Ornithine racemase uses a catalytic cysteine

Abstract

Salmonella enterica serover typhimurium expresses an ornithine racemase (OrnR), located in a small operon containing amino acid transporter homologues and D-ornithine/d-lysine decarboxylase. This OrnR is part of a clade that has high identity (>75 %) with other sequences in enterobacteria, but is distant (~40 % identity) from another clade of OrnR found in anaerobic Firmicutes and actinobacteria. OrnR is a member of the pyridoxal-5′-phosphate (PLP)-dependent alanine racemase (AlaR) superfamily (Fold III), but sequence alignments show that it lacks the catalytic Tyr acid/base of the other racemases. However, the sequences show a conserved cysteine. OrnR shows hyperbolic kinetics in the L → D direction, but exhibits sigmoid kinetics in the D → L direction, with n = 1.8. The structure of OrnR predicted by AlphaFold3 was complexed with PLP-DL-ornithine in silico, minimized, and subjected to molecular dynamics simulation. The resulting structure shows that Cys-164 is in the active site, about 4 Å from the Cα of the external aldimine of L-ornithine, while the NZ of Lys-36 is located about 4 Å below Cα on the D-face. The C164A mutant enzyme has no measurable racemization activity, and does not exchange the α-H of L- or D-ornithine in D₂O, consistent with the function of Cys-164 as an acid/base catalyst. These data are consistent with a concerted mechanism for the racemization. Gel filtration of OrnR shows a mixture of dimeric and tetrameric assemblies.

1. Introduction

Ornithine racemase (OrnR) was first discovered in the catabolic pathway for L-ornithine in Clostridium sticklandii [1,2] (Eq. 1). This pathway allows C. sticklandii to utilize anaerobic fermentation of L-ornithine as a carbon and energy source for growth. Enzymes of the pathway include OrnR, D-ornithine aminomutase, 2,4-diaminopentanoate dehydrogenase, and 2-amino-4-oxopentanoate thiolyase, resulting in acetyl-CoA and D-alanine (Eq. 1). This pathway is widely distributed in Firmicutes, some of which are pathogenic, like Clostridiodes difficile, so it may represent a new target for antibiotic drugs. Recently, we discovered another OrnR in Salmonella enterica serovar typhimurium (STM) encoded by STM2358 [3] (Eq. 2). This enzyme is located in the genome in a small operon, together with a D-ornithine/d-lysine decarboxylase (STM2360, DOKDC) and amino acid transporter homologues [3]. OrnR is a member of the alanine racemase (AlaR) superfamily (Fold III) of pyridoxal-5′-phosphate (PLP) dependent enzymes, which primarily contain racemases and decarboxylases [4], as well as a few aldolases [5]. Other racemases in this family include arginine racemase (ArgR) [6], lysine racemase (LysR) [7], and “broad specificity” amino acid racemase (BsR) [8]. However, while AlaR is highly specific for alanine, ArgR, LysR, OrnR and BsR have somewhat overlapping substrate specificities [6-8]. STM OrnR exhibits highest activity with ornithine, much less activity with arginine, and no detectable activity with lysine [3]. We have now used AlphaFold3 to predict the structure of STM OrnR, which shows that Cys-164 is located in a position to serve as a catalytic acid/base for the racemization reaction, in contrast to AlaR and the other racemases, which use a tyrosine as an acid/base catalyst, in addition to the active site lysine that forms a Schiff base with the PLP [9]. Mutation of Cys-164 of STM OrnR to alanine results in an enzyme with undetectable racemase activity, confirming that OrnR uses a Cys/Lys acid/base pair, rather than a Tyr/Lys acid/base pair, to catalyze the racemization reaction. The structural and catalytic properties of STM OrnR and its C164A variant are reported herein.

2. Experimental methods

2.1. Materials

L-Ornithine monohydrochloride was from Fisher Scientific, and D-ornithine monohydrochloride was from Alfa Aesar. Sodium 2,4-dinitro-benzenesulfonate (DNBS) was obtained from TCI America (Product No: D0821). Other materials were obtained from standard commercial sources.

