MEMBRANE-BOUND D-LACTATE DEHYDROGENASE
Biochemical and Biophysical Studies on the Interaction of a Membrane-Bound Enzyme, D-Lactate Dehydrogenase from Escherichia coli, with Phospholipidst Leslie W.-M. Fung,g E. A. Pratt,' and Chien Ho*,*
ABSTRACT:Purified, detergent-free membrane-bound D-lactate dehydrogenase from Escherichia coli W 3 110trpA33 has been used for the study of lipid and protein interactions. DE52 column chromatography of enzyme mixed with 32P-labeled phospholipids of E. coli and electron microscopy of E. coli phospholipid vesicles in the presence and absence of enzyme have been used to demonstrate the association of lipid and protein. Enzymatic activity is increased fivefold by Triton X- 100 and phospholipids of E. coli, tenfold by lysolecithin. A very large lipid to protein ratio is required for maximum enhancement of activity at room temperature. Preincubation of enzyme and lipid for 10 min at 37 "Cincreases activity still further. The Michaelis-Menten constant, K,, of the purified enzyme is the same in aqueous solution and in Triton X- 100 but is increased by addition of E. coli phospholipids or ly-
solecithin. Magnetic resonance techniques have been used to investigate the structure and environment of the phospholipids of E. coli in the presence and absence of enzyme. When enzyme is added to E. coli phospholipids, the 3'P nuclear magnetic resonance of the lipids is shifted downfield by about 2.8 parts per million and the intensity increases by an order of magnitude. The 'H nuclear magnetic resonance signals of E. coli phospholipids appear to be better resolved in the presence of enzyme than in its absence, with the methyl and methylene signals each shifting slightly upfield and resolving into two signals. These results suggest that there are interactions between phospholipids and D-lactate dehydrogenase and these lipid-protein interactions are essential for optimal enzymatic activity.
S t u d i e s of isolated cytoplasmic membrane vesicles of Escherichia coli strain M L 308-225 have shown that the transport systems of many amino acids and sugars are coupled to the oxidation of D-lactate to pyruvate, catalyzed by the membrane-bound enzyme, D-lactate dehydrogenase (D-LDH)' (Kaback, 1974). D-LDH of E. coli has been purified and characterized for strains M L 308-225 (Kohn & Kaback, 1973; Futai, 1973) and W3110trpA33 (Pratt et al., 1979). It has been shown that the purified D-LDH from strain M L 308-225 can be incorporated into membrane vesicles of a mutant with defective D-LDH (strain M L 308-225 dld-3) to restore Dlactate dependent respiration and transport processes (Reeves et al., 1973; Short et al., 1974; Futai, 1974). This finding suggests the capability of purified D-LDH to interact with other membrane proteins and lipids in an active form. In fact, the enzymatic activity of D-LDH is enhanced by the addition of phospholipids. W e have studied the effects of total phospholipids extracted from E. coli, as well as single-component synthetic phospholipids, on the enzymatic activity of the purified, detergent-free D-LDH of E . coli strain W 3 1 1OtrpA33. Kinetic parameters of the dehydrogenasecatalyzed reaction have also been measured. A preliminary biochemical investigation of the D-LDH of E. coli strain M L 308-225 has been reported by Tanaka et al. (1976). Our biochemical and biophysical studies on this enzyme confirm and extend their results.
The functional significance of each membrane component in active transport processes is not clear. In particular, how lipid molecules influence the action of membrane-bound enzymes remains to be understood. Many recent studies represent attempts to understand lipid-protein interaction on a molecular basis (for example, see Coleman, 1973; Lucy, 1974; Cronan & Gelmann, 1975; Gulik-Krzywicki, 1975; Lee, 1975; Linden & Fox, 1975; Masoro, 1977; Korenbrot, 1977; Gennis & Jonas, 1977; Gent and Ho, 1978). The lipid-protein interactions appear to be both complex and heterogeneous. For example, the enzymatic activity of glycerol-3-phosphate dehydrogenase is not affected by manipulating the unsaturated fatty acid content of fatty acid auxotrophs of E. coli, unlike glycerol 3-phosphate and 1-acylglycerol-3-phosphateacyltransferases (Mavis & Vagelos, 1972). In a delipidated membrane system, addition of total cell lipids restores the activity of succinic-ubiquinone reductase of E. coli to normal, while neither phospholipids nor neutral lipids alone can do so (Esfahani et al., 1972). The activity of purified, detergent-free D-LDH of the cytoplasmic membrane of E. coli is enhanced by lipid or lipid-like molecules, and yet no specificity for a particular kind of lipid is indicated (Tanaka et al., 1976; present work). These and many similar results imply that lipid-protein interactions are important for both the organization and regulation of enzymatic activities in membranes and suggest that studies of lipid-protein interactions a t a molecular level are necessary to understand the reaction mechanisms involved. The elucidation of the mechanism of membrane function relies on the precise molecular understanding of membrane components and of their association.
From the Department of Biological Sciences, Faculty of Arts and Sciences, University of Pittsburgh, Pittsburgh, Pennsylvania 15260. Received April 18,1978. Supported by research grants from the National Institutes of Health (GM-18698 and RR-00292) and the National Science Foundation (PCM 76-21469). L.W.-M.F. was a recipient of a National Research Service Award, National Institutes of Health (GM-05164), during 1975-1977. This paper was presented in part at the 21st Annual Meeting of the Biophysical Society, Feb 15-18, 1977, New Orleans, Louisiana, and at the 61st Annual Meeting of the Federation.of American Societies for Experimental Biology, April 3-8, 1977, Chicago, Illinois. $Present address: Department of Chemistry, Wayne State University, Detroit, Michigan 48202. 'Present address: Department of Biological Sciences, Mellon Institute of Science, Carnegie-Mellon University, Pittsburgh, Pennsylvania 15213.
