryptophan (TDO) or indoleamine dioxygenase (IDO) is mediated by a transient ferric-superoxide or “oxy” intermediate.

The reactions of PrnA, PrnC, and PrnD are readily rationalized on the basis of the relatively well-characterized transformations catalyzed by flavin-dependent halogenases^{7, 8} and mononuclear N-oxygenases.⁹ However, the underlying mechanism for the complex ring rearrangement and decarboxylation catalyzed by PrnB to furnish MCAP has remained enigmatic. Crystallographic characterization of several PrnB orthologs ^{10, 11} has revealed significant structural homology with other members of the histidine-ligated heme-dependent aromatic oxygenase (HDAO) family.^{12, 13} Prototypical HDAOs, previously designated as the tryptophan dioxygenase (TDO) family typically incorporate both atoms from atmospheric dioxygen into aromatic substrates through a dioxygenase reaction.^{14, 15} The recent reclassification of the family owes to the discovery of structurally verified (e.g. TyrH,¹⁶ SfmD¹⁷), or probable HDAO members on the basis of sequence similarity,^18-21 that perform monooxygenation on aromatic substrates.

Several mechanisms for PrnB have been proposed but the most commonly invoked pathway involves the reactive intermediate that is formed when the reduced heme-iron coordinates molecular oxygen, termed the “oxy“ or ferric-superoxide species.^{11, 22} The initial step for this proposed mechanism is primarily grounded in the known chemistry of the founding members of the HDAO superfamily and is typified by TDO and indoleamine (IDO) dioxygenases.^{14, 22} Investigation of the TDO and IDO reactions through transient kinetic approaches has revealed that both utilize a short-lived ferric-superoxide intermediate as a primary reactant to initiate substrate dioxygenation through what has been described as electrophilic or radical addition to the indole moiety. This general trajectory is shown in Figure 1B. As a result, a transient Fe⁴⁺-oxo species, termed compound II, is generated that is then poised to enable the installation of the second oxygen.^{14, 23, 24} The structural similarity of PrnB to IDO/TDO was previously used as primary grounds to suggest that analogous intermediates and reaction steps are involved, even if molecular oxygen is not formally incorporated into the final MCAP product. However, in vitro studies by the Naismith ¹¹ and Wang ²⁵ laboratories have shown that while ferrous PrnB appears to readily bind dioxygen to generate a ferric-superoxide adduct, the decay of this complex does not lead to product generation in the presence of 7-Cl-Trp. Although alluding to a very different mechanism than prototypical HDAOs, further dissection of this unique trajectory has been significantly limited by the inability to reconstitute the reaction in vitro and under defined turnover conditions. To date, PrnB has only been found to be active when supplemented with crude lysate from the host,^{10, 11} presenting significant questions regarding the substrate, transformation, mechanism for C2-N1 cleavage, and further steps needed to fashion MCAP.

In this work, we have begun to unravel the oxygen activation mechanism and (co)substrate requirements of PrnB using a combination of steady-state and rapid kinetics approaches. The ordered binding of substrate and O₂ to ferrous PrnB result in the formation of an extremely long-lived ferric superoxide intermediate that is characterized by optical and Mössbauer spectroscopies. Despite similarity in electronic structure with analogous intermediates formed by TDO and IDO, the PrnB-oxy intermediate is completely unreactive with the presumptive 7-Cl-Trp substrate. However, in vitro turnover activity is restored upon supplementation of a surrogate biological or chemical redox donor. Taken together, the data indicates that PrnB exhibits a co-substrate dependence and oxygen activation pathway that differs from other HDAOs that have been characterized to date. These studies highlight the biochemical basis for functional expansion of the HDAO framework, and more broadly, blurs previously held divisions between histidine and cysteine-ligated heme biocatalysts.

EXPERIMENTAL PROCEDURES

Reagents

Oxygen, nitrogen, argon, and hydrogen gases were supplied by Arc3. ⁵⁷Fe was purchased from Cambridge Isotope Laboratories, Inc. 7-Cl-Tryptophan was obtained from Chemscene, LLC. The MCAP standards were synthesized according to literature protocols.

PrnB expression and purification

A gene encoding an E. coli codon-optimized and N-terminal hexahistidine tagged PcPrnB (accession ID: A0A0D5XZW7) was synthesized and subcloned into a pet28 vector, between NdeI and BamHI, by Genscript. The full optimized coding sequence follows:

ATGGGCAGCAGCCATCATCATCATCATCACAGCAGCGGCCTGGTGCCGCGCGGCAGCCATATGGAACGCACCTTGGACCGGGTAGGCGCATTCACAGTCACGCACGCTGCCGTGGCGGCCTGTGATCCGTTGCAGGCGCGCGCGCTCGTTCTGCAGCTGCCGGGCCTGAACCGTAACAAGGACGTGCCGGGCATCGTCGGCTTGTTGCGCGAGTTCCTCCCGGCGCGCGGCGTGCCCTCCGGCTGGGGCTTCGTCGAAGCCGCCGCCGCGATGCGGGACATCGGGTTCTTCCTGGGGTCGCTCAAGCGGCACGGACACGAGCCCGCGGAGGTGGTGCCCGGCCTTGAGCCGGTGCTGCTCGACCTGGCGCGCGCGACCGACCTGCCGCCGCGCGAGACACTCCTGCATGTGACGGTCTGGAACCCCGCGGCGGCCGACGCGCAGCGCAGCTACACCGGGCTGCCCGACGAAGCGCACCTGCTCGAGAGCGTGCGCATCTCGATGGCGGCCCTCGAGGCGGCCATCGCGGTGACCGTCGAGCTGTCCGATGTCTCCCTGCGGTCGCCCGCGTTCGCGCAAGGGTGCGACGAGTTGGAAGCCTATCTGCAGAAAATGGTCGAATCGATCGTCTATGCGTACCGCTTCATCTCGCCGCAGGTCTTCTACGATGAGCTCCGCCCCTTCTACGAACCGATTCGAGTCGGGGGCCAGAGCTACCTCGGCCCCGGTGCCGTAGAAATGCCCCTCTTCGTGCTGGAGCACGTGCTGTGGGGCTCGCAATCGGACGACCAAGCTTATCGAGCATTCAAAGAGACGTACCTGCCTTACGTGCTTCCCGCGTACAGGGCGGTCTACGCTCGGTTCTCCGGAGAGCCGGCGCTCATCGACCGCGCGCTCGACGAGGCGCGAGCGGTCGGTGCGCGGGACGAGCACGTCCGGGCTGGGCTGACAGCCCTCGAGCGGGTCTTCAAGGTCCTGCTGCGCTTCCGGGCGCCTCACCTTAAATTGGCGGAGCGGGCGTACGAAGTCGGGCAAAGCGGCCCCACAATCGGCAGCGGGGGGTACGCGCCCAGCATGCTCGGTGAACTACTCACGCTCACGTATGCCGCGCGGTCCCGCATCCGCGCCGCGCTCGACGAATCCTGA

The protein was heterologously expressed in Escherichia coli BL21 (DE3) cells grown in LB media. The original pChuA expression vector was obtained from Addgene (plasmid #42539) and was a kind gift from Alan Jasanoff.²⁶ The ChuA gene was subsequently subcloned into pet21b. Co-expression of PrnB with the heme-importer ChuA was found to substantially improve heme loading. Cells from an overnight culture were used to inoculate 2 L shaker flasks of LB medium with antibiotic selection provided by kanamycin (50 μM) and ampicillin (100 μM). At an optical density OD₆₀₀ ~ 0.1, the cells were supplemented with 50 μM FeCl₃, 25 μM delta-aminolevulinic acid, and 5 mg/L hemin. When cells reached an OD₆₀₀ ~ 0.6, the temperature was lowered from 37 to 18°C, allowed to equilibrate temperature for 1 hour, and cells were induced with 200 μM IPTG and incubated for an additional 16 hours. Cells were harvested by centrifugation (9000xg,10 min) and either stored at −80°C for future use or resuspended directly in 300 mL of buffer A (50 mM Tris pH 8.0, 100 mM NaCl) containing 10 mM imidazole. The cells were lysed via sonication with a Branson sonifier with continuous stirring on ice. The crude lysate was clarified by centrifugation (35000xg, 45min) and the supernatant was loaded onto a nickel nitriloacetic (Ni-NTA) column equilibrated in buffer A. The column was subsequently washed with >10 column volumes of buffer A supplemented with 20 mM imidazole. Protein was eluted from the column with buffer A containing 250 mM imidazole. Fractions containing PrnB were pooled and dialyzed twice against buffer A for a minimum of 4 hours at 4°C. Protein concentration was determined based on the extinction coefficient ε₄₀₅ = 115 mM⁻¹ cm⁻¹determined by the pyridine hemochrome assay.²⁷ Protein homogeneity was verified based on the Reizenstahl (R_z) ratio (A_{406 nm}/A_{280 nm} ≥ 2) and by SDS-PAGE following purification (Figure S1). Purified PrnB was concentrated to 0.5-1.5 mM with 30 kDa centrifugal filters (Millipore) before being aliquoted, flash-frozen, and stored at −80°C for future use.

Equilibrium substrate binding

The binding of 7-Cl-Trp to ferric or dithionite reduced ferrous PrnB (~10 μM) was monitored by following the absorbance changes at 412 nm or 426 nm for the ferric and ferrous enzymes respectively. The absorbance changes were normalized and plotted against [7-Cl-Trp]. Due to the high-affinity (i.e. low dissociation constant (K_D)), data were fit using the Morrison quadratic expression²⁸ that accounts for the depletion of the substrate due to binding to the enzyme.

PrnB substrate-binding kinetics

PrnB protein stocks were prepared at a concentration of 10 μM. The protein was transferred into a Coy anaerobic chamber and allowed to incubate while gently stirring for 2 hours. A small excess of anaerobically prepared sodium dithionite was introduced, and the ferrous protein was transferred via a Hamilton gastight syringe to the stopped flow that was made anaerobic by sodium dithionite washes followed by rinsing with an anaerobic buffer. Substrate binding to the ferrous enzyme was monitored by rapidly mixing with different concentrations of 7-Cl-Trp prepared in buffer A. The ferrous enzyme was mixed with substrate at 4 °C using an Applied Photophysics SX20 stopped flow equipped with either a photodiode array detector (PDA) or photomultiplier tube (PMT) attachment. The absorbance changes at 558 nm were adequately fit to a single exponential expression and the k_obs versus [7-Cl-Trp] plots revealed a linear dependence. The slope and intercept of the plot were used to determine the on-rate (k_on) and off-rate (k_off) respectively.

