The M, = 160,000 epidermal growth factor (EGF) receptor in A431 cells is partially cleaved during mem- brane isolation to a M, = 145,000 polypeptide contain-.
Proteolytic Cleavageof Epidermal Growth Factor Receptor A Ca"-DEPENDENT, SULFHYDRYL-SENSITIVE PROTEOLYTIC SYSTEM IN
A431 CELLS* (Received for publication, March 10,
1982)
Dan Cassel and Luis Glaser From the Department of Biological Chemistry, Division of Biology and Biomedical Sciences, Washington Uniucrsity School of Medicine, St. Louis, Missouri 63110
membrane preparations (3), and that the EGF receptor itself The M, = 160,000epidermalgrowthfactor(EGF) receptor inA431 cells is partially cleaved during mem- is the major substrate for the EGF-dependent phosphorylabrane isolation to a M, = 145,000 polypeptide contain- tion (8-10). We have shown that ["'SIATPyS is an effective kinase and ing both EGF binding and phosphate acceptor sites. We substrate for the EGF receptor-associated protein of show that the proteolytic degradation of the EGF re- that the thiophosphorylated receptor resists the action ceptor depends upon the presence of Ca2+ in the me- membrane-bound phosphatase(s)(11). Thus, the phosphoryldium used to scrape the cells from the substratum. Only ation and thiophosphorylation reactions provide a convenient the high molecular weight form of the receptor is de- means for efficient labeling and identification of the receptor tected in membranes prepared in the absence ofCaz+. in plasma membranes. Ca2+-dependent proteolysis occurs rapidly(tlIz 5 min) Using either affinity labelingwith I2"I-EGFor EGF-dependfollowing cell scraping. Proteolysis results ina de- ent phosphorylation, several investigators have observed mulcrease in EGF-dependent phosphorylation of the receptiple forms of the receptor in membrane preparations. These tor while retaining EGF binding capacity. In addition, include preparations from A431 cells (3, 8 ) , 3T3 and HF-15 membranes containing the uncleaved form of the re- cells (€9, human placenta(6), and normal ratkidney cells (12). ceptor reveal a substantial increase in EGF-dependent Linsley and Fox (13) have shown that A431 membrane prepphosphorylation of proteins with M, 80, 89, and 185 arations contain varying proportions of proteins of M , = 115, X lo3. In the presence of Ca", addition of iodoacetic acid to the scraping mediumstrongly inhibits receptor 145, and 160 X lo3which could be labeled with ""I-EGF; only fragmentation, whereas other inhibitors (phenylmeth- the M , = 145 and 160 X lo3proteins were substrates for EGFylsulfonyl fluoride, leupeptin, and pepstatin) have no dependent phosphorylation. By contrast, only a single M , = effect. The results implicate arole for a Ca2+-dependent,150-170 X 10' component was detected uponlabeling of intact cells with '2"II-EGF, suggestingthat the receptoris subjected SH-sensitive protease in EGF receptordegradation. to partial proteolytic cleavage duringmembrane isolation. Prevention of proteolysis yields membranepreparaa membrane tions with highly active EGF-dependent kinase system. O'Keefe et al. (14) have been able to obtain preparation from human placenta containing only the high molecularweight receptor byemployingacombination of protease inhibitors (EDTA, PMSF, and IAA).However, it Epidermal growth factorstimulatesthe proliferation of has not been reported which of these inhibitors was effective many cell types both in vivo and in cultured cells (1). The in preventing receptor degradation. effects of EGF' on cellular functions are mediated througha Due to its apparently ubiquitous presence in different cell specific receptor in the plasma membrane. Specific binding types and itsspecificity for the EGF receptor, the proteolytic sites possessing high affinity for EGF have been detected in system that cleaves the receptor is of interest. Characterizavarious cells and cell membranepreparations (1) and are tion of the proteolytic system could also provide the means to present in very large amounts in the humanA431 epidermoid avoid proteolysis and thus obtain membranepreparations carcinoma cell line (2, 3). The EGF binding entity has been containing the intact receptor. In this paper, we show that identified through the covalent attachment of '"I-EGF which proteolytic degradation of the EGF receptor during isolation occurs spontaneously and can be enhanced by cross-linking of membranes from A431 cells is Ca'+-dependent and can be agents. Using these methods, components of apparent M , of eliminated by removal of Ca2+ from themedium. Some char150-190 X 10' have been detected on SDS gels by different acteristics of this proteolytic degradation aredescribed. investigators in various cells (4-8). Recently it has been discovered that EGF enhances protein MATERIALS AND METHODS phosphorylation catalyzed by kinase activities present in the Sodium iodoacetate, leupeptin, pepstatin, and PMSF were pur* This research was supported by National Institutes of Health chased from Sigma. ["SS]ATPyS, [y'"P]ATP, and ENHANCE were
-
-
Grants GM 18405,GM 28002,and National Science Foundation Grant PCM 8011973.Tissue culture medium was obtained from the Washington University Basic Cancer Center, supported by Grant CA 16217. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. ' The abbreviations used are: EGF, epidermal growth factor; IAA, iodoacetic acid; ATPyS, adenosine 5'-O-(thiotriphosphate); TES, N (tris[hydroxymethyl]methyl-2-amino)ethanesulfonic acid; PMSF, phenylmethylsulfonyl fluoride;CMF-Hanks, calcium and magnesiumfree Hanks balanced salt solution; SDS, sodium dodecyl sulfate.
