Molecular Characterization of a Novel N-Acetylneuraminate Lyase from Lactobacillus plantarum WCFS1

Abstract

N-Acetylneuraminate lyases (NALs) or sialic acid aldolases catalyze the reversible aldol cleavage of N-acetylneuraminic acid (Neu5Ac) to form pyruvate and N-acetyl-d-mannosamine (ManNAc). In nature, N-acetylneuraminate lyase occurs mainly in pathogens. However, this paper describes how an N-acetylneuraminate lyase was cloned from the human gut commensal Lactobacillus plantarum WCFS1 (LpNAL), overexpressed, purified, and characterized for the first time. This novel enzyme, which reaches a high expression level (215 mg liter⁻¹ culture), shows similar catalytic efficiency to the best NALs previously described. This homotetrameric enzyme (132 kDa) also shows high stability and activity at alkaline pH (pH > 9) and good temperature stability (60 to 70°C), this last feature being further improved by the presence of stabilizing additives. These characteristics make LpNAL a promising biocatalyst. When its sequence was compared with that of other, related (real and putative) NALs described in the databases, it was seen that NAL enzymes could be divided into four structural groups and three subgroups. The relation of these subgroups with human and other mammalian NALs is also discussed.

Sialic acid aldolase or N-acetylneuraminate lyase (NAL; EC 4.1.3.3) is a class I aldolase that catalyzes the cleavage of N-acetylneuraminic acid (sialic acid; Neu5Ac) to pyruvate and N-acetyl-d-mannosamine with an equilibrium that favors Neu5Ac cleavage. The enzyme plays an important role in the regulation of sialic acid metabolism in bacteria (37, 41, 42). Neu5Ac lyase has been found in pathogenic as well as nonpathogenic bacteria (1) and in mammalian tissues. Neu5Ac lyase also catalyzes the reverse aldol condensation reaction and has been used in this way to synthesize sialic acid and some of its derivatives from pyruvate and N-acetyl-d-mannosamine (18, 44-46). Interest in this aspect of the enzyme's activity has increased, with growing appreciation of the role of sialic acid in controlling biomolecular interactions, particularly at the cell surface.

N-Acetylneuraminate lyase has previously been cloned from Escherichia coli (2, 32, 33), Clostridium perfringens A99 (41), Haemophilus influenzae (25), Trichomonas vaginalis (28), and Pasteurella multocida (24). In addition, X-ray structures of NAL and some mutants from E. coli and H. influenzae have also been solved (7, 11, 12, 19, 20, 23). All the N-acetylneuraminate lyases cloned up to the present moment come from human pathogens, because of the ability of these species to utilize the carbon sources present in the mucus-rich surfaces of the human body. Lactobacillus plantarum is a Gram-positive lactic acid bacterium, a commensal of the human gastrointestinal tract, which has a long history of safe use in many fermented sausages and fermented lactic products. Lactobacillus has also been used as a probiotic supplement or as an approved food additive that enjoys “generally recognized as safe” (GRAS) status (Title 21 of the Code of Federal Regulations [21 CFR] and the FDA Office of Premarket Approval list of microorganisms). These characteristics make the Lactobacillus plantarum genome ideal for obtaining new recombinant enzymes, such as N-acetylneuraminate lyase, which can be safely used in industrial production processes.

Based on the above, this paper describes the cloning, overexpression, and a detailed characterization of an N-acetylneuraminate lyase gene (nanA) from Lactobacillus plantarum WCFS1 (LpNAL). LpNAL was found to be remarkably stable at basic pHs compared with most studied NALs from Escherichia coli (EcNAL). Its topological and phylogenetic study also revealed the presence of four different bacterial NAL groups and several subgroups.

MATERIALS AND METHODS

Strains, plasmids, and chemicals.

Lactobacillus plantarum WCFS1 was from the NCIMB collection (number 8826). The pET 52 3C/ligation-independent cloning (LIC) kit was from Novagen (EMD Bioscience Inc., Madison, WI). The DNeasy tissue extraction kit, QIAquick PCR purification kit, and QIAprep spin miniprep kit were from Qiagen (Valencia, CA). Pfu DNA polymerase was from Stratagene (La Jolla, CA). N-Acetyl-d-mannosamine and N-acetylneuraminic acid were from Nacalai Tesque (Kyoto, Japan). Other reagents were from Sigma.

