Nitrilase of Rhodococcus rhodochrous J1

Conversion into the active form by subunit association


T. Nagasawa, Department of Biomolecular Science, Faculty of Engineering, Gifu University, Gifu-Yanagido, 501-1193, Japan, Tel.: + 81 58 293 2647, Fax: + 81 58 2932647, E-mail:


Nitrilase-containing resting cells of Rhodococcus rhodochrous J1 converted acrylonitrile and benzonitrile to the corresponding acids, but the purified nitrilase hydrolyzed only benzonitrile, and not acrylonitrile. The activity of the purified enzyme towards acrylonitrile was recovered by preincubation with 10 mm benzonitrile, but not by preincubation with aliphatic nitriles such as acrylonitrile. It was shown by light-scattering experiments, that preincubation with benzonitrile led to the assembly of the inactive, purified and homodimeric 80-kDa enzyme to its active 410-kDa aggregate, which was proposed to be a decamer. Furthermore, the association concomitant with the activation was reached after dialysis of the enzyme against various salts and organic solvents, with the highest recovery reached at 10% saturated ammonium sulfate and 50% (v/v) glycerol, and by preincubation at increased temperatures or enzyme concentrations.

Nitrile-converting enzymes have attracted substantial interest as biocatalysts in preparative organic chemistry because of their capability to convert readily available nitriles into the corresponding higher-value amides and acids [1]. One example is the high molecular mass nitrile hydratase from Rhodococcus rhodochrous J1, which is currently employed in the large-scale industrial production of acrylamide [2–5] and nicotinamide [6]. Furthermore, nitrilases have a great potential as biocatalysts for the direct hydrolysis of nitriles to the corresponding acids and have advantages over the chemical hydrolysis, such as milder pH and temperature conditions and the absence of by-products. Most nitrilases exhibit a preference towards aromatic nitriles [7–14], but a few aliphatic nitrile-converting nitrilases have been described [15–17].

The isovaleronitrile-induced nitrilase of R. rhodochrous J1 [18], which was previously purified [19] and characterized regarding the suitability of ε-caprolactam as an inducer [20], has been studied genetically [21,22] and its biotechnological applicability has been investigated [23–26]. It has been shown to hydrolyze a wide range of aromatic nitriles, but did not act on aliphatic nitriles [19]. This contrasted with high turnover rates in the hydrolysis of acrylonitrile and methacrylonitrile to the corresponding acids by ε-caprolactam-induced resting cells [26]. In the present study, we have studied the cause of this inconsistent catalytic activity of different enzyme preparations.

Materials and methods


DEAE–Sephacel, phenyl–Sepharose CL-4B and the low molecular mass markers for SDS/PAGE were obtained from Pharmacia, Uppsala, Sweden. Unless otherwise stated, all other chemicals were from Wako, Osaka, Japan.

Microorganisms and culture conditions

R. rhodochrous J1, which has been previously identified [6], was cultivated in a 500-mL shaking flask containing 60 mL of 15 g·L−1 glucose, 7.5 g·L−1 sodium glutamate, 1 g·L−1 yeast extract, 0.5 g·L−1 KH2PO4, 0.5 g·L−1 K2HPO4, 0.5 g·L−1 MgSO4. 7 H2O and 5 g·L−1ε-caprolactam (pH 7.2). After 120 h reciprocal shaking at 28 °C, cells were harvested by 10 000 g centrifugation at 4 °C and washed twice with 50 mm potassium phosphate buffer (pH 7.8). A 10-fold concentrated cell suspension in this buffer was used for resting cell experiments.

Enzyme assay

The nitrilase activity was assayed at 20 °C in a 2-mL reaction volume containing 25 mm potassium phosphate buffer (pH 7.8), 10 mm benzonitrile or 25 mm of other nitriles such as acrylonitrile, and an appropriate amount of biocatalyst. After 10 min, the reaction was stopped by the addition of 0.2 mL 2 m HCl, and the enzyme products were quantified by HPLC or gas chromatography. One unit of nitrilase activity was defined as the amount of enzyme catalyzing the formation of 1 µmol benzoate per min. In activity tests using enzyme preincubated with benzonitrile, the amount of benzoate was corrected for the endogenous benzoate in the enzyme sample formed during preincubation. The activity towards all other substrates besides benzonitrile was determined after dialysis of the enzyme against buffer A containing 10% ammonium sulfate and 50% (m/v) glycerol.

