Trapping conformational states of a flavin-dependent N -monooxygenase in crystallo reveals protein and flavin dynamics

,

SidA gene, the inhibition of an equivalent host enzyme is not a restricting factor during treatment or prophylactic use.
SidA is essential for A. fumigatus virulence because it is required for siderophore-dependent iron acquisition (8).Iron is an essential nutrient for growth and proliferation of organisms.Iron in the host organism is sequestered by iron-binding proteins, making it a restricted nutrient for A. fumigatus and other invasive organisms to acquire (9,10).Fungi and other infectious organisms have specific metabolic pathways to produce iron-chelating molecules called siderophores (9).Siderophores are peptide-based scaffolds containing iron-chelating groups such as carboxylates, catecholates, and hydroxamates (11), and are used by microorganisms to extract iron from low-iron environments (9,12).SidA catalyzes the hydroxylation of L-orni- thine (L-Orn) to N 5 -hydroxyornithine ( S1 Scheme 1) (12,13), which is the first step in the biosynthesis of hydroxamate-containing siderophores ( ZSI Fig. S1).This is a key step, because the hydroxyl group added in the SidA reaction is part of the ironchelating hydroxamate group of the mature siderophore ( ZSI Fig. S1).Without the SidA step, the siderophore lacks three of the six oxygen ligands to iron, which would significantly reduce the affinity.The essential role of SidA in siderophore biosynthesis makes it an attractive drug target to impair iron uptake by pathogens (8).
SidA is an N-hydroxylating flavin-dependent monooxygenase (NMO) and belongs to a family of enzymes that hydroxylate soft nucleophiles such as the primary amines of lysine and ornithine (14).SidA is a class B flavin monooxygenase, which is characterized by both a high selectivity for its substrate and by NADP 1 remaining bound throughout most of the catalytic cycle ( S2 Scheme 2) (as opposed to the class A family which are more promiscuous and release NADP 1 upon reduction of the flavin) (15)(16)(17).SidA contains a tightly bound FAD, which is oxidized in the ligand-free resting state.The catalytic cycle is initiated by the binding of NADPH, resulting in the reduction of the FAD (Scheme 2) (2,14).The FAD red -NADP 1 AQ:F complex reacts with molecular oxygen to form a C4a-hydroperoxyflavin (FAD OOH ), with NADP 1 remaining bound to the enzyme to stabilize the FAD OOH and thereby preventing H 2 O 2 release (i.e.uncoupling) (16).L-Orn is then hydroxylated by FAD OOH , and finally N 5 -hydroxyornithine and NADP 1 are released to complete the cycle (18).
Several crystal structures of SidA and the related ornithine hydroxylases, PvdA and KtzI, have been determined in various redox and ligation states ( ZSI Table S1) (19)(20)(21)(22).PvdA and KtzI are 37.3% and 33.7% identical in sequence to SidA, respectively, and the three enzymes share a conserved fold and tetrameric structure.The protein fold is composed of three domains: an ornithine binding domain and two Rossmann fold domains, one of which binds FAD and the other NADP(H).The domains are arranged to bring together the re-face of isoalloxazine of the FAD, the nicotinamide of NADP(H), and the N5 atom of ornithine for catalysis.These structures have provided information about residues that contribute to substrate and cofactor recognition and insight into the catalytic mechanism.
Motion of the flavin is an important aspect of some flavoenzymes.For example, it has been well established in p-hydroxybenzoate hydroxylase, and other related enzymes, that substrate binding induces flavin motion to an exposed position, where it can react with NADPH (16,23).Flavin reduction then induces motion back to the active site, where catalysis can occur.Also, the flavins in some flavin switch proteins exhibit large redox-linked conformational changes (24).Evidence for flavin dynamics is also emerging for ornithine hydroxylases (22).The SidA structures all show the FAD in a conserved wellordered state, which will hereafter be termed the in conformation.In this conformation the si-face of the isoalloxazine is bur-ied and the re-face is primed for catalysis.All previously determined SidA structures show the flavin in this in position, regardless of oxidation state or bound ligands ( Table S1).The in position of the flavin was also observed in PvdA structures in both the oxidized and reduced states ( Table S1) (19)(20)(21).In contrast, Setser et al. (22) described a novel out conformation of the oxidized flavin in KtzI in which the re-face is buried and not available for catalysis.They also showed that the flavin redox state controls movement between the in and out conformations ( ZSI Table S1).This work suggested that flavin dynamics may be an important aspect of the catalytic mechanism of ornithine hydroxylases.
Here  S2).The structures reveal information about the role of dynamics in the SidA catalytic mechanism, which could be exploited to identify potential new binding sites for inhibitors.

