The Neuropeptide Processing Enzyme EC 3.4.24.15 Is Modulated by Protein Kinase A Phosphorylation*

The metalloendopeptidase EC 3.4.24.15(EP24.15) is a neuropeptide-metabolizing enzyme expressed predominantly in brain, pituitary, and testis, and is implicated in several physiological processes and diseases. Multiple putative phosphorylation sites in the primary sequence led us to investigate whether phosphorylation effects the specificity and/or the kinetics of substrate cleavage. Only protein kinase A (PKA) treatment resulted in serine phosphorylation with a stoichiometry of 1.11 ± 0.12 mol of phosphate/mol of recombinant rat EP24.15. Mutation analysis of each putative PKA site, in vitro phosphorylation, and phosphopeptide mapping indicated serine 644 as the phosphorylation site. Phosphorylation effects on catalytic activity were assessed using physiological (GnRH, GnRH1–9, bradykinin, and neurotensin) and fluorimetric (MCA-PLGPDL-Dnp and orthoaminobenzoyl-GGFLRRV-Dnp-edn) substrates. The most dramatic change upon PKA phosphorylation was a substrate-specific, 7-fold increase in both K m andk cat for GnRH. In both rat PC12 and mouse AtT-20 cells, EP24.15 was serine-phosphorylated, and EP24.15 phosphate incorporation was enhanced by forskolin treatment, and attenuated by H89, consistent with PKA-mediated phosphorylation. Cloning of the full-length mouse EP24.15 cDNA revealed 96.7% amino acid identity to the rat sequence, and conservation at serine 644, consistent with its putative functional role. Therefore, PKA phosphorylation is suggested to play a regulatory role in EP24.15 enzyme activity.

Intracellular communication is a vital regulator of the fundamental processes of metabolism, growth, and differentiation in all organisms. Neuropeptides are involved in autocrine, paracrine, and endocrine signaling allowing cells to communicate without necessarily requiring close synaptic proximity (1). Neuropeptides are unique from other neurotransmitters in that peptides lack classical reuptake mechanisms for recycling components into the cell and terminating action. Instead, neuropeptide-metabolizing enzymes are required to extinguish the signaling action of neuropeptides. The zinc metalloendopeptidase EC 3.4.24.15 (EP24.15) 1 exhibits characteristics of both metabolizing and processing enzymes, and has multiple peptide substrates. One substrate, GnRH, is of critical physiological importance in reproduction. Inhibition of EP24.15 activity has been demonstrated in vivo in rat models resulting in an increased half-life of the hormone by decreased GnRH degradation and subsequent augmentation of the luteinizing hormone surge (2)(3)(4). Other important EP24.15 substrate targets include neurotensin, where inhibition of EP24.15 in mice prolonged forepaw licking latency (5), bradykinin (6), somatostatin [1][2][3][4][5][6][7][8][9][10][11][12][13][14] (7), and nociceptin (8). EP24.15 also processes Met-and Leu-enkephalin from the enkephalin-containing peptides (9), and the specific EP24.15 inhibition increased Metenkephalin antinociception in rodents (10). The enzyme has also been implicated in regulation of the cleavage of amyloid-␤ peptide (11), the aberrant processing of which has been linked to Alzheimer's disease pathogenesis.
The regulation of EP24.15 action upon its substrates is achieved by unique elements such as thiol activation (12). Another regulatory mechanism may be phosphorylation, which represents an important means of neural extracellular signal transduction and biological response modulation (13). Indeed, modulation of protease activity by phosphorylation has been demonstrated in proteasomes (14) and more recently, in the caspase family of proteases (15,16). Of particular interest, treatment of rat pheochromocytoma cells (PC12) by a cAMP analogue decreased the specific activity of soluble EP24.15 (17). This would suggest modulation of EP24.15 protein by PKA phosphorylation, or by another kinase activated through PKA (18,19). Indeed, the amino acid sequence of EP24.15 contains PKA, CKII, and PKC consensus phosphorylation motifs (20), suggesting that the enzyme may be a kinase substrate in mammalian cells. To build on the indirect observation of possible kinase influences on EP24.15 activity in rat PC12 cells (17), we sought to determine the role of phosphorylation upon EP24. 15, specifically examining its effect on neuropeptide hydrolysis. Similarly, it would be important to determine if phosphorylation is a conserved event in other neuroendocrine/peptide hydrolysis model systems, such as in AtT-20 mouse pituitary cells.