from New England Nuclear. EGF was prepared by the method of Savage and Cohen (15)and was labeled with 1'" as described (16). Protein was determined by the method ofLowry et al. (17)using bovine serum albumin as standard. A431 cells were propagated in Dulbecco's modified Eagle's medium containing 10% calf serum, 100 units/ml of penicillin, and 100 pg/ml of streptomycin. Membranes were prepared essentially as described by Thom et al. (18).Cells in 150 cm2 dishes were washed twice with calcium and magnesium-free Hanks' balanced salt solution containing 20 mM 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid, pH 7.3 (CMF-Hanks). Cells were scraped from duplicate dishes using sharpedged policemen in 10 ml of ice-cold CMF-Hanks containing various
9845
9846
Ca2+-dependent Proteolysis
of EGF Receptor
additions to be tested as proteolysis stimulants and inhibitors. The cellsuspensionswerequicklytransferredtocentrifuge tubes immersed in ice. Onehour following scraping,the cells were collected by centrifugation (400 X g for 5 min at 4 "C), then processed for membrane preparationas described (18)using Ca2+-freemedia throughout Mr the procedure. Membranes were labeled by the EGF-dependent phosphorylation or thiophosphorylation reactions essentiallyas previously described (11). Reaction mixtures in a final volume of40 pl contained 1.2 mM MnC12, 0.5mM thiophosphate, 0.125 mg/ml of bovine serumalbumin, 25 mM N-(tris[hydroxymethyl]methyl-2-amino)ethanesulfonicacid, pH7.4,40 p~ adenosine 5'-(/3,y-imino) triphosphate, 5 p~ of eithsr [y"P]ATP or ["'S]ATPyS (2-5 X 10" cpm), 0.2 pg of EGF and 20-40 pg of A431 membranes. Reaction was for20 and 30 min at 0 "C in the presence of [y"P]ATP and [''SlATPyS, respectively, and was terminated by addition of 30 pl of5% SDS/12% 2-mercaptoethano1/30% glycerol. Following 20-min incubation at 60 "C, 2O-pl samples were taken for determination of the incorporation of radioactivity by precipitation with trichloroacetic acid in the presence of y-globulins added as protein carrier (for details see Ref. 11). A duplicate 20-pl aliquot was taken for SDS gel electrophoresison 5-15% linear acrylamidegradient gel (11). Gels were stained withCoomassieblue, destained, dried, and autoradiograms were prepared onpreflashed KodakXAR-5 film. Molecularweightmarkersincludedmyosin FIG. 1. Polyacrylamide SDS gel electrophoresis of labeled (200,000), /3-galactosidase (116,500), transferrin (81,000), bovine serum proteins from A431 membranes prepared in the presence or albumin (68,000), catalase (60,000),ovalbumin (43,000), trypsinogen absence of Ca2+ and Mg2*. Cells were scraped from the substratum in the absenceorpresence of 1 mM each ofCa" or Mg'+, then (24,000). and lysozyme (14,000). "'I-EGF binding was determined accordingto Carpenter et al. (3). processed for membrane preparation. Membranes were labeledwith Incubation mixtures( 5 0 pl) contained 25 mM N-(tris[hydroxymethyI] [y-92P]ATPin the absenceorpresence of EGFandsubjectedto methyl-2-amino)ethanesulfonic acid, pH 7.4, 0.125 mg/ml of bovine electrophoresis. Left, Coomassie blue (CB) staining (40 pg of memserum albumin, 12 ng of"'1-EGF (90,000cpm/pmol) and 2.5 pg of branes per lane); right, autoradiography. membraneprotein.Incubation was for 10 min at 25"Cand the amount of membrane-bound '*'I-EGF was determined following filtration through EGWP Millipore filters. Nonspecific binding, deter- labeled in membranes prepared with Ca" ( M , 19-20, 40, cleavmined in the presence of 2.5 pg of unlabeled EGF,was approximately and 94 X 10'). These bands might represent proteolytic 2% of the total binding. age products of the receptor or other membrane proteins which are substrates for EGF-dependent phosphorylation. RESULTS Careful inspection of the Coomassie blue-stained gels reveals Effect of ea2+ on EGF Receptor Proteolysis-The proceseveral changes in minor components which parallelthe above dure of Thom et al. (18)which is commonly used for prepa- mentioned differences intheautoradiographypattern beration of A431 membranes involves scraping of the cells from tween membranes prepared in the presence or absence of the substratum in isotonic medium containing 1 mM each of Ca2+. Ca2+ and M 8 + ions, followed by hypotonic lysis and memFig. 