2.2. Enzymes

OrnR from S. typhimurium LT2 (STM2358) and D-ornithine/d-lysine decarboxylase (STM2360) were prepared as described previously [3]. Inducible L-ornithine decarboxylase (iODC) from S. typhimurium LT2 was prepared with the synthesized gene in pET-28+, optimized for expression in E. coli, and with a C-terminal hexaHis-tag, obtained from Twist Bioscience. E. coli BL21(DE3) was transformed with the plasmid, and the cells were grown in Studier autoinduction medium [10] with 200 mg/L kanamycin for 24 h at 37 °C for expression. The cells were collected by centrifugation and suspended in 30 mL 0.05 M potassium phosphate, pH 7.0, 0.3 M NaCl, 0.1 mM pyridoxal-5′-phosphate. The cells were disrupted by sonication for 4 min in 1 min intervals, with cooling on ice, followed by centrifugation for 90 min at 4000 ×g. The supernatant was applied to a BioRad Ni-MAC column, then the column was washed at 1 mL/min with buffer containing 0.02 M imidazole until the 280 nm absorbance returned to baseline. The iODC was then eluted with a linear gradient of buffer from 0.02 M to 0.2 M imidazole. The peak fractions eluting at about 0.1 M imidazole were then concentrated in a centrifugal filter with 30 kDa cutoff, diluted with column buffer without imidazole, concentrated, and frozen at −70 °C.

C164A OrnR was prepared by PCR based mutagenesis with the method of Liu and Naismith [11] using Phusion DNA polymerase (New England Biolabs). The primers, with the sequences shown below, were obtained from MWG Operon. The location of the mutagenic codon is indicated by underlines. The PCR conditions started with an annealing temperature of 66 °C for 30 s, and decreased by 1 °C each cycle until it reached 55 °C, with an extension temperature of 72 °C for 3 min. Twenty additional cycles were

Forward:

CAACTTTAATGCTTACGGCACCGTGTTGCCAACGGTAAAAAATG.

Reverse:

GTAAGCATTAAAGTTGGTGGCCAGTCCATATAATTCCAGAAC.

performed with an annealing temperature of 55 °C for 30 s and extension temperature of 72 °C for 3 min, followed by a final 10 min at 72 °C. The PCR product was used to transform competent E. coli GC-5 cells (Genesee Scientific). The cells were grown overnight in LB supplemented with 100 μg/mL ampicillin. The resulting plasmids were submitted to MWG Operon for sequencing. The plasmid with the confirmed mutation was then used to transform E. coli BL21(DE3) for expression. The C164A mutant OrnR was then purified as described previously for wild-type OrnR [3]. PAGE analysis on 12 % Tris-glycine gels showed that both wild type and C164A OrnR are >90 % pure (Fig. S1).

2.3. Activity assay of OrnR

OrnR activity was measured by coupling the racemization product with either D- or L-ornithine decarboxylase, and measuring the putrescine product by derivatization with sodium 2,4-dinitrobenzenesulfonate (DNBS) (Phillips et al., manuscript in preparation). The racemase (6 μg) was incubated in a total volume of 250 μL with various [Orn] from 0.5 to 26 mM, 0.05 M buffer (pH 6–9.2), and 0.8 mM PLP for 20 min at 37 °C in a 1.5 mL microcentrifuge tube. The coupling enzyme activities were added in ~100-fold excess to ensure that the reaction was rate limited by the racemase. It should be noted that D-ornithine decarboxylase shows no detectable activity with L-ornithine [12]. The reactions were stopped by addition of 250 μL of 0.1 M NaOH, and 250 μL of 40 mM DNBS. The mixture was incubated at 65 °C for 1 h, and then extracted with 750 μL of toluene. The absorbance of the toluene layer was read at 346 nm to determine the 1,6-bis(2,4-dinitrophenyl)putrescine formed, in a Cary100 or Thermo EvolutionPro UV/Vis spectrophotometer. Standards were run with putrescine under the same conditions. The kinetic parameters with L-ornithine were determined by fitting to Eq. 3, while the parameters with D-ornithine were determined by fitting to Eq. 4, where n is the Hill coefficient and S0.5 is the substrate concentration at half maximal activity.

$$\mathrm{v}={{\mathrm{V}}_{\max }}^{\ast }[\mathrm{S}]∕(K_{\mathrm{m}}+[\mathrm{S}])$$ (3)

$$\mathrm{v}={{\mathrm{V}}_{\max }}^{\ast }{[\mathrm{S}]}^{\mathrm{n}}∕({{\mathrm{S}}_{0.5}}^{\mathrm{n}}+{[\mathrm{S}]}^{\mathrm{n}})$$ (4)

2.4. Size-exclusion chromatography of OrnR

Size-exclusion chromatography was performed with an NGC system chromatograph from BioRad, using an ENrich SEC 650 10 × 300 mm column. The buffer was 0.05 M potassium phosphate, pH 7.0, at a flow rate of 1 mL/min, operating at 4 °C. Samples were 1 mg/mL in 0.05 M potassium phosphate, pH 7.0 (100 μL), and the absorbance was monitored at 280 nm.

