ROLE OF Pzz PURINERGIC RECEPTORS IN ATP ...

0022-1767/92/14910-3372$02.00/0 THEJ O U R N A L OF IMMUNOI.OCY Copyrght C 1992 by The American Association of lrnrnunologlsts

Vol. 149.3372-3378. No. 10. November 15. 1992 Printed in U.S.A.

ROLEOF Pzz PURINERGIC RECEPTORS IN ATP-MEDIATED KILLINGOF TUMOR NECROSISFACTOR (TNF)-SENSITIVE AND TNF-RESISTANT L929 FIBROBLASTS PAOLA PIZZ0,2* MARTA MURGIA,' ANNALISA ZAMBON,+ PAOLA ZANOVELLO,' VINCENZO BRONTE,' DANIELA PIETROBON,* AND FRANCESCO Di VIRGILIO* * From the 'National ResearchCouncil Centerfor the Study of the Physiologyof Mitochondria and Instituteof General Pathology and the 'Chair oflmrnunology, University of Padova, I t a l y . and the 'Instituteof General Pathology, Universityof Ferrara, I t a l y

Two closely related cell lines were characterized known as "programmed cell death" (3.4).The implication in their responses to extracellular ATP (ATP,): the of ATP, as a causative agent of apoptosis is particularly fibroblast cell line L929 a and TNF-resistant variant intriguing because on one hand apoptosis is known to L929/R. Both lines showed ATP,-activated inoccur during a number of embryogenetic, developmental, creases in intracellular Ca2+, inward current, and hormone-dependent, and immunomediated processes (5, sustained depolarizationof the plasma membrane, 6);on the other,ATP is thought tobe released in thebody cell responses compatible with activation of puri- on several occasions,such as platelet aggregation, shock, nergic receptors of the P2y. P ~ x or , P2z subtype: and probably also cell-mediated cytotoxic reactions (7however, only the L929/R variant was susceptible 9).It is thereforelikely that body cells are exposed to high to ATPddependentearlypermeabilizationofthe of physiologicand plasma membrane to hydrophilic solutes of M, be- local ATP, concentrations in a number pathologic conditions. low 900, a response uniquely caused by the activation of Pzzreceptors. Both cell types were suscepti- The cytotoxic activity ofATP, is not the result of a ble to the cytotoxic effect of ATP,, but killing of the nonspecific perturbation of the phospholipid bilayer or L929/R variant required much shorter incubations to chelation of extracellular divalent cations, but seems in the presenceof this nucleotide. Morphologic ex- to be rather because of a specific interaction with memaminationofATP,-challengedL929andL929/R brane receptors. In fact, a number of laboratories have cells showed that cell death occurredby two alter- shown that a few cell types are insensitive to exceedingly native mechanisms: colloido-osmotic lysis or apop- high (several millimolar) ATP, concentrations (2, 4, 10): tosis. Occurrenceof apoptosiswasconfirmedby furthermore, ATP,-resistant mutants have been selected agarose gelanalysis of cellular DNA. Although ATP, from susceptible cell lines (11, 12). caused afast mobilizationof intracellular Ca2+, nei- It is well established that ATP, can interact with at ther colloido-osmoticlysis nor apoptosis were Ca2+ least four different cell surface purinergic receptors (Pz dependent. Our results show that the L929/R var- receptors): Pzt, present on platelets: Pzx.Pzr, and Pz. reiant, but not the L929 parental fibroblast cell line, PZz ceptors, which are present on many other cells (7, 13). expresses functional purinergic receptors of the subtype. The presence of Pzz receptors confers to Although the pharmacology of these receptors is still L929/R cells enhanced susceptibility to ATP,-me- confusing because of the lack of specific agonists or antagonists (14). a convenient classificationcan be made diated cytotoxicity. The cytotoxic activity of extracellular ATP (ATPJ3 toward several cell types recently has rekindled interest and raised intriguing questions on the role of this nucleotide in the immune system (1, 2). It has been proposed that, depending on the cell target and the experimental conditions, ATP, can cause cell death by two different mechanisms: colloido-osmotic lysis or apoptosis.also Received for publicationMay 14. 1992. Accepted for publication September2, 1992. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked aduertlsernent in accordance with 18 U.S.C. Section 1734 solely to indicate thisfact. ' This work was supported by the Italian Ministry for Scientific Research [Grants MURST 40% and 60%).the ItalianNational Research Council [SpecialProjects BTBS, RantNo. 89.00236.70. and ACRO; Target Project Biology and Pathology of Calcium, Grant No.90.01264.CT14) and Associazione Italiana per la Ricerca sul Cancro (AIRC). Address correspondence and requests for reprints to Dr. Paola Pizzo, Institute of General Pathology.Via Trieste 75, 1-35121, Padova. Italy. Abbreviations used in this paper:ATP,. extracellular ATP:BzATP, 3'-O-[benzoyl)-benzoyl-ATP; [Ca2+],. cytoplasmic free Ca" concentration.