New crystal form reveals a new flavin conformation
The structure of ligand-free oxidized SidA in a new P2 1 AQ:H crystal form was determined at 2.09Å resolution.This structure represents the resting state of the enzyme, prior to the binding of NADPH.The color of the crystal used for data collection was yellow, consistent with the oxidized state, and no reducing agent was included during crystallization and cryoprotection.The electron density map clearly indicated the conformation of the FAD ( F1 Fig. 1A).Surprisingly, the isoalloxazine conformation is much different from all previous SidA structures (Fig. 1B).In the new conformation, the re-face of the isoalloxazine is buried by the protein, whereas the si-face stacks against Tyr 324 (Fig. Trapping conformational states of SidA in crystallo 1B).Thus, neither face is available for catalysis.The new structure also shows protein conformational changes that accompany movement of the isoalloxazine.The largest of these involves Tyr 324 , which moves by 10 Å.The new conformation of the FAD is similar to the out conformation described for KtzI (22).The movement of the FAD between the in and out conformations appears to involve rotations around two dihedral angles, one of ;90°around the C19-C29 bond and another of ;180°around the N10-C19 bond (Fig. 1C).The net result is that the pyrimidine edge moves by 9.5 Å, although the ribityl chain hydroxyl groups are left undisturbed.

Flavin reduction or NADP(H) binding induces movement of the FAD to the in conformation
Crystal soaking and co-crystallization were used to identify conditions that induce movement of the FAD to the in conformation observed in all previous SidA structures.Soaking crystals of oxidized, ligand-free SidA with sodium dithionite and L-Orn bleached the yellow color of the crystals, consistent with flavin reduction.In the resulting structure, the isoalloxazine has rotated to the in conformation and Tyr 324 has shifted away from the active site ( F2 Fig. 2B).Electron density for L-Orn is strong and indicates binding in the expected conformation ( ZSI Fig. S2C).Similarly, SidA crystals grown in the presence of NADP 1 also show the in conformation of the active site (Fig. 2C).The in conformation was modeled at 100% occupancy in three chains of the asymmetric unit, whereas the density was consistent with dual occupancy of the in and out conformations in chain C (53% in, 47% out).Electron density for NADP 1 was strong in all four chains, and the ligand was modeled in the conformation observed in other SidA structures ( ZSI Fig. S2B).Finally, a crystal of the SidA ox AQ:I -NADP 1 complex was soaked with NADPH and L-Orn.Soaking bleached the yellow color of the crystals, indicating that NADPH had displaced NADP 1 and reduced the flavin.The structure shows that the in conformation was maintained in all subunits (Fig. 2D).
Altogether, the structures show that either flavin reduction or cofactor binding is sufficient to promote movement of the FAD from out to in, and that Tyr 324 moves in concert with the FAD ( F3 Fig. 3A).We note that Tyr 324 is part of the Tyr-loop (Asn 323 -FTyr 324 -Ser 325 ), which is highly conserved among ornithine hydroxylase sequences (see alignment in ZSI Fig. S3).The transition from out to in dramatically changes the structural relationship between Tyr 324 and the FAD.In the out conformation, the two make intimate contact of 3.3 Å (Fig. 3B), whereas they are 9.4 Å apart in the in conformation (Fig. 3C).The stacking of Tyr 324 against the isoalloxazine in the out conformation appears to block the isoalloxazine's path to the in conformation, necessitating a concerted 10-Å movement of Tyr 324 to allow the FAD to rotate around its C19-C29 and N10-C19 bonds into the in conformation.Movement of Tyr 324 is also required for the binding of NADPH.In the out conformation of the active site, the stacking  pose of Tyr 324 occupies the nicotinamide binding site, necessitating a conformational change to allow cofactor binding (Fig. 3D).

