Structural basis for substrate specificity and mechanism of N-acetyl-D-neuraminic acid lyase from Pasteurella multocida

Abstract

N -Acetylneuraminate lyases (NALs) or sialic acid aldolases catalyze the reversible aldol cleavage of N-acetylneuraminic acid (Neu5Ac, the most common form of sialic acid) to form pyruvate and N-acetyl-D-mannosamine (ManNAc). Although equilibrium favors sialic acid cleavage, these enzymes can be used for high-yield chemoenzymatic synthesis of structurally diverse sialic acids in the presence of excess pyruvate. Engineering these enzymes to synthesize structurally modified natural sialic acids and their non-natural derivatives holds promise in creating novel therapeutic agents. Atomic resolution structures of these enzymes will greatly assist in guiding mutagenic and modeling studies to engineer enzymes with altered substrate specificity. We report here the crystal structures of wild-type Pasteurella multocida N-acetylneuraminate lyase and its K164A mutant. Like other bacterial lyases, it assembles into a homotetramer with each monomer folding into a classic (β/α)₈ TIM barrel. Two wild-type structures were determined; in the absence of substrates, and trapped in a Schiff base intermediate between Lys164 and pyruvate, respectively. Three structures of the K164A variant were determined: one in the absence of substrates and two binary complexes with N-acetylneuraminic acid (Neu5Ac) and N-glycolylneuraminic acid (Neu5Gc), respectively. Both sialic acids bind to the active site in the open-chain ketone form of the monosaccharide. The structures reveal that every hydroxyl group of the linear sugars makes hydrogen bond interactions with the enzyme and the residues that determine specificity were identified. Additionally, the structures lend some clues in explaining the natural discrimination of sialic acid substrates between the P. multocida and E. coli NALs.

Introduction

Sialic acids are negatively-charged α-keto acids with a nine-carbon backbone (1–3). More than 50 structurally distinct sialic acids have been found in nature. Among these, N-acetylneuraminic acid (Neu5Ac) is the most common form and has been found in bacteria, human, animals and other higher vertebrates. In comparison, N-glycolylneuraminic acid (Neu5Gc), which has an extra oxygen atom at the C-5 N-glycolyl group compared to the N-acetyl group in Neu5Ac, is biosynthesized by non-human animals. Humans do not synthesize Neu5Gc due to the lack of an active CMP-N-acetylneuraminic acid hydroxylase (CMAH) that is responsible for producing CMP-Neu5Gc from CMP-Neu5Ac (4). Nevertheless, humans can incorporate dietary Neu5Gc (from cow milk or more abundantly from red meat) and present them on cell surface glycoconjugates. All human individuals have varied levels of anti-Neu5Gc antibodies, which are named as xeno-autoantibodies (5). Cancer cells can overexpress Neu5Gc and some cancer patients have elevated levels of anti-Neu5Gc antibodies against Neu5Gc-containing tumor associated carbohydrate antigens (TACAs) (6).

The broad distribution of sialic acids in nature and the remarkable structural diversity make sialic acids strongly influential in cell biology. The carboxyl group of sialic acids is typically deprotonated under physiological pH, giving sialic acids a net negative charge. The monosaccharides are commonly found at the non-reducing terminal position of cell surface glycans of higher eukaryotes. Sialic acids are therefore key recognition sites and are among the first molecules to be encountered in cellular interactions. Due to these characteristics, sialic acids are involved in important intracellular and intercellular processes such as molecular recognition, cell-cell communication, bacterial and viral infection, and tumor metastasis (2, 3, 7, 8).

The growing significance of sialic acid function has made understanding the function and mechanism of enzymes involved in the biosynthesis and metabolism of sialic acids very important. N-Acetyl-D-neuraminate lyase (NAL; EC 4.1.3.3), also known as sialic acid aldolase, catalyzes the reversible aldol condensation of N-acetyl-D-mannosamine (ManNAc) and pyruvate to produce Neu5Ac, with the equilibrium favoring Neu5Ac cleavage (9–14). NAL is the archetype member of a superfamily of enzymes that share unifying mechanism in the reaction pathway by forming a Schiff base between a conserved lysine residue in a TIM (β/α)₈-barrel to carbon C2 of an α-keto acid moiety of the substrate (15). Other members in the family include: dihydrodipicolinate synthetase, D-5-keto-4-deoxyglucarate dehydratase, trans-O-hydroxybenzylidenepyruvate hydrolase-aldolase, trans-2′-carboxybenzalpyruvate hydratase-aldolase, and 2-keto-3-deoxygluconate aldolase (15–18).