Spectroelectrochemistry

Redox potential measurements of PrnB in the presence or absence of saturating 7-Cl-Trp were performed in an anaerobic Coy glovebox at room temperature. A 30 μM solution of PrnB was deoxygenated by purging with argon and subsequent stirring in the anaerobic chamber for 45 minutes. The following redox mediator dyes were added prior to the start of the experiment: 0.2 μM of riboflavin (−235 mV), and 2 μM of each 2-Hydroxy-p-naphthoquinone (−145 mV), methylene blue (+10 mV), phenazine ethosulfate (+55 mV), phenazine methosulfate (+82 mV), and benzoquinone (+684mV). A Honeycomb spectroelectrochemical cell (Pine Research) was used to perform the experiments. The anaerobic solution of PrnB was added to a 1.7 mm pathlength cuvette with a gold Honeycomb electrode and Ag/AgCl reference electrode. The setup was connected to a Pine WaveNow wireless potentiostat. Potentials were applied in 20-30 mV increments and spectra were recorded on an Avantes spectrophotometer after equilibrium was reached at each applied potential, as determined by stabilization of the spectra. The percentage of oxidized PrnB was calculated at each potential which were corrected by +197 mV for the Ag/AgCl reference. Midpoint potentials of PrnB in the presence and absence of substrate were determined by fitting the data to the Nernst equation:

$$\mathrm{E}={\mathrm{E}}_0-25\text{ln}\frac{[\text{ox}]}{[\text{red}]}$$

Oxy-ferrous formation and decay kinetics

The sample preparation of ferrous PrnB is described above. In the case of the substrate-bound ferrous PrnB, 7-Cl-Trp was added anaerobically and in excess (>20-fold relative to the protein). The ferrous protein (~10 - 20 μM) was mixed rapidly with oxygen-saturated buffer at 4°C, and PDA traces were collected over a period of 1500 – 8000 s. Reaction timecourses of oxy-ferrous formation were monitored in Soret band region (411 nm) for the substrate-free protein or the Q-band region (580 nm) for the substrate-bound protein respectively. Decay rates were fit at 424 nm. Time-courses for were accurately fit to the following general exponential expression with well-separated reciprocal relaxation times which is appropriate for a series of first-order reactions:

$${\mathrm{A}}_{\mathrm{t}},\text{obs}={\mathrm{A}}_{\max }+\sum _{\mathrm{i}=1}^{\mathrm{n}}{\mathrm{a}}_{\mathrm{i}}{\mathrm{e}}^{-\mathrm{t}∕{\mathrm{t}}_{\mathrm{i}}}$$

${\mathrm{A}}_{\mathrm{t}}$ is the measured absorbance, ${\mathrm{A}}_{\max }$ is the maximum absorbance, ${\mathrm{a}}_{\mathrm{i}}$ is the spectral amplitude of phase $\mathrm{i}$, $\mathrm{t}$ is time (s), and $1∕{\mathrm{t}}_{\mathrm{i}}$ is reciprocal relaxation time (s⁻¹), and n is the number of phases. In the absence of 7-Cl-Trp, the timecourse can be fit to a two-summed (n=2) exponential expression. O₂-concentration studies were conducted under pseudo-first order (> 10 molar equivalents of O₂ relative to protein). The first (much faster) phase exhibits an O₂-concentration dependence and corresponds to oxy-ferrous PrnB formation, whereas the slow O₂-independent phase (k₂) corresponds to the rate of autoxidation. The formation phase (k_obs) versus O₂ plots revealed a linear dependence. In this scenario, the y-intercept and slope of the plot corresponds to the off- (k_off) and on-rate (k_on) of O₂-binding, and the apparent K_D^O2 is provided by k_off/k_on. When the substrate is present, fitting instead required a three-summed exponential expression. In this case, none of the three phases exhibit any appreciable O₂ concentration dependence. Possible interpretations are discussed in the Results section.

