Enzymatic Hydrolysis of Organic Cyclic Carbonates*

Ethylene carbonate, a cyclic organic carbonate widely used industrially, is toxic when metabolically converted to ethylene glycol. A rat liver enzyme active in catalyzing the ring opening has been purified to electrophoretic homogeneity and found to be active in the hydrolysis of ethylene, vinylene, and propylene carbonates to CO 2 and the respective glycols. Neither thiocarbon- ates nor open chain carbonates served as substrate nor did a variety of esters, lactams, lactones, and related heterocycles. The enzyme was active, however, with imides and appears to be identical to rat liver imidase. The identification was confirmed by copurification of enzyme activities, by similarities in the pattern of inhibi-tion, and by the reactivity with a polyclonal antibody directed against the enzyme purified here. The systemic toxicity observed with ethylene carbonate, a cyclic organic carbonate that serves as a major intermediate in the plastics industry, was recognized as being similar to the effects of ethylene glycol (1). When the cyclic carbonate was fed to rats, ethylene glycol accumulated as did more oxidized metabolic products (1). number catalyzing the hydrolysis of ethylene carbonate to form 1 m mol of ethylene glycol/min under standard assay conditions. Specific activity is in terms of units of activity/mg of enzyme. Protein was determined by the method of Brad- ford (8) using bovine serum albumin (Pierce) as standard and the Bio-Rad protein assay reagent (Bio-Rad).

We report the isolation in homogeneous form of an enzyme from rat liver that catalyzes the hydrolysis of ethylene carbonate as well as certain other cyclic organic carbonates but that is inactive toward linear organic carbonates and carboxyesters. Of specific interest is the finding that the isolated enzyme appears to be identical to one variously described as dihydropyrimidinase (EC 3.5.2.2), hydantoinase, dihydropyrimidine hydrase, and dihydropyrimidine aminohydrolase, an enzyme previously identified as a nonspecific imidase active with a large number of cyclic and linear imides (4).

MATERIALS AND METHODS
Organic carbonates and their analogues were from Aldrich; sources of additional candidate substrates have been presented (4). Unless otherwise noted, other reagents and enzymes were from Sigma.

Electrophoresis
Polyacrylamide gel electrophoresis was performed with an X-cell Minigel Cell (Novex, San Diego) using the manufacturer's precast isoelectric focusing gels (pH 3-10 range) as well their 14% Tris-glycine gels; the latter were used with SDS by the method of Laemmli (5). Isoelectric focusing was conducted at 100 V for the first h, 200 V for the second h, and 500 V for the last 30 min. Protein standards for SDS, gradient-native, and isoelectric focusing gels were purchased from Novex, Amersham Pharmacia Biotech, and Bio-Rad, respectively.
The isoelectric focusing pattern was assessed after incubating 1 mg/ml of enzyme under the following conditions: with 60 units/ml of potato acid phosphatase at pH 5.0 for 90 min, with 20,000 units/ml of calf intestinal mucosa alkaline phosphatase at pH 8.0 for 90 min (both enzymes from Boehringer Manheim), or with 1 unit/ml of Clostridium perfringens neuraminidase at pH 5.5 for 60 min.
Covalently bound carbohydrate was measured by G. Gilbert Ashwell using a Dionex PA-1 ion exchange system; this technique is capable of detecting 1 mol of carbohydrate/mol of enzyme (6).

