In the Absence of RidA, Endogenous 2-Aminoacrylate Inactivates Alanine Racemases by Modifying the Pyridoxal 5′-Phosphate Cofactor

Abstract

Members of the RidA (YjgF/YER057c/UK114) protein family are broadly conserved across the domains of life. In vitro, these proteins deaminate 3- or 4-carbon enamines that are generated as mechanistic intermediates of pyridoxal 5′-phosphate (PLP)-dependent serine/threonine dehydratases. The three-carbon enamine 2-aminoacrylate can inactivate some enzymes by forming a covalent adduct via a mechanism that has been well characterized in vitro. The biochemical activity of RidA suggested that the phenotypes of ridA mutant strains were caused by the accumulation of reactive enamine metabolites. The data herein show that in ridA mutant strains of Salmonella enterica, a stable 2-aminoacrylate (2-AA)/PLP adduct forms on the biosynthetic alanine racemase, Alr, indicating the presence of 2-aminoacrylate in vivo. This study confirms the deleterious effect of 2-aminoacrylate generated by metabolic enzymes and emphasizes the need for RidA to quench this reactive metabolite.

INTRODUCTION

The RidA (previously YjgF/YER057c/UK114) family of proteins is highly conserved and widely distributed in all domains of life (1–3). Past studies with Salmonella enterica and other organisms identified phenotypes that resulted from loss of RidA (1, 3–7). The diversity of phenotypes made it difficult to assign a unifying role for RidA. Results from numerous genetic and biochemical analyses led us to suggest that the role of RidA was to eliminate reactive metabolites that could inhibit key enzymes or catalyze anomalous chemical reactions (5, 6, 8). This model was bolstered by the findings that in ridA strains of S. enterica, (i) the activity of transaminase B (IlvE) was decreased by a posttranslational mechanism (5) that depended on the dehydration of serine by threonine dehydratase IlvA (9) and (ii) 2-aminocrotonate derived from threonine facilitated anomalous synthesis of the thiamine intermediate phosphoribosylamine by anthranilate phosphoribosyl transferase (TrpD) (10, 11).

RidA protein family members deaminate the 3- and 4-carbon enamines 2-aminoacrylate (2-AA) and 2-aminocrotonate, respectively, in vitro (2). These enamines are intermediates in the pyridoxal 5′-phosphate (PLP)-dependent mechanism of serine/threonine dehydratases (e.g., IlvA). The metabolites are rapidly deaminated to ketoacids in aqueous solution, even in the absence of RidA (12). The instability of enamine metabolites and the high rate with which they are hydrated nonenzymatically to product appeared to negate a need for the enamine deaminase activity of the conserved RidA protein family in the cell.

The in vitro demonstration that RidA had deaminase activity suggested that enamine intermediates produced by cellular enzymes (e.g., serine/threonine deaminases) might be relevant in the cellular environment, despite their rapid hydrolysis in aqueous solution. Furthermore, if the role of RidA was to quench reactive enamines, cells lacking RidA would accumulate enamines. This hypothesis was recently supported with the in vitro demonstration that the activity of threonine dehydratase (IlvA), using serine as the substrate, inactivated transaminase B (IlvE) in the absence of RidA (9). The same study showed that a modified IlvE could be isolated from a ridA mutant background but not from a wild-type strain. These data confirmed the role of RidA in protecting enzymes from posttranslational modification and identified an enzyme responsible for generating a relevant reactive metabolite (9).

Certain β-substituted amino acids (e.g., serine O-sulfate and β-chloroalanine) can act as suicide substrates for certain PLP enzymes. In the active sites of relevant enzymes, these substrates are enzymatically converted to the RidA substrate, 2-AA. Once in the active site, 2-AA can perform a nucleophilic attack and covalently modify the active-site PLP cofactor, thus rendering the enzyme inactive (13–18). Two mechanisms for inactivation have been described: (i) a 2-AA/PLP adduct formed as an external aldimine can be attacked by a nonlysine nucleophilic residue in the active site (15, 16, 19–22), or ii) the 2-AA attacks the PLP bound to the active-site lysine, generating a 2-AA/PLP adduct that can be recovered as a pyruvate/PLP adduct after denaturing the protein (Fig. 1) (14, 17, 23). The reactivity of 2-AA characterized in these in vitro studies suggested that 2-AA could be a relevant reactive metabolite in strains lacking RidA.

