Kinetic Analysis of a Mammalian Phospholipase D

Abstract

In mammalian cells, phospholipase D activity is tightly regulated by diverse cellular signals, including hormones, neurotransmitters, and growth factors. Multiple signaling pathways converge upon phospholipase D to modulate cellular actions, such as cell growth, shape, and secretion. We examined the kinetics of protein kinase C and G-protein regulation of mammalian phospholipase D1 (PLD1) in order to better understand interactions between PLD1 and its regulators. Activation by Arf-1, RhoA, Rac1, Cdc42, protein kinase Cα, and phosphatidylinositol 4,5-bisphosphate displayed surface dilution kinetics, but these effectors modulated different kinetic parameters. PKCα activation of PLD1 involves N- and C-terminal PLD domains. Rho GTPases were binding activators, enhancing the catalytic efficiency of a purified PLD1 catalytic domain via effects on K_m. Arf-1, a catalytic activator, stimulated PLD1 by enhancing the catalytic constant, k_cat. A kinetic description of PLD1 activation by multiple modulators reveals a mechanism for apparent synergy between activators. Synergy was observed only when PLD1 was simultaneously stimulated by a binding activator and a catalytic activator. Surprisingly, synergistic activation was steeply dependent on phosphatidylinositol 4,5-bisphosphate and phosphatidylcholine. Together, these findings suggest a role for PLD1 as a signaling node, in which integration of convergent signals occurs within discrete locales of the cellular membrane.

Phospholipase D(PLD²; EC 3.1.4.4.) enzymes are phosphodiesterases that hydrolyze phospholipids to phosphatidic acid (PA) and their free head groups. In mammals, the principal substrate is phosphatidylcholine (PC), and the production of PA has broad biological impact (1). PA regulates physical properties of cellular membranes, acts as a second messenger to alter the activity of many enzymes and proteins (2), and can be further metabolized to diacylglycerols and lysophosphatidic acid by lipid phosphate phosphohydrolase and phospholipase A₂, respectively. Diacylglycerols derived from PC are important cellular signaling molecules, and lysophosphatidic acid is released as an intercellular messenger that affects many cell types. Recent evidence has supported a role for phospholipase D in exocytosis, cell proliferation, membrane trafficking, migration, and tumor formation (3, 4).

PLD has been cloned from animals, fungi, plants, and bacteria (5). Two mammalian PLD genes (PLD1 and PLD2) have been identified, and their products occur as several splice variants. The mammalian isozymes have a conserved primary sequence and domain structure but are differently regulated by upstream signals. Both enzymes are members of the PXPH-PLD subfamily that have putative pleckstrin homology (PH) and phox homology (PX) domains in tandem at their N termini (6) and are hypothesized to have pseudodimeric catalytic domains with invariant HXKX₄D motifs (7, 8).

Because of the critical roles of PLD and its products, the enzymatic activity of PLD is tightly regulated by a variety of hormones, neurotransmitters, growth factors, cytokines, and other cellular signals. The PLD1 isozyme is under elaborate control in vitro and in vivo (1). Phosphatidylinositol 4,5-bisphosphate (PIP₂) is an essential PLD1 activator, and most characterized eukaryotic PLD isoforms are regulated by PIP₂. Phosphatidylinositol 3,4,5-trisphosphate can partially substitute for PIP₂, but other acidic phospholipids are ineffective or nearly so (9, 10). Bulk interaction between PLD and lipid vesicles is dependent on PIP₂ in vitro (11–14). Whereas multiple PLD domains bind polyphosphoinositides directly (Fig. 1A), PIP₂ interaction with a conserved polybasic region within a C-terminal PLD catalytic subdomain appears to be responsible for regulation by this lipid (12–14).

FIGURE 1. Purification and activity of rat phospholipase D1b.

A, schematic representation of full-length, rat PLD1b (amino acids 1–1036; top) and rPLD1b.d311 (amino acids 312–1036; bottom). N-terminal His₆ (6H) tags were added to both constructs to facilitate purification. PLD1.d311 retains both catalytic repeats, but PH and PX domains have been replaced with a maltose-binding domain (MBD) from E. coli. Sequences involved in intermolecular interactions are outlined above (13, 16, 20, 42, 43). B, phospholipase D1 protein and recombinant PLD activators were purified as described under “Experimental Procedures.” Samples (1.5 µg of total protein) were analyzed by SDS-PAGE using a 4–20% denaturing gel and stained by Coomassie Blue. Purified proteins were verified by immunoblot (data not shown), and immunoreactive bands are indicated by asterisks where necessary. C, specific activities of full-length (PLD1; left) and truncated (PLD1.d311; right) phospholipase D1 were determined by [³H]phosphatidylcholine hydrolysis in vitro using standard methods (31). Enzymatic activity of purified phospholipase D (5–10 nm) is measured in the presence of 10 µm GTPγS alone (Basal) or reconstituted with GTPγS and maximally effective concentrations of purified activators. Data are presented as mean initial rates ± S.E.

Current work suggests that there are distinct yet interacting binding sites for major regulators. Mutational studies have identified PLD1 domains and amino acid sequences involved in interactions with many PLD1 effectors (Fig. 1A). PLD1 activity is regulated by conventional protein kinase C isozymes (PKCα, -β, and -γ). PKCα activates PLD1 via direct protein-protein interactions between PKC regulatory domains (15) and multiple PLD domains (16, 17). PLD1 is also regulated by members of the Rho and Arf subfamilies of the Ras GTPase superfamily. RhoA activates PLD through direct interaction with a C-terminal PLD catalytic subdomain (18–20). Members of the Arf subfamily enhance PLD activity in vitro (21) and in vivo (22), but a direct interaction has not been demonstrated.

A novel PLD1 expression construct allowed high level expression and efficient purification of highly active enzyme (>100-fold enhanced yields). Sufficient amounts of PLD1 now permit detailed characterization of PLD1 activity in vitro. To understand how allosteric modulators regulate catalysis, we performed kinetic analyses of regulated PLD1 activity. We report that PLD1 effectors enhanced PLD activity differently. PKCα activation of PLD1 involves N- and C-terminal PLD domains. Rho GTPases were binding activators, and Arf-1 was a catalytic activator. When PLD1 is simultaneously stimulated by multiple effectors, the combined response was greater than the sum of the responses to individual effectors (23, 24). A kinetic description of PLD1 activation by allosteric modulators reveals a mechanism for apparent synergy between activators. Consistent with this interpretation, we found that Arf-1, a catalytic activator, was required for synergistic activation of PLD1.

