THE JOURNAL OF BIOLOGICAL CHEMISTRY Vol. 266, No. 2, pp ... · THE JOURNAL OF BIOLOGICAL CHEMISTRY 8 1991 by The American Society for Biochemistry and Molecular Biology, Inc. Vol.

THE J O U R N A L OF BIOLOGICAL CHEMISTRY 8 1991 by The American Society for Biochemistry and Molecular Biology, Inc.

Vol. 266, No. 2, Issue of January 15, p p . 958-965,1991 Printed in U.S.A.

Receptor and G Protein-mediated Responses to Thrombin in HEL Cells*

(Received for publication, June 19, 1990)

Lawrence F. BrassSBV, David R. Manning[[**, Alison G . WilliamsSII, Marilyn J. Woolkalis(($$, and Mortimer PonczSB From the Departments of $Medicine, §Pathology, I(Pharmacology, and §§Pediatrics, University of Pennsylvania, Philadelphia, Pennsylvania 19104

Thrombin is believed to activate platelets via cell surface receptors coupled to G proteins. In order to better understand this process, we have examined the interaction of thrombin with HEL cells, a leukemic cell line that has served as a useful model for studies of platelet structure and function. In HEL cells, as in platelets, thrombin stimulated inositol trisphosphate (IP,) formation and suppressed cAMP synthesis. Both events were inhibited by pertussis toxin with 50% inhibition occurring at a toxin concentration that ADP-ribosylated 60% of the Gi, subunits present in HEL cells. IPS formation was also stimulated by a second serine protease, trypsin. The trypsin response was identical to the thrombin response in time course, magnitude, and pertussis toxin sensitivity, suggesting that a similar mechanism is involved. Agonist-induced changes in the cytosolic-free Ca2+ concentration were used to test this hypothesis. Both proteases caused a transient increase in intracellular calcium [Ca2+Ii that could be inhibited with D-phenylalanyl-L-prolyl-L-ar- ginine chloromethyl ketone thrombin. Exposure to either protease desensitized HEL cells against subse- quent increases in [Ca2+Ii and IP3 caused by the other, although responses to other agonists were retained. This loss of responsiveness persisted despite repeated washing of the cells and the addition of hirudin. Com- plete recovery occurred after 20 h and could be pre- vented with cycloheximide. These observations sug- gest that 1) HEL cell thrombin receptors, like those on platelets, are coupled to phospholipase C and adenylyl- cyclase by pertussis toxin-sensitive G proteins, 2) the G proteins involved are equally accessible to pertussis toxin in situ, 3) when access is limited to the outside of the cell the response mechanisms for thrombin and trypsin are similar, if not identical, despite the broader substrate specificity of trypsin, 4) both proteases cause persistent changes that may involve proteolysis of their receptors or associated proteins, and 5 ) desensitization of the thrombin response occurs at a step no later than the activation of phospholipase C and requires protein synthesis for recovery.

* This work was supported in part by National Institutes of Health Grants HL40387 and CA39712 and the American Heart Association. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1 Established Investigator of the American Heart Association. To whom all correspondence should be addressed Hematology-Oncology Section, Silverstein 7, University of PA, 3400 Spruce St., Philadel- phia, PA 19104. Tel.: 215-662-3910; Fax: 215-662-7617.

** Established Investigator of the American Heart Association. $$ Recipient of a Special Investigatorship Award from the South-

eastern Affiliate of the American Heart Association.

Thrombin evokes biological responses from a variety of vascular and perivascular cells, including platelets, megakar- yocytes, endothelial cells, fibroblasts, and vascular smooth muscle. Although these responses have been studied exten- sively, a number of questions remain unanswered. For exam- ple, in platelets thrombin has been shown to initiate fibrino- gen receptor expression, aggregation, and granule secretion. These responses are known to involve the products of phos- phoinositide hydrolysis and arachidonate metabolism and are enhanced by thrombin’s ability to suppress cAMP formation. However, the cell surface receptor that interacts with throm- bin has not been identified and the mechanism by which thrombin receptors interact with intracellular effectors has not been fully defined.

Although thrombin receptors have not been isolated, some of their properties can be predicted from existing information. Based upon dose-response curves and binding studies, plate- lets contain several hundred thrombin receptors per cell with at least nanomolar affinity (1, 2). These receptors are located on the cell surface and are believed to interact with phospho- lipase C and adenylylcyclase via one or more guanine nucleo- tide-binding proteins or G proteins,’ a family of heterotrimeric regulatory proteins associated with cell membranes (3). In platelets it has been shown that thrombin stimulates high affinity GTPase activity and that activation of phospholipase C by thrombin requires GTP (reviewed in Refs. 4 and 5). It has also been shown that thrombin’s ability to stimulate phosphoinositide hydrolysis (6-9) and suppress cAMP for- mation (10,ll) can be inhibited by pertussis toxin, a bacterial enzyme which ADP-ribosylates some G protein a subunits. Collectively, these observations suggest that thrombin recep- tors are relatively low in number and high in affinity, and are coupled by one or more pertussis toxin-sensitive G proteins to phospholipase C and adenylylcyclase.

In many respects, cellular responses to thrombin resemble those to other potent agonists. However, unlike other ago- nists, thrombin is a serine protease, a family of enzymes that includes trypsin as well as proteases with greater substrate selectivity (12). Although trypsin and other proteases can mimic the effects of thrombin on platelets (12-15), the role of proteolysis in these events has not been clearly defined. Platelet activation by thrombin requires active enzyme. Re- agents such as diisopropyl fluorophosphate F and D-phen- ylalanyl-L-prolyl-L-arginine chloromethyl ketone (PPACK)

The abbreviations used are: G protein, GTP-binding regulatory protein; PPACK, D-phenylalanyl-L-prolyl-L-arginine chloromethyl ketone; HEPES, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; EGTA, [ethylenebis(oxyethylenenitrilo)]tetraacetic acid; [Ca2+]i, in- tracellular Ca2+; Gpp(NH)p; guanosine 5’-(P,y-imido)triphosphate; NPY, neuropeptide Y; IPS, inositol trisphosphate; SDS-PAGE, SO-

dium dodecyl sulfate-polyacrylamide gel electrophoresis.