EXPERIMENTAL PROCEDURES
Materials-Reagents were purchased from Sigma unless otherwise noted.

SDS-and Native-PAGE
Samples for SDS-PAGE were heated in 2 ϫ sample buffer at 65°C for 5 min. The proteins were separated on an 8% SDS-polyacrylamide gel as described by Laemmli (23). Native gels were run similarly except SDS and ␤-mercaptoethanol were omitted from both the sample buffer and the polyacrylamide gel, and samples were not heated. After electrophoresis, the gels containing radiolabeled EP24.15 were exposed to film or to a phosphor screen (Molecular Dynamics, Sunnyvale, CA) for quantitation.
For PKC-PKC was purified from Rat 6 fibroblast cell lines overexpressing PKC ⑀ (calcium independent) or PKC ␤ (calcium dependent) (gift of Dr. Robert Krauss, Mount Sinai School of Medicine) and assays were performed as described elsewhere (24,25), using 3 g of EP24.15; each condition was assayed in triplicate. EGF receptor peptide (RKRTLRRL) served as positive control.
Each kinase was incubated with or without 200 M ATP (800 M ATP for PKC assays) (Roche Molecular Biochemicals, Indianapolis, IN), with the addition of trace amounts (0.04 mCi, 3000 Ci/mmol) of [␥-32 P]ATP (New England Nuclear, Boston, MA). Under the assay conditions, there was no incorporation in the absence of kinase. Reactions for time course experiments were terminated by the addition of equal volumes of 0.5 mM EDTA (to prevent chemical, non-enzymatic phosphorylation from occurring upon heating) (26), sample loading buffer containing ␤-mercaptoethanol, and heating to 65°C for 10 min. Samples were then electrophoresed on an 8% SDS-PAGE gel, dried, and exposed to film or to a phosphorscreen for PhosphorImager analyses.

Quantitation of Incorporated Phosphate
Prior to inactivation of the kinase assays (described above), a 2-l aliquot was spotted onto P81 cellulose phosphate paper (Life Technologies, Grand Island, NY) (representing total counts). The sample was allowed to dry, washed with 75 mM phosphoric acid (specific incorporation) (4 ϫ 25 ml, where no more label was eluted), counted, and moles of phosphate incorporated per mole of EP24.15 Ϯ S.E. calculated. To ensure phosphorylation saturation, controls included: decreasing substrate concentration, increasing kinase concentrations, increasing ATP concentration, and examining time course reactions by PhosphorImager analyses for time-dependent saturation of signal. There was no incorporation in the absence of kinase. Additionally, after saturation of phosphate incorporation (90 min, 10 units of PKA), an additional 10 units of PKA was added, demonstrating that saturation was not due to kinase depletion.

Kinetic Determinations Using the Fluorimetric Substrates QFS and QF7
EP24.15 enzymatic activity was determined under discontinuous assay conditions with the quenched fluorescent substrate QFS (27) and QF7 (28), as described previously with modifications. The non-phosphorylated EP24.15 enzyme (control) used for all kinetic determinations underwent identical kinase reaction conditions (described above), except that ATP was excluded. All determinations were done using two independent protein preparations and two independent phosphorylation reactions. Total substrate hydrolysis was less than 10%. 6.8 ng of EP24.15 (either phosphorylated or non-phosphorylated) was incubated at 37°C with varying amounts of QFS or QF7 (4.4 -17.6 M) in a final volume of 635 l. Kinetic parameters (K m , V max , k cat , and k cat /K m ) were evaluated using the double-reciprocal plot method of Lineweaver and Burk (29).

