L-Ribulose 5-phosphate 4-epimerase from Aerobacter aerogenes. Evidence for a role of divalent metal ions in the epimerization reaction.

Abstract l-Ribulose 5-phosphate 4-epimerase from Aerobacter aerogenes was inactivated by treatment with EDTA and was reactivated to varying extents by the addition of divalent metal ions in the order: Mn++ g Co++ g Ni++ g Ca++ g Zn++ g Mg++. When optimal Mn++ was present, the homogeneous enzyme had a specific activity of 70 µmoles-1 min-1 mg of protein at 28° and pH 7.2. This value is about five times greater than that displayed by the crystalline enzyme as isolated and assayed in the absence of added metal ion. In other mechanistic studies, l-ribulose 5-phosphate 4-epimerase was found to be stable to treatment with sodium sulfite and arsenite in the presence of a thiol compound. It was also stable to sodium borohydride in the presence or absence of substrate. Further, a reaction of tetranitromethane with the enzyme-substrate complex could not be detected. Possible mechanisms for l-ribulose 5-phosphate 4-epimerase are discussed.

When optimal Mn++ was present, the homogeneous enzyme had a specific activity of 70 pmolesl mine1 mg of protein at 28' and pH 7.2. This value is about five times greater than that displayed by the crystalline enzyme as isolated and assayed in the absence of added metal ion.
In other mechanistic studies, L-ribulose 5-phosphate 4epimerase was found to be stable to treatment with sodium sulfite and arsenite in the presence of a thiol compound.
It was also stable to sodium borohydride in the presence or absence of substrate.
Further, a reaction of tetranitromethane with the enzyme-substrate complex could not be detected.
L-Ribulose 5-phosphate 4-epimerase from Aerobacter aerogenes, which catalyzes the interconversion of L-ribulose-5-P and Dxylulose-5-P, is unique among 4-epimerases in that it neither contains nor requires NAD+ for catalysis (1). In addition, no evidence was found for the presence of chromophoric substances in the crystalline enzyme, nor did additional cofactors have an influence on the activity (1). In contrast, there is substantial evidence that epimerization by UDP-glucose 4-epimerase involves an oxidation-reduction mechanism utilizing NAD+ as the electron acceptor and donor (2)(3)(4)(5)(6).
Thus, if n-ribulose5P 4epimerase catalyzes a similar oxidation-reduction reaction, another as yet unrecognized electron acceptor must perform the function of NBD+.
It has also been observed that there is no kinetic isotope effect when n-[4-T]xylulose 5-phosphate is used as substrate (7). This is in contrast to UDP-glucose 4-epimerase where a normal isotope effect is observed (8). In this connection, it may be * Michigan Agricultural Experiment Station Journal Article 5698.
$ Present address, Department of Pharmacology, University of Wisconsin Medical Center, Madison, Wisconsin 53706. significant that the substrates, n-ribulose-5-P and n-xylulose-5-P, are open chain carbohydrates lacking a nucleotide moiety and possessing a carbonyl group two carbon atoms removed from the epimerization site. Undoubtedly, this confers chemical properties on the substrate which are considerably different from those of nucleotide sugars. For these reasons, mechanisms of 4-epimerization of L-ribulose-5-P other than oxidation-reduction have been considered.
The results presented in this paper indicate that L-ribulose-5-P ii-epimerase requires divalent metal ions for activity, and that different divalent metal ions activate to varying estents. In addition, an exploration of a number of mechanistic possibilities involving oxidation-reduction, or carbanion and carbonium ion formation, gave negative results.

MATERIALS
Chemicals-L-Ribulose-5-P was prepared according to the procedure of Anderson (9). Spectra-pure sulfate salts of Mn++, Mg+f, Ni*, Zn++, and Co++ were obtained from Johnson, Mathley and Co., Ltd. Chloride salts of the divalent metal ions were obtained from Mallinckrodt, Inc. Tris base was obtained from Sigma Chemical Co.
