Revealing the intricate mechanism governing the pH-dependent activity of a quintessential representative of flavoproteins, glucose oxidase

Abstract

Glucose oxidase (Gox), a prototypical flavoprotein, exhibits diverse industrial applications in glucose sensing and gluconic acid production. Its enzymatic activity is pH-dependent, with maximum activity observed at approximately neutral pH but less than 5% of peak activity at pH ≤ 3.0. However, the underlying mechanism governing these pH-dependent changes in activity remains elusive. Therefore, our objective was to investigate conformational alterations in Gox across different pH levels for engineering purposes. Our mutagenesis results suggest that protein degradation does not primarily contribute to the enzyme's pH-dependent activity. Fluorescence spectroscopy findings reveal subtle influences of pH on Gox's conformation while maintaining a similar overall microenvironment. Furthermore, the crystal structure and molecular dynamics simulations reveal that alterations in pH have a significant impact on the conformation of His514, a crucial catalytic residue for Gox function. These changes also result in structural variations within the substrate-binding pocket for both flavin adenine dinucleotide (cofactor) and β-d-glucose (substrate) between pH 6.0 and 2.5. Consequently, under acidic conditions (pH 2.5), β-d-glucose exhibits unstable binding within this pocket, leading to rapid dissociation from the active site. In summary, our findings underscore the intimate relationship between the conformational dynamics of His514 and the pH-dependent reaction mechanism, offering valuable insights for engineering acid-active Gox variants.

Graphical abstract

1. Introduction

Oxidases, peroxidases, dehydrogenases, and reductases (which belong to the oxidoreductase family) facilitate electron transfer in living organisms from oxidizing agents to reducing agents [1]. By catalyzing oxygen-insertion, hydride-transfer, proton-extraction reactions, and other critical steps, these oxidoreductases play essential roles in various biological activities such as redox catalysis and DNA repair [2]. Therefore, comprehending the fundamental mechanisms underlying the catalytic reactions mediated by oxidoreductases is of utmost importance for understanding life-sustaining activities and advancing protein engineering in bioenergy applications [[3], [4], [5]]. Flavoproteins represent a specific subclass of oxidoreductases that utilize flavin mononucleotide (FMN) or flavin adenine dinucleotide (FAD) as redox cofactors. Due to their ability to facilitate a wide range of redox processes through diverse chemical mechanisms, flavoproteins have garnered significant research attention and are valuable tools for biotechnological applications [6,7].

Glucose oxidase (Gox; EC 1.1.3.4), a prototypical flavoprotein, finds extensive applications in both industry and biotechnology. The tightly but noncovalently bound cofactor FAD serves as a redox carrier during Gox reactions [8]. By utilizing oxygen as an electron acceptor, Gox efficiently and specifically generates gluconic acid and H₂O₂ through a ping-pong bi-bi reaction mechanism [9]. This catalytic process can be divided into reductive and oxidative half-reactions [10,11]. In the reductive step, the substrate β-d-glucose undergoes transformation to d-glucono-δ-lactone followed by nonenzymatic hydrolysis to form gluconic acid while simultaneously reducing the cofactor FAD to FADH₂. During the oxidation step, FADH₂ is reoxidized to FAD upon reacting with molecular oxygen, resulting in the production of H₂O₂. The reaction process exhibits a remarkable specificity towards β-d-glucose, displaying an exceptionally high rate constant (approximately 10⁶/M/s) compared to other oxidases. This exceptional reactivity has earned it the moniker of “Ferrari” among oxidases [11]. Consequently, Gox-based applications have been extensively developed for various purposes such as glucose sensors in cancer diagnosis, environmental monitoring, and oxidation reactions. Additionally, they find utility in gluconic acid production. These versatile applications are widely employed across diverse industries including chemical manufacturing, oral hygiene products, medical diagnostics, biotechnology research, food processing, and feed production [[12], [13], [14]].

Gluconic acid, approved as safe by the US Food and Drug Administration and utilized as a cement additive, is projected to exceed 80 million USD in market value by 2024 [15]. The production of gluconic acid relies on Gox catalysis with glucose serving as a primary substrate [16]. However, Gox activity diminishes with increasing hydrogen peroxide (H₂O₂) concentration and accumulation of gluconic acid during the reaction. To counteract the negative effects of H₂O₂, cross-linked Gox variants and multi-enzyme complexes containing catalase have been employed to mitigate detrimental impacts on Gox activity and stability [1]. It has been observed that the accumulation of gluconic acid creates a strongly acidic environment (i.e., pH decreasing from 5.5 to 2.75 and lower), which significantly hampers Gox activity and gluconic acid yield. De-acidification can be achieved by introducing alkali solutions; however, this approach poses challenges in terms of environmental protection and production costs [12]. Therefore, comprehending the mechanism underlying the pH-dependent activity of this “Ferrari” enzyme (Gox) holds great significance for engineering improved variants with enhanced activities at low pH levels, thereby facilitating their application in industrial settings.

