Biochemical and Cell Biological Properties of the Human Prohormone Convertase 1/3 Ser357Gly Mutation: A PC1/3 Hypermorph

Abstract

Satiety and appetite signaling are accomplished by circulating peptide hormones. These peptide hormones require processing from larger precursors to become bioactive, often by the proprotein convertase 1/3 (PC1/3). Several subcellular maturation steps are necessary for PC1/3 to achieve its optimal enzymatic activity. Certain PC1/3 variants found in the general population slightly attenuate its enzymatic activity and are associated with obesity and diabetes. However, mutations that increase PC1/3 activity and/or affect its specificity could also have physiological consequences. We here present data showing that the known human Ser357Gly PC1/3 mutant (PC1/3^S357G) represents a PC1/3 hypermorph. Conditioned media from human embryonic kidney-293 cells transfected with PC1/3^WT and PC1/3^S357G were collected and enzymatic activity characterized. PC1/3^S357G exhibited a lower calcium dependence; a higher pH optimum (neutral); and a higher resistance to peptide inhibitors than the wild-type enzyme. PC1/3^S357G exhibited increased cleavage to the C-terminally truncated form, and kinetic parameters of the full-length and truncated mutant enzymes were also altered. Lastly, the S357G mutation broadened the specificity of the enzyme; we detected PC2-like specificity on the substrate proCART, the precursor of the cocaine- and amphetamine regulated transcript neuropeptide known to be associated with obesity. The production of another anorexigenic peptide normally synthesized only by PC2, αMSH, was increased when proopiomelanocortin was coexpressed with PC1/3^S357G. Considering the aberrant enzymatic profile of PC1/3^S357G, we hypothesize that this enzyme possesses unusual processing activity that may significantly change the profile of circulating peptide hormones.

Peptide hormones and neuropeptides are major hormonal regulators of hunger and satiety (1). Circulating neuroendocrine peptides need to be processed from larger precursors to become bioactive (2). The serine proteases prohormone convertase 1/3 (PC1/3) and prohormone convertase 2 (PC2) are two major processing enzymes for precursor proteins, and both are highly expressed in the nervous and in neuroendocrine systems (3).

Many studies have shown that certain heterozygote polymorphisms result in a mild PC1/3 hypomorph phenotype that is strongly correlated with obesity in ethnically diverse patient populations (4–6). The correlation between obesity and PC1/3 loss of function is further supported using a mutant mouse model. Mice expressing PC1/3 containing an Asn222Asp mutation exhibit obesity, hyperphagia, and defective proinsulin processing, mimicking the human phenotype (7). In contrast, rare but severe PC1/3 loss-of-function variants are associated with an autosomal recessive disorder that presents in neonates with severe generalized malabsorptive diarrhea and other subsequent age-dependent systemic endocrinopathies (8, 9).

The maturation steps of PC1/3 and PC1/3 activity are finely regulated subcellular processes wherein the cellular pH gradient plays a critical role (10, 11). PC1/3 is initially synthesized as an inactive proPC1/3 94-kDa zymogen, composed of a prodomain, a catalytic domain, a P domain common to all convertases (and thought to be involved in the control of catalytic activity), and a carboxyl-terminal domain, which is specific to each type of convertase. ProPC1/3 is rapidly converted into an 87-kDa PC1/3 by the autocatalytic removal of the N-terminal propeptide in the endoplasmic reticulum (12). C-terminal processing of PC1/3 to the 66-kDa carboxyl-terminal domainless form occurs mainly in the secretory granules and the trans-Golgi network (13). PC1/3 then enters into secretory granules, in which it is costored with its prohormone substrates. The removal of the C terminus strongly affects the kinetics, stability, and even the specificity of PC1/3 (14, 15).

PC1/3 was first cloned in 1990 from a mouse insulinoma library (16) and subsequently from a human pituitary cDNA library (17). The coding region of PC1/3 (mouse vs human) exhibits the highest homology (98%) within the catalytic segment of the molecule (residues 84–399) (17). Most of the biochemical characterization studies on PC1/3 have been carried out using mouse PC1/3 (18). In 1992, a human PC1/3 variant was cloned from lung tumor cells that contains a different residue at position 357 in the catalytic domain; glycine replaces serine (Ser357Gly) in this sequence (19). This variant (rs1050622) is exceedingly rare, with an allelic frequency in the general population of less than 0.00014 (20), and there are no reports about the effect of this substitution on the enzymatic activity of PC1/3. Moreover, functional assessment of all human variants reported thus far have used the clone containing the Ser357Gly polymorphism (4, 5, 8, 9, 22–25). We here present the enzymatic characterization of the human PC1/3 Ser357Gly variant (PC1/3^S357G). Our data indicate that this mutant possesses profound biochemical differences from the PC1/3^WT enzyme, which have the potential to cause a misbalance in peptide production relating to energy systems in patients bearing this mutation.

