Calmodulin Binds to the Cytoplasmic Domain of Angiotensin-converting Enzyme and Regulates Its Phosphorylation and Cleavage Secretion*

The rate of cleavage secretion of the enzymatically active ectodomain of angiotensin-converting enzyme (ACE) is regulated by tyrosine phosphorylation of the protein and by the phorbol ester, phorbol 12-myristate 13-acetate (PMA), an activator of protein kinase C. Here, we report that both calmodulin inhibitor (CaMI) and calmodulin kinase inhibitor could also enhance cleavage secretion of ACE. This effect was accompanied by the dissociation of calmodulin from a specific region within the cytoplasmic domain of ACE to which it had been bound. The same domain of ACE was phosphorylated, and both CaMI and PMA caused dephosphorylation of ACE as well. Mass spectrometric and mutational analyses identified Ser730 as the only phosphorylated residue in the cytoplasmic domain of ACE. The Ser730 → Ala mutant of ACE was not phosphorylated, but it still bound calmodulin, and its cleavage secretion was enhanced by both CaMI and PMA. Similarly, when Ser730 was replaced by the phosphoserine mimetic, Asp, cleavage secretion of the resultant mutant remained susceptible to the enhancing effect of CaMI and PMA. These results demonstrate that, although CaMI and PMA can enhance both cleavage secretion of ACE and its dephosphorylation, the two effects are not mutually interdependent.

ding, the "sheddases," are also a member of transmembrane protein family (7). Some of the proteolytic events are carried out by ADAM (a disintegrin and metalloproteinase) family of proteases (9). Tumor necrosis factor-␣-converting enzyme (TACE, also known as ADAM 17), was the first secretase with known physiological substrate (tumor necrosis factor-␣) to be isolated and characterized (10,11). TACE is also involved in shedding of a number of transmembrane proteins, including L-selectin (12), ␤-amyloid precursor protein (13), epidermal growth factor receptor ligands (14,15), and growth factor receptors (16 -18).
ACE (also known as kininase II and CD143) is a central component of the renin-angiotensin system, which regulates blood pressure, electrolyte balance, and fluid homeostasis. ACE participates in blood pressure regulation by converting inactive decapeptide angiotensin I to vasoactive octapeptide angiotensin II and by inactivating the vasodilator bradykinin. ACE exists as two isoforms, namely somatic ACE (sACE) and germinal ACE (gACE), which are generated from the same gene with tissue-specific choice of transcription from two alternative promoters (19). Somatic ACE, the larger isoform (170 kDa), is expressed in vascular endothelial cells, renal epithelial cells, neuronal cells, and macrophages and is involved in regulation of blood pressure and renal function. Germinal ACE, the smaller isoform (116 kDa), is expressed only in maturing sperm cells and is involved in male fertility. Studies on tissue-specific expression of ACE revealed that expression of sACE in vascular endothelial tissue is essential for maintaining normal blood pressure (20). Both the isoforms of ACE are expressed on the cell surface as a type I ectoprotein, with a short cytoplasmic domain, a transmembrane domain, and a long extracellular domain containing the active site(s), and are cleaved on the membrane proximal region to release the enzymatically active extracellular domain in the body fluid, including serum (21)(22)(23). Our previous studies on rabbit gACE demonstrated that ACE shedding is a regulated event, which can be stimulated by phorbol esters (PMA) and inhibited by specific metalloprotease inhibitor (Compound 3/TAPI II) (22,24), the distal ectodomain of ACE plays a vital role in ACE shedding (25), and the ACE molecule is cleaved at 663 RS 664 on the membrane-proximal region to generate the soluble form (22). The protease responsible for ACE cleavage is the putative ACE secretase, a metalloprotease, which is different from TACE (23,26) and still remains unidentified.
Ectodomain shedding of ACE and other physiologically important ectoproteins appears to be an important cellular function, but the regulation of this process is not yet fully understood. We have shown earlier that gACE is associated with specific cellular proteins such as immunoglobulin-binding protein and protein kinase C isozymes and this association can regulate the cleavage secretion of ACE (27). In the present study, we identified another cellular protein, calmodulin (CaM), which binds to the cytoplasmic domain of ACE and regulates its shedding.
Recently we also showed that tyrosine phosphorylation of the ectodomain of ACE regulates ACE shedding (28). Here we show that ACE is also phosphorylated on a specific serine residue of the cytoplasmic domain. However, in contrast to tyrosine phosphorylation, serine phosphorylation of ACE does not directly regulate the rate of its ectodomain shedding.
Cell Culture and Transfection-ACE89, a mouse epithelial cell line, permanently transfected with rabbit gACE expression vector, was maintained in Dulbecco's modified Eagle's medium with 10% fetal bovine serum, antibiotics, and 10 M CdCl 2 (30). HeLa cells were maintained in a similar media (except that CdCl 2 was omitted) and transiently transfected with rabbit gACE using the Lipofectamine reagent (Invitrogen) following the manufacturer's instructions. Primary human umbilical vein endothelial cells (HUVECs) were isolated as described before (31). Isolated cells were maintained in MCDB 105 (Sigma), supplemented with 20% fetal bovine serum, 0.009% heparin, and 0.015% endothelial cell growth factors on fibronectin-coated flasks, and used from passage 2 for experiments.