Determination of Enzyme Inhibitor Constants for cFP-AAF-pAB
The inhibition constant of EP24.15 for the specific active site-directed inhibitor, cFP-AAF-pAB, was determined with 25 ng of EP24.15 (either phosphorylated or non-phosphorylated) incubated at 37°C with QFS (4.4 M final concentration) and varying concentrations of cFP-AAF-pAB (0 -100 nM) in reaction buffer (125 mM NaCl, 0.3 mM dithiothreitol, 25 mM Tris-HCl, pH 7.5) in a final volume of 635 l. Reactions were terminated after 30 min by the addition of 115 l of 0.5 M sodium formate, pH 3.5. EP24.15 activities were determined as described above and the inhibition constant or K i for phosphorylated and non-phosphorylated enzyme was evaluated using the method of Dixon (30).

CNBr Cleavage of Phosphorylated EP24.15
In vitro phosphorylated EP24.15 was separated from free [␥-32 P]ATP by Sephadex G75 gel filtration (Amersham Pharmacia Biotech, Piscataway, NJ). Crystalline cyanogen bromide was prepared in 70% trifluoroacetic acid, and cleavage performed (34). Cleavage fragments were chromatographed on an analytical (0.6 ϫ 50 cm) Bio-Gel P10 Fine (Bio-Rad) column eluted isocratically in 0.1 M ammonium bicarbonate (pH 8.0) under denaturing conditions at 0.5 ml/min, 0.5-ml fractions were collected and radioactivity quantitated in a scintillation counter. Mass estimates of cleavage fragments were deduced by a plot of the relative elution constant (v e /v o ) versus log(MW) of calibrated molecular weight standards prior to, and after the sample was chromatographed.

Proteolytic Cleavage of Phosphorylated EP24.15 and MALDI-TOF Mass Spectrometry
Trypsin and endoproteinase Lys-C (Roche Molecular Biochemicals, Indianapolis, IN) were reconstituted and incubated (as per the manufacturer's instructions) with in vitro phosphorylated EP24.15. The digested (phospho)peptides were desalted using a ZipTip (Millipore, Bedford, MA), divided into aliquots, and resuspended in 5 l of 10 mg/ml ␣-cyano-4-hydroxycinnamic acid in 50% acetonitirile, 0.1% trifluoroacetic acid, with angiotensin as an internal standard. To further enhance detection of phosphorylated peptides, a portion of the peptide mixture was dissolved in a 1:1 1 mM ammonium citrate/matrix solution (35). Analyses were performed at the Howard Hughes Medical Institute, Columbia University Protein Facility (directed by Dr. Mary Ann Gawanowicz) on a MALDI-TOF mass spectrometer (Voyager-DE RP, PE-PerSeptive Biosystems) in the linear mode. Each mass spectrum was averaged from a minimum of 300 measurements. Controls for the cleavage of phosphorylated peptides included, in parallel, PKA without ATP added to EP24.15, and PKA alone to determine background signal. Only unique fragments generated upon phosphorylation of EP24.15 were considered in the analyses. Proteolytic cleavage sites used for mapping and miscuts were generated by the ExPASy Molecular Biology Server of the Swiss Institute of Bioinformatics (36,37).

