Long Chain Ceramides Activate Protein Phosphatase-1 and Protein Phosphatase-2A

The search for potential targets for ceramide action led to the identification of ceramide-activated protein phosphatases, which include protein phosphatase-2A (PP2A) and protein phosphatase-1 (PP1) with roles in regulating apoptosis and cell growth. Thus far, in vitro studies on ceramide-activated protein phosphatases have been restricted to the use of short chain ceramides, limiting the extent of mechanistic insight. In this study, we show that the long chaind-erythro-C18-ceramide activated PP2A (AB′C trimer), PP2Ac (catalytic subunit of PP2A), and PP1γc and -αc (catalytic subunits of PP1γ and -1α isoforms, respectively) 2–6-fold in the presence of dodecane, a lipid-solubilizing agent, with 50% maximal activation achieved at approximately 10 μm d-erythro-C18-ceramide. The diastereoisomers ofd-erythroC18-ceramide,d-threo-, andl-threo-C18-ceramide, as well as the enantiomericl-erythro-C18-ceramide, did not activate PP1 or PP2A, but they inhibited PP1 and PP2A activity. The addition of phosphatidic acid decreased the basal activity of PP1c but also increased the stimulation byd-erythro-C18-ceramide from 1.8- to 2.8-fold and decreased the EC50 ofd-erythro-C18-ceramide to 4.45 μm. The addition of 150 mm KCl decreased the basal activity of PP1 and the dose ofd-erythro-C18-ceramide necessary to activate PP1c (EC50 = 6.25 μm) and increased the ceramide responsiveness up to 10–17-fold. These studies disclose stereospecific activation of PP1 and PP2A by long chain natural ceramides under near physiologic ionic strengths in vitro. The implications of these studies for mechanisms of ceramide action are discussed.

Several lines of evidence have suggested ceramide as an important regulator of various stress responses and growth mechanisms. First, formation of ceramide from the hydrolysis of sphingomyelin or from de novo pathways is observed in response to inducers of stress such as tumor necrosis factor-␣, ␥-interferon, 1,␣-25-dihydroxyvitamin D 3 , interleukin-1, ultraviolet light, heat, chemotherapeutic agents, FAS antigen, and nerve growth factor (1)(2)(3)(4)(5)(6)(7)(8)(9). Second, the addition of exogenous ceramide or the enhancement of cellular levels of ceramide induces cell differentiation, cell cycle arrest, apoptosis, or cell senescence in various cell types (4,10,11). Third, the action of ceramide relates mechanistically to key regulators of growth such as the retinoblastoma gene product (Rb), 1 caspases, Bcl-2, and p53 (12)(13)(14)(15)(16)(17)(18). Fourth, studies in yeast have demonstrated an essential role for sphingolipids in many stress responses where ceramide may function in the adaptation to heat (19,20). Finally, studies with knock-out mice lacking acid sphingomyelinase or with fumonisin B1, an inhibitor of ceramide synthesis, have disclosed necessary roles for ceramide in several pathways of growth regulation (21,22).
These emerging roles of ceramide necessitate a mechanistic understanding of ceramide action, This goal has led to the identification of several candidate ceramide-regulated enzymes, including ceramide-activated protein kinase and ceramide-activated protein phosphatase (CAPP) (23,24). CAPP was first identified as a member of the 2A class of serine/ threonine phosphatases (PP2A) (22)(23)(24). Recently, we have also demonstrated that protein phophatase-1 (PP1) is a target for ceramide. 2 The specificity for CAPP activation in vitro closely resembled the specificity for various cellular activities of ceramide such as apoptosis (23,24).
Possible direct downstream targets for these CAPP enzymes include c-Jun, protein kinase C␣, and Rb, which has been shown to function in ceramide-dependent cell cycle arrest pathways (3,5,13). Recent studies have demonstrated Rb as a specific substrate for PP1 in vitro and as an in vivo target for CAPP with dephosphorylation of Rb resulting from ceramide treatment in MOLT-4 cells. 2 Also in Molt-4 cells, protein kinase C␣ has been demonstrated to be regulated by PP2A and not PP1 in response to ceramide, demonstrating substrate specificities between CAPP enzymes (25). Galarreta and co-workers (26) demonstrated in vivo and in vitro that ceramide leads to dephosphorylation of c-Jun, and okadaic acid inhibited this effect in A431 cells. Therefore, c-Jun, protein kinase C␣, and Rb are likely candidates for direct substrates of CAPP in mediating ceramide effects.