Enzymes-L-Ribulose-5-P 4-epimerase from A. aerogenes (constitutive for n-arabinose operon, uracil auxotroph designated "u-i-") and n-xylulose-5-P phosphoketolase from Leuconostoc mesenteroides were purified by procedures previously reported (1). A triose phosphate isomerase-ol-glycerol phosphate dehydrogenase mixture was obtained from Calbiochem. METHODS L-Ribulose-5-P 4-Epimerase Assay-The 4-epimerase was assayed by two methods designated as "continuous" and "twostep." The continuous assay involved the conversion of Lribulose -5 -P to oc-glycerol phosphate with the concomitant oxidation of NADH utilizing phosphoketolase, triose phosphate isomerase, and a-glycerol phosphate dehydrogenase as coupling enzymes (1). The two-step assay involved the epimerization of n-ribulose-5-P to n-xylulose-5-P in Tris-Hepesl buffer, pH 8.0, in the absence of coupling enzymes. The 4-epimerase was then inactivated by the addition of acetic acid and heating in a boiling water bath for 1 min. The pH was readjusted to 7.0 by the addition of ammonium hydroxide, and an aliquot of the assay mixture was assayed for n-xylulose-5-P by measuring the amount of P;TADH osidized upon enzymatic conversion of Dxylulose-5-P to cY-glycerol phosphate in the presence of phosphoketolase, triose phosphate isomerase, and or-glycerol phosphate dehydrogenase (10). Only small amounts of metal or chelator from the incubation with epimerase was carried over in the analytical step for n-xylulose 5-phosphate.
In separate controls it was shown that both the metal ions and chelators used affected the n-xylulose j-phosphate values less than 10%.
Where indicated, all traces of metal ions were removed from the reagents and glassware used in the first step of the twostep assay. The glassware was soaked overnight in 4 N HCl and then was extensively rinsed with double quartz-distilled water. The buffers used in the assay were passed over Chelex resin and the pH was adjusted with solid Tris base. The Lribulose-5-P was also passed through a Chelex column, and the pH was adjusted to 6.0 with Tris base prior to use.
One unit of enzyme activity is defined as the amount required to epimerize 1 pmole of L-ribulose-5-P per min at 28" and pH 7.2.
Protein concentration was determined using &0:&0 ratio method of Warburg and Christian (11).
Removal oj EDTA by Chromatography on Sephadex G-25-EDTA was removed by passing the enzyme through a Sephadex G-25 (0.6 X 11 cm) column. The column, of sufficient length to clearly separate a mixture of blue dextran and 32P, had previously been washed free of metal ions with 10e2 M EDTA followed by extensive washing with double quartz-distilled water. Prior to use, the column was equilibrated with Tris-Hepes buffer which had been freed of metal ions by passage through Chelex resin. ,411 of the glassware used had been soaked overnight in 4 N HCl and extensively washed with double quartz-distilled water.
Tests for Lipoate and Cystine-L-Ribulose-5-P &epimerase (0.4 pmole, 85% pure) was incubated at room temperature with 10-l M, lo+ M, and lo+ M Wa2S03 in 0.05 M Tris-Hepes buffer, pH 8.0. The activity remaining after 30 min was determined using the continuous assay as described under "Nethods." Alternatively, the enzyme was incubated as above with lo-" JI mercaptoethanol or lop3 1\1 dithiothreitol. After 30 min, sodium arsenite was added to a concentration of 10-l nr to lo-" 11. Aliquots of the enzyme were removed after 5 min and the activity determined in the continuous assay.

Metal
Ion Activation lcflect of Xefal Chelators on L-Ribulose-5-P 4-lfpitwrasc lctivity-L41thougli L-ribuloxe-5-P 4-epimerase does not, require added organic or met,nl cofactors for activity, it is possible that tightly bound metal ions may participate in catalysis. As n first test of this hypothesis, the 4-epimerase was incubated with a series of metal chelators for various incubation times ( Fig. 1). A wide variety of responses was observed including both inhibition and stimulation of activity.
Complete inhibition was obtained only with 1OV hf EDTA.
In contrast, an initial activation was obtained with either dithiothreitol, 1 m&f o-phenanthroline, or 1 M mercaptoethanol. These results suggest that the enzyme, as isolated, may bind a variety of metals including those species which inhibit.
Thus, EDTA may inactivate by removing all metal ions, whereas other metal chelators such as o-phenanthroline may activate by preferentially complesing certain inhibitory metal ions. This possibility is supported by the fact that the stability constants for Mn-o-phenanthroline or Mn-8-hydroxyquinoline chelabes are two or more decades lower than those for heavy metals, whereas the Mn-EDTA stability constant is extremely high and in the same range as those for the heavy metals (12). However, the possibility that the met'al chelators may nonspecifically activate or inactivate by means other than removal of a metal ion must be considered.