Fig 1.

Mechanism of 2-aminoacrylate inhibition in Alr. 2-Aminoacrylate (e.g., generated from β-chloroalanine) can inhibit the alanine racemases DadX and Alr by attacking the internal aldimine formed by the PLP cofactor bound to the active-site lysine. The resulting enzyme is covalently modified and inactive due to the 2-aminoacrylate-modified PLP (2-AA/PLP). Upon base denaturation of the 2-aminoacrylate-modified enzyme, an adduct of pyridoxal 5′-phosphate and pyruvate (pyruvate/PLP) is released (18).

This study was initiated to determine if 2-AA inactivation of PLP enzymes occurred in vivo, despite the short half-life of this enamine. To test this hypothesis, we chose to investigate alanine racemases in S. enterica, with an emphasis on the biosynthetic enzyme Alr. Inactivation of Alr by aminoacrylate via suicide substrates has been studied thoroughly in vitro and occurred by the second mechanism described above (18). The data herein showed that in the in the absence of RidA there is aminoacrylate-mediated inactivation of alanine racemase. To our knowledge, this is the first report of the isolation of a PLP adduct formed by endogenously generated 2-aminoacrylate, and it supports the hypothesis that the RidA family of proteins attenuates this enamine stress.

MATERIALS AND METHODS

Bacterial strains, media, and chemicals.

The strains used in this study are derivatives of Salmonella enterica serovar Typhimurium LT2, and the genotypes are shown in Table 1. MudJ refers to the Mud1734 transposon (24). Insertion-deletion mutations of the dadX and alr genes were generated by recombination methods described previously, whereby the gene was replaced by the cat⁺ gene or kan⁺ from plasmid pKD3 or pKD4, respectively (25). A high-frequency general transducing mutant of bacteriophage P22 was used for transductional crosses required for strain construction (26). Plasmid pJF2 is a derivative of pBAD24 (27) and contains a gene fusion of alr and the gene encoding the self-cleaving intein chitin affinity tag. This fusion protein was functional and supported alanine racemase activity that could be measured prior to cleaving the tag.

Table 1.

Bacterial strains

Strain	Genotypea
DM3480	ridA3::MudJ
DM9404	Wild type
DM13674	Wild type/pJF2
DM13675	ridA3::MudJ/pJF2
DM13756	dadX121::cat ridA3::MudJ
DM13760	dadX121::cat
DM14178	alr-51::cat
DM14179	alr-51::cat ridA3::MudJ
DM14180	dadX121::cat alr-52::kan

^aMudJ refers to the Mud1734 transposon (24). The plasmids are derivatives of pBAD24 (27).

Minimal medium was no-carbon E (NCE) or no-carbon no-nitrogen (NCN) medium supplemented with 1 mM MgSO₄ (28) and 22 mM d-glucose. In the case of NCN, the nitrogen source was as indicated. Difco nutrient broth (NB) (8 g/liter) with NaCl (5 g/liter) was used as a rich medium. Difco BiTek agar was added (15 g/liter) for solid medium. When required for plasmid maintenance, ampicillin was added to minimal and nutrient media at 15 and 150 mg/liter, respectively. Unless noted, all chemicals and enzymes were purchased from Sigma-Aldrich Co. (St. Louis, MO).

Construction of pJF2.

The alr gene was amplified from S. enterica LT2 with primers JMF64 ( 5′-TAGCATATGCAAGCGGCAACAGTCGT-3′ ) and JMF65 ( 5′-TAGCTCGAGATCAATATACTTCATCGCCACCCT-3′ ) using Herculase II polymerase (Agilent Tech.). Following digestion with NdeI and XhoI, the gene fragment was cloned into pTYB2 (Impact kit; New England BioLabs) to create pJF1, a plasmid containing a gene fusion of alr and the gene encoding the self-cleaving intein chitin affinity tag. Plasmid pJF1 was cut with XbaI and PstI to excise the fragment containing alr and the affinity tag. The resulting fragment was cloned into plasmid pBAD24 (27) cut with NheI and PstI to create pJF2. The final construct was verified by sequencing the ligation junctions.