Previously, a thermostable variant Gox_M8 from Aspergillus niger was obtained through eight rounds of rational design [17,18]. The exceptional economic significance of Gox_M8 is highlighted by its highest reported T_m value (69.5 °C) among all Gox variants [9]. While the structure-thermostability relationship of Gox has been extensively investigated, the underlying molecular mechanisms governing its pH-dependent functions remain largely elusive. This study aims to elucidate the pH-dependent activity mechanism of Gox by investigating the impact of pH-induced changes on its enzymatic properties, aspartyl-prolyl peptide bond hydrolysis, fluorescence spectroscopy, and conformational dynamics. Additionally, an analysis of the substrate-binding pocket of Gox_M8 will be conducted. We found that the pH had a significant impact on the conformation of His514, a critical catalytic residue, as well as the binding pocket for both FAD cofactor and substrate β-d-glucose. Consequently, at pH 2.5, dissociation of β-d-glucose occurred. These findings offer valuable insights into the pH-dependent catalytic mechanism of Gox and establish a solid foundation for enhancing its industrial applications.

2. Materials and methods

2.1. Gene mutagenesis and heterologous expression

The recombinant plasmid pPIC9-gox_m8, containing the gox_m8 gene, was previously constructed in our study [17]. Recombinant plasmids harboring gox_m8 variants were generated using the pPIC9-gox_m8 plasmid as a template through Fast Mutagenesis System Kit (Vazyme Biotech, Nanjing, China). The methylation of the initial plasmids was eliminated by treating the polymerase chain reaction products with DMT enzyme. Subsequently, these plasmids from different mutagenesis reactions were individually transformed into chemically DMT-competent Escherichia coli cells. After verification and linearization of the recombinant plasmids with BglII restriction enzyme, electroporation (Bio-Rad, Hercules, CA, USA) was employed to transform the linearized plasmids into Pichia pastoris GS115 competent cells. Transformants exhibiting superior Gox activity were selected for large-scale fermentation and purified according to previous protocols [18].

2.2. Steady-state enzyme kinetics

The Gox-activity assay was performed in a 1.5 mL reaction system comprising of 9 U of horseradish peroxidase, 1 mg/mL glucose, 2.5 mg/mL o-dianisidine in 100 mM McІlvaine buffer (pH 6.0), and 50 µL of appropriately diluted Gox_M8 solution. The reaction was initiated by the addition of Gox_M8, followed by incubation at 30 °C for 3 min. To terminate the reaction, 1 mL of 2 M H₂SO₄ was added. The absorbance at wavelength 540 nm was measured to record the reaction mixture's response. One unit of Gox activity was defined as the quantity of enzyme that oxidized d-glucose at a rate of one µM per minute under standard conditions (pH 6.0, 30 °C). The kinetic parameters for Gox_M8 were determined using non-linear regression analysis by fitting to the Michaelis-Menten plot under standard assay conditions ranging from glucose concentrations between 5 and 200 mM. K_m and V_max values were obtained through this analysis approach. All assays were conducted in triplicate to ensure accuracy and reliability.

2.3. Structure determination and refinement

Gox_M8 (20.0 mg/mL) was subjected to crystallization using the hanging-drop vapor-diffusion method, employing commercial available crystallization buffer kits (Hampton Research, Riverside, CA, USA). In brief, 2 µL of Gox_M8 solution was combined with an equal volume of reservoir solution and allowed to equilibrate against 500 µL of the same reservoir solution. The resulting X-ray diffraction dataset for Gox_M8 was automatically processed using HKL-2000 software [19].

2.4. Setup of the theoretical model

The initial holoenzyme structures at various pH values were obtained from the Protein Data Bank (https://www.rcsb.org/). All structures were simulated in their oxidized form as holoenzymes (Gox + FAD) complexed with β-d-glucose and molecular oxygen. The construction of the holoenzyme/β-d-glucose complex was performed using the CHARMM-GUI Ligand Reader and Modeler [20].

All simulations were performed using the GROMACS package [21], employing the CHARMM36 force field [22]. The holoenzyme/β-d-glucose complex was solvated in a cubic box using the TIP3P water model [23] with dimension of 9.22 nm × 9.22 nm × 9.22 nm, and periodic boundary conditions were applied along all three dimensions. To achieve a physiological saline concentration of 150 mM KCl, K⁺ and Cl⁻ ions were added to the system. The LINCS algorithm was utilized to calculate bond lengths involving hydrogen atoms accurately [24]. Long-range electrostatic interactions were treated using the Particle Mesh Ewald method [25,26], while van der Waals interactions had a typical distance cutoff of 1.28 nm imposed on them. A time step of 2 fs was employed for movement-integration purposes. To maintain a coupling temperature of 300 K, a Berendsen thermostat with a constant time of 0.1 ps was implemented throughout the simulation process [27]. Additionally, pressure control at 1 atm was achieved by applying a coupling time of 1 ps and setting an isothermal compressibility value of 4.6 × 10⁻⁵ bar⁻¹. The model was initially subjected to energy minimization using the steepest descent algorithm, followed by a 500 ns productive simulation. For each model, three independent trajectories were considered with randomly assigned initial velocities.

