gtp hydrolysis by adp-ribosylation factor is dependent on ... · 10758 . phospholipids regulate arf...

THE JOURNAL OF B I O ~ I C A L CHEMISTRY Vol. 269, No. 14, Issue of April 8, pp. 10768-10763, 1994 Printed in U.S.A.

GTP Hydrolysis by ADP-ribosylation Factor Is Dependent on Both an ADP-ribosylation Factor GTPase-Activating Protein and Acid Phospholipids*

(Received for publication, December 7, 1993, and in revised form, February 1, 1994)

Paul A. Randazzo and Richard A. Kahn From the Laboratory of Biological Chemistry, Developmental Therapeutics Program, Division of Cancer Deatment, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892

ADP-ribosylation factor (ARF) is a 91-kDa GTP binding protein that regulates eukaryotic membrane traffic. Both the binding and hydrolysis of GTP by ARF have been shown to be necessary for this function. However, purified mammalian ARI? lacks intrinsic GTPase activity (<0.0015 min-'1. We document the presence, in bovine brain extracts, of a protein with the predicted proper- ties for an ARF GTPase-activating protein (ARF GAP). This activity was highly dependent on phospholipids. An acid phospholipid fraction from bovine brain (containing primarily phosphatidylinositol 4,s-bisphosphate (PIP,), phosphatidylinositol 4-phosphate, phosphatidylinositol, and phosphatidylserine) had no effect on intrinsic GTPase activity of purified ARF but increased the ARF GAP activity of bovine brain homogenates about 8-fold. This dependence on acid phospholipids was retained after >100-fold purification of ARF GAP, making it, likely, an inherent property of this reaction. PIP, alone stimulated ARF GAP activity up to 30-fold with a half-maximal effect at 100300 but had no effect on the GTPase rate of ARF alone. Phosphatidyl- inositol 4-phosphate was also active but had only 50% of the maximal effect and twice the EC, of PIP,. Phos- phatidylserine, phosphatidylethanolamine, phosphatidylcholine, phosphatidylinositol, and diacylglycerol either alone or in the presence of A R F GAP do not stimulate ARF GTPase activity.

ARF proteins have been identified recently as regula- tors of phospholipase D. The product of the phospholipase D reaction, phosphatidic acid, stimulated ARF GAP approximately 6-fold and reduced the PIP, concentration needed for GAP stimulation about &fold. The substrate of phospholipase D, phosphatidylcholine, inhibited ARF GAP activity, but this inhibition seen with phosphatidylcholine was partially reversed by phosphatidic acid. Afeedback loop for the coordinate regulation of phospholipase D and ARF activities is proposed.

ADP-ribosylation factors (ARF)' are a family of GTP binding proteins, including both the ARF and ARF-like proteins (1-3). The first ARF was identified and purified on the basis of its activity as a cofactor for the cholera toxin-catalyzed ADP ribo-

payment of page charges. This article must therefore be hereby marked * The costs of publication of this article were defrayed in part by the

"advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

GTPase-activating protein; PIP,, phosphatidylinositol 4,5-bisphos- The abbreviations used are: ARF, ADP-ribosylation factor; GAP,

phate; PIP, phosphatidylinositol 4-phosphate; PI, phosphatidylinositol; PS, phosphatidylserine; PC, phosphatidylcholine; PA, phosphatidic acid; DMPC, L-a-dimyristoylphosphatidylcholine; MOPS, 4-mor- pholineethanesulfonic acid.

sylation of G, (4, 5). Members of the ARF family have been identified in every eukaryotic cell examined and have been found to be essential genes in the yeast, Saccharomyces cerevisiae (ARFl andARF2; 6) and in Drosophila melanogaster (ad; 7). The physiological role of at least some ARF proteins appears to be the regulation of membrane traffic (reviewed in Refs. 3 and 8).

The ability of ARF proteins to regulate specific steps in the exocytic or endocytic pathways is intimately associated with its activation state, which in turn is a direct result of its ability to bind and hydrolyze guanine nucleotides. ARFs possess a number of membrane transport-related activities, identified through the use of in vitro protein transport and related assays. Many of these assays, including those designed to measure endoplasmic reticulum-Golgi transport (9), intra-Golgi transport (10, ll), endosome-endosome fusion (12, 131, and nuclear vesicle fusion (14, 151, have been found to be blocked by the addition of GTPyS when cytosol was present. Indeed, in two of these cases (intra-Golgi transport (11) and nuclear vesicle fusion (15)) the cytosolic factor that confers sensitivity to GTPyS has been purified and shown to be ARF. Data from live cells that support the proposed roles for ARF in the functioning of the endoplasmic reticulum, Golgi, and endosomes have come from the regulated expression of a GTPase-deficient mutant of ARFl in cultured normal rat kidney cells (16). The expression of the (Q71L)ARFl mutant, homologous to the oncogenic (Q61L)ras p21 mutant, was shown to correlate with the loss of Golgi stacks, the expansion of the endoplasmic reticulum lu- men, and the inhibition of fluid phase endocytosis. Continued expression of this allele was lethal to cells. These findings not only reveal likely important cellular roles for ARF at these organelles but also document the importance of GTP hydrolysis to terminate an ARF signal that otherwise is lethal to cells.