958

Receptor and G Protein-mediated Responses to Thrombin 959

which block the active site of thrombin have little or no effect on thrombin binding, but block aggregation and secretion (1, 16), as does leupeptin, another protease inhibitor (15,17). On the other hand, proteolysis of any of the known substrates for thrombin on the platelet surface, including glycoprotein V, does not appear to be required for platelet activation (18).

In summary, current evidence suggests that thrombin evokes biological responses from platelets and other cells via receptors that are coupled to G proteins and raises the possi- bility that proteolysis plays a role in this process. The goal of the present studies was to clarify this interaction. To facilitate certain types of studies and to avoid the technical problems created by the inability of pertussis toxin to penetrate the platelet plasma membrane, the observations were made with HEL cells. HEL cells are a human leukemic cell line that retains a number of features of the megakaryocyte/platelet lineage, including the ability to synthesize platelet factor 4 and membrane glycoproteins Ib, IIb, and IIIa, all of which have been shown to be identical to their counterparts in platelets (19-21). Our previous studies show that HEL cells undergo phosphoinositide hydrolysis in response to thrombin (22) and contain the same three members of the Gi, family of G proteins that are present in platelets, and in the same relative proportions (23). The present studies clarify the role of G proteins and proteolysis in cell activation and desensi- tization by thrombin and demonstrate the utility of HEL cells as a model system in which to study these events.

MATERIALS AND METHODS

Inositol Phosphate Formation in Intact HEL Celk-HEL cells growing in RPMI 1640 medium (Gibco) supplemented with 10% fetal calf serum were resuspended at 5 X IO5 cells/ml in inositol-free RPMI medium containing 10% dialyzed fetal calf serum and incubated for 22 h with my~-[~H]inositol(lO pCi/ml, American Radiolabeled Chem- icals). When noted, pertussis toxin (List Biological Laboratories) was added for the final 16 h of the [3H]inositol loading period. Afterwards, the HEL cells were resuspended in serum-free RPMI supplemented with 10 mM HEPES, pH 7.3, a t a cell concentration of 1 X 106/ml and allowed to equilibrate for 1 h at 37 "C. A typical reaction mixture had a final volume of 600 pl of which the HEL cell suspension represented 540 pl and any additions such as thrombin were made in the remaining 60 pl. Reactions were terminated by the addition of 300 pl of 12% perchloric acid containing 3 mM EDTA, 1 mM dieth- ylenetriaminepentaacetic acid, and 15 pg of phytic acid. Adenosine, AMP, ADP, and ATP were added as internal standards. The per- chloric acid extract was neutralized and the [3H]inositol phosphates present were analyzed by high performance liquid chromatography on a Whatman Partisil-SAX column using a linear gradient from 0 to 1.5 M ammonium formate, pH 3.7. Peaks of radioactivity corre- sponding to the various [3H]inositol phosphates were detected by scintillation counting and identified using appropriate standards.

Inositol Phosphate Formation by HEL Cell Membranes-HEL cells radiolabeled with [3H]inositol were resuspended at 5 X lo6 cells/ml in homogenization buffer containing 10 mM Tris-HC1, 1 mM MgC12, 5 mM EDTA, 10 mM benzamidine, 20 pg/ml leupeptin, 0.1% (v/v) aprotinin, and 0.9 mM phenylmethylsulfonyl fluoride, pH 8.0. The cells were lysed in a Dounce-style homogenizer at 4 "C with >95% cell disruption confirmed by microscopic examination. A membrane pellet was prepared by sequential centrifugation at 250 X g for 5 min followed by 31,400 X g for 15 min and then resuspended in 1 mM EGTA with 50 mM HEPES, pH 7.0, at a final protein concentration of 1-2 mg/ml. The radiolabeled membranes were used immediately after preparation.

To assay ['H]inositol phosphate formation, 50 p1 of radiolabeled

NaC1, 50 mM KC1, 50 mM HEPES, 5 mM LiCl, 10 mM MgC12, 1.5 membranes were added to 100 pl of reaction buffer containing 10 mM

mM EGTA, and 2 mM ATP, pH 7.0. Additions such as thrombin, CaC12, and GTP were made in a further 50 p l to bring the final volume to 200 ~ 1 . Unless otherwise noted, the final concentrations of EGTA and GTP were 1 mM and 25 p ~ , respectively. The free Ca2' concentration in the reaction mixture was estimated using the micro- computer program, FreeCal (24). At the end of the incubation period (usually 4 min at 37 "C), the reaction was terminated by the addition

of 200 pl of 12% perchloric acid. After removal of any precipitate by sedimentation, the extract was neutralized and analyzed by high performance liquid chromatography as described above.

cAMP Formation-The formation by HEL cell membranes was measured using [cY-~'P]ATP as a substrate (25). In the studies com- paring cAMP formation and phosphoinositide hydrolysis in the same membrane preparation, the conditions of the cAMP assay were mod- ified to match those used in the phosphoinositide hydrolysis assay. The membranes were prepared from cells incubated in inositol-free RPMI, but without [3H]inositol. Typically, 31 p1 of membranes (1-2 mg/ml) were added to 62 pl of the reaction buffer used in the inositol phosphate measurements supplemented with [o~-~'P]ATP (28 pCi/ ml). Additions such as thrombin, CaC12, isobutylmethylxanthine, and GTP were made in a further 32 pl of bring the final volume to 125 p1. The final concentrations were: 1 mM EGTA, 0.3 mM CaC12, 0.4 mM isobutylmethylxanthine, 1 mM ATP, 25 p M GTP. The amount of cAMP present in intact HEL cells was measured by radioimmuno- assay (Du Pont-New England Nuclear).

Detection of Pertussis Toxin Substrates-HEL cell membranes were prepared as described above. Platelet membranes were prepared by nitrogen cavitation (9). Afterwards, the membranes were incubated for 60 min at 30 "C with 20 pg/ml pertussis toxin (preactivated with 20 mM dithiothreitol plus 0.125% SDS for 30 min at 30 "C), 20 pM NAD, ["PINAD (10,000 cpm/pmol), 1 mM EDTA, 5 mM dithiothre- itol, 10 mM thymidine, 0.2 mg/ml bovine serum albumin, and 10 mM HEPES, pH 8. The proteins were separated by SDS-polyacrylamide gel electrophoresis (26) with a 5% stacking gel and an 11% separating gel. Radioactivity was detected by autoradiography and quantitated by laser densitometry.