2.5. Data analysis

GROMACS tools were employed for trajectory analysis. The MM-GBSA method [[28], [29], [30]], based on NAMD [31], was utilized to calculate the binding free energy, and statistical analysis along with standard deviation analysis were performed on all simulation trajectories. Additionally, GROMACS was used for conducting Principal Component Analysis (PCA), employing its built-in tools to statistically analyze all simulation trajectories. Specifically, the process involved several steps. Initially, all simulations were aligned to a common set of initial coordinates using the heavy atoms of key residues (Tyr66, Arg93, Val104, Asn105, Gly106, Thr108, Arg174, Thr329, Gln345, Glu410, Asp414, Asp422, Trp424, Arg510, Asn512, Tyr513, His514, and His557) as reference points. Subsequently, the simulation coordinates of residues from all systems (pH 6.0 and pH 2.5) and replicates were concatenated and utilized to fit a transformation function. This fitted transformation function was then employed to reduce the dimensionality of heavy atoms in each system's residue coordinates within the principal component (PC) space. Importantly, to ensure consistent eigenbasis for principal components across systems, all system coordinates were transformed into the same PC space prior to conducting PCA comparisons between them. Additionally, the criteria for determining hydrogen bonds were as follows: the angle formed between the hydrogen atom and the donor-acceptor should be less than 30°, and the H…O bond length should be less than 0.35 nm. The VMD 1.9.3 [32] software was utilized for visualizing structural figures.

3. Results and discussion

3.1. Gox activity was strictly pH-dependent

Most forms of Gox exhibit optimal activity at a neutral pH (pH 5.0–7.0, Table S1). Similarly, Gox_M8 demonstrates its highest level of activity within the pH range of 5.0–6.0 but shows less than 5% peak activity at pH 3.0 (Fig. 1a). However, Gox_M8 displays remarkable stability across a broad pH spectrum, ranging from acidic to alkaline conditions, and retains over 90% of its maximum activity after 1 h incubation at pH levels between 3.5 and 8.0 at 37 °C (Fig. 1b). Notably, even after an hour-long incubation period at pH 3.0, Gox_M8 still retains approximately 74% of its original activity; however, it only maintains around 7% activity at pH levels as low as 2.5. These results indicate that the enzymatic activity of Gox_M8 is strictly dependent on the surrounding pH environment and exhibits marginal functionality under extremely acidic conditions (pH < 3.0). The relationship between the ratio of maximum velocity (V_max) to Michaelis–Menton constant (K_m) was assessed across a wide range of pH values from 3.0 to 8.0. The V_max/K_m profile decreases significantly towards both low and high extremes in terms of acidity or alkalinity with limiting slopes close to 1 and –1 (Fig. 1c), thus allowing for fitting data into Eq. 1 [33,34]: log(V_max/K_m) = C – log(1 + [H⁺]/K₁ + K₂/[H⁺]) where C is a constant and K₁ and K₂ are the acid-base equilibrium constants of the ionizing group involved in the catalytic reaction.

Fig. 1.

Biochemical characterization of Gox_M8. (a) pH profile of Gox_M8. (b) pH stability of Gox_M8 at pH 2.0–8.0 over 1 h at 37 °C. The residual activity was determined in 10 mM McIlvaine buffer (pH 6.0) for 3 min. (b) pH-dependence of log(V_max/K_m). The individual data points represent the fits of the initial velocities at different β-d-glucose concentrations (5–200 mM). The data's curve corresponds to an iterated match to Eq. 1 for log(V_max/K_m).

The observed pK values for the acidic and alkaline sides were determined to be 3.88 ± 0.61 (pK₁) and 6.97 ± 0.62 (pK₂), respectively. The absence of ionizable groups in the substrate β-d-glucose at these specific pH levels suggests that the activity of Gox is reliant on the ionization of crucial residues within the enzyme structure, highlighting its dependence on essential residue ionization for optimal functionality.

3.2. Hydrolysis of the aspartyl-prolyl peptide bond was not the primary cause for the pH-dependence of Gox

In enzymatic reactions, alterations in pH can disrupt the essential hydrogen bonding and electrostatic properties necessary for maintaining a stable structure and conformation of an enzyme, ultimately resulting in enzyme inactivation [35]. To investigate the impact of different pH values on Gox_M8 stability, sodium dodecyl–sulphate polyacrylamide gel electrophoresis (SDS–PAGE) analysis was performed using 12% (w/v) gels (Fig. 2a). Gox_M8 exhibited stability at pH 6.0 (∼78 kDa), while degradation commenced at pH 5.0. Notably, under acidic conditions (pH 2.0–4.0), four distinct bands (∼72, 55, 38, and 32 kDa) emerged as evidence of Gox_M8 deterioration. After Gox_M8 was separated by SDS-PAGE at pH 3.0, it was transferred onto a polyvinylidene fluoride membrane for subsequent analysis. The five prominent bands were excised from the membrane and subjected to N-terminal sequencing using the Edman degradation method (Fig. 2b). Bands I, II, IV, and V exhibited an identical N-terminal sequence as that of full-length Gox_M8: Glu-Ala-Glu-Ala-Tyr-Val-Glu-Phe. However, through N-terminal sequencing analysis, band III revealed a distinct cleavage site: Pro-His-Gly-Val-Ser-Met-Phe-Pro. These findings unequivocally demonstrate that acidic conditions induce degradation of Gox_M8 between Asp206 and Pro207. Previous data has demonstrated that aspartyl-prolyl peptide bonds exhibit exceptional lability, particularly in pH-sensitive proteins where they can undergo cleavage under acidic conditions [36]. To further investigate the impact of aspartyl-prolyl peptide bond degradation on the pH dependence of Gox_M8, a substitution of valine for residue Asp206 was performed (Fig. S1). Subsequent experimentation confirmed that the D206V variant displayed comparable stability to Gox_M8 at pH 6.0 (Fig. 2c). However, it exhibited reduced degradation at pH 3.0, indicating an enhanced resistance to acidic conditions conferred by the D206V variant. Notably, despite this improved resistance, the D206V variant retained similar (or slightly lower) maximal activity compared to Gox_M8 after incubation for 1 h at 37 °C and pH levels of both 2.75 and 3.0 (Fig. 2d). It is worth mentioning that within the Gox_M8 peptide chain there exist three additional aspartyl-prolyl peptide bonds: Asp9-Pro10, Asp440-Pro441, and Asp449-Pro450 (Fig. S1). Changing each aspartic acid residue to alanine individually at all three aspartyl-prolyl peptide bond sites did not result in any alteration of the pH-dependent activity, compared to that observed for Gox_M8 (data not shown). These findings strongly suggest that the hydrolysis of the aspartyl-prolyl peptide bond is not the primary underlying factor contributing to the pH-dependent activity exhibited by Gox_M8.

