THE J O U R N A L OF BIOLOGICAL CHEMISTRY

Printed m U. S. A. Vol 256, No. 18, Issue of September 25, pp. 9755-9761, 1981

Purification of the putA Gene Product A BIFUNCTIONAL MEMBRANE-BOUND PROTEIN FROM SALMONELLA TYPHIMURIUM RESPONSIBLE FOR THE TWO-STEP OXIDATION OF PROLINE TO GLUTAMATE*

(Received for publication, December 22, 1980)

Rolf Menzel and John Roth From the Department of Biology, University of Utah, Sal t Lake City, Utah 84112

In this paper we report the purification of a protein which is able to catalyze both the proline oxidase and the pyrroline-5-carboxylic acid dehydrogenase activi- ties necessary for the oxidation of proline to glutamic acid. The purification involves the preparation of a crude membrane pellet, detergent solubilization, am- monium sulfate fractionation, and DEAE-chromatog- raphy. We are able to obtain an essentially pure prep- aration (>95% pure) after only a 52-fold purification, demonstrating that the protein is a major protein in cells fully induced for proline utilization. Both proline oxidase and pyrroline-5-carboxylic acid dehydrogenase activities co-purity throughout our purification. Veloc- ity sedimentation of the purified protein demonstrates that both proline oxidase and pyrroline-5-carboxylic acid dehydrogenase activities co-sediment. Early in the purification procedure we are able to detect two species of protein which have both proline oxidase and pyrro- line-5-carboxylic acid dehydrogenase activities. Our procedure purifies only the larger molecular weight species. The purified protein is a dimer composed of identical 132,000-dalton subunits. Analysis of mutants defective for proline utilization demonstrate that the bifunctional enzyme is the putA gene product.

The oxidation of proline to glutamic acid proceeds by the sequential action of an oxidase and a dehydrogenase in animal tissue (1,2) and bacteria (3,4). These two enzymatic steps, as well as a nonenzymatic Schiffs base formation, are shown in Fig. 1. The reaction catalyzed by proline oxidase is dependent on molecular oxygen and it has been shown that inhibitors of oxidative electron transport inhibit this reaction ‘(2). The reaction product, pyrroline-5-carboxylic acid, is in sponta- neous equilibrium with y-glutamic acid semialdehyde by an intramolecular Schiff s base reaction (5). y-Glutamic acid sem- ialdehyde is then oxidized by pyrroline-5-carboxylic acid de- hydrogenase to glutamic acid in an NAD’-dependent reaction.

Proline oxidase activity has been demonstrated to be a mitochondrial enzyme in the liver tissue of several mammals (6-8). Such a subcellular location of the oxidase activity is consistent with the functional requirement to interact with an electron transport chain. Pyrroline-5-carboxylic acid dehy- drogenase has also been demonstrated to be associated with mitochondria in rat liver (9), ox liver (lo), and plant seedlings (11). In Escherichia coli both proline oxidase and pyrroline-

* A preliminary report of these results has appeared as abstract K112 in the abstracts of the 1977 Annual Meeting of the American Society for Microbiology. 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.

5-carboxylic acid dehydrogenase have been shown to be as- sociated with a particulate fraction following breakage of the cells ( 12).

Recently Ratzkin and Roth (13) have analyzed a large number of mutants defective for the oxidation of proline to glutamic acid in Salmonella typhimurium. These mutants define the putA gene. Most mutants in the putA gene are defective for both proline oxidase and pyrroline-5-carboxylic acid dehydrogenase activities. This suggests that the putA gene encodes both the proline oxidase and pyrroline-5-carbox- ylic acid dehydrogenase enzyme activities.

In this paper we report the purification of the putA gene product of S. typhimurium. The putA gene product is a bifunctional 132,000-dalton peptide which possesses both pro- line oxidase and PCA‘ dehydrogenase enzyme activities.

EXPERIMENTAL PROCEDURES

Materials

Nucleotides-NAD’ and NADP’ (recrystallized from acetone) were purchased from Sigma. FMN’ and FAD’ were of the highest grade available from Sigma.

Chromatography-DEAE-Sephadex A-25 and coarse Sephadex G- 50 were purchased from Pharmacia.

Chemicals-Tween 20, Tween 80, and Brij 58 were products of the Atlas Chemical Industries, Inc. Cacodylic acid, iodonitrophenyl tet- razolium, phenazine methosulfate, ethylene glycol, glycerol, and Trizma base were purchased from Sigma. Sodium dodecyl sulfate and dimethyl suberimidate were purchased from Pierce. Coomassie bril- liant blue R-250, acrylamide, bisacrylamide, N,N,N’,N’-tetramethyl- ethylenediamine, and ammonium sulfate were purchased from Bio- Rad.

Pressure Dialysis-The 50-ml Amicon pressure cell and a PM-30 dialysis filter were purchased from Amicon Corp.

