Crystal Structure of Glycerol Dehydrogenase from Klebsiella pneumoniae

Abstract

Glycerol dehydrogenase (GlyDH) plays a crucial role in the glycerol metabolism pathway by catalyzing the oxidation of glycerol to dihydroxyacetone (DHA). Previous studies of GlyDH have predominantly focused on unraveling the structural features of the active site and its binding interactions with ligand. However, the structural details of GlyDH in complex with both NAD⁺ and the substrate bound have remained elusive. In this study, we present the crystal structures of Klebsiella pneumoniae GlyDH (KpGlyDH) in the absence and presence of NAD⁺ at a resolution of 2.1 Å. Notably, both structures reveal the binding of the substrate, ethylene glycol, to the zinc ion. Interestingly, a significant change in the coordination number of the zinc ion is observed, with three in the absence of NAD⁺ and four in its presence. These findings shed light on the structural aspects of GlyDH and its interactions with NAD⁺ and the substrate.

INTRODUCTION

Glycerol dehydrogenases (GlyDH; glycerol:NAD⁺ 2-oxidoreductase, EC 1.1.1.6) have been identified in various organisms, such as bacteria,¹ Neurospora,² yeast,³ and mammals.⁴ These enzymes are categorized into three classes based on the specific site of glycerol oxidation and the specific cofactor they require. Under anaerobic conditions, numerous microorganisms utilize glycerol as a carbon source. The key enzyme responsible for this metabolic conversion is GlyDH, which simultaneously reduces NAD⁺ to NADH during the oxidation of glycerol (Fig. 1).⁵

Figure 1. Glycerol oxidation scheme by glycerol dehydrogenase using NAD⁺ as the hydride ion acceptor during glycerol oxidation.

This enzymatic reaction is the first step of the glycerol metabolism pathway where glycerol is converted to dihydroxyacetone (DHA). DHA is then phosphorylated by DHA kinase and used in the glycolytic pathway for further degradation. GlyDH has also been demonstrated to catalyze poly-hydroxyl species, indicating a broad substrate specificity for polyol.^1a,6 GlyDH also holds significant value in various industrial applications. For example, the DHA, produced by GlyDH, utilizes as a precursor for various chemical products. These include antifreeze fluids like 1,2-propylene glycol, as well as various biologically active compounds such as drug, pesticides, and sweeteners.⁷ Moreover, GlyDH is also the one of the key enzymes in the metabolic pathways involved in the production of biofuels like ethanol, 1,3-propanediol, 2,3-butanediol, and ethanol.⁸ Considering that glycerol is a major byproduct of the biofuel production process, research focusing on glycerol metabolism has been increased in recent years. Intensive research has been concentrated on Klebsiella and Clostridium, which are particularly efficient glycerol utilizers.⁹ Furthermore, Klebsiella pneumoniae possesses two genes (dhaD and gldA) encoding GlyDH, both genes are complementary as a result of gene duplication.¹⁰ This implies that K. pneumoniae is an excellent candidate for investigating the biochemical and biophysical characteristics of GlyDH. This research also aims to enhance the viability and efficiency of the biofuel industry.¹¹

GlyDH (GldA) from Klebsiella pneumoniae (KpGlyDH) has been characterized in previous studies.¹² It exhibited distinct preference for NAD⁺ over NADP⁺ and showed maximum activity at pH 8.6 and pH 10.¹² Furthermore, KpGlyDH displayed the highest specificity constant (k_cat/K_m) value for glycerol compared to 2,3-butanediol and ethylene glycol.¹² Many structural analyses of GlyDH from various microorganisms complexed with glycerol or NAD⁺ has been reported.¹³ The structure of GlyDH consists of two domains separated by a deep cleft. Within this cleft lies the active site, which accommodates the metal ion, NAD⁺, and the substrate. To gain insights into the catalytic mechanism of GlyDH, it is crucial to investigate the structure in the presence of both the cofactor and the substrate. Notably, while our manuscript was being prepared, a crystal structure of Tris-bound GlyDH from Escherichia coli in the absence or presence of NAD⁺ was published.^13a The Tris molecule is a known competitive inhibitor with an IC₅₀ of 2 mM.¹⁴