Rabbit Lung Extract-Extracts of rabbit lungs were obtained using previously described methods (22). 35 S Labeling of Cells, Immunoprecipitation, and Quantification of Cleavage Secretion-Confluent ACE89 or transiently transfected HeLa cells were first incubated in methionine/cysteine-free medium for 1 h, and then labeled with [ 35 S]methionine/cysteine (PerkinElmer Life Sciences) 0.5 h; the label was chased for the time periods as indicated in figures. Trifluoperazine and other agents were added to the medium as detailed in the figure legends and under "Results." At the end of the chase period, ACE was immunoprecipitated from the cell extracts (cells were washed with cold phosphate-buffered saline and extracted in radioimmune precipitation assay buffer containing 50 mM Tris, pH 7.4, 150 mM NaCl, 1% Triton X-100, 1% SDS, and 1 mM EDTA) and the culture media with anti-ACE. Immunoprecipitated ACE was then analyzed by 8% SDS-PAGE, followed by autoradiography (25). The dried gels were subjected to PhosphorImager analysis, to quantify the amount of ACE present in the cell extract and the culture medium. Cleavage secretion was calculated as the amount of medium ACE over the total ACE present in cell extracts and culture medium. Underglycosylated ACE was, however, present in the immunoprecipitate, not considered for cleavage secretion, because it was not cleaved in the medium. 32 P Labeling of Cells-Confluent ACE89 or transfected HeLa cells were incubated in phosphate-free culture media for 1 h and then labeled with [ 32 P]orthophosphate (50 Ci/60-mm plate, PerkinElmer Life Sciences) for 16 h and then treated with CaMI or PMA, as detailed under "Results." Labeled cells were extracted in an alternative lysis buffer containing 50 mM Tris, pH 7.6, 150 mM NaCl, 0.1% Triton X-100, 1 mM sodium orthovanadate, 10 mM sodium fluoride, 5 mM sodium pyro-phosphate, and 10 mM ␤-glycerophosphate in the presence of protease inhibitor mixture (Sigma). ACE was immunoprecipitated by anti-ACE, and the immunoprecipitate was resolved by SDS-PAGE, followed by autoradiography. The phosphorylated ACE was quantified by Phospho-rImager analysis.
ACE Activity Measurement-ACE activity was assayed by using hippuryl-L-histidyl-L-leucine (Hip-His-Leu) as the substrate and measuring fluorometrically the His-Leu liberated at 5 mM Hip-His-Leu (29).
Immunoprecipitation by Anti-phosphoserine/Threonine Antibody-Serine-phosphorylated ACE in the ACE89 cell extracts (in the alternative lysis buffer containing protease inhibitor mixture) was immunoprecipitated by rabbit anti-phosphoserine/threonine as per the manufacturer's instructions, and the immunoprecipitate was resolved by 8% SDS-PAGE and immunoblotted with anti-ACE.
Western Blotting-Membrane-bound and -secreted ACE was immunoblotted with anti-ACE. Cell-bound cytoplasmic peptide of ACE, from the solubilized membranes of ACE89 cells, was resolved in 4 -20% gradient gel (Bio-Rad) and Western blotted with anti-carboxyl-terminal peptide antibody. For calmodulin immunoblotting the anti-ACE immunoprecipitates were resolved in 15% SDS-PAGE and transferred to nitrocellulose membranes and blotted with anti-calmodulin (anti-CaM).
Purification and Carboxyl-terminal Sequencing of Soluble ACE-Secreted ACE was purified from culture media of either untreated or PMA-treated ACE89 cells, using Lisinopril-Sepharose affinity chromatography as described before (32). The eluted pure ACE was resolved by 8% SDS-PAGE, and it appeared as a single band, when stained with Coomassie Brilliant Blue. 165 g (1.5 nmol) of purified ACE was digested with Carboxypeptidase Y (Sigma) at a substrate to enzyme molar ratio of 35:1, in a digestion buffer containing 100 l of pyridyl acetate and 100 l of 0.2% Rapigest (a denaturing agent, Waters). The reaction mixture was incubated at 37°C, and samples (15 l) taken at various time points were used for amino acid analysis. 2 nmol of bovine serum albumin was used as a control for the analysis.
Purification of Cell-bound Carboxyl-terminal Peptide of ACE from ACE89 Cells-Confluent ACE89 cells (150 mm) were incubated with PMA (500 nM) in serum-free medium for 1 h. The cells were harvested after washing twice with serum-free medium and suspended in 5 ml of 50 mM Tris-HCl buffer, pH 8.0, containing 1 mM phenylmethylsulfonyl fluoride. After three cycles of rapid freezing and thawing at 37°C, the suspension was homogenized in glass-glass homogenizer and centrifuged at 25,000 ϫ g for 1 h. The pellet was resuspended in 5 ml of homogenization buffer containing 0.1% Triton X-100. After 30 min on ice, the samples were similarly centrifuged, and the supernatant containing the carboxyl-terminal peptide of ACE was used for further purification. Purification of the carboxyl-terminal peptide was performed using the procedure described in a previous study (24). In short, the above preparation of carboxyl-terminal peptide was passed through an immunoaffinity column, prepared by purified IgG from anti-carboxylterminal peptide polyclonal antiserum, and coupled to CNBr-activated Sepharose 4B (Sigma), using the manufacturer's instructions. The antibody-coupled Sepharose was washed as mentioned previously (24) and eluted with 0.1 M glycine-HCl