INTRODUCTION

D-amino acids are important components of eubacteria, as they constitute parts of the fundamental tetrapeptide chain in murein of the cell wall (1). D-Alanine is formed from L-alanine by alanine racemase (EC) which is quite widespread in eubacteria (2). Other D-amino acids, such as D-aspartate or D-glutamate, are formed either by transamination from D-alanine to the appropriate -keto acids or by racemization of the enantiomeric L-amino acids (3, 4, 5, 6, 7). Alanine racemase requires pyridoxal 5-phosphate as a cofactor. Racemization proceeds through the formation of an aldimine Schiff base between the substrate alanine and pyridoxal 5-phosphate (8). However, several other amino acid racemases, including those for glutamate and aspartate, are cofactor independent (6, 7, 9, 10, 11, 12). In these racemases, thiol groups of cysteine residues serve as catalytic bases in the proton transfer reactions.

There are few reports describing endogenous D-amino acids in eukaryotes or archaea. Although it is proposed that the biosynthesis of D-serine in worms and insects involves amino acid racemases (13), these enzymes have not been found in invertebrates or mammals. Several groups have identified D-serine, D-aspartate, and D-glutamate in mammals, including humans (14, 15, 16). The biosynthetic pathways of these D-amino acids remain unknown.

Murein is the only cell wall polymer that forms rigid cell walls in eubacteria. However, archaea have a variety of cell wall and cell envelope polymers (17, 18). Many archaea (including all crenarchaeotes, euryarchaeotes, halophiles, and methanogens) have an outer envelope (or S-layer) composed of hexagonally or tetragonally arranged proteins or glycoproteins that are easily disintegrated by mechanical shearing or detergents. In methanogens, the structure of this polymer is similar to that of murein, and it is called ``pseudomurein.'' Pseudomurein differs from murein in several respects. No D-amino acids are present in pseudomurein. Thus, neither D-amino acids nor amino acid racemases have been found in archaea.

Hyperthermophiles which grow optimally at over 80 °C have been isolated (19, 20). Most of them are archaea. Since the hyperthermophiles are located in deep branches close to the root of the phylogenic tree based on 16/18 S rRNA (21), it is suggested that hyperthermophiles retain some of the physiological or biochemical features of early life forms. On the other hand, they produce highly thermostable enzymes which are industrially useful.

Here, we describe the cloning of the gene encoding aspartate racemase from the hyperthermophilic archaeum, Desulfurococcus strain SY (22), which was inadvertently cloned during the search for homologues of hsp70 stress proteins in hyperthermophilic archaea. This is the first archaeal amino acid racemase described to date.

EXPERIMENTAL PROCEDURES

Restriction and modification enzymes were the products of Takara Shuzo Co., Ltd. Taq DNA polymerase and other reagents for the polymerase chain reaction (PCR)¹ were also obtained from Takara Shuzo Co., Ltd. Nylon membranes and the ECL random prime labeling and detection system were from Amersham. D-Amino acid oxidase from pig kidney was the product of Biozyme Laboratories, Ltd. Horseradish peroxidase was purchased from Wako Pure Chemical Industries, Ltd. All other chemicals were of analytical grade.

All plasmids were propagated in Escherichia coli DH5 (Life Technologies, Inc.). The recombinant aspartate racemase was expressed in E. coli BL21(DE3) (23). E. coli was cultured aerobically at 37 °C in 2 × YT media supplemented with ampicillin (50 µg/ml).

The hyperthermophilic archaeum, Desulfurococcus strain SY (22), was cultured at 90 °C in 5-liter glass bottles as described (43). Total DNA was prepared as described by Kagawa et al. (24).

PCR was performed using total DNAs from several hyperthermophilic archaea. Oligonucleotide primers (5-CCI-GA(CT)-GA(AG)-GTI-GTI-GC-3 and 5-TC-IGC-(AG)TC-(CT)TT-IAC-CAT-3) were designed from the consensus sequences of hsp70s. An aliquot of 100 ng of D. SY total DNA and 100 pmol of each primer were included in 100 µl of the reaction mixture, which was incubated at 94 °C for 10 min, then the DNA was amplified by 40 thermocycles of 60 s at 94 °C, 120 s at 50 °C and 150 s at 72 °C. The amplified fragment was subcloned into the PCR^TMII vector (Invitrogen, TA Cloning Kit) and sequenced.

