Molecular Cloning of Endopin 1, a Novel Serpin Localized to Neurosecretory Vesicles of Chromaffin Cells

Serpins represent a diverse class of endogenous protease inhibitors that regulate important biological functions. In consideration of the importance of regulated proteolysis within secretory vesicles for the production of peptide hormones and neurotransmitters, this study revealed the molecular identity of a novel serpin, endopin 1, that is localized to neurosecretory vesicles of neuropeptide-containing chromaffin cells (chromaffin granules). Endopin 1 of 68–70 kDa was present within isolated chromaffin granules. Stimulated cosecretion of endopin 1 with chromaffin granule components, [Met]enkephalin and a cysteine protease known as “prohormone thiol protease,” demonstrated localization of endopin 1 to functional secretory vesicles. Punctate, discrete immunofluorescence cellular localization of endopin 1 in chromaffin cells was consistent with its secretory vesicle localization. Endopin 1 contains a unique reactive site loop with Arg as the predicted P1 residue, suggesting inhibition of basic residue-cleaving proteases; indeed, trypsin was potently inhibited (K i (app) of 5 nm), and plasmin was moderately inhibited. Although endopin 1 possesses homology with α1-antichymotrypsin, chymotrypsin was not inhibited. Moreover, endopin 1 inhibited the chromaffin granule prohormone thiol protease (involved in proenkephalin processing). These results suggest a role for the novel serpin, endopin 1, in regulating basic residue-cleaving proteases within neurosecretory vesicles of chromaffin cells.

processing of proneuropeptides is critical in the biosynthesis of active neuropeptides, endogenous serpins in vivo may potentially influence proneuropeptide processing. Such endogenous serpins should be localized to secretory vesicles, an important subcellular site of proneuropeptide processing, storage, and secretion (9,10). In addition, the serpin should possess specificity for proteases cleaving at basic residues that represent the processing sites of proneuropeptides (6 -9). Serpins possess a reactive site loop (RSL) 1 that mimics the substrate cleavage site (1,2); therefore, a secretory vesicle serpin that may potentially inhibit proneuropeptide processing should possess a basic residue (Arg or Lys) as the active site P1 residue of the RSL domain. The P1 residue represents the amino acid at the NH 2terminal side of the cleaved peptide bond (1,2). Therefore, the goal of this study was to identify a serpin that is present in neurosecretory vesicles and that possesses specificity for inhibiting basic residue-cleaving proteases.
Neuroendocrine chromaffin cells of the adrenal medulla synthesize, store, and secrete several neuropeptides including [Met]enkephalin (11,12) and neuropeptide Y (13). Importantly, secretory vesicles from chromaffin cells have been utilized as a model system for defining protease and protease inhibitor components involved in proneuropeptide processing (6,14). Specifically, secretory vesicles from chromaffin cells (known as chromaffin granules) contain proteases of several mechanistic classes that are involved in proneuropeptide processing. The cysteine protease "prohormone thiol protease" (PTP) has been characterized as a proenkephalin processing enzyme for the production of enkephalin-related opiate peptides (15)(16)(17). Chromaffin granules also contain the subtilisin-like PC1/3 and PC2 (PC ϭ prohormone convertase) enzymes that represent members of the subtilisin-like prohormone convertases that are involved in processing a variety of proneuropeptide and proprotein substrates (6 -8, 18). Moreover, a yapsin-like aspartyl protease is present in chromaffin granules and pituitary secretory vesicles and appears to be involved in proenkephalin and pro-opiomelanocortin processing (19 -21). These proteases are similar in their specificity for cleaving proneuropeptides at paired basic residues.
Previous studies demonstrated that chromaffin granules contain ␣ 1 -antichymotrypsin-like (ACT-like) immunoreactivity (22), a serpin that typically inhibits chymotrypsin (23). ACT also inhibits the cysteine protease PTP involved in proenkephalin processing in chromaffin granules (22). Since anti-ACT antibodies cross-react with related serpins (serine protease inhibitors) (23), the structural identity of the chromaffin cell ACT was unknown. Initial studies identified a partial AM ACT cDNA that represented a neuroendocrine-specific serpin (24). However, a full-length AM ACT cDNA is required for functional analyses.