on the basis of the intracellular effector pathway activated by each receptor (13).Pzx receptors behave as ligand-operated cation channels; Pzy receptorsare linked to inositol 1.4.5-triphosphate generation and Ca2+mobilization from intracellularstores: Pzz receptors form transmembrane aqueous pores, permeable to solutes of M, up to 900,which shows intriguing similarities with the gap junctionalprotein connexin-43 (15). The precise assignment of a given ATP,-triggered response toa receptor subtypeis as yet uncertain: in particular,it is unclear which receptorsand intracellular second messengers mediate ATP,-dependent cell death. During a n investigation on the cytotoxic effects of ATP,, we observed a striking differencein the susceptibility to killing between the mouse fibroblast cell line L929 and the TNF-resistant variant L929/R. In fact, ATP,, at a concentration of 1 to 2 mM, caused in these latter cell populations a n early swelling and, aftera 2- to 4-h incubation, a release of cytoplasmic markers (lactic dehydrogenase and 51Cr-labeledmacromolecules) indicative of colloido-osmotic lysis. On the contrary, in the

3372

ATP-DEPENDENT CYTOTOXICITY IN L929 FIBROBLASTS

former cells, ATP,, at the same concentrations, did not trigger early increases in cell volume, and induced release of intracellular macromolecules only after several hours of treatment. Morphologic and biochemical analysis revealed that ATP, also caused apoptosis of L929 and L929/ R cells. We took advantage of the differential sensitivity to ATP,, of these two closely related cell lines to identify the purinergicreceptors andthe earlyintracellular changes leading tocolloido-osmotic lysis or apoptosis. Both cell lines showed ATP, triggered fast mobilization of Ca2+ from intracellular stores, a response typical of activation of P,, receptors. However, Plyreceptors do not seem to be involved in ATP,-dependent toxicity because maximal [Ca2+], releases were triggered by nontoxic ATP, concentrations a s well as by other nucleotides that did cell lines, ATP, not causecell death. Furthermore, in both also triggered inward ionic currents and plasma membrane depolarization, responses that canbe triggered by P2xor PZzreceptors. However, only in the L929/R cells, in addition to Ca2+mobilization and membranepotential changes, ATP, also caused a n early permeabilization of the plasma membrane to low molecular weight aqueous solutes, a response associated to activation of the P2, purinergic receptor ( 1 6). Our data suggest that L929 and L929/R fibroblasts are equipped with both P2, and PZx receptors. The L929/R variant, in addition, also expresses receptors of the PZz type. Both celllines are susceptible tothe cytotoxic action ofATP,, therefore, as PPy receptors appeared not to be involved in ATP,-dependent cytotoxicity, we conclude that both PZxand PPzreceptors can mediate cell death. Activation of the P2z receptor promotes rapid cell death via opening of large plasma membranepores; activation of the PZx receptor causes cell death via a less defined mechanism that involves the opening of a plasma membrane ion channel. The PlX receptor is likely responsible for ATP,-triggered apoptosis of L929 cells lacking the PZz receptor. Combined expression of PZxand PZzreceptors on the plasma membrane of L929/R fibroblasts on the one hand potentiates ATP,-promoted apoptosis and, on the other, makes thesecells also susceptible to colloidoosmotic lysis. MATERIALS AND METHODS

Cells and materlals. L929 and L929/R fibroblasts were a kind gift of Dr. A. Mantovani (Mario Negri Institute for Pharmacologic

3373

the expression: 100 x

(experimental release - spontaneous release) (maximum release - spontaneous release)

For DNA fragmentation analysis, 2 X lo6 cells were incubated a t 37°C in 35-mm wide Petri dishes in RPMI medium containing 10% FCS and treatedwith ATP for the indicated times. At the end of the incubation, cells were transferred to 15-ml Falcon tubes and lysed by adding in sequence 200 pl of lysis buffer [containing 2 M NaCI, 50 mM Tris-C1. 10 mM EDTA), 1% SDS. and 250 fig of Proteinase K (Boheringer Mannheim. Milano, Italy). DNA extraction was carried out by the salting-out method as follows: 400 plof 5 M NaCl were added to the lysate, which was shakenvigorously and centrifuged a t 800 g for 15 min. The supernatant wascollected, transferred to 15ml Falcon tubes, and5 ml of absolute ethanolwere added. DNA was then washed with 70% ethanol.resuspended in TE (10 mM Tris-C1. 0.2 mM Na-EDTA, pH 7.5) and treated with RNase at 37°C for 30 min. DNA was finally run on 1 % agarose gels containing 0.5 pg/ml of ethidium bromide and visualized under UV light. Measurement of (Ca2']l. Changes in [Ca"], were measured with the fluorescent indicator fura-$/AM as described previously (18). Briefly, 5 x lo5cells were loaded with 4 pM fura-2/AM and incubated in a thermostat-controlled (37°C)and magnetically stirred fluorimeter cuvette [Perkin Elmer LS5, Perkin-Elmer Corp., Norwalk. CT). Excitation and emission wavelengths were 340 and 500 nm; excitation slit width was 2.5 nm, and emission slit width was 10 nm. [Ca"], levels were calculated according to thegeneral formula: [C~'+]I = Kd(F - F r n d / [ F r n a x - F ) , where rC, is the dissociation constant forCa'+ binding, 224 nM under our experimentalconditions; F is the fluorescence of the intracellular indicator: F,,, is the fluorescence after lysis of the cells with 0.05% Triton X-100 in the presence of 6.6 mMEGTA and 40 mM Tris (final pH 8.5). and F,,, is the fluorescence of the lysed cells after addition of 1 mM Ca2+.The inhibitorof organic anion transport sulphinpyrazone was routinely added during [Ca2+],measurements a t a concentration of 250 pM to prevent fura-2 excretion and sequestration into subcellularorganelles (19). Fluorimetricmeasurement of plasmamembranepotential. Changes in plasma membrane potential were measured with the fluorescentdye bis( 1. 3-diethylthiobarbiturate) trimethineoxonal (bis-oxonol) at the wavelength pair 540 to 580 nm, as previously described (20). Experimentswere carried out in a Perkin-ElmerLS5 spectrofluorimeter equipped with a thermostat-controlled (37°C)cuvette holder and magnetic stirrer. Fluorescence microscopy. Microscopic observations were performed with a n Olympus (IMT-2or BH-2) (Olympus Optical Co., LTD. Tokio, Japan) microscope equipped with a 40X objective. Patch clamp recordings. Whole-cell currents 121) were recorded with a Dagan 3900 patch-clamp amplifier (Dagan Corporation. Minneapolis, MN) and stored and analyzed on a PDP 11/73 computer. Signals were sampled a t 5 KHz and filtered at 1 KHz (8-pole Bessel filter, -3dB). Standard Ca'+-supplemented saline was used as bath solution. The pipette solution [internal solution) contained[in 1 mM): 130 KCI, 3 MgCI,, 10 Hepes, 1 EGTA, pH 7.4. ATP was applied by bath perfusion. Complete bath perfusion required 15 to 25 s. All recordings were obtaineda t room temperature. RESULTS

