Lipid and Peptide Control of Phosphatidylinositol 4-Kinase IIα Activity on Golgi-endosomal Rafts*

The most abundant and widely expressed mammalian phosphoinositide kinase activity is contributed by phosphatidylinositol 4-kinase IIα (PI4KIIα). In this study we demonstrate that PI4KIIα is a novel GTP-independent target of the wasp venom tetradecapeptide mastoparan and that different mechanisms of activation occur in different subcellular membranes. Following cell membrane fractionation mastoparan specifically stimulated a high activity Golgi/endosomal pool of PI4KIIα independently of exogenous guanine nucleotides. Conversely, GTPγS stimulated a low activity pool of PI4KIIα in a separable dense membrane fraction and this response was further enhanced by mastoparan. Overexpression of PI4KIIα increased the basal phosphatidylinositol 4-kinase activity of each membrane pool, as well as the mastoparan-dependent activities, thereby demonstrating that mastoparan specifically activates this isozyme. Both mastoparan and M7, at concentrations known to invoke secretion, stimulated PI4KIIα with similar efficacies, resulting in an increase in the apparent Vmax and decrease in Km for exogenously added PI. Mastoparan also stimulated PI4KIIα immunoprecipitated from the raft fraction, indicating that PI4KIIα is a direct target of mastoparan. Finally we reveal a striking dependence of both basal and mastoparan-stimulated PI4KIIα activity on endogenous cholesterol concentration and therefore conclude that changes in membrane environment can regulate PI4KIIα activity.

The wasp venom peptide mastoparan stimulates signaling by G i and G 0 heterotrimeric G-proteins by enhancing the rate of dissociation of bound GDP, thereby allowing GTP to bind. Increased G i and G 0 signaling then activates a variety of cell-type-dependent events, including secretion and ion transport.
PI4KII␣ localizes to membranes of the Golgi-endosomal system (10,11) where it provides the phosphatidylinositol 4-phosphate (PI4P) that is required to recruit the AP-1 clathrin adaptor complex to membranes of the trans-Golgi network (TGN) (11). Two membrane fractions containing PI4KII␣ activity have been identified using sucrose density gradient ultracentrifugation (12) and found to possess differing levels of intrinsic kinase activity both in intact membranes and in immunoprecipitates prepared from detergent lysates (13). The higher activity pool is found in membranes that possess the raft-like properties of high buoyancy in sucrose density gradients and resistance to detergent solubilization (12)(13)(14). As reported by Barylko et al. (15), rafting of PI4KII␣ is likely to arise from palmitoylation of a cysteine-rich sequence within the kinase domain (residues 174 -178, CCPCC), but whether or not a particular raft lipid environment is required for the high PI4KII␣ activity is not known. The lower activity pool is less buoyant and contains the bulk of the enzyme (12,13). Previous studies of the activation of PI4KII by mastoparan have assumed a dependence on G-protein activation (8,9); however experiments have hitherto not addressed whether mastoparan activates either or both enzyme pools nor whether activation of either pool by mastoparan can occur directly.
The established mechanisms by which mastoparan activates signaling involve the formation of an amphipathic ␣-helix from the random conformation present in free solution. The basic, helical peptide mimics the effector region of cognate G-protein-coupled receptors (16) and of calmodulin-binding proteins (17), leading to activation of G i/0 and calmodulin, respectively. Two distinct modes of binding have been identified, one in which the peptide binds only to the target protein and another in which the peptide binds concomitantly to a membrane and a target protein. For example, in aqueous solution mastoparan binds directly to p67-phox (7) and to calmodulin. Mastoparan binds to calmodulin in a helical conformation with low nanomolar affinity (18). In contrast binding to purified G i/0 -proteins is weak, but micromolar affinities are afforded by the presence of membrane lipids (19); this phenomenon arises from the fact that mastoparan adopts an amphipathic helical conformation at the lipid-aqueous interface (20,21).
