Production of Bioactive Peptides in an In Vitro System

Abstract

An in vitro system for the preparation of bioactive peptides is described below. This system couples three different post-translational modification enzymes, prohormone convertases (PCs), carboxypeptidase E, and peptidyl α-amidating enzyme, to transform recombinant precursors into bioactive peptides. Three different precursors, mouse proopiomelanocortin (mPOMC), rat proenkephalin (rPE) and human proghrelin were used as model systems.

The conversion of mPOMC and rPE to smaller peptide products was measured by radioimmunoassay. After optimization of the system, excellent efficiency was obtained: about 85% of starting mPOMC was converted to des-acetyl α-MSH. For proenkephalin, 75% and 96% yields were obtained for the opioid peptides Met-RGL and Met-enk respectively. Cell-based assays demonstrated that in vitro-generated des-acetyl α-MSH successfully activated the melanocortin 4 receptor. Proghrelin digestion was used to screen the specificity of PC cleavage as well as to confirm the cleavage site by mass spectroscopy. Mature ghrelin was produced by human furin, mouse prohormone convertase 1, and human prohormone convertase 7, but not by mouse prohormone convertase 2.

These results demonstrate that our in vitro system 1) can produce peptides in quantities sufficient to carry out functional analyses; 2) can be used to determine the specificity of proprotein convertases on recombinant precursors; and 3) has the potential to identify novel peptide functions on both known and orphan GPCRs.

Introduction

Peptide hormones are synthesized as relatively large precursors that require activation via post-translational processing events. The initial processing event is endoproteolytic cleavage at single/double basic residues by prohormone convertases (PCs), calcium-dependent subtilisin-like serine proteases in the secretory pathway [1,2]. Intermediate-sized peptides generated by PC cleavage then undergo exoproteolytic cleavage at C-terminal basic residues by carboxypeptidases E (CPE; ref.[3,4])). In certain cases, additional modifications, such as C-terminal amidation, N-terminal acetylation, pyroglutamylation, and/or sulfation are required after proteolysis for the peptides to become biologically active [5-8]. While amidation is common, with about 50% of bioactive peptides being amidated, other modifications occur less frequently [9,10].

Recent searches for novel bioactive peptides have involved reverse pharmacological approaches to deorphanize specific GPCRs [11-13]. In this method, peptide ligands for a specific receptor are purified from tissue extracts using competition in assays against an orphan GPCR and a radioactive ligand, usually a drug. However, this method requires some advance knowledge of the tissue localization of the bioactive peptide as well as the appropriate radioactive ligand. Since genomic DNA sequencing has been completed in many species [14-16] , another strategy has been to use bioinformatics analyses to predict novel peptides from novel precursor sequences [17,18]; peptide bioactivity has then been tested using synthetic peptides [19,20]. However, it is often unclear whether the activity of a novel peptide is regulated by post-translational modifications such as amidation or N-terminal cyclization. More importantly, there is no completely accurate algorithm to predict which single/double basic sites within a precursor will actually be cleaved [17,18]. For instance, the presence of obestatin, a potentially bioactive peptide consisting of 23 amino acids (∼2.5 kDa) within proghrelin, was predicted from cleavage site usage [19]. However, questions remain regarding the actual bioactivity of this peptide; recent reports do not support the idea that the obestatin peptide activates the specific receptor, GRP39, for which it was described to be a ligand [21,22]. Therefore, a need exists for a suitable process to generate physiologically relevant forms of novel peptides. In this paper, we present the development of an in vitro system for the preparation of bioactive peptides using recombinant precursors.