Cloning of the LpNAL gene.

The cloning and transformation techniques used were essentially those described by Sambrook et al. (36). Lactobacillus plantarum WCFS1 cells were collected from an overnight culture of MRS medium and provided as the source of the N-acetylneuraminic acid lyase gene, denoted nanA. The 879-bp gene was amplified by PCR using forward primer 5′-CAGGGACCCGGTATGAGTAAAAAACTATTGTATGCAGCCCAAATG-3′ and reverse primer 5′-GGCACCAGAGCGTTGTTTTTGAAGTATTTTTCGTAAATCGCCG-3′ (5′ LIC extension sites are italicized). The resulting PCR product was purified and treated with LIC-qualified T4 DNA polymerase (+dATP), annealed to the C-terminal 10× His-tagged pET52 3C/LIC vector and transformed into E. coli Novablue Giga competent cells. A selected clone harboring the correct sequence, denoted pET52-LpNAL, was transformed into E. coli BL21(DE3)pLys competent cells (Novagen).

Expression and purification.

The above-described E. coli cells harboring the recombinant plasmid pET52-LpNAL were grown for 4 h at 37°C in 400 ml of LB medium containing ampicillin-chloramphenicol (Amp-Chlor) before being transferred to a 5-liter fermentor (Sartorius), containing 4 liters of Terrific broth supplemented with antibiotics. This culture was allowed to grow for 3 h at 37°C and then induced by adding 1 mM isopropyl-β-d-thiogalactoside (IPTG) for 12 h at 30°C with constant stirring and oxygenation. The culture was diafiltered through a 500-kDa membrane (GE Healthcare, Uppsala, Sweden) and cleaned with 50 mM phosphate buffer (pH 8.0) containing 0.5 M NaCl. Cells were disrupted using a homogenizer (MiniZetaII, Netzsch), and the cell debris was harvested by centrifugation. The recovered supernatant (crude extract) was treated with 3 U/ml DNase I (Sigma) to remove nucleic acids and then centrifuged for 20 min at 6,000 × g.

The purification was performed in three steps, starting with tangential ultrafiltration with a 100-kDa cutoff membrane on a QuixStand system (GE Healthcare), followed by a heat shock treatment (65°C, 50 min), as previously described for other NALs (15). After centrifugation at 40,000 × g, the resulting supernatant was purified by Ni²⁺-chelating affinity chromatography (ÄKTA Prime Plus, GE Healthcare) onto a HisTrap phenyl FF column (GE Healthcare, Uppsala, Sweden). The bound enzyme was eluted with a linear imidazole gradient up to 250 mM in 50 mM phosphate buffer (pH 8.0). The fractions containing the aldolase activity were pooled, desalted, concentrated, and stored at −80°C.

Gel filtration (Superdex 200; Amersham Bioscience, Uppsala, Sweden) was used to confirm the homogeneity and the molecular weight of the purified enzyme. The column was equilibrated in 50 mM phosphate buffer (pH 7.0) containing 0.15 M NaCl. Superdex 200-purified LpNAL was used for the protein melting experiments. In addition, the molecular weight was determined using the high-pressure liquid chromatography-electrospray ion trap (HPLC-ESI) system (30). The protein concentration was determined using Bradford's reagent (Bio-Rad) and bovine serum albumin (BSA) as the standard.

Enzyme assay.

Aldolase cleavage was determined both spectrophotometrically and chromatographically (HPLC). The first method measured the decrease in absorbance at 340 nm corresponding to the oxidation of NADH produced by lactate dehydrogenase (LDH) when pyruvate appeared as a consequence of the hydrolysis of Neu5Ac into such compound and of ManNAc produced by LpNAL (1, 43). The standard reaction medium (1 ml) for the above-described assay, which was carried out in a Shimadzu UV-2401 PC spectrophotometer, contained 150 μM NADH, 0.5 U LDH, 10 mM Neu5Ac, and 1.5 μg of purified LpNAL in 20 mM phosphate buffer (pH 7.0). A control assay without Neu5Ac was also carried out to determine the presence of any other NADH-consuming enzymes. This hydrolytic activity was also measured from the increase of area of the ManNAc peak, in the same reaction conditions but without NADH and LDH, using an HPLC-evaporative light scattering detection (ELSD) II chromatograph (Shimadzu), an amino-UK column (Imtakt Co., Japan), and a mobile phase (58% acetonitrile, 42% 50 mM ammonium acetate) running at 0.4 ml/min at 60°C. Under these conditions, the retention times (R_T) for Neu5Ac and ManNAc were 10.3 and 4.2 min, respectively. One unit of activity was defined as the amount of enzyme required to cleave 1 μmol of Neu5Ac releasing 1 μmol of ManNAc in 1 min (HPLC) or consuming 1 μmol of NADH in 1 min at pH 7.0 and 37°C. The standard reaction medium for the synthetic reaction followed the above-described HPLC conditions and consisted of 100 mM ManNAc, 30 mM pyruvate, and 87 μg purified LpNAL in 20 mM phosphate buffer (pH 7.0). One enzymatic unit was defined as the amount of enzyme required to synthesize 1 μmol of Neu5Ac per minute under the above conditions. Kinetic parameters were obtained after three repeated experiments.