Conformational changes associated with ligand binding
By capturing the first structure of SidA in the absence of NADP(H) and L-Orn, we are able to describe the protein conformational changes that accompany cofactor and substrate binding.In addition to Tyr 324 described above, notable movements involve Asn 323 and Ser 325 of the Tyr-loop.Ser 325 rotates 130°around its x Conformational changes in residues near the Tyr-loop are also observed.Asn 293 (near Asn 323 ) rotates to hydrogen-bond to the backbone of L-Orn (Fig. 4, A to B and C to D).Also, Arg 279 (near Ser 325 ) rotates to ion-pair with the adenosine phosphate of the cofactor (Fig. 4, A to C).

Comparison of SidA and KtzI structures
The resting states (oxidized, ligand-free) of SidA ox and KtzI ox (Protein Data Bank Entry 4TM3) share similarities in the  Trapping conformational states of SidA in crystallo   positioning of the active site residues and the flavin ( F5 Fig. 5, A, E,  and H).In both structures, the oxidized flavin is in the out position with the re-face of the FAD buried and the si-face stacking with a tyrosine (Tyr 324 in SidA, Tyr 276 in KtzI).This shows a conserved active site structure in the resting state of the two ornithine hydroxylases.
The pre-turnover reduced states of SidA and KtzI are also very similar.This state is represented by the structures of SidA red -NADPH-Orn and KtzI red -NADP-Orn (Protein Data Bank Entry 4TLX).In both structures, the flavin is in and the Tyr-loop has been displaced by the NADP(H) cofactor (Fig. 5,  C, G, and J).
Curiously, the structures of the oxidized enzymes with NADP 1 are different.SidA ox -NADP 1 has the FAD in (Fig. 5B), whereas KtzI ox -NADP 1 -Orn has the FAD out (Fig. 5F).In both structures the Tyr-loop makes an obligatory shift away from the active site to avoid a steric clash with NADP 1 .These two structures reveal different factors that induce the movement of the isoalloxazine ring.In SidA, we see that the oxidized FAD moves in with NADP 1 binding and the nicotinamide of NADP 1 stacks with the re-face of the isoalloxazine.This suggests that for SidA, cofactor binding can induce the movement of the FAD into the active conformation.In KtzI, the FAD remains out even upon cofactor binding.The authors attributed the out position of the flavin to its oxidized state and suggested that reduction was required to move the FAD to the in conformation (22).Thus, the timing of flavin motion may differ in the two enzymes.As discussed below, our results suggest that the hydride transfer step in SidA could occur with the FAD in an out conformation.

Site-directed mutagenesis
We investigated residues that the structures suggest could be important in movement between the in and out active site conformations.The stacking residue Tyr 324 was mutated to Ala (Y324A) and Phe (Y324F).Mutation to Ala appears to negatively impact protein folding/solubility, because we were unable to recover any soluble Y324A protein for analysis.Y324F was soluble and exhibited kinetic parameters substantially different from WT.The k cat measured in the oxygen consumption assay was decreased 3-fold, whereas the K m AQ:K values for L-Orn and NADPH were increased by ;10-fold ( T2 Table 2).Thus, the catalytic efficiencies (k cat /K m ) of Y324F are ;30 times lower than WT.We note that removing the hydroxyl of Tyr 324 impacts the kinetic parameters much more than removing the side chain of Asn 323 , a residue that hydrogen-bonds to both NADPH and L-Orn (N323A, Table 2).In particular, the catalytic effi-ciency of N323A for NADPH is essentially unchanged from WT, whereas that of Y324F is 35 times lower.
We also investigated His 91 with the mutant variant H91A.His 91 consistently hydrogen-bonds with the ribityl chain in both the in and out conformations of the FAD ( ZSI Fig. S4).The ribityl chain conformation is identical in both conformations, suggesting that residues that interact with the ribityl may be important for limiting the conformational space of the FAD.Mutation of His 91 to Ala causes no change in k cat and only a modest 2.5-fold increase in the K m values (Table 2).However, the coupling efficiency of H91A is decreased to 28%, meaning that for every mole of oxygen consumed, only 0.28 mole of HO-N5-Orn is formed.For comparison, the coupling efficiencies of the WT enzyme and Y324F are close to 100% (Table 2).

