Expression of Mouse Proopiomelanocortin in an Insulinoma Cell Line REQUIREMENTS FOR @-ENDORPHIN PROCESSING*

Proopiomelanocortin (POMC) is a neuroendocrine precursor protein which is processed at paired basic amino acids in a tissue-specific manner. To study this phenomenon, a vaccinia virus recombinant, which di- rects the synthesis of mouse POMC (VV:mPOMC) was constructed and used to infect epithelial (BSC-40) and endocrine (Rin m5F) cell lines. Bona fide mPOMC was produced in both cell types and @-endorphin immuno- reactivity was secreted in a nonregulated manner from BSC-40 cells and in a regulated manner from Rin m5F cells. Although the precursor was not cleaved to smaller &MSH or &endorphin immunoreactive peptides in BSC-40 cell extracts, Rin m5F cells produced primarily authentic y-lipotropin and des-acetyl 8-endorphinl-31. Furthermore, production of these pep- tides was restricted to the regulated secretory pathway in Rin m5F cells. Site-directed mutagenesis was then used to change the inefficiently recognized Lys-Lys potential cleavage site near the carboxyl terminus of &endorphin to Lys-Arg. Expression of the mutant precursor in Rin m5F cells resulted in the synthesis of both des-acetyl and ,&endorphin1-2e. Most neuroendocrine peptides are derived from larger precursor proteins or prohormones. Peptides are excised from the prohormone by endoproteolysis, usually

Proopiomelanocortin (POMC) is a neuroendocrine precursor protein which is processed at paired basic amino acids in a tissue-specific manner. To study this phenomenon, a vaccinia virus recombinant, which directs the synthesis of mouse POMC (VV:mPOMC) was constructed and used to infect epithelial (BSC-40) and endocrine (Rin m5F) cell lines. Bona fide mPOMC was produced in both cell types and @-endorphin immunoreactivity was secreted in a nonregulated manner from BSC-40 cells and in a regulated manner from Rin m5F cells. Although the precursor was not cleaved to smaller &MSH or &endorphin immunoreactive peptides in BSC-40 cell extracts, Rin m5F cells produced primarily authentic y-lipotropin and des-acetyl 8endorphinl-31. Furthermore, production of these peptides was restricted to the regulated secretory pathway in Rin m5F cells. Site-directed mutagenesis was then used to change the inefficiently recognized Lys-Lys potential cleavage site near the carboxyl terminus of &endorphin to Lys-Arg. Expression of the mutant precursor in Rin m5F cells resulted in the synthesis of both des-acetyl and ,&endorphin1-2e.
Most neuroendocrine peptides are derived from larger precursor proteins or prohormones. Peptides are excised from the prohormone by endoproteolysis, usually at pairs of basic amino acids (e.g., Lys-Arg, Lys-Lys, Arg-Arg) (1). In many instances, further modification of the cleavage products, such as carboxyl and/or amino shortening (2), amidation (3), and N-acetylation (4), must also occur for bioactive peptides to be generated.
A number of prohormones contain the sequence of more than one bioactive peptide. A well-studied example of this type of prohormone (polyprotein) is proopiomelanocortin (POMC).' POMC is expressed in the anterior and neurointermediate lobes of the pituitary as well as in several regions of the brain and other tissues and is processed in a tissue-specific manner to produce a variety of peptide hormones (5). In the pituitary, these are primarily ACTH, P-LPH, and amino-terminal fragment in the anterior lobe, and smaller amino-terminal fragments, a-MSH, CLIP, T-LPH, and several forms of @-endorphin in the neurointermediate lobe. The mechanisms responsible for tissue-specific processing of the prohormone are unknown, but several factors may be involved, including: (i) selective expression of distinct processing enzymes, (ii) differential compartmentalization of either the protease(s) or the precursor, and (iii) differential modification of the prohormone to regulate cleavage site accessibility.
For peptide hormone in which secretion can be controlled by external stimuli (regulated release), the proteolytic maturation steps occur once the precursor has been committed to the regulated secretory pathway (6, 7). In contrast, prohormone that fails to enter the regulated secretory pathway is not cleaved and is secreted in bulk-flow carrier vesicles via the constitutive pathway (8). It is not clearly understood which structural features of prohormones may serve as determinants for targeting the precursor to the correct cellular compartment for cleavage (9) or for recognition by processing endopeptidases (10). However, several naturally occurring mutants of proinsulin (11) and proalbumin (12,13) have demonstrated that the integrity of the paired basic amino acid target sequence is critical for proteolysis. Additionally, sequences (14) or posttranslational modifications (15) proximal to cleavage sites may also influence processing.
