Intracellular processing of epidermal growth factor and its effect on ligand-receptor interactions.

When normal human fibroblasts are brought to a steady state with 125I-labeled epidermal growth factor (125I-EGF), greater than 90% of the radioactivity is intracellular. We investigated this material to determine whether the 125I-EGF is intact or degraded. Our results show that 125I-EGF is rapidly processed after internalization and can be resolved into four peaks by native gel electrophoresis. These different forms were isolated and tested for their ability to bind to cell-surface EGF receptors. The first processed form was fully capable of binding to EGF receptors, but the second processed form could not. The third form was a collection of small degradation products. We calculated that at steady state about 60% of internalized "125I-EGF" was in a form still able to bind to EGF receptors. We then investigated the ability of different reported inhibitors of EGF "degradation" to block the processing of EGF. Although inhibitors of cathepsin B (leupeptin, antipain, N alpha-p-tosyl-L-lysine chloromethyl ketone, and chymostatin) were able to inhibit the release of monoiodotyrosine from treated cells in a time- and concentration-dependent manner, they had little effect on the processing step that apparently inactivates 125I-EGF. In contrast, agents that raised intravesicular pH, such as methylamine and monensin, inhibited the initial steps in EGF processing as well as the later steps. Low temperatures inhibited the transfer of 125I-EGF to the lysosomes and inhibited the conversion of EGF to a nonbindable form, but had little effect on the initial processing. We conclude that the intracellular processing of EGF is a multistep process that is initiated prior to lysosomal fusion, involves cathepsin B activity, and requires an acidic pH. In addition, many of the protease inhibitors that have been utilized to investigate the role of EGF degradation in mitogenesis do not block the conversion of EGF to a form that is apparently unable to interact with its receptor.

