Oxygen Exchange between Acetate and the Catalytic Glutamate Residue in Glutaconate CoA-transferase from Acidaminococcus fermentans

The exchange of oxygen atoms between acetate, glutaryl-CoA, and the catalytic glutamate residue in glutaconate CoA-transferase from Acidaminococcus fermentans was analyzed using [18O2]acetate together with matrix-assisted laser desorption/ionization time of flight mass spectrometry of an appropriate undecapeptide. The exchange reaction was shown to be site-specific, reversible, and required both glutaryl-CoA and [18O2]acetate. The observed exchange is in agreement with the formation of a mixed anhydride intermediate between the enzyme and acetate. In contrast, with a mutant enzyme, which was converted to a thiol ester hydrolyase by replacement of the catalytic glutamate residue by aspartate, no 18O uptake from H2 18O into the carboxylate was detectable. This result is in accord with a mechanism in which the carboxylate of aspartate acts as a general base in activating a water molecule for hydrolysis of the thiol ester intermediate. This mechanism is further supported by the finding of a significant hydrolyase activity of the wild-type enzyme using acetyl-CoA as substrate, whereas glutaryl-CoA is not hydrolyzed. The small acetate molecule in the substrate binding pocket may activate a water molecule for hydrolysis of the nearby enzyme-CoA thiol ester.

The strict anaerobic bacterium Acidaminococcus fermentans is able to grow with glutamate as the sole source of energy. The fermentation of glutamate proceeds via the hydroxyglutarate pathway yielding ammonia, carbon dioxide, acetate, butyrate, and hydrogen as products (1). With the exception of the oxidation of glutamate to 2-oxoglutarate followed by reduction to (R)-2-hydroxyglutarate, almost all other transformations in this pathway are carried out on the CoA-ester level (2). Hence the activation of (R)-2-hydroxyglutarate is a key step in this metabolic pathway (3).
The catalytic action of CoA transferases has been suggested to proceed via a mechanism outlined in Fig. 1 (6,7). By nucleophilic attack of a glutamate residue of the enzyme at the carbonyl of the donor acetyl-CoA, a mixed anhydride between the enzyme and acetate is formed. The transiently liberated CoAS Ϫ anion reattacks the glutamate carbonyl carbon to form the product acetate and an enzyme-CoA thiol ester. The second product, (R)-2-hydroxyglutaryl-CoA, is formed by repetition of the steps outlined above. It should be noted that during one catalytic cycle, one oxygen atom is transferred from the acetate to the glutamate residue. Consecutive turnovers lead to a complete equilibration of all oxygen atoms involved in the reaction.
The catalytic glutamate residue in glutaconate CoA-transferase has been identified as amino acid 54 of the smaller ␤-subunit (␤Glu-54). The thiol ester between coenzyme A and the glutamate residue has been reduced with sodium boro[ 3 H]hydride to the corresponding alcohol, which was identified as 2-amino-5-hydroxy [5-3 H]valeric acid within a tryptic peptide (4,5). The enzyme was crystallized, and its structure has been solved recently at 2.55 Å resolution. The crystallographic data (8) as well as sequence alignments (9) also confirmed that glutamate ␤54 is the catalytic residue in glutaconate CoA-transferase.
The important role of glutamate ␤54 has been demonstrated by site-directed mutagenesis experiments (10). Whereas the replacement of glutamate ␤54 by alanine (␤E54A) or asparagine (␤E54N) completely abolished the transferase activity, the glutamine (␤E54Q) mutant retained a remarkable residual activity (1% of the wild type). By incubating this enzyme with both substrates for 20 h at room temperature, the glutamine was completely converted to glutamate yielding a fully active CoA-transferase. Changing glutamate 54 to aspartate converted the enzyme to a thiol ester hydrolyase (9). It has been suggested that the missing methylene group in the side chain of aspartate as compared with glutamate allows a water molecule to occupy the space between the substrate and the carboxylate of aspartate ␤54. Activated by the carboxylate of aspartate as the general base, this water molecule should be able for a nucleophilic attack at the thiol ester carbonyl carbon and cleavage of the thiol ester bond. Alternatively, a mechanism requiring the transient formation of a mixed anhydride between aspartate and the donor carboxylic acid might be possible.  (5). The crystallization step, however, was replaced by chromatography on Q-Sepharose. The molar concentrations of the hetero-octameric enzyme were calculated with the molecular mass of the heterodimeric subunit (65 kDa).
