activated protein phosphatase (CAPP) was found to share several properties .... sequence was based on the published sequence of the human B, subunit (13).
Vol. 268, No. 21, Issue of July 25. pp. 15523-15530,1993 Printed in U.S.A.
THEJOURNAL OF BIOLOGICAL CHEMISTRY 0 1993 by The American Society for Biochemistry and Molecular Biology, Inc.
Ceramide Activates Heterotrimeric Protein Phosphatase 2A* (Received for publication, January 29, 1993, and in revised form, April 7, 1993)
Rick T. DobrowskySS, Craig Kamibayashili, Marc C. Mumbyli, and Yusuf A. HannunSIJ From the +Departmentsof Medicine and Cell Biology, Duke University Medical Center, Durham, North Carolina 27710 and the TDepartment of Pharmacology, University of Texas Southwestern Medical Center, Dallas, Texas 75235-9041
Ceramide activates a cytosolic protein phosphatase 3). Tumor necrosis factor-a also induces sphingomyelin hyU937 present in rat glioma T9 cellsand rat brain. Ceramide- drolysis andceramideproductionwithin15minin activated protein phosphatase (CAPP) was found to monoblastic leukemia cells (4). In rat T9 glioma cells, nerve share several properties with protein phosphatase 2A growth factor stimulates sphingomyelin hydrolysis and cer(PPPA) leading to the hypothesis that ceramide may amide production within10-15 min.’ These observations supdirectly activate PP2A.PPZA was isolated as a heter- port the presence of a “sphingomyelin cycle” which is actiotrimer (AB’C, AB,$), heterodimer (AC), or free C vated in response to various agonists generating the potential subunit, and the effect of ceramide on the catalytic lipid second messenger, ceramide (1). activity wasassessed. Cz-ceramide, 5-20 pM, activated The characterization of ceramide as the major metabolite heterotrimeric PP2A up to 3.5-fold but had no effect of agonist-induced sphingomyelinhydrolysissuggests that on the activity of AC or C. Ceramides possessing hexceramide may mediate, in part, some of the biologic effects of anoyl, decanoyl, and myristoyl but not stearoyl acyl chains also activated heterotrimeric PP2A. Ceramide theses cytokines, hormones, and growth factors. Studies utiactivation of heterotrimeric PPZA required the pres- lizingceramide analogs have demonstratedthat ceramide indeed possessesbioactivity andcan directly inhibit cell ence of a B subunit since trypsinization or heparin cell differentiation (5, 61, down-regulate treatment abolished ceramide activation. Activation of growth,induce heterotrimeric PPZA was specific for ceramide be- expression of c-myc mRNA? induce apoptosis in U937 cells (4),andactivate NFKB in permeabilized Jurkat cells (7). cause related sphingolipids had no effect. Moreover, dihydro-Cz-ceramide,whichlacksthe trans double Intriguingly, a closely relatedcompound, dihydroceramide, bond in the sphingoid base, inhibited AB‘C activity by which lacks the trans double bond, was inactive in inducing >90% at 10 pM. The specificity of activation of AB’C apoptosis (4). This degree of specificity suggeststhat ceramide and AB,C by stereoisomers of Cz-ceramide was found may interact with a specific cellular target. to differ. Whereas activation of AB’C by either DLprotein phosphatase Recently, a ceramide-activated erythro- orthreo-Cz-ceramide was similar,AB,C was (CAPP)3 present in cytosol of rat T9 glioma cells and rat activated by either D- or r.-erythro-Cz-ceramide but brain has been identified and proposed as a putative molecular not by the threo isomers. CAPP isolated from T9 cells target for ceramide (8, 9). CAPP is specifically activated by was most effectively activated by D-erythro-Cz-cerPM ceramides of varying hydrophobicities and is not 1-20 amide. CAPP was found to possess two peaks of cerstimulated by either sphingosine or sphingomyelin (8).CAPP amide activated phosphatase activity. The initial peak of activity wascoincident with the elution of AB’C and shares severalbiochemical properties with PPBA, Le., it is was stimulated 1.8-fold by 20 p~ Cz-ceramide. A sec- cation-independent, inhibitedby okadaic acid with an IC,,, of ond peak of phosphatase activity was negligible in the 0.5-1 nM, and insensitive to inhibition by inhibitor 2 (8, 9). absence of ceramide but was stimulated 5.5-fold by 20 However, ceramide does not activate the catalytic subunitof pM Cz-ceramide. These results support the hypothesis PPSA (9) suggesting that CAPP may be related to heterotrimod- meric or heterodimeric forms of PP2A . that ceramide is a specific lipid second messenger ulating heterotrimeric PP2A activity. PP2A has been isolatedfrom numerous tissues asa heterotrimeric complex (ABC) possessing one structural (A), one regulatory (B), and one catalytic subunit (C) (10-14). The A subunit exists as a M , = 60,000-65,000 protein while the C Ceramide has recently been identified aasdirect metabolic subunit possesses a M , = 36,000-38,000 (10, 11). Biochemical product of agonist-induced sphingolipid metabolism inseveral studies have also identified several distinct B subunits with cell types. For example, inHL-60 human leukemiacells, 1,25- M , = 55,000 (B), 54,000 (B’), and 74,000 (B”) (12-15). Addidihydroxycholecalciferol(vitamin D3),tumor necrosis factor- tionally, based on the cDNAs, several isoforms exist for the a, andy-interferon induce sphingomyelin hydrolysis with B subunits ( a , /3, y ) (16). The physiological significanceof subsequent generation of ceramide and phosphocholine (1- thisheterogeneityiscurrentlyunknown. However, recent B subunit regulates catalytic activity studies indicate that the * This work was supported in part by National Institutes of Health (17, 18).As such, and substrate specificity of the AC complex Grants GM43825 (to Y. H.), HL 31107 (to M. C. M.), and by a scholarship from the PEW Foundation. The costs of publication of association of the various B subunits with AC may differenthis article were defrayed in part by the payment of page charges. tially regulate the activity of this complex. It has been sug~
~~
~
This article must therefore be hereby marked “aduertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. 5 Recipient of National Research Service Award AG05531. 11 Mallincrokdt Scholar. To whom correspondence should he addressed Dept. of Medicine, Div. of Hematology/Oncology, Box 3355, Duke University Medical Center, Durham, NC 27710. Tel.: 919-6842449; Fax: 919-681-8253.