When normal human fibroblasts are brought to a steady state with l2'I-1abeled epidermal growth factor (12"I-EGF), greater than 90% of the radioactivity is intracellular. We investigated this material to determine whether the "'I-EGF is intact or degraded. Our results show that ' T -E G F is rapidly processed after internalization and can be resolved into four peaks by native gel electrophoresis. These different forms were isolated and tested for their ability to bind to cellsurface EGF receptors. The first processed form was fully capable of binding to EGF receptors, but the second processed form could not. The third form was a collection of small degradation products. We calculated that at steady state about 60% of internalized "lz'I-EGF" was in a form still able to bind to EGF receptors. We then investigated the ability of different reported inhibitors of EGF "degradation" to block the processing of EGF. Although inhibitors of cathepsin B (leupeptin, antipain, Nu-p-tosyl-L-lysine chloromethyl ketone, and chymostatin) were able to inhibit the release of monoiodotyrosine from treated cells in a time-and concentration-dependent manner, they had little effect on the processing step that apparently inactivates "'I-EGF. In contrast, agents that raised intravesicular pH, such as methylamine and monensin, inhibited the initial steps in EGF processing as well as the later steps.
Low temperatures inhibited the transfer of l2'I-EGF to the lysosomes and inhibited the conversion of EGF to a nonbindable form, but had little effect on the initial processing. We conclude that the intracellular processing of EGF is a multistep process that is initiated prior to lysosomal fusion, involves cathepsin B activity, and requires an acidic pH. In addition, many of the protease inhibitors that have been utilized to investigate the role of EGF degradation in mitogenesis do not block the conversion of EGF to a form that is apparently unable to interact with its receptor.
The interaction of epidermal growth factor (EGF') with target cells is a complex process which involves binding, * This research was supported by National Institutes of Health Grants AM-30534, CA-12306, and HL-27148 and in part by a grant from the R. J. Reynolds Foundation. 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.
To whom all correspondence should be addressed. 'The abbreviations used are: EGF, epidermal growth factor; CHAPS, 3-[(3-cholamidopropyl)~methylammonio]-l-propanesulfonate; HF, human foreskin fibroblasts; TLCK, Nu-p-tosyl-L-lysine chloromethyl ketone; Hepes, 4-(2-hydroxyethyl)-l-piperazineethanesulfonic acid; EGTA, ethylere glycol his(@-aminoethyl ether)-N,N,N',N'-tetraacetic acid; PMSF, phenylmethylsulfonyl fluoride. internalization, and degradation of the hormone (1,2), as well as phosphorylation and processing of its cell-surface receptor (3-5). Although studies have shown that EGF binding to cellsurface receptors is necessary for the mitogenic response (6, 7), little is known about the role of these other events in the biological response. For example, it is not clear whether internalized EGF participates in the generation of the mitogenic signal. Some investigators have approached this issue by utilizing various inhibitors of EGF degradation and then determining their effect on EGF-stimulated mitogenesis (8-10). These studies have led to conflicting conclusions regarding the role of intracellular EGF since some inhibitors appear to either facilitate (8) or inhibit (9) some actions of EGF while other inhibitors have no apparent effect (10). In all of these studies, the ability of the inhibitors to block EGF degradation was assessed mainly by their ability to block the release of mono-and di['251]iodotyrosine from treated cells. However, information is lacking about intermediate forms of EGF that might be formed during its intracellular processing, the potential activity of these intermediates, and the effect of inhibitors on their formations. These issues are important in view of our previous findings that during the induction of the mitogenic response, up to 92% of EGF associated with responsive human fibroblasts is in an intracellular compartment (11,121. These considerations prompted us to examine the nature of intracellular '=I-EGF and to evaluate its potential biological activity. For these analyses, we employed native gel electrophoresis which enabled us to identify several processed forms of EGF. We also investigated the ability of previously reported "inhibitors" of EGF degradation to block the different steps in EGF processing. The potential activity of the different forms was evaluated by analyzing their ability to interact with cell-surface EGF receptors. We found that the processing of EGF is initiated prior to delivery of the ligand to the lysosomes and that some agents that block the complete breakdown of EGF to amino acids do not prevent intermediate steps that apparently destroy its ability to bind to the EGF receptor. EGF degradation on the intracellular processing pathway. Each inhibitor is shown next to the step in the processing pathway that is the most sensitive to it. When two steps are almost equally sensitive, then the inhibitors are shown next to both steps. hibition points are those that seem to be the most sensitive to the specified treatments. Monensin is shown above the peak IV to iodotyrosine conversions, since we found in a number of experiments that treatment of cells with low concentrations of monensin induced an accumulation of peak IV EGF (data not shown). Since this effect was somewhat variable, we have placed a question mark next to this step. Nevertheless, it is apparent that the inhibition of the breakdown of lZ5I-EGF into amino acids can occur at a number of different steps in the processing pathway. However, since only peak I and peak I1 forms of EGF appear to retain their ability to bind to the receptor (Fig. 2), inhibiting the breakdown of lZ5I-EGF to amino acids is not equivalent to inhibiting the inactivation of the molecule.

