Structural-based mutational analysis of d-aminoacylase from Alcaligenes faecalis DA1

Abstract

d-Aminoacylase is an attractive candidate for commercial production of d-amino acids through its catalysis in the zinc-assistant hydrolysis of N-acyl-d-amino acids. We report here the cloning, expression, and structural-based mutation of the d-aminoacylase from Alcaligenes faecalis DA1. A 1,007-bp PCR product amplified with degenerate primers, was used to isolate a 4-kb genomic fragment, encoding a 484-residue d-aminoacylase. The enzyme amino-terminal segment shared significant homology within a variety of enzymes including urease. The structural fold was predicted by 3D-PSSM to be similar to urease and dihydroorotase, which have grouped into a novel α/β-barrel amidohydrolase superfamily with a virtually indistinguishable binuclear metal centers containing six ligands, four histidines, one aspartate, and one carboxylated lysine. Three histidines, His-67, His-69, and His-250, putative metal ligands in d-aminoacylase, have been mutated previously, the remaining histidine (His-220) and aspartate (Asp-366) Asp-65, and four cysteines were then characterized. Substitution of Asp-65, Cys-96, His-220, and Asp-366 with alanine abolished the enzyme activity. The H220A mutant bound approximately half the normal complement of zinc ion as did H250N. However, the C96A mutant showed little zinc-binding ability, revealing that Cys-96 may replace the carboxylated lysine to serve as a bridging ligand. According to the urease structure, the conserved amino-terminal segment including Asp-65 may be responsible for structural stabilization.

N-acyl-d-Amino acid amidohydrolases (d-aminoacylases) catalyze the hydrolysis of N-acyl-d-amino acids to produce the corresponding d-amino acids, which are intermediates in the preparation of pesticides, bioactive peptides, and antibiotics. Recently, various d-amino acids have been found in bacteria, plants, and mammals, and their physiological functions have received increased attention. Production of l-amino acids by optical resolution using l-aminoacylase immobilized on DEAE-Sephadex has been used in industry. Therefore, production of d-amino acids by d-aminoacylase has commercial importance.

To search for new d-aminoacylases suitable for d-amino acid production, microorganisms in various soils were screened. Two d-aminoacylases from Alcaligenes denitrificans DA181 (Tsai et al. 1988; Yang et al. 1992) and A. faecalis DA1 (Yang et al. 1991; Tsai et al. 1992) were isolated and showed higher specific activity and better stereospecificity than those from Pseudomonas and Streptomyces (Sugie and Suzuki 1978; Kubo et al. 1980). Two other d-aminoacylase-producing strains, A. denitrificans MI-4 (Sakai et al. 1991), and A. xylosoxydans A-6 (Moriguchi et al. 1993), were subsequently screened. Three genes encoding for the A-6 d-aminoacylases have been cloned and expressed in Escherichia coli (Wakayama et al. 1995a,b,c). These A-6 d-aminoacylases share 45%–50% sequence identity to one another, but no significant homology with l-aminoacylase.

The Alcaligenes d-aminoacylases have been shown to contain two zinc ions per molecule (Yang et al. 1992; Wakayama et al. 1995b). In zinc metalloenzymes, His, Cys, and Asp are the ligands most frequently involved in metal binding. To address the zinc ligands, Wakayama and co-workers mutated four of the eight conserved histidines in A-6 d-aminoacylase, and demonstrated that His-67 and His-250 were essential for the enzyme activity (Wakayama et al. 2000). Both H69I and H69N mutants were overproduced in the insoluble fraction. The decreased metal binding of H250N mutant suggested that His-250 is essential for catalysis through zinc binding. The similar circular dichroism spectra for the wild-type and mutant enzymes indicated that no significant conformational change was caused by mutations. However, other residues involved in the metal ligation and the enzyme catalysis remain unclear.

Because of more thermostability and higher affinity to DEAE resins than the DA181 d-aminoacylase, the DA1 d-aminoacylase is more suitable for optical resolution of N-acyl-dl-amino acids. To elucidate the structure-function relationship of the DA1 d-aminoacylase, the gene was first cloned, and expressed in E. coli. Some putative active-site residues were then identified based on the computer-aided structural prediction. To ascertain the structural prediction, site-directed mutagenesis was performed, and each mutant protein was purified and characterized.

