Biochimica ct Bioptwsica Acta, 1120(1992) 49-58

© 1992 Elsevier Science Publishers B.V. All rights reserved 0167-4838/92/$05.00

49

BBAPRO 34140

Association of a-phosphatidylinositol-specific phospholipase C with phospholipid vesicles Chong jun Xu and Gary L. Nelsestuen Department of Biochemistry, Unicersity of M#mesota. St. Paul. MN (USA)

(Received 26 August 1091)

Key words: Calcium-bindingprotein; Protein-membrane interaction; Phosphatidylinositolcycle; PIP,; Phospholipase C; Calcium; Second messenger The alpha isoform of phosphatidylinositol-specific phospholipase C (a-PI-PLC, M r 62000) was purified from bovine brain. Enzyme activity was dependent on calcium, sodium cholate and showed the anticipated specificity for the phosphatidylinositols. Calcium interaction with this protein, investigated by gel filtration chromatography, showed no detectable binding at calcium concentrations adequate to activate the enzyme. Association of a-PI-PLC with phospholipid vcsicles was studied by light scattering, fluorescence energy transfer and gel-filtration chromatography. The enzyme readily associated with vesicles of high charge density, with vesicles of crude acidic phospholipids and with PIP 2. Interaction was characterized by a rapid association followed by slower addition of more protein to the phospholipjd. Complexes containing 20-30 percent protein (by weight) were readily obtained. Calcium had only a small effect on this interaction. The protein-phospholipid complexes appeared to bind less calcium than a similar amount of phospholipid alone. Thus, a-PI-PLC did not appear to be a calcium-binding protein in either its free or membrane-associated states. Although a-PI-PLC showed the highest propensity to bind to phospholipids, a number of other proteins also associated with phospholipids under the conditions used. Thus, whether or not the observed interaction of a-PI-PLC with membranes was specific and biologically important or whether it was a process common to many proteins, was not known. Knowledge of this interaction may enhance our understanding of possible mechanisms for protein-membrane interactions in general.

Introduction Calcium is a universal second messenger in virtually every biological system that has b e e n examined. One way by which calcium response elements can be cataloged is by the type o f calcium binding site. T h e most abundant forms o f calcium response elements known contain calcium binding structures related to that found in parvaibumin. Many reviews o f this topic are available (cf Refs. 1 and 2). T h e calcium binding site of these proteins consists of the ' E - F hand' structure, a combination o f two helixes (the E and F helixes of

Abbreviations: BSA, bovine serum albumin; EDTA, ethylenediaminetetraacetic acid; EGTA, ethylene glycerol bis(/3-aminoethyl ether) N. N. N', N'-tetraacetic acid; PS, phosphatidylserine; PC, phosphatidylcholine; PE, phosphatidylethanolamine; PI, phosphatidylinositol; PIP2. phosphatidylinositol4,5-bisphosphate; PLC, phospholipase C; a-PI-PLC, the form of PLC purified for this study; PKC, protein kinase C. Correspondence: Gary L. Nelsestuen, Department of Biochemistry, 1479 Gortner Ave., St. Paul. MN 55108, USA.

parvalbumin) and an intervening loop. Earlier, Kretsinger postulated that all intracellular calcium binding proteins might be based on the E-F hand [3,4]. T h e latter postulation seems inconsistent with more recent developments which show that several calciumresponsive enzymes have no sequence analogous to the E-F hand. Examples include protein kinase C (PKC) and phospholipase C (PLC). In addition, the annexin family of proteins [5] bind to phospholipids in a calcium-dependent manner, but contain no E-F hand structures. T h e annexin family o f proteins contain a repeating domain described as the endonexin fold [6] and this may constitute a calcium binding structure. However, the number of calcium ions bound to these proteins does not appear to be closely related to the number of repeats of the endonexin fold [7]. In addition, the X-ray crystal structure of annexin V showed only partial participation o f amino acids from the endonexin structure in calcium binding [8]. Calcium binding to the annexins has b e e n r e p o r t e d to be highly d e p e n d e n t on the presence o f phospholipid so that calcium binding sites on the free protein are probably not fully ex-

50 pressed [9]. Calcium sites in annexin V have been detected in protein crystals soaked in 5{) mM calcium [8]. This high level of metal ion may identify relatively low affinity interactions. Thus, whether or not the sequence of the endonexin fold can be used as a predictor of calcium binding to a protein, in the same manner that the E , F hand sequence is used to predict calcium binding, is largely unknown. The enzyme activities of PKC and PLC are both calcium.sensitive, but these proteins do not contain E-F hand sequences or endonexin fold sequences. Thus, the emerging picture might suggest the existence of a large number of different calcium binding structures. However, measurement of calcium binding to many of these calcium-responsive proteins has not been reported and it is not clear that these proteins actually bind calcium. For example, recent studies with PKC and other calcium-dependent phospholipid-binding proteins showed that the free proteins bound no detectable calcium, even at levels much above intracellular calcium [7,10]. Their calcium binding activity was dependent on the presence of phospholipids. It seemed likely that the calcium sites were formed at the protein-phospholipid interface. PKC and related proteins therefore do not constitute calcium binding proteins in the formal sense even though they are calcium response elements in the cell. The phosphatidylinositol-specific PLC enzymes are a family of proteins, many of which are calcium-sensitive. A surprising aspect of these proteins is that, although the substrate is a membrane component, they are commonly purified as soluble proteins from the cytosol. Some reports suggest that PLC will bind to bilayer membranes in a manner adequate to survive gelifiltration chromatography [ll]i b u t few quantitative aspects of this interaction were investigated. To our knowledge, no direct calcium binding studies have been reported. Therefore, whether the PLC enzymes are calcium binding proteins in the strict sense, like calmodulin, or whether they bind calcium in the phospholipid-dependent manner described for PKC, is not known. This study was initiated in order to examine these questions. The results were somewhat surprising, a-PIPLC from bovine brain did not bind detectable levels of calcium, even though its activity was sensitive to calcium, a-PI-PLC interacted readily with certain phospholipid membranes in a calcium-independent process. The biological importance of this interaction is discussed.