Stability assays.

A heat stability assay was carried out by incubating the enzyme at various temperatures (21) using a PCR thermocycler (TGradient, Biometra). Aliquots of 100 μl were taken at different times, cooled on ice, and spectrophotometrically measured in the standard reaction medium.

Melting curves to determine protein unfolding were obtained with the fluorescent dye Sypro orange (Molecular Probes). Sypro orange is a hydrophobic, environmentally sensitive fluorophore that is quenched in aqueous solutions but which binds to exposed hydrophobic surfaces of an unfolding protein, leading to a sharp increase in fluorescence emission as a function of temperature. Thermally induced unfolding is an irreversible process that follows a typical two-state model with a sharp transition between the folded and unfolded states, where T_m is defined as the midpoint of the protein-unfolding transition temperature. The T_m values obtained with this method correlate well with those obtained by other biophysical methods, such as circular dichroism (CD) or differential scanning calorimetry (DSC) (13). The assay was carried out in Milli-Q water or buffer containing 10× Sypro orange (emission at 530 nm and excitation at 490 nm), using a model 7500 fast reverse transcription (RT)-PCR machine (Applied Biosystems). The time/temperature control of the PCR machine was adapted from Malawaski et al. (27), performing 70 steps of 1 min each and raising the temperature by 1°C steps, from 20 to 90°C. Independent experiments were conducted with a minimum of three replicates per condition. This technique was also used to determine the pH stability dependence of temperature.

Structural, topological, and phylogenetic analysis.

BLAST searches were used to identify homologues of N-acetylneuraminate lyase (4). The sequences were aligned using ClustalW (40) and ESPript (16). Protein sequences were three-dimensionally (3-D) modeled with Geno3D (34) and ModWeb (35), and their topological drawings were obtained with TOPS (29). MEGA 4.0 was used to create and visualize phylogenetic trees (39). Bootstrap values were obtained after 1,000 generations.

RESULTS

Amino acid sequence comparison.

The deduced amino acid sequence of the L. plantarum N-acetylneuraminate lyase (LpNAL) showed significant identity with those of other bacterial species in the database. Sequence alignment indicates that LpNAL has 60%, 54%, 51%, and 38% amino acid sequence identity with the N-acetylneuraminate lyase from Clostridium perfringens, Trichomonas vaginalis, Haemophilus influenzae, and E. coli K-12 (GenBank accession numbers Q9S4K9, AAB42182, P44539, and AAC76257), respectively. LpNAL is more closely related to Haemophilus influenzae NAL (Protein Data Bank [PDB] code 1F5Z) than to the other crystallized NAL from E. coli (PDB code 1NAL).

In addition, sequence alignment (Fig. 1) revealed that LpNAL contained conserved residues forming the characteristic active site of the NAL subfamily (Fig. 1, filled triangles), the catalytic lysine at position 163 (K163), a tyrosine at position 135 (Y135), and the conserved specific substrate (Neu5Ac) binding motif, which includes the GxxGE motif and a group of amino acids (D189, E190, and S206) involved in the carbohydrate moiety binding. The GxxGE motif, situated between positions 45 and 49 (LpNAL numbering), is involved in the binding of the carboxylate group of the α-keto moiety of the substrate, and xx are usually S and/or T. These last two amino acids, together with Y135 and a water molecule, are involved in the hydrogen bond network with pyruvate (7, 23).