Materials and Methods

Expression vector construction/mutagenesis

A FLAG-tagged human PC1/3^S357G vector, a kind gift of J. W. Creemers (23) was used as a DNA template for site-directed mutagenesis to change the GGC(Gly) ≥ AGC(Ser) to obtain a wild-type PC1/3 (PC1/3^WT) (GenScript USA Inc). Both vectors were then subcloned by PCR using Pfu-Ultra HF DNA polymerase (Stratagene) using specific primers incorporating an XbaI restriction site followed by a Kozak translation initiation consensus sequence (GCCACC-ATG) at the 5′ end, and a XhoI restriction endonuclease site at the 3′ end following the stop codon. For the truncated human (h) PC1/3–66 kDa forms, a reverse primer containing a stop codon was inserted after residue 616, followed by an XhoI restriction endonuclease site. The amplified fragments were purified, digested with XbaI/XhoI and subcloned into the same sites of the pcDNA3.1(−) expression vector (Invitrogen). PC1/3^Q665E+S690T and PC1/3^S357A vectors were obtained via site-directed mutagenesis using the PC1/3^WT vector as a template (GenScript USA Inc). A human PC2 vector, a kind gift of Robert Day (University of Sherbrooke, Sherbrooke, Canada), was used as a template for PCR subcloning. The PCR subcloning method for PC2 was the same as used for PC1/3. This new PC2^WT vector, pCDNA3.1(−)/hPC2^WT, was also used as a DNA template for the site-directed mutagenesis (performed by GenScript USA Inc). All mutations were verified by sequencing of the entire PC1/3 or PC2 cDNA insert.

Cell transfection

Human embryonic kidney (HEK) or Neuro2A cells were transfected with plasmids encoding human PC1/3^WT or PC1/3^S357G (or their truncated forms) in triplicate wells using Fugene HD (Promega). After a 24-hour incubation in growth medium, Opti-MEM (Invitrogen) containing 100 μg/mL bovine aprotinin (Desert Biologicals) was added to each well. Cells were incubated for an additional 24 hours before conditioned medium and cells were harvested.

For constructing the Neuro2A stable expression of proCART-EGFPm, 2 × 10⁶ cells were transfected with 5 μg of proCART-EGFPm vector (26) and 15 μL of Fugene HD (Promega). After 48 hours of expression, the medium was replaced with fresh media supplemented with 0.5 mg/mL G418 sulfate. The selection media were replaced every week until colonies were detected. Screening of positive clones was performed using fluorescence microscopy for green fluorescent protein (GFP).

Enzyme assay

Enzymatic activity of secreted PC1/3 was measured in triplicate in 50-μL reactions in a 96-well polypropylene plate containing 25 μL of conditioned medium and final concentrations of 200 μM fluorogenic substrate (p-Glu-Arg-Thr-Lys-Arg-aminomethylcoumarin; Peptides International), 100 mM sodium acetate (pH 5.5), 2 mM CaCl₂, 0.1% Brij 35, and a protease inhibitor cocktail (final concentrations: 1 μM pepstatin, 0.28 mM L-1-tosylamino-2-phenylethylchloromethyl ketone [TPCK], 10 μM E-64, and 0.14 mM Nα-tosyl-L-lysine chloromethyl ketone [TLCK]). For pH and calcium experiments, the enzyme assay buffer pH and calcium concentration were altered, according to the figure legends. For PC2 assays, the pH buffer was 5.0. Reaction mixtures were incubated at 37°C and fluorescence measurements (380 nm excitation, 460 nm emission) were taken under kinetic conditions every minute for 150 minutes in a SpectraMax M2 microplate reader.

Western blotting

Proteins within the conditioned media were precipitated by adding 3 volumes of methanol and stored at −20°C. After thawing, samples were centrifuged at 14 000 rpm for 30 minutes at 4°C and the pellet resuspended in 100 μL sample buffer (SDS-PAGE sample buffer supplemented with β-mercaptoethanol and urea); 100 μL sample buffer was added directly to cell monolayers. All samples were boiled for 5 minutes before storage at −20°C. Cells and medium samples were thawed, reboiled, and subjected to SDS-PAGE followed by Western blotting using rabbit antiserum directed against the amino terminus of mature mouse PC1/3 (27). For hPC1/3 blotting, blots were then probed with antirabbit peroxidase antibody (31460; Thermo Scientific). For GFP blotting, anti-GFP antibody (sc-9996; Santa Cruz Biotechnology) and antimouse peroxidase antibody (A5278; Sigma) were used. For Western blotting of hPC2, two rabbit antibodies were used together: LSU7 [antiserum against the mouse PC2 mature N terminus, made against a peptide containing the first 10 amino acids of mature mouse (m) PC2 conjugated to keyhole limpet hemocyanin, as in the C-terminal PC2 antiserum described elsewhere (28)]; and LSU26 (antiserum against the mouse PC2 propeptide) (29). Visualization of immunoreactive protein was accomplished using the SuperSignal West Dura extended duration substrate kit (Thermo Scientific) and either film or a ChemiDoc MP imaging system (Bio-Rad Laboratories).

Pulse-chase/immunoprecipitation

For the pulsed samples, Neuro2A cells previously transfected with PC1/3WT or PC1/3S357G vectors were starved for 20 minutes in Cys- and Met-free medium (RPMI 1640; MP Biomedicals) and then incubated in labeling media (Cys and Met deficient medium with 0.25 mCi/well of ³⁵S-Met/Cys Translabel; MP Biomedicals) for 20 minutes. After rinsing the monolayers with normal RPMI 1640, cell lysates were extracted in boiling buffer (50 mM Na-phosphate, pH 7.4; 1% sodium dodecyl sulfate; 50 mM β-mecaptoethanol; 2 mM EDTA), boiled and diluted in immunoprecipitation (IP) buffer (0.1 M sodium phosphate; 1 mM EDTA; 0.1% Triton X-100; 0.5% Nonidet P-40; 1.0 mM phenylmethylsulfonyl fluoride; and 0.9% NaCl, pH 7.4). For the chased samples, the cells were rinsed, incubated with chase media containing normal levels of cysteine and methionine for 2 hours, and the medium collected and one fifth volume of 5× concentrated IP buffer lacking NaCl was added. For the chased cell lysates, the extraction procedure was identical to the pulsed samples. All samples were immunoprecipitated using N-terminally directed anti-PC1/3 antiserum (27) and protein A-Sepharose beads. Immunoprecipitated samples were separated by SDS-PAGE and radiolabeled PC1/3 was detected using a Storm Phosphoimager (Molecular Dynamics).