Alanine racemase assay. (i) Crude extract.

The assay for alanine racemase activity was adapted from an established protocol (29). Briefly, 5-ml cultures were grown in minimal medium to an optical density at 650 nm (OD₆₅₀) of 0.8 and harvested by centrifugation (8,000 × g, 15 min). Cells pellets were resuspended in 0.9 ml of NaCl (145 mM) and permeabilized by the addition of Popculture reagent (Novagen) to a final concentration of 2% (vol/vol). The alanine racemase assay mixture contained 10 to 810 μl of permeabilized cells and glycine buffer (50 mM, pH 9.0) in a total volume of 1 ml; l-alanine was added to a final concentration of 100 mM to initiate the reaction. The mixture was incubated at 37°C, and samples were removed at 0 and 30 min and heated to 85°C for 5 min to terminate reactions. The resulting d-alanine concentrations were determined by conversion to pyruvate using d-amino acid oxidase (Sigma-Aldrich). The reaction mixture for d-alanine determination contained 900 μl of inactivated sample as described above, d-amino acid oxidase (0.24 U), FAD⁺ (2.2 μM), bovine liver catalase (10 U) (Sigma-Aldrich), and glycine buffer (50 mM, pH 9.0) in a total volume of 1 ml. The reaction mixtures were continuously shaken and incubated at 37°C for 1 h. To the 1-ml reaction mixture, 250 μl of 0.1% (wt/vol) dinitrophenol hydrazine in 2 M HCl was added, and the reaction was allowed to proceed at room temperature for 15 min, after which 1 ml of NaOH (1.5 M) was added. After a brief incubation at room temperature, absorbance at 520 nm was measured. A time course determined that d-alanine concentrations increased linearly over 60 min. Absolute activity values varied between experiments done on different days, but the ratio between strains was consistent. Cells were solubilized in sodium hydroxide (1.5 M), and the bicinchoninic acid (BCA) protein assay (Thermo Scientific Pierce) was used to determine total cell protein versus bovine serum albumin. Assays results were normalized to total cell protein. Data given are the means and standard deviations of three biological replicates for strains DM3480, DM9404, DM13756, DM13760, DM14178, DM14179, and DM14180 and of two biological replicates for strains DM13674 and DM13675.

(ii) Purified protein.

Racemization of d-alanine to l-alanine by pure protein was assayed by modifying a previously described protocol (23). Briefly, the assay mixture contained 50 mM glycine (pH 9), 1.2 mM NAD⁺, 20 mM d-alanine, 0.1 unit Bacillus subtilis l-alanine dehydrogenase (Sigma-Aldrich) and approximately 0.4 μg of Alr in a total volume of 105 μl. Reduction of NAD⁺ was monitored by the increase in absorbance at 340 nm. The extinction coefficient for NADH in 50 mM glycine (pH 9.0) was determined experimentally (ε = 4,000 absorbance units [AU] M⁻¹ cm⁻¹) and used to calculate the rate of l-alanine production. Reactions were performed at room temperature, and the specific activities are reported as μmol alanine per min per mg. The BCA protein assay (Thermo Scientific Pierce) was used to measure protein concentration using bovine serum albumin as a reference.

Protein purification.

Two liters of fresh minimal medium was inoculated with 50 ml of overnight cultures of strains containing pJF2 grown on minimal medium with shaking at 37°C. When the cultures reached an OD₆₅₀ of 0.5, l-(+)-arabinose was added to a final concentration of 0.2% (wt/vol) to induce alr expression. Cells were harvested by centrifugation (15 min at 9,000 × g) when the OD₆₅₀ of the culture was between 2 and 2.5; the resulting cell pellets were frozen at −80°C. Pellets were resuspended in HEPES buffer (20 mM, pH 8.5) containing sodium chloride (100 mM), and cells were broken with a French pressure cell (2 passes at 10,342 kPa). After clarification by centrifugation (45 min at 48,000 × g), the supernatant was loaded onto chitin resin (5 ml), and protein purification proceeded as per the manufacturer’s instruction (Impact; New England BioLabs). After elution, protein was concentrated and dialyzed into 30 mM potassium phosphate buffer, pH 8.3. Yields were approximately 1 mg Alr per liter of starting culture.