NAL is present in various mammalian tissues and in both pathogenic and non-pathogenic bacteria (19–23). While NAL functions primarily to regulate sialic acid metabolism in mammalian cells, some microorganisms use NAL to catabolize sialic acid for a carbon and energy source (24–31). NAL belongs to the class I aldolase family, which is characterized by a TIM-barrel fold and aldol condensation proceeding through the formation of a Schiff base between a conserved lysine residue and the substrate pyruvate. In contrast, class II aldolases differ in reaction mechanism by which intermediates are stabilized by a metal cofactor (e.g. Zn⁺) (13, 32, 33).

An important application of NAL has been in vitro chemoenzymatic synthesis of Neu5Ac and its derivatives, especially sialic acid derivatives requiring enantioselective aldol condensation that are difficult to produce by traditional chemical synthesis (9, 10, 27, 34). NAL can also be used to determine sialic acid concentrations by coupling to lactate dehydrogenase or pyruvate oxidase (9, 35). Likewise, NAL coupled with sialidase can be used to quantitate the total amount of sialic acid in glycoproteins, glycolipids, polysialic acids, and on cell surfaces (9, 36).

The use of a capillary electrophoresis assay method, which is able to directly monitor the reactions of NAL proceeding in both directions of aldol condensation and cleavage, showed that Pasteurella multocida aldolase (PmNAL), and not EcNAL (Escherichia coli), can utilize an O-methylated derivate 5-O-methyl-ManNAc efficiently for the synthesis of the corresponding O-methyl sialic acid 8-O-methyl-Neu5Ac (Neu5Ac8OMe) (9). Neu5Ac8OMe is a naturally occurring sialic acid found in gangliosides from starfish and sperm eggs of teleost. Chemical synthesis of Neu5Ac8OMe is difficult and low yielding, but recently it has been efficiently synthesized through the utilization of PmNAL and the 5-O-methyl-ManNAc derivate (9, 37). PmNAL has also been used for the synthesis of several other C8-modified sialic acids, such as Neu5Gc8Me, Kdn8Me, and Kdn8Deoxy, which may serve to understand the biological functions of 8-OMe modifications of sialic acid and sialosides that occur in nature (37).

The X-ray crystal structures has been previously determined for N-acetyl-D-neuraminate lyase from E. coli (EcNAL) (13, 19), H. influenzae (HiNAL) (16), and S. aureus (SaNAL) (38). To gain more insight on the substrate specificity of the NALs from different organisms and to assist in structure-based protein engineering, we set out to determine the high resolution crystal structure of N-acetyl-D-neuraminate lyase from Pasteurella multocida and compare it to the structures of other NALs. Here we present the crystal structures of PmNAL in both wild-type and mutant forms. Structures of the wild-type PmNAL are in the native form and complexed with pyruvate. PmNAL K164A mutant was used to study sialic acid binding and crystal structures were determined in ligand-free form and in complexed forms with Neu5Ac and Neu5Gc, where they bound to the active site in the open-chain ketone form.

Materials and Methods

Site-Directed Mutagenesis

Cloning of wild-type NAL were performed as previously reported (9). Site-directed mutagenesis to create the K164A variant was generated using full-length PmNAL plasmid as a template and the primers PmNAL_K164A_F (5′-CCAAAAGTTTTAGGGGTGGCCTTTACCGCGGGTGATTTCTACTTATTAGAGCGCTTG-3′) and PmNAL_K164A_R (5′-CAAGCGCTCTAATAAGTAGAAATCACCCGCGGTAAAGGCCACCCCTAAAACTTTTGG-3′). Polymerase chain reaction for mutagenesis was performed in a 50 μl reaction mixture containing full-length PmNAL plasmid (1 μg), forward and reverse primers (1 μM each), 10x PCR buffer (5 μl), 10x MgCl₂ (5 μl), dNTP mixture (1 mM), and 5 U (0.1 μl) of Taq DNA polymerase. The reaction mixture was subjected to 30 cycles of amplification with an annealing temperature of 55°C. The PCR product was transformed into chemically competent TOP10 E. coli cells (Invitrogen) and DNA sequencing performed by Davis Sequencing (Davis, CA).