GH4C1 cell transfections and RIA

For cotransfection, 1.0 μg of hPC1 vector was transfected into each well of a six-well plate (500 000 cells) of GH4C1 cells in the presence of either 1.5 μg of rat proopiomelanocortin (POMC; cloned into pCDNA3.1) or a human proglucagon vector (pTT5-proglucagon, a kind gift of E. Lee, FivePrime Therapeutics, Inc). After 24 hours of incubation in growth medium, Opti-MEM containing 100 μg/mL bovine aprotinin was added to each well. Cells were incubated for an additional 24 hours before conditioned medium and cells were harvested. Media samples were centrifuged and supernatants saved frozen for later RIA analysis. Cells were extracted with 1 mL of 0.1 M HCl supplemented with 0.1 mg/mL BSA, collected by scraping, and centrifuged at 14 000 rpm for 5 minutes. The supernatant was lyophilized and resuspended in 1000 μL of Opti-MEM, centrifuged again, and the resulting clear supernatant saved for RIA analysis.

RIAs were carried out according to protocols described previously (30). For α-MSH assays, the polyclonal anti-α-MSH antiserum was commercially purchased from Chemicon and used at a final dilution of 1:90 000. This antibody is specific for α-MSH with no cross-reactivity with ACTH (31). Media and cell extracts were subjected to assay in duplicate. ¹²⁵I-Labeled α-MSH was prepared by the chloramine-T method originally described by Hunter and Greenwood (32). For determination of ACTH derived from POMC processing, samples were analyzed using a commercial ACTH ELISA kit (21-ACTHU-E01; ALPCO); this two-site assay exhibits no cross-reactivity with α-MSH. For glucagon assays, samples were subjected to assay in duplicate using a commercially available glucagon RIA kit (GL-32K; EMD Millipore). The antiglucagon antiserum used in this kit recognizes processed glucagon but not its precursors, proglucagon and oxyntomodulin. Samples were assayed using the RIA protocol according to the manufacturer's instructions. Radioactivity was determined using a Wallac 1470 Wizard γ-counter.

Results

Evolutionary PC1/3 data were available for analysis. Alignment of the primary amino acid sequences from various species showed that serine 357 (S357) in PC1/3 is completely conserved from Homo sapiens to Danio rerio (Figure 1A). Serine 357 is located between the oxyanion asparagine at residue 309 and the catalytic triad serine at residue 382 within the catalytic domain; no examples of PC1/3s containing glycine at 357 are found in the species sequenced to date.

Figure 1.

Serine 357 is conserved in PC1/3. A, Alignment of PC1/3s from several species shows that serine in position 357 is highly conserved. Amino acid sequences from the National Center for Biotechnology Information database used were as follows: Homo sapiens (NP_000430.3), Mus musculus (NP_038656.1), Rattus norvegicus (NP_058787.1), Danio rerio (NP_001131134.1), Gallus gallus (XP_003643108.1), Chrysemys picta bellii (XP_005284059.1), Xenopus tropicalis (XP_002933476.2), Pseudopodoces humilis (XP_005533963.1), and Tursiops truncatus (XP_004321922.1). For multiple PC1/3 sequence alignment, we used ClustalW2 web software (21). Asterisks show conserved residues among species. B, The PC1/3^S357G model is based on the known furin structure; this model was obtained courtesy of Than et al (33). The catalytic domain is represented in purple and the P domain in light blue. The catalytic triad (D167–H208-S382) is shown in black spheres and the oxyanion hole N309 with a black stick. G357 is shown with orange spheres. Two calcium ions are shown as yellow spheres. A tetrapeptide substrate is depicted in red.

A human PC1/3 model, constructed by Than and colleagues (33) and based on the structure of human furin (34) is shown in Figure 1B. The human PC1/3 amino acid sequence used to construct this model contained a glycine in position 357 (G357). In this model, G357 (shown in orange) exhibits a spatial location close to the catalytic triad (Figure 1B), but not within the substrate binding pockets; given the fact that PC2 contains an asparagine in this position, the PC2-like specificity acquired by the G357 mutant (vide infra) is difficult to explain and must await an actual prohormone convertase crystal structure.

Western blot experiments show that the replacement of S357 by glycine greatly increases the ratio of the PC1/3 (74 kDa + 66 kDa)/87 kDa forms present in conditioned media from Neuro2A cells transfected with hPC1/3-encoding vectors (Figure 2A₁), indicating increased autoconversion. Enzymatic behavior was also affected by this substitution. Both wild-type and mutant hPC1/3 exhibited normal exponential enzymatic behavior using the p-ERTKR-amc fluorogenic substrate (Figure 2). When the same amount of cDNA of each hPC1/3 vector was transfected, the enzymatic activity of the mutant was decreased compared with wild-type (Figure 2A₂); however, when corrected for total expression levels (Figure 2A₁), it is likely that the specific activity of PC1/3^S357G is enhanced relative to PC1/3^WT. In conditioned media obtained from transfected HEK293 cells, similar results were obtained; as expected, due to the lack of regulated secretory granules, in HEK cell media, the levels of the 66-kDa form of PC1/3^WT were undetectable (Figure 2B), whereas 66 kDa PC1/3 protein was always seen for PC1/3^S357G.

Figure 2.