Activity of L-R&lose-5-P 4-Epimerase after Removal oj EDTA --In order to determine whether inactivation of the L-ribulose-5-P 4-epimerase by EDTA was due to chelation of metal ions or to binding of EDTA to the enzyme, it was necessary to determine the activity of the treated enzyme after removal of the EDTA.
For this purpose, the enzyme was inactivated by incubation with 10e2 M EDTA for 1 hour at room temperature. The EDTA was then separated from the enzyme by passage through a Sephadex G-25 column as described under 'Ylethods." All buffers used to elute the enzyme from the column and used in the enzyme assay were treated to remove trace contaminations of metal ions as described under "Methods." The enzyme activity recovered from the Sephades column varied from 0 to 10yO of the initial activity.
In the experiment cited in Table I no activity remained.
In other cases where low activity remained, it was not ascertained how much was attributable to inaccuracies of the two-step assay, incomplete removal of metal, or recontamination by metal during passage 30% through Sephadex. At any rate when metal ions were added to TABLE I the first step of the two-step assay, the actirit,y was greatly increased; the increase depended upon t,he metal ion species pres-Divalent metal ion activation of z-ribdose-5-P .$-epirtrerasc ent as detailed below. The 4-epimerase (85y0 pure) m-as dialyzed overnight. against 0.05 M Tris-Hepes buffer, pH 8.0, incubat.ed for 1 hour wit.h 10-* Divalent Metal Ion Specificity-To det,ermine the activating M EDTA, and passed through a Sephadex G-25 column (0.6 x 11 capability of various metal ions, the 4-epimerase was dialyzed cm) which had been washed free of cations with EDTA and overnight against 0.05 M Tris-Hepes buffer, pH 8.0, treated with equilibrated with 0.05 M Tris-Hepes buffer, pH 8.0. Activity was EDTA, freed of EDTA on Sephadex G-25, and assayed in the determined in the two-step assay containing spectra-pure metals presence of varying quantities of specific divalent metal ions as at the levels indicated in the table. Precautions were taken to described under "Methods." The metal salts used were freshly remove the contaminating metals from the glassware and the prepared solutions of spectrographically analyzed metal salts reagents as described under "Methods." containing less than 5 iprn of most other metals. Under the conditions and concentrations used no precipitation of Mn was observed either in reagents or incubation mixtures.
Since, in a preliminary test, the same activity was obtained when the enzyme was previously incubated with 10m3 ~1 Co++ for 0, 10, or 30 min, the enzyme was not previously incubated with metal ions prior to assaying.
Rather, metal ions and substrate were incubated to allow temperature equilibration of the assay mixture, and the reaction was started by the addition of a very small volume of the 4-epimerase.
The results in Table I show that dialysis against Tris-Hepes buffer resulted in an activity loss of about a-fold, presumably due to loss of metal ion. Following treatment with EDTA and passage through Sephades G-25 no activity remained.
After incubation with metals the highest 4-epimerase activity was obtained with Mn++, and this activation occurred at the lowest divalent cation concentration.
A 17-fold stimulation over the activity present in the dialyzed preparation was observed. The MnS04 concentration was almost optimal at 1Cr5 M (17-fold versus B-fold stimulation at 1OF M), whereas 10V4 M NiS04 and 10e3 11 or higher MgS04 were required for the maximum activation. Similar activating effects were obtained using Cl-salts of metal ions, thus indicating that a specific anion is not required.
To show the importance of EDTA treatment in obtaining full act.ivation as described above, the 4-epimerase (90% pure) was dialyzed for 2 hours against 0.05 in barbital buffer, pH 8.0, then incubated for 1 hour with 1OF M Co++, IIn++, Zn++, and MgClz salts, and assayed in the two-step assay. Contaminating metals were not removed from the glassware or reagents.
The results presented in Table II indicate that only Mn++ can stimulate 4-epimerase which had not been treated with EDTA.
However, only a a-fold stimulation was obtained, indicating that Mn++ was not able to activate completely without prior EDTA treatment. These results suggest that various nonactivator divalent cations are bound to the Mn++ binding site of the 4-epimerase as isolated.
These dissociate slowly even in the presence of Mn++, as reported for phosphoglucomutase by Ray (13). Specific Activity of Crystalline L-Ribulose-5-P &E'pimerase in Presence of Mn++--Since the preceding results strongly indicated that L-ribulose-5-P 4-epimerase was activated by metal ions, Mn++ being the most active, it was necessary to redetermine the specific activity of homogeneous Mn+" 4-epimerase.
