Synergistic Binding of the Halide and Cationic Prime Substrate of the l-Lysine 4-Chlorinase, BesD, in Both Ferrous and Ferryl States

An aliphatic halogenase requires four substrates: 2-oxoglutarate (2OG), halide (Cl− or Br−), the halogenation target (“prime substrate”), and dioxygen. In well-studied cases, the three non-gaseous substrates must bind to activate the enzyme’s Fe(II) cofactor for efficient capture of O2. Halide, 2OG, and (lastly) O2 all coordinate directly to the cofactor to initiate its conversion to a cis-halo-oxo-iron(IV) (haloferryl) complex, which abstracts hydrogen (H•) from the non-coordinating prime substrate to enable radicaloid carbon-halogen coupling. We dissected the kinetic pathway and thermodynamic linkage in binding of the first three substrates of the l-lysine 4-chlorinase, BesD. After 2OG adds, subsequent coordination of the halide to the cofactor and binding of cationic l-Lys near the cofactor are associated with strong heterotropic cooperativity. Progression to the haloferryl intermediate upon addition of O2 does not trap the substrates in the active site and, in fact, markedly diminishes cooperativity between halide and l-Lys. The surprising lability of the BesD•[Fe(IV)=O]•Cl•succinate•l-Lys complex engenders pathways for decay of the haloferryl intermediate that do not result in l-Lys chlorination, especially at low chloride concentrations; one identified pathway involves oxidation of glycerol. The mechanistic data imply that (i) BesD may have evolved from a hydroxylase ancestor either relatively recently or under weak selective pressure for efficient chlorination and (ii) that acquisition of its activity may have involved the emergence of linkage between l-Lys binding and chloride coordination following loss of the anionic protein-carboxylate iron ligand present in extant hydroxylases.


S3
AmSO4. The lysate supernatant was concentrated to a manageable volume before overnight dialysis in 100 mM HEPES, pH 7.5 to remove the remaining AmSO4.
The desalinated lysate was concentrated after dialysis to <100 mL and loaded onto a selfpacked Q column (26/100, 600 mL resin) for ion exchange purification with an AKTA Purifier FPLC system (GE Healthcare). The protein was eluted using a gradient from 0-50% buffer A (100 mM HEPES, pH 7.5, 50 mM MgSO4) to buffer B (100 mM HEPES, 1 M MgSO4, pH 7.5) with a final wash of 100% buffer B. Fractions containing BesD were pooled and concentrated via centrifugal concentrators. Purified BesD was dialyzed against a solution of 5 mM EDTA and 100 mM HEPES, pH 7.5 buffer for 4+ hours and two additional dialysis steps without EDTA to remove exogenous metals. Protein purity was accessed by SDS-PAGE (as shown in Figure S17) and protein concentration was determined by its absorbance at 280 nm using a calculated molar absorptivity of 50,420 M -1 cm -1 (https://web.expasy.org/protparam/). Typical protein yields ~250 mg BesD per liter of rich LB medium.

Anion Exchange Chromatography to Remove Chloride Ions from Lysine Stocks
To better control chloride concentrations under various conditions, it was necessary to remove the chloride ions from commercial lysine stocks. While protiated L-lysine (Sigma-Aldrich) was available as a free amino acid, i.e. chloride-free, no lysine deuterated isotopes could be found without HCl counterions.
The d4-4,4,5,5-L-Lys•2HCl stock (Cambridge Isotopes) was dissolved in a 25:75 mixture of 1M NaOH:100 mM HEPES, pH 7.5 buffer to a final concentration of 0.5 M d4-L-Lys, pH 6-8. Final pH was confirmed to be neutral via pH strips (EMD Millipore). A 5-mL FF Q-column (Cytiva) was prepared by rinsing the column with 25 mL of 1 M ammonium acetate, pH 6.5, followed by 25 mL of H2O via syringe or peristaltic pump. The high excess of the weaker-binding acetate ions outcompeted and exchanged with the resin-bound chloride ions. 1 The d4-L-Lys stock was loaded onto and washed off the column with 10 mL H2O and the flowthrough was collected. Chloride ions in the L-lysine solution were expected to bind to the Q resin and release the previously resinbound acetate ions. The flowthrough solution was flash frozen in LN2 and lyophilized for >18 hours. The solid was resuspended in H2O as d4-4,4,5,5-L-Lys •2NaAcetate. Removal of chloride ions was confirmed by addition of a high concentration of treated lysine stock to BesD-Fe II -2OG complex in a UV-Vis spectrophotometer, as shown in Figure S18. Lack of perturbation of the Fe II -2OG MLCT at 520 nm indicates little to no chloride ions present in the lysine stock.
Absorption Spectroscopy UV-Visible spectra were measured using an Agilent 8453 UV-visible spectroscopy system housed in an MBraun (Stratham, NH) anoxic chamber. Titrations of the various analytes were carried out using two identical cuvettes; one for the sample of interest and a control lacking Fe II for subtraction during the analysis. Spectra were collected after each addition of an aliquot of the concentrated stock titrant solution. Each experimental spectrum was corrected for dilution by the titrant stock solution and the initial spectrum ([titrant]= 0 M) was subtracted from the dilutioncorrected spectrum. The absorbance at 800 nm was subtracted from each spectrum and set to zero and, if required, were baseline corrected via a point-based cubic spline in Kazan Viewer. 2