’ R. T. Dobrowsky and Y. A. Hannun, unpublished observation.
R. A. Wolfe, R. T. Dobrowsky, L. M. Obeid, and Y. A. Hannun, manuscript in preparation. The abbreviations used are: CAPP, ceramide-activated protein phosphatase; PPZA, protein phosphatase 2A; MBP, myelinbasic protein.
15523
Ceramide Activates Heterotrimeric PP2A
15524
gested that B subunit inhibition of the catalytic activity of AC results in a low basal level of activity of ABC in uiuo (11). However, to date, no physiologic activators of t h e ABC complex have been convincingly described. The similarities between CAPP and PP2A led t o the hypothesis that ceramide-activated phosphatase may be a specific subtype of PP2A. We therefore examined the effect of ceramide on the phosphatase activityof highly purified preparations of various PP2A subtypes, i.e., AB'C, AB$, AC, and C, and compared these activities withCAPP isolated fromT9 glioma cells a n d rat brain. Our data suggests that, in contrast is a specific lipid activator of PP2A. to other lipids, ceramide Further, ceramide activation requires the presence of the B subunit. The role of heterotrimeric PP2A as a potential molecular target for the action of ceramide is discussed. EXPERIMENTALPROCEDURES
Materials [y-32P]ATPwas obtained from Du Pont-New England Nuclear. Sphingosine, histone type 111-S, myelin basic protein, bovine serum albumin, catalytic subunit of cAMP kinase, partially dephosphorylated casein, dihydrosphingosine, trypsin, soybean trypsin inhibitor, and sphingomyelin were purchased from Sigma. Cell culture media and fetal bovine serum were products of Life Technologies, Inc. Methods Cell Culture-T9 rat glioma cells were maintained and cultured as previously described (7). Briefly, cells were maintained in Dulbecco's modified Eagle's medium, 10% fetal calf serum and grown in an atmosphere of 5% CO, at 37 "C. Partial Purification of CAPP-Confluent cultures of T9 cells were trypsinized, washed with phosphate-buffered saline, andresuspended to 1.1 X lo7 cells/ml in 10 ml of homogenization buffer (20 mM TrisHCI, pH 7.4, 1 mM EDTA, 0.1 mM EGTA, 1 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride, 0.5 mM dithiothreitol, 2.5 pg/ml leupeptin, and 2 pg/ml pepstatin A). Cells were disrupted by nitrogen cavitation at 350 p s i . for 10 min a t 0 "C and theresulting homogenate centrifuged at 100,000 X g for 1 h at 4 "C. Cytosol (1.7 mg/ml, 13.6 mg) was adsorbed to 1 ml of packed DEAE-Sephacel for 30 min at 4 "C and packed intoa1-mlfast protein liquid chromatography column. The column was washed with Buffer A (20 mM Tris-HC1, pH 7.4, 1 mM EDTA, 0.1 mM EGTA, 1 mM benzamidine, 0.5 mM dithiothreitol, 10% glycerol) until the returned to base line, and the protein was step eluted with 100,200, and 300 mM NaCl in Buffer A. Fractions were assayed for CAPP activity as described below. CAPP activity primarily eluted in the 300 mM salt fraction (0.22 mg/ ml, 1.99 mg) which wasdiluted 3-fold with Buffer A and applied to a Mono-Q HR 5/5 column (Pharmacia LKB Biotechnology Inc.). Protein was eluted with a 25-ml linear gradient from 100 to 300 mM NaCl at a flow rate of 0.5 ml/min. Fractions of 0.5 ml were collected and assayed for CAPP activity. Active fractions (0.06 mg/ml, 0.24 mg) were aliquoted, snap frozen in liquid Nz, and stored at -80 "C. CAPP was partially purified from rat brain by a similar procedure except that the 300 mM NaCl fraction was first applied to a column of heparin-Sepharose. The flow through was collected and subsequently applied to Mono-Q. Protein concentrations were determined by Coomassie Blue binding using bovine serum albumin asthe standard (19). Protein Phosphatase Preparations-Heterodimeric PP2A (AC) and the catalytic subunit of PP2A (C) were purified from bovine heart or brain as described elsewhere (12). HeterotrimericPP2A (AB'C, AB$) were purified from bovine cardiac and brain tissues, respectively, using the same procedure (12). The identity of the Mr= 54,000 B subunit in the bovine brain PP2A preparation was determined by reactivity with antisera generated against a synthetic peptide whose sequence was based on the published sequence of the human B, subunit (13). The M,= 54,000 B subunit in the cardiac preparation was distinct from B, since it did not cross-react with antisera against the B, subunit. The tentative classification of this subunit as B' is based on itsapparent molecular weight and elution from anion exchange columns, which most closely resemble those of the PP2Ao form described in rabbit skeletal muscle (15). The two preparations
