Functional role of residues involved in substrate binding of human 4-hydroxyphenylpyruvate dioxygenase.

4-Hydroxylphenylpyruvate dioxygenase (HPPD) catalyzes the conversion of 4-hydroxylphenylpyruvate (HPP) to homogentisate, the important step for tyrosine catabolism. Comparison of the structure of human HPPD with the substrate-bound structure of A. thaliana HPPD revealed notably different orientations of the C-terminal helix. This helix performed as a closed conformation in human enzyme. Simulation revealed a different substrate-binding mode in which the carboxyl group of HPP interacted by a H-bond network formed by Gln334, Glu349 (the metal-binding ligand), and Asn363 (in the C-terminal helix). The 4-hydroxyl group of HPP interacted with Gln251 and Gln265. The relative activity and substrate-binding affinity were preserved for the Q334A mutant, implying the alternative role of Asn363 for HPP binding and catalysis. The reduction in kcat/Km of the Asn363 mutants confirmed the critical role in catalysis. Compared to the N363A mutant, the dramatic reduction in the Kd and thermal stability of the N363D mutant implies the side-chain effect in the hinge region rotation of the C-terminal helix. The activity and binding affinity were not recovered by double mutation; however, the 4-hydroxyphenylacetate intermediate formation by the uncoupled reaction of Q334N/N363Q and Q334A/N363D mutants indicated the importance of the H-bond network in the electrophilic reaction. These results highlight the functional role of the H-bond network in a closed conformation of the C-terminal helix to stabilize the bound substrate. The extremely low activity and reduction in Q251E's Kd suggest that interaction coupled with the H-bond network is crucial to locate the substrate for nucleophilic reaction.

Human HPPD is a dimeric protein with two domains per monomer [3,4]. The active site is located in the C-terminal domain [17]. It possesses a conserved motif of 2-His-1-carboxylate facial triad, which is buried in a β-barrel structure at the active site for ferrous ion coordination [18][19][20][21][22][23]. Regarding the different coordination positions in the active site, the three metal-binding ligands have different catalytic roles [24]. The active site is covered by the terminal α-helix, which has a flexible conformation and functions as a gate to shield the substrate from the solvent during catalysis [19,25].
Spectroscopy, computation, and crystal structure studies have revealed that the HPP substrate forms a bidentate coordination with the ferrous ion through the carboxylate and α-keto groups of the substrate [23,26,27]. The association with the substrate activates dioxygen for oxidative decarboxylation [26,28]. This reaction generates 4-hydroxyphenyl acetate (HPA) and a highly reactive Fe(IV)-oxo intermediate, followed by aromatic ring hydroxylation in HPA and migration of the acetyl group to form HG [4,[29][30][31][32]. The side chain binding mode of HPA is critical for the oxidative reaction by the Fe(IV)-oxo on the aromatic ring of the substrate [27,30,31]. The 4-hydroxyl group in the aromatic side chain of HPP is essential for electron delocalization [30,33]. A single mutant of P214T or N241S in S. avermitilis HPPD generated a new product of oxepinone or quinolacetic acid in which the two residues were located in the substrate-binding pocket, implying the involvement of an epoxide or arenium cation intermediate in the catalytic reaction [34,35]. The model of substrate binding in P. fluorescens HPPD indicated that the 4-hydroxy group interacts with Gln272 and Gln286 residues. The interactions were confirmed with site-directed mutagenesis and QM/MM theoretical calculation [21,33].
Recently, the substrate-bound structure of A. thaliana HPPD was elucidated (PDB code: 5XGK), and, in contrast to a previous report, the 4-hydroxyl group of the substrate was discovered to Downloaded from http://portlandpress.com/biochemj/article-pdf/doi/10.1042/BCJ20210005/912974/bcj-2021-0005.pdf by guest on 01 June 2021 Biochemical Journal. This is an Accepted Manuscript. You are encouraged to use the Version of Record that, when published, will replace this version. The most up-to-date-version is available at https://doi.org/10.1042/BCJ20210005 4 form a hydrogen bond with the Asn423 side chain on the C-terminal helix [23]. Although site-directed mutagenesis also indicated the role of Asn423 in substrate binding and catalysis, the binding mode of the substrate remains unclear due to the conformation of the active site.
The crystal structures of A. thaliana HPPD displayed a similar conformation in the presence of the substrate or inhibitors or in the absence of the substrate (PDB code: 1TFZ and 1SQD), particularly for the C-terminal helix [22,23] thaliana and Z. mays HPPD [19,36]. The differences might arise from low sequence similarity between different species and consequently affect the contact of the C-terminal α-helix with the active site that stabilizes the closed conformation of the protein (Fig. S1) [36].
No substrate-bound structure of human HPPD has been elucidated. Comparison of the structure of human HPPD (PDB code: 3ISQ) with the substrate-bound structure of A. thaliana HPPD revealed notably different orientations of the C-terminal helix (Fig. 1A) [20,23]. The C-terminal helix in human HPPD had a closed conformation. Accordingly, the substrate-binding mode in the active site of human HPPD may differ from that of A. thaliana HPPD [36]. A simulation of the substrate-bound human HPPD revealed the interaction of the hydrogen bonds between the 4-hydroxyl group of the substrate and the side chains of Gln251 and Gln265 [25]. A notable hydrogen bond network bridged by the Gln334 side chain was observed, which interacted with the Glu349 side chain (the ferrous ion binding ligand), Asn363 side chain from the C-terminal helix, and the carboxyl group of HPP (Fig. 1B). These residues are strictly conserved across the HPPD families, although a low sequence similarity (~24%) was noted (Fig.   S1). To reveal the role of these residues in catalysis, we performed site-directed mutagenesis and simulation analysis, which indicated the critical role of Gln251 in stabilizing the orientation of the aromatic side chain of HPP by participating in the initial nucleophilic reaction. Asn363 interacted with the substrate in the Q334A mutant and preserved relative activity and substrate binding affinity of the enzyme. The substitution of Asn363 confirmed its critical role in substrate binding and catalysis. The Asn363-substrate interaction also highlights the functional role of the

