d-Amino acids play widespread structural, functional and regulatory roles in organisms. These d-amino acids often arise through the stereoinversion of the more plentiful l-amino acids catalysed by amino acid racemases and epimerases. Such enzymes are of interest since many are recognized targets for the development of drugs or may be employed industrially in biotransformation reactions. Despite their enzyme–substrate complexes being diastereomers, some racemases and epimerases exhibit a kinetic pseudo-symmetry, binding their enantiomeric or epimeric substrate pairs with roughly equal affinities and catalyzing their stereoinversion with similar turnover numbers. In other cases, this kinetic pseudo-symmetry is absent or may be ‘broken’ by substitution of a catalytic Cys by Ser at the active site of cofactor-independent racemases and epimerases, or by altering the Brønsted base of the catalytic dyad that facilitates deprotonation of the Cys residue. Moreover, a natural Thr-containing l-Asp/Glu racemase was discovered that catalyses ‘unidirectional’ substrate turnover, unlike the typical bidirectional racemases and epimerases. These observations suggest that bidirectional Cys–Cys racemases may be re-engineered into ‘unidirectional’ racemases through substitution of the thiol by a hydroxyl group. Catalysis by such ‘unidirectional’ racemase precursors could then be optimized further by site-directed mutagenesis and directed evolution to furnish useful enzymes for biotechnological applications.

Over the past three decades, there has been increasing recognition that d-amino acids play widespread structural, functional and regulatory roles in organisms. For example, d-Ala, d-Glu, d,l-(meso)-diaminopimelate and d-Asp are important components of peptidoglycan found in cell walls of eubacteria. Branched-chain d-amino acids regulate processes in bacteria, such as remodelling of cell walls and disassembly of biofilms. d-Ser plays a role in bacterial virulence, and may be found in the d-Ala–d-Ser dipeptide that replaces the d-Ala–d-Ala dipeptide present in bacterial peptidoglycan, thereby conferring resistance to vancomycin. The biological roles of d-amino acids are not limited to bacteria. For example, free d-Ser and d-Asp are present in a variety of species, ranging from bacteria to mammals. Both amino acids play a role in brain function, and d-Asp plays an important role in the reproduction of vertebrates.

One principle source of d-amino acids is through the stereoinversion of the more abundant l-amino acids catalysed by amino acid racemases and epimerases (eqn. 1). These enzymes are of interest since many are recognized targets for the development of drugs or may be employed industrially in biotransformation reactions. Typically, amino acid racemases and epimerases may be either cofactor-independent or utilize pyridoxal 5′-phosphate (PLP) as a cofactor, although such enzymes from the enolase superfamily catalyse the racemization or epimerization of N-acylamino acids and dipeptides with the aid of a divalent cation. Alanine, arginine and serine racemases, as well as isoleucine epimerase, are examples of enzymes requiring PLP as a cofactor. Formation of a Schiff base with the amino group of the substrate helps to acidify the α-carbon (Figure 1a). On the other hand, aspartate racemase (AspR), glutamate racemase (GluR), proline racemase (ProR), diaminopimelate epimerase (DAPE) and 4-hydroxyproline epimerase (HypE) all catalyse the racemization or epimerization of their respective substrates without the aid of PLP (Figure 2). The active sites of these PLP-independent amino acid racemases contain a pair of highly conserved cysteine residues that function as enantiospecific Brønsted bases to effect catalysis via a two-base mechanism (Figure 1b). The thiolate of Cys A acts as a Brønsted base (pKa ~8–10) to abstract the α-proton from the l-amino acid substrate (nominal pKa ~29) to generate an aci-carboxylate/carbanionic intermediate, which is subsequently protonated from the opposite side by the thiol group of Cys B to generate the corresponding d-amino acid. In the opposite reaction direction, the roles of the Cys residues are reversed. X-ray crystal structures show that the two Cys residues are located on the opposite sides of the active site, and that the enantiomeric or epimeric substrates bind individually in a mirror image packing arrangement with a pseudo-mirror plane formed by the three non-hydrogen substituents on the stereogenic centre undergoing inversion of configuration.