Standard Assay
Ethylene carbonate was adopted as the standard assay substrate, and its hydrolysis was followed by estimating the formation of ethylene glycol. The product was oxidized by periodate to form 2 mol of formaldehyde, which was allowed to react with chromotropic acid to yield a colored product (7).
Enzyme in a total volume of 0.1 ml was incubated in 100 mM sodium pyrophosphate at pH 8.1 with 5 mM ethylene carbonate for 1 h at 25°. The reaction was terminated by transfer to an ice bath and addition of 0.1 ml of 10 N H 2 SO 4. After adding 0.2 ml of 10 mM potassium periodate and incubation at 37°for 30 min, 0.25 ml of 0.1 M sodium arsenite was added, and the reaction mixture was maintained at room temperature for another 10 min. A 150-l aliquot of each sample was mixed with 900 l of chromotropic acid reagent. (The reagent was prepared from 200 mg of 4,5-dihydroxy-2,7-naphthalene disulfonic acid in 110 ml of water that was cooled in ice and to which 90 ml of concentrated H 2 SO 4 were slowly added). The sample was heated at 95°for 30 min, and its absorbance was measured at 570 nm in cuvettes of 1-cm light path. Under these conditions, a change in A 570 of 0.619 represents the formation of 2.0 mol of formaldehyde from 1.0 mol of ethylene glycol. The reaction is linear with time and with enzyme concentration for measurements of absorbance changes of less than 1.0. (Note that the yield of formaldehyde by periodate oxidation is 1 mol/mol from either propylene carbonate or vinylene carbonate).
A unit of activity is defined as the amount of enzyme catalyzing the hydrolysis of ethylene carbonate to form 1 mol of ethylene glycol/min under standard assay conditions. Specific activity is in terms of units of activity/mg of enzyme. Protein was determined by the method of Bradford (8) using bovine serum albumin (Pierce) as standard and the Bio-Rad protein assay reagent (Bio-Rad).
* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

Continuous Spectrophotometric Assay
The standard assay is effective only with substrates that yield a compound with vicinal nucleophilic groups as products (7). That is not the case for products derived from the linear carbonates of ethyl, propyl, and allyl alcohols that we followed by the formation of the respective alcohol. The alcohols were measured by coupling the hydrolysis of the carbonate with horse liver alcohol dehydrogenase and yeast aldehyde dehydrogenase and measuring the formation of NADH. Although ethylene glycol was not an effective substrate in this dehydrogenasecoupled assay, the formation of each of the other noted alcohols could be readily measured. Included in a total volume of 1.0 ml were 0.1 M Tris-HCl at pH 8.0, 2 mM NAD, 0.2 unit of horse liver alcohol dehydrogenase, and 0.2 unit of yeast aldehyde dehydrogenase. After 3 min, 0.04 units of the organic carbonate hydrolyzing enzyme was added followed by 5 mM test substrate. The course of the reaction was followed at 340 nm with a Shimadzu UV1201 spectrophotometer. An absorbance change of 12.2 represents the formation of 1 mol of acid from the respective alcohol.

Other Enzyme Assays
Evidence of stereoselectivity was sought by monitoring hydrolysis of propyl carbonate by circular dichroism spectrometry with a Jasco J-500C spectropolarimeter (4). Enzyme incubation conditions were the same as for the standard spectrophotometric assay except for the increased volume, 3 ml, of the reaction vessel.
Hydrolysis of L-2-oxothiazolidine-4-carboxylate was measured by thiol formation as assayed with DTNB 1 (9). In a total volume of 1 ml, 2 g of enzyme in 100 mM sodium pyrophosphate at pH 8.1 were incubated for 1 h at 25°with 5 mM substrate and 0.1 mM DTNB; changes in absorbance were monitored at 412 nm (⑀ 412 ϭ 14,150 M Ϫ1 cm Ϫ1 ).
Luminol as a possible substrate was examined by chemiluminescence of the expected product in the presence of peroxidase (10). Enzyme, 10 g, was incubated in 100 mM sodium pyrophosphate at pH 8.1 with 10 M luminol. After 1 h, 10 l of 0.1 mM hydrogen peroxide were added along with 20 units of horseradish peroxidase, and luminescence was measured with an SLM 8000 spectrofluorometer.
The possibility of hydrolysis of cyclic nucleotides was examined by incubating candidates for substrate at 1 mM with 2 g of enzyme in a total volume of 0.1 ml. After incubation at room temperature for 1 h, aliquots were applied to cellulose thin-layer chromatography plates containing a fluorescent indicator (Eastman Kodak). Plates were developed with 2-propanol:ammonium hydroxide:water (70:15:15), and the nucleotides were visualized under UV light. Carboxyesterase activity was measured with 4-nitrophenyl acetate as substrate (11).

Molecular Weight Determination
Centrifugation was carried out with a Beckman XLA analytical centrifuge and a Ti-60 rotor using a 12-mm double sector cell containing 3-mm columns of protein solution (300 g/ml in 20 mM Tris-HCl at pH 7.8, 0.1 M NaCl, with or without 1 mM DTT) and diffusate (12) layered over FC-47 fluorocarbon. Exact values of loading absorbance at 280 nm were obtained from initial scans taken just after speed had been reached. Rotor speeds of 5000 and 7000 rpm were employed at temperatures of 5 and 25°C. Radial scans were performed at 6 -8-h intervals; equilibrium was judged to have been reached when the calculated difference between successive scans was a flat line, indistinguishable from zero within experimental error. The enzyme was prepared by dialysis overnight against two changes of Tris-HCl at pH 7.8. A partial specific volume for the protein of 0.73 ml/g and a buffer density of 1.003 g/ml were used.