DISCUSSION
After polypeptide hormones bind to the surface of their target cells, they are eventually internalized and degraded (1, 31-34). However, a number of investigators have observed a finite time interval between internalization of polypeptide ligands and initiation of their degradation (14,19,32,35). The events that occur during the intracellular transit of these molecules to the lysosomes are generally unknown as is the role(s) of these events in controlling the biological response of the cells. A number of other investigators have studied the intracellular processing of hormones such as insulin (32, 36) and choriogonadotropin (33,37) in an effort to clarify these questions. However, an understanding of the role of intracellular ligands in hormone action requires a quantitative description of both their distribution and their "activity." It is the intracellular activity of a ligand that is the most difficult to determine since it depends on the presence of the appropriate receptor system, a bindable ligand, and environmental conditions appropriate for hormone-receptor interactions.
In the present study, we have documented the intracellular processing of '"I-EGF and have also measured the effect of this processing on the ability of EGF to bind to its receptor. Subsequent to the completion of this study, a report appeared documenting the intracellular processing of EGF in 39). That report showed that the initial processing of lZ5I-EGF involves the removal of several amino acid residues from the carboxyl terminus of the molecule, but it did not address the effect that the processing had on ligandreceptor interactions (39). The results of our study in general support the findings of Planck et al. (39) in that the migration position of the processed lZ5I-EGF on native gels is very similar to that reported for EGF in which up to 5 residues have been removed from the carboxyl terminu~.~ In addition, we have analyzed the molecular weights of the different processed forms of lZ5I-EGF using high performance liquid H. S. Wiley and D. J. Knauer, unpublished observations. chromatography under reducing conditions and in the presence of 6 M guanidine hydrochloride (results not shown). Those results indicate that peak I1 and peak I11 are virtually indistinguishable from native EGF with respect to molecular weight while peak IV EGF is a collection of small peptides. This is consistent with the initial processing involving a very small alteration in the number of amino acid residues of the protein as would occur if only a few amino acids were removed from the carboxyl terminus. However, the exact nature of the processing which occurs after EGF internalization in human fibroblasts can be established only after sufficient amounts of material for direct protein analysis are isolated. Such studies are currently in progress.
The initial step in the processing of "'I-EGF that we observed apparently occurs after binding and internalization since the addition of lZ5I-EGF directly to solubilized cells did not result in any alteration in the molecule (Fig. 1). In addition, when we bound "'I-EGF to the surface of cells at 0 "C and then homogenized and fractionated the cells on Percoll gradients, all of the label sedimenting with the plasma membranes was intact lZ5I-EGF (data not shown). We found that initial processing of the "'1-EGF occurred within 5 min after entry into the cell (Fig. 1). This is prior to the time that internalized EGF has been observed to enter the lysosomal compartment (19,40). Indeed, when we blocked the transfer of EGF to the lysosomes by lowering the incubation temperature, the initial processing of the EGF to peak I1 material was unaffected (Fig. 8). However, the conversion of peak I into peak I1 was pH-sensitive, as would be expected for a lysosomal process. It has been reported that there is a rapid acidification of endocytic vesicles (41). If there were pHsensitive proteases or processing enzymes localized in endocytic vesicles, then the acidification of these vesicles might activate them. Alternatively, monensin and methylamine may act by inducing vesicular swelling (24, 25) which could then inhibit the transfer of "'I-EGF to vesicles where processing occurs.
In contrast to the initial processing of lZ5I-EGF, the conversion of peak I1 material to peak I11 and peak IV material had all the properties of a lysosomal process. There was a delay between the initial entry of EGF into cells and the initiation of degradation which is similar to the delay that has been observed between the endocytosis of EGF and the transfer to the lysosomes (19,40). Inhibitors of the lysosomal protease cathepsin B (ie. leupeptin) inhibited the later steps in EGF processing (Fig. 5). These later steps were also more sensitive to agents that raised intravesicular pH than the initial processing steps (Fig. 5). Finally, blocking the transfer of lZ5I-EGF to the lysosomes by lowering the incubation temperature of the cells also blocked the later steps in processing. Thus, it seems that the initial steps in EGF processing occur very shortly after internalization while the subsequent steps occur after the transfer of EGF to the lysosomes.