0006-2960/79/0418-03 17$01 .OO/O
I Abbreviations used: D-LDH, D-lactate dehydrogenase; PL, E . coli phospholipid; PE, phosphatidylethanolamine; PG, phosphatidylglycerol; D M P C , dimyristoylphosphatidylcholine; DPPC, dipalmitoylphosphatidylcholine;DMPE, dimyristoylphosphatidylethanolamine;DPPE, dipalmitoylphosphatidylethanolamine;LL, lysolecithin (mixture); IysoPC, lysophosphatidylcholine (pure); IysoPE, lysophosphatidylethanolamine; L/P, lipid to protein ratio; TX, Triton X-100; cmc, critical micelle concentration.
0 1979 American Chemical Society
As an approach to these problems, many lipid model systems have been studied by magnetic resonance techniques (for example, see Lee, 1975; Cronan & Gelmann, 1975; Gent & Ho, 1978, and references therein). From these studies, a large amount of knowledge has been developed on the structure, motion, and physical properties of the lipid molecules in model systems. Further, magnetic resonance techniques have been used for providing molecular information on biological membrane systems. However, only a relatively few lipidprotein systems have been examined by these techniques (for example, see Jost et al., 1973; Stier & Sackmann, 1973; Warren et al., 1974a; Dehlinger et al., 1974; Jost et al., 1977). These studies, mainly by electron paramagnetic resonance spectroscopy employing spin labels, have shown that there is a layer of boundary or annulus lipid molecules tightly coupled to the membrane proteins. In the cytochrome oxidase system, a small fraction of phospholipid molecules, about 0.17 to 0.21 mg of lipids per mg of protein, is immobilized by cytochrome oxidase molecules (Jost et a]., 1977). I n the sarcoplasmic reticulum membrane, 0.15 mg of phospholipids per mg of protein is immobilized (Nakamura & Ohnishi, 1975). A much larger amount of boundary lipids was observed for proteolipid apoprotein of bovine myelin (Curatolo et al., 1977). More recently, Gent & Ho ( I 978) have observed boundary lipids in E . coli membranes by I9F nuclear magnetic resonance ( N M R ) . These findings imply that the boundary lipids have intimate interactions with membrane proteins and may modulate the activities of membrane-bound enzymes. In biochemical studies, where the activities of lipid-free enzymes have been monitored as a function of added lipid concentrations, a relatively large amount of lipid is required to achieve maximum enhancement of activity. For example, a 20-fold excess by weight of sarcoplasmic reticulum lipid to protein was needed to obtain a steady-state level of calcium ion accumulation (Warren et al., 1974b). More than 17-fold excess by weight of phosphatidylserine to protein was used to restore the respiratory control of delipidated cytochrome oxidase (Eytan et al., 1976). A large amount of excess phospholipids is required to enhance the activity of D-LDH (Tanaka et al., 1976; present work). It thus appears that the functions of membrane proteins are influenced not solely by the boundary lipid molecules but also by the bulk lipid molecules present in the system. These bilayer or bulk lipid molecules have been shown to be in equilibrium with the boundary lipids (Jost et al.. 1977; Gent & Ho. 1978). However, little molecular information is available on the bilayer lipid molecules in lipid-protein systems. We have focused our attention on part of these lipid molecules and studied some of their molecular properties. ‘H and 31PN M R techniques have been used to investigate the interactions between E . coli phospholipids and the membrane-bound enzyme of E . coli, D-laCtate dehydrogenase. Experimental Section D-hCtate Dehydrogenase. E. coli W3110trpA33 was grown to exponential phase aerobically as described in the preceding paper (Pratt et al., 1979). D-LDH was isolated and purified in the presence of Triton X-100 (Rohm and Haas), which was later removed by acetone precipitation to give D-LDH aggregates in aqueous buffer solution (Pratt et al., 1979). The detergent can also be removed by anion-exchange (Whatman DE52) column chromatography. However, the method of acetone precipitation was used throughout the investigation due to its simplicity and reproducibility. N o detectable Triton, as determined by measuring the concentration of Tritonammonium-cobalt-thiocyanate complex in dichloroethylene
FLYG, PRATT, AiiD HO
spectrophotometrically (Garewal, 1973), was observed in the final preparation. D-LDH enzymatic activity was measured by the increase in absorbance a t 570 nm of dimethylthiazolyldiphenyltetrazolium bromide being reduced by DLDH and D-lactate in the presence of phenazine methosulfate. Detailed assay procedures were described previously (Pratt et al., 1979). The purified enzyme was stored at 4 O C in 0.05 M phosphate buffer at pH 7.2. E . coli Phospholipids. E . coli W3110trpA33 was grown at 37 “ C to exponential phase as described in the above section. Since the E. coli phospholipid (PL) composition and the fatty acid chains of the phospholipids can be modified by growth conditions (Marr & Ingraham, 1962; Cronan, 1968; Linden & Fox, 1975; Okuyama et al., 1977), all conditions, including optical densities at the times of inoculation and harvest, as well as the temperature during growth, were kept constant throughout the course of the investigation. Phospholipids of E . coli were extracted from fresh cells by the modified Bligh and Dyer procedures (Bligh & Dyer, 1959; Kates, 1972). Thin-layer chromatography (TLC) on silica gel (type 0 of New England Nuclear) with chloroform-methanol-water (65:25:4) as solvent (Ames, 1968) was performed routinely during lipid preparation. The fatty acids in the phospholipids were determined as their methyl esters by gas chromatography (Perkin-Elmer Model 900). The proportions of each phospholipid and the fatty acid content are in good agreement with the published results for E . coli K12 (Ames, 1968; Fox, 1972), i.e., about 70-80% phosphatidylethanolamine (PE), 10-1 5% phosphatidylglycerol (PG), and 5-100h cardiolipin. The major fatty acids are palmitic (16:0), palmitoleic (16:1), and vaccenic (18:l) acids. We would like to emphasize the fact that, in this work, the enzyme and the phospholipids were purified from the same source, E . coli W3110trpA33, grown to exponential phase. 