Proc. Natl. Acad. Sci. USA Vol. 79, pp. 2260-2263, April 1982 Biochemistry

Cleavage of honeybee prepromelittin by an endoprotease from rat liver microsomes: Identification of intact signal (precursor processing//membrane /melittin) CHRISTA MOLLAY, ULRIKE VILAS, AND GUNTHER KREIL Institute for Molecular Biology, Austrian Academy of Sciences, Billrothstrasse 11, A5020 Salzburg, Austria Communicated by Paul D. Boyer, January 19, 1982

ABSTRACT It has previously been shown that rat liver mi- tivities, on a sucrose-deoxycholate gradient. This endoprotease crosomes contain a proteolytic that cleaves honeybee pre- cleaves prepromelittin at a single peptide bond, the pre- promelittin to yield promelittin. This enzyme has now been fur- pro junction. The-present communication describes the isola- ther purified by centrifugation on a sucrose-deoxycholate gradient tion ofthe second reaction product, intact signal peptide, from and then reconstituted into phospholipid vesicles. Incubation of such in vitro assays. prepromelittin with vesicles in the presence of melittin yields, in addition to promelittin, a hydrophobic peptide. The latter could be isolated by extraction with 1-butanol and paper electrophoresis MATERIALS AND METHODS in 30% formic and was shown to be intact signal peptide by analysis of peptic fragments and automated Edman degradation. Tritium-labeled amino ofthe highest specific activity cur- The microsomal enzyme is thus an endoprotease that hydrolyzes rently available were purchased from the Radiochemical Centre prepromelittin exclusively at the pre-pro junction. The precision (Amersham, England). Phospholipids were obtained from Lipid of this cleavage of an insect preprotein by a rat liver enzyme in- Products (Nutfield Nurseries, England), synthetic dipeptides dicates that we are dealing with the ubiquitous eukaryotic signal were from Bachem AG (Bubendorf, Switzerland), and other peptidase. chemicals were from standard suppliers. Phospholipase-free melittin was prepared from lyophilized bee venom (Firma The precursors of secreted polypeptides contain an amino-ter- Mack, Illertissen, Federal Republic of Germany) as described minal signal sequence that is not present in the final product (14). The fragment melittin1_19 was isolated by limited (1, 2). In eukaryotic cells, cleavage of signal is a co- of melittin with (15). translational reaction that takes place early in the assembly of Preparation of Prepromelittin. Total RNA from queen b -e the polypeptide chain (3, 4). The that catalyzes this venom glands was translated in a cell-free system from wheat reaction, tentatively called signal peptidase, is neither tissue germ in the presence of one labeled and 19 unlabeled amino nor species specific, and correct processing of different pre- acids as described (16). Aliquots of these incubation mixtures in vitro by, e.g., dog pancreas microsomes (5) or in vivo were used without further fractionation for the cleavage assays. in oocytes ofXenopus laevis (6) has been shown. Several groups Partial Purification of Signal Peptidasei. Rat liver rough of workers have, moreover, described the posttranslational microsomes were prepared according to Sandberg et aL (17), cleavage of signal peptides in the presence of microsomes and suspended in 1 M KCV/0.02% deoxycholate, and again collected detergents (7-10). These studies showed that signal peptidase by centrifugation. They were then dissolved in 0.4% deoxy- is an intracellular membrane-bound enzyme that, interestingly, cholate and layered onto a step gradient of 2-0.2 M sucrose/ is not inactivated by any ofthe typical protease inhibitors (11). 0.2% deoxycholate as described by Kaderbhai and Freedman In all these-experiments, only one of the reaction products, the (18). After centrifugation for 17 hr at 80,000 X g in a Beckman secreted polypeptide, could be detected, so that the question SW 41 rotor, the fraction present in the 1.4 M sucrose layer was of whether signal peptides are cleaved en bloc or in fragments collected. This fraction was added to dry lipid (phosphatidyl- remained unanswered. Evidence has, however, been pre- choline/phosphatidylethanolamine, 2.5:1; 14 mg ofphospho- sented that in vivo signal peptides, ifthey ever exist as separate lipid/1 mg of protein) and dialyzed for 4Whr against 0.1 M entities, must be extremely short-lived (12). Tris'HCVl1 mM o-phenanthroline, pH 8.0. The reconstituted We have investigated the cleavage ofhoneybee prepromelit- vesicles were collected by centrifugation. tin by rat liver microsomes and have shown that, in the presence Assay for Signal Peptidase. Labeled prepromelittin was in- ofdeoxycholate, promelittin with the correct NH2 terminus can cubated for 2 hr at 250C in dialysis buffer with reconstituted be generated in a slow reaction (8, 13). As reported briefly, an membranes containing 20-50 Ag ofprotein and melittin (1 moV 10-fold purification of a protease could be achieved after so- 50 mol ofphospholipid). In control samples, vesicles or melittin lubilization ofmicrosomal membraneswith detergents and sub- were omitted. The samples were then adjusted to pH 8.5 with sequent reassembly of membrane proteins with added phos- dilute NH3 and extracted three times with an equal volume of pholipids bydialysis (13). Even in these reconstituted membranes, 1-butanol. Under these conditions, promelittin remains in the this enzyme, which we consider to be the signal peptidase, is aqueous layer while uncleaved prepromelittin accumulates at acryptic enzyme requiringfor activity the presence ofdetergent the interface (16). or, as we have found, of melittin. Product Analysis. The purification of promelittin and its Signal peptidase from rough microsomes ofrat liver has now characterization by enzymatic hydrolysis and stepwise Edman .been partially purified, apparently free ofother proteolytic ac- degradation of the fragment derived from its amino end have been described (8). Essentially the same methods can be used The publication costs ofthis article were defrayed in part by page charge for analysis of prepromelittin (19). payment. This article must therefore be hereby marked "advertise- The butanol layer was applied to Whatman 3MM paper and ment" in accordance with 18 U. S. C. ยง1734 solely to indicate this fact. purified by high-voltage paper electrophoresis in 30% formic 2260 Downloaded by guest on October 2, 2021 Biochemistry: Mollay et al. Proc. Natl. Acad. Sci. USA 79 (1982) 2261

