Mechanisms and Specificity of Phenazine Biosynthesis Protein PhzF

Phenazines are bacterial virulence and survival factors with important roles in infectious disease. PhzF catalyzes a key reaction in their biosynthesis by isomerizing (2 S,3 S)-2,3-dihydro-3-hydroxy anthranilate (DHHA) in two steps, a [1,5]-hydrogen shift followed by tautomerization to an aminoketone. While the [1,5]-hydrogen shift requires the conserved glutamate E45, suggesting acid/base catalysis, it also shows hallmarks of a sigmatropic rearrangement, namely the suprafacial migration of a non-acidic proton. To discriminate these mechanistic alternatives, we employed enzyme kinetic measurements and computational methods. Quantum mechanics/molecular mechanics (QM/MM) calculations revealed that the activation barrier of a proton shuttle mechanism involving E45 is significantly lower than that of a sigmatropic [1,5]-hydrogen shift. QM/MM also predicted a large kinetic isotope effect, which was indeed observed with deuterated substrate. For the tautomerization, QM/MM calculations suggested involvement of E45 and an active site water molecule, explaining the observed stereochemistry. Because these findings imply that PhzF can act only on a limited substrate spectrum, we also investigated the turnover of DHHA derivatives, of which only O-methyl and O-ethyl DHHA were converted. Together, these data reveal how PhzF orchestrates a water-free with a water-dependent step. Its unique mechanism, specificity and essential role in phenazine biosynthesis may offer opportunities for inhibitor development.

. pK a determination of DHHA. 75 ml of 10 mM racemic synthetic DHHA were titrated with 100 mM NaOH and the pH was monitored. The method was validated by titrating L-glycine and malonic acid and comparing to pK a values reported in the literature. Differences to these values were used as standard errors for the pK a values of DHHA.     The density is displayed at 2 σ. Figure S7. Hypothetical mechanism of a charge-induced sigmatropic [1,5]-hydrogen shift in PhzF. The activation barrier for the [1,5]-hydrogen shift was found at 22 kcal/mol in QM/MM calculations and no feasible mechanism for the initial deprotonation of the 3-hydroxy group could be identified.  while DFT methods BP86 for broader investigation 16,17 and B3LYP for final path refinement 12 with 6-31G* basis set and 6-31G** basis sets 18 were used for the QM part. QM/MM calculations were performed using electrostatic embedding and link atoms as implemented in the pDynamo library were introduced to saturate the valence at the QM/MM boundary. The QM-treated active site model comprised of the substrate DHHA (zwitterionic form) and side chains of neighboring polar aminoacids, namely E45 (both protonated and deprotonated version), H74, D208, S213 and a crystal water molecule 407. The rest of the protein and the surface water layer were present in the MM region. Harmonic restraints were applied on MM atoms beyond 8 Å from any QM atom, i.e. the QM region was surrounded by 8 Å of flexible MM layer, between 8 -16 Å the force constants of the restraints linearly increased from 0 to 12 kcal/mol and beyond 16 Å the restraints were set to maximal force constant 12 kcal/mol. The initial reaction state was derived from the CHARMM-hydrogenated crystal coordinates by geometry optimization. The potential energy surface scans followed by geometry optimizations were then used for fundamental investigation of possible reaction pathways. In a typical scan, a hydrogen atom was moved in discrete 0.2 Å steps breaking a bond to its donor and subsequently forming a new bond with its acceptor. After each step, the position of the hydrogen in question was fixed, while the rest of the QM/MM model was optimized. After moving the hydrogen to a typical bond distance from the acceptor, the geometry was minimized without restraints applied on the hydrogen in question. An RMS gradient of less than 0.01 kcal/(mol * Å) was used as a minimization convergence criterion. The promising scan-derived pathways were refined by a modified Nudge Elastic Band (NEB) method 19 and the Conjugate Peak Refinement (CPR) method 20 . Alternatively, CPR was also used to find the reaction pathway between two stable states (eg. adduct and intermediate state) de novo. The procedure described above was carried out using the BP86 DFT functional with as the QM method. Since BP86 is known to underestimate the energies while producing reasonable geometries, energetically feasible pathways were further optimized using CPR with the computationally more demanding hybrid B3LYP as QM method. Vibration frequency calculations were performed on the B3LYP level as well. To make the frequency calculations computationally feasible, the MM region was kept completely fixed. Vibration frequencies were used to verify the minima and transition states of the final reaction coordinate and furthermore for the calculation of zero point energy correction. The kinetic isotope effect of the initial reaction step was calculated from the difference in vibrational frequencies between normal and C3-deuterated variant of DHHA and the first transition state 21 .