Characterization of cofactor content.

Cofactors bound to Alr were released from the protein as described elsewhere (17). Briefly, KOH (30 mM final concentration) was added to purified Alr (∼2 mg/ml in 30 mM potassium phosphate buffer, pH 8.3) and incubated at room temperature for 10 min. The protein was precipitated by adding potassium phosphate buffer (500 mM, pH 7.0) until the solution reached neutral pH or, alternately, by adding 10% trifluoroacetic acid until a protein precipitate was formed. The precipitated protein was removed by centrifugation (3 min at 16,000 × g), and the cofactor content of the supernatant was analyzed by high-performance liquid chromatography (HPLC). The separations were performed on a Shimadzu HPLC equipped with a Luna 5μ C₁₈ column (250 by 4.60 mm; 5 μm) (Phenomenex) using a 2-step isocratic method with a flow rate of 0.8 ml/min as follows: 0 to 5 min with 100% buffer A (0.06% [vol/vol] trifluoroacetic acid) and 5 to 18 min with methanol-buffer A (3:97). Between each run, the column was washed for 10 min with methanol-buffer A (60:40). The eluant was monitored at 305 nm using a photodiode array detector (Shimadzu SPD-M20A). Areas under the peaks were integrated using Shimadzu LabSolutions software (version 5.42). Authentic pyridoxal 5′-phosphate (>98% pure; Sigma-Aldrich) and pyruvate/PLP served as standards. Pyruvate/PLP was synthesized as described previously (30), purified using HPLC, and lyophilized. H¹ nuclear magnetic resonance (NMR) was performed by the National Magnetic Resonance Facility in Madison, WI, on a 600.14-MHz Bruker DMX NMR spectrometer. The H¹ NMR spectrum of the synthesized compound contained the diagnostic peaks of the previously published spectrum for pyruvate/PLP (17) (H₂O-D₂O [80:20] at pH 3.5,): δ 2.624 (s), 5.037 (d, J = 7.8 Hz), 7.525 (dd, J = 238.8 Hz, J′ = 16.8 Hz), 8.075 (s). 4,4-Dimethyl-4-silapentane-1-sulfonic acid (DSS) was used as an internal standard. The previously published extinction coefficient (A_max = 408 nm, ε = 8,000 cm⁻¹ M⁻¹) of pyruvate/PLP was used to determine the concentration at pH 7.0 (23). Data given are triplicate measurements of a representative purification represented as mean ± standard deviation.

MS of cofactor.

Cofactors were released and separated as described above. Eluant corresponding to pyruvate/PLP was collected and twice lyophilized to remove trifluoroacetic acid. Samples were analyzed at the University of Wisconsin School of Pharmacy Mass Spectrometry Facility. Samples were diluted in 50% acetonitrile and analyzed using electrospray ionization (ESI) infusion mass spectrometry (MS) in the negative-ion mode with a Bruker Maxis 4G UHR-ToF mass spectrometer.

RESULTS AND DISCUSSION

ridA mutant strains are compromised in l-alanine utilization.

Bacterial alanine racemases are PLP-containing enzymes that catalyze the racemization of l-alanine to d-alanine and vice versa by the mechanism shown in Fig. 2. Like that of Escherichia coli, the S. enterica genome encodes both a biosynthetic alanine racemase and a catabolic alanine racemase, Alr and DadX, respectively (31). Although the proteins are homologous, their roles in vivo are distinct due to the differing regulation of their expression. The former (Alr) is required to convert l-alanine to d-alanine, an essential precursor of the peptidoglycan layer in the cell wall, and is constitutively expressed at a low level (32). In contrast, l-alanine can be used by S. enterica as the sole nitrogen source by a pathway that requires the catabolic alanine racemase, DadX (Fig. 2).

Fig 2.