Although short chain ceramides have clearly been demonstrated to activate the PP2A trimer (ABЈC), PP2A catalytic subunit (PP2Ac), and PP1 catalytic subunit (PP1c) in vitro, 2 a major problem has been the inability to utilize more natural long chain ceramides. In the present study, we overcome this problem of delivering long chain ceramides to both of the CAPP enzymes, PP1 and PP2A. We also characterize the enzymes under various physiological environments related to ceramide responsiveness.
Preparation of 32 P-Phosphorylated MBP-Myelin basic protein was labeled in a 0.5-ml reaction containing 1 mg of MBP, 50 mM Tris-HCl, pH 7.4, 90 mM MgCl 2 , 0.1 mM cold ATP, 5 mM dithiothreitol, 10 mM ␤-mercaptoethanol, 40 Ci of [␥-32 P]ATP (0.2 mM), and 125 units of protein kinase A. After the components were mixed, the reaction was incubated at 37°C for 2 h. 170 l of a 100% trichloroacetic acid solution was added, and labeled MBP was precipitated on ice for 20 min. The precipitated substrate was washed twice with acetone at Ϫ20°C, airdried, and reconstituted in 1 ml of 50 mM Tris-HCl, pH 7.4. The specific activity of 32 P-labeled MBP was ϳ22 Ci/mg.
Solubilization of Long Chain Ceramides and Phosphatidic Acid-C 18 -ceramide was brought up in a 2% dodecane-EtOH solution. The solution was dissolved by incubation at 37°C. After solubilization, the ceramide solution was kept at 37°C until addition to reaction tube. Egg yolk phosphatidic acid in chloroform was dried under nitrogen and resuspended in 50 mM Tris-HCl, pH 7.4, by ultrasonication.
Phosphatase Assays-Reactions were carried out in 13 ϫ 100-mm borosilicate glass tubes. Solubilized ceramides were added to tubes containing PP1␣c in Buffer A (50 mM Tris-HCl, pH 7.4) with the EtOH concentration not exceeding 1%. Stock enzyme was diluted to 1units/ml in Buffer A, and 10 milliunits was added to each tube. Components were preincubated for 5 min at 30°C. Reaction were initiated with 0.005 ml of 32 P-labeled myelin basic protein (1 mg/ml) in Buffer A. After 20 min at 30°C, the assay was terminated by the addition of 0.1 ml of 1 mM KH 2 PO 4 in 1 N H 2 SO 4 followed by the addition of 0.3 ml 2% ammonium molybdate. After 10 min, 1 ml of isobutanol:toluene (1:1) was added, and each reaction was vortexed for 10 s. The reactions were centrifuged at 1000 ϫ g for 10 min, and an aliquot of the upper organic phase was removed, mixed with scintillant, and counted. PP1␥c, PP2A trimer, and PP2Ac were assayed as described for PP1␣c. For PP1␥c and PP1␣c, 1 unit of activity is defined as the amount of enzyme that will hydrolyze 1.0 nmol of p-nitrophenyl phosphate/min at 30°C, pH 7.0. For PP2A trimer and the catalytic subunit, 1 unit of activity is defined as the amount of enzyme that will hydrolyze 1.0 nmol of phosphorylase/min at 30°C, pH 7.0.