PIP₂ had concentration-dependent effects on membrane association of N-terminally truncated rat PLD1b (amino acids 312–1036) and biphasic effects on PLD activation. Surprisingly, synergy between activators was steeply dependent upon PIP₂ concentration. A narrow range of PIP₂ concentration and a combination of Arf and binding activators are required to produce synergistic activation of PLD1. Cellular locales where these signaling molecules converge on appropriate membranes experience dramatic remodeling of the lipid membrane, hydrolyzing PC to PA.

These studies address open questions about the complex prerequisites for PLD activation. A greater understanding of activation mechanisms may help to pharmacologically dissect signaling pathways that converge upon PLD1.

EXPERIMENTAL PROCEDURES

Materials

The chemicals used for all experiments were of the highest grade. Myristic acid was purchased from Sigma; n-octyl-β-d-glucopyranoside (β-OG) and guanosine GTPγS were from Calbiochem. Phosphatidylethanolamine (PE), PC, and PIP₂ were purchased from Avanti Polar Lipids. [methyl-³H]PC was purchased from PerkinElmer Life Sciences.

Partial Purification of PLD1

The full open reading frame of rat phospholipase D1b (25) in the pBlueBacHis2B (Invitrogen) baculovirus transfer vector was modified to generate an N-terminal FLAG epitope by site-directed mutagenesis as described in the QuikChange™ method (Stratagene). Baculovirus was generated by cotransfection with linearized AcMNPV viral DNA (Invitrogen) in Spodoptera frugiperda Sf21 cells. Recombinant virus was isolated by three rounds of plaque selection and amplified in Sf21 cells adapted to suspension culture in Trichoplusia ni Medium-Formulation Hinks with 10% fetal bovine serum. Plasmid constructs and baculovirus DNA were sequenced to verify coding regions. Adherent Sf21 cultures (7 × 10⁸ cells) were infected at a multiplicity of infection of 1. After 72 h, cells were harvested by centrifugation and resuspended in 10 ml of Lysis Buffer A (50 mm sodium phosphate buffer, pH 7.5, 250 mm NaCl, 15 mm imidazole, 1 mm MgCl₂, 1% (w/v) β-OG, 0.5 mm DTT, 2 mm phenylmethylsulfonyl fluoride, Complete protease inhibitor mixture (Roche Applied Science)). Cells were disrupted by sonication for 6 × 10 s at 6 watts root mean square on ice. Insoluble material was removed by centrifugation (40,000 × g for 30 min at 4 °C), and lysate was further clarified (100,000 × g for 1 h at 4 °C). Recombinant PLD1 protein (124 kDa) was purified over a HiTrap chelating HP nickel iminodiacetic acid column (GE Biosciences) in Buffer A (50 mm sodium phosphate buffer, pH 7.5, 200 mm NaCl, 30 mm imidazole, 1 mm MgCl₂, 1% (w/v) β-OG, 0.5 mm DTT), eluting at 180 mm imidazole in a linear imidazole gradient. Pooled fractions were desalted by gel filtration and stored at −80 °C in 10% glycerol.

Purification of PLD1.d311

Rat PLD1.d311 (amino acids 312–1036) was amplified by PCR from rPLD1b (26) and subcloned into pENTRd-TOPO (Invitrogen) in frame with an N-terminal His₆/maltose-binding domain (malE) fusion protein amplified from pSV282 (provided by Laura Mizoue, Vanderbilt Center for Structural Biology). PLD1b.d311 was transferred to pDEST8 baculovirus expression vectors by homologous recombination using GATEWAY^® methods (Invitrogen). Baculovirus encoding the PLD1.d311 construct was generated in Sf21 insect cells using the Bac-to-Bac^® system (Invitrogen) and screened for optimal expression conditions. Plasmid constructs and baculovirus DNA were sequenced to verify coding regions. Adherent Sf21 cultures (1 × 10⁹ cells) were infected at a multiplicity of infection of 0.1–0.5 in T. ni Medium-Formulation Hinks with 10% fetal bovine serum. After 72 h, cells were harvested by centrifugation and resuspended in 20 ml of Lysis Buffer A. Recombinant PLD1.d311 protein (131 kDa) was purified as described for full-length rPLD1b above. Purified protein eluted from HiTrap chelating HP nickel iminodiacetic acid columns at 170 mm imidazole in a linear imidazole gradient. Pooled fractions were immediately desalted over a Sephadex G-25 Superfine gel filtration column (GE Biosciences) and exchanged to Buffer B (30 mm Hepes, pH 7.5, 150 mm NaCl, 1 mm DTT. Aggregated protein and minor contaminants were resolved from the purified fractions by size exclusion chromatography over a 16/60 Superdex 200-pg column (GE Biosciences). PLD1.d311 (>95% pure) eluted with a retention volume of 68 ml. Pooled fractions were concentrated to >1 mg/ml by ultrafiltration, frozen in 10% glycerol, and stored at −80 °C. Purified PLD retained high enzymatic activity for 2–3 days at 4 °C and for >1 year at −80 °C.

Purification of Myristoyl-ADP-ribosylation Factor-1

Human Arf-1 was coexpressed with human N-myristoyltransferase 1 in BL21(DE3) Escherichia coli. Four-liter cultures were grown in LB broth (150 µg/ml carbenicillin and 50 µg/ml kanamycin) at 37 °C and 250 rpm. At A₆₀₀ = 0.7, myristic acid was added to 50 mg/liter. Expression was induced at A₆₀₀ = 0.9 with 0.5 mm isopropyl-1-thio-β-d-galactopyranoside. Cultures were incubated for an additional 3 h at 27 °C and harvested by centrifugation. Cells were resuspended in 10 ml of lysis buffer (20 mm Tris-Cl, pH 8, 20 mm NaCl, 1 mm MgCl₂, 100 µm GDP, 1 mm DTT, 10 mg/ml lysozyme, 2 mm phenylmethylsulfonyl fluoride, Complete protease inhibitor mixture) and lysed by sonication (6 × 30-s pulses at 6 watts on ice). Insoluble material was removed by centrifugation (40,000 × g for 30 min at 4 °C), and lysate was further clarified (100,000 × g for 1 h at 4 °C). Supernatant was diluted to 50 ml in Buffer C(20 mm Tris-Cl, pH 8, 20 mm NaCl, 1 mm MgCl₂, 1 mm EDTA, 100 µm GDP, 1 mm DTT). Clarified lysate was applied to an 85-ml DEAE-Sepharose FF column (GE Biosciences) and eluted at 110 mm NaCl in a linear NaCl gradient. Active fractions were identified by activation of PLD in vitro and concentrated to 5 ml by ultrafiltration. Recombinant Arf-1 was applied to a 26/60 Superdex 75-pg column (GE Biosciences) in Buffer D (20 mm Tris-Cl, pH 8, 150 mm NaCl, 1 mm MgCl₂, 1 mm DTT). Purified (~95%) Arf-1 eluted with a retention volume of 196 ml. Active fractions were concentrated to 0.5 mg/ml by ultrafiltration, frozen in 5% glycerol, and stored at −80 °C.

