Identification of flavin adenine dinucleotide and heme in a homogeneous spermidine dehydrogenase from Serratia marcescens.

Abstract Spermidine dehydrogenase has been purified 5,000-fold to homogeneity from extracts of Serratia marcescens. The molecular weight is 76,000; no evidence has been obtained for subunits. The spectrum of the enzyme is characteristic of a heme protein. Both iron-protoporphyrin IX and flavin adenine dinucleotide have been identified in the enzyme in a molar ratio of heme to FAD to enzyme protein of approximately 0.8:0.9:1.0.


Identification of Flavin Adenine Dinucleotide and Heme in a Homogeneous
Spermidine Dehydrogenase from Serratia marcescens* (Received for publication, May 27, 1970) CELIA WHITE TABOR AND PATRICIA DEAL KELLOGG From the Laboratory of Biochemical Pharmacology, National Institute of Arthritis and Metabolic Diseases* National Institutes of Health, Bethesda, Maryland iTOOl SUMMARY Spermidine dehydrogenase has been purified 5,000-fold to homogeneity from extracts of Serratia marcescens.
The molecular weight is 76,000; no evidence has been obtained for subunits.
The spectrum of the enzyme is characteristic of a heme protein. Both iron-protoporphyrin IX and flavin adenine dinucleotide have been identil?ed in the enzyme in a molar ratio of heme to FAD to enzyme protein of approximately 0.8:O.g: 1.0.

NH&H&H&H~NHCH&H&H&H~NHZ + Spermidine
H&-CHZ NH&H&H&HeNH2 f Hp'!? AH (1) \N// 1,3-Diaminopropane A'-Pyrroline Earlier work in this laboratory showed that this enzyme was largely particulate in crude extracts (5) ; a method for solubilization and partial purification was described, and a requirement for an electron acceptor was demonstrated. The finding that the enzyme was inactivated by acid ammonium sulfate precipitation and was partially reactivated by prolonged preliminary incubation with flavin adenine dinucleotide indicated that FAD was a cofactor (5). In the present studies the enzyme has been purified to homogeneity. Its molecular weight is 76,000. We have found that both FAD and the heme, iron-protoporphyrin IX, are present in the pure protein. The molar ratios of enzyme to heme to flavin are approximately 1: 0.8 : 0.9.
* A preliminary report of this work was presented at the meeting of the American Society of Biological Chemists in April 1968 (1).