Protein expression and purification
PhzF from Pseudomonas fluorescens 2-79 was expressed and purified as described previously 22 . In brief, the protein (UniProt-entry: Q51792 (PHZF_PSEFL)) was cloned into a pET15b expression vector (Novagen) and overexpressed in E. coli Rosetta (DE3) pLysS (Novagen) in LB-(Luria Bertani) medium after induction with 0.5 mM Isopropyl β-D-1thiogalactopyranoside (IPTG) at an OD 600 of 0.6 at 20 °C over night. Purification of the Nterminal hexahistidine-tagged protein followed a standard two-step procedure starting from immobilized Ni 2+ -affinity chromatography followed by a size-exclusion step in assay (50 mM sodium phosphate pH 7.5) or crystallization buffer (20 mM TRIS/HCl pH7.5, 150 mM NaCl) supplemented with 10% (v/v) glycerol. Purified protein was concentrated in the respective buffer, snap-frozen in liquid nitrogen and stored at -80 °C until further usage. Site-directed mutagenesis was performed with the QuikChange protocol (Stratagene).

Enzyme assay
Enzyme kinetic parameters were determined in an Infinite® 200 microplate reader (Tecan Group Ltd.) with UV-Star® 96-Well microplates (Greiner Bio-One) at 25°C. After determination of a suitable pH value ( Figure S3A), the assay was performed in 50 mM sodium phosphate pH 7.5, 1% (v/v) DMSO using up to 1 mM DHHA in a final volume of 100 µL and the reaction initiated by the addition of 40 nM PhzF dimer. Substrate depletion was followed at 275 nm for 20 minutes after reaching a linear phase, using an experimentally determined extinction coefficient of 6500 M -1 cm -1 for DHHA. Typical UV-time traces are shown in Figure S3B. The enzyme concentration was increased to 200 and 1000 nM to obtain measurable rates for 2S,3S-3-deutero-DHHA and racemic O-Et-DHHA, respectively. All experiments were performed in triplicate and rate constants were derived by fitting to a Michaelis-Menten model in GraFit5 (Erithacus Software Ltd.). Enzyme kinetic parameters are listed in Table S1.  25 and maximum likelihood refinement in phenix.refine 26 of the PHENIX software suite 27 . Ligand restraints were generated using the program phenix.elbow 28 of the latter software suite. With respect to the apparent resolution of 1.7 Å structural flexibility was modelled using TLS (Translation/Libration/Screw) refinement 29 . 8 TLS groups were identified by the program phenix.find_tls_groups, implemented in phenix.refine. In the last step of refinement, non-protein residues (water and ligands) were attributed to their nearest TLS group using a script developed in our group (Reichelt & Blankenfeldt, unpublished). Final structure validation was done with MolProbity 5 . Diffraction data and coordinates have been deposited in the Protein Data Bank 30 (PDB entry 5IWE for the complex with O-Et-DHHA). Full data collection and refinement statistics are shown in Table S2.