In this study, molecular cloning revealed a full-length cDNA whose deduced primary structure represents a novel serpin, endopin 1. Its presence in endocrine tissues is specified by "endo" and its serpin-like characteristic is indicated by "pin," thus designating this new serpin as endopin 1. Endopin 1 was present within isolated chromaffin granules and was secreted from chromaffin cells in culture. Homology alignments with other serpins predicted Arg as the P1 residue, thus suggesting inhibition of basic residue-cleaving proteases. Indeed, recombinant endopin 1 inhibited trypsin and plasmin that cleave at basic residues, as well as the PTP proenkephalin-processing enzyme. This is the first demonstration of a novel, endogenous serpin that is secreted from chromaffin cells and that possesses specificity for inhibiting basic residue-cleaving proteases. These results predict a role for serpin regulation of proteolysis within neurosecretory vesicles that synthesize peptide neurotransmitters and hormones.

EXPERIMENTAL PROCEDURES
Molecular Cloning of Endopin 1 cDNA from Bovine Adrenal Medulla-Molecular cloning of endopin 1 cDNA was achieved by isolating overlapping genomic and cDNA fragments (fragments A-C, Fig. 1). A partial 1001-bp adrenal medulla cDNA with homology to ␣ 1 -antichymotrypsin (ACT) (AM ACT) was previously obtained (fragment A, Fig.  1) (25). An overlapping 410-bp DNA fragment (fragment B) was generated by PCR of bovine genomic DNA isolated from hypothalamus (as described previously (25)), utilizing a 5Ј-primer of 5Ј-GCCTCCAGCAA-CACTGACTTCGCCT-3Ј (primer a, Fig. 1), that was complementary to the consensus sequence Leu-Ala-Ser-Ser-Asn-Thr-Asp-Phe-Ala of several serpins (25)(26)(27) and a 3Ј-primer of 5Ј-TTCTTCACATAGTCATT-TATTAGACTCCT-3Ј (primer b, Fig. 1). The resultant 410-bp fragment was subcloned into the pCR TM 2.1 vector (Invitrogen, Carlsbad, CA) and subjected to DNA sequencing as described previously (25). A 2.3-kb genomic DNA (fragment C) that includes the 5Ј-region of the endopin 1 cDNA (fragment C, Fig. 1) was identified in genomic blots probed with fragment B (performed as described previously (25)); fragment C was subcloned into the pUC18 vector and subjected to DNA sequencing with the primer 5Ј-AAGGCTATGGAGACACTCA-3Ј (primer c, Fig. 1). The overlapping DNA fragments A, B, and C provided the nucleotide sequence of the endopin 1 cDNA, GenBank TM accession number AF125526. The deduced amino acid sequence of endopin 1 was compared with serpins using BLAST at NCBI to search GenBank TM . In addition, search of the human EST data base indicated homology of endopin 1 with ␣ 1 -antichymotrypsin; however, identical matches to the endopin 1 sequence were not found, since sequencing of the human genome has not been completed.
RT-PCR of Endopin 1 from Bovine Adrenal Medulla mRNA-DNA sequencing of the endopin 1 cDNA obtained by RT-PCR of bovine adrenal medulla mRNA confirmed that the overlapping cDNA and genomic clones represented endopin 1. First strand cDNA synthesis utilized oligo(dT) 20 as primer and Superscript II and was followed by PCR with primers 5 and 2 ( Fig. 2) consisting of 5Ј-CATATGCTCCCA-GAAAATGTGGTG-3Ј and 5Ј-GGATCCCTAGGCTTCACTGGGGTT-GGT-3Ј, respectively. The 1165-bp RT-PCR product was subcloned into the pCR TM 2.1 vector and subjected to DNA sequencing.
Production of Antisera Against Endopin 1-Rabbit antisera were generated against a synthetic peptide corresponding to residues 3-18, ENVVVKDQHRRVDGHT, of endopin 1. This peptide was synthesized with Cys-Tyr at the NH 2 terminus, as the peptide CYENVVVKDQHR-RVDGHT (by Phoenix Pharmaceuticals, Inc., Mountain View, CA), to allow conjugation to keyhole limpet hemocyanin protein via the Cys residue. The peptide conjugate was injected into rabbits at monthly intervals, and antisera were collected 2 weeks after each injection. Antisera titers were assessed by enzyme-linked immunosorbent assays that measured antibody binding to peptide antigen, coated onto 96-well microtiter wells, as described previously (28). Immunoglobulins from antisera were purified by protein A-Sepharose affinity chromatography according to the manufacturer's protocol (Amersham Pharmacia Biotech).