Expression and Purification of PmNAL

Expression and purification of wild-type PmNAL and K164A mutant was performed as previously reported (9). Plasmids with PmNAL insert were selected and transformed into BL21 (DE3) chemical competent cells. E. coli BL21 (DE3) containing the recombinant plasmid in pET22b(+) vector was cultured in LB-rich medium (10 g/L tryptone, 5 g/L yeast extract, and 10 g/L NaCl) supplemented with 100 μg/mL ampicillin. The cell cultures were grown to an OD_600nm of 0.8–1.0, induced with 0.1 mM IPTG (isopropyl-1-thio-β-D-galactopyranoside), and incubated at 20 °C for 24 h with shaking at 250 rpm. Purification of His₆-tagged PmNAL protein was by affinity chromatography using the ÄKTA FPLC system (GE Healthcare) equipped with a HisTrap_FF 5-mL column. The bacterial cells were harvested by centrifugation at 4000 g for 20 min. Cell pellets were resuspended in 20 mL of lysis buffer (Tris-HCl, pH 8.0, 100 mM, 0.1% Triton X-100) containing lysozyme (100 μg/mL) and DNaseI (3 μg/mL) at 37 °C for 50 min with vigorous shaking. The cell lysate was collected by centrifugation at 12,000 g for 30 min and the supernatant (lysate) applied to a HisTrap FF 5 mL column (GE Healthcare). The column was then washed with 10 volumes of binding buffer (5 mM imidazole, 0.5 M NaCl, 20 mM Tris-HCl, pH 7.5), 15 volumes of washing buffer (30–50 mM imidazole, 0.5 M NaCl, 20 mM Tris-HCl, pH 7.5), followed by 8 volumes of elute buffer (200 mM imidazole, 0.5 M NaCl, 20 mM Tris-HCl, pH 7.5). The fractions containing the purified enzyme were collected, dialyzed against Tris-HCl buffer (20 mM, pH 7.5) containing 10% (v/v) glycerol, and stored at 4 °C. Protein concentration was determined in a 96-well plate using a bicinchoninic acid protein assay kit (Pierce Biotechnology, Rockford, IL, USA) with bovine serum albumin as a protein standard. The absorbance of each sample was measured at 562 nm by a BioTek SynergyTM HT Multi-Mode Microplate Reader. The expression profiles of PmNAL were analyzed by 12% sodium dodecyl sulfate–polyacrylamide gel electrophoresis (SDS-PAGE). The purified protein exhibited a molecular mass of about 33 kDa, matching well with the calculated masses of the translated His₆-tagged proteins of 33.7 kDa (39).

Crystallization of PmNAL

PmNAL was crystallized in three different conditions. The wild-type PmNAL crystals grew by handing drop in 21% polyethylene glycol (PEG)-1000, 150 mM NaCl, 100 mM Na₂HPO₄-KH₂PO₄, pH 6.2. Wild-type PmNAL crystals were soaked with 50 mM pyruvate for 24 h to obtain the wild-type PmNAL binary structure with pyruvate. PmNAL K164A mutant in ligand-free form was grown by hanging drop in 30% PEG-200, 100 mM NaCl, and acetate, pH 4.5. Crystals of PmNAL K164A mutant bound to either Neu5Ac or Neu5Gc were grown in 38% PEG-300, 0.01 M CaCl₂, 0.1 M sodium cacodylate, pH 6.5. Sialic acid concentration was 5 mM.

Data Collection, Model Building, and Refinement

X-ray diffraction data for all crystals except for the K164A mutant complexed with Neu5Gc were collected at Stanford Synchrotron Radiation Laboratory (SSRL) beam lines at 100 K. The SSRL data were indexed and integrated with MOSFLM, and then scaled with SCALA. Diffraction data for the K164A mutant complexed with Neu5Gc were collected on a rotating anode home X-ray source at 85 K. These data were processed and scaled with the PROTEUM software suite (Bruker AXS, Madison, WI). All data collection and structure refinement statistics are given in Table 1.

Table 1.