Results and Discussion

Cloning of DA1 d-aminoacylase

A 1,007-base PCR product was first amplified from the chromosomal DNA of A. faecalis DA1, and used as a probe to isolate a 4-kb genomic DNA. After subcloning, a 1,957-base DNA fragment, encoding a 484-residue d-aminoacylase in an open reading frame (ORF), was obtained. A probable ribosome-binding sequence, GGAG, was present seven bases upstream of the putative start codon. However, sequences similar to the E. coli –35 sequence (TTGACA) and the –10 sequence (TATAAT) were not found. The G+C content of the ORF was 67.6%, in agreement with high G+C contents of the Alcaligenes genomes. The nucleotide sequence has been submitted to the GenBank database under the accession number AF332548.

Sequence similarity search showed that the DA1 d-aminoacylase shares high sequence identity to d-aminoacylase (85%), N-acyl-d-aspartate amidohydrolase (56%), and N-acyl-d-glutamate amidohydrolase (47%) from A. xylosoxydans A-6. None of these enzymes have been extensively studied.

Structural prediction

To identify the potential active site residues in the absence of a crystal structure for d-aminoacylase, sequence homology was sought through the structural databases revealing that residues 8–96 in DA1 d-aminoacylase displayed putative homology to residues 66–159 of urease. Subsequent screening of the urease demonstrated that this segment was significantly conserved within a variety of enzymes including dihydroorotase, allantoinase, hydantoinase, adenine deaminase, aryldialkylphosphatase, and N-isopropylammelide isopropyl amidohydrolase (Fig. 1 ▶).

Fig. 1.

The conserved amino-terminal segment in a variety of enzymes. Identical amino acids are shown in white with the black background, whereas conservative amino acids are shaded gray. The accession numbers are indicated for each sequence. (URE1) Urease from K. pneumoniae; (ADEC) adenine deaminase from S. meliloti; (PYRC) dihydroorotase from M. loti; (ADP) aryldialkylphosphatase from Nocardia; (HYDA) d-hydantoinase from A. tumefaciens; (ALN) allantoinase from E. coli; (CREC) creatinine deaminase from M. loti; (CODA) CodA1 cytosine deaminase from S. meliloti; (ATZC) N-isopropylammelide isopropyl amidohydrolase from Pseudomonas; (ATZA) atrazine chlorohydrolase from Pseudomonas; (HUTI) imidazolone-5-propionate hydrolase from Halobacterium; (FWDA) formylmethanofuran dehydrogenase from M. jannaschii.

The structural fold of DA1 d-aminoacylase was then predicted by the program 3D-PSSM (Kelley et al. 2000). Several structural models with more than 95% certainty were generated on the basis of the crystal structures of Klebsiella aerogenes urease (Jabri et al. 1995) and E. coli dihydroorotase (Thoden et al. 2001), containing a common TIM (β/α)₈ barrel. Structural similarity of both enzymes was subsequently verified by the program VAST (Gibrat et al. 1996), revealing many similar triosephosphate isomerase (TIM)-fold proteins. Because the d-aminoacylases were shown to contain two zinc ions (Yang et al. 1992; Wakayama et al. 2000), only the proteins containing two metal ions were selected for detailed structural comparison by using the program INSIGHT II (Molecular Simulation Inc.). In spite of the apparent lack of sequence homology, the eight β-strands of the TIM domains of urease, dihydroorotase, phosphotriesterase (Benning et al. 2001), and phosphotriesterase homology protein (Buchbinder et al. 1998) are superimposable, with a virtually indistinguishable binuclear metal centers (Fig. 2 ▶). The conversed metal ligands, four histidines, one aspartate, and one carboxylated lysine (or one glutamate), are virtually identical in the spatial positions. For example, superposition of urease and dihydroorotase showed that the backbones of 56 residues (224 atoms) in the eight β-strands overlay with 1.24 Å root-mean-square (rms) deviation, whereas the 124 nonhydrogen atoms of the two metal ions and the six ligands with 0.96 Å rms deviation. The high degree of global structure and the metal center similarity of phosphotriesterase, adenosine deaminase, and urease have been noted once these structures were solved (Jabri et al. 1995). Holm and Sander (1997) subsequently compared these three protein structures and discovered a novel α/β-barrel amidohydrolase superfamily, in which the metal ligands, four histidines and one aspartate, are strictly conserved.

Fig. 2.

Detection of conservation of the binuclear metal centers by structural alignment. (A) A close-up view from the top of the α/β barrel. The backbone structures of K. aerogenes urease (1FWD), E. coli dihydroorotase (1J79), and E. coli phosphotriesterase homology protein (PHP, 1BF6) are shown in pink, gray, and blue, respectively, with the metal ligands displayed as stick representations in red, green, and yellow. The eight β-strands are superimposable, with the conserved metal ligands are virtually identical in the spatial positions. (B) The detailed comparison of the coordination geometries for the binuclear metal centers with ligated residues in which the residue numbering is labeled in the same color for each protein. Some PHPs such as E. coli PHP uses a glutamate instead of a carboxylated lysine residue for bridging both metal ions.