PC1/3^S357G exhibits increased autoconversion to the 66-kDa truncated form. Samples from conditioned media obtained from transfected Neuro2A (A) or HEK293 (B) cells were collected and analyzed by Western blotting (top panels) and enzymatic assay (bottom panels). Cells were transfected with vectors encoding full-length PC1/3^WT, full-length PC1/3^S357G, or empty vector. The antiserum used for Western blotting recognizes all PC1/3 forms. Enzymatic activity assay shows AMC release with respect to time. The lane on the right shows recombinant mPC1 protein, used as a positive control. C, Neuro2A cells were transfected with vectors encoding full-length PC1/3^WT or full-length PC1/3^S357G, radiolabeled, and cell extracts and media subjected to immunoprecipitation. A 20-minute pulse of medium containing radioactive methionine followed by a 2-hour chase with medium containing cold methionine is shown. ProPC1/3 is detected at 94 kDa, PC1/3 at 87 kDa, and the truncated form of PC1/3 at 66 kDa.

Metabolic labeling of similarly transfected Neuro2A cells was also used to examine the synthesis of the two enzymes. A pulse-chase experiment confirmed the increased ratio of the PC1/3 (74 kDa + 66 kDa) to 87 kDa forms, not only in the extracellular medium obtained from PC1/3^S357G-expressing cells but also in cell extracts (Figure 2C). The prior subcellular maturation step converting the proPC1/3 zymogen (94 kDa) to PC1/3 (87 kDa) was identical for both PC1/3 enzymes, suggesting that intramolecular propeptide cleavage is not affected by the Ser357Gly substitution.

The 66-kDa form of PC1/3 is known to be much more active than the parent 87-kDa form of PC1/3 (14). In consideration of the large differences in the maturation rates of the wild-type and mutant PC1/3s, we constructed truncated 66-kDa forms to study the enzymatic profile of the Ser357Gly mutation in a manner independent of autoconversion. Both PC1/3 truncated forms were constitutively secreted from HEK293 cells and conditioned medium was used for enzymatic assays. PC1/3^WT-66 kDa exhibited a linear enzymatic profile, achieving a steady state after a short lag phase (10 min). However, PC1/3^S357G-66 kDa exhibited a continuously exponential enzymatic profile (Figure 3A; Western blots are shown below each panel in Figure 3).

Figure 3.

PC1/3^S357G shows exponential enzyme kinetics and alterations in pH and calcium requirements. A, Samples from conditioned media HEK293 cells were collected and analyzed by Western blotting and enzymatic assay. HEK293 cells were transfected with vectors encoding the PC1/3^WT truncated form or the PC1/3^S357G truncated form. The conditioned media were used at pH 5.5 and 2 mM calcium. An enzymatic activity assay shows AMC release with respect to time. B, For pH experiments, the buffer used was 100 mM Bis-Tris + 100 mM sodium acetate, and the different pHs were obtained using 5 N acetic acid. Calcium was fixed at 2 mM in all buffers. C, For calcium experiments, 0.9 mM EDTA (final concentration) was added to conditioned media for neutralization of the calcium present in the medium. The calcium curve was then made in regular reaction buffer at pH 5.5. AMC release per minute was determined at 60 minutes of enzymatic reaction. A′, B′, and C′, Expression levels of truncated forms of PC1/3 (Western blotting). Samples were obtained from the experiments shown in panels A, B, and C, respectively.

A possible explanation for the greater amount of autoprocessing in the mutant PC1/3 observed in cells is an alteration in the enzyme pH optimum. PC1/3^S357G-66 kDa activity exhibited a neutral pH optimum: PC1/3^WT-66 kDa showed the normal pH optimum of 5.5, whereas PC1/3^S357G-66 kDa activity peaked at 6.5–7.0 (Figure 3B). At all levels of pH studied, PC1/3^S357G-66 kDa exhibited a higher enzymatic activity than wild-type PC1/3. Additionally, we tested the calcium dependence of activity at pH 5.5. Both enzymes exhibited strong calcium dependence; however, PC1/3^S357G-66 kDa functioned well at only 8 μM calcium [∼3 pmol/min, 7-amino-4-methylcoumarin (AMC) release]; PC1/3^WT-66 kDa required more than 2 mM calcium levels to achieve a similar performance (Figure 3C).

The LLRVKR peptide is a relatively specific competitive inhibitor for PC1/3 (35). When PC1/3^S357G-66 kDa was preincubated with this inhibitor peptide, enzymatic activity was inhibited but not to the extent of the wild-type enzyme (Figure 4). For PC1/3^WT-66 kDa, only 15 nM LLRVKR were necessary to decrease the enzymatic rate (Figure 4A), whereas for PC1/3^S357G-66 kDa, 60 nM LLRVKR was necessary to inhibit enzyme activity at similar PC1/3 amounts and times (Figure 4B). The truncated forms used in these assays are shown in Figure 4C.

Figure 4.

PC1/3^S357G shows increased tolerance to inhibitor peptide. Samples from conditioned media of HEK293 cells were collected and analyzed by Western blotting and enzymatic assay. The HEK293 cells were transfected with vectors encoding either the PC1/3^WT truncated form (A) or the PC1/3^S357G truncated form (B). A and B, Enzymatic activity assays show AMC release with respect to time. Samples were preincubated with LLRVKR at different concentrations prior to adding the fluorogenic peptide substrate (pERTKR-AMC). The standard buffer containing pH 5.5 and 2 mM calcium was used. C, Expression levels of the truncated forms of PC1/3 used in this experiment (Western blotting of parallel reaction mixtures).