Purification of Geranylgeranylated RhoA, Rac1, and Cdc42

Baculoviruses encoding N-terminal His-tagged human RhoA, Rac1, and Cdc42 have been described previously (27). Adherent Sf21 cultures (3 × 10⁸) cells were infected at a multiplicity of infection of >1. After 72 h, cells were harvested by centrifugation and resuspended in Lysis Buffer B (50 mm sodium phosphate buffer pH 7.5, 300 mm NaCl, 5 mm MgCl₂, 1% (w/v) β-OG, 10 µm GDP, 2 mm phenylmethylsulfonyl fluoride, Complete protease inhibitor mixture). Cells were disrupted by sonication for 6 × 10 s at 6 watts on ice. Lysate was clarified by centrifugation at 100,000 × g at 6 °C for 60 min. Recombinant GTPases were purified over 1-ml HiTrap chelating HP nickel iminodiacetic acid columns (GE Biosciences) and eluted with linear imidazole gradients. Active fractions (300–360 mm imidazole) were identified by activation of PLD in vitro and exchanged to a storage buffer containing 25 mm phosphate buffer, pH 7.5, 150 mm NaCl, 5 mm MgCl₂, and 0.5% (w/v) β-OG) over a Sephadex G-25 Superfine gel filtration column (GE Biosciences). Purified (>95%), highly active geranylgeranyl-RhoA, Rac1, and Cdc42 were frozen in 5% glycerol and stored at −80 °C.

Partial Purification of Protein Kinase Cα

Rat PKCα was expressed in baculovirus-infected Sf21 cells and purified essentially as described for PKCβII in Ref. 28.

Binding of PLD1.d311 to Sucrose-loaded Vesicles

The sucrose-loaded vesicle binding assay was adopted, with minor modifications, from the procedure of Buser and McLaughlin (29). Large unilamellar vesicles were prepared by extrusion (Lipex Biomembranes, Inc., Vancouver, Canada) of lipid dispersions through two 0.1 µm polycarbonate filters (Nuclepore). Lipid composition was identical to standard in vitro assay preparations (see below) or modified to replace PIP₂ with PE. Vesicles were loaded with 176 mm sucrose, 50 mm Hepes, pH 7.5, 3 mm EGTA, 3 mm MgCl₂, 3 mm CaCl₂. Sucrose-loaded vesicles were washed and resuspended in an isotonic buffer containing KCl. Sucrose-loaded vesicles were incubated with 10 nm PLD1.d311 for 30 min at 25 °C and sedimented by ultracentrifugation at 100,000 × g for 45 min at 20 °C. PLD1 present in supernatant (PLD_sup) and pellet (PLD_pellet) was estimated by immunoblot, and vesicle-associated PLD1.d311 was calculated according to Equation 1,

$$\mathrm{P}\mathrm{L}{\mathrm{D}}_{\text{vesicle}}=\frac{(\beta )\mathrm{P}\mathrm{L}{\mathrm{D}}_{\text{pellet}}(\beta -1)\mathrm{P}\mathrm{L}{\mathrm{D}}_{\text{sup}}}{\alpha +\beta -1}=\frac{(B_{\text{max}}){[\mathrm{P}\mathrm{L}]}_T}{K_s^A+{[\mathrm{P}\mathrm{L}]}_T}$$ (Eq. 1)

where α is the fraction of sedimented vesicles determined by scintillation counting and β is the fraction of PLD immunoreactivity (horseradish peroxidase-conjugated αFLAG-M2 antibody; Sigma) in the supernatant fraction in the absence of lipid (accounts for PLD that precipitates without lipid). [PL]_T is the total concentration of lipid, and $K_s^A$ is a dissociation constant describing bulk association with lipid.

Measurement of PLD Activity in Vivo

Activity assays were performed as described in Ref. 30. Hemagglutinin-tagged RhoA.G14V was obtained from Guthrie Research Institute (available on the World Wide Web at www.cdna.org). Rat PLD1b and PLD1.d311 were subcloned into pcDNA3.1 (Invitrogen). PLD1 and activators were transiently expressed in COS7 cells by liposome transfection (Fugene6; Roche Applied Science). PLD constructs (0.6 µg of DNA) were cotransfected 1:1 with RhoA.G14V or empty vector as ballast. Transfected cells were serum-starved (18 h) and treated (15 min) with phorbol myristate acetate (100 nm) as indicated. Expression of heterologous proteins was verified by SDS-PAGE and immunoblot.

Measurement of PLD Activity in Vitro

Activity assays were performed with exogenous substrate as described previously (31). Briefly, PLD activity was measured by the release of [methyl-³H]choline from [choline-methyl-³H]dipalmitoyl-PC. 1–10 nm PLD was reconstituted with phospholipid vesicle substrates as described under “Results and Discussion,” typically composed of 10 µm dipalmitoyl-PC, 100 µm PE (bovine liver), 6.2 µm PIP₂ (porcine brain), and 1.4 µm cholesterol. Lipid solutions were dried under a gentle stream of nitrogen and then resuspended in 100 mm Hepes, pH 7.5, 160 mm KCl, 6 mm EGTA, 0.2 mm DTT. Small unilamellar vesicles were prepared by bath sonication (6 × 1-min intervals at 80 watts). All assays were conducted for 30 min at 37 °C in 50 mm Hepes, pH 7.5, 80 mm KCl, 3 mm EGTA, 0.1 mm DTT, 3.6 mm MgCl₂, 3.6 mm CaCl₂, and 10 µm GTPγS. Reactions were stopped by the addition of trichloroacetic acid and bovine serum albumin. Free [methyl-³H]choline was separated from precipitated lipids and proteins by centrifugation and was analyzed by liquid scintillation counting. The enzymatic reactions were linear with time and protein concentration. Initial rates were determined from measurements between 5 and 25% PC hydrolysis. Data are presented as mean initial enzymatic rates ± S.E. (nmol of PC hydrolyzed/min/mg of PLD) measured in 3–12 independent experiments performed in duplicate.

