A Phospholipid-requiring Enzyme, Malate-Vitamin K Reductase

Malate-vitamin K reductase has been purified to near homogeneity from Mycobacterium phlei. The purified enzyme is dependent upon an added phospholipid for activity and requires FAD as a cofactor. The enzymatic activity was found to be affected by the degree of enzyme aggregation. At high salt concentrations the enzyme existed in a monomeric form which was more active than the aggregated form. The enzyme was reversibly aggregated into a less active form by either dilution or dialysis against a buffer of low salt concentration. An enzyme-phospholipid complex was isolated by glycerol gradient centrifugation. It is suggested that a phospholipid binding site (or sites) seems to be involved in the aggregation-disaggregation process. The molecular weight of the monomeric form was determined to be 53,000 by Sephadex G-ZOO chromatography and 51,000 by sodium dodecyl sulfate gel electrophoresis, whereas the aggregated form had a molecular weight of approximately 164,000, as estimated by Sephadex G-ZOO.

A requirement of phospholipid for enzymatic activity has been demonstrated with many enzymes (1, 2). The pioneering work of Fleischer et al. (3) showed the importance of phospholipid in the electron transfer activity of mitochondria.
The electron transfer decreased after extraction with an organic solvent but was restored by the addition of an extracted lipid fraction. P-Hydroxybutyrate dehydrogenase has been shown to be dependent on lipid for activity (4,5). Lecithin is specifically required, whereas many other phospholipids and neutral lipids are wit.hout effect. A requirement for lipid for enzymatic activity has been investigated with the membrane-bound enzymes in mitochondria (6-lo), microsomes (ll-15), and bacteria (16-18). Phospholipids have also been shown to be necessary for prothrombin activation (19) and rhodopsin regeneration (20).
* This work was supported by grants from the National Science Foundation (GB 32351X), the Nat,ional Institutes of Health (AI 05637), and the Hastings Foundation of' the University of Southern California School of Medicine.
$ Visiting Research Associate, on leave of absence from the Institute for Enzvme Research. School of' Medicine. Tokushima University, Tokushima, Japan.
The sonic extracts from dlycobacterium phlei contain an unique enzyme, malate-vitamin K reductase, which requires added phospholipid for activity (21-23). The partial purification and general properties of this enzyme were reported (24). In the present paper malate-vitamin K reductase was further purified and other characteristics of the enzyme, especially with respect to phospholipid binding, are described.

EXPERIMEKTAL PROCEDURE
Materials-DEAE-cellulose was obtained from Sigma Chemical Co. Hydroxylapatite was the product of Hio-Rad Laboratories. Sephadex G-200 and QAE-Sephadex A-50 were purchased from Pharmacia Fine Chemicals.
Asolectin (soy bean phospholipid) was obtained from Associated Concentrates and purified by the method of Kagawa and Racker (6). All other chemicals were of reagent grade.
Growth of Bacteria--M. phlei (ATCC 354) was grown as previously described (25). Cells were sonically disrupted and fractionated into particulate and supernatant fractions (26). The supernatant fraction was used as the starting material for purification of the enzyme. Assay of Enzyme Activity-Malate-vitamin K reductase activity was measured spectrophotometrically in a cuvette of l-cm light path with a Cary model 14 recording spectrophotometer (24). The reaction system contained 100 pmoles of Tris-HCl (pH 7.4), 40 pmoles of KCl, 12.5 nmoles of FAD, 0.24 pmole of MTT,' 1.1 pmoles of vitamin K1 (sonically dispersed with 2.5 mg of Asolectin), and an enzyme sample in a final volume of 1.5 ml. The reaction was started with m-malate (50 pmoles) and MTT reduction was followed spectrophotometrically at 565 nm. A millimolar extinction coefficient of 15.0 was used for MTT (24). PMS (0.66 pmole) and DCIP (0.15 pmole) were also used as final electron acceptor.