Fig. 2.

pH-dependent degradation of Gox_M8. (a) SDS-PAGE analysis of Gox_M8 at different pH values. M indicates a set of protein molecular-weight standards, and 1, 2, 3, 4, 5, and 6 represent Gox_M8 at pH values of 2.0, 2.5, 3.0, 4.0, 5.0, and 6.0, respectively. (b) Diagram of Gox_M8 at pH 3.0 after being transferred to a polyvinylidene fluoride membrane. Five major bands were removed from the membrane and underwent N-terminal sequencing via Edman degradation. Two lanes symbolize two iterations. (c) SDS-PAGE analysis of Gox_M8 and its variant D206V at a pH of 6.0 or 3.0. M indicates a set of protein molecular-weight standards; 1 and 2 indicate the Gox_M8 and its variant D206V at pH 6.0, respectively; 3 and 4 indicate the Gox_M8 and its variant D206V at pH 3.0, respectively. (d) pH stability of Gox_M8 and its variant D206V after incubation at pH 2.75 or 3.0 (37 °C) for 1 h. The residual activity was measured in McIlvaine buffer (pH 6.0, 30 °C, 3 min).

3.3. pH influenced the conformational state population of Gox, but a similar microenvironment was maintained

To investigate the influence of pH on the structural characteristics of Gox, we conducted fluorescence spectroscopy analysis using a microplate reader (Synergy H1, BioTek Instruments, Inc.) to examine the fluorescence of 1-anilino 8-naphthalene sulfonic acid (ANS), FAD, tryptophan, and tyrosine. Remarkable variations in the conformations and structures of Gox_M8 were observed under different pH conditions. The extrinsic fluorophore ANS can effectively detect the molten globule state by strongly binding to cationic groups present in polypeptides and proteins through ion-pair formation [37]. As shown in Fig. 3a, no discernible differences were observed in fluorescence intensity at pH levels 7.0 and 8.0. The ANS fluorescence intensities exhibited a 122% increase at pH 3.0 and an 8% decrease at pH 8.0 compared to the optimal pH value (pH 6.0), indicating that significant conformational changes were induced in Gox_M8 by alterations in pH. The ${\lambda }_{max}$ value was observed at 450 nm with excitation at 390 nm under optimal conditions (pH 6.0). No blue or red shift occurred in the emission maximum of Gox_M8 when the pH was lowered to 3.0 or raised to 8.0, suggesting that the major exposure of internal non-polar groups (and hence surface) remained consistent with that observed under optimal conditions (pH 6.0) [38].

Fig. 3.

Structural changes in Gox_M8, analyzed by measuring emission spectra at various pH values. (a) For ANS-fluorescence measurements, emission data were recorded from 420 to 600 nm at 390 nm of excitation. (b) For FAD-fluorescence measurements, emission data were recorded from 480 to 580 nm at 450 nm of excitation. For (c) tryptophan and (d) tyrosine fluorescence measurements, emission data were recorded from 300 to 400 nm at 292 nm and 274 nm of excitation, respectively. Inset: relative fluorescence unit versus pH. Three independent tests were carried out in each case and the standard deviation was lower than 2.5%.

FAD, which reflects the distinctive environmental characteristics of isoalloxazine, serves as a natural indicator for investigating the dynamic microenvironment of the flavin fluorophore in flavoproteins [39]. Gox_M8 exhibited strong fluorescence at 520 nm without any shift in the emission wavelength peak across a pH range of 3.0–8.0 (Fig. 3b). However, the fluorescence intensity decreased when the pH was lowered to 3.0 and increased when raised to 8.0. The variation in FAD-fluorescence intensity observed in Gox could be attributed to FAD dissociation from the enzyme [40]. These results suggest that pH influences an essential conformational change required for FAD binding to the enzyme.