Analysis of Kinetic Data

Apparent dissociation constants and rate constants were determined from best fit parameters by nonlinear regression (sum of squares) using Equation 2 (32).

$${\frac{d[\text{choline}]}{\mathrm{d}t}|}_{(t~0)}=\frac{k_{\text{cat}}{[\mathrm{P}\mathrm{L}]}_T{[\mathrm{P}\mathrm{C}]}_0[\mathrm{P}\mathrm{L}\mathrm{D}]}{K_s^A{K}_m^B+K_m^B{[\mathrm{P}\mathrm{L}]}_T+{[\mathrm{P}\mathrm{L}]}_T{[\mathrm{P}\mathrm{C}]}_0}$$ (Eq. 2)

Concentration-response data were fit to Equation 3, where MAX and MIN refer to maximum and minimum rates, n_H is the Hill coefficient, and [a] represents the concentration of an allosteric effector.

$$\frac{d[\text{choline}]}{\mathrm{d}t}=\frac{\mathit{\text{MIN}}+(\mathit{\text{MAX}}-\mathit{\text{MIN}})}{1+{10}^{(n_{\mathrm{H}}\text{log}\mathrm{E}{\mathrm{C}}_{50}-\text{log}[\mathrm{a}])}}$$ (Eq. 3)

Synergy is expressed as a ratio (κ) of the response to combined activators relative to the responses to individual activators (Equation 4). The degree of activation (ε) (33), describes the increase in PLD catalytic rate due to the effects of individual activators (a and b) or activators in combination (a,b).

$$\kappa =\frac{{\epsilon }_{(a,b)}}{{\epsilon }_{(a)}+{\epsilon }_{(b)}}$$ (Eq. 4)

Calculations were performed using Prism version 4.0 (GraphPad Software).

RESULTS AND DISCUSSION

The present study reports the first detailed kinetic description of regulated phospholipase D activity toward PC substrate. Mammalian PLD activity is regulated by elaborate networks of upstream signaling molecules in vivo, and PLD1 is directly activated by multiple classes of proteins and lipids in vitro (1). We evaluated basic components of this signaling network by reconstituting purified PLD1 with varied amounts of substrate and allosteric modulators. Steady-state kinetic analysis of PLD activity revealed important differences between PLD1 effectors. Activators exploit different properties of PLD catalysis to enhance activity. Distinct modes of activation provide a mechanistic basis for synergistic relationships between PLD1 effectors. A systematic investigation of PLD1 activation revealed several strict requirements for synergy between activators. Synergy requires appropriate lipid composition and particular combinations of protein effectors. These requirements are revealed in each of the kinetic terms that describe PLD1 activation. Phosphoinositides regulate the association of PLD1 and the lipid bilayer. This form of activation is represented by the dissociation constant, $K_s^A$ Synergy also requires modulation of k_cat by Arf-1 and modulation of K_m by PKC or Rho GTPases.

Prior studies have characterized the enzymatic activities of bacterial and plant PLD enzymes in detail (34, 35). The inability to generate sufficient amounts of purified mammalian PLD has hindered the biochemical characterization of this enzyme. A novel PLD1 truncation mutant with greatly enhanced expression and stability was purified to homogeneity from recombinant sources. PLD1.d311 lacks N-terminal domains but retains full enzymatic activity (Fig. 1). Unexpectedly, this catalytic domain fragment was regulated by all classes of PLD allosteric modulators.

Purification of Phospholipase D1

Full-length PLD1b (amino acids 1–1036; Fig. 1A) and an N-terminally truncated PLD1b (amino acids 312–1036) were expressed in baculovirus-infected insect cells and purified as described under “Experimental Procedures.” Both proteins efficiently bound to immobilized nickel columns, but full-length PLD1 failed to purify in a homogeneous manner over subsequent chromatographic steps. N-terminally truncated PLD1 (PLD1.d311) was expressed as a maltose binding domain fusion protein and purified by metal chelation and size exclusion chromatography. Typical yields of purified PLD1.d311 (30–40 mg/liter insect cells) were much greater than those of full-length PLD1 (0.2–0.5 mg/liter insect cells). Protein kinase Cα, RhoA, Rac1, and Cdc42 were purified from baculovirus infected insect cells and recombinant Arf-1 was purified from bacteria. Purified proteins were evaluated by SDS-PAGE followed by colloidal Coomassie staining (Fig. 1B) and immunoblot.

Regulated Activity of Phospholipase D1

To examine the enzymatic properties of purified PLD, we used standard in vitro methods (31) to assay activity toward dipalmitoyl-PC substrate. PLD was reconstituted with purified effectors and lipid vesicles containing radiolabeled substrate. PLD1.d311 retained the full enzymatic activity of full-length PLD1 and was regulated by all classes of PLD activators (Fig. 1C). PLD1.d311 exhibited slightly elevated basal activity in the absence of activators, similar to other N-terminally truncated PLD mutants (17, 36).

Recent work has mapped major sites of interaction with PKCα to the extreme N terminus and PH domain of PLD1 (Fig. 1A). Surprisingly, potent activation by PKCα was still observed when these major PKC binding sites were deleted from the N terminus of PLD1. Maximal PLD1.d311 response to PKCα was 16% of the maximal response of full-length PLD1 to PKCα (Fig. 1 and Table 1). Monomeric G-protein activators were equally effective toward PLD1 and PLD1.d311, both in terms of potency and maximal activation.

TABLE 1. Activation of phospholipase D1 in vitro.

Activity of purified phospholipase D1 (1 nm) was stimulated by purified activators. Concentration-response data were analyzed by nonlinear regression.

Activator(s)	PLD1 (full-length)		PLD1.d311
	EC50	Max	EC50	Max
	nm	nmol/min/mg	nm	nmol/min/mg
Unstimulated		7 ± 1		13 ± 1
PKCα	32 ± 1	204 ± 17	56 ± 22	34 ± 4
Arf-1		>500		>500
RhoA	29 ± 2	44 ± 5	67 ± 43	37 ± 4
Rac1	5 ± 2	17 ± 1	19 ± 19	31 ± 6
Cdc42	9 ± 4	23 ± 2	30 ± 20	23 ± 1
Arf-1 + RhoAa	346 ± 21	346 ± 21
Arf-1 + PKCαa	731 ± 24	731 ± 24
PKCα + RhoAb	14 ± 4	177 ± 9
PKCα + Arf-1b	22 ± 12	716 ± 25

^aConcentrations of Arf-1 were varied, and RhoA or PKCα was held constant at 300 nm.