2.5. Absorbance and fluorescence spectra of OrnR

Absorbance spectra were collected with a Thermo EvolutionPro UV–visible spectrophotometer. The enzyme was 1 mg/mL (18 μN) in 0.05 M potassium phosphate, pH 8.0. Fluorescence spectra were collected on an OLIS RSM-1000 spectrophotometer. The enzyme (3 μN) was in 0.05 M potassium phosphate, pH 8.0 at ambient temperature (~22 °C). Excitation was at either 290 or 420 nm, with 5 nm slits, and emission was collected with 5 nm slits. The data were collected with 150 data points, with 0.5 s integration.

2.6. NMR spectra of OrnR reactions

The ¹H NMR spectra were acquired using a Bruker Avance operating at 400 MHz. The samples contained 6 mg of D- or L-ornithine monohydrochloride, 0.2 mM PLP, in 600 μL of 0.1 M potassium phosphate, pD 8, with 0.6 to 1 mg of wild type or C164 A OrnR. Spectra were collected immediately after addition of enzyme, and daily for up to 1 week.

2.7. Bioinformatics of OrnR

The DNA sequence of STM2358 from S. typhimurium LT2 (NC_003197.2) was submitted to the EFI-EST website (https://efi.igb.illinois.edu/efi-est/) [13] using the UniProtKB database with the default E-value of 5 for the SSN Edge calculation, and a maximum of 5000 sequences. A sequence alignment score of 60 was applied to ensure complete sequences. The resulting file was opened with Cytoscape 3.10.3 to generate the sequence similarity network. The sequence alignment was then submitted to the EFI-GNT website (https://efi.igb.illinois.edu/efi-gnt/) to create the genetic neighborhood [14].

2.8. Structure and molecular dynamics of OrnR

An AlphaFold3 [15] model of OrnR is in the AF-database, as the monomer (AF-Q8ZNC6-F1-v6). The dimeric AlphaFold3 model of OrnR was created using the AlphaFoldServer (http://alphafoldserver.com) website from the translated amino acid sequence of STM2358 (Q8ZNC6), since Fold III PLP-dependent enzymes are known to form obligatory homodimers for activity [16], or as a homotetramer. The initial model obtained from AlphaFold3 did not contain the PLP cofactor. The model was aligned in COOT [17] with the structure of AlaR from Geobacillus stearothermophilus containing 5′-phosphopyridoxal-L-alanine bound in the active site. The AlaR ligand was copied and merged with the aligned structure of OrnR. The OrnR PLP-Ala complex was then edited in YASARA [18] to create the structures of PLP-L-ornithine and PLP-D-ornithine in the active sites. Hydrogens were then added to the complex, the protonation state was equilibrated at pH 7.4 or 9.0 with YASARA, and it was put into a cubic cell of water with 0.9 % NaCl, 10 Å larger than the protein. The parameters for bound PLP-L-ornithine and PLP-D-ornithine were generated by YASARA. The structure was energy minimized, and molecular dynamics were performed by YASARA, with the YAMBER3 force field, under periodic boundary conditions, at 1 atm and 298 K. Timesteps were 2.5 fs, and snapshots were collected at 100 ps, for a total of 15 to 18 ns. The figures showing structures were prepared with the open source version of PYMOL.

3. Results

3.1. Bioinformatics of OrnR

The amino acid sequence of STM OrnR was submitted to the EFI-EST web site [13] to calculate a sequence similarity network containing 945 sequences. The results show that the OrnR sequences can be separated into several clusters (Fig. 1A). STM OrnR is located in sequence cluster 1 (red circles), along with 493 other sequences. The previously studied C. sticklandii OrnR is located in a second large cluster of 201 sequences, which is linked to a number of smaller subclusters. We then examined the genomic context of the clusters (Fig. 1B) to determine the function of the OrnRs. The OrnRs in cluster 1 (red block in top 6 rows in Fig. 2B), primarily found in enterobacteria, are all found in a small operon, containing a DOKDC orthologue (blue/brown block), and amino acid transporter and permease homologues (pink and mauve blocks), as we found previously for S. typhimurium [3]. The physiological function of this pathway in these cells is unknown at present, but it is likely that it has a biosynthetic purpose, since the end product is putrescine. There is no evidence that this operon is induced under acidic conditions or is involved in the acid stress response, although there is an inducible L-ornithine decarboxylase in Salmonella that is part of the acid stress response [19]. The OrnRs in the other large cluster are found in the previously described operon for catabolic anaerobic fermentation of L-ornithine (red block, bottom 5 rows in Fig. 2B). The microorganisms in this cluster are mostly anaerobic Firmicutes, actinobacteria and thermotogae [2].

Fig. 1.