Immunology
Antibody to the enzyme, purified as described below and homogeneous upon SDS-gel electrophoresis, was prepared by Lofstrand Labs Ltd. (Gaithersburg, MD) 10 weeks after the initial inoculation with five booster doses, yielding a titer of greater than 1:32,000. The antibody was purified by salting out (between 0.3 and 0.5 saturation with ammonium sulfate), dialysis against 20 mM Tris-HCl at pH 8.5, and by elution with a 0 -0.5 M NaCl gradient in Tris-HCl at pH 7.8 from a DEAE-Sepadex column. Antibody diluted with 100 mM potassium py-rophosphate at pH 7.8, was incubated with enzyme at 4°overnight, and the suspension was centrifuged in an Eppendorf centrifuge at the same temperature. The supernatant fluid was tested for enzyme activity.

Purification
Livers from young male Sprague-Dawley rats (150 -200 g) were shipped in DriIce by Pel-Freeze Biologicals (Rogers, AZ) and stored for periods of up to 3 months at Ϫ80°C. Unless otherwise noted, all procedures were conducted in a cold room at 4°C or in an ice bath. The pH of all buffers refers to measurements taken after equilibration at room temperature.
Step 1: Extract-About 300 g of liver were partially thawed and suspended in 3-fold their volume of Buffer A (20 mM Tris-HCl containing 1 mM EDTA, 1 mM DTT, 1 mM phenylmethylsulfonylfluoride, 1 nM pepstatin A, and 2 g/liter antipain adjusted to pH 8.0 with HCl). Each half of this suspension was homogenized for 30 s in a Waring blender and then filtered though a double layer of cheesecloth. The combined filtrate was centrifuged for 20 min at 16,000 ϫ g, and the residue was discarded.
Step 2: Salting Out-Ammonium sulfate, 25.2 g/100 ml was added to the solution, which was stirred for 30 min. The suspension was centrifuged at 16,000 ϫ g, and the supernatant fluid was treated with an additional 10.8 g of ammonium sulfate/100 ml. The resultant precipitate was collected by centrifugation (16,000 ϫ g, 20 min), suspended in 100 ml of Buffer A, and dialyzed overnight against three changes of 1, 1, and 4 liters, respectively, of the same buffer.
Step 3: DEAE-Sephacel-The enzyme solution was applied to a column (4.5 ϫ 50 cm) of DEAE-Sephacel (Amersham) that had been equilibrated with Buffer A and was washed with an additional liter of the buffer. Protein was eluted with a linear salt gradient of 3.0 liters of the same buffer containing from 0 to 0.4 M NaCl (Fig. 1, DEAE). Active fractions of approximately 7 ml each were pooled (fractions 290 -320).
Step 4: Red Agarose-The enzyme from Step 3 was applied directly to a column (3.2 ϫ 17 cm) of Red Agarose (Sigma) that had been equilibrated with Buffer B (Buffer A without EDTA). The column was subsequently washed with 150 ml of Buffer B and eluted with a linear salt gradient in Buffer B ranging from 0 to 1 M NaCl. Active enzyme was collected, pooled ( Fig. 1, RA), concentrated to 20 ml with an Amicon Diaflo PM10 membrane, and dialyzed overnight against three changes of, respectively, 1, 1, and 2 liters of Buffer C (10 mM potassium phosphate containing 1 mM DDT, 1 mM phenylmethylsulfonyl fluoride, 1 nM pepstatin A, and 2 g/liter antipain adjusted to pH 6.8).
Step 5: Hydroxyapatite-The dialysate was applied to a column (2.0 ϫ 20 cm) of hydroxyapatite gel (Sigma) that had been equilibrated with Buffer C. After washing the column with 100 ml of Buffer C, enzyme was eluted with 400 ml of a linear salt gradient of Buffer C and Buffer C made 0.4 M in potassium phosphate at pH 6.8 (Fig. 1, HA). Pooled, active fractions were concentrated by ultrafiltration to 5 ml with a PM10 membrane and dialyzed against Buffer E (Buffer C at pH 6.5 made 1 mM in EDTA).
Step 6: CM-Sepharose-The dialyzed protein solution was loaded onto a column (1.0 ϫ 15 cm) of CM-Sepharose previously equilibrated with Buffer E. The column was washed with 30 ml of Buffer E and eluted with a 120-ml linear salt gradient from 0 to 400 mM NaCl in Buffer E (Fig. 1, CM).
Step 7: Sephacryl-After concentration to 1 ml, the enzyme solution was used to charge a column (2 ϫ 80 cm) of Sephacryl S-400 (Amersham) that had been equilibrated with a solution of 20 mM Tris-HCl at pH 7.8, 0.1 M NaCl, 1 mM DDT, 0.1 mM EDTA, and the three proteinase inhibitors. The column was developed with the aforementioned Tris buffer, resulting in the elution of the enzyme close to the front.