Enzyme purification

All steps of the enzyme purification, which has been shortened compared to a previously described procedure [19], were performed at 4 °C using 10 mm potassium phosphate buffer, pH 7.0, containing 1 mm dithiothreitol (buffer A) unless otherwise specified. Centrifugation was performed for 30 min at 20 000 gε-caprolactam-induced cells (15.3 g dry mass obtained from 3 L culture broth) were suspended in 500 mL 100 mm potassium phosphate buffer, pH 7.0, containing 1 mm dithiothreitol, disrupted by 30 min ultrasonication (19 kHz, Insonator 201 m, Kubota, Tokyo) and centrifuged. After fractionation with (NH4)2SO4 (20–40% saturation) and centrifugation, the 40% supernatant, designated as partially purified enzyme, was dialyzed against buffer A containing 10% (v/v) ethanol and loaded on a DEAE–Sephacel column (2.5 × 25 cm) equilibrated with this buffer. The enzyme was eluted at 150 mm KCl in a linear 0–600 mm gradient in this buffer, and all 5-mL fractions had 0.5 mL buffer A containing ammonium sulfate at 90% saturation added and was assayed against the two substrates. The enzyme pool was dialyzed against buffer A plus 10% (v/v) ethanol and 10% saturated ammonium sulfate and applied to a phenyl–Sepharose CL-4B (column 2.5 × 25 cm) equilibrated with this buffer. The nitrilase activity was eluted by buffer A containing 40% (v/v) ethyleneglycol, and tested as described above. The purified enzyme was dialyzed against buffer A containing 10% ammonium sulfate and 50% (m/v) glycerol and stored at −20 °C.

Molecular mass determination

The molecular mass of the native enzyme was estimated by HPLC with a TSK G-3000SW column (0.75 × 60 cm, Toyo Soda, Tokyo, Japan) at a flow rate of 0.7 mL·min−1 using 100 mm potassium phosphate buffer, pH 7.0, plus 0.2 m NaCl as eluent. The molecular mass was calculated from a linear regression curve obtained from the mobilities of the standard proteins acyl-CoA oxidase (600 kDa), glycerol dehydrogenase (390 kDa) both from Toyobo, Osaka, Japan, and glutamate dehydrogenase (290 kDa), lactate dehydrogenase (142 kDa), enolase (67 kDa), adenylate kinase (32 kDa) and cytochrome c (12.4 kDa) from Oriental Yeast, Tokyo, Japan. The sedimentation equilibrium was determined according to the method of Van Holde & Baldwin [27] in a Spinco Model-E Beckman using an An-G rotor and double cells of different side-wedge angles. Multicell operations were carried out for five enzymes samples of different initial concentration ranging from 0.9 to 4.7 mg protein·mL−1 in the presence of 7.5 mm phenylacetonitrile to accomplish the enzyme association. Phenylacetonitrile was found to associate with the enzyme to a similar extent as benzonitrile. The rotor was centrifuged at 8059 g for 20 h, and interference patterns were photographed at intervals of 60 min. The relationship between the concentration of the enzyme and the fringe shift was determined using synthetic boundary cells. Furthermore, a method to determine the molecular mass based on the turbidimetric measurable scattering of light of distinct wavelength by large molecules such as proteins, which offers real-time kinetics, was performed [28,29]. The measurement was made at 633 nm using a dynamic light scattering spectrophotometer DLS-700 and a differential refractometer RM-102 (Otsuka Electronics, Tokyo, Japan). The solutions were filtered through 0.2-µm membrane filters prior to light scattering in a 1.2-mm diameter cylindrical cell. The refractive index increment was 0.18 mL3·g1. The averaged molecular mass m was calculated from the 90° angle scattering data using the equation m = Ro·Kc−1 with Ro the Rayleigh ratio, K the light scattering constant, and c the protein concentration determined photometrically at 280 nm using an extinction coefficient of 0.93 mg−1·cm−1·mL−1.

