Role of a secondary coordination sphere residue in halogenation catalysis of non-heme iron enzymes

. Chemo- and regio-selective catalysis of C( sp 3 )-H halogenation reaction is a formidable goal in chemical synthesis. 2-oxo-glutarate (2OG) dependent non-heme iron halogenases catalyze selective chlorination/bromination of C-H bonds and exhibit high sequence and structural similarities with non-heme iron hydroxylases. How the secondary coordination sphere (SCS) of these two enzyme systems differentiate and determine their reactivity is not understood. In this work, we show that tyrosine placement in the SCS of non-heme iron halogenases have a huge impact on their structure, function, and reactivity. We discover that a tyrosine mutant (F121Y) in SyrB2 halogenase undergoes post-translational oxidation to dihydroxyphenylalanine (DOPA) physiologically. A combination of spectroscopic, mass-spectrometric, and biochemical studies show that the DOPA modification in SyrB2 renders the enzyme non-functional. Further bioinformatics analysis suggests that halogenases, unlike hydroxylases, have a conserved placement of phenylalanine at position 121 to preclude such unproductive Overall, this demonstrates the importance of the in controlling the structure and enzymatic activity of non-heme iron halogenases. Our results have significant implications towards the design of small-molecule and protein-based halogenation


Introduction
2OG-dependent non-heme iron enzymes (NH-Fe) catalyze a diverse range of reactions ranging from epoxidation to stereo-inversion. 1 From natural product biosynthesis to cellular signaling and metabolism, the reactions catalyzed by this superfamily of enzymes serve crucial roles biologically. 2,3 Two subgroups of this superfamily are the hydroxylases and halogenases.
Hydroxylases catalyze the insertion of a hydroxyl functional group into unactivated aliphatic C-H bonds while halogenases catalyze the insertion of a chloride/bromide functionality into such bonds. The difference in reactivity of the two enzymes is enabled by differences in their ironbinding primary coordination sphere. Hydroxylases bind iron via a 2 His/1 Carboxyl (Asp/Glu) facial triad analogous to halogenases except that in halogenases the carboxyl ligand is replaced by a non-polar Ala or Gly residue. 1,4 This amino acid substitution in halogenases enables the steric and electrostatic accommodation of an iron coordinated halide anion that is essential for halogenation reactivity. 5 Both hydroxylases and halogenases have been suggested to operate via a similar catalytic mechanism (Fig S1). 6,7 Once 2OG, iron, and substrate are bound to the enzyme, a conformational change is triggered that facilitates molecular oxygen binding to the iron center. Next, oxidative decarboxylation of the 2OG co-substrate results in the cleavage of oxygen's O-O bond and the formation of a high-valent ferryl intermediate (Fe IV =O). The ferryl intermediate abstracts a hydrogen atom from the substrate, resulting in a ferric-hydroxyl (Fe III -OH) species and a substrate radical. At this point, the mechanisms for the hydroxylases and halogenases diverge. In hydroxylases, the hydroxyl group bound to the iron will "rebound" and combine with the substrate radical to give a hydroxylated product. In halogenases, the protein positions the halide ligand closer to the substrate radical than the hydroxyl ligand. This favors radical combination with the halide ligand resulting in a halogenated product, though a minor hydroxylated product is also possible through hydroxyl rebound.
This work focuses on a well-characterized halogenase, SyrB2, which chlorinates/brominates the side chain methyl group of its substrate Thr. 8,9 Amino acid Thr is delivered to the active site of SyrB2 via partner protein SyrB1's covalently attached phosphopantetheine (Ppt) group. 8 When Thr is substituted with non-native amino acid substrates like alpha-aminobutyric acid or norvaline, SyrB2 primarily hydroxylates them. 10 These studies and others have demonstrated the importance of SyrB2's interaction with Thr and other amino acid substrates to influence halogenation/hydroxylation reaction outcomes. [10][11][12][13][14][15] While substrate-protein interactions are wellunderstood, it's unclear how iron's secondary coordination sphere (SCS; amino acid residues that interact with the ligands bound to iron) governs reaction outcome in SyrB2 and other halogenases. From a broader perspective, while the impact of the SCS on the structure and function of heme iron enzymes have been extensively studied, little is known about how nonheme iron enzymes utilize SCS to control their reactivity. [16][17][18][19][20] In this work, we investigate the importance of iron's SCS in controlling SyrB2 halogenase's reactivity. Specifically, we establish the role of a hydrophobic Phe121 residue that is ~4 Å away from iron's chloride ligand towards configuring halogenase activity in SyrB2. Mutating this residue to Tyr facilitates its intracellular post-translational modification to a dihydroxyphenylalanine (DOPA) residue that renders SyrB2 non-functional. Beyond SyrB2, the F121 residue is highly conserved in halogenases, unlike hydroxylases, signifying the differential nature of the SCS in the two enzyme classes. Overall, our studies emphasize the importance of the SCS in controlling catalytic activities of non-heme iron enzymes, particularly halogenases and hydroxylases.
Site directed mutagenesis was performed on the SyrB2 plasmid to form the F104Y and F121Y mutations using the Phusion™ Site Directed Mutagenesis Kit from Thermo Scientific. The forward and reverse primers (5' to 3') for the F104 mutation were: GAGTTCTATCCCAAATATCCGGGCG and TTTGGGATAGAACTCGGTACGCCAG respectively. The forward and reverse primers (5' to 3') for the F121Y mutation were: CACCTACGCCAATGCCTCCGGCA and CATTGGCGAAGGTGTCGGCCTGG respectively.
A three-step denaturation, annealing, and extension protocol was used for both mutants with an annealing temperature of 67.6°C for F104 and 72.0°C for F121Y. All PCR experiments were performed using a Axygen Maxygene II instrument. All PCR products were transformed into DH5α cells (Thermo Scientific) for plasmid expression. Single mutant DNA sequences were confirmed using Sanger Sequencing at the UMN Genomics Center.