Protein Standards-Bovine serum albumin was purchased from Worthington. Purified E. coli RNA polymerase was a gift from M. Chamberlin.

Methods

Bacteria-The following strains of S. typhimurium LT2 were used: the wild type LT2, a diploid strain TT1868 (this strain contains a Salmonella episome whose construction is described elsewhere (14)) and mutants putA736 and putPA523 (13). For enzyme preparation the strain TT1868 was grown in the PSN media of Ratzkin and Roth (13) in a 100-liter fermentor. Cells were grown overnight from a 2- liter innoculum to a density of Ah:<, = 1.2 at 30 “C with vigorous agitation and aeration (IO liters of air/min). For mutant analysis and a demonstration of putA induction, the NCE media of Berkowitz et al. (15) was supplemented as described in the text and cells were grown a t 37 “C with gyratory shaking.

Stock Solutions for Enzyme Assays-The reaction mix for the enzyme assays described in the text were made from the following stock solutions. Trizma base was dissolved in 40% ethylene glycol and 1% Tween 20 to a concentration of 0.4 M. The pH of this solution was

I The abbreviations used are: PCA, pyrroline-5-carboxylic acid; INT, iodonitrophenyl tetrazolium; SDS, sodium dodecyl sulfate.

9755

9756 Purification of the putA Gene Product

PART A PROLINE OXIDASE

PART 8 NONENZYMATIC w.cp

PART C PCA DEHYDROGENASE

FIG. 1. Pathway for the degradation of proline to glutamic acid.

adjusted to either 8.9 or 7.8. INT was dissolved in boiling water at a concentration of 3.2 mg/ml with stirring. Following 5 min of stirring on a hot plate the solution was filtered on a Whatman No. 1 fdter with a vacuum. This solution is stored in a dark bottle. Phenazine methosulfate was dissolved in cold water to a concentration of 0.4 mg/ml and stored in a dark bottle. Gelatin was dissolved in hot water to a concentration of 2 mg/ml. Proline is diluted from a stock 1 M solution. L-PCA is maintained as a stock solution of 80 mM in 6 M HC1; both HCI and H20 are removed to dryness by vacuum evapo- ration. PCA is then dissolved in buffer immediately prior to use. Stock solutions of NAD' at 20 mg/ml are adjusted to pH 7.0 and stored frozen in small aliquots. All other solutions are stored at 4 "C for periods up to 1 month prior to use in the assays.

G Buffer-G buffer is 70 mM Tris-HC1 with 30% (v/v) glycerol. This buffer is made immediately prior to use from a stock 1 M Tris- HCl buffer (pH 8.2), glycerol, and deionized H20. The G buffer is then scanned for absorption in the 320-240 nM wavelength region. Certain batches of glycerol contain unknown material which absorbs light in this region. G buffer with an A ~ M L > 0.02 is discarded as such batches prove to give anomalous results in the DEAE-chromatography.

Column Chromatography-The G-50 desalting column used in this work was prepared from coarse Sephadex G-50. The Sephadex was prepared by swelling in 4 liters of 70 m~ Tris-HCl, pH 8.2, for 3 days in the cold. The Sephadex was then poured into a 500-ml column (3.5 X 54 cm) and packed with 2 liters of the same buffer with 0.02% sodium azide. Prior to use in enzyme purification the column is rinsed with 2 liters of G buffer in which the enzyme is then desalted as described below. After use the column is washed with 3 liters of 70 mM Tris-HCI with 0.02% sodium azide in which it is stored.

The DEAE-column used in this enzyme purification is prepared by boiling DEAE-Sephadex A-25 in 500 ml of G buffer for 2 h. The column material is then poured into a 100-ml column (2.32 cm) and packed at 4 "C with 500 ml of cold G buffer. The column is prepared immediately prior to use and chromatography is performed as de- scribed below. After use the column material is discarded and a new column is used for each purification.

Synthesis of L-PCA-L-PCA is synthesized by the periodate oxi- date of allo-1-hydroxylysine and purified by chromatography on Dowex AG 8 as described by Wu and Seifter (16). Following elution from the Dowex column in 0.5 N HCI, PCA is concentrated by vacuum evaporation at room temperature until the solution is 6 N HCI. At this point the solution is stored at 4 "C until required for enzyme assay. Occasionally the solution takes on a yellowish color which is removed by activated charcoal treatment. The appearance of the yellowish color and its subsequent removal by charcoal results in a less than 5% decrease in the PCA concentration as determined by reaction with 0-aminobenzaldehyde (16). The PCA solution is treated as described above prior to use in enzyme assays. Analysis of synthe- sized PCA on a Beckman amino acid analyzer demonstrated that all the ninhydrin-reactive material is present in a single peak.