Identification and Sequencing of Mouse EP24.15 cDNA
A ZapII mouse pituitary cDNA library was purchased (Stratagene, La Jolla, CA) containing ϳ1.5 ϫ 10 6 recombinants and was screened as described previously with modifications (38). After plaque formation, plates were overlaid first with one nitrocellulose filter (BA82 Schleicher and Schuell, Keene, NH), for 1 min, and then a second filter was placed for 2 min. These duplicate filters were probed with the entire coding sequence for the rat EP24.15 cDNA (36), labeled by random priming (Rediprime II, Amersham Pharmacia Biotech). Filters were washed stringently to 0.1 ϫ SSC, 0.1% SDS, 65°C. From the first screen (ϳ2.8 ϫ 10 6 plaques), 27 positive plaques were identified. Subsequent rescreening produced 14 positive plaques which were purified, and placed through a tertiary screen. Bluescript plasmid containing the cloned inserts were excised by superinfection with R408 helper phage (36). cDNA inserts ranged in size from 1.1 to 2.3 kilobases. Restriction analysis and probing with 5Ј and 3Ј random primed labeled cDNA fragments indicated that one of the clones contained a full-length insert. This clone was sequenced in both directions using the ABI Bigdye terminator sequencing kit (PerkinElmer Life Sciences, South Plainfield, NJ). Data were analyzed using ABI Sequencing Analysis 3.3 (PerkinElmer Life Sciences), and Sequencher 3.1.1 (Gene Codes, Ann Arbor, MI) computer programs. Sequence alignments were made with the CLUSTAL program (39).

EP24.15 Is a Substrate for PKA Phosphorylation-Examina-
tion of the EP24.15 primary sequence revealed putative consensus phosphorylation sites (20), including those for PKA, PKC, and CKII (refer to Fig. 5). These motifs suggested that EP24.15 could be a potential substrate for phosphorylation. To test this hypothesis, rat recombinant EP24.15 and control substrates were incubated with various protein kinases. Whereas the incubation of EP24.15 with PKC (both ␤ and ⑀ isotypes) and CKII did not cause incorporation of phosphate into the protein (versus positive control subtrates) (Fig. 1A), incubation with PKA yielded a rapid and saturable incorporation of phosphate at the correct mass of 77 kDa (Fig. 1B). The phosphorylated residue was confirmed to be serine by phosphoamino acid analysis (Fig. 1C). This incorporation yielded an overall stoichiometry of 1.11 Ϯ 0.12 mol of phosphate/mol of EP24.15, consistent with one primary site of phosphate incorporation. A time course of phosphorylation performed (see "Experimental Procedures") from 15 to 240 min indicated that saturation of incorporated label occurred within 90 min (data not shown) under the assay conditions. To ensure saturable labeling conditions, components of the kinase reaction were varied in an attempt to increase the molar ratio (Table I). Doubling the ATP concentration, PKA concentration, and priming the reaction with additional PKA at 90 min did not effect the stoichiometry of the reaction. These results suggested that the system was at saturation for subsequent kinetic analyses.
Phosphorylation of EP24. 15  was observed, nearly identical to the result seen with QF7. Similarly, a 38% decrease in the specificity constant was observed for GnRH 1-9 (p Ͻ 0.09). The specificity constant of neurotensin also decreased, and indicated the same trend, but was not significant.
Strikingly, phosphorylation caused a 7-fold increase in the K m and k cat (with the corresponding increase in V max ) parameters measured for GnRH (Table II). The possibility that the 7-fold increases observed with the kinetic values for GnRH (not seen with GnRH 1-9 ) were a function of a newly formed charge interaction between the COOH-terminal residue on the peptide and the phosphate group on EP24.15 was investigated. Another EP24.15 substrate, bradykinin (RPPGFSPFR), and a synthetic bradykinin analog containing an amide-blocked COOH terminus (RPPGFSPFR-NH 2 ) were used for similar kinetic analyses. The carboxyl-terminal charge of these substrates were analogous to GnRH 1-9 and GnRH, respectively. Following phosphorylation, the K m of EP24.15 for bradykinin increased from 3.8 to 13 M upon phosphorylation, while for bradykinin-amide it decreased from 82 to 50 M, and thus did not replicate the findings for GnRH.