One of the aims of this study was to determine if internalized EGF was capable of occupying its receptor. We found that the processing of peak I to peak I1 EGF did not significantly alter its ability to bind to the EGF receptor (Fig. 3). In contrast, we found no measurable binding of peak I11 material to EGF receptors. The peak IV material could not be tested for its ability to bind to the cell surface since it was lost from the dialysis tubing. However, it is doubtful that it would bind in view of the inability of the larger peak I11 material to bind. From the present results, we can calculate that at steady state approximately 67% of the total radioactivity associated with human foreskin fibroblast cells is present in a form that can bind to the EGF receptor. Of this total, 13% is associated with the surface of the cells, with the remainder in an internal compartment. We can thus conclude that a large majority of the EGF that is associated with responsive cells at steady state is in an intracellular compartment and is capable of binding to the EGF receptor.
It may be significant that the processing of EGF to a form incapable of interacting with its receptor occurs only after the transfer of the ligand to the lysosomes. It has been reported that the rate at which EGF is transferred to the lysosomes in human fibroblasts is significantly different from that observed for other internalized ligands (29). Thus, the termination of the ligand-receptor complex may be regulated by the rate of lysosomal transfer and not by the rate of its internalization. Another candidate for terminating the EGF-receptor complex is an acidification of the intracellular vesicle (41) which could cause dissociation of EGF from the receptor (15). Recently, we determined the minimum concentration of EGF inside endocytic vesicles by simultaneously measuring the internalization of '251-EGF and the internalization of '251-polyvinylpyrrolidone (42). These studies showed that the concentration of EGF in vesicles is at least 10 pg/ml when the extracellular concentration is 1 ng/ml, a 10,000-fold increase. In view of the extremely high concentrations of hormone present in intracellular vesicles, a combination of the two mechanisms may be required. Thus, the acidification of the vesicle could cause dissociation of the EGF and activation of proteases.
Cleavage of EGF could then prevent reoccupancy of the receptor.
An unexpected finding of this study was that many agents that have been previously employed to inhibit the intracellular breakdown of EGF do not prevent intracellular processing of the molecule. As a single example, leupeptin has been used previously to determine the role of intracellular degradation of EGF in the hormonal response (10). Those studies found no significant effect of leupeptin on the cellular response to the hormone, although a large intracellular accumulation of radiolabeled ligand was observed. Our results are entirely consistent with these findings in that leupeptin causes an accumulation of internalized radioactivity in treated cells (Fig. 4). However, our analysis of the effect of leupeptin on the intracellular processing of EGF indicates that the inhibitor is much more effective at preventing the breakdown of peak I11 form of EGF than it is at preventing the conversion of peak I1 to peak I11 forms (Fig. 5). Since we found that the peak I11 form of EGF is essentially devoid of binding activity, these results indicate that leupeptin is not efficient in preventing the intracellular conversion of EGF into a form that lacks the ability to occupy its receptor. Thus, one cannot necessarily equate the ability of agents to prevent either the release of m~no['~~I]iodotyrosine or the conversion of the molecule to a low molecular weight form with their ability to prevent breakdown of internalized ligands.
Several important questions remain regarding intracellular EGF. Is the intact intracellular EGF still associated with its receptor and, if so, is this complex important in the response of cells to the hormone? Since EGF and its receptor are internalized together (5, 19), it seems likely that receptor occupancy could persist after internalization. Indeed, studies on the intracellular route of EGF tagged with ferritin indicate that the EGF-receptor complex persists for a significant length of time (19). Thus, these internal hormone-receptor complexes could potentially give rise to certain cellular re-sponses. However, we have recently demonstrated that at least some of the cellular responses to EGF are triggered at the cell surface (42), as has also been demonstrated in the case of human choriogonadotropin (43) and thrombin (44). Nevertheless, EGF triggers a wide variety of cellular responses that differ in their timing and magnitude, and it is possible that the signal for some of these might be generated intracellular. This seems possible in view of the finding that the vast majority of intact EGF that is associated with human foreskin fibroblast cells during the induction of mitogenesis is intracellular (11, 12). Although this study does not prove that intracellular EGF is still associated with its receptor and generating a signal, our results do demonstrate that such a possibility is worth serious consideration.   quantitative.