If purified mammalian ARFl or ARF3 has GTPase activity, it is below the lower limit of detection of conventional GTPase assays, typically <0.0015 min". Given the absence of intrinsic GTPase activity by purified ARFlp and ARF3p, the importance of GTP hydrolysis by ARF in in vitro transport assays and live cells and the precedent established with p21 ras, the existence in cells of a factor that is capable of stimulating the intrinsic GTPase activity of ARF, is predicted (5). We cannot predict whether such a GTPase-activating protein (GAP) serves as an effector, as has been proposed for phospholipase C and G, (171, or is purely inhibitory, as has been proposed for ras GAP and the IRA1 and IRA2 gene products in yeast (18-21). The nature of an ARF GAP and its enzymatic characterization are likely to have important implications to our understanding of the molecular mechanisms involved in the regulation of membrane transport and phospholipase D activity (8, 22).

In this paper we document the existence of an ARF GAP activity in bovine brain extracts. Attempts to purify this activity were hampered by unexplained variability in activity until

10758

Phospholipids Regulate ARF GAP 10759

it was determined that ARF GAP activity is highly dependent on acid phospholipids. The lipid requirements for nucleotide exchange on ARF and ARF GAP activity were investigated and dissected to reveal a specificity of ARF GAP activity for PIP, and PA. This complex interrelationship between ARF activities and specific phospholipids is speculated to have specific physi- ologic roles in cell regulation.

EXPERIMENTAL PROCEDURES Materials-Recombinant ARFl and recombinant myristoylated

ARFl (myrARF1) were prepared as described (23,24). ARF was purified from bovine brain as described (5) and is a mixture of primarily ARFl and ARF3. PI-specific phospholipase C, purified from squid axons as described: was the generous gift of John K. Northup (National Institute of Mental Health, National Institutes of Health). Recombinant p21H" was a generous gift from Alex Papageorge and Doug Lowy (National Cancer Institute) (26). [a-32PlGTP and [36S]GTPyS were purchased from DuPont NEN. L-a-Dimyristoylphosphatidylcholine, an acid phospholipid fraction from bovine brain (phosphoinositides, catalog no. P-6023), phosphatidylinositol from bovine liver, phosphatidylinositol 4-phosphate from bovine brain, phosphatidylinositol 4,5-bisphosphate from bovine brain, phosphatidylserine from bovine brain, phosphatidylcholine from bovine brain, phosphatidylethanolamine from bovine brain, phosphatidic acid prepared from egg, synthetic 1-stearoyl-2-arachi- donoylphosphatidic acid, synthetic 1-stearoyl-2-arachidonoyldiacylglyc- erol, Triton X-100, trypsin, soybean trypsin inhibitor, and GTP were purchased from Sigma. GTPyS, pyruvate kinase, and phospho(eno1)- pyruvate were obtained from Boehringer-Mannheim. ARF GAP Assay-ARF GAP activity was determined by a modifica-

tion of the assay described by Xu et al. (27) for ras GAP in which a single round of GTP hydrolysis was measured. ARF, at a concentration of 1-3 p, was loaded with [a-32PlGTP by incubation for 30 min in 25 nm HEPES pH 7.4,lOO nm NaC1,2.5 nm MgCl,, 1 nm EDTA, 1.5 m~ (0.1%) Triton X-100, 1.25 unitdml pyruvate kinase, 3 n m phospho(eno1)pyruvate, 25 n m KCl, 1 nm dithiothreitol, and 0.3 p [a-32PlGTP. Of the labeled nucleotide that bound ARF, greater than 80% and usually greater than 90% was GTP. The concentration of ARF.GTP formed was typically 20-40 n~ when using recombinant proteins and 50-100 n~ when using ARF purified from bovine brain. The preloaded ARF was diluted 1:s into a reaction solution containing 25 nm HEPES pH 7.4,2.5 nm MgCl,, 100 nm NaCl, 1 nm GTP, 1 m~ dithiothreitol, 1.5 nm Triton X-100, and other additions as indicated. Reactions were initiated by the addition of ARF.[a-32PlGTP and terminated after 0 and 5 min (unless otherwise indicated) by diluting the samples into ice-cold 10 nm Tris, pH 8, 100 nm NaCl, 10 nm MgCl,, 1 nm dithiothreitol. Protein-bound nucleotide was trapped on nitrocellulose filters as described (24), and the nucleotide was then extracted from the filters into 1 ml of 2 M formic acid. This treatment was shown to cause the release of 95100% of the filter-bound radiolabel without changing the nature of the radiolabeled nucleotide. Nucleotides were fractionated by chromatography on poly- ethyleneimine cellulose plates developed in 1 M LiC1, 1 M formic acid. 3ZP-containing nucleotides were subsequently visualized and quantified using a PhosphorImager (Molecular Dynamics). When ARF was ex- cluded from the loading reaction, [a-3?lGTP recovered from the nitrocellulose filters was less than 0.2% of that recovered when ARF was included. Hence, trapping the radiolabeled GTP on filters is a faithful means of monitoring the nucleotides bound to ARF. Less than 10% of the protein-bound GTP dissociated during the ARF GAP reaction, and the inclusion of unlabeled GTP in the GAP reaction eliminates the possibility of free [U-~~PIGXP binding to any protein during the ARF GAP assay. Conversely, a t least 90% of the protein-bound GTP lost could be accounted for by an increase in protein-bound GDP.