Research, Milano. Italy) and weregrown in RPMI medium (Flow Extracellular ATP is cytotoxic. Incubation in the presFCS. Labs, Inc., Irvine. UK) supplemented with 10% heat-inactivated ence ofATP,, was cytotoxic to both L929 and L929/R L-GLUTAMINE ( z x M finalconcentration), HEPES (10' M final cells, as judged from release of intracellular "Cr-labeled concentration). 100IU of penicillin/ml and 100 pg of streptomycin/ ml. Experiments were performed a t 37°C. either in theabove men- macromolecules or the cytoplasmic enzyme lactic dehytioned RPMI medium or in saline medium containing 125mM NaCl, drogenase. The two cell types, however, widely differed 5 mMKCI, 1 mMMgSO,, 1 mMKH'PO,. 5.5 mM glucose, 20 mM in their sensitivity to this nucleotide. In fact, lactic deHEPES/NaOH. pH 7.4, and, when indicated,1 mM CaCI2;this latter hydrogenase release from L929/R cells started within l medium is hereafter referred to as standard saline. Cytotoxlclty a n d DNAfragmentatlon assay. Cell lysis was as- to 2 h after theaddition of ATP, and reached 35%of total sessed by measuring releaseof either thecytoplasmic enzyme lactic content after 4 h (Fig. 1). On the contrary, after a similar dehydrogenase or 51Crfrom prelabeled cells. Lactic dehydrogenase incubation inthe presence of ATP,, practically no enzyme activity of cellular supernatants was measured as previously de- release from L929 cells was observed. Similar results scribed ( 1 7). For 51Crrelease assay.cells were labeled forl h a t 37°C in 0.1ml of DMEM plus 5%FCS containing 100pCi of 51Cr(NEN Du were obtained by monitoring cell death by specific 51Cr Pont de Nemours Italy. Florence, Italy), as previously described (3). release (inset in Fig. 1 ) . Although ATP,-dependent cytoLabeled cells (104/well) were incubated for various lengths of time toxicity in L929 cells was delayed compared with the with different concentrations ofATP, in duplicate wells of round L929/R variant, the percentage of lactic dehydrogenase bottom microtiter plates (Sterilin, Teddington, Mlddlesex, UK) in a final volume of 0.2 ml. At the endof the incubation, the plates were and specific 5'Cr release peaked to very close values in both cell types after a prolonged ( 1 2 h) stimulation with centrifuged, and 0.1 ml of supernatant was withdrawn for counting, The percentage of specific 51Crrelease was calculated according to this nucleotide (Fig. 2,and inset). ATP, threshold (1.75