The fluorescence intensity of Trp and Tyr residues was further analyzed to characterize the structural and dynamic properties of Gox_M8 (Fig. 3c, 3d). The highest fluorescence-intensity values for residues Trp and Tyr were recorded at pH 4.0–5.0. Gradual decrease in relative fluorescence intensity with decreasing pH down to 3.0 indicated conformational alterations in Gox_M8. Additionally, the emission wavelength maxima of residues Trp and Tyr unchanged as Gox_M8 encountered low pH values, suggesting a non-polar microenvironment for aromatic amino acids was maintained despite the dependence of Gox_M8 conformation on pH value.

3.4. Side-chain conformation of the catalytic residue His514 was affected by the pH condition

To investigate the underlying mechanism behind the pH-dependent enzymatic activity observed in the aforementioned experiments, we endeavored to determine the X-ray crystal structures of Gox_M8 under varying pH conditions. Crystals with lamelliform, prismoid, irregular-sphere, and needle-like morphologies were successfully obtained for data collection after 2–7 days in a crystallization buffer ranging from pH 3.5 to 7.0 (Fig. S2). However, only prismoid-like crystals yielded high-resolution data. Notably, diffraction data was collected from a single crystal exhibiting an intense yellow color under specific conditions: 0.1 M HEPES buffer at pH 7.0, supplemented with 0.2 M sodium chloride and 12% (w/v) polyethylene glycol monomethyl ether 2000; this was performed on the BL10U2 beamline at Shanghai Synchrotron Radiation Facility (SSRF), maintaining a temperature of 100 K. The Gox_M8 crystal belonged to space group P4₁2₁2 and possessed unit-cell parameters of a = b = 126.45, c = 193.36 Å (Table S2). Subsequently, the crystal structure of Gox_M8 was determined through molecular replacement technique employing the initial search model derived from the crystal structure of Gox obtained from A. niger (PDB entry 1CF3) [41]. The refinement process yielded a resolution of 2.08 Å, resulting in an impressive final working R value of 16.5% and a free R value of 20.3%, demonstrating high accuracy and reliability in our findings.

Gox_M8, a representative enzyme belonging to the glucose/methanol/choline oxidoreductase family, is constituted by a homodimeric arrangement of two identical subunits. Each subunit tightly binds a noncovalent cofactor FAD, which functions as a redox carrier during the catalytic process. Similar to other canonical forms of Gox, Gox_M8 possesses seven N-linked Asn-glycosylation sites (Asn87, Asn159, Asn256, Asn276, Asn353, Asn386, and Asn471). The overall structure of Gox_M8 primarily comprises two domains (Fig. 4). The FAD-binding domain exhibits a three-layer sandwich architecture with an irregular β-sheet and six short α-helices. Additionally, it features a long loop containing a short antiparallel β-sheet that collectively formulates a narrow channel for FAD occupancy. The substrate-binding domain consists of a large six-stranded antiparallel sheet, surrounded by six helices and an additional, shorter helix. In particular, in the active site, an oxygen molecule was discovered between the catalytic residue His514 and the isoalloxazine group of FAD, resembling the crystal structures of A. niger Gox variants (PDB entry 5NIT and 5NIW) [11]. Upon superimposing Gox_M8 onto all known Gox structures solved at pH 5.1–7.5 (Fig. S3), a significant conformational change was observed between the stationary region and catalytic residue His514 in the dynamic region (Fig. 4). When comparing Goxs from A. niger (1CF3, pH 5.6) and Penicillium amagasakiense (1GPE, pH 7.4) [41], it was observed that the active site of Gox_M8 (pH 7.0) is situated within a pocket defined by Glu410, His514, His557, and FAD. At pH 5.6, the residues His559 and Glu412 are in close proximity to the molecular oxygen while the side chain of catalytic residue His516 (χ₁ = 259°, χ₂ = 188°) is oriented towards His559. Conversely, at pH 7.0, the side chain of catalytic residue His514 (χ₁ = 297°, χ₂ = 165°) is directed away from the molecular oxygen. Finally, at pH 7.4, the side chain of catalytic residue His520 (χ₁ = 281°, χ₂ = 76°) is positioned towards the isoalloxazine of FAD. The ping-pong bi-bi mechanism of Gox relies heavily on the protonation state of His514, which plays a crucial role in facilitating the oxidative half-reaction by progressively enhancing oxygen binding and reactivity through successive single-electron transfers [42]. Our findings demonstrate that substitution of His514 with Ala-leads to complete loss of enzyme activity in Gox_M8. Furthermore, this pivotal histidine residue exists in both catalytic (240° < χ₁ < 360° and 150° < χ₁ < 210°) and non-catalytic states, where alterations to its geometry and chemical properties render the active site unsuitable for simultaneous proton and hydride transfer [11]. The observed variations in conformation under different pH conditions confirm that Gox activity is strictly dependent on pH.

Fig. 4.

Crystal structure of Gox_M8. Structural superimposition of Gox_M8 (pH 7.0, cyans) with the known structures of Gox from P. amagasakiense (1GPE, pH 7.4, yellow) and A. niger (1CF3, pH 5.6, rose).