Synthesis of Acetyl-, Glutaryl-, and Propionyl-CoA-The CoA thiol esters of acetic, glutaric, and propionic acid were prepared from the corresponding anhydrides and CoASH by the method of Simon and Shemin (11). The acyl-CoAs were desalted using Sep-Pak TM C 18 cartridges (Millipore  (12). The 18 O content of the sodium acetate (99%) was determined by gas chromatography/mass spectrometry analysis of the methyl ester synthesized with diazomethane.
Kinetics of the Oxygen Exchange-Glutaconate CoA-transferase (10 g/ml) was added to a mixture containing 100 mM sodium phosphate, pH 7.0, 1 mM glutaryl-CoA, and 1 mM sodium [ 18 O 2 ]acetate. Before the addition of the enzyme and after various time intervals, aliquots were taken and stopped by the addition of 1 volume of 8 M guanidinium hydrochloride, pH 7.5. The samples were applied on LiChroCART columns (4 ϫ 250 mm, 5 m, Merck) equilibrated with 0.1% (v/v) trifluoroacetic acid. The CoA derivatives were separated by a linear gradient from 10 to 20% (v/v) acetonitrile for 20 min at 20°C and monitored at 260 nm. The eluting CoA derivatives were collected and analyzed by MALDI-TOF MS (see below). The CoA derivatives were quantified using an internal calibration.
Labeling of Glutaconate CoA-transferase with [ 18 O 2 ]Acetate-The enzyme (2 nmol) was incubated for 2 min at 37°C in 200 l of 100 mM sodium phosphate, pH 7.0, 1 mM glutaryl-CoA, and 1 mM sodium [ 18 O 2 ]acetate. The reaction was stopped by the addition of 1 volume of 8 M guanidinium hydrochloride, pH 7.5, and applied to reverse-phase HPLC as described. A second sample was incubated as described and separated from low molecular mass compounds by size exclusion chromatography using a Sephadex-G25 column (NAP-10, Amersham Pharmacia Biotech) equilibrated with 100 mM sodium phosphate, pH 7.0, and 100 mM [ 16 O 2 ]acetate. Glutaryl-CoA was added to the enzyme at a final concentration of 1 mM and incubated for another 2 min at 37°C.
Cleavage of the Tryptic Peptide ␤6 by Endoproteinase Asp-N-The tryptic peptide ␤6 was lyophilized, redissolved in 100 l of 1 M guanidinium hydrochloride, 100 mM Tris-HCl, pH 7.5. Endoproteinase Asp-N (100 ng) was added, and the digestion was allowed to proceed for 16 h at 37°C. The resulting peptides were separated by reverse-phase HPLC as described above.
Carboxyl-terminal Sequencing-The peptide ␤6A was dissolved in 20 mM sodium acetate, pH 6.6, yielding a final concentration of approximately 10 M. The peptide solution (2 l) was mixed with 2 l of carboxypeptidase Y (0.1 mg/ml) and incubated for 20 min at 25°C. The reaction was stopped by the addition of 2 l of 10% (v/v) trifluoroacetic acid. The individual molecular mass values of the generated peptides were determined by MALDI-TOF MS (see above).
Determination of the Hydrolyase Activity of Glutaconate CoA-transferase-The enzyme (0.77 nmol) was incubated at 25°C in 1 ml of 100 mM potassium phosphate, pH 7.0, 100 mM NaCl, 1 mM 5,5Ј-dithio-bis-(2-nitrobenzoate), and 200 M acyl-CoA in either the presence or the absence of 100 mM sodium acetate. The change in absorbance at 412 nm was recorded for 10 min.