To analyze the effect of the S357G mutation on processing prohormone substrates, we initially chose proCART as a PC1/3 processing model. proCART is a natural substrate of both PC1/3 and PC2 in neuronal and endocrine cells (36). We stably transfected Neuro2A cells with proCART fused to enhanced green fluorescent protein (EGFP) (proCART-EGFP) and used these stable cells for PC1/3 cDNA transient transfections. Processing sites described for proCART consist of a monobasic site (R₉) and two dibasic sites, KR₄₁ or KK₄₇ (36). Because processing occurs within the proCART N-terminal portion of the fusion protein, the EGFP tag remains attached to the resulting products. When PC1/3 was transiently expressed, GFP-immunoreactive detection showed three bands: the precursor (full length proCART_1–89-EGFP), 37 kDa; another band of 36 kDa (proCART_10–89-EGFP processed at the monobasic site); and a third band of approximately 33 kDa (proCART_42–89-EGFP and/or proCART_48–89-EGFP processed at dibasic sites) (Figure 5C). We could not discriminate between these smaller products. The 37-kDa species was always the predominant form, and processed forms were not detected after transfecting with empty vector. However, when PC1/3^WT was expressed, proCART-EGFP processed forms were present in the conditioned media (36 kDa and 33 kDa). Interestingly, when PC1/3^S357G was expressed, only 33 kDa proCART-EGFP was robustly shown in the conditioned media (Figure 5A, top panel). Curiously, in the same samples, Western blotting using anti-PC1/3 antiserum showed only PC1/3^WT expression (Figure 5A, bottom panel).

Figure 5.

A proCART fusion protein is processed more extensively in Neuro2a cells by PC1/3^S357G. Neuro2A/proCART stable cells were transfected with cDNAs encoding PC1/3^WT or PC1/3^S357G full-length (A) or their respective truncated forms (B). The negative control used was empty vector and the positive control was an mPC2 vector. Conditioned media were used for Western blotting analysis. Upper panels show the expression levels of proCART-EGFPm forms by Western blotting using an antibody against GFP. Bottom panels show the expression levels of PC1/3 by Western blotting. C, proCART-EGFP processing scheme. WB, Western blot.

When we used the PC1/3–66 kDa truncated forms for the same experiment, Western blotting using anti-GFP antiserum replicated the results observed with the PC1/3 full-length experiments (Figure 5B); PC1/3^S357G-66 kDa exclusively produced the 33-kDa proCART processed form. As a positive control, we tested the effect of the PC2 activity in these cells. When a mouse PC2-expressing vector was transfected, the resultant PC2 activity exclusively produced the 33-kDa proCART processed form, similar to PC1/3^S357G (Figure 5B, top panel). In this series of experiments, it was possible to detect the PC1/3 66-kDa forms, confirming their similar expression levels (Figure 5B, bottom panel). Figure 5C summarizes the processing of this fusion protein by the two enzymes.

Considering the PC1/3^S357G PC2-like activity observed during proCART processing, we decided to explore other possible alterations in substrate specificity caused by this mutation. POMC is extensively processed by PC1/3 and PC2 (37). α-MSH is an anorexigenic peptide produced from POMC only via PC2 activity (37). For this experiment, we examined α-MSH production in a somatotroph cell line, GH4C1, that expresses neither POMC nor PC2 and exhibits only low expression of PC1/3 (38). A rat POMC vector was coexpressed with the various PC1/3 vectors, and 48 hours after transfection, media and cell extracts were analyzed using a RIA specific for α-MSH. The results showed that PC1/3^S357G significantly increases α-MSH production (Figure 6A). In both cell extracts and media, the α-MSH levels were increased 2–3 times when the PC1/3^S357G enzyme was expressed, as compared with PC1/3^WT. When PC1/3^WT was expressed, the α-MSH level was indistinguishable from the empty vector control.

Figure 6.

POMC and, to a lesser extent, proglucagon are more extensively processed to α-MSH and glucagon by PC1/3^S357G. GH4C1 cells were cotransfected with either POMC or proglucagon with the PC1/3 full-length enzymes indicated. Cell extracts and conditioned media were analyzed for α-MSH (A), ACTH (B), or glucagon (D). Data were analyzed using a one-way ANOVA followed by a post hoc Bonferroni test. *, P <.05; **, P < .01; ***, P < .001. Panels C and E show POMC (C) and proglucagon (E) processing schema.

We also analyzed ACTH levels using a specific two-site ELISA kit (ALPCO). Conversely to α-MSH, ΑCΤΗ levels were decreased when the PC1/3^S357G enzyme was expressed (as compared with PC1/3^WT) (Figure 6B). Thus, in this system, PC1/3^S357G also exhibits a PC2-like specificity. These results are summarized in Figure 6C. Glucagon is another peptide hormone thought to be produced exclusively by PC2 (39, 40). In GH4C1 cells we observed a small increase in stored cellular glucagon but not in conditioned media, when proglucagon was coexpressed with PC1/3^S357G rather than with PC1/3^WT (Figure 6D). Glucagon production was significantly increased independently of the PC1/3 type used, as compared with the empty vector control (Figure 6D). A diagram summarizing glucagon cleavage is shown in Figure 6E.

In addition, we reanalyzed one of the most common human polymorphisms contributing to human obesity, the tandem single-nucleotide polymorphisms rs6234 (Gln665Glu) and rs6235 (Ser690Thr), in the wild-type PC1/3 context (ie, lacking Ser357Gly) to compare with previous data. Enzymatic data were obtained using conditioned medium from cells transfected with PC1/3^WT vs PC1/3^Q665E-S690T. In Neuro2A and HEK293 cells, PC1/3^WT and PC1/3^Q665E-S690T showed identical enzymatic behaviors (Figure 7), with no significant difference in specific activity in either Neuro2A cell media (PC1/3^WT: 1.02 ± 0.75 pmol/min per PC1/3 expression unit; PC1/3^Q665E/S690T: 0.68 ± 0.13 pmol/min per PC1/3 expression unit; mean ± SD, n = 3. Figure 7A) or in HEK293 cell media (PC1/3^WT: 2.77 ± 0.63 pmol/min per PC1/3 expression unit; PC1/3^Q665E/S690T: 1.91 ± 0.20 pmol/min per PC1/3 expression; mean ± SD, n = 3 wells; Figure 7B).