Protein Expression
SyrB1, SyrB2, SyrB2 mutants, and Sfp were all expressed using the same method. Plasmids containing the proteins were transformed into BL21(DE3) cells (Thermo Scientific) for protein overexpression. An overnight culture (~50 mL 2XYT broth, 0.05 mg/mL kanamycin) was grown at 37°C for ~16 hours at 220 RPM. Ten mL of the overnight culture was used to inoculate 1 L cultures (2XYT, 0.05 mg/L kanamycin) in 2.8L baffled flasks. A single drop of Antifoam 204 (Sigma) was added to each culture to aid in oxygen diffusion. Cultures were grown at 37°C until an OD600 of 0.6-0.8 was reached. Cultures were then placed in an ice bath for 15-20 minutes to cool. Isopropyl β-D-1-thiogalactopyranoside (IPTG, GoldBio) was add ed to a final concentration of 0.2 mM to induce protein overexpression. After 18 hrs of shaking at 18°C, cell pellets were harvested by centrifugation, flash frozen in liquid nitrogen, and stored at -20°C.
Typical yields were ~8-10 g cell pellet/L culture. Overexpression for TycF followed the same protocol except the [IPTG] = 1 mM and the expression time was 72 hours.

Protein Purification
All buffers and solutions described herein were prepared in MilliQ water from a Barnstead GenPure water filtration system (Thermo Scientific) with a resistivity of at least 18.2 MFor all proteins, thawed cell pellets were resuspended in 5 mL wash buffer/g of cells (wash buffer = 50 mM HEPES (Sigma: Biocertified), 300 mM NaCl (Sigma: BioUltra), 5 mM imidazole (Sigma: ReagentPlus), pH = 7.5). Resuspension was supplemented with Pierce™ Protease Inhibitor, EDTA-free tablets (Thermo Scientific). After resuspension, cells were lysed by sonication and centrifuged to remove any non-soluble cell debris (20,000 RPM, 20 min, 4°C).
Supernatant was filtered with 0.43 micron syringe filters and loaded onto a wash-buffer equilibrated 5 mL HisTrapFF column (Cytiva, 2 mL/min binding rate) using an AKTA Start protein purification system (Cytiva). After sample application, the column was washed with 15 CV of wash buffer to elute any non-specific binding proteins. Elution occurred using a linear gradient from 0% to 100% elution buffer (elution buffer = 50 mM HEPES, 100 mM NaCl, 250 mM imidazole, pH = 7.5). For TycF, an isocratic elution was performed by washing the column with 10 CV of 30% elution buffer then 10 CV of 100% elution buffer. Relevant protein fractions, as determined by SDS-PAGE analysis, were pooled. Sfp and TycF were dialyzed against RXN buffer (20 mM HEPES, pH =7.5) overnight, filtered, concentrated using Amicon™ Ultra 10 kDa MWCO centrifugal filter units (Millipore Sigma), flash frozen in liquid nitrogen, and stored at -80°C.
SyrB2 and its mutants were buffer exchanged into RXN buffer three times using Amicon™ Ultra Preparation of SyrB1 after initial immobilized metal affinity chromatography protein is described in the next section. All protein concentrations were determined using the absorbance at 280 nm using molar extinction coefficients from the ProtParam tool. 21