For the greatest experimental reliability for a single round reaction, assays were performed such that 2575% of the ARF-bound GTP was hydrolyzed to GDP. With this extent of substrate hydrolysis, activity is not linearly related to product formation or substrate loss. We, therefore, used an integrated rate equation to compare activity under different conditions. Samples were taken at 0 and 5 min for the typical experiment. As the substrate concentration was much less than the Kd for the substrate, the integrated rate equation is approximated by (ln(SdS)) = V,,,;t/K,, where So is the initial substrate concentration and S is the substrate concentration at time = t. Activity expressed as ln(SdS) was linearly related to the amount of protein added to the GAP reaction and to the time of incubation up to 10 min. Activity is reported

J. Mitchell, J. Gutierrez, and J. Northup, submitted for publication.

as ln(S,,/S) min" pg of protein". For time courses, GDP formed is expressed as a fraction of ARF.GTP present at time = 0 (GDPIGTP,). The variation between duplicate samples was less than 10% for activity expressed either way.

Partial Purifiation of ARF GAP-Bovine brains were homogenized in 4 volumes of 10 n m Tris, pH 8.0,10% sucrose at 4 "C. All subsequent steps were carried out at 2-8 "C. The homogenate was filtered through cheesecloth and centrifuged at 20,000 x g for 60 min. The particulate fraction was resuspended in homogenization buffer and centrifugation was repeated. The pellet from this second centrifugation, referred to as P20, was extracted with 10 volumes of 750 m~ NaCl in 10 nm Tris, pH 8 , l nm MgCI,, 1 nm dithiothreitol. The extract was dialyzed against 10 volumes of 10 nm Tris, pH 8, 25 nm NaCl, 1 nm MgCl,, 1 m~ dithiothreitol and applied to a DEAE-Sephacel column, equilibrated with the same buffer. The column was developed in a 25300 nm NaCl gradient. The peak fractions were pooled and applied to a hydroxylapatite column that was developed in a 10-300 m~ K P , gradient. Peak fractions were pooled. Specific activity was increased 80-300-fold over that of the brain extract. Chromatography ofARF GAP on DEAE-Sephacel and hydroxylapatite has been performed 3 times with very similar results. A single peak of AFtF GAP activity was observed in each case.

Miscellaneous-PIP, was hydrolyzed with PI-specific phospholipase C by incubating 15 pg of PIP, with 2 pg of squid PI-specific phospholipase C in 10 nm MOPS, pH 7.0, 100 m~ NaC1, 2.5 nm MgCl,, 500 p~ CaCl,, 500 p EGTA, 1 nm dithiothreitol, and 0.1% Triton X-100 in a volume of 50 pl for 30 min at 37 "C. Prior to use in the ARF GAP assay, the lipid mixture was adjusted to pH 7.5, and the EGTA concentration was adjusted to 1 n m .

Proteins extracted from the P20 fraction (50 pg) with high salt were treated with either 500 ng of trypsin or 500 ng of trypsin with 2.5 pg of soybean trypsin inhibitor in 25 m~ HEPES pH 7.4, 750 nm NaCl, and 2.5 nm MgC1, in a total volume of 50 pl for 30 min at 37 "C. Soybean trypsin inhibitor was then added to samples that lacked it.

[35SIGTPyS binding to ARF was determined by the nitrocellulose filter assay as described with the no-protein control subtracted as back- ground (24).

Protein was determined by Amido Black assay (28) using bovine serum albumin as a standard.

RESULTS

When assaying ARF GAP, some phospholipid andor detergent must be present as A R F l and ARF3 require phospholipids andor detergents both for nucleotide exchange and for stabili- zation of the ARF.GTP formed. Previously, this requirement has usually been satisfied by the inclusion of mixed micelles containing (--1:l) dimyristoylphosphatidylcholine (DMPC) and sodium cholate. In early studies examining bovine brain homogenates for ARF GAP activity, little activity was seen when DMPC was included in the loading reaction and even less was seen when DMPCIcholate was included in the GAP assay to stabilize ARF-GTP (not shown). To test the possibility that DMPC may be interfering with the ARF GAP assay, we looked for an alternate means for loading ARF with radionucleotide. Triton X-100 was found to allow ARF to bind GTP$ to the same extent and at similar rates as did DMPCkholate (data not shown) and was used in subsequent studies for both loading ARF with GTP and stabilizing the substrate ARF'sGTP.

Bovine Brain Homogenates Contain a Phospholipid-dependent ARF GAP Activity-Purified mammalian ARFl or ARF3 lack GTPase activity, whether measured in the presence of DMPC and cholate or in Triton X-100. However, when a crude extract from bovine brain was included in the assay, a small but reliable increase in GTP hydrolysis was observed. This activity was observed only when the substrate, ARF.GTP, was prepared in Triton X-100 and not when it was assayed in the presence of DMPCkholate. Attempts to enrich for this activity by column chromatography on DEAE-Sephacel resulted in a high degree of variability between experiments. The sensitivity of the ARF GTPase activity to DMPC prompted an examination of the effects of other phospholipids. Strong cooperative effects were observed between the activity detected in crude brain homogenates and a commercially available acid phospholipid fraction

10760 Phospholipids Regulate ARF GAP TABLE I

Effect of acid phospholipids on ARF GAP activity in bovine brain ARF GAP activity was determined and is expressed as In(SdS) as

described under “Experimental Procedures.” In experiment 1, hydrolysis of ARF-bound GTP was monitored in the ARF GAP assay either alone or in the presence of 1.2 mg/ml particulate fraction from bovine brain (P20), 1 mg/ml crude acid phospholipids (phosphoinositides), or both as indicated. In experiment 2, salt-extracted ARF GTPase activator was prepared as described under “Experimental Procedures.” The ARF GAP activity was determined in the salt extract (SE; 80 pg/ml) of bovine brain membranes either alone or after the extract was treated with trypsin, with inactivated trypsin (trypsidsoybean trypsin inhibitor (STI)), or by thermal denaturation (95 “C). Details of the salt ex- traction and trypsin and thermal treatments are described under “Ex- perimental Procedures.” In experiment 3, hydrolysis of GTP bound to ARF was determined in the presence of either PIP, (250 p ~ ) , partially purified ARF GAP (43 pg/ml), ARF GAP plus PIP,, ARF GAP plus PIP, treated with squid axon phosphoinositide-specific phospholipase C (PLC), or ARF GAP plus PIP, treated with boiled squid axon phosphoinositide-specific phospholipase C (denatured PLC). PLC digestion of PIP, was performed as described under “Experimental Procedures.”