A. Sequence similarity networks for OrnR. B. Genomic context of OrnR. In enterobacteria, OrnR is shown in red, and DOKDC in blue/brown, while the pink and mauve blocks are amino acid transporter homologues. In Firmicutes and actinobacteria, OrnR (red block) is in the catabolic pathway in Fig. 1.

Fig. 2.

Sequence alignments of OrnR and AlaR. The conserved residues, including the active site lysine and arginines, have a yellow background and are indicated with asterisks. The catalytic cysteine of OrnR and tyrosine of AlaR are indicated by pound symbols. Residues in red are identical, and residues in blue are similar. C. sticklandiiOrnR, Genbank CAQ42981.1; C.difficileOrnR, CAJ67274.1; K.pneumoniaeOrnR, CAH3371893; STMOrnR, CAD5309703.1; C.freundiiOrnR, CAI9388532.1; C.difficileAlaR, KPI47058.1; GeobacillusAlaR, WP_020958452.1; Nereida_sp.OrnR, XHY13704.1; S.aureusOrnR, PPE54257.1.

The sequence alignment for selected OrnRs with AlaRs is shown in Fig. 2. Although the active site lysine and arginines in the PLP-binding site of the racemases are strictly conserved in all sequences, the essential catalytic tyrosine acid/base of AlaR, ArgR and LysR (Fig. 2, indicated by #) is notably missing in OrnRs. The overall sequence identity of OrnR with AlaR is low, only about 25 %. Furthermore, the sequences of OrnR in cluster 1 have >75 % identity with each other, but they have only ~40 % identity with the sequences in the catabolic OrnR cluster. However, all the OrnR sequences in both clades show a conserved cysteine residue, corresponding to Cys-164 in STM OrnR (Fig. 2, indicated by #). This cysteine is substituted by threonine in AlaR. The OrnR in Staphlococcus aureus contains the conserved cysteine, but it is distantly related to both these clades, and is involved in biosynthesis of the siderophore, staphyloferrin A [20]. Although a tyrosine residue immediately follows the conserved cysteine in the OrnR of S. typhimurium and related enterobacteria, it is not conserved in all the sequences of the catabolic OrnR, and it is replaced by phenylalanine, leucine, or isoleucine in about 10 % of the sequences. In S. aureus OrnR, the tyrosine is replaced by phenylalanine (Fig. 2, bottom row), and in Nereida sp. OrnR, it is substituted by leucine (Fig. 2, next to last row). In contrast, the cysteine is found in all 945 of the sequences identified in the sequence similarity network, including those in the smaller clusters in Fig. 1A. Thus, this tyrosine cannot be a catalytic residue in OrnR.

3.2. Structure of OrnR

The amino acid sequence of STM OrnR was submitted to AlphaFold3 [14] to predict the three-dimensional structure. Since the members of the AlaR family are known to form obligate dimers, with the active site created at the monomer-monomer interface [16], we initially instructed AlphaFold3 to construct the structure as a homodimer. The resulting structure exhibits the typical (β/α)₈ TIM-barrel core of the Fold III PLP-dependent enzymes (Fig. 3A). The alignment of the active sites of the predicted structure of STM OrnR (cyan) with the crystal structures of AlaR of Geobacillus stearothermophilus [9] (PDB ID: 1L6F, green) and Clostridium difficile [21] (PDB ID: 4LUS, yellow) is shown in Fig. 3B. The RMSD of the Cα backbone of chain A of OrnR with 1L6F is 5.11 Å and for 4LUS is 6.03 Å. This structure shows that the catalytic acid/base Tyr-265 of G. stearothermophilus AlaR is replaced in space with the Cys-164 in STM OrnR, as also seen in the sequence alignments in Fig. 2. The strictly conserved active site residues, Lys-36, Arg-129 and Arg-220, which are directly involved in PLP-binding, are present, but, surprisingly, the active site histidine, which π-stacks with the PLP pyridine ring in members of the Fold III family, is replaced by Asn-161 in STM OrnR (Fig. 3B), which aligns in the sequences with the histidine (Fig. 2). This is an unusual case of a PLP-dependent enzyme that apparently does not have an aromatic side chain π-stacking with the pyridine ring of the PLP. There is a phenylalanine residue in STM OrnR at position 162, but it is pointed away, about 10 Å from the PLP, in the predicted structure, and it is not conserved in OrnR sequences, since in cluster 2 sequences it is replaced by leucine.

Fig. 3.