Experimental procedures Phosphatidyl-(2-3H)-inositol-4,5-bisphosphate (3 H_ PIP2; specific activity 1.0 Ci/mmol), L-3-phosphatidyl(2-3H)-inositol (~H-PI; specific activity 17.2 Ci/mmol)

were purchased from Amersham; 45CaCl2 (20.51 m C i / m g ) was purchased from Dupont New England Nuclear; phosphatidylinositol 4,5-bisphosphate (PIP2) , phosphatidylinositol (PI), bovine brain phosphatidylserine (PS), egg yolk phosphatidylcholine (PC), L-aphosphatidylethanolamine-dipalmitoyI-N-dan syl (dansyI-PE), bovine brain extract type 1 (Foich fraction 1), bovine serum albumin (BSA), fibrinogen, alkaline phosphatase, DEAE-cellulose (Coarse Mesh), Sephacryl S-300, hydroxyapatite and heparin (170 U / m g ) were purchased from Sigma; Mono-Q HR (5/5) was from Pharmacia; Factor-Xa was purified by published procedures [12] and was a kind gift from Ruth Schwalbe. Other chemicals were from Sigma in analytical grade.

PLC purification PLC was purified to homogeneity from bovine brain by modification of previously published procedures [13,14]. Briefly, eight whole fresh bovine brains were each homogenized in 1 l of 20 mM Tris-HCl buffer containing l0 mM benzamidine, 5 mM EDTA, 10 mM EGTA, 5 0 / ~ g / m l PMSF and 0.3%/3-mercaptoethanol (pH 8.0). The homogenate was centrifuged for l h at 5000 × g . The supernatant was incubated overnight with 4 1 of a slurry of DEAE-cellulose which had been equilibrated with 20 mM Tris-HCI, l mM EDTA, 5 mM fl-mercaptoethanol and 0.5 mM PMSF, pH 7.5 (buffer A). The DEAE-cellulose gel was packed into a column and was washed with buffer A until the A2sonm of the effluent was less than 0.05. A linear gradient of 12 ! of buffer A and 12 I buffer A containing 0.4 M NaCI was developed at a flow rate of 240 m l / h . 25 ml fractions were collected and assayed for activity. A broad peak of PLC activity (specific activity: 0.4 t t m o l / m i n per mg, total yield of approx. 3200 units; 1 unit = 1 / z m o l / m i n ) was pooled and dialyzed overnight against four changes of buffer A containing 0.1 M NaCl (buffer B). The sample was applied to a Heparin-Sepharose 4B column (600 ml), prepared by the procedure of Parikh et al. [15], which was preequilibrated with buffer B. After washing with buffer B to remove unbound protein, the column was eluted with a 4 1 linear gradient of 0.1-1 M NaCI in buffer B at a flow rate of 120 m l / h . The active fractions were pooled (specific activity: 0.9 /~mol/min per mg, total of 2000 units) and concentrated to about 20 ml by pressure dialysis. The concentrated sample was then applied to a column (2.5 x 120 cm) of Sephacryl S-300 that had been equilibrated with 20 mM Tris-HCl, 1 mM EDTA, 0.3% /3-mercaptoethanol and 0.15 M NaCI, pH 7.5. The column was eluted with the same buffer at a flow rate of 20 m l / h . The 8 ml fractions were collected and assayed for activity. A sharp peak of PLC activity (specific activity: 3.1 # m o l / m l , total of approx. 210 units) was collected and dialyzed against 20 mM potas-

51 sium phosphate buffer (pH 7.3) containing 1 mM EDTA, 5 mM 2-mercaptoethanol and 0.5 mM PMSF (buffer C). The dialyzed sample was applied to a hydroxyapatite column (150 ml) equilibrated with buffer C. After washing the column with buffer C to an effluent of A2s0, m less than 0.05, the column was eluted with a 500 ml linear gradient of 20-500 mM potassium phosphate in buffer C. The 5 ml fractions were collected and assayed for activity. The fractions containing PLC activity were collected and concentrated to about 10 ml by pressure dialysis. This sample was then dialyzed against buffer A without /3mecaptoethanol (buffer D). The dialyzed sample (specific activity: 8.3 /zmol/min per mg, total of approx. 65 units) was applied to a Mono-Q HR (5/5) cation exchange HPLC column that had been equilibrated with buffer D at a flow ratg of 0.25 ml/min. The column was eluted with a 15 ml linear gradient of 0-0.5 M NaC1 in buffer D at a flog, rate of 0.5 ml/min. Fractions (0.5 ml) were collected, assayed for activity and analyzed by polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate [16]. PLC activity was eluted at approx. 10 ml after the start of the gradient (about 0.3 l~ NaCl) and had a specific activity 12 /zmol/min per mg with a final yield of 1.5 mg protein.

activity titration with calcium, calcium-EGTA buffers were used to control the free calcium concentration [17]. For preincubation experiments, the reaction was initiated by addition of calcium.

Phospholipid cesicle preparation Small unilamellar vesicles were prepared according to the methods of Huang [18] and Bazzi and Nelsestuen [19]. Briefly, the appropriate phospholipids were mixed in organic solvent and dried under a stream of N2 gas. Tris-HCl buffer (10 mM, pH 7.5) was added. The solutions were votexed vigorously for about 5 min, and sonicated with a microprobe for 3 min (2 s bursts with 5 s intervals). After sonication, the vesicles were applied to a column of Sepharose 4B (1.5 × 40) which was previously equilibrated with 50 mM Tris-HCl (pH 7.5) buffer containing 100 mM NaC! and eluted with the same buffer. 2 ml fractions were collected and their turbidity at 260 nm was measured. The fractions corresponding to small unilamellar vesicles were pooled and concentrated by pressure dialysis. Phospholipid concentrations were determined from organic phosphate assay [20] using a phosphorus to phospholipid weight ratio of 1:25 and sodium phosphate as the standard [21].