Catalytic mechanism of alanine racemase and pathway for l-alanine utilization. Both Alr and DadX catalyze the reversible racemization between d- and l-alanine. The catalytic mechanism of PLP-dependent alanine racemases is illustrated here and has four states: I, PLP cofactor is bound to the enzyme via an internal aldimine bond; II, l-alanine attacks the aldimine linkage, displacing the lysine, and forms an external aldimine; III, the abstraction of the α-proton results in a quinonoid intermediate; and IV, the addition of the α-proton on the opposite side of the molecule results in the formation of d-alanine external aldimine which can be released to obtain structure I again. In l-alanine catabolism, l-alanine is racemized to d-alanine via the catabolic alanine racemase, DadX. d-Alanine is then oxidized via d-amino acid dehydrogenase, releasing ammonia and pyruvate.

Growth curves were determined using l-alanine as the sole nitrogen source (Fig. 3A). As expected, the growth of strains containing a deletion of alr was similar to that of wild-type strains, while strains containing a deletion of dadX were unable to grow. ridA mutant strains grew poorly using l-alanine as a nitrogen source, displaying a rate and final density that were midway between those of the wild type and a dadX mutant. Each of the strains was proficient at growth when d-alanine was provided as the sole nitrogen source, a condition that circumvented the requirement for racemase activity (Fig. 3B). These results suggested that alanine racemase activity contributed by DadX was compromised in a ridA strain, but they did not address an impact of RidA on the activity of Alr.

Fig 3.

ridA mutants are compromised in the utilization of l-alanine as a nitrogen source. Growth curves were determined by monitoring optical density (OD) at 650 nm. The indicated strains were grown in minimal nitrogen-free medium (NCN) supplemented with l-alanine (5 mM) (A), d-alanine (5 mM) (B), or l-alanine (5 mM) with isoleucine (0.3 mM) (C). Data represent growth curves determined in triplicate; values are means ± standard deviations.

Alanine racemase activity is compromised in ridA mutant strains.

Several relevant strains were grown in minimal medium, the cells were permeabilized, and alanine racemase activity was assayed. The effect that the lack of RidA had on each isozyme was addressed by assaying isogenic strains expressing one or the other racemase (Table 2). In strains with a functional DadX and lacking RidA, alanine racemase activity was ∼70% of that measured in an isogenic strain with a functional RidA. These data allowed the conclusion that DadX was compromised in the absence of RidA. In the strains lacking DadX, the low level of racemase activity contributed by Alr made it difficult to address whether Alr was similarly compromised in the absence of RidA.

Table 2.

Alanine racemase activity in permeabilized cells

Strain	Relevant genotype	Alanine racemase activity
		Sp act (nmol d-Ala min−1 mg−1)a	% of wild-type activity
DM9404	Wild type	61.6 ± 7.4
DM3480	ridA	42.7 ± 2.5	69
DM14178	alr	66.0 ± 1.5	107
DM13760	dadX	2.3 ± 0.2	3.7
DM14179	alr ridA	41.0 ± 1.6	66
DM13756	dadX ridA	2.1 ± 0.3	3.5
DM14180	alr dadX	1.2 ± 0.2	2.0

^aAlanine racemase activity in permeabilized cells was assayed as described in the Materials and Methods. Shown are averages and standard deviations for biological triplicates.

Alr was expressed as an intein-chitin binding domain fusion to overcome the difficulties associated with the low level of activity from chromosomally encoded Alr. When the fusion protein was overexpressed, alanine racemase activity in a ridA strain was 62% of that found in the wild-type strain (371 ± 28 and 230 ± 36 nmol d-alanine min⁻¹ mg⁻¹, respectively). The difference between the racemase activities in a ridA strain and a wild-type strain indicated that the status of RidA affected the activity of Alr in addition to that of DadX.

Isoleucine addition corrects the growth defect of RidA-deficient strains.

Previous work showed that an activity of the biosynthetic serine/threonine dehydratase IlvA caused partial inactivation of the branched-chain transaminase IlvE in a ridA strain (5). To address whether IlvA activity was involved in modulating the activity of alanine racemase in the absence of RidA, isoleucine was added to the medium. The addition of isoleucine corrected the growth defect of a ridA strain using l-alanine as a nitrogen source but had no effect on the growth of a dadX strain (Fig. 3C). Isoleucine decreases IlvA activity through a feedback inhibition mechanism, and this result suggested that IlvA activity was required for inactivation of the alanine racemases in strains lacking RidA.