ARF GAP activity

P20 0.01 0.05

Acid phospholipids 0.01 P20 + acid phospholipids

2 0.38

No addition SE

0.0 0.48

Trypsin-treated SE 0.04 TrypsidSTI-treated SE SE (95 “C)

0.45 0.04

GAP 0.01 0.03

GAP + PIP, 0.74 GAP + PIP, treated with PLC 0.05 GAP + PIP, treated with 0.74

Experiment Condition

1 No addition

3 PIP,

denatured PLC

from bovine brain, enriched in PIP,, PIP, PI, and PS. No intrinsic GTPase activity was detected either in the presence or absence of the added acid phospholipids (Table I). The (low) level of ARF GTPase activity detected in bovine brain homogenates incubated in the absence of added phospholipids (Table I) was increased 5-10-fold by the inclusion of acid phospholipids (1 mg/ml) enriched in PIP, (Table I). Phosphoinositides present as liposomes or as mixed micelles with either Triton X-100 (1.5 mM) or sodium cholate (2 mM) had the same effect on the ARF GTPase stimulating activity of brain homogenate, whereas sodium cholate or Triton X-100 alone had no effect (not shown). Because of the highly reproducible solubilization of phospholipids in Triton X-100, in subsequent studies, lipids were usually added to the incubations as lipidldetergent micelles.

Fractionation of the brain homogenate revealed that ARF GTPase stimulator activity was present in both soluble (20- 40%) and particulate fractions (6040%). The particulate activity could be largely (7040%) extracted with 0.75 M NaCl. The specific activity of the salt-extracted material was 7-10-fold enriched over that of the homogenate. The ARF GTPase activator in the salt extract was thermolabile, with greater than 90% of the activity lost after a 5-min incubation at 95 “C (Table I). The activity was also protease-sensitive as trypsin treatment resulted in the loss of 90% of the activity while the same incubation in the presence of soybean trypsin inhibitor resulted in no such loss of activity (Table I). This activity will be referred to as ARF GTPase-activating protein, or ARF GAP. Salt-extracted ARF GAP activity was further purified by sequential chromatographies on DEAE-Sephacel and hydroxylapatite. The activity eluted as a single peak from each column (not shown). The resolved ARF GAP had a 80-300-fold higher specific activity than the starting material. ARF GAP was inhibited by ARFlpeGTPyS (Kj = 2 p ~ ) but not by ARFlp-GDP (Kj >

- 0.20 cn 1 I c .- E 0.15 - [PIP,] IpMI 0 125 250 500

v)

v)

- \ 0

Y

5 0.10 - I

z c .- .- c >

a a a c9 0.00 -I

0 250 500 750 1000 1250 1500

[Phospholipid] (pM) hc. 1. Concentration dependence of added phospholipids on

ARF GAP activity. ARF GAP activity in 43 pg/ml partially purified ARF GAP was determined in the presence of the indicated concentrations of PIP, (filled circle), PIP (filled square), PI (open square), PA (filled triangle), PS (open circle), diacylglycerol (filled inverted triangle), or PC (open triangle). GTPase activity was determined and expressed as In(SdS) min” pg of protein”, as described under “Experi- mental Procedures.” Data are representative of 10 experiments with PIP, and three experiments with the other phospholipids. Inset, phos- phorimage of the protein-bound guanine nucleotides from incubations with 86 pg/ml ARF GAP and the indicated concentrations of PIP,.

60 p ~ ) . Extensive tests of specificity of the ARF GAP for different low molecular weight GTP binding proteins were not performed, but under these conditions the resolved ARF GAP preparation had no effect on the GTPase activity of purified ras p21 (not shown).

PIP2 Is a Potent and Specific Activator of ARF GAP- Chromatography on DEAE and hydroxylapatite is an efficient means of resolving endogenous acid phospholipids from ARF GAP activity. The ARF GAP resolved by these two steps had a greater dependence on phospholipids than did the activity in the bovine brain homogenate. The specificity of the phospholipid stimulation of ARF GAP activity was examined using this resolved preparation. PIP, alone stimulated the ARF GAP activity approximately 20-fold (Fig. 1, filled circles). The EC,, was between 100 and 300 p~ in 10 different experiments, and the maximal stimulation was observed at PIP, concentrations of 250-500 p ~ . At higher concentrations of PIP,, the stimulatory effect was consistently seen to diminish. The explanation or significance of this decrease is not known. PIP was also found to be a good activator although to only half the extent of PIP, and with only half the potency (Fig. 1, filled squares). PA, prepared from either egg phosphatidylcholine (PC) (Fig. 1, filled triangles) or synthetic 1-stearoyl-2-arachidonoyl-PA (not shown), increased the ARF GAP activity about 5-fold at the highest concentration tested (1.4 mM) without having achieved a clear maximal effect. Diacylglycerol (Fig. 1, filled inverted triangles), PS (Fig. 1, open circles), or PI (Fig. 1, open squares) had little, if any, effect on ARF GAP activities. PC (Fig. 1) decreased the small amount of activity observed in the absence of PIP, or PA.