The mechanism by which mastoparan binds to PLD2, Rho, NDPK, glycogen phosphorylase, PLA2, and PI4KII has not yet been characterized; although NDPK, PLA2, and PI4KII may be activated indirectly, as a direct interaction with mastoparan has not been established. As with calmodulin and p67-phox, the interaction between mastoparan and Rho and between mastoparan and glycogen phosphorylase occurs in solution (5), whereas binding to PLD2, like the interaction with heterotrimeric G-proteins, has only been observed for the membrane-bound protein (1,22). This investigation began with the question of whether mastoparan and GTP regulate the two main pools of PI4KII␣ activity via a common G-protein-dependent mechanism. Our results indicated that activation of PI4KII␣ by mastoparan occurs via both GTP-dependent and GTPindependent mechanisms in different membrane fractions. We also found that GTP-independent activation of PI4KII␣ in buoyant mem-branes occurs directly. Finally we reveal for the first time that the PI4KII␣ activity in buoyant membranes shows a striking dependence on cholesterol concentration. The results are discussed in terms of the role of rafting in the regulation of PI signaling.

EXPERIMENTAL PROCEDURES
Materials-G-protein agonists, pertussis toxin, wortmannin, and mastoparans were purchased from Calbiochem (Nottingham, UK). GDP, cGMP, and PI were bought from Sigma. Anti-PI4KII antiserum was prepared and purified as previously described by us (12). Anti-G␣ i1/2 and anti-G␣ 0 were from Santa Cruz. [␥-32 P]ATP was obtained from Amersham Biosciences. Dulbecco's modified Eagle's medium, fetal calf serum, and penicillin/streptomycin were purchased from Invitrogen.
PI4K Activity Assays-PI4K assays in the presence of endogenous and exogenous PI were performed as previously described (14). Unless otherwise indicated, all assays were carried out using endogenous PI as substrate. Reaction products were separated by thin layer chromatography and visualized on a Typhoon 9400 PhosphorImager (Amersham Biosciences). Quantitative data were obtained within the linear range of the instrument using ImageQuant Software (Amersham Biosciences). Data analysis and curve fitting were performed using Prism 4 software (GraphPad, San Diego, CA). Michaelis-Menten kinetic analyses were performed by non-linear regression curve-fitting using Prism software (GraphPad) and compared using the unpaired Student's t test where significance was attained at p Ͻ 0.05. For classical Michaelis-Menten kinetic analyses it is understood that both enzyme and substrate are freely diffusible. In this study, substrate PI is present in micellar form or membrane-bound and is not freely diffusible; hence the resultant V max and K m values presented here are strictly apparent values.
Note that we have previously shown that these assays are linear for time points of up to 30 min (13). In addition we have shown that these assays reflect the intrinsic activity of the enzyme and do not reflect differences in endogenous substrate concentration between gradient fractions (13). Assays were carried out on equal volume aliquots from density gradient fractions. All mastoparan and guanine nucleotide additions were on isolated membrane fractions.
For the 4C5G inhibition experiments the monoclonal antibody was used at a final concentration of 1 g/ml. For G i/0 inhibition, cells were incubated for 20 h with pertussis toxin at a concentration of 100 ng/ml. Subcellular Fractionation-A431 and HT1080 cells were maintained at 37°C in a humidified incubator at 10% CO 2 . Cells were cultured in Dulbecco's modified Eagle's medium containing Glutamax, 10% fetal calf serum, 50 IU/ml penicillin, and 50 g/ml streptomycin. Separation of A431 and HT1080 cell post nuclear supernatants on a continuous 10 -40% (w/v) sucrose gradient was performed as previously described (12,13). Gradient fractions were harvested from the top of the ultracentrifuge tube with the first and least dense fraction designated fraction number 1. Fractions 1-4 of the gradient contained cytosol, and fractions 7-12 contained membranous organelles. The distributions of organelle membranes were determined using Western blotting to identify a range of organelle marker proteins (12,13).
Cholesterol Depletion and Repletion Studies-Membrane fractions were treated for 20 min in the presence of methyl-␤-cyclodextrin (M␤CD) to remove raft-associated cholesterol. Cholesterol loading was carried out by the addition of 2.5 mM cholesterol:M␤CD complex at a ratio of 10:1 cholesterol:M␤CD prepared as previously described (23).