Calculation of the 18 O Content from Mass Spectra-The natural isotopic distribution of large molecules such as acyl-CoAs and peptides was calculated solving the binomial equation for the isotopic distribution. This calculation was aided by the isotope pattern calculator provided by the University of Sheffield (UK) at the Sheffield ChemPuter web site (http://www.shef.ac.uk/chemistry/chemputer/isotopes.html). The 18 O overlay was simulated by solving the binomial coefficient for different numbers of exchange sites. The distribution pattern for the enrichment of 18 O is given by the general formula, The measured mass spectra were integrated, and the measured distributions were fitted to simulated stick spectra minimizing the least squares deviation between the intensities of measured and simulated distributions. These evaluations were performed using the "Solver" facilities provided by Excel 97. The experimental errors of the measurements were determined by doubling the value of 2 for the nonlinear least square fits.

RESULTS
The time course of the oxygen exchange was investigated with catalytic amounts (0.1 M) of glutaconate CoA-transferase and initial concentrations of 1 mM for [ 18 O 2 ]acetate and glutaryl-CoA. At various time intervals, samples were subjected to HPLC analysis and subsequent MALDI-TOF MS of the CoA derivatives. As evident from the concentrations of acetyl-CoA and glutaryl-CoA (Fig. 2C), a chemical equilibrium of the reaction was reached after approximately 15 min with K eq ϭ 0.64 Ϯ 0.04. This value is close to 0.77, which can be calculated from the pK values of glutarate (pK 1 ϭ 4.34) and acetate (pK ϭ 4.75). In addition to acetyl-and glutaryl-CoA, increasing amounts of free CoA were detected by HPLC, whereas neither acetyl-CoA nor glutaryl-CoA exhibited significant rates of hydrolysis in control experiments.
As demonstrated in Fig. 2, the mass spectra of acetyl-CoA ( Fig. 2A) and glutaryl-CoA (Fig. 2B) showed clearly resolved single isotopic peaks. The mass spectra were integrated, and the relative signal intensities for the isotopic distribution were calculated. Alignments of these distributions to simulations allowed the determination of the 18 O content of the acyl-CoAs. Minimizing the least squares deviation between simulation and measured data, the 18 O content of the analytes could be calculated with an experimental error of less than 4% (see insets in Fig. 2, A and B). The isotopic equilibrium, however, was reached much more slowly than the chemical equilibrium after approximately 120 min. The 18 O contents at equilibrium for acetyl-CoA (45 Ϯ 3%) and glutaryl-CoA (41 Ϯ 3%) agreed with the calculated value of 39.4%.
To analyze the incorporation of 18 O into the catalytic glutamate residue ␤Glu-54 of glutaconate CoA-transferase, 20 M of enzyme was incubated with 1 mM unlabeled glutaryl-CoA and 1 mM [ 18 O 2 ]acetate for 5 min, followed by HPLC separation (Fig. 3). The 18 O label of the isolated acetyl-CoA (44 Ϯ 3%) and glutaryl-CoA (43 Ϯ 3%), determined by MALDI-TOF MS, indicated that an isotopic equilibrium of the oxygen atoms participating in the reaction was reached.
The separated subunits of glutaconate CoA-transferase were identified by their molecular mass values obtained by MALDI-TOF MS. For the ␣-subunit a molecular mass of 35,573 Ϯ 12 Da was determined in agreement with the value of 35,568 Da predicted from the sequence (Fig. 4A). For the ␤-subunit, a molecular mass of 29,018 Ϯ 18 Da (theoretical 29,017 Da) was obtained. However, a second signal at m ϩ 752 Da was found (Fig. 4B). In incubations with glutaryl-CoA but without acceptor carboxylate, the higher molecular mass signal was the only visible one. The mass difference observed confirmed the covalent binding of a CoA-molecule to the catalytic glutamate according to the reaction scheme shown in Fig. 1.