Preparation of variant HPPD
Single and double mutants of Q334A, Q334N, N363A, N363D, N363Q, Q334A/N363D, Q334A/N363Q, and Q334N/N363Q were generated using the QuikChange mutagenesis system (Stratagene, San Diego, CA, USA) with the vector of pTrc-HPPD as template [25]. Table S1 presents the sequences of primers that were used for PCR. The mutant-containing vectors were transformed into E. coli DH5-competent cells, and the desired mutations were confirmed through complete DNA sequencing.

Protein purification
Overexpression and purification of variant HPPD were performed as reported previously

CD studies
The CD spectra were recorded in a Jasco J-815 spectropolarimeter equipped with a Peltier temperature controller accessory. Proteins (0.2 mg/mL in 20 mM sodium phosphate buffer, pH 7.4) in a 1-mm-path-length cell were measured in the far-UV region (180-260 nm) at 25 °C . All spectra were averaged from five accumulations and were buffer-corrected. The fractions of secondary structure were estimated through the DICHROWEB server (http://www.dichroweb.cryst.bbk.ac.uk/html/home.html) using the CDSSTR method [38,39]. The ellipticity at 222 nm was recorded for the thermostability of the secondary structure from 25 to 85 °C at a heating rate of 1 °C /min. The mid-point temperature (Tm, °C ) for the protein unfolding transition was calculated using the sigmoidal fit of the curve.

ITC measurements
The binding affinity of the HPP substrate with the wild-type and mutant HPPD was

DSF
The DSF experiment was performed using a thermocycler (C1000 Thermocycler, CFX96