E64c) and was concentrated. For immunoprecipitation, the sample was incubated with anti-endopin 1 serum (1:20 final dilution) at 4°C for 18 h and with protein A-Sepharose (50 l of slurry) at 4°C for 2 h. After centrifugation at 15,000 ϫ g for 10 min, the pelleted resin was washed in 50 mM Tris-HCl, pH 7.4, 150 mM NaCl, 5 mM EDTA, and 0.2% Triton X-100. The washed pellet was heated in 30 l of 2ϫ sample buffer for SDS-PAGE gels at 95°C for 10 min, and the immunoprecipitate was analyzed by autoradiography of SDS-PAGE gels, as described previously (15). PTP and [Met]enkephalin in secretion media were assessed, as described previously (29).
Endopin 1 in Isolated Secretory Vesicles of Adrenal Medulla (Chromaffin Granules)-Chromaffin granules from bovine adrenal medulla were isolated as described previously (15,16). This procedure results in purified chromaffin granules that are free from lysosomes and other cellular organelles (31). Electron microscopy of isolated chromaffin granules was performed, as described previously (29). A concanavalin A-Sepharose-bound fraction of chromaffin granules was prepared as described previously (15). Aliquots of the ConA-bound fraction were deglycosylated with endoglycosidase F (Roche Molecular Biochemicals) and were analyzed by Western blots (performed as described previously (17)) with the anti-endopin 1 IgGs (1:800 final dilution).
Expression of NH 2 -terminal His-tagged Endopin 1 in E. coli and Purification-The endopin 1/pET19b(ϩ) construct was transformed into BL21(DE3) E. coli, as described previously (16). Cells from a 1-liter culture were resuspended in 10 ml of His-Bind R buffer (5 mM imidazole, 0.5 M NaCl, 20 mM Tris-HCl, pH 7.9) containing a protease inhibitor phenylmethylsulfonyl fluoride and sonicated on ice. The sample was centrifuged at 100,000 ϫ g for 90 min, and the soluble fraction was subjected to affinity chromatography with the His-Bind R resin as described by the manufacturer (Novagen, Madison, WI). Protein concentration was determined by the Bradford method (Bio-Rad), and the purity of the recombinant protein was confirmed by SDS-PAGE, peptide microsequencing, and amino acid composition analyses (Harvard Microchemistry Facility).
Evaluation of Endopin 1 Inhibition of Target Proteases-To test inhibition of target proteases by endopin 1, several serine and cysteine proteases were assayed in the presence of endopin 1. Serine proteases were assayed as described previously for trypsin (23) Purified NH 2 -terminal His-tagged endopin 1 was preincubated with each protease for 15 min at room temperature, and peptide-MCA substrates were added to 100 M final concentration (100 l/assay). After incubation under buffer conditions for each protease, proteolytic activities were monitored by the production of fluorescent aminomethylcoumarin, as described previously (15,31). SDS-stable complexes of endopin 1 and trypsin were assessed by Western blots with anti-human ACT serum on non-reducing SDS-PAGE gels, as described previously (22).
Inhibition of the PTP by Endopin 1-The PTP was purified from chromaffin granules as described previously (15). PTP was preincubated with NH 2 -terminal His-tagged endopin 1 (1-200 g/ml) in PTP assay buffer for 15 min at room temperature, and PTP activity was assayed with 35 S-enkephalin precursor as substrate with incubation at 37°C for 2 h. PTP activity was measured by the production of acid FIG. 2. Endopin 1 cDNA, nucleotide and deduced primary structure. The nucleotide sequence and deduced primary sequence of the endopin 1 cDNA were predicted from the overlapping cDNA and genomic DNA fragments of endopin 1 (fragments A, B, and C, Fig. 1). The asterisk indicates the predicted P1 active site Arg residue. A signal sequence is predicted to include the first 24 residues. The first residue (indicated as residue ϩ1) of the mature endopin 1 is predicted based on known NH 2 -terminal residues of bovine trypsin inhibitor and elastase inhibitor serpins. Potential Asn-Xaa-Thr/Ser glycosylation sites are indicated by Asn. Primers a and b used for PCR of genomic fragment B and primer c used for DNA sequencing of the 2.3-kb genomic fragment C are indicated. Primers 1-7 are illustrated by arrows (with incorporation of restriction sites whose nucleotide sequences are indicated at 5Ј-ends of primers) and were used to subclone the endopin 1 cDNA into the pET(19)ϩ vector.