SDS Gels-Electrophoresis in polyacrylamide slab gels (12 cm X 13 cm x 1.5 mm) and staining with Coomassie brilliant blue was performed as described by Ames (17). Following destaining, gels are dried under a vacuum between two sheets of dialysis membrane. When required, the dried gels were scanned with a densitometer.

Natiue Gels-Electrophoresis in polyacrylamide tube gels (10 X 0.8 cm) was performed using the same buffers and acrylamide solutions

as were used in the SDS gels except for the omission of the 0.1% SDS. Acrylamide concentrations were varied as required. Gels were run at 2 mA/tube. When the marker dye, bromphenol blue, had migrated within 1 cm of the bottom of the gel, the electrophoresis was termi- nated and gels were stained for either protein or enzyme activity. Gels were stained at 37 "C for oxidase activity in a solution containing 0.2 M proline, 1 mM INT, 16% ethylene glycol, and 0.6 M Tris-HC1 at pH 8.4. The gels were stained for 15 min to 2 h as required. Staining was terminated by washing the gels with deionized H20 and then trans- ferring them to a 7% acetic acid solution. A control gel for substrate independent dye reduction was stained with the omission of the 0.2 M proline. Gels were stained at 37 "C for dehydrogenase activity in a solution containing 4 mM L-PCA, 1 mM INT, 0.015 mg/ml of phenazine methosulfate, 2 m~ NAD', 16% ethylene glycol, and 0.6 M Tris-HC1 at pH 7.6. The gels were stained for 15 min to 2 h as required. Staining was terminated by washing the gels with deionized H20 and then transferring them to a 7% acetic acid solution. Control gels for substrate independent dye reduction were stained with the omission of the 4 mM L-PCA for one gel and omission of the 2 mM NAD' for the second control gel.

Protein Determination-Protein concentrations were determined by a Coomassie dye binding assay (18) using commercially available reagents (Bio-Rad). Bovine serum albumin was used as a standard. Different protein standard solutions were made in each buffer used since background values varied from buffer to buffer.

Velocity Sedimentation-The sedimentation behavior of the putA protein was analyzed by sedimentation on a 20 to 40% glycerol gradient containing 70 mM Tris-HC1 at pH 8.2. A 0.2-ml sample containing 0.02 mg of the purifiedputA protein and 0.5 mg of catalase was layered on a 4.8-ml linear gradient. Sedimentation was performed at 45,000 rpm for 18 h in a Beckman SW 50.1 rotor at 4 "C. Fractions containing 0.2 ml were collected from the bottom of the gradient. The putA protein was located by both oxidase and dehydrogenase assay. Catalase was located by protein assay.

Oxidase Assay-To measure proline oxidase activity we follow the reduction of the electron accepting dye iodonitrophenyl tetrazolium by monitoring the absorbance at 520 nM. In the assay 0.005 to 0.050

0.2 M proline, 16% ethylene glycol (v/v), 0.4% Tween 20 (v/v), 0.16 ml of enzyme is added to a final reaction volume of 1.0 ml containing

M Tris-HC1 (pH 8.9), 0.04 mg of gelatin, and 0.5 mM INT. The reaction is initiated by the addition of enzyme and incubated at 23 "C. The reactions are terminated by the addition of 0.1 ml of 2 N HCI. All reactions are corrected for a proline-independent background of INT reduction. For a final A R ~ ~ between 0.05 and 0.70 the reaction is linear with protein added and with time to 20 min. We use EM = 11.5 X 10' liters M" cm"' (19) as the extinction coefficient at 520 nm for reduced INT. We define 1 unit of oxidase activity as 1 pmol of INT reduced/ h. The details of the reaction characteristics are described in the accompanying paper (23).

Dehydrogenase Assay-To measure pyrroline-5-carboxylic acid dehydrogenase activity we follow the reduction of the electron ac- cepting dye INT as in the oxidase reaction. In this assay, 0.005 to 0.50 ml of enzyme is added to a final reaction volume of 0.7 ml with 4 mM ~-pyrroline-5-carboxylic acid, 16% ethylene glycol (v/v), 0.4% Tween 20 (v/v), 0.16 M Tris-HC1 (pH 7.8), 0.01 mg of phenazine methosulfate, 1 mM INT, and 0.04 mg of gelatin. The reaction is initiated by the addition of enzyme and incubated at 23 "C. The reaction is terminated by the addition of 0.1 ml of 2 N HCI. All reactions are corrected for both enzyme and substrate independent dye reductions. For a final Asao between 0.05 and 0.70 the reaction is linear with protein and time to 30 min. We define 1 unit of dehydrogenase activity as 1 pmol of INT reduced/h. The details of the reaction characteristics are de- scribed in the accompanying paper (23).