In the presence of both ARF GAP and either PIP, or PA, ARF hydrolyzed GTP with first-order kinetics after a 1-2-min lag (see Fig. 2). If either ARF GAP or phospholipid were absent, little or no GTP hydrolysis was detected during the 5-min incubation (Fig. 2, A and B ). If the reactants are incubated for 5 min at 30 “C before lipid is added, no GTP hydrolysis was observed until the lipid was added, and the rate of hydrolysis was then found to be nearly identical to that observed when lipid was added at time 0. Thus, phospholipids are required for


b I / I

0 2 4 6 0 10

n 0.05

0.00 0 2 4 6 0 10

Time (mid FIG. 2. Time course of GTP hydrolysis by ARF in the presence

of ARF GAP and PIP, (A) or PA (B) . The ARF GAP assays were performed in a total volume of 100 pl with the indicated additions. At the indicated times, 12-pl samples were taken, and the products were

pressed as the fraction of GTP bound to ARF at time 0 (GTP,) hydro- analyzed as described under “Experimental Procedures.” Data are ex-

lyzed to GDP during the incubation (GDPIGTP,). A, PIP,. Additions were none (open circles), PIP, (250 p, open squares), partially purified ARF GAP (43 pg/d, closed circles), or ARF GAP plus PIP, (closed squares). In addition, ARF.[a-32P]GTP was incubated with ARF GAP for 5 min prior to the addition of PIP, (250 p, triangles) as indicated by the arrowhead. B, PA. Additions are none (open circles), l-stearoyl-2- arachidonoyl-PA(1.4 m ~ , open squares), partially purified ARF GAP (86 pg/ml, closed circles) and PA plus ARF GAP (closed squares). As in A, ARF+(r-3ZPlGTP was incubated with ARF GAP for 5 min prior to the addition of PA (triangles) as indicated by the arrowhead.

ARF GAP activity rather than simply preventing the rapid denaturation of one or more of the reactants.

The identity of PIP, as the activator of the ARF GAP reaction was confirmed using purified PI-specific phospholipase C from squid giant axon. Treatment of the PIP, preparation with this PI-specific phospholipase C for 30 min at 37 “C resulted in the loss of 90% of the ARF GAP stimulatory activity (see Table I). Thermal denaturation of the PI-specific phospholipase C pre- vented loss of activity. These data also indicate that inositol phosphates, products of the PI-specific phospholipase C reaction, are not activators and confirm that PIP, is the specific activator in the ARF GAP-stimulated hydrolysis of GTP bound to ARF.

Effects on ARF GAP Activity of PA, PC, and PIP2-When the effects of individual phospholipids were studied, PIP, had the greatest effect on ARF GAP activity. PA (430 or 720 p ~ ) , which modestly stimulated ARF GAP by itself (see above; Fig. 11, increased the sensitivity of ARF GAP to PIP, @-fold. In the experiment shown in Fig. 3, 720 p~ PA lowered the EC,, for PIP, from 150 to 25 p ~ . PA also appeared to decrease the maximal rate of hydrolysis achievable in the presence of PIP,,

0 100 200 300 400 500

[PIP21 (PMI FIG. 3. Effect of PA, PS, and PC on PIP,-stimulated ARF GAP

activity. The activity of partially purified ARF GAP (43 pg/ml) was determined in the presence of the indicated concentration of PIP, and either no further additions (circles), 1-stearoyl-2-arachidonoyl-PA (720 p, squares), PS (680 p, triangles), or PC (700 p, inverted triangles). Data are representative of three experiments.

although this effect was small (5-30% in five experiments). Neither up to 620 p~ PS (Fig. 3) nor PI (not shown) had any effect on the concentration dependence for the stimulation of ARF GAP by PIP,.

The stimulations of ARF GAP activities by PIP, and PA were additive. In the absence of PIP,, the maximal concentration of PA tested (720 p ~ ) increased GAP activity about 7-fold (Fig. 4, open circles). In the same assay, PIP, (50 p ~ ) alone stimulated ARF GAP activity 6-fold. However, when added together 720 p~ PA and 50 p~ PIP, stimulated the activity 30-fold. The effect of PA on PIP,-stimulated ARF GAP activity was concentration- dependent (Fig. 4, open squares). Combinations of PIP, (50 p ~ ) with either PS or phosphatidylethanolamine, at concentrations up to 1.2 mM, had the same effect on ARF GAP as did PIP, alone. As seen in Fig. 3, PC (700 p ~ ) caused a small but reproduc-

ible inhibition of ARF GAP activity observed over a wide range of PIP, concentrations. This observation was explored in more detail by examining the dependence of PIP,-stimulated ARF GAP on the PC concentration. In the presence of 250 p~ PIP,, PC inhibited the activity in a concentration-dependent manner with an EC,, of approximately 1 m~ (Fig. 5). At the highest concentration tested (2.8 m ~ ) , PC inhibited the ARF GAP activity measured by 85%. The extent of the inhibition observed with PC varied with PIP, concentration. In 700 p~ PC the ARF GAP activity observed in the presence of 100 and 50 p~ PIP, was decreased by 75 and 90%, respectively. Similar results were obtained with bovine brain PC or purified DMPC. The effect of DMPC was independent of detergent. Similar degrees of inhibition by DMPC were observed when the assay was performed in the presence of an acid phospholipid fraction in the absence or presence of 2.5 m~ sodium cholate (not shown). Notably, the previous standard binding mixture for AFW protein included both 3 m~ DMPC and 2.5 m~ sodium cholate.