3374


L929 and L929/R cells (Fig. 3, arrows in panels A and

B, respectively). Control L929 and L929/R cells incubated in the absence ofATP,, never showed signs of apoptosis. but some necrotic cells were occasionally observed (Fig. 3, panels C and D ) . The experiments reported in the previous figures suggest that ATP, can cause cell death by two different mechanisms: osmotic lysis and apoptosis. To better characterize these two pathways leading to cell death, we further investigated the biochemical and morphologic changes triggered by ATP,. /i 5 15 30 60 120 i t - 2:o The biochemical hallmark of apoptosis is cleavage of Time ( m i n l cellular DNA into fragments that are discrete multiples Ftgure 1 . Time course of l x t i c dehydrogenase and 5'Cr release from of 200 bp. Fig. 4, upper panel, shows a n agarose gel cells. For lactic dehydrogenase ( L D H ) release, cell monoATP-pulsed layers. 0.4 x 106/ml. were incubated at 37°C In 24-well Costar dishes in electrophoresis of ethidium bromide-stained DNA exstandard Ca2+-supplementedsaline in the presence or absence of 5 mM tracted from ATP,-treated and control L929 and L929/R ATP,. At the end of the various incubation times. the supernatants were cells. DNA was purified from these cells after a n incuremoved, centrifuged and enzyme content measured. Lactic dehydrogenase release is expressed as a percentage of total content determined by bation of 14 h in the presence of 1 mM (lanes 1 and 5 ) . lysing an equal amount of cells with 0.1 % Triton-Xl00. 0. L929 cells: A, L929/R cells: 0 . L929 cells incubated in the presence ofATP,: and A, 2.5 mM (lanes 2 and 6 ) and 5 mM ATP, (lanes 3 and 7): L929/R cells incubated in the presence of ATP,. For specific 51Crrelease lanes 4 and 8 refer to control L929 and L929/R cells, ATP, was 5 mM: other experimental conditions were as described in respectively, incubated for 14 h in the absence of ATP,. Materials and Methods. Although DNA cleavage in discrete fragments was detected at the higher ATP,, concentration (5 mM) in both cell lines, intensity of low molecular weight DNA bands by densitometric analysis was 4- to 5-fold higher in the L929/R variant than in L929 parental cells (Fig4.. lower panel). In control cells, DNA was remarkably intact and ran asa single band of above 20.000 bp. Changes in [Ca2+liand plasma membrane perrneability caused by extracellular ATP. In several cell types, ATP, is known to trigger early changes in [Ca2+jj via the activation of Ply purinergic receptors. Because [Ca"'], increases havebeen widely implicated as intracellular mediators in several cytotoxic reactions, we examined the I correlation between ATP,-triggered [Ca"'], changes and 3.3 3.6 3.9 IO~[ATTP]pm cell death. A s shown in Figure 5, and as already described Figure 2. ATP,, dose-dependence of lactic dehydrogenase and 5'Cr re- in several other cell types (22, 23), ATP, caused in both L929 and L929/R cells a transient [Ca2'], increase, lease. For lactic dehydrogenase ( L D H )release, cell monolayers in 24-well Costar dishes (0.4x 106/ml were incubated for 12 h at 37°C in Caz+- mainly the result of Ca2+release from intracellular stores. supplemented standard saline in the presence of increasing ATP, concenas it was not modified by chelation of extracellular Ca2+. trations: 0. L929 cells: and A, L929/R cells. For specific 51Crrelease, cells were incubated for 12 h in the presence of increasing ATP, concentraA s shown by the subsequent addition of ionomycin, the tions: other experimental conditions were as described in Materials and ATP,,-mobilizable pool was of the same order of magniMethods. tude of the total ionophore-mobilizable intracellular Ca2+ mM) and ED50 (2.2 mM) for lactic dehydrogenase release pool. Maximal Ca2+release was elicited by 500 pM ATP,,, were about the same in both cell types. ATP, dose-de- a nucleotide concentration largely subthreshold for cydissociation pendence for 5'Cr release was, on the contrary, slightly totoxicity, even after a 24-h incubation. The shifted tolower concentrations in L929/R compared with in thedose-response curvesfor Ca2+mobilization and cell L929 cells (ATP,,ED5,,of 2.5 and 3.3 mM, respectively). death and theCa2+independence of ATP,-mediated lysis and apoptosis suggest that intracellularfactorsother Chelation of extracellular divalent cations with excess EDTA decreased both the ATP, threshold and theED50, than Ca2+ must be involved in the cytotoxic effect of ATP,. indicating that, in analogy to results obtained in other In many cell types, ATP, is known to cause depolaricell systems, the active form is likely to be ATP4- the zation of the plasma membrane through activation of fully dissociated form of ATP (not shown). It recently has been shown that ATP, is a trigger of either the Pzx receptor, which coincides with a ligandapoptosis in a number of cell types, therefore we inves- activated ion channel with a cut-off of about M, 200 (24, 25). or the PPzreceptor, identified with a n aqueous pore tigated whether this mechanism was also operative in L929 and L929/R cells. Apoptosis can be identified by that allows hydrophilic molecules up to M, 900 to permorphologic examination and by agarose gel analysis of meate across the plasma membrane and gain access to cellular DNA. Fig. 3 shows fluorescence photographs of the cytoplasm (26). L929 and L929/R cells responded to ATP, with a sustained depolarization of the plasma memATP,,-treated L929 and L929/R cells stained with the nuclear stain ethidium bromide. This dye is slowly per- brane, preceded at lower ATP, stimulatory doses, by a meant across the intact plasma membrane: if, however transient hyperpolarization, probably the resultof open' channels activated by the fastATP,-dependent the membrane is damaged or leaky, it enters the cells, ing of K binds to DNA, and allows clear identification of the nu- [Ca'+],increase (Fig. 6).Plasma membranedepolarization clear clumpstypical of apoptosis caused by ATP,, in both was greatly decreased by replacement of extracellular ~~~~

3375


Ftgure 3. Mfect of ATP, on nuclear morphol-

ogv of L929 and L929/R cells stained wlth ethfd-

ium bromide. L929 ( p a n e l s A and C). and L929/ R cells [ p a n e l s Band Dl in %well costar dishes, 2 X lO"/ml. were incubated for 18 h at 37°Cin RPMl medium plus 10%FCS in the presence of ATP. (2.5 and 5 mM in panels A and B. respec-