EXPERIMENTAL PROCEDURE
Enzyme Assays Several procedures have been devised which use different electron acceptors.
A. Potassium Fe&cyanide Assay-For our standard assay, spermidine dehydrogenase was assayed spectrophotometrically in a Gilford recording spectrophotometer by following the decrease in absorbance at 400 rnp due to the reduction of potassium ferricyanide.
The mixture, which was used for routine assays, consisted of 0.2 M potassium phosphate buffer (pH 7.2), 1 mM potassium ferricyanide, and 0.5 mM spermidine hydrochloride; this was stable at 4" for at least a week. One milliliter of this solution was mixed with 10 to 50 ~1 of enzyme (0.02 to 1.0 unit, see below) in a cuvette with a l-cm light path. Two equivalents of ferricyanide were reduced per mole of spermidine. The molar extinction coefficient of potassium ferricyanide is 0.96 x lo3 at 400 rnp. One unit of enzyme catalyzed the oxidation of 1 pmole of spermidine in 1 min at 37".
B. Phenazine Methosulfate Assay-The enzymatic oxidation of spermidine, using phenazine methosulfate as the electron acceptor, was assayed by the reaction of one of the products, Al-pyrroline, with o-aminobenzaldehyde. The 2,3-trimethylene-1,2-dihydroquinazolinium derivative formed (3,(5)(6)(7)(8)(9)) has a molar extinction coefficient of 1.86 X 1Oa at 435 mp. The incubation mixture consisted of 0.1 mM phenazine methosulfate, 0.8 mM o-aminobenzaldehyde, 1 mu spermidine hydrochloride, 0.2 M potassium phosphate (pH 6.0), enzyme, and water in a final volume of 1 ml. The incubation was carried out in a l-ml cuvette with a l-cm light path at 25" for 30 min; the cuvette was inverted at 5-min intervals to aerate the contents. The increase in absorbance at 435 rnp was measured over a period of 30 min in a Beckman DU spectrophotometer.
C. Dichloroindophenol Assay-The oxidation of spermidine was measured by the decrease in absorbance at 600 rnp due to reduction of dichloroindophenol.
The incubation mixture consisted of 0.2 M potassium phosphate (pH 6.5), 20 ~1 of a freshly prepared solution containing 2 mg of dichloroindophenol (sodium salt) per ml, 0.02 mM spermidine, and enzyme in a final volume of 1 ml. The incubation mixture was rapidly stirred and a zero time absorbance was read immediately against a blank cuvette containing the incubation mixture without the 5424 dye. The reaction rate was followed at 600 rnp, at 25", in a Gilford recording spectrophotometer.
The molar extinction coefficient of the dichloroindophenol at pH 6.5 is 1.84 x lo4 (lO).l D. Phenazine Methosulfate-Nitro Blue Tetrazolium Assay-This assay for qualitative determinations of enzyme activity on analytical disc gels was a modification of the method described by Katzen and Schimke (11) for dehydrogenase assays on starch gels. A mixture consisting of 0.1 M potassium phosphate (pH 7.2), 1 mM phenazine methosulfate, 1 mM spermidine, and 200 pg per ml of nitro blue tetrazolium was prepared immediately before use. After electrophoresis of the protein on an analytical gel (see below), the gel was incubated in the dark in a test tube containing the assay mixture for 15 to 30 min at 25". Excess dye was removed by repeated washing with water in the dark; the enzyme bands retained a deep purple stain. The gels were stored in 7.5% acetic acid after the staining procedure.
Protein Determinations-Protein was routinely determined by the Lowry procedure (12), using bovine serum albumin as the standard.
With the purified enzyme, the protein value obtained by the Lowry procedure agreed within 10% with the value determined by a refractometric method with the use of Rayleigh interference optics.2 For the calculations of the protein concentration by the latter method (13), we used the value of 1.875 x NY3 for the specific refractive increment of the enzyme protein (14). The assumption that the presence of the heme does not significantly affect this value (15, 16) is supported by the published value of 1.9415 x 10e3 for hemoglobin (16) and by the value of 1.955 x low3 for catalase (determined by Dr. William CarrolP) . A 1% solution of the purified enzyme (based on the refractometric method) had an Azso of 16.0 f 0.9.
Proteins on analytical acrylamide gels were stained by a modification of the procedure of Chrambach et al. (17). The gels were immersed in a 0.1% Coomassie blue solution in 7.5% acetic acid after fixation with 10% trichloracetic acid; excess dye was removed by washing repeatedly with 7.5% acetic acid.
Absorption Spectra-Absorption spectra were measured in a Cary model 11 recording spectrophotometer.
Other spectrophotometric measurements were determined with a Gilford spectrophotometer.
Heme Determinafion-Heme was determined by the pyridine hemochromogen method of Appleby and Morton (18) ; the procedure was modified to use 80 pg of protein (about 1 nmole) in a final volume of 0.5 ml. The spectrum of the pyridine hemochromogen derivative was determined with the Cary model 11 spectrophotometer using an expanded scale calibrated from 0 to 0.1. 1 Bachrach and Oser (4)  For the quantitative determination of the heme content, we used the molar extinction coefficient of the characteristic absorption peak of the pyridine hemochromogen derivative of ironprotoporphyrin IX (~556 = 33,200; Appleby and Morton (18)). With pure enzyme preparations, the 414-rnp absorbance of the native enzyme was also used directly to compare the heme content of various preparations.
Assay for Flavin Adenine Dinucleotide-FAD was determined fluorometrically by a modification of the method of Bessey,Lowry,and Love (19) using an Aminco Bowman spectrofluorometer. The procedure was carried out in a final volume of 0.1 ml, using a square quartz cuvette with a O.l-ml capacity; satisfactory determinations were obtained on 10 pg of pure enzyme, containing approximately 130 pmoles of FAD. Gel Electrophoresis-Analytical disc gel electrophoresis was carried out at pH 8.9, as described by Davis (20). Columns were prepared with a 7.5% acrylamide separating gel (6 cm) and a 3.1% acrylamide spacer gel (1 cm). The enzyme was routinely applied to the columns in a 10% sucrose solution. Electrophoresis was performed at 5 ma per gel column for about 45 min at 4". Preparative disc gel electrophoresis was carried out as described in the "Canalco Prep-Disc Instruction Manual" (21).