Redox partner expression and purification

For spFd expression, a pET-28a(+) plasmid containing the gene encoding for hexahistidine-tagged protein was transformed into E. coli BL21(DE3) cells harboring the ISC system to enable iron assembly, which was a generous gift from the Dean lab.²⁹ Cultures were grown at 37 °C until they reached an OD₆₀₀ of 0.6-0.8, upon which the temperature was reduced to 20°C and 100 μM L-cysteine and 100 μM FeCl₃ were added. Fd expression was induced with the addition of 0.2 mM IPTG. SpFdR was expressed by transforming the pET-28a(+) plasmid containing the hexahistidine-tagged gene encoding spFdR into E. coli BL21(DE3) cells and plating on LB agar plates containing 50 μg/mL kanamycin. An overnight starter culture was used to inoculate 1L cultures of LB media containing 50 μg/mL kanamycin. Cultures were grown at 37 °C until they reached an OD₆₀₀ of 0.8-1, upon which the temperature was reduced to 20 °C and 50 μM riboflavin was added. spFdR expression was induced by the addition of 0.2 mM IPTG. Purification of spFd and spFdR was performed using Ni-NTA affinity chromatography using 50 mM Tris pH 7.5 100 mM NaCl (buffer B). Proteins were dialyzed against 50 mM Tris pH 8.0 and concentrated with Millipore centrifugal filters (10K MWCO). The purity of Fd was determined by the A₄₂₀/A₂₈₀ ratio and exceeded 0.4. The purity of FdR was determined by the A₄₅₄/A₂₈₀ ratio. Rat cytochrome b₅ (b₅), and rat cytochrome P450 reductase (CPR), were expressed and purified using published protocols.^{30, 31} Protein concentrations were determined using the following extinction coefficients: ε_420nm = 9.8 mM⁻¹ cm⁻¹ (sp Fd), ε_454nm = 10 mM⁻¹ cm⁻¹ (sp FdR), ε_410nm = 130 mM⁻¹ cm^{−1 30}(b5), ε_455nm = 23.5 mM⁻¹ cm^{−1 31}(CPR).

Mössbauer sample preparation and spectroscopy

⁵⁷Fe-PrnB was expressed using M9 minimal media and Fe⁵⁷ metal, dissolved in aqua regia, was added upon induction to a final concentration of 50 μM. Hemin was excluded from the growth medium. The yield of protein and extent of heme loading (R_Z~1.1) decreased using this protocol. After purification, ⁵⁷Fe-PrnB was made anaerobic, reduced, and desalted using a PD10 column (Cytiva) and concentrated to 1 mM. To generate the enzyme-substrate complex, 2 mM substrate was added to before aliquoting and diluting with an equal volume of anaerobic buffer. The ferrous protein was either flash-frozen in Mössbauer cups using liquid N₂ or mixed 1:1 with oxygen-saturated buffer (1.95 mM in O₂ pre-mix) and incubated for 40 s at 4 °C prior to flash-freezing. Custom spectrometers equipped with Janis Research SuperVaritemp dewars (Wilmington, MA) were employed to measure the Mössbauer spectra, allowing for studies with an applied magnetic field up to 8.0 T. Spectral simulations were conducted using the WMOSS software package (SEE Co., Edina, MN), with isomer shifts referenced to Fe metal at 298 K. The preparation of all Mössbauer figures was carried out using SpinCount software.³²

Synthesis of monodechloroaminopyrrolnitrin (MCAP)

MCAP was prepared following a previously reported procedure according to the scheme shown below.³³ NMR spectra can be found in the Supporting Information.

¹H NMR (500 MHz, d₆-DMSO) δ 11.06 (br s, 1H), 7.10 (d, J = 7.0 Hz, 1H), 7.10 (d, J = 7.0 Hz, 1H), 7.03 (s, 1H), 6.89 (s, 1H), 6.61 (dd, J = 6.5 and 6.5 Hz, 1H), 6.3 (s, 1H), 4.83 (br s, 2H); ¹³C NMR (125 MHz, d₆-DMSO) δ 140.6, 127.8, 126.3, 123.6, 120.1, 118.6, 118.2, 117.4, 116.3, 107.3. HRMS (ESI) calcd for C₁₀H₉ClN₂ [M+H]⁺ 193.0527, found 193.0528. The analytical and NMR data, shown in Figure S2, were in accordance with those previously reported.^{33, 34}

Activity screening

Stocks of CPR, b₅, spFdR, and spFdR with spFd were prepared at 20 μM. A PrnB stock (10 μM) was premixed with 4 mM 7-Cl-Trp and was divided into two stocks, where either bovine catalase (~15,000 unit/mL) was added or omitted. A 2x stock of an NADPH regenerating system (Promega) was brought to room temperature and prepared immediately prior to the initiation of the reaction. Each final reaction (100 μL total) consisted of 2.5 μM PrnB, 5 μM of redox partners, and 1x of the NADPH generating system to initiate the reaction. Reactions with ascorbate or H₂O₂ were prepared at similar PrnB concentrations and included 10 mM ascorbate and 10 μM methylene blue or H₂O₂ at various concentrations as indicated. Additional peroxide reactions were tested (varying [H₂O₂] or the rate of addition to prevent possible degradation) but did not result in appreciable changes in turnover number. Reactions were incubated for 1 hour at room temperature before being quenched with 100 μL acetonitrile and were vortexed for 30 seconds prior to HPLC analysis.

Steady State Activity

To determine the activity of purified PrnB, steady-state turnover reactions were conducted using spinach ferredoxin at various concentrations. The concentration of spFdR was maintained at 20 μM, and catalase was added to prevent peroxide-mediated heme degradation during the reactions. Reactions were set up with 1 μM PrnB pre-incubated with 200 μM 7-Cl-Trp, and all necessary components were added to the reaction stocks. The reactions were initiated by the addition of an equal volume of premixed NADPH regenerating system and allowed to proceed for 30 minutes. At the end of the incubation period, the reactions were stopped by the addition of an equal volume of acetonitrile and vortexed. A similar experimental setup was used to evaluate the activity of PrnB as a function of 7-Cl-Trp concentration, except that the ferredoxin and reductase concentrations were fixed at 5 μM and 2 μM, respectively. In the indicated reactions, catalase was added to a concentration of (~15,000 unit/mL) to prevent enzyme inactivation.

