Purification and Properties of Glycerol Dehydrase*

Abstract Glycerol dehydrase, a coenzyme B12-dependent enzyme, from Aerobacter aerogenes, has been purified to a stable, inactive, homogeneous complex with hydroxocobalamin. The inactive complex can be converted to the active holoenzyme by replacement of hydroxocobalamin with coenzyme B12 in the presence of Mg++ and SO3=. The molecular weight of the hydroxocobalamin apoenzyme complex is 188,000 and 1 mole of hydroxocobalamin is bound to 1 mole of the apoenzyme. There is a significant difference between the spectra of the dissociated and the protein-bound hydroxocobalamin. The greater stability of the protein hydroxocobalamin complex as compared to the apoenzyme was shown by various methods. Complete dissociation of the apoenzyme is caused by low pH, Na+, or EDTA, whereas the hydroxocobalamin apoenzyme complex does not dissociate under the same conditions. EDTA markedly inhibits the activity of the enzyme whereas other chelating agents have little or no inhibitory effect. Partial reversal of the inhibitory effect of EDTA is achieved by increased concentrations of glycerol or by adding an excess of either subunit, A or B. The experimental evidence suggests that the effect of EDTA on the apoenzyme does not involve a divalent cation.

absence of substrate.
The inactive enzyme contains hydroxocobalamin firmly bound to the apoenzyme.
The inactive complex containing hydroxocobalamin can also be obtained by addition of hydroxocobalamin to solutions of the apoenzyme AB. Other cobamide derivatives can also form inactive complexes with AB (3,4). The intact apoenzyme is required for the binding of the cobamide derivatives since no binding of cobamide derivatives by the separate subunits, A or B, was observed. Since the cobamide derivatives appear to be firmly bound to the active site, conditions for the replacement of the Blz derivatives by DBCC to restore the active holoenzyme were studied.
Furthermore, these complexes thus provided a means of studying the number of active centers cm the enzyme.
The apoenzyme and the separate subunits, A and B, are very unstable compared to the complex formed from AB and hydroxocobalamin.
Therefore, the enzyme was purified as the complex, AB .BltOH. This paper describes improved methods for the purification of the enzyme as the inactive AB .B120H complex and presents further studies on the properties of the enzyme. fMethods-Hydroxocobalamin was obtained from Merck, dl-tryptophan from Nutritional Biochemicals, hexadecyltrimethylammonium bromide from Eastman Kodak, and p-chloromercuribenzoate from Calbiochem. Sephadex G-25 and G-50 were obtained from Sigma, and hydroxylapatite from Bio-Rad.
The crystalline DBCC was a gift from the Yamanouchi Pharmaceutical Company, Japan. Labeled DBCC and hydroxocobalamin were isolated from Propioni bacterium shermanii cultures grown on a medium containing 60CoC12 (5). One nanomole of radioactive cobalamin had 3100 cpm. Bacteriological reagents were obtained from Difco. Reagents for disc electrophoresis were obtained from Canalco.
All other reagents used were analytical grade. Determination of Enzyme Activity-The method of Smiley and Sobolov (6), which involves conversion of the P-hydroxypropionaldehyde to acrolein, was used to determine the P-hydroxypropionaldehyde produced in the glycerol dehydrase assay. The optical density was proportional to the concentration of acrolein over the range of 0.5 to 3.0 pmoles per sample.
The assay mixture contained 0.06 M glycerol, 0.1 M potassium phosphate (pH 8.6)) 0.05 M potassium sulfite, 1 nmole of DBCC, and apoenzyme (AB) containing 0.03 to 0.10 unit of glycerol dehydrase activity in a total volume of 1 ml. The mixture was Issue of July 10,1970 Schneider, Larsen, Jacobson, Johnson, and Pawellciewicx 3389 incubated for 30 min at 30". The reaction was started by the addition of DBCC and stopped with 1 drop of HCl. One unit of BB activity is defined as the amount of AB which produces 1 pmole of P-hydroxypropionaldehyde per min. The activity of the AB .BrzOH was estimated by adding to the previously described assay mixture 30 pmoles of magnesium acetate and an -4B.BizOH preparation instead of AB, containing 0.004 to 0.012 unit of activity.