were composed of a single type of B subunitas judged by immunoblot analyses with the two antisera.4 Preparation of ~ 2 P I P h o s p h o ~ l a t eSubstrate~-[~~P]Phosphohid stone was prepared as previously described (8). 32P-Myelin basic protein and [32P]caseinwere prepared by phosphorylation with the catalytic subunit of cAMP kinase. The reaction contained 50 mM Tris-HCI, pH 7.4, 10 mM MgCI,,0.1 mM [y- 32P]ATP (2-7 X IO3 cpm/pmol), 0.5 mM dithiothreitol, 10 mM P-mercaptoethanol, 0.04% Brij 35,5 mg/ml MBP or casein, and 250 units/ml catalytic subunit CAMP kinase in a volume of 0.5 ml. Reactions were incubated for 2 h at 37 "C and quenched with 0.17 mlof ice-cold 100%trichloroacetic acid. Tubes were placed on ice for 30 min and centrifuged at 15,000 X g for 10 min at 4 "C. The precipitate was washed with 2 X 1-ml aliquots of -20 "C acetone and resuspended in 1 ml of water. To aid solubilization of [32P]casein,20 p1 of 2 M Tris were added. Concentration of labeled substrates was based on the specific activity of the [y32 PIATP. Typical preparations of 32P-MBP,[32P]casein,and ["PI phosphohistone contained 3-9 nmol of Pi/mg, 2.5-5 nmol of Pi/mg, and 0.5-1 nmol of Pi/mg, respectively. Assays wereperformed so that hydrolysis did not exceed 20% of total substrate added. Phosphatase Assay-Dephosphorylation reactions contained 50 mM Tris-HCI, pH 7.4, 1 mM EDTA, 10-200 ng of protein, and the indicated concentration of radiolabeled substrate in a final volume of 0.1 ml. Assays utilizing [3ZP]phosphohistoneas substrate also contained 40 pg/ml histone. All lipid solutions were prepared fresh and dissolved in 10%ethanol to a concentrationof 0-200 p~ immediately prior to addition to assays. Ten-microliter aliquots were added to reactions giving 0-20 p M lipid and a final ethanol concentration of 1%.Reactions were typically run for 10 min at 37 "C unless indicated otherwise. Reactions were terminated by addition of 0.1 ml of 1 mM KH,PO, in 1 N HzS04.Released ["P04]i was quantitated by scintillation spectrometry after addition of 0.3 ml of 2% ammonium molybdate and extraction of the phosphomolybdate complex with 1 ml of to1uene:isobutanol (l:l, v/v). CAPP activity was quantitated as the difference in activity in the presence versus absence of ceramide. One activity unit was defined as liberation of 1 nmol of PJmin. Subunit Dissociation of of AB'C and CAPP-Partially purified CAPP from rat brain (20 pg) and AB'C (0.8 pg) were incubated with trypsin in 50 mM Tris-HCI, pH 7.4 at a ratio of 25:l (w/w) for either l h a 37 t "C or 4 hat 4 "C (18).Reactions were terminated by addition of a 10-fold excess of soybean trypsininhibitor andan aliquot removed for activity measurements. Control incubations contained either enzyme plus buffer or trypsin inhibitor. In some experiments 30 pg/ml heparin was added to AB'C (0.8 pg) and incubated for 4 h at 4 "C after which an aliquot was removed foractivity measurements. Synthesis of Stereoisomers of Cz-ceramide-C,-, CC-, CW, C14-, and Cls-ceramide were prepared by reaction of sphingosine with the appropriate fatty acyl anhydride and purified as described elsewhere (8). Stereoisomers of C,-ceramide were prepared from enantiomerically pure erythro or threo sphingosine as previously de~cribed.~ RESULTS
Effect of Cn-ceramide on Dephosphorylating Activityof PP2A Subtypes-Thesimilarities of CAPP t o PP2A led t ot h e hypothesis that this class of protein phosphatase may be responsive to ceramide. We therefore incubated highly purified preparations of PP2A in the presence of increasing concentrations of ceramide. Results depicted in Fig. 1 indicate that heterotrimeric forms of PP2A (AB&, AB'C) were stimu l a t e d u p t o1.7-fold b y 20 P M Cz-ceramide. However, heterodimeric PP2A (AC) or the catalytic subunit were essentially unaffected. In comparison, CAPP from T9 cells showed a 2.4-fold increase in responseto 20 g M Cz-ceramide. Heterotrimeric and heterodimeric PPBA are readily sepaTOfurther confirm rated by chromatography on Mono-Q (18). that ceramide activates heterotrimeric but not heterodimeric PP2A, purified AB'C and AC wereseparatelychromatographed on a Mono-Q column and fractions were assayed for ceramide-activated phosphatase activity.As indicated in Fig. 2, onlyheterotrimeric PP2A wasresponsivetoceramide.
(c)
C. Kamibayashi and M. C. Mumby, unpublished observations. Bielawska, R. T. Dobrowsky, R.A. Wolfe, D. Liotta, L. M. Obeid, and Y. A. Hannun, submitted for publication.
PP2A
Heterotrimeric Activates Ceramide
C, -Ceramide, pM
FIG.1. Activation of various phosphatase preparations by Cz-ceramide. Phosphatase preparations were diluted in Buffer A and incubated under standard conditions for 10 min at 37 ”C in the presence of 0, 1, 5, 10, 15, and 20 p M C2-ceramide using 0.5 p M [”PI phosphohistone as substrate. Reactions were terminated, and the released [3zP04]iwas quantitatedas described under “Methods.” M in 10% Lipids were prepared as 10, 50, 100, 150, or 200 ~ L solutions ethanol immediately prior to addition to the assays and diluted 10fold giving the indicated concentrations in 1%ethanol. Results are expressed as percent of initial activity in the absence of C2-ceramide. Data are mean S.D. of triplicate determinations and are representative of three experiments. Initial activities (in milliunits/ml) were: (B) CAPP, 7.5; (0)AB’C, 27.3; (0)AB,C, 31.8; (A) AC, 4.3; and (A) C, 20.3.