Reagents and analytical methods for the synthesis of DHHA and DHHA derivatives
Reagents and solvents were purchased from Sigma-Aldrich, Alfa, Aesar, ABCR, Fisher Scientific, Acros Organics, Roth or VWR and dried and/or degassed as required.
Pig liver esterase (PLE) was purchased from Fluka as a technical precipitate in a half saturated (NH 4 ) 2 SO 4 solution. 100 mM NaH 2 PO 4 /Na 2 HPO 4 buffer at pH 7.6 was added to provide a constant pH.
Analytical thin layer chromatography (TLC) was carried out on Merck TLC silica gel aluminium sheets (silica gel 60, F254, 20 x 20 cm). Flash column chromatography was performed on silica gel 60 from Acros Organics (particle size 35 -70 µm). 1 H-, 13 C and 19 F-NMR spectra were recorded on a Bruker AVANCE III 300 or a Varian/Agilent Unity Inova 500 spectrometer. Chemical shifts were referenced to tetramethylsilane as internal standard or to the residual proton and carbon signal of deuterated solvents, which were purchased from euriso-top®. CDCl 3 was neutralized by filtering through activated aluminium oxide, basic type 5016A, 58 Å, particle size: 150 mesh, Brockmann Grade I from Acros Organics.
HRMS spectra were recorded on a Waters Micromass GCT Premier system either directly or after gas-chromatographic ionization on a Hewlett Packard BC 7890A system, using electron impact (EI) for ionization. Molecule ions were analyzed by a time-of-flight mass analyzer in the positive mode.
Melting points were determined on a Mel-Temp melting point apparatus from Electrothermal, specific optical rotations were measured at 589 nm on a Perkin Elmer Polarimeter 341.

Synthesis of DHHA (9)
Enantioselective synthesis of DHHA in form of a TFA salt 9 in analogy to the published synthesis of Steel 31 .

(1R,2S,3S,4S)-3-((tert-Butoxycarbonyl)amino)-7-oxabicyclo[2.2.1]hept-5-ene-2carboxylic acid (6) 31
Five 250 mL round-bottom flasks equipped with a Teflon-coated magnetic stirring bar were charged with a solution of 984 mg (3.47 mmol, 1.0 eq) bicyclic ester rac-5 in 40 mL Et 2 O. After the addition of 70 mL NaH 2 PO 4 /Na 2 HPO 4 buffer (pH 7.6, c = 100 mM) and 3.6 mL PLEprecipitate (in half saturated (NH 4 ) 2 SO 4 , unknown activity), the yellowish two-phasic mixture was stirred with 200 rpm in the closed flask at RT for 22 h until the non-hydrolyzed enantiomer of compound ent-5 reached an e.e. value between 94-96 % (E = 180). The yellowish reaction mixture was phase separated and the aqueous phase was washed with Et 2 O (2 x 600 mL). In order to obtain a better phase separation the mixture was centrifuged. The yellowish aqueous layer was acidified with 50 mL of saturated KHSO 4 to pH 1-2 and the product was extracted with DCM (4 x 500 mL). The combined yellowish organic layers were dried over MgSO 4 , filtered and the solvent was removed under reduced pressure on a rotary evaporator. For purification the brownish crude material was triturated in 15 mL cyclohexane:EtOAc = 2:1 (v/v), collected by filtration, washed with cold cyclohexane:EtOAc = 2:1 (v/v) (2 x 2.0 mL) and the resulting colorless powder was dried under high vacuum.
In order to obtain more precise e.e. values, this procedure was repeated with 2.10 g (7.41 mmol, 1.0 eq) of the enantiomerically enriched bicyclic ester 5 (e.e. = 96 %) as starting material. After reaching an e.e. = 38 % of non-hydrolyzed ester ent-5, the work-up as well as the purification was performed as described above.