Ca2* Measurements-HEL cells were resuspended at a concentra- tion of 1.75 X lo6 cells/ml in RPMI containing 10% fetal calf serum and 20 mM HEPES, pH 7.3, and allowed to equilibrate for 30 min. Afterwards, 5 p~ Fura-2/AM (Molecular Probes) was added for 1 h at 37 "C before the cells were washed and resuspended in RPMI medium with 10% fetal calf serum. After an additional l -h equilibra- tion period, the cells were washed again and resuspended in RPMI with 10% fetal calf serum at 3 X lo6 cell/ml. Fluorescence were detected in a Perkin-Elmer spectrophotometer with excitation set at 340 nm and emission set at 510 nm. Approximate values for [Ca"], were calculated using an assumed ko of 224 nM (27).

Miscellaneous-Human a-thrombin and PPACK-inactivated thrombin were a gift from Dr. John Fenton I1 (Division of Labora- tories and Research, New York State Department of Health, Albany, NY). The activity of the a-thrombin was typically 3000-4000 units/ mg. The PPACK-thrombin was >95% inactivated. Bovine pancreatic trypsin treated with ~-tosylamido-2-phenylethyl chloromethyl ketone was obtained from Worthington. Hirudin and S2238 were obtained from Sigma. S2238 hydrolysis was assayed as previously described (17).

RESULTS

Previous studies have demonstrated that thrombin stimu- lates phosphoinositide hydrolysis in intact HEL cells (22). As a preliminary step for the present studies, the involvement of a G protein in this response was examined using membranes prepared from HEL cells radiolabeled with [3H]inositol. In the absence of GTP, thrombin caused little, if any, phosphoi- nositide hydrolysis (Fig. 1, left). In the presence of GTP, thrombin-induced [3H]inositol phosphate formation in- creased approximately &fold. GTP alone had a smaller effect. Dose-response curves showed that the effects of thrombin and GTP were half-maximal at approximately 5 nM and 5 pM, respectively (not shown). The nonhydrolyzable GTP an- alog, Gpp(NH)p, stimulated inositol phosphate formation to a level exceeding that produced by GTP alone and approach- ing that of thrombin plus GTP (Fig. I), as did NaF plus AlC1, when added to intact HEL cells (not shown). AlF; is thought to stimulate G proteins directly (28). Finally, as has been observed in other cells in which G proteins activate phospho- lipase C, thrombin plus GTP lowered the Ca2+ concentration required for phosphoinositide hydrolysis in HEL cell mem- branes into the range of Ca2+ concentrations normally found in unstimulated HEL cells (Fig. 1, right). These findings

960 Receptor and G Protein-mediated Responses to Thrombin

2 2000

g ? 1500

g 1000

E

0 - 500

0

I 1,4,5-IP3

Thrambtn

2000

1500

io00

-8 -7 -6 -5 -4 -3 DCa

+ CaCI$GTA

FIG. 1. Phosphoinositide hydrolysis by HEL cell mem- branes. In the studies shown on the left, membranes prepared from HEL cells radiolabeled with [3H]inositol were incubated for 4 min at 37 "C with EGTA (I mM) alone or with EGTA, CaC1, (0.3 mM), and combinations of thrombin (50 nM), GTP (25 p ~ ) , and Gpp(NH)p (500 p ~ ) as indicated. The free Ca2+ concentration in the presence of EGTA plus CaC1, was approximately 190 nM. [3H]Inositol phosphate formation was measured. The results shown are the mean f S.E. of five studies. In the studies shown on the right, radiolabeled mem- branes were incubated with or without thrombin (50 nM) and GTP (25 p ~ ) plus EGTA (1 mM) and sufficient CaCl, to give the approx- imate free Ca2+ concentrations indicated. The results shown are the mean f S.E. of three studies.

resemble those obtained in other cells, including platelets, and support the conclusion that a G protein mediates the interaction between thrombin receptors and phospholipase C in HEL cells.

Pertussis Toxin-In some types of cells agonist-induced phosphoinositide hydrolysis can be blocked with pertussis toxin. In general, this inhibition is cell- and agonist-specific with the toxin being effective in some cells with some agonists and not with others. Among other possibilities, this variability has been ascribed to differences in the G proteins and recep- tors which interact with phospholipase C. To examine the effect of pertussis toxin on thrombin responses, HEL cells were incubated overnight with 200 ng/ml pertussis toxin, a concentration that maximally ADP-ribosylates the available pertussis toxin substrates (see below and Ref. 23). In un- treated cells thrombin stimulated [3H]1,4,5-IP3, formation which peaked within 10 s and was followed by slower accu- mulations of 1,3,4-IP3, inositol bisphosphate, and inositol phosphate. This response was markedly diminished in the toxin-treated cells (Fig. 2).

Pertussis toxin also suppressed the inhibition of forskolin- stimulated cAMP formation normally observed when throm- bin is added to HEL cell membranes. The rate of cAMP formation in control HEL cell membranes rose from a mean of 25 pmol/min/mg in the absence of forskolin to 686 pmol/ min/mg with 20 p~ forskolin. When added 30 s before the forskolin, thrombin (50 nM) inhibited this increase by 27 f 2% (mean f S.E. n = 4), but had no effect on membranes prepared from toxin-treated HEL cells (1 f 5% inhibition). Therefore, in HEL cells, as in platelets, pertussis toxin sup- presses both of the G protein-mediated responses to thrombin, suggesting that in each case the G protein involved is a substrate for the toxin.

1 ,3,4-IP3

Control

40:u 0 10 20 30 40 50 60 0 I 10 20 30 40 50 60

600

500

400

300

200

1W

0 :

n - 0

5000 om

4000

- -

3000

2000

1000

0

Seconds

FIG. 2. The effect of pertussis toxin on thrombin-induced phosphoinositide hydrolysis in intact HEL cells. [3H]Inositol- labeled HEL cells were grown overnight with (open symbols) or without (closed symbols) 200 ng/ml pertussis toxin, resuspended, and then incubated with 50 nM thrombin. The results shown are the mean f S.E. of four studies. IP,, inositol bisphosphate; ZP,, inositol phos- phate.