PC did not change ARF GAP activity observed in the presence of PA. The additivity of PA and PIP, was also retained in the presence of PC, although it was diminished. As shown in Fig. 4, PC (700 p ~ ) had no effect on the PA-stimulated activity (compare open and closed circles). While the maximal ARF GAP activity observed in the presence of PC was 55% less than in its absence, the interdependence of ARF GAP on PIP, and PA was still observed (closed squares).

10762 Phospholipids

0 150 300 450 600 750

[PA1 ipM1 FIG. 4. Effect of PIP, on the stimulation ofAW GAP by PA. ARF

GAP activity of partially purified ARF GAP (43 pg/ml) with the indicated concentration of PA was determined with either no further addition (open circles), PC (700 p ~ , closed circles), PIP, (50 p ~ , open squares), or PIP, plus PC (closed squares). The data are representative of two experiments in the presence of PC and four experiments in absence of PC.

0.12 I

Regulate ARF GAP

0.00 0 500 1000 1500 2000 2500 3000

[PC] ipM1

FIG. 5. Effect of PC on PIP,-stimulated ARF GAP activity. Par- tially purified ARF GAP (43 &ml) was incubated with 250 p~ PIP, and the indicated concentration of PC. The assay was performed as described under “Experimental Procedures.” Data are representative of three experiments.

DISCUSSION We have documented the existence in bovine brain extracts

of a protein capable of stimulating ARF proteins to hydrolyze GTP. ARF GAP activity was found to be highly dependent on PIP, and sensitive to both PA and PC. These specific phospholipid sensitivities are likely to explain the previous failures to observe ARF GAP activity and illustrate the critical nature of the interaction between ARF proteins and the lipid bilayer. The data presented above demonstrate extensive sensitivities of ARF activities to the lipid components of membrane. The recent identification of ARF as a regulator of phospholipase D activity (22) illustrates in turn that ARF is capable of modu- lating the nature of the phospholipids in membranes. Together, it appears that ARF proteins may be a key component in the dynamic two-directional signaling between the constituents of a lipid bilayer and proteins acting in or on that membrane.

The existence of an ARF GAP was first suggested when purified ARF proteins were found to lack any detectable GTPase activity (5). However, determination of the protein sequence of mammalian ARF proteins led to the observation that ARFs are as structurally related to the heterotrimeric G protein a sub- units as they are to members of the Ras superfamily (29). As these two families of proteins employ distinct mechanisms for

regulating the rates of GTP hydrolysis, it was unclear which mechanism was employed by ARF. With the demonstration of the existence of an ARF GAP it is clear that ARF proteins are more ras-like in this regard. However, ARF is quite different from ras p21 in having available a number of in vitro assays for activity. We are now in a position to quickly determine if ARF GAP serves as an effector for the ARF signaling pathway or as simply a down regulator to terminate an ARF signal.

How is the finding that ARF GAP activity is highly dependent on PIP, understood within our current knowledge of the biology of ARF proteins? Two independent ARF related activities, ARF GAP and phospholipase D, have now been shown to require PIP, (22). One possibility is that this is a way for the cell to integrate phospholipase C and phospholipase D signaling pathways. For example, activation of phospholipase C by G, may result in the lowering of local PIP, concentrations, which in turn would lead to lowered ARF GAP activity with resultant increases in AFtF activity. Such a scenario is currently only hypothesized but could explain the suggested involvement of both a heterotrimeric G protein and ARF proteins in membrane traffic (30). A reductionist view suggests that the site of PIP, action is likely at the common element in the two reactions, ARF. Preliminary results in our laboratory support the hypoth- esis that PIP, binds directly to ARFs.~ Certainly, the specificity, dose dependence, and different effects of the individual lipids tested indicate that the ARF GAP can be regulated at this level. These data cannot be explained by a “nonspecific” surface charge effect of different lipid head groups. Specific effects of phosphoinositides on protein activity has been documented for numerous other proteins including profilin (311, cofilin (321, gelsolin (33), and p-calpain (34). A role for phosphoinositides in membrane traffic or vesicle targeting has been suggested from independent genetic studies in the yeast S. cerevisiae. One of the original 23 complementation groups of secretion defective genes, secll, has been found to encode a PI transfer protein. Loss of SEC14 resulted in a block in invertase secretion and an accumulation of Golgi-derived vesicles (35, 36). Suppressor analysis of mutants of secll identified several enzymes in the CDP-choline biosynthetic pathway for PC (37). Strains carry- ing two of these suppressors of secl4, sacl-, and bsd2-1, were found to be inositol auxotrophs (38). A role for phosphoinositides in vesicle targeting was suggested by the observation that VPS34, a gene required for protein sorting to vacuoles in S. cerevisiae, was found to encode the catalytic subunit of a PI-3 kinase (39). Together these data suggest that the local regulation of phospholipid metabolism is a critical aspect of membrane traffic that is distinguishable from bulk lipid flow. Al- though a detailed molecular mechanism for the action of PIP, in ARF action or membrane traffic is lacking, the data presented above provide further, albeit indirect, evidence of a role for phosphoinositides in the protein secretory pathway.