Neither the mass accuracy nor the resolution obtained in the spectra of the ␤-subunit was sufficient to determine 18 O uptake. Hence, the protein was reduced with dithiothreitol, carboxymethylated, and digested with trypsin. The carboxymethylation was required to obtain the ␤Glu-54-containing peptide in good yields. It is well established that a significant loss of Cys-containing peptides is frequently observed if no chemical modification of the thiol group is performed. The peptide containing the catalytic glutamate residue was purified by reverse-phase HPLC on a Supelcosil LC-318 (4.6 ϫ 250 mm, 5 m). The monoisotopic molecular mass of that peptide (2744.6 Ϯ 1.5 Da) was, within the experimental range of accuracy, identical to the expected mass of 2744.3 Da, but it showed a clearly visible broadening of the isotopic distribution, which was only poorly resolved (data not shown). Therefore, the peptide was digested for a second time using endoproteinase Asp-N. As shown in Fig. 5, the undecapeptide containing amino acids 48-DCHIIVESGLM-58 gave well resolved mass spectra. The observed monoisotopic molecular mass of 1273.9 Ϯ 0.5 Da was in agreement with the theoretical mass of 1274.6 Da. The natural isotopic distribution was clearly overlaid by an 18 (Table I).
In addition to the peptide signals at 1274 Da, varying amounts of an oxidized form of the peptide at 1290.8 Ϯ 1.2 Da were found. The mass difference of 16 Da suggested an oxidation of the methionine 58 to the sulfoxide. Further analysis showed that this oxidation was inevitable during the preparation of peptide samples for MALDI-TOF MS. Using a carboxylterminal cleavage with carboxypeptidase A and MALDI-TOF MS read-out of the peptide ladder, the C-terminal sequence VESGLM of the peptide was confirmed. The oxidation site was located on methionine 58 and the 18 O label of the peptide was found in glutamate 54 (data not shown). The 18 O contents in the non-oxidized and the oxidized forms of the peptide were found to be identical within the experimental accuracy. However, the isotopic distribution of the oxidized form of the peptide was always slightly disturbed by the sodium adduct of the non-oxidized peptide.
As evident from Fig. 2, wild-type glutaconate CoA-transferase also catalyzed the slow hydrolysis of glutaryl-CoA. Further analysis revealed that this hydrolytic activity was significantly stimulated by 100 mM acetate (114 milliunits/mg as compared with Ͻ2 milliunits/mg, respectively), whereas the hydrolysis of acetyl-CoA was not further enhanced (105 milliunits/mg as compared with 98 milliunits/mg). Propionyl-CoA was hydrolyzed more slowly than acetyl-CoA, and a stimulation by 100 mM acetate was found (54 and 118 milliunits/mg, respectively) (Table II). It should be noted that the apparent K m values observed for the hydrolysis of different acyl-CoA derivatives were about 500 times lower (67-69 M, Table II) than the apparent K m value for acetate (26 mM) in the CoA-transferase reaction in the presence of 100 M glutaryl-CoA (4).
In this paper we have demonstrated that a mechanism-based introduction of 18 O label can be followed during several steps of analysis to the level of the labeled amino acid and that quantitative results can be obtained from the integrated spectra. These results encouraged us to investigate the hydrolysis reaction catalyzed by wild-type glutaconate CoA-transferase. As expected, the chemical hydrolysis of the HPLC-purified thiol ester between ␤Glu-54 and CoA in H 2 18 O at pH 8.6 led to the incorporation of almost exactly 1.0 18 O into the catalytic glutamate residue (47 Ϯ 3%, Table I). By enzymatic hydrolysis of FIG. 4. Mass spectrometry of the subunits of glutaconate CoA-transferase. The mass spectra of glutaconate CoA-transferase ␣-subunits (A) and ␤-subunits (B) were recorded with sinapinic acid as matrix using an accelerating voltage of 25,000 V, 89 -90% grid voltage, and a 300-ns delay time in the linear mode of the instrument. The 100% intensity scale refers to 600 counts in A and 6000 counts in B, respectively. The calculated molecular mass values are 35,568 Da for the ␣-subunit and 29,018 Da for the ␤-subunit.