Regulation of PI4KII␣ by Mastoparan and GTP-
The distribution of PI4P synthesis using endogenous PI as substrate was examined in subcellular fractions prepared from A431 cells and separated as previously reported on a 10 -40% continuous sucrose gradient (12,13). For the purposes of this study we concentrated on fractions 7-12 of the gradient, which contain all the PI4KII activity. As we reported previously, PI4KII activity was found to peak in the buoyant membrane fractions 8 -10 (12, 13), which contain late endosomes, TGN and also a pool of activated PI4KII␣ (12,13). To investigate whether or not mastoparan ( Fig. 1A) could enhance PI4KII activity in the absence of G-protein activation, mastoparan (10 M) was added to each membrane fraction in the absence of GTP. PI4K assays revealed that mastoparan elicited a highly reproducible 3-4-fold increase in PI4P generation in the more buoyant region of the gradient that contained activated PI4KII␣ (Fig. 1, B and C) but had little effect on denser fractions that contained the bulk of the PI4KII␣ and PI4KII␤ protein (12,13). The M7 analogue of mastoparan (Fig. 1A) is a more potent activator of G i/0 -proteins (24) but gave a similar amount of activation to mastoparan. In contrast to the clear enhancement of the peak of PI4K activity by mastoparan, the addition of GTP␥S (100 M) did not significantly affect the PI4K activity in this region of the gradient (Fig. 1C). However GTP␥S did enhance PI4P generation in denser fractions, which contained secretory vesicles, plasma membrane, rough endoplasmic reticulum, lysosomes, and PI4KII␤ (12,13). Compared with the mastoparan response, the stimulatory effect of GTP␥S was found to be more variable (range 0.2-2-fold) and labile, as it was lost more rapidly on storage than the mastoparan responsiveness. Consequently, in response to the starting question concerning the mechanism by which PI4KII␣ is regulated by mastoparan and GTP, these data indicate that different modes of regulation occur in different subcellular membranes.
Activation of PI4KII␣ in buoyant membranes by mastoparan in the absence of exogenous GTP suggested that the previously assumed mediators of PI4KII regulation by mastoparan, namely the G i/0 heterotrimeric G-proteins (8,9) and Rho family small G-proteins (8), are not in fact required in these membranes. In contrast the stimulation of PI4KII activity by GTP␥S in denser membranes indicated the presence of a G-protein-activated PI4KII activity in this region of the gradient, which was maintained in the presence of wortmannin (Fig. 1D). Unexpectedly however, co-addition of M7 and GTP␥S gave rise to a level of PI4K activation in fraction 12 that was only equivalent to the sum of the responses observed when each reagent was added independently (Fig.  1D). The effects of M7 and GTP␥S on PI4KII in fraction 12 did not therefore demonstrate the synergy expected for G-protein activation (16). Furthermore, the inability of mastoparan to affect PI4KII␣ in the dense membranes indicates that the peptide did not affect the endogenous PI, which is present in buoyant and dense membranes.
Effect of Mg 2ϩ Concentration-One concern at this stage was that the millimolar levels of Mg 2ϩ used for the PI4K assay may have interfered with the mastoparan activation of G-proteins in the buoyant membranes, as work by Higashijima et al. (16) had demonstrated that mastoparan-stimulation of GTP exchange for GDP on G i/0 is inhibited by millimolar Mg 2ϩ . However in buffer containing 500 M Mg 2ϩ , mastoparan, but not GTP␥S, robustly stimulated PI4P synthesis (Fig. 2). Note that as these assays were carried out in the presence of wortmannin (5 M), the observed stimulation by mastoparan was selective for PI4KII activity. Furthermore, in control experiments, we observed that the profile of PI4K activity across the gradient was unaffected when PI4K assays were carried out in the presence of 500 M MgCl 2 , although the absolute levels of PI4P generation were reduced as would be expected (25,26) (data not shown). Furthermore GDP␤S, a potent G-protein antagonist did not affect mastoparan-dependent PI 4-kinase activity (Fig. 2), thereby confirming G-protein-independent activation in the buoyant fraction.