^bConcentrations of PKCα are varied, RhoA was held constant at 300 nm, and Arf-1 was held constant at 1 µm.

Full-length PLD1 and PLD1.d311 were both strongly activated by Arf-1·GTPγS(Fig.1C). Arf-1 stimulated PLD1 activity in a concentration-dependent manner, and its effects did not saturate at even 10 µm Arf-1 (data not shown). Other activators were more than 30-fold more potent than Arf-1 (Fig. 2A). No other activator equaled the degree of activation elicited by Arf-1. Three members of the Rho GTPase subfamily potently stimulated PLD1 but with low maximal stimulation. GTPγS-loaded RhoA, Rac1, or Cdc42 stimulated PLD1 and PLD1.d311 2–5-fold above basal activity (Table 1). Normal (n_H ≈ 1) concentration-dependent effects were observed for each of the activators tested, although very high (µm) concentrations of PKCα, RhoA, Rac1, or Cdc42 had reduced ability to activate PLD1 (Fig. 2A).

FIGURE 2. Regulated activity of PLD1b.

Increasing concentrations of purified activators were reconstituted with partially purified, full-length phospholipase D1 (5 nm). Phosphatidylcholine hydrolysis was measured under standard conditions described under “Experimental Procedures.” A, Arf-1, PKCα, RhoA, Rac1, and Cdc42 were examined for their ability to activate PLD1. B, simultaneous activation of PLD1 by increasing concentrations of PKCα and maximally effective concentrations of Arf-1 or RhoA. C, synergistic activation of PLD1 by increasing concentrations of Arf-1 and maximally effective concentrations of PKCα or RhoA. Specific activities are presented as mean initial rates ± S.E.

Synergy between PLD1 Activators

At maximally effective concentrations, RhoA and PKCα stimulated PLD1 activity no more than PKCα alone. At lower PKCα concentrations, the combined effects of PKCα and 100 nm RhoA were roughly equal to the sum of the effects of each activator alone (Fig. 2B). The combined effect of PKCα and Arf-1 exceeded the effects of either activator alone, indicating synergy (Fig. 2B). Co-stimulation with Arf-1 enhanced the maximal response to PKCα but did not change PKCα potency toward PLD1. Arf-1 (1 µm) synergized with PKCα to stimulate PLD1 100-fold above basal activity, more than twice the maximal response predicted for an additive effect.

Arf-1 also synergized with RhoA to activate PLD1 (Fig. 2C). Arf-1 (1 µm) and RhoA (100 nm) combined to enhance PLD1 activity almost 50-fold, twice the response predicted for an additive effect. Arf-1 exhibited synergistic relationships with all PLD1 effectors tested. Combinations without Arf-1 did not produce synergistic responses.

PLD1.d311 Activity in Vivo

The N terminus of PLD1 contains several membrane-targeting domains that determine the subcellular localization and trafficking of the enzyme in COS7 cells (13). In the PLD1.d311 mutant, N-terminal PX and palmitoylated PH domains have been removed. Despite lacking these domains, PLD1.d311 was highly active when expressed in COS7 fibroblasts (Fig. 3). Consistent with our findings in vitro, PLD1.d311 activity was stimulated by RhoA in vivo, but its response to PKC was blunted. To examine Rho-dependent stimulation of PLD in cultured cells, full-length PLD1 or PLD1.d311 was transiently coexpressed with a constitutively activated RhoA mutant (RhoA.G14V). Phorbol myristate acetate stimulation of COS7 cells activates PLD1, presumably via regulation of classical PKC isoforms (1). Full-length PLD1 was expressed at levels about one-third of PLD1.d311 expression and exhibited robust stimulation by RhoA.G14V (1.6-fold over control) and by 4β-phorbol 12-myristate 13-acetate (1.6-fold over control).

FIGURE 3. Activity of PLD1 and PLD1.d311 in vivo.

Recombinant phospholipase D was transiently expressed in COS7 cells, and activity was measured in vivo by the transphosphatidylation of radiolabeled endogenous substrate to 1-butanol. PLD constructs (0.6 µg of DNA) were cotransfected 1:1 with RhoA.G14V or empty vector as ballast. Transfected cells were serum-starved (18 h) and treated (15 min) with phorbol myristate acetate (100 nm) as indicated. Phosphatidylbutanol production is presented as mean rates ± S.E. for three independent experiments performed in triplicate. ^**, significant (p < 0.01) increases relative to control (vector + RhoA.V14); ^***, a significant (p < 0.01) increase relative to control (vector + 4β-phorbol 12-myristate 13-acetate (PMA)).

Steady-state Kinetic Analysis of PLD1.d311

A detailed kinetic characterization of PLD1 activity was performed to gain insight into the regulation of the enzyme. It has previously been shown that mammalian PLD preparations obey Michaelis-Menten kinetics toward substrate (37–39), but allosteric modulation of PLD kinetics has not been studied. We report the first use of saturation kinetics to characterize the activation of PLD1 by purified effectors. The ability to generate large quantities of PLD1.d311 permitted us to perform comprehensive enzymatic studies.

Previous studies suggest that many PLD enzymes exhibit interfacial behavior (39–41). Macroscopic Michaelis constants are dependent upon interactions with substrate and with bulk lipid. Surface dilution kinetic models account for aggregated substrates, describing both three-dimensional interactions with lipid interfaces and two-dimensional surface interactions with specific substrate (32). Because detergents strongly inhibit PLD1 in vitro (10, 25, 31, 37, 38) and participate in the transphosphatidylation reaction as nucleophiles (40), kinetic parameters cannot be calculated from mixed detergent-lipid micelle experiments.

Association of phospholipase D and the phospholipid interface was examined by measuring the binding of purified PLD1.d311 to sucrose-loaded phospholipid vesicles (Fig. 4). Vesicle-bound PLD1.d311 was separated from free enzyme by ultracentrifugation and analyzed by immunoblot (29). The catalytic domain of PLD1 (PLD1.d311) displayed high affinity for phospholipid surfaces. The PIP₂ content of the vesicles was crucial for the binding interaction with PLD1.d311. For vesicles prepared with 5 mol % PIP₂, we report a dissociation constant, $K_s^A$, of about 2 µm bulk lipid. Sucrose-loaded vesicles prepared without PIP₂ bound PLD1.d311 weakly. Prior reports demonstrated that other phosphoinositides can partially substitute for PIP₂ in promoting PLD association with lipid vesicles and that membrane association is dependent upon conserved arginine residues within a C-terminal polybasic region (11–14). Since N-terminal PX and PH domains are removed in the PLD1.d311 construct, membrane association depends upon PIP₂ inter-action with catalytic domains.