The conserved metal ligands in DA1 d-aminoacylase

According to the structural prediction, the potential conserved metal ligands in DA1 d-aminoacylase are His-67, His-69, His-220, His-250, and Asp-366. Three histidines in A-6 d-aminoacylase have been mutated (Wakayama et al. 2000). In this study, His-220 and Asp-366 in DA1 d-aminoacylase were changed into alanine, resulting in the complete loss of enzyme activity (Table 1). Mutation of the metal ligands in hamster dihydroorotase H15G, H17G, H134A, H158A, D230G, and D230N also resulted in no detectable activity (Williams et al. 1995). Together with the previous mutation studies, we hypothesized that His-67, His-69, His-220, His-250, and Asp-366 are the conserved metal ligands in the DA1 d-aminoacylase. Replacement of these ligands may destroy the metal-binding sites, which are essential for the enzyme activity, as shown by the crystal structure of the urease H134A mutant (Park et al. 1996).

Table 1.

Characterization of DA1 d-aminoacylase mutants

	kcat (min−1)	Km (mM)	kcat/Km (min−1 mM)	Zinc content (g • atom/mol)
w.t.	2.76 × 105	0.63	4.38 × 105	1.43 ± 0.09
C144A	2.85 × 105	0.82	3.46 × 105	1.43 ± 0.11
C207A	1.94 × 105	0.53	3.69 × 105	1.44 ± 0.12
C308A	1.94 × 105	0.95	2.04 × 105	1.43 ± 0.10
C96A	n.d.	n.d.	n.d.	0.23 ± 0.03
D65A	n.d.	n.d.	n.d.	1.27 ± 0.08
H220A	n.d.	n.d.	n.d.	0.69 ± 0.07
D366A	n.d.	n.d.	n.d.	1.42 ± 0.09

n.d. = the enzyme activity was too low to determine.

The cysteine mutants

The novel α/β-barrel amidohydrolase superfamily has been divided into two subsets (Holm and Sander 1997). To date the best-characterized members of the superfamily include urease, dihydroorotase, and phosphotriesterase with a binuclear center, and adenosine deaminase and cytosine deaminase with a mononuclear center. The critical structural element for the binuclear subset is a carboxylated lysine serving as a bridging ligand. However, there is no lysine residue between residues 84–251 in the DA1 d-aminoacylase. Because cysteine residues are the ligand frequently involved in zinc binding, they were analyzed by several methods described below.

As shown in Table 2, the enzyme activity was relatively sensitive to 4-(chloromercuri) benzoic acid [PCMB] and N-ethylmaleimide (NEM), whereas 5,5′-dithio-bis (2-nitrobenzoic acid) [DTNB] caused moderate inhibition, and E-64 had no significant effect. In addition, the enzyme inactivation by NEM was enhanced in the presence of 10 mM EDTA, implying that zinc binding could protect the NEM inactivation. The sensitivity of DA1 d-aminoacylase activity to PCMB or NEM treatment suggested that cysteinyl residues might play a role in the enzyme activity, probably involved in the metal coordination.

Table 2.

Effects of thiol-specific reagents on enzyme activity

	Residual activity
Reagents	1 mM	10 mM
No addition	100	100
PCMB	23.9	3.6
DTNB	64.8	61.1
E-64	96.4	97.6
NEM	73.7	24.8a
NEM +10 mMEDTAb	44.7	11.5a

^a Only 5 mM NEM was added.

^b Incubation with 10 mM EDTA alone for 20 min did not affect the enzyme activity.

To investigate the roles of the cysteines in enzyme activity, mutants were constructed by site-directed mutagenesis in which the four residues were substituted by either Ala, Ser, or Thr. Only the mutant carrying the Cys-96 substitution was catalytically inactive, and the other mutants retained the similar specific activity as the wild-type enzyme (Table 1).