In the present study, we present the crystal structures of KpGlyDH in the absence or presence of NAD⁺, shedding light on the structural characteristics. Our structures also include the substrate-bound conformation, utilizing an ethylene glycol molecule. Furthermore, we conducted an analysis of the substrate tunnel in both dimeric and octameric form of GlyDH, revealing potential correlation between the oligomeric state and cooperative effects on catalytic activity. Through these investigations our study expands our understanding of its structure and the underlying mechanisms involved.

EXPERIMENTAL

Crystallization and X-ray Data Collection

KpGlyDH was expressed and purified as described previously.¹² The concentration of the protein solution for crystallization trials was 9 mg ml^-1. Initial crystallization screening was performed using the sitting drop vapor diffusion method at 295 K with crystallization reagent kits supplied by Hampton Research (USA) and Qiagen (Germany), respectively. The crystals for the data collection were grown with the hanging drop vapor diffusion method by mixing 2.0 µl of protein solution with an equal amount of reservoir solution containing 0.7 M ammonium tartrate dibasic and 0.1 M sodium acetate (pH 4.6) and equilibrating against 0.5 ml of the reservoir solution. For data collection, a crystal was coated with a cryoprotective solution with an additional 30% (v/v) ethylene glycol, and directly flash-cooled in liquid nitrogen before data collection. Additionally, the crystal was coated with a cryoprotective solution of the same composition, along with the addition of 1 mM zinc chloride and 5 mM NAD⁺, to obtain the NAD⁺-bound GlyDH structure. X-ray diffraction data were collected at 100 K using a Dectris Pilatus 3 6 M detector on the beamline 11C at the Pohang Accelerator Laboratory in Republic of Korea. X-ray diffraction data were indexed, integrated, and scaled using HKL-3000 program.¹⁵

Structure Determination and Refinement

The initial model was determined by molecular replacement with Phaser in the CCP4 suite using Escherichia coli GlyDH (RCSB Protein Data Bank ID: 5ZXL) as a search model.¹⁶ Subsequent manual modeling and refinement was performed iteratively using Coot and Refinement in the PHENIX suite.¹⁷ The refinement statistics are presented in Table S1. Coordinates and structure factors for this study have been deposited in the Protein Data Bank under the following accession codes: 8K1G for four ethylene glycol-bound GlyDH structure; and 8K1H for both ethylene glycol- and NAD⁺-bound GlyDH structure. These are publicly available as of the date of publication. Figs were prepared using PyMol (PyMOL Molecular Graphics System, Version 2.5.0 Schrödinger, LLC).

RESULTS AND DISCUSSION

Overall Structure of Ethylene Glycol-bound KpGlyDH in the Absence of NAD⁺

The crystal structure of full-length KpGlyDH in the absence of NAD⁺ was determined using molecular replacement at a resolution of 2.1 Å. Data collection and refinement statistics are summarized in Table S1. One protomer of KpGlyDH was found in the asymmetric unit, while the octamer KpGlyDH was observed through crystallographic symmetry operations (Fig. S1). It comprises of 9 β-strands, 13 α-helices, and 2 3₁₀ helices, forming two domains with a deep cleft (Fig. 2). The N-terminal α/β domain (residues 1–161) consists of a parallel 6-stranded β sheet (β4↑ β3↑ β5↑ β6↑ β9↑ β2↑ with an additional β-strand (β1↑) from the adjacent protomer, and a flanked by 5 α-helices (Fig. 2 and S1b). Two β-strands (β7, β8) are separated by a β hairpin. The C-terminal domain (residues 162–367) comprises two subdomains forming a bundle of α-helices. During the iterative refinement process, several blobs of the electron density maps (Fo-Fc) were observed near the active site (Fig. 2c). The first blob of the electron density map in the deep cleft between N- and C-terminal domain represented a zinc ion coordinated to two ethylene glycol molecules (EG1 and EG2), a water molecule, and His271. The ethylene glycol molecules were derived from as a cryoprotectant for X-ray data collection.