Figure 7.

PC1/3^Q665E+S690T and PC1/3^WT show identical enzymatic activity in both Neuro2A and HEK293 cells. Samples of conditioned media obtained from transfected Neuro2A (A) or HEK293 (B) cells were collected and analyzed by Western blotting and enzymatic assay. Cells were transfected with vectors encoding full-length PC1/3^WT, full-length PC1/3^Q665E+S690T, or empty vector. The antiserum used for Western blotting recognizes all PC1/3 forms. Enzymatic activity depicts AMC release with respect to time. A′ and B′, Expression levels of full-length PC1/3^WT or PC1/3^Q665E+S690T (Western blotting). Samples correspond to the experiments shown in panels A and B, respectively.

To determine the specific contribution of the glycine residue to PC1/3^S357G activity, we also tested a Ser357Ala mutant in enzymatic assays. In HEK293 cell-conditioned media, PC1/3^S357A showed significantly less enzymatic activity than PC1/3^WT (PC1/3^WT: 4.28 ± 0.8 pmol/min per PC1/3 expression unit; PC1/3^S357A: 1.42 ± 0.33 pmol/min per PC1/3 expression unit; t test; **, P < .01, mean ± SD, n = 3 wells; Figure 8A). Furthermore, similarly to PC1/3^WT, only the 87-kDa form was present in media obtained from PC1/3^S357A-transfected cells (Figure 8A′). Because the hypermorph phenotype observed using the Ser357Gly substitution was not replicated using Ser357Ala, we interpret this to mean that a glycine residue is required.

Figure 8.

Glycine specificity of the Ser357Gly mutation: Ser357Ala decreases PC1/3 activity, whereas the Asn357Gly mutation increases PC2 activity. A, Samples of conditioned media obtained from transfected HEK293 cells were analyzed by Western blotting and enzymatic assay. Cells were transfected with vectors encoding full-length PC1/3^WT, full-length PC1/3^S357A, or empty vector. The antiserum used for Western blotting recognizes all PC1/3 forms. Enzymatic activity depicts AMC release with respect to time. Panel A′, Expression levels of full-length PC1/3^WT or PC1/3^S357A (Western blotting); samples correspond to the experiments shown in panel A. Panel B, Samples of conditioned media obtained from transfected Neuro2A cells were collected and analyzed by Western blotting and enzymatic assay. Cells were transfected with vectors encoding either full-length PC2^WT, full-length PC2^N357G, or empty vector. The antisera combination used for Western blotting recognizes all PC2 forms. Enzymatic activity depicts AMC release with respect to time. Panel B′, Expression levels of full-length PC2^WT or PC2^N357G (Western blotting); media and lysate samples correspond to the experiments shown in panel B.

We then extrapolated these findings to PC2. Wild-type PC2 contains an asparagine residue in this position; we therefore constructed Asn357Gly mutant protein (PC2^N357G). Neuro2A cells were transfected with PC2 ^WT or PC2^N357G, and conditioned media were analyzed using a PC2 enzymatic assay (Figure 8B). Interestingly, a robust increase in the enzymatic activity of PC2^N357G was observed over PC2^WT (Figure 8B′). PC2^N357G-transfected cells also showed an increased ratio of mature 66 kDa PC2 to the 76-kDa proPC2 form in the conditioned medium; however, this was not evident in the lysate samples. Note that these enzymatic data have not been normalized to PC2 expression due to the significant differences in proPC2/PC2 processing (Figure 8B′).

Discussion

We here describe biochemical and cellular differences exhibited by the human PC1/3^S357G mutation. We find that this PC1/3^S357G mutant exhibits profound differences with respect to wild-type PC1/3, both in the posttranslational activation events occurring during transit through the regulated secretory pathway and in its general enzymatic properties, including specificity.

Many prior studies of human PC1/3 have used the Ser357Gly mutant as a wild-type prototype to study mutants associated with human pathologies (4, 5, 9, 20, 22–24, 33). However, the effect of this particular mutation on PC1/3 performance has apparently never been examined. Among the various human prohormone convertases (furin, PC1/3, PC2, PC4, PC5, PC7, and PACE4 (paired basic amino acid cleaving enzyme 4), a structure-based amino acid sequence alignment shows that only PC2 and PC7 have an amino acid other than serine in this position (asparagine and glycine, respectively) (33). When the amino acid sequences of PC1/3s present in the various species cloned to date are compared, residue 357 is always serine. Therefore, we assume that PC1/3^S357G should be considered a mutant rather than an innocuous variant.

Our studies show that PC1/3^S357G exhibits a higher conversion rate into the PC1/3–66-kDa form. Although metabolic labeling indicates that proPC1/3 (94 kDa) conversion to PC1/3 (87 kDa) is identical between PC1/3^WT and PC1/3^S357G, the subsequent C-terminal maturation step is clearly enhanced in PC1/3^S357G during its transit through the secretory pathway. In addition, Western blotting of samples from conditioned media always showed that PC1/3^S357G exhibits greatly increased conversion into the PC1/3–66-kDa form, independently of whether the cell line used contains a regulated secretory pathway. This observation is also evident in the literature in descriptions of human PC1/3^S357G expression patterns in both neuroendocrine and constitutive cells, although conversion to the smaller form is more robust in endocrine cell lines containing a regulated secretory pathway (4, 5, 9, 20, 22–24). On the contrary, wild-type PC1/3, when expressed in cells lacking a regulated secretory pathway, does not generate (or generates very small quantities) the PC1/3–66-kDa form (41). Indeed, PC1/3 normally requires cell types with a regulated secretory pathway for adequate intracellular substrate and self-processing (42). The greatly enhanced conversion of PC1/3^S357G into the 66-kDa form observed here is an unusual subcellular process in that it occurs independently of the presence of secretory granules.