All four SH groups in the wild-type enzyme under the denatured condition could be titrated by DTNB, whereas only two SH groups were titratable under the native condition (Table 3), suggesting that none of the four cysteines is engaged in disulfide bridges. Neither Cys-96 nor Cys-308 appear to be solvent-accessible, perhaps due to location in the interior or ligation to metal ions. In addition, C144A, C207A, and C308A mutants contain similar zinc contents as the wild-type enzyme. However, the inactive C96A mutant has little zinc-binding ability (Table 1). The predicted structural fold revealed that Cys-96 is located at the loop between β2 and α2 in the TIM barrel and may be able to coordinate to both divalent cations. Therefore, it seems plausible that Cys-96 replaces the carboxylated lysine for the metal ligation and that the shift in the sequence position might compensate for the longer length of the carboxylated side chain. This is the first example of a cysteine residue that coordinates to zinc ions in this superfamily. Therefore, DA1 d-aminoacylase may define a novel subset of the amidohydrolase superfamily.

Table 3.

Determination of free SH groups in DA1 D-aminoacylase

	Free SH group/protein
Enzyme	In 2% SDS	Native protein
w.t.	3.9 ± 0.3	2.2 ± 0.1
C96A	2.9 ± 0.2	2.5 ± 0.2
C144A	2.9 ± 0.1	1.0 ± 0.1
C207A	3.1 ± 0.2	1.1 ± 0.1
C308A	3.4 ± 0.1	2.1 ± 0.1

The circular dichroism spectra of four mutants (C96A, C207A, H220A, and D366A) overlaid nearly with that of wild-type enzyme, as did those for A-6 d-aminoacylase (Wakayama et al. 2000). The similar circular dichroism spectra indicated that the mutations do not affect the enzyme global structure. Combining with the results above, we proposed that the protein ligands are His-67, His-69, Cys-96, and Asp-366 for the first zinc ion, whereas Cys-96, His-220, and His-250 for the second zinc (Fig. 3 ▶). The d-aminoacylases from A. denitrificans DA181 and A. xylosoxydans A-6 were shown to contain two zinc ions per enzyme molecule (Yang et al. 1992; Wakayama et al. 2000). However, the zinc content of the purified recombinant DA1 d-aminoacylase was measured to be between 1.3 and 1.5 g • atom per mole of enzyme (Table 2). The lower zinc content may be due to some metal loss during the protein purification. Cys-96 may contribute the most toward the interactions with zinc ions among the ligands because the mutant C96A shows the least zinc-binding ability. The zinc content of the H220A mutant was nearly half of that of wild-type enzyme, as H250N in the A-6 d-aminoacylase, suggesting that His-220 and His-250 contribute more interactions with the second metal ion because of only three protein ligands (Fig. 3 ▶). On the other hand, neither H67N mutant in A-6 d-aminoacylase nor D366A mutant in DA1 d-aminoacylase lost the first zinc binding, perhaps due to more ligands. In addition to ligate the metal ion, the conserved aspartate Asp-366 is believed to serve as a general base in the enzyme catalysis (Benini et al. 1999; Benning et al. 2001; Thoden et al. 2001).

Fig. 3.

The proposed bi-zinc center in the DA1 d-aminoacylases on the basis of structural prediction and mutational studies.

The conserved amino-terminal segment in a variety of enzymes

Sequence similarity search revealed that the amino-terminal segment of DA1 d-aminoacylase displayed significant conservation within a variety of enzymes (Fig. 1 ▶). All these enzymes may belong to the α/β-barrel amidohydrolase superfamily because of containing the four conserved histidines and one aspartate. According to the crystal structure of K. aerogenes urease, most of the conserved residues in the amino-terminal segment are involved in the structural stabilization. In particular, the carboxylate group of Asp-132 in urease (Asp-65 in d-aminoacylase), near the metal ligand His-134 (His-67 in d-aminoacylase), interacts with the main chain amine groups of Thr-133, His-134, and Ser-359 (next to the conserved Asp-360 (Asp-366 in d-aminoacylase)), and the side chain of Ser-359. Replacement of this conserved aspartate residue, D65A in DA1 d-aminoacylase, and D13N in hamster dihydroorotase (Williams et al. 1995), resulted in an inactive enzyme, perhaps due to structural destabilization of the metal-binding sites and the proper orientation of the general base aspartate.

In summary, combination of structural comparison and mutational studies, we propose a binuclear zinc center for the DA1 d-aminoacylase, which may define a novel subset of the α/β-barrel amidohydrolase superfamily. Determination of the protein structure by x-ray diffraction methods is under investigation to verify our structural prediction and approach the zinc-assistant catalysis.

