Introduction

A variety of free d-amino acids have been detected in animal tissues. As a result, many researchers have attempted to elucidate their physiological functions (Hamase et al. 2002; Radkov and Moe 2014). The distribution and physiological functions of d-serine and d-aspartate have been intensively studied. Free d-serine is present in the frontal brain, and is an endogenous co-agonist of the N-methyl-d-aspartate receptors (Hamase et al. 2002; Wolosker et al. 1999). Free d-aspartate is found in mammalian tissues and plays key roles in neuronal and endocrine functions (Katane and Homma 2011). In invertebrate animal species, the presence and possible function of d-aspartate, d-serine, and d-alanine have been reported (Abe et al. 2005; Corrigan and Srinivasan 1966; D’Aniello 2007; Preston 1987; Saitoh et al. 2012; Rosenberg and Ennor 1961). In addition to these studies, a prior report from our group has demonstrated that d-arginine is found in the annelid worm Sabellastarte indica (Uda and Suzuki 2007).

The formation of d-amino acids from l-amino acids is catalyzed by amino acid racemases. (Jiraskova-Vanickova et al. 2011; Radkov and Moe 2014; Yoshikawa et al. 2009; Abe et al. 2006; Wang et al. 2011). Amino acid racemases are divided into two groups, pyridoxal-5′-phosphate (PLP)-dependent and PLP-independent enzymes, with distinct reaction mechanisms (Yoshimura and Esak 2003). In animals, the amino acid racemase genes have been reported in a limited number of species; PLP-independent alanine racemase from the shrimp, PLP-dependent aspartate racemase (AspR) or serine racemase (SerR) from the ark shell Scapharca broughtonii, the sea slug Aplysia californica, and mammalian species (Yoshikawa et al. 2009; Abe et al. 2006; Wang et al. 2011).

Recently, we cloned and characterized eleven mammalian serine racemase homologs from eight species in eight different invertebrate phyla (Uda et al. 2016; Katane et al. 2016). All of these homologs showed SerR activity and/or AspR activity. Phylogenetic tree showed that animal SerR and AspR are not separated by their particular racemase functions and form a serine/aspartate racemase family cluster. The coral Acropora millepora, Pacific oyster Crassostrea gigas and tiger shrimp Penaeus monodon have two paralogous genes, SerR and AspR, and we considered that these paralogous genes have evolved independently by gene duplication at their recent ancestral species. Moreover, we found interesting the three consecutive residues at position 150–152, which form a loop structure and are closest to the substrate l-serine binding site (Fig. 1). The three residues are well conserved as Pro/His/Glu, Pro, and Phe/Tyr/Asn, respectively, in all SerR enzymes. However, these three residues are replaced with two or three serine residues in all animal AspRs and we referred to the corresponding substitution of two to three consecutive serine residues as the “triple serine loop” region. We have proposed that the triple serine loop region may be responsible for the large aspartate racemase activity and the evolution of AspR from SerR. The importance of the triple serine loop region was indicated by previous site-directed mutagenesis studies. The P151S mutant enzyme of Dictyostelium discoideum SerR showed 2.5–4 times higher SerR activity than wild-type (Ito et al. 2013) and the same mutant of mouse SerR increased the SerR activity without an effect on the K m value (Foltyn et al. 2005). However, it is not clear whether these P151S mutants have AspR activity.

Fig. 1
figure 1

The overall structure (a) and binding site for the substrate l-Ser b in Schizosaccharomyces SerR. Three-dimensional structure of Schizosaccharomyces SerR (Goto et al. 2009; PDB accession code 2ZR8) with the PLP and substrate l-Ser that is a model since the electron density is not available for the whole side chain, was constructed using a SwissPdbViewer (http://kr.expasy.org/spdbv/)

In this study, we investigated the role of triple serine loop region in several SerRs and AspRs using site-directed mutagenesis. To accomplish this, we constructed mutant enzymes which contain a substitution or an introduction of the triple serine loop region in Acropora, Crassostrea and Penaeus AspR or SerR. These mutants showed a large change in AspR activity. Generally, the presence of the triple serine loop region in both AspRs and SerRs led to greater AspR activity while removing the triple serine loop region resulted in almost complete loss of AspR activity. Furthermore, we introduced serine residues in all combinations at position 150–152 in mouse SerR. These mutants revealed the function of individual serine residues in the triple serine loop region.