*
15525
mide also activated a phosphatase in crude cytosol isolated from T9 glioma cells. Heterotrimeric PP2A (AB,C) was similarly activated by ceramide containing various acyl chains (Fig 3). N-Acyl substitution of sphingosine with increasingly hydrophobic fatty acids shifted the potency for activation of AB,$ by ceramide such that Cl0- > CS- > C2-ceramide. N Acyl substitution of sphingosine with either myristic or stearic acyl chains resulted in a progressive loss of activation uersus CIo-ceramide.Indeed, Cls-ceramide failed to activate AB,C at all concentrations tested. These results clearly demonstrate that more hydrophobic ceramide molecular species are capable of activating PP2A. The inability of CIS-ceramide,which is more representative of naturally occurring ceramides, to activate AB,C may be related to poor solubility or formation of lipid aggregates which may not be recognized by the enzyme. Preliminary experimentsindicated that ceramide activation of heterotrimeric PPBA was sensitive to theinitial activity of the enzyme preparation. Incubation of decreasing amounts of AB,C (Fig. 4A) or AB’C (Fig. 4B)with ceramide resulted in an increasing extent of activation such that at7.1 milliunits/ ml, 20 PM ceramide activated AB,C by 280%; approximately 2-fold greater than at aninitial activity of 31.8 milliunits/ml (Fig. 4A).However, aplateau was reached after which increasing dilution did not further enhance ceramide stimulation (Fig. 4B).This enzyme-dependent effect explains the modest increase in activity for AB,C and AB’C in Fig. 1, since these assays were performed at a higher initial activity (27-32 milliunits/ml). The ratio of lipid to enzyme similarly affected CAPP activity from T9 cells (data not shown). The reason for the effect of enzyme concentration on ceramide activation was not determined. However, the effect of enzyme dilution Acyl Chain Length
-2
-C+-
Acetyl
Hexanoyl
Y
400
300 0.40 -
0.30 . 0.20
’
0.10
’
0.00
200
100
0
10
20
30
40
50
60
0 Praccion X
FIG.2. Chromatographic behavior of ceramide-activated phosphatase. AB’C (0,O) and AC (0,H) were applied to a MonoQ column and eluted with a linear gradient of0-400 mM NaCl in Buffer A over 60 ml a t a flow rate of 1 ml/min. One-ml fractions were collected and 75 p1 assayed under standard conditions for 30 min a t 37 “C in the absence ( 0 , O ) or presence (0,B) of 20 p M Cz-ceramide using 0.5 p M 32P-MBP assubstrate. Reactions were terminated and released [32P04]tquantitated asdescribed under “Methods.” These studies clearly demonstrate that ceramide activates heterotrimeric but not other forms of PP2A. Previously, we demonstrated that both CS- and Cls-cera-
0
0.01
0.1
1
to
Ceramide, pM FIG.3. Ceramide activation ofAB,C is affected by acyl chain length. Phosphatase preparations were diluted in Buffer A to 8 milliunits/ml and incubated under standard conditions for 10 min at 37 “C in the presence of the indicated concentrations of ceramides using 0.5 p~ 32P-MBP assubstrate. Reactions were terminated and released [32P04Jiquantitated as described under “Methods.” Results are expressed as percent of activity in the absence of ceramide. Data are the means of triplicate determinations from a single experiment which is representative of results obtained in three independent experiments.
Cerarnide Heterotrirneric Activates
15526
PP2A
4001
0 0
a,
E (AB'C)
5
10
15
20
25
1
B-(PEA-AC)
C2Ceramide, pM
FIG. 4. Ceramide activation of AB,C and AB'C is affected by protein concentration. Phosphatase preparations were diluted in Buffer A to give the indicated initial activity and incubated under standard conditions for 10 min at 37 "C in the presence of 0, 1,5, 10, 15, and 20 pM C,-ceramide using 0.5 pM [32P]phosphohistone asthe substrate. Reactions were terminated and released [32P0,], quantitated as described under "Methods." Results are expressed as percent of activity in the absence of ceramide. Data are the means f S.D. of triplicate determinations and are representative of two experiments. Initial activities (in milliunits/ml) were A , AB&: (0)31.8, (0)17.6, ).( 5.6; B, AB'C: ( 0 )11.3, (0)5.4,).( 1.5.
on ceramide activation underscores thenecessity of utilizing enzyme preparationsof approximately equal activity to compare quantitatively the effect of ceramide between various heterotrimeric phosphatases. The above data raised the possibility thatthe lack of ceramide activation of AC and C may also be related to the ratio of lipid to enzyme. Fig. 5 shows that, regardless of initial activity, neitherAC nor C show any dose-dependent ceramide activation. Moreover, we have previously observed that cer-
C2€eramide, 1M
FIG. 5. Ceramide does not activate the C and AC forms of PPSA. Phosphatase preparations were diluted in Buffer A to give the indicated initial activity and incubated under standard conditions for 10 min a t 37 "C in the presence of 0,5,10, and 20 @iM Cs-ceramide using 0.5 p~ [32P]phosphohistoneas the substrate. Reactions were terminated and released [32P04]iquantitated as described under "Methods." Results are expressed as percent of activity in the absence of ceramide. Data are mean f S.D. of triplicate determinations and are representative of two experiments. Initial activities (in milliunitsl ml) were A , C subunit: (0)12.9, (0)6.6, (a)4.3; B, AC: (0)5.1, (0) 3.3, (a)1.8.