.+

m

tively]or in its absence (panelsC and D). At the ethldium I bromide end of this incu;bation. 20 & was added. and cells were analyzed by fluorescence microscopy. Bar = 10 sm.

a

-LL

-140 -110

iTP

tlono

'ATP

tlono !An

Figure5. ATP.-triggered [Ca*+], changes. Cells. 0.5 x 106/ml.were loaded wlth 4 r M fura-a/AM. suspended in &%-free standard saline and I ATP. and 0.5 p M ionomycin (lono).Trace Q. challenged with 500 LM L929 cells: trace b. L929/R cells. om

cm

Figure 4. DNA fragmentation caused by ATP.. Upper panel, lanes 1 , 2. and 3,L929 cells incubated in the presence of 1.2.5. and 5 mMATP.; lanes 5.6,and 7. L929/R cells incubated in the presence of 1. 2.5. and 5 mM ATP,,: lanes 4 and 8. control L929 and L929/R. respectively. Cells were incubated for 14 h. L o w r p a n e l . densitometric analysis of lanes 3. 4. 7. and 8. The experiment shown is representative of three othersthat gave similar results. Other experimental details are reported in Materfals and Methods.

Na+ with choline and completely inhibited in a fully sucrose-substituted medium (not shown). These experiments clearly suggested that depolarization dependend on a n inward current carried by sodium, as an outward

chloride current should be unaffected by replacement of extracellular Na", and indeed should be greatly potentiated in sucrose medium. To further characterize ATP,triggered membrane conductance changes, we performed a patch-clamp study that confirmed that ATP, caused in both cell lines an inward cationic current consistent with the fast opening of ion channels (Fig. 7).Currents activated quickly, and in some cases the initial rapid burst was followedby a slower increase. Furthermore, the current inactivated slowlyor not at all, but decayed quickly when ATP, was removed. Current densities (PA/ pF) were very close in the two cell lines (21.42 & 7.16

3376

ATP-DEPENDENT CYTOTOXICITY IN L 9 2 9 FIBROBLASTS

tration of 500 pM, a s they arereported to be several-fold more potent activators of the Pzxreceptorthan ATP (29). BzATP was used at a maximal concentration of 1.7 mM, being 5-fold more potent than ATP as a n activator of the PtZ receptor. A s shown in Figure 10, these analogues on one handlacked cytotoxic activity, and on the otherwere unable to preventATP,-induced lactic dehydrogenase release. Pmin

DISCUSSION

t_

FLgure 6. ATP,-dependent plasma membrane potentialchanges. Cells, 0.5 x 106/ml. were incubated in CaZ+-free standard saline containing 100 nM bis-oxonol. ATP, was 1 mM and gramicidin D (Gram D) 500 nM. Traces a and b refer to L929 and L929/R cells, respectively.

and 16.5 k 7.19 for L929 and L929/R cells, respectively), indicating that they had a similar plasma membrane distribution of ATP-activated channels. In ATP,-treated fibroblasts, the seal quickly deteriorated, thus making impossible a full characterizationof membrane permeability changes by the patch-clamp technique. Although patch clamp experiments showed the presence of ATP,activated ion channels in both cell lines, they did not allow to discriminate whether theywere of the Pzxor Pzz type. We thus investigated whether L929 and L929/R fibroblasts expressed the ATP,-gated pore (PZzreceptor) by measuring cellular uptakeof water-soluble extracellular markers of M , below 900 [e.g.lucifer yellow, M,463) after a transient (5-to 15-min) exposure toATP,. In the L929 line exposure to 5 mM, ATP, did not cause anydetectable uptake of the marker, thus ruling out the presence of a functional ATP,-activated pore and suggesting that plasma membrane depolarization in thiscell type is sustained by a Pzxreceptor (Fig. 8,panel b).On the contrary, cells of the L929/R variant were nearly 100% permeabilized by ATP,, as routinely observed in cells equipped with the ATP,-gated pore (Fig. 8,panel d ) .Besides lucifer yellow, ethidium bromide, a fluorescent markernormally impermeantacrosstheplasmamembrane,was also taken up by ATP,-treated L929/R but not by L929 cells after a 5-min incubation (not shown). The time course of uptake of lucifer yellow by L929/R cells was rapid,being near maximal already at 15 min, similarly to previous data obtainedwith mouse macrophages(27). while a negligible uptake by L929 cells could be observed only after 120 min (Fig. 9). There are circumstantial indications that theATP analogues a,@-methylene-ATPand @,ymethylene-ATP may act as selective agonists for the PzX. whereas BzATP might be specific for the Pzzreceptor (13). although this latter compound has also been claimed to selectively label the P2y receptor (28).a,@-methylene-ATP and @,y-methylene-ATPwere used at a maximal concen-