Single turnover Activity

A 1 mM PrnB stock was made anaerobic and desalted as described previously before addition of excess substrate. An anaerobic stock of reduced spFd was similarly prepared. PrnB was oxygenated for 30 s and then reacted with buffer or reduced spFd at various concentrations before the sample was quenched and analyzed for MCAP formation.

HPLC Analysis

Samples quenched in acetonitrile were thoroughly vortexed and centrifuged at 18000xg for 45 minutes. The supernatant was collected for HPLC analysis. Samples were analyzed using a Shimadzu Nexera 40 series High-Performance Liquid Chromatography (HPLC) system and separation was performed on an Agilent Zorbax Extend-C18 column (4.6 x 50 mm, 1.8 μm) . The mobile phase was 35 % acetonitrile, 1% formic acid (v:v) and the flow rate was 0.4 mL/min. Metabolites were detected using Agilent Technologies 6120 single quadrupole or 6530 quadrupole time of flight mass spectrometer under electrospray ionization in positive ion mode (ESI+). The MCAP product was verified via comparison (retention time and m/z) to an authentic MCAP standard. Product quantitation was performed by comparison of the extracted ion chromatograms at (m/z = 193) with an MCAP standard curve.

Electron transfer kinetics

Concentrated solutions of reduced spFd were prepared as previously described. The solutions were then diluted to 30 μM spFdR and 25, 50, or 100 μM spFd with 50 mM Tris, pH 8, 100 mM NaCl. A solution of PrnB (10 μM) was made anaerobic and 500 μM 7-Cl-Trp was added. PrnB was rapidly mixed with reduced spFd, and reduction of PrnB was monitored by the formation of the Fe²⁺ heme at 426 nm.

RESULTS AND DISCUSSION

Regulation of substrate binding in a PrnB ortholog from P. chlorographis

Previous biophysical and structural studies of PrnB from Pseudomonas fluorescens BL915 (described hereafter as PfPrnB) utilized a P. fluorescens expression system due to the reported difficulties in heterologous expression in E. coli.^{10, 11, 35} Further modifications included the elimination of three surface cysteines to prevent protein aggregation and/or enable crystallization. With an eye towards enabling the high-concentration spectroscopic studies described below, we selected a putative PrnB ortholog from the known pyrrolnitrin producer Pseudomonas chlorographis (PcPrnB). Overall, PcPrnB has 95 % sequence identity to PfPrnB with variations that are exclusively localized to the protein surface (Figure S3). The ortholog fortuitously lacks several of the problematic surface cysteines. The sequence identity and genomic context of PcPrnB, with the gene placed within a similarly structured operon as that found for P. fluorescens, suggests it is functionally equivalent (Table S1). Co-expression of hexahistidine-tagged PcPrnB with the heme importer ChuA ³⁶ in E. coli resulted in high yields of purified protein with a Reinheitszahl (R_Z) ratio (A₄₀₇/A₂₈₀ >2) indicative of a high-level of heme incorporation (Figure S1).

To ensure that PcPrnB binds the presumptive 7-Cl-L-Trp substrate in a similar fashion as PfPrnB, the optical absorbance changes associated with binding to the ferric or ferrous protein were followed through titration studies. For the ferric enzyme, a shift of the Soret maximum (from 407 to 412 nm), and an accompanying decrease in the Q-band absorption at 630 nm, occurs that has been attributed in several PrnB orthologs, as well as other HDAOs as deriving from a shift from an admixture of high- (HS) and low-spin (LS) species to a predominantly LS hexacoordinate form (Figures 2A, S4). In the case of IDO, these changes are thought to arise from the formation of a low-spin, hydroxide bound heme,³⁷ while in PrnB, they have been attributed to direct ligation of the substrate amine to the heme-iron, as verified both through structural and spectroscopic (e.g. resonance Raman (rR) and electron paramagnetic resonance (EPR)) studies.^{10, 25} Binding of 7-Cl-Trp to the ferrous protein was also probed. The addition of 7-Cl-Trp to PcPrnB results in a shift of the Soret band from 431 nm to 425 nm and a marked increase in its intensity (Figures 2B, S4) with the concomitant emergence of prominent Q-bands at 530 and 560 nm. These features are consistent with assignment as a hexacoordinate ferrous heme, consistent with recent rR studies by Li et al.,²⁵ and markedly differ from other HDAOs, which are typically pentacoordinate in the reduced state.^{38, 39} Fitting of the titration data to a quadratic Morrison expression revealed a high-affinity with a measured K_D = 0.13 ± 0.03 μM and 1.9 ± 0.3 μM for the ferric and ferrous proteins respectively.

Figure 2:

Optical spectroscopic changes associated with 7-Cl-tryptophan binding to PcPrnB in the ferric (A) and ferrous (B) states. The optical changes in the Soret region were used to measure substrate dissociation constants (K_D) in titration studies (inset) through fitting to a Morrison equation. (C) Spectroelectrochemical measurements of PcPrnB in the substrate-free (red) and substrate-bound (blue) states. Open and closed symbols refer to titrations done in a reductive or oxidative direction respectively.