Materials and Methods

Preparation of recombinant precursors

The preparation of rat proenkephalin (rPE) from Chinese hamster ovary (CHO) cell-conditioned medium has been described previously [23]. For overexpression of mouse POMC (mPOMC) and human proghrelin in E.coli, the coding region for each gene (lacking the signal peptide) was inserted into the pQE-30 vector (Qiagen, Chatsworth, CA) and pProEX (Invitrogen, Carlsbad, CA), respectively. Recombinant precursors were synthesized as His-tagged proteins and then purified on Ni-resin (Sigma, Saint Louis, MO). Proteins were eluted with a buffer containing 20 mM Hepes (pH 7.4), 150 mM NaCl, and 500 mM imidazole. For His-tagged proghrelin, TEV protease (Invitrogen) digestions were carried out to remove the His-tag overnight at 4°C in 500 ul reactions containing phosphate buffer (pH 7.5) and 10 units of TEV protease. The reaction mixture was then subjected to Ni-resin chromatography to remove His-tagged proteins with the untagged precursor remaining unbound. The flow-through was collected and concentrated using a Microcon ultrafiltration device with a YM-3 membrane (Millipore, Bedford, MA). For POMC, the eluted protein from the Ni resin purification was purified by reverse-phase chromatography on a C4 column (GRACE Vydac, Hesperia, CA) with an acetonitrile gradient in 0.1% trifluoroacetic acid, then lyophilized and resupended in PBS. The yield of each purified protein was 2 mg non-His-tagged proghrelin and 1.4 mg His-tagged POMC per 250 ml of culture. Proteins were stored at -20°C until used.

In vitro modification reactions

Recombinant mouse PC2 (mPC2) was purified from the conditioned medium of overexpressing CHO cells by the procedures described previously [23]. Human CPE (hCPE) and human peptidyl α-amidating enzyme (hPAM) were a kind gift of R&D Systems (Minneapolis, MN), and Unigene (Fairfield, NJ), respectively; pilot experiments were performed with amidating enzyme provided by Dr. Betty Eipper (Univ. Connecticut). Two micrograms of either rPE, mPOMC and ACTH (1-39) were digested with mPC2 (1 μg) in 50 μl reactions containing 100 mM sodium acetate (pH 5.0), 5 mM CaCl₂ and 0.1% Brij-35 (Sigma, Saint Louis, MO) at 37°C for 0, 1, 2, 4, 6 and 16 h. In the subsequent CPE reaction to trim off C-terminal basic residues, 10 μl of a mixture containing 100 ng hCPE, 100 mM sodium acetate (pH 5.0), 5 mM CaCl₂ and 12 mM CoCl₂ (2 mM final concentration) was added to the mPC2 reaction. The CPE reaction was then carried out for 6 h at 37°C in 60 ul. In order to amidate ACTH (1-15), the CPE reactions were diluted to 400 ul with PAM reaction buffer containing 700 ng hPAM, 150 mM MES (pH 6.0). 2.5 μM CuSO_4, 0.1 mg/ml catalase (Roche, Indianapolis, IN), and 0.5 mM ascorbate. The amidating reaction was then performed at 37°C for 3 h. The conversion of precursors to mature peptides was measured by radioimmunoassay (RIA) with α-MSH polyclonal antiserum (Chemicon, Temecula, CA), or with Met-enk [24] or Met-RGL polyclonal antisera [25]. Data were analyzed using GraphPad PRISM 4 (GraphPad Inc., San Diego, CA), and the data were fit to a sigmoidal dose-response curve. Additionally, 10-fold scaled up reactions were performed to generate sufficient des-acetyl α-MSH (1.6 kDa) for testing bioactivity. After the modifying reactions, des-acetyl α-MSH was isolated on a C18 column in order to verify product identity by co-elution with a des-acetyl synthetic standard (des-acetyl α-MSH; Sigma, Saint Louis, MO). The des-acetyl α-MSH concentration was measured using RIA, and the products stored at -20°C until further analysis. The mass of des-acetyl α-MSH was confirmed by using MALDI-TOF mass spectroscopy. The spectra were acquired in linear mode on an Applied Biosystems Voyager System 1181. A total of 200 scans were averaged for each spectrum using an accelerating voltage 20 kV, 94% grid voltage and 0.05% guide voltage. Alpha-cyano-4-hydroxycinnamic acid (Aldrich, Milwaukee, USA) was prepared as a saturated solution in 50% acetonitrile and 0.1% TFA, and used as the matrix for this analysis. For the acquisition of spectra from 500-5000 Da, 2 ul of sample was mixed with 2 ul of alpha-cyano-4-hydroxycinnamic acid solution, and 1 ul of the mixture was loaded onto the MALDI plate. Des-acetyl α-MSH (M + 1620.71) purchased from Sigma was used as the control peptide for this experiment.