To the enzyme solution, sufficient solid sucrose was added to make the final concentration equal to 10%. The enzyme solution was placed directly on the separating gel and current (5 ma per gel column) was applied for 2 hours. The gel was fixed with trichloracetic acid and stained for protein with Coomassie blue, as described above (17). The molecular weight of enzyme which had been dialyzed against sodium dodecyl sulfate was determined from the mobility obtained by the sodium dodecyl sulfate gel electrophoresis as described above, using bovine serum albumin, bovine hemoglobin, and ovalbumin as standard proteins (22).
Analytical Ultracentrifugatior?-Studies were carried out in a Spinco model E analytical ultracentrifuge, equipped with either Rayleigh interference optics, or with a photoelectric scanner (23). Enzyme solutions were dialyzed for 18 hours against 0.1 M potassium phosphate buffer, pH 7.2.
Determination of Molecular Weight-The molecular weight of the native enzyme was calculated from the data of the sedimentation equilibrium experiments (24-26), and from the diffusion coefficient and sedimentation velocity experiments (26). Sedimentation equilibrium centrifugation of the native enzyme was performed according to the method of Yphantis (25). Protein solutions contained approximately 0.5 mg per ml, and column heights of 3 mm were used. The solution was centrifuged in a double-sector cell at 24,000 rpm at 4" to 7" until equilibrium was reached (21 to 28 hours) ; Rayleigh interference optics were used. The molecular weight of enzyme which had been denatured by dialysis for 18 hours at 25" against 6 M guanidine Spernzidine Dehydyogenase from Xerratia marcescens Vol. 245,No. 20 hydrochloride-O.1 M 2-mercaptoethanol was calculated from the data of the sedimentation equilibrium centrifugation in 6 M guanidine-0.1 M 2-mercaptoethanol at 25". The experimentally determined ti was used for these calculations, with corrections for volume changes due to temperature (27) and to high guanidine concentration (28). The partial specific volume (fi) used for the calculation of the molecular weight was determined according to the method of Edelstein and Schachman (29). In this method, enzyme dissolved in D&80 and enzyme dissolved in water were simultaneously centrifuged to equilibrium in two double-sector cells; the data from this run, obtained by Rayleigh interference optics, were used to calculate the 0. To prepare the enzyme for the centrifugation, the solution was concentrated to approximately 4.4 mg of protein per ml and dialyzed overnight against 0.1 M potassium phosphate buffer, pH 7.2. In order to obtain the maximum enrichment with D&80,200 ~1 of 0.1 M potassium phosphate, pH 7.2, were dried in each of four test tubes in a vacuum desiccator.
The DQ*O used had 98.8% of deuterium and 97.47% of IsO. The final enrichment used for centrifugation was 89.8% of deuterium, and 88.6% of l*O. The B was also calculated from the amino acid composition of the enzyme (26, 30).
Density Determination-The density determination of 0.1 M potassium phosphate buffer, pH 7.2, was performed in a standard capillary neck pycnometer (10 ml) at 25". The density used for the 6 M guanidine-0.1 M 2-mercaptoethanol solution was that determined by Kawahara and Tanford (31); the density for the D&*0 solution was calculated by the method of Edelstein and Schachman (29). Viscosity values were obtained from the International Critical Tables. Determination of Diflusion Coeficient-The diffusion coefficient was determined by centrifugation in an artificial boundary cell. A solution of pure enzyme, containing 4.0 mg per ml by the fringe count, was centrifuged at 5,000 rpm, at 3.7", for 4 hours. The fringe spreading was measured according to the technique of Longsworth (32) .2 Determination of Sedimentation Velocity-Sedimentation velocity experiments were carried out according to the method of Schachman and Edelstein (23) using an ultracentrifuge equipped with a photoelectric scanning optical system. Recordings of the sedimentation pattern were obtained with an 8-mm column of pure enzyme solution containing 380 pg per ml of protein.
The solvent was 0.1 M potassium phosphate, pH 7.2. The centrifugation was carried out at 60,000 rpm and 19.8" for 44 min while recording at 280 rn+ In order to demonstrate definitively that the moiety with the 414-rnp absorbance peak was an integral part of the protein molecule, we also determined the sedimentation velocity of this 414-rnp absorbing material. Although it was not possible to obtain readings at this wave length, readings at 404 rnp were suitable for monitoring the sedimentation velocity of this moiety. For this purpose, after 44 min, without interrupting the centrifugation, the lens was repositioned and the wave length changed so that recordings were obtained at 404 rnp during an additional 24 min at the same speed and temperature.
An additional centrifugation was performed with a dilution of the same enzyme, containing 127 pg per ml, at 60,000 rpm, 18.9", for 48 min; because of the di1ut.e solution used, the recording was at 235 mp.
Amino Acid Ana!yses-Amino acid analyses were determined with a Beckman amino acid analyzer3 on two independent preparations of pure enzyme. The cysteic acid residues were determined after performic acid oxidation (33). The samples were prepared by anaerobic hydrolysis for 22 hours at 110" in redistilled 6 N HCl.
Isoelectric Point-The isoelectric point of the enzyme was determined by the electrophoresis of a 300-fold purified enzyme in the LKB electrofocusing apparatus, using a 440-ml jacketed column (No. 8100-20) packed with a buffer gradient of LKB ampholytes from pH 3.0 to pH 10.0 (34). Enzyme (5 mg of protein; specific activity 27) was added to the ampholine-sucrose mixture which was layered in the middle of the column.
Electrophoresis was carried out at 440 volts for 40 hours; the current was 17 ma at the onset and 3 ma at the end of the run. The column was cooled during the run by pumping water at 4" through the jacket.
The gradient was then allowed to run out of the column; fractions of 5 ml were collected and assayed for activity and pH.
Sucrose Density Gradient Centrifugation-The centrifugation was performed according to the method of Martin and Ames (35).
N , N'-Bis(3-aminopropyl)-l , a-propanediamine tetrahydrochloride was synthesized by Dr. Herbert Tabor by a modification of the method described for the synthesis of spermine (38). All other materials were obtained commercially.
Guanidine hydrochloride (ultrapure) was purchased from Mann Research Laboratories, Inc.