Materials and methods

Chemicals and materials

All chemicals were reagent grade and obtained from Merck Millipore (MA, USA), Sigma-Aldrich (MO, USA), Tokyo Kasei (Tokyo, Japan), Wako (Osaka, Japan), or Peptide Institute, Inc. (Osaka, Japan).

Site-directed mutagenesis of Penaeus monodon, Crassostrea gigas and Acropora millepora SerR and AspR and Mouse (Mus musculus) SerR

The pET30b overexpression constructs for Penaeus monodon, Crassostrea gigas and Acropora millepora SerR and AspR and Mouse SerR (Uda et al. 2016), were used as templates for PCR-based site-directed mutagenesis (Suzuki et al. 2003). Primers used for construction of mutants for SerRs and AspRs are listed in supplemental Table S1. The mutations were introduced using the KOD+ DNA polymerase (TOYOBO, Tokyo, Japan) under the following conditions: 35 cycles of 94 °C for 15 s, 60 °C for 30 s and 68 °C for 7 min. The PCR products were treated with DpnI and T4 polynucleotide kinase (Takara, Kyoto, Japan) and were self-ligated using T4 ligase. The cDNA insert was completely sequenced to confirm that only the intended mutations were introduced.

Expression of all AspRs and SerRs genes in Escherichia coli

Recombinant proteins with a His tag at the C-terminal end were expressed in Escherichia coli host BL21 (DE3) by induction with 0.5 mM IPTG when OD600 reached a 0.6 absorbance value and then were cultured at 25 °C for 36 h. The cells were sonicated in lysis buffer (50 mM NaH2PO4, 300 mM NaCl, pH 8.0) with a Microson XL 2000 sonicator at 4 °C for 10 cycles of 1 min at 15 W with 30 s between each cycle. After sonication, the suspensions were centrifuged at 15,000×g for 30 min at 4 °C. The resultant soluble recombinant proteins were purified by affinity chromatography with Ni-NTA Superflow (QIAGEN, CA, USA) using open column (Muromac column S; Muromachi Kagaku Kogyo Ltd., Tokyo, Japan, 50 mm; diameter, 5 mm). All recombinant proteins appeared as a single band on SDS-PAGE. The purified enzyme concentrations were determined using the Pierce BCA-protein assay reagent kit (Novagen, WI, USA) with the bovine serum albumin as a standard, were shown in the supplemental Table S2.

Enzyme assays

The enzyme assay mixture for both racemase and dehydratase activities contained 50 mM Tris–HCl (pH 8.0), 1 mM ATP, 1 mM DTT, 1 mM MgCl2, 0.025 mM PLP, 30 μL of purified recombinant enzyme, and a substrate l- or d-isomers of serine and aspartate at various concentrations, in a final volume of 0.3 mL. The substrate and enzyme concentrations used in the kinetic assays were shown in the supplemental Table S2. Each condition was repeated independently at least three times and the negative control reactions were carried out in the absence of enzyme solution or with heat-inactivated enzymes. The reactions were performed at 30 °C for 0.5–16 h and stopped by boiling for 5 min. The specific racemase and dehydratase activities were measured using 10 mM l- or d-amino acids. The kinetic parameters, K m and k cat values of SerR and AspR were determined from the Lineweaver–Burk plots.

The racemase assay and detection of resultant l- or d-amino acids were performed as previously described (Uda et al. 2016). The resultant pyruvate by the serine dehydratase (SDH) activity was determined by end-point analysis using lactate dehydrogenase-catalyzed reduction of pyruvate to lactate by NADH as described previously (de Miranda et al. 2002; Stri´šovský et al. 2003). The reaction mixture contained 100 mM Tris–HCl (pH 7.5), 200 μM NADH, 2.5 U lactate dehydrogenase and a 100 or 200 μL aliquot of pyruvate-containing sample in a final volume of 500 μL. The reactions were performed at 37 °C for 30 min. The decrease in absorbance at 340 nm was measured with an Ultrospec 6300 Pro spectrophotometer (GE Healthcare Bio-Science Corp., Uppsala, Sweden). The amount of pyruvate produced from l- or d-serine was calculated based on a standard curve for pyruvate.