Although less dramatic than the PIP, effects, the opposing effects of PA and PC on ARF GAP activity were likely of biological significance in light of the recent evidence that ARF is an activator of phospholipase D (22). The findings that the substrate for phospholipase D, PC, inhibits ARF GAP activity while the product, PA, decreased the requirement for PIP, and thus increased ARF GAP activity, lead us to propose a negative feedback loop mechanism for the regulation of phospholipase D activity. Activation of ARF would lead to increased phospholipase D activity and the production of PA and choline. While the resulting small decrease in PC concentration is less likely to be regulatory for ARF GAP, the fractional change (increase) in PA is proposed to lead to increased ARF GAP activity, hydrolysis of GTP on ARF, and cessation of the phospholipase D

P. A. Randazzo, J. Luftig, and R. A. Kahn, unpublished observation.


activating signal. PA has been found to regulate several other enzymes (40-42) the most recent and notable example being phospholipase C-y (42). In this case, PA was found to decrease the K,,, for PIP, by 10-fold and decrease the Hill coefficient from 2.5 to 1. The data presented offer new insights into potential means of regulating the essential ARF signaling pathway that can now be examined in biological membranes and live cells.

The specificity of the ARF GAP characterized above for different ARF or ARF-like proteins is not yet known. The possibility exists that more than one protein will be found with GAP activity toward ARFl or other ARF proteins, as has been shown for Ras proteins (18, 21, 43, 44). It is clear that the resolved ARF GAP preparation is not a ras GAP, as it lacks such activity. Indeed, ras GAP has been reported to be inhibited by several phospholipids, including PA and PIP, which are stimulatory for ARF GAP (25, 45, 46). Previously characterized GTPase-activating proteins active on other low molecular weight GTP binding proteins lack any phospholipid dependence. Thus, ARF GAP is unlikely to have been characterized previously. Future tests of the specificity of this activity for different ARF and ARF-like proteins will likely contribute to our understanding of the distinct or overlapping roles for the different ARF isoforms expressed in the same cell.

With the identification of second messengers, such as CAMP or Ca2+, and later the CAMP- or Ca2+-dependent kinases, these signaling pathways were established and readily quantifiable. Methodologies for assessing the local concentrations of specific phospholipids in specific intracellular membranes are crude at best and hamper efforts to construct more detailed molecular models for membrane dynamics. Consider the added difficul- ties that must be considered when the second messengers pro- duced, e.g. PA or diacylglycerol, both produce a signal across a membrane but also alter the composition of that membrane. While we do not know the physiological concentration of PIP, at a bud emerging from the endoplasmic reticulum or medial- Golgi cisterna, we can begin to make informed speculations about the forces brought to play in the regulation of that proc- ess by understanding the details of the biochemistry of the purified components in the signaling system responsible. We believe that the results of Brown et al. (22) and those reported here offer exciting new directions in attempts to understand the regulation of membrane dynamics and traffic.

Acknowledgments-We thank John K. Northup, Michael A. Wallace, Takeshi %mi, Alex Brown, Paul C. Sternweis, and Alex Papageorge for helpful discussions and Alex Brown and Paul Sternweis for sharing the

Northup (squid axon phospholipase C) and Douglas Lowy and Alex results concerning phospholipase D prior to publication. John K.

Papageorge ( ~ 2 1 ~ " ) generously provided the purified proteins to assist in this work. The continued support of the Developmental Therapeutics Program in the Division of Cancer Treatment is also gratefully acknowl- edged.

REFERENCES 1. Kahn, R. A. (1993) in GTPases in Biology (Dickey, B. F., and Birnbaumer, L.,

2. Clark, J., Moore, L., Krasinskas, A., Way, J., Battey, J., Tamkun, J., and Kahn,

3. Rosenwald, A. G., and Kahn, R. A. (1994) in GTPase-controlled Molecular

eds) pp. 529-542, Springer-Verlag, Berlin

R. A. (1993) Proc. Natl. Acad. Sci. U. S. A. 90,8952-8956

Machines (Luini, A,, and Corda, D., eds), in press

4. Kahn, R. A,, and Gilman, A. G. (1984) J. Biol. Chem. 259,62264234 5. Kahn, R. A,, and Gilman, A. G. (1986) J. Biol. Chem. 261,7906-7911 6. Steams, T., Kahn, R. A,, Botstein, D., and Hoyt, M. A. (1990) Mol. Cell. Biol. 10,

6690-6699 7. Ta~nkun, J. W., Kahn, R. A., Kissinger, M., Brizuela, B. J., Rulka, C., Scott, M.

8. Kahn, R. A,, Yucel, J. K., and Malhotra, V. (1993) Cell 76, 1045-1048 P., and Kennison, J. A. (1991) Proc. Natl. Acad. Sci. U. S. A. 88,3120-3124

9. Balcb, W. E., Kahn, R. A., and Schwaninger, R. (1992) J. Biol. Chem. 267,

10. Melancon, P., Glick, B. S., Malhotra, V., Weidman, P. J., Serafina, T., Gleason,

11. Taylor, T. C., Kahn, R. A., and Melancon, P. (1992) Cell 70, 69-79 12. Mayorga, L. S., Diaz, R., Colombo, M. I., and Stahl, P. D. (1989) Cell Regul. 1,