32P-Labeled Phospholipids. 32P,5 mCi (New England Nuclear), in a liter of low phosphorus medium (Hershey & Chase, 1952) containing M tryptophan was used for the growth of E . coli W3110trpA33. 32P-labeled phospholipids were isolated from labeled cells as described above. 32Pradioactivity was counted in a Packard liquid scintillation spectrometer (Model 3380) with Liquifluor in toluene and Triton X- 100 (2:l by volume) as the scintillation “cocktail”. Single-Component Phospholipids. Synthetic L-a-dimyristoylphosphatidylcholine (DMPC), DL-a-dipalmitoylphosphatidylcholine (DPPC), L-a-dimyristoylphosphatidylethanolamine (DMPE), and L-a-dipalmitoylphosphatidylethanolamine (DPPE), L-a-phosphatidyl-DL-glycerol (PG), cardiolipin, egg lecithin, lysolecithin (mixture) (LL), ~ - a lysophosphatidylcholine (pure) (IysoPC), and ~ - a - l y s o phosphatidylethanolamine (IysoPE) were obtained from commercial sources (Applied Science Lab. Inc., Calbiochem, and Sigma) and used without further purification. Samples checked by TLC showed a single spot, suggesting at least 95% purity. Phospholipid Vesicles. EDTA, 1 mM, 0.15 M NaCI, and 10 mg of PL, DMPE, or DPPE were added to 1 mL of freshly prepared 0.1 M sodium carbonate buffer a t p H 11 (titrated with N a O H ) . Tris or phosphate buffer at p H 7.1 was used instead of the carbonate buffer for other lipid systems. The mixture was vortexed and sonicated to give a pale blue solution of lipid vesicles. The sonication was performed shortly before the assay experiments, For N M R studies, D 2 0 was used instead of H 2 0 . Lysophospholipids were not sonicated, as micelles of lysophospholipids form on vortexing. In some experiments, sonication was performed under nitrogen. There
VOL. 18, N O . 2, 1979
MEMBRANE-BOUND D-LACTATE DEHYDROGENASE
was no difference in experimental results in the presence or absence of nitrogen during sonication. Column Chromatography with Anion Exchanger. DE52 resin was equilibrated with column buffer (0.05 M potassium phosphate buffer a t pH 7.1). D-LDH (0.05 mg) was added to 1 m L of sonicated 32P-labeled phospholipid solution containing 1 mg of phospholipids. This solution containing lipid-protein complexes was loaded onto a 0.5 X 8 cm DE52 column. A stepwise elution procedure was used: washing with several column volumes of buffer solution followed by elution with the same buffer solution containing 0.3, 0.325, 0.35, or 0.4 M NaC1. Fractions were collected when elution with salt solution was started. I n the control experiment, D-LDH was absent. All procedures were carried out at 23 "C. Phospholipid-Enzyme Complexes. We have used the same method for associating lipid and D-LDH molecules to form complexes for both biochemical and biophysical studies in order to be able to correlate the findings of the two different approaches. The method essentially involves mixing detergent-free D-LDH with sonicated lipid solution a t room temperature (Fleischer et al., 1966). For the N M R studies, an 8-mg D-LDH sample was evaporated to dryness under a gentle stream of nitrogen gas. The freshly sonicated lipid solution containing 8 mg of lipid in D 2 0 buffer was added to the dry film of D-LDH. Gentle shaking allowed D-LDH to dissolve and resulted in a clear yellowish solution. All samples were prepared immediately before the N M R experiments. Enzymatic activity of D-LDH was assayed and lipid composition was checked by TLC before and after the experiments. D-LDH was fully active after the N M R experiments and no decomposition of the phospholipids was seen. Electron Microscopy. Electron micrographs were obtained using a Phillips electron microscope (Model 300). Complexes were prepared as described in the previous section, with the D-LDH to lipid weight ratio 1:l or molar ratio 1:lOO. Specimens were prepared by agar filtration and negatively stained with uranyl acetate at p H 3.5. NMR Experiments. 'H and 31PN M R spectra were obtained on the M P C - H F 250 superconducting spectrometer with operating frequencies at 250 and 101.2 MHz, respectively. The ambient temperature of the probe was 27 "C. The 31P N M R spectra were obtained without proton decoupling. Usually, the experiments were carried out in the following sequence: first 31PN M R and then 'H N M R spectra were measured using the same lipid sample. After the 'H N M R , the lipid sample was immediately resonicated for 10 min before being added to the dry film of D-LDH prepared as described earlier. 'H and 31PN M R experiments were then performed on the freshly prepared lipid-enzyme complexes. The time required for changing the spectrometer from one nucleus to the other was about 10 min. Multiple scans were accumulated for each spectrum and the signal-to-noise ratio was further enhanced by the N M R correlation technique (Dadok & Sprecher, 1974). The proton chemical shifts are referenced with respect to the residual water proton signal in the samples and the 31Pchemical shifts are referenced with respect to the 31Presonance of 85% H3P04. The N M R measurements were repeated several hours later and similar results were obtained, indicating that the lipid and enzyme molecules were in equilibrium during the N M R experiments. Results Lipid-Protein Complexes. W e have used column chromatography with an anion exchanger to demonstrate the association of lipid molecules with purified D-LDH. D-LDH is absorbed on a DE52 column and can be eluted with a 0.25
No D-LDH W i t h D-LDH
S O ._
1: Elution patterns of 32P-labeled E . coli phospholipids and D-LDH from a DE52 column with 0.05 M potassium phosphate buffer at pH 7.1 containing 0.4 M NaC1. FIGURE
Table I : Effect of E. coli Lipid on D-LDH Activity-Concentration, Preincubation, and Dilution preincubation mixturea ~
1000 1000 1000 100 100 10 10 10
protein pg of lipid
6 6 6 6 6 6 6 6
1000 200 20 2 0 100 20 4 10 2 2 1 0.2
0.12 0.12 0.12 0.12 0.12 0.6 0.12 0.024 0.6 0.12 1.2 0.6 0.12
lipid/ of protein
sp act. (units/ mg)
8333 1667 167 17 0 167 167 167 17 17 1.7 1.7 1.7
220 160 85 48 36 24 1 152 111 109 63 66 62 32
pg of pg
10 50 250 10 50
5 10 50
The indicated amounts of E. coli phospholipid and D-LDH were mixed and incubated at 37 "C for 10 min. The mixtures were then diluted as indicated into the assay solution and assayed immediatelv at 23 "C.