Ppt and Thr attachment to SyrB1
After immobilized metal affinity chromatography, SyrB1 was buffer exchanged into RXN buffer three times using Amicon™ Ultra 50 kDa MWCO centrifugal filter units (Millipore Sigma). To attach the Ppt cofactor, SyrB1 (100 µM) was incubated with coenzyme A (1 mM, Sigma: cofactor for acyl transfer), MgSO4 (5 mM, Sigma: BioReagent), and phosphopantetheinyl transferase Sfp (5 µM) for 90 minutes at room temperature. The reaction mixture was concentrated using the same centrifugal filter units and then diluted to ~8 mL with reaction buffer. After filtering, the protein was loaded into the superloop of an AKTA Pure protein purification system. The same elution method as described for the SyrB2 proteins was performed except that the flow rate was adjusted to 0.8 mL/min. Relevant fractions from size-exclusion chromatography were pooled and concentrated using centrifugal filter units. To append threonine to the protein, SyrB1-Ppt (100 µM) was incubated with MgSO4 (5 mM), ATP (10 mM, Sigma: Grade I, ≥99%, from microbial), and L-threonine (10 mM, Sigma: reagent grade, ≥98% (HPLC)) in RXN buffer for 30 min at room temperature.
After this step, it's crucial that SyrB1-Ppt-Thr remain at cool temperatures to prevent spontaneous hydrolysis of the thioester bond that links threonine to PPT-SyrB1. The amino acidloaded protein was then concentrated using the previous centrifugal filter unit, and buffer exchanged four times with RXN buffer to remove excess reaction components. The protein was aliquoted, flash frozen in liquid nitrogen, and stored at -80°C.

UV-Vis measurements
All UV-Vis measurements were recorded on either a Cary 8454 UV-Vis Spectrophotometer or Cary 4000 UV-Vis Spectrophotometer. Measurements with the Cary 4000 were performed in double beam mode with a S/N ratio of 10000, a timeout of 0.2 s, and a spectral bandwidth of 2 nm. All spectra were background corrected at 900 nm.

EPR measurements
All SyrB2 samples (WT and F121Y/DOPA) were prepared at a final concentration of 450 µM in a volume of 280 µL. Samples were transferred to EPR Tubes (Wilmad) and frozen in a methanol/dry ice bath followed by a liquid nitrogen bath. Both samples were stored in a liquid nitrogen dewar until use. EPR measurements were performed on a Bruker Elexsys E-500 spectrometer with an E500-T-DU Digital upgrade equipped with an Oxford ESR-910 liquid He cryostat. EPR collection parameters for F121Y/DOPA SyrB2 are : microwave frequency = 9.6 GHz; microwave power = 1.69 mW, modulation amplitude = 9 G, modulation frequency = 100 kHz, temperature = 2 K. EPR collection parameters for WT SyrB2 were: microwave frequency = 9.6 GHz, microwave power = 210 W, modulation amplitude = 5 G, modulation frequency = 100 kHz, temperature = 2 K.

DOPA nitration/oxidation reaction assay
The following procedure was adapted from the method of Waite et al. 22  Sigma: ACS Reagent) was prepared in MilliQ water. The denatured SyrB2-F121Y/DOPA protein (300 µL) was mixed with 1 equivalent (v/v) of the nitration reagent forming a yellowcolored solution. 1 M NaOH (400 µL) was immediately added to oxidize the nitrated species and form the pink-colored dinitro-quinone complex. The UV-Vis spectrum of the reaction was then recorded on a Cary 4000 instrument using the parameters described above.