amide has no effect on the activity or inhibition by okadaic acid of the C subunit isolatedfrom ratbrain (9). Taken together, we conclude thatthese PP2A subtypesarenot responsive to ceramide. Ceramide Activation of AB'C and CAPP Requires the B Subunit-Treatment of AB'C with trypsin or heparin specifically dissociates the B subunit from the heterotrimer and
Ceramide Activates Heterotrimeric PP2A
15527
AB'C also showed a similar degree of specificity to activation by ceramide but not sphingosine (data not shown). Additionally, using 32P-MBP as substrate, sphingosine was unable to activate the phosphatase activity of AB'C at 20 pM (Table 1).Interestingly, dihydro-C2-ceramidealso was ineffective at activating AB'C. Several sphingolipids, phospholipids, diolein, and oleic acid were also ineffective activators of AB'C (Table 1). Although the activation of AB'C by ceramide appears dependent on enzyme concentration, the various lipids tested showed no ability to activate AB'C at either of the enzyme concentrations tested. These data suggest that ceramide is a relatively specific lipid activator of AB'C. Previous work from our laboratory indicatedthat dihydroceramide lacks the bioactivity profile characteristic of ceramide (4).As indicated in Table 1,dihydroceramide was also unable to activate AB'C. Indeed, dihydro-Cz-ceramide actually inhibited the activity of AB'C in a dose dependent manner such that 10 FM dihydro-C2-ceramideinhibited phosA phatase activity by 290% (Fig. 7). These datastrongly suggest that activation of PP2A by ceramide is dependent upon the presence of a double bond in the sphingoid backbone as well as a minimum hydrophobic character. Effect of Stereoisomers of C1-ceramide on PPZA and CAPP Activity-Given the specificity of activation associated with the presence of the transdouble bond of ceramide, it was also of interest to determine if some degree of stereochemical specificity existed for ceramide activation of PPZA, Ceramide possesses two chiral centers with four possible stereochemical configurations (D- or L-erythro and D- or L-threo). However, ceramides present in biologic sources are primarily of the Derythro configuration (20). Enantiomerically pure stereoisomers of C2-ceramide were prepared from the corresponding sphingosines and tested for their ability to activate AB'C, Control Trypain Heparin AB,C, and CAPP from T9 cells. All stereoisomers of C2B ceramide activated AB'C to approximately 5-6-fold of control activity (Fig. 8A). However, the effect of Ce-ceramide stereoisomers on the phosphohistone phosphatase activityof AB& demonstrated greater specificity. Both D- and L-erythro-Cqceramide activated phosphohistone phosphatase activity of . under identiAB,C approximately 2-fold at 20 p ~ However, cal conditions, both D- and L-threo-Cz-ceramide were poor activators (Fig. 8B). That AB$ is not stimulated by D- or Lthreo-Cz-ceramide and possesses a nonhomologous subunit structure with B' (13) may indicate that theB subunitconfers some level of specificity for the interaction of PP2A with stereoisomers of ceramide. Interestingly, CAPP isolated from T9 cells was found to be more responsive to stimulation by D-erythro-C2-ceramide,with other stereoisomers being up to 2-fold less active (Fig. 8C).The increased activity of Derythro-C2-ceramide versus other stereoisomers on the activation of CAPP was not substrate-dependent, in that similar results were obtained using 32P-MBP (data notshown). The Conaol Tryp~in Tryp Inh ltyp+TrypInh initial activity of all phosphatases was approximately equal, FIG. 6. Trypsin and heparin treatment of AB'C and CAPP suggesting that the differential response to the C2-ceramide from rat brain abolishes ceramide activation. AB'C (0.8 fig, A ) stereoisomers may reside in differences in the composition of and CAPP (20 pg, B ) were incubated with trypsin in 50 mM Tris- the B subunit between the two phosphatase preparations. HCI, pH 7.4, at a ratio of 25:l (w/w) for 1 h at 37 "C or 4 h at 4 "C, CAPP Activity from T9 Cells Is Distinct f r o m AB'C and respectively. Addition of a 10-fold excess of soybean trypsin inhibitor AB,C-The above data suggested that CAPP activity from terminated the reaction, and an aliquot was removedfor activity measurements. Control incubations contained either enzyme plus T9 cells may be distinct from both AB'C and AB& To buffer or trypsin inhibitor. Heparin treatment of AB'C was for 1 h ascertain if CAPP isolated from T9 cells is associated with a t 4 "C using 30 gg/ml heparin. Assays were performed under stand- AB'C or AB,C, partially purified enzyme preparations were ard conditions in the presence of 0 or 20 W M C2-ceramide for 10 min rechromatographed on a Mono-Q column, and fractions were at 37 "C using 0.5 p M [32P]phosphohistoneor 32P-MBP assubstrate. assayed for ceramide activated phosphatase. Fig. 9 shows the Phosphatase activity (solidbar, lower stack) is expressed asthe percent of control incubation (buffer only). Ceramide-stimulated elution of ceramide-activated phosphatase isolated from T9 activity (striped bar, upper stack)is expressed as the percent activity cells in relation to purified AB'C and AC from bovine heart. in the absence of ceramide for each treatment. Two peaks of ceramide activated phosphatase were observed.