cAMP assay

HEK-293 cells stably expressing human MC4 receptor (hMC4-R) were a kind gift of Dr. Roger Cone [26]. The cells were plated at 5×10⁴ cells per well into 96-well plates. cAMP assays were carried out approximately 18 h after plating. The cells were stimulated with 0.1, 1.0, 5.0, 10, and 100 nM des-acetyl α-MSH produced via our method in 50 μl assay buffer (DMEM containing 10 mM Hepes, pH 8.0, 0.5 mM 3-isobutyl 1-methylxanthine, and 0.1 mg/ml BSA) for 2 h at 37°C. The Parameter™ cyclic AMP immunoassay kit (R&D systems, Minneapolis, MN) was used to determine the level of intracellular cAMP stimulated by des-acetyl α-MSH. Cells were lysed in 150 ul of kit lysis buffer, and 100 ul of cell lysate was used to measure the cAMP levels. Data were analyzed using GraphPad PRISM 4, and the data were fit to a sigmoidal dose-response curve.

Proghrelin digestion

We purified mouse PC1 (mPC1) and human furin (hfurin) from conditioned media taken from CHO-overexpressing cells using procedures described previously [27-29]. Human PC7 (hPC7) was a kind gift of R&D systems (Minneapolis, MN). Proghrelin (2 μg) digestion was carried out with 2.5 units of mPC1, mPC2, hPC7, and hfurin in a 50 μl reaction. The reactions contained 100 mM sodium acetate (pH 5.5 for PC1 or pH 5.0 for PC2), 5 mM CaCl₂ and 0.1% Brij-35. For furin, 100 mM Hepes, pH 7.0, 5 mM CaCl₂ and 0.1% Brij-35 was used, and for PC7 the reaction buffer contained 20 mM Bis-Tris, pH 6.5, 1 mM CaCl₂ and 0.1 % Brij-35. Reactions were incubated at 37°C for the indicated times. One unit of PC activity is equal to the amount of the enzyme that is required to cleave 1 pmol/min of the pRTKR-aminomethyl coumarin (Peptides International, Lexington, KY) fluorogenic substrate. After the reactions, the products were separated on 15% SDS-polyacrylamide gels and stained with Coomassie blue.

Identification of proghrelin-derived peptides

Mass spectroscopy was used to identify peptide products obtained from proghrelin. The process of proghrelin digestion was as described above. For mass spectroscopy, reactions were terminated by adding TFA to a final concentration of 0.1%. The mixtures were then purified using C18 ZipTip™ columns (Millipore, Bedford, MA), vacuum-dried, and resupended in 20 μl of 5% acetonitrile and 5% formic acid. The sample solutions were directly infused at a flow rate of 700 nL/min into a LTQ mass spectrometer (Thermo Electron, San Jose, CA), equipped with a nano-electrospray interface. The instrument was operated under the data-dependent acquisition mode. The five most intense peaks were selected from the precursor ion scan and fragmented automatically via collision-induced dissociation (CID). The normalized collision energy was set at 35% with the activation Q set at 0.25. The tandem mass spectra of peptides were processed by the “Qual Browser” in “Xcalibur 1.4 SR1” (Thermo Electron, San Jose, CA). The fragment ions on the peptide tandem mass spectra were manually annotated based on the theoretical fragment masses and charge states generated by “Mass-Product” in “ProteinProspector” [30].