Growth of Organism and Preparation of Extract
S. marcescens (ATCC 25179)4 has been used as the source of the enzyme.
Cells were grown at 30" for 18 hours, with vigorous aeration in a salts-citrate medium (39). No additional carbohydrate was added, since the cells used the citrate in the medium as a carbon source. Glucose, succinate, or a rich medium decreased the total enzyme yield. The addition of N-(S-aminopropyl)-1 ,3-propanediamine to the minimal medium increased the yield of enzyme 3-fold.
Spermidine could replace N-(3aminopropyl)-1 ,3-propanediamine, but was not used routinely because of the expense. Attempts to increase the yield of enzyme by the use of spermidine (lWa M) as the sole carbon source resulted in a 6-fold increase in the specific activity of the enzyme over that obtained with minimal medium, but growth of the cells was very slow. Use of spermidine as the sole nitrogen source resulted in good growth of the culture with only a 3-fold increase in specific activity. N-(3-Aminopropyl)-l , 3-diaminopropane trihydrochloride (lOea M) did not support growth when used as either the sole carbon or the sole nitrogen source. The addition of 1,3-diaminopropane had no effect on the yield of enzyme.
Under our standard conditions, the cells were a deep red (3,5). Growth without aeration or growth at 37" resulted in cells which contained + to f as much enzyme activity as our standard cultures; these cultures were essentially colorless.
We also tested another strain of S. marcescens (Nima), and two mutant strains of Nima (WF, 933),4 which cannot make the charactersistic red pigment, prodigiosin (40). The enzyme activity of the crude extracts of all these strains was approximately the same as in our standard strain.
Cells were grown in a vat containing 320 liters of Vogel-Bonner (39) medium, 30 ml of N+aminopropyl)-1 ,3-propanediamine, and 57 ml of concentrated HCl; 4 liters of a stationary culture were used as the inoculum.
The culture was incubated for 17 to 20 hours at 30" with vigorous aeration; antifoam was added as needed. The culture was then cooled, and the cells were harvested in a refrigerated Sharples centrifuge. 5 The wet paste (approximately 650 g) was suspended in 6 times its weight of distilled water and was passed once through a Gaulin homogenizer at 12,000 to 13,000 p.s.i. During this process, the temperature rose to about 25"; therefore, the extract was rapidly cooled in crushed ice after the homogenization.
About 50% of the enzyme in this extract sedimented when an aliquot was centrifuged at 100,000 x g for 1 hour. The homogenization was then repeated three times; care was taken to cool the homogenate to 04" after each treatment.
The extract was then centrifuged in a refrigerated Sharples centrifuge. At this stage in most preparations, 70 to 80% of the total cellular activity was soluble after centrifugation at 100,000 x g for 1 hour. When solubilization was less complete (as in the purification reported in Table I), the sediment was resuspended in 6 times its weight of distilled water, was again passed through the homogenizer four times, and centrifuged in a Sharples centrifuge. The supernatant fractions were pooled and could be stored at -20" for several months without loss of activity.
The use of water for the suspending medium was essential to obtain soluble enzyme; in the presence of phosphate buffer, the yield of soluble enzyme was low. The use of water instead of phosphate buffer (41) obviated the requirement for EDTA and an alkaline pH which we found necessary for solubilization in our earlier studies (5).