With PMS and DCIP as the electron acceptors, 2.5 mg of (sonically dispersed) Asolectin were added. A unit of activity is defined as 1.0 pmole of final acceptor reduced per min at room temperature.
Specific activity is in units per mg of protein.
Phospholipid Preparations-Phospholipid (Asolectin) was dispersed by a 30-min sonication as described previously (24). This preparation was used without further centrifugation. Phospholipid or phospholipid and vitamin K1 were also prepared by the method of Fleischer and Klouwen (27). 7485 Chemical Determinations-Protein concentration was determined by the method of Lowry et al. (28), with bovine serum albumin as the standard.
Phospholipid was estimated accordng to the method of King (29).
Electrophoresis-Polyacrylamide gel electrophoresis was carried out according to the method of Weber and Osborn (30) in the presence of 0.17" SDS. Samples were usually treated with 1 70 SDS, 1 70 mercaptoethanol, and 4 M urea for 45 min at 45". The gels were stained overnight with 0.05% Coomassie blue in 5O70 methanol-l070 acetic acid solution and then destained, first with the same concentration of the methanol-acetic acid solution and then with 7% acetic acid. Analytical gel electrophoresis in 5y0 gel was carried out by the method of Davis (31).

Glycerol
Density Gradient Centrijugation-G lycerol gradient centrifugation was performed by the method of Martin and Ames (32). A 0.1.ml sample was layered on top of a 4.5.ml linear gradient of glycerol (5 to 2570 v/v) in a 20 mM Tris-WC1 (pH 82-0.5 mM EDTA solution.
At the bottom of the gradient, was placed 0.3 ml of 50% glycerol.
The tubes were centrifuged at 38,000 rpm for 18 hours at 4-6" in a Peckman SW 50.1 rotor.
Gel filtration was performed by descending chromatography and fractions of 3.0 ml were collected.
The distribution coefficient (K,,) equals ve -vo/vt -~0 and was experimentally determined for each protein standard (33). Amino Acid Analysis-Samples were dialyzed overnight against water and then were hydrolyzed in 6 x HCl at 110" for 18 hours in oacuo. The hydrolyzed material was analyzed with a Joel-5AH automatic amino acid analyzer.

RESULTS
Puri$cation-Nucleic acids and inactive proteins were removed from the starting material by precipitation with streptomycin. A 3f0 volume of 10% streptomycin sulfate solution was added dropwise with continuous stirring to crude supernatant fraction (300 to 350 ml). After additional stirring for 20 min the extract was centrifuged at 10,000 x g for 20 min. The supernatant was diluted with an equal volume of cold water and EDTA was added to a final concentration of 0.5 mM. The precipitate formed between 30 and 50% ammonium sulfate saturation was collected by centrifugation, dissolved in 100 ml of cold 10% glycerol, and dialyzed against 4 liters of 10 70 glycerol overnight.
The dialyzed material was applied to a DEAE-cellulose column (4 x 55 cm) equilibrated with a 10% glycerol and 20 mM Tris-HCl (pH 8.2) solution.
The column was washed with the buffer until the eluate was colorless.
The gradient elution was accomplished by the use of a linear gradient of KC1 from 0 to 0.35 M in the same buffer. The total volume of the gradient was 2 liters.
The eluate was collected in 15-ml samples and the active fractions were found in Fractions 84 to 122. The active fractions were pooled and then precipitated by the addition of ammonium sulfate (5070 saturation).
The precipitate was collected, dissolved in 20 ml of a solution of 10 mM potassium phosphate buffer (pH 6.8) and 10% glycerol.
In order to obtain enough protein for further purification the fractionation procedure was repeated with another batch of starting material.
The two batches obtained following DEAE-cellulose fractionation were combined and dialyzed against 1 liter of a 10 mM phosphate buffer (pH 6.8) and 10% glycerol solution.