Halogenation Assays
SyrB2 or SyrB2-F121Y/DOPA (100 µM) was added to RXN buffer containing NaCl (10 mM), 2OG (1 mM). For the WT system, ferrous ammonium sulfate (Sigma: BioUltra, in 2.5 mM H2SO4) was added to the reaction, and the solution was pipette mixed. SyrB1-PPT-Thr was added to the rxn to a final concentration of 100 µM, and the rxn was pipette mixed again. The total reaction volume was 200 µL, and all steps so far were performed on ice. One volume equivalent of O2 saturated buffer was added to the reaction, and each solution was pipette mixed.
All reactions were transferred to an Eppendorf Thermomixer ® and spun at 300 RPM for 10 minutes and were then transferred to ice. The reactions were loaded into pre-washed (100 mM NaOH then RXN buffer) 0.5 mL Amicon™ Ultra 10 kDa MWCO centrifugal filter units, concentrated to ~50 µL, then diluted to 500 µL, and concentrated again. This washing step was performed two more times. All centrifugation steps occurred at 4°C to prevent thioester  Table S1; low-mass resolution (Q2), 15 V; high-mass resolution (Q2), 15 V; ion energy (Q2) 3.5 V.

Bioinformatics Analysis
To observe the sequence conservation position 121, the amino acid sequence for SyrB2 was used as a query for a BLAST search in the non-redundant protein database with a e -5 cutoff, resulting in 846 hits. Sequences were aligned using the MUSCLE algorithm and visualized using the AliView software. Results were filtered between the halogenases (HXA/G motif, 186 sequences) and hydroxylases (HXD/E motif, 660 sequences). Each of the sequence groups were subjected to cluster analysis using the CD-HIT web server to remove redundant sequences of 90% similarity (resulting in 132 halogenase sequences and 503 hydroxylase sequences). Individual halogenase/hydroxylase sequences were realigned with MUSCLE. LOGOS plots were generated for each cluster using WebLogo (https://weblogo.berkeley.edu/logo.cgi).

SyrB2 variant
A visual inspection of SyrB2's active site reveals two hydrophobic Phe residues that were ~4 Å away from iron's ligands (Fig. 1A). 5 Specifically, the F104 residue is 3.8 Å from a water ligand and F121 residue is 3.9 Å away from the chloride ligand. To assess if the hydrophobic nature of these residues was important, tyrosine mutants of both residues, i.e. F104Y and F121Y SyrB2, were constructed, expressed, and purified. Upon purification, WT SyrB2 and F104Y exhibited no signatures in the visible range of the UV-Vis spectrum, implying that both halogenases purify without iron bound to the active site (Fig. 1B). In contrast, SyrB2-F121Y purified as a bluegreen solution with a λmax at 650 nm (Fig 1B). This blue-green chromophore has been observed in a few NHFe proteins previously, and has been postulated to originate from a ligand-to-metal charge transfer band (LMCT) of a DOPA-Fe 3+ species. [23][24][25][26][27] We hypothesize that such a species has formed from intra-cellular oxidation of the F121Y mutant and will refer to this SyrB2 variant as F121Y/DOPA. To further investigate the iron oxidation state in F121Y/DOPA, we attempted to oxidize the protein using hydrogen peroxide (Fig. S2). Even after incubation with 100 equivalents of peroxide, no shift in the LMCT band was observed, which implies that the iron center is not in the ferrous state. To further characterize the electronic structure of the iron center, we investigated the F121Y/DOPA protein with EPR spectroscopy. The X-band EPR spectrum shows a distinct signal at g ~ 4.29 with additional features at g ~ 5.03 and 9.06 (Fig 2A). The signal at g ~ 4.29 is a hallmark of high-spin ferric iron species in a highly orthorhombic environment (E/D = 0.33). [28][29][30][31] The broad shoulder at g ~ 5.03 and small peak at g ~ 9.06 could be originating from a different Fe 3+ -bound species in a more rhombic environment (E/D = 0.2).
A variation in coordination number of the ferric-DOPA species in F121Y/DOPA could explain the heterogeneity of species in the EPR spectrum. Nevertheless, no EPR signals are observed in the WT SyrB2 spectrum confirming the absence of ferric iron in this protein sample (Fig 2A).
While EPR spectroscopic studies confirm the presence of a high-spin ferric center in F121Y/DOPA sample, similar signals have been previously observed in a ferric-phenolate complex (i.e. a tyrosinate residue binding to Fe 3+ ) as well. 29 However, high-spin ferric-phenolate species exhibit UV signatures blue-shifted (550 nm) relative to the F121Y/DOPA species (Fig   1B). 28,29 To exclude the possibility of ferric-tyrosinate and confirm the presence of DOPA posttranslational modification in F121Y/DOPA sample, we executed a reaction that specifically targets DOPA amino acids. 22 Nitration/oxidation of the DOPA amino acid results in the formation of a di-nitro-DOPA-quinone species with a distinct UV-Vis signature at 500 nm (Fig.   S3A). We performed this reaction on different concentrations of L-DOPA free amino acid to develop a standard curve, which would quantify the extent of DOPA modification in F121Y/DOPA (Fig. S3B). After denaturation of F121Y/DOPA in acidic buffer, the nitration/oxidation reaction was performed, and the UV-Vis spectrum was recorded (Fig 2B). A peak at 510 nm was present in the product, consistent with other DOPA-proteins treated with this and succinate (116 Da) cannot account for this mass differential, other ligands may be present bound to the ferric-DOPA active site. We propose that a non-native ligand, pyruvate (87 Da), is bound to ferric ion in the active site via its alpha-keto functional group. 29 The active site is possibly quite perturbed with the formation of the ferric-DOPA species (F121 is located ~6 Å from the iron center in the WT crystal structure) and it's feasible that the protein's affinity for various ligands has changed. Given the ubiquity of pyruvate in the cellular environment (390 µM), the overexpressed protein would have an abundance of this ligand available to bind. 33 The rest of the unaccounted mass (18 Da) is explained by an aqua ligand, which is present in the WT crystal structure and numerous other non-heme iron structures. 5 Overall, using a combination of spectroscopic, mass-spectrometric, and biochemical studies, we were able to confirm the formation of DOPA post-translational modification in SyrB2 halogenase.