This mixture was incubated for 4 hours at 30". One unit of AB .BrtOH activity is defined as the amount of enzyme which produces 1 pmole of /3-hydroxypropionaldehyde per min. The conversion of 1 unit of AB into AB .Br20H yields 0.08 unit of the enzyme activity.
Estimation of Activity of Subunits, A and B-The activity tests on the subunits were performed in the way described for the apoenzyme except that an excess (0.2 to 0.4 unit) of B was added to 0.03 to 0.10 unit of A for the determination of A. An excess of A (0.2 to 0.4 unit) was added to 0.03 to 0.10 unit of B for the determination of B. Determination of Cobalumin-Concentrations of DBCC and hydroxocobalamin were determined from the millimolar extinction coefficient of the dicyanate at 580 rnp produced by treating a sample with potassium cyanide (e = 10,100) (7).
The concentration of cobamide derivative bound to AB was determined enzymatically by the method of Pawelkiewicz and Schneider (4) under the following conditions. AB.Bi20H containing 50 to 400 pmoles of BleOH in 1 ml of 1 M acetic acid was boiled for 4 min and cooled, and 1 ml of 1 M KOH was added. Aliquots of 0.05 to 0.20 ml of this mixture were added to tubes containing 0.1 unit of crude, dialyzed AB and 100 pmoles of potassium phosphate, pH 8.6, in a final volume of 0.80 ml. hfter 80-min incubation at room temperature, 50 pmoles of glycerol, 50 pmoles of potassium sulfite, and 1 nmole of DBCC were added to a final volume of 1 ml.  (10) with a linear 5 to 20% (w/v) sucrose gradient and a total volume of 4.7 ml. The Beckman Spinco model L 2-65B with an SW-50L rotor was used in these studies.
Optical Instruments-A Gilford model 240 and a Beckman DB were used for the spectra and the protein determinations.
A Bausch and Lomb Spectronic 20 was used for the routine enzyme assays. The scanning diagrams of the polyacrylamide gels were done on a Goertz Electra densitometer purchased from Gelman Instrument Company. Organism and iMedia-A. aerogenes (PZH, strain 572) was used to produce the enzyme. The streak plate method was used to prove that only one strain of A. aerogenes was present in the culture. 2 The organism was grown as described previously (l), with the following medium being used in the present work: 10 g of Difco yeast extract, 20 g of casein hydrolysate, 14 g of K2HPO,, 6 g of KH2POI, 5 g of MgSO+ 0.5 g of MnSO+ 10 g of K&04, 200 ml of glycerol, and distilled water to obtain a final volume of 8.5 liters.
The organisms were harvested with a Sharples centrifuge.
The organism harvested at this stage contained a relatively small amount of enzyme activity.

Induction and Preparation of Crude AB
One hundred fifty grams of wet weight of the organism were uniformly suspended in 1 liter of induction media with the following composition: 10 g of K2SO4, 100 ml of glycerol, 2 g of Difco yeast. This suspension was incubated for 1 hour at 25" with constant stirring.
A pH of 8.0 to 9.5 was maintained during the incubation by frequent additions of 2 M KOH. The bacteria were ruptured by sonic oscillation followed by centrifugation.

Preparation and Puri$cation of AB . BlzON
The K2HP04 concentration was adjusted to 0.2 M, pH 8.6, and hydroxocobalamin, at a level of 0.3 nmole per unit of AB, was added to the dialyzed solution.
Complete conversion of AB to AB .B1%OH was accomplished by incubation for 30 min at 30", as evidenced by complete loss of AB activity.
Step I-Hexadecyltrimethylammonium bromide was added to the BB .BlzOH solution dropwise with constant stirring until a concentration of 140 mg of the bromide per 100 ml of protein solution was obtained.
After 30 min the inactive precipitate was removed by centrifugation.