Figure 1

‘Two-base’ mechanisms of pyridoxal 5′-phosphate-dependent (a) and cofactor-independent (b) amino acid racemases and epimerases

Figure 1

‘Two-base’ mechanisms of pyridoxal 5′-phosphate-dependent (a) and cofactor-independent (b) amino acid racemases and epimerases

Figure 2

Reactions catalysed by amino acid racemases (GluR, AspR and ProR) and epimerases (DAPE and HypE). Racemases and epimerases catalyse the inversion of a single stereocenter in substrates containing one or several stereocenters, respectively.

Figure 2

Reactions catalysed by amino acid racemases (GluR, AspR and ProR) and epimerases (DAPE and HypE). Racemases and epimerases catalyse the inversion of a single stereocenter in substrates containing one or several stereocenters, respectively.

The pKa of the Cys residues may be lowered by two different mechanisms. In some cases, there is a catalytic dyad present. For example, in ProR from Trypanosoma cruzi, His 132 and Asp 296 act as Brønsted bases to deprotonate Cys 130 and 300, respectively. Most GluRs utilize an Asp–Cys dyad for the deprotonation of d-Glu and a His–Cys dyad (and possibly, a Glu–His–Cys triad) for the deprotonation of l-Glu. Furthermore, the deprotonation of the Cys residues may be facilitated by the location of the Cα atom of the substrate and Cys sulphur atoms at the N-terminal ends of short α-helices. This directs the positive dipole of the α-helix towards the thiol group and the α-carboxyl group, which perturbs the pKa of the Cys thiol downwards (to ~6–7) and stabilizes the aci-carboxylate of the intermediate, respectively. A similar role has been suggested for the α-helices in DAPE, which has no Brønsted bases located adjacent to the reactive Cys residues (Figure 3).

Figure 3

Representative structures of PLP-independent amino acid racemases and epimerases. (a) The structures of the monomers of GluR (PDB ID 5HJ7, bound d-Glu), AspR (PDB ID 5HJ7, bound citrate), ProR (PDB ID 1 W61, bound pyrrole-2-carboxylate), DAPE (PDB ID 2GKE, covalently bound l,l-aziDAP) and HypE (PDB ID 4 J9X, bound (4R)-hydroxy-l-proline) are shown in cartoon representation. The active site Cys residues and ligands are shown in ball and stick representation. α-Helices that direct a positive dipole towards the active site Cys thiol groups are shaded in dark green (ProR) or dark blue (DAPE and HypE). (b) The structures of the monomers of the unidirectional l-AspR/GluR from Escherichia coli O157 (PDB ID 5HRC, pink), unidirectional l-AspR/GluR from E. coli BL21(DE3) (PDB ID 5ELM, light purple), AspR from Microcystis aeruginosa (PDB ID 5WXY, light cyan) and AspR from Pyrococcus horikoshii (PDB ID 1JFL, light brown) are superposed and shown in cartoon representation. The stereoview of the active site shows the Brønsted acid–base catalysts in ball and stick representation. l-Asp bound to l-AspR/GluR from E. coli O157 (PDB ID 5HRC) is shown for reference. In both panels, the atoms are coloured as yellow (sulphur), blue (nitrogen), red (oxygen) and grey or the colour of the corresponding monomer (carbon).

Figure 3

Representative structures of PLP-independent amino acid racemases and epimerases. (a) The structures of the monomers of GluR (PDB ID 5HJ7, bound d-Glu), AspR (PDB ID 5HJ7, bound citrate), ProR (PDB ID 1 W61, bound pyrrole-2-carboxylate), DAPE (PDB ID 2GKE, covalently bound l,l-aziDAP) and HypE (PDB ID 4 J9X, bound (4R)-hydroxy-l-proline) are shown in cartoon representation. The active site Cys residues and ligands are shown in ball and stick representation. α-Helices that direct a positive dipole towards the active site Cys thiol groups are shaded in dark green (ProR) or dark blue (DAPE and HypE). (b) The structures of the monomers of the unidirectional l-AspR/GluR from Escherichia coli O157 (PDB ID 5HRC, pink), unidirectional l-AspR/GluR from E. coli BL21(DE3) (PDB ID 5ELM, light purple), AspR from Microcystis aeruginosa (PDB ID 5WXY, light cyan) and AspR from Pyrococcus horikoshii (PDB ID 1JFL, light brown) are superposed and shown in cartoon representation. The stereoview of the active site shows the Brønsted acid–base catalysts in ball and stick representation. l-Asp bound to l-AspR/GluR from E. coli O157 (PDB ID 5HRC) is shown for reference. In both panels, the atoms are coloured as yellow (sulphur), blue (nitrogen), red (oxygen) and grey or the colour of the corresponding monomer (carbon).