Materials and methods

Cloning of DA1 d-aminoacylase

Isolated d-aminoacylase (300 μg) from A. faecalis DA1 was digested by 5 μg of clostripain (Sigma) in 0.5 mL of 100 mM Tris-HCl buffer (pH 7.6) for 4 h at 37°C. The digest was injected into a C-18 reverse phase-HPLC column (4.6 by 250 mm, Vydac) and eluted with a linear gradient of acetonitrile with 0.1% trifluoroacetic acid. One of the isolated peptides and the intact protein were subjected to amino-terminal sequencing on a protein sequencer (PE Biosystems).

Two primers, 5′-CARCCNGAYGCNACNCCNTTYGA (R: A+G, N: A+T+G+C, Y: T+C), and 5′-ARYTCNGGNACNACRT CRTAYTT, were then designed based on partial protein sequences. A DNA fragment was first amplified out from the A. faecalis DA1 chromosome, and used as a probe to screen the DNA library, which was constructed by ligating the Sau3AI-digested DNA fragments into the vector pGEM-7Zf(−). E. coli XL-1 Blue competent cells were transformed, and selected on an agar plate containing ampicillin, IPTG, and X-gal. The plasmid isolated from a positive transformant was sequenced on a DNA sequencer (PE Biosystems).

Another two primers 5′-AGGGATCCCATGTCCCAGCCCGA CGCC and 5′-TCTAAGCTTTCAGGCTCCGGCCCGGTTGAG GACGCGGCC, were designed based on the cDNA sequence to amplify the gene. The PCR products were cloned into the pQE30 (Qiagen) vector, and the resulting plasmid (pQE-DA1) was confirmed and transformed into E. coli M15.

Protein purification

The DA1 d-aminoacylase gene was expressed in E. coli grown at 37°C on LB media. One-liter cultures of transformed bacteria were grown for 4–5 h before induction by addition of 2 mL of 0.5 M IPTG. Cultures were then grown for another 5 h at 37°C before harvest. Cell pellets were resuspended in 100 mL of cold sodium phosphate buffer (50 mM, pH 7.8) with 300 mM NaCl and lysed by sonication. After removal of cellular debris by centrifugation at 27,000 g at 4°C for 30 min, the supernatant was applied to a Ni-NTA (Qiagen) column (2.6 by 4.5 cm), pre-equilibrated with buffer A (50 mM sodium phosphate buffer and 300 mM NaCl at pH 7.0). After washing with 100 mL of 75 mM imidazole in buffer A, the protein was eluted with a 75-mL linear gradient of 75–500 mM imidazole in buffer A. The protein fractions were pooled, and concentrated to 8 mL by ultrafiltration, then applied to a Fractogel HW-50S column (2.6 by 85 cm) eluted with buffer B (50 mM Tris-HCl buffer at pH 7.8). The enzyme fractions were collected and loaded to a HiTrap Q (Pharmacia Biotech) column (1.6 by 2.5 cm), pre-equilibrated with buffer B. After washing with 15 mL of buffer B containing 150 mM NaCl, the enzyme was eluted using a 70-mL linear gradient of 150–250 mM NaCl in buffer B. The protein was dialyzed against buffer B, and the concentrations were determined by a convention protein-dye-binding method using bovine serum albumin as the standard.

Site-directed mutagenesis and characterization of the mutants

Site-directed mutagenesis was carried out by the PCR overlap extension technique (Higuchi et al. 1988) using the mismatching synthetic oligonucleotide primers. The fragments, containing the gene with the desired mutations, were subsequently digested with BamHI and HindIII and ligated into the BamHI/HindIII-treated pQE vector. The mutations were confirmed by DNA sequencing.

d-Aminoacylase activity was measured using N-acetyl-d-methionine as substrate coupled to a d-amino acid oxidase enzyme assay as previously described (Tsai et al. 1988). For the inhibition assay, each thiol-specific reagent was incubated with the enzyme for 20 min at 30°C in buffer B; the residual activity was then measured. The metal contents were detected by a Perkin Elmer AAnalyst 800 atomic absorption spectrophotometer in flame mode. Before metal determination, the enzyme samples were dialyzed against a metal-free buffer. All values were the means ± S.E.M for three experiments.

The 5,5′-dithio-bis(2-nitrobenzoic acid) (DTNB) titration of cysteinyl residues was performed in 100 mM phosphate buffer (pH 7.2) containing 8 μM enzyme and 2 mM DTNB. The change in absorbance at 412 nm was recorded and the total free SH content was calculated assuming a molar absorption coefficient (ɛ_m) of 14,150 under native conditions and of 13,700 in the presence of 2% SDS (Riddles et al. 1983). All values were the means ± S.E.M for three experiments.