RESULTS
Based on the standard assay with ethylene carbonate as substrate, the enzyme responsible for Reaction 1 was purified to electrophoretic homogeneity by essentially conventional means in yields of 4 -5% from rat liver. The enzyme is a homotetramer with subunits of identical size that were estimated to be 55,000 Da by SDS gel electrophoresis; a single band of protein was seen (Fig. 2, lane 1) in the same position whether or not the enzyme was heated with 1% 2-mercaptoethanol (5). Using sedimentation equilibrium centrifugation, an M r of approximately 220,000 was estimated; the M r remained the same when equilibrium was established in the presence or in the absence of 1 mM DTT.
Effects of pH-The enzyme was most stable at slightly alkaline pH but not below neutrality. At the pH of the purification buffer, 7.8, activity was unaffected by storage of the purified enzyme at 4 and Ϫ80°C for 3 months. Storage at pH 6.3 and 4°C led to a 14% loss of activity after 3 days.
As shown in Fig. 3, the pH optimum for the enzyme was at approximately 8.1 after correction for spontaneous hydrolysis. A correction was also required for the spontaneous hydrolysis of the other two active cyclic carbonates that were substrates.
Redox Effects-Although the purified enzyme presented as a single protein band on both SDS and gradient native gels, gel isoelectric focusing displayed multiple protein bands between a pH 7.2 and 7.4 (Fig. 2, lane 2). Treatment of the protein with bovine alkaline phosphatase, potato acid phosphatase, or Clostridium neuraminidase did not alter the pattern (not shown), and the determination of the presence of less than 1 mol of carbohydrate/mol of protein suggested that glycosylation was not responsible for the multiple bands. The multiple band pattern could be manipulated, however, with oxidizing and reducing reagents. In the presence of 1 mM DTT, major bands were apparent between pH 7.2 and 7.5. In contrast, 5 mM GSSG yielded at least nine protein bands with P i values between 6.3 and 7.2; the shift to lower P i was enhanced progressively over the course of a 3-h exposure to GSSG (lanes 3 and 4). A similar shift to lower P i was observed with 1 mM DTNB (lane 5) or 2 mM hydrogen peroxide (lane 6) as oxidizing agents. The effect of GSSG could be reversed by subsequent incubation with DTT (not shown).
Changes in P i were reflected in enzyme activity. Incubation with 2 mM DTT resulted in dramatic loss of activity (Fig. 4) that was irreversible if 20 mM DTT was used (not shown). Incubation of enzyme with 1 mM GSSG or 1 mM DTNB, however, resulted in a slight but distinct increase in activity. Observations of such changes in activity as the result of formation of mixed disulfides with glutathione have been made (13-16) and, in one instance, were found, due to the formation of a disulfide bond between two protein cysteine residues (16).
Carbonates as Substrates-The enzyme was capable of hydrolyzing each of three cyclic carbonates: ethylene carbonate, vinylene carbonate, and propylene carbonate. Stoichiometry was estimated for ethylene carbonate; at completion, 0.99 and 1.00 mol of ethylene glycol, respectively, were formed from 1.0 mol of ethylene carbonate in two trials.
When linear carbonates were used, diethyl, dipropyl, and allylmethyl carbonates were not substrates at concentrations of 1 and 10 mM. Also inactive in the same concentration range were certain related compounds that included the 2Ј,3Ј-cyclic nucleotides of adenine, uridine, and cytidine, esters and such heterocycles as (R)-pantolactone, tetrahydrofuran-2-carboxyl- ate, patulin, oxindole, 5-oxotetrahydrofuran-2-carboxylate, luminol, and ethylene and vinylene trithiocarbonates. The enzyme did not display esterase activity with 4-nitrophenyl acetate or ethyl acetate nor phosphatase activity with 4-nitrophenyl phosphate. Hydrolysis of ethylene carbonate was not inhibited by the serine esterase inhibitor paraoxon at 3 mM.