2 illustrates the effect of the additionof different Ca2+ brane purification. These membrane preparations oftencon- concentrations to the scraping medium on receptor fragmentain two forms of the EGF receptor(Mr 160 and 145 X lo3) tation, as determined by scintillation counting of individual that are substrates for EGF-dependent phosphorylation. Fig. bands from SDS gels of membranes incubated with [y-"PI 1 shows that both formsof the receptor are present in mem- ATP in the presence of EGF. While 50 p~ Ca2' had little branes prepared from cells that have been scraped in the effect, the presence of Ca2' concentration of 250 pM or more presence of 1 mM Ca2'. By contrast, only the high molecular causes a remarkable decrease in the M , = 160 x lO'-labeled weight phosphorylated receptor is detected in membranes receptor and increase in the M , = 145 X 10' fragment. The from cells that have been scraped either in the absence of small amount of label found in the M , = 145 X lo3 region in Ca2' or in the presence of 1 mM Mg2+. The Ca2'-dependent the absence of Ca2+ was not furtherreduced by the addition alteration in the distribution of radioactivity among the M , of 1 mM EDTA to the scraping medium (not shown). This is = 160 and 145 X lo3bands can be correlated with specific consistent with the observed dependency of the proteolytic alteration in Coomassie blue-staining of proteins with similar fragmentation upon the presenceof substantial Ca2+concenmolecular weight, whereas the staining pattern of other mem- trations in the medium (Fig. 2).Note that the total amount of brane proteins essentially is unchanged (Fig. 1).These findings labeled M , = 160 and 145 X lo3 proteins decreases by up to establish a Ca" dependency for the proteolytic fragmentation 40% a t t h ehigher Ca2' concentration. Possible interpretations of the EGF receptor that takesplace following cell scraping. for this result areconsidered below (see Fig. 4). It is noteworthy that additionof Ca2+to isolated membranes Effect of Inhibitors on Ca2+-dependent Proteolysis-To containing theuncleaved receptor (i.e. prepared in the absence characterize further the proteolytic system, we tested the of Ca") does not result in receptor fragmentation (data not effect of addition of protease inhibitors to scraping media with shown).Thus,theproteolyticsystem is eitherabsent or or without added Ca2+ on receptor fragmentation (Fig. 3). In inactive in the purified membranes. the presenceof Ca2+,IAA effectively inhibited the generation T h e absence or presence of Ca2+ in the scraping medium of the M r = 145 X lo3 receptor fragment. Addition of IAA to also effects thelabeling pattern of proteins other than the M , the Ca2'-containing medium prevented some of the alterations 145-160 X lo3bands (Fig. 1 ) : Two bands withM, 80 and 89 in the labeling pattern of other membraneproteins, including X lo3 as well as a M, = 185 X lo3 componentaremore the decrease in the amount of 32Pin the M , 80-89 X l o i intensely labeled in membranes prepared in the absence of bands. T h e inhibition of receptor fragmentation was halfCa2'. On the other hand, several bands are more intensely maximal at -0.1 mM IAA and reached -90% at 4 mM IAA 'The pattern of these minor components is more evident in Fig. 3 (data not shown). Membranes prepared using Ca2'-free mewhere longer exposure time was applied duringautoradiography. dium with or without IAA revealed similar extent of EGF-