Calcium bbrding PLC actirity assay PLC activity was assayed according to the method described by Katan and Parker [14] with a minor modification. Briefly, substrate PIP, ( l l . l mg) was mixed with 3H-PIP2 (25 ttCi) and dried from chloroform/methanol solution under a stream of N 2 gas. It was then resuspended in 5 ml of l0 mM Tris-HCi, pH 7.5 solution. After dispersal by mixing on a VWR Vortexer 2 for 5 min, the mixed solution was sonicated for 3 min at 4°C with a model W385 Sonicator using a ~ U 419 microprobe. Preparation of other substrates such as pI/aH-PI was the same as described for PIP2/ 3H-PIP2. The assay was carried out in a total vol. of 100/~! containing 20 mM Tris-HCl (pH 6.6), 100 mM NaCi, 0.6% sodium cholate, 0.5 mM EGTA, 0.6 mM CaCl 2, 5 mM/~-mercaptoethanol, 40/zg bovine serum albumin and 10 t~l of PIPJ3H-PIP2. The final concentration of PIP 2 in the reaction mixture was 220/~M, containing 0.05/zCiSH-PIP2 . The reaction was initiated by the addition of a-PI-PLC samples, except as noted elsewhere. The reaction mixture was incubated at 37°(2 for 10 min (except as noted in the figure legends) and the reaction was stopped by adding 0.5 ml of chloroform: methanol: concentrated HCI (100:100:0.6) followed by 0.15 ml of 1 M HCI. The mixture was vortexed vigorously for at least 20 s and centrifuged at 1 3 0 0 0 x g for 5 rain. A 400 ttl aliquot of the upper phase was moved and radioactivity was measured in a liquid scintillation counter. For

Calcium binding to tr-PI-PLC was measured by the method of Hummel and Dreyer [22] as described by Bazzi and Nelsestuen [10]. This method utilized gelfiltration chromatography at 4°C on a Sephacryl S-300 column (1.0 × 30) which was equilibrated with 20 mM Tris-HCl (pH 7.25), 10% glycerol, 100 mM NaCI, 0.5 mM dithiothreitol and 0.5 m g / m l BSA containing 20 /~M CaCI 2. Standard calcium solution was made by diluting 45CAC12 with standard CaCI 2 solution to produce the desired calcium concentration and specific activity (36800 cpm/nmol). The standard CaCI 2 solution was made by dissolving anhydrous CaCO 3, that has been heated at 150°C overnight to remove traces of water, in dilute HCI (the pH of the final solution was adjusted to 5.5). BSA was added to the column equilibration buffer in order to reduce nonspecific adsorption of protein to the column matrix, and to stabilize the free protein [23]. Approx. 0.4 mg a-PI-PLC in 0.8 ml of equilibration buffer was applied to the column and eluted with the same buffer at a flow rate about 1 ml/h. Fractions (18 drops) were collected. The calcium concentration in each fraction was determined by counting the radioactivity of 45CACI2. a-PI-PLC activities in each fraction were assayed by the standard assay method.

Protein-phospholipid binding Three methods were used to measure the interaction of a-PI-PLC with phospholipid. First, binding was

2 measured by gel-filtration chromatography on Sephacryl S-300. In the presence of calcium, the experiment was the same as the Hummcl-Dreyer method mentioned above which contained 20 # M Ca -'+. Crude acidic phospholipid (0.8 mg) made of bovine brain extract type l (Folch fraction l) containing about 5060% PS was mixed with 0.4 mg of a-PI-PLC in the equilibration buffer and incubated at room temperature for 30 min. The sample was applied to the Sephacryl S-300 column at 4°(? at a flow rate of l ml/h. Again, a-PI-PLC activity in each fraction was assayed by the method described above. The association of a-PI-PLC with phospholipid vesicles was also measured by fluorescence energy transfer methods as described by Bazzi and Nelsestuen [19]. The fluorescence originated from tryptophan residues in the protein (excitation at 280 nm) which can transfer excitation energy to a dansyl group in the phospholipid vesicle (emission at 520 nm, with a 500 nm cutoff filter) when the two components are associated, c~-PI-PLC (15/~g) and phospholipid (15 #g) containing Dansyl-PE were initially mixed in 1.5 mi of 50 mM Tris-HCl (pH 7.5) buffer containing 100 mM NaCl plus the calcium concentration given in the figure legends. Fluorescence energy transfer was expressed as I / I o, where 1o refers to the fluorescence intensity of dansyl labeled phospholipid alone (excited at 280 nm), while 1 refers to the fluorescence intensity of the protein-phospholipid complex. In every case, control experiments showed that the increased fluorescence in the sample containing protein arose from energy transfer rather than from changes in quantum yield. That is, direct excitation of the dansyl group at 350 nm showed no increase in emission at 520 nm as the protein bound to the membranes. The third method of measuring PLC-phospholipid interaction was by light scattering intensity as described by Bazzi and Nelsestuen [10]. For small particles, light scattering intensity is proportional to molecular weight [24]. Light scattering intensity was measured at 320nm and is expressed as 1/Io, where ! is the light scattering intensity of the protein-phospholipid mixture and I o is the intensity of the vesicles alone. The very small light scattering intensity from the free protein was subtracted as a background. All of the light scattering and fluorescence energy transfer measurements were made at 25°C in a Hitachi-Perkin-Elmer Model MPF-44A fluorescence spectrophotometer.

Electrophoresis and protein concentration assay Polyacrylamide gel electrophoresis in the presence of sodium dodecyl sulfate was performed on a slab gel (7.5% acrylamide) according to Laemmli [16]. Protein was stained by 0.25% Coomassie brilliant blue R-250 and destained by 7% acetic acid, 10% methanol solution. The molecular weight standards used in the gel

were: myosin (205000), //-galactosidase 016000), phosphorylase b (97400), bovine plasma albumin (66000), egg albumin (45000) and carbonic anhydrase (29000). The protein concentration of the a-PI-PLC preparation was determined by the method of Bradford [25] using BSA as the standard. Results

Specificity and cofactor requirements of PLC The isolation procedure and molecular weight (see below) of the enzyme isolated for this study suggested that it corresponded to a-PI-PLC. However, due to the unusual calcium-interaction properties reported below, it was necessary to establish that this enzyme displayed the typical calcium-sensitive activity and had the reported substrate specificity. The results shown in Fig. 1 demonstrate that this was the case. This enzyme hydrolyzed PIP 2 in a calcium-dependent manner. The calcium dependence showed a response at l /.tM calcium, 12

A

IO 8

,~,

0

"~

5

I

I

I

I

I

I

I

I

I

I

.01

.1

I

B

3 -1 -,.-

2

""

I

.m

,.a 2 C

O

0 0

.0001

.001

i

tO

!00

Ca z÷ {mM) Fig. I. Effect of Ca2÷ concentration on the hydrolysis of phosphatidylinositols by a-PI-PLC. Panel A, the assay was carried out at 37°C for 10 rain in 20 mM Tris-HCI (pH 6.6), 100 mM NaCI, 0.6% sodium cholate, 5 mM//-mercaptoethanol, 0.4 mg/ml BSA, 220/~M PIP2, 0.05 /zCi 3H-PIP2 and the calcium concentration indicated. Panel B, experiments were the same as in A except PI was the substrate and 0.2% sodium cholate was the detergent. Panel C, experiments were as in A except no sodium cholate was in the solution and the reaction time was 30 min. The symbols represent PIP2 (o), and Pl as the substrates (o).