Active site pseudo-symmetry

Unlike most enzymes which exhibit exquisite stereochemical preference with respect to their substrates, racemases and epimerases furnish a notable exception to this generalization since they must bind and catalyse the reaction of both substrate enantiomers and epimers, respectively. One of the most remarkable features of these enzymes is that some have evolved to bind their enantiomeric or epimeric substrate pairs with roughly equal affinities (as typically measured by their Km values) and to catalyse their stereoinversion with similar turnover numbers (kcat). Considering that the enzyme–substrate complexes are diastereomeric, this kinetic pseudo-symmetry is striking and suggests that evolution has led to a functional asymmetry at the active sites of these enzymes. Overall, however, the ratio of catalytic efficiencies (i.e., kcat/Km) for racemases operating in the ld and the dl reaction directions should be unity since this corresponds to the equilibrium constant (eqn. 2) as dictated by the Haldane relationship (eqn. 3).

formula
(1)
Keqapp=[D][L]
(2)
Keqapp=(kcat/Km)LD(kcat/Km)DL
(3)

Indeed, most experimentally observed Keq values determined for racemases using the Haldane equation reside between 0.5 and 2.0.

The observation of kinetic pseudo-symmetry among some racemases and epimerases raises two interesting questions: (1) given that evolution has led to homochirality among proteins (i.e., composed of l-amino acids), how do such racemases and epimerases, which lack mirror symmetry, bind and catalyse the reactions of their enantiomeric and epimeric substrates with near equality? and (2) can the bidirectional interconversion of stereoisomers be ‘broken’ to favor unidirectional conversion of the stereoisomers? To address these questions, our discussion will focus on cofactor-independent amino acid racemases and epimerases.1

Structural studies have furnished valuable insights into how chiral racemases or epimerases cope with their enantiomeric or epimeric substrates. In addition to having two symmetrically disposed Cys residues at their active sites, the structures of AspR and GluR monomers contain two compact α/β domains with the active site nestled between the two domains (Figure 3). Typically, these domains are related by a pseudo-twofold symmetry axis that passes roughly through the Cα of the bound substrate and superposes the catalytic Cys residues. The domains of AspR from Pyrococcus horikoshii OT3 exhibit a pseudo-mirror-symmetric spatial arrangement of the amino acids at the active site. Consequently, the active site may be regarded as two moieties specific for each enantiomeric substrate. The C-terminal domain of the AspR monomer can be expressed and purified, indicating that it has high structural stability and suggesting the existence of an ancestral domain. Indeed, it has been proposed that the PLP-independent amino acid racemases likely arose via a gene duplication event followed by gene fusion and subsequent adaptation to bind and catalyse the reaction of enantiomers in a mirror-symmetric fashion.

Although the structures of monomers of ProR, DAPE and HypE differ from those of GluR and AspR, all show a bidomain structure (Figure 3) wherein the domains are structurally homologous and related by a pseudo-twofold symmetry axis that passes through the reaction centre. Interestingly, a high degree of residue homology often resides adjacent to the Cys residues. Overall, this high degree of symmetry at the active site contributes to the pseudo-symmetry observed for some racemases and epimerases with respect to their kinetic parameters.

Altering the active site residues

Is it possible to engineer racemases and epimerases that exhibit unidirectional turnover of one enantiomeric or epimeric substrate over the other, thereby breaking the pseudo-symmetry in some cases? Presumably, the pKa of one of the Brønsted acid–base catalysts could be changed either by substitution of an active site Cys by an alternative amino acid residue or by altering the identity of dyad amino acids that serve as Brønsted bases to facilitate deprotonation of the catalytic Cys residues.