A. Overlay of the AlphaFold3 predicted structure of OrnR (cyan) with the crystal structure of AlaR from Geobacillus thermophilus (PDB ID: 1L6F) (green) and Clostridiodes difficile (PDB ID: 4LUS) (yellow). B. Crossed-eye stereo view of the overlay of the active sites of OrnR (cyan) AlaR from Geobacillus thermophilus (PDB ID: 1L6F) (green) and Clostridiodes difficile (PDB ID: 4LUS) (yellow) The residue numbers are those of STM OrnR. The ligand in the AlaR structures is 5′-phosphopyridoxyl-L-alanine.

We performed size exclusion chromatography with wild-type and C164A OrnR to confirm the presence of the dimer (Fig. S2). The main peak for both wild type and mutant OrnR was observed at 13.8 mL, which has an estimated volume of 65 kDa (inset, Fig. S2), consistent with the predicted molecular weight of the dimer of 82 kDa. Surprisingly, a shoulder is seen for both wild type and C164A OrnR at 12.5 mL, which corresponds to a predicted molecular weight of 130 kDa (inset, Fig. S2), which is probably due to the tetramer, with molecular weight of 164 kDa. The relative amount of the dimer and tetramer is approximately 70:30 for both proteins. As a result of this finding, AlphaFold3 was used to model OrnR as a tetramer. The resulting tetrameric model shows interactions between a loop composed of residues 277–283 with the loop of residues 162–169 from another dimer (Fig. S3A). The 277–283 loop is absent from sequences of AlaR (Fig. 2). The dimer-dimer interface has a contact surface of 1068 Ų, compared to the monomer-monomer contact surface of 3134 A², as calculated by PISA [22]. Although the vast majority of AlaR family members form only dimers, these dimer-dimer contacts in OrnR are similar to those observed in the crystal structure of AlaR2 from Oenococcus oeni (PDH ID: 3HUR), which forms a tetramer at pH 3.5 (Fig. S3B).

3.3. Molecular dynamics of OrnR

We created models of the PLP external aldimine of L- and D-ornithine and attempted to dock them with YASARA into the active site of the OrnR AlphaFold3 model, but we were not successful, because the active site is blocked by the loop containing residues 277–283, possibly because AlphaFold3 predicted the structure of a closed conformation. We then manually modeled the PLP external aldimine of L- or D-ornithine into the two active sites of the model, guided by overlaying the structure of OrnR with G. stearothermophilus AlaR with 5′-phosphopyridoxyl-L-Ala bound (PDB ID: 1L6F) (Fig. 3B). The structure of the resulting OrnR-PLP-DL-ornithine complex, containing L-ornithine in one active site and D-ornithine in the other active site, was then submitted to steepest descent energy minimization, followed by 18 ns of molecular dynamics, with YASARA, using the YAMBER3 force field, at pH 7.4 and 9. The energy of the L-ornithine complex at pH 7.4 stabilized within 2 ns (Fig. S4 A), and there is no significant change in secondary structure (Fig. S4B). The RMSD compared to the initial structure also stabilized after 1 ns (Fig. S4C), and ligand movement continued for about 10 ns (Fig. S4D). The largest values of RMSF in both chains A and B are at residues 11, 173, 280 and 338 (Fig. S4E), indicating that these regions are highly mobile.

We next substituted Cys-164 in silico with alanine in the starting model at pH 7.4, and performed molecular dynamics on the mutant enzyme complexed with PLP-DL-ornithine (Fig. S5). The results are very similar to those with wild type OrnR. The energy stabilizes after about 1 ns (Fig. S5A), and there are negligible changes in secondary structure (Fig. S5B). The RMSD of the protein (Fig. S5C) and the ligands (Fig. S5D) also shows little change after 1 ns. Hence, we see very little change in the structure of wild-type and C164 OrnR after molecular dynamics (Fig. 4A), with a Cα backbone RMSD of 0.246 Å compared to the wild type model, showing that the mutation of Cys-164 to alanine does not significantly affect the global structure. The lowest energy structure of the PLP-L-Ornithine complex shows that the conserved Lys-36, Arg-129 and Arg-220 residues are located near the ligand (Fig. 4B). Similar to AlaR, Arg-220 donates a hydrogen bond from NE to N1 of the PLP, and from NH2 to OD1 of Asn-161. Similarly, ND1 of His-166 is located to accept a hydrogen bond from NH2 of Arg-219 in AlaR of G. stearothermophilus. Arg-129 donates a hydrogen bond from NH2 to O-3′ of the PLP and the carboxylate O of L-ornithine. Cys-164 is located with SG in position (~4 Å) to abstract the α-H of the L-ornithine external aldimine, and NZ of Lys-36 is 4 Å from the α-C of the bound L-ornithine on the opposite face, in a position to reprotonate the ornithine. The PLP phosphate accepts hydrogen bonds from NH of Gly-222, Gly-223, and Cys-201, as well as SG of Cys-201 and ND2 of Asn-200. It is interesting that there is no anionic residue side chain located near the δ-N of the L-ornithine in the complex, but after MD it is located in hydrogen bonding distance (3.1 Å) from the phosphate, and within 4 Å of Tyr-349, possibly allowing for a cation-π interaction (Fig. 4B). The homologous Tyr-354 in B. sterearothermophilus AlaR is a hydrogen bond donor to the phosphate [9], but we find that Tyr-349 in OrnR is about 4 Å from the closest phosphate oxygen. In the other chain, the structure of the PLP-D-ornithine complex is similar, but the carboxylate of the D-ornithine is pointing above the PLP ring plane, while the carboxylate of the L-ornithine is nearly planar with the PLP (Fig. 4C). Arg-219 is not in a position to donate a hydrogen bond to the carboxylate of the bound D-ornithine. At pH 7.4, Cys-164 is probably protonated, and at pH 9 it is deprotonated, so we performed molecular dynamics at pH 9 to see the effect of ionization of Cys-164 on the complex structure (Fig. S6). However, there was no significant difference in the structures after molecular dynamics, with a Cα RMSD of 0.308 between the final structures at pH 7.4 and 9.