The dialyzed enzyme was applied to a hydroxylapatite column (2.1 X 20 cm) which had been equilibrated with the same buffer. The column was extensively washed with a solution containing 80 rnM potassium phosphate (pH 6.8) and 10% glycerol until the absorbance of the eluate at 280 nm decreased below 0.1 O.D. The enzyme was eluted with a linear gradient from 80 to 160 mM phosphate in a total volume of 800 ml. The eluate was collected in lo-ml samples and the active fractions (Fractions 46 to 70) were con centrated first in an Amicon Diaflo cell equipped with a PAI-30 membrane and then with a collodion bag. The enzyme solution was completely colorless at this step.
The concentrated enzyme was then applied to a Sephadex G-200 column (2.6 x 90 cm) equilibrated with a 0.3 M KCl, 20 mM Tris-HCl (pH 8.2), and 10 70 glycerol solution.
Gel filtration was performed as descending chromatography with a flow rate of 15 to 20 ml per hour. Fractions of 3 ml were collected and the active fractions (Fractions 108 to 126) were concentrated as described above and 20 mM Tris-HCl (pH 8.2) in 10% glycerol was added to obtain a 0.1 M KC1 concentration.
After the column was washed with several column volumes of the equilibrated buffer, the elution was started by increasing the concentration of KC1 to 0.2 M. The active fractions were pooled and concentrated as described above. The active material chromatographed essentially as one component as will be discussed below.
A summary of the purification procedure is given in Table I. The specific activities of the final step varied from 50 to 80 units per mg of protein.
A 30 to 5070 higher activity was obtained with phospholipid preparations prepared by the method of Fleischer and Klouwen (27). As can be seen in Fig. 1, gel electrophoresis in the presence of SDS revealed a single major band with two faint bands. Judged from relative color intensity measured with a Gilford gel scanner, the enzyme was approximately 85% pure. E$'ects of Glycerol and KC&In an attempt to isolate the enzyme, it was found that the composition of the buffer solution used to dissolve the enzyme had a pronounced effect on the enzymatic activity.
The enzyme was dissolved in different compositions of buffer solution at 4" and assayed for activity at the times indicated (Fig. 2). As shown in Fig. 2, 20 70 glycerol and 0.35 M KC1 have a stabilizing effect on the enzyme.
After hydroxylapatite column chromatography, the enzyme preparation isolated in the absence of glycerol was chromatographed on a Sephadex G-200 column (2.6 x 50 cm). The column was equilibrated with three different buffer solutions: Solution 1, 5 mM Tris-HCl (pH 8.2) in 0.02 M KCl; Solution 2, 10% glycerol in 20 mM Tris-HCl (pH 8.2); Solution 3, 0.3 M KC1 and 10% glycerol in 20 mM Tris-HCl (pH 8.2). With 5 mM Tris-HCl buffer in 0.02 M KC1 the enzyme activity pattern had a broad distribution, appearin, 0' in almost all of the fractions (Fig. 3A).
Such a broad activity pattern suggests either that the enzyme is polydisperse or that the enzyme has a high affinity for other protein.
n'ine micrograms of the enzyme were treated and subjected to electrophoresis as described under "Experimental Procedure." Electrophoresis w'as carried out for 3 hours at room temperature. A single activity peak was observed upon the addition of 10% glycerol to the buffer (Fig. 3B). The addition of 0.3 M KC1 to the glycerol-Tris buffer produced a dramatic change in the gel filtration pattern (Fig. 3C). The specific activity was increased from that observed with 10% glycerol and buffer (Fig. 3B). At the same time, the peak activity migrated with the more slowly eluting fractions, indicating conversion of the enzyme to a smaller molecular weight species. Although 0.25 M sucrose gave essentially the same result as 10% glycerol, the specific activity was only slightly improved.
It should be noted that when the enzyme was isolated in the presence of 10% glycerol, gel filtration on a column equilibrated with 20 mM Tris-HCl (pH 8.2) gave exactly the same pattern as a column equilibrated with 10% glycerol in 20 mM Tris-HCl (pH 8.2). The gel filtration studies indicate: (a) the enzyme is associated with other proteins and can be dissociated in the presence of glycerol or sucrose, (b) the enzyme exists in an aggregated form which is disaggregated in the presence of a high salt concentration, (c) the disaggregated form is much more active than the aggregated form.