Probing the out active site of SidA with docking
The out conformation of the active site is potentially a new target for inhibitor discovery.We used a fragment-mapping approach (25,26) to identify potential ligand-binding hot spots.In this approach, sixteen small organic molecules were exhaustively docked to the protein, and the areas where the probe molecules cluster are potential ligand-binding sites.
A control fragment-mapping calculation performed on the in structure (Protein Data Bank Entry 6X0I) yielded a large cluster of probes in the NADPH binding site ( F6 Fig. 6A).In particular, the probes bound preferentially to the nicotinamide dinucleotide site.A smaller cluster of probes corresponded to the d-amino group of L-Orn (Fig. 6A).This control calculation shows that exhaustive docking of small organic molecules has the potential to identify bona fide ligand-binding sites in SidA.
An analogous fragment-mapping calculation performed on the out conformation also yielded two clusters (Fig. 6B).A large, dense cluster of probes was found in a pocket surrounded by Tyr 324 , Gln 102 , and Arg 238 .Note this pocket is separated from the FAD isoalloxazine by Tyr 324 .The unique conformation of the Tyr-loop makes this pocket exclusive to the out state.A more diffuse cluster of probes was found on the surface of the protein near the pyrimidine edge of the FAD (Fig. 6B, cluster 2).This potential binding surface does not exist in the in state, because the pyrimidine edge of the isoalloxazine is inaccessible (Fig. 6C).In contrast, the pyrimidine edge pokes out of the NADPH entrance in the out state (Fig. 6D).
The two potential ligand-binding sites of the out conformation were explored further by docking larger compounds.A total of 3035 compounds with molecular weights in the range of 150-300 g/mol were docked to each site.Histograms of the docking energies for the top poses of each compound to the   S3).The physicochemical properties of the top 10 compounds are similar for the two sites ( ZSI Table S5).In fact, one compound appeared in the top 10 lists of both sites (Fig. 7, B and C).
Viewing the docked compounds provides insight into the size, shape, and chemical characteristics of the potential binding sites.Site 1 is a true pocket that surrounds the ligand (Fig. 7B).The pocket is connected to two tunnels, one leading to the protein surface and another leading to the FAD ribityl chain.Some of the docked compounds have functional groups that access the latter tunnel and interact with the ribityl 39-hydroxyl.Site 1 has many side chains available for interacting with ligands.The constellation of side chains is diverse, including aromatic, nonpolar, polar, and charged groups (Fig. 7B).
Site 2 is rather different.It is a shallow cleft, and the docked ligands have significant solvent-exposed surface area (Fig. 7C and ZSI Fig. S7).Fewer side chains interact with the docked ligands, and the diversity of side chains in the cleft is lower than that of site 1 (Fig. 7C, right).Two binding modes were observed among the top 10 compounds.In one mode, the compound accesses the N5 edge of the FAD isoalloxazine (Fig. 7C), whereas in the other mode, the compound covers the pyrimidine edge of the FAD ( ZSI Fig. S7).

Discussion
Here we described a new crystal structure representing the resting state of SidA.The active site is different from all other SidA structures in that the FAD has rotated so that the active face of the isoalloxazine (re-face) is buried by the protein and the si-face is stacked against the eponymous residue of the conserved Tyr-loop.The resting conformation may offer new opportunities for inhibitor discovery.The docking of probe molecules suggested a preference for a binding site exclusively found in the resting state.Furthermore, this site was lined with residues advantageous for design of inhibitors specific to this conformational state.We also showed by mutagenesis that the stacking residue, Tyr 324 , is essential for proper catalytic function.
Although atypical for SidA structures, the out FAD conformation is very similar to a structure of the resting state of the ornithine hydroxylase KtzI (22).Setser et al. (22) referred to the motion of the FAD as "flapping," although as mentioned above, the true molecular motion likely involves two dihedral angle rotations (Fig. 1C).The observation of this conformational change in two ornithine hydroxylases with only 34% identity suggests that it may be a conserved dynamic feature of ornithine hydroxylases.
The role of flavin dynamics in the catalytic mechanism remains to be determined.Setser et al. (22) have proposed that  it helps eject spent NADP 1 from the active site following turnover to allow another round of catalysis.Surprisingly, the determinants of flavin dynamics may be different in KtzI and SidA.In KtzI, the oxidation state of the FAD seems to be the key factor, with the reduced FAD always in and the oxidized FAD always out.In SidA, reduction of the FAD also promotes the in state (Fig. 2B), but so does co-crystallization of the oxidized enzyme with NADP 1 (Fig. 2C).Thus, it appears that for SidA, both flavin reduction and cofactor binding may be determinants of the FAD conformation.These results suggest that the role of flavin dynamics in the mechanism is complicated and remains to be unraveled.
Because the out conformation is a feature of the resting enzyme, it presumably plays a role in the first step of catalysis, when the FAD is reduced by NADPH (Scheme 2).One possibility is that NADPH reduces the flavin while it is in an out conformation.This would require movement of the Tyr-loop, which protects the N5 of the FAD in the out state.The reduced flavin would then rotate to the in position in unison with NADP 1 shifting into the conformation seen in the crystal structures, which is thought to be optimal for stabilization of FAD OOH .Hydride transfer from NADPH to an out-like FAD could potentially address the conundrum of the stereochemistry of hydride transfer for class B NMOs.All crystal structures of class B monooxygenases complexed with NADP(H) imply proS stereochemistry (e.g.Fig. 2, C and D), yet kinetic studies indicate proR stereochemistry (13,20,27,28).Capturing structures of class B NMOs with NADP(H) bound and the FAD in out-like conformations represents a new challenge in flavoenzyme structural biology.