Alignment of amino acid sequences of AspR and SerR and construction of a phylogenetic tree

A multiple sequence alignment of AspRs and SerRs was generated by the MUSCLE program with default parameters (Edgar 2004) using MEGA 7 (Tamura et al. 2013). Throughout this paper, the amino acid sequence numbering of the fission yeast Schizosaccharomyces pombe SerR was used. The best-fit model for the maximum likelihood (ML) analysis was searched by MEGA 7. The ML tree was constructed using MEGA with best-fit model (LG + G).

Results and discussion

Substituting the triple serine loop region in AspRs enhances serine racemization

We previously reported the presence of both SerR and AspR genes in Acropora millepora (Cnidaria), Crassostrea gigas (Mollusca) and Penaeus monodon (Arthropoda) (Uda et al. 2016). Each animal AspR gene has a higher sequence homology with the SerR in same species compared to other AspRs, and has evolved independently by gene duplication at its recent ancestral species (Uda et al. 2016) (Fig. 2). To test our hypothesis that the triple serine loop region is involved in conferring strong aspartate racemase activity of AspR, we replaced the two or three serine residues at amino acid positions 150–152 of Acropora, Crassostrea and Penaeus AspR with the corresponding residues of SerR in the same species (Fig. 3). We previously reported that the Acropora, Crassostrea and Penaeus AspRs had strong racemase activities against l- and d-aspartate and weak but significant activities against l- and d-serine (Uda et al. 2016) (Table 1). Furthermore, in this study, we showed the Crassostrea and Acropora AspRs had weak SDH activity, which generates pyruvate and ammonia from serine (Table 1).

Fig. 2
figure 2

Phylogenetic tree based on the amino acid sequence of AspRs and SerRs. The ML tree was constructed using the MEGA (Tamura et al. 2013). The ML bootstrap values are shown at the branching point. The following sequences were used for alignment and phylogenetic tree; Penaeus monodon AspR (GenBank Accession Nos. LC041007) and SerR (LC041008), Crassostrea gigas AspR (EKC33218) and SerR (LC041009), Acropora millepora AspR (JT020910) and SerR (JT017883), Caenorhabditis elegans SerR (NP_492318), Scapharca broughtonii AspR (BAE78960), Mus musculus SerR (NM_013761), Arabidopsis thaliana SerR (NP_192901), Oryza sativa SerR (NP_001053521), Dictyostelium discoideum SerR (XP_636213), Milnesium tardigradum SerR (EZ759262), Dugesia japonica SerR (LC041006) and Schizosaccharomyces pombe SerR (NP_587715). The stars indicate occurrence of the evolution of the AspR from the SerR following gene duplication

Fig. 3
figure 3

Amino acid sequence alignment around the triple serine loop region. The triple serine loop region at amino acid position 150–152 in Schizosaccharomyces pombe SerR are indicated by an asterisk. The highly conserved residues are boxed in black

Table 1 Comparison of kinetic constants of SerRs and AspRs

The S150P–S151P–S152Y (SSS to PPY) mutation of Crassostrea AspR drastically decreased the specific AspR activity to <0.04% of the wild-type, whereas significantly increased the specific SerR and SDH activities (6–13-fold) (Table 1; Fig. 4). The large change in the AspR activity in the mutant enzyme was caused by approximately 400-fold decrease in k cat rather than by 22–32-fold decrease in K m (Table 1). In SerR activity, the mutant enzyme had a 2.6-fold decrease in K m for d-serine and 5.2 and 1.7-fold increase in k cat for l-serine and d-serine, respectively, while there is no change in the affinity for l-serine as compared to wild-type (Table 1). Thus, the mutant enzyme showed similar substrate affinity for aspartate and serine, but showed 13–20-fold higher k cat, 12–14-fold higher k cat/K m and 16–27-fold higher specific activity for serine than aspartate in racemase reaction (Table 1). Thus, we observed that the SSS to PPY mutation of Crassostrea AspR changed the substrate specificity from aspartate to serine.