Other analytical methods

Aromatic and aliphatic unsaturated acids were analyzed by a Shimadzu LC-5 A HPLC with a RP C18 column (4.6 × 150 mm, M & S Instruments) and a 10-mm KH2PO4, pH 2.8 (H3PO4)/acetonitrile (3 : 2, v/v) eluent at a flow rate of 1 mL·min−1 monitored at 220 nm. Aliphatic, saturated acids were determined by gas chromatography using a Shimadzu GC-4 CM equipped with a flame ionization detector and a 80/100 mesh Gaskuropack 56 (Gaskurokogyo, Tokyo, Japan) column, and N2 as carrier gas at a flow rate of 50 mL·min−1. SDS/PAGE was performed in 12.5% (m/v) polyacrylamide gels [30] and proteins in the gel were stained with Coomassie Blue R-250. Protein was quantified by the Bradford method [31] using BSA as standard. The N-terminal amino-acid sequence of the enzyme was determined by Edman degradation using a Perkin Elmer Applied Biosystems 470 A gas-phase amino-acid sequencer. The antiserum of the isovaleronitrile-induced R. rhodochrous J1 nitrilase and the Ouchterlony plates were prepared as previously described [19].


Enzyme purification

The enzyme purification resulted in the same yields whether acrylonitrile or benzonitrile was used as substrate, indicating that R. rhodochrous J1 contains only one nitrilase for both the aliphatic and the aromatic substrate. This was supported by the finding that the nitrilase activity for both substrates always eluted as a single peak, and additionally by SDS/PAGE analysis showing a single band of the purified enzyme with a molecular mass of 40 kDa consistent with previous SDS/PAGE results [19]. Gel filtration experiments revealed a molecular mass of 80 kDa confirming the assumption of a homodimeric structure of the purified, nontreated enzyme.

Immunological and molecular comparison of isovaleronitrile- and ε-caprolactam-induced nitrilase

Acrylonitrile was converted by resting cells of ε-caprolactam-induced R. rhodochrous J1 [26], but not by the purified enzyme obtained from isovaleronitrile-induced cells [19]. In order to rule out that the differing substrate specificities were due to two differently inducible enzymes, immunodiffusion tests were performed. The antiserum reactive to the isovaleronitrile-induced nitrilase cross-reacted with the purified, ε-caprolactam-induced nitrilase to the same extent as with the purified isovaleronitrile-induced enzyme itself, suggesting that isovaleronitrile-induced and ε-caprolactam-induced nitrilase were identical. Furthermore, the sequence of the 16 N-terminal amino acids of the differently induced enzymes were identical, and the differently induced enzymes showed the same molecular mass on both subunit and holoenzyme level as shown by SDS/PAGE and HPLC gel filtration, respectively. This lack of unambiguous evidence strongly suggested that isovaleronitrile- and ε-caprolactam-induced nitrilase activities are based on a single enzyme.

Recovery of the purified enzyme

In contrast to resting cells, the purified enzyme dialyzed against buffer A did not hydrolyze acrylonitrile. The activity of the purified enzyme toward acrylonitrile was recovered after dialysis of the enzyme against a number of salts and organic solvents (Table 1). The highest recovery was obtained in buffer A containing 50% (v/v) glycerol plus 10% saturated ammonium sulfate. A higher content of ammonium sulfate resulted in the precipitation of the enzyme. The nitrilase activity towards acrylonitrile was also regained at higher enzyme concentrations without glycerol. Furthermore, enzyme recovery was reached by 30 min incubation at increased temperatures of 35 °C, and after enzyme preincubation with 10 mm benzonitrile. The recovery by benzonitrile and higher temperatures was observed even at low enzyme concentrations of 30 µg·mL−1. However, no recovery of the activity was obtained after preincubation with benzoic acid, acrylonitrile, acrylic acid, or without benzonitrile.