Overexpression of PI4KII␣ Leads to Increased Basal and Mastoparanstimulated Activity-To verify that mastoparan was activating the PI4KII␣ isozyme and not activating a secondary pool of the PI4KII␤ enzyme or inhibiting a PI4P phosphatase, we investigated the effect of increasing the expression this enzyme. HT1080-PI4KII␣ cells stably expressed a GFP-PI4KII␣ fusion protein and were chosen because they afforded a measurable increase in PI4KII␣ expression without the disruption of the Golgi body 3 that has been observed in other cells (11) or any measurable change in G␣ i levels as shown by Western blotting (Fig.  3A). The level of GFP-PI4KII␣ in these cells was typically 2-4-fold higher than endogenous PI4KII␣ as determined by Western blotting with anti-PI4KII␣ antiserum (Fig. 3B). If mastoparan did activate PI4KII␣ we expected to find increases in the control and mastoparanstimulated activities. There was indeed an increase in the effects of mastoparan on the profile of PI4K activity in subcellular fractions obtained from these cells, and typically, HT1080-PI4KII␣ cells had a 2-3-fold increase in the peak of PI4K activity relative to non-transfected cells. Furthermore, in line with our previous reports using different epithelial cell lines (12,13), the peak of PI4K activity in this fibrosarcoma cell line was almost completely inhibited by the monoclonal antibody 4C5G but insensitive to wortmannin, thereby confirming the activity as PI4KII (Fig. 3C). As observed with A431 cells, addition of either mastoparan or its M7 analogue (10 M) to membranes prepared from HT1080 or HT1080-PI4KII␣ cells resulted in a similar 3-4-fold increase in the peak of PI4K activity (Fig. 3D). The observed changes in activity in response to mastoparan were therefore increased by PI4KII␣ overexpression (Fig. 3E), thereby confirming that mastoparan and M7 stimulate PI4KII␣ activity. Finally, as with the basal activity, the mastoparan response was inhibited by the 4C5G monoclonal antibody (Fig. 3F).
Lack of Synergy between Mastoparan and Guanine Nucleotides-Although G i/0 -proteins have previously been reported to stimulate PI4K activity (8,9), G-protein-independent activation by mastoparans was a   FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 novel finding. Subsequent experiments therefore focused on the effect of mastoparan on the buoyant, activated pool of PI4KII␣, which localizes to fractions enriched in late endosomal and TGN markers. We first returned to investigate more closely whether there was any evidence at all for synergy between mastoparan and guanine nucleotides by taking advantage of the higher activities observed with the HT1080-PI4KII␣ cells. We found that addition of GTP␥S (100 M), GTP (100 M), GDP (10 M), or cGMP (10 M) either on their own or in the presence of mastoparan had a slight inhibitory effect on PI4P generation by buoyant membranes (Fig. 4A). However the apparent inhibition by guanine nucleotides was found not to be statistically significant when assessed in further experiments. These results confirmed that neither GTP nor its metabolites are required for the observed stimulation of the PI4KII activity in buoyant membranes by mastoparan. Moreover, preincubation of cells with pertussis toxin had no apparent effect on the stimulation of PI4K activity by mastoparan in the buoyant membranes (Fig. 4B). However similar to heterotrimeric G-protein activation (19), the M17 analogue (Fig. 1A) did not significantly enhance PI4K activity (Fig. 3F).

Regulation of PI4KII␣
Quantification: Mastoparan Dose Response and PI4KII␣ Enzymatic Activity in Buoyant Membranes-Dose responses to both mastoparan and M7 were obtained for the PI4K activity in buoyant membranes prepared from HT1080-PI4KII␣ cells (Fig. 4C). We found that mastoparan and M7 stimulated this PI4KII␣ activity with EC 50 s of 12.9 Ϯ 3.1 M (n ϭ 3) and 7.2 Ϯ 4.1 M (n ϭ 3), respectively. The similar potencies of mastoparan and M7 observed in this study provide further evidence that these peptides do not stimulate PI4KII␣ via G i/0 activation. It is also FIGURE 3. Increased PI4KII␣ expression elevates the response to mastoparan. A, Western blot demonstrating that HT1080-PI4KII␣ cells (GFP-PI4KII␣) express similar levels of G␣ i1/2 relative to HT1080 cells (WT). Protein concentrations of cell lysates were normalized using the Bradford reagent. The endoplasmic reticulum structural protein calnexin was used to demonstrate that protein levels were equal in both samples. B, Western blot showing typical distribution of GFP-PI4KII␣ and endogenous PI4KII␣ in gradient fractions 7-12 from HT1080 cells stably transfected with GFP-HT1080. C, effects of the PI4KII-specific inhibitory monoclonal antibody 4C5G (1 g/ml F) and the PI4KIII-specific inhibitor wortmannin (5 M ࡗ) on PI4P synthesis from GFP-HT1080 fractions using endogenous PI substrate (untreated control f). D, effect of mastoparan (10 M ࡗ) and M7 (10 M F) on PI4P generation by HT1080 cells expressing GFP-PI4KII␣ using endogenous PI substrate (untreated control f). E, comparison of responses to mastoparan (10 M) on peak PI4K activity fractions for control (open bars) and HT1080 cells stably expressing GFP-PI4KII␣ (filled bars). Note that the experimental results in each subsection in this figure were derived from cell fractionations carried out on different days, which gives rise to the slight differences in PI4K activity profiles. F, effect of M17 (10 M) on basal PI4P production and inhibitory antibody 4C5G on mastoparan-stimulated PI4P production.