Fig. 4
figure 4

Comparison of the K m (a), k cat (b) and k cat/K m (c) values of the wild-type (WT) and mutant of Penaeus monodon AspR, Crassostrea gigas AspR, Acropora millepora AspR, and Mus musculus SerR

Acropora AspR which has Pro-Ser-Ser residues at positions 150–152 (Fig. 3), showed the highest substrate affinity and k cat values for aspartate of all known AspRs (Uda et al. 2016). Acropora AspR and Crassostrea AspR shared only 47% amino acid sequence identity, and evolved independently from different SerR lineages. Nonetheless, S151P–S152F (PSS to PPF) mutant of the Acropora AspR, resulting in Pro-Pro-Phe residues at positions 150–152, exhibited similar substrate kinetics to the Crassostrea AspR SSS to PPY mutant. In fact, the PSS to PPF mutant of the Acropora AspR almost lost its AspR activity (Table 1; Fig. 4). In contrast, the mutated residues had no or little effect on SerR activity and had a 6.4-fold increase on l-Ser dehydratase activity in k cat/K m value (Table 1; Fig. 4). Thus, the PSS to PPF mutant of the Acropora AspR changed substrate specificity from aspartate to serine in the racemase reaction which is in accordance with the data from SSS to PPY mutant of Crassostrea AspR. Furthermore, we introduced S150H–S151P–S152Y (SSS to HPY) mutation in Penaeus AspR, which evolved from another SerR lineage, but attempts to overexpress the mutant gene were not successful.

Introducing the triple serine loop region into SerRs promotes aspartate racemization

Crassostrea and Penaeus SerR had sufficient SerR activity and SDH activity to determine their kinetic parameters; however, these SerR activities were significantly lower than other animal SerRs (Uda et al. 2016). In Marsupenaeus (Penaeus) japonicus, which is closely related to Penaeus monodon, some d-amino acids have been found in several tissues, but d-serine is not detected in any tissues (Okuma et al. 1995; Yoshikawa et al. 2011). Moreover, Katane et al. (2016) reported that Caenorhabditis elegans SerR, which showed little SerR activity, is not critical for serine metabolism in vivo, using a gene deletion mutant. These facts indicate that Crassostrea and Penaeus SerRs probably do not function as a SerR enzyme in vivo.

Acropora SerR had about 100 times higher k cat/K m values of l- and d-serine racemase activity than Crassostrea and Penaeus SerRs and showed insignificant SDH activity and undetectable AspR activity (Table 1; Fig. 4). The SerRs from Crassostrea, Penaeus and Acropora share only 44–58% sequence identity with each other and have different enzyme properties. Nevertheless, P150S–P151S–Y152S (PPY to SSS) mutant of Crassostrea SerR, H150S–P151S–F152S (HPF to SSS) mutant of Penaeus SerR and P150S–P151S–F152S (PPF to SSS) mutant of Acropora SerR, which have the triple serine residues at amino acid positions 150–152, produced similar results in the present study. In these mutant enzymes, the SDH activity was greatly reduced and SerR activity was hardly changed, whereas the AspR activity was dramatically increased (Table 1; Fig. 4). Thus, those mutants showed similar or higher substrate affinity for l- and d-aspartate than l- and d-serine and showed 11–683-fold higher k cat and 28–351-fold higher k cat/K m values for l- and d-aspartate than l- and d-serine racemization (Table 1). Taken together, our results indicate that introducing the triple serine loop region into a SerR changes it to an AspR, and substitution of the triple serine loop region in an AspR changes it to a SerR.