Data collection and Refinement Statistics

X-ray Source	Wild-type SSRL BL 7-1	Wild-type w/ pyruvate Schiff Base SSRL BL 7-1	K164A SSRL BL 7-1	K164A w/ Neu5Ac SSRL BL 7-1	K164A w/ Neu5Gc Rotating Anode
Wavelength (Å)	0.97607	0.97607	0.97946	0.97946	1.54
Space Group	P21	P21	P21	C2221	C2221
Cell parameters	a = 80.9 Å	a = 80.8 Å	a = 90.1 Å	a = 81.7 Å	a = 82.2 Å
	b=113.5 Å	b=114.1 Å	b=147.7 Å	b=150.2 Å	b=150.0 Å
	c = 82.2 Å	c = 82.0 Å	c = 112.0 Å	c = 113.1 Å	c = 112.5 Å
	β = 111.35°	β = 111.74°	β = 98.51°
Monomers/ASU	4	4	8	2	2
VM (Å/Da)	2.67	2.66	2.79	2.63	2.63
Resolution (Å)	1.85 (1.90-1.85)	2.10 (2.15-2.10)	1.75 (1.80-1.75)	1.90 (1.95-1.90)	1.85 (1.94-1.85)
No of reflections	372,657 (22,725)	282,694 (21,074)	1,072,927 (73,656)	199,068 (14,798)	573,486 (39,412)
No. of unique reflections	112,841 (7,869)	79,123 (5,837)	287,971 (20,895)	54,641 (4,027)	59,552 (7,821)
Completeness (%)	95.8 (90.4)	98.3 (97.9)	99.0 (97.4)	99.2 (99.8)	100.0 (100.0)
Rmergea (%)	5.5 (31.8)	8.4 (32.5)	7.3 (45.9)	3.1 (50.6)	7.4 (42.8)
I/σ	9.3 (2.7)	14.0 (3.4)	13.6 (3.0)	25.7 (2.9)	18.0 (2.5)
Refinement Statistics
Resolution (Å)	1.85	2.10	1.75	1.90	1.85
No. of reflections (F>0)	107,159	75,078	273,403	51,885	56,427
Rfactorb (%)	17.2	17.8	17.5	16.7	17.8
Rfreeb (%)	20.9	22.0	20.8	20.0	21.6
RMS bond length (Å)	0.015	0.018	0.015	0.017	0.017
RMS bond angle (°)	1.353	1.842	1.671	1.761	1.853
Ramachandran Plot Statistics
Most Favorable (%)	1145 (98.7%)	1138 (98.8%)	2302 (98.9%)	570 (98.3%)	571 (98.1%)
Allowed (%)	15 (1.3%)	14 (1.2%)	24 (1.0%)	10 (1.7%)	11 (1.9%)
Disallowed (%)	0 (0%)	0 (0%)	2 (0.1%)	0 (0%)	0 (0%)
Asymmetric Unit Content
Non-hydrogen atoms	10,594	10,169	21,389	4,949	5,388
Waters	1,206	799	2,785	253	645
PDB ID	4IMC	4IMD	4IME	4IMF	4IMG

^aR_merge = [Σ_hΣ_i|I_h − I_hi|/Σ_hΣ_iI_hi] where I_h is the mean of I_hi observations of reflection h. Numbers in parenthesis represent highest resolution shell.

^bR-Factor and

^cR_free = Σ||F_obs| − |F_calc|| / Σ|F_obs| × 100 for 95% of recorded data (R-Factor) or 5% data (R_free)

The PmNAL wild-type structure was solved by molecular replacement using the previously solved NAL structure from Haemophilus influenzae (PDBID: 1f5z) with the program PHASER (40). Subsequent PmNAL structures were solved using the wild-type PmNAL structure as a phasing model. Atomic model building was carried out with the molecular graphics program COOT (41). The structure was refined with the program REFMAC5 (42) using 95% of the measured data as a target function. Noncrystallographic symmetry restraints and TLS parameters were included during refinement. The final R-factor and R-free along with the quality of the models based on PROCHECK are listed in Table 1.

Results

Overall structures

Each monomer consists of 293 residues and is characterized by a TIM-barrel (β/α)₈-fold (43) followed by an additional three alpha-helices at the C-terminus (αI, αJ, αK) (Figure 1A). For the wild-type protein, four monomers of the crystallographic asymmetric unit assemble into the biological functional tetramer with 222 symmetry and overall dimensions ~52 Å × 40 Å × 35 Å (Figure 1B). The ligand-free PmNAL K164A mutant contains two tetramers in the asymmetric unit, while the structures of K164A mutant with bound sialic acids, only contain two monomers in each asymmetric unit, with a crystallographic two-fold generating the biological functional tetramer.

Figure 1.