important to note that mastoparan activates PI4KII␣ at physiologically relevant concentrations that also activate secretion (27). There were however some differences in the overall efficacies of mastoparan and M7 in stimulating PI4KII␣ activity. In particular the maximum response to M7, 6.4 Ϯ 1.9-fold over basal (n ϭ 3), was attained at a concentration of ϳ20 M and then declined at higher concentrations. The determination of the potency of M7 was therefore derived from M7 concentrations in the range 0 -20 M. In contrast the maximal response to mastoparan, 7 Ϯ 0.8-fold over basal (n ϭ 3), appeared to plateau between 20 and 200 M. Consequently mastoparan and M7 are not equally efficacious at concentrations greater then 20 M.
The effects of mastoparan on the enzymatic properties of the activated pool of PI4KII␣ were also quantified. By varying the concentration of exogenously added PI (0 -200 M) we were able to quantify changes in the rate of PI4P production by PI4KII␣ in intact membranes (Fig. 5).
In the presence of mastoparan (10 M), the apparent V max for the reaction was enhanced 3.6-fold (545 Ϯ 39 PhosphorImager units/h for control membranes versus 1967 Ϯ 171 PhosphorImager units/h in the presence of mastoparan). The apparent K m for PI was also affected significantly by mastoparan (9.8 Ϯ 3.1 M for control membranes versus 1.7 Ϯ 0.3 M in the presence of mastoparan). Therefore, mastoparan appeared to enhance the rate of PI4P production at saturating and subsaturating concentrations of PI.
Mastoparan Activates PI4KII␣ Isolated from Membrane Rafts-The different responses of PI4KII␣ in buoyant and dense membrane pools indicated that either a modification of the enzyme itself or a secondary molecule was localized to the buoyant membranes. Indeed M7 (50 M) failed to elicit any increase in the kinase activity of a bacterially expressed GST-PI4KII␣ fusion protein over the range of PI concentrations shown in Fig. 6A, supporting the notion that activation of PI4KII␣ by mastoparan requires either a particular modification of PI4KII␣ itself (direct activation) or a particular membrane environment (indirect activation). These assays required the presence of Triton X-100 because the recombinant enzyme was otherwise completely inactive. In contrast PI4KII␣ immunoprecipitated from the buoyant fraction was active in the absence of Triton X-100 and was stimulated by mastoparan (Fig.  6B). In the presence of Triton X-100 the basal activity was enhanced over 10-fold, and no response to mastoparan was detectable under these conditions.
PI4KII␣ Activity Is Dependent on Cholesterol Concentration-The presence of PI4KII␣ and PI in buoyant membrane rafts has been inferred from several previous studies (13,14,28,29). To establish whether or not the organization of PI4KII␣ and PI in rafts is required for the observed regulation by mastoparan, the effects of cholesterol depletion were investigated. Fig. 7A illustrates that treatment of the buoyant membrane fraction with M␤CD, which selectively removes endogenous cholesterol from membranes, caused a drastic inhibition of mastoparan-stimulated PI4KII␣ activity (IC 50 1.9 Ϯ 0.1 mM, n ϭ 3). However proportionally similar amounts of inhibition by M␤CD were also observed using unstimulated preparations (IC 50 4.2 Ϯ 0.1 mM, n ϭ 3) Data is presented as mean Ϯ S.E. of three separate determinations. Note that the responses to either mastoparan alone or co-addition of mastoparan with any of the guanine nucleotides were not significantly different. B, mastoparan-stimulated PI4P production in the buoyant fraction prepared from cells treated overnight with pertussis toxin (100 ng/ml). C, dose responses for PI4P generation, using endogenous PI as substrate, to either mastoparan (f), M7 (F), or GTP␥S (ࡗ) on peak PI4KII␣-activity membrane fractions from HT1080 cells.  FEBRUARY 17, 2006 • VOLUME 281 • NUMBER 7 and using endogenous or exogenous PI substrate, indicating that cholesterol is a direct regulator of the enzyme itself rather than operating via a secondary mastoparan target or PI. This idea was further substantiated by using cholesterol-loaded M␤CD to replenish cholesterol in membranes. In these experiments the inhibition caused by cholesterol depletion was reversed (Fig. 7B). Furthermore cholesterol also seemed to augment the basal level of PI4P synthesis indicating that cholesterol levels may regulate the concentration of PI4P in rafts.