Role of individual serine residues at position 150–152

After having demonstrated the importance of the triple serine loop region in AspR activity, we attempted to elucidate the function of individual serine residues in the triple serine loop region using mouse (Mus musculus) SerR. We introduced serine residues in all combinations at position 150–152 in mouse SerR, and the resultant seven mutant enzymes showed various changes in enzyme activity. It was well known that the mammalian SerR had racemase and dehydratase (SDH) activities against l- and d-serine, and recently we and Ito et al. have disclosed that AspR activity also existed in mouse SerR using recombinant enzyme (Uda et al. 2016; Ito et al. 2016). The single amino acid substitution mutants, H150S, P151S and N152S, revealed several effects on activity. The Ser150 substitution decreased all enzyme activities, especially the SDH activity, while Ser151 dramatically increased the SerR and AspR activities with no change on SDH activity. An increase in the AspR activity was observed after introducing Ser152, while there was little or no change in the other activities (Table 1; Fig. 4). Three mutant enzymes, H150S–P151S (HPN to SSN), H150S–N152S (HPN to SPS), and P151S–N152S (HPN to HSS), indicated that the effect of introducing two serine residues into the triple serine loop region was roughly equivalent to that of the single amino acid substitution (Table 1). The HPN to HSS mutant enzyme showed higher AspR activity than SerR activity and showed the lowest K m and the highest k cat and k cat/K m values for l- and d-aspartate among all mutant enzymes of mouse SerR (Table 1). However, this mutant also demonstrated very high SDH activity due to higher substrate affinity and k cat/K m value (Table 1). The H150S–P151S–N152S (HPN to SSS) mutant, which has the complete triple serine loop region, completely lost the SDH activity and showed about a fivefold and a tenfold decrease in k cat/K m value of AspR activity and SerR activity, respectively, compared to the HPN to HSS mutant (Table 1). Thus, the HPN to SSS mutant changed its function to AspR. These results suggest that changing the main substrate from serine to aspartate in the racemase reaction is caused by introducing at least Ser151 and Ser152, and addition of the third serine residue at position 150 further enhances the enzyme specificity for aspartate due to a decrease in SerR and SDH activities. This conclusion is also supported by the kinetic parameters of wild-type and mutant Acropora enzymes. Acropora AspR, which has Pro-Ser-Ser residues at positions 150–152, showed higher AspR and SDH activity than other AspRs, having the complete triple serine loop region. This result was observed in the case of HPN to HSS mutant of mouse SerR (Table 1; Fig. 4). Additionally, P150S mutant of Acropora AspR decreased the AspR activity relative to the wild-type and lost the SDH activity, which was also observed with the HPN to SSS mutant of mouse SerR. This type of activity switch occurred not only with the AspR enzymes but also was observed between the P151S–F152S and PPF to SSS mutants of Acropora SerR. It is notable that all of the ten enzymes that had a Ser residue at position 150 completely or almost lost the SDH activity without exception (Fig. 4). The observation indicates that the amino acid position 150 in the triple serine loop region is very important for the SDH activity.

All mutations at amino acid positions 150–152 in SerR and AspR bring about dramatic change in the k cat value for racemase activity, while the mutations had little effect on K m values. This observation suggests that the amino acid residues are not involved in the substrate binding but in catalysis of the racemase reaction. The crystal structure of Schizosaccharomyces pombe SerR (Goto et al. 2009; Fig. 1) indicates that these amino acid residues form a flexible loop and are closest to the substrate l-serine binding site, but it is not clear how the residues are involved in catalysis. Further studies, such as crystallographic analysis of AspR, are needed to fully resolve this issue.

Evolution of AspR from SerR

We propose that the ancestral gene of the serine/aspartate racemase family is SerR. Our hypothesis is supported by the following two observations. First, in the molecular phylogenetic tree (Fig. 2), the SerR from plants, slime molds and fission yeast are sister groups to the animal serine/aspartate racemase family cluster. Second, it is more reasonable to assume that Acropora, Crassostrea and Penaeus AspR evolved independently from SerR, which would require three independent evolution events, than to assume that all animal SerRs evolved independently from AspR, which would require seven such events (Fig. 2). We suggest that the evolution of AspR from SerR occurs in three steps. First, gene duplication of the original SerR gene; second, introduction of the two serine residues at position 151 and 152, resulting in a drastic increase of the AspR activity; and third, formation of the complete triple serine loop region by additional introduction of serine residue at position 150, resulting in enhanced the enzyme specificity for aspartate. In fact, Aplysia AspR and several AspR homologs have Pro150–Ser151–Ser152 residues, while all SerR have no serine residues at position 150–152 (Uda et al. 2016). Moreover, two animal SerR homologs from polychaete worm Capitella teleta have Pro150–Pro151–Tyr152 and Pro150–Ser151–Ser152 residues, respectively, in spite of the fact that these homologs differ only by 13 amino acid residues, indicating that the gene duplication and introduction of Ser151 and Ser152 occurred very recently (Uda et al. 2016).

Conclusion

Previously, we proposed that the AspR-specific triple serine loop region at amino acid positions 150–152 may be responsible for the large AspR activity (Uda et al. 2016). To test this hypothesis, we prepared and characterized fourteen mutants in this region of animal SerRs and AspRs. The activity of mutant enzymes indicated that the triple serine loop region plays an essential role in contributing the AspR activity. The presence or absence of the triple serine loop region can decide whether an enzyme would act preferentially as an AspR or SerR. Moreover, we revealed that changing the enzyme property from SerR to AspR is caused by introducing Ser151 and Ser152 at a minimum, and adding the third serine residue at position 150 enhances the enzyme specificity for aspartate due to a decrease of the SerR and SDH activities, suggesting that the AspR gene has evolved from the SerR gene by acquisition of the triple serine loop region.