(trichloroacetic acid)-soluble radioactivity and by assessing cleavage of the 35 S-enkephalin precursor substrate by autoradiography of SDS-PAGE gels, as described previously (15).

Molecular Cloning of the cDNA Encoding the Novel Serpin
Endopin 1-Molecular cloning of the full-length coding sequence of the endopin 1 cDNA, a novel AM serpin with homology to ACT, was achieved through genomic and cDNA cloning approaches to obtain overlapping DNA segments (Fig. 1). Initially, a partial 3Ј-segment of the endopin 1 cDNA was isolated (fragment A, Fig. 1a) (24). An overlapping cDNA fragment (fragment B, Fig. 1a) was generated by PCR with genomic DNA as template, a 5Ј-primer corresponding to a consensus sequence LASSNTDFA present in several ACT-like serpins (25)(26)(27) (primer a, Fig. 1, a and b), and a 3Ј-primer corresponding to the partial cDNA (primer b, Fig. 1a). The LASSNTDFA sequence represents a consensus sequence within the NH 2 -terminal region of the serpins bovine liver L2 ACT (25), human liver ACT (23), bovine trypsin inhibitor (26), and goat contrapsin (ACT) (27) (Fig. 1b). PCR generated the predicted 410-bp fragment B ( Fig. 1a) that should reside within a single exon, based on the gene structures of human ACT and human ␣ 1 -antitrypsin (39). To obtain the 5Ј-region of the cDNA, genomic blots probed with fragment B identified a 2.3-kb band (fragment C, Fig. 1) obtained by RsaI digestion of bovine genomic DNA. DNA sequencing of fragment C with a primer corresponding to a 5Јregion of fragment B (primer c, Fig. 1a) determined the sequence of the 5Ј-region of the endopin 1 cDNA.
To determine that the endopin 1 cDNA sequence (Fig. 2) corresponds to endogenous endopin 1 mRNA, deduced from overlapping DNA fragments, RT-PCR of bovine adrenal medulla RNA (using primers 2 and 5, Fig. 2) amplified the cDNA segment encoding the mature endopin 1 (without signal sequence). DNA sequencing indicated that the cDNA obtained by direct RT-PCR was identical to the endopin 1 cDNA sequence deduced from overlapping DNA fragments (Fig. 2).
The nucleotide sequence of the endopin 1 cDNA (Fig. 2) contains 1509 bp with an open reading frame of 411 residues. The deduced primary sequence indicates a protein of 46,203 daltons with an NH 2 -terminal signal sequence corresponding to the first 24 residues, based on the hydrophobic nature of this domain that is characteristic of signal sequences (40). The signal sequence suggests that endopin 1 may be routed to the secretory pathway and may be secreted from cells. The pre-  (23), and mouse contrapsin (m. Con.) (42). Dots represent identical residues, and dashes represent gaps. The reactive site loop region is indicated by the box, with the predicted P1 residues (underlined) indicated by the asterisk. The underlined residues 3-18 represent the synthetic peptide used as antigen to generate anti-endopin 1 serum. The percent identity (% Identity) in primary sequence of endopin 1 segments with those of bovine L2 ACT, human liver ACT, and mouse contrapsin are indicated in the right margin. Secretion of PTP into media, induced by nicotine and KCl, was assessed by Western blots with anti-PTP serum, as described previously (29). PTP was analyzed in media from control cells, nicotine-treated cells, and from cells treated with KCl. c, secretion of [Met]enkephalin. [Met]Enkephalin secreted into the media from control, nicotine-, and KCltreated cells was measured by radioimmunoassay, as described under "Experimental Procedures." dicted mature endopin 1 without the signal sequence, with Leu at the NH 2 terminus (indicated as the ϩ1 residue, Fig. 2), possesses a calculated molecular mass of 43,656 daltons, and a pI (isoelectric point) of 5.3. Potential glycosylation sites are indicated by consensus Asn-Xaa-Ser(Thr) glycosylation sites (41).