FIG. 1.

Multiple sequence alignment for L. plantarum WCFS1 (LpNAL) and related N-acetylneuraminate lyases. ESPript outputs (16) obtained with the sequences from the Swiss-Prot databank and aligned with ClustalW (40). Sequences are grouped according to similarity. The enzyme showed 51% sequence identity with NAL from H. influenzae (GI 162960935), 37% with NAL from E. coli (GI 49175990), 29% with NAL from S. pneumoniae (GI 169832377), and 31% with NAL from Yersinia pestis (GI 16120353). Residues strictly conserved across NAL enzymes have a dark background. Symbols above blocks of sequences represent the secondary structure, springs represent helices, and arrows represent β-strands. The residues forming the active site are indicated by small black triangles.

Cloning, overexpression, and purification of LpNAL.

The nanA gene from Lactobacillus plantarum WCFS1 was inserted into the pET52 3C/LIC vector, a vector which provides His₁₀-tagged recombinant protein, through a ligation-independent cloning system (see Materials and Methods). The DNA sequence of the cloned gene showed no mutations compared to the nanA gene sequence reported for L. plantarum genomic strain WCFS1 (GenBank accession no. NC_004567.1). The recombinant clone with the highest expression rate was induced with 1 mM IPTG in a 5-liter fermentor at 30°C for 12 h with vigorous shaking and oxygenation.

LpNAL was purified from E. coli cells by a three-step procedure consisting of a 100-kDa ultrafiltration step, 65°C heat shock step, and Ni²⁺ chelating affinity chromatography onto a HisTrap FF column. After these three steps (see Fig. S1 in the supplemental material), the enzyme was pure, as shown by SDS-PAGE (see Fig. S1, line 3), with an 11.8-fold purification and a 46% recovery. LpNAL showed a specific activity of 7.65 U/mg for the hydrolysis of Neu5Ac at 37°C and pH 7.0 (see Table S1 in the supplemental material). Up to 215 mg of Ni²⁺ column-purified LpNAL could be obtained from 1 liter of E. coli BL21(DE3)pLys culture, which is 2.2-fold higher than the best expression level reported for NAL (94 mg liter⁻¹ cell culture) (46). The molecular mass of purified protein was determined by gel filtration (132 kDa) and by HPLC/ESI/ion trap (37,317 Da) (data not shown), confirming the homotetrameric nature of LpNAL.

Biochemical characterization of recombinant LpNAL.

The activity of LpNAL was pH dependent, being active over a broad pH range, from pH 5.0 to 9.0, in both synthetic (Fig. 2 A, squares) and hydrolytic (Fig. 2A, circles) directions. The optimum pH of the enzyme in both directions was around pH 7 to 7.3, which is similar to values described for other NALs, such as those from E. coli K1 and C600 (1, 14), the native and the recombinant NAL from C. perfringens (22, 31), and the recombinant NAL from P. multocida (24).

FIG. 2.

Effect of pH toward LpNAL activity. (A) pH profile for the Neu5Ac synthesis (▪) and hydrolysis (•) activity of LpNAL determined by HPLC. Assay conditions at 37°C were 100 mM mannosamine, 30 mM pyruvate, and 87 μg enzyme for synthetic assay and 10 mM Neu5Ac and 1.5 μg for hydrolytic assay. The buffers (20 mM) used were sodium acetate (pH 5 to 5.5), sodium phosphate (pH 6 to 8), Tris-HCl (pH 8.5 to 9), and glycine (pH 9.5 to 11.5). (B) pH stability profile in the synthetic direction. Samples were analyzed after 15 days of incubation at 37°C in the different pH media to determine residual activity under the standard reaction, using 100 mM mannosamine and 30 mM pyruvate as substrates. Buffers used were the same as above.

Interestingly, LpNAL maintained 5 to 10% activity in the synthetic direction above pH 11.0. The synthetic activity of LpNAL above pH 10 is a remarkable feature of this NAL compared with those of E. coli (24) and P. multocida (24) and combats one of the industrially relevant drawbacks of this aldolase for the chemo-enzymatic synthesis of Neu5Ac from GlcNAc (39). The enzyme was also stable at basic pHs, where it maintained around 80% residual synthetic activity after 15 days of incubation (Fig. 2B). The residual activity after incubation at pH 11 was also close to 60%. This is quite remarkable stability compared with the stability of EcNAL, which at an alkaline pH showed a substantial loss of activity after 8 h (30% at pH 10.5 and 75% at pH 11.3) (9). This fact, together with the above-mentioned higher activity at alkaline pH, reinforces the possibility of using LpNAL to produce Neu5Ac at basic pHs, where the chemical isomerization of GlcNAc to ManNAc is more favorable (8, 9, 26).