In order to distinguish between the G proteins mediating activation of phospholipase C and inhibition of adenylylcy- clase by thrombin, a comparison was made between the con- centration of pertussis toxin required to block each of these responses to thrombin and the extent of ADP-ribosylation occurring at that toxin concentration. The assays were per- formed with membranes from HEL cells grown overnight with pertussis toxin. An indirect measure of the extent of ADP-ribosylation that occurred in the intact cell was obtained by incubating membranes prepared from pertussis toxin- treated HEL cells with additional pertussis toxin plus ["PI NAD to [32P]ADP-ribosylate any toxin substrates remaining unmodified. Membrane proteins were then analyzed by SDS- PAGE and autoradiography.

The results are shown in Fig. 3A. An approximately 41- kDa band of [32P]ADP-ribosylated protein was observed which co-migrated with the pertussis toxin substrates in plate- lets. As in platelets (9), this band can be resolved by isoelectric focusing into a set of radiolabeled proteins with similar mo- lecular weights, but different isoelectric points (23). All of these are members of the Gi, family (23). [32P]ADP-ribosyla- tion was half-maximal a t approximately 0.4 ng/ml pertussis toxin. The extent of [32P]ADP-ribosylation correlated closely with the extent of inhibition of thrombin's effects on 1,4,5- IP3 formation and forskolin-stimulated cAMP formation (Fig. 3B). Over a range of toxin concentrations, both responses were suppressed to the same extent. The same correlation was found between ADP-ribosylation and inhibition of cAMP formation by epinephrine (not shown). Therefore, these data suggest that the a subunits of the G proteins which mediate these responses are similar in their sensitivity to pertussis toxin and that loss of function is proportional to the extent of ADP-ribosylation of the various forms of G,.

Thrombin and Trypsin-Serine proteases other than thrombin can activate thrombin-response cells. Trypsin, for example, mimics many of the effects of thrombin on platelets (15, 29, 30). Fig, 4 compares trypsin and thrombin-induced inositol phosphate formation in HEL cells pretreated with or without pertussis toxin. In cells not exposed to pertussis toxin, trypsin stimulated inositol phosphate formation with a time

Receptor and G Protein-mediated Responses to Thrombin 961

[IAP] 0 0.1 0.4 1 4 0 nglm

0 0.01 0.1 1 10 100

I Petlussis Toxin ] (nglml)

FIG. 3. Correlation between ADP-ribosylation and loss of thrombin responses in pertussis toxin-treated HEL cells. A, membranes were prepared from HEL cells grown overnight in me- dium containing pertussis toxin. Afterwards, the membranes were incubated with dithiothreitol-activated pertussis toxin (5 pg/ml) and ['"PINAD and then analyzed by SDS-PAGE and autoradiography. Platelet membranes incubated with pertussis toxin and ["PINAD are shown for comparison. B, phosphoinositide hydrolysis, inhibition of CAMP formation, and the extent of ["PIADP-ribosylation were meas- ured in membranes prepared from pertussis toxin-treated HEL cells. The data for 1,4,5,-IP( formation (open circles) are the mean of two to five studies expressed as percent of the response to 50 nM thrombin in the absence of pertussis toxin (340 cpm/100 pg of protein). The results shown for suppression of CAMP formation (stippled squares) are the mean of nine studies expressed as percent inhibition of forskolin-stimulated CAMP formation (342 pmol/min/mg of protein). The extent of ["PIADP-ribosylation (solid circles) was determined as in A. The results are expressed as a fraction of the radioactivity found in the membranes prepared from cells that were not pretreated with pertussis toxin (n = 3).

course and magnitude identical to that for thrombin. Pertussis toxin inhibited the response to both proteases to the same extent.

A comparison was also made between the effects of throm- bin and trypsin on cAMP formation. When added to HEL cell membrane preparations, trypsin, like thrombin, inhibited adenylylcyclase activity. However, the extent of inhibition was far greater: under the conditions cited earlier in which thrombin inhibited forskolin-stimulated cAMP formation by 27%, trypsin (10 pg/ml) inhibited forskolin-stimulated cAMP formation by 88 +. 4%. Pertussis toxin abolished the inhibition caused by thrombin, but had no effect on the inhibition caused by trypsin. Presumably this difference between thrombin and trypsin reflects the broader substrate specificity of trypsin and is due in part to proteolysis of G, and/or adenylylcyclase by trypsin (31-33). In intact HEL cells, where access of the enzymes is limited to the outside of the cell, the effects of thrombin and trypsin on cAMP formation are identical (see below).

These observations suggest that thrombin and trypsin are capable of evoking responses from HEL cells by similar, if not identical, mechanisms. Changes in the cytosolic-free Ca2+ concentration in HEL cells loaded with Fura-2 were used to test this hypothesis further (Fig. 5). Both proteases caused a concentration-dependent increase in [Ca"];. Since this in- crease was only partially inhibited by removing extracellular Ca2+ with EGTA (not shown), it presumably reflects a com-

1000 - 10000

8000

6000

4000

2000 - 0 10 20 30 40 50 60

Seconds FIG. 4. Comparison of thrombin and trypsin-induced phos-

phoinositide hydrolysis. [3H]Inositol-labeled HEL cells grown overnight with (open symbols) or without (closed symbols) 200 ng/ml pertussis toxin were incubated with 50 pg/ml trypsin (triangles) or 50 nM thrombin (circles). The results shown are the mean of three studies. IP,, inositol bisphosphate.

Thrombin

Trypsin

2 7 E.

0.1

i?- 4

PPACK-Thr Thr 4 4

(MonMI (IOnM)

D.

H.

FIG. 5. Inhibition of HEL cell responses to thrombin trypsin by active site-inhibited thrombin. Fura-2-loaded HEL cells were incubated with thrombin (top) or trypsin (bottom) in the presence or absence of PPACK-thrombin at the final concentrations shown.

962 Receptor and G Protein-mediated Responses to Thrombin

bination of intracellular Ca2+ mobilization by IP3 and en- hanced Ca2+ influx. As was noted in astrocytoma cells (34), thrombin treated with the active site blocking reagent PPACK had no effect by itself, but partially inhibited the HEL cell response to native thrombin. PPACK-thrombin also inhibited the response to trypsin. In both cases the inhibition was concentration dependent. At a molar ratio of approxi- mately lO:l, PPACK-thrombin caused a small decrease in the trypsin response (Fig. 5, F versus E ) . At 201, the inhibition was more pronounced (Fig. 5, H versus G).