There is increasing interest in the biologic actions of extracellular nucleotides that are recognized as important mediators of cell-to-cell communication (1-4, 30). Among them, ATP is particularly interesting because it is released from several cells during physiologic and pathologic responses and acts as a stimulantfor a wide variety of different cells. A number of responses to ATP,. spanning from stimulation of cell proliferation to inductionof cell death, havebeen reported (31);this makesit difficult to identifya clear cut physiologic rolefor this nucleotide. One of the most intriguing actions of ATP, is its cytotoxic activity against severalcell targets. Cytotoxic effects of ATP, have been reported in mast cells, macrophages, fibroblasts, lymphocytes, and a number of cultured cell lines. Different cell types alsodiffer with respect to ATP, ED50, which can be as low as a few micromolar in mast cells, and ashigh 1 to 2mM in fibroblasts. Furthermore, in some cases, abrief (10 to 15min) exposureis sufficient to irreversibly damage the cells, whereas in other cases, much longer incubations are needed (see Ref. 16, for recent review). A few cell surface receptors endowed with the capacity of triggering cell death are known, e.g. the glutamate receptor specific for N-methyl D-aspartate (NMDA receptor), which mediate neuronal cell death (excitotoxicity) (32), orthe CD3/T cell receptor, which causes apoptosis of immature T lymphocytes (33),but we have very little information aboutthe intracellular transduction systems associated with their activation. The results presented in this paper indicate that two closely related fibroblast cell lines, L929 and the TNFresistant variantL929/R, express purinergic receptors of the Pzyand Pzxtype. In addition, L929/R cells also express purinergicreceptors of the Pzz type. Bothcell types undergo apoptosis when exposed to ATP,: in addition, L929/R fibroblasts on one hand aremore susceptible to apoptosis than their TNF-sensitive counterpart and, on the other, alsoundergo colloido-osmotic lysis. These observations suggest to us that the Pzxreceptor mediates ATP,-promoted apoptosis, while the Pzz receptor potentiates apoptosis and also cause necrotic lysis. On the contrary, we exclude an involvement of the Payreceptor

20[a ;;i-n L929

l.91911

ATP

ATP UI

sm

ATP 51yI

300

Figure 7. Effect of ATP, on whole-cell current of L929 and L929/R cells. Whole-cell current at -100 mV was measured every 4 s. Re5 mM 4 s. ATP. returned Withina similarly creasedbath peated the rapid perfusions time course, current in less with the than current

to controllevelupon wash-out ofATP,.Inward cationic currents are plotted with positive values.

=

5

loo

0

L 0

L

b

1 e TIME W I N I

0

2

1 TIME

3 (MINI

4

z

15

3377

IN L 9 2 9 FIBROBLASTS

ATP-DEPENDENTCYTOTOXICITY

40

60

30

5

120 Tim. (mln)

FLgure 9. Time course of ATP,-dependent lucifer yellow uptake. Cell monolayers (0.L929: A, L929/R) in 24-well Costar dishes were incubated at 37°C in &'+-supplemented standard saline in the presence of 5 mM ATP,. Five minutes before the sampling, lucifer yellow (LY)at a concentration of 0.5 mg/ml was alsoadded to thewells. Lucifer yellow was added just before the sampling to avoid possible artifacts as a result of dye uptake by pinocytosis. Lucifer yellow uptake is expressed as a percentage of stained cells.

a

I

I

r 40-

2 -

r

?1

t 0 a

z 3020

-

10 -

0-

Figure 10. Effect of ATP analogues on lactic dehydrogenase release. For lactic dehydrogenase (LDH)release, cell monolayers In 24-well Costar dishes (0.4 x 106/ml were incubated for 12 h a t 37°C in Ca2+-supplemented standard saline with the following additions: 1 . none: 2. 5 mM ATP,: 3. 1 mM BzATP: 4. 1.7 mM BzATP: 5. 300 pM P.y-methylene-ATP: 6.500 r M &y-methylene-ATP: 7, 500 pM a.b-methylene ATP: 8. 500 pM a.B-methylene-ATP plus 5 mM ATP: and 9. 500 pM &y-methylene-ATP plus 5 rnM ATP. Open bars. L929 fibroblasts: solid bars. L929/R fibroblasts. Data are average of duplicate determinations from one single experiment representative of three others thatgave similar results.

heart myocytes in the range from 0.1 to 5.0 mM when acting on the cytoplasmic side of the membrane (34). Opening of such a large pore in unpaired cells would be expected to cause osmotic swelling and lysis, as observed with other well known membrane-perturbing agents reFigure 8. Effect of ATP,, on lucifer yellow uptake by L929 and L929/ R cells. Cell monolayers in 24-well Costar dishes were incubated a t 37°C leased by bacteria, suchas a-staphylotoxin, or generated in standard saline containing Caz+and 0.5 mg/rnl lucifer yellow in the during the immune response. such as the complementpresence or absence of ATP,. Panels a and c. control L929 and L929/R cells, respectively: panels b and d. L929 and L929/R cells incubated in derived membrane attack complex or perforin. This is the presence of 5 mM ATP, for 5 mln. respectively. Bar = 20 pm. indeed the case,as indicated by morphologicexamination and increase in cell volume of ATP,-stimulated L929/R on the basisof the following evidence: 1) ATP,-resistant cells, thus suggesting that expression of Pz. receptors cell clones have been selectedthat retain functional Pay permits colloido-osmotic lysis of cells that would otherreceptors (23); 2)Pzyreceptors can be activated,besides wise die by apoptosis. ATP,, by other purine and pyrimidine nucleotides that, The susceptibility of L929/R cells to undergo cell death however, do nottrigger cell death ( 10): and 3)intracellular by two different mechanismsdoes not depend on a gross Ca'+ mobilization, a typical Pz,-associated response, is qualitative heterogeneity of this cell population, because, induced in L929 and L929/R fibroblasts by ATP, concen- a s shown by lucifer yellow uptake experiments, more trations lower than those necessary totrigger cell death. than 90%of ATP,-pulsed L929/R cells became permeable Recently, Beyer and Steinberg ( 15) have proposed that to lucifer yellow, thus providing clear evidence for the the ATP,-induced pore is related to the gap junctional presence of a functional P,, receptor in the overwhelming protein connexin-43.To this respect, itis of interest that majority of these cells. It cannot be excluded that whether ATP can directly affect junctional conductanceof paired cell death occurs by lysis or apoptosis also depends on