M NaCl solution (Pratt et al., 1979). Lipid molecules alone cannot be eluted even with a 0.4 M NaCl buffer solution (Figure 1). However, in the presence of D-LDH, some lipids can be eluted with buffer solution containing 0.4 M NaC1. At lower NaCl concentrations, results were inconsistent, but a t 0.4 M NaC1, we have consistently obtained an elution pattern as shown in Figure 1, with about a 4:l weight ratio of lipid to protein recovered. Lipid Activation. Table I presents the effects of phospholipids of E. coli W3110trpA33 on the enzymatic activity of D-LDH under different experimental conditions. Different lipid concentrations give different degrees of activation. When we compare the assay where the lipid to protein weight ratio (L/P) is over 8000 with that where no lipid is present, we find a sixfold increase in activity. Decreased L / P gives decreased activation. For example, when L/P is 1667, 167, or 17, the activation is 4.4, 2.4, and 1.3, respectively. Table I also shows that preincubation of lipids and protein at 37 "C for 10 min increases activity over that of the same ratio of lipid to protein without preincubation. Prolonging the incubation did not
B I oc H E M I sT R Y
F L h G . PRATT, AND HO
Table 11: Effect of Lipid on D-LDH Activity-Preincubation Followed by Addition of Lipid 300
of kg of lipid/ lipid protein protein
0 100 100
0.12 0.12 6 6 6
0 0 17 17 1.7
sp act. lipid/ (units/ protein protein me)
1000 200 1000 200 200
0.12 0.12 0.12 0.12 0.12
8333 1667 8333 1667 1667
354 218 372 316 211
' Thc indicated amounts of E. coli phospholipid and D-LDH wcrc mixed and incubated at 37 "C for 10 min. The mixtures were then diluted into the assay solution, containing additional phospholipid, and assayed immediately at 23 "C.
L i p i d C o n c e n t r a t i o n , gm/ml
2: Effect of concentration of lipids and lipid-like molecules on activity of D-LDH. Lipids and Triton at the concentrations indicated were added to 0.2 kg of D-LDH in the assay solution. Activity FIGLRE
was assayed immediately at 23 'C.
Table 111: Activation of D-LDH by Lipids and Lipid-Like Molecules
DPPE DPPC PG cardiolipin egg lecithin PL DMPC DMPE 1ysoPE 1ysoPC LL TX
1 2 4 4 4 4 4
after incubation of lipid at 60 " C b
Lipids at a concentration of 0.1% or Triton X-100 at a concentration of 0.6% were mixed with 0.2 pg of D-LDH in the assay solution and the activity was immediately assayed at 23 "C. Lipids were equilibrated for 30 min at 60 "C and then added to 0.2 pg of D-LDH in the assay solution.
increase the activity further, as shown also by Tanaka et al. (1976). However, dilution of the preincubated lipid-protein mixture before assay decreases this effect. As shown in Table 11, when enzyme is preincubated either alone or with relatively small amounts of lipid, additional lipid added at the time of assay increases activity beyond that of the same ratio of lipid to protein without preincubation. We have also studied the activation of the enzyme with several different pure synthetic phospholipids. Table 111 summarizes the different degrees of activation of D-LDH activity by these lipid and lipid-like molecules. At room temperature, 0.1% DPPE and DPPC have little effect when the activities are compared with those in buffer. However, 0.1% PG, cardiolipin, egg lecithin, DMPC, and D M P E give four- to fivefold increases in D-LDH activity. Triton molecules (0.6%) show similar activation. Lysophosphatides give larger effects with a ninefold activation for 0.1% LL. The D-lactate concentration was 10 m M in these experiments. When the lipid solutions were equilibrated to 60 OC before assaying, a 2.8-fold increase was observed for DPPE and 3.5-fold for DPPC. However, little change was observed for P L of E . coli and DMPE. W e have also studied the lipid activation as a function of lipid concentration of several lipid and lipid-like molecules. As expected, increase in concentration gives increase in activation, as shown in Figure 2. A Hill plot [Le., log (V/(Vma, - V ) ) vs. log lipid concentration (L)] demonstrates linearity (Figure 3), with slopes (Hill coefficient) close to two for PL ( I .8), DPPC (2.2), and LL (1.8). With Triton X-100 mol-
FlGtiRE 3: Double log plots of [ V / ( V,,, - 131 vs. [L], where Vis the activity with maximum at V,,, and L is the concentration of TX, LL, DPPC, or PL in g / m L of 1ipid:o-LDH systems.