TO determine %state of intracellular 1251-~P. we incubated HP cells Continuously with I-EGP, and analyzed Cel1-assOCiated radioactivity at surface bound EGP with acetic acid. l T p results of one of these experiments indicated time intervals by native gel electrophoresis after removing cell
is shorn in Piguro 1. The input I-EGP migrated as a single peak in our electmphpretic system. when the cells were incubated with the hormone for 5 .in at 37 C, the recovered intracellular radioaCtivif$5was distributed between two peaks: a major peak that eo-migrated with input I-EGF and a second peak hat migrated slightly ahead of it. when the i n t r a c e l w a r radioactivity was ililarlv analyzed after a I or 5 h incubation with I -E G P there were at .+asL four pebks of radioactivity. (Fig  1). For convenience w e have named ne peaks I, 11, 111 and Iv in order of increasing mobility ' the gel system. A a k I Corresponded to the migration position Of .the input '"1-EGP while peak IV corresponded to the dye front and thus could contain a number Of unreeolved :omwnsnts.

I-EGF.
If we assume that a11 Of the peaks represent I-EGP or stoichiometric Cleava roducts then we can calculate the relative molar distribution Of th.9API-monoibdoryro.ins-conraining ratios Of peaks I, 11, I11 and IV are 0.20:0.37:0.29:0.14. fragment-. With this assumption w e calculate that a t steady s t a r e the molar /i

Figure 1. T h e e l e c t r o p h o r e t i c p a t t e r n o f 1 2 5 1 -f f I p s h i f t s a f t e r internalization.
HP cells r e r e exposed to 2.5 XlfQ-g M I-EGP at 37OC for the lengths Of ti-indicated in each panel. m y I-EGP associated with the cell surface was removed (151 and the cells were then diBso1ved and subjected to native gel electrophoresis as described in Experimental PrOCedures. The gels were then sliced and the radiactivity quantitated. The sample run on the control q& (top painel) was obtained by dissolving a set Of cells and then addina I-EGP directlv to the solution.
In the bottom .anel are the d&ign;tione assigned to s&h peak.
The ~resence of these cell orocessed forma of EGP raised the question about their ability to bind to the EGP receptor. Since EGP and its~receptor are apparently internalized together and appear to remain associated for soma time (19). it was poseible that the observed processing could be a mechanism to tsrrinate EGP-receptor interaction and possib$% the hormone aignol. TO measure the binding of these processed form. of 51-EGP. we isolated them by oremrative a d electrwhoresis. Since Deaks I and I1 migrated v e r y close cells with I-EGP for two hours and then with unlabeled EGP for 20 mi".

topither, yi5sought ti sliminate most if the former peak-by incubating the T h i s f a c i l i t a t e d t h e c~n v e r s i o n O f p e a k I t o p e a k I1 (unpublished observations).
After elution of the Deak8 from the DreDaratiYe gel, the fractions comprising the individual psbks were pocl.d,-Foncentratsd~and then dialyzed against binding medium. The rmcovery Of material from peaks I1 and 111 was nearly quantitative (greater than 951). However, greater than 981 Of the material in peak IV was lost during dialyais indicating that its molecular m i g h t uaa less than the 3500 nr Cutoff of the dialysis tubing. When *ample* electrophoresis, the radioactivity was observed to migrate in the appropriate of the purified peak I1 and I11 material. were examined by analytical gel positions on the gel as single bands. we did not detect any contamination of the peak I1 material with intact EGP. indicating that we effeCtiYely resolved t h e w two caponents. of k for the peak I1 material was somewhat higher than the inputmI-EGP.

u-d to be identical to that of the original I-EGP.
B~w e v e r , value wbiolf could indicate a lowered affinity Of the material for the EGP receptor.
On the other hand. the peak I11 material had no -.surable ability to bind to the S F r.EeptOr at ligand concentration we Used, indicating that this processed form of '*I-EGF has lost its ability to transmit a signal through the EGP receptor. (-*-I, the peak I1 material (-+I or the peak I11 material (-A-) to bind to tho cell surface. The lines drawn through the data points were Calculated using ttie nonlinear curve-fitting EompUter program described in the text. The cal$,ul_afed V-~luss for the p q n d 0:p-r yqgciation rate constants were 9 . 5 x 10 II sec and calculatpp dilpociation rate conrrtants were 7.9 X 1 0 ' soc for "I-EGP and 1.0 X 10 I4 aec for I-EGP and peak I1 material pspe_'ltivelp hile the 1.6 X 10 Bec for peak I1 material. The amount of binding Observed with the peak I11 material was not significantly above the amount Of non-specific binding Observed (appmqfmately 250 cpm). The concentration of ligand used in a11 cases vas 3.3 x 10 n.