To examine whether the alterations in the kinetic parameters of EP24.15 for GnRH were caused by a change in active site accessibility due to phosphorylation, the K i was determined with an EP24.15-specific active site-directed inhibitor, cFP-AAF-pAB, the design of which is based on the hydrophobic and spatial considerations of the GnRH peptide structure (40). The K i of the nonphosphorylated enzyme was 24 Ϯ 0.5 versus 21 Ϯ 0.7 nM upon PKA phosphorylation, values consistent with both the recombinant enzyme (38) and rat brain-purified enzyme (40).
Determination of the PKA Phosphorylation Site-To determine which serine was the primary site for PKA phosphorylation, a series of mutant enzymes were designed, expressed, and assayed. Each of the putative PKA sites were systematically altered by site-directed mutagenesis, generating the following mutants: S98A, S106A, S172A, S288A, S398A, S522A, and S644A. Under identical PKA reaction conditions, phosphate incorporation in the mutants was indistinguishable from wild type, with the exception of the S644A mutant, which revealed negligible phosphate incorporation ( Fig. 2A). PhosphorImager quantification of the bands indicated an approximate 95% decline in the incorporation of phosphate (after normalization of the Coomassie stain using scanning densitometry) as compared with the wild type enzyme. To ensure that the amino acid substitution did not cause perturbations with respect to the correct folding of the protein (41), all of the mutants were analyzed by native gel electrophoresis. No changes in mobility nor protein expression were observed as compared with the wild type enzyme (data not shown).
To further confirm that the phosphorylation site on EP24.15 was serine 644, PKA phosphorylated enzyme was prepared using [␥-32 P]ATP and then cleaved by cyanogen bromide. This treatment yielded a phosphorylated 2-kDa fragment which was detectable by scintillation counting of fractions eluted from a size exclusion chromatography column calibrated before and after the CNBr-cleaved fragments were separated (Fig. 2B). The 2-kDa fragment uniquely and unambiguously assigned the fragment to residues 640 -657 (1987 daltons) (Fig. 2C).
Furthermore, phosphorylated and non-phosphorylated enzyme was prepared (see "Experimental Procedures"), and subjected to specific (trypsin and Lys-C) proteolytic cleavage and mass analysis by MALDI-TOF mass spectrometry. Lys-C digestion yielded a peptide whose mass (2484 daltons) corresponded to the fragment 637-659 with the addition of a ϩ80dalton phosphate moiety (Fig. 2C). Trypsin digestion of PKAphosphorylated EP24.15 yielded a peptide whose mass (1846 daltons) corresponded closely to fragment 643-659 with the addition of a ϩ80-dalton phosphate moiety (an actual addition of 84 daltons) (Fig. 2C). This fragment also included the serine at residue 644, a consensus PKA site. Other miscut fragments from the analysis of trypsin digestions also indicated serine 644-containing fragments with the addition of an 80-dalton moiety. These miscut fragments included fragment 637-664, the mass of which measured 3162 (the theoretical mass of 3085 ϩ 77 daltons), as well as fragment 643-675, the mass of which measured 3687 (the theoretical mass of 3606 ϩ 81 daltons) (Fig. 2C). No ϩ80-dalton adducts to corresponding fragments were noted elsewhere in the spectra of either the trypsin or Lys-C proteolytic digestions.  3A). The labeled band was extracted from the gel and subjected to phosphoamino acid analysis which revealed only serine phosphorylation (Fig. 3B), consistent with serine/threonine kinase action. To test if EP24.15 is PKA phosphorylated in vivo, PC12 cells were subjected to a kinase activation and inhibition paradigm (Fig. 3C). Forskolin stimulation resulted in a 38% increase in 32 P labeling. Preincubation with the PKA-selective inhibitor H89 dropped 32 P labeling to 58% of basal (vehicle) levels, despite the presence of forskolin. Western blot autoradiograms showed no change in the protein expression levels secondary to the treatments (data not shown).