Results

Preparation of bioactive peptides from rPE and mPOMC

Since PC2 activity was inhibited by both cobalt and ascorbate, which are essential ingredients in the CPE and PAM reactions respectively [31,32], we could not perform simultaneous reactions, ie endoproteolytic, exoproteolytic and amidation reactions in a single tube. Therefore, each modification step was first optimized as shown in Fig. 1. Bioactive peptides from POMC and rPE were produced by sequential reactions, as shown in Figure 1. Figure 2 depicts the efficiency of the production of des-acetyl α-MSH, Metenk-RGL, and Met-enk from their respective precursors. Approximately 85% of ACTH and mPOMC were converted to des-acetyl α-MSH (935 and 170 nM in 400 ul reactions, respectively). In addition, 75% and 96% yields were obtained for Met-RGL (875 nM) and Met-enk (4.9 μM, 4 copies in rPE) from rPE in 60 μl reactions. One of the goals of our research was to produce sufficient quantities of peptide for bioactivity testing; our yield of des-acetyl α-MSH in the ten-fold scaled-up reaction, 680 pmol (4 ml of 170 nM peptide), appears to be ample for this purpose. In order to confirm that in vitro synthesized peptides show bioactivity against GPCRs, in vitro synthesized des-acetyl α-MSH was applied to HEK 293 cells expressing the hMC4-R and activation assessed using a cyclic AMP assay (Fig. 3). The results showed that des-acetyl α-MSH generated via our method could successfully stimulate the accumulation of intracellular cAMP level via hMC4-R activation in a dose-dependent manner. The EC₅₀ value calculated with the dose-response curve was 3.3 nM, which approximates values obtained previously [26,33]. Additionally, MALDITOF mass spectroscopy was used to confirm the molecular mass of des-acetyl α-MSH (Table 1). The mass was determined as M + 1620.4, similar to that of the standard peptide des-acetyl α-MSH (M + 1620.71; Sigma). These results demonstrate that the accumulation of cAMP was induced through hMC4-R activation by des-acetyl α-MSH.

Figure 1. The process of in vitro preparation of bioactive peptides.

Figure 2. Time course of liberation of α-MSH, Met-RGL, and Met-enk from their precursors.

Two micrograms of mPOMC and rPE were sequentially digested with mPC2 and hCPE, respectively. In additionally, hPAM reactions were performed using ACTH- and POMC-derived peptides to generate des-acetyl α-MSH. The details of each reaction were described in Materials and Methods. Peptide products were measured by specific RIAs. Panel A, Met-enk; panel B, Met-enk-RGL; panels C and D, des-acetyl α-MSH derived from ACTH and POMC, respectively. Each sample was tested in duplicate at each time point. Results are shown as the percentage of maximal conversion.

Figure 3. α-MSH generated by in vitro processing can efficiently activate the MC4-receptor.

HEK-293 cells expressing MC4-receptors were treated with different concentrations of des-acetyl α-MSH produced by our method. After 2 h, the levels of intracellular cAMP were measured using the Parameter™ cyclic AMP immunoassay kit. Each sample was tested in duplicate at each concentration.

Table 1.

Comparison of expected and observed molecular masses of generated peptides

precursor	peptides	M.W. (Da)
		expected	observed
human proghrelin
	Gly-Ala-Met-Ghrelin-Arg	3503.01	3505.1
	Obestatin-Gly-Lys	2546.84	N.Da
	C-terminal proghrelin- derived peptide	7268.01	7290.5
mouse POMC
	des-acetyl α–MSH	1622.79	1620.4

^a“N.D.”: not detected by mass spectroscopy

Determination of the specificity of PC cleavage on proghrelin

The specificity of PC cleavage on proghrelin was examined using mPC1, mPC2, hPC7 and hfurin. The results showed that mPC1, as well as hfurin and hPC7, could cleave proghrelin to produce mature (non-octanoylated) ghrelin (Fig. 4B). However, the mPC2 reaction did not generate any digestion products, even in the presence of 15 units of mPC2 (data not shown), indicating that this precursor does not serve as a substrate for mPC2. Using a constant amount of enzymatic activity (as assessed against a fluorogenic substrate) of hfurin, mPC1, and hPC7 for proghrelin digests, mPC1 and hfurin exhibited almost equal efficiency, but hPC7 was much less active. Interestingly, a 2.8 kDa product with a size corresponding to obestatin, previously identified as another potentially bioactive mature peptide within proghrelin [19], was never observed in Coomassie-stained gels (Fig. 4B). Both His-tagged and non-His-tagged precursors were equally well cleaved (data not shown).