-
-
-
-
Ca2+-dependent Proteolysis
of
EGF Receptor
9847
Time Course of Cap++-dependent Proteolysis-To study the time course of the proteolytic fragmentation, the cells were scraped from the substratum at 0-4 "C in the presence of 1 mM Ca2+ and proteolysis was stopped a t different times by addition of 4 mM IAA. In this experiment, the receptor was labeled by the EGF-dependent incorporation of :"S from ["SI ATPyS (11).In Fig. 4A, incorporation of intomembrane proteins was determined by gel electrophoresis. The results show that the decreasein the M , = 160 X 10'and increase in the M , = 145 X 10:'substrates for EGF-dependent thiophosphorylation occur rapidly following removal of the cells from the dish. The decrease in the M , = 160 X 10' substrate is steeper and proceeds for longer periods than the increase in I _ " 1 . the M , = 145 X 10' labeled protein, resulting in a net decrease 4 c' " b b 5 0.25 I in total ( M , = 145 + 160 X 10')labeled proteins. A possible Ca2' In scrap~nqmedlum, mM explanation for these findings is that the quantity of the M , FIG. 2. Effect of addition of different Ca'+ concentrations = 145 X 10:'receptor fragment is under a quasi-steady state during cell scraping on EGF-dependent phosphorylation of the M,= 160 and 145 X lo3proteins in the membrane preparations. representing its generation from the M , = 160 X 10'receptor Membranes were prepared from A431 cells scraped in the presence of and i t s further degradation by the proteolytic system. Alterbe a better Ca'+ at the concentrations indicated. After phosphorylation of the natively, the M , = 160 X 10' receptormight membranes with [y-'I'P]ATI' in the presence of EGF and SIX-gel substrate for EGF-dependent thiophosphorylation than the electrophoresis, the M , = 160 and 145 X 10:' bands were excised from M , = 145 X 10'polypeptide. the gel and the amount of radioactivity determined. Increasing incubation times following cell scraping in the presence of Cap+resulted in a significant loss of EGF-dependent thiophosphorylation activity in the membrane preparainhibitor tions. By contrast, the number of "'I-EGF binding sites reCa2' - t mained unchanged (Fig. 4B).This indicates that, while the -V proteolytic fragments of the receptor retain EGF binding, Some of them are depleted of phosphate acceptor sites and/or kinase activity. Apossible receptor fragmentwith these properties has been detected by Linsley and Fox in A431 mem185Kbranes 160K4
-
I )
145KF 9 4L 89K80KF
350h
B
300
40K200 250r'=-
L--"
20K-
I 2 3 4 5 6 7 8 9 1 0 FIG. 3. Effect of inhibitors added to the scraping medium on EGF receptor proteolysis. Cells were scraped from the dish in the presence orabsence of 4 mM Ca" and,whereindicated, in the presence of inhibitors (1.4.4 = 4 mM, PMSF = 0.1 mM, leupeptin ( L e u ) , andpepstatin (Pep) = 10 @g/ml). Lane 5, * = cellswere incubated for 15 min a t 4 "C with IAA and Ca", then washed with CMF-Hanks' and scraped in the presence of Ca'+. Membranes were labeled with [y-"'P]ATP + EGF and subjected to SDS-gel electrophoresis and autoradiography.
dependent phosphorylation (Fig. 3) and "'I-EGF binding (not shown) indicating that under these conditions IAA does not interfere with receptor function. When cells were treated with IAA and then washed and scraped in the presence of Cap+, the resulting membranes contained significant amount of the M , = 145 X 10' polypeptide (Fig. 3, lane 5). This indicates that the proteolytic system is localized within the cell and becomes effectively exposed to IAA only during cell scraping. Unlike IAA, PMSF (0.1mM), leupeptin, and pepstatin (10 pg/ml each) failed to inhibit the Cap+-dependent generation of the M , = 145 X 10'fragment, and also did not prevent the alteration in the autoradiographic patternof the lower molecular weight polypeptides (Fig. 3).