Fig. 4.

A. OrnR complexed with the external aldimine of PLP with L- and D-ornithine after 18 ns of molecular dynamics. Green, wild-type; cyan, C164A. B. Crossed-eye stereo view of the active site of OrnR and C164A OrnR complexed with the external aldimine of PLP with L-ornithine. C. Crossed-eye stereo view of the active site of OrnR and C164A OrnR complexed with the external aldimine of PLP with D-ornithine. The dashed lines indicate potential hydrogen bonds for wild-type OrnR.

We next performed molecular dynamics on the tetrameric structure of OrnR with L- and D-ornithine bound to the PLP at pH 7.4 (Fig. S7). As with the dimer, the energy stabilized rapidly, and the complex was stable for the 18 ns simulation (Fig. S7A). There is no change in the secondary structure content (Fig. 7B). The RMSD of the protein and ligands is stable (Fig. 7C, D), and the RMSF of the residues is similar to the dimer, with the loop containing residue 280 showing high mobility (Fig. 7E).

Fig. 7.

¹H NMR spectra of OrnR reactions. A. Reaction of wild-type OrnR with L-ornithine. Black, spectrum of L-orntihine without OrnR. Red, spectrum of L-ornithine incubated with wild-type OrnR for 1 day. B. Reaction of C164A OrnR with L- and D-ornithine. Black, L-ornithine without OrnR. Red, spectrum of L-ornithine incubated with C164A OrnR for 5 days. Blue, spectrum of D-ornithine incubated with C164A OrnR for 5 days.

3.4. Reaction of OrnR

In the present work, we developed a new sensitive quantitative assay of OrnR, by coupling it with D-ornithine decarboxylase in the L → D direction and with L-ornithine decarboxylase in the D → L direction. We used sodium 2,4-dinitrobenzenesulfonate (DNBS) to derivatize the resulting putrescine formed, measuring the 1,6-bis-(2,4-dinitrophenyl) putrescine at 346 nm after toluene extraction. The full details of this method will be published in full elsewhere (Phillips et al., manuscript in preparation). The pH dependence of the reaction in the L → D direction shows a maximum at pH 9.2 (Fig. S8). The reaction at pH 9.2 and 37 °C shows hyperbolic kinetics in the L → D direction (black circles, Fig. 5), with k_cat = 0.21 ± 0.01 s⁻¹, K_m = 5.6 ± 0.7 mM, and k_cat/K_m = 37.5 ± 3.2 M⁻¹ s⁻¹. Remarkably, the catabolic OrnR from C. sticklandii has k_cat = 980 s⁻¹ and k_cat/K_m = 1.3 × 10⁶ M⁻¹ s⁻¹ in the L → D direction [1]. In contrast, STM OrnR shows sigmoid kinetics in the D → L direction, with k_cat = 0.87 s⁻¹, S_0.5 = 12.8 ± 2.7 mM, and n = 1.77 ± 0.25. Although these catalytic parameters are rather low for STM OrnR, they are comparable to those of some other racemases [23]. The cooperativity suggests that D-ornithine may bind to an allosteric site, and it is also possible that an allosteric ligand may activate the enzyme in vivo. This cooperative behavior of D-ornithine suggests that the enzyme has two conformations, which preferentially react with either L- or D-ornithine, with the equilibrium of the free enzyme strongly favoring the L-reactive form of the enzyme, since at concentrations below 3 mM L-ornithine reacts faster (Fig. 5). We note that it has been proposed that proline racemase, which is not a PLP-dependent enzyme, exhibits two conformations, specific for the enantiomers of proline [24], but it does not show cooperativity. It is not possible to measure K_m directly in the D → L direction for OrnR because of the cooperativity; however, since the Haldane relationship demands that k_cat/K_m be equal in both directions for a racemase, we can estimate that the K_m for D-ornithine must be about 23 mM. Under these reaction conditions, we could not detect significant activity above background with C164A OrnR with the DNBS assay. Based on the detection limits of the DNBS assay, we can estimate that the C164A variant of OrnR has activity reduced at least 300-fold.