Protein cloning, expression, and purification
An N-terminally truncated SidA construct was generated by deleting the coding sequence for the first 28 residues of the enzyme (SidAD1-28).The truncated gene was cloned into pET15b (Novagen).In the pET15b plasmid, SidAD1-28 is expressed with an N-terminal His-tag followed by a thrombin cleavage site.This truncation was chosen because these residues are disordered in previously published structures of SidA (20,21).We note these N-terminal residues are not present in some other ornithine hydroxylases, including KtzI and PvdA (see alignment in ZSI Fig. S3).SidAD1-28 and full-length SidA were expressed in Escherichia coli and purified using previously described methods (20).In brief, the protein was expressed in Turbo BL21 (DE3) cells (Genlantis C302020, Fisher Scientific) overnight at 25 °C.The cells were harvested by centrifugation and lysed by sonication in a buffer containing 25 mM HEPES (pH 7.5), 300 mM NaCl, 25 mM imidazole, and 25 mg/ml each of lysozyme, DNase I, and RNase.The protein was purified over a nickel-nitrilotriacetic acid column, washed with 120 mM imidazole, and eluted with 300 mM imidazole.The His-tag was cleaved from fulllength SidA using thrombin as described in previous methods (20) AQ:M .The His-tag was not cleaved from SidAD1-28.SidAD1-28 and full-length SidA for crystallization were dialyzed into a storage buffer containing 25 mM HEPES (pH 7.5) and 100 mM NaCl, flash-frozen in liquid nitrogen, and stored at 280 °C.

Mutagenesis and steady-state kinetics
The WT gene in pET15b was used as a template to make the following amino acid substitutions: Y324A, Y324F, and H91A.Site-directed mutagenesis was performed following the QuikChange system (Agilent Technologies).The genes were sequenced to ensure that the mutations were incorporated.Protein expression and purification was performed as previously described (20).In general, the protein yield was similar to WT SidA, with the exception of Y324A, which was insoluble.The purified proteins were flash-frozen in liquid nitrogen and stored at 280 °C.
The enzyme activity was monitored by measuring either the rate of oxygen consumption or the rate of formation of hydroxylated L-Orn.The rate of oxygen consumption was measured using an Oxygraph system (Hansatech Instruments).Reactions consisted of a 1-ml volume of 100 mM sodium phosphate, pH 7.5, at 25 °C.Hydroxylated ornithine was monitored using a variation of the Csaky iodine oxidation assay (20,27).The standard assay buffer contained 104 ml of 100 mM sodium phosphate (pH 7.5).NADPH was held constant at 0.5 mM when L-Orn was varied for SidA, H91A (0.1-10 mM), and Y324F (0.1-50 mM).When NADPH was varied (0.01-1.0 mM), L-Orn was present at 10 mM for SidA and H91A and at 60 mM for Y324F.The data were fit to the Michaelis-Menten equation to estimate kinetic parameters.

