Intracellular Processing of Epidermal Growth Factor THE COOH-TERMINAL OF '251-EPIDERMAL GROWTH FACTOR*

products by reverse-phase high performance liquid chromatography as described for the preparation of a-EGF. Under the milder digestion conditions, 2 new peaks were detected by absorp- tion of 280 nm light. One of the peaks was shown to be the COOH-terminal pentapeptide of EGF by digestion with aminopeptidase and analysis of the released amino acids after precolumn derivatization with o-phthalaldehyde (11). The protein comprising the second peak was the remaining 48 amino acids of EGF, EGF14; analysis of the amino acids released by carboxypeptidase A (Sigma) and carboxypep- tidase B was consistent only with cleavage adjacent to the arginine moiety at position 48. Similarly, Savage et al. (6) reported that mild trypsin digestion of undenatured EGF selectively yielded EGF14. Isoelectric focusing analysis showed that EGF14 had a PI of 4.35. Longer times of trypsin digestion of EGF resulted in a third protein peak eluting from the reverse-phase column. The protein in this third peak had a PI of 4.3 and COOH-terminal amino acids consistent with the protein being EGFI45. Both '=I-EGF14 and '251-EGF145 were purified by preparative isoelectric focusing for use as standards in this study. The identities of the minor trypsin fragments with PI values 4 were not determined. These fragments may have resulted from slight oxidation or trace amounts of another proteolytic activity.

tions such as by phosphorylation OT limited proteolysis. The purpose of this investigation is to determine the precise nature of the EGF modification reactions.
We employed protein mapping techniques with proteolytic enzymes of known specificity in order to localize regions on the EGF molecule which are modified by intracellular processes. Since such studies clearly require a very pure ligand, we used a-EGF that had been purified to near homogeneity with reverse-phase high performance liquid chromatography as the final purification step (4). After an iodination reaction, the labeled EGF was purified by isoelectric focusing in agarose gels (5). Only the T -E G F species with a PI of 4.55, the same as unreacted a-EGF, was used in these studies. Interpretation of our data has been aided by the amino acid sequence of mouse EGF originally published by Savage et al. (6) (Table I) and recently confirmed at the genetic level by Gray et al. (7).
We found that upon binding and internalization into Rat-I fibrobIasts, Iz5I-EGF underwent a series of proteolytic cleavages near the COOH-terminal end of the protein.
The most abundant EGF species, a-EGF, was labeled with using chloramine-T (10). The iodinated form of EGF with an isoelectric point of 4.55, corresponding to the PI of unlabeled a-EGF, was purified by preparative isoelectric focusing on agarose gels as previously described (5). The specific activity of the '%I-EGF was approximately 500 mCi/mg. Isoelectric Focusing-Isoelectric focusing was performed with horizontal 1% agarose gels as previousIy described (2,5). The pH range of the gels was approximately from 3.0 to 6.0. The isoelectric points of the various '=I-EGF species were determined from standard curves generated with the aid of pI marker proteins (Federal Marine Colloids, Inc.) detected by Coomassie blue R250 staining. The location of the '*'I-protein was determined by autoradiography of analytical gels after drying and of preparative gels without drying (2).
Purification of internalized 12'i-EGF Species-Ten 10-cm culture dishes of confluent Rat-1 cells (1 X 10' cells/dish) were rinsed with Dulbecco's modified Eagle's medium. To each dish, 4 ml of binding media containing 3.4 X lo6 cpm/ml of '=I-EGF and 1OmM methytamine (to permit accumulation of internalized 'Y-EGF products) were added, and the cultures were incubated at 37 "C for 3 h. The dishes were rinsed 6 times in cold Hank's balanced salt solution plus 1 mg/ ml of bovine serum albumin and the bound EGF was extracted from the cells with 5 ml/dish of cold 0.05 M HCI for 18 h. The dishes were rinsed with 5 ml of cold 0.05 M HCl, and the acid-soluble proteins were combined, extensively dialyzed against cold water, and lyophilized. The various internalized lz6I-EGF species were separated by preparative isoelectric focusing as described for the purification of    T o determine the identity of the major tryptic fragments of '"I-EGF, we first digested EGF in the absence of bovine serum albumin and separated the products by reverse-phase high performance liquid chromatography as described for the preparation of a-EGF. Under the milder digestion conditions, 2 new peaks were detected by absorption of 280 nm light. One of the peaks was shown to be the COOHterminal pentapeptide of EGF by digestion with aminopeptidase and analysis of the released amino acids after precolumn derivatization with o-phthalaldehyde (11). The protein comprising the second peak was the remaining 48 amino acids of EGF, EGF14; analysis of the amino acids released by carboxypeptidase A (Sigma) and carboxypeptidase B was consistent only with cleavage adjacent to the arginine moiety a t position 48. Similarly, Savage et al. (6) reported that mild trypsin digestion of undenatured EGF selectively yielded EGF14. Isoelectric focusing analysis showed that EGF14 had a PI of 4.35. Longer times of trypsin digestion of EGF resulted in a third protein peak eluting from the reverse-phase column. The protein in this third peak had a PI of 4.3 and COOH-terminal amino acids consistent with the protein being EGFI45. Both '=I-EGF14 and '251-EGF145 were purified by preparative isoelectric focusing for use as standards in this study.