The L-ribulose-5-P 4-epimerase was twice crystallized as previously reported (1). The second cryst,als were at least 98% pure as determined by polyacrylamide gel electrophoresis. The enzyme solution was then incubated with lo-* M EDTA, and the EDTA was removed by passage through a Sephadex G-25 column as before. The metal-free enzyme was incubated with 10-k x MnS04 (spectra-pure) and assayed with the two-step assay to which 10m5 bf MnS04 had been added. A specific ac- Mechanistic Studies E$ect of Arsenite and Su&ite---Since the 4-epimerase is devoid of NAD+, it was considered possible that the epimerization process may involve an oxidation-reduction mechanism using enzyme-bound oxidized lipoic acid or cystine as an electron acceptor.
If this were true, either arsenite or sulfite should inhibit the 4-epimerase since dihydrolipoate and cysteine irreversibly react with sulfite.
Accordingly, L-ribulose-5-P 4-epimerase (85% pure) was incubated with 10-l M, 10e2 M, low3 M, and 10B4 M sodium sulfite or sodium arsenite with and without prior incubation with either 10e2 M mercaptoethanol or 10e3 M dithiothreitol to reduce any Activation of L-Ribulose 5-Phosphate Q-Epimerase Vol. 247, No. 10 disulfide bonds, as described under "Methods." The epimerase was inactivated no more than 10% in any of these experiments, as determined in the continuous assay.
Eject of Sodium Borohydride Treatment-Less than 20% of the epimerase activity was lost on incubation with NaBH4 in the presence or absence of substrate using the method of Ingram and Wood (14). In addition, when the 4-epimerase was incubated with borohydride in the presence of L-ribulose-5-P and lop4 M Co&, there was less than a 10 $$ loss in activity.
Likewise, the 4-epimerase at pH 6.5 in 0.05 M phosphate buffer was not inactivated by borohydride either in the presence or absence of substrate.
The conditions used in these experiments are routinely used by others in this laboratory to obtain complete inactivation of 2-keto-3-deox@phosphogluconic aldolase in the presence of pyruvate, or of the pyridoxal phosphate-containing L-threonine dehydrase.
Test for Carbanion Intermediate-Carbanions react with tetranitromethane with the liberation of nitroformate which absorbs at 350 nm. Christen and Riordan (15,16) have used this reagent to demonstrate the presence of a carbanion intermediate in both the yeast (Class II) and the muscle (Class I) fructose diphosphate aldolase-catalyzed reactions. Thus, it is reasonable to assume that if 4-epimerization of L-ribulose-5-P and n-xylulose-5-P were proceeding via a carbanion intermediate, it should be detected by tetranitromethane.
Nitroformate was produced in the presence of either L-ribulose 5-phosphate alone or the 4-epimerase (80% pure) alone in imidazole, glycylglycine, and Tris buffers; pure 4-epimerase did not react with tetranitromethane in Tris buffer. However, the rate of the reaction with L-ribulose-5-P plus the 4-epimerase was not significantly greater than the sum of the individual rates of reaction.
In addition, increasing the amount of enzyme did not increase the rate of the nitroformate formation.
In a control with FDP aldolase, it was possible to obtain a net increase in 350 nm absorbance with the rate of reaction being dependent upon the aldolase concentration. 2. Proposed dealdolization-aldolization mechanism for L-ribulose-5-P 4.epimerase.
M depicts a divalent metal ion in the active site and B: indicates a base function in the active site. atom 4 of the substrate.
However, previous results have indicated that NAD+ is not present and is not required for enzyme activity (1). In addition, there were not chromophoric groups as are characteristic of many prosthetic groups and coenzymes. Since the 4-epimerase requires only the addition of metal ions for activity, the putative oxidation-reduction mechanism would have to involve only metal ions and the constituent amino acids.
Cystine and lipoic acid have reduction potentials comparable to that of NAD+ and, thus, could participate in the epimerization reaction.
Presumably, the disulfide form would be required. Since sulfhydryl groups are often readily oxidized by air, the oxidized form could predominate in the active site. Although lipoic acid absorbs at 330 nm, its extinction coefficient is too low to have been readily detected in previous spectral studies (1). However, the evidence discussed below tends to eliminate the SH-disulfide oxidation-reduction mechanism.