releases the B subunit inhibition of the catalytic activity of AC (18). As such, if ceramide activation requiresthe presence of the B subunit, dissociation via trypsinization or heparin treatment should abolish the response of AB'C to ceramide and increase base line activity. Indeed, treatment of AB'C with either trypsin or heparin totally abolished the activation of AB'C by ceramide (Fig. 6A). Moreover, trypsinization of partially purified CAPP from rat brain also resultedina decrease in ceramide activation from 3.7- to 1.1-fold (Fig. 6B). These results strongly indicate that ceramide activation of heterotrimeric PP2A requires the presence of the B subunit. Specificity of Ceramide Activation of Heterotrimeric PP2AUsing [32P]phosphohistoneas substrate, CAPP is not activated by the related sphingolipids, sphingosine and sphingomyelin, nor other unrelated lipids such as diolein and phosphatidylcholine (8).Using [32P]phosphohistone assubstrate,
1
15528
PP2A
Ceramide Heterotrimeric Activates
TABLEI Effect of various lipids on activation of MBPphosphataseactivity of AB ‘C AB’C was incubated under standard conditions with 0.4 p~ 32P-MBPfor 10 min at 37 “C in the presence or absence of the indicated lipids. Data are expressed as percent of activity in the absence of lipid, and values are means f S.D. of triplicate determinations. Activity in the presence of 20 p~ C2-ceramidewas 164 f 15% and 139.6 f 3.3% a t 39 milliunits/ml and 86 milliunits/ml, respectively. Percent of control
Lipid
AB‘C 39
AB‘C 86
5 PM
Dihydro-C2-ceramide Sphingosine Dihydrosphingosine Sphingomyelin Dioleoyl-phosphatidylcholine Phosphatidylethanolamine Diolein Oleic acid
20 p M
rnilliunits/rnl
ND
50.1 f 24.1 2.5 83.5 k104.1 14.7 98.7 f 13.3 f 7.5 65.771.4 60.2 f104.8 15.9 65.1 f 1.7 ND” ND
f 6.5 f 44.3 115.7 f75.4 42.1 f 16.6 f52.6 11.2
136.1 f 13.9 26.4 ND
5 PM
20 p M
rnilliunits/rnl
71.0 f 15.8 53.6 105.1 I 21.8 f107.9 15.1 49.4 k 43.5 11.7 I 27.1 f 8.7 1.4 f 1.26 33.1 f 8.0
f 7.6 65.3 f 8.7 f 25.1 I 17.0 58.4 I 20 43.3 I 7.7 O I O
2.3 f 1.3
ND, not determined.
(8, 9). That CAPP shared many biochemical properties with PP2A suggested that this class of phosphatases may be activated by ceramide. The current results support the hypothesis that ceramide canactivatePP2A.Importantly,activation requires the presenceof a B subunit since 1) neither AC nor C were significantly affected by ceramide, 2) chromatography of AB‘C and AC found ceramide activation only associated with the heterotrimer, and 3) dissociation of the B subunit of AB’C by trypsinization or heparin treatment abolished ceramide activation. Severalstudieshaveshown ahigh degree of homology between Aand C subunits of various PP2A preparationsfrom very divergent species (10,11, 13).However, the B subunit is comprised of at least three distinct families of proteins with several possible genetic isoforms (16). Previous studies have indicated that the B subunit of heterotrimeric PP2A modulates the catalytic activity of the holoenzyme as well as the substrate affinity (17, 18).We extend the regulatory role for 10 0 .1 1 this subunit by proposing that its presence is necessary for the activation of heterotrimeric PP2A by ceramide. Our data DihydrO-CzCerunide, additionally suggest that functionaldifferences exist between FIG. 7. Dihydro-Cz-ceramide inhibits AB’C. AB’C (6.7 mil- B subunits in theirresponsiveness and specificity to ceramide liunits/ml) was incubated under standard conditions in the presence and itsstereoisomers. That theB subunits of AB’C and AB,C of 0, 0.1, 1, and 10 / r dihydro-C2-ceramide ~ for 10 min at 37 “C using 0.5 p~ 32P-MBP assubstrate. Reactions were quenched and released are distinct proteins with little homology suggests that any a distinctsubunit [32P0,]i extractedas described under “Methods.” Data are presented functional difference isassociatedwith as thepercent of activity in the absence of dihydro-C2-ceramideand composition. are mean f S.D. of triplicate determinations. Results are representThemechanism bywhich ceramidealtersthecatalytic ative of three experiments. Activity in the presence of 20 p M CZ- activity of heterotrimeric PP2A is currently unknown, but ceramide was 19.3 milliunits/ml. several possibilities exist. First, since dissociation of the B subunit from the heterotrimer is associatedwith an increase The initial peak eluted at 200 mM NaCl and corresponded to in activity (18), ceramidemayinduce subunit dissociation. the elution positionof AB’C (arrow).This activitywas stim- However, experiments to prove this point were inconclusive ulatedapproximately 1.8-fold by 20 PM ceramide.Addia partial (datanotshown).Second, ceramidemayinduce tionally, asecondpeak of ceramide activated phosphatase dissociation of the B subunit thereby releasingB subunit eluted immediately after AB’C but prior toAC, which eluted inhibition of catalytic activity. Reconstitution experiments at 230 mM NaC1. This peak of activity eluted at 210 mM indicate that the B subunit can associate directly with only NaCl,showedvery low basal activity, and was stimulated the AC complex (17, 18). Since a short segment of the B approximately 5.5-fold by 20 PM ceramide. Similarresults for the interaction were obtained in comparing the chromatographic behavior of subunit (-7 kDa) appears to be important AB$ and CAPP (data not shown). These data suggest that of this subunit with the AC complex (18), it is possible that partially purified preparations of CAPP from T9 cells contain ceramide induces some conformational change in this interaction thereby affecting catalytic activity of the heterotrimer. at least twoceramide activated protein phosphatases; one Third,althoughourdataindicatethat ceramide cannot which is closely related to AB‘C and the other showinga directly activate AC or C subunits, ceramidemay interact distinct chromatographicbehavior. with these subunits in the intact heterotrimer thereby affectDISCUSSION ing the associationof the B subunit with AC. This report extends our previous observations on the exist- Theactivation of PP2A by ceramide was foundto be influenced by acyl chain composition as well as the ratio of ence of a cytosolic ceramide-activated phosphatase activity
15529
Ceramide Activates HeterotrimericPP2A
FIG. 8. Effect of stereoisomers of C2-ceramide on activation of AB'C, AB$, and CAPP from T9 glioma cells. Phosphatase preparationswere diluted in Buffer A to the indicated initial activity and incubated in the presence of various concentrations of stereoisomers of Cz-ceramide for 5 min a t 3 7 "C using 0.5 ~ L M[.32P]phosphohistone(A, B ) or [32P]caseln (C) as substrate. Reactions were quenched, and the released [32P04]iextracted as described under "Methods." Results are expressed as the percent of activity in the absence of lipid and are means -t S.D. of triplicate determinations. Effect of D-erythro-C2-ceramide (O),L-erythro-C2-ceramide(01,Dthreo-C2-ceramide (A),and L-thre0-Gceramide (A) on A , AB'C (4.0 milliunits/ ml); B , AB,C (7.1 milliunits/ml); and C, CAPP from T9 glioma cells (3.5 milliunits/ml) .