Table 1. Recovery of the nitrilase activity by salts and organic solvents. A 1.6-mL solution containing organic solvents and/or salts in buffer A was added to 0.4 mL of the partially purified enzyme (20.8 mg·mL−1) previously dialyzed against buffer A. After storage on ice for 2 days, 50 µL of the enzyme solution was used for the standard activity assay. Activities were calculated from the amount of acid product formed after 10 min reaction time and determined by HPLC. The sensitivity of the assay was 1 nmol·min−1·mL−1.
 Activity (µmol·min−1·mL−1)
Preincubation mediumBenzonitrileAcrylonitrile
Buffer A2.60
+ 10% saturated (NH4)2SO45.00.21
+ 20% saturated (NH4)2SO48.20.68
+ 30% (v/v) glycerol4.30
+ 50% (v/v) glycerol6.90.15
+ 30% (v/v) glycerol +10% saturated (NH4)2SO410.52.3
+ 50% (v/v) glycerol +10% saturated (NH4)2SO412.66.3
+ 10% (v/v) ethanol5.20
+ 10% (v/v) ethanol +10% saturated (NH4)2SO413.64.0
+ 10% (v/v) acetone5.90.22
+ 10% (v/v) acetone +10% saturated (NH4)2SO411.50.16
+ 30% (v/v) glycerol + 5% (w/v) NH4Cl5.20.37
+ 30% (v/v) glycerol + 5% (w/v) MgSO4·7 H2O6.50.19
+ 30% (v/v) glycerol + 5% (w/v) KCl4.40.25

The purified enzyme dialyzed against buffer A converted benzonitrile with an initial time lag in its velocity (Fig. 1A). After dialysis against buffer A containing 50% (v/v) glycerol and 10% saturated ammonium sulfate, the enzyme acted on both benzonitrile and acrylonitrile without showing a time lag in the hydrolysis of benzonitrile (Fig. 1B).

Figure 1.

Timecourse of the nitrilase-catalyzed conversion of benzonitrile and acrylonitrile. The initial nitrilase velocity was followed by HPLC after acrylonitrile or benzonitrile were added to the standard reaction mixture at 20 °C containing 30 µg·mL−1 purified enzyme previously dialyzed against buffer A (A) or buffer A containing 50% (v/v) glycerol and 10% saturated ammonium sulfate (B).

The purified enzyme dialyzed against buffer A converted aromatic nitriles, but not aliphatic, unsaturated nitriles (Table 2). The nitrilase activity toward aliphatic, unsaturated nitriles was recovered by dialysis against ammonium sulfate and glycerol. However, aliphatic nitriles containing the cyano group bound to a saturated carbon were converted neither by resting cells nor by the recovered enzyme, as shown with acetonitrile, propionitrile and n-butyronitrile. With resting cells, the highest turnover rate was found with acrylonitrile, whereas the best substrate of the (NH4)2SO4-recovered enzyme was benzonitrile.

Table 2. Comparison of the nitrilase substrate specificity of resting cells and the purified enzyme in the presence and absence of ammonium sulfate. The specific activity toward benzonitrile, which was taken as 100%, corresponded to either 3.65 U·mg−1 resting cells (dry weight), 2.6 U·mg−1 purified enzyme previously dialyzed against buffer A [–(NH4)2SO4], or 14.6 U·mg−1 purified enzyme dialyzed against buffer A including 10% saturated ammonium sulfate [+(NH4)2SO4].
 Relative activity (%)
  Purified enzyme
SubstrateResting cells –(NH4)2SO4+(NH4)2SO4
3-Cyanopyridine  95.7  63  75.9
Acrylonitrile128    058.8
Methacrylonitrile9.20    07.45
Crotononitrile27.1    1.6811.5

Enzyme association

The purified enzyme dialyzed against buffer A revealed a single peak in HPLC gel filtration corresponding to a molecular mass of 80 kDa. Under recovery conditions such as dialysis of the enzyme against buffer A containing 20% saturated ammonium sulfate and preincubation with 10 mm benzonitrile, a second peak corresponding to a 410-kDa nitrilase polymer, probably a decamer, was observed (Fig. 2). The polymer predominated when 5 mm benzonitrile was present in the HPLC eluent, however, using the standard HPLC eluent lacking benzonitrile, the dimer peak was more pronounced due to the enzyme dilution during gel filtration. Because unambiguous distinction between an octamer, decamer and dodecamer by gel filtration was not possible, sedimentation and light scattering analyses were performed. By sedimentation equilibrium analysis, assuming a partial specific volume of 0.74 mL·g−1, the molecular mass of the polymerized enzyme was determined to 412 kDa. Based on a subunit molecular mass of 40 kDa determined by SDS/PAGE [19] and a mass of 40 189 Da deduced from the DNA sequence [21], a decameric structure of the associated nitrilase seemed to be most reasonable. Further molecular mass determinations by light scattering, which were in good agreement with the values obtained by HPLC gel filtration, confirmed that the association of the dimer to the proposed decamer was caused by increased temperatures (Fig. 3). Intermediate oligomeric nitrilase forms could not be detected by gel filtration, ultracentrifugation or light scattering. Furthermore, previous results of the enzyme purification never showed activity in more than one eluted protein peak [19], also suggesting that other enzyme oligomers besides a dimer and a decamer were not stable. Assuming an equilibrated interconversion of five dimers and one decamer, the equilibrium constant Keq of the dimer association was estimated as the ratio (concentration of the decamer)/(concentration of the dimer)5. The standard free-energy change Δ was calculated as –RT lnKeq with R the gas constant and T the absolute temperature. The association was favored at increased temperatures, concomitant with higher equilibrium constants (Keq = 1.1 × 1023, 2.1 × 1025 and 2.9 × 1027 L−4·mol4 at 10, 20 and 30 °C, respectively) and lower values of the free-energy change (ΔG° = –120, −140, and −160 kJ·mol−1 at 10, 20 and 30 °C, respectively). The positive value of the enthalpy change ΔH° = 450 kJ·mol−1 and the entropy change ΔS° = 2.0 kJ·K−1·mol−1, obtained from the temperature dependence of ΔG° and the relationship ΔG° = ΔH° – TΔS°, indicated that the association is an entropy-driven process.