First, borohydride should reduce the disulfide bond with loss in activity. Concerning the possibility that the disulfide may have been quickly reoxidized prior to or during the assay, it has been observed that activity was not lost on incubation with 1 M mercaptoethanol for 1 hour followed by assay in the presence of 0.05 M mercaptoethanol; that is, under conditions which are usually sufficient to reduce and maintain the integrity of disulfide groups. Although 50% of the activity was lost on incubation with mercaptoethanol for an additional hour, the activity was not recovered on passage through the Sephadex column, suggesting that the activity loss was due to some phenomenon other than reduction of a disulfide bond at the active site. Second, arsenite should have reacted with the reduced disulfide and caused inactivation, and third, sulfite should have reacted with the disulfide group to form the stable sulfur-sulfonated derivative. No data have been obtained to indicate that an indolenine intermediate derived from tryptophan functions as the electron acceptor in the 4-epimerization as reported by Schellenberg for alcohol dehydrogenase (17).
Consequently, the previous results (1) and those presented herein are not consistent with the electron-acceptor being NSD+, cobamide coenzyme, lipoate, cystine, or an oxidized indolenine derivative of tryptophan.
In the absence of any substantial evidence for participation of an oxidizing group on the enzyme, it is necessary to consider other mechanisms for epimerization such as: A Snz inversion is not considered probable since McDonough and Wood (18) previously reported no isotope incorporation into L-ribulose-5-P and n-xylulose-5-P when the epimerization was conducted in Hzl*O. The mechanism proposed in Fig. 2 depicts a metal ion-assisted aldolytic cleavage in a manner strictly analogous to the Schiff base mechanism (19,20). The metal ion chelates with the carbonyl group (and possibly a hydroxyl group) and serves as an electrophile.
A base on the enzyme surface acting as a nucleophile impinges on the C-4 hydroxyl group. In the ensuing rearrangement of electrons, C-3-C-4 cleavage occurs and C-3 takes on carbanion character.
In completion of the rearrangement, a metal-oxygen bond is formed at C-2 along with a double bond at C-2-C-3.
These intermediates would be analogous to the eneamine and ketamine intermediates in the Schiff base mechanism.
In this scheme, it is not intended to favor a discrete as opposed to a concerted mechanism.
If the characteristics of the epimerase are such that (a) the carbanion of dihydroxyacetone from carbon atoms 1, 2, and 3 cannot be discharged by a proton as in the case of transaldolase (21, 22) and (b) the glycolaldehyde phosphate moiety does not readily dissociate, it would follow that the carbon-carbon bond would immediately re-form.
If there were a high probability that the bond between C-4 and the hydroxyl group would reform cis or trans in this process, epimerization would be observed. In such a mechanism, L-ribulose 5-phosphate 4-epimerase would, in fact, be a special kind of trausaldolase to the extent that the dihydroxyacetone moiety does not dissociate. It would differ in that the other fragment, glycolaldehyde phosphate, is bound and precludes other aldehydes functioning in dihydroxyacetone transfer reactions.
If this hypothesis is correct, the carbanion intermediate would be very short lived because the proximity of glycolaldehyde phosphate would favor condensation.
In this connection, the reaction of tetranitromethane in fructose diphosphate aldolaseand transaldolase-catalyzed reactions may be observable because dissociation of glyceraldehyde a-phosphate allows access to the carbanion intermediates.
An alternative mechanism would be dehydration-redehydration by acid-base catalysis as shown in Fig. 3. The first step would involve a base-catalyzed removal of the proton on C-3 leaving either a carbanion at C-3 or a double bond between C-2 and C-3. The presence of the meOa1 ion in the active site would facilitate removal of the C-4 hydroxyl group in a manner pro-the limitation on this mechanism is that the same proton and hydroxyl groups removed must be involved in the reverse reaction.
In this connection, Rose (25) has produced evidence with phosphoglucoisomerase that the intramolecular transfer of a proton can be faster than equilibration with the surrounding medium.
Thus, it is conceivable that the proton removed from C-3 becomes bound to the enzyme and is not free to diffuse into the medium.
The hydroxyl group would probably be chelated by the metal ion in a position where it would be readily accessible to both sides of C-4 but not to the medium.
Neither dealdolization-aldolization nor dehydration-redehydration can participate in the mechanism of the other carbohydrate 4-epimerases since the substrates of all other 4-epimerases do not possess a free carbonyl group which could participate in the mechanism.