Figure 4. Proghrelin is cleaved by multiple PCs, but not by PC2.

Panel A: Alignments of proghrelin. Pro; prosequence. Panel B: Digestion of His-tagged proghrelin (2 μg) was carried out with indicated enzymes. All samples were electrophoresed on the same gel; some lanes are omitted for clarity. The cleavage products are indicated with asterisks.

To clarify the exact peptides generated by PC cleavage, mass spectroscopy analyses were carried out on furin-treated proghrelin. We observed that only two products, ghrelin and a C-terminal peptide, were generated in the reaction. Therefore, only one cleavage site was used; this was determined to be LQPR↓ALAG (Fig. 4A, 5 and Table 1). Again, a peptide with the mass of obestatin was not observed, indicating that it is unlikely to be generated by prohormone convertases.

Figure 5. PC cleavage of proghrelin produces ghrelin, but not obestatin.

(A) Tandem mass spectrum of [Ghrelin 1-31]⁴⁺ at m/z 877.45 (predicted average m/z 877.02): (B) Tandem mass spectrum of [Ghrelin 32-97]⁵⁺ at m/z 1458.6 (predicted average m/z 1458.4). All major peaks correspond to the peptide bond cleavage; b ions are fragments containing peptide N-termini while y ions are those containing peptide C-termini. Tandem mass spectra confirm that PCs cleave proghrelin between Arginine 31 and Alanine 32, and yield a smaller fragment of 3505.1 Da and a larger fragment of 7290.5 Da.

Discussion

The data presented here demonstrate that we have successfully processed specific prohormone precursors to mature peptides using an in vitro system. Since ligands for receptors are normally active in the nanomolar range, the amounts of product generated by the above reaction should be sufficient to carry out further studies to test ligand and receptor matching. For example, the quantity of peptides produced by our method (4 ml of 170 nM des-acetyl α-MSH) can activate 136 wells of receptor-containing cells, given the use of 50 ul/well of 100 nM peptide.

Identification of accurate precursor cleavage sites is critical to a comprehensive understanding of how bioactive peptides are generated from precursors by regulated proteolysis. Although several programs exist for PC cleavage site prediction [17,18], a high prediction accuracy of proteolytic cleavage sites is still lacking because of the current lack of experimentally verified sites. For example, when the cleavage sites within POMC were predicted using the ProP 1.0 server (http://www.cbs.dtu.dk/services/ProP/) and NeuroPred (http://neuroproteomics.scs.uiuc.edu/neuropred.html), which represent the algorithms commonly used to predict the cleavage sites at single/double basic residues, quite different results are obtained (Table 2; [34-38]). Not only do the predictions differ from each other, but they also differ significantly from experimentally verified cleavage sites. These discrepancies support the continuing need to obtain accurate cleavage information experimentally for accurate synthesis of novel peptides. The collection of experimentally determined PC cleavage sites will also assist in building better algorithms in the future.

Table 2.

Differential prediction of cleavage sites in POMC using two different algorithms.