E DNP Lys Glu Leu

20 A2 Al

2

10

I

C. 0 2 4 6 8 Or CUL 0. 0 2 4 6 8 10 12 14 16 18 20 22 "0 cd

1

4 2 0 2 4 0 2 4 6 8 10 12 14 16 18 20 22 Distance from origin, cm

FIG. 1. Analysis of the butanol extract from an experiment with prepromelittin labeled with [3H]alanine. (A) High-voltage paper electrophoresis in 30% formic acid. The butanol extracts from a sample ( ) and a control incubated without melittin (----) were applied to Whatman 3MM paper and fractionated at 30 V/cm for 2 hr. The dried paper was then cut into sections and the radioactivity present in each was determined in a liquid scintillation counter (toluene-based scintillator). (B) Electrophoretic separation of the peptic digest. Radioactive material present at the origin of A was hydrolyzed with pepsin overnight, and the digest (- ) was then fractionated by paper electrophoresis at pH 4.8. ----, Control. (C) Electro- phoretic separation of peptic fragments at pH 2.0. Neutral peptides (-) from the electrophoresis at pH 4.8 (B) were recovered by elution and rerun at lower pH. Two peptides, designated Al and A2, were present in about equal amounts. ----, Control. (D) Electrophoresis of peptides Al (-) and A2 (-) after Edman degradation. The two fragments were recovered by elution, shortened by one residue by Edman degradation, and again sep- arated by paper electrophoresis at pH 2.0 with synthetic dipeptides as references. MEL (1-19), melittin-119; E-DNP-Lys, E-dinitrophenyllysine.