FIGURE 4. Association of PLD1 catalytic domains with phospholipid vesicles is dependent upon PIP₂.

PLD1.d311 binding to sucrose-loaded vesicles was performed using the method of Buser and McLaughlin (29). Vesicles were prepared with and without PIP₂ (5 mol%). Bound PLD1.d311 was separated from free enzyme by ultracentrifugation. Supernatant and pellet fractions were analyzed by SDS-PAGE and immunoblot. Data are presented as means ± S.E. for three independent experiments.

These data demonstrate that a limiting step in PLD1 catalysis (i.e. partitioning to the lipid interface) is saturated at ~10 µm bulk lipid. Interestingly, conditions that promote maximal (>98%) binding of PLD to phospholipid vesicles do not stimulate PLD activity to maximal rates. This indicates that activation of PLD1 involves effects on both bulk and interfacial binding steps.

The surface binding model (32) describes a two-step association of PLD1 with its substrate, where PLD1 nonspecifically binds the vesicle surface before specifically binding substrate. To examine subsequent interfacial events in PLD1 catalysis, total lipid concentration was held constant at saturating levels (116 µm, >98% bound enzyme).

Kinetic parameters of PLD catalysis were determined by initial velocity experiments measuring dipalmitoyl-PC hydrolysis in PE-dominated vesicles, as described under “Experimental Procedures.” Phospholipase activity was dependent on the surface concentration of substrate (Fig. 5 and Table 2) and the experimental data fit Michaelis-Menten relationships by least squares and Eadie-Hofstee analyses (r² > 0.7). All reported kinetic constants were determined by nonlinear regression analysis. The interfacial Michaelis constant, $K_m^B$, describes two-dimensional surface interactions between PLD and its substrate, PC.

FIGURE 5. Specific phospholipase activities of purified PLD1 and PLD1.d311 were assayed with increasing concentrations of substrate.

Phospholipase activity was measured in vitro in the exogenous substrate assay (31) modified to vary the PC content of the lipid vesicles. PIP₂ and total lipid concentrations were held constant while the PC/PE ratio was adjusted to achieve the indicated substrate concentration. A, activity of purified, full-length PLD1 was measured in the presence of 10 µm GTPγS (unstimulated) or reconstituted with GTPγS and PKCα, Arf-1, or RhoA. Right, Eadie-Hofstee linear transformation of the same data. B, activity of purified PLD1.d311 was measured in the presence of 10 µm GTPγS (unstimulated) or reconstituted with GTPγS and PKCα, Arf-1, RhoA, Rac1, or Cdc42. Right, Eadie-Hofstee linear transformation of the same data. Concentrations of activators were 150 nm Arf-1, 300 nm PKCα, 300 nm RhoA, 300 nm Rac1, and 300 nm Cdc42.

TABLE 2. Effects of allosteric activators on K_m and k_cat values of purified phospholipase D1 toward PC.

Apparent K_m and k_cat were calculated as described under “Experimental Procedures.”

Activator(s)	PLD1 (full-length)			PLD1.d311
	$K_m^B$	kcat	$k_{\text{cat}}/K_m^B$	$K_m^B$	kcat	$k_{\text{cat}}/K_m^B$
	mol %	min−1	mol %−1 min−1	mol %	min−1	mol %−1 min−1
Unstimulated	33 ± 12	4 ± 1	0.1	32 ± 8	6 ± 1	0.2
PKCα	6 ± 4	56 ± 10	9	<1	5 ± 1	11
Arf-1	19 ± 6	16 ± 3	0.8	19 ± 4	22 ± 2	1.2
RhoA	13 ± 3	11 ± 1	0.8	11 ± 3	10 ± 1	0.9
Rac1				9 ± 2	8 ± 1	0.9
Cdc42				8 ± 2	6 ± 1	0.8
Arf-1 + PKCα				<1	26 ± 2	48
Arf-1 + RhoA				2 ± 1	27 ± 2	12
Arf-1 + Rac1				4 ± 2	22 ± 3	5
Arf-1 + Cdc42				3 ± 2	22 ± 3	6
PKCα + RhoA				1 ± 1	6 ± 1	6
PKCα + Rac1				1 ± 2	6 ± 1	6
PKCα + Cdc42				2 ± 1	7 ± 1	4

Apparent $K_m^B$ values for dipalmitoyl-PC were about 33 mol % (39 µm) in the absence of protein activators, similar to the reported value of 42 µm for PLD purified from bovine kidney (37). A V_max value of 32 nmol of PC hydrolyzed min⁻¹ mg⁻¹ PLD1 was obtained for the unstimulated enzyme. PLD1.d311 was only slightly activated, with a V_max of 46 nmol min⁻¹ mg⁻¹. PLD1 activators enhanced catalytic efficiency, k_cat/K_m, both by reducing $K_m^B$ and by enhancing catalytic potential, k_cat (Table 2). PLD1 activators could be discriminated based on their effects on these kinetic parameters.

PKCα had profound effects upon PLD1 kinetics, enhancing catalytic efficiency 75-fold. A mixed activator, PKCα produced dramatic effects on both $K_m^B$ and k_cat. A maximal rate (V_max) of 453 nmol of min⁻¹ mg⁻¹ was calculated for PKCα-stimulated PLD1. PLD1.d311 maximal rate was not enhanced by PKCα. Kinetic analyses reveal that PKCα regulates the catalytic efficiency of PLD1.d311 via binding activation (33). Whereas marked effects on $K_m^B$ remain, deletion of N-terminal PLD1 domains are reflected in a loss of PKCα effects on k_cat. Whereas full effects of PKCα on PLD1 activity require both N- and C-terminal domains, C-terminal interactions with PKCα had marked effects on PLD1.d311 catalysis (Tables 1 and 2). These experiments provide new insights into PKCα regulation of PLD1.