Step II-The supernatant was dialyzed against 0.05 M K2HPOI, pH 8.6, followed by concentration on dry Sephadex G-50 to a protein concentration of 10 mg per ml. The K2HP0, concentration was adjusted to 0.5 M, pH 8.6. The AB.Br20H solution was heated in glass tubes at 72" for 5 min. After centrifugation the inactive precipitate was discarded and the supernatant was dialyzed against 0.05 M K~HPOI, pH 8.6, followed by concentration with dry Sephadex G-50 to a protein concentration of 5 mg per ml.
Step ZZZ-The AB .Br20H solution was dialyzed against 0.01 M potassium acetate, pH 5.3, for 8 hours at 0". The inactive precipitate was removed by centrifugation and the pH of the supernatant was adjusted to 8.6 with 2 M KOH.
Step IV-The AB .Br20H solution was applied to a hydroxylapatite column (3 x 5 cm) which had been equilibrated with a solution of 0.1 nmole of hydroxocobalamin in 0.005 M K2HP04, pH 8.6. Chromatographic separation was achieved by stepwise elution with K2HP04 buffer, pH 8.6. The following concentrations of K2HPOd were used: 0.005, 0.01, 0.02, 0.04, and 0.1 M. 2 We are indebted to Dr. Glen Bulmer of the Department of Microbiology for the strain analysis of the organism by the streak plate method. Step V-Final purification was achieved by acrylamide electrophoresis according to Davis (11). Approximately 2 mg of a concentrated solution of AB .B120H were applied to a column 9 x 100 mm. The electrophoresis was performed in the Canalco analytical apparatus with 3 ma per tube for 3 hours at 4" with an electrode buffer containing 14.4 g of glycine and 3 g of Tris per liter.

Properties of Glycerol Dehydrogenase
The stacking and separating gels contained, respectively, 5% and 10.2% acrylamide and 0.09% and 0.185% bis. Riboflavin was used as the catalyst to avoid the inactivation of the Approximately 100 pg of protein were fractionated with polyacrylamide gel for each step of purification checked. The 10% gel and the electrophoretic procedure were done according to Davis (11).
AB .B120H by ammonium persulfate. The migration of a reddish orange AB .BlzOH band could readily be observed during the electrophoresis.
Two other colored bands were also resolved. A red band of free hydroxocobalamin preceded the AB .&tOH and an unidentified orange band followed the AB .B120H. Very sharp resolution was attained with this electrophoretic technique. The electrophoretic separation was considered to be optimal when the AB .B120H had migrated 15 mm into the separating gel. The AB .B120H band was separated from the gel column as a narrow disc with a surgical scalpel. A number of these discs were homogenized with an equal amount of dry Sephadex G-25 with a mortar and pestle. A few drops of 2 M K~HPOI, pH 8.6, were added to facilitate the homogenization. The AB .B120H was extracted from this homogenized mixture with 0.1 M KzHPO,, pH 8.6, for 10 min, followed by centrifugation and a second extraction.  2. Spectra of the protein-bound and dissociated cobalamin. The optical density of a homogenous preparation of AB. BuOH in 0.1 M KC1 (2.34 mg per ml; specific act.ivity, 3.0 units per mg) was measured against a reference containing only apoenzyme protein prepared from ABvB~~OH as described in the text (-). An identical sample, which had been adjusted to pH 3.0 with HCl for 3 min and neutralized by KOH, was run against the same reference (---). Imet, difference spectrum between the two samples described above.

Influence of pH on Stability of AB . BlzOH
The complex is quite stable in the pH range 5.0 to 9.5. Below pH 5.0, AB .BlzOH dissociates irreversibly, liberating BlzOH which can be readily separated from the protein by filtration through a Sephadex G-100 column.
The dissociation of the BItOH from the protein can be observed visually since the color changes from a reddish orange to a distinct pink which remains after neutralization of the acidified solution. Fig. 2 shows the spectra of AB .B120H and an acid-treated sample of the complex. After dissociation of BizOH from the protein there is a shift of the peak at 361 rnp to 355 rnp and a disappearance of two distinct maxima at 480 and 505 mp. The cobalamin compound was not identified by chromatography, but the spectrum of the dissociated B12 is identical with a spectrum of hydroxocobalamin obtained under the same conditions.