(A) PmNAL monomer. Ribbon drawing of wild-type PmNAL monomer represented by a rainbow color scheme from blue (N-terminus) to red (C-terminus). Each monomer consists of a TIM barrel (β/α)₈-fold followed by an additional three alpha-helices at the C-terminus (αI, αJ, αK). Secondary structure elements are labeled. Displayed in the active site is the Lys164-pyruvate Schiff base intermediate shown with sticks in yellow colored carbon atoms. (B) PmNAL tetramer with each subunit displayed in a different color. The tetramer possesses D2 symmetry with access to the four active sites from the central cavity.

All subunits in the ligand-free wild-type tetramer are structurally similar and superpose with a root-mean-square deviation (rmsd) range between 0.085 – 0.173 Å for 293 α-carbons. For the wild-type pyruvate-bound superposition, the range in rmsd is between: 0.089 – 0.137 Å for individual monomer superpositions within the asymmetric unit. For the ligand-free K164A mutant, the rmsd superpositions range from 0.104 – 0.196 Å for 8 monomers in the asymmetric unit, while the K164A mutant with Neu5Ac and Neu5Gc bound the monomers superimpose with rmsd of 0.146 Å and 0.163 Å respectively.

Wild-type PmNAL Binary Structure with Pyruvate

The mechanism of NAL has been proposed to proceed through the formation of a Schiff base between a conserved lysine residue (Lys163) and the substrate (44). Crystals of the wild-type PmNAL were soaked with 50mM pyruvate for 24 hours in the reservoir buffer. The soaked crystals were transferred to a cryogenic solution, containing reservoir buffer supplemented with 30% ethylene glycol and 50mM pyruvate, for a few seconds and flash frozen in liquid nitrogen prior to data collection. The resulting electron density map clearly showed that the Nζ of Lys164 is conjugated as a Schiff’s base to C2 of pyruvate (Figure 2A). The carboxylate group of the pyruvate hydrogen-bonds to main-chain nitrogens of Ser47 and Thr48. Additionally, O1 of the pyruvate hydrogen-bonds to Oγ1 of Thr48 (Figure 2B). Similar interactions are seen in the pyruvate complexes with EcNAL (13) and SaNAL (38). Ser47–Thr48 is part of a highly conserved GSTGE motif. Glu50 of this motif points away from the active site where it ion-pairs with conserved Lys255.

Figure 2.

(A) Schiff base intermediate in the active site of PmNAL. The aldol reaction is mediated by conserved residue Tyr136, and an ordered water molecule, likely coming from the oxygen of the pyruvate, is also shown. The electron density represented by blue mesh contoured at 1σ showing continuous density linking C2 of pyruvate and Nζ of Lys164. (B) Hydrogen-bonding network around the pyruvate moiety forming a Schiff base with Lys164. Hydrogen-bond interactions between the Schiff base intermediate and nearby residues are represented by yellow dashed lines with the corresponding distances shown in Ångstroms. The carboxyl group of the pyruvate forms hydrogen bonds to the main-chain amide nitrogens of Ser47 and Thr48 as well as Oγ1 of Thr48. Two conformations of Phe189 are shown near the active site.

Tyr136 on strand βe lies nearly parallel to the Schiff base intermediate and is involved in a hydrogen-bonding network for the Schiff base reaction. Tyr136 is proposed to assist in abstracting a proton from Lys164 and transferring it to the carbonyl C2 of pyruvate with the release of a water molecule (16). The water molecule is observed only in the A subunit of the tetramer (Figure 2). This hydrogen-bonding network is consistent with the proposed mechanism of Schiff base formation illustrating the release of a water molecule during the reaction. Tyr136 also hydrogen bonds to Ser47, part of a GSTGE motif, suggesting Ser47 may play a role in catalysis, possibly helping to position Tyr136.

The electron density map shows two alternate conformations for Phe189 in the pyruvate bound structure (Figure 2B), while only one conformation is seen in the ligand-free structure. In the ligand-free structure, as well as the sialic acid-bound structures (below), the Phe189 points towards the active site. In fact, in the K164A mutant structure with bound sialic acid, it makes van der Waals contact (~4.5 Å) with C7 of sialic acid. In the pyruvate-bound wild-type structure, the other conformation of Phe189 swings away from the active site, where it becomes buried between Met192, Phe170, and the main chain of Gly168. This results in moving the main chain up slightly (~1 Å) at Ala167. Considering this Phe189 mobility is only observed in the pyruvate-bound enzyme and not the ligand-free or full substrate bound (sialic acid) structures, suggests that Phe189 movement, may help assist in catalysis during the Schiff base formation of Lys164 with substrate.