I

3378

ATP-DEPENDENT CYTOTOXICXTY IN L929 FIBROBLASTS

intrinsic factors such as, for example, the cell cycle or purinergic receptor density onthe individual cells. The ATP,,concentrations needed to trigger the cytotoxic response are, in the cells used in this study and in a number of other cell types, in the millimolar range. This is generally believed to be a relevant obstacle for advocating an in vivorole for ATP,, as a cytotoxic agent. However, it is intriguing to notethat all the ecto-ATPases so far characterized have a Km for ATP, between 300 and 500 p M (35, 36). One wonders why these ATP-degrading enzymes have such a low affinity for ATP, unless they are devised to handle high concentrations of this nucleotide, and therefore protect the cell from its injurious effects.ATP is known tobe stored inmolar concentration within the dense granules of platelets and the cytoplasmic concentration in most cell types is usually 5 to 10 mM; therefore, it is not unlikely that millimolar ATP,, concentrations could be reached at sites of extensive platelet aggregation or cell death, especially if ATP release occurs in a restricted environment. Finally, the inverse correlation between TNF resistance and ATP,, sensitivity observed in the L929 and in the L929/R variant fibroblasts may not be coincidental, in the light of the suggested identification of the ATP,,-gated pore with the gap junctional protein connexin-43 (15) and the proposed correlation between expression of gap junctions andTNF resistance (37).

Acknowledgments. We wish to thank Professors T. Pozzan and D. Collavo for helpful advice and continuous encouragement. We are also indebted to Dr. A. Mantovani (Mario Negri Research Institution)for gift of TNF-resistant L929 cells. REFERENCES 1. Di Virgilio, F., P. Pizzo. P. Zanovello, V. Bronte, and D. Collavo. 1990a. Extracellular ATP as a possible mediator of cell-mediated cytotoxicity. Irnrnunol. Today, 11:274. 2. Filippini. A., R. E. Taffs, T. Agui, and M. V. Sitkovsky. 1990a. EctoATPase activity in cytolytic T-lymphocytes. Protection from the cytolytic effects of extracellular ATP. J. Biol. Chern. 265334. 3. Zanovello, P..V. Bronte, A. Rosato. P. Pizzo. and F. Di Virgilio. 1990. Responses of mouse lymphocytes toextracellular ATP. 11. Extracellular ATP causes cell type-dependent lysis and DNA fragmentation. J. Irnrnunol. 145: 1545. 4. Zheng. L. M.,A. Zychlinsky, C-C. Liu. D. M. Ojcius, and J. D-E. Young. 1991. Extracellular ATP as a trigger for apoptosis or programmed cell death. J. Cell S f o f .1 12:279. 5. Wyllie, A. H., J. F. K. Kerr, and A. R. Currie. 1980. Cell death: the significance of apoptosis. Int. Reu. Cytol. 68:251. 6. Duke, R.C.. R. Chervenak, and J. J. Cohen. 1983. Endogenous endonuclease-induced DNA fragmentation: a n early event in cellmediated cytolysis. Proc. Natl. Acad. Sci. USA 8033361. 7. Gordon, J. L. 1986. Extracellular ATP: effects, sources and fate. Biochern. J. 233~309. 8. Trams, E. G., H. Kaufmann, and G. Burnstock. 1980. A proposal for the role of ecto-enzymes and adenylates in traumatic shock. J. Theor. Biol. 87:609. 9. Filippini, A.. R. E. Taffs, andM.V. Sitkovsky. 1990b. Extracellular ATP in T-lymphocyte activation: possible role in effector function. Proc. Natl. Acad. Sci. USA 87:8267. 10. Di Virgilio. F..V. Bronte, D. Collavo. and P. Zanovello. 1989. Responses of mouse lymphocytes to extracellular adenosine 5’-triphosphate. Lymphocytes with cytotoxic activity are resistant to the permeabilizing effects of ATP. J. Irnrnunol. 143: 1955. 1 1. Kitagawa, T., and Y. Akamatsu. 1986.Control of membrane permeability by external ATP in mammalian cells: isolation of a n ATP resistant variant from Chinese hamster ovary cells. Biochfrn. BLo-