DOPA modification in SyrB2 forms an off-cycle, catalytically inactive species
Next, we tested the impact of DOPA post-translational modification on the halogenation catalysis of SyrB2. To that end, we developed halogenation assays for SyrB2 with its native substrate, Thr, which is delivered to SyrB2's active site while covalently bound to the ppt group of SyrB1. Chlorinated, hydroxylated (minor product), or unreacted threonine were liberated from SyrB1 after incubation with TycF thioesterase. 34 These products were first derivatized with 6aminoquinolyl-N-hydroxysuccinimidyl carbamate (AQC) and then analyzed by reversed-phase UPLC/Multiple-Reaction-Monitoring-MS/MS for detection. WT-SyrB2, when subjected to halogenation assay, reveals AQC-Cl-Thr as the major product, eluting slightly after the unreacted AQC-Thr (Fig 3A). Isotopic ratios of AQC- 35 Cl-Thr to AQC-37 Cl-Thr were shown to be 3:1, consistent with the natural abundance of chlorine, and providing validation to the identity of chlorinated Thr (Fig S5). The AQC-OH-Thr minor product was also observed, eluting prior to the unreacted AQC-Thr, though with much less intensity than AQC-Cl-Thr. SyrB2-F121Y/DOPA, on the other hand, when subjected to similar halogenation assays (with or without the reductant ascorbate), showed no chlorinated or hydroxylated products (Fig 3B, S6-7). The only analyte observed corresponds to the unreacted Thr substrate (Fig 3B, Fig S6-7). While a low-intensity peak is present close to the noise at 2.52 min. in the AQC- 35 Cl-Thr channel, no such peak is seen at the same elution time in the AQC-37 Cl-Thr channel suggesting no chlorinated product was formed. Even in the presence of ten equivalents of ascorbate, no products were observed (Fig S7). With these observations, we postulate that the ferric-DOPA SyrB2 species is a catalytically inactive, off-cycle species. We hypothesize that the modification of tyrosine to DOPA itself isn't necessarily the reason why F121Y/DOPA is catalytically inactive. It's the binding of ferric iron by the catecholate group of DOPA that results in an enzymatic species that's not part of the overall catalytic cycle. The DOPA121 is possibly bound to the ferric iron in a bidentate fashion which would result in large distortions in the active site, since F121 is located ~6 Å away from iron in the WT active site. To test this hypothesis, we performed UV-Vis titration of chloride into SyrB2-F121Y/DOPA and observed no perturbation in the LMCT band eluding to the fact that DOPA121 residue is binding in a position where chloride would bind or that distortion of the active site prevents chloride binding (Fig S8).
Furthermore, it's unclear whether both histidine ligands would still be bound to ferric iron, as they would be needed for proper iron positioning in the active site for native activity. Saturation of coordination sites by DOPA121 would also prevent cofactors like 2OG and oxygen from binding to the iron site. Occupancy of coordination sites by pyruvate, as shown in our MS studies, would also exacerbate this problem. Overall, the sequestration of ferric iron by the modified DOPA residue abolishes any catalytic activity of SyrB2. While we've shown that F121 residue plays a significant role in preventing unproductive oxidation in SyrB2, we wanted to assess the impact of this residue across similar proteins. Using the SyrB2 sequence, we performed a BLAST analysis and sorted hits into hydroxylases (HXD/E motif) or