Effect of D&ion-Dilution experiments were performed to provide further information on the effect of the medium or microenvironment on the enzyme activity. in different media and then assayed immediately for activity. As shown in Table II, dilution of the enzyme in Tris-HCl buffer or water produced a decrease in specific activity.
Whereas, when the enzyme was diluted with a mixture containing KCl, glycerol, and Tris-HCl buffer, the enzyme specific activity was higher. Moreover, when the enzyme was diluted in a phospholipid dispersion, the specific activity was higher than that observed in the presence of the KCl, glycerol, and Tris-HCl buffer.
The enzyme activity could be partially inactivated by the removal of KC1 and reversibly reactivated in the presence of KClglycerol as shown in Fig. 4. A maximum reactivation was attained after 2 hours of incubation at 4'. An analysis following glycerol density gradient centrifugation determined that the reactivation was dependent upon the KC1 concentration and was accompanied with conversion of the enzyme to a more slowly sedimenting fraction (Fig. 5). These results suggest that the enzyme is spontaneously aggregated upon dilution in a salt-free buffer, and that the aggregated enzyme is reversibly disaggregated into a more active form at a slower rate. Moreover, after dilution in phospholipids, the enzyme seems to rapidly form an active enzyme-phospholipid complex having much higher activity.

Phospholipid
Binding-The requirement for phospholipid for enzymatic activity suggests that a specific binding of phospholipid to enzyme may well be necessary for formation of an active complex.
Evidence for complex formation was shown by the use of glycerol gradient centrifugation (Fig. 6). When the enzyme alone was placed on a gradient, the enzyme sedimented as a single symmetrical peak. A more rapidly sedimenting peak was found after the enzyme and phospholipids were mixed and then centrifuged.
In the presence of 0.5 M KCl, this rapidly sedimenting peak disappeared and the enzyme was detected at the same position as the enzyme alone. Because of the insolubility of vitamin K1, vitamin Kl-MTT were replaced with PMS-DCIP to detect activity without phospholipids (Fig. 6B). The activity profiles without added phospholipids indicated that substantial activity was detected in rapidly sedimenting frac-   tions, while slowly sedimenting fractions were essentially inactive in the absence of phospholipids. Both of the fractions showed higher activity after the addition of phospholipids.
The relative ratios of phospholipids to the enzyme changed the sedimentation pattern as shown in Fig. 7. At low phospholipids to enzyme ratios two activity peaks were observed (Fig.  7, D and E), while only slight changes in the activity pattern were observed with lower amounts of phospholipids (Fig. 7, B and C). Addition of excessively large amounts of phospholipids resulted in the sedimentation of almost all of the enzyme to &he bottom fractions (Fig. 719. The activity without the added phospholipids in the rapidly sedimenting fractions paralleled the activity with phospholipids.
Thus, no definite ratio of phospholipid to protein could be obtained for the formation of the complex.
il[olecular Weight-The molecular weight was determined by gel filtration with a calibrated column of Sephadex G-200 (with standard protein markers) to be 53,000 in the presence of KC1 (Fig. 8). Calculation of molecular weight from density gradient centrifugation in Fig. 6A gave a value of 60,800 in the presence of 0.5 M KCl.
With gel filtration in the absence of KCl, the 7491 molecular weight was calculated to be 164,000 (Fig. 9). As can be seen in Fig. 3B, the elution pattern was rather broad as compared to the one in the presence of KCl. These results are compatible if it is assumed that the enzyme may exist in different degrees of aggregation.