3.5. Conformation of the Gox cofactor FAD was pH-dependent

To elucidate the pH-dependent reaction mechanism of Gox, we conducted 500-ns molecular-dynamics simulations to compare the pH-induced alterations in the substrate-binding pocket of Gox. To obtain the accurate conformation of Gox_M8 bound with β-d-glucose, we replicated the structure of d-glucono-1,5-lactone from the crystal structure of glucose dehydrogenase from Aspergillus flavus (PDB entry 4YNU) during structural superimposition and subsequently modified the ligand accordingly (Fig. S4). To comprehensively understand the pH-dependent reaction mechanism, we compared the dynamic properties of the Gox complex (holoenzyme + β-d-glucose + molecular oxygen) at both pH 6.0 and pH 2.5. The protonation state of Gox_M8 was predicted using H++ server (http://newbiophysics.cs.vt.edu/H++/), revealing that at pH 6.0 and pH 2.5, its total charge was –9 and 46, respectively. The root-mean-square deviation (RMSD) values of the heavy atoms of Gox_M8 (Fig. S5) were calculated to determine the equilibrium state of the entire simulation system, which was achieved after 250 ns. Analysis of three trajectories conducted at pH 6.0 demonstrated that the RMSD profiles exhibited minimal fluctuations within a narrow range, approximately 0.21–0.27 nm, indicating a highly stable structure. In contrast, at pH 2.5, the initial enzyme structure displayed greater flexibility as evidenced by significant variations in the RMSD profiles ranging from 0.40 nm to 0.58 nm. To identify regions with higher flexibility, we analyzed the root-mean-square fluctuation (RMSF) value for each residue in Gox_M8 (Fig. S6). Notably, residues 53–71, 82–89, 101–116, 142–179, 194–237, and 489–497 comprising loops and parts of α-helices were observed to be destabilized at pH 2.5. These findings highlight that Gox_M8 exhibits enhanced structural stability under neutral conditions at pH 6.0.

In the Gox-catalyzed reaction, a hydride is transferred from β-d-glucose to the N5 atom of FAD, resulting in the formation of an intermediate known as FAD-glucose-I. The residues His514 and His557 act as acidic and basic groups, respectively, facilitating the transfer of a proton and electron from β-d-glucose to the flavin moiety [43]. These previous findings suggest that FAD directly participates in Gox catalysis by serving as a redox carrier. To assess whether interactions between FAD and Gox_M8 at different pH values play a crucial role in catalysis, we initially determined the number of heavy atom contacts between them (Fig. S7). At pH 6.0, there was an average of 304 stable heavy atom contacts with minor fluctuations observed, indicating a stable binding between FAD and Gox_M8. In contrast, at pH 2.5, there were 280 contacts with significant fluctuations observed, suggesting a less stable structure. This difference was further confirmed by evaluating variations in hydrogen bond (H-bond) numbers between Gox_M8 and FAD (Fig. S8). Gox_M8 and FAD exhibited a higher number of H-bonds at pH 6.0 compared to pH 2.5, indicating a stronger binding affinity between them at the former pH. This enhanced stability of FAD in Gox_M8 suggests that the binding pocket of Gox_M8 was more compact at pH 6.0 than at pH 2.5, as supported by our simulation results showing a smaller radius of gyration (Rg) for the FAD isoalloxazine group and three residues (His514, Glu410, and His557) involved in its intimate binding with Gox_M8 (Fig. 4). The Rg remained consistently stable throughout the 500-ns simulation period at pH 6.0, confirming the tight interaction between FAD and these three residues (Fig. S9). Conversely, an increase in Rg observed at pH 2.5 indicated a looser structure between them under acidic conditions. These findings are consistent with our fluorescence spectroscopy data demonstrating that the binding conformation between FAD and Gox_M8 is influenced by changes in pH (Fig. 3b).

Next, our focus shifted towards investigating the correlation between FAD and the catalytic residue His514. The position of His514 was monitored by analyzing the distance between the isoalloxazine ring of FAD and the side chain of His514, as well as the relative orientation. At pH 6.0, we observed a concentrated conformational distribution with a distance of 0.54 nm and an angle maintained at approximately 145° (Fig. 5a). This finding highlights a highly focused conformational state. However, at pH 2.5, we observed a scattered conformational landscape with distances ranging from 0.62 to 1.07 nm and angles spanning approximately from 85° to 176° (Fig. 5b), indicating significant instability within the Gox_M8-FAD complex under acidic conditions (pH 2.5).

Fig. 5.

Density maps plotted as a function of distance versus angle for Gox_M8 (His514) and FAD (N5). The data shown were collected at (a) pH 6.0 and (b) pH 2.5.

Furthermore, the holoenzyme–β-d-glucose complex exhibited enhanced structural stability at pH 6.0. To determine the binding affinity of the Gox_M8-FAD complex (referred to as ΔG_bind), we employed the Molecular Mechanics-Generalized Born Surface Area (MM-GBSA) theory. The average ΔG_bind value was calculated to be –69.11 ± 4.67 kcal/mol at pH 6.0 (Table 1). In contrast, at pH 2.5, the average ΔG_bind was found to be –60.62 ± 1.57 kcal/mol. These results consistently indicate that the binding energy of the Gox_M8-FAD complex was significantly higher at pH 6.0 across all simulation trajectories compared to pH 2.5, corroborating our aforementioned structural analyses. To quantitatively investigate the energetic origin of Gox_M8 and FAD, we deconstructed the total ΔG_bind value into physical components, including the internal energy of the Gox_M8–FAD complex (E_internal), the electrostatic-interaction energy (E_ele), the van der Walls interaction energy (E_vdW), and the solvation free energy (G_sol). Our analysis revealed that at pH 6.0, protonation played a crucial role in promoting stable H-bond formation between Gox_M8 and FAD through Met559···O3, Ser289···O7, Ser101···O1, Thr108···N4, Glu48···O6 interactions facilitated by E_ele. Interestingly, there were no significant differences observed in E_vdW between pH 6.0 and pH 2.5. Furthermore, it was observed that G_sol encompasses both polar and non-polar solvation energies with a higher positive value detected at pH 6.0 compared to pH 2.5. These results strongly suggest that the conformational state of cofactor FAD within Gox is dependent on changes in pH levels.