Redox potential measurements were used as an additional means to discriminate the substrate binding mode of PcPrnB from other HDAOs. As documented for both IDO and TDO, the binding of their respective substrates often results in an increase in the midpoint redox potential (E°_M) which serves to stabilize the catalytically active ferrous form for successive turnovers.^{12, 37} In the case of PcPrnB, the inclusion of saturating amounts of 7-Cl-Trp resulted in a 150 mV decrease in the apparent midpoint potential E°_M (from −74 to −230 mV) (Figure 2C). This substrate-linked decrease in the apparent potential mirrors that previously found for PfPrnB (80 mV decrease) ¹⁰, emphasizing PrnB as an outlier within the HDAO superfamily.

PrnB forms a long-lived oxy-complex

We next sought to characterize, based on analogy to the HDAO dioxygenases, the next presumptive step of the catalytic cycle. The reaction of the ferrous form of the enzyme with O₂ was followed by stopped flow (SF) absorption spectroscopy. Upon rapid mixing of ferrous substrate-free PrnB with excess dioxygen, the enzyme rapidly transforms (within 10 ms with 100 μM O₂) to an intermediate with Soret band at 416 nm and Q-band maxima at 542 nm and 578 nm respectively, characteristics that are similar to other oxyferrous HDAOs,^{23, 24, 40-42} and prior studies on PrnB ^{11, 25} (Figure 3A). The formation phase could be adequately fit to a single exponential expression (Figure S5). Varying O₂ yielded a linear concentration dependence for oxy-PrnB formation that is consistent with a straightforward bimolecular process. The resulting fits of the slope and intercept of the plot yielded a k_on and k_off of 2.64 ± 0.09 μM⁻¹ s⁻¹ and 71.2 ± 10.9 s⁻¹ and a resultant K_D^app for O₂ ~ 30 μM (Figures 3A). The measured affinity is a similar magnitude as that estimated for K_M ^O2 obtained in steady-state turnover studies of TDO ⁴² and IDO,⁴³ as well as binding of the O₂ mimic cyanide (CN) ligand to substrate-free PfPrnB.¹¹

Figure 3:

Photodiode array spectra of oxygen-binding to substrate-free (A) and 7-Cl-Trp bound (B) ferrous PrnB at 4°C. The rate of formation at 411 nm and 580 nm for the enzyme (E) and enzyme substrate (E-S) complexes respectively were fit and plotted against O₂ concentration (inset). Representative time traces and results from non-linear fitting are shown in Figures S5 and S6. The O₂ concentration dependence was fit to a linear expression in the case of the substrate-free enzyme, yielding a K^D_O2 ~ 30 μM.

In the presence of 7-Cl-Trp, the ferrous enzyme reacts with O₂ to form a very similar species (Figure 3B). However, the rate for its formation is over three orders of magnitude slower (Figure S6). Examination of the rate of formation versus O₂ under accessible pseudo-first order conditions (≥ 100 μM in O₂) no longer reveals a significant concentration dependence (Figure 3B, inset). This suggest that either the K_D for O₂ is substantially lower in the presence of 7-Cl-Trp (and thus inaccessible within the experimental constraints), or that a fast and essentially irreversible step (e.g. dissociation of the substrate-amine from the heme-iron) precedes the formation of the ternary oxy-complex. In previous examinations of cyanide (CN) binding to PfPrnB via isothermal calorimetry, Naismith and colleagues invoked a sequential binding mechanism for formation of an enzyme:substrate:CN complex.¹¹ A similar ordered binding event has also been cited for the regulation of O₂ binding to human indoleamine dioxygenase (hIDO),⁴⁴ and more recently by Li et al for two presumptive PrnB orthologs.²⁵

Given the surprising stability of the presumptive PrnB-oxy complex, we performed Mössbauer spectroscopy on an ⁵⁷Fe enriched E:S complex before and after exposure to O₂ to further confirm its assignment and ascertain if it differs from other oxy-HDAO species. A comparison of the Mössbauer spectra taken in the absence (0 mT) and presence of a high (7 mT) magnetic field of the ferrous and oxygenated forms is shown in Figure 4. The spectra of ferrous PrnB can be fit to a single doublet, comprising ~98% of the total Fe, with δ = 0.48 mm/s and ΔE_Q = 1.07 mm/s (Figure 4 and B). The values are indicative of a hexacoordinate LS ferrous site with an S = 0 spin-state and similar to the parameters observed for other ferrous hexacoordinate cytochromes (e.g. bis-His cytochrome c3 and His/Met cytochrome peroxidase (Table 1) ^{45, 46} and comport with the recent PrnB rR study.²⁵ Upon exposure to dioxygen, the Mössbauer transforms nearly quantitatively (92 % of the total Fe) to a new species with δ = 0.27 mm/s and ΔE_Q = 2.33 mm/s (Figure 4, spectra C and D). These features, and in particular, the δ that lies between 0.2 - 0.3 mm/s, are diagnostic for heme ferric-superoxide (S=0) complexes and similar to other histidine-ligated oxy-ferrous heme complexes (Table 1).^{41, 47, 48}

Figure 4:

Mössbauer spectroscopy of ⁵⁷Fe PrnB. Dithionite reduced and 7-Cl-Trp bound ferrous protein in the absence (A) or presence (B) of a 7 mT magnetic field. Spectra of PrnB after mixing the ferrous enzyme-substrate complex with an equal volume of O₂-saturated buffer in the absence (C) or presence (D) of a 7 mT magnetic field.