PmNAL K164A Mutant with Sialic Acid

The wild-type structure with pyruvate complexed in a Schiff base to Lys164 provided insight to the pyruvate moiety binding to the active site, but not the ManNAc. Subsequent wild-type crystal soaks with ManNAc followed by X-ray diffraction data collection did not yield any electron density to the ManNAc moiety in the active site (data not shown). This is most likely due to the wild-type enzyme having a low binding affinity to the ManNAc, the product in the lyase reaction. Crystals of wild-type enzyme soaked with Neu5Ac resulted in the crystals dissolving. Therefore, to inhibit activity and to further analyze the interactions between substrate and residues in the active site, a PmNAL K164A variant was constructed.

The ligand-free PmNAL K164A crystallized with two tetramers in the asymmetric unit in space group P2₁. The structure was determined for the K164A variant in the ligand-free form at 1.75 Å resolution (Table 1). PmNAL K164A complexed with Neu5Ac crystallized under different crystallization conditions resulting in a different space group (C222₁), with two monomers in the asymmetric unit (one of the crystal 2-folds generates the functional tetramer) (Table 1). Structures were determined at 1.90 Å and 1.85 Å resolution for Neu5Ac- and Neu5Gc-bound respectively.

The crystal structure of the PmNAL K164A variant in the ligand-free state shows very little structural change compared to the wild-type. Superpositions of K164A monomers have an rmsd range of 0.23 – 0.26 Å (293 equivalent alpha-carbons) onto the ligand-free wild-type structure. Binding Neu5Ac or Neu5Gc into the active site has little structural difference as measured by alpha-carbon main chain superpositions (rmsd range 0.24 – 0.30 Å) to ligand-free wild-type and K164A variant. Therefore, crystals of wild-type enzyme dissolving during Neu5Ac soaks, suggest subtle conformational changes occur either upon substrate binding or during catalysis, which result in breaking lattice contacts.

In the co-crystal structure of PmNAL K164A with Neu5Ac, electron density clearly defined the Neu5Ac binding to the active sites in both subunits in the crystallographic asymmetric unit (Figure 3). The Neu5Ac binds to the enzyme in the linear ketone form with the pyruvate moiety binding in the same location as seen in the wild-type pyruvate Schiff base intermediate structure, where it makes similar contacts to the enzyme. The ManNAc portion of the Neu5Ac hydrogen bonds to highly conserved side chains of Thr48, Thr166, Asp190, Glu191, Ser207, and Tyr251, and main chain atoms in Ser47, Thr48, Gly188, Asp190, and Ser207 (Figure 4). Every hydroxyl oxygen of Neu5Ac participates in at least one hydrogen bond to the enzyme. The N-acetyl group points up out towards the solvent where the oxygen atom hydrogen-bonds the side chains of Thr48 and Tyr251 (Figure 3B). The nitrogen atom of the N-acetyl group does not interact with the enzyme.

Figure 3.

Active site of PmNAL K164A with sialic acids bound. (A) 2Fo-Fc electron density map contoured at 1σ is shown in blue mesh modeled around the Neu5Ac (sticks with white carbon bonds). (B) Active site showing potential hydrogen bonds between Neu5Ac and protein atoms. (C) 2Fo-Fc electron density map contoured at 1σ showing the configuration of Neu5Gc bound to PmNAL K164A. (D) Active site showing potential hydrogen bonds between Neu5Gc and enzyme.

Figure 4.

Flatten two-dimensional representation of Interactions between PmNAL K164A and Neu5Ac. Covalent bonds are represented by solid green lines for Neu5Ac and tan lines for protein residues. Hydrogen bonds are shown as black dashed lines. van der Waals contacts are drawn as hashed red lines around an atom pointing in the direction of the corresponding interacting atom. Diagram was generated with the program LIGPLOT (48).