Arf-1·GTPγS was equally effective toward PLD1 and PLD1.d311. In both cases, Arf-1 enhanced PLD catalytic efficiency with only minor effects on $K_m^B$. A catalytic activator (33), Arf-1 increased PLD1 catalytic potential (k_cat) in a concentration-dependent manner. 150 nm Arf-1 increased the maximal rate of PLD 4-fold, whereas $K_m^B$ values were barely reduced, ~40% (Table 2). 10 µm Arf-1 enhanced k_cat values more than 10-fold (data not shown). To allow direct comparisons between activators and to compare synergistically activated conditions, Arf-1 concentrations were set at 150 nm throughout these experiments. These conditions allowed accurate estimates of initial enzymatic rates where PLD activity was within the linear range of the assay (PC hydrolysis between 5 and 25% of total substrate).

Rho family GTPases regulated PLD1 catalysis via allosteric interactions that promote binding activation. It is understood that RhoA activates PLD1 through direct interaction with a C-terminal PLD catalytic subdomain (18–20). Our own data show that RhoA-dependent activation of PLD1 is not altered by deletion of N-terminal domains (Table 2). Maximally effective concentrations of RhoA·GTPγS enhanced the catalytic efficiencies of PLD1 and PLD1.d311 equally. The primary kinetic effect of RhoA was on the apparent interfacial Michaelis constant, $K_m^B$. Effects of closely related GTPases, Rac1 and Cdc42, were also examined (Fig. 5B). Like RhoA, Rac1 and Cdc42 were binding activators. $K_m^B$ values were reduced 3- to 4-fold, whereas k_cat values remained essentially unchanged. Binding activators stimulated PLD1 activity to limiting rates even at low concentrations of PC, shifting the substrate dependence curve leftward without increasing the maximal rate (see also Table 2).

Synergistic Activation of PLD1 Is a Product of Catalytic Activation and Binding Activation

Kinetic analyses of PLD1.d311 activity revealed that synergistic responses were hybrid responses, composed of the kinetic properties of each activator. Arf-1 and PKCα combined to activate PLD1.d311 with catalytic activator effects on k_cat and binding activator effects on K_m. The substrate dependence curve was shifted leftward, reflecting a 50-fold reduction in $K_m^B$, and vertically, reflecting a 4-fold increased k_cat (Fig. 6, Table 2). Moreover, the combined effects of PKC and Arf on $K_m^B$ values were equivalent to the effects of PKC alone, and the effects of combined activators on k_cat values were equivalent to the effects of Arf alone. Together, these effects led to cooperative effects on catalytic efficiency (k_cat/K_m) and led to synergy between these activators.

FIGURE 6. Synergistic activation of PLD1 is dependent upon Arf-1.

Substrate dependence of purified PLD1.d311 was examined as described in the legend to Fig. 5. A, phospholipase activity was measured without activators or stimulated by PKCα, Arf-1, or PKCα and Arf-1. RhoA (B), Rac1 (C), and Cdc42 (D) were combined with Arf-1 or PKCα to stimulate PLD1.d311.

Rho GTPases did not synergize with PKCα. Kinetic properties of PKCα-stimulated PLD1.d311 were not significantly changed by the addition of RhoA, Rac1, or Cdc42 to the reaction mixture (Fig. 6 and Table 2). Arf-1 synergized with RhoA, Rac1, and Cdc42 (Fig. 6, B–D and Table 2). Arf-1 contributed enhanced catalytic potential, whereas Rho GTPases contributed reduced $K_m^B$ values. As shown in Fig. 6B, simultaneous stimulation by Arf-1 and RhoA led to a synergistic response with properties of each activator. $K_m^B$ values were reduced 14-fold, and k_cat was enhanced 4-fold. PLD1.d311 initial rates reached levels nearly 3-fold greater than can be explained by the additive effects of RhoA and Arf-1. Synergy between Arf-1 and Rac1 or Cdc42 was slightly less robust. Combinations of Arf-1 and Rac1 (Fig. 6C) or Arf-1 and Cdc42 (Fig. 6D) stimulated PLD1.d311 to levels 2-fold greater than can be explained by additive effects.

PIP₂ Is an Essential PLD1 Activator

The activity of purified PLD1.d311 was measured over a range of PIP₂ concentrations, and the parameters of the assay are described in relevant figure legends and under “Experimental Procedures.” In the absence of PIP₂, PLD1.d311 was inactive and did not respond to activators. PIP₂ stimulated PLD1.d311 activity in a concentration-dependent manner (Fig. 7). In the absence of protein activators, PLD activity was detected at less than 2 mol %PIP₂ and achieved maximal velocity at ~5 mol % PIP₂. As shown in Fig. 4, binding of PLD1.d311 to phospholipid vesicles required PIP₂, and binding was essentially complete (>98%) for vesicles prepared with 5 mol % PIP₂. Together, these data are consistent with the interpretation that PIP₂ activates PLD1.d311 catalysis by recruiting the enzyme to the membrane.

FIGURE 7. Synergy is dependent upon Arf-1 and PIP₂.

Specific phospholipase activity of PLD1.d311 was assayed with increasing concentrations of PIP₂. Phospholipase activity was measured in vitro in the standard assay (31) modified to vary the PIP₂ content of the lipid vesicles. PC and total lipid concentrations were held constant while the PIP₂/PE ratio was adjusted to achieve the indicated PIP₂ concentration. Phospholipase activity of purified PLD1.d311 (10 nm) was measured in the presence of 10 µm GTPγS alone (unstimulated) or reconstituted with GTPγS and activators at maximally effective concentrations. A, phospholipase activity was stimulated by PKCα, RhoA, or PKCα and RhoA. B, PKCα was combined with Rac1 or Cdc42. C, PLD.d311 activity was stimulated by Arf-1 combined with PKCα, RhoA, Rac1, or Cdc42.

Regulated PLD activity was also steeply dependent on PIP₂. PIP₂ potentiated the effects of all PLD1 activators (Fig. 7). Activity rose sharply with increasing PIP₂ concentrations. PIP₂ effects were biphasic toward PKCα-, RhoA-, Rac1-, and Cdc42-stimulated PLD1.d311 activity. Maximal enzymatic rates were achieved at 8 – 12 mol % PIP₂, and concentrations greater than 35 mol % rendered PLD1.d311 unresponsive to activators. RhoA, Rac1, and Cdc42 stimulate PLD1.d311 equally and have identical PIP₂-dependent profiles (Fig. 7A). Rho GTPases did not synergize with PKCα to activate PLD1.d311 (Fig. 7B). At optimal PIP₂ concentrations (8–12 mol %), effects of PKCα and Rho GTPases were nearly additive. At other concentrations of PIP₂, PKCα effects dominated.