acid. The electropherogram was cut into sections, and the ra- Automated Edman Degradation. The cleaved signal peptide dioactivity present on each was determined in a liquid scintil- could also be partially purified by applying the butanol layer lation counter. The section corresponding to the origin was sub- to Whatman 3MM paper and then eluting with water to remove sequently soaked in dilute HCl (pH 1.8) containing 0.7-1 mg soluble radioactive contaminants. Signal peptide could subse- of pepsin/ml and incubated overnight at 30'C. The digest was quently be recovered by elution in sufficient yield with the sol- then dried and examined for the presence of peptic fragments vent mixture used for paper chromatography. The dried ma- ofthe signal peptide ofprepromelittin. This involved separation terial was dissolved in 0.4 ml of formic acid and subjected to by high-voltage paper electrophoresis at pH 2.0 (10% acetic acid automated Edman degradation in the presence of 3 mg of the adjusted with formic acid) or at pH 4.8 (1% pyridine/1% acetic nonprotein carrier Polybrene and 4 mg of apomyoglobin as de- acid) or by paper chromatography (butanol/acetic acid/water, scribed by Beyreuther et al. (20). 4:1:2) using suitable reference peptides as markers. Radioactive fragments were recovered by elution with 1% acetic acid/1% RESULTS 1-butanol, subjected to Edman degradation, and characterized Incubation of prepromelittin with rat liver microsomes in the by one of the above-mentioned separation methods. presence of detergent has previously been shown to yield pro- Downloaded by guest on October 2, 2021 2262 Biochemistry: Mollay et al. Proc. Natl. Acad. Sci. USA 79 (1982) melittin in a slow hydrolytic reaction (8, 13). After extraction cycle. Fragment Al thus has the structure X-Ala-Leu. Frag- with 1-butanol, this cleavage product can be purified from the ment A2 was also recovered by elution and shortened by one aqueous phase by paper chromatography and subsequently residue by Edman degradation. The resulting dipeptide is iden- characterized by enzymatic digestion and Edman degradation tical to Tyr-Ala on paper electrophoresis at pH 2.0 (Fig. 1D) and of the NH2-terminal fragment. Butanol extraction is a very ef- paper chromatography (data not shown). After a second Edman fective fractionation step, leaving no trace of proteolytic frag- step, free alanine remained in the aqueous layer. Peptide A2 ments characteristic of the proregion of either prepro- or pro- was therefore X-Tyr-Ala and corresponded to the COOH ter- melittin in the organic phase. minus of the signal peptide. By using this time-consuming but very specific assay, we From the aqueous phase ofthe same experiment, promelittin have partially purified a protease from rat liver microsomes that was isolated and labeled alanine could be detected in positions liberates promelittin from prepromelittin. Conditions have 1 and 7 of the NH2-terminal fragment (data not shown). The been worked out in which 30-40% ofthe radioactive precursor labeled promelittin/signal peptide liberated ratio in this ex- is cleaved in a 2-hr incubation period. If this enzyme is an en- periment was close to one, so that further cleavage of signal doprotease, the second reaction product, intact signal peptide, peptide is apparently minimal. should also be detectable. In our search for this apolar peptide, By using a similar approach, we analyzed the butanol layer we were guided by our previous experience with achymotryptic from experiments with prepromelittin labeled with [3H] fragment ofmelittin, melittin-19 (15). This peptide is comprised and patterns resembling the ones shown in Fig. 1 A and B were of one and 18 amino acids with an hydrophobicity index obtained. The material at the neutral region ofthe electropho- (21) of 2.1, similar to the signal peptide ofprepromelittin (one resis at pH 4.8 was recovered by elution, subjected to one cycle lysine and 20 amino acids with an index of2.9). Melittinj p9 can of Edman degradation, and then fractionated by descending be readily extracted into 1-butanol at slightly alkaline pH and paper chromatography. Three dipeptides were present (Fig. 2), then purified by high-voltage paper electrophoresis in 30% of which Ti and T3 have COOH-terminal tyrosine and comi- formic acid. Under these conditions, this water-insoluble pep- grate with Ser-Tyr and Val-Tyr, respectively. Most important, tide migrates slowly toward the cathode. Consequently, butanol a peptide indistinguishable from Tyr-Ala was also observed and extracts ofprepromelittin samples that had been incubated with the tyrosine was shown by Edman degradation to be NH2 ter- a signal peptidase preparation were fractionated in 30% formic minal. Thus, in experiments with labeled alanine and tyrosine, acid. An experiment of this type, using prepromelittin labeled apeptic fragmentwith the structure X-Tyr-Ala could be isolated with [3H]alanine, is shown in Fig. 1. We could not detect from the butanol phase. Since, withprepromelittin, pepsin does any labeled fragment migrating like melittin-19 but noted in- not cleave after the Tyr-Ala sequence, this peptide must rep- stead the presence of much more radioactivity at the origin of resent the COOH end of the peptide present at the origin of the pherogram for assay than for the control sample (Fig. 1A). the electrophoresis in formic acid. As elution of large hydrophobic peptides in good yield is dif- For analysis ofthe NH2-terminal region ofthe cleavage prod- ficult, the material present at the origin was directly digested uct, experiments with labeled , lysine, and phenyl- with pepsin and the soluble peptic fragments were subse- alanine were carried out. Afterpepticdigestion, electrophoresis quently fractionated by paper electrophoresis at pH 4.8 (Fig. at pH 4.8 showed in all instances the presence ofabasic peptide 1B). This clearly showed the absence of peptic fragments de- that was most likely Met-Lys-Phe. However, Edman degra- rived from the proregion, which contain several alanine resi- dation of this peptide gave ambiguous results since large dues and migrate with about two-thirds the mobility of free amounts of radioactive lysine and could be ex- . The neutral material was recovered by elution tracted after the first cycle. and then separated by paper electrophoresis at pH 2.0, yielding To establish that the fragment liberated from prepromelittin two tripeptides, designated Al and A2 (see Fig. 1C). The faster is indeed intact signal peptide, two sequenator runs were car- (i.e., the smaller) ofthese, Al, was indistinguishable from Ala- ried out with material labeled with [3H], an Leu after one step of Edman degradation and >90% of the ra- that gives particularly high incorporation in the wheat-germ dioactivity could be extracted into butyl acetate after the second system. The butanol layer from such assays was applied to Whatman 3MM paper and eluted for several hours with water. The paper strip was then dried and eluted overnight with bu- tanol/acetic acid/water. This yielded sufficient material for the