The identities of the minor trypsin fragments with PI values below 4 were not determined. These fragments may have resulted from slight oxidation or trace amounts of another proteolytic activity.
Fractionation of pI 4.35 Species by Reverse Phase Chromtography-The PI 4.35 species was bound to a Beckman ultrapore RPSC (C-3) column (4.6 X 75 mm) equilibrated in acetonitrile/20 mM trifluoroacetic acid, 15:85. The bound protein was eluted by raising the acetonitrile proportion by l%/min for 30 min. The flow rate was 1 ml/min, and I-min fractions were collected. The ' "I content of each fraction was measured in a y counter. The samples in fractions 5 to 9 and 10 to 14 were pooled to form the PI 4.35a and PI 4.35b preparations, respectively.

RESULTS
A typical isoelectric point distribution of intracellular '9-EGF species 40 min after internalization is shown, along with a sample of the original 12'I-EGF, in the isoelectric focusing gel pictured in Fig. 1A. Clearly, the '"1-EGF has been converted to a few discrete species with more acidic isoelectric points. T o investigate the nature of the changes we first separated the major intracellular lZ5I-EGF species by preparative isoelectric focusing. The most abundant iodinated species, with PI values of 4.55, 4.35, 4.2, and 4.0, were obtained with very little cross-contamination (Fig. 1B).
The cell-associated "'I-species which focused at pH 4.55 had the same PI as the starting lZ5I-EGF. The PI 4.55 species  (Fig. 2). Apparently, both proteins had COOH-terminal arginine moieties; EGF contains no lysine moieties. Likewise, the 12SI-products resulting from either mild or extensive trypsin digestion of the two proteins were indistinguishable by isoelectric focusing. Thus, the PI 4.55 species appears to be unmodified '2sII-EGF. The PI 4.2 species appeared on the cell surface within 10 min after binding of '"I-EGF (3). The autoradiograph pictured in Fig. 3 shows that the PI 4.2 species had the same isoelectric point as 12'I-EGF after treatment with carboxypeptidase B to remove the COOH-terminal arginine moiety.
Furthermore, the PI 4.2 species was insensitive to treatment with carboxypeptidase B. Thus, the PI 4.2 species appears to have been formed by cleavage of the COOH-terminal arginine moiety (see Table I for amino acid sequence). The penultimate amino acid, leucine, may also have been removed. Since the leucine side chain is uncharged, its presence or absence cannot be detected with isoelectric focusing gels. The major product from mild digestion of "'1-EGF with Staphylococcus uureus V8 protease, which should release the COOH-terminal dipeptide, is indistinguishable on the isoelectric focusing gels from the carboxypeptidase B product (data not shown). In contrast, the glutamate moiety a t position 51 must not have been removed, since this would have appreciably raised the pI of the protein. A comparison of the tryptic peptides generated from the pI 4.2 species and '?-EGF suggests that only the COOH-terminal portion of EGF was altered in the formation of the p1 4.2 species; the labeled tryptic peptides were identical.
The second modified ""I-EGF species to appear in cells had a p1 of 4.35 and was found in vesicles with the same buoyant density as clathrin-coated vesicles, Golgi, and endoplasmic reticulum (3). Upon digestion of the p1 4.35 species with carboxypeptidase B, under conditions where '*"I-EGF reacted quantitatively, most the labeled protein underwent a p1 shift indicative of the loss of the COOH-terminal arginine moiety, but a portion of the labeled protein was resistant to the treatment (Fig. 4). The resistant portion of the ~14.35 species did not change p1 even after prolonged incubation in the presence of carboxypeptidase B. Therefore, the ~14.35 species as resolved by isoelectric focusing must be heterogeneous.
The two forms of the ~14.35 species were partially separated by reverse-phase chromatography on a Beckman RPSC (C-3) column and termed pI4.35a and pI4.35b species, respectively. The equality of the isoelectric points of the two species is evident in Fig. 5. When the two p1 4.35 preparations were incubated with carboxypeptidase B, about half of the labeled protein in the pI 4.35a sample shifted p1 while essentially none of the p1 4.35b species was altered. Thus, the p1 4.35a species had a COOH-terminal arginine moiety and was contaminated with the p1 4.35b species which did not terminate in arginine. lz51-EGF characteristically forms broad peaks upon reverse-phase chromatography' so we did not try at this time to improve resolution of the two species.