To determine the presence of subunit structure, the enzyme was subjected to gel electrophoresis in the presence of SDS. The enzyme was incubated in a solution of 1% SDS-l% mercaptoethanol and 4 M urea for 45 min at 45". The relative electrophoretic mobility of the treated enzyme as well as those of standard proteins were used for a plot of the log molecular weight versus mobility (Fig. 10). The molecular weight calculated from such a standard curve was 51,000. Treatment of the enzyme at 100" for 2 min in 1 y0 SDS, 1 y0 mercaptoethanol-10 mM phosphate (pH 7.0) or dialysis overnight at room temperature against the same solution did not change the 1 n * mobility of the enzyme. Thus it is highly unlikely that the enzyme is composed of subunits.
The higher activity in the presence of KC1 indicates that the monomeric form of the enzyme with a molecular weight of 53,000 has a free binding site (or sites) available for phospholipids. However, in the absence of KC1 the enzyme aggregates via phospholipid binding site (or sites), resulting in a decrease in the effective binding sites. The aggregated form appears to exist predominantly as a trimer. Absorption Spectrum and Amino Acid Compositions-The The procedure was the same as that in Fig. 6, except that the amount of the enzyme was 25.4 pg in each tube, and the amount of phospholipids in micrograms was varied : 0 (A), 5 (B) , 10 (C), 25 (ZI), 75 (E), and 250 (F) FIG. 8 ( Fro. 11. Absorption spectrum of malate-vitamin K reductase. The protein concentration was 1.08 mg per ml in 0.3 M KC1 and 10% glycerol in 20 mM Tris-HCI (pH 8.2). The reference cuvette contained 0.3 M KC1 and 10% glycerol in 20 mM Tris-HCI (pH 8.2). sorption spectrum exhibits a maximum at 280 nm with a shoulder at 290 nm (Fig. 11). Since the enzyme requires FAD and vitamin K for activity (24,35), the presences of these cofactors were examined with the purified enzyme.
No spectral evidence was found for the presence of a flavin or quinone with 1.08 mg of purified enzyme even with an expanded full scale of 0.01 optical density on an Aminco DW-2 spectrophotometer.
The results of amino acid analyses are shown in Table III. Because of the difficulty in obtaining large quantities of purified enzyme only one analysis was obtained.
It is of interest to note that tyrosine, phenylalanine, and half-cystine were not detected or were present in very small amounts, and that there was a relatively high content of nonpolar amino acids. Calculation of the polarity of the enzyme (36) gave a value of 42.7%, which is a value intermediate between that for soluble proteins (47 * 6%) and that for membrane-bound proteins (below 40%).

DISCUSSION
A considerable number of enzymes have been reported to have a phospholipid requirement for enzymatic activity, but only a few have been purified and characterized (1, 2). The protein phospholipid interaction can be considered as an important factor in understanding membrane structure and related functions (37-39).
Although natural membrane systems such as mitochondria, microsomes, erythrocytes, and bacteria, have been the subject of extensive research, a simple model system for membranes containing a protein and phospholipid may provide a tool for obtaining basic information about membrane-bound enzymes.

Malate-vitamin
K reductase has been characterized as an enzyme requiring phospholipids for enzymatic activity. Enzyme activity was not observed with the purified enzyme in the absence of phospholipids, while with less purified preparations, 3 to 5% of full activity was obtained without the addition of phospholipids.
The determination of phospholipid content in the fractions obtained from different steps of purification gave values of 166, 8.4, and 2.3 pg of phospholipid per mg of protein in ammonium sulfate, DEAE-cellulose, and Sephadex fractions, respectively.
Asolectin, whose major components are lecithin and phosphatidylethanolamine (40), has been shown to be the most effective phospholipid; individual lipids alone or cardiolipin are less effective (24). Differential centrifugation of sonicated cell extract was used to localize malate-vitamin K reductase. The enzyme activity was predominantly found in the supernatant fraction with some residual activity (resistant to 0.15 M KC1 washing) in the particulate fraction. But ghost preparations, obtained by the treatment of cells with lysozyme in the presence of glycine, contained more than 85% of the enzyme activity, which can then be released by sonic oscillation (41). These results suggest that the enzyme is loosely bound to the cytoplasmic membrane and is not a "soluble cytoplasmic" enzyme. According to the classification of a membrane protein by Singer and Nicolson (42), this enzyme may be classified as a "peripheral" protein.