30
20
TIME
OF I O 0 ADDITION, MIN
FIG.4. Time course of EGF receptor fragmentation following scraping of cells from the substratum. At different times after scraping of cells in the presence of 1 mM Ca". proteolysis was stopped by addition to the scraping medium of concentrated IAA solution to a final concentration of 4 mM. Membranes were then isolated and tested for receptor functions. A , membranes were labeled with ["S]ATI'pS + EGF, and the amountof the '"S-labeled M , = 145 and I60 X IO' forms of the receptor was determined following separation by SDS-gel electrophoresis,B , A, total amountof trichloroacetic acid-precipitable :'"Sin membranes treated with ['"S1ATPpS + E G F B, '"'I-EGF binding to the membrane preparations determined at a saturating EGF concentration (see "Materials and Methods"). The same membrane preparations were used for theexperiments illustrated in A and B.
9848
Ca2+-dependent Proteolysis
of EGF Receptor
method of Thom et al. (18) employed by us; possibly these proceduresmightcausean increasedinitial disruption of cellular organization resulting in the mobilization of Ca2+from internal pools and activation of proteolysis. Yet another possibility is that the Ca'+ dependency of proteolysis is not an absolute requirement but rather depends upon the experimental conditions. By eliminating proteolysis we have been able to obtain A431 membrane preparations with highly active EGF-dependent phosphorylation. These membranesalso reveal a significant increase in the phosphorylation of two proteins of M , 80-89 X lo3(Figs. 1 and 3). This is of interest since a protein with similar molecular weight has been found to undergo a rapid EGF-dependent phosphorylation in intact cells (21) and is considered as a possible mediator of the ECF response. Our results may suggest that both the receptor and theM , 8089 X lo3 proteins can be cleaved by the sameCa'+-dependent SH-sensitive proteolytic system, suggesting a possible relationship among these proteins and the receptor. After completion of this work, a paper by Cohen et al. (22) has appeared documenting the presence of uncleaved EGF DISCUSSION receptor in shed plasma membrane vesicles from A431 cells. Our results suggest that EGF receptor fragmentation in The absence of receptor proteolysis in thevesicles is inagreeA431 cells that havebeen scraped from the substratumis due ment with our findings of the absence of Ca2*-dependent to aCa"-dependent proteolytic process. Furthermore, the proteolysis in purified membranes suggesting that the protesensitivity of the proteolysis to ratherlow IAA concentrations olytic system is not localized in the plasma membrane. implicates the involvement of an SH-sensitive protease, while the ineffectiveness of PMSF, leupeptin, and pepstatin seems REFERENCES to rule out several types of serine proteases and the lysosomal 1. Carpenter, G . , and Cohen, S. (1979) Annu. Rev. Biochem. 48, cathepsins B and D. The following data make itunlikely that 193-216 proteolysis is mediated by the CaZf-dependent proteases that 2. Fabricant, R.N., DeLarco, J . E., and Todaro, G . J . (1977) Proc. are known to be present in serum. 1) Serum proteases that Natl. Acad. Sci. U. S. A . 74, 565-569 are Ca"-dependent are not SH-sensitive(19). 2)We find that 3. Carpenter, G., King, L. E., and Cohen, S . (1979) J. Biol. Chem. treatment of cells with IAA followed by washing does not 254,4884-4891 prevent the generation of receptor fragmentsupon subsequent 4. Das, M., and Fox, C. F. (1978) Proc. Natl. Acad.Sci. U. S. A . 75, 2644-2648 scraping of the cells in the presenceof Ca2+(Fig. 3). Since the IAA treatment should have irreversibly inactivated externally 5. Baker, J., Simmer, R. L., Glenn, K. C., and Cunningham, D. C. (1979) Nature 278, 743-745 adsorbed proteases, we conclude that the proteolytic system 6 . Hock, R. A., N e x ~E., , and Hollenberg, M. D. (1979) Nature 277, is localized within the cell. We have so far been unable to 403-405 determine the subcellular localization of the protease. Addi7. Wrann, M. M., and Fox, C. F. (1979) J. Biol. Chem. 254, 8083tion ofCa'+ to membranes prepared without Ca2+ does not 8086 result in receptor fragmentation, indicating that the protease 8. Linsley, P. S., and Fox, C. F. (1980) J . Supramol. Struct. 