DISCUSSION
The regulation of cellular PI4P synthesis is of central importance to many events in the cell including the provision of PI4P for phospholipase C signaling, formation of secretory vesicles from the TGN and endocytosis. The major PI4P biosynthetic activity in most mammalian cells is provided by PI4KII␣. In this study we show that the amphiphilic peptides mastoparan and M7 can activate a pool of PI4KII␣ associated with a buoyant subcellular fraction enriched in TGN-derived vesicles, late endosomes, and p97/valosin-containing protein-rich endoplasmic reticulum (12). Although mastoparan has been shown previously to enhance PI4KII activity in preparations of secretory vesicles (8,9), in these studies the authors concluded that mastoparan was acting to regulate PI4KII through G i/0 activation. Despite this conclusion, in these previous studies there was no reported addition of guanine nucleotides with the mastoparan, a condition that would be necessary if mastoparan was acting as a promoter of GTP exchange on the ␣-subunit of G i/0 . The results presented here demonstrate that the mastoparan and M7 stim-ulation of PI4KII␣ can occur completely independently of G-protein activation. In addition, GTP, GTP␥S, nor the metabolites GDP and cGMP affected PI4KII activity on membranes either on their own or in conjunction with mastoparan. Furthermore, preincubation of cells with pertussis toxin, a bacterial enzyme that inactivates the ␣-subunit of G i/0 through ADP-ribosylation, also had no affect on PI4KII activity in this membrane fraction. These results demonstrated that mastoparan does not activate PI4KII␣ on late endosomal and TGN membranes via G i/0 activation.
In contrast to the behavior of the activated pool of PI4KII␣ present in buoyant membranes, a weakly GTP-sensitive PI4K activity was associated with a dense membrane fraction that contained plasma membrane, rough endoplasmic reticulum, and secretory vesicles. It is important to stress that the wortmannin insensitivity of the mastoparan and GTP␥S responses allowed contributions by G-protein-dependent type III PI 4-kinases that are wortmannin-sensitive (30) to be excluded. However we have shown previously that this gradient fraction is enriched in PI4KII␤ (13); hence the GTP-sensitive response could represent an effect on this particular isoform. Consistent with this view Wei et al. (31) have shown that PI4KII␤ can be regulated by the small G-protein Rac. Nonetheless, overexpression of PI4KII␣ increased the GTP-independent and GTP-dependent response across the whole density gradient, indicating that the PI4KII␣ isozyme is predominantly responsible for the observed GTP-dependent activity in dense membranes. The unexpected inability of M7 to induce a synergistic activation of the GTP␥S response in this region of the gradient (fraction 12), suggests that in dense membranes the mastoparan and GTP-dependent responses occur via biochemically independent pathways. FIGURE 6. Mastoparan directly activates PI4KII␣ immunoprecipitated from rafts. A, the M7 peptide has no effect on purified PI4KII␣. Recombinant GST-PI4KII␣ was purified from Escherichia coli and PI4P activity measured in response to varying concentrations of PI in the presence (f) or absence (f) of M7 (50 M). Data are presented as the mean Ϯ S.E. of three separate determinations. B, mastoparan stimulates immunoprecipitated PI4KII␣. PI4KII␣ was immunoprecipitated using anti-PI4KII␣ antiserum from RIPA-solubilized rafts prepared from HT1080 cells. Equal aliquots were then assayed in triplicate for PI4K activity in the presence or absence of Triton X-100 (0.2%) and mastoparan (10 ⌴) and PI (50 ⌴). Analysis of the dose-responses for mastoparan and M7-stimulated PI4P synthesis by the active PI4KII␣ fraction also points to the importance of a mechanism other than G i/0 activation. It is important to note that the EC 50 values we obtain for mastoparan and M7 (12.9 Ϯ 3.1 and 7.2 Ϯ 4.1 M, respectively) are also in the physiological range reported for mastoparan-induced secretion obtained in other studies, an effect that had been attributed to heterotrimeric G-protein activation (27). We found that both