Enzyme Purification

Preparation of Cells-Bacteria of S. typhimurium LT2 strain TT1868 were used as a source of enzyme in this study. Strain TT1868 is diploid for the put genes. Cells were grown as described above under ''Bacteria'' and harvested by centrifugation at 4 "C. The cells were washed twice with 0.85% (w/v) NaCl and suspended to a density of 28 g wet weight/40 ml of 0.5 M cacodylic acid, pH 6.8. These cells can be used immediately for purification or may be frozen and stored at -70 "C indefinitely. All purification steps are carried out at 4 "c.

Fraction I: Crude Extract-The cell suspension above (40 ml) is passed through an Amicon French pressure cell twice at 10,OOO to 12,000 p s i . These ruptured cells are then diluted to a volume of 200 ml with cold (4 "C) 0.5 M cacodylic acid, pH 6.8, mixed thoroughly,

Purification of theputA Gene Product 9757

and centrifuged for 10 min at 5000 rpm in a Sorvall SS-34 rotor. The supernatant is carefully decanted (Fraction I).

Fraction ZZ: Washed Membranes-Fraction I is diluted to a vol- ume of 200 ml with 0.5 M cacodylic acid pH 6.8 and centrifuged for 8 min at 110,OOO X g. The pellet at this point is very loose and it is important to decant the supernatant carefully, allowing some of the supernatant to remain behind to avoid losing the pellet. The super- natant is discarded and the pellet is resuspended to a volume of 200 ml in cold 0.5 M cacodylic acid, pH 6.8. The resuspended pellet is then centrifuged for 8 min at 110,OOO X g. After this second centrifugation the pellet is very tight, the supernatant may be poured off quickly, and any excess supernatant is removed from the centrifuge tube with a lint-free tissue.

Fraction IIZ: Detergent Extract-Fraction I1 is resuspended to a volume of 100 ml in 0.1 M Tris-HCI, pH 8.2, with 0.1% Tween 20. The resuspended pellet is spun at 110,OOO X g for 10 min. The supernatant containing the solubilized enzyme is decanted (Fraction 111). For the determination of oxidase and dehydrogenase activities a small aliquot of the supernatant is immediately brought to a concentration of 3 0 8 (v/v) glycerol. The remainder of the supernatant is immediately precipitated with ammonium sulfate as described in step IV. (For the preparation of washed membranes having the membrane-bound oxi- dase coupled to electron transport the pellet may be resuspended in 0.5 M cacodylic acid at pH 6.8.)

Fraction ZV: Ammonium Sulfate Precipitation and Sephadex G- 50 Desalting-To precipitate the enzyme, Fraction I11 is brought to 50% saturation by dropwise addition of an equal volume of a saturated ammonium sulfate solution over a 5-min period. The precipitation is carried out at 4 “C with stirring. Stirring is continued for 5 min after the addition of ammonium sulfate. The supernatant is then centri- fuged for 5 min at 10,000 rpm in a Sorvall SS-34 rotor to collect the precipitate. The supernatant is discarded and the pellet is redissolved in 10 ml of G buffer (30% (v/v) glycerol, 70 mM Tris-HC1, pH 8.2)

E w 2

50 I O0 Column

Fraction

0.2

0.1

FIG. 2. DEAE-chromatography of the putA protein. For the conditions of elution, see the text. Oxidase (A) and dehydrogenase (0) were determined for 2 0 4 aliquots as described under “Methods.” Ala,, (0) was used as a method of monitoring column protein. The A ~ w ) values are determined with an accuracy of kO.01 A units.

with 0.5% Tween 20. This material is immediately applied to a 500 ml of Sephadex G-50 column (3.5 X 53 cm) from which 5-ml fractions were collected at a flow rate of -300 ml/h. Details of the column’s preparation are given above. The peak fractions of enzyme activity elute with the column void volume and are easily identified by their yellowish appearance. Those fractions accounting for approximately 80% of the activity recovered from the column are pooled and spun at 110,OOO X g for 15 min. The insoluble material is discarded and the supernatant is saved (Fraction IV).

Fraction V: DEAE-Column Chromatography-Fraction IV is ap- plied to a 100-ml (2 X 32 cm) DEAE-Sephadex A-25 column equili- brated with G buffer. The enzyme application, all subsequent column rinses, and the salt gradient are run at a flow rate of 35 ml/h. Following the application of the enzyme, the column is rinsed first with 150 ml of G buffer containing 0.1% Tween 20 and then with 200 ml of G buffer with 60 m~ KC1. Chromatography is performed with a 500-ml linear salt gradient from 60 to 160 mM KC1 in G buffer; 3-ml fractions are collected (Fig. 2). All fractions with >12 units oxidase/ ml are pooled (Fraction V).

The peak fractions from the DEAE column (-40 ml) are concen- trated in an Amicon pressure dialysis cell with a PM-30 filter. Follow- ing a volume reduction to 3 to 4 ml, the enzyme is dialyzed for 24 h against four changes of 1 liter of G buffer. After concentration and dialysis of the enzyme to remove KCI, the activity is stable for a period of weeks when stored at 0 “C. For good enzyme yields, it is necessary to carry out all the purification steps as rapidly as practical.