14,441459 is either absent or inactive in the purified membranes. Also, 9. King, L. E., Carpenter, G . , and Cohen, S . (1980) Biochemistry 19, preliminary attempts to obtain proteolytic cleavage products 1524-1528 of the receptor in membranes prelabeled with ["5S]ATPyS by 10. Cohen, S., Carpenter, G., and King, L. E. (1980) J. Biol. Chem. incubation with cell extracts in the presence of Ca2+ have not 255,4834-4842 been successful. 11. Cassel, D., and Glaser, L. (1982) Proc. Nutf. Acad. Sci.U. S. A . 79,2231-2235 In agreement with previous observations (13) we find that receptorfragmentationoccwsrapidlyafterthe cells are 12. Fernandez-Pol, J. A. (1981) J. Biol. Chem. 256,9742-9749 13. Linsley, P. S., and Fox, C. F. (1980) J. Supramol. Struct. 14,461scrapedfromthedish (Fig. 4). Conceivably thescraping 471 procedure may disrupt membrane permeability barriers, re14. O'Keefe, E. J., Battin, T. K., and Bennet, V. (1981) J. Supramol. sulting in the entry of Ca2' into the cells and activation of Struct. 15, 15-27 proteolysis. Whether such changes in Ca2+fluxes play a role 15. Savage, C. R., and Cohen, S . (1972) J . Biol. Chem. 247, 76097611 in receptor metabolismin the intactcell is notknown. Cellular 16. Lieberman, M. A., Rothenberg, P., Raben, D. M., and Glaser, L. proteasesthatareCa2+-dependentandSH-sensitivehave (1980) Biochem. Biophys. Res. Commun. 92,696-702 been found in various cells and tissues (reviewed in Ref. 20), 17. Lowry, 0. H., Rosebrough, N. J., Farr, A. L., and Randall, R. J. but their function in normal cell processes remains obscure. (1951) J. Biol. Chem. 193,265-275 Regardless of the significance of receptor fragmentation, 18. Thom, D., Powell, A. J., Lloyd, C. W., and Rees, D. A. (1977) Biochem. J. 168,187-194 the present findings make it possible consistently to obtain A431 membranes that are essentially free of receptor frag- 19. Jackson, C.M., and Nemerson, Y. (1980) Annu. Reu. Biochem. 49, 765-811 (8, 14) have ments. It should be noted that other investigators Tanaka, K., Hatanaka, M. and Murakami, T. (1981) observed EGF receptor fragments in membrane preparations 20. Murachi, T., Adu. Enzyme Regul. 19,407-424 despite the absence of added Ca2' in the media used during 21. Hunter, T., and Cooper, J. A. (1981) Cell 24, 741-752 the initial steps of membrane preparation. The procedures 22. Cohen, S., Ushiro, H., Stoscheck, C., and Chinkers, M. (1982) J . used by these investigators are somewhat different from the Biol. Chem. 257, 1523-1531
In order to simplify the procedurefor determination of EGF receptor fragmentation, we have attempted to reconstitute the proteolytic system. Membranes prepared in the absence EGF as described of Ca" were labeled with ["S]ATPyS under"Materialsand Methods," andthen washed. These washed membrane preparations retain most of the incorporated "'S label due to the resistance to phosphatase of the thiophosphorylated EGF receptor(11).The 35S-labeledpreparations were then incubated with A431 cell fractions in the presence of 4 mM Ca2+ and receptor fragmentation was determined by SDS-gel electrophoresis. These experiments were essentially negative and only a marginal increase in the ratio of M , = 145 X 10'/Mr = 160 X lo3 labeled proteins in the membranes could be detected. These results were obtained when thiophosphorylated membranes were incubated with complete A431 homogenates or scraped cells suspensions or with thehigh speed supernatants from cell homogenates (data not shown). It therefore appears that the proteolytic degraa certain degree of dation of thereceptordependsupon cellular integrity.
+
-
-