Pur$ication
A typical purification is presented (Table I). In some preparations, however, when an early step failed to yield the expected purification, an additional DEAE-cellulose or calcium phosphate column was included in the procedure.
All of the steps were carried out at O-4", except where otherwise noted.
Step 1 Q A unit is the amount of enzyme that oxidizes 1 rmole of spermidine in 1 min, using Assay A.
b Protein content was assayed by the Lowry procedure (12) in all steps except the last; after gel electrophoresis it was assayed by the extinction at 280 rnp (see "Experimental Procedure").
with a linear gradient formed from 2 liters of 0.1 M KC1 in 0.005 M potassium phosphate, pH 7.2, in the mixing flask, and 2 liters of 0.5 M KC1 in 0.005 M potassium phosphate in the reservoir. The flow rate was approximately 300 ml per hour. The enzyme usually was eluted after 1500 ml of buffer had passed through the column.
Fractions containing enzyme with a specific activity greater than 0.25 unit per mg of protein (approximately 1000 ml) were pooled.
The active fractions from four columns were combined and used for Step 2.
Step 2. Calcium Phosphate Cellulose Gel Adsorption and Elution-Calcium phosphate cellulose gel was prepared by the method of Price and Greenfield (43). A pad, 2.5 cm high x 16.5 cm in diameter; was prepared by pouring the calcium phosphate cellulose suspension onto a 16.5-cm sintered glass funnel; suction from a water pump was applied until all of the supernatant fluid was removed.
The pooled DEAE eluates from Step 1 were poured onto the pad, and suction from a water pump was applied to remove the liquid rapidly.
The filtrate, which contained approximately 80% of the activity, was collected in a suction flask immersed in ice slush. The pad was then washed with 500 ml of cold 0.01 M potassium phosphate, pH 7.2. Essentially all of the enzyme was recovered in the combined filtrate and wash. These combined filtrates were adsorbed to a second calcium phosphate cellulose pad prepared in the same way; the pad was washed with 500 ml of cold 0.01 M potassium phosphate buffer, pH 7.2. Approximately 80% of the enzyme was recovered in the combined filtrates.
The protein was then precipitated by the addition of 51.6 g of solid ammonium sulfate (Mann's special enzyme grade) per 100 ml (to 80% saturation). The precipitate was collected by centrifuga6ion at 10,000 x g at 4" for 20 min and dissolved in 400 ml of 0.01 M potassium phosphate, pH 7.2. This solution was dialyzed against 10 volumes of 0.005 M potassium phosphate, pH 7.2, for 5 hours, and then overnight against 10 volumes of fresh buffer.
Step 3 Stepwise elution was carried out with potassium phosphate buffers, pH 7.2; 5-ml fractions were collected.
The column was washed first with 100 ml of 0.07 M buffer, and then was eluted successively with 100 ml of 0.075 M, 100 ml of 0.08 M, 200 ml of 0.085 M, and 100 ml of 0.09 M buffer. The fractions with the best specific activity (usually eluted with 0.085 M buffer) were pooled and concentrated by ultrafiltration in a collodion bag apparatus. 6 This ultrafiltration sometimes resulted in an additional a-fold increase in specific activity.
The enzyme was dialyzed against 100 volumes of 0.001 M potassium phosphate, pH 7.2, for 12 hours.
Step 4. Hydroxylapatite Chromatography-Hydroxylapatite, prepared by the method of Levin (44), was gently packed into a column (2.5 x 16 cm) and was washed with 0.001 M potassium phosphate, pH 7.2. The concentrated dialyzed enzyme from Step 3 was adsorbed onto the column; the column was eluted first with 50 ml of 0.005 M potassium phosphate, pH 7.2, followed by a linear gradient formed from 100 ml of 0.005 M potassium phosphate, pH 7.2, in the mixing vessel, and 100 ml of 0.05 M buffer in the reservoir.
Fractions of 1 ml were collected; the activity was eluted after 48 ml of the gradient had been collected.
Step 5. Preparative Acrylamide Disc Gel Electrophoresis-The material obtained in Step 4 was subjected to preparative disc gel column electrophoresis in a Canalco jacketed column PD-2/70 in a 4" cold room, with the water jacket filled, but without additional cooling.
The concentrated enzyme from Step 4 was brought to 10% sucrose concentration by adding solid sucrose; a drop of indicator dye (bromphenol blue) was added. This solution was layered on a column consisting of a l&cm stacking gel and a 3-&m separating gel, pH 8.8, prepared according to the Canalco prep-disc formulation (21). A current of 8 ma was applied until the protein moved into the separating gel. The current was then increased to 10 to 15 ma. Current was applied for approximately 5 hours. Elution of the column was carried out continuously with 5% sucrose-O.38 M Tris-HCl, pH 8.8, while the current was applied at a flow rate of 2 to 3 ml per min. The deep red band formed by the enzyme was eluted approximately 1 hour after the dye. The 10 fractions with the highest ratios of units to 280.rnp absorbance were pooled and then concentrated with dialysis against 0.1 M potassium phosphate, pH 7.2, in an ultrafiltration collodion sac (Table I).