Relevant to the nature of this association with the cytoplasmic membrane is the fact that the residual activity in the particulate fraction was completely released upon washing with either 10% glycerol or 0.25 M sucrose (data not shown).
A bacterial coupling factor containing latent adenosine triphosphatase activity is associated with the particulate fraction from M. phlei and was also solubilized by the use of sucrose washing (43). Such an effect of sucrose or glycerol on a membrane-associated enzyme was also shown with acetyl-CoA carboxylase (44) and squalene synthetase (45). It is interesting to note that acetyl-CoA carboxylase was found to be slightly stimulated by phospholipid and was postulated to be attached weakly to a microsomal membrane (44). Thus, another type of chemical interaction, different from electrostatic bonding, seems to be an important force in associating these enzymes with the membranes.
In the present studies, complex formation between malatevitamin K reductase and phospholipids has been successfully demonstrated.
The lipid-protein complex dissociates completely into the separate components in the presence of KCl.
It appears that electrostatic interaction is an important factor in the association between the enzyme and phospholipid.
However, the enzyme activity associated with the particulate fraction can be released only by glycerol or sucrose, and not by KCl.
This could be explained if the enzyme is not directly bound to phospholipid in the natural system, for example, if another protein (or proteins) is involved in the binding.
The effect of high KC1 concentration on enzyme activity showed another aspect of enzymes with a requirement for phospholipid binding.
The enzyme may undergo a KC'-dependent aggregation disaggregation phenomenon; the monomeric form (mol wt, 51,000 to 53,000) exists in the presence of KC1 and the polymeric form (mol wt, 164,000) in the absence of KCl.
Another important phenomenon is that the enzymatic activity is correlated with the aggregation state; that is, the monomeric form has more than twice as much activity as thepolymeric form. As shown in Fig. 6, the enzyme binds phospholipids, giving an enzyme-phospholipid complex which was dissociated with KCl. This was further verified by dilution experiments which showed that a higher specific activity was obtained upon dilution either in KCl-glycerol or in phospholipid. This provides a strong basis for the concept that the phospholipid binding site (or sites) and the aggregation site (or sites) are the same or are very closely located in the enzyme.
There have been many examples of the tendency of proteins isolated from membranes to aggregate. This aggregation can be prevented by lipids, detergents, alkaline pH, or by sonication.
Thus, membrane proteins seem to have a hydrophobic site (or sites) which interact with each other or with lipids (46)(47)(48).
SDS-gel electrophoresis, under various conditions, shows that the enzyme is a single polypeptide chain with a molecular weight of 51,000. All membrane bound enzymes do not exhibit a requirement for phospholipid.
It is thus of interest that although malate-vitamin K reductase is not a tightly bound membrane protein it requires phospholipid for activity. The presence of "hydrophobic segments" or "phospholipid-binding segments" was clearly shown in microsomal cytochrome b5 (46), glycoprotein from red blood cell membranes (49), and apo high density lipoprotein (50) in the single protein.
Such a submolecular differentiation in the membrane protein is discussed in connection with a lipid-globular protein mosaic model of membrane (38,51). In view of present concepts of the mosaic model of membrane structure (38,42,51), studies on the interaction of active enzymes with lipid bilayers and phospholipid vesicles (liposomes) will be important.
However, since only a few enzymes capable of interacting with phospholipids have even been isolated and purified, relatively few such studies have been performed. Glycosyltransferases (16,52), @hydroxybutyrate dehydrogenase (4, 5), cytochrome c (53,54), and adenosine triphosphatase from Streptococcus .fuecalis (55) are the most representative systems that have been studied.
Studies with such enzymes have shown that upon formation of a lipoprotein complex, changes in electrical 7493 conductance (55) and surface pressure (52) are observed.
Conformational changes in the protein were also suggested as a mechanism for protein phospholipid interaction (50,54).