Functional Localization of Endopin 1 in Secretory Vesicles, Demonstrated by Cosecretion with [Met]Enkephalin and PTP
from Chromaffin Cells-The signal sequence of endopin 1 suggests that it may be routed to secretory vesicles. Therefore, secretion of endopin 1 from chromaffin cells was examined to demonstrate functional localization within secretory vesicles. Secretion of metabolically labeled 35 S-endopin 1 from chromaffin cells was induced by nicotine and KCl depolarization, and the media were analyzed for endopin 1 by immunoprecipitation and SDS-PAGE. Nicotine and KCl induced the secretion of 68 -70-kDa endopin 1 (Fig. 4a). Importantly, the secretion of endopin 1 was accompanied by the simultaneous release of PTP, a proenkephalin-processing enzyme, and [Met]enkephalin from chromaffin cells (Fig. 4, b and c). Secretion of approximately 10 -14% of cellular secretory vesicles occurred upon stimulation by nicotine or KCl depolarization, measured by the cellular content of [Met]enkephalin after secretion. These results demonstrate the cosecretion of endopin 1 with [Met]enkephalin and PTP, indicating the localization of endopin 1 within functional secretory vesicles.
Endopin 1 Glycoprotein in Isolated Secretory Vesicles (Chromaffin Granules) of Adrenal Medulla-Functional secretion of endopin 1 predicted that it would be localized within isolated secretory vesicles of adrenal medulla (chromaffin granules) purified by sucrose density gradient centrifugation (31). The integrity and homogeneity of these purified vesicles was demonstrated by electron microscopy (EM) (Fig. 5a). These isolated granules appear similar to those observed by EM in situ in chromaffin cells (43). Western blots demonstrated the presence of 68 -70-kDa endopin 1 in chromaffin granules (Fig. 5b); con-trol immunoblots with preimmune serum showed no bands. Furthermore, endopin 1 was not detected in lysosomes isolated from bovine adrenal medulla (31) (data not shown). These results demonstrate that endopin 1 is present in chromaffin granules, secretory vesicles of adrenal medulla.
Consensus glycosylation sites predicted that endopin 1 may be a glycoprotein. The presence of endopin 1 in the ConA-bound fraction of chromaffin granules demonstrated the glycoprotein nature of endopin 1 (Fig. 5c). The 68 -70-kDa endopin 1 glycoprotein resembles the molecular weight and glycoprotein nature of many serpins (44). Endopin 1 was sensitive to deglycosylation by endoglycosidase F that hydrolyzes N-linked oligosaccharides of the N-acetylglucosamine type, with hydrolysis of high mannose type structures to a lesser extent (45). The diversity of N-linked oligosaccharides of glycoproteins, combined with the specificity of endoglycosidase F for a few types of oligosaccharides, is consistent with partial deglycosylation of endopin 1, compared with the endopin 1 polypeptide backbone (from cDNA sequence) with calculated mass of 43.6 kDa.
Endopin 1 Immunofluorescence Cytochemistry in Chromaffin Cells-Immunofluorescence cytochemistry demonstrated the discrete pattern of endopin 1 immunostaining in primary cultures of chromaffin cells (Fig. 6), which parallels the punctate immunostaining pattern of the secretory vesicle components PTP (Fig. 7a), a proenkephalin-processing enzyme, and its product [Met]enkephalin (Fig. 7b). The majority of chromaffin cells possess endopin 1, which is consistent with the localization of [Met]enkephalin and PTP to these cells. The discrete pattern of endopin 1 immunofluorescence staining in the cell body areas, neuritic-like extensions, and absence in nuclei is consistent with localization to secretory vesicles of chromaffin cells. In addition, the discrete pattern of endopin 1 immunostaining resembled that of chromogranin A (data not shown), another marker for neuroendocrine secretory vesicles (46,47). The discrete pattern of endopin 1 localization in chromaffin cells, like the immunostaining pattern of the secretory vesicle components [Met]enkephalin and PTP, is consistent with the presence of endopin 1 in secretory vesicles.