Cleavage site P1↓′1	Algorithms		Comments	Experimentally proofed In vivo
	ProP	NeuroPred
RACK↓LDLS		Y		Never reported
NPRK↓YVMG			POMC (1-49)/γ-MSH	Cleaved by PC2 only Ref. 36
FGPR↓NSSS		Y		Never reported
AQRR↓AEEE	Y	Y	γ-MSH/JP	PC1 and 2 Ref. 36, 38
EGKR↓SYSM		Y	JP/α-MSH	PC1 and 2 Ref. 36, 38
RWGK↓PVGK		Y		Never reported
GKKR↓RPVK		Y	α-MSH/CLIP	Favored by PC2 Ref. 36, 38
KKRR↓PVKV		Y		Never reported
RPVK↓VYPN		Y		Never reported
EFKR↓ELEG	Y	Y	CLIP/γ-LPH	Favored by PC1 Ref. 34, 36
DAEK↓DDGP			β-LPH/β-MSH	Never reported
KDKR↓YGGF	Y	Y	γ-LPH/β-endo	Favored by PC2 Ref. 34, 36
AHKK↓GQ			C-terminal to β-endo	Cleaved by PC2 only Ref. 35, 36, 38

ProP 1.0 server (ProP) and NeuroPred were used to predict the cleavage sites in POMC. “Y” indicates the predicted cleavage sites using each program.

Digestion of proghrelin with multiple PCs demonstrated that in vitro digestion can be used to screen the relative specificities of various PCs on recombinant peptide precursors. Our data also indicate that both bacterially-expressed as well as CHO cell-expressed POMC were well-recognized by PC1 and PC2 (data not shown), indicating that bacterially-expressed precursors supply sufficient structural information for accurate cleavage. The information obtained from these in vitro experiments should be helpful in understanding how specific processing enzymes regulate peptide content as well as their associated biological activities. In addition, our in vitro approach is able to provide information about potential convertase activity on precursors for which little processing data is available. For example, we found that several convertases could process proghrelin, with furin and PC1 being the most active enzymes. In confirmation, Zhu X et, al [39] have recently reported that PC1 must be the natural enzyme which produces mature ghrelin within the endocrine cells of the stomach, since PC1 knockout mice do not produce this peptide.

We were unable to observe the production of obestatin using any convertase. The production of this peptide might require the activity of other proteases, perhaps from blood, since the accumulation of C-terminal fragments of ghrelin in plasma has been reported [40]. However, it is unlikely that the amidation reaction could take place in the circulation, since peptide amidation of peptide hormones is commonly thought to occur prior to secretion.

Given the large number of precursors emerging from genome sequencing efforts, increasing efforts are now being devoted to combinatorial and collective approaches to the analysis of peptide function [14-16]. Reverse pharmacological methods have been used to identify the ligands for orphan receptors [11-13]. However, this approach requires purifying bioactive molecules from tissue extracts, a laborious procedure. Additionally, solid phase peptide synthesis is also commonly used to synthesize bioinformatically-predicted bioactive peptides. This approach is likely to be suitable to prepare known bioactive peptides, but, as mentioned above, the biochemical information to predict accurate processing sites is still lacking, and erroneous predictions can result in the synthesis of unnatural peptides. An alternative in vitro method has been reported which employs a combination of cell-free translated precursors and subcellular fractions to generate the final peptide products; this method was also shown to be capable of producing GPCR-activating peptides [41]. However, disadvantages of the cell-free translation approach include the potential destruction of peptide products by the microsomal membrane fraction; and the low yield, only about 0.7 pmol for the 50 ul reaction (about 100-fold lower than our system [41]).

In conclusion, we suggest that our in vitro system could potentially be useful in the preparation of bioactive peptides for functional studies as well as for peptidomics research. The use of the highly limited-specificity prohormone convertases to process precursors (as opposed to more promiscuous proteolytic enzymes such as trypsin) will ensure “restriction processing” of precursors only at natural cleavage sites. In combination with other tools, such as powerful high-throughput in vitro protein expression [42-44], and analysis systems such as mass spectroscopy and cell-based assays, large numbers of novel, physiologically relevant peptides can theoretically be produced. We are now attempting to add other common recombinant modifying enzymes, such as glutaminyl cyclase, in our processing reactions in order to refine our system to generate the widest possible complement of physiologically-relevant peptides from each recombinant precursor.