To be sure that the inhibition by PPACK-thrombin was not due to contamination of the preparation with residual PPACK, combinations of thrombin, PPACK-thrombin, and trypsin were prepared at the final concentrations shown in Fig. 5. Proteolytic activity was measured using the chromo- genic substrate, S2238. Under these conditions, PPACK- thrombin had no activity by itself and failed to inhibit hy- drolysis of the chromogenic substrate by thrombin and trypsin (not shown). Therefore, the results with the Fura-2-loaded HEL cells suggest that thrombin and trypsin work at the same site and that active enzyme is required.

Desensitization-Thrombin's ability to cause phosphoino- sitide hydrolysis in platelets and other cells is subject to desensitization: readdition of thrombin after an initial re- sponse fails to produce a second response (34-37). It is not known whether this loss of responsiveness reflects alterations in the thrombin receptor or the failure of post-receptor events. Changes in [Ca2+Ji and IP3 were used to address this issue. Readdition of thrombin to HEL cells following an initial response failed to evoke a second increase in the cytosolic Ca2+ concentration, as did addition of trypsin after thrombin or thrombin after trypsin (Fig. 6). Under the same conditions the increases in [Ca2++Ii caused by neuropeptide Y (NPY) and epinephrine were retained, although with some variability in magnitude from experiment to experiment (compare Fig. 6

A. 1 rnin

0.1 1 3 4 4 Thr Tryp €pi

B.

4 4 4 TVP Thr €pi

D.

4 4 4 Thr Tryp NPY

4 4 4 Tryp Thr NPY

4 4 NPY Thr

F.

4 €pi

FIG. 6. Desensitization of thrombin and trypsin-induced changes in cytosolic calcium. HEL cells were loaded with Fura-2 and then incubated with 50 nM thrombin (Thr), 50 pg/ml trypsin ( T u p ) , 100 nM neuropeptide Y, or 100 p~ epinephrine (Epi) at the times indicated.

with the data in Fig. 9C). In no case did prior exposure to thrombin abolish a subsequent response to NPY or epineph- rine, as it routinely did to trypsin or a second addition of thrombin (Figs. 6, 8A, and 9).

Whether the loss of the Ca2+ response is associated with a failure of IP3 formation was also tested. For these studies, HEL cells were incubated with either thrombin or NPY for 20 min, sufficient time for [3H]inositol phosphate levels to return to baseline. Afterwards, the cells were rechallenged with thrombin, trypsin, or NPY (Fig. 7). As would be expected if desensitization affects phospholipase C activation, throm- bin and trypsin caused little, if any, [3H]IP3 formation in cells previously exposed to thrombin. Prior exposure to NPY, on the other hand, had no effect on a subsequent response to thrombin (Fig. 7) and the response to NPY, although smaller than the initial response to thrombin or trypsin, was undi- minished in thrombin-desensitized cells (not shown).

The desensitization caused by thrombin proved to be both rapid in onset and persistent. In the studies shown in Fig. 8, hirudin was used to inactivate thrombin. Hirudin is a protein originally isolated from leeches which forms an irreversible complex with thrombin, blocking its catalytic site and its ability to bind to substrates (38,39). Hirudin does not inhibit trypsin and has been shown to rapidly dissociate thrombin

"11 I I

istagonist: Buffer Thrombin NPY Buffer Thrombin

2nd agonist: Thrombin Trypsin

FIG. 7. Desensitization of thrombin and trypsin-induced inositol phosphate formation after exposure to thrombin. HEL cells loaded with [3H]inositol were preincubated with buffer, 50 nM thrombin, or 100 nM neuropeptide Y for 20 min ("1st agonist") before adding 50 nM thrombin, 50 pg/ml trypsin, or 100 nM neuropeptide Y ("2nd agonist") as indicated. [3H]Inositol phosphate formation was measured 2 min later and is expressed as the increment in total IP3 (1,4,5-IP3 + 1,3,4-IP3) present (mean f S.E., n = 3).

1

2 0.5 $; 0.3

- 2 0.2 d g 0.1 &

0.05

A. 1 min

L 4 4 Thr Tryp

C.

4 4 4 Jhr H Tryp

E.

A 4 4 4

H Thr Jryp

D.

4 4 4 Thr H JVP

FIG. 8. The effect of hirudin on desensitization after throm- bin. Thrombin (Thr) (10 nM or =1 unit/ml), trypsin (Tryp) (25 pg/ ml), and hirudin (H) (10 units/ml) were added to Fura-2-loaded HEL cells at the times indicated.

Receptor and G Protein-mediated Responses to Thrombin

from its binding sites on platelets, even when added as long as 15 min after thrombin (40). Addition of hirudin prior to thrombin blocked the increase in [Ca2+]i and prevented the loss of trypsin responsiveness that would have otherwise occurred (Fig. 8B). Addition of hirudin after the thrombin response was complete failed to prevent the loss of the trypsin response even when the cells were incubated with the hirudin for several minutes before the trypsin was added (Fig. 8D) , as did washing the cells repeatedly and then incubating them with hirudin for 1 h (not shown). Addition of the hirudin at the peak of the Ca2+ transient resulted in a more rapid return to baseline and partially prevented the loss of trypsin respon- siveness (Fig. 8C).

These results suggest that desensitization requires catalyt- ically active thrombin, occurs shortly after thrombin addition, and persists after thrombin is removed. The duration of desensitization is addressed in the studies shown in Figs. 9 and 10. HEL cells were incubated with thrombin for 10 min and then washed. When tested 1 and 4 h later, the thrombin- treated cells failed to respond to either a second addition of thrombin (Fig. 9, B and D ) or to trypsin (not shown). How- ever, after incubation overnight the thrombin response re- turned to normal, as did the trypsin response (Figs. 9, E and 10, A and C ) . Inclusion of cycloheximide to retard protein synthesis inhibited recovery by >90% (Fig. 10).