53 but maximum activity at almost 100 p M calcium. This type of calcium responsiveness has been reported previously [26]. An important observation was that the activity was highly dependent on calcium; reactions of PIP 2 with a-PI-PLC without calcium showed virtually no substrate hydrolysis in 50 rain (see below). Overall, these properties indicated that the enzyme used for subsequent measurements was similar to other calcium-sensitive PLC preparations. Substrate specificity also indicated similarity to other phosphatidylinositol-specific PLC enzymes. In general, PIP 2 (Fig. IA) was favored over PI (Fig. 1B). An interesting point, consistent with previous reports [ 2 7 ] , w a s that specificity toward inositol phosphates was reversed when detergents were omitted. That is, without detergent, the activity toward PI was higher than toward PIP, (Fig. IC). In addition, the calcium requirement seemed highly dependent on the nature of the substrate and whether or not detergent was included in the assay.

Calcium binding to PLC Calcium binding to a-PI-PLC was assessed at 20 # M calcium using the Hummel-Dreyer technique [22]. This calcium concentration was adequate to activate the enzyme (Fig. 1) and, if calcium sites were limited to the protein itself, should be adequate to detect bound calcium. However, the results showed no excess calcium in the fractions containing protein (Fig. 2). The limits of detection were estimated from the difference between the average and standard deviation for fractions 26 to 30 (fractions with protein) minus fractions 6 to 10 (column background without protein). The difference corresponded to 0.07 + 0.42/.t M calcium over the background. That the estimated error was greater than the mean difference between the background and samples may suggest that error estimation was generous. Nevertheless, these values could be used to calculate a

24

E 2.5 - 2.0

22

i.s

~_0 "-'=

~:I. I-0 ~.0.5

tqe: 18 (,D

~0.0

'16 5

10

15

20

25

3O

35

40

Fraction Number Fig. 2. Determination of calcium binding to a-PI-PLC by the Hummel-Dreyer technique, a-PI-PLC (0.4 rag) was incubated with 0.8 ml of buffer at room temperature for 30 rain and was applied on a Sephacryl S-300 column (I.0x30)equilibrated and eluted with 20 mM Tris-HCl (pH 7.25), 10% glycerol, I00 mM NaCI, 0.5 mM dithiothreitol, 0.5 mg/ml BSA and 20 #M CaCIz containing 45Ca2+. T h e flow rate w a s 1 ml/h. Fractions (0.5 ml) were collected and assayed for enzyme activity (e), and for 45Ca2" determination Co}.

minimum stoichiometry of calcium bound per protein. The average protein concentration in fractions containing protein was 2.2 # M so that the amount of calcium bound to a-PI-PLC corresponded to 0.03 +0.19 moi/mol. The protein concentration in the column fractions was estimated from activity eluted from the column using specific activity of 12 units/mg of pure protein (Fig. 1A). Thus, a-PI-PLC did not appear to constitute a calcium-binding protein in the formal sense in that there was no calcium binding to the free protein at intracellular calcium levels.

Interaction of PLC with phospholipids and calcium binding to the complex The interaction of many proteins with phospholipids can readily be detected by light scattering changes. This technique has been used for the vitamin K-dependent plasma proteins [28], other blood clotting proteins [29], protein kinase C [10] and its related proteins [7), myelin basic protein [30] and serum amyloid P component [31]. Another method for determining proteinmembrane binding is fluorescence energy transfer from tryptophan residues in the protein to a dansyl group attached to the phospholipids of the membrane. This method showed signal changes for all of the cases listed above except for most of the vitamin K-dependent plasma proteins. These proteins apparently bind in a manner that does not place tryptophan residues near to the membrane surface. Neither light scattering (data not shown) nor fluorescence energy transfer (Fig. 3B) detected interaclion of PLC with membranes of pure phospholipids containing 20% monoanionic lipids. Addition of 1.0 mM calcium also failed to generate a detectable interaction (data not shown). Light scattering intensity was sensitive to a digital change of about 3%, corresponding to that expected for binding 1 protein ( M r 62000) to a small unilamellar vesicle ( M r 4 × 106). Thus, protein binding to membranes with low charge density was minimal. In addition, these results showed that a-PIPLC was not analogous to PKC. The latter protein bound to phospholipids of this composition in a calcium-dependent manner [19]. A previous report indicated that PLC did associate with membranes and co-chromatographed with vesicles on gel-filtration chromatography [11). Two differences of the earlier experiment were that the ratio of phospholipid to protein was high (approx. 150: 1) and that the membranes used contained high charge density (70%PI, 30%PE). These conditions may detect low affinity associations. To obtain a better comparision, a-PI-PLC was mixed with phospholipid vesicles containing high charge density. Results with purified phospholipids (10% dansyl-PE, 90% PS) showed a slow interaction that was detected by fluorescence energy. transfer (data not shown). The velocity and the extent

54 1.6

A g

z.4

E 1.2

..~ _a l.(~

4

6O

80

Time (rain.)

1.2

N 2.0

1.o I 0

2{)

4C) Time (rain.)