Potassium hexamethyldisilazide (KHMDS) (7) 34
An oven-dried, evacuated and argon purged 100 mL three-necked round-bottom flask equipped with a Teflon-coated magnetic stirring bar, gas inlet adapter, reflux condenser and bubbler was charged with 2.36 g (20.6 mmol, 1.1 eq) KH suspension (35 % (w/w) in mineral oil). It was dispersed in 10 mL anhydrous n-hexane, the supernatant was removed and the remaining solid was washed with n-hexane (3 x 5 mL) as described above. Afterwards the greyish solid was carefully dried under high vacuum, diluted with 20 mL anhydrous THF and 3.8 mL (18.4 mmol, 1.0 eq) 1,1,1,3,3,3-hexamethyldisilazane were added in one portion. The greyish suspension was ultrasonicated in an oil bath for 4 h under argon atmosphere. After a short induction time, heavily gas bubbling was observed and a yellowish suspension was formed. Subsequently, the yellowish suspension was allowed to stand at RT under argon atmosphere for 15 h and the supernatant was transferred into an oven-dried, evacuated and argon purged 80 mL Schlenk flask by cannuling. It can be stored in a freezer at -24 °C under argon atmosphere for several weeks without significant degradation, but has to be titrated before use (see below). Concentration: 0.76 M (after titration with 1.0 M 2-butanol (in toluene) and 2-(6-butyl-1,6dihydropyridin-2-yl)pyridine (12) in THF as indicator solution).

Synthesis of d-DHHA (20)
Synthesis of d-DHHA in form of a TFA salt 20 in analogy to the published synthesis of DHHA (9) by Steel. 31

2-Bromofuran (15) 36
An oven-dried, evacuated and argon purged 250 mL three-necked round-bottom flask equipped with a Teflon-coated magnetic stirring bar, gas inlet adapter, 100 mL dropping funnel, internal thermometer and bubbler was charged with a solution of 15.3 g (225 mmol, 2.0 eq) furan in 40 mL anhydrous DMF. Afterwards a deeply orange solution consisting of 20.0 g (112 mmol, 1.0 eq) N-bromosuccinimide (NBS) in 60 mL anhydrous DMF was added via the dropping funnel at RT over a period of 50 min. The temperature did not exceed 35 °C, and the brownish reaction mixture was vigorously stirred at RT for additional 5 h. Afterwards the brown solution was concentrated on a rotary evaporator (10 mbar, 35 °C) to remove the excess of unreacted furan. The remaining solution was purified by water steam distillation. Therefore the storage vessel filled with H 2 O was heated in an oil bath to 140 °C, the brown solution, however, in a second oil bath to 105 °C. Due to contamination with furan, the first few drops of condensate were discarded. The collected colorless condensate was washed with H 2 O (1 x 30 mL) to remove residual DMF and finally the colorless, clear product was stored in an inert 25 mL Schlenk flask over dried K 2 CO 3 under argon atmosphere in a freezer at -20 °C for several weeks.

Ethyl (1R,2S,3S,4S)-3-((tert-butoxycarbonyl)amino)-7-oxabicyclo[2.2.1]-hept-5-ene-2carboxylate-4-d (rac-17)
An oven-dried 250 mL three-necked round-bottom flask equipped with a Teflon-coated magnetic stirring bar, gas inlet adapter and bubbler was charged with a colorless solution of 3.65 g (12.5 mmol, 1.0 eq) bicyclic Diels-Alder compound 16 in 110 mL EtOD and afterwards cooled in an ice-water bath to 0 °C. Consecutively, 25.8 mL conc. DCl (~38 % (w/w) in D 2 O, 99.5 atom% D) and 24.5 g (375 mmol, 30.0 eq) Zn/Cu couple (max. 3 % Cu) were added in small portions to the colorless solution in an argon counter flow. Immediately after the addition of the Zn/Cu couple intense D 2 gas formation was observed. After 30 min of vigorously stirring at 0 °C, the ice-water bath was removed and the grey suspension was additionally stirred at RT for 10 h. It was filtered through a pad of anhydrous Celite ® (diameter: 3.0 cm, height: 4.0 cm) and the filter cake was washed with EtOD (2 x 20 mL). The filtrate was collected in an oven-dried 500 mL round-bottom flask equipped with a Teflon-coated magnetic stirring bar. 52.4 mL (300 mmol, 24.0 eq) DIPEA and 4.91 g (22.5 mmol, 1.8 eq) Boc 2 O were added successively and the resulting colorless suspension was vigorously stirred at RT for 16 h. Subsequently it was carefully concentrated under high vacuum, the colorless solid residue was diluted with 600 mL EtOAc, washed with H 2 O (1 x 600 mL) and the cloudy, colorless aqueous phase was reextracted with EtOAc (2 x 600 mL). The combined yellowish organic layers were washed with saturated NaHCO 3