Table IV: Kinetic Parameters of D L D H in the Presence and Absence of Liaid and Lipid-Like Moleculesa
TX PL LL
250 25 0 5 00
0.42 0.67 1.43
x 10-~ x 104 x
a PL and LL at a concentration of 0.1% and Triton X-100 at a concentration of 0.6% were mixed with 0.2 j.q of D-LDH in the assay solution. Activity was assayed immediately at 23 "C.
ecules, the slope equals 1.13 above the critical micelle concentration (cmc, 0.01%; Rohm & Haas, 1976) and 3.45 below the cmc. Kinetic Parameters of D-LDH. Double reciprocal (Lineweaver-Burk) plots of activity ( V ) vs. D-lactate concentration (S) in the presence of E . coli PL, LL, TX, and buffer alone were obtained (data not shown). Table IV shows the values of the Michaelis-Menten constant (K,) and maximum velocity (V,,,) obtained from these plots. V,,, values increase fivefold in T X and P L solution; LL shows a tenfold increase. The K , values are also different: 0.42 m M in H 2 0 and TX, 0.67 m M in PL, and 1.43 m M in LL. The concentration of T X was 0.58% and of both P L and L L 0.1%. Electron Microscopy. Figure 4 shows electron micrographs of sonicated phospholipids of E. coli in the presence and
MEMBRANE-BOUND D-LACTATE DEHYDROCENASI
1 8 , N O , 2,
H z f r o m HDO
The 250-MHz 'H NMR spectra of sonicated phospholipids of E . coli with and without D-LDH in D20with 0.15 M NaCl and I mM EDTA at pH 11 and 21 "C. Lipid to protein ratio is I:]by weight. FIGURE 5:
FIGURE 6 : The 101.2-MHz "P NMR spectra of sonicated phospholipids in DiO with 0.15 M NaCl and I mM EDTA at 27 T.
FIGURE 4; Elcctron micrograph of sonicated phospholipids of E . coli in H 2 0 with 0.15 M NaCl and I m M EDTA at pH I I. (a) Without n-LDH and (b) with D-LDH.
absence of D-LDH ( L / P = I ) . Sonicated phospholipids show large lipid aggregates (Figure 4a). However, these aggregates disappear in the presence of D-LDH (Figure 4b). N M R Specrroscopy. High-resolution 250-MHz 'H N M R spectra were obtained on the sonicated E. coli phospholipid dispersions (Figure 5 ) . indicating the existence of sharp and well-resolved signals, similar to those of the well-characterized 'H N M R spectra of D M P C (Chapman et al., 1968). The sonicated E. coli phospholipid dispersions also gave highresolution 101.2-MHz 3'P N M R spectra (Figure 6). E. coli phospholipids, with PE as the major component (Ames, 1968;
F I G U R E 7: The 101.2-MHr "P NMR spectra of sonicated phospholipids of E . coli with and without D-LDH in D 2 0 with 0.15 M NaCl and 1 mM EDTA at pH I I and 21 "C. Lipid to protein ratio is I:l by weight. The signal to noise ratio in the spectrum of PL plus D-LDHis an order of magnitude greater than that without the protein.
Fox, 1972: present work), have a ) ] P chemical shift similar to that of D M P E but different from that of D M P C (Figure 6). Since the head groups of the two molecules, D M P E and DMPC, are not the same, different )IP N M R chemical shifts would be expected. In the spectra of sonicated E. coli phospholipid dispersions, it appears that we have observed only a small fraction of the lipid molecules in the sample. Quite a large fraction of the E . coli phospholipid molecules is probably still in the form of large aggregates, which will give very broad signals. We have prepared lipid dispersions without sonication and obtained, with different spectrometer settings, a broad, asymmetric powder-type spectrum (results not shown).