Working with the PC1/3–66-kDa truncated forms permitted us to compare the enzymatic behaviors of PC1/3^WT and PC1/3^S357G without the complication of this differential processing. Truncated PC1/3–66 kDa is clearly the more active and unstable form of PC1/3 (14, 15). We observed that the 66-kDa truncated form of wild-type human PC1/3 exhibits linear enzymatic behavior after achievement of the steady-state phase, an observation previously reported for the same form of mouse PC1/3^WT (43). Interestingly, PC1/3^S357G-66 kDa exhibits quite altered enzymatic behavior in that it continues to show exponential kinetics similar to the parent 87-kDa form. Thus, it is impossible to directly compare kinetic parameters such as Vmax between PC1/3^S357G-66 kDa and PC1/3^WT-66 kDa. However, at the end of our long (2 h) enzymatic assays, PC1/3^S357G-66 kDa always yielded more fluorescent product than PC1/3^WT-66 kDa. Thus, PC1/3^S357G exhibits a novel kinetic behavior that is distinct from PC1/3^WT.

We found that human PC1/3^WT-66 kDa achieves its maximal enzymatic activity at pH 5.5 and requires millimolar calcium levels; similar characteristics have previously been reported for mPC1/3–66 kDa (14, 43, 44). The pH within the secretory granules is thought to be between 5.0 and 5.5 (45) and to contain greater than 20 mM calcium (46); this is the organelle in which PC1/3 completes its activation and is costored with its prohormone substrates (11). However, the mutant PC1/3^S357G-66 kDa achieves its maximal enzymatic activity at a much more neutral pH, 6.5, and requires only micromolar calcium levels. These findings suggest that PC1/3^S357G may be able to operate efficiently in earlier compartments than the wild-type enzyme. We speculate that PC1/3^S357G could be activated to the truncated form earlier than wild-type enzyme, beginning its conversion to the PC1/3–66 kDa form during transit through the endoplasmic reticulum and Golgi and subsequently initiating prohormone substrate processing prior to entering secretory granules.

The fact that it possesses a neutral pH optimum supports the idea that the PC1/3^S357G mutant could function prematurely within the endoplasmic reticulum and subsequent compartments. It has been previously shown that the formation of the 66-kDa form of PC1/3 can occur in endoplasmic reticulum extracts at pH 5.5 and 10 mM calcium (11). PC1/3 exhibits a highly sophisticated pH-dependent activation mechanism, which takes place during transit through the regulated secretory pathway (10); the mutant PC1/3^S357G would be expected to lose this compartment-specific activation control. In addition, the increased tolerance of PC1/3^S357G for the LLRVKR peptide, a PC1/3-specific competitive inhibitor (35), could indicate yet an additional level of regulatory loss. This peptide inhibitor mimics normal substrate dibasic sites and is present in the C-terminal sequence of proSAAS, a potent endogenous PC1/3 inhibitor (47, 48). When proSAAS is overexpressed in AtT-20 cells, it reduces PC1/3 conversion to the fully mature 66-kDa form (47, 49). We speculate that PC1/3^S357G may be more resistant to regulation by proSAAS within the regulated secretory pathway. Collectively, our findings support the idea that the Ser357Gly substitution likely alters the subcellular regulation of PC1/3 at many levels that are critical for proper peptide hormone production and secretion.

The aberrant biochemical profile of PC1/3^S357G is concordant with the changes in prohormone processing specificity detailed in this report. PC1/3^S357G exhibits a PC2-like activity in proCART, POMC, and to a minor extent, in proglucagon. For proCART, PC1/3 activity is mainly associated with monobasic processing, yielding an intermediate proCART form (9 kDa); PC2 cleaves at dibasic sites within proCART, yielding the anorexigenic cocaine- and amphetamine regulated transcript (CART) peptide (∼ 5 kDa) (36). Full-length PC1/3^S357G and the 66-kDa truncated form exclusively produce the anorexigenic CART peptide, similar to PC2; however, the PC1/3^WT full-length and the 66-kDa truncated species produce the intermediate and anorexigenic CART peptide forms. These results also support the idea that the PC1/3 C-tail domain itself does not influence cleavage specificity but robustly affects enzymatic efficiency, reducing substrate processing (15, 50). Thus, PC1/3^S357G may be considered a hypermorph.

Curiously, when the PC1/3^S357G full-length vector was transfected into Neuro2A/proCART-EGFPm stable cells, we could not detect its expression in conditioned medium by Western blotting. However, when the PC1/3^S357G-66 kDa truncated vector was transfected, the truncated enzyme was detected normally. Of note, this phenomenon was observed only in Neuro2A/proCART-EGFPm-transfected cells and was not observed in untransfected Neuro2A cells. It has been reported that the PC1/3 C-tail domain influences the subcellular targeting of PC1/3 (51–53) and is able to modulate both activity (15, 54) and stability (14). In addition, it has been suggested that more efficient processing of PC1/3 into the 66-kDa truncated forms, observed in certain PC1/3 variants associated with obesity, could affect the availability of PC1/3 activity (7, 38). We surmise that when PC1/3^S357G is expressed in the context of stable Neuro2A/proCART-EGFPm cells, in which the enzyme is coexpressed with a large quantity of substrate, its maturation to the 66-kDa form is accelerated, and it experiences a dramatic change in intracellular stability. However, further experiments will be necessary to test this idea.