Figure 2. Overall structure of ethylene glycol (EG)-bound KpGlyDH. (a,b) Overall structure of EG-bound KpGlyDH. Four EG molecules are depicted as stick models, and a zinc ion is represented as a gray ball. (c) Closed-up view of the active site. The stick models for the four EG molecules are superimposed with polder Fo-Fc electron density maps contoured at 3.0 times RMSD (green). The crystal structure of Bacillus stearothermophilus GlyDH (PDB ID: 1JQA) was superimposed with KpGlyDH, and the glycerol molecule is shown as a purple-colored stick model representation (RMSD 0.880). The crystal structure of Escherichia coli GlyDH (PDB ID: 8GOA) was superimposed with KpGlyDH, and three Tris molecules are displayed as yellow-colored stick model representations (RMSD 0.330). (d) The crystal structures of Bacillus stearothermophilus GlyDH (PDB ID: 1JQA) and Escherichia coli GlyDH (PDB ID: 8GOA) was superimposed with KpGlyDH. EG, glycerol, and Tris molecules are shown as stick models, and a zinc ion is represented as a gray ball.

The first ethylene glycol molecule, referred to as EG1, was coordinated to the zinc ion at a distance of 2.5 Å (Fig. 2c and S2a). The position of EG1 was previously identified in crystallographic studies as the substrate or inhibitor binding site. The second ethylene glycol molecule, termed as EG2, was also coordinated to the zinc ion at a distance of 2.5 Å (Fig. 2c and S2a). In the structure of the homologous GlyDH, the carboxylate group of Asp171 is coordinated to a zinc ion. The distances between the zinc ion and the hydroxyl group of the ethylene glycol molecule were slightly extended. The third ligand of the zinc ion was the NE2 atom of His271 at a distance of 2.1 Å (Fig. 2c and S2a). The last ligand for the zinc ion was His254 in the homologous structure,^13a,13c,13g but the side chain of His254 in our structure swung out and formed a hydrogen bond with the third ethylene glycol molecule, referred to as EG3. The distance of the hydrogen bond between EG3 and His254 is 3.1Å (Fig. 2c and S2a). The last ethylene glycol molecule, termed as EG4, formed a hydrogen bond with His257 and Lys274 (Fig. 2c and S2a). The site of EG4 was identified as the binding site for the Tris molecule, while the binding sites of EG2 and EG3 were not previously identified. Ethylene glycol molecule was characterized as a substrate for KpGlyDH, having a similar binding affinity to the glycerol molecule, but the specific constant (k_cat/K_m) was 30 times lower than that of the glycerol molecule.¹² By comparison to the homolog structures, our crystal structure of KpGlyDH represented an open form, identified as the β-harpin conformation, consisting of two β-strands (β7 and β8), whereas glycerol-bound structure of Bacillus stearothermophilus GlyDH in the absence of NAD⁺ (PDB ID: 1JQA) adopted a closed form (Fig. 2d).^13a,13g The alterable β-hairpin conformation has been demonstrated to contribute to cofactor binding and catalytic efficiency.^13a In conclusion, we observed two ethylene glycol molecules coordinated to a zinc ion, and the other two ethylene glycol molecules may represent the pathway for the substrate. Furthermore, our structure, in conjunction with previous structures, supports the idea of structural flexibility in the β-hairpin.