EP24.15 Is Phosphorylated by PKA in Vivo in Rat PC12 and Mouse AtT-20 Cells-To build on earlier findings (17) and con-
Because murine AtT-20 cells are an important cell biological model for the study of EP24.15 regulation (27,31), we sought to determine whether PKA phosphorylation of EP24.15 was conserved across species, and specifically in this mouse model. Mouse pituitary AtT-20 cells were incubated with [ 32 P]orthophosphate and EP24.15 was immunoprecipitated (Fig. 3D). These cells also revealed only serine phosphorylation on EP24.15 protein (Fig. 3E). Again, a kinase activation and inhibition paradigm was employed (Fig. 3F) to determine if PKA is a kinase acting on EP24.15. Upon stimulation of AtT-20 cells with forskolin, a 35% increase in 32 P labeling was observed. When the cells were preincubated with the PKA-selective inhibitor H89, the 32 P labeling dropped to 17% of basal (vehicle). In parallel, it was confirmed by scanning densitometry of the Western blot autoradiograms that there was no change in the protein expression due to the treatments (data not shown).  Fig. 4). On the nucleic acid level, the mouse and rat coding sequences were 92.9% identical. The mouse and rat amino acid sequences shared 96.7% identity and 97.1% similarity (Fig. 5) and both species encoded a protein of 687 amino acids. The structural features are as reported for the rat form of the enzyme (38), and serine 644 was conserved (Fig. 5).

DISCUSSION
Whereas there are many possible mechanisms by which peptidase activity may be regulated, the presence of numerous putative phosphorylation sites on EP24.15 led us to examine whether phosphorylation plays a role in its modulation. Of the kinases studied, only PKA elicited significant phosphorylation (Fig. 1B). The effects of the in vitro PKA phosphorylation of EP24.15 on enzyme kinetics with different substrates was assessed. The 46, 44, and 40% decline in the k cat /K m observed for QF7 and QFS, and GnRH 1-9 , respectively, suggested a possible mechanism of down-regulation of EP24.15 activity upon phosphorylation. These kinetic changes are the same magnitude elicited by phosphorylation as observed in other systems where there was further amplification through signal transduction. A phosphorylation-induced 50 -60% decrease in activity has been reported for caspase-9 (15), which has significant downstream consequences in the apoptotic cascade.
There are compelling physiological data (2-4) establishing EP24.15 as the primary GnRH processing enzyme in vivo. For example, upon intracerebroventricular administration of a specific EP24.15 inhibitor, the half-life of GnRH increased 8-fold (2). Additionally, peripheral infusion of the EP24.15 inhibitor augmented the GnRH-dependent luteinizing hormone surge in rats (4). GnRH secretion from the hypothalamus in a pulsatile fashion is critical for the proper release of luteinizing hormone and follicle stimulating hormone from the pituitary, and the resultant control of mammalian reproduction (42). As such, understanding the kinetic aspects of GnRH degradation at the release site by EP24.15 is fundamentally important in understanding GnRH pulse waveform regulation. In this study, there was a 7-fold increase in K m and k cat (and correspondingly, V max ) observed with GnRH upon EP24.15 phosphorylation. These changes imply that the affinity of GnRH binding was reduced upon phosphorylation, but once bound, the substrate appears to be turned over more rapidly. It is possible that the phosphorylated enzyme has the versatility to expediently handle large increases in GnRH concentration (during the large increase in amplitude, concomitant with GnRH pulsatile release), without becoming rapidly saturated at these substrate concentrations. The insulin peptide concentration in secretory vesicles has been measured to be approximately 40 mM just prior to release (43). In another study (44), the concentration of neuropeptide achieved at the site of release at the synapse has been postulated to approach ϳ10 mM in synaptic vesicles. Therefore, it is quite plausible that peptide concentrations in the synapse at the point of vesicular release can approach the in vitro K m range of ϳ100 M described here and by others (45), as well as 1 mM upon phosphorylation. Overall, these findings suggest that phosphorylation can have a significant impact on EP24.15 activity by preventing its saturation at the time of pulsatile release, and hence, GnRH hydrolysis. If EP24.15 exists in the hypophysial portal blood both in a phosphorylated and non-phosphorylated state, the effective substrate concentration range of this enzyme would be substantially broadened.