A. Control (1 hr.) B. Thrombin-treated (1 hr.)

0.1 J c1 4 Thr

4 Thr

C. Control (4 hrs.) D. Thrombin-treated (4 hrs.)

4 4 c

Thr NPY 4 4 Thr NPY

E. Control (22 hrs.) F. Thrombin-treated (22 hrs.)

4 Thr

;a 4 Thr

FIG. 9. Duration of the desensitization caused by thrombin. HEL cells were loaded with Fura-2, incubated for 10 min with or without 50 nM thrombin, and then washed twice to remove the agonist. Changes in Fura-2 fluorescence in response to 50 nM throm- bin (Thr) , 50 pg/ml trypsin (Tryp), and 100 nM neuropeptide Y were measured 1,4, and 22 h later.

A. Control

’ 1 C. Control

963

6. Cycloheximide

c Thr

D. Cycloheximide

FIG. 10. The effect of cycloheximide on recovery from de- sensitization. HEL cells were incubated with 50 nM thrombin for 10 min, washed twice, and then incubated for 20 h with or without cycloheximide (10 pg/ml) before being loaded with Fura-2 and stim- ulated with 50 nM thrombin (Thr) or 25 pg/ml trypsin (Tryp).

DISCUSSION

Thrombin is a tightly regulated plasma protease which, when activated by factor Xa, is able to stimulate a variety of cells, including platelets. Using HEL cells as a model, the present studies have examined the interaction between thrombin, its receptors, and the G proteins that regulate phospholipase C and adenylylcyclase. In particular, we have focused upon the properties of the G proteins involved, the role of thrombin’s proteolytic activity, and the mechanism of cell desensitization caused by thrombin.

G Proteins-Although previous studies with platelets have provided ample evidence that thrombin activates phospholi- pase C via G proteins, it is not clear which G proteins are involved, in part because platelets are refractory to the entry of pertussis toxin unless permeabilized with a detergent such as saponin (6). Most of the available evidence suggests that thrombin-induced phosphoinositide hydrolysis is inhibited by pertussis toxin (6-9, 41). However, some studies have shown that the toxin activates platelets (42) and stimulates phos- pholipase C (43). HEL cells provided a means to readdress this issue. Our results show that in HEL cells, as in platelets, thrombin can stimulate phospholipase C and inhibit ade- nylylcyclase in a pertussis toxin-sensitive manner. Phosphoi- nositide hydrolysis requires GTP and submicromolar concen- trations of Ca2+ and can be evoked by adding nonhydrolyzable analogs of GTP to HEL cell membranes or fluoride to intact HEL cells. Furthermore, the pertussis toxin concentration required to inhibit phosphoinositide hydrolysis in HEL cells is well below the >1 pg/ml necessary to demonstrate inhibi- tion in saponin-permeabilized platelets (6).

These observations suggest that in HEL cells, as in plate- lets, a pertussis toxin-sensitive G protein mediates the inter- action between thrombin receptors and phospholipase C. This also appears to be the case in Chinese hamster CCL39 fibro- blasts (44,45), chick embryonic heart cells (46), and vascular smooth muscle cells (47). On the other hand, in IIC9 Chinese hamster embryonic fibroblasts (48), 3T3 fibroblasts (49),

964 Receptor and G Protein-mediated Responses to Thrombin

UMR 106-H5 osteosarcoma cells (50), and human umbilical vein endothelial cells (51), thrombin-induced phosphoinosi- tide hydrolysis has been reported to be unaffected by pertussis toxin under conditions in which other effects of the toxin could be demonstrated. Whether these differences between cells reflect differences in methodology, differences in the thrombin receptor, or differences in the G proteins that interact with thrombin receptors remains to be determined. However, two tentative conclusions can be made from cur- rently available data. The first has to do with the identity of the G protein that regulates phospholipase C. At present, there are only three known G proteins in platelets and HEL cells that are substrates for pertussis toxin. Based upon [35S] methionine labeling, in situ hybridization, and immunoblot- ting, their relative order of abundance appears to be GiOz >> Gins > Giml (23). Recent studies with Gi,-specific antisera have implicated Gi,, in the regulation of adenylylcyclase in platelets (52) and NG108 cells (53). The same approach has not been applied to phospholipase C activation, but our previous stud- ies show that thrombin receptors in platelets are coupled to members of the Gi, family in addition to Gimz (9). Although this does not constitute proof, the limited repertoire of avail- able pertussis toxin-sensitive G proteins in platelets and HEL cells implies that one of the Gi, family members may interact with phospholipase C.

A second conclusion is suggested by a comparison that was made between the extent of ADP-ribosylation and the inhi- bition of thrombin responses. The expectation was that dif- ferences in the G proteins which interact with phospholipase C and adenylylcyclase would translate into differences in the toxin concentration required to block these interactions. In fact, such differences were not found. At each pertussis toxin concentration both responses to thrombin were inhibited to the same extent. Furthermore, there was a 1:l correlation between ADP-ribosylation and impairment of function with both thrombin and epinephrine. This contrasts with a pre- vious study in which it was found that angiotensin’s ability to suppress cAMP formation in rat liver extracts was unaf- fected until >80% of the available pertussis toxin substrate was ADP-ribosylated (54). Although the results with liver cells could reflect the presence of an excess number of angio- tensin receptors and/or G proteins, our studies suggest that in HEL cells there may not be “spare” G proteins capable of substituting for those that have been ADP-ribosylated.

Receptors, Proteolysis, and Desensitization-The interac- tion between thrombin and its receptors was examined in two ways. First, by comparing HEL cell responses to thrombin with those to trypsin and, second, by re-examining the phe- nomenon of desensitization. Trypsin is of interest because it has been shown to duplicate the effects of thrombin on platelets by stimulating high affinity GTPase activity (55), activating phospholipase C (15, 29), inhibiting adenylylcy- clase (30), and causing aggregation and secretion (13, 14). This suggests that trypsin, like thrombin, activates platelets in a specific manner and that one or more G proteins is involved. This conclusion is extended by the present studies. As in platelets, trypsin was found to stimulate phosphoinosi- tide hydrolysis in HEL cells. The time course and magnitude of this response were similar to that caused by thrombin. Furthermore, pertussis toxin inhibited phospholipase C acti- vation by both proteases to the same extent and, as was recently reported by Jones et al. (34), proteolytically inactive PPACK-thrombin blocked the response to trypsin as well as thrombin.