Crystals were also grown for the K164A mutant in the presence of N-glycolylneuraminic acid (Neu5Gc), where the methyl group on the acetyl group of Neu5Ac is hydroxylated forming a glycolyl group. Again, electron density clearly defines the sugar binding to the enzyme in the linear ketone form (Figure 3C). The conformation and binding interactions are essentially identical to Neu5Ac binding. The only exceptions are the hydroxyl of the glycolyl group hydrogen bonds to Thr48 and Tyr251 in place of the carbonyl oxygen of the amide group (Figure 3). While Thr48 is strictly conserved, Tyr251 is not as conserved. In E. coli, this residue is a Phe, while in mammals this residue is typically Ser or Pro. Previously it was shown that both E. coli NAL and PmNAL catalyzed the formation of Neu5Gc similarly well (9), suggesting this Tyr251 interaction with the acetyl/glycolyl group is not essential. Another difference is observed in the pyruvate moiety where the electron density suggests it is more planar in Neu5Gc compared to Neu5Ac (Figure 3), which was also confirmed by refinement using simulated annealing protocols in PHENIX (45). All the hydroxyls of Neu5Gc participate in hydrogen bond interactions with the enzyme as in Neu5Ac binding.

Structural differences and substrate specificity

Previously we have shown that PmNAL can catalyze the synthesis of 8-O-methyl-Neu5Ac by coupling 5-O-methyl-ManNAc and pyruvate, while the enzyme from E. coli is incapable of catalyzing this reaction (9). Moreover, while PmNAL can catalyze this synthesis, it does so much less efficiently compare to ManNAc and other ManNAc derivatives (9). Inspection of the structures gives some clues for this specificity. The PmNAL K164A structure with Neu5Ac bound reveals that the hydroxyl oxygen (O8) linked to carbon 8, which is carbon 5 in ManNAc, makes a hydrogen bond to the main chain amide nitrogen of Asp190 and is tightly packed between the side chains of Asp190, Glu191, and Phe189 (Figure 4). In this structure, it appears there is little room to accommodate a methyl group off of O8 in Neu5Ac. However, the PmNAL wild-type structure with pyruvate bound in a Schiff base to Lys164 revealed flexibility in Phe189, which occupies two conformations. One conformation is similar to that seen with Neu5Ac bound, and one where Phe189 swings away from Neu5Ac binding site and packs in a more hydrophobic pocket, allowing for more space near O8 (Figure 2B). In the E. coli enzyme, the structurally equivalent residue is Tyr190, whose larger side chain would clash with the main chain near residue 164, preventing it to swing away from the Neu5Ac binding site (Figure 5). Additionally, the pocket that would accommodate Tyr190 in the E. coli enzyme is smaller due to the presence of Asp188, which is Ala187 in the PmNAL structure (Figure 5). All 21 known crystal structures of the EcNAL deposited in PDB were examined and Tyr190 is situated in the same position in all structures, laying adjacent to the Neu5Ac binding site as seen in Figure 5. Attempts to co-crystallize and soak in 8-O-methyl-Neu5Ac into the PmNAL K164A mutant proved unsuccessful yielding only weak-discontinuous electron density in the active site (data not shown), suggesting weaker binding, confirmed by the low catalytic efficiency (9).

Figure 5.

Comparison of active site pockets between PmNAL and EcNAL. Stereoview showing the wild-type PmNAL Schiff base complex (white), superposed onto the PmNAL K164 Neu5Ac complex (green) and the EcNAL (PDBID: 3LBC) structure (magenta). Phe189 in PmNAL takes on two conformations in the pyruvate-bound Schiff base, which might allow for modifications on O8 in the PmNAL enzyme. This residue is Tyrosine in the E. coli enzyme, which would likely not be able to accommodate similar multiple conformations due to the hydroxyl residue and the presence of Asn188 occupying the same pocket.

Discussion

The biological importance of sialic acids and the key role of sialic acids in pathogenic infections have made NAL an enzyme of high interest of study. Structural analysis of NAL helps to elucidate the details of the reaction pathway, which can provide insight into the mechanism of other members of the NAL subfamily, since the enzymes share a conserved α-keto acid binding motif suggestive of a common reaction chemistry for Schiff base formation. Previously reported structures of NAL have indeed demonstrated a Schiff base formation during aldol addition, where the Nζ of the strictly conserved residue Lys164 covalently binds to C2 of the pyruvate moiety, followed by a proposed substrate-assisted catalysis mediated by a conserved residue Tyr136 and an ordered water molecule to form sialic acid from pyruvate and ManNAc (16, 46). The NAL structures presented here are wild-type PmNAL and the enzyme trapped in a Schiff-base complex with pyruvate, without the need for reduction with sodium borohydride (21, 23). Additionally, three PmNAL K164A mutant structures were determined; mutant alone, mutant complexed with Neu5Ac, and complexed with Neu5Gc. This PmNAL enzyme is similar in structure and topology compared to previously reported aldolase crystal structures from other organisms. The PmNAL structures also show conserved residues in the active site and corroborate the proposed mechanism for catalysis.