Fig. 5.

Kinetics of OrnR reaction with L-ornithine (black circles) and D-ornithine (red circles) at pH 9.2. The lines are the result of fitting with the parameters given in the text. The same amount of OrnR (0.6 μM) was used in both assays.

3.5. UV–visible and fluorescence spectra of OrnR

The UV–visible absorption spectrum of OrnR exhibits peaks at 278 nm, from the protein, and 418 nm, with a shoulder at about 330 nm, arising from the bound PLP (Fig. 6A). The ratio of the 278 nm to 418 nm absorbance is 6.6. Since the calculated monomeric absorption coefficient of OrnR is 32,890 M⁻¹ cm⁻¹, based on the amino acid composition, this gives an estimated absorption coefficient of 4980 M⁻¹ cm⁻¹ for the 418 nm band. This is in excellent agreement with published values of ε for PLP aldimines of 5000–6000 [25], showing that the enzyme contains approximately 1 mol PLP per subunit. C164A OrnR shows peaks at 330 and 392 nm from the bound PLP.

Fig. 6.

UV–visible absorption and fluroescence spectra of wild type OrnR. A. UV–visible absorption spectrum of OrnR (18 μN) in 0.05 M potassium phosphate, pH 8.0. B. Fluorescence emission spectrum of OrnR (3 μN) is 0.05 M potassium phosphate, pH 8.0, with excitation at 290 nm. C. Fluorescence excitation spectrum of OrnR (3 μN) in 0.05 M potassium phosphate, pH 8.0, with emission at 500 nm.

On excitation of the protein band at 290 nm, the fluorescence emission spectrum of wild-type OrnR exhibits peaks at 340 nm, from the tryptophan, and 500 nm, from the PLP (Fig. 6B). The tryptophan emission peak at 340 nm indicates that the tryptophans are partially exposed to solvent [26]. The 500 nm emission band indicates that Förster resonance energy transfer (FRET) is occurring between the tryptophan and

PLP. This suggests that tryptophan residues must be close to the bound PLP, since FRET efficiency drops rapidly with distance with 1/r⁶. The reported critical distance for FRET from tryptophan to PLP in glycogen phosphorylase is 21 Å [27]. To clarify this, we collected the excitation spectrum, with emission set at 500 nm, shown in Fig. 6C. The excitation spectrum shows peaks at 275, 330 and 420 nm. The 330 nm excitation peak overlaps well with the tryptophan emission, thus accounting for the FRET. Since excitation of the peak at 330 nm shows emission at 500 nm, this band is likely due to an enolimine tautomer of the internal aldimine (Scheme 1), rather than pyridoxamine, which absorbs at 325 nm and emits at about 400 nm. It is known that PLP enolimines undergo rapid excited state proton transfer to give the ketoenamine prior to emission [28].

Scheme 1.

Tautomerism of PLP aldimines.

3.6. NMR spectra of OrnR reactions

Racemization of amino acids in D₂O can be followed by ¹H NMR spectroscopy, since the α-CH exchanges with the solvent [29]. Previously, we used this technique to follow the reaction of wild-type OrnR by ¹H NMR [3]. We have now used ¹H NMR to determine if C164A OrnR has ability to exchange the α-CH of L- and D-ornithine (Fig. 7). In Fig. 7A, the reaction of wild-type OrnR with L-ornithine results in a reduction in the intensity of the triplet at 3.75 ppm arising from the α-CH after 1 day. The β-CH₂ at 1.8 ppm also collapses to a triplet with the exchange of the α-CH, but the γ-CH₂ multiplet at about 1.6 ppm and the δ-CH₂ triplet at 3 ppm are unchanged. Integration of the wild type spectrum shows that the α-CH resonance at 3.75 ppm is about 80 % exchanged. In contrast, the incubation of C164A OrnR with either L- or D-ornithine shows no detectable exchange even when incubated for 5 days (Fig. 7B). The multiplet in the spectra centered at ~3.5 ppm comes from glycerol, added to stabilize the enzyme for freezing.