Long Chain Ceramide Activates PP1 and PP2A in Vitro-
Cell-permeable short chain ceramides have been shown to activate protein phosphatase-1 and -2A in vitro (22)(23)(24). 2 The difficulty in delivering long chain ceramides in vitro and in vivo stems from solubility problems, because long chain neutral lipids are poorly soluble in aqueous environments. To overcome this problem of delivery in vitro, we mixed long chain ceramides with 2% dodecane, resulting in a final reaction concentration of 0.02% dodecane and variable concentrations of ceramide (29). Dodecane alone did not have any effect on phosphatase activity, but solubilized D-erythro-C 18 -ceramide activated PP1␣c (240% of control, EC 50 ϭ 8.75 M), PP1␥c (190% of control, EC 50 ϭ 11.5 M), PP2Ac (179% of control, EC 50 ϭ 11.25 M), and PP2A trimer (ABЈC) (580% of control, EC 50 ϭ 10.6 M) in a dose-dependent manner (Fig. 1). Initial activation was not observed until D-erythro-C 18 -ceramide concentrations reached 7.5 M with saturation occurring at 12.5-15.0 M D-erythro-C 18 -ceramide for 10 milliunits of enzyme ( Fig. 1). A nonspecific inhibition of both PP1 and PP2A en-zymes was observed at lower doses (3 and 5 M) of long chain ceramides (see below) ( Fig. 1).
Activation of PP1 and PP2A Is Stereospecific-To examine the stereospecificity of the phosphatase activation by long chain ceramides, we tested stereoisomers of C 18 -ceramide solubilized in dodecane (Fig. 2, A-D). D-Erythro-C 18 -ceramide (2S,3R) is the naturally occurring ceramide, with L-erythro-C 18ceramide (2R,3S) being its mirror image. The diastereoisomers, the threo conformations, differ from erythro conformations in that the C-2 configuration relative to C-3 is in a cis-conformation. In the erythro conformation, the groups are in a transconformation. We found that, similar to D-erythro-C 18 -ceramide, each stereoisomer inhibited the phosphatases at low doses (3-5 M), but only D-erythro-C 18 -ceramide activated each phosphatase at 7.5-15 M (Fig. 2, A-D).
Phosphatidic Acid Inhibits PP1 Enzyme Activity but Decreases the Dose of D-erythro-C 18 -Ceramide Needed to Activate PP1-In other studies, we have shown that PA inhibits protein phosphatase-1 but not PP2A activity in vitro. 3 Thus, we examined whether PA could regulate ceramide effects on PP1. Fig. 3 depicts PP1␥c activity in the presence of 500 nM phosphatidic acid and increasing doses of D-erythro-C 18 -ceramide. The addition of PA to the reaction lowered the basal activity of protein phosphatase-1 as expected, with 500 nM PA causing 95% inhibition of PP1␥c (Fig. 3). The inhibition of PP1␥c by low concentrations of long chain ceramides was relieved in the presence of PA, and the maximum stimulation by ceramide was enhanced from 374.3 to 581.0 fmol of P i released/min (Fig. 3), such that in the presence of PA, the fold stimulation by ceramide increased up to 2.8-fold. Importantly, lower doses of D-erythro-C 18 -ceramide were able to activate PP1␥c with an EC 50 of 4.45 M. PP1␣c behaved in a manner similar to PP1␥c, with an increase in maximal stimulation from 472.8 to 738.8 fmol of P i released/ min and a decrease in the EC 50 to 5.5 M (data not shown). Stereospecificity was still retained for both PP1␣c and PP1␥c under these conditions, further demonstrating the specificity of D-erythro-C 18 -ceramide (data not shown).
Near Physiological Ionic Strength Relieves the Inhibition by Unnatural Ceramides and Increases D-Erythro-C 18  to bind PP1 subunits under near physiological ionic strengths (30). Therefore, we examined whether 150 mM KCl affected ceramide responsiveness and PA inhibition. The addition of KCl lowered the dose of D-erythro-C 18 -ceramide necessary for activation of each phosphatase and increased the maximal stimulation of PP1␥c from 374.3 to 1753.3 fmol of P i released/ min (EC 50 ϭ 6.25 M) (Fig. 4A). Similarly, activation of PP1␣c was increased from 472.8 to 4137 fmol of P i released/min (EC 50 ϭ 5.55 M), PP2Ac from 352.6 to 1319.9 fmol of P i released/min (EC 50 ϭ 6.75 M), and PP2A (ABЈC) trimer from 1152.5 to 2403.4 fmol of P i released/min (EC 50 ϭ 6.25 M) (data not shown). The addition of salt also decreased the basal activity of each phosphatase by 50%, and therefore the fold stimulation was even higher. For example, ceramide increased the fold stimulation of PP1␥c from 1.9-to 8.9-fold in the presence of 150 mM KCl when compared with control activity in the absence of 150 mM KCl. The actual fold stimulation increased to 17.8-fold when compared with control activity in the presence of 150 mM KCl. Even low concentrations of KCl were able to increase ceramide responsiveness, as the addition of 15 mM KCl increased maximal stimulation of PP1␥c to 1103.3 fmol of P i released/min and decreased the EC 50 to 9.0 M (Fig. 4A).