halogenases (HXA/G motif). After removing redundant hits, 635 protein sequences were aligned, and LOGOS plots were generated (Fig 4A). For hydroxylases, the prominent amino acid in position 121 is a redox-active tyrosine. In halogenases, on the other hand, the redox-inactive phenylalanine is strongly conserved at position 121. Since mutation of phenylalanine to tyrosine at this position in SyrB2 has led to an off-cycle species, it's intriguing that the hydroxylases have evolved to have a redox-active tyrosine proximal to the reactive iron center. We believe the differential residue conservation at position 121 for hydroxylases and halogenases is due to the isomeric flexibility of the reactive ferryl intermediate (Fig 4B). 14,15,[35][36][37]   electrostatically accommodates halide binding to the iron center. As a consequence, the protein's ability to "lock" the reactive intermediate into a single isomer is lost. Three different isomers for the haloferryl intermediate have been proposed for halogenase systems (Fig 4B). One is analogous to hydroxylases and exists as trigonal bipyramidal structure with the oxo ligand and a histidine ligand defining the axial plane (1). Another isomer is structurally similar to 1 but swaps the positions of the oxo and halide ligands (2). The last isomer is best described as a heavily distorted trigonal bipyramidal structure, with the halide ligand and the coordinating oxygen of the succinate ligand defining the axial plane (3). Given this flexibility, any one of these intermediates could be reactive across the 2OG-dependent halogenase enzymes. Since reactive haloferryl intermediates exist in the timescales of hundreds of seconds, 11 halogenases must prevent a non-productive isomer from damaging the protein or leading to an unreactive enzyme intermediate. Evolution of the redox-inactive phenylalanine residue at position 121, therefore, prevents an unproductive oxidation from occurring, maintains catalytic activity, and allows the halogenase to align a productive ferryl isomer for its halogenation function.

Conclusions
We show that the incorporation of a redox active tyrosine residue at position 121 in SyrB2 results in further post-translational modification to DOPA. This DOPA residue binds ferric iron with very high affinity and results in a catalytically inactive, off-cycle enzymatic species. The ferric-DOPA species likely forms as a result of an unproductive oxidation in F121Y-SyrB2 (Scheme 1). After H-atom abstraction of the hydroxyl group of Y121, the newly formed hydroxyl group on the iron center will rebound and incorporate into the ring structure, eventually forming DOPA121. This residue can become deprotonated and bind ferric iron from the cellular environment, forming the blue-green chromophore we observe upon purification of this protein (Fig. 1B). To avoid such unproductive oxidations, halogenases tend to keep redox-inactive residues in their SCS, especially in position 121. The placement of redox-inactive residues in the halogenases is especially important due to multiple isomers of ferryl intermediates, as compared to the hydroxylases. This study also explores the function of H-bonding in the active sites of halogenases, and their implications on catalysis. Our results serve as a warning to others attempting to engineer non-heme iron enzymes for non-native functions, [38][39][40] as placement of redox-active residues can result in catalytically non-functional enzymes.