The cloning of a variety of prohormone genes and/or cDNAs coupled with gene transfer technology permits a systematic study of the processing of these precursor proteins in heterologous cell types (16). In addition, structural alterations can be introduced into these prohormones to define the parameters involved in cleavage site recognition or targeting to the cellular compartment containing the processing endopeptidase(s).
In this communication, we describe the use of a vaccinia virus expression system to study the proteolytic maturation of mouse POMC (mPOMC) in two heterologous cell lines. In addition, the importance of the amino acid composition of a potential cleavage site was determined by expression of a mutant mPOMC generated by site-directed mutagenesis.
Vaccinia virus strain WR was used in these studies. VV:mPOMC

3545
Processing of P-Endorphin in Insulinoma Cells was constructed as described (21 Antibodies and Radioimmunoassay (RIA)-All synthetic peptide standards were obtained from either Peninsula Laboratories or Bachem Inc., with the exception of rat insulin, which was obtained from Novo Biolabs. Monoiodinated porcine insulin for RIA was obtained from Du Pont-New England Nuclear. N-Acetylated (Ac) camel P-endorphinl-2.i and monkey p-MSH were iodinated using Na"'I (Du Pont-New England Nuclear), by the chloramine-T method (22). Guinea pig anti-rat insulin (Linco Research Inc.) was raised against rat insulin but is 30% cross-reactive with the prohormone. p-Endorphin RIAs were performed using a rabbit antiserum (FF) raised against human P-endorphin. The antibody is directed against the midportion of the molecule and cross-reacts equally with Ac and des-Ac p-endorphins 1-31, 1-27, and 1-26. It also cross-reacts, although to a lesser extent, with larger 6-endorphin-containing peptides, including P-LPH and POMC. The 8-MSH RIAs were performed using a rabbit antiserum (Molly) raised against monkey p-MSH. Antiserum Molly cross-reacts with mouse y-LPH, 6-LPH, and POMC. RIAs were performed as follows: samples were incubated overnight at 4 "C with primary antiserum and iodinated peptide in 200 111 of 50 mM Na2HP04, 100 mM NaCl, 10 mM NazEDTA, 50 mM NaN3, 0.1% BSA, 0.1% Triton X-100, p H 7.4 (RIA buffer I). Primary antibodies were precipitated with an excess of either goat anti-guinea pig IgG (Linco Research Inc.) or goat anti-rabbit IgG (Arne1 Products) in 0.5 ml of 1.2 M (NH4)2S0,, 10 mM Na2HP04, pH 7.5, in the presence of 0.2% nonimmune rabbit (or guinea pig) serum (NRS) at 4 "C for 30 min. Standards were assayed in triplicate and samples in duplicate.
POMC Purification-BSC-40 cells were grown on 15-cm plates in MCDB 202 medium. Confluent cells were infected with VV:mPOMC at a multiplicity of infection (m.0.i.) of 2 and incubated in serum-free MCDB 202 containing 50 pg/ml ovalbumin for 24 h. Medium was then collected and adjusted to 50 mM MES, pH 6.0, and 20% acetonitrile (ACN, J. T. Baker Chemical Co. high performance liquid chromatography (HPLC) grade), and centrifuged 30 min at 53,000 X g (r,,") in a Beckman SW28 rotor to remove particulate matter. The clarified media was loaded onto an FPLC cation exchange column (Pharmacia LKB Biotechnology Inc. FPLC Mono S HR 5/5) at a flow rate of 0.6 ml/min. POMC was eluted with a linear NaCl gradient (from 50 to 500 mM over 30 min at 0.6 ml/min) in 20% ACN, 50 mM MES, pH 6. POMC eluted at 270 mM NaCI. Fractions containing POMC, as determined by @MSH immunoreactivity (IR), were pooled, adjusted to 0.1% trifluoroacetic acid (Pierce Chemical Co. HPLC grade) and applied to a C4 HPLC column (Vydac 214TP54) for reversed phase chromatography. POMC eluted in a two-step linear gradient (from 12.8 to 20% ACN over 2 min, immediately followed by 20 to 39.2% ACN over 75 min) in 0.1% trifluoroacetic acid with a flow rate of 1 ml/min. POMC eluted at 36% ACN.
POMC fractions were pooled, lyophilized to 50% of their original volume, and diluted 2-fold in water, and HPLC grade heptafluorobutyric acid (HFBA, Pierce Chemical Co.) was added to a final concentration of 0.13%. This sample was reapplied to the C4 column and eluted with a linear gradient (28 to 47% ACN over 80 min) in 0.13% heptafluorobutyric acid with a flow rate of 1 ml/min. POMC eluted at 44.5% ACN. Aliquots of POMC-containing fractions were lyophilized to dryness, resuspended in 62.5 mM Tris-HC1, pH 6.8, 2% sodium dodecyl sulfate (SDS), 10% glycerol, 5% p-mercaptoethanol, and 0.001% bromphenol blue (SDS-SB), and resolved on a 12.5% acrylamide/O.l% bisacrylamide gel with a 5% acrylamide/0.13% bisacrylamide stacking gel as described (23) (SDS-PAGE). Silver staining demonstrated about 90% purity of POMC at this stage. 200 pmol of POMC were lyophilized and resolved by SDS-PAGE as described above and electrophoretically transferred to Immobilon@ PVDF membrane (Millipore) (24). POMC was localized on the membrane by staining with Coomassie Brilliant Blue R250 and the band was excised and subjected to nine rounds of sequencing in an Applied Biosystems 470A automated protein sequencer as described (25).