When D-LDH was added to sonicated E . coli phospholipid samples, marked changes were observed in the 31PN M R spectra (Figure 7). The j’P resonance is shifted downfield by about 2.8 ppm and the intensity increased by an order of magnitude. We have not done intensity calibration measurements on these samples and therefore the data will be discussed only in qualitative terms. I n the ’H N M R spectra of this lipid-protein system, the methyl signal at 3.9 ppm and the methylene at 3.5 ppm upfield from water appear to shift upfield slightly and each resolves into two signals upon addition of D-LDH (Figure 5). I n general, the proton N M R signals of E . coli phospholipid in the presence of D-LDH appear to be better resolved than those in the absence of D-LDH. N o spectral changes in either ” P or ‘HN M R spectra were observed when we substituted a nonmembrane protein, human adult carbonmonoxyhemoglobin, for D-LDH. Discussion Many studies of the effects of lipids on membrane-associated functions of E. coli utilize unsaturated fatty acid auxotrophs to monitor enzymatic activities either in whole cells or in isolated membranes [for example, see review of Fox (1972) and of Cronan & Gelmann (19791. Another approach in studying the lipid effect is to reconstitute delipidated membranes by adding purified lipid [for example, see Razin (1 972)], The results of these experiments all indicate that lipids play an important role in the functioning of membrane enzymes. The allotopic phenomenon (Racker, 1967) is often observed when results with purified enzymes are compared with those of the same enzymes in intact membranes, perhaps indicating the importance of molecular spatial arrangement to enzymatic activities or perhaps arising from intermembrane-component interactions which regulate the enzymatic activity. In order to distinguish these effects, a direct approach is to determine the effects of lipids on purified membranebound enzymes. D-LDH of E . coli strain M L 308-225 has been isolated and its enzymatic activity assayed in the presence of detergent and of phospholipids (Tanaka et al., 1976). We have used phospholipids extracted from E. coli as well as pure synthetic phospholipids to study the lipid effect on the enzymatic activity of purified, detergent-free D-LDH from a different strain of E . coli, W3110trpA33. The removal of detergent is essential, as it is likely that detergent molecules compete with lipid molecules at the binding sites of the enzyme. Lipid-Protein Complexes. Our results show that aggregates of detergent-free D-LDH interact with lipid or lipid-like molecules to form complexes in an aqueous medium. It is difficult to define the exact number of lipid molecules associated with one enzyme molecule, especially with the present semiquantitative data. Our biophysical studies indicate definite interaction between lipid and enzyme at a molar ratio of 100 to 1 (weight ratio of 1 to 1). At this ratio phospholipid molecules isolated from E. coli demonstrate a new magnetic environment in the presence of D-LDH. The biophysical studies were done a t rather high p H in order to obtain lipid vesicles; however, neither the lipid nor the protein molecules were affected, as described in the Experimental Section. The electron micrographs as well as the magnetic resonance spectra show that the enzyme “solubilizes” the large lipid aggregates in aqueous solution. This has also been shown by column chromatography, where our results indicate that the complexes, comprising up to four times more phospholipid by weight than enzyme, absorb onto the anion-exchange resins and can be eluted by high ionic strength buffer solution (Figure 1). This result also indicates the involvement of electrostatic interaction
A N D HO
in the lipid-protein association. The enzymatic activity assays indicate a much larger amount of lipid, several orders of magnitude larger than used for the biophysical studies, is required to fully enhance the activity of D-LDH. Enhancement of D-LDH Activity by Phospholipids. Our present data, and those of Tanaka et al. (1976), clearly demonstrate that E. coli PL, as well as several different kinds of lipid and lipid-like molecules, can enhance the enzymatic activity of D-LDH. The mechanism of activation and the nature of the lipid-protein interactions appear to be quite complex. Even though different kinds of lipid molecules enhance the activity to different extents, the enhancement of D-LDH activity does not seem to be lipid specific. For example, PG, egg lecithin, and D M P E have quite different chemical compositions, and yet they have a similar enhancing effect on the activity of D-LDH (Table 111). Triton molecules do not have a phosphate head group and lysoPE has only one acyl chain, but they both enhance the activity to a similar extent. Therefore, different degrees of activation do not seem to arise from the structure of lipid molecules, with a specific requirement for particular alkyl chains or lipid head groups, but from their physical state. The observation that lysophosphatides give the highest activation suggests that the fluidity of the phosphatides may influence their potential for stimulating D-LDH activity. Lysophosphatides have demonstrated their unique ability for solubilization of membrane particles (Rydstrom, 1976; Peterson & Deamer, 1977). LysoPE was also found to give the highest stimulatory activity for pyruvate oxidase (Cunningham & Hager, 1971a,b). The concept that fluidity of phosphatides influences their ability to enhance enzymatic activity is further substantiated by the results of the experiments in which lipids which have phase transition temperatures above room temperature, such as DPPE and DPPC, require preincubation a t 60 “ C to fully enhance enzyme activity. Similarly Jonas et al. (1 977) have reported interaction of human or bovine A-I apolipoprotein with DMPC only at temperatures above the phase transition temperature. We have found that even a t a constant lipid-protein ratio, dilution of the lipid-protein complexes results in a decrease in enzymatic activity, as shown in Table I. Since the lipidprotein complexes are in equilibrium with lipid and protein molecules in solution, upon dilution the equilibrium shifts away from the formation of the complexes and thus lower enzymatic activity is obtained. The purified, delipidated D-LDH molecules form aggregates in aqueous solution and possibly exist in a structure with constraints due to various noncovalent interactions. The enzyme molecules in this “constrained” structure bind D-lactate in a manner different from enzyme associated with phospholipid molecules. It is believed that the phospholipid activation reflects a conformational change in the enzyme molecules resulting in a return to their native conformation, as found in a membranous type of environment. In order to investigate further the mode of action of lipid molecules on the enzymatic activity of D-LDH, we have studied the kinetic behavior of the interaction of D-lactate with D-LDH in the presence and absence of lipid and lipid-like molecules. These systems do not seem to follow enzyme action as described by Michaelis-Menten kinetics. However, the data can be described by the double logarithmic form of the Hill equation (a measure of deviation from standard MichaelisMenten kinetics) as indicated in Figure 3, with similar slopes for the plots of PL, DPPC, and LL. The plot for Triton X-100 is different, and there is also a lower K, value in the presence