60

I

i

0.1 0.3 0.5 0.7 0.9 NaCI (M)

Fig. 3. Association of a-PI-PLC with phospholipids. Panel A, light scattering intensity of the protein-vesicle mixture (I)relative to that of the vesicles alone (I~,) was followed after addition of 15 /~g a-PI-PLC to 15 ~ g of crude acidic phospholipids (11}0% brain extract fraction l) in !.5 ml of buffer containing 50 mM Tris-HCI (pH 7.5), 100 mM NaCI and 0.1 mM Ca- . Panel B, fluorescence intensity ( I ) relative to that of the vesicles alone (Io) was followed after addition of 15 /.tg a-P1-PLC to 15 # g of crude phospholipids (bovine brain extract type 1 :Dansyl-PE, 90: 10) in the presence of 0.1 mM Ca 2+ (o) or 0.5 mM EGTA (o). The result from a similar experiment for vesicles ( 15 p.g) of PS: PC: DansyI-PE (20: 70: 10) is also shown ( m ). Panel B (right portion) also shows the effect of NaCI. Control experiments showed that neither high salt nor calcium influenced the fluorescence of the vesicles over the time scale of the experiment. Calcium was added to the phospholipids just before measurements were initiated~ Panel C, lefL The time-course of light scattering at 320 nm after addition of 15/zg a-PI=PLC to 15/zg of PIP~ in 1.5 ml of 50 mM Tris-HCI (pH 7.5), 100 mM NaCI buffer is shown in the presence of 0.6% cholate sodium with 0.1 raM Ca -~+ ( I ) or with 0.5 mM EGTA (ra). Also shown is the time-course of fluorescence energy transfer after addition of 15 ~zg a-PI-PLC to 15 /zg of PIPz:DansyI-PE (90: 10) in 1.5 ml of 50 mM Tris-HCI (pH 7.5), 100 mM NaCI buffer in the presence of 0.1 mM Ca :+ (o) or 0.5 mM EGTA (©). Panel C, right. The effect of NaCI on fluorescence energy transfer is shown.

of interaction were greater with membranes of crude acidic phospholipids (Fig. 3A). Nevertheless, the resuits with pure phospholipids suggested that impurities in the crude phospholipids were helpful but not essential to this interaction. In any event, the rate of PLCphospholipid binding was easily detected by light scattering and fluorescence energy transfer (Fig. 3A, B). Since the molecular weight ratio of the protein-phospholipid complex to that of the phospholipid alone is nominally equal to the square root of I / l o [24], the complexes formed in Fig. 3 contained about 20% pro-

tein. This phospholipid association was at least partially reversed by high salt (Fig. 3B, right panel). a-PI-PLC also associated with pure PIP2 in a timedependent manner (Fig. 3C). This interaction might be of greater interest since this is the most active substrate. Thus, charge density of the membrane appeared to be an important factor for PLC-membrane binding and PIP2 may simply maximize this parameter. The small influence of calcium on these protein-phospholipid interactions (Fig. 3A-C) may have been due to the fact that enzyme will hydrolyze some of the lipid when calcium is added and thereby change its structure. In this case, less than 30% of the a-PI-PLC was dissociated by 1 M NaCl (Fig. 3C, right panel). Detergent enhanced the association of a-PI-PLC with PIPs as detected by light scattering (Fig. 3C). Gel-filtration chromatography of a-PI-PLC that had interacted with crude acidic phospholipids showed that the enzyme activity eluted with the phospholipid vesicles at the exclusion volume of the column (Fig. 4). Several experiments were run and the recovery of enzyme activity was variable. That is, some experiments showed quantitative recovery of activity, most of which was associated with the phospholipids (Fig. 4). However, when calcium was omitted from the buffer, only about 58% of the activity was recovered (data not shown). This suggested that the enzyme may be inactivated to some degree by incubation with membranes of high charge density. The protective effect of calcium may arise from reduction of the effective charge density of the membranes. In order to demonstrate that the protein itself was present in the phospholipid peak, SDS gel electrophoresis was conducted (Fig. 4, inse~t). The results showed a molecular mass of about 62 kDa for the initial protein and showed the presence of t h e same protein in the phospholipid peak. Thus, light scattering, fluorescence energy transfer, activity measurements and acrylamide gels all showed interaction of o~-PI-PLC with membranes containing high levels of acidic phospholipids. Calcium binding to the PLC-phospholipid complex was examined by the Hummel-Dreyer technique at 20 /zM calcium (Fig. 5A). The results showed a peak of calcium eluting with the protein-phospholipid complex. However, this peak was not greater than that found in experiments containing only phospholipids. The difference between the five fractions containing protein (fraction 26 to 30) and those same fractions in the column containing only phospholipid vesicles was -1.64 /zM with a standard deviation for five background fractions of + 0.26 # M calcium. In other words, the protein-phospholipid complex bound less calcium than the phospholipid alone. This would be consistent with cationic protein groups binding to and neutralizing some of the membrane charge, thereby reducing calcium binding to the phospholipids. When the PLC-

55 (a)

(b)

Association of other proteins with membranes contabling high charge density

(c)

2 m

E o

E

0

0

10

20 Fraction Number

30

40

Fig. 4. Gel filtration of a-PI-PLC-phospholipid on Sephacryl S-300. Samples containing either 0.4 mg a-PI-PLC (el or 0A mg a-PI-PLC plus 0.8 mg phospholipids (bovine brain extract type I. m) were incubated in 0.8 ml of buffer at room temperature for 3t) min and then chromatographed on a Sephacryl S-300 gel filtration column (I.0 × 30 cm). The buffer, elution and assay conditions were as in Fig. 2. Insert. SDS-gel electrophoresis are shown for (a), a-PI-PLC alone, (b) the molecular weight markers (MW 2{15000; 116000; 97400; 66000:450{10 and 29000) and (c) 40 p l of fraction 2ll from the column elution profile.

phospholipid complex from fraction 19 in Fig. 5A was rechromatographed on a Sephacryl S-300 column in the presence of 2 mM EGTA (Fig. 5B), only about 10% of the PLC activity dissociated from the phospholipid. This result showed that calcium was not necessary for maintaining the PLC-membrane complex. The peak of bound calcium was offset by a trough in the elution profile of the column. As reported previously [7], columns used for phospholipid chromatography generated a somewhat irregular trough of calcium that was broad, extending between fractions 47-61 of the elution profile of the column used in Fig. 5A (not shown). The delayed nature of this trough (the calcium-EDTA complex eluted at fractions 41-42, Fig. 5B) and its breadth may result from phospholipid that becomes bound to the column matrix and buffers the column with respect to free calcium concentration, resulting in a more gradual and delayed trough. This irregularity was independent of protein and did not appear to interfer with analysis of the peak of bound calcium [7]. However, analysis of the trough area was considered an ineffective means of estimating bound calcium.

