Serial Femtosecond Zero Dose Crystallography Captures a Water‐Free Distal Heme Site in a Dye‐Decolorising Peroxidase to Reveal a Catalytic Role for an Arginine in FeIV=O Formation

Abstract Obtaining structures of intact redox states of metal centers derived from zero dose X‐ray crystallography can advance our mechanistic understanding of metalloenzymes. In dye‐decolorising heme peroxidases (DyPs), controversy exists regarding the mechanistic role of the distal heme residues aspartate and arginine in the heterolysis of peroxide to form the catalytic intermediate compound I (FeIV=O and a porphyrin cation radical). Using serial femtosecond X‐ray crystallography (SFX), we have determined the pristine structures of the FeIII and FeIV=O redox states of a B‐type DyP. These structures reveal a water‐free distal heme site that, together with the presence of an asparagine, imply the use of the distal arginine as a catalytic base. A combination of mutagenesis and kinetic studies corroborate such a role. Our SFX approach thus provides unique insight into how the distal heme site of DyPs can be tuned to select aspartate or arginine for the rate enhancement of peroxide heterolysis.


Cloning of DtpB and construction of a plasmid for over-expression in Escherichia coli
The gene encoding for DtpB in Streptomyces lividans 1326 (SLI_7409) consists of 948 nucleotides and encodes for a protein of 316 amino acids. Genomic DNA was used together with the following forward and reverse primers with flanking NdeI and Hind III restriction sites (underlined): F-5'-CTGCCATATGATGGGCGGAGAAGTCGAGGAACC-3' and R-5'-CTTAAGCTTTCAGGGCCGAGCGGAGAGGTCCTC-3' to amplify dtpB by PCR. The resulting PCR product was ligated into a pET28a (Kan r ) vector (Novagene), to create an Nterminal His6-tag construct (pET28dtpB). Several clones underwent DNA sequencing (Source Bioscience) to the confirm the expected sequence.

Over-expression and purification of DtpB
The pET28dtpB (Kan r ) vector was transformed into E. coli BL21 (DE3) cells. Overnight precultures (low salt LB medium; Melford) were successively used to inoculate 1.4 L of high salt LB medium (10 g tryptone, 10 g sodium chloride, 5 g yeast extract per litre) with 50 mg ml -1 kanamycin and grown at 37°C, 180 rpm. At an OD600 of 1.2, 5-aminolaevulinic acid (0.25 mM final concentration) and iron citrate (100 µM final concentration) were added consecutively for their use as a haem-precursor and iron supplement. Cultures were then induced by adding isopropyl β-D-thiogalactopyranoside (IPTG; Melford) to a final concentration of 0.5 mM and carbon monoxide (CO) gas bubbled through the culture for 20-30 s. Flasks were then sealed and incubated for a further 18 h at 30 °C and 100 rpm. Cells were harvested via centrifugation (10,000 g, 10 min, 4 °C) and the cell pellet resuspended in 50 mM Tris/HCl, 500 mM NaCl (Fisher) and 20 mM imidazole (Sigma) with the pH adjusted to 8 (Buffer A). The resuspended cell suspension was lysed using an EmulsiFlex-C5 cell disrupter (Avestin) followed by centrifugation (22,000 g, 30 min, 4 °C). The clarified supernatant was loaded onto a 5-ml nickel-nitrilotriacetic acid-Sepharose column (GE Healthcare) equilibrated with Buffer A and eluted by a linear imidazole gradient using Buffer B (Buffer A with 500 mM imidazole). The peak eluting at ~ 50 % Buffer B contained DtpB and was pooled and concentrated using a centricon (VivaSpin) with a 10 kDa cut-off at 4 °C followed by loading onto a PD-10 column (Generon) equilibrated in 20 mM NaPi, 100 mM NaCl, pH 7 to remove imidazole. The eluant was further concentrated for application to a S200 Sephadex column (GE Healthcare) equilibrated with 20 mM NaPi, 100 mM NaCl, pH 7. A major peak eluted at ~ 55 ml consistent with a monomer species with fractions assessed by SDS-PAGE then concentrated and stored at -20 °C.

Site directed mutagenesis
Single amino acid substitutions were created using the QuikChange protocol (Stratagene). The pET28dtpB was used as template and the following forward and reverse primers (Sigma) were used to create mutations that would result in the D152A and R243A mutants, respectively;

Sample preparation
DtpB was exchanged into a desired buffer using a PD-10 column (Generon) and concentrated using centrifugal ultrafiltration devices (Vivaspin GE Healthcare). Enzyme concentration was determined by UV-visible spectroscopy (Varian Cary 60 UV-visible spectrophotometer) using an extinction coefficient (ε) at 280 nm of 18,575 M -1 cm -1 . H2O2 solutions (Sigma-Aldrich) were prepared from a stock with the final concentration determined spectrophotometrically using an e = 43.6 M -1 cm -1 at 240 nm. [1] Stopped-flow absorbance spectroscopy Transient kinetics of the interaction of H2O2 with ferric DtpB and variants was performed using a SX20 stopped-flow spectrophotometer (Applied Photophysics, UK) equipped with a diodearray multi-wavelength unit and thermostatted to 25 o C. DtpB solutions (10 µM before mixing) were prepared in 50 mM sodium acetate, 150 mM NaCl, pH 5 and mixed with a series of H2O2 concentrations (ranging from 10 -600 µM for wild-type; 40 -200 µM for D152A; 500 -5000 µM for R243A; before mixing). The overall spectral transitions were monitored and fitted to models in the Pro-K software (Applied Photophysics, UK) to yield pseudo-first order rate constants for Compound I (kobs1) formation.