The above results indicate that not a11 Of the internalized 1251-EGP is other inVestigatOCB in which inhibitors of BOP 'degradation' have been used to intact. This finding is particularly significant in the Context of studies by probe the rois of intracellular BOP in the q p r + t i o n of cellular resp~nses. Since those studies used the release Of mono1
Illodotyrosins from cells as did not CreYent Droceesing of EGP into non-bindabls forms. TO clarify the criterion for EGP degradation, it is possible that the inhibitors utilized questions-raised by previous studies on the role of intracellular EGP and tb g a i n additional insight into the mechanisms of EGP proceasinq, we examined a number Of previously reported inhibitors Of BOP 'degradation.. To assess the role Of 1~808omal enzmes in EGP degradation, we treated HI cells with a variety Of pblypeptids inhibitors that-affect l y s o s o~l enzymes (leupeptin, antipain, pepstatin. chylostatin and aprotinin; 2 0 ) . As shown in Pig. 3, leupaptin was the most effective in inhibiting tpq5release of antipsin and Chymostatin were also affective. All of these inhibitore have radioactive degradation products from cells incubated with I-EGP, but been reported to inhibit capthesin B activity (20.21). Significantly. their relative effectiveness in inhibiting capthesin B activity (leupeptin > antipain > c h w s t a t i n l is the s-as their relative ability to Inhibit the release of EGP dearadation Droducts from treated cells 120). In Contrast.

Laupeptin
BEcent ligand as concentration I Iliodotyrosines
One of the surprising effects of monensin 1 . 1 1 its apparent ability to prevent the conversion of peak I BOP to peak I 1 EGP as well as the ather steps. This indioatea that a11 of the steps in the intracellular prOCe8sing of EGP are dependent upon an acidic intraresicular pH.
BOweveCr the venl rapid time course Of appearance of the peak I1 EGP (lese than 5 Din. Fig. 1) and the relative inability Of the lysos0Dal pr0tea.e inhibitor leupeptin t0 block this step in EGP proceseing [Pig. 41, indicate that this step Occurs prior to the delivery Of the ligand to the lysosomes. The processing doen not seem to occur at the cell surface since the initial proceaeing step was blocked when internalization was inhibited by phenylarsine oxide (results not sholnl. Thus it seemed possible that an acidic enviro-nt in the endmytic vesicle itself was renpon-ible for the initial proceseing of ffiP.
If thie is the case. then blockina the transfer Of EGP to the lvsosombn should have little effect on the iniii.1 processing step, but should-strongly inhibit the later step..

It han been reported tlat lowering the incubation t-cature
Of cells determine the temperature at which lysomomdl tranafer was blocked in our strain of HPlqglla, we .ruined the effect of incubation temperatura OD the transfer of I-EGF to lysos-s using mr'co11 gradient fractionation. Cells were incubated with labeled ligand for 5 Din at different temperatures and then chased with unlabeled ligand for am additional hour. They "era than homogenized and fractionated OD Perco11 gradients. Re*Ultm&om this experiment are presented in Pig. 6

any of the prgoeesing steps, r e incubated cells with I-EGP for 5 minuter at
To determine if blocking the transfer Of e C P to f& lysos-~ would block 37, 27 and 20 C and then Chased with unlabeled EGP for an additional hour at the same temperatures. The internalized ligand 1 0 s then extracted and and asaayed for tho different processed f o q These data are n h a n in Pig. 7 and appearance O f the peak I l l and peak Iv t o m e of the molecule. However. the indicate that blocking the transfer Of 51-KGP to the lyaosomea prevents the processing Of tho EGP into the Peak I1 form was unaffected. indicating that the initial processing of the ligand occurs prior to ita delivery to the 1YLlOS".  Table 111. These results indicate that methylamine is the most effective inhibitor of the peak I to pask I1 conversion while lower temperatures are the moat effective inhibitors of the peak I1 to peak 111 conversion. Both laupeptin and TICK were the most effective inhibitors of the peak Ill to Peak IV conversion.

TABLE I11
Effect of different inhibitors on the processing of 1 2 5 1 -~P Cells were treated with indicated concentration Of inhibitor for 2 hrg prior to the ezperirnt. '"I-ffiP was added at a concentration of 2.5 x 10-M for 5 min with the inhibitors and then replaced with medium containing the name concentration of inhibitors and unlabeled BOP. After an additional hour incubation, cells we=-. rinsed and dissolved in detergent and fractionated by native gel electrophoresis a* described in Materials and Methods. The gels were sliced and the relative amounts of radioactivity Coligcating with the different proceseed f o M of EGP 1 . 1 1 determined.