Previous studies with several proteins have shown calcium-independent protein-membrane binding. For example, calcium-independent interaction of several vitamin K-dependent proteins with membranes containing high charge density is well known [32]. Calcium-independent association of prothrombin with phospholipid vesicles of high PS content also caused protein inactivation [33]. This interaction appeared to occur through the thrombin region of the protein rather than through the amino terminal portion of prothrombin [34]. The latter participates in calcium-dependent protein-membrane binding. Whether or not this process is of biological interest has been questioned [32]. Inagaki et al. [35] reported that incubation of PKC with membranes resulted in protein inactivation. Huang and Huang [36] recently reported that various isoforms of PKC interacted with membranes of 100% PS and became inactive. This interaction appeared to involve the

4(I

3

211

2

I0 ~

I .~

10

20

30

40

B

~ 0.15

:

I o.,o 0.05

1

u-

K_~___ 10

20

_" "! --i!-?--2---_~_~.00 30 40 50 60 70 Fraction Number

Fig. 5. Calcium interaction with the PLC-phospholipid complex. Panel A, samples containing either 0.4 mg of ot-PI-PLC plus 0.8 mg of phospholipid (bovine brain extract type 1) or 0.8 mg of phospholipid alone were incubated in 0.8 ml of the equilibration buffer used in Fig. 2 for 30 min at room temperature, Other procedures were as described in Fig. 2. PLC activity (O) in each fraction was assayed and the calcium concentration (as ~SCaZ+) in each fraction was

determined for vesicles alone (el and for the PLC-vesicle mixture (o). Panel B, fraction 19 from the elution profile in panel A was rechromatographed on the same Sephacryl S-300 column with the same buffer except that 20 gM CaCI2 was replaced by 2 mM EGTA. PLC activity ([]1 and calcium concentration (*) were determined.

56 1.3

-I. L.

t.-

1.2

tNI

1.1

0

t~_ 1.0 0

I

lo

I

I

~

I

20 3o 40 Time (min.)

e

I

50

60

Fig. 6. Binding of several proteins to phospholipid vesicles as measured by fluorescence energy transfer. 15 /.tg of Factor-Xa (e), alkaline phosphatase (o) or fibrinogen ( [] ) were added to a solution containing 15 #g of acidic phospholipids (10% DansyI-PE, 90% bovine brain extract type 1) in 1.5 mi of 50 mM Tris-HCl (pH 7.5), 1O0 mM NaCI and 0.5 mM EGTA. Fluorescence intensity of the sample (!) relative to that of the vesicles alone ( I o) is shown.

catalytic domain of PKC rather than the regulatory domain and therefore appeared to be very different from calcium-dependent PKC-membrane association. This interaction and corresponding inactivation of PKC was not observed when membranes of low charge density were used [36]. Thus, it appears that many proteins interact with membranes that have high charge densities. To demonstrate this general feature for several proteins, their interaction with phospholipid was examined. The results showed various levels of interaction, depending on the protein (Fig. 6). Blood clotting factor Xa interacted readily when the membrane contained high charge density. Previous studies showed that fac-

Influence of PLC-phospholipid interaction on enzyme activity The occasional loss of PLC activity during incubation with phospholipids of high charge density indicated that protein-membrane interaction may inactivate the enzyme. Alternatively, time-dependent association with PIP 2 (Fig. 3C) might correspond to an activation step so that preincubation of PLC with substrate before addition of calcium may enhance a-PI-PLC activity toward PIP 2. However, preincubation of a-PIPLC with PIP 2 did not enhance activity (Fig. 7). If anything, the activity decreased with time. An important secondary aspect of these experiments was the high degree of calcium dependence of enzymatic re,~ction. That is, no detectable product was released during the calcium-free preincubation period. Release of IP 3 into the aqueous extraction only began when calcium was added. Discussion

0.8

~0.4 -1

~., 0.2 J

0.(

tor IXa, a related protein, was even more susceptible to this type of interaction [32]. Alkaline phosphatase interacted with the phospholipids as well. Only fibrinogen failed to show detectable interaction on the time scale of these experiments. Parallel experiments using light scattering intensity to measure these proteinmembrane interactions showed similar signal changes (data not shown). In other experiments, mixtures of these phospholipids plus bovine serum albumin were incubated for 1 h and then chromatographed on gelfiltration columns to separate phospholipid from free protein. A substantial amount of albumin was found in the phospholipid peak (data not shown). Thus, interaction of proteins with membranes of high charge density seemed to be quite common but occurred at variable rates.

r'l

--"-°'---h~..... 10

u 20

[] 30

40

50

Time (rain.) Fig. 7. Effect of preincubation on PLC activity, a-PI-PLC (50 ng) was incubated at room temperature with 22.2 # g of PIP 2 in 100 p.I of 20 mM Tris-HCl (pH 6.6), 100 mM NaCI, 5 mM fl-mercaptoethanol, 0.4 mg/ml BSA and 0.5 mM EGTA. At the time indicated, the enzymatic reaction was initiated by adding calcium to 0.6 mM (e). The system without Ca -'+ (12) and the system with Ca -'+ but without a-PI-PLC (o) were used as controls.

It is thought that many hormones, neurotransmitters and related agonists exert some of their effect by activating phospholipase C which breaks down PIP 2 in the plasma membrane of the target cell. This produces two intracellular messages, inositol-l,4,5-triphosphate (IP3), which releases intracellular Ca 2÷, and diacylglycerol (DAG), which activates PKC (see reviews 37, 38, 39, 40). While many studies have shown that the in vitro activity of PLC is highly dependent on Ca 2+ [27,41,42], the nature of the calcium effect is unknown. A surprising aspect of PLC is that it is primarily a cytosolic enzyme while its substrate is a membrane component. It seems clear that, at some point, the enzyme combines in some manner with its substrate and becomes a membrane-associated protein. This study examined the calcium- and membrane-binding properties of one PLC isozyme (a-PI-PLC, M r 62000). Its molecular weight and enzyme activity appeared