f
1
, I 5 10 15 20 25 0
ceramide to protein. Activation of phosphatase was evident with acetoyl, hexanoyl, decanoyl, and myristoyl substitutions of the sphingoid base. It is interesting that as thehydrophobicity of the amide-linked hydrocarbon increased, the ECso for maximum activation decreased. This effect was apparent with substitutions up to Clo-ceramide, after which the extent of activation decreased with more hydrophobic ceramide species. Surprisingly, C18-ceramidewas found to be ineffective at activating purified PP2A. We previously demonstrated that C18-ceramidecould stimulate phosphatase activity from crude cytosol, albeit to a lesser degree than C2-and C6-ceramide (8). The reason for this discrepancy is unknown but may be related to the higher protein concentration used in assays with crude cytosol versus purified enzyme, i.e.,0.5-1 gg versus 10-200 ng protein, the presence of lipids in crude cytosolic preparations which may aid solubilization of Cls-ceramide, the poor solubility of C18-ceramidein dilute ethanol solutions, or the formation of lipid aggregates which may not provide a properinterface for interaction of lipid with enzyme. In support of the latter possibility is the decrease in activation seen with higher concentrations of CI4-and C6-ceramide (data not shown).
CZCeramide, pM
CZCeramide, pM
of lipid Ceramide activation of PPSA increased as the ratio to protein increased, suggesting that the interface between in ceramide protein and lipid is a critical parameter optimizing activation of PPZA. The use of ethanolic solutions of lipid in the presentstudy may be the reason for the expression of this effect. Providing a more stable lipid to protein interface, such as utilizing either lipid vesicles or mixed micellar methodologies, may eliminate this effect and allow for a more systematic evaluation of the critical factors in ceramide activation of PP2A. Current work is directed at optimizing ceramide activation of PPZA under conditions which more closely approximate the potential in vivo interaction of this cytosolic enzyme with membrane lipid. Ceramide appears to be a specific activator of AB'C, at least with respect to othersphingolipids, phosphatidylcholine, phosphatidylethanolamine, the neutrallipid dlolein, and oleic acid. This specificity was not protein dependent in that activation was not observed at either protein concentration tested. These results support that ceramide may serve as a specific intracellular lipid activator of this subtype of PP2A. An intriguing aspect of the specificity studies is the effect of C2-ceramide versus dihydro-C2-ceramide on PP2A activity.
Heterotrimeric Activates Ceramide
15530
PP2A
more latent phosphatase activity. Several investigators have noted the presence of latent heterotrimeric phosphatase activities from various preparations. These activitieshave usually been expressedafter freezing and thawing in the presence of thiol reagents (15,22). Our data suggest that ceramide may be an endogenous activator of these latentactivities. However, differences do exist in the chromatographic behavior of the latent phosphatase activity isolated from T 9 cells uersus those isolated from muscle (15, 22). Ceramide is emerging as a novel lipid second messenger capable of influencing the growth and differentiationof cells 0 21 0 (5, 6). The results of several investigators support the exist10 20 30 40 50 ence of a signaling pathway involving agonist-induced sphinFraction # gomyelin hydrolysis and ceramide generation (1-4, 7). We FIG. 9. Chromatographic behavior of CAPP from T9 glioma have demonstrated that ceramide can activate a cellularphoscells uersus AB’C from bovine heart. AC, AB’C, or CAPP were phatase activity (8, 9) and that heterotrimeric PP2Amay be diluted into 100 el of Buffer A and chromatographed separately on a Mono-Q HR 5/5 fast protein liquid chromatography column. The a proximal molecular target mediating further downstream flow rate was 1 ml/min, and protein was eluted with a linear gradient effects of ceramide, i.e., c-myc down-regulation, apoptosis, of 150-250 m M NaCl in Buffer A over 40 min. One-ml fractions were and cell differentiation (2, 4-6). Further, ceramide-induced collected and 50 p1 assayed for MBP phosphatase activity under down-regulation of c-myc can be dramatically attenuated by standard conditions in the absence (0)or presence (0)of 20 F M Cg- okadaic acid.’ Taken together, these results provide strong ceramide for 15 min at 37 “C. Data show the elution of purified evidence that ceramide-inducedc-myc down-regulationis mephosphatase standards (arrows) in relation to elution of phosphatase diated, at least in part, by an okadaicacid-sensitive phosphaactivity from the partially purified CAPP preparation. tase. These data provide increasing evidence that ceramide interacts with endogenous phosphatases and thereby affects Previous studies indicate that dihydro-C?