Tyr Leu

X Ser Tyr Tyr-Ala Val Tyr

T1 T2 T3 6 2 - 0.4 U

11 AF1FLFJ Q Ki~~LLI- 0 0 4 8 12 16 20 24 28 Distance from origin FIG. 2. Paper chromatography of peptic fragments labeled with 0 2 i 8 10 [3H]tyrosine. The butanol extract was processed as described in Fig. 12 14 1 A and B. After electrophoresis at pH 4.8, the neutral peptides were Degradation cycle recovered by elution, subjected to one cycle of Edman degradation, and then fractionated by descending paper chromatography using authen- FIG. 3. Automated Edman degradation of signal peptide labeled tic dipeptides as references. with [3H]leucine. Downloaded by guest on October 2, 2021 Biochemistry: Mollay et al. Proc. Natl. Acad. Sci. USA 79 (1982) 2263

l I I 10 I l l 20 NH 2 *Met-Lys-Phe-Leu-Val-Asn-Val-Ala-Leu-Val-Phe-Met-Val-Val- -Ile-Ser- -Ile- -Ala- 30 --Ala-Pro-Glu-Pro-Glu-Pro-Ala-Pro-Glu-Pro-...... FIG. 4. Amino acid sequence of the pre- and part of the proregion of prepromelittin. The sequence is taken from ref. 19; residues 6 and 17 were identified in unpublished experiments. I, Peptide bonds hydrolyzed by pepsin; -, cleavage site of signal peptidase (between residues 21 and 22). Residues identified in the cleaved signal peptide are as follows: ----, positions assigned to leucine by sequenator runs; -, alanine and tyrosine residues identified by analysis of peptic fragments and residues inferred from characteristics of dipeptides labeled with [3H]alanine or [3H]tyrosine. Several other residues in the cleaved signal peptide have also been identified (unpublished experiments).