57 similar to a PLC isozyme recently purified from bovine brain [43]. Although its enzyme activity depended on Ca 2+, gel-filtration chromatography did not detect Ca z÷ binding to PLC, even at a concentration adequate to activate the enzyme. Furthermore, PLC that had become associated with phospholipids failed to bind Ca 2+ to a detectable level. Thus, PLC did not appear to be a calcium,binding protein which associated first with calcium and underwent a conformational change that altered its activity or interaction with other proteins. Indeed, it appears that the only intracellular calciumbinding motif that has been unambiguously demonstrated to use this mechanism is the 'E-F" hand structure [3,4]. Primary sequence showed no 'E-F" hand in a-PI-PLC and only one possible "E-F' hand structure in all of the PI-PLC isoforms [44]. t~-PI-PLC also did not correlate with a type of calcium-response element illustrated by PKC [7,10] and possibly the lipocortins [9]. These proteins bind calcium in a manner that is highly dependent on phospholipid. Even though a-PI-PLC interaction with phospholipid was easily observed, no additional calcium binding by the complex was detected. PLC therefore must respond to calcium in a manner that is considerably more subtle than these other systems. The fact that the calcium concentration needed for PLC activation varied with the type of substrate and with the presence of detergent in the assay may suggest a necessary ealcium-substrate interaction. In fact, Ca -'+ has a higher affinity for PIP~ than other phospholipids [45,46] and the Ca 2÷ requirement in the PLC activity assay increased as the substrate concentration increased [47]. Whether calcium is an important regulator of PLC u n d e r in vivo circumstances is not known. In some experiments, PLC activity has not appeared to be sensitive to changes in intracellular calcium [48,49]. This might suggest that, either the enzyme is always activated with calcium or that other factors in the cell, such as G-proteins [50-53], serve to activate PLC in a manner that is independent of intracellular calcium. The current study, which showed no obvious or common type of interaction with calcium, could not distinguish whether the calcium-dependent properties of PLC were unique to the in vitro assay or whether they represent an in vivo process. The second major goal of this study was to examine the interaction of PLC with phospholipids, a process that was easily detected. It was possible that this membrane association represented an important process that may help to explain the existence of two PLC protein populations in vivo, one associated with the membrane fraction and the other in the cytosoi [48]. That is, under the appropriate conditions, cytosolic PLC may become membrane-bound in the manner

detected in this study. This alteration in PLC might be analogous to tl~e interaction of PKC with phospholipids to form a calcium-independent membrane-bound form of the enzyme that displays altered cofactor requirements [54]. Unlike PKC, however, membrane-associated a-PI-PLC still required Ca z+ for activity. The nature of the protein-membrane interaction is not known. Reversal of binding by high salt did not necessarily indicate an ionic mechanism of interaction. While high salt would disrupt ionic interactions between protein and phospholipid, it should also help relax ionic repulsion between the charged head groups of the phospholipids, which might allow tighter packing of the head groups and an increased membrane surface pressure. This might eliminate weak hydrophobic contact of protein with the hydrocarbon region of the membrane. Peterson et al. [55] have recently presented a preliminary report describing interaction between the delta isoform of PI-PLC and phospholipids. That interaction also occurred in a calcium-independent manner and may be analogous to the interaction that we document for t~-PI-PLC. Once again, further studies will be needed to determine possible differences in the observed isoforms and the types of forces involved in protein-membraae binding. While demonstration of a PLC-membrane interaction is intriguing and interaction is an unavoidable aspect of the catalytic process, further studies will be necessary to determine if the interaction documented in this study is biologically important. An argument favoring its importance is that, among the proteins tested, a-PI-PLC showed the highest propensity to form a membrane-bound state. However, blood clotting factors Xa (Fig. 6) and IXa [32) also interact readily with membranes with high charge density. There is no clear evidence suggesting that calcium-independent association of factors Xa and IXa with phospholipids is of biological importance. By analogy, simple observation of a protein-membrane interaction by PLC does not prove a biological role. Perhaps a leading problem for interpretation of the membrane-binding phefiomenon is the possibility that virtually any protein will associate to some degree with membranes of high charge density. When placed at a membrane interface with high electrostatic potential, proteins may denature to some extent and become membrane-associated. This might be analogous to the more well-documented denaturation and interaction of proteins at the air-water interface [56, 57]. Thus, whether the high propensity of a-PI-PLC to interact with phospholipids, in the manner shown in this study, represents an important biological function, is not established. Further studies are needed. In any event, the association of a-PI-PLC with membranes may be useful for other purposes. For

58 e x a m p l e , this i n v e s t i g a t i o n d e m o n s t r a t e d m e t h o d s for producing abundant and easily-observ~le complexes of a-PI-PLC with phospholipids. Future studies can identify t h e t y p e s o f forces a n d t h e r e g i o n o f t h e p r o t e i n t h a t is involved. S i n c e a - P I - P L C d o e s n o t a p p e a r t o c o n t a i n a p e p t i d e s e q u e n c e , such a s a t r a n s m e m b r a n e helix, t h a t is a n a l o g o u s t o k n o w n m e m b r a n e - a s s o c i a t i o n p e p t i d e s , it a p p e a r s t h a t p r o t e i n m e m b r a n e i n t e r a c t i o n s m a y b e s u p p o r t e d by a v a r i e t y of peptide sequences. Further studies with proteins such as f a c t o r X a a n d a - P I - P L C m a y i d e n t i f y p e p t i d e motif(s) t h a t h a v e high p r o p e n s i t y to i n t e r a c t w i t h membranes.

Acknowledgements W e a r e g r a t e f u l to Dr. M o h a m m a d Bazzi for his advice on m a n y t e c h n i c a l m a t t e r s . W e a l s o t h a n k D r . R u t h S c h w a l b e for p r o v i d i n g F a c t o r X a . T h i s w o r k w a s s u p p o r t e d in p a r t by g r a n t s H L 1 5 7 2 8 a n d G M 3 8 8 1 9 from t h e N a t i o n a l I n s t i t u t e o f H e a l t h .