Radiolabeling and Immunoprecipitation-Cells grown in 35-mm plates were infected at a m.0.i. of 5 and incubated 6 h in culture medium. Medium was then removed and cells were washed and incubated 30 min at 37 "C in 2 ml of methionine-free MEM (GIBCO). This was replaced by 0.5 ml of methionine-free MEM containing 350 gCi of (35S]methionine (Du Pont-New England Nuclear, 1100 Ci/ mM), and cells were incubated at 37 "C for 1 h. Cells were harvested in 0.5 ml of 1 M acetic acid (pH 1.9 with HCl), 1 mM phenylmethylsulfonyl fluoride (harvest buffer) and stored at -70 "C until used. Cell extracts were probe sonicated 20 s at 40 watts and clarified in a microcentrifuge. Aliquots of labeled extracts were mixed with similarly treated unlabeled BSC-40 carrier extracts, neutralized with Tris base, and precipitated with 4 volumes of acetone at -20 "C overnight. Pelleted precipitates were resuspended in 450 pl of 150 mM NaC1, 5 mM EDTA, 50 mM Tris-HC1, pH 8.0, 0.5% Nonidet P-40, 0.01% NaN3, 0.1% SDS (NET-N) and preincubated with 2 pl of NRS and 25 111 of protein A-Sepharose 6MB (Sigma) (preequilibrated with NET-N) with gentle mixing at 4 "C, >1 h. The protein A-Sepharose was removed by centrifugation and the supernatants incubated with gentle mixing overnight at 4 "C with an affinity purified anti-ACTH antibody, which cross-reacts with intact precursor (Van Go; obtained from the laboratory of Dr. Ed Herbert). Antigenlantibody complexes were adsorbed to 25 pl of protein A-Sepharose at 4 "C with gentle mixing for 1 h. These were pelleted and washed in 1 ml of NET-N containing 500 mM NaCl, 1 mg/ml ovalbumin, and 0.25% sodium deoxycholate (NET-NOND). Pellets were resuspended in NET-NOND and gently mixed for 15 min a t 4 "C. These washed pellets were resuspended in NET-N, transferred to new tubes, and washed once more in NET-N. Pellets were resuspended in 25 p1 of SDS-SB, heated 5 min in a boiling water bath, and resolved by SDS-PAGE as described above. Gels were fixed and stained in Coomassie Brilliant Blue R-250 and impregnated with EN3HANCE (Du Pont-New England Nuclear) for fluorography.
Gels were then dried and placed under film.
Regulated Secretion-Cells were infected with virus at a m.0.i. of 1 and incubated in culture medium 16 h. This medium was removed and cells were rinsed once in a large volume of serum-free medium containing 0.07% BSA. Serum-free medium (1 ml for 35-mm and 2 ml for 60-mm plates) containing 0.07% BSA with or without secretagogues (100 nM phorbol 12-myristate 13-acetate (PMA (Sigma) or 50 PM forskolin (Behring Diagnostics)) was added to each plate, and cells were incubated at 37 "C for 1 h. Medium was collected and assayed directly by RIA, or trifluoroacetic acid was added to 0.1% final concentration and samples were clarified by centrifugation for injection directly on the C4 column. Peptides were resolved in the trifluoroacetic acid/ACN gradient described above. 25-p1 aliquots or less of 1-min fractions were assayed directly by RIA. Larger volumes were aliquoted into RIA tubes and lyophilized to dryness before being assayed.