phys. Acta. 860:185. 12. Steinberg, T.H., and S. C. Silverstein. 1987. Extracellular ATPpromotes cation fluxes in the 5774 mouse macrophage cell line. J. Biol. Chern. 262:31 18. 13. Watson. S . . and A. Abbott. 1991. Receptor Nomenclature Supplernent. Trends Pharrnacol. Scf. 12:1. 14. 0’ Connor. s. E.. I. A. Dainty. and P. Leff. 1991. Further subclassification ofATP receptors based on agonist studies. Trends PharrnQCOl. S C i . 12:137. 15.Beyer, E. C.. and T.H. Steinberg. 1991. Evidence that the gap junctionproteinconnexin-43 is the ATP-induced pore of mouse macrophages. J. Biol. Chern. 266:7971. 16. Steinberg. T. H.. and F. Di Virgilio. 199 1.Cell-mediated cytotoxicity: ATP a s a n effector and therole of the target cells. Curr. Op. Irnrnunol. 3:71. 17. Picello, E., P. Pizzo. and F.Di Virgilio. 1990. Chelation of cytoplasmic Ca” increases plasma membrane permeability in murine macrophages. J. Biol. Chern. 2655635. 18. Di Virgilio. F.,D. Milani, A. Leon, J. Meldolesi, and T. Pozzan. 1987. Voltage-dependent activation and inactivation of calcium channels in PC12 cells. J. Bfol. Chern. 262:9189. 19. Di Virgilio, F., T. H. Steinberg, and S . C. Silverstein. 1990b. Inhibition of fura-2 sequestration and secretion with organic anion transport blockers. Cell Calcium 11:57. 20. De Togni. P., G. Cabrini, and F. Di Virgilio. 1984. Cyclic AMP inhibition of fMet-Leu-Phe-dependent metabolic responses in human neutrophils is not due to its effects on cytosolic Ca2+.Bfochern. J. 224~629. 21. Hamill, 0. P.. A. Marty. E.Neher, B. Sakmann, and F. J. Sigworth. 1981. Improved patch-clamp techniques for high-resolution current recording from cells and cell-free membrane patches. PJugers Arch. 391:85. 22. Dubyak, G. R.. and M. B. De Young. 1985. Intracellular Ca2+mobilization activated by extracellular ATP in Ehrlich ascites tumor cells. J . B i d Chern. 260:10653. 23. Greenberg, S.,F. Di Virgilio, T. H. Steinberg, and S . C. Silverstein. 1988. Extracellular nucleotides mediate Ca2+fluxes in 5774 macrophages by two distinct mechanisms. J. Biol. Chern.263:10337. 24. B e a n , B. P., and D.D. Friel. 1990. in ‘lon Channels” (T. Narahashi ed.). Plenum Press, p 169. 25. Pizzo. P.,P. Zanovello, V. Bronte, and F. Di Virgilio. 1991. Extracellular ATP causes lysis of mouse thymocytes andactivates a plasma membrane ion channel. Biochern. J. 274: 139. 26. Gomperts. B.D. 1983. lnvolvement of guanine nucleotide-binding protein in the gating of Caz+by receptors. Nature 30654. 27. Steinberg, T. H.,A. S. Newman. J. A. Swanson, and S. C. Silverstein. 1987. ATP4- permeabilizes the plasma membrane of mouse macrophages to fluorescent dyes.J. Biol. Chern. 262:8884. 28. Gonzales, F. A., D-J. Wang, N-N. Huang, and L. A. Heppel. 1990. Activation of early events of the mitogenic response by a P2y purinoceptor with covalently bound 3‘-0-(-4-benzoyl)-benzoyladenosine 5’-triphosphate. Proc. Natl. Acad. Sci. USA 87~9717. 29. Cusack, N. J., and S. M. 0. Hourani. 1990. Subtypes of P,-purinoceptors. Studiesusinganalogues ofATP. Ann. N. Y . Acad. Sci. 603: 1 72. 30. Burnstock. G., and C. Kennedy. 1986. A dual function for adenosine 5’4riphosphate in the regulation of vascular tone. Excitatory cotransmitter with noradrenaline from perivascular nerves and locally released inhibitory intravascular agent. Circ. Res. 58:319. 31. Dubyak, G.R., and J. S. Fedan (eds.). 1990. Biological actions of extracellular ATP. Ann. N. Y . Acad. Sci.603, p 1 . 32. McDermott, A. B., and N. Dale. 1987. Receptors, ion channels and synaptic potential underlying the integrative actions of excitatory amino acids. Trends Neurosci. 10:280. 33. Smith, C. A., G. A. Williams, R. Kingston, E. J. Jenkinson, and J. J. T.Owen. 1989. Antibodies to CD3/T-cell receptor complex induce death by apoptosis in immature T cells in thymic cultures. Nature 337:181. 34. Sugiura. H.. J. Toyama, N. Tsuboi, K. Kamiya, and I. Kodama. 1990. ATP directly affects junctional conductance between paired ventricular myocytes isolated from guinea pig heart. Circ. Res. 66: 1095. 35. De Pierre, J., and M. L. Karnovsky. 1974. Ecto-enzymes of the guinea pig polymorphonuclear leukocytes. 1. Evidence for a n ectoadenosine monophosphatase, adenosine triphosphatase. and p-nitrophenyl phosphatase. J. Biol. Chern. 249:7111. 36. Torres, M..J. Pintor. and M. T. Miras-Portugal. 1990. Presence of ectonucleotidases in cultured chromaffin cells: hydrolysis of extracellular adenine nucleotides. Arch. Bfochern. Biophys. 279:37. 37. Fletcher, W. H., W. W. Shiu. T. A. Ishida. D. L. Haviland. and F. F. Ware. 1987. Resistance to the cytolytic action of lymphotoxin and tumor necrosis factor coincides with the presence of gap junctions uniting target cells. J. Irnrnunol. 139:956.