MEMBRANE-BOUND D-LACTATE DEHYDROGENASE
of Triton (Table IV). Therefore, even though TX and PL give similar V,,, values, the molecular interaction of T X with D-LDH is probably different from that of phospholipids. Our data are similar to those of Cunningham & Hager (1971b) on pyruvate oxidase. They have suggested that phospholipids act as allosteric effectors for pyruvate oxidase. A separate theoretical treatment has also been attempted to interpret the nonhyperbolic behavior of the interaction of enzymes with soluble, amphiphilic lipids by use of separate kinetic parameters for monomers and micelles (Gatt & Bartfai, 1977a,b). N M R Studies. W e have used N M R spectroscopy to study the interaction between D-LDH and the total phospholipids isolated from E . coli a t a molar ratio of approximately 100 to 1, We have compared several molecular properties of the phospholipid molecules in the presence and absence of protein. The major E . coli phospholipid is PE. 31PN M R studies on lipid crystals and lamellar dispersions of PE have shown broad asymmetric powder spectra (McLaughlin et al., 1975; Kohler & Klein, 1976; Seelig & Gally, 1976; Kohler & Klein, 1977) which we have also seen with dispersed P L of E . coli (unpublished results). The asymmetric spectra suggest that lipid molecules experience relatively rapid rotation about the long axis of the molecules but have restricted motion about axes perpendicular to this axis of rotation (McLaughlin et al., 1975). N M R studies of sonicated PE samples at various pH values have suggested that the charge of the primary amine of the head group has a marked effect on the packing and mobility of the PE molecules (Michaelson et al., 1974). At neutral pH, sonicated P E molecules in aqueous sample probably have molecular packing similar to that in lamellar dispersions. Multilamellar systems with large lipid aggregates give restricted motion and orientation for each individual molecule. However, a t high pH, when the P E is negatively charged and the electrostatic attraction is reduced, the molecular packing becomes looser than at neutral pH. Sonication a t p H 1 1 produces heterogeneous lipid aggregates, among which are some small vesicles. Our electron micrograph (Figure 4a) shows that indeed the sonicated E. coli phospholipid sample at p H 1 1 consists of small vesicles as well as large multilayer liposomes (aggregates) of heterogeneous sizes and shapes. The high-resolution N M R spectra of sonicated phospholipids of E. coli or P E at pH 11 are similar to those of sonicated PC molecules a t neutral pH, which are known to be associated with small vesicles. They have sharp resonances for the terminal methyl protons and the chain methylene protons of the fatty acid chains in 'H N M R and the phosphate head group in 3'P NMR, each with specific chemical shifts (Chapman et al., 1968; Lee et al., 1974; McLaughlin et al., 1975; Bergelson & Barsukov, 1977). Furthermore, the isotropic motion due to the tumbling of vesicles will also give motional narrowing in the line width (McLaughlin et al., 1975). Upon addition of D-LDH to the sonicated lipid solution at a concentration of 1 protein molecule to 100 lipid molecules, the sharp "P N M R signal shifts downfield and the intensity increases without change in the line width (Figure 7). The increase in intensity of the 31Psignal indicates that many more, if not most, of the lipid molecules, under the influence of the newly introduced lipid-protein interaction, have head-group mobility fast enough on the N M R time scale to give motional narrowing, similar to that observed in small vesicles. The 2.8-ppm downfield chemical shift indicates a specific lipidprotein interaction. Our data suggest that most of the head groups are in a new environment and/or new orientation with
NO. 2, 1979
relatively large motional freedom. The lipid-protein interaction is strong enough for the PE molecules to overcome the electrostatic and hydrophobic interactions that cause them to pack tightly in multilayer form in aqueous solution in the absence of D-LDH. The sharp resonances in the 'H N M R spectrum (Figure 5 ) indicate relatively unrestricted motion for the hydrocarbon chains. Thus, upon addition of D-LDH to E . coli phospholipids, we observe changes in both 31Pand 'H resonances of the lipids. D-LDH acts like a detergent to reduce the size of phospholipid aggregates. The smaller aggregates would be expected to produce the observed alterations in 31Pand 'H N M R spectra. Our electron microscopy results also suggest the disappearance of large aggregates in the D-LDH-phospholipid system (Figure 4b). Our present results are similar to those reported by Novosad et al. (1976) on the effect of a major apoprotein constituent of human very low density lipoprotein, apoC-111, on DMPC vesicles. They found that when apoC-I11 interacts with DMPC vesicles, the particles are smaller than the original vesicles. In conclusion, our results show greater mobility of the lipid molecules in the lipid-D-LDH (1:l by weight or 1:lOO molar ratio) system than with lipid alone and suggest that in the presence of protein molecules the lipid-lipid and lipid-water interactions are being modulated by lipid-protein interactions. When associated with the protein molecules, the lipid molecules are probably in a more favorable environment than when they are in aqueous solution. These results imply that membrane components, lipid and protein, will interact spontaneously to minimize the free energy of the system and to increase the enzymatic activity of D-laCtate dehydrogenase. Acknowledgments W e wish to thank Judith A. Flowers for her excellent technical assistance. We are indebted to Dr. Sam C. M. To for taking the electron micrographs. References Ames, G. F.-L. (1968) J . Bacteriol. 95, 833. Bergelson, L. D., & Barsukov, L. I. (1977) Science 197, 224. Bligh, E. G., & Dyer, W. J. (1959) Can. J . Biochem. Physiol. 37, 911. Chapman, D., Fluck, D. J., Penkett, S. A., & Shipley, G. G. (1968) Biochim. Biophys. Acta 163, 255. Coleman, R. (1973) Biochim. Biophys. Acta 300, 1. Cronan, J. E. (1968) J . Bacteriol. 95, 2054. Cronan, J. E., & Gelmann, E. P. (1975) Bacteriol. Rev. 39, 232. Cunningham, C. C., & Hager, L. P. (1971a) J . Biol. Chem. 246, 1575. Cunningham, C. C . , & Hager, L. P. (1971b) J . Biol. Chem. 246, 1583. Curatolo, W., Sakura, J. D., Small, D. M., & Shipley, G. G. (1 977) Biochemistry 16, 23 13. Dadok, J., & Sprecher, R. F. (1974) J . Magn. Reson. 13, 243. Dehlinger, P. J., Jost, P. C., & Griffith, 0. H. (1974) Proc. Natl. Acad. Sci. U.S.A. 71, 2280. Esfahani, M., Crowfoot, P. D., & Wakil, S. J. (1972) J . Biol. Chem. 247, 725 1. Eytan, G. D., Matheson, M. J., & Racker, E. (1976) J . Biol. Chem. 251, 6831. Fleischer, B., Casu, A., & Fleischer, S. (1966) Biochem. Biophys. Res. Commun. 24, 189. Fox, C. F. (1972) in Membrane Molecular Biology (Fox, C . F., & Keith, A. D., Eds.) pp 345-385, Sinauer Associates Inc., Publishers, Stamford, CT.