Simulations
The complexed structure of human HPPD (PDB code: 3ISQ) [20] was modeled using Perdew-Burke-Ernzerhof (PBE) functional. The spin multiplicity was set as "smart" in the DMol3 settings, and thermal smearing of electronic occupancy was used in the calculation. The other parameters used in the calculation were run in a default setting.
Molecular dynamics simulations were performed on the complexed structure of HPPD in which an explicit periodic boundary solvation model was applied to the structure. The system includes water molecules and counter ions in the orthorhombic shape, with a minimum distance between the protein and box boundary of 7 Å . Harmonic restraint was set for all atoms in the protein complex followed by energy minimization of the whole system for 1000 steps of the steepest descent and 2000 steps of the adopted basis Newton-Raphson algorithm. The system was heated from 50 to 300 K for a 4-ps simulation, and a 10-ps simulation was ran at 300 K for equilibration. The final steps for production simulation were conducted in a constant temperature and pressure type with no restraints. The particle mesh Ewald method was used to treat the long-range electrostatics, and the shake algorithm was applied to constrain covalent bonds involving in the aromatic hydroxylation and side chain migration of the HPA, and the hydroxylation at the benzylic carbon of the HPA, respectively [3,42]. Although the sequence identity between human HPPD and A. orientalis HMS is less than 30%, the overall structures are highly similar (~2.3 Å in the Cα rms deviations) [41]. The coordination of Fe(III)-O2 •and HMA in HPPD structure was modified to Fe(IV)-oxo and HPA, followed by a rigid body translation and rotation to enhance the interaction of the 4-hydroxy group of HPA with the side chains of

Construction of mutant HPPD and protein purification
To investigate the effect of the hydrogen bond network of Glu349-Gln334-Asn363 and interactions of Gln265 and Gln251 with the substrate related to the function of HPPD, the following single and double mutants were constructed by site-directed mutagenesis: Q334A, Q334N, N363A, N363D, N363Q, Q251E, Q265E, Q334A/N363Q, Q334N/N363Q, and Q334A/N363D.
The recombinant wild-type and mutant HPPD were overexpressed and purified through an anion exchanger, hydrophobic interaction, and size exclusive chromatography to near homogenous, as judged by SDS-PAGE (Fig. S2). Far-UV circular dichroism (CD) spectroscopy was used to analyze the secondary structure of the variant purified proteins. Those proteins displayed a similar spectrum and estimated secondary structure contents, suggesting that the overall secondary structures of the proteins were not affected by mutation ( Fig. S3 and Table S2).

Activity of wild-type and variant HPPD
The specific activity of HPPD was analyzed with the oxygen consumption and HG product formation in the reaction. The results revealed that the activity of the Q334N and Q334A mutants was approximately 4% and 50% of the wild-type enzyme, respectively, suggesting the importance of the Gln334 in HPPD activity (Table 1). However, the significant loss in the specific activity for mutants of N363Q, N363A, and N363D (approximately 10%, 30%, and 10% of those of the wild-type enzyme, respectively) indicated the important role of this residue in the function. For the Q251E and Q265E mutants, the observed activity was less than 0.5% and 5% of that of the wild type, respectively; this suggested that the interaction from Gln251 is crucial in the function of HPPD.
The activities for double mutants of Q334N/N363Q, Q334A/N363Q, and Q334A/N363D were approximately 5%, 10%, and 14% that of the wild-type enzyme, respectively, as analyzed with the oxygraph assay (Table 1). When the activity was measured using the HPLC assay, it was found that the activities of Q334N/N363Q and Q334A/N363D were approximately 3-and 10-fold lower than those measured using the oxygraph assay ( Table 1). The differences in the activity of the two assays suggested an uncoupled reaction. From the HPLC elution profile, a new peak that eluted at approximately 13 mL in addition to the HG product (eluted at approximately 10 mL) was observed for the Q334N/N363Q and Q334A/N363D mutants (Fig. 2).
The elution site of this peak was identical to that of HPA. The specific activity calculated from the peak area of HPA was approximately 0.02 ± 0.02 nmol/min/µg for those double mutants.

Steady-state kinetic analysis
The steady-state kinetics of the variant HPPD as a function of HPP concentration was determined using the oxygen consumption assay (Table 2 and

Isothermal titration calorimetry analysis
The binding properties of the HPP substrate with wild-type and mutant HPPD were analyzed with isothermal titration calorimetry (ITC). The measurements were performed by titrating HPP into the solution, which contained an equal molar ratio of monomeric enzyme with cobalt ion. The Co 2+ ion was used in this binding experiment to eliminate the precipitation of Fe 2+ ion in the buffer solution, which can interfere with the heat measurement. Titration of HPP into wild-type HPPD-Co 2+ complex revealed an exothermic reaction (Fig. 3). The resulting heat data obtained by integrating the area of each injection was well fitted to a one-site binding model. suggesting the crucial role of these residues in substrate binding.