Overall Structure of Ethylene Glycol-bound KpGlyDH in the Presence of NAD⁺

Subsequently, we determined the crystal structure of KpGlyDH in the presence of NAD⁺, which was added during the cryoprotection process, at a resolution of 2.1 Å (Table S1). The structure of the enzyme exhibited minimal differences compared to the structure in the absence of NAD⁺, with the root-mean-square deviation (RMSD) of 0.34. The crystal structure of KpGlyDH in the presence of the cofactor revealed the coordination of an ethylene glycol molecule to a zinc ion, as well as the presence of a tartrate ion in the second binding site for the substrate (Fig. 2a and 2b). The tartrate ion was derived from crystallization reagents. During the iterative refinement process, the electron density maps (Fo-Fc) revealed the presence of NAD⁺ (Fig. 3c). Upon further analysis, we determined that the cofactor was fully occupied, and engaged in hydrogen bonding interactions with Asp37, Lys95, Thr116, Ser119, Leu127, and Tyr131 (Fig. 3c). Interestingly, the coordination of the zinc ion reverted to the known structure reported elsewhere (Fig. 3d and S3).¹³ It is noteworthy that the occupancy of the zinc ion in the cofactor-deficient structure was lower (0.22) compared to the fully occupied zinc ion in the cofactor-bound structure. This disparity in occupancy cannot be disregarded, as it may impact the zinc coordination. In the cofactor-bound structure, the distances between the zinc ion and ligands (Asp171, His254, His271, and ethylene glycol molecule) in the cofactor-bound structure fell within the range of 1.9 to 2.2 Å, which is the average distance observed in the protein structures¹⁸ (Fig. 3d and S2b).

Figure 3. Overall structure of NAD⁺-bound KpGlyDH. (a,b) Overall structure of NAD⁺-bound KpGlyDH. NAD⁺, ethylene glycol (EG), and tartrate molecule (TLA) are depicted as stick models, while a zinc ion is represented as a gray ball. (c) Closed-up view of the NAD⁺-binding site. The stick model for NAD⁺ is superimposed with polder Fo-Fc electron density maps, contoured at 4.0 times RMSD (green). Key interacting residues are displayed as stick models. (d) Closed-up view of the active site. The stick models for ethylene glycol and tartrate molecules are shown, superimposed with polder Fo-Fc electron density maps contoured at 3.0 times RMSD (green). Water molecules surrounding a tartrate ion are represented as red balls.

Additionally, a tartrate ion from the crystallization reagents was discovered at the second binding site for the substrate, forming hydrogen bonds with Lys95, His257, His271, Lys274, and water molecules (Fig. 3d and S2b). Due to its larger size, the tartrate cannot be accommodated in the substrate site near the zinc ion. The second binding site for the substrate, observed with tartrate in this study and with Tris in a prior study, represents the substrate pathway.^13a Additionally, our structure demonstrates that GlyDH prefers smaller substrates of less than C3 due to limited access to the active site, as indicated by the capture of tartrate in the second binding site. In summary, the crystal structure of KpGlyDH in the presence of NAD⁺ was presented here, featuring the zinc ion-bound ethylene glycol, the substrate, and the tartrate as a substrate-mimic situated in the second substrate binding site, approximately 8 Å away from the zinc ion.

Comparison Between EG-bound KpGlyDH Structures in the Absence and Presence of NAD⁺

As mentioned above, the structural disparities between EG-bound KpGlyDH structures in the absence and presence of NAD⁺ extend beyond zinc coordination. The occupancies of zinc in both structures were estimated during the structure refinement. In the NAD⁺-bound structure, the coordination of zinc involves the participation of four ligands, while three ligands are coordinated to zinc in the NAD⁺-unbound structure (Fig. 4). The most distinguished feature in both structures is His254, which has a side chain that rotates 102° outward to form hydrogen bonds with the ethylene glycol in NAD⁺-unbound structure (Fig. 4b). In addition, the distance between Asp171 and the zinc is 3.0 Å in NAD⁺-unbound structure, while it is 1.9 Å in NAD⁺-bound structure (Fig. 4b). While ethylene glycol, labelled as EG2, coordinates to zinc, both His254 and Asp171 are retracted. The dynamics in the active site require further studies. In summary, the comparison between EG-bound KpGlyDH structures in the absence and presence of NAD⁺ reveals unexpected dynamics of zinc coordination, providing valuable insights into the structural adaptability under varying conditions.