Alr protein purified from ridA cells is modified.

The above data were consistent with a posttranslational modification, derived from an intermediate in the dehydratase reaction of IlvA, contributing to the decreased alanine racemase activity in a ridA strain. In one scenario, the accumulation of endogenous 2-AA in ridA strains would inactivate Alr via the 2-AA/PLP adduct previously characterized in vitro (Fig. 1). If this was the case, a pyruvate/PLP adduct would be released from Alr when the protein was denatured (18, 23). Alr was purified to >95% homogeneity from ridA⁺ and ridA strains grown in minimal medium. The two Alr protein samples, and an additional control sample, were denatured, and the released cofactor(s) was separated by HPLC. The protein samples were (i) Alr purified from a wild-type (ridA⁺) genetic background, (ii) Alr purified from a wild-type (ridA⁺) background and partially inactivated by incubation with β-chloroalanine, and (iii) Alr purified from a ridA strain. The cofactor content of each protein sample is depicted in the HPLC chromatograms shown in Fig. 4. When monitored at 305 nm, two peaks were observed. The first, larger peak had a retention time of 10.3 min and was present in all three samples. The second peak was significantly smaller, had a retention time of 14.2 min, and was present only in the sample purified from a ridA strain and the sample treated with β-chloroalanine. The sample derived from the wild-type strain had no detectable absorbance (≤500 AU) at 305 nm between 13.5 and 14.5 min.

Fig 4.

Alr isolated from ridA strain contains multiple cofactors. (A to C) Protein isolated from a wild type strain (A), wild-type protein partially treated with β-chloroalanine (B), and protein isolated from a ridA mutant (C) were denatured to release the cofactor, and protein-free supernatant was analyzed by HPLC. (D and E) The UV-visible spectra of authentic PLP (D) and pyruvate/PLP (E) were compared to spectra from the peaks of ridA mutant-isolated Alr at pH 2. Solid lines, spectra from authentic species; dashed lines, ridA mutant-isolated spectra.

The similarity between the chromatograms of the sample treated with β-chloroalanine and the sample from a ridA strain suggested that the second peak was the pyruvate/PLP adduct derived from 2-AA/PLP during base denaturation of the protein (Fig. 1) (17, 18, 23). Retention time, coinjection of authentic compounds, and UV-visible spectra (Fig. 4D and E) suggested that the peak eluting at 10.3 min was PLP and the peak eluting at 14.2 min was a pyruvate/PLP adduct. This assignment was confirmed by mass spectrometry (MS). The eluant from the peak at 14.2 min in both the β-chloroalanine-treated Alr and ridA-isolated Alr contained the MS signatures of the synthesized adduct (Fig. 5).

Fig 5.

Alr isolated from a ridA strain contains 2-aminoacrylate-modified cofactor. Authentic pyruvate/PLP (A), the 14.2-min HPLC eluant from denatured β-chloroalanine-treated Alr (B), and the 14.2-min HPLC eluant from denatured Alr isolated from a ridA mutant strain (C) were subjected to mass spectral analysis. All contained a major species at m/z 316.02278 (±5 ppm) corresponding to the mass of pyruvate/PLP and another species (inset) at m/z 157.50699 (±5 ppm) corresponding to the charged (−2) molecule.