Table 1:

Comparison of PrnB Mössbauer spectroscopic parameters with various histidine ligated heme proteins in the ferrous and oxygenated states

Ferrous hemeproteins
protein	substrate	ΔEQ (mm/s)	δ (mm/s)	Ref
PrnB	7-Cl-Trp	+1.07	0.48	This work
cmTDO	Trp	3.43x	0.99	41
		2.39x	0.92
		1.28x	0.41
Myoglobin	None	−2.22	0.92	49
Hemoglobin	None	−2.4	0.925	49
		1.07a,x	0.46
Cytochrome c3 b	None	0.98x	0.43	46
Cytochrome c peroxidase c	none	+1.23	0.46	45
Oxy hemeproteins
protein	substrate	ΔEQ (mm/s)	δ (mm/s)	Ref
PrnB	7-Cl-Trp	−2.33	0.27	This work
cmTDO	Trp	2.28x	0.25	41
Sperm whale Mb	None	−2.29	0.26	50
Horse Mb	None	−2.36	0.26	50
Human Hb	None	−2.21	0.27	50
HRP	None	−2.31	0.23	47
P450cam	d-camphor	−2.15	0.31	51

^aThe second set of doublets represents ~10% of the total iron and is proposed to represent a population of the protein with a distal histidine ligating the heme iron.

^bCyt c3 is hexacoordinate and ligated by two histidine ligands.

^cCytC peroxidase is hexacoordinate and ligated by a histidine and a methionine.

^xThe sign of ΔE_Q is not determined.

The most remarkable facet of the PrnB-oxy adduct is its profound stability (t_1/2~1000 s) particularly in the presence of substrate. Rather than accelerating its decay, the inclusion of saturating amounts of 7-Cl-Trp results in a modest (~ 3-fold) decrease in the rate of oxy- decomposition (Figure 5). Although the prolonged lifetime could in principle be attributed to the presence of stabilizing interactions from the enzyme (i.e. an oxyanion hole formed by the protein backbone) and substrate, the kinetic behavior for the PrnB-oxy complex significantly differs from other HDAOs where the analogous intermediate represents the primary oxidizing species that reacts with substrate.^{14, 23, 41, 52} Examination of the PrnB-oxy decomposition process reveals that it decays directly to the ferric-enzyme, a behavior that is consistent across the tested O₂ concentration range. In IDO and TDO, completion of the catalytic cycle restores the ferrous-resting state, which is then poised to react with O₂ for subsequent turnover. As a result, the corresponding kinetics of TDO/IDO oxy-complexes can exhibit multiphasic formation kinetics and sometimes reveal the presence of additional intermediates indicative of a successful turnover event (e.g. Fe²⁺ or Compound II).^{40, 41, 53}. In contrast, the decay process of oxy-PrnB, which decays directly to the ferriheme in the presence of 7-Cl-Trp, suggests there are no additional intermediate(s) that accumulate. Rather, the absence of any discernible spectral signature for MCAP (λ_max ~ 320nm), nor chemical detection of the product, further confirms that the intermediate decomposes through an unproductive autoxidative process.

Figure 5:

Decay of the PrnB-oxy intermediate in the (A) presence, and (B) absence of excess 7-Cl-Trp. The decomposition kinetics at 424 nm are shown in the inset and fit to a single (A) or two-summed (B) exponential expression. Data and fits are shown in red and black respectively.

Addition of a redox donor to oxy-PrnB restores activity.

Unlike CYPs that return to the ferric resting-state at the completion of catalysis, TDO and IDO regenerate the ferrous state following substrate dioxygenation. As a result, only the initial input of an electron is required to initiate successive rounds of turnover. A physiological redox partner for TDO has yet to be identified. However, there is some evidence that cytochrome b₅ can act as a redox donor for IDO.^{54, 55} Reappraisal of the catalytic activity of the HDAO hydroxylases SfmD ¹⁷ and TyrH ⁵⁶ reveal that O₂ and ascorbate, or H₂O₂, can both support product formation. However, much higher turnover is observed with the latter. IDO can also use a peroxide-shunt reaction for the monooxygenation of indoles.⁵⁷ At least some of the differences in cofactor preference is likely attributed to the necessity to access a more activated intermediate (e.g. Fe³⁺-(hydro)peroxo or Fe⁴⁺-oxo) required for C-H activation. However, some interchangeability may be due to the generation of an oxy-complex from the reaction of Compound II with excess H₂O₂ through a catalase mechanism.^{16, 17, 58} Various oxidants and redox partner systems were tested with PrnB and reaction products were measured using liquid chromatography through comparison to an MCAP standard. The reaction of PrnB with H₂O₂ only yielded a meager and sub-stoichiometric (<1 % MCAP relative to the enzyme) amount of product (Figure 6A). Likewise, a combination of several chemical (ascorbate/dye mediator) and biological surrogate redox systems (e.g. cytochrome b₅, cytochrome P450 reductase (CPR), and NADPH) and O₂ used for other HDAOs failed to produce any appreciable MCAP. However, a turnover system consisting of spinach ferredoxin (spFd), spinach ferredoxin reductase (spFdR), and NADPH resulted in the production of a new metabolite with a retention time and m/z = 193 (m/z −46 relative to 7-Cl-Trp) consistent with assignment as MCAP (Figures 6A, 6B). Omission of individual components of the turnover system resulted in a loss of activity, suggesting that a full redox chain is required (Figure 6B). The activity obtained by the spFd/spFdR/NADPH system was further enhanced (~5-fold) by the addition of catalase, strongly suggesting that O₂ (rather than H₂O₂) was used as a co-substrate to furnish MCAP. This is consistent with rapid mixing studies of PrnB and H₂O₂ that have been shown to lead to heme degradation.²⁵ Analysis of the rate of MCAP formation, either as a function of substrate or spFd concentration, revealed saturation kinetics. Notably, the apparent K_M for spFd (K_M ~ 20 μM) is similar to dissociation constants that have been measured for other redox partner:heme (e.g. ferredoxin:P450) pairs, which are typically in the low μM range (Figure 6C).⁵⁹