The optimal temperature of LpNAL was measured in both synthetic and hydrolytic directions by HPLC, showing an optimal temperature of 60°C for synthetic activity and up to 70°C for the hydrolytic reaction (data not shown), which is close to that described for EcNAL (∼80°C) in the latter reaction (1). Both temperatures are far from the optimal growth temperature of Lactobacillus plantarum WCFS1, 30 to 37°C, at which temperature range kinetic parameters of NAL are usually determined (1, 24, 44).

The above results are in agreement with the following thermostability studies, suggesting that the LpNAL structure is very temperature stable, maintaining 80% of cleavage activity after 48 h at 60°C (data not shown). However, at higher temperatures (70°C), activity decreases to less than 10% in 8 h (Fig. 3 inset, open squares). Such thermal resistance is a common feature of NALs (1). This rapid inactivation of LpNAL at high temperatures was minimized up to 8-fold by the use of stabilizers, such as ammonium sulfate (Fig. 3, inset, triangles) and, specially, hydroxyectoine (Fig. 3, inset, circles).

FIG. 3.

Study of thermal stability of LpNAL. Melting temperature curves of purified enzyme (1 μg) were obtained in Milli-Q water (○), in 100 mM sodium phosphate buffer (□), and in the presence of additives such as 1 M hydroxyectoine (•), 1 M ammonium sulfate (▵), and 0.4 M ammonium sulfate (▴) in the above buffer. Assays were performed in a real-time PCR apparatus with 10× Sypro orange. Inset, effect of additives on the relative activity of LpNAL incubated at 70°C. Phosphate buffer (20 mM) without any of the following additives (□) and in the same buffer with the following additives: 1 M hydroxy-ectoine (•), 1 M ammonium sulfate (▵), and 0.4 M ammonium sulfate (▴). Aliquots were removed every hour and measured spectrophotometrically using the coupled assay at 37°C.

This thermal stability was further confirmed by a thermal melt assay, a T_m (±standard deviation) value of 71.5 ± 0.1°C being obtained for LpNAL in Milli-Q water (Fig. 3). The presence of a buffer solution (100 mM sodium phosphate, pH 7.0) stabilized the enzyme, raising the T_m value up to 76.5 ± 0.2°C. This effect was also evident using the above stabilizers (1 M hydroxyectoine and 1 M and 0.4 M ammonium sulfate). Hydroxyectoine raised the T_m of LpNAL up to 78.9 ± 0.1°C.

Thermal melt curves were also used to understand the pH stability described above (Fig. 2B). LpNAL showed great thermal stability from pH 6 to 10, with the T_m rising to ∼74 to 76.5°C, with fluctuations of between +5°C and −7°C compared with the T_m in Milli-Q water (Fig. 4). Thermal stability falls quickly above pH 10 (T_m at pH 10.5 was 64 ± 0.1°C). Figure 4 also shows, for the first time, that NAL stability toward pH was not only pH dependent but also buffer dependent, since different buffers, discrete buffers (Fig. 4, black bars), or continuous buffers (citric acid-HEPES-CHES) (Fig. 4, gray bars) had different stabilizing effects on the enzyme (Fig. 4). Glycine had the greatest stabilizing affect at alkaline pHs. This result is consistent with the data obtained in the conventional pH stability assay (Fig. 2B), confirming the potential of this enzyme in alkaliphilic biotechnological processes.

FIG. 4.

Effect of pH on melting temperatures of LpNAL. Assays were performed in a real-time PCR apparatus with 10× Sypro orange, 1 μg LpNAL, and two sets of 100 mM buffers. Discontinuous buffer solutions (black bars) were sodium acetate (pH 5 to 5.5), sodium phosphate (pH 6 to 7.5), Tris-HCl (pH 8 to 9), glycine-NaOH (pH 9.5 to 10), and CAPS (pH 10.5). A broad-pH-range citric acid-HEPES-CHES (100 mM) buffer system (gray bars, pH 5 to 10) was used. Differences in T_m were calculated by subtracting Milli-Q water T_m values.