4. Discussion

4.1. Structure of OrnR

The AlphaFold3 predicted structure of OrnR is a homodimer of two PLP-dependent Fold III subunits, similar to AlaR (Fig. 3). The PLP is bound in the N-terminal (βα)₈-barrel, and the C-terminal 7-stranded β-sheet contains one of the mobile loops, residues 277–283, which extends over the active site of the other subunit. The active site residues are highly conserved with AlaR, except for the replacement of the catalytic tyrosine in AlaR, and other racemases, with a cysteine, which is apparently unique to OrnR. The catalytic cysteine is located on a different structural loop than the other racemases in the AlaR family, suggesting OrnR may have evolved independently by convergent evolution. Molecular dynamics of the predicted OrnR structure complexed with PLP-L-ornithine showed that the complex structure is stable, and the lowest energy structure has the acid/base catalysts, Lys-36 and Cys-164, in position about 4 Å above and below the Cα. The differences in the kinetics of reaction of L- and D-ornithine (Fig. 5) suggest that different conformations react with the enantiomeric substrates. We do not see evidence for different conformations of the protein from the molecular dynamics, but the PLP-D-ornithine ligand structure does have a significantly different conformation than the PLP-L-ornithine structure (Fig. 4B, C). OrnR contains three tryptophan residues, 133, 212, and 363, and in the model, they are located in partially exposed positions near the surface (Fig. S9). This is consistent with the observed fluorescence emission at 340 nm (Fig. 6B). Furthermore, Trp-212 is located only 13 Å from the PLP, ideally located for FRET (Fig. S9).

The SEC chromatogram of wild type and C164A OrnR show evidence of a dimer-tetramer equilibrium (Fig. S2), which is very rare in Fold III PLP enzymes. The tetramer was modeled by AlphaFold3 and subjected to molecular dynamics, showing a stable complex. The dimer-dimer contacts are with the mobile loop of residues 277–283, which interacts with the loop containing the Cys-164. Remarkably, the 277–283 loop contacts the other monomer in the dimer, but in the tetramer, this loop comes from the other dimer (Fig. S3A). However, there is at least one example of AlaR forming a tetramer (PDB ID: 3HUR) (Fig. S3B).

4.2. Mechanism of OrnR

The mechanism of OrnR can proceed through either a stepwise mechanism, with a carbanionic quinonoid intermediate, or a single step concerted mechanism. AlaR is proposed to use a stepwise mechanism, based on observation of a peak at 510 nm in the visible spectrum of the alanine complex of the R219E mutant [30], suggesting a quinonoid intermediate. Furthermore, AlaR catalyzes the elimination of halides from β-haloalanines and 3-fluorovinylglycine, as expected if there is a carbanionic intermediate [31,32]. In contrast, we observe no measurable exchange of the α-hydrogen of either D- or L-ornithine during incubation with C164A OrnR in D₂O (Fig. 7B). If the reaction proceeded by a stepwise mechanism with a quinonoid intermediate, D-ornithine could still undergo α-hydrogen exchange and/or transamination catalyzed by C164A OrnR, since the catalytic base that removes the α-proton on the D-face, Lys-36, is still present. In this regard, Y265A and Y265S mutants of AlaR were found to catalyze transamination of D-alanine, but not L-alanine, consistent with a stepwise mechanism [33]. In contrast, extended incubation, up to 18 h, of wild type or C164A OrnR with D- or L-ornithine did not show any evidence for transamination products by UV–visible spectroscopy (data not shown). Thus, the proposed concerted mechanism for OrnR is shown in Scheme 2. The change from a stepwise mechanism to a concerted mechanism may be the result of changing the tyrosine acid/base catalyst to a cysteine. We note that the pyridine nitrogen of the PLP is not protonated in Fold III enzymes, so the quinonoid intermediate is destabilized compared to Fold I PLP dependent enzymes. It should also be noted that a concerted proton transfer mechanism for OrnR may still be asynchronous, but without a discrete quinonoid intermediate. In fact, QM/MM calculations of the reaction of proline racemase are consistent with an asynchronous concerted mechanism [34].

Scheme 2.

Proposed mechanism of OrnR.

4.3. Conclusions

Unique among the members of the AlaR superfamily, OrnR uses a cysteine residue as an acid/base catalyst for the racemization reaction. Furthermore, the catalytic reaction appears to proceed by a concerted proton transfer, in contrast to AlaR, which uses a stepwise mechanism with a transient quinonoid intermediate. We also see evidence for a dimer-tetramer equilibrium in OrnR.

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.bbapap.2025.141120.

Data availability

Data will be made available on request.