HPLC Analysis of Peptides in Cell Extracts-Cultures were infected with recombinant vaccinia at a m.0.i. of 1 and incubated at 37 "C 16 h. Medium was removed from cells, which were then washed once in fresh warm medium, scraped in harvest buffer and stored at -70 "C. Extracts were thawed, probe sonicated 20 s, clarified by centrifugation, lyophilized to dryness, resuspended in 0.1% trifluoroacetic acid, 12.8% ACN, injected on the Cq column, and resolved with the trifluoroacetic acid/ACN gradient described above. Fractions from the C, column, which were further resolved by cation exchange HPLC, were diluted with 5 volumes of 10 mM HPLC grade NH40Ac (J. T. Baker Chemical Co.) pH 4.5, in 25% ACN and injected onto a PolyAspartic Acid WCX column (The Nest Group, No. 1850-102). &Endorphin samples were resolved with a two-step linear NH40Ac pH 4.5 gradient (from 10 to 140 mM in 20 min, immediately followed by 140 to 300 mM in 10 min) in 25% ACN. Elution times for des-Ac P-endorphins 1-26, 1-27, and 1-31 were 32, 35, and 45 min respectively. The corresponding acetylated forms eluted much earlier. y L P H samples were resolved with a linear NH40Ac pH 4.5 gradient (increasing from 10 to 500 mM in 30 min) in 25% ACN. Mature y-LPH eluted a t 27 min, whereas the form containing a Lys-Arg carboxyl extension eluted at 32 min. Both gradients maintained a flow rate of 0.5 ml/min. Oneminute fractions were aliquoted into RIA tubes and lyophilized to dryness. 200 $1 of 3% NH,OH (final concentration in HzO) was added to each tube and samples were again lyophilized to dryness before assaying by RIA.
Site-directed Mutagenesis and Construction of VV:K233R-mPOMC"pMSKU16 containing the mPOMC cDNA (26) was partially digested with StuI endonuclease and protruding ends made blunt with T 4 DNA polymerase. The plasmid was digested to completion with EcoRI and cloned into M13-mp18 cut with BamHI (which was made blunt with T4 DNA polymerase) and EcoRI. The cDNA insert was now lacking all but 20 base pairs of its 5"untranslated sequences and was flanked by regenerated BamHI and EcoRI restriction sites. Single primer oligonucleotide-directed site-specific mutagenesis was performed as described (27) using a 15-mer oligonucleotide (5'-CACAAGAGGGGCCAC-3'). The mutagenized cDNA was excised with BumHI and EcoRI and cloned into the vaccinia insertion plasmid pVV3 (28). Marker rescue was performed as described (29) to construct the vaccinia expression vector VV:K233R-mPOMC.
SDS-PAGE of @-Endorphin Peptides-C4 column eluates were lyophilized to dryness and resuspended in 10 mM (pH 6.8 with Tris), 2.5% SDS, 5% 8-mercaptoethanol, and 0.02% bromphenol blue, heated 3 min in a boiling water bath, and resolved on a polyacrylamide slab gel by a modification of the method of Swank and Munkres (30). The separating gel contained 17% acrylamide/l.l% bisacrylamide, 0.1% SDS, 0.1 M H3P04 (pH 6.8 with Tris), 0.07% ammonium persulfate (w/v), and 0.7% TEMED. The stacking gel contained onehalf the concentrations of all components except for the ammonium persulfate and TEMED, which were unchanged. Electrode buffer was 0.1 M H3P04 (pH 6.8 with Tris), 0.1% SDS, and electrophoresis was carried out a t 10 V/cm. Lanes with molecular weight standards (tryptic fragments of myoglobin (Sigma) and synthetic @-endorphin) were stained in Coomassie Brilliant Blue R-250. Lanes with samples were cut into 2.5-mm thick slices and incubated 3.5 h a t 55 "C in 0.5 ml of RIA buffer I, vortexing occasionally. Aliquots were assayed directly for @-endorphin or @-MSH IR.

RESULTS
Expression of Mouse POMC in Heterologous Cells-A recombinant vaccinia virus which directs the synthesis of mouse proopiomelanocortin (mPOMC) was constructed by previously described marker rescue techniques (21). T o verify that this vaccinia recombinant (VVmPOMC) directs the synthesis of the prohormone, BSC-40 (a monkey kidney epithelial line) and Rin m5F (a rat insulinoma line) cell cultures were either mock infected, or infected with wild type vaccinia (VV:WT) Qr VVmPOMC at a m.0.i. of 1. Cultures were harvested 16 h after injection and @-endorphin immunoreactivity (IR) quantitated by RIA. Only the VV:mPOMC-infected cells produced significant levels of @-endorphin IR (1.2 and 0.5 pmol of /3endorphin IR/lOe cells in BSC-40 and Rin m5F cell extracts, and 4.5 and 1.2 pmol/1O6 cells in the culture medium of BSC-40 and Rin m5F cells, respectively). The authenticity of the prohormone synthesized in these cells was confirmed by SDS-PAGE of mPOMC which had been metabolically labeled with [3sSS]methionine and immunoprecipitated with an affinitypurified anti-ACTH antibody (Fig. IA). Both VV:mPOMCinfected BSC-40 and Rin m5F cells, but not the corresponding VV:WT-infected cultures, produced the same two size isoforms of the prohormone (32 and 36 kDa) as are found in the anterior pituitary corticotroph line, AtT-20. This correlates well with earlier studies that demonstrated two major size isoforms of mPOMC in uiuo, due primarily to differences in glycosylation (31,32). Correct removal of the 26 amino acid signal peptide (33) ("pre" sequence) was demonstrated by amino-terminal sequencing of 200 pmol of purified prohormone isolated from media of VV:mPOMC-infected BSC-40 cells (see "Materials and Methods"). The first nine amino acids of this precursor correspond to residues 27-35 of the preprohormone as deduced from the cDNA sequence (26) (Fig. 1B).