Kinetic parameters.

Kinetic parameters were determined for both enzymatic cleavage and Neu5Ac synthesis (see Table S2 in the supplemental material). The K_m for Neu5Ac cleavage (1.8 ± 0.1 mM) was lower than the reported values for E. coli NAL (2.5 ± 0.3 mM) (24), P. multocida (4.9 ± 0.7 mM) (see Table S2) (24), and C. perfringens (2.8 to 3.2 mM) (22, 38). This LpNAL value, together with the k_cat/K_m value obtained (5.6 mM⁻¹ s⁻¹), points to a catalytic efficiency for this new recombinant enzyme toward hydrolysis similar to that of other NALs reported previously for the same substrate (from 3 to 4 mM⁻¹ s⁻¹) (see Table S2) (24).

On the synthetic side, LpNAL showed K_m values of 160 ± 5 mM and 19.9 ± 0.3 mM for ManNAc and pyruvate, respectively. These values are also slightly lower than those described for EcNAL (180 ± 10 mM and 22 ± 1 mM) and P. multocida NAL (220 ± 30 mM and 23 ± 1 mM), respectively (see Table S2 in the supplemental material) (24). As occurs in Neu5Ac hydrolysis, LpNAL was also efficient for the synthesis of Neu5Ac (see Table S2), showing k_cat/K_m values of 0.03 mM⁻¹ s⁻¹ for ManNAc and 0.11 mM⁻¹ s⁻¹ for pyruvate. A similar catalytic efficiency has been reported for other NALs (see Table S2) (24).

Structural and topological analysis.

The 3-D structure of crystallized H. influenzae NAL (HiNAL) (PDB code 1F5Z; 51% identity) (7) was selected as a template by Geno3D (34) to create an LpNAL model (see Fig. S2A in the supplemental material). The modeled enzyme shows the typical (β/α)₈-barrel (see Fig. S2B), as is usual in the NAL subfamily, which includes NAL, dihydrodipicolinate synthetase (DHDPS), d-5-keto-4-deoxyglucarate dehydratase (KDGDH), trans-o-hydroxybenzylidene-pyruvate hydratase-aldolase (HBPHA), and other related enzymes (7). These enzymes share a structural framework but catalyze different reactions in separate biochemical pathways. Within the subfamily, NAL catalyzes the aldol cleavage of Neu5Ac to form ManNAc and pyruvate via a Schiff base intermediate, as mentioned above.

DISCUSSION

This paper describes the expression of a recombinant Lactobacillus plantarum NAL which shows good synthetic activity not only at its optimum pH but also at basic pHs. This, together with its thermostability and pH stability at such pHs, underlines the potential biotechnological application of this new NAL for the synthesis of sialic acid and its derivatives, as has been demonstrated for other NALs (10, 15, 17).

In addition, the three-dimensional structure model created for LpNAL (see Fig. S2 in the supplemental material) used HiNAL as a template due to the higher sequence identity (Fig. 1). In order to confirm this relationship, a phylogenetic analysis was expanded to other putative NAL sequences found in the databases (Fig. 5). Distribution of the putative NAL protein sequence is curiously limited among bacteria. Interestingly, apart from a few aquatic bacteria (Photobacterium profundum, Pseudoalteromonas haloplanktis, Shewanella pealeana, Psychromonas, and Vibrio), the NAL gene is present only in commensal or pathogenic bacteria related with humans (3). Based on this analysis, four structural groups of NALs were clear (Fig. 5). The first group included enzymes from Gram-negative genera, Haemophilus, Actinobacillus, and Pasteurella, as well as Gram-positive genera, Lactobacillus, Clostridium, Staphylococcus, and Mycoplasma, and also fusobacteria (Fig. 5; see also Fig. S3 in the supplemental material). Group 2 (Fig. 5) includes NAL sequences from E. coli, Shigella spp., and Salmonella enterica, which all are human gastrointestinal pathogens and have high sequence identities (∼90 to 100%) (see Fig. S4 in the supplemental material). Streptococcus genera, some species of Lactobacillus (L. sakei), Clostridium (C. bolteae, C. hylemonae), and other human gut commensal bacteria, such as Ruminococcus gnavus and Dorea formicigenerans, make up group 3 (Fig. 5; see also Fig. S5 in the supplemental material). Finally, group 4 is formed by NAL proteins with low sequence identity from members of the family Vibrionaceae, Shewanellaceae, Psychromonadaceae, and Pseudoalteromonadaceae and some species of genera Bacteroides, Parabacteroides, and Capnocytophaga (Fig. 5; see also Fig S6, S7, and S9 in the supplemental material). These four groups agree with the phylogenetic analysis of NAL sequences (3), which divides NALs into four lineages (named I, II, III, and IV), which correspond to groups 1 to 4 described in this paper.