A glutaconate CoA-transferase, in which the catalytic glutamate has been replaced by aspartate by site-directed mutagenesis, has been shown to be a thiol ester hydrolyase rather than a coenzyme A-transferase. Although the replacement of glutamate by aspartate is predicted to generate an additional cleavage site for endoproteinase Asp-N, this new cleavage site was apparently not used under the experimental conditions. If this enzyme was incubated in the presence of either [ 18 O 2 ]acetate and glutaryl-CoA or glutaryl-CoA in H 2 18 O, no 18 O was found in the corresponding undecapeptide even after complete hydrolysis of glutaryl-CoA. In addition, the formation of an enzyme-CoA thiol ester was never observed with this enzyme.

DISCUSSION
The incubation of [ 18 O 2 ]acetate and glutaryl-CoA in the presence of glutaconate CoA-transferase from A. fermentans redistributed the 18 O label, initially present only in acetate, among all seven of the participating oxygen atoms. Whereas the 18 O content of the three oxygen atoms of glutaryl-CoA increased immediately, the decrease in the 18 O content of the single oxygen atom of acetyl-CoA was slightly delayed. This observation is in agreement with the statistical proposal for the exchange reaction. Initially, the reaction proceeds predominantly  The enzyme was incubated as indicated, and the undecameric peptide containing the catalytic residue was isolated and analysed by MALDI-TOF MS as described in the text. The 18 O uptake was calculated from the integrated mass spectra as described in the text.
The normalized relative signal intensities were calculated from the integrated mass spectra (n ϭ 3-5). The value given for the monoisotopic mass (m) corresponds to the signal of the monoisotopic peak at m/z ϭ 1275.6 Da for the wild type and at m/z ϭ 1261.7 Da for the mutant. The following single isotopic peaks have been numbered ϩ1 to ϩ5. The distribution is normalized on the signal intensity of the highest peak.
b The calculated 18 O content of these samples was smaller than the experimental error. c The signal distribution of this peptide did not align to an equal exchange distribution for both carboxylate oxygen atoms. The introduction of only one oxygen, whereas hydrolysis increases the exchange to 93% and is close to the 18 (Table I). The 18 O label has been found to be localized on both oxygen atoms of glutamate residue 54 within the ␤-subunit of the enzyme, which forms the enzyme-CoA thiol ester intermediate as predicted by the mechanism outlined above (Fig. 1). Only one oxygen atom of the catalytic glutamate residue became labeled with 18 O by chemical hydrolysis of the isolated thiol ester between CoA and the wild-type enzyme in H 2 18 O.
For the ␤E54D mutant of glutaconate CoA-transferase, it was proposed that because of the shortened side chain of aspartate as compared with the original glutamate residue, a water molecule might occupy the space between the aspartate and the acyl-CoA in the active center (9). Acting as a general base, the carboxylate group of aspartate has been suggested to activate the water for hydrolysis of the thiol ester, which is in agreement with the absence of any 18 O in the ␤Asp-54 carboxylate. This result, however, cannot exclude the intermediacy of a mixed anhydride between ␤Asp-54 and acetate, which is specifically attacked by H 2 18 O at the acetyl carbonyl (Fig. 6). Interestingly, the slow hydrolysis of acetyl-CoA in H 2 18 O catalyzed by the wild-type enzyme led to a significant 18 O incorporation into the ␤Glu-54 carboxylate. In this case at least three mechanisms are possible that cannot be distinguished because of the subsequent much faster equilibration of the four oxygen atoms of acetate and ␤Glu-54 by CoA-transfer. The three mechanisms are as follows: (i) hydrolysis of acetyl-CoA by assistance of ␤Glu-54 carboxylate; (ii) hydrolysis of either carbonyl of the mixed anhydride; or most likely, (iii) the acetate catalyzed hydrolysis of the enzyme-thiol ester.