Often, when enzymologists seek to identify the catalytic Cys residues present at the active sites of amino acid racemases or epimerases, either one or both Cys residues are substituted by an Ala residue using site-directed mutagenesis. In general, replacement of a single Cys by Ala yields a catalytically inactive enzyme in both the ld and dl reaction directions. However, the Cys residue in these variants may retain either all or some of its ability to act as a Brønsted base to deprotonate alternative substrates in an enantiospecific manner (Figure 4). Alternatively, conservative Cys→Ser substitutions are sometimes employed to study catalysis by amino acid racemases. While such substitutions can yield variants that exhibit enantiospecific deprotonation of alternative substrates similar to the Cys→Ala substitutions, the Ser-containing variants often retain some racemase activity. Indeed, this was the case for the Cys→Ser variants of GluR from Lactobacillus fermentum, which also exhibited ‘breaking’ of pseudo-symmetry with their natural substrates to favour turnover of one enantiomer over the other at increased pH. The catalytic efficiency (kcat/Km) was reduced ~1200- and ~600-fold in both reaction directions for the C73S and C184S variants, respectively, at the pH optimum of 8.0. A striking observation, however, was made when the pH–rate profile for the Cys→Ser variants was examined with d-Glu as the substrate. The value of kcat/Km for the C73S variant increased ~10-fold from pH 7.0 to 8.7, consistent with a shift in the optimal pH to a value >9. On the other hand, the kcat/Km values for the C184S variant exhibited very little variation from pH 7.0 to 8.7. In fact, the values of (kcat/Km)C73S/(kcat/Km)C184S changed from ~0.5 at pH 7 to ~ 1.0 at pH 7.8 and ~4.2 at pH 9. Although the pH dependencies of the kcat/Km values for the Ser variants in the ld reaction direction were not examined due to assay obstacles, the authors anticipated that the behaviour would be reversed with C184S exhibiting greater efficiency as the pH is increased.2 Thus, the C73S variant exhibited ‘unidirectional’ behaviour at higher pH values. This intriguing observation foreshadowed the discovery of a natural ‘unidirectional’ l-Asp/Glu racemase.

Figure 4

Examples of enantioselective reactions catalysed by Cys→Ala and Cys→Ser variants of GluR and DAPE. The C73A variant of GluR from Lactobacillus fermentum (LfGluR) preferentially catalyses the elimination of HCl from (2R,3R)-3-chloroglutamate, while the C184A variant preferentially catalyses the elimination of HCl from (2S,3S)-3-chloroglutamate. The C73S variant of LfGluR preferentially catalyses the elimination of water from l-N-hydroxyglutamate, while the C184S variant preferentially catalyses the elimination of water from d-N-hydroxyglutamate. Although the C73A and C217A variants of DAPE from Haemophilus influenzae (HiDAPE) were both inactive as epimerases, they were able to catalyse the elimination of HF from d,l-3-fluoro-DAP and l,l-3-fluoro-DAP, respectively. On the other hand, in addition to catalysing the elimination of HF from the 3-fluoro-DAP diastereomers, the C73S and C217S variants were also able to catalyse the epimerization of DAP at rates approximately 2%–3% of that exhibited by wild-type DAPE.

Figure 4

Examples of enantioselective reactions catalysed by Cys→Ala and Cys→Ser variants of GluR and DAPE. The C73A variant of GluR from Lactobacillus fermentum (LfGluR) preferentially catalyses the elimination of HCl from (2R,3R)-3-chloroglutamate, while the C184A variant preferentially catalyses the elimination of HCl from (2S,3S)-3-chloroglutamate. The C73S variant of LfGluR preferentially catalyses the elimination of water from l-N-hydroxyglutamate, while the C184S variant preferentially catalyses the elimination of water from d-N-hydroxyglutamate. Although the C73A and C217A variants of DAPE from Haemophilus influenzae (HiDAPE) were both inactive as epimerases, they were able to catalyse the elimination of HF from d,l-3-fluoro-DAP and l,l-3-fluoro-DAP, respectively. On the other hand, in addition to catalysing the elimination of HF from the 3-fluoro-DAP diastereomers, the C73S and C217S variants were also able to catalyse the epimerization of DAP at rates approximately 2%–3% of that exhibited by wild-type DAPE.