The PC2-like substrate specificity profile that PC1/3^S357G exhibits for proCART is also observed for other prohormone substrates. We found that PC1/3^S357G, but not PC1/3^WT, robustly increases α-MSH levels in POMC-cotransfected GH4C1 cells. However, α-MSH normally arises from processing of POMC exclusively via PC2 activity (13, 37). GH4C1 cells exhibit poor α-MSH production when POMC is overexpressed, either alone or in the presence of PC1/3^WT. These findings, which support the specificity of the RIA, are in concordance with the inability of PC1/3 to produce α-MSH (37, 55) and the extremely low PC1/3 expression found in GH4C1 cells (56). The large increase in α-MSH levels observed in the presence of PC1/3^S357G provides additional support for the idea that PC1/3^S357G acquires PC2-like specificity. In addition, the concomitant decrease in ACTH and increase in α-MSH levels produced by PC1/3^S357G activity strongly suggest that α-MSH production is favored over ACTH production. This enzymatic behavior could cause a drastic change in the balance of satiety/appetite peptidergic signals in patients carrying the Ser357Gly mutation. α-MSH is a potent anorexigenic peptide at a central level (57), and a patient with high circulating α-MSH levels could potentially present hypophagia. Unfortunately, a clinical phenotype of patients with this mutation is not available.

A third example of altered specificity is glucagon, which is processed from proglucagon, mainly via PC2 activity (39, 40). GH4C1 cells generate only small amounts of glucagon when proglucagon is overexpressed in the absence of convertases. We found that transfection of both PC1/3^S357G and PC1/3^WT yielded a significant amount of immunoreactive glucagon; because PC1/3^S357G was more efficient than PC1/3^WT in generating this peptide, this again supports a PC2-like specificity profile. However, these glucagon results also suggest that PC1/3^S357G does not entirely mimic PC2 specificity. The small increase in glucagon compared with the large increase in α-MSH produced by PC1/3^S357G suggests another level of substrate specificity and supports the idea that the production of only certain peptides would be affected by the PC1/3^S357G substitution in a patient bearing this mutation.

Collectively these findings support the idea that PC1/3^S357G lacks normal subcellular regulation and will produce anomalous substrate processing. This mutation was originally described in a human bronchial carcinoid tumor (19), which represents a tumor of neuroendocrine origin; since this original report, PCSK1 expression has been confirmed in human bronchial tumors (58) as well as in certain, but not all, small-cell lung cancers and cell lines (59, 60). Patients with bronchial carcinoid tumors can sometimes present with Cushing's syndrome due to excessive ACTH production (see reference 61 and references therein); by enhancing the cleavage of ACTH to α-MSH, the Ser357Gly mutation would be expected to be protective against this phenomenon. However, the frequency of occurrence and the possible physiological effect of the Ser357Gly mutation in patients bearing bronchial carcinoid tumors is currently unknown.

Because mild impairment in PC1/3 enzymatic function is strongly associated with human obesity in a variety of different genetic association studies (4–6), one prediction arising from the identification of the Ser357Gly variant as a PC1/3 hypermorph is that this allele might be found in greater frequency in lean vs obese cohorts. To our knowledge, this prediction has not yet been tested. Given its low allelic frequency, large cohorts could be required to obtain statistically significant results; however, a positive result could support the idea that obesity is directly related to the enzymatic function of this important precursor processing enzyme.

A previous study did not find significant differences in the enzymatic activity of the two most common obesity PC1/3 polymorphisms (the tandem variant Gln665Glu/Ser690Thr) in the standard fluorogenic assay using human PC1/3 constructs in the Ser357Gly background (5). In this study we show that even when these polymorphisms are expressed in a true wild-type context, there is no significant difference between wild-type and Gln665Glu/Ser690Thr PC1/3 in fluorogenic substrate hydrolysis. These results are consistent with those found by Mbikay et al (38), who studied this tandem polymorphism (also in a true wild type vector background) and found that neither POMC processing nor PC1/3 processing was affected. It is possible that future experiments using other precursor substrates will illuminate substrate-processing differences which clarify the genetic association of the Gln665Glu/Ser690Thr polymorphism with obesity. In addition, it will be necessary to systematically reexamine other known human variants for possible interactions with the Ser357Gly mutation because it is likely that other variants (especially those in domains other than the C terminal tail) may exhibit differential processing when analyzed in a true wild-type background (8, 9).

To understand the specific contribution of the glycine residue to the Ser357Gly effect, we also examined the activity of the homologous Ala mutation. However, the Ala mutation resulted in the exact opposite effect of the Gly mutation: a robust decrease in PC1/3 activity (and no effect on PC1/3 C-tail processing). Interestingly, the Gly hypermorph effect could even be extrapolated to PC2: substituting a Gly for the Asn which is normally present at the homologous position, also resulted in an increase in activity. We speculate that the lack of a side chain at this particular position (ie, glycine) may confer additional conformational flexibility to residues near the catalytic site, thus resulting in increased enzymatic activity.

In conclusion, we have shown here that the alteration of a single residue within the PC1/3 sequence is able to radically shift the enzymatic properties of this enzyme to a less enzymatically stringent phenotype. We speculate that the resulting shift in substrate specificity may have physiological consequences in the processing of peptide hormone precursors and a potential phenotype of leanness; the test of this speculation must await the identification of a human patient in which precursor processing can be analyzed.