Structural stability of HPPD
The structural stability of the variant enzymes under thermal stress was analyzed by changing the ellipticity at 222 nm in CD spectroscopy. It showed a two-state unfolding transition as a function of temperature for those proteins (Fig. S5). Compared with the wild-type protein, the Tm values of Q334A, N363D, Q334A/N363Q, Q334A/N363D, and Q251E were reduced by 3, 4, 3, 7, and 2 °C , respectively (Table 4), suggesting reduced stability of the secondary structure by those substitutions.
Thermal denaturation of HPPD in the absence and presence of ferrous ions was investigated using differential scanning fluorometry (DSF), which measures the fluorescence enhancement of the probe upon binding to the exposed hydrophobic surface of the unfolded protein (Fig. S6) [44].
The estimated Tm values of the wild-type enzyme were 52.6 °C and increased by approximately 2 °C in the presence of metal ions ( However, the estimated Tm values of the Q334A, N363D, and Q334A/N363D mutants in the presence of ferrous ions were still smaller than that of the wild type by approximately 3, 4, and 5 °C , respectively. Thus, in the active site, these mutants underwent a conformational change to reduce the structural stability.

Simulation of wild-type and mutant HPPD
Spectroscopic analysis revealed the charge transfer feature upon the binding of HPP to HPPD-Fe +2 [26,31,45]. The bidentate coordination of HPP with the iron would activate dioxygen for nucleophilic reaction on substrate [45]. To reveal the effect of residue replacement on substrate binding and the reaction mechanism, a model of HPPD with a bound Fe(III)-O2 •group and HPP substrate was generated. This model was used to demonstrate the nucleophilic attack of the activated dioxygen at the carbonyl carbon of HPP. MD simulation was applied to the models that were geometrically minimized using QM/MM calculation. After the simulation, the stable conformations from the wild type and the Q334A, N363D, and Q265E mutants were selected to run another QM/MM minimization, which included additional residues of Gln251, Gln265, Gln334, and Asn363 in the quantum region (Fig. 4, Fig. S7 and S8). The other mutants were excluded from further analysis due to the geometrical change of the active site at the beginning of the MD simulation.
The spin population, bond length and charge on the iron and dioxygen as shown in Table S3 suggested an oxygen activation state in the six-coordinated complex, consistent with the studies using DFT and QM/MM calculation for 2-oxoglutarate-dependent and cysteine dioxygenase enzymes [28, 46,47]. The optimized geometries of wild-type or mutant HPPD with a bound Fe(III)-O2 •group showed a suitable orientation of HPP substrate for the nucleophilic reaction ( Fig. 4 and S8). The bond length for Fe-O1 and O1-O2, and the charge and spin state on the iron and dioxygen were similar between wild-type and mutant enzymes (Table S3)  In the Q334A mutant, the Asn363 side chain was oriented to interact with the carboxyl groups of HPP so that the aromatic side chain of HPP could direct the 4-hydroxyl group to interact with Gln265 and Gln251 and thus stabilize the substrate binding (Fig. 4B). In this mode, the peroxide anion, which was approximately 2.4 Å from the carbonyl carbon of HPP, demonstrated a possible catalytic reaction for the Q334A mutant (Table 1). However, due to the interaction, the Asn363 side chain obliged to lose the interactions with the carbonyl group of Leu332 and Leu323 in the β-sheets of the active site ( Fig. 4A and B).
The protonated form of the substituted residue in the N363D mutant might be due to the hydrophobic environment around the substituted residue (Fig. 4C). Although the interaction of Gln334 with Asp363 was lost, it still retained the H-bond interactions with Glu349 and HPP. In this binding mode, the 4-hydroxyl group of the substrate interacted with Gln251 and the peroxide anion was approximately 2.67 Å from the carbonyl carbon of HPP, demonstrating the possible catalytic reaction for the N363D mutant ( Table 1).
The protonated form of the substituted residue was also observed in the Q265E mutant, highlighting the necessity of active site hydrophobicity for substrate binding (Fig. 4D).
Interactions of the H-bond network were retained, which stabilized the bidentate-chelated substrate for the reaction by the peroxide anion at approximately 2.54 Å to the carbonyl carbon of HPP in the Q265E mutant model. However, the aromatic side chain of HPP was oriented, which led the 4-hydroxyl group to interact with Gln251.