Table 1.

Binding energetics^a(kcal/mol) of the Gox_M8–FAD complex, as calculated by MM/GBSA.

Trajectory		Einternal	Eele	EvdW	Gsol	∆Gbind
pH 6.0	1	0	–194.22	–78.90	203.60	–69.53 ± 0.24
	2	0	–197.06	–79.62	213.48	–63.20 ± 0.38
	3	0	–253.97	–82.16	261.53	–74.60 ± 0.31
pH 2.5	1	0	–58.45	–82.46	81.92	–59.00 ± 0.25
	2	0	–64.15	–86.85	90.84	–60.16 ± 0.22
	3	0	–46.03	–85.77	69.08	–62.71 ± 0.31

^aE_internal denotes the internal energy of the complex. E_ele represents electrostatic and E_vdw represents van der Waals interactions. G_sol is the solvation free energy obtained from the solvent-accessible surface area model and generalized Born model.

3.6. pH-induced conformational changes occurred in the substrate binding pocket of Gox

During catalysis, Gox typically forms intimate interactions with the substrate β-d-glucose [11,41]. Our simulations revealed a clear pH-dependent binding of β-d-glucose to Gox_M8, particularly involving residues 53–71 and 101–116 located around the binding pocket (Fig. S6). By calculating the number of atomic contacts between Gox_M8 and β-d-glucose, we observed stable binding of β-d-glucose to the binding pocket of Gox_M8 at pH 6.0 throughout the entire 500-ns simulation (Fig. S10). The overall shape of the binding pocket remained nearly unchanged during the simulation, providing appropriate space for substrate binding. In contrast, significant conformational changes in the pocket were observed at pH 2.5 (Fig. 6a). This change resulted in a rapid dissociation from the pocket (Fig. S10), with the H-bonds between Gox_M8 and β-d-glucose playing crucial roles in stabilizing substrate binding. Consequently, we conducted an analysis of the changes in the number of H-bonds present in both simulated models over time (Fig. S11). Remarkably, the complex system exhibited a higher number of H-bonds at pH 6.0 compared to pH 2.5, consistent with our previous observations indicating that the complex displays greater structural stability at pH 6.0. Therefore, we hypothesize that interactions between Gox_M8 and β-d-glucose were disturbed under low-pH conditions, potentially influencing the catalytic activity of Gox. To further investigate this phenomenon, we conducted calculations to determine the interaction energy between the holoenzyme (Gox_M8+FAD) and β-d-glucose under varying pH conditions. As shown in Fig. S12, at pH 6.0, the binding of the holoenzyme and β-d-glucose resulted in a higher release of energy, with average values of –59.06, –58.02, and –41.30 kcal/mol across three trajectories. Conversely, at pH 2.5, during the initial stage, a strong interaction energy was observed between the holoenzyme and β-d-glucose; however, as β-d-glucose dissociated from its binding pocket, the energy value gradually decreased and fluctuated around zero.

Fig. 6.

Changes of Gox's substrate-binding pocket induced by pH. (a) Close-up views of the substrate β-d-glucose in the reaction pocket of Gox_M8 at pH 6.0 (orange) and pH 2.5 (purple). β-d-Glucose is represented using an orange ball and stick model. FAD and key residues are shown using stick representations, with nitrogen atoms shown in blue and oxygen atoms shown in red. (b) Distances (left) and frequencies of H-bonds (right) between β-d-glucose and key residues or FAD in the molecular-dynamics simulations at pH 6.0 (red) and pH 2.5 (blue).

In the present study, our primary focus is on the crucial residues of Gox_M8 that directly participate in interactions with β-d-glucose within its native conformation, specifically residues Tyr66, Arg93, Arg510, Asn512, and His514. Notably, at acidic conditions (pH 2.5), the catalytic residue His514 undergoes protonation, leading to a disruption of hydrogen bonding with the substrate and an observed gradual increase in their spatial separation (Fig. 6b). Conversely, at pH 6.0, His514 forms a stable H-bond with the substrate for approximately 43.9% of simulation time. Our simulations further reveal an enhanced propensity for β-d-glucose to interact via a H-bond with Tyr66 at pH 6.0 due to its closer proximity to the substrate molecule. However, at pH 2.5, Tyr66 gradually relocates away from the center of the binding pocket resulting in an inability to form H-bonds. Additionally, at pH 6.0, the reduced distance between β-d-glucose and Arg93 leads to a persistent formation of H-bonds with Arg93, occurring for approximately 42.3% of the simulation time. Furthermore, there is a higher frequency of interactions between β-d-glucose and Arg510 at pH 6.0 compared to pH 2.5, with the distance between the amide group of Arg510 and O3/O4 of β-d-glucose remaining relatively stable and contributing to a consistent formation of H-bonds with an average length of 1.75 Å. Interestingly, Asn512 forms H-bonds with β-d-glucose at an average distance of approximately 1.12 Å for about 80% of the total simulation time. But, at pH 2.5, alterations in the orientation of the side chain of Arg510 and repositioning of the amide group in Asn512 lead to decreased substrate stability as it moves away from its pocket within an established H-bond network. It was observed that the cofactor FAD facilitated binding sites for the β-d-glucose substrate. Specifically, at pH 6.0, a stable H-bond formed between O11 of FAD and β-d-glucose. However, under acidic conditions (pH 2.5), the distance between β-d-glucose and FAD gradually increased, leading to failure to establish H-bonds.