Futai, M. (1973) Biochemistry 12, 2468. Futai, M. (1974) Biochemistry 13, 2327. Garewal, H . S. (1973) Anal. Biochem. 54, 319. Gatt, S., & Bartfai, T. ( I 977a) Biochim. Biophys. Acta 488, 1.
Gatt, S.,& Bartfai, T. (1977b) Biochim. Biophys. Acta 488, 13. Gennis, R . B., & Jonas, A. (1977) Annu. Rec. Biophys. Bioeng. 6, 195. Gent, M. P. N., & Ho, C. (1978) Biochemistry 17, 3023. Gulik-Krzywicki, T.( 1 975) Biochim. Biophys. Acta 415, 1. Hershey, A. D., & Chase, M. (1952) J . Gen. Physiol. 36, 39. Jonas, A., Krajnovich, D. J., & Patterson, B. W. (1977) J . Biol. Chetn. 252, 2200. Jost, P. C., Griffith, 0. H., Capaldi, R . A., & Vanderkooi, G . (1973) Proc. Natl. Acad. Sci. U.S.A. 70, 480. Jost, P. C., Nadakavukaren, K. K., & Griffith, 0. H. (1977) Biochemistry 16, 3 110. Kaback, H. R. (1974) Methods Enzymol. 31, 698. Kates, M. ( 1 972) in Laboratorjs Techniques in Biochemistry and Molecular Biology (Work, T. S . , & Work, E., Eds.) Vol. 3, pp 267-610, North-Holland, Amsterdam. Kohler, S. J., & Klein, M. P. (1976) Biochemistrj>15, 967. Kohler, S. J.. & Klein, M. P. (1977) Biochemistry 16, 519. Kohn, L. D., & Kaback. H. R. (1973) J . Biol. Chem. 248, 7012. Korenbrot. J . 1. (1977) Annu. Rec. Physiol. 39, 19. Lee, A. G. (1975) Prog. Biophys. Mol. Biol. 29, 3. Lee, A. G . , Birdsall, K. J . M.,& Metcalfe, J . C. (1974) Methods Membr. Biol. 2, 1-1 56. Linden, C. D., & Fox, C. F. (1975) Acc. Chem. Res. 8, 321. Lucy, J . A . (1974) FEBS Lett. 40, S-105. Marr, A . G . , & Ingraham, J. L. (1962) J . Bacteriol. 84, 1260. Masoro, E. J . (1977) Annu. Rec. Physiol. 39, 301. Mavis. R. D., & Vagelos, P. R. (1972) J . Biol. Chem. 247, 652.
FUNG, PRATT, AND HO
McLaughlin, A. C., Cullis, P. R., Hemminga, M. A., Hoult, D. I., Radda, G. K., Ritchie, G . A., Seeley, P. J., & Richards, R . E. (1975) FEBS Lett. 57, 213. Michaelson, D. M., Horwitz, A. F., & Klein, M. P. (1974) Biochemistry 13, 2605. Nakamura, M., & Ohnishi, S. (1975) J . Biochem. ( T o k y o ) 78, 1039. Novosad, Z., Knapp, R . D., Gotto, A. M., Pownall, H. J., & Morrisett, J . D. (1976) Biochemistry 15, 3176. Okuyama, H., Yamada, K., Kameyama, Y., Ikezawa, H., Akamatsu, Y., & Nojima, S. (1977) Biochemistry 16,2668. Peterson, S . W., & Deamer, D. W. (1977) Arch. Biochem. Biophys. 179, 218. Pratt, E. A., Fung, L. W.-M., Flowers, J. A., & Ho, C. (1979) Biochemistry 18 (preceding paper in this issue). Racker, E. (1967) Fed. Proc., Fed. A m . SOC.Exp. Biol. 26, 1335. Razin, S. (1972) Biochim. Biophys. Acta 265, 241. Reeves, J. P., Hong, J.-S., & Kaback, H. R. (1973) Proc. Natl. Acad. Sci. U.S.A. 70, 1917. Rohm & Haas (1976) Surfactants and Dispersants: Handbook of Physical Properties, Rohm & Haas, Philadelphia, PA. Rydstrom, J. (1 976) Biochim. Biophys. Acta 455, 24. Seelig, J., & Gally, H . U . (1976) Biochemistry 15, 5199. Short, S . A,, Kaback, H. R., & Kohn, L. D. (1974) Proc. Natl. Acad. Sci. U.S.A. 7 1 , 1461. Stier, A., & Sackmann, E. (1973) Biochim. Biophys. Acta 31 1, 400. Tanaka, Y., Anraku, Y., & Futai, M . (1976) J . Biochem. ( T o k y o ) 80, 82 1 . Warren, G. B., Toon, P. A , , Birdsall, N . J. M., Lee, A. G., & Metcalfe, J. C. (1974a) Biochemistry 13, 5501. Warren, G . B., Toon, P. A., Birdsall, N. J. M., Lee, A. G., & Metcalfe, J. C. (1974b) Proc. Natl. Acad. Sei. U.S.A. 71, 622.