Criteria of Purity
All of the recoverable activity moved as one sharp red band on preparative disc gel electrophoresis. Analytical disc gel electrophoresis at pH 8.8 (20) showed that this enzyme, when freshly prepared, had only one band of activity, determined by the nitro blue tetrazolium assay, with an RF of 0.67; only one band of protein with the same RP was found on a parallel gel. Additional evidence for the homogeneity of the enzyme was obtained from the linear character of the plots of the data from the sedimentation equilibrium and sedimentation velocity experiments presented below (Figs. 2, 3, and 4).

Stability of Enzyme
The pure enzyme could be stored at -20" in 0.1 M potassium phosphate buffer, pH 7.2, for over 6 months with less than 10% loss of activity.
Storage of the enzyme for 3 or 4 months at -20" resulted in the appearance of multiple bands of activity which migrated more slowly than the major band. Analytical disc gel electrophoresis of this material showed that these additional bands usually contained less than 5% of the protein.
The enzyme was stable when heated for 5 min at 58". However, 1 min at 65" resulted in 65% inactivation, and 3 min at 65" resulted in complete inactivation.

Absorption Spectrum of Pure Xpermidine Dehydrogenase
The pure enzyme shows one major peak of absorption at 272 rnp and another at 414 rnp (A272:A414 is 1.35) (Fig. 1, Curve A).
Hydrodynamic Properties of Purijed Dehydrogenase Fig. 2 shows the data obtained from a high speed sedimentation equilibrium centrifugation of a pure preparation of spermidine dehydrogenase. The linearity of the data indicates that, with respect to molecular weight, only one species of protein was present.
The partial specific volume, determined as described in "Experimental Procedure" (29)) was 0.738 (Fig. 3). Determination of the B from the amino acid content (see below) (26) gave a value of 0.736.
The molecular weight determinations, calculated from the data of the sedimentation equilibrium centrifugations of Fig. 2 and Fig. 3, Curve A, were 78,000 and 76,000. The value obtained from the sedimentation velocity centrifugation and the diffusion coefficient (see below) was 79,000 (24, 26).
The molecular weight calculated from the data obtained from centrifugation in a sucrose density gradient (35)) using catalase, glutathione reductase, and hexokinase as markers, was 76,000.
The sedimentation velocity of the enzyme was determined by scanning at 280 rnp to follow the protein and at 404 rnp to follow the 414-rnp absorbing material (23). A straight line of identical slope was obtained from both sets of data, indicating that the sedimentation velocity was identical at both wave lengths. Thus, the 404-rnp absorbing moiety and the protein sedimented as a unit (Fig. 4). The sedimentation coefficient (Sag+,) was 4.85 S. At a low protein concentration, with scanning at 235 rnp, the sedimentation coefficient was essentially the same (s~~,~ = 5.04 S) as in the concentrated solution.
The diffusion coefficient of the enzyme (D2E.J was 7.45 x 10-T cm2 per sec.
In addition to the studies on native enzyme described above, the molecular weight was determined for enzyme that had been denatured by dialysis against 6 M guanidine-0.1 M %-mercaptoethanol for 24 hours at 25". Sedimentation equilibrium centrifugation was then carried out at 25" and 24,000 rpm in 6 M guanidine. The molecular weight was found to be 70,000.
When the sample which had been denatured with guanidine was examined spectrally, almost all of the 414-rnp absorbance was absent, suggesting that the heme had been lost during the denaturing process. To study whether or not the removal of cofactors was associated with a small decrease in the molecular weight, we used other procedures which removed some or all of the heme and flavin. One sample of enzyme was dialyzed against 1 M potassium phosphate, pH 7.2, for 2 hours and then overnight against 0.1 M phosphate buffer. About 50% of the 414-rnk absorbance was removed by this dialysis. High speed sedimentation equilibrium centrifugation was carried out on this preparation; calculation of the molecular weight led to a value of 65,500.
Another sample was subjected to acid ammonium sulfate precipitation to remove the FAD, as described in the legend to Fig. 6, and then dialyzed overnight against 0.1 M potassium phosphate buffer, pH 7.2. A molecular weight of 68,000 was determined from sedimentation equilibrium data. The molecular weight of the holoenzyme from which this preparation was derived was 74,000. In all of these determinations, the experimentally determined fl of 0.738 was used, with the corrections necessary for guanidine concentration or temperature.
Similar results were obtained when the molecular weight of the protein was determined by electrophoresis on sodium dodecyl sulfate gels. The enzyme, denatured by dialysis against sodium dodecyl sulfate (as described in "Experimental Procedure"),

Spermidine
Dehyclrogenase from Serratia marcescens Vol. 245,No. 20   Q The values given here are the average of three analyses on two independent preparations of pure enzyme recalculated for a molecular weight of 76,000. Two of the assays were performed on aliquots from the same preparation; one of these was hydrolyzed in the presence of 4% (v/v) thioglycollate, as described by Matsubara and Sasaki (45). In thisanalysisthevalue for prolinewasapproximately 25% greater than in the other runs. The third analysis was performed on another preparation after precipitation with 5oj0 trichloracetic acid. There was no significant difference between the values obtained for the two independent preparations of enzyme. Although the Matsubara and Sasaki procedure was satisfactory for the determination of tryptophan in chymotrypsin, we were not able to use it for the determination of the tryptophan content of spermidine dehydrogenase. With this enzyme, other products developed during the hydrolysis which chromatographed with tryptophan on the amino acid analyzer column, and which resulted in an atypical absorption after reaction with ninhydrin.
*Determined by analysis of an independent preparation of spermidine dehydrogenase, after performic acid oxidation (33). The values for most of the amino acids on the acidic column in this run were similar to those reported in this table.
showed a 50% reduction in the 414-rnp peak. On electrophoresis in sodium dodecyl sulfate gels, two approximately equal bands were observed, with calculated molecular weights of 60,000 and 68,000. In this experiment, less than 5% of the total protein was also found in minor bands of lower molecular weight.

Isoelectric Point
The isoelectric point of the enzyme, determined by electrofocusing according to the method described in "Experimental Procedure," was at pH 4.5.

Amino Acid Analysis
The amino acid composition of the spermidine dehydrogenase is summarized in Table II.