the maximal responses and EC 50 values for mastoparan and M7 are not significantly different. This is not the case for G i/0 regulation, where the EC 50 for M7 is 4-fold lower than for mastoparan and where maximal stimulation is twice that of mastoparan. M7 has one less positive charge than mastoparan because the Lys-Ile residues at positions 12 and 13 in mastoparan are substituted by Ala-Leu in M7. For the activation of G-proteins by M7 the change from Lys to Ala at position 12 is thought to cause the much-enhanced G i/0 responsiveness (19). This structure-function relationship does not hold true for the stimulation of membrane-associated PI4KII␣ by M7 where the change in amino acid sequence has no apparent effect. However in common with the G i/0 observations (19), we do find that M17, the control mastoparan analogue with a positive charge added to the hydrophobic side of the mastoparan helix, does not significantly enhance PI4KII␣ activity. Our results, taken together, demonstrate that the positive charge at residue 12 of mastoparan is not important for the activation of PI4KII␣, a clear difference from the G i/0 determinants. With this new information it should now be feasible to consider more rational approaches to the design of organelle-specific activators of PI4KII␣ and ultimately, the development of selective reagents with which to investigate the functional role of PI4P in intracellular vesicle trafficking.
In this study the high and low activity forms of PI4KII␣ reveal compartment-specific modes of regulation by mastoparan. The observed differences may arise from intrinsic differences in PI4KII␣ or extrinsic differences in membrane composition. Intrinsic differences in the enzymatic properties of the two forms have been previously characterized; the low activity form of PI4KII␣ present in the membrane fractions enriched in secretory vesicles and early endosomes has a lower apparent V max for PI4P generation and a higher apparent K m for PI than the PI4KII␣ present in the membrane fractions enriched in late endosomal and TGN membranes. It is interesting to note that the kinetic parameters that we have determined for a purified, bacterially expressed GST-PI4KII␣ fusion protein are consistent with the properties of the low activity form. As the GST-PI4KII␣ was neither acylated nor bound to a membrane, correct modifications and targeting of PI4KII␣ may be essential elements in the acquisition of mastoparan sensitivity and high enzyme activity. The maintained sensitivity to mastoparan following immunoprecipitation indicated a direct mechanism of activation and identified PI4KII␣ as a new target of mastoparan that links mastoparan to PI signaling in Golgi-endosomal rafts.
In previous work we reported that the low buoyant density membrane fraction containing highly active PI4KII␣ displays biochemical properties consistent with a subpopulation of membrane rafts (12,14,28,32). Furthermore Barylko et al. (15,29) have found that PI4KII␣ undergoes palmitoylation, a well defined raft-targeting modification. Consequently, it is possible that the observed increased efficacy of mastoparan stimulation in the TGN/late endosomal fraction may require the presence of PI4KII␣-rich lipid rafts on the surface of these organelles.
Finally, although PI4KII␣ is intrinsically activated in the buoyant membranes, presumably via one or more specific covalent modifications, we have found a striking dependence of kinase activity on the endogenous cholesterol level. The virtual ablation of activity following cholesterol depletion indicates that PI4KII␣ activity is critically dependent on its immediate membrane environment. This observation is consistent with the existence of PI4KII␣ in cholesterol-rich rafts, and with studies on intact cells in which decreased PIP levels were found in buoyant membrane fractions prepared after treatment of A431 cells with M␤CD (33) or after treatment of Madin-Darby canine kidney cells with filipin (34). These results lead to the exciting hypothesis that cholesterol, which is known to affect vesicle trafficking in the Golgi-endosomal system, is a physiological regulator of PI4KII␣ activity.