Crystallization of SidAD1-28
A new crystal form was identified by screening SidAD1-28 using the Crystal Screen 1 and 2 kits and the Index kit (Hampton Research).Initial screening was done with 8 mg/ml SidAD1-28 in the presence of 1 mM NADP 1 and 15 mM L-Orn.Small, yellow rod-shaped crystals formed in a condition containing 20% PEG-3350, 0.1 M MgCl 2 , and 50 mM HEPES (pH 7.5), so this condition was chosen for optimization.Crystal optimization was performed in 24-well plates using the hanging drop vapor diffusion method.Replacing MgCl 2 with various salts and optimizing the concentrations of PEG-3350 and HEPES led to the discovery of conditions that yielded larger, more reproducible crystals, which could be grown using either SidAD1-28 or full-length SidA, although typically better crystals resulted using SidAD1-28.Yellow rod-shaped crystals (0.1-0.2 mm) were reproducible in 17-21% PEG-3350, 0.1 M HEPES (pH 7.5), and 0.1 M calcium acetate.During crystal optimization, crushed seeds from the initial hit were used to streak-seed new drops, using a horse hair.The drop ratio was optimized, resulting in an optimal drop ratio of 2:1 (protein: reservoir).All crystal solutions were prepared with 0.05% sodium azide to prevent fungal growth.
The crystals used in data collection were grown using a reservoir solution of 20% PEG-3350, 0.1 M HEPES (pH 7.5), and 0.1 M calcium acetate, at 20 °C.Microseeding from crushed crystals grown during the optimization stage was important for improving the crystal quality.The hanging drops used to produce crystals for X-ray diffraction consisted of 4 ml SidA (8-10 Trapping conformational states of SidA in crystallo mg/ml) and 2 ml of the reservoir solution.The volume of the reservoir solution was 500 ml.Upon mixing, the protein solution immediately appeared heavily precipitated, and 24 h later the drops appeared clear.Crystals usually appeared after 2-3 days and grew along the streak-seed line.We note that if microseeding was not used, very large (;1 mm), irregularly shaped crystals that diffracted poorly resulted.The SidA ox ligand-free structure was determined from a crystal of SidAD1-28 grown in the absence of L-Orn or NADP (H).The ligand-free crystals were yellow, consistent with the oxidized state of the FAD.The crystals were prepared for lowtemperature data collection by soaking in the cryo-buffer (15% PEG-200, 20% PEG-3350, 0.1 M HEPES, pH 7.5, 0.1 M calcium acetate), followed by flash-cooling in liquid nitrogen.
The SidA ox -NADP 1 structure was obtained by co-crystallizing full-length SidA with 1 mM NADP 1 and 15 mM L-Orn (note that L-ornithine was not built into the active site due to weak density).The SidA ox -NADP 1 crystals were yellow, confirming the oxidized state of the FAD.The crystals were cryoprotected as described above.
The SidA red -NADP(H)-Orn structure was obtained by cocrystallizing SidAD1-28 with 1 mM NADP 1 and 15 mM L-Orn, and then soaking the crystals in the cryo-buffer supplemented with 100 mM NADPH and 15 mM L-Orn.Upon soaking, the crystals changed from yellow to colorless, demonstrating that the FAD had been reduced by NADPH.
The SidA red -L-Orn complex was obtained by soaking crystals of oxidized SidAD1-28 in the cryo-buffer supplemented with 100 mM sodium dithionite and 50 mM L-Orn.Upon soaking, the crystals changed from yellow to colorless, demonstrating that the FAD had been reduced by sodium dithionite.
X-ray diffraction, data collection, and refinement X-ray diffraction data were collected in shutterless mode at beamlines 24-ID-C and 24-ID-E of the Advanced Photon Source (Pilatus 6M and Eiger 16M detectors, respectively), and beamline 4.2.2 of the Advanced Light Source (Taurus-1 CMOS detector).The data were integrated and scaled using XDS (29).Intensities were converted to amplitudes using Aimless (30).The data sets for SidA ox , SidA ox -NADP 1 , and SidA red -NADP (H)-Orn were processed in space group P2  S2.The initial phases were determined by molecular replacement.Phaser (31) was used for molecular replacement calculations with a search model derived from chain A of a structure of SidA complexed with NADP 1 and L-Orn in the space group I222 (Protein Data Bank Entry 4B63) (20).All heteroatoms were removed from the model prior to molecular replacement.PHENIX (32) was used for refinement and Coot (33) was used for model building.Simplified molecular-input line-entry system strings for L-Orn, and NADP 1 were used as the input to eLBOW (34) to generate the ligand coordinates and restraint files used during refinement.Structure validation was performed using MolProbity and the wwPDB validation service (35,36).Refinement statistics are summarized in Table 1 and listed in detail in ZSI Table S2.