Targeting of mPOMC to Regulated Secretory Pathway-In AtT-20 cells, mPOMC-derived peptides are secreted via the regulated secretory pathway (34). Since localization of the precursor to this pathway appears to be required for correct processing to occur (7, 8), it was important to ascertain the targeting of the prohormone synthesized in cells infected with VV:mPOMC. The Rin m5F cell line provided a good model system for this study because it has a well-characterized regulated secretory pathway (35,36).
The effect of vaccinia infection on regulated secretion was first determined by examining the secretagogue-stimulated release of insulin from mock-, VVWT-, and VVmPOMCinfected Rin m5F cells (Fig. 2, A and B ) . Although the virus suppressed the absolute levels of released endogenous hormone, as compared to mock-infected cultures (approximately %fold, 16-h postinfection), the extent of stimulation (5.5-fold) by the phorbol ester PMA was not affected, demonstrating that the regulated secretory pathway remains functional in vaccinia-infected cells. T o determine whether mPOMC was correctly sorted into this pathway, secreted @-endorphin IR was also quantitated in this experiment (Fig. 2C). As expected, only the VV:mPOMC-infected cells produced significant levels. Additionally, a 2.7-fold increase in secretion of @-endorphin IR over control levels was observed in the presence of PMA, demonstrating that virally produced mPOMC can be targeted into the correct secretory pathway in a heterologous environment. When similar experiments were performed in BSC-40 cells, however, secretion of @-endorphin IR could not be stimulated with either PMA or forskolin (secretagogues acting via the protein kinase C and CAMP-dependent pathways, respectively) (data not shown). This suggests that a regulated secretory pathway is lacking in this cell line; thus mPOMC must exit BSC-40 cells by way of the constitutive pathway only.
Processing of mPOMC in Heterologous Cells-The profi- ciency of BSC-40 and Rin m5F cells in processing the prohormone was examined by fractionating extracts of VV:mPOMC-infected cultures on a C4 HPLC column (described under "Materials and Methods"). Since the processing of the endogenous mPOMC in AtT-20 cells has been well characterized (37,38), an extract of these cells was first run as a standard and column fractions assayed for both P-endorphin and (3-MSH IR (Fig. 3B). As there is considerable heterogeneity in the posttranslational modifications ( i e . glycosylation, phosphorylation, and sulfation) of the aminoterminal and ACTH domains of the precursor, analysis was greatly simplified by focusing on the products of the P-LPH domain of mPOMC. These are P-LPH, 7-LPH, and des-Ac P-endorphinl-sl in AtT-20 and anterior lobe corticotrophs, and r-LPH, Ac P-endorphinl-p7 and Ac P-endorphinl-26 in intermediate lobe melanotrophs (Fig. 3A).
The peptide which eluted at 48 min was identified as 6e n d~r p h i n~-~~ by coelution with synthetic camel P-endorphinl-sl from the C4 column. The P-MSH IR peak at 41 min was determined to be y-LPH (the only P-MSH sequence in mPOMC is located at the carboxyl terminus of -y-LPH). Since synthetic standards were not available for this peptide, it was identified by its apparent molecular weight of 4.6 kDa on an SDS-polyacrylamide gel (as compared to 4.4 kDa, the calculated molecular weight of mouse y-LPH) (data not shown). The peptide which eluted a t 55 min had both @-endorphin and 0-MSH IR and ran as a single species of molecular mass 8.2 kDa on SDS-PAGE, identifying it as P-LPH (calculated molecular mass of 8.1 kDa). The intact precursor eluted a t 69-71 min. SDS-PAGE analysis of these fractions demon-  Fig. 1 were present in this peak (data not shown). The p-MSH IR peptide eluting at 20 min was not identified in the present study. Using the vaccinia expression system, processing of mPOMC in BSC-40 and Rin m5F cells was next examined. Cells were harvested 16 h after infection with VV:mPOMC, extracts resolved by reversed phase HPLC with a C4 column, and fractions assayed for both p-endorphin and P-MSH IR. In agreement with our earlier findings with human proenkephalin (39), BSC-40 cells were found to be incapable of processing mPOMC to smaller p-endorphin-and p-MSHcontaining peptides (Fig. 3C) (21). In contrast, Rin m5F cells processed the precursor almost completely to peptides which coelute with authentic y-LPH and des-Ac &-end0rphinl-3~ (Fig. 30). The identities of these peptides were confirmed by cation exchange chromatography using AtT-20 y-LPH and synthetic P-endorphinl-s1 as standards (see "Materials and Methods").