A comparison of the effects of thrombin and trypsin on cAMP formation in HEL cells and platelets illustrates two

additional points. First, access to the inner membrane surface allows trypsin, with its broader substrate specificity, to pro- duce effects not seen with thrombin. For example, when thrombin was added to HEL cell (or platelet) membranes, CAMP formation was inhibited in a pertussis toxin-sensitive Gi-dependent manner. On the other hand, when trypsin was added to HEL cell membranes, cAMP formation was essen- tially abolished and the extent of inhibition was unaffected by pertussis toxin. In this case, inhibition is presumably due to proteolysis of G, and/or adenylylcyclase (31-33). Second, as is described in greater detail elsewhere,’ the Gi-mediated inhibition of adenylylcyclase observed with HEL cell mem- brane preparations is not apparent when thrombin is added to intact HEL cells. Instead, thrombin increases rather than decreases cAMP levels in intact HEL cells, just as it does in human umbilical vein endothelial cells (56). This increase in cAMP formation is the opposite of what occurs in platelets stimulated by thrombin and represents a departure from the analogy of thrombin responses between HEL cells and plate- lets. It is notable, however, that in all three types of cells the effects of thrombin and trypsin are the same: in intact plate- lets both proteases stimulate phosphoinositide hydrolysis and inhibit cAMP formation; in intact HEL cells and endothelial cells, both proteases stimulate phosphoinositide hydrolysis and raise cAMP levels.

The phenomenon of desensitization can serve as a useful marker for receptor-mediated events. Desensitization of thrombin responses has been reported in a variety of cells, includingplatelets (35), fibroblasts (36), endothelial cells (37), and astrocytoma cells (34). Conceivably, it could be due to an alteration of either the receptor itself, perhaps by proteolysis or phosphorylation, or to part of the post-receptor mechanism that mediates the agonist response. In the present studies, the onset and duration of desensitization were examined as keys to understanding the molecular basis for the interaction between thrombin and its receptors. In HEL cells, as in other cells, prior exposure of the intact cells to thrombin or trypsin precluded a further response to either protease. Desensitiza- tion was, therefore, reciprocal, as well as homologous. How- ever, it was not global since the cells retained their respon- siveness to at least two other agonists, neuropeptide Y and epinephrine. NPY has been shown to cause phosphoinositide hydrolysis in HEL cells, although not as vigorously as throm- bin (57). The mechanism by which epinephrine increases the cytosolic Ca2+ concentration in HEL cells is unknown, but it does not appear to involve phosphoinositide hydrolysis (22).

The time required for desensitization to occur and its duration after thrombin is removed provides additional clues to its etiology. Desensitization required active thrombin and did not occur in the presence of hirudin, which forms a tight complex with thrombin. The timing of the addition of hirudin and thrombin was critical. When added before thrombin, hirudin blocked changes in the cytosolic Ca2+ concentration and prevented desensitization. When added after thrombin, hirudin diminished both the primary thrombin response and thrombin-induced desensitization to an extent that depended upon the interval between the addition of the thrombin and hirudin. Partial inhibition of both was seen when the hirudin was added within the first minute after thrombin, suggesting that the two processes are related and that thrombin needs to be present for more than the few seconds necessary for detectable phosphoinositide hydrolysis to occur. A similar conclusion can be drawn from studies by Jaffe et al. (37) in which PPACK, rather than hirudin, was added to terminate the action of thrombin on endothelial cells. A delay before

L. Brass and M. Woolkalis, manuscript in preparation.

Receptor and G Protein-med

desensitization is complete could reflect either a requirement for sustained receptor occupation or variability in the re- sponse time of individual cells within the total population. If the latter, the more slowly responding cells may not have time to become activated before the thrombin was scavenged by hirudin, leaving them capable of responding to trypsin. When the HEL cells were washed repeatedly after exposure to thrombin, their ability to respond to thrombin and trypsin recovered only after more than 4 h, with full recovery requir- ing up to 20 h and preventable with cycloheximide.

These observations suggest several conclusions. The first is that thrombin and trypsin activate phospholipase C in HEL cells by similar, if not identical, mechanisms. Although tryp- sin is clearly able to activate signal transduction mechanisms in broken cells as a “nonspecific” proteolytic event (the pres- ent studies and Refs. 29 and 31-33), considerable evidence, including the effects of pertussis toxin, suggests that it is also able to interact with cell-surface receptors on intact cells in a specific manner. Based upon the identity of responses to thrombin and trypsin in intact HEL cells and platelets, the observed reciprocal desensitization, and the ability of PPACK-thrombin to inhibit responses to both thrombin and trypsin, the receptor is probably the same as for both pro- teases.

Finally, based upon its duration and persistence, desensi- tization appears to arise from an irreversible change in the signal transduction mechanism at a step up to or including the activation of phospholipase C. One possibility is that both this defect and the initiation of thrombin-stimulated events involve proteolysis of the receptor. If so, then the ability of cycloheximide to retard recovery suggests that the return of thrombin responsiveness requires re-synthesis of the affected protein, a mechanism for cell recovery unavailable in platelets.

Acknowledgments-We express our gratitude to Elizabeth Bel- monte, Calvin Shaller, and Marie Elena Hudzicki for their expert technical assistance.

REFERENCES 1. Tollefsen, D. M., Feagler, J. R., and Majerus, P. W. (1974) J.

2. Harmon, J. T., and Jamieson, G. A. (1985) Biochemistry 24, 58-

3. Gilman, A. G. (1987) Annu. Reu. Biochem. 56, 615-649 4. Brass, L. F. (1988) ZSZAtlas of Science: Biochem. 1 , 343-349 5. Brass, L. F., and Manning, D. R. (1991) Trends Cardiouasc. Med.,

6. Brass, L. F., Laposata, M., Banga, H. S., and Rittenhouse, S. E.

7. O’Rourke, F., Zavoico, G. B., Smith, L. H., Jr., and Feinstein, M.

8. Banga, H. S., Walker, R., Winberry, L., and Rittenhouse, S. E.

9. Brass, L. F., Woolkalis, M. J., and Manning, D. R. (1988) J. Biol.

10. Aktories, K., Schultz, G., and Jakobs, K. H. (1983) Naunyn-

11. Aktories, K., and Jakobs, K. H. (1984) Eur. J. Biochem. 145,

12. Furie, B., and Furie, B. C. (1988) Cell 53, 505-518 13. Davey, M. G., and Luscher, E. F. (1967) Nature 216,857-858 14. Martin, B. M., Feinman, R. D., and Detwiler, T. C. (1975)