Altering the catalytic dyad

The observation that some of the Cys→Ser variants retain activity, albeit markedly reduced relative to the wild-type enzyme, is consistent with the presence of the adjacent Brønsted bases that normally facilitate deprotonation of the Cys thiol. Hence, partial deprotonation of the Ser hydroxyl group (pKa ~ 16) may occur despite its pKa being greater than that of the thiol group (pKa ~10) by ~6 units. Moreover, substitution of the Brønsted bases adjacent to the catalytic Cys residues can yield a marked asymmetry of the kinetic parameters. For example, for the D7S variant of Aquifex pyrophilus GluR, (kcat/Km)ld/(kcat/Km)dl = 3.5, while for the H186N variant of L. fermentum GluR, (kcat/Km)ld/(kcat/Km)dl = 2.0. Glu 147 of A. pyrophilus GluR is located at the tip of an interdigitating loop that inserts into the active site from an adjacent monomer. The Oε1 atom of Glu 147 is 4.2 Å from the S atom of Cys 178 and is stabilized by the Nε2 atom of His 180 on the same monomer. Intriguingly, for the E147N variant, (kcat/Km)ld/(kcat/Km)dl = 46! Thus, in addition to the Cys→Ser substitution perturbing the kinetic pseudo-symmetry, altering the dyad residues may achieve a similar outcome.

A naturally occurring Cys–Thr racemase

In 2015, the genome of E. coli BL21(DE3) was shown to encode an l-Asp/Glu-specific racemase, designated as EcL-DER. Intriguingly, this enzyme acted only on either l-Asp or l-Glu as the substrates, but not their corresponding enantiomers. The 1.8-Å resolution X-ray crystal structure of the dimeric EcL-DER was similar to that of Asp/Glu racemase from P. horikoshii OT3. Indeed, the enzyme contained two superimposable α/β domains, i.e., the N- and C-terminal domains, that formed a pseudo-mirror-symmetry at the active site with one striking difference: rather than containing two antipodal Cys residues, EcL-DER contained a Cys and a Thr as the Brønsted acid–base catalysts (Figure 3b). Substitution of the active site Thr residue by Cys restored the bidirectional activity, although activity in the dl reaction direction was only 20% of that observed in the ld reaction direction.

The X-ray crystal structures of EcL-DER from E. coli O157 complexed with l-Asp or d-Asp revealed that the α-H of l-Asp was directed at the sulfhydryl group of Cys 197, while the α-H of d-Asp was directed at the hydroxyl group of Thr 83. This mirror image-binding orientation of the enantiomers was in accord with the observation that the enzyme catalyses only conversion of l- to d-Asp. The observation that this enzyme with an SH/OH pair at its active site exhibits ‘unidirectional’ racemase activity in the ld reaction direction implies that (kcat)ld > (kcat)dl and, based on the Haldane relationship, (Km)l-substrate > (Km)d-substrate.

Conclusions

The discovery of a natural Thr-containing, unidirectional l-Asp/Glu racemase and the pseudo-symmetry ‘breaking’ by racemases resulting from either Cys→Ser substitutions or substitutions of the catalytic dyad residues suggest that bidirectional Cys–Cys racemases may be re-engineered as ‘unidirectional’ racemases through substitution of the thiol by a hydroxyl group. Catalysis by such ‘unidirectional’ racemase precursors could then be optimized further by site-directed mutagenesis and directed evolution. Interestingly, bacterial homologues with other unbalanced active site pairs have also been identified, including Cys–Ser, Cys–Asp, Cys–Asn, Cys–Ala and Cys–Gly pairs. Functional characterization of these proteins could furnish important insights for the development of ‘unidirectional’ racemases for biotechnological applications.

Further reading

  • Ahn, J.W., Chang, J.H., and Kim, K.J. (2015) Structural basis for an atypical active site of an l-aspartate/glutamate-specific racemase from Escherichia coli. FEBS Lett. 589, 3842–3847. DOI: 10.1016/j.febslet.2015.11.003

  • Bearne, S.L. (2020) Through the looking glass: chiral recognition of substrates and products at the active sites of racemases and epimerases. Chem. Eur. J. 26, 10367–10390. DOI: 10.1002/chem.201905826

  • Buschiazzo, A., Goytia, M., Schaeffer, F., et al. (2006) Crystal structure, catalytic mechanism, and mitogenic properties of Trypanosoma cruzi proline racemase. Proc. Natl. Acad. Sci. USA103, 1705–1710. DOI: 10.1073/pnas.0509010103