As with many of the substrates for the enzymes of detoxication (3), K m values for the cyclic carbonates were in the millimolar range, although V max values were substantial. Thus, for ethylene, vinylene, and propylene carbonates, the respective K m values were 26, 37, and 4 mM, and the respective V max values were 65, 56, and 8 mol min Ϫ1 mg Ϫ1 .
Since the product of propylene carbonate has an asymmetric center, evidence was sought for stereoselective hydrolysis. Under conditions previously employed for measuring circular dichroism with imidase (4), no evidence could be obtained for the stereoselective appearance of an asymmetric product over the course of 18 h of incubation with enzyme.
Imides as Substrates-Although a variety of heterocyclic compounds were tested and found inactive as noted above, one group, the imides, was effective as substrates. Phthalimide, the standard assay substrate for imidase, as well as 3,4-pyridyldicarboximide and 6-methyl dihydrouracil, were each substrates for the carbonate-hydrolyzing enzyme exactly as presented for rat liver imidase (4). The K m for phthalimide had previously been found to be 3 mM with imidase (4) and was 3.5 mM for the enzyme purified here.
When used as competitive inhibitors of propylene carbonate hydrolysis, phthalimide had a K i of 5 mM, and N-carbamoyl-␤alanine, another substrate for the imidase reaction, had a K i of 0.2 mM. Although ethylene carbonate appeared to be a competitive inhibitor of phthalimide hydrolysis, a K i could not be accurately estimated at these high concentrations, i.e. greater than 0.1 M.
Comparison of Imidase and Carbonate Hydrolase-Examination of Table I will confirm that the two enzyme activities were purified concomitantly and that, in each case, only a single protein species was apparent at each step of purification (Fig. 1). A dose-response curve with rabbit antibody raised against the electrophoretically homogeneous rat carbonate hydrolase was identically inhibitory to both carbonate and imide hydrolysis (data not shown). After overnight incubation at 4°C, a precipitate could be removed, leaving the supernatant solution without activity; upon resuspension of the antigen-antibody precipitate, both activities were observed. DISCUSSION Examination of the substrate specificity of the rat liver enzyme responsible for hydrolysis of cyclic carbonates led to the evaluation of compounds bearing other functional groups. Examples of linear organic carbonates, simple esters, lactones, lactams, and a number of a variety of heterocyclic compounds were inactive as substrates with the exception of imides. Several lines of evidence point to the identity of the newly isolated cyclic carbonate-hydrolyzing activity to an enzyme that this laboratory had previously described as an imidase (4), an enzyme that also plays a normal physiological role as an imidase in the pyrimidine salvage pathway (17). Thus, hydrolytic activity toward imides and carbonates co-purified throughout the various steps in the purification process ( Table I). The isolated enzyme appears to be the only protein with either type of catalytic activity in rat liver; no second peak of active protein was apparent at any stage of purification (Fig. 1). Purified polyclonal rabbit antibody raised against the homogeneous enzyme inactivated both activities in a similar manner. Finally, an imide and a cyclic carbonate were competitive inhibitors of carbonate hydrolase and imidase, respectively; the K i for phthalimide with ethylene carbonate and this preparation of the enzyme, 5 mM, was close to the K m obtained for phthalimide, 3 mM, with the imidase purified by either the original method (4) or the significantly different method presented here.
The mechanism previously outlined for the hydrolysis of imides (4) appears to apply equally to the activity toward cyclic organic carbonates. The reaction would be expected to take the form of protonation of the carbonyl group of the carbonate, thereby providing a strong electrophilic center for the addition of water. Upon such addition, ring opening would be followed by elimination of CO 2. The finding of this enzyme in rat liver provides a metabolic pathway for the conversion of cyclic organic carbonates to their respective glycols.