Figure 6:

MCAP formation using various redox donor combinations and O₂, or H₂O₂ with PrnB and 7-Cl-Trp in the presence and absence of catalase (A). Extracted-ion chromatograms monitored at m/z = 193 of the reactions containing the indicated turnover system components and comparison to a synthetic MCAP standard, top trace (B). Kinetics of MCAP formation upon variation of spFd or 7-Cl-Trp (C and inset). Yield of MCAP formed after mixing the PrnB:O₂:7-Cl-Trp ternary complex with buffer or chemically reduced spinach ferrodoxin (D).

To further evaluate the origin for the restorative effect of spFd, the ternary PrnB:7-Cl-Trp:O₂ complex was generated and either mixed with buffer or spFd. As anticipated from the transient kinetics experiments, HPLC analysis revealed that no MCAP was formed from autoxidation. The addition of oxidized spFd to oxyPrnB also failed to restore activity, confirming that MCAP formation did not originate simply from an allosteric effect imposed through spFd binding. In contrast, the addition of reduced spFd resulted in MCAP in a concentration dependent manner (Figures 6D). This result confirms that a reduction of the oxy-complex is required for turnover. To provide further evidence that the surrogate redox donor served as a competent electron transfer reagent, reduced spFd was anaerobically mixed with substrate-bound ferric PrnB. The appearance of reduced PrnB, and electron transfer rate (k_ET approaching 100 s⁻¹ at high concentrations of spFd) shows that spFd can serve as a redox role (Figure S7).

Conclusions

Mechanistic Implications.

Ever since its discovery several decades ago, PrnB has been viewed as an outlier within the HDAO family. However, the biochemical basis for its deviation from prototypical HDAO dioxygenases, and more recently identified hydroxylases, has remained obscured by the inability to reconstitute its activity under defined conditions. We have shown here that PrnB does not absolutely require an unidentified cofactor (nor downstream biochemical process) to initiate catalysis. Taken together, the recovery of activity when a reduced biological cofactor is supplied to oxy-PrnB in single turnover experiments and demonstrated ability of spFd for competent electron transfer to the ferriheme, is consistent with a process involving reduction of the oxy-complex to enable catalysis. Although the turnover or PrnB (~20/hr) is quite sluggish compared to those of many other HDAOs, this inefficiency is very likely attributed to the use of a non-native redox partner that is highly uncoupled (~95 %) to H₂O₂ and ultimately leads to the inactivation of PrnB (Figure S8). Unfortunately, a native redox-donor system is not found within the PrnB operon, and wider screening efforts to find a more compatible biological redox donor system are clearly warranted to improve turnover efficiency.

The requirement for a second electron transfer step would in many ways mirror the scenario found in cytochrome P450s, where such a process enables the proton-linked formation of potent reactive electrophilic species (e.g. hydroperoxo ⁶⁰ or oxo ⁶¹). A very early catalytic scheme for PrnB involving a process that involves reduction of the oxy-complex is shown in Figure 7. Liu and colleagues have implicated oxidants downstream of oxy, such as Fe(III)-OOH or an Fe(IV) oxo, may be operant in the native hydroxylation as well as non-native C-F activation reactions catalyzed by the HDAO TyrH. ¹⁶ For PrnB, it has been proposed that cryptic oxygen installation (e.g. at the indole C3) followed by ring rearrangement, decarboxylation and loss of the installed oxygen, could be envisioned to furnish MCAP. While entirely plausible, a carbocationic mechanism that results from hydrogen atom transfer (via a Compound I-like intermediate) can also be envisioned (Figure S9). Either trajectory would benefit from the formation of an intermediate with more oxidizing potential than oxyPrnB.

Figure 7:

Proposed catalytic scheme for PrnB, where S indicates the 7-Cl-Trp substrate. Reduction of the oxy-complex and further protonation step(s) would yield a hydroperoxo or oxo species that serves as an electrophilic oxidant to initiate MCAP formation.

Supplementary Material

NMR spectra of MCAP standard, additional optical and stopped flow absorption data and fitting, possible catalytic mechanisms of PrnB.