  • Fischer, C., Ahn, C., and Vederas, J.C. (2019) Catalytic mechanism and properties of pyridoxal 5′-phosphate independent racemases: how enzymes alter mismatched acidity and basicity. Nat. Prod. Rep. 36, 1687–1705. DOI: 10.1039/c9np00017h

  • Glavas, S., and Tanner, M.E. (2001) Active site residues of glutamate racemase. Biochemistry40, 6199–6204. DOI: 10.1021/bi002703z

  • Glavas, S., and Tanner, M.E. (1999) Catalytic acid/base residues of glutamate racemase. Biochemistry38, 4106–4113. DOI: 10.1021/bi982663n

  • Hwang, K.Y., Cho, C.S., Kim, S. S., et al. (1999) Structure and mechanism of glutamate racemase from Aquifex pyrophilus. Nat. Struct. Biol. 6, 422–426. DOI: 10.1038/8223

  • Koo, C. W., Sutherland, A., Vederas, J. C., and Blanchard, J. S. (2000) Identification of active site cysteine residues that function as general bases: Diaminopimelate epimerase. J. Am. Chem. Soc. 122, 6122–6123. DOI: 10.1021/ja001193t

  • Liu, X., Gao, F., Ma, Y., et al. (2016) Crystal structure and molecular mechanism of an aspartate/glutamate racemase from Escherichia coli O157. FEBS Lett. 590, 1262–1269. DOI: 10.1002/1873-3468.12148

  • Liu, L., Iwata, K., Kita, A., et al. (2002) Crystal structure of aspartate racemase from Pyrococcus horikoshii OT3 and its implications for molecular mechanism of PLP-independent racemization. J. Mol. Biol. 319, 479–489. DOI: 10.1016/S0022-2836(02)00296-6

  • Pillai, B., Cherney, M.M., Diaper, C.M., et al. (2006) Structural insights into stereochemical inversion by diaminopimelate epimerase: an antibacterial drug target. Proc. Natl. Acad. Sci. USA103, 8668–8673. DOI: 10.1073/pnas.0602537103

  • Tanner, M.E., Gallo, K.A., and Knowles, J.R. (1993) Isotope effects and the identification of the catalytic residues in the reaction catalyzed by glutamate racemase. Biochemistry32 3998–4006. DOI: 10.1021/bi00066a021

  • Usha, V., Dover, L.G., Roper, D.L., and Besra, G. S. (2008) Characterization of Mycobacterium tuberculosis diaminopimelic acid epimerase: paired cysteine residues are crucial for racemization. FEMS Microbiol. Lett. 280, 57–63. DOI: 10.1111/j.1574-6968.2007.01049.x

  • Visser, W.F., Verhoeven-Duif, N.M., and de Koning, T.J. (2012) Identification of a human trans-3-hydroxy-l-proline dehydratase, the first characterized member of a novel family of proline racemase-like enzymes. J. Biol. Chem. 287, 21654–21662. DOI: 10.1074/jbc.M112.363218

Acknowledgements

I thank the Natural Sciences and Engineering Research Council (NSERC) of Canada for a Discovery Grant (grant no. RGPIN-2016–05083) supporting our research on racemases and epimerases.

Author information

graphic

Stephen L. Bearne is a professor of enzymology in the Departments of Biochemistry & Molecular Biology and Chemistry at Dalhousie University, Halifax, Nova Scotia. Research from his laboratory has focused on understanding the role of enzyme binding determinants in transition state stabilization and developing general strategies for the design of inhibitors of racemases and epimerases, and other enzymes of therapeutic interest. Email: sbearne@dal.ca

1.

The epimerization domains of non-ribosomal peptide synthetases, as well as radical S-adenosylmethionine-dependent enzymes, enzymes utilizing dehydration–hydrogenation processes, and peptide epimerases can catalyse the unidirectional epimerization of amino acids residing in peptides (e.g., MurL is a peptide epimerase that catalyses the unidirectional, ATP-dependent conversion of UDP-MurNAc-l-Ala-l-Glu to UDP-MurNAc-l-Ala-d-Glu).

2.

Of course, this would lead to the Haldane equation no longer yielding Keq = 1, which could arise if an iso kinetic mechanism was operating.

Published by Portland Press Limited under the Creative Commons Attribution License 4.0 (CC BY-NC-ND)