Microscale Thermophoresis
Affinity measurements for free L-lysine binding to BesD was carried out using a NanoTemper Tech. Monolith NT.115 microscale thermophoresis (MST) instrument. A stock of BesD was labelled with a fluorescent moiety via coupling to protein-based amines using the protein labeling kit RED-NHS 2 nd generation (NanoTemper Tech.). MST samples were prepared in a anaerobic chamber. The fluorophore labelled BesD (RED-BesD) was supplemented to the protein binding mixture. The protein solution was aliquoted and added to an L-Lys concentration series in a 1:5 mixture for a final concentration of 50 nM RED-BesD, 0.5 mM BesD, 0.5 mM Fe II , 2.5 mM 2OG, 0.05% TWEEN, and 100 mM HEPES, pH 7.5. L-lysine concentration ranges varied based on the concentration of chloride present in each series to better incapsulate the binding range. Unlabeled BesD was added to the samples to prevent unincorporated ferrous ions from causing fluctuations in the fluorescence signal due to quenching. Samples were loaded into capillary tubes and sealed with wax on both ends to prevent sample oxidation. Samples were analyzed by MST using 100% MST power and 100% excitation power at 650 nm and emission at 670 nm.

Stopped-Flow Absorption Spectroscopy
Stopped-flow absorption experiments were performed on an Applied Photophysics Ltd. (Leatherhead, UK) SX-20 stopped-flow spectrophotometer housed inside a glovebox. Reactions were carried out at 5°C in a single-mixing configuration with a 1 cm (or 0.2 cm, for reactions in Figure 6D) pathlength and photomultiplier tube (PMT) detector. Wavelengths were selected/isolated from the broadband light source before the reaction cell via a monochromator. Specific experimental details are provided in the figure legends.

LC-MS Activity Assays as a Function of Chloride Concentration
Reactions to assess the activity of BesD under single-and multi-turnover conditions were carried out. Multi-turnover assays contained 1 mM substrate (L-Lys or d4-L-Lys), 0.1 mM (NH4)2Fe(SO4)2•6H2O, 0.1 mM BesD, 5 mM Na22OG, 5 mM ascorbic acid, and NaCl at varying concentrations. Reactions were initiated by addition of 2OG, mixed thoroughly to allow for proper oxygenation in ambient air, and allowed to proceed for 20 min at room temperature before quenching with a 2-fold dilution in methanol. After quenching, internal standards of the opposite substrate isotopologue (L-Lys for d4-L-Lys reactions, d4-L-Lys for L-Lys reactions) and sodium d4succinate was added to the mixture before filtration. The samples were immediately transferred and analyzed by LS-MS, as described below.
Single turnover assays contained 1 mM substrate (L-Lys or d4-L-Lys), 0.5 mM (NH4)2Fe(SO4)2•6H2O, 0.7 mM BesD, 0.375 mM Na22OG, and NaCl at varying concentrations. Reactions were prepared inside an anaerobic chamber and initiated by the addition of O2saturated buffer (5°C) for a final concentration of 0.6 mM. The reactions were allowed to proceed for 5 min before quenching with a 2-fold dilution in methanol. After quenching, internal standards of the opposite substrate isotopologue (L-Lys for d4-L-Lys reactions, d4-L-Lys for L-Lys reactions) and sodium d4-succinate was added to the mixture before filtration. The samples were immediately transferred and analyzed by LS-MS, as described below.
Liquid chromatography coupled to mass spectrometry (LC-MS) analysis was carried out on a 1200 series LC system connected to a 6400 series triple quadrupole mass spectrometer (Agilent Technologies). Analysis of L-Lys-based products by LC-MS was performed by injection (5 µL) of reaction mixture onto a SeQuant ZIC-HILIC (3.5 µM, 150 x 2.1 mm; EMD Millipore) column with a mobile phase consisting of 10:90 (v/v) MeCN:H2O, 10 mM NH4HCO2, pH 5.0 (A) and 90:10 (v/v) MeCN:H2O, 10 mM NH4HCO2, pH 5.0 (B). Separation of elutants was carried out with a linear gradient of 95% B to 30% B over 25 min followed by a linear gradient of 30% B to 95% B over 5 min at a flow rate of 0.2 mL/min. Samples were analyzed in positive-ionization mode.
Evaluation of succinate and 2OG present in reactions was performed by injection (2 µL) onto an Extend-C18 (1.8 µM, 50 x 4.6 mm; Agilent) column with a mobile phase consisting of 0.1% formic acid in H2O (A) and MeCN (B). An isocratic method of 5% B at 0.3 mL/min was utilized to separate elutants over a 5 min period. Samples were analyzed in negative-ionization mode.