Figure 4. Structural superposition of a zinc binding site. The color codes are consistent with those in Fig. 1 and Fig. 2. (a) Zinc binding site of EG-bound KpGlyDH in the absence of NAD⁺. (b) Zinc binding site of NAD⁺-bound KpGlyDH. The stick models of NAD+- unbound state are shown in transparent for comparison.

Substrate Tunnel Analysis of Octamer GlyDH

In order to comprehend the significance of the octameric form of GlyDH, we conducted further investigations. While previous structural studies, including our own, have reported the presence of an octameric structure,^12,13g the relationship between this oligomeric state and the enzyme’s activity remains unclear. The dimeric form of GlyDH is formed through crystallographic 2-fold symmetry, while octamer arises through crystallographic 4-fold symmetry from the dimer. An assembly analysis of GlyDH, conducted using PDBePISA, revealed that the free energies of dissociation (ΔG^diss) were 5.9 kcal mol^-1 for the dimer and 24.5 kcal mol^-1 for the octamer.¹⁹ This indicates that the octameric GlyDH is a tetramer of dimers. Furthermore, we performed an analysis of the substrate tunnel within the octameric GlyDH using CAVER 3.0.3, a plugin integrated into the PyMol software.²⁰ To initiate the analysis, the starting point of the substrate tunnel was set as an input parameter. Given that the ethylene glycol molecule in our structure is coordinated to the zinc ion and resided in the well-known substrate binding site, it was reasonable to designate it as the starting point for the substrate tunnel. Other parameters were set to their default values. The substrate tunnel, highlighted in red (Fig. 5), represented a pathway within one protomer, while the other pathway traversed through two active sites in the dimeric GlyDH and extended into the second substrate binding site in the opposite protomer, highlighted in blue (Fig. 5). This result aligns with our crystal structure, which shows the presence of the ethylene glycol molecule, referred to as EG2, at the backside of the zinc ion. Furthermore, this finding suggests a potential correlation between the oligomeric status of GlyDH and cooperative effects on catalytic activity, whether positive or negative. In summary, we have identified substrate tunnels within the dimeric GlyDH, representing independent four channels that exist within the octameric GlyDH structure.

Figure 5. Overall structure of the octamer KpGlyDH, generated through crystallographic symmetry operations. The substrate tunnels, highlighted in blue and red, were analyzed using the CAVER 3.0.3 plugin on PyMol. A zinc ion is represented as a gray ball, while NAD⁺ molecule is represented as a stick model.

CONCLUSION

In this study, we presented the crystal structure of ethylene glycol-bound KpGlyDH both in the absence and presence of NAD⁺, as well as conducted an analysis of the substrate tunnel within the octameric GlyDH. The crystal structures of KpGlyDH revealed the presence of the ethylene glycol molecules coordinated to a zinc ion in the active site. In the absence of NAD⁺, the structure showed that two ethylene glycol molecules replaced His254 and Asp171 in coordinating the zinc ion. On the other hand, in the presence of NAD⁺, the crystal structure displayed an ethylene glycol-bound structure along with other ligands, such as Asp171, His254, and His271, coordinating the zinc ion. Additionally, the tartrate ion occupied the second binding site for the substrate which was identified in the prior study.^13a Furthermore, the substrate tunnel analysis within the octameric GlyDH unveiled independent four channels, including pathways within each protomer and interactions between active sites in the dimeric form. This finding suggests potential cooperativity in the catalytic activity of the octameric GlyDH.

Collectively, our findings provide valuable insights into the structural characteristics and functional implications of ethylene glycol-bound KpGlyDH, highlighting the interplay between the oligomeric state, NAD⁺ binding, substrate binding sites, and catalytic activity. Further investigations will contribute to a comprehensive understanding of GlyDH and its biological functions.