15. Ruggiero, M., and Lapetina, E. G. (1985) Biochem. Biophys. Res.

16. Harmon, J . T., and Jamieson, G. A. (1986) J. Biol. Chem. 2 6 1 ,

17. Brass, L. F., and Shattil, S. J. (1988) J. Biol. Chem. 2 6 3 , 5210-

Biol. Chem. 249,2646-2651

64

in press

(1986) J. Biol. Chem. 261,16838-16847

B. (1987) FEES Lett. 2 1 4 , 176-180

(1988) Biochem. J. 252,297-300

Chem. 263,5348-5355

Schmiedebergs Arch. Pharmacol. 324 , 196-200

333-338

Biochemistry 14, 1308-1314

Commun. 131 , 1198-1205

15928-15933

5216

!iated Responses to Thrombin 965

18. Bienz, D., Schnippering, W., and Clemetson, K. J. (1986) Blood

19. Poncz, M., Surrey, S., LaRocco, P., Weiss, M., Rappaport, E., Conway, T., and Schwartz, E. (1987) Blood 6 9 , 219-223

20. Zimrin, A. B., Eisman, R., Vilaire, G., Schwartz, E., Bennett, J. S., and Poncz, M. (1988) J. Clin. Znuest. 81, 1470-1475

21. Poncz, M., Eisman, R., Heidenreich, R., Silver, S. M., Vilaire, G., Surrey, S., Schwartz, E., and Bennett, J. S. (1987) J. Biol. Chem. 262,8476-8482

22. Michel, M. C., Brass, L. F., Williams, A., Bokoch, G., LaMorte, V. J., and Motulsky, H. J. (1989) J. Biol. Chem. 264 , 4986- 4991

23. Williams, A,, Woolkalis, M. J., Poncz, M., Manning, D. R., Gewirtz, A., and Brass, L. F. (1990) Blood 7 6 , 721-730

24. Brass, L. F., and Shattil, S. J. (1989) Methods Enzymol. 169 ,

25. Salomon, Y., Londos, C., and Rodbell, M. (1974) Anal. Biochem.

26. Laemmli, U. K. (1970) Nature 227,680-685 27. Grynkiewicz, G., Poenie, M., and Tsien, R. (1985) J. Biol. Chem.

28. Sternweis, P. C., and Gilman, A. G. (1982) Proc. Natl. Acad. Sci.

29. Lazarowski. E. R.. and LaDetina, E. G. (1990) Arch. Biochem.

68,720-725

355-371

58,541-548

260,3440-3450

U. S. A. 79,4888-4891

Biophys. 276, 265-269 . .

30. Jakobs. K. H.. and Grandt. R. (1988) Eur. J. Biochem. 172.255- 260

Chem. 253,2921-2926

, . ,

31. Anderson, W. B., Jaworski, C. J., and Vlahakis, G. (1978) J. Biol.

32. Hughes, R. J., and Ayad, S. R. (1980) Biochim. Biophys. Acta

33. Stiles, G. L., and Lefkowitz, R. J. (1982) J. Biol. Chem. 257,

34. Jones, L. G., McDonough, P. M., and Brown, J. H. (1989) Mol. Phmmacol. 36,142-149

35. Huang, E. M., and Detwiler, T. C. (1987) Biochem. J. 242 , 11- 18

36. Paris, S., Magnaldo, I., and Pouyssegur, J. (1988) J. Biol. Chem. 263,11250-11256

37. Jaffe, E. A., Grulich, J., Weksler, B. B., Hampel, G., and Watan- abe, K. (1987) J. Biol. Chem. 262,8557-8565

38. Bagdy, D., Barabas, E., Graf, L., Peterson, T. E., and Magnusson, S. (1976) Methods Enzymol. 45 , 669-678

39. Stone, S. R., and Hofsteenge, J. (1986) Biochemistry 2 5 , 4622- 4628

40. Tam, S. W., Fenton, J . W. 11, and Detwiler, T. C. (1979) J. Biol. Chem. 254,8723-8725

41. Nakashima, S., Hattori, H., Shirato, L., Takenaka, A., and No- zawa, Y. (1987) Biochem. Biophys. Res. Commun. 148 , 971- 978

42. Banga, H. S., Walker, R. K., Winberry, L. K., and Rittenhouse, S. E. (1987) J. Biol. Chem. 2 6 2 , 14871-14874

43. Lapetina, E. G. (1986) Biochim. Biophys. Acta 8 8 4 , 219-224 44. Paris, S., and Pouyssegur, J . (1986) EMBO J. 5, 55-60 45. Chambard, J . C., Paris, S., L’Allemain, G., and Pouyssegur, J.

46. Chien, W. W., Mohabir, R., and Clusin, W. T. (1990) J. Clin.

47. Huang, C., and Ives, H. E. (1989) J. Bwl. Chem. 264,4391-4397 48. Raben, D. M., Yasuda, K., and Cunningham, D. D. (1987) Bio-

49. Murayama, T., and Ui, M. (1985) J. Biol. Chem. 260,7226-7233 50. Babich, M., King, K. L., and Nissenson, R. A. (1990) Endocri-

51. Brock, T. A., and Capasso, E. L. (1989) Am. Rev. Respir. Dis. 140 , 1121-1125

52. Simonds, W. F., Goldsmith, P. K., Codina, J., Unson, C. G., and Spiegel, A. M. (1989) Proc. Natl. Acad. Sci. U. S. A. 86, 7809- 7813

53. McKenzie, F. R., and Milligan, G. (1990) Biochem. J . 267, 391- 398

54. Pobiner, B. F., Hewlett, E. L., and Garrison, J. C. (1985) J. Biol. Chem. 260, 16200-16209

55. Jakobs, K. H., and Aktories, K. (1988) Biochem. J . 249,639-643 56. Woolkalis, M. J., Manning, D. R., and Brass, L. F. (1990) J. Cell

57. Motulsky, H. J., and Michel, M. C. (1988) Am. J . Physiol. 255 ,

630,202-209

6287-6291

(1987) Nature 326 , 800-803

Znuest. 85,1436-1443

chemistry 26,2759-2765

nology 126,948-954

Biol. 111 , 215a (abstr)

E880-E885