Identijication of Flavin Adenine Dinucleotide in Spermidine Dehydrogenase
In earlier studies, Campello, Tabor, and Tabor (5)  Pure spermidine dehydrogenase (0.1 ml containing 8 rg of protein with a specific activity of 400 units per mg) was added at 0" to 0.15 ml of a solution of bovine serum albumin (1.7 mg per ml). To this was added 0.25 ml of 3 M KBr, followed by 0.5 ml of 4 M (NHJzSOd mixed with 0.027 ml of 1 N H&04. The precipitate which formed was centrifuged rapidly (1 min at 10,000 rpm at 0') and was dissolved immediately in 0.1 ml of 0.1 M potassium phosphate, pH 7.2. A portion, 0.05 ml, was added to 0.05 ml of 1 M K2HP04 to bring the pH to 8.3, and was stored at room temperature for 23 hours. After this preliminary incubation, 0.05 ml (Sample A) was cooled at 0", and sufficient FAD was added to give a final FAD concentration of 0.1 mM. A second portion (Sample B) was cooled and used for the assay without further treatment.
The activities of the enzyme solutions with FAD (Sample A) and without FAD (Sample U) were assayed at various time intervals to determine the enzymatic activity.
In this study, we have been able to demonstrate similar results with pure enzyme by adding carrier bovine serum albumin and using the conditions described by Strittmatter (46). Reactivation of 13 '% was observed within 2 min, and of 25% within 1 hour after the addition of FAD (Fig. 5, Curve A).
Additional evidence for the presence of FAD was obtained from the spectrum of the soluble cofactor in the supernatant fraction after acid ammonium sulfate precipitation of the pure protein.
The spectrum (Fig. 6, Curve A) closely resembles that of FAD (Fig. 6, Curve B). The spectrum of the apoenzyme (Fig. 6, Curve C) resembles that of an oxidized heme (see below) (47). In the spectrum of the native enzyme (Fig. 1, Curve A) one cannot find spectral evidence for the flavin since the flavin absorbance is masked by the prominent 414-rnp peak.
The FAD content of four independent preparations of the pure enzyme was also determined fluorometrically, as described under LLExperimental Procedure." The average value found was 0.86 mole of FAD per mole of enzyme (using the observed molecular weight of 76,000) (Table III).
No FMN was present.

Ident$cation of Heme in Spermidine Dehyclrogenase
The visible spectrum of the oxidized enzyme is characteristic of a heme (Fig. 1). After addition of substrate without any electron acceptor the major peaks shift to 265 and 427 rnp (A265:A427 is 0.99). Two small distinct peaks also appeared at 530 and 562 rnp (Fig. 1, Curves A and B). The apoenzyme formed by acid ammonium sulfate precipitation retained the spectrum of the oxidized heme; however, the peak in the ultra- The soluble supernatant fraction was diluted to 400~1 with HzO. The spectra of both fractions were determined with a Cary model 11 recording spectrophotometer, using a slide wire calibrated from 0 to 0.1 absorbance.
The spectrum from 250 to 400 rnp was obtained with the use of an ultraviolet light source; from 400 to 600 rnp, with the use of the tungsten light source. Curve A, soluble supernatant fraction; Curve B, standard FAD in a solution containing the same KBr, (NHJzSOd, and H&O4 concentrations as in the supernatant fraction; and Curve C, precipitated protein, dissolved in phosphate buffer. violet shifted from 272 to 276 rnp (Fig. 6C). In order to demonstrate that the heme remains in a constant ratio to the active protein during purification, we determined the ratio of the 414-rnp absorption peak to enzymatic activity over an 11-fold increase in purification.
The ratio was 2.75 in a 450-fold purified enzyme and 1.62 after the maximum purification (5000-fold purification). The ratios of absorbance at 414 rnp to enzyme activity for two independent preparations of pure enzyme were 1.62 and 1.70. To identify the type of porphyrin present, and to determine the amount present, we formed the pyridine hemochromogen derivative of the pure enzyme (see "Experimental Procedure") (18). The spectrum of this derivative (Fig. 7) was identical to that formed with iron-protoporphyrin IX (48). From the molar extinction coefficient of the pyridine hemochromogen derivative of iron-protoporphyrin IX (18, 48), we determined the number of moles of heme per mole of enzyme. In three of the preparations used for the flavin determination there was an average of 0.78 mole of heme per mole of enzyme (Table III).