Secretory Pathway Specific Processing of mPOMC-For both AtT-20 and pancreatic fl cells, it has been shown that processing of the endogenous prohormone occurs in the regulated secretory pathway (6,7). To demonstrate that mPOMC processing in Rin m5F cells similarly occurs in the regulated secretory pathway, we analyzed the mPOMC-derived products secreted in 1 h by these cells in the presence and absence of secretagogues (Fig. 4). In the absence of secretagogues, primarily intact precursor was found in the medium (Fig. 44).
In the presence of 100 nM PMA and 50 p M forskolin, processed forms (p-end0rphin~_3~ and y-LPH) were the principle secreted products (Fig. 4B). Importantly, a similar amount of unprocessed mPOMC was secreted from both control and stimulated cells. This demonstrated very efficient processing in the regulated secretory pathway of the Lys-Arg cleavage sites in the p-LPH domain of mPOMC. Limited, if any, proteolytic maturation of the prohormone could be detected in the constitutive secretory pathway. Cleavage Site Requirements-Because Rin m5F cells, like AtT-20 and anterior pituitary corticotrophs, synthesize primarily the nonacetylated, noncarboxyl shortened form of pendorphin (ie. p-end~rphin,-~l, by cleaving L y~"~-A r g 2~~ in the p-LPH domain of mPOMC), this cell line appears to be incapable of either N-acetylation or efficiently cleaving the L y~~~' -L y s '~~ target site near the carboxyl terminus of the precursor (Ly~'~-Lys'~ in p-endorphin, see Fig. 3A). There are several possible explanations for inefficient proteolysis of the Lys-Lys site, including: (i) proximity of the site to the carboxyl terminus (only two amino acids away) or other structure of the precursor may interfere with enzyme binding, or (ii) the composition of the paired basic amino acid target site itself may be critical for recognition by the processing endopeptidase. To address this question, site-directed mutagenesis was used on the mPOMC cDNA to substitute an arginine codon for the lysine codon corresponding to position 29 in pendorphin (LYS'~~ in mouse prePOMC), thereby changing the Lys-Lys cleavage site to Lys-Arg (see "Materials and Methods"). This mutant cDNA was cloned into the vaccinia expresmelanotrophs ( N I L ) . E , cell extract was prepared from a 6-cm plate of AtT-20 cells and resolved by C4 HPLC with the acetonitrile gradient in 0.1% trifluoroacetic acid as described under "Materials and Methods." One-minute fractions were collected and aliquots assayed for @-endorphin ( U ) and P-MSH (W) IR. C, a 15cm plate of BSC-40 cells was infected with VV:mPOMC at a m.0.i of 1. After a 16-h incubation at 37 " C , a cell extract was prepared and resolved on the C4 column as described above. D, A 15-cm plate of Rin m5F cells was infected with VVmPOMC and extracts analyzed as described above. Note that the unidentified @-MSH IR species in AtT-20 extracts eluting at 20 min is also produced by Rin m5F cells. cells were collected and analyzed as described in Fig. 2 and under "Materials and Methods." All experimental conditions were performed in triplicate and 8-endorphin IR quantitated in the medium from each of the culture dishes to verify reproducible stimulated secretion (data not shown). The medium from the three dishes of each experimental condition was pooled, resolved on the C4 column, and assayed for @endorphin (M) and P-MSH ( U ) IR as described in Fig. 3. A , control medium. B, medium containing 100 nM PMA and 50 p~ forskolin.
sion system, and the resultant virus (VVK233R-mPOMC) used to infect BSC-40 and Rin m5F cultures. As expected, only unprocessed K233R-mPOMC was found in BSC-40 cell extracts, coeluting with the native prohormone from the C4 column (data not shown). However, the peptide profile from extracts of VV:K233R-mPOMC-infected Rin m5F cells (Fig. 5A) was distinct from that of the native precursor. The y-LPH molecules synthesized from both the native and mutant precursors were identical, demonstrating consistent efficient cleavage at the Lys-Arg sites flanking this sequence in the prohormone. In contrast to the processing of native mPOMC, however, the p-endorphin IR peak from the C4 column shifted from 48 min ( & -e n d~r p h i n~-~~) for VV:-mPOMC (Fig. 3 0 ) to 51 min (the elution time of des-Ac pendorphinl-27) for VV:K233R-mPOMC. In addition, the 55 min peak (P-LPH) had more &endorphin than p-MSH IR. This is in contrast to the p-LPH peak obtained frox either AtT-20 cells or VV:mPOMC-infected Rin m5F cells, which had more p-MSH IR (compare Figs. 5A and 3 0 ) .