RESULTS

Purification-The purification of the putA gene product is summarized in Table I. A 52-fold purification of both oxidase and dehydrogenase activities is achieved; an observation con- sistent with the bifunctional nature of the putA gene product. The data given in Table I is that of a typical purification. Yields of up to 7.5% have been obtained. The specific activity of the purified oxidase varies from 140 to 180 units/mg of protein from batch to batch. The ratio of oxidase to dehy- drogenase activities vary from 1.4 to 1.7. Such variation we feel results from the differential stability of enzyme activities during the purification.

In Table I we note that during purification the ratio of oxidase to dehydrogenase activity varies from 1.3 to 1.8. Such variation, we believe, is due to the limited reproducibility of our assay method (+lo%; see accompanying paper (28)) and the fact that the protein has more than one form (see Fig. 6 and “Discussion”). In spite of the minor discrepancies, we feel the data in Table I shows the co-purification of two activities. In Fig. 2 it is readily apparent that both enzymatic activities co-chromatograph with the only protein peak to elute from the column during the course of the 60 to 160 mM KC1 gradient.

In Fig. 3 we show an SDS-polyacrylamide gel of purified enzyme run at several protein concentrations. The gel in Fig. 3 and several duplicate gels have been scanned with a densi- tometer. The results indicate that essentially all the protein (-97%) is in a single molecular weight band. Of the minor bands present none accounts for >I% of the total protein. We have verified these conclusions by running both denaturing and non-denaturing gels at 4,8, and 12% polyacrylamide (data not shown).

TABLE I Purification of the putA gene product

Oxidase actlvitv Dehvdroeenase activitv Ratio ox- Step

I ”

Total protein ~~~~l Units/mg idase/de- Purifica-

Total Units/mg hydro- units protein units protein genase

tion

mg q -fold

Fraction I: crude extract 2800 8200 2.9 4500 1.6 1.8 Fraction 11: washed membranes 344 3900 11.3 2800 8.1 1.4 47 3.9 Fraction 111: detergent extract 84 3000 35.8 1800 21.4 1.7 36 12 Fraction IV: (NHs)SOd/G-50 45 2000 44.4 1500 33.3 1.3 24 15 Fraction V: DEAE 2.7 412 152 281 104 1.5 5 53

9758 Purification of the putA Gene Product

FIG. 3. SDS-polyacrylamide gel electrophoresis ofputA pro- tein purified as described in this paper. 3.6, 0.6, and 0.05 pg of protein were run. Details of the electrophoresis are given under “Methods.”

We have run SDS-polyacrylamide gels of purified enzyme with molecular weight standards. Fig. 4 shows that the puri- fied putA protein has a peptide molecular weight of 132,000 when run in a 6% polyacrylamide gel. Estimates have also

been made at 4 and 8% polyacrylamide. These estimates agree within +5% of the 132,000 estimate (data not shown).

Velocity Sedimentation of Purified Enzyme-Enzyme pu- rified by the procedure described above was sedimented through a glycerol gradient. The data in Fig. 5 demonstrate that both oxidase and dehydrogenase activities co-sediment. This result supports our conclusion that the oxidase and dehydrogenase reside in a single molecular species. We have also attempted to determine a sedimentation value for our purified protein. Unfortunately the sedimentation behavior is not completely reproducible with respect to standard proteins. Although the two activities always co-sediment, their posi- tions vary slightly with respect to standard proteins. Our purified protein has run slightly ahead, slightly behind, and precisely with catalase in different glycerol gradients. In the experiment presented in Fig. 5 our purified protein ran slightly ahead of catalase.

Two Active Forms of the Enzyme-Fig. 6 shows that two forms of theputA gene product are present after solubilization. The gels shows that both form I and form I1 are able to catalyze the proline-dependent reduction of iodonitrophenyl tetrazolium. Further both form I and form I1 are able to carry out an oxidation of PCA which is NAD+-dependent. In aputA deletion mutant (putPA523) grown under identical conditions neither form I or form I1 are present (data not shown). The

’2 601 X

I 3I 40 02 0.4 0.6 0.0 Rf

FIG. 4. Calibration curve for the estimation of molecular weights from SDS-polyacrylamide gels. Procedural details are described under “Methods.” Mobilities were determined relative to bromophenol blue by densitometer tracings of dried gels. BSA, bovine serum albumin; RNAP, RNA polymerase.

m

Fraction Number FIG. 5. Velocity sedimentation of the purified putA protein.

Procedural details are given under “Methods.” The purified putA protein was located by both oxidase (A) and dehydrogenase (0) assays.