Expression of Recombinant Endopin 1 and Target Proteases: Characterization of Trypsin Inhibition-To characterize the target protease specificity of endopin 1, recombinant endopin 1 was expressed in Escherichia coli as a fusion protein with an NH 2 -terminal polyhistidine tag. The segment of the cDNA corresponding to the mature endopin 1 protein (without signal sequence), beginning with LPENV . . . as the NH 2 terminus (Fig. 2), was subcloned into the pET-19b(ϩ) vector. IPTG induced the expression of the 46,439-dalton NH 2 -terminal Histagged endopin 1 protein in E. coli (Fig. 8a). The purified NH 2 -terminal His-tagged endopin 1 (obtained by a nickel af- finity column) was observed as a 46-kDa band on SDS-PAGE gels and Western blots. Some endopin 1 also appeared as a 92-kDa band in anti-endopin 1 Western blots, possibly representing a dimer of endopin 1, since serpins (48) such as ␣ 1antitrypsin (49) and PAI-2 (50) are well known to polymerize spontaneously.
Recombinant endopin 1 (purified) completely inhibited trypsin at a molar ratio of inhibitor to enzyme of 12:1 (Fig. 8b). Trypsin cleaves at the Arg residue (P1 position) of the substrate Z-Arg-Arg-MCA. Endopin 1 was a less effective inhibitor of plasmin (25% inhibition was observed at a molar ratio of 12:1 for inhibitor:enzyme) which cleaves at the basic Lys residue of the substrate Boc-Glu-Lys-Lys-MCA. However, endopin 1 did not inhibit chymotrypsin or elastase when tested at the same molar ratio. In addition, subtilisin A was not inhibited, even when tested at a high molar ratio of inhibitor to enzyme of 134:1. Thus, although endopin 1 possesses primary sequence homology with ␣ 1 -antichymotrypsin, endopin 1 does not inhibit chymotrypsin or proteases that cleave at Phe or Ala hydrophobic residues. These results demonstrate the protease-specific nature of endopin 1 to inhibit an Arg-or Lys-cleaving protease(s), such as trypsin or plasmin.
Endopin 1 inhibited 80% of trypsin activity at a molar ratio of inhibitor:enzyme of 1:1 (Fig. 8c). The nearly complete inhibition of trypsin indicates an approximate stoichiometry of one inhibitor molecule interacting with one trypsin molecule. Inhibition of trypsin was concentration-dependent; kinetic assessment by the equation tected on non-denaturing SDS-PAGE gels by the retarded electrophoretic mobility of endopin 1 as a band of approximately 55 kDa (detected by endopin 1 Western blots, Fig. 8d, lane 5). These complexes were observed at a molar ratio of inhibitor: enzyme of 2:1. Endopin 1 thus resembles serpins in its ability to form complexes with a target protease. At a molar ratio of inhibitor:enzyme of 1:1, a complex represented by a band of approximately 50 kDa was observed (Fig. 8d, lane 3); it is known that upon serpin-protease complex formation, the protease cleaves the serpin to result in a complex of slightly lower apparent molecular weight (23). Thus, the complexes of slightly different mobilities may represent intact serpin-protease and cleaved serpin-protease complexes.
In contrast to potent inhibition of trypsin by endopin 1 (K i(app) of 5 nM), plasmin was only moderately inhibited with a K i(app) of approximately 0.8 M (data not shown). Further tests showed that endopin 1 possesses remarkable selectivity for trypsin, compared with other basic residue-cleaving proteases. Endopin 1 (at molar ratios of endopin 1:protease of 20:1) did not inhibit thrombin, furin, cathepsin B, cathepsin L, or papain proteases that cleave at Arg residues (data not shown). These results demonstrate that among several basic residue-cleaving proteases, endopin 1 possesses selectivity for inhibiting certain basic residue-cleaving proteases, such as trypsin and plasmin.