Several factors could account for the additional @-endorphin IR found in the K233R-mPOMC peak eluting at 55 min: (i) the p-endorphin antibody may have an increased affinity for the mutant p-LPH, or (ii) there may be a second pendorphin IR species eluting at the same position. To distinguish between these possibilities, an aliquot of the material eluting at 55 min was resolved by SDS-PAGE, and eluates of and 0-MSH (W) as described in Fig. 3. Arrows indicate the elution positions of: a, AtT-20 y-LPH; 6, P-endorphinl-al; c, fl-endorphinl_,,; and d, both 6e n d~r p h i n~.~~ and p-LPH. B , SDS-PAGE of IR peptide from VV:K233R-mPOMC-infected Rin m5F cells eluting at 55 min from the C4 column. An aliquot (about 1 ng) of the 55-min C4 column fraction was lyophilized to dryness, resolved by SDS-PAGE, and eluted from gel slices as described under "Materials and Methods." Aliquots of each eluate were assayed in duplicate for both P-endorphin (W-W) and 8-MSH (0-0) IR. Arrows indicate the migration of molecular weight standards, as determined by staining with Coomassie Brilliant Blue: a, 30 kDa; b, 17.2 kDa; c, 14.6 kDa; d, 8.2 kDa; e, 6.4 kDa; and f, P-endorphinl.n.i. When analyzed by this method, AtT-20 P-LPH migrates at position d.
gel slices were assayed for @-endorphin and @-MSH IR (Fig.  5B). Although there is a @-endorphin and @-MSH IR peptide, which migrates at the correct MW for 0-LPH, most of the @endorphin IR runs at the same position as @-endorphin. This peptide was tentatively identified as des-Ac since synthetic @-endorphinl-Z6 elutes at 55 min from the C4 column.
T o confirm the identities of both of the @-endorphin species synthesized from K233R-mPOMC in Rin m5F cells, these peptides were further analyzed by cation exchange chromatography. Again, the major and minor &endorphin IR species coeluted with synthetic des-Ac P-endorphinl-p7 and @e n d~r p h i n~_~~, respectively (data not shown), implying efficient proteolytic processing of the mutant Lys-Arg cleavage site.

DISCUSSION
In this study, a recombinant vaccinia virus was used to express mouse POMC in epithelial (BSC-40) and endocrine (Rin m5F) cell lines. Using this expression system, high levels of intracellular and secreted mPOMC (or mPOMC-derived peptides) were obtained only 16 h postinfection. In agreement with studies on human proenkephalin (39), BSC-40 cells were found to be incapable of storing mPOMC for stimulated secretion or of processing the precursor to smaller @-endorphinor @-MSH-containing peptides (Fig. 3C). Moreover, the prohormone secreted by these cells was not degraded in the culture medium, greatly facilitating large scale purification for sequence analysis. In addition, because synthesis occurred in mammalian, rather than in bacterial cells, this mPOMC should contain similar posttranslational modifications of the precursor as occur in uiuo, making it ideal for use as an authentic substrate for in uitro characterization of putative prohormone processing enzymes.
Other fibroblast and epithelial cell lines have also been found to be incapable of processing a variety of prohormones and have therefore been termed "maturation deficient" (40)(41)(42). It is interesting to note that, while only intact POMC is found intracellularly in gene transfer experiments with COS-1 (42), BSC-40, or P388D1 cells,'the protein secreted by these cell lines, with the exception of BSC-40, is degraded in the culture medium. These results emphasize a major advantage of the broad host range of a vaccinia expression system; a variety of different cell lines can be rapidly screened to identify those which express the heterologous protein in the desired manner.
In contrast to BSC-40 cells, the Rin m5F cell line was found to both efficiently target mPOMC into the regulated secretory pathway and process the prohormone to bona fide @-endorphin-and @-MSH-containing peptides (Fig. 3 0 ) ; thus, Rin m5F cells are "maturation proficient." Targeting of mPOMC to the regulated secretory pathway was determined by the extent of stimulation of secretion of @-endorphin IR (2.7-fold) by second messenger analogues. This stimulation, however, was somewhat less than was obtained for the endogenous hormone insulin (5.5-fold) (Fig. 2). There are several possible explanations for the apparent difference. (i) The two prohormones may be packaged into distinct vesicle populations, as has been observed in anterior pituitary somatomammotrophs (43), with the insulin-containing granules being more susceptible to stimulation. (ii) Since basal secretion consists primarily of intact precursor whereas media from stimulated cells contains mainly mature peptides, the observed differences in stimulation may reflect differential affinities of the @-endorphin and insulin antisera for precursors and processed products. (iii) Sorting into the regulated secretory pathway may be less efficient for mPOMC than for insulin. Indeed, a greater stimulation of insulin secretion can be obtained in AtT-20 cells stably transfected with the gene for human insulin than for the endogenous mPOMC-derived peptides (44,45). This has been attributed to a greater intrinsic "sorting index" of proinsulin than mPOMC. The relative "sorting indices" of mPOMC and proinsulin may thus carry over to Rin m5F cells, a system complementary to the AtT-20 studies.