DNA was manipulated by the standard procedures (25). Total DNA of D. SY was digested with several restriction enzymes, resolved by electrophoresis on a 1.0% agarose gel, then transferred onto nylon membranes (Hybond N⁺, Amersham). The blots were hybridized with fluorescein isothiocyanate-labeled probes and detected using the ECL^TM random prime labeling and detection system (Amersham).

The total DNA was digested with the restriction enzymes that were selected by Southern blotting, then separated by 1% agarose gel electrophoresis. A fragment identified by Southern blotting was purified, then subcloned into pUC119 (26) or pUC18 (27). This DNA library was transferred onto nylon membranes (Hybond N⁺) and screened by colony hybridization. The selected gene was digested with the appropriate restriction endonucleases and subcloned into pUC18. The plasmid DNAs were prepared using Qiagen^TM and sequenced. DNA was sequenced by means of an ALF II^TM DNA Sequencer (Pharmacia Biotech Inc.) using an AutoRead^TM Sequencing Kit (Pharmacia). The sequences were then analyzed using the software Genetyx-MAC (S.D.C.). The program ``gap'' in the GCG package (Genetics Corporation Group) was used to estimate homology scores.

An oligonucleotide primer (5-AGG-AAT-TCA-TAT-GGC-CGA-GA-3) was designed to generate an NdeI digestion site at the initiation codon. It also contains an EcoRI site at the 5 terminal for subcloning. Another primer (5-TT-AAG-GAG-CGC-GCG-GTG-GTA-3) corresponding to the region immediately downstream of the PstI site in the coding sequence was synthesized. Using the primer set, a 432-bp DNA fragment corresponding to the 5-terminal half of the coding sequence was amplified by PCR (30 cycles of 60 s at 95 °C, 30 s at 50 °C, and 120 s at 72 °C) from the plasmid, pAB101. The amplified fragment was digested with EcoRI and PstI, subcloned into pUC18, and sequenced. Among several clones sequenced, pAB501, which was free from mutation, was selected and further manipulated. The NdeI/PstI fragment of pAB501 was excised and subcloned into pET21c with a PstI/EcoRI fragment of pAB401 corresponding to the 3-terminal end of the gene. Thus, the entire coding sequence was cloned downstream of the T7 promoter and the Shine-Dalgarno sequence of pET21c in pAB503.

A reaction mixture (100 µl) containing 0.1 M potassium phosphate buffer (pH 8.0), 1 mM L-amino acid, and the crude extract of E. coli BL21(DE3) harboring pAB503 or D. SY was incubated at the specified temperature. Thereafter, the mixture was cooled to 37 °C, then mixed with 400 µl of 30 mM 4-aminoantipyrine, 30 mM N-ethyl-N-(2-hydroxy-3-sulfopropyl)-m-toluidine, 2.5 units of horseradish peroxidase, 100 milliunits of D-amino acid oxidase, and 0.1 M potassium phosphate buffer (pH 8.0), then incubated for 30 min at 37 °C. D-Amino acid oxidase oxidizes D-amino acids to produce hydrogen peroxide which reacts with 4-aminoantipyrine and N-ethyl-N-(2-hydroxy-3-sulfopropyl)-m-toluidine, to form a condensation product with an absorption peak at 555 nm (28). Considering the D-amino acids formed during the incubation at 37 °C, the specific activity at each temperature was calculated. One unit was defined as the activity required to racemize 1 µmol of amino acid per 1 min.

SDS-PAGE was performed as described by Laemmli (29). The protein concentration was determined by the method of Bradford (30), using bovine serum albumin as the standard.

RESULTS

While examining the roles of heat shock proteins in the thermotolerance of hyperthermophiles, we intended to clone the genes coding heat shock proteins from hyperthermophilic archaea. A set of primers was designed from the consensus sequences of hsp70 stress proteins and used to PCR amplify the total DNA of several hyperthermophilic archaea. Among those tested, a fragment of about 400 bp was amplified from the DNA of D. SY. However, the sequence of the amplified DNA was not homologous to hsp70s, but rather similar to the gene for aspartate racemase from Streptococcus thermophilus IAM 10064 as shown later. Since no amino acid racemases or D-amino acids had been found in archaea, we further studied this fragment.