Figure 2.

Elution profile of the nitrilase in HPLC gel filtration. HPLC analysis of the purified nitrilase (58 µg) was performed after dialysis of the enzyme against buffer A (A) or buffer A plus 50% (v/v) glycerol and 10% saturated ammonium sulfate (B), and after 30 min preincubation in buffer A containing 10 mm benzonitrile (C). Both the standard eluent (left) and the eluent plus 5 mm benzonitrile (right) were used. 

Figure 3.

Effect of temperature and enzyme concentration on the nitrilase assembly. Different enzyme concentrations were incubated for 30 min in buffer A at different temperatures prior light scattering analysis for the determination of the enzyme molecular mass. 

Light scattering also revealed polymerization of the nitrilase after enzyme dialysis against 10% saturated ammonium sulfate and 50% (v/v) glycerol, preincubation with 10 mm benzonitrile (Fig. 4) or at higher enzyme concentrations (Fig. 5). Other aromatic nitriles like phenylacetonitrile, 4-toluenenitrile, 4-chlorobenzonitrile (both 10 mm) induced the association to a similar extent as did benzonitrile. While phenylacetonitrile induced the association, it was barely converted by the nitrilase. No association was observed by enzyme preincubation with 10 mm benzoic acid, 4-methylbenzoic acid, acrylonitrile, acrylic acid, methacrylonitrile, or acetonitrile.

Figure 4.

Effect of HgCl2 and benzonitrile on the association of the enzyme. A solution of 36 µg·mL−1 of the enzyme was incubated in buffer A without addition, in buffer A plus 10 mm benzonitrile, or in buffer A plus 10 mm benzonitrile and 250 µm HgCl2. After different incubation times, the molecular mass of the nitrilase was determined by light scattering. 

Figure 5.

Effect of the enzyme concentration and HgCl2 on the association and activity of the nitrilase. After 20 min preincubation at 20 °C and 160 r.p.m. shaking with different enzyme concentrations in 50 mm potassium phosphate buffer, pH 7.0, containing 0.1 m NaCl, in the presence and absence of 250 µm HgCl2, the nitrilase association and the activity were measured by light scattering analysis and the standard activity test, respectively. 

Recovery of the nitrilase correlated with its subunit association as shown in Fig. 5 for increased enzyme concentrations.

Effect of HgCl2 on the nitrilase assembly and activity

Because of a well-documented thiol-dependence of nitrilases [9–12,19,21], 250 µm HgCl2 totally inhibited the enzyme activity, but did not affect the enzyme assembly at higher enzyme concentrations (Fig. 5). At diluted enzyme concentrations (36 µg·mL−1), the enzyme association was inhibited by preincubation with 10 mm benzonitrile plus 250 µm HgCl2 (Fig. 4), indicating that HgCl2 inhibits not only the activity but also the nitrilase association.


The reversible subunit association of enzymes has been repeatedly described. For example, a pigeon liver malic enzyme, which is either composed of a monomer or a dimer, was associated with the fully active tetramer by basic pH and high temperatures or ionic strength, but was dissociated by increased ammonium sulfate concentrations of 0.2 m[32]. Further examples are a RecA protein, which self-assembled with increased ion and protein concentrations [33], and the dissociated form of a dehydrogenase, that aggregated to the active dodecamer by high enzyme concentrations [34].