Purification of the putA Gene Product 9759

:JX:DASE DEt1YDHOSEYA’;E

PCA -Pia + .’:A i YAD -NAD ~ UA;

U 1

F

FIG. 6. Native gel electrophoresis of Tween 20 extracts of LT2 stained for both proline oxidase and pyrroline dehydro- genase activities. Material run on these gels is identical with Frac- tion 111 material (Table I). Details of the electrophoresis and the activity stains are given under “Methods.” Omissions of substrates are as indicated in the figure. Cells were grown in NCE media of Berkowitz et al. supplemented with 0.4% succinate and 0.2% proline (15).

mutant does, however, show the same substrate-independent iodonitrophenyl tetrazolium reduction bands present in the control gels of the wild type.

In order to obtain molecular weight estimates of form I and form 11, polyacrylamide gels were run having different acryl- amide concentrations. The migration of oxidase and dehy- drogenase activities was compared to standard proteins of known molecular weight. Slopes of the log of the relative mobility uersus per cent acrylamide were calculated for forms I and I1 and for the standard proteins by the least squares method of linear regression. This data has been plotted ac- cording to the method of Hedrick and Smyth (20). The data shown in Fig. 7 assign to form I1 a molecular weight in the range of 135,000 to 165,000 and indicate that form I has a molecular weight in the range of 210,000 to 270,000. Enzyme purified by the procedure given above has been subject to the same analysis and was shown to be identical with form I (data not shown).

Proline Oxidase is Membrane- bound-The catalytic prop- erties of proline oxidase require an interaction with an electron transport chain which uses oxygen as a terminal electron acceptor. It has been demonstrated that a variety of inhibitors of electron transport inhibit proline oxidase (2). The stoichi- ometric consumption of oxygen concomitant with proline oxidation has also been reported (12). Consistent with such a functional requirement, we find that proline oxidase is asso- ciated with a particulate fraction which is easily pelleted. In our purification procedure we pellet approximately 50% of the enzyme activity (longer periods of centrifugation will pellet all the activity). Others have also reported that proline oxidase is a particulate activity in E. coli (12).

Although the association of proline oxidase with a particu- late electron transport chain is obligate to its function, this association is quite weak following the rupture of cells. Caco- dylic acid is unique among the buffers we have used in its ability to keep all the enzyme membrane associated. The use of Tris-HC1, Tris-malate, Tris-borate, Triethanolamine, or phosphate buffers all result in solubilizing approximately 50% of the enzyme. The addition of small amounts (0.5%) of any of a number of nonionic detergents results in solubilization of 100% of the enzyme in the cacodylic acid buffer as well as the other buffer systems.

The behavior of the oxidase solubilized by a variety of detergents in a number of buffers was examined in a native

MW X 10-3Doltom FIG. 7. Molecular weight estimates by native gel electropho-

resis. Fraction 111 material (Table I) was run at 5, 6, 7, and 8% polyacrylamide and stained for oxidase and dehydrogenase activity as described in Fig. 6. Ovalbumin, catalase, and apoferritin were run on parallel gels and stained with Coomassie as described under “Methods.” Mobilities were determined by densitometer tracings. Slopes of RF versus per cent acrylamide were determined by the least squares method of linear regression. The data is plotted according to the method of Hedrick and Smyth (20). Error bars indicate 1 S.D. in the determination of the slopes. 245 f 30K. 245,000 & 30,000, 150 & 15K, 150,000 f 15,000,450K, 450,000,240K, 240,000; 47K, 47,000.

gel system identical to that used in Fig. 6. The two forms of the enzyme seen in Fig. 6 are intrinsic properties of the enzyme. The same two enzymatically active forms were seen regardless of the buffers or detergents used. Further, when all the proline oxidase is coupled to electron transport (cells ruptured in the cacodylic acid buffer without detergent), it is bound to membrane particles too large to enter a polyacryl- amide gel. (No enzyme is seen in the gel, but it is readily apparent at the top of the stacking gel.)