Inhibition of the Proenkephalin-processing Enzyme PTP by Endopin 1-The presence of endopin 1 in chromaffin granules where proenkephalin processing occurs suggests that endopin 1 may inhibit PTP which cleaves proenkephalin at paired basic residues within these granules (15)(16)(17). In vitro enkephalin precursor processing assays showed that endopin 1 inhibited PTP at molar ratios of inhibitor:enzyme of approximately 40:1 and demonstrated a K i(app) of 1.0 M for inhibition of PTP (Fig.  9a). The cleavage of 35 S-enkephalin precursor was inhibited by up to 70% inhibition (at 4.4 M), demonstrated by SDS-PAGE gels and autoradiography (Fig. 9b). Endopin 1 inhibition of PTP is consistent with the specificity of PTP for cleavage at paired basic residues with Arg in the P1 position (15,16). Overall, these results demonstrate that endopin 1 is a novel, secretory vesicle serpin with specificity for inhibiting basic residue-cleaving proteases. DISCUSSION The physiological importance of proteolytic processing of proneuropeptides into active peptide neurotransmitters and hor- mones in neuroendocrine functions predicts that endogenous protease inhibitor(s) may be involved in regulating proneuropeptide processing. Serpins represent a diverse class of protease inhibitors that control proteases participating in key biological functions. Consideration of serpins as a regulator(s) of proneuropeptide processing requires first that the serpin should be secreted from regulated secretory vesicles to demonstrate localization to secretory vesicles where proteolytic processing of proneuropeptides occurs. Second, the serpin should inhibit basic residue-cleaving proteases since proneuropeptides are processed at paired basic residues. In this study, molecular cloning revealed the structural identity of a novel serpin, endopin 1, that is localized to secretory vesicles, demonstrated by functional secretion from chromaffin cells that synthesize several neuropeptides including enkephalins. Further evidence consistent with the vesicular localization of endopin 1 was indicated by its presence in isolated secretory vesicles of chromaffin cells and discrete immunofluorescence localization in chromaffin cells. Importantly, endopin 1 possesses specificity for inhibiting proteases cleaving at basic residues but not proteases cleaving at non-polar or hydrophobic residues. Consist-ent with the target protease specificity of endopin 1, endopin 1 inhibited the proenkephalin-processing enzyme PTP (prohormone thiol protease) that cleaves at paired basic residues. These results demonstrate the presence of a novel serpin, endopin 1, in secretory vesicles of chromaffin cells. The subcellular location of endopin 1 suggests that it could potentially regulate basic residue-cleaving proteases that participate in the conversion of protein precursors into active peptide hormones and neurotransmitters. This novel serpin is termed "endopin" based on its localization in neuroendocrine tissues of adrenal medulla and pituitary (and its absence in liver) (24) and on its sequence homology to the serpin family of protease inhibitors including ACT. During the course of this study, a second adrenal medulla serpin cDNA with sequence homology to ACT was identified that possesses a different target protease specificity. For this reason, "endopin 1" designates the unique serpin of this study and the second "endopin 2" will be described in a subsequent report. Molecular cloning of the adrenal medulla endopin 1 cDNA was achieved by obtaining overlapping genomic DNA and cDNA fragments. The complete cDNA sequence derived from these overlapping fragments was confirmed by RT-PCR of poly(A ϩ ) RNA and DNA sequence analysis. Importantly, the deduced primary sequence of endopin 1 indicates that it is a distinct serpin with 50 -70% homology to human and bovine liver ACTs (23) and only 30 -35% homology to ␣ 1 -antitrypsin (52). Notably, the RSL of endopin 1 possesses Arg as the predicted P1 residue, indicating that endopin 1 may inhibit Argcleaving proteases such as trypsin. The P1 residue is an important feature of serpins in determining their target protease(s). For example, human liver ACT, with Leu as the P1 residue, inhibits chymotrypsin, but mutagenesis of the P1 residue to Arg alters its target protease specificity, resulting in inhibition of trypsin (23). Therefore, the RSL of endopin 1 predicts that it may inhibit Arg-cleaving proteases such as trypsin or proneuropeptide processing enzymes.
Recombinant NH 2 -terminal His-tagged endopin 1 potently inhibited trypsin (K i(app) of 5 nM) that cleaves at Arg residues in the P1 position of the cleaved peptide bond (the peptide bond P1-P1Ј is cleaved). Endopin 1 was less effective in inhibiting plasmin that cleaves at Lys residues (K i(app) of 0.8 M). The lack of inhibition of several other basic residue-cleaving proteases, including thrombin, furin, cathepsin B, cathepsin L, and papain, indicates the remarkable specificity of endopin 1 for certain basic residue-cleaving proteases. Moreover, endopin 1 did not inhibit chymotrypsin, elastase, or subtilisin A which cleave at hydrophobic residues.
Serpins typically form SDS-stable complexes with their target proteases. Formation of endopin 1 complexes with trypsin was observed at a molar ratio of inhibitor to enzyme of 2:1, demonstrated by the retarded electrophoretic mobility of endopin 1 complexed with trypsin on non-denaturing SDS-PAGE gels. These results demonstrate the serpin nature of endopin 1.