13. Lenhard, J. M., Kahn, R. A,, and Stahl, P. D. (1992) J. Biol. Chem. 267,

14. Boman, A. L., Delannoy, M. R., and Wilson, K. L. (1992) J. Cell Biol. 116,

15. Boman, A. L., Taylor, T. C., Melancon, P., and Wilson, K L. (1992) Nature 358,

16. Zhang, C.J . , Rosenwald, A. G., Willingham, M. C., Skuntz, S., Clark, J., and

17. Berstein, G., Blank, J. L., Jhon, D.-Y., Exton, J. H., Rhee, S . G., and Ross, E.

18. Ballester, R., Marchuk, D., Boguski, M., Saulino, A., Letcher, R., Wigler, M.,

19. Tanaka, K., Nakafuku, M., Satoh, T., Marshall, M. S., Gibbs, J., Mataumoto,

20. Tanaka, K, Nakafuku, M., Tamanoi, F., Kaziro, Y., Matsumoto, K, and Toh-e,

22. Brown, H. A,, Gutowski, S., Moomaw, C. R., Slaughter, C., and Stemweis, P. C. 21. Lowy, D. R., and Willumsen, B. M. (1993) Annu. Reu. Biochem. 62,851-891

23. Weiss, O., Holden, J., Rulka, C., and Kahn, R. A. (1989) J. Biol. Chon. 264,

24. Randazzo, P. A,, Weiss, O., and Kahn, R. A. (1992) Methods Enzymol. 219,

25. Yu, C.-L., Tsai, M.-H., and Stacey, D. W. (1990) Mol. Cell. Biol. 10,66834689 26. DeClue, J. E., Stone, J. C., Blanchard, R. A,, Papageorge, A. G., Martin, P.,

Zhang, K. E., and Lowy, K. R. (1991) Mol. Cell. Biol. 11,31324138 27. Xu, G., Lin, B., Tanaka, IC, Dunn, D., Wood, D., Gesteland, R., White, R.,

Weiss, R., and Tamanoi, F. (1990) Cell 63, 835441 28. Schafier, W., and Weissman, C. (1973) Anal. Biochem. 56,502-514 29. Sewell, J. L., and Kahn, R. A. (1988) Proc. Natl. Acad. Sci. U. S. A. 85,462CL

30. Donaldson, J. G., Kahn, R. A., Lippincott-Schwartz, J., and Klausner, R. D.

31. Lassing, I., and Lindberg, U. (1985) Nature 314, 472-474 32. Yonezawa, N., Homma, Y., Yahara, I., Sakai, H., and Nishida, E. (1991) J. Biol.

33. Yu, F.-X., Wun, H.-Q., Janmey, P. A,, andYln, H. L. (1992)J. Biol. Chem. 267, 14616-14621

34. Saido, T. C., Shibata, M., Takenawa, T., Murofushi, H., and Suzuki, K. (1992) J. Biol. Chem. 267,24585-24590

35. Bankaitis, V A,, Aitken, J. R., Cleves, A. E., and Dowhan, W. (1990) Nature 347,561562

36. Salama, S. R., Cleves, A. E., Malehom, D. E., Whitters, E. A., and Bankaitis, V. A. (1990) J. Bucteriol. 172,451&4521

37. Cleves, A. E., McGee, T. P., Whitters, E. A,, Champion, K. M., Aitken, J. R., Dowhan, W., Goebl, M., and Bankaitis, V. A. (1991) Cell 64. 789-800

38. Whitters, E. A., Cleves, A. E., McGee, T. P., Skinner, H. B., and Bankaitis, V. A. (1993) J. Cell Biol. 122, 79-94

39. Schu, P. V., Takegawa, K., Fry, M. J., Stack, J. H., Waterlield, M. D., and E m ,

13053-13061

M. L., Orci, L., and Rothman, J. E. (1987) Cell 51, 1053-1062

113-124

13047-13052

281-294

512-514

Kahn, R. A. (1994) J. Cell Biol., in press

M. (1992) Cell 70,411418

and Collins, F. (1990) Cell 63,851-859

K., Kaziro, Y., and Toh-e, A. (1990) Cell SO, 803407

A. (1990) Mol. Cell. Biol. 10, 43034313

(1993) Cell 75,1137-1144

21066-21072

36%369

4624

(1991) Science 254, 1197-1199

Chem. 266, 17218-17221

40. Bocckino, S. B., Wilson, P. B., and Exton, J. H. (1991) Proc. Natl. Acad. Sci. S. D. (1993) Science asO,88-91

41. Moritz,A., DeGraan, P. N. E., Gispen, W. H., and Wirtz, K W. A. (1992) J. Biol.

43. Trahey, M., Way, G., Halenbeck, R., Rubinfeld, B., Martin, G., Ladner, M., 42. Jones, G. A., and Carpenter, G. (1993) J. Biol. Chem. 268,20845-20850

242, 1697-1700 Long, C., Crosier, W., Watt, K., Koths, K., and McCormick, F. (1988) Science.

44. Maekawa, M., Nakamura, S., andHattori, S. (199315. Biol. Chem. 268,22948- 22952

45. Tsai, M.-H., Yu, C.-L., and Stacey, D. W. (1990) Science 260,982-985 46. Tsai, M.-H., Roudebush, M., Dobrowolski, S., Yu, C.-L., Gibbs, J. B., and Sta-

U. S. A. 88, 6210-6213

Chem. 267, 7207-7210

cey, D. W. (1991) Mol. Cell. Biol. ll, 2785-2793

gtp hydrolysis by adp-ribosylation factor is dependent on ... · 10758 . phospholipids regulate arf...

Documents