The Purified Protein Is the putA Gene Product-Fig. 8A shows an SDS-polyacrylamide gel of a detergent extract (step 3 material, Table I) from S. typhimurium grown under a variety of conditions. Material in the left lane (Fig. 8A) is derived from wild type cells grown on a noncatabolite repress- ing media without proline. For the center lane (Fig. 8A), LT2 was grown on a noncatabolite repressing media with proline. For the right lane (Fig. 8A) LT2 was grown on a catabolite repressing medium containing proline. The arrow indicates the position at which the purified enzyme runs. Fig. 8A shows that a band corresponding to the mobility of the purified enzyme is subject to induction by proline and catabolite repression by glucose. TheputA gene product’s enzyme activ- ities, proline oxidase and pyrroline-5-carboxylic acid dehy- drogenase, have been shown to be subject to proline induction and catabolite repression (13). SDS-polyacrylamide gels of detergent extracts from wild type cells (LT2) and an amber mutant (putA736) grown under noncatabolite repressing con- ditions with proline present are shown in Fig. 8B: the left lane shows that the M, = 132,000 band is present in the wild type, the center lane shows that the amber mutant lacks this peptide, and the right lane shows purified enzyme as a refer- ence. Taken together, all the gels in Fig. 8 suggest that the purified protein is theputA gene product. Although we cannot rule out that the putA gene is a regulatory gene required for the production of the 132,000-dalton peptide, we choose to believe that the putA gene codes for the 132,000-dalton pep- tide. Over 50 mutants with defects in proline oxidase and pyrroline-5-carboxylic acid dehydrogenase all map at theputA locus which defines a single complementation group (13, 21). If the structural gene for the 132,000-dalton peptide mapped elsewhere, we should have recovered mutants of this gene. Failurq to find a second class of mutants implies that we failed

9760 Purification of the putA Gene Product

,

PART A PART B FIG. 8. SDS-polyacrylamide gel electrophoresis of detergent

extracts. Fraction 111 material (Table I) from the sources indicated was electrophoresed as described under “Methods.” Part A: 4 pg of Fraction I11 material from LT2 (wild type cells) was applied to the gels. Left lane, cells grown in NCE media supplemented with 0.4% sodium succinate; center lane, cells grown in NCE media supple- mented with 0.48 (w/v) succinate and 0.28 (w/v) L-proline; right lune, cells grown on NCE media supplemented with 0.4% (w/v) succinate, 0.2% (w/v) L-proline, and 0.5% (w/v) D-glucose. The arrow indicates the mobility of the purified putA protein. Part B, strains LT2 andputA 736 were grown on NCE media supplemented with 0.4% (w/v) succinate and 0.2% (w/v) L-proline. 4 pg of Fraction I11 material (Table I ) was applied. Left lane, material from the wild type cells (LT2); center lane, material from an amber mutant (putA736); right lane, 2 pg of purified putA protein.

to identify this hypothetical locus‘ because mutations there are unconditionally lethal: a highly unlikely proposition. The putA gene almost certainly encodes the bifunctional degra- dative enzyme.

DISCUSSION

We have shown that the product of the putA gene is a 132,000-dalton polypeptide. This 132,000-dalton peptide is weakly bound to the membrane. This binding, however, is functional since it couples proline oxidation to oxygen reduc- tion, implying some interaction with a membrane-bound elec- tron transport chain. The enzyme may be solubilized with the aid of nonionic detergents.

Following solubilization, the oxidase appears in two forms; form 11, 135,000 to 160,000 daltons, and form I, 210,000 to 270,000 daltons. Both forms catalyze both the oxidation of proline and the NAD’-dependent oxidation of pyrroline-5-

carboxylic acid. The purification described in this text purifies form I. We have shown that the purified putA gene product is 97% pure as estimated by both denaturing and nondenatur- ing polyacrylamide gels. No single contaminating species ac- counts for >1% of the total protein. During the purification, both proline oxidase and pyrroline-5-carboxylic acid dehy- drogenase activities are co-purified.

The purification described in this text results in the prepa- ration of a protein which is essentially homogenous following a 52-fold purification. This indicates that the putA gene prod- uct accounts for approximately 2% of the cell’s protein in diploid cells which are fully induced. SDS-polyacrylamide gels of crude extracts and membranes are consistent with this conclusion (data not given). TheputA gene product is a major protein in cells which are utilizing proline.

Velocity sedimentation data of the purified protein (form I) suggests a molecular weight near the size of catalase. Native gel analysis by the method of Hedrick and Smyth (20) give our purified protein a molecular weight of 240,000 f 30,000. We have also attempted to obtain data about the subunit structure of the purified protein using dimethyl suberimidate cross-linking. At high concentrations of protein and high concentrations of cross-linker a new high molecular weight species of 260,000 was observed.2 Repeated efforts failed to detect >3% of the protein in the new high molecular species and we could not therefore, infer conversion of the 132,000- dalton monomer to a higher molecular weight species. We interpret our data as indicating that the purified form I is a dimer of identical 132,000-dalton subunits. The form I1 protein encountered after solubilization, we believe, represents an active monomer. We speculate that the enzyme exists as a monomer in the membrane and that the dimeric association is a result of hydrophobic surfaces “sticking” together follow- ing solubilization. We find a peptide molecular weight of 132,000 and a dimer molecular weight of 264,000 consistent with the resolution of our data from SDS gels, native gels, and velocity sedimentation.