Ceramide stereospecificity for each phosphatase was completely retained (Fig. 4B). Furthermore, L-erythro-C 18 -ceramide and PA retained an inhibitory effect, but the IC 50 increased from approximately 100 nM to 1 M in the presence of 150 mM KCl (data not shown). Inhibition of CAPPs by D-and L-threo-C 18 -ceramides was relieved under these conditions of 150 mM KCl (Fig. 4B). Also, the addition of 150 mM NaCl had the same effects as 150 mM KCl by both increasing D-erythro-C 18 -ceramide responsivenes and relieving nonspecific inhibi- Effects of Cations on PP1 and PP2A Activity and Ceramide Responsiveness-To examine whether the activation of PP1 and PP2A by long chain ceramides was dependent on or affected by cations, we preincubated the phosphatases with the different cations thought to be bound to the phosphatase metal binding site, Zn 2ϩ , Co 2ϩ , Fe 2ϩ , Fe 3ϩ , and Mn 2ϩ (31). Fig. 5 depicts the effects of these cations on PP1␥c activity. Unlike PP1␣c, PP1␥c is supplied without preincubation with Mn 2ϩ and was therefore used for these studies. Similar to other laboratories, we found that Mn 2ϩ increased PP1␥c activity by 200% (Fig. 5) (31). PP2A trimer and PP2Ac behaved in a similar manner, demonstrating a 175% increase in activity when preincubated with Mn 2ϩ (data not shown). Pre-binding PP1␥c with Mn 2ϩ increased the V max with ceramide from 1753.3 to 3201.3 fmol of P i /min but had no effect on the overall fold stimulation by ceramide (Fig. 5). Mn 2ϩ also had no effect on inhibition by PA (data not shown). Incubation with Co ϩ2 acted to increase basal activity by 25%, but it had no effect on ceramide responsiveness (Fig. 5). Incubation with Fe ϩ2, Fe ϩ3 , or Zn ϩ2 completely inhibited PP1 activity and ceramide activation (Fig. 5). Preincubation of the enzyme with Mn 2ϩ could not rescue the Zn 2ϩ and Fe 3ϩ inhibition (data not shown). DISCUSSION Ceramide-activated protein phosphatase was initially identified as a phosphatase-activated by cell-permeable D-erythro-C 2 -ceramide, and at least two phosphatases, PP1 and PP2A, are now known to respond to short chain ceramides (23, 24 -26, 33). Prior to this study, naturally occurring ceramides had not been shown to activate protein phosphatases. In this study, we demonstrate that, once solubilized with dodecane, D-erythro-C 18 -ceramide activates two CAPP enzymes, PP1 and PP2A. This activation was very specific, as the unnatural stereoisomers of D-erythro-C 18 -ceramide did not activate PP1 nor PP2A. These observations are important for several reasons. First, specificity for ceramide is now clearly demonstrated for the long chain ceramide, and this stereochemical specificity is better defined with the long chain isomers than with the short chain ones. Second, a reproducible assay has now been developed that allows the study of CAPPs under more physiologic conditions (natural ceramides and near physiological ionic strength). Third, a specific ceramide-binding/interaction site is now inferred to be present on the catalytic subunit of at least PP1 and PP2A.
Our system demonstrates strict stereospecificity for D-erythro-C 18 -ceramide, because neither its enantiomer, L-erythro-C 18 -ceramide, nor its diastereomers, D-and L-threo-C 18 -ceramide, increased CAPP activity. The fact that even the mirror image of naturally occurring D-erythro-C 18 -ceramide, L-erythro-C 18 -ceramide, did not activate the phosphatases shows that the interaction is very specific and not an environmental effect on the enzymes.