Strictly conserved Lys164 forms a Schiff base intermediate to pyruvate in wild-type PmNAL, which is crucial for activity. The PmNAL K164A mutant is unable to form a Schiff base thus producing a non-functional enzyme, but has allowed the structure determination of two stable binary complexes with Neu5Ac and Neu5Gc bound to the active site. These structures provide the first glimpses of the NAL enzyme bound to native substrates providing high-resolution details of active site residues and their interaction with the substrate. The sialic acid binds to the C-terminal side of the TIM barrel in the open-chain ketone form of the sugar, in contrast to the more stable cyclic hemi-ketal form prevailing in solution.

Attempts to determine structures of both wild-type and K164A mutant with the six-carbon sugar substrate ManNAc proved unsuccessful (data not shown). However, this is consistent with the proposed mechanism where pyruvate was found to be a weak competitive inhibitor, but ManNAc was not an inhibitor (21, 23), suggesting ManNAc does not bind in the absence of pyruvate. Although PmNAL can catalyze the synthesis of 8-O-methyl-Neu5Ac from pyruvate and 5-O-methyl-ManNAc (37), numerous crystal soaks were not able to yield a structure with it bound in the active site. Yet, comparing PmNAL to the E. coli enzyme, which cannot catalyze the production of Neu5Ac8OMe, did suggest potential movement of Phe189 to accommodate the modified substrate. However, this remains to be proven.

One goal in sialic acid research is to utilize these NALs to chemoenzymatically synthesize novel sialic acid derivatives, which can be used to produce therapeutic compounds and biotechnology tools in carbohydrate studies. This work presents detailed structures of the PmNAL, which has proven to be a promising biocatalyst for synthesis (9). The structures presented here provide atomic details on substrate recognition and specificity. A greater understanding in the detailed binding specificity of NAL will help to further advance the production of novel sialic acids as well as sialic acid analogues that are difficult to produce by traditional chemical synthesis.

Structure-guided enzyme design of E. coli NAL has already shown modest success in increasing the enzyme specificity of the ManNAc analog 2,3-dihydroxy-4-oxo-N,N dipropylbutanamide (DHOB) to produce (5R,6R)-7-(dipropylamino)-4,5,6-trihydroxy-2,7-dioxoheptanoic acid (DPAH) (47). In this engineered enzyme, Williams et al., mutated Glu192 to smaller more hydrophobic residues, which improved catalytic efficiency for DHOB (compared to wild-type). Glu192 is equivalent to Glu191 of PmNAL, which hydrogen bonds to hydroxyl oxygens O8 and O9 at the end of Neu5Ac (Figure 3). In the DHOB analog, this end of the molecule contains the bulkier hydrophobic propyl groups, which would require an altered active site that contains a smaller more-hydrophobic side chain to accommodate the more hydrophobic substrate. Mutating Glu192 to smaller hydrophobic residues did not alter k_cat significantly but decreased K_m for DHOB (47) suggesting this residue is simply involved in substrate binding and does not influence catalysis.

These studies show promise in being able to engineer NAL to synthesize unique modified sialic acid derivatives. Given the required chemistry of pyruvate to form the Schiff base, and the carboxylate group of pyruvate that hydrogen bonds to main-chain amide nitrogens, the tolerance of the NAL enzyme in accepting modified pyruvate will be limited. However, producing sialic acids with modifications at carbons 5–9 from the corresponding ManNAc derivatives holds great promise. The structures presented here provide atomic resolution details on the conformation of the ManNAc moiety and the residues that interact with it. These structures offer the framework to guide future engineering experiments to synthesize unique sialic acid analogs. Such derivatives can allow a better understanding of sialic acid and its extensive functions in biology, and may also serve as potential therapeutic targets for fighting against cancer and infectious diseases.

Abbreviations

NAL

N-acetylneuraminate lyase or sialic acid aldolase

Neu5Ac

N-acetylneuraminic acid

Neu5Gc

N-glycolylneuraminic acid

RMSD

root mean squared deviation