References I Cheung, W.Y. 119801 Science 207, 19-27. 2 Klee, C.B.. Crouch, T.H. and Richman, P.G. (19801 Annu. Rev. Biochem. 49, 489-515. 3 Kretsinger, R.H. (19761 Annu. Rev. Biochem. 45, 239-266. 4 Kretsinger, R.H. (198111CRC Crit. Rev. Biochem. 8. 119-174. 5 Klee, C.B. 11988) Biochemistry 27. 6645-6653. 6 Geisow, M.J., Fritsche, U., Hexham, J.M. Dash, B. and Johnson. T. 119861 Nature 320, 636-638, 7 Bazzi, M.D. and Nelsestuen, G.L. 11991) Biochemistry 3{). 071979. 8 Huber, R., Schneider, M., Mayr, l.,Romiscb, J. and Paques, E.P. (199111 FEBS Lett. 275, 15-17. 9 Crompton, M.R., Moss, S.E. and Crumpton, M.J. (19881 Cell 55. I-3. l0 Bazzi, M.D. and Nelsestuen, G.L. 119901 Biochemistry 29. 76247630. !1 Wilson, D.B., Bross, J.E,, Hofmann, S.L. and Majerus, P.W. 119841J. Biochem, Chem. 259, 11718-11724. 12 Pletcher, C.H, and Nelsestuen. G.L.II983) J. Biol. Chem. 258, 1086-1091. 13 Meldrum, E., Katan, M. and Parker, P. 11989) Eur. J. Biochem. 182, 673-677. 14 Katan, M. and Parker. P.J. (1987j Eur. J. Biochem. 168, 413-418. 15 Parikh, I., March, S. and Cuatrecasas, P. 119741 Methods Enzytool. 34, 77-102. 16 Laemmli, U.K. 11970) Nature 227, 680-685. 17 Raaflaub, J. 119601 Methods Biochem. Anal. 3. 301-325. 18 Huang. C. (1969) Biochemistry 8, 344-352. 19 Bazzi, M.D. and Nelsestuen, G.L. 11987) Biochemistry 26, !15122. 20 Chen, P.S,, Toribara, T.Y. and Warner, H. (1956) Anal. Chem. 28, 1756-1758. 21 Bazzi, M.D. and Nelsestuen, G.L. (1989b) Biochemistry 28, 93179323. 22 ttummel, J.P. and Dreyer, W.J. 11962) Biochim. Biophys. Acta 63, 530-532.

23 Bazzi, M.D. and Nelsestuen, G.L. (1989a) Biochemistry 28, 35773585. 24 N¢lsestuen, G.L. and Lim. T.K. 11977) Biochemistry 16, 41644171, 25 Bradford, M.M. (19761 Anal. Biochem. 72, 248-254. 26 Fukui, J.. kutz, R.J. and Lowenslein, J.M. (19881 J. Biol. Chem. 263, 177311-17737. 27 Banno, Y., Yu, A., Nakashima, T., tlomma. Y., Takenewa, T. and Nozawa, Y. (19901 Biochem. Biophys. Res. Commun. 167, 396-40 I. 28 Welsch, DJ., Pletcher, C.H. and Nelsestuen, G.L. 119881 Biochemistry 27, 4933-4938. 29 Greip, M.A., Fujikawa, K. and Nelsestuen, G.L. (19861 Biochemistry 25, 6688-6694. 311 Lampe. P.D., Wei. G.J. and Nelsestuen, G.L. 11983) Biochemistry 22, 1594-1599. 31 Schwalbe, R.A.. Dahlhach, B. and Nelsestuen, G.L. 119901 J. Biol. Chem. 265, 21749-21757. 32 Nelsestuen, G.L., Kisiel, W. and DiScipio, R.G. (19781 Biochemistry 17. 2134-2138, 33 Bull, R.K., Jevons, S. and Barton, P.G. 11972) J. Biol. Chem. 247, 2747-2754. 34 Prigent-dachary, J., Lindhoat, T., Boisseau, M.R. and Dufourcq, J. (1989) Eur. J. Biochem. 181,675-677. 35 Inagaki, M., Hagiwara, M., Saitoh. M. and Nidara, It. (1986) FEBS lett. 2112, 277-281. 36 Huang, K.P. and Huang, F.L. 119901J. Biol. Chem. 265, 738-744, 37 Exlon, J.H. 11990) J. Biol. Chem. 265, 1-4. 38 Majerus, P.W., Ross. T.S., Cunningham. T.W., Caldwell, K.K., Jefferson, A.B. and Bansal, V.S. (1990) Cell 63, 459-465, 39 Low, M.G. and Saltiel, A.R. (19881 Science 239, 268-275. 40 Waite, M. (1985)J. Lipid Res. 26, 1379-1388, 41 Banno, Y., Nakashima, S. and Nozawa, Y, (1986) Biochem. Biophys. Res. Commun. 136, 713-721. 42 Low, M.G., Carroll, R.C. and Cox, A.C. (1986) Biochem. J. 237, 139-145. 43 Tompkins, T.A. and Moscarello, M, 11991) J. Biol. Chem. 266, 4228-4236. 44 Bairoch, A. and Cox, J.A. 119911) FEBS len. 269, 454-456. 45 Hayashi, K., Muhleisen M., Probst, W. and Rahmann, H. (1984) Chc~ Phys. Lipid. 34, 317-332. 46 Toner. M, Vai,o, G., McLaughlin, A. and McLaughlin, S. 11988) Biochemistry 27, 7435-7443. 47 Sagawa, N., Bleasdale, J.E. and Di Renzo, G,C. 11983) Biochim. Biophys. Acta 752, 153-161. 48 Behl, B., Sommermeyer, H., Goppelt-Strube. M. and Resch. K. 119881 Biochim. Biophys. Acta 971, 179-188. 49 Plantavid, 3.4~., Rossignol, L., Chap, H. and Douste-Blazy, L. 11986) Biochem. Biophys. Acta. 875, 147-156. 50 Smrcka, A.V., Hepler, J.R., Brown, K,O. and Sternweis, P.C. ( 1991) Science 25 i, 804-807. 51 McAtee, P. and Dawson. G. 11990)J. Biol. Chem. 265, 6788-6793. 52 Exton, J.H. 11988) Rev. Physiol. Biochem. Pharmacol. 111, 117224. 53 Baldassare, J.J. and Fisher, G.J. 119861 Biochem. Biophys. Res. Commun. 137, 801-805. 54 Bazzi, M.D. and Nelsestuen. G,L. 11988) Biochemistry 27, 75897593. 55 Peterson, A.A., Rebecehi, M.J. and McLaughlin, S. (19911 Biophys. J. 59, 509a. 56 Miller, I.R. and Ruysschaert, J.M. 11971)J. Colloid and Interface Science 35, 340-345. 57 Quinn, P.J. and Dawson, R.M.C. (1970)Biochem. J. 116, 671-680.