Previously, two nitrilases have been reported to have an association/dissociation equilibium, as shown by HPLC gel filtration. A nitrilase from Rhodococcus (formerly Nocardia) sp. NCIB 11216 resembled our enzyme with regard to a substrate preference for benzonitrile and a time-dependent benzonitrile-induced activation and association of a 47-kDa monomer to a 560-kDa dodecamer, accelerated by increased temperature and enzyme concentration [9,10]. Furthermore, a Rhodococcus ATCC 39484 nitrilase demonstrated a substrate-induced activation accompanied by the aggregation of 40-kDa subunits to form a 560-kDa complex [35]. We have used light scattering in order to study the association of the Rhodococcus rhodochrous J1 nitrilase in more detail. Besides the polymerization effect of benzonitrile and increased temperature or enzyme concentration, which was also described for the NCIB 11216 enzyme, we have found a number of salts and organic solvents that trigger the association of the Rhodococcus rhodochrous J1 nitrilase. Furthermore, the NCIB 11216 enzyme was not described to convert aliphatic nitriles such as acrylonitrile, even after association, and no relation between the substrate specificity and the compounds which have an association effect has been described. The previously reported inconsistency in acrylonitrile hydrolysis by the purified Rhodococcus rhodochrous J1 enzyme and resting cells [19,26] was explained here by the finding that acrylonitrile is a substrate only; it is not capable of assembling the enzyme. The previously reported activity of purified Rhodococcus rhodochrous J1 nitrilase towards a wide range of aromatic, but not aliphatic, nitriles [19], was probably due to an aromatic substrate-induced association, in accordance with the aggregation effect of benzonitrile and some other aromatic nitriles described here. Furthermore, the similarity of enzyme in resting cells and the enzyme dialyzed against glycerol and ammonium sulfate in terms of stability and substrate specificity might suggest that the decameric enzyme predominates in cells with high protein and ion concentrations, i.e. the in vitro conditions of high protein and salt concentrations probably resembling cytoplasmatic conditions in vivo. In another report on a nitrilase from a Fusarium species it has been shown that in addition to dimeric and dodecameric forms of a nitrilase, many enzymes with intermediate aggregation states exist that also show nitrilase activity [36]. However, we could not observe stable intermediate enzyme forms during the nitrilase assembly, and we have only found nitrilase activity in the dimer and the proposed decamer. The nitrilase assembly of the dimer to the decamer followed a reactivation after incubation with benzonitrile, increased salt, organic solvent or enzyme concentrations, or at higher temperatures. All these conditions increase the probability that enzyme dimers meet via a hydrophobic interaction. For example, increased enzyme or salt concentrations lead to dehydration of the enzyme and therefore an enhanced hydrophobic environment. The hydrophobic effect resulting from the presence of aromatic nitriles such as benzonitrile might change the conformation of the nitrilase exhibiting hydrophobic enzyme sites, thereby enabling nitrilase assembly; the inability of aliphatic nitriles to aggregate the enzyme might be due to the lack of an hydrophobic aromatic moiety. However, the mechanism of benzonitrile-induced assembly might differ from the one at higher enzyme concentrations, as the association at higher enzyme concentrations was not inhibited by HgCl2, but the benzonitrile-induced aggregation was inhibited by HgCl2. It is known that benzonitrile interacts with thiol groups in the catalytic enzyme site during nitrile hydrolysis. The inhibition of thiol groups by HgCl2 might hinder the binding of benzonitrile and therefore also the enzyme assembly. The participation of thiol groups in the substrate-induced association from a dimer to a tetramer has also been reported for an l-lysine ε-dehydrogenase [37]. Further studies of the Rhodococcus nitrilase are planned to study more systematically the ability of other compounds besides benzonitrile to induce enzyme association and to elucidate the molecular basis of the reversible association.


M. W. was supported by the Japanese Society for the Promotion of Science.


  1. Enzymes: nitrilase (EC, nitrile hydratase (EC, l-lysine ε-dehydrogenase (EC 1.4.1) (S)-malate:NADP+ oxidoreductase (oxaloacetate decarboxylating) (EC