Localization of endopin 1 within secretory vesicles where proneuropeptide processing occurs was predicted based on the presence of a signal sequence at the NH 2 terminus of endopin 1. The presence of endopin 1 in functional secretory vesicles in chromaffin cells was demonstrated by regulated secretion of 35 S-endopin 1 (by immunoprecipitation) induced by nicotine and KCl depolarization. Moreover, endopin 1 was cosecreted with the secretory vesicle components [Met]enkephalin and PTP. Nicotinic receptor and KCl depolarization are known to induce secretion of the contents of regulated secretory vesicles from chromaffin cells (53). Secretion predicted the presence of 68 -70-kDa endopin 1 in isolated chromaffin granules (secretory vesicles). The glycoprotein nature of endopin 1 in chromaffin granules was demonstrated by its binding to a concanavalin A-Sepharose lectin affinity column and by its sensitivity fo deglycosylation by N-glycosidase F. Furthermore, immunofluorescence cytochemistry endopin 1 demonstrated its discrete cytoplasmic and neuritic pattern of subcellular localization (with no staining in the nucleus); this staining parallels that of the secretory vesicle components [Met]enkephalin and PTP, a proenkephalin processing enzyme. These results thus demonstrate the presence of endopin 1 within functional secretory vesicles of chromaffin cells.
The cysteine protease PTP has been identified as a proenkephalin processing enzyme in chromaffin granules (15)(16)(17). The parallel cleavage specificity of PTP for paired basic residues and the specificity of endopin 1 to inhibit basic residuecleaving proteases suggested that endopin 1 could inhibit PTP. Indeed, endopin 1 inhibited PTP cleavage of recombinant enkephalin precursor in in vitro processing assays. Effective inhibition of PTP occurred at micromolar concentrations of en-dopin 1, with an estimated K i(app) value of 1.0 M endopin 1. The in vivo levels of endopin 1 in chromaffin granules are estimated at approximately 10 M, based on the sensitivity of the anti-endopin 1 antibody to detect endopin 1 in chromaffin granules. Thus, the estimated in vivo micromolar levels of endopin 1 appear to be compatible with inhibition of PTP by endopin 1 in chromaffin granules.
Endopin 1 inhibition of PTP is consistent with partial proenkephalin processing in vivo. Complete inhibition would be contrary to the observed partial proenkephalin processing in vivo. Moreover, transient inhibition is predicted, since serpin and protease interactions are followed by proteolytic cleavage and inactivation of the serpin. PTP cleaved endopin 1 to generate a 4-kDa fragment that is consistent with cleavage in the vicinity of the RSL domain (data not shown). These results suggest that the in vivo inhibition of PTP may be transient. For these reasons, endopin 1 may retard proenkephalin processing in vivo.
Endopin 1 may inhibit other secretory vesicle proteases with cleavage specificities for basic residues, such as members of the subtilisin-like prohormone convertases (PC) (7,8). Chromaffin granules contain the subtilisin-like PC1/3-and PC2-processing enzymes (18,54). Predicted interactions of endopin 1 with PC1/3 and PC2 are consistent with the basic residue cleavage specificities of these PC enzymes. It is known that the subtilisin-like furin protease is inhibited by the engineered ␣ 1 -antitrypsin Portland (␣ 1 -antitrypsin PDX) serpin, containing Arg-Ile-Pro-Arg as the P4 to P1 residues within the reactive site loop (54 -57). However, the engineered ␣ 1 -antitrypsin PDX variant is not present in vivo. It will therefore be of interest to assess effects of the naturally occurring endopin 1 on the proteolytic activities of PC1/3, PC2, and related subtilisin-like processing enzymes. This is the first demonstration of a novel, endogenous secretory vesicle serpin with specificity for basic residue-cleaving proteases, consistent with proneuropeptide processing occurring at paired basic residues. Endopin 1 is an effective inhibitor of the basic residue-cleaving PTP-processing enzyme that is present in chromaffin granules for proenkephalin processing. These findings provide a possible mechanism for the limited processing of proenkephalin in adrenal medulla in vivo (11,12,58). Overall, the results from this study provide insight into serpin control mechanisms that may regulate neurosecretory vesicle proteases, including proneuropeptide-processing enzymes.