The cofactor isolated from the Alr from wild-type and ridA strains was quantified by integrating the area under each peak shown in Fig. 4. The total amounts of cofactor isolated from the samples were similar (19.61 and 18.09 nmol mg⁻¹, respectively). These values indicated that the concentration of PLP-populated enzyme was not significantly different between the two samples. However, a reproducible difference in the composition of cofactor was measured in the samples from the wild-type (19.61 ± 0.14 nmol mg⁻¹ [PLP]) and ridA (15.44 ± 0.05 and 2.65 ± 0.07 nmol mg⁻¹ [PLP and pyruvate/PLP, respectively]) strains. Further, the pyruvate/PLP accounted for approximately 15% of the cofactor released from the Alr sample isolated from the ridA mutant strain. Activity levels in the two Alr protein samples were consistent with these data. First, the Alr protein isolated from the strain lacking RidA was ∼15% less active than the protein purified from the wild type (2.19 ± 0.06 and 2.56 ± 0.16 μmol l-alanine min⁻¹ mg⁻¹, respectively). Furthermore, the addition of PLP to the activity assay mixture increased the specific activities of both preparations proportionally (2.56 ± 0.11 and 2.91 ± 0.10 μmol l-alanine min⁻¹ mg⁻¹ for proteins from the ridA and wild-type strains, respectively). The ∼15% difference in activity between the two protein samples was determined to be statistically significant (P = 0.021 and P = 0.014), although it was not as extensive as the ∼40% difference in Alr activity measured in crude extracts. The disparity between the ratios of activities measured in crude extract and with purified protein may suggest that the 2-AA-modified protein is unstable in vivo or was lost in the purification. Taken together, these data indicate that a modification generated by endogenous 2-AA exposure contributed to the decreased activity of Alr in a strain lacking RidA.

Conclusions: in vivo-generated 2-aminoacrylate is deleterious in the absence of RidA.

In bacteria and other organisms, there are multiple PLP-dependent enzymes known to generate 2-AA as part of the biochemical mechanisms, including serine/threonine dehydratases (30, 33), cysteine desulfhydrases (34), tryptophanase (35), and tryptophan synthase (36), among others. While the reactivity of aminoacrylate has been emphasized by studies in vitro, the potential for this and similar enamines to cause significant damage in vivo has only recently come to light based on work with RidA (2, 9, 10). The significance of 2-AA as a metabolic stressor was not appreciated because of the short half-life of this metabolite in aqueous solution (∼1.5 s) and the resulting expectation that it would not persist in the cellular environment (12). The demonstration that RidA was a 2-aminoacrylate deaminase, coupled with the multiple phenotypes of a ridA strain, suggested that endogenously generated 2-AA could pose a significant risk in vivo. However, rigorous evidence of 2-AA-induced damage in vivo and the mechanism by which it occurs was lacking. This study was initiated to define the consequences of 2-AA accumulation in the cell at a molecular level.

This work made use of the molecular characteristics of an adduct formed by exposing Alr to 2-AA in vitro. The isolation of a pyruvate/PLP adduct from a cellular protein provided the first direct evidence of 2-AA accumulation in vivo and is the first isolation of a pyruvate/PLP adduct formed in vivo by the mechanism characterized for Alr in vitro (Fig. 1). Despite the presence of the same modification on Alr protein that was inactivated in vivo or in vitro, the difference between the two mechanisms is worth noting. Inactivation of the enzyme in vitro required the formation of 2-AA from β-chloroalanine in the active site of Alr (17, 18, 23). In contrast, in this study the source of the 2-AA was from a metabolic enzyme(s) in the cell. A recent report showed that 2-AA generated by one enzyme, threonine dehydratase, was able to enter and inactivate a second enzyme, transaminase B, in vitro (9). The relevant study provided evidence that the same mechanism occurred in vivo, consistent with the results herein indicating that endogenously generated reactive 2-AA can modify proteins. The covalent PLP adduct formed on IlvE, both in vivo and in vitro, was distinct from the one reported here (9).

PLP-dependent enzymes in addition to transaminase B and alanine racemases are likely to be inhibited by 2-AA-mediated inactivation in ridA mutants. With the described examples, two different PLP adducts have been identified, both of which were previously characterized in vitro. It remains to be seen whether additional distinct 2-AA-generated modifications exist in vivo. Future studies to identify the breadth of the enzymes affected in ridA strains have potential to provide insights into characteristics that make an enzyme susceptible to modification (e.g., active-site configuration) via cellular 2-AA.

In combination with other reports, the data herein emphasize that endogenously generated 2-AA persists free in the cellular milieu long enough to enter distinct active sites. Reconciliation of this conclusion with the characteristic half-life of enamines in aqueous solutions (12) requires that our understanding of the cellular milieu be reassessed. Genetic dissection of this system will complement biochemical/biophysical efforts that continue to define characteristics of the cellular environment.