We next sought to understand the nature of the biophysical changes conferred by phosphorylation of EP24.15. None of the kinetic changes outlined (Table II) would appear to be the result of limited substrate accessibility to the active site, given the nearly identical K i determinations with phosphorylated and non-phosphorylated EP24.15 for the active site-directed inhibitor cFP-AAF-pAB. We further hypothesized that the major alterations in kinetic parameters observed with GnRH as compared with GnRH 1-9 , were perhaps due to a direct interaction of the phosphate group of the enzyme with the carboxylterminal amide of the substrate peptide (this glycine-amide being absent in GnRH 1-9 ), where there might be repulsion and conformational alterations with respect to the carboxyl moiety in GnRH [1][2][3][4][5][6][7][8][9] . In this context, we examined the kinetics of both phosphorylated and non-phosphorylated EP24.15 with bradykinin-amide and bradykinin in a manner identical to that for GnRH and GnRH 1-9 , respectively. The K m of EP24.15 for bradykinin increased upon phosphorylation, while for bradykininamide it decreased, not in agreement with the relative changes seen with GnRH and GnRH [1][2][3][4][5][6][7][8][9] . This would suggest that the major changes observed in the kinetic parameters for GnRH upon EP24.15 phosphorylation are not likely a charge-induced phenomenon. There may exist other, as yet unknown, binding mechanisms which differentiate between the two substrates (GnRH and GnRH 1-9 ). More importantly, it indicates that the phosphorylation of EP24.15 alters its neuropeptide kinetic profile and substrate specificity.
Systematic site-directed mutagenesis of all putative serine PKA phosphorylation sites to alanine, indicated serine 644 to be the primary site of in vitro phosphorylation ( Fig. 2A). To confirm the serine 644 phosphorylation site, cyanogen bromide cleavage was utilized since there are few overlapping putative phosphorylated peptides by mass. A labeled 2000-dalton fragment correlated uniquely to the cyanogen bromide fragment 640-657 (mass 1987), containing serine 644 (Fig. 2B). We continued further mapping of the phosphorylated protein with trypsin and Lys-C utilizing MALDI-TOF mass analyses. As anticipated, trypsin digestion yielded a complex spectra containing miscuts due to the phosphorylation blocking enzyme access to the adjacent Lys/Arg residues, but clearly indicated peptides with ϩ80-dalton adducts (phosphoryl groups) not seen in the non-phosphorylated enzyme (Fig. 2C). The most complete cleavage with trypsin yielded the 643-659 fragment. This fragment contained an internal arginine (647), situated amino to a proline residue, a combination of residues which is known to cleave very inefficiently with trypsin. Furthermore, limited proteolysis with endoproteinase Lys-C yielded spectra with a full cut fragment (637-659), consistent with serine 644 phosphorylation (Fig. 2C). Serine 644 was conserved between rat and mouse, a finding paralleled by the in vivo data demonstrating that PKA is contributing to EP24.15 phosphorylation in both rat and mouse cell lines. In the human EP24.15 sequence, 6 of 7 putative PKA sites are conserved, but not the serine 644 consensus site. Nonetheless, when human M17 neuroblastoma cells were subjected a similar PKA kinase activation/inhibition scheme, EP24.15 likewise was phosphorylated by PKA (data not shown). Importantly, in a fashion similar to rat and mouse (Fig. 3), the PKA phosphorylation of EP24.15 still occurs in vivo in human cells.