We are at this time unable to eliminate the possibility that two different peptides of identical molecular weight are pres- ent in the purified protein. The existence of form I1 (mono- meric weight with both activities) implies that, if peptides of identical molecular weight were present, then they would also have identical net charges at pH 8.8. This seems unlikely. We have examined the genetic complementation ofputA mutants and were able to demonstrate that the putA mutants define a single complementation group (21). We would have expected two complementation groups had two different peptides been encoded by the putA gene. Recently the putA gene has been cloned on the plasmid pBR322 using recombinant DNA tech- nology.” The putA functions (proline oxidase activity and pyrroline-5-carboxylic dehydrogenase activity) were localized to a 4.1-kilobase fragment of DNA. Such a fragment has a maximum coding capacity of an approximately 150,000-dalton peptide. We conclude that the putA gene codes for a bifunc- tional 132,000-dalton polypeptide.

Scarpulla and Soffer have reported the purification of an E. coli membrane-bound flavoprotein which catalyzes the oxi- dation of proline to pyrroline-5-carboxylic acid (22). They report that their purified preparation consists of a 124,000- dalton SDS polypeptide which is present as a dimer in its native state, a result which agrees with our observation for the S. typhirnuriurn enzyme. Scarpulla and Soffer (22) report that their preparation does not, however, catalyze the pyrro- line-5-carboxylic dehydrogenase reaction. We feel that this is

R. Menzel, unpublished results. :’ R. Menzel, manuscript in preparation.

Purification of the putA Gene Product 9761

due to the harsher procedures they have used to purify their enzyme. We have found that the inclusion of glycerol in our purification buffer is essential for maintenance of the full activity of both proline oxidase and pyrroline-5-carboxylic acid dehydrogenase. The fact that these workers report that the addition of exogenous FAD is required for maximum oxidase activity supports our contention that their enzyme preparation has undergone some inactivation. (FAD fails to stimulate further our preparation.) Although organism differ- ences might explain the different results, we feel this is not the case. The limited work we have done with E. coli prepa- rations demonstrate the E. coli and S. typhimurium enzymes are identical on native gels.*

In an accompanying paper (23), we examine the kinetic parameters of the purified putA protein’s proline oxidase and pyrroline-5-carboxylic acid dehydrogenase activities. We also demonstrate that the putA protein is a flavoprotein with the flavin group involved in the oxidase but not the dehydrogenase reaction.

REFERENCES 1. Strecker, H. J., and Mela, P. (1955) Biochem. Biophys. Acta 17,

2. Strecker, H. J . (1971a) Methods Enzymol. 7, 251-263 580-586

4. Dendinger, S., and Brill, W. J . (1970) J. Bacteriol. 103, 144-152 5. Vogel, H. J., and Davis, B. (1952) J. Am. Chem. SOC. 74, 109-112 6. Taggart, J. V., and Krakaur, R. R. (1949) J. Biol. Chem. 177,641-

653 7. Kramar, R., and Fitscha, S. (1970a) Enzymologia 39, 101-106

Kramar, R. (196713) Enzymologia 33, 33-37 8. Scriver, C. R., and Efron, M. L. (1972) in Metabolic Basis of

Inherited Disease (Stanbury, J. B., Wyngaarden, J. B., and Frederickson, P. S., eds) p. 351, McGraw-Hill, New York

9. Brunner, G., and Neupert, W. (1969) FEBS Lett. 3, 283-286 10. Strecker, H. J . (1960) J. Biol. Chem. 235, 3218-3223 11. Stewart, C. R., and Lai, E. Y. (1974) Plant Sci. Lett. 3, 173-179 12. Frank, L., and Ranhand, B. (1964) Arch. Biochem. Biophys. 107,

13. Ratzkin, B., and Roth, J . (1978) J. Bacteriol. 133, 744-754 14. Chumley, F., Menzel, R., and Roth, J . R. (1979) Genetics 91,639-

15. Berkowitz, D., Hushon, J. M., Whitfield, H. J., Roth, Jr., J., and

16. Wu, G. Y., and Seifter, S. (1975) Anal. Biochem. 67,413-420 17. Ames, G. F.-L. (1974) J. Biol. Chem. 249,634-644 18. Bradford, M. M. (1976) Anal. Biochem. 72,248-253 19. Johnson, A. B., and Strecker, H. J . (1962) J. Biol. Chem. 237,

325-331

655

Ames, B. N. (1968) J. Bucteriol. 96,215-220

1876-1882 20. Hedrick. J . L., and Smyth, A. J. (1968) Arch. Biochem. Biophys.

21. Menzel, R., and Roth, J . R. (1981) J. Mol. Biol. 148, 21-44 22. Scarpulla, R. C., and Soffer, R. L. (1978) J. Biol. Chem. 253,

126, 15-23

5997-6001 Strecker, H. J . (1960b) J. Biol. Chem. 235, 2045-2050

441-447 23. Menzel, R., and Roth, J. R. (1981) J. Biol. Chem. 256, 9762-9766 3. Frank L., and Rybicki, P. (1961) Arch. Biochem. Biophys. 95, . . . . -