150 nm Arf-1 stimulated PLD.d311 activity 4 – 6-fold above basal activity, with a PIP₂ dependence profile very similar to the profile of unstimulated PLD1.d311 (Fig. 7, A and C). Arf-1 strongly synergized with all other activators (Fig. 7C). PLD1 activators synergistically activated PLD1.d311 only under optimal PIP₂ concentrations. Synergy between Arf-1 and PKCα was greatest at 8 mol % PIP₂, whereas synergy between Arf-1 and Rho GTPases was greatest at ~5 mol %.

PIP₂ performs multiple roles in PLD1 activation. Clearly, PIP₂ is required for recruitment of PLD catalytic domains to the membrane interface (11–14, 43). PLD1 mutations that disrupt a PIP₂ binding site led to impaired membrane association and catalytic activity (<5% of wild-type activity in vitro and in vivo) (13). In the surface-binding kinetic model, this effect of PIP₂ occurs at the first binding step and is characterized by the binding constant, $K_s^A$. PIP₂ may also affect interfacial events, orienting the catalytic site at the membrane or stabilizing activated PLD conformations (14). In addition, PIP₂ may activate PLD via independent effects on substrate conformation (44) and on PLD1 activators. All PLD1 activators bind PIP₂ in lipid membranes and are them-selves activated by PIP₂ (45–47). Through multiple mechanisms, PIP₂ levels determine the activity of PLD1 and determine synergistic interactions between activators.

The proposed kinetic interpretation of synergism between PLD1 activators has several important corollaries. At limiting concentrations of substrate ([PC]₀ ≪ K_m), initial velocity of the PLD-catalyzed reaction is directly proportional to initial substrate concentration (v₀ ≈ [PC]₀V_max/K_m). This relationship is consistent with observations that synergy between activators is pronounced at low PC concentrations. At high substrate concentrations ([PC]₀ ≫ K_m), the reaction displays first-order kinetics and approaches maximal velocity (v₀ ≈ V_max). Under such conditions, binding activators have little effect on enzymatic rates and minimal capacity to synergize with other activators.

In a cellular context, binding activators like Rho, Rac, and Cdc42 may have little or no effect on PLD1 activity in PC-rich membrane domains. At nearby membrane subdomains with reduced PC content, the same activators may have dramatic stimulatory effects on PLD1 activity. The proposed kinetic model of PLD1 activation predicts these counterintuitive responses.

Signal Integration by PLD1

Input from multiple PLD1 activators produces responses in a context-dependent manner. As a coincidence detector, PLD1 can constrain activity to cellular locales where these signaling molecules converge. Synergy requires appropriate concentrations of substrate, PIP₂, catalytic activators, and binding activators. Working together, these molecules can elicit substantial PLD1 activity, hydrolyzing PC to generate PA.

A descriptive parameter, denoted κ, was generated to describe functional interactions between PLD effectors (Equation 4). With this formalism, κ is a ratio of the response to combined activators relative to the sum of the responses to individual activators: ε_(a,b) = κ(ε_(a) + ε_(b)). Degree of activation, ε_(a), relates PLD functional output (enzymatic rate) to a signaling input (effector concentration) for a given effector, a (33). ε_(a,b) denotes the enzymatic response to simultaneous stimulation by effector a and effector b. κ values of unity indicate strictly additive effects (ε_(a,b) = ε_(a) + ε_(b)). Synergistic effects produce κ values greater than 1 (combined response greater than the sum of the effects of individual activators, ε_(a,b) > ε_(a) + ε_(b)). A κ value of 4 represents a synergistic response, 4 times greater than an additive response.

PKCα did not interact with Rho, Rac, or Cdc42 to produce synergistic responses (κ ≤ 1, not shown). Combinations including Arf-1 led to synergy (κ > 1) only when PC and PIP₂ concentrations were optimal (Fig. 8). Synergistic activation is related to differences in K_m values between individual activators; synergy is greatest at substrate concentrations less than K_m for Arf-1-stimulated PLD (~19 mol % PC). Consequently, synergy was greatest at the lowest substrate concentrations tested (Fig. 8A).

FIGURE 8. Synergistic effects of PLD1 activators.

Synergy is expressed as a ratio (κ) of the response to combined activators relative to the sum of the responses to individual activators. Additive responses have κ values of unity (dashed lines). A, synergy between effectors was evaluated over a range of substrate (PC) concentrations. B, synergistic responses to effectors were evaluated over a range of PIP₂ concentrations. Concentrations of activators were 150 nm Arf-1, 300 nm RhoA, 300 nm Rac1, 300 nm Cdc42, and 300 nm PKCα.

Arf-1 + PKCα and Arf-1 + RhoA displayed the strongest synergistic interactions, reaching rates 3.5-fold greater than those predicted for additive effects (Fig. 8A). Rac1 and Cdc42 synergized with Arf-1 to generate responses with κ values of ~2. Differences between the properties of RhoA, Rac1, and Cdc42 became apparent when these activators were combined with Arf-1. Functional differences between Rho GTPases reveal subtle differences in intermolecular interactions between PLD1 and these activators.

Synergy was steeply dependent on PIP₂ levels. Synergy between Arf-1 and PKCα was greatest at ~8 mol % PIP₂, and synergy was sensitive to small changes in PIP₂ surface concentration (Fig. 8B). A 2-fold increase or decrease in PIP₂ levels eliminated synergy altogether. Synergy between Arf-1 and Rho GTPases was similarly PIP₂-dependent, but these combinations required less PIP₂ (3 mol % compared with 8 mol %).

These requirements for PLD1 regulation are consistent with the proposed role of the enzyme inmembrane trafficking and exocytosis. PLD1, Arf-1, PKCα, and Cdc42 localize in late endosomes and trans-Golgi structures (48–50). Secretory functions require PLD, Arf-1, PIP₂, and Cdc42 (51–53). Extensive cross-talk between signaling networks coordinates PLD1 activators. Golgi-associated GTPase-activating proteins coordinate the activation state of Rho, Cdc42, and Arf (54, 55). PLD, PKC, Arf, and Rho subfamily GTPases activate PIP₂ synthesis (56). PIP₂ stimulates PLD1 and PLD1 activators (45–47).

An important further development in this area is greater definition and structural characterization of the sites at which PLD1 activators interact with the enzyme. This and the elucidation of the structure of this enzyme will enable understanding of the molecular mechanisms by which activators alter the catalytic mechanism.