PCR might have amplified DNA from contaminating bacteria, or the amplified DNA fragment might be part of the gene which is not structurally or functionally related to amino acid racemases. Thus, the gene corresponding to the amplified DNA was searched in the total D. SY DNA.

We identified the appropriate restriction enzymes to clone the gene by means of Southern blotting using the amplified DNA fragment as a probe. HindIII was selected because a single fragment of about 3-kbp, which is large enough to include the entire aspartate racemase gene, hybridized to the probe.

Total D. SY DNA was digested with HindIII, and a fragment of about 3-kilobase pairs was subcloned into pUC119. This gene library was screened and the plasmid, pAB101, was generated. The fragment contained an open reading frame that was highly homologous to the aspartate racemase gene, but devoid of the 3-terminal sequence of the gene.

Therefore, we cloned the remaining portion of the gene using a 344-bp PstI/HindIII fragment of pAB101, corresponding to the 3-terminal of the open reading frame in the clone, as a probe. The cloned pAB401 contained about 1.8 kilobase pair of PstI/HincII fragment covering the 3-terminal of the gene (Fig. 1A).

The sequence of a 1440-bp DNA region including a 705-bp open reading frame, encoding a 235-residue polypeptide, is shown in Fig. 1B. The underlines indicate the binding sites of the PCR primers used for the screening. Although a ribosomal binding sequence (Shine-Dalgarno sequence) was found just before the initiation codon, no putative archaeal promoter sequence was found. The estimated molecular weight of the encoded protein was 25,977. It shared considerable homology (35.2% identity and 63.1% similarity in amino acid sequence) with the aspartate racemase of the lactic eubacterium, S. thermophilus (31) (Fig. 2A).

The amino acid sequence was also homologous to those of glutamate racemases from Bacillus sphaericus,² E. coli (32, 33), Lactobacillus brevis (34), Lactobacillus fermenti (35), and Pediococcus pentosaceus (36). The homology scores were 28.2, 24.0, 21.6, 27.6, and 27.7%, respectively.

Two cysteine residues, which are thought to be the catalytic centers of these cofactor independent racemases, as well as the surrounding amino acid sequences, were highly conserved among these enzymes (Fig. 2B).

To confirm that the cloned gene really encodes the aspartate racemase, the gene was expressed in E. coli using the T7 polymerase expression system (23). The entire open reading frame of the gene was subcloned into the expression vector pET21c, yielding pAB503 (Fig. 3A). The reading frame was preceded by an E. coli ribosomal binding site under the control of the viral T7 promoter.

E. coli strain BL21(DE3), which expresses T7 polymerase by the induction of IPTG, was transformed with pAB503. The transformant expressed a protein of about 26 kDa (Fig. 3B) 6 h after adding 2 mM IPTG. The molecular weight coincided well with the value predicted from the nucleotide sequence. The amount of the expressed protein was about 10% of the total protein in E. coli. The expressed racemase was soluble and remained stable after an incubation at 70 °C for 30 min (Fig. 3C). The purity of the aspartate racemase was estimated to be more than 95%. Thus, this procedure would be very effective for purifying this hyperthermophilic aspartate racemase.

The amino acid racemase activity of the heated cell-free extract of E. coli BL21(DE3) harboring pAB503 was measured, using alanine, aspartate, and glutamate as the substrates (inset of Fig. 4). At 37 °C, the extract showed a little aspartate racemase activity. Since the E. coli alanine racemase was heat-denatured, this activity was absent. At 70 °C, the extract showed high aspartate racemase activity. However, neither alanine nor glutamate was racemized even at this temperature. The racemization activity of alanine or glutamate at 70 °C was none or at most 0.3% of that of aspartate racemase. These results showed that the encoded protein is a thermophilic amino acid racemase that is highly specific for aspartate.