Although the low but significant CoA-ester hydrolyase activity has been recognized much earlier (4), this current work shows that hydrolysis is due to an intrinsic property of the enzyme and subject to catalysis by acetate. Whereas glutaryl-CoA was not hydrolyzed (Ͻ2 milliunits/mg), hydrolyase activity was observed (114 milliunits/mg) in the presence of 100 mM acetate exhibiting an apparent K m ϭ 69 M acetate (Table II). Further analysis revealed that acetyl-CoA is hydrolyzed at the same rate and is not affected by the addition of acetate, whereas propionyl-CoA behaves in a manner in between acetyland glutaryl-CoA. The low apparent K m for acetate observed in the hydrolysis of acyl-CoA substrates is comparable with the low apparent K m -values observed in CoA-ester hydrolysis by the ␤E54D mutant and for the specific acetyl-CoA-dependent exchange of 1.0 3 H from [2,4-3 H]glutaconate with the solvent water (14). These low K m values show that the enzyme is able to bind these substrates very tightly. In all three reactions no complete CoA-transfer occurred. The unusually high K m -values  in the complete CoA-transfer are in accordance with the hypothesis of White and Jencks (7) that CoA-transferases are able to convert binding energy into rate enhancement. If one adds the ϳ370-fold increase in K m for acetate and the ϳ170-fold increase in K m for acetyl-CoA (14), the transition state could be lowered by 27 kJ/mol. It has been shown by reduction with NaBH 4 (4) and by MALDI-TOF MS (this paper) that incubation of the CoA-transferase with glutaryl-CoA alone results in up to 100% conversion of the enzyme into the stable CoA-ester. Addition of acetate results in two reactions, CoA-transfer and hydrolysis. If hydrolysis requires a proceeding CoA-transfer, then no difference should be observed in the K m for both processes. Hence, it appears likely that acetate stimulates the hydrolysis of the enzyme CoA-ester intermediate by activating a water molecule. One of the possible two carboxylate binding sites should be large enough to accommodate water together with the general base acetate or, to a lesser extent, with propionate (Fig. 7). The hydrolysis at the stage of the enzyme-CoA ester rather than acyl-CoA is in accordance with the high reactivity of the former toward NaBH 4 . Probably the enzyme-CoA ester is somehow distorted, which makes the carbonyl group more accessible for nucleophiles.
In E. coli 3-methyladenine-DNA glycosylase II (DNA glycosidase, AlkA) the catalytic aspartate residue 238 of the enzyme was proposed to activate a water molecule for hydrolysis of the N-glycosidic bond of the substrate (15). In agreement with this mechanism, it was found that the exchange D238N by mutagenesis completely abolished the hydrolyase activity of the protein. The same holds true for the mutants of glutaconate CoA-transferase (9,10). Whereas the ␤E54Q mutant is slowly converted to the active wild-type CoA-transferase in the presence of both substrates, no reactivation to a hydrolyase was found for the ␤E54N mutant. In a mechanism-based reactivation of the ␤E54Q mutant, the substrate carboxylate, e.g. glutaconate, may slowly form a mixed anhydride with the glutamine residue, and ammonia is formed. The anhydride is further hydrolyzed or more likely aminolyzed, yielding the active, glutamate ␤54-containing enzyme. In contrast, the ␤E54N mutant is not converted to the hydrolyase in a similar, mechanism-based fashion, probably because the anhydride between the aspartate and the glutaconate residue cannot be formed due to a gap between the asparagine ␤54 and the enzyme-bound dicarboxylate.
The following lines of evidence support the activation of a water molecule for hydrolysis of acyl-CoA by the ␤E54D mutant of glutaconate CoA-transferase rather than formation of a mixed anhydride intermediate: 1) no 18 O incorporation from H 2 18 O or [ 18 O 2 ]acetate was found for the ␤E54D mutant enzyme in the presence or absence of acyl-CoA; 2) no reactivation to a hydrolyase activity was found for the ␤E54N mutant, whereas the ␤E54Q mutant could be reactivated to the active wild-type enzyme in the presence of substrates; and 3) a specific acyl-CoA hydrolyase activity with acetate as cofactor was found for the wild-type enzyme.
Taking all of these findings together, it appears far more likely that the carboxyl group of aspartate can act as a general base in the activation of a water molecule, allowing the direct hydrolysis of the acyl-CoA substrate rather than the intermediate formation of a mixed anhydride, which is further hydrolyzed.