The p1 4.35 species had the same isoelectric point as a trypsin fragment consisting of amino acids 1 to 48 of lz51-EGF, termed 1251-EGF,-,, (Fig. 5). The pI4.35a species underwent the same pI shift as '251-EGFl-~ upon digestion with carboxypeptidase B. Thus, it appears that the pI4.35a species and EGF,-,, are identical. The p1 4.35b species may have resulted from the incomplete removal of the tryptophan moieties at positions 49 and 50. In support of this interpretation is the protein profile from the RPSC column indicating that the 4.35b species was more hydrophobic than the p1 4.35a species. Furthermore EGFr-48 was the principal product formed from EGF by very mild trypsinization conditions (Figs. 2 and 3) and both ~14.35 species were resistant to mild trypsin digestion (Fig. 5). More vigorous trypsinization conditions generated the same labeled peptides from the 12'1-EGFieaa and the p1 4.35 species.
A pulse-labeling experiment, similar to those described in the accompanying paper (3), indicated that the percentage of intracelluar "'1 associated with the p1 4.35a species reached a peak between 20 and 40 min after a 5-min labeling period with '*"I-EGF. In contrast, the percentage of '*'I associated with the pI4.35b species remained constant for 4 h. Therefore, the p1 4.35a species and not the p1 4.35b species appears to be an intermediate in the processing scheme.
The ~14.0 species is the last to appear and the longest lived of the major intracellular '*"I-EGF species. It is found in dense, lysosomal-like organelles ( the same isoelectric point as aliquots of the PI 4.35a species that had been digested with carboxypeptidase B to remove the COOH-terminal arginine moiety (Fig. 6). Thus, the PI 4.0 species could have been formed in the cell by removal of the COOH-terminal arginine or arginine-leucine moieties of the PI 4.35a species. Accordingly, the data in Fig. 6 indicate that the PI 4.0 species did not have COOH-terminal arginine. If the PI 4.0 species consisted of EGF amino acids 1 to 46 or 47, extensive trypsin digestion might be expected to generate a PI 4.3 species that we did not observe. However, the rate of hydrolysis by trypsin between positions 45 and 46 of the PI 4.0 species may have been greatly reduced by the acidic aspartate moiety adjacent to the arginine moiety and by the proximity of the arginine moiety to the COOH terminus.

DISCUSSION
We have previously reported that 1251-EGF is chemically altered as it traverses through recipient cells and that these alterations can be detected by isoelectric focusing (2, 3). It was not feasible to purify adequate quantities of the individual internalized "'I-EGF species for direct amino acid sequencing. However, we were able to localize the regions of change with peptide mapping techniques. We found that the "'I-EGF was sequentially cleaved near its COOH-terminal end to generate the altered species. The points of cleavage are indicated with the amino acid sequence of mouse EGF in Table I  Intracellular Cleo few minutes after '251-EGF binds to cell surface receptors, the PI 4.2 species was formed by removal of 1 or 2 amino acid residues. This first cleavage occurred at least in part at the cell surface but also may have occurred after internalization (3). As the PI 4.2 species was transported through the cells in vesicles, an lZ5I-EGF form with a PI of 4.35 was produced by cleavage adjacent to the arginine moiety at position 48. The final major intracellular species, which is located in lysosomallike organelles (3), had a PI of 4.0 and had lost at least the arginine moiety at position 48. The PI values described for each species agree well with the PI values calculated based on the presumed amino acid composition and textbook values for the pK values of the charged groups. We cannot eliminate the possibility of additional modifications within lZ51-EGF that do not appreciably change the isoelectric point. Both EGF and the EGF receptor appear to be cleaved upon binding of EGF at the cell surface (13, 14); EGF is cleaved more after internalization and the receptor also may be further cleaved. Any of the resulting fragments could conceivably serve as a second messenger to convey a signal within the cell. Another possibility, not necessarily exclusive from the latter, is that some of the biological responses could result from the activation of a tyrosine protein kinase activity upon binding of EGF to its receptor protein (12). Because removal of the COOH-terminal pentapeptide of EGF reduces the affinity of the EGF for its receptor (3,15), this cleavage may cause EGF to separate from the receptor protein and thereby inactivate the receptor protein kinase activity. The freed EGF receptor may then migrate to another site within the cell or be degraded.
During the course of EGF binding and internalization, a wage of "'1-EGF 3057 diversity of signals may be generated to induce early and late cellular responses, including the modulation of de ~O U O EGF receptor synthesis. Coordination of these responses may result from intracellular signals produced as a result of proteolytic cleavage of EGF and its receptor. Further experimentation is necessary to resolve this issue.