EPR spectroscopy and simulation
Wilmad SQ EPR tubes (Wilmad Glass, Buena, NJ) with OD = 4.05 ± 0.07 mm and ID = 3.12 ± 0.04 mm (mean ± range) were used. Samples frozen in a set of these tubes yielded very similar intensities of EPR signals; with only ~1-3% random error. All EPR spectra were measured on a Bruker EMX EPR spectrometer (X-band) at a modulation frequency of 100 kHz. A Bruker resonator ER 4122 (SP9703) and an Oxford Instruments liquid helium system were used to measure the low-temperature (10 K) EPR spectra. EPR spectra of a blank sample (frozen water) measured at the same set of instrumental conditions were subtracted from the DtpB spectra to eliminate the background baseline EPR signal. Spectra deconvolution into two components and measurements of the intensities of these components in the time dependence set of samples were performed by using the procedure of spectra subtraction with variable coefficient. [2] Quantitative estimates of the concentrations of the paramagnetic centres were performed by comparison of the second integrals simulated EPR signals with a reference to known total concentration of the ferric haem in the sample. Simulation was performed by WinEPR SimFonia (Bruker).

Preparation of time series for EPR spectroscopy
A time series of DtpB samples following activation by H2O2 was created in two ways. The first procedure required the addition of a stock solution of H2O2 (~10 mM) to a DtpB sample (40 µM) to give a 1:10 ratio (DtpB:H2O2), from which an aliquot was drawn and frozen in methanol kept on dry ice (~195 K). This method provided the freezing time (i.e. the reaction time) from 11 s and upwards. The second procedure to enable sub 10 s sample preparation required the stock H2O2 solution to be inserted into plastic tubing connected to the syringe used to draw the DtpB sample from the EPR tube, which was subsequently loaded back to the EPR tube and frozen, providing freezing times from 4 s. Microspectrophotometry and X-ray data collection at 100 K Crystals of ferric and H2O2 soaked DtpB were cryo-protected in mother liquor containing 20% w/v glycerol and flash-cooled in liquid nitrogen. X-ray diffraction and single crystal spectroscopic data at 100 K were collected at the Swiss Light Source (SLS) beamline X10SA.

Crystallisation and H2O2 soaking of DtpB crystals
The MS3 on-axis microspectrophotometer [3] was used to measure absorbance spectra of ferric DtpB and H2O2 soaked DtpB crystals in the range of 450 to 700 nm. Each spectrum was the result of 50 accumulations of 100 ms exposures. Spectra were measured prior to and following X-ray data collection and a dose limit was selected such that minimal changes occurred to the spectrum of the ferric and ferryl forms during data collection from each crystal. A multi-crystal approach was performed to obtain a complete low-dose composite dataset for the ferric and ferryl DtpB structures. A total of 21 (ferric DtpB) and 13 (ferryl DtpB) spectroscopicallyvalidated diffraction data wedges were merged using the in house go2gether.com script in the XDS package. [4] X-ray absorbed doses were estimated using Raddose-3D. [5] To reflect the beam profile used, a weighted average of doses calculated for top hat and Gaussian profiles was calculated and doses were estimated for the range of crystal dimensions used.

Serial femtosecond X-ray crystallography (SFX)
Silicon fixed-target chips with either 12 or 14 µM apertures at their narrowest opening and a nominal capacity of 25,600 were loaded with 200 µl of microcrystal suspension within a humidity enclosure and sealed between two layers of 6 µm thick Mylar. We have described this approach in more detail previously. [6] SFX data were measured at SACLA beamline BL2 EH3 using an X-ray energy of 11 keV, a pulse length of 10 fs, beam size 1.6 x 1.6 µm and a repetition rate of 30 Hz. Chips were translated within the interval between X-ray pulses, ensuring that the chip had stopped at the centre of each crystal position (the centre of the aperture) and was exposed only once to X-rays, before moving to the next pulse interval. Data was typically collected from all 25,600 positions on a chip in < 15 min using the SACLA MPCCD detector. [7] SFX data were processed using the CHEETAH pipeline [8] and CrystFEL [9] with scaling and merging using the Partialator program.

DtpB structure determination and refinement
The ferric SFX structure was solved by molecular replacement using MrBUMP [10] and BUCCANEER. [11] The search model identified by MrBUMP for molecular replacement was 3QNR. [12] Initial refinement of the structure was carried out in PHENIX, [13] using torsion-angle simulated annealing to eliminate model bias and completed using Refmac5 [14] in the CCP4i2 suite. [15] Model building between refinement cycles was performed in Coot. [16] Riding hydrogen atoms were added during refinement. The ferryl DtpB SFX structure ferric and ferryl 100 K structures were determined from a starting model of the ferric SFX structure, with the same simulated annealing and refinement procedures used. No restraints were placed on the Fe-N e2 His and Fe-O distances. All structures were validated using the Molprobity server, [17] the JCSG Quality Control Server and tools within Coot. [16] Bond length error estimates were obtained using atomic diffraction precision index values obtained from the Online_DPI server. [18] A summary of data collection and refinement statistics are given in Table S1 and   Table S2, respectively. Table S1: SFX (ambient temperature) and composite (100 K) X-ray crystallography data processing for DtpB in space group P212121. Values in parenthesis refer to the outermost resolution shell. The effective absorbed X-ray dose for the SFX structures is assigned as zero due to the femtosecond duration of the X-ray pulse. As is standard practice for SFX data, [19] the metrics Rsplit and CC1/2 are used to assess data quality and resolution limit in place of conventional crystallographic metrics such as Rmerge or I/sd(I).