Molecular Weight Determination and Molar Ratio in AB.BlzOH
The molecular weight of the purified AB .Bi20H is 188,000 as determined by the sucrose gradient centrifugation method (10) with alcohol dehydrogenase as a standard.
The partial specific volume of the AB.BrQOH calculated from its amino acid composition is 0.735 cma per g (Table II) (12). The specific data and details are presented in Fig. 3 and "Materials and Methods." As indicated in Table III, the results show that 1 mole of BlzOH is bound to 1 mole of the protein, AB.

Factors Necessary for Reactivation of AB. BlzOH
The replacement of BrzOH by DBCC on AB is essential to reactivate AB+BizOH.
As the data in Table IV illustrate, sulfite, magnesium ion, and DBCC are necessary for the reactivation. Fig. 4 depicts the absolute requirement and concentration effect of K$SOs. The optimal concentration of magnesium ion (Fig. 5) (30 mM) for the reactivation of AB is relatively high compared to most magnesium-requiring systems. However, it should be emphasized that the enzymatic activity of AB does not require added magnesium ion. Since many bacterial species contain an enzyme system capable of converting BlzOH to DBCC, the possibility that BizOH bound to the apoenzyme could be converted directly to DBCC on the apoenzyme and thus produce an active holoenzyme was investigated (14). Crude AB.Bi20H was incubated with ATP, Mg++, FMNH2,    Further proof of an exchange reaction is given in Table V. The  liberation of G°Co-labeled BlzOH from labeled AB.B120H is 4 times greater in the presence of magnesium than the control without magnesium. The quantity of 6OCo in the labeled AB. BlzOH could not be increased if the labeled AB.BltOH was incubated with Y&labeled DBCC. The specific activity of the labeled BlzOH was the same as the labeled DBCC.
Therefore, the results from this experiment also suggest that replacement of BlzOH by DBCC occurs at the same site on the apoenzyme. Based on the slopes of the curves in Fig. 6 Issue of July 10, 1970 Xchneider, Larsen, Jacobson, Johnson, and Pawelkiewicx 3393 ditions (Fig. 7). Previous work (4) showed that the thermal lability of AB is probably due to the properties of subunit A since subunit A has minimal stability at 70" whereas B is relatively stable under the same conditions (see inset in Fig. 7). Since previous work (2) had shown the effects of the sodium and the potassium ions on the dissociation and association of the apoenzyme, the effects of these 2 ions on the stability of AB .Bi20H were investigated.
As shown in Fig. 8  AB.BlzOH, 0.024 unit, in potassium or sodium phosphate, pH 8.6, at indicated concentrations in a volume of 0.1 ml, was heated for 5 min at 78" and cooled, and the activity was determined as usual. The activity was expressed relative to the unheated controls containing the same concentrations of phosphate. Because of the possible involvement of a divalent metal ion in the protein subunit structure or enzyme activity (or both), the influence of the following chelating agents on the activity of the enzyme was investigated: 8-hydroxyquinoline, 1, IO-phenanthroline, a,&-dipyridyl, diethyldithiocarbamate, glycine, histidine, 2,4-pentanedione, 1,4-diaminobutane, inositolhexaphosphoric acid, salicylic acid, and EDTA.
EDTA and salicylic acid were the only chelating agents tested which had a significant inhibitory effect. Fig. 9 illustrates the effect of increasing concentrations of EDTA on enzyme activity. This figure also shows that the inhibitory influence of EDTA could be completely reversed by Mg++ in adequate concentrations.