Docking calculations
Ligand-binding hot spots were identified using an approach similar to FTMap, in which small organic molecules are docked to the protein (25,26).Our implementation of the method is as follows.Sixteen probe molecules were docked to the in and out conformations of SidA using the Autodock Vina 1.1.2(37): ethanol, isopropanol, isobutanol, acetone, acetaldehyde, dimethyl ether, cyclohexane, ethane, acetonitrile, urea, methylamine, phenol, benzaldehyde, benzene, acetamide, and N,Ndimethylformamide. To explore all reasonable probe binding pockets, a large search box was defined that encompassed the known binding sites for the FAD, NADP(H), L-Orn, and the connecting solvent-accessible areas.Solvent molecules, NADP (H), and L-Orn were removed from the in and out structures to prepare them for docking; FAD was retained.The structures were converted to PDBQT AQ:N format with the AutoDockTools program version 1.5.6 (38).Three-dimensional coordinates of probe molecules were generated with Open Babel version 2.3.2 (39).Docking was performed with nine poses for each probe and maximum exhaustiveness of the search box.
The hot spots of the out conformation were characterized further by docking larger compounds.The compounds were selected from ChEMBL 27 (40) to have no Lipinski violations and molecular weight in the range of 150-300 g/mol.This filter identified 311,108 compounds.A random selection of ;1% of these compounds (3035) was used for docking.For each site, Autodock Vina was run with 10 poses per compound and maximum exhaustiveness.The physicochemical properties of compounds were calculated with SwissADME (41).

Figure 1 .
Figure 1.New FAD conformation revealed by the P2 1 crystal form.A, electron density for the FAD in the oxidized ligand-free P2 1 SidA structure.The cage represents a polder omit map (4s).B, superposition of the oxidized ligand-free P2 1 SidA structure (yellow) and oxidized SidA complexed with L-Orn in space group I222 (cyan, Protein Data Bank Entry 4B69).C, proposed dihedral angle rotations that move the FAD isoalloxazine from the in to the out conformation.
AQ:J angle upon NADP(H) binding to enable hydrogen bonding with the adenosine phosphate of NADP(H) ( F4 Fig. 4, C and D).The conformational changes for Asn 323 are more complex than a simple side-chain rotation.The binding of either L-Orn or NADP 1 is accompanied by stabilization of Asn 323 .Electron density for this residue is very weak in three of the subunits of the SidA ox structure, implying high mobility in the absence of ligands.In contrast, Asn 323 has strong electron density and a well-defined conformation when either NADP 1 or L-Orn is present.The stabilization of Asn 323 likely reflects the hydrogen bonds it forms with the nicotinamide ribose of NADP(H) and the d-amino group of L-Orn (Fig. 4, B-D).

Figure 2 .
Figure 2. Conformational changes in the active site of SidA.All structures are shown in the same orientation.Oxidized FADs are colored yellow; reduced FADs are colored gray.A, ligand-free oxidized SidA showing the out conformation.B, dithionite-reduced SidA complexed with L-Orn (in conformation).C, oxidized SidA co-crystallized with NADP 1 (in conformation).D, NADPH-reduced SidA complexed with NADPH and L-Orn (in conformation).Electron density maps are shown inZSI

Figure 3 .
Figure 3. Movements of Tyr 324 and the FAD are coordinated in SidA.A, summary of the major conformational changes between out (yellow) and in (gray) states of SidA.The arrows show the direction of movement upon either flavin reduction or NADP 1 binding.B, distance between Tyr 324 and the FAD in the resting (out) conformation of the active site.C, distance between Tyr 324 and the FAD in the active (in) conformation of the active site.D, superposition of the resting enzyme conformation (yellow) and the NADPH complex (green) showing how Tyr 324 blocks the nicotinamide site.