The aspartate racemase activity increased with temperature from 37 °C to 90 °C as shown in Fig. 4. This result confirmed that the gene originated from a hyperthermophile.

To reconfirm that D. SY really expresses the gene, we measured the aspartate racemase activity in the crude extract of D. SY. Aspartate racemase activity in the extract was about 0.004 unit/mg of protein at 70 °C.

DISCUSSION

Aspartate racemases have been found only in lactic acid bacteria (7). Staudenbauer and Strominger (37) have described the incorporation of D-aspartic acid into murein in S. faecalis and L. casei (37). Only the aspartate racemase of the lactic acid bacterium, S. thermophilus, had been purified and cloned (7, 31). Although D. SY is evolutionarily distinct from eubacteria, the aspartate racemase of D. SY was highly homologous to that of S. thermophilus (Fig. 2A). Both of these aspartate racemases are highly related to glutamate racemases (Fig. 2B).

Several amino acid racemases, including proline racemase, diaminopimerate epimerase, glutamate racemase, and aspartate racemase, are cofactor independent and use thiol groups of cysteine residues as bases (6, 9, 10, 12, 28). Amino acid racemization proceeds by means of one- and two-base mechanisms. The reaction by pyridoxal 5-phosphate-dependent amino acid racemases proceeds by the one-base mechanism (8). Cofactor independent amino acid racemases achieve racemization by the two-base mechanism (9, 10, 12, 28, 38). In the latter, an -hydrogen of an amino acid is abstracted on one face as a proton while a proton is incorporated on the other face. Thus, two cysteine residues are required for the reaction by these amino acid racemases. Proline racemase and aspartate racemase consist of two identical subunits. It is likely that the active sites of these racemases are formed at the interface of two identical subunits, each of which provides one catalytic cysteine residue. However, glutamate racemase exits as monomer. Thus, the two catalytic cysteine residues are in one subunit (38). The cysteine residues and surrounding amino acid sequences are highly conserved among all cloned and sequenced aspartate and glutamate racemases (Fig. 2B). Although Cys-197 of the S. thermophilus aspartate racemase is not essential for catalysis (28), the cysteine residue and surrounding amino acid sequence are also conserved in D. SY aspartate racemase. The regions surrounding the conserved core sequences of aspartate racemases are quite distinct from those of glutamate racemases. Since both aspartate racemases from S. thermophilus and D. SY are highly specific for aspartate, the regions around the conserved core sequences might be related to substrate specificity.

The recombinant D. SY aspartate racemase expressed in E. coli was extremely stable (Fig. 4). This enzyme would be useful for studying the structure and function of cofactor-independent amino acid racemases.

It has been believed that archaea contain neither D-amino acids nor amino acid racemases. However, the results of this study refuted this notion. The functions of the aspartate racemase and the produced D-aspartate in D. SY are of interest. This hyperthermophile is an obligate heterotroph that requires amino acids for growth (39). Although most amino acids spontaneously racemize very slowly, aspartate does so rapidly at high temperature (D/L ratio 0.2/day, at 106.5 °C) (40, 41). Thus, the L-aspartate synthesized by autotrophs rapidly racemizes in the vent. This suggests that the aspartate racemase functions to uptake D-aspartate formed in the vent. In addition, D. SY might contain D-aspartate as a cell component. Since cell wall or cell envelope polymers of archaea are very diverse compared with those of eubacteria, it is possible that the cell wall or envelope of D. SY contains D-aspartate like eubacteria.

Endogenous free D-amino acids have been found in mammals (13, 14, 15). D-Serine is localized in the brain, and it is thought to be the endogenous ligand for the glycine site of the N-methyl-D-aspartic acid receptor (42). Although free D-aspartate and D-glutamate have been found in mammalian tissues, their functions and biosynthetic pathways remain unknown. The fact that homologous aspartate racemases have been found in evolutionarily distinct organisms suggests that the D-aspartate in mammals is also produced by an aspartate racemase similar to that of D. SY.

The nucleotide sequence showed that one of the oligonucleotides used for PCR hybridized to the noncoding region. This means that the gene was amplified by chance and that there is no functional or structural relationship between this aspartate racemase and hsp70s.