sequenator analysis, the results of which are shown in Fig. 3. the electrophoresis in formic acid to monitor the extent of hy- Labeled leucine was found in positions 4 and 9 of the cleaved drolysis. This should facilitate the further purification and fragment, which corresponds to the NH2 terminus of intact characterization ofthis interesting protease and also make pos- prepromelittin. sible a search for the enzyme(s) that degrades signal peptides The sequence of the pre- and part of the proregion of pre- in vivo. Since attempts to isolate these entities from intact cells promelittin is shown in Fig. 4 with the residues identified in have not been successful (12), it seems likely that a mechanism the peptide present in the butanol layer indicated. The results for the rapid hydrolysis of these hydrophobic, possibly mem- show that cleavage of prepromelittin by the protease from rat brane-perturbing, peptides exists. Evidence for such an activity liver yields, besides promelittin, intact signal peptide. in E. coli has been presented (23), and it will be interesting to compare the routes of cleavage and disposal of signal peptides DISCUSSION in bacteria and eukaryotic cells. By using rat liver microsomes as source we have material, par- We thank Prof. K. Beyreuther (University of Cologne), who kindly tially purified a proteolytic enzyme that liberates promelittin carried out the automated Edman degradation. This work was sup- from prepromelittin. As judged by polyacrylamide gel electro- ported by Grant 3810 from the Austrian Fonds zur Forderung der phoresis, the preparation still contains three major and several Wissenschaftlichen Forschung. minor proteins and at least one additional purification step will be necessary before any molecular quantities for this enzyme 1. Blobel, G., Walter, P., Chang, C. N., Goldman, B., Erickson, A. H. & Lingappa, V. R. (1979) in Symposium of the Society of can be described. However, only a single proteolytic activity Experimental Biology (Great Britain), eds. Hopkins, C. R. & appears to be present in these fractions, which made possible Duncan, C. J. (Cambridge Univ. Press, London) 33, 9-36. the experiments described here. 2. Kreil, G. (1981) Annu. Rev. Biochem. 50, 317-348. The 1-butanol extraction procedure was originally developed 3. Blobel, G. & Dobberstein, B. (1975)1. Cell Biol 67, 835-851. to separate promelittin from melittin labeled in vivo, and it was 4. Palmiter, R. D., Gagnon, J., Ericsson, L. H. & Walsh, K. A. (1977) J. Biol. Chem. 252, 6386-6393. subsequently shown that prepromelittin accumulates at the in- 5. Shields, D. & Blobel, G. (1977) Proc. Natl. Acad. Sci. USA 74, terface (16). This simple procedure can now be used to separate 2059-2063. the cleaved signal peptide from the starting material and the 6. Lane, C. D., Champion, J., Haiml, L. & Kreil, G. (1981) Eur.J. other reaction products. The subsequent fractionation by paper Biochem. 113, 273-281. 7. Jackson, R. C. & Blobel, G. (1977) Proc. Natl. Acad. Sci. USA 74, electrophoresis in formic acid removes all the charged low-mo- 5598-5602. lecular-weight products, including >90% of the total radioac- 8. Kaschnitz, R. & Kreil, G. (1978) Biochem. Biophys. Res. Coin- tivity present in the butanol extract. From the material re- mun. 83, 901-907. maining at the origin of this electrophoresis, peptic fragments 9. Strauss, A. W., Zimmerman, M., Mumford, R. A. & Alberts, A. could be isolated that are clearly derived from the signal peptide W. (1980) Ann. NY Acad. Sci. 343, 168-179. 10. Thibodeau, S. N. & Walsh, K. A. (1980) Ann. NY Acad. Sci. 343, ofprepromelittin. In particular, since the tripeptide X-Tyr-Ala, 180-191. detected in experiments with labeled alanine or tyrosine, is 11. Jackson, R. C. & Blobel, G. (1980) Ann. NY Acad. Sci. 343, absent from peptic digests of prepromelittin (19), it must be 391-404. derived from the COOH terminus ofthe cleaved signal peptide. 12. Habener, J. F., Rosenblatt, M., Dee, P. C. & Potts, J. T. (1979) J. Biol Chem. 254, 10596-10599. Automated Edman degradation further showed that we are in- 13. Kreil, G., Mollay, C., Kaschnitz, R., Haiml, L. & Vilas, U. (1980) deed dealing with the complete pre-piece and no evidence for Ann. NY Acad. Sci. 343, 338-346. frayed termini could be observed. The enzyme we have en- 14. Mollay, C., Kreil, G. & Berger, H. (1976) Biochim. Biophys. Acta riched from rat liver microsomes is thus an endoprotease that 426, 317-324. cleaves prepromelittin only at the pre-pro junction. We con- 15. Mollay, C. (1976) FEBS Lett. 64, 65-68. 16. Suchanek, G. & Kreil, G. (1977) Proc. Nati Acad. Sci. USA 74, sider it highly probable that this enzyme is the signal peptidase, 975-978. but similar experiments with at least one additional preprotein 17. Sandberg, P. O., Marzella, L. & Glaumann, H. (1980) Exp. Cell are necessary to establish this firmly. Res. 130, 393-400. Signal peptidase has previously been isolated from Esche- 18. Kaderbhai, M. A. & Freedman, R. B. (1980) Biochim. Biophys. It Acta 603, 366-370. richia coli by Zwizinski and Wickner (22). seems that the 19. Suchanek, G., Kreil, G. & Hermodson, M. A. (1978) Proc. NatL bacterial enzyme also is an endoprotease, since fragments that Acad. Sci. USA 75, 701-704. might have been cleaved prepeptides have been observed on 20. Beyreuther, K., Ehring, R., Overath, P. & Wright, J. K. (1980) polyacrylamide gels (22). The methods we describe for the iso- in Methods in Peptide and Protein Sequence Analysis, ed. Birr, lation ofcleaved signal peptide ofprepromelittin should be ap- C. (North-Holland, Amsterdam), pp. 199-212. 21. Segrest, J. P. & Feldman, R. J. (1974)J. Mol. Biol. 87, 853-858. plicable for other cases also. Moreover, a simplified assay pro- 22. Zwizinski, C. & Wickner, W. (1980) J. Biol. Chem. 255, cedure for signal peptidase can now be developed using, for 7973-7977. example, the amount of radioactivity remaining at the origin of 23. Silver, P., Watts, C. & Wickner, W. (1981) Cell 25, 341-345. Downloaded by guest on October 2, 2021