FIG. 5.

Phylogenetic analysis of the bacterial NAL subfamily. Bacterial NALs used in this study (see text for details) are phylogenetically divided into four groups. E. coli dihydrodipicolinate synthetase (PDB code 1DHP) was used as the outgroup. Only bootstrap values below 95 are shown. The phylogenetic tree was constructed by Mega 4.0 (39). Details of ESPript alignment outputs (16) are shown on the right side of the figure. An adapted drawing of the NAL active site (7) is shown in the upper left corner of the figure in order to clarify the meaning of the blocks (LpNAL numbering). H, Haemophilus; P, Pasteurella; C, Clostridium; L, Lactobacillus; St, Staphylococcus; E, Escherichia; S (flexneri, sonnei, boydii), Shigella; Sa, Salmonella; S (pneumoniae, gordonii, sanguinis, zooepidemicus, pyogenes, dysgalactiae), Streptococcus; S pseudotub, Streptococcus pseudotuberculosis; Ca, Capnocytophaga; B, Bacteroides; Ho, Homo; Su, Sus; M, Mus; V, Vibrio; Y, Yersinia; G, group; Sg, subgroup.

However, taking into account this analysis, it was not possible to clearly understand the heterogeneity of group 4. In order to clarify this, the phylogenetic relationships between the NAL sequences used above and the active center residues (Fig. 5, top left) were also analyzed. Three blocks of the active center are described: carboxylate-binding zone (sequence GxxGE), sugar-binding zone (sequence GxDE), and the aldol-cleavage zone (sequences KxT/Sx and xxG/ST). Groups 1 and 2 are closely related in the active center sequence, with just small changes (1 or 2) in the sequences considered, in agreement with the phylogenetic tree (Fig. 5, right; see also Fig. S3 and S4 in the supplemental material). However, although related to group 1 and 2, group 3 has its own fully conserved active center signature (Fig. 5, right; see also Fig. S5 in the supplemental material). Finally, in group 4, the diversity is more evident in the active center sequence (Fig. 5, right), in which four clear subgroups can be described, including subgroup 4.1, which includes Capnocytophaga spp. (see Fig. S6 in the supplemental material); subgroup 4.2, which includes Bacteroides spp. (see Fig. S7 in the supplemental material); and subgroup 4.4 (see Fig. S9 in the supplemental material), which includes Vibrio spp. and Yersinia spp. When animal NAL sequences were considered (see Fig. S8 in the supplemental material), they clustered together within group 4 (Fig. 5), forming subgroup 4.3 but not a different clade (3). Subgroup 4.4 is closely related to human NAL, as revealed by the similarities between both active center sequences (GTTGE/KFxx/GVDE/VGST), indicating lateral gene transfer between bacteria and humans. This was previously described for human NAL (5), but no direction of the transfer was assigned. This new phylogenetic and sequence analysis supports the hypothesis of transfer of NAL from group 4 bacteria to mammals.

In conclusion, the N-acetylneuraminate lyase from Lactobacillus plantarum WCFS1 was cloned in E. coli, overexpressed, and purified to obtain a stable 132-kDa homotetrameric protein. The enzyme was able to synthesize and cleave Neu5Ac with catalytic efficiencies similar to those of the best previously described NALs. This enzyme also shows good stability and activity under alkaline pH conditions and a wide range of temperatures (up to 70°C). In addition to these biotechnologically useful characteristics, the phylogenetic analysis of LpNAL, and other related bacterial NAL from the databases, led to a new classification of bacterial NALs into four groups and three subgroups (4.1, 4.2, and 4.4), as has been carried out for other enzymes, such us lipases/esterases (6). Among these groups, subgroup 4.4 was related with some mammalian NALs (subgroup 4.3), including that of humans.