Characteristics of Enzyme Reaction
The enzyme reaction rate, carried out with a pure preparation, was linear with respect to time and enzyme concentration.
The rate of oxidation did not vary significantly over a range of spermi- For details see "Experiment,al Procedure." dine concentrations from 0.0005 to 0.1 mM. Thus, the K, for spermidine was too low to determine (less than 0.0005 mM); the pH optimum was at 7.2. Two equivalents of ferricyanide were reduced per mole of spermidine.
The K, for spermine in the standard incubation mixture (pH 7.2) was 0.02 InM.
The pH optimum was 8.8; at this pH the K, was approximately 0.05 mM. Two equivalents of ferricyanide were reduced per mole of spermine, suggesting that spermine was oxidized at only one of the secondary amine groups. Monoacetylspermidine B, N, N'-bis(3-aminopropyl)-l , 3-propanediamine, and N-(3-aminopropyl) -1, 3-propanediamine, which are similar to spermidine in structure, are oxidized more slowly than spermidine, even when present in much higher concentrations (Table IV) and under saturating conditions. However, the following secondary amines were not oxidized at all when present in the concentration range from 0.5 mM to 5 mM: dibutylamine, monoacetylspermi- and proline were all inactive as substrates. None of these compounds inhibited spermidine oxidation. The turnover number, using Assay A (see "Experimental Procedure"), is 30,000 moles of spermidine per min per mole of enzyme.
We have been able to show that a natural electron acceptor, horse heart cytochrome c, can serve as an electron acceptor for this reaction.
A reaction mixture containing 0.01 pmole of cytochrome c, 0.1 pmole of spermidine, 5 pmoles of potassium phosphate, pH 7.2, and 2.5 units of enzyme (specific activity, 265 units per mg) in a final volume of 1 ml, was incubated at 37" for 30 min. The ASSO increased from 0.100 at zero time to 0.243. In order to determine the completeness of the reduction, 2 mg of sodium dithionite were added at the end of the incubation (49). The final As50 of 0.260 after dithionite addition indicated that over 90% reduction of the cytochrome c had resulted from the enzymatic oxidation of spermidine.
The increase in absorbance at 550 rnp was found to be proportional to the amount of spermidine added.
Increments of 1 pmole of spermidine were added at 3-min intervals to a l-ml incubation mixture containing 0.01 pmole of cytochrome c, 5 pmoles of potassium phosphate buffer, pH 7.2, and 2.5 units of enzyme (specific activity of 265 units per mg). After each addition, the reaction went to completion in less than 30 sec. The AScO readings were 0.105 before the addition of spermidine, and 0.130, 0.158, 0.182, 0.205, and 0.215 after successive additions of spermidine.
Under these conditions, the reduction of cytochrome c was much slower than of ferricyanide but could be used to follow the reaction.
We were unable to couple the reaction to DPN, ferredoxin, or coenzyme Q under similar conditions.
The monoacetylspermidine B (with the acetyl group on the primary amine of the 3-carbon chain) is active as a substrate.
Quinacrine, added to a reaction mixture at 0.2 InM final concentration, resulted in 80% inhibition.
Isoniazid, iproniaeid, cyanide, sodium azide (each at 1 mM), and borohydride reduction did not inhibit the enzyme. DISCUSSION Spermidine dehydrogenase is found in S. murceacens cultures even when grown in the absence of added spermidine.
The amount of enzyme activity can be increased several fold by the addition of spermidine or N-(3-aminopropyl)-l , 3-diaminopropane to the growth medium.
About 50% of the enzyme is found in a particulate form in water extracts of X. murcescens. The natural factors which couple this enzyme to molecular oxygen are lost upon solubilization, and the soluble enzyme has a requirement for an added electron acceptor. During purification, a heme and FAD remain with the enzyme; in a homogeneous preparation, 5000-fold purified, we find a molar ratio of heme to flavin to enzyme of 0.8:O.g: 1. We have been able to resolve the flavoprotein into an inactive apoenzyme, which still contains the heme; this undergoes about 25y0 reactivation upon the addition of FAD.
We have also been able to remove the heme by dialysis against 1 M potassium phosphate, but have not yet been able to reconstitute an active enzyme from the apoenzyme lacking the heme.
The molecular weight determinations from sedimentation equilibrium centrifugation, sedimentation velocity studies, and sucrose density gradient centrifugation yield a value of 76,000 i 3,000. No conclusive evidence for subunits has been obtained. However, suggestive evidence for the loss of a small polypeptide has been obtained by the decreased molecular weight (65,000 to 70,000) observed after dialysis of the holoenzyme against 1 M phosphate, guanidine, or sodium dodecyl sulfate, and after dialysis of the flavin apoenzyme against 0.1 M phosphate.
The spectrum of the pure enzyme contains two major peaks, one at 272 rnp and one at 414 rnp, and two minor peaks at 530 mp and 560 rnp. When substrate is added to the enzyme in the absence of an electron acceptor, the 414-, 530-, and 560-rnp peaks (which are characteristic of a heme) show the shifts which occur upon reduction of a heme; the 272-rnp peak also shifts to 265 mp. It is difficult to find evidence of the FAD in the spectrum because of the very strong absorbanceat 414 rnp. However, after resolution of the enzyme, a 276-rnp peak is found instead of the 272-rnp peak of the holoenzyme; this is consistent with the loss of a flavin moiety absorbing at 260 mp.
The presence of both a heme and a flavin has been described in two other pure enzymes: crystalline yeast cytochrome bz (18)) which contains a heme and FMN, and sulfite reductase of Escherichia coli, which contains a heme and both FMN and FAD (50). Both cofactors have also been found in a partially purified nitrate reductase from Neurospora cra.ssa (51). FAD has been shown to be a cofactor for tyramine oxidase of Xarcina Zutea (52) and for putrescine oxidase of Micrococcus rubens (53), and has been implicated as a cofactor for tissue monoamine oxidase in mammals and histaminase from hog kidney by some authors (54, 55), but not by others (56-58).
Spermidine dehydrogenase is the only amine oxidase that has been shown to contain both FAD and a heme, and to require an added electron acceptor for activity.