Homology modeling studies of EP24.15 based on the bacterial enzymes, thermolysin, and neutral protease (previously solved to atomic resolution by x-ray diffraction), indicate the presence of a 4-helix bundle structural motif in the carboxylterminal 80 -95 residues. 2 This motif has been previously modeled by homology to the related metalloenzymes enkephalinase and angiotensin converting enzyme (46). In this model, serine 644 would reside near the carboxyl end of the second helix in a 4-helix bundle, a structural motif present in many proteins (reviewed in Ref. 47). The closest distance approximation of serine 644 to the active site zinc is approximately 17 Å, seemingly too far for a direct steric effect in the active site. This interpretation is consistent with the unchanged EP24.15 inhibitor (cFP-AAF-pAB) K i data upon phosphorylation, although it is possible a longer range conformational change in the protein is modulated through this structural motif.
As our initial studies characterizing the phosphorylation and inhibition of EP24.15 activity had been performed in vitro with recombinant enzyme, therefore it was important to determine if phosphorylation of EP24.15 occurs in mammalian cells. Interestingly, an earlier study in rat pheochromocytoma (PC12) cells treated with cAMP analogues showed a decrease in the soluble specific activity of EP24.15 without a decrease in the amount of EP24.15 protein (17). We extend this observation by specifically demonstrating decreases in EP24.15 enzyme activity by PKA phosphorylation in vitro, and by demonstration of the PKA phosphorylation of EP24.15 in PC12 cells. We further explored whether the PKA phosphorylation of EP24.15 is conserved in a commonly used mouse neuroendocrine cell model. Utilizing the AtT-20 mouse pituitary cell line, our studies indicated that EP24.15 can be phosphorylated by PKA. As was the case in the PC12 cells, EP24.15 phosphorylation was enhanced by forskolin treatment, and inhibited by the PKAselective inhibitor H89 (48) concomitant with forskolin treatment. The complete cDNA cloning of mouse EP24.15 (Fig. 4) and alignment with the rat sequence (Fig. 5) indicated the perfect conservation of the PKA phosphorylation site, serine 644.
It is also possible that phosphorylation may regulate EP24.15 function by subcellular targeting and/or expression at the plasma membrane (27), the nucleus (4, 49), or other cellular locations, either directly or via protein-protein interactions. For example, enkephalinase (EC 3.4.24.11), a related enzyme, can be phosphorylated by casein kinase II, and subsequently co-associates with a tyrosine-phosphoprotein complex in Nalm 6 (lymphoblastic leukemia) cells, suggesting a role for this peptidase in signal transduction pathways (50). Previously, a faster migrating form of EP24.15 was found to be present on the plasma membrane of AtT-20 cells (27). By performing labeling studies of AtT-20 cells with [ 35 S]Met/Cys (detecting all forms) performed in parallel with [ 33 P]orthophosphate labeling (providing higher resolution than 32 P), this faster migrating form of the enzyme was not a phosphorylated form of EP24.15 (data not shown). Interestingly, serine 644 resides within a potential 14-3-3-binding protein consensus site (reviewed in Ref. 51). 14-3-3 proteins have been demonstrated to interact with various signaling proteins through phosphoserine motifs, one of which is RXpSXXXP found in several proteins (52), exactly matching the motif at serine 644 of EP24.15. Thus, phosphorylation may actually represent a modification which induces multidimensional modulations of EP24.15: both spatiotemporal as well as kinetic. Now that the site of phosphorylation has been identified, future studies can address the role of phosphorylation and its influence on trafficking using appropriate mutant EP24.15 expression vectors.
In summary, the present report demonstrates phosphorylation of EP24.15 by PKA on serine 644. Importantly, EP24.15 is phosphorylated by PKA in both mouse and rat species which both share this conserved phosphorylation site. This results in an alteration of neuropeptide hydrolyzing activity indicating phosphorylation as a possible physiological regulator of EP24.15 activity.