-ceramide does not phosphoprotein metabolism and cellularregulation. Future activate cytosolic phosphatase activity isolated from yeast studiesare being directed at determiningthemechanistic (21)and fails to induce apoptosisin U937 cells (4).The basis of ceramide activation and elucidating thespecific role current studies show that dihydro-C2-ceramide notonly fails of protein phosphatase activation in the expression of certo activate AB’C but acts as an inhibitor of phosphatase amide-mediated biology. activity. These data ascribe a high degree of specificity to the double bond of C2-ceramide and suggest that the desaturation Acknowledgments-We thank Drs. S. Shenolikar and J. Fishbein of the sphingoid backbone may modulate ceramide mediated for critical review of the manuscript and Marsha Haigood for secretarial assistance. bioactivity.However, the role of the double bondinthe mechanism of activation of PP2A remains unknown. REFERENCES The synthesis of thevariousstereoisomers of ceramide 1. Okazaki, T., Bell, R. M., and Hannun,Y. (1989) J . Biol. Chem. 264,1907619080 provided additional information on the specificity of ceramide 2. Kim, M.-Y., Linardic, C., Obeid, L., and Hannun, Y. (1990) J . Biol. Chem. activation of heterotrimericPP2A.Takentogether,these 266,484-489 3. Dressler, K.A., Mathias, S., and Kolesnick, R. N. (1992) Science 246, results suggest that D-erythro-C2-ceramide can activate all 1715-1717 heterotrimericphosphatasesandsubstratestested. Addi4. Obeid, L. M., Linardic, C. M., Karolak, L. A., and Hannun, Y. A. (1993) Science 259, 1769-1771 tionally, L-erythro-Cz-ceramide appeared as effective as the 5. Okazaki, T., Blelawska, A,, Bell, R. M., and Hannun, Y. A. (1990) J. Biol. Chem. 265, 15823-15831 D-enantiomer in stimulating AB’C and AB,C but less effec6. Bielawska, A., Linardic, C. M., and Hannun, Y. A. (1992) FEBS LRtt. 307, tive at stimulating partially purified CAPP from T 9 glioma 211-214 7. Schiitze, S., Potthoff, K., Machleidt, T., Berkovic, D., Wiegmann, K., and cells. SimilartoCAPP from T 9 cells, ceramide activated Kronke, M. (1992) Cell 71, 765-776 cytosolic phosphatase from yeast is more potently activated 8. Dobrowsky, R. T., and Hannun, k. A. (1992) J. Biol. Chem. 267, 50485051 by D-erythro-C2-ceramide than its L-enantiomer (21). Addi9. Dobrowsky, R. T., and Hannun, Y. A. (1993) Adu. Lipid Res. 25,91-104 tionally, D- andL-erythro-Cz-ceramide showdistinct bioactiv- 10. Cohen, P. (1989) Annu. Reu. Biochem. 58,453-508 S., and Nairn, A. C. (1991) Ado. Second Messenger Phosphoity profiles in yeast,i.e., L-erythro-C2-ceramide is less effective 11. Shenolikar, protern Res. 23, 1-121 in inhibitinggrowth compared to the natural enantiomer (21). 12. Mumby, M. C., Russell, K. L., Garrard, L. J., and Green, D.D. (1987) J. Biol. Chem. 262,6257-6265 As such, some formsof cellular ceramide-activated phospha- 13. Mayer, R., Hendrix, P., Cron, P., Matthies, R., Stone, S. R., Goris, J., Merlevede, W., Hofsteenge, J., and Hemmmgs, B. A. (1991) Brochemlstry tase activity may be more specifically activated by D-erythro30,3589-3596 ceramide. 14. Waelkens, E., Goris, J., and Merlevede, W. (1987) J. B i d . Chem. 262, 1049-1059 CAPP isolated from T 9 cells appears to becomposed of at 15. Tung, H. Y., Alemany, S., and Cohen, P. (1986) Eur. J. Biochem. 148, least two ceramide-stimulated phosphatases. One activity ap253-263 pears tobe similar toAB’C based on chromatographic behav- 16. Kamibayashi,,C., Lickteig, R. L., Estes, R., Walter, G., and Mumby, M. C. (1992) J. B~ol.Chem. 267,21864-21872 ior. The second activity was chromatographicallydistinct 17. Imaoka, T., Imam, M., Usui, H., Kinohara, N., and Takeda, M. (1983) J. Bid. Chem. 258, 1526-1535 from AB’C and AB,C and appeared more sensitive to cer18. Kamibayashi, C., Estes, R., Slaughter, C., and Mumby, M. C. (1991) J. amide, in that its activity was negligible in the absence of Biol. Chem. 266, 13251-13260 19. Bradford, M. M. (1976) Anal. Biochem. 272,248-254 ceramide. This result agrees with our initial observation on 20. Fujino, Y., and Zahin, I. (1962) J. Biol. Chem. 2 3 7 , 2069-2074 Fishbein, J. D,, Dobrowsky, R. T., Bielawska, A,, Garrett, s.,and Hannun, 21. the isolation of CAPP from T9 cells (8).Thus, although our Y.A. (1993) J. Eiol. Chem. 268,9255-9261 data indicate that different forms of heterotrimeric PP2A can 22. Yang, S.-D., Yu, J.-S.,and Fong, Y:L. (1986) J. Bid. Chem. 261, 55905596 be stimulated by ceramide, CAPP may be associated with a 5
:I
AB‘C I
I
1 300