Propyl trifluoromethanesulfonate (34)
An oven-dried, evacuated and argon purged 50 mL Schlenk flask equipped with a Tefloncoated magnetic stirring bar was charged with a colorless solution of 1.33 mL (16.5 mmol, 1.10 eq) pyridine in 15 mL anhydrous DCM and afterwards cooled in an acetone/dry ice bath under argon atmosphere to -20 °C. To this cooled, colorless solution 2.65 mL (15.8 mmol, 1.05 eq) Tf 2 O were added via a syringe and septum over a period of 10 min, which resulted in the formation of a colorless suspension. This suspension was additionally stirred at -20 °C in the acetone/dry ice bath under argon for 10 min. Afterwards 1.12 mL (15.0 mmol, 1.00 eq) n-PrOH were added dropwise via a syringe over a period of 10 min, the cooling bath was removed and the colorless suspension was stirred under argon at RT for 15 min. It was filtered through a Schlenk frit, the filter cake was washed with anhydrous DCM (2 x 10 mL) and the colorless filtrate was carefully concentrated in the vacuum of an oil pump to approximately 5 mL, which resulted in the precipitation of a colorless solid. Subsequently the colorless suspension was treated with 30 mL anhydrous n-pentane and again filtered through a second Schlenk frit in an oven-dried, evacuated and argon purged 80 mL Schlenk flask. The filter cake was washed with anhydrous n-pentane (2 x 5 mL) and the solvent of the colorless filtrate was carefully removed in the vacuum of an oil pump. Finally, the brownish, oily residue was dried at 5 mbar for 5 min. It was immediately used in the propylation step without further purification.

Ethyl (5S,6S)-6-((tert-butoxycarbonyl)amino)-5-propoxycyclohexa-1,3-diene-1carboxylate (rac-35)
An oven-dried, evacuated and argon purged 50 mL two-necked round-bottom flask equipped with a Teflon-coated magnetic stirring bar and gas inlet adapter was charged with a colorless solution of 425 mg (1.50 mmol, 1.0 eq) racemic ester rac-8 in 10 mL anhydrous Et 2 O and to the colorless solution 695 mg (2.00 mmol, 2.0 eq) Ag 2 O as well as 3.0 g 3 Å MS were added in an argon counter flow, respectively, which resulted in the formation of a black suspension. In parallel 2.88 g (15.0 mmol, 10.0 eq) of the freshly prepared propyl trifluoromethanesulfonate (34) were dissolved in 7.5 mL anhydrous Et 2 O under argon in the same Schlenk flask, in which this compound was dried. This solution was added to the black suspension via a syringe and septum in one portion, the Schlenk flask was rinsed with anhydrous Et 2 O (1 x 2.5 mL) and the black suspension was vigorously stirred in the closed flask at RT for 120 h. Afterwards it was filtered through a pad of Celite ® (diameter: 3.0 cm, height: 4.0 cm), the filter cake was washed with EtOAc (4 x 50 mL) and the solvent was removed on a rotary evaporator. Finally, the orange, viscous crude material was purified via flash column chromatography (80 g SiO 2 , 25.0 x 2.5 cm, eluent: cyclohexane:EtOAc = 9:2 (v/v), R f = 0.24) and the resulting yellowish, highly viscous liquid was dried under high vacuum.