In studying the optimal environment for delivery of long chain ceramides to CAPP, we initially found, as did Hirabayashi and co-workers (29) in cell studies, that dodecane acted as a useful vehicle for solubilizing long chain ceramides. We also found that the addition of near physiological levels of KCl to the reaction reduced inhibition of PP1 by both PA and long chain ceramides while enhancing ceramide activation of each phosphatase studied. Only PA and L-erythro-C 18 -ceramide remained inhibitory under these conditions, although the dose necessary for inhibition of PP1 was increased approximately 10-fold. Importantly, stereospecificity was still retained such that D-erythro-C 18 -ceramide was very effective, whereas its enantiomer, L-erythro-C 18 -ceramide, lacked any activity. At this point, it is not clear why increasing the salt decreases the inhibition of PP1 by PA and low doses of long chain ceramides while enhancing the responsiveness of PP1 and PP2A to D-erythro-C 18 -ceramide. It is assumed that the enzyme's conformation is affected under physiological salt conditions such that with no salt, there is easier access to many lipids nonspecifically, whereas under near physiological ionic strength, interactions are more specific.
These findings also suggest that a specific binding site for D-erythro-C 18 -ceramide is present on both the PP1 and PP2A catalytic subunits. The stereospecificity of this activation suggests a direct and specific interaction of ceramide with the catalytic subunits of these phosphatases. Because the enantiomer and diastereomers of D-erythro-C 18 -ceramide do not activate the phosphatases, the orientation of the C-2 and C-3 carbons relative to the sphingolipid backbone is suggested to be important for proper interaction and binding of ceramide to CAPP.
Another intriguing observation emerged with the finding that the PP2A heterotrimeric complex was activated to a greater extent than the catalytic subunit alone. This finding suggests that the A and B subunits of trimeric PP2A, which are important for compartmentalization and substrate specificity (38,39), may also impart a conformation that allows for greater ceramide stimulation. Alternatively, ceramide may also interact with the A and B subunits to impart greater phosphatase activity. A third possibility is that A and B preferentially interact with a form of C that is more responsive to ceramide. These possibilities are currently under investigation.
The dual action of PA and D-erythro-C 18 -ceramide on PP1 carries important implications for the physiological environment in which PP1 resides. CAPP phosphatases such as PP1 and PP2A have roles in apoptosis and cell cycle arrest. Therefore, a role for PA, a product of phospholipase D, may be hypothesized in the suppression of phosphatase activity until the proper stimuli (e.g. ceramide) are generated to protect the cell from undergoing premature apoptosis or inhibition of growth.
Establishing the conditions necessary for delivery of long chain ceramides to PP1 and PP2A allowed us to examine the influence of cations on PP1 activity and ceramide responsiveness. We demonstrated, as had others, that preincubation with Mn 2ϩ increased PP1c basal activity (31). We also found that Mn 2ϩ increased PP2Ac and PP2A trimer basal activity. Mn 2ϩ did not, however, affect ceramide responsiveness. We also examined other cations of which Zn 2ϩ , Fe 3ϩ , and Fe 2ϩ potently inhibited PP1 and PP2A activity in vitro; ceramide was not able to overcome this inhibition of PP1 and PP2A. The inhibition may be the result of allosteric effects as suggested by Schlender and co-workers (31). The influence of Zn ϩ2 inhibiting apotosis is well known (36,37), and our observations suggest yet another target for this anti-apoptotic cation.
In this study, we have demonstrated that both PP1 and PP2A are activated by long chain ceramides and that this activation is stereospecific. We have also demonstrated that PA and salt have important effects on this activation by lowering the ceramide concentration necessary for achieving phosphatase activation. We also show that Fe 2ϩ/3ϩ and Zn 2ϩ inhibit PP1 and PP2A activity, with ceramide not able to overcome cation inhibition. Clearly, this study demonstrates several new avenues of CAPP regulation and that physiologic environments can enable these enzymes to respond to natural ceramide rapidly and effectively.