Figure 4 .
Figure 4. Conformational changes in the active site of SidA are coordinated through the Tyr-loop.A, ligand-free SidA (green) with oxidized flavin (yellow).The FAD is out.B, L-Orn (blue) bound to SidA (orange), with the flavin reduced (gray) by sodium dithionite.The FAD is in.C, NADP 1 (green) bound to SidA (pale blue) with oxidized flavin (yellow).The FAD is in.D, L-Orn (blue) bound to SidA (pink), with the flavin reduced by NADPH (green).The FAD is in.

Figure 5 .
Figure 5.Comparison between equivalent SidA and KtzI structures.The top row shows SidA structures representing the following states: A, FAD ox-ligand-free, B, FAD ox -NADP 1 , C, FADred-NADP(H) and L-Orn, and D, FAD red AQ:Q L-Orn.The middle row shows KtzI structures representing the following states: E, FAD ox -ligand-free (Protein Data Bank Entry 4TM3), F, FAD ox -NADP 1 (Protein Data Bank Entry 4TLZ), and G, FAD red -NADP(H) (Protein Data Bank Entry 4TLX).H, overlay of panel A and panel E. I, overlay of panel B and panel F. J, overlay of panel C and panel G.In all the panels, the Tyr-loop is shown as sticks, FAD ox is yellow, FAD red is gray, and NADP(H) is green.

Figure 7 .
Figure 7. Characterization of potential ligand-binding sites with docking.A, histograms of docking scores for the top poses of 3035 compounds docked to site 1 (black) or site 2 (red) of the out conformation of SidA.B, The 10th ranked compound for site 1.C, the fifth-ranked compound for site 2. Note this is the same compound shown in panel B.

Figure 6 .
Figure 6.Ligand-binding hot spots from fragment mapping.A, results of docking 16 small organic molecules (pink sticks) to the in conformation of SidA (Protein Data Bank Entry 6X0I).The NADP 1 molecule was removed prior to docking but is included here (in green) to show that the probes cluster on the nicotinamide dinucleotide binding site.The smaller cluster on the right (near the pyrimidine ring of the FAD) corresponds to the location of the d-amino group of L-Orn.B, results of docking 16 small organic molecules (pink sticks) to the out conformation of SidA (Protein Data Bank Entry 6X0H).C, surface representation of the in conformation of SidA (Protein Data Bank Entry 6X0I).The FAD is shown in yellow spheres.The opening to the active site is the entrance for NADPH.D, surface representation of the out conformation of SidA (Protein Data Bank Entry 6X0H).The orientation of this view is the same as that in panel C.
Figure(s): F1-7,S2 ART: RA120.014750Page: 8 Total Pages: 13 ID: yadavr Time: 17:22 I Path: //mbnas01.cadmus.com/home$/yadavr$/SB-JBCJ200421 1 and have unit cell dimensions of a = 77-85 Å, b = 153-157 Å, c = 89-91 Å, and b = 109-111°.The asymmetric unit contains a tetramer of SidA.Note the a-axis varies appreciably depending on the ligand and redox state.The space group for SidA red (reduced with sodium dithionite) is also P2 1, but apparently soaking in sodium dithionite induced a large change in the unit cell parameters to a = 105.9,b = 155.0,c = 146.85,and b = 91.01°,and the asymmetric unit contains two tetramers of SidA.Data processing statistics are summarized inTable 1 and listed in detail in ZSI Table we report a new crystal form of SidA that enabled us to capture a new conformation of the FAD and investigate the dependence of the FAD conformation on redox state and ligand binding.Four crystal structures of SidA in different redox and ligation T1Table 1 and ZSI Table

Table 1
Summary of X-ray diffraction and data collection statistics for four crystal structures of SidA in different redox and ligation states a Values for the outer resolution shell of data are given in parenthesis.c 5% test set.c From MolProbity.The percentile ranks (PR) for Clashscore and MolProbity score are given in parentheses.

Table 2
Kinetic parameters a for WT and mutant variants of SidA The kinetic parameters, k cat and K m , were obtained from oxygen consumption assays.The table shows the mean and standard deviation of at least two technical replicates.
a b Catalytic efficiency (k cat /K m ) of WT divided by that of variant.c Data from Robinson, et al. (21).