Other divalent ions such as Ca+f, Mn++, Zn++, and Co++ also reverse the EDTA inhibitory effect. Table VII contains data showing that glycerol reverses the inhibition by EDTA at the specified concentrations. The inhibition by EDTA and salicylic acid can also be overcome by adding an excess of either protein subunit as shown by the data presented in Table VIII. Dissociation of AB by EDTA EDTA at a concentration of 0.01 M and 0.1 M glycerol and KzHPOl cause the complete dissociation of the apoenzyme, AB, into the protein subunits, A and B, on Sephadex G-100, as shown in Fig. 10, whereas a control without EDTA yields 90% of the undissociated form, AB. Removal of EDTA by dialysis prior to gel filtration also yields the undissociated form of AB. In contrast to the effect produced on AB, AB .B,OH is not dissociated on a Sephadex G-100 column equilibrated with EDTA under the same conditions. Magnesium acetate, 30 pmoles, was added as indicated.
The enzyme activity was then determined with the standard assay. The results were expressed as the activity remaining relative to the control which had no EDTA and magnesium. For similar reasons, it is difficult to isolate the holoenzyme (15,16).
The development of a technique which reactivated AB .B120H has made it possible to isolate a form of the enzyme which is relatively stable as compared to the apoenzyme. The differences in stability and other properties of AB .B120H and AB are summarized in Table IX. Although AB . BlzOH is more stable than the apoenzyme or the apoenzymecoenzyme combination, it is still sufficiently sensitive so that a number of the usual enzyme purification methods cannot be used as, for example, prolonged gel filtration on Sephadex G-200 or ion exchange chromatography.
The conversion of AB .B120H to active holoenzyme can be illustrated by the following exchange reaction:  to -SH inhibitors (Table VI) in contrast to AB suggests that a sulfhydryl is involved in the reduction and possible binding of BIZOH. A similar interpretation was given by Lee and Abeles (16) on diol dehydrase.
Thus far, attempts in this laboratory to recover and isolate the pure apoenzyme have not been sufficiently successful to report at this time. Further work will be published on the isolation of A or B from the pure AB.Bl%OH by selective denaturation of either subunit. As previously shown (4), the formation of AB .B120H is a stoichiometric reaction. Inhibition data obtained under standardized conditions repeatedly showed that 1 nmole of BlzOH inactivates 6.3 units of AB and produces 0.5 unit of AB .B120H which is equivalent to 1 mole of hydroxocobalamin per mole of AB. This also agrees with the results obtained by the cobalt activation analysis presented in Table III.
The experimental data suggest that the active holoenzyme has a 1:l molecular ratio with respect to the DBCC and ,4B. However, since the experimental approach has been indirect because of the extreme lability of the holoenzyme, the possibility of multiple active sites on the apoenzyme cannot be eliminated.
The sharply contrasting effects of the Na+ and the K+ ions on the reversible dissociation of AB into subunits, A and B, are an observation that has not been previously published for other enzymes. Admittedly, a number of enzymes are known in which the Na+ and the K+ have an antagonistic effect on activity (19), but these effects have not been related to the dissociation of the enzyme into subunits.
The usual interpretation of the inhibition of enzyme activity by EDTA is that a divalent metal ion is essential for optimal enzyme stability or activity (or both).
However, in this case the use of a number of chelating agents and exhaustive dialysis after EDTA treatment has thus far yielded no evidence indicating that a divalent cation is essential for enzyme activity.
The requirement for Mg++ is limited to the exchange reaction involved in the conversion of the inactive AB .BlzOH to the active holoenzyme.
Further evidence for a nonmetal effect of EDTA is the reversible dissociation of the AB into subunits by EDTA, presented in this paper. The reversal of the EDTA inhibition by the addition of either subunit in excess suggests a reversible association of inhibitor with subunits.
Previous work (2) showed that glycerol increases the affinity of the subunits for each other, thus shifting the equilibrium toward the associated form, which may explain the reversal of the EDTA effect by glycerol.
Several observations have been published on the prevention of association of proteins by EDTA which were not related to its chelating properties.
Sh 1 b a a y and Lauffer (20) found that EDTA retarded polymerization of tobacco mosaic virus protein. Yue, Noltmann, and Kuby (21) found that EDTA completely inhibited the TPN-induced association of the apoprotein of glucose B-phosphate dehydrogenase, but had no inhibitory effect on enzymatic activity.