HPLC analysis of the media from Rin m5F cells demonstrated stimulated secretion of only processed forms of mPOMC (Fig. 4). Because the C4 chromatogram of these secreted mPOMC-derived peptides reflects that of cell extracts, processing in the regulated secretory granules is implicated. Conversely, the major form of mPOMC secreted from control cells was unprocessed precursor, indicating that minimal maturation takes place in the constitutive secretory pathway. This apparent coupling of proteolytic processing and targeting to the regulated secretory pathway is analogous * B. A. Thorne and G . Thomas, unpublished results. to the sorting and secretion of mPOMC in .
These results are consistent with the observed processing of the endogenous proinsulin at Lys-Arg and Arg-Arg sites, as well as with results from expression of other heterologous prohormones in insulinoma cells (40,41,46). Although the precise site of cleavage within the target sites has yet to be determined for mPOMC in Rin m5F cells, correct removal of the extending Lys and Arg residues by an amino and/or carboxypeptidase must occur to produce both bona fide y L P H and P-endorphinl-al. The specificity of the Lys-Arg and Arg-Arg specific endoproteases thought to be involved in proinsulin processing in insulin-producing tumors would suggest that cleavage occurs on the carboxyl side of the second basic amino acid (47). The presence of high levels of carboxypeptidase H activity (or enkephalin convertase, an enzyme which specifically removes carboxyl extended basic amino acids (48)) in Rin m5F cells is consistent with this mechanism.* Processing of (3-LPH to @-endorphin and y L P H is much more efficient in Rin m5F cells than in anterior pituitary corticotrophs or AtT-20 cells (compare Fig. 3, D and B ) . In addition, preliminary results indicate efficient cleavage within the ACTH domain of the molecule. Thus, the proteolytic processing of mPOMC in Rin m5F cells is more reminiscent of processing in intermediate pituitary melanotrophs. In contrast to the intermediate lobe, however, which primarily synthesizes forms of P-endorphinl-*-i and P-endorphinl-26 by cleaving Lys232-Lys233 in mPOMC, the predominant form of P-endorphin IR produced by Rin m5F cells is P-endorphinl-31. To determine whether the sequence of the target site rather than structural constraints of the precursor precluded cleavage near the carboxyl terminus of the @-endorphin domain, Lys-Lys was replaced with Lys-Arg through the use of site-directed mutagenesis. Expression of the mutant precursor (K233R-mPOMC) in Rin m5F cells resulted in efficient synthesis of P-endorphinl-27, while very little /'3-end0rphin~-~~ accumulated in the cells (Fig. 5). This demonstrates that sequence of the paired basic amino acids, rather than just position of the potential cleavage site within the precursor, can influence prohormone processing.
In addition to P-endorphinl-2-i, about 10% of the processed peptide in VV:K233R-mPOMC-infected Rin m5F cells was in the form of P-endorphinl-*6. If cleavage indeed occurred on the carboxyl side of the Lys-Arg site, these results would indicate efficient removal of extending Arg and Lys residues, but inefficient removal of His (at position 27 in P-endorphin) by an endogenous carboxypeptidase H-like activity. This would correlate well with in vitro carboxypeptidase H activity studies, which demonstrate much less efficient (but detectable) removal of carboxyl terminal His versus Arg or Lys from synthetic substrates (49). In the intermediate pituitary and in several regions of the brain, however, the levels of the various forms of P-endorphinl-P7 and P-endorphinl_2s are very similar (50). The difference in the extent of carboxyl shortening of P-endorphin in these tissues as compared to Rin m5F cells may possibly be explained by a much longer exposure time of the mPOMC endoproteolytic cleavage products to carboxypeptidase H activity in vivo than in our transient expression system.
In conclusion, we have expressed mouse POMC in heterologous cell lines using a recombinant vaccinia virus. Maturation deficient BSC-40 cells provided the means for synthesizing sufficient levels of intact precursor to permit isolation for biochemical studies, whereas the maturation proficient Rin m5F cells were employed to demonstrate that composition of the paired basic amino acid target sequence can be a critical factor governing cleavage site utilization. Using this as a model system, the requirements of potential cleavage sites for recognition by processing enzymes and the role of specific prohormone sequences for targeting to the regulated secretory pathway can be more precisely defined.