Discussion
The binding of the substrate in the structure of human HPPD remains elusive, although the substrate-bound structure of A. thaliana HPPD was elucidated recently [20,23]. The orientations of the C-terminal helix in the two proteins affect the active site conformation for substrate binding. The C-terminal helix functions as a gate to monitor the active site integrity for catalysis [19,25]. We proposed a possible substrate-binding model based on the structure of human HPPD.
The substitution and simulation experiments demonstrated the vital role of Gln251 in the initial nucleophilic reaction by interacting with the 4-hydroxy group of the substrate (Table 1 and Fig.   4). The functional role of the H-bond network in substrate binding and catalysis was also identified (Tables 3 and 4). Among them, Asn363 from the C-terminal helix played a crucial role by contributing to the interaction to stabilize substrate binding for catalysis. [27]. It indicated that Gln251's interaction with the 4-hydroxyl group of HPP is crucial to locate the substrate in a precise position for the nucleophilic reaction by the peroxide anion [48]. Q251E mutant stability increased in the presence of ferrous ion, suggesting a conformational change to accommodate the bidentate binding of the substrate in the active site. The result supported the critical location of this residue at the entrance of the active site, which forms a bifurcate interaction with the C-terminal helix and substrate to bury the active site (Fig. 4A). A reduction in the binding affinity and activity was observed for the Q265E mutant, in which the 4-hydroxyl group of HPP interacted with the Gln251 in the simulation model ( Fig. 4D and Table 3).
Simulation models revealed that the 4-hydroxyl group of HPP interacted with both of the Gln265 and Gln251 or rotated to interact with the Gln251 performed a suitable conformation for nucleophilic reaction (Fig. 4). It suggests that both residues are required to stabilize the bound substrate for catalysis, but the role of Gln251 is crucial.
Because Gln334 is at the center of the hydrogen bond network, it links the ferrous ion binding center, terminal helix, and substrate in the active site of HPPD. The lower thermostability of the Q334A mutant compared with that of the wild type in the presence of ferrous ion indicates its critical role in stabilizing the active site conformation (Table 4). Notably, the relatively high activity and substrate-binding affinity preserved in this mutant suggested the involvement of Asn363 in substrate binding and catalysis (Tables 1 and 3). This was supported by the simulation model, which demonstrated the interaction of Asn363 with the substrate in the Q334A mutant (Fig. 4B). The role of Gln334 in substrate-binding stabilization but not catalytic function was consistent with the report in A. thaliana HPPD [33]. The Asn363-substrate interaction directs the 4-hydroxyl group of HPP to interact with Gln251 and Gln265, in which the HPP showed a suitable conformation for nucleophilic reaction by the activated dioxygen (Fig. 4B). The interaction also implied the role of the C-terminal helix, which forms a closed conformation of the active site for the H-bond interactions to stabilize the substrate binding for catalysis. This may explain why the C-terminal helix in a different conformation exhibits a different substrate binding from that reported in the structure of A. thaliana HPPD [23].
Previous report had indicated that the C-terminal helix gates the entrance of substrate into the active site through rotation of the hinge region around Asn363 [19]. It suggested the functional role of this residue in the open-closed state of the active site. Asn363 substitution decreased the catalytic efficiency, suggesting its critical role in catalysis (Tables 2 and 3). In the simulation model, Asn363 interacted with the β-sheet in the β-barrel structure of the active site and with the Gln334 from the C-terminal α-helix to link the H-bond network (Fig. 4A) interactions of Gln334 with the substrate and Glu349 were retained (Fig. 4C). The substitution result indicated that the interaction was crucial to stabilize the substrate for the initial nucleophilic reaction. However, the dramatic reduction in substrate-binding affinity and thermal stability of the N363D mutant but not for the N363A mutant suggested that the active site conformation was changed to reduce substrate-binding stability and increase the substrate dissociation rate (Tables 3 and 4). The result also implies the effect of the side chain structure of this residue with the hinge region rotation of the C-terminal helix. The conformation of the C-terminal helix may affect the substrate binding and structural stability [19,25,36].
The catalytic efficiency, substrate-binding affinity, and thermal stability were not recovered through the double mutation of Gln334 and Asn363, suggesting that an appropriate arrangement of the two residues is critical for a functional H-bond network. Notably, HPA production from the