Freeze-Quench Mössbauer Spectroscopy
Freeze-quench Mössbauer samples were prepared according to previously published procedures. 3 Mössbauer spectra were recorded on a spectrometer from SEECO (Edina, MN) equipped with a Janis SVT-400 variable-temperature cryostat. The reported isomer shift is given relative to the centroid of the spectrum of α-iron metal at room temperature. External magnetic fields were applied parallel to the direction of propagation of the γ-radiation. Simulations of the Mössbauer spectra were carried out using WMOSS spectral analysis software from SEECO (www.wmoss.org, SEE Co., Edina, MN).
Samples were generated from anoxic solutions of 1.8 mM BesD, 1.5 mM 57 Fe(II), 16 mM 2OG, 10 mM or 60 mM d4-L-Lys, and 100 mM or 3.0 M NaCl in a butter solution of 100 mM HEPES, pH 7.5. These samples were mixed at 5 °C with an equal volume of the buffer that had been saturated with O2 (~1.8 mM). This reaction mixture was allowed to incubate for the varying reaction times indicated in Figure 6 and subsequently frozen by injection into cryogenically cooled (-150 °C) 2-methylbutane (for reaction times <30 seconds) or by pipetting into a Mössbauer cell cooled on a metal block that was in contact with LN2 (for reaction times >30 seconds). Handquench EPR samples were made in parallel under same conditions as Mössbauer samples. Samples were hand-mixed and transferred to EPR tubes before freezing in LN2 at appropriate quench times.

Electron Paramagnetic Resonance Spectroscopy
EPR samples were transferred into custom X-band EPR tubes (Quartz Scientific, Inc., Fairport Harbor, Ohio), and continuous-wave X-band EPR spectra were recorded on a Magnettech MS5000X spectrometer equipped with an Oxford Instruments ESR-900 continuous flow cryostat and an Oxford Instruments ITC-300 temperature controller. Data were collected for vanadyl-based samples at 35 K from 250 to 450 mT and iron-based samples were collected at 10 K from 50 to 425 mT. Spectra were collected with modulation amplitude of 1 mT, microwave power of 0.1 mW, and a microwave frequency of 9.43516 GHz.
BesD-vanadyl samples were prepared with 0.5 mM vanadyl sulfate, 0.6 mM BesD, 10 mM sodium succinate, and 0.4 M sucrose (cryosolvent) in buffer containing 100 mM HEPES, pH 7.5. Concentrations of L-Lys and NaCl for samples are denoted in figures. Analysis of the L-Lys/NaCl titration data was performed using three reference spectra multiplied by a fractionation coefficient in a linear regression to determine the amount of each species in the spectrum. This was performed in the software package Igor Pro 9 (WaveMetrics, Inc., Lake Oswego, OR, USA.) The collected spectra were baseline corrected by subtraction of a fitted polynomial in Kazan Viewer. 2 The baseline-corrected EPR spectra were simulated with the EasySpin toolbox 4 within MATLAB.
Analysis of L-Lysine and Chloride Cooperativity in Fe II -BesD-2OG Complex As the affinity of L-lysine (L-Lys) and chloride to Fe II -BesD-2OG are independently weak but relatively stronger in the presence of the opposite cosubstrate, it is assumed that they bind in a positively cooperative manner. To model this, a square scheme has been constructed as demonstrated below: This value should be <<1 for strong, positive cooperativity (as observed for chloride binding in the absence/presence of L-lysine in Figure 1). Due to the thermodynamic cycle, only three parameters (K Cl II , K Lys II , and a II ) are required to characterize the equilibria among the relevant species. While the first binding events of just L-lysine or just chloride can be determined by a simple binding curve in the absence of the other cosubstate, the parameters of the synergistic second binding event require a more complicated analysis. For example, the titration curve of chloride in the presence of a static concentration of L-lysine, reveals the apparent KD for chloride binding, K Cl,app II , by monitoring the metal-to-ligand charge-transfer (MLCT) band at 520 nm exhibited specifically by a chloride ion bound to a Fe II -BesD-2OG complex (i.e., BesD-Cl and BesD-Cl-Lys, from the above scheme). This constant can be expressed as: Using the definitions of K Lys II , K Cl II , and a II K Cl II K Lys II , this expression can be substituted and rearranged: Since L-Lys is a weak and strong binder in the absence and presence of Cl-, we assume that a II , such that a II K Lys II ≫1: This expression allows for the determination of the cooperative binding step, a II K Lys II , by measuring Analysis of productive and unproductive ferryl intermediate decay To extract the formation and decay rates for the ferryl intermediate in BesD when two reaction outcomes are present, a step-wise, branching pathway is invoked, as the ferryl can either decay productively or unproductively. Typically, the ferryl intermediate can be analyzed using an expression derived from a simple A to B to C model (shown above), as only the ferryl intermediate (B) absorbs appreciably at 318 nm. In the case of BesD with deuterated substrates, the ferryl appears to decay via two different pathways: either to a ferrous product (C) or ferric product (D) state. The ferric state has an appreciable absorbance at 318 nm overlapping with the ferryl LMCT and requires a more complex expression to account for the mixed absorbance of ferryl intermediate and ferric product. The model used for this stepwise branching pathway is demonstrated below: The overall decay of the ferryl intermediate (B) is a combination of both productive and unproductive pathways and is represented by kdecay.
The integrated rate laws for the four states are as follows:

S8
As stated previously, the absorbance at 318 nm as a function of time is defined as a combination of ferryl (B) and ferric (D) multiplied by their respective extinction coefficients.
Substitution and rearrangement leads to the final analytical expression: From this expression and the experimental data measured at A318, kdecay can be determined for the ferryl intermediate and used for further analysis (see below).

Analysis of d4-L-lysine and NaCl Cooperativity in [Fe IV =O]-BesD Complex
The decay rate of the ferryl intermediate in BesD can be affected by both concentration of chloride and d4-L-lysine. At low concentrations of both substrates, the ferryl decays faster and there is a net positive absorbance in the baseline at 318 nm, indicative of the ferryl uncoupling and ending in irreversibly formed ferric state that has an increased absorbance at 318 nm. As [d4-L-Lys] and [Cl -] increase, the ferryl decay rate slows and the DA318 in the baseline after ferryl decay is smaller, indicating less uncoupling. Based on these observations and what has been observed in the ferrous state, the following scheme representing the mechanism after oxygen activation by the BesD quinary complex was constructed (with the presence of BesD implicitly inferred): Similar to the ferrous cooperativity scheme, K Lys IV and K Cl IV represent the dissociation constants for L-lysine and chloride, respectively, and a IV is the cooperativity coefficient in the ferryl state. Two rate constants are assigned to the observed two pathways by which the ferryl decays: the rate of uncoupling, kunc, and the rate of productive decay, kD. From this scheme, an analytical solution can be derived to express the five parameters that dictate ferryl decay in BesD. As the ferryl absorbance at 318 nm originates from the ligand-to-metal charge transfer (LMCT) band of the oxo ligand and the Fe IV , we can assume that each state of the ferryl has a relatively equal extinction coefficient. The experimentally observed DA318 is therefore an observation of the total ferryl, as defined below: The differential rate law for total ferryl decay can be written as: The decay of the ferryl only occurs from two of the four ferryl states, Fe IV -Cl-Lys, and Fe IV . As such, this can be rewritten as: If there are fast equilibria for L-Lys and Clbinding with respect to the decay rates, the following assumptions can be made, Using the previous definitions, the rate of decay can be rewritten and substituted,  Figure 5) to determine the rate of uncoupling, kunc, the rate of productive decay, kD, the dissociation constants for L-lysine and chloride, K Lys IV and K Cl IV , and the cooperativity coefficient for L-lysine and chloride binding, a IV .

BesD-VO-Lys Sim
Figure S14: Reference spectra and simulations of the three different vanadyl species present in L-Lys and NaCl titration data in Figure S14. (A) Overlay of the three reference spectra with low and high field wings emphasized (insets). Experimental and simulated spectrum for (B) BesD-VO, (C) BesD-VO-Lys, and (D) free vanadyl with no protein added to the sample. (E) Simulation parameters for all three species. gStrain is a line shape broadening term used for each g-component.