β-d-glucose formed more intimate interactions with Gox_M8 at pH 6.0 than at pH 2.5. Different interaction networks under different pH conditions are expected to affect the catalytic activity of the enzyme. To quantify the changes in the Gox_M8 conformational dynamics, we performed principal component analysis (PCA) for key residues (Tyr66, Arg93, Val104, Asn105, Gly106, Thr108, Arg174, Thr329, Gln345, Glu410, Asp414, Asp422, Trp424, Arg510, Asn512, Tyr513, His514, and His557) in the enzyme-binding center based on the H-bonds contributions of six trajectories (Fig. S13; totaling 3-µs in duration during the simulations). The free energy landscape was obtained from the PCA, and two-dimension profiles were created using the two main components of the simulation as the reaction coordinate. These residues clearly occupied a larger conformational space at pH 2.5 than at pH 6.0 (Fig. 7). The conformational changes in the abovementioned residues, and the distinct binding modes in the enzyme active center, are responsible for the primary disparity between the two states. The binding pocket exhibits a wide and extended conformation, with the corresponding energy map revealing two discernible conformations at pH 2.5. In contrast, the binding pocket of Gox_M8 maintains a stable structure at pH 6.0, indicating that the movement of the catalytic pocket follows a constrained and highly clustered pattern due to stable substrate binding.

Fig. 7.

Free-energy landscapes of key residues in the substrate-binding pocket of Gox_M8 at pH 6.0 and pH 2.5. The unit of the potential of the mean force was kJ/mol. The figure shows a representation of the molecular-dynamics trajectories projected into the two most important principal components (PC1 and PC2) at pH 6.0 and pH 2.5. PC1 (x-axis) differentiates pH 2.5 (low PC1 values) and pH 6.0 (high PC1 values).

4. Conclusion

In summary, the mechanism underlying the pH-dependent activity of Gox was elucidated through comprehensive investigations including pH-induced changes in enzyme properties, aspartyl-prolyl peptide bond hydrolysis, fluorescence spectroscopy, crystal-structure analysis, and examination of the substrate-binding pocket of Gox_M8. Gox_M8 exhibited strict pH-dependent and retained < 5% of its peak activity at pH 3.0. However, under acidic conditions, aspartyl-prolyl peptide bond hydrolysis was not identified as the primary cause for the observed pH-dependent activity in Gox_M8. Our fluorescence spectroscopy results revealed that while the overall microenvironment remained similar, the conformation of Gox_M8 varied with pH. Notably, the side-chain conformation of catalytic residue His514 was influenced by changes in pH. Specifically, at pH 2.5, a distinct alteration in the substrate-binding pocket involving cofactor FAD, substrate β-d-glucose, and Gox_M8 was observed compared to that at pH 6.0. At neutral pH (6.0), β-d-glucose stably bound to the holoenzyme without significant disruption to catalytic-pocket conformation due to strong hydrogen bonding interactions. Furthermore, a stronger binding energy for the Gox–FAD complex at pH 6.0 compared to that of at acidic conditions (pH 2.5) by employing the MM-GBSA theory. Due to spatial structural changes in key residues within this region at pH 2.5, β-d-glucose could not bind stably and rapidly dissociated from its binding pocket, which can be potentially affecting enzyme catalysis efficiency. These findings provide insights into understanding how Gox's reaction profile is controlled by varying environmental acidity levels and offer a foundation for designing protein engineering strategies aimed at adjusting optimal working range based on desired applications.

Data availability

Data will be made available on request.

Declaration of competing interest

The authors declare that they have no conflicts of interest in this work.

Biographies

Tao Tu (BRID: 06181.00.51691) is currently a professor at State Key Laboratory of Animal Nutrition and feeding, Institute of Animal Sciences, Chinese Academy of Agricultural Sciences. He obtained his Ph.D. in biochemistry and molecular biology from Chinese Academy of Agricultural Sciences in 2016. From 2016 to 2020, he was a researcher at the Feed Research Institute of the Chinese Academy of Agricultural Sciences. His research focuses on the structure-function relationship of enzyme.

Huoqing Huang is a professor at the Institute of Animal Science of the Chinese Academy of Agricultural Sciences. He received his Ph.D. degree in biochemistry and molecular biology from the Chinese Academy of Agricultural Sciences in 2009. From 2009 to 2020, he was a researcher at the Feed Research Institute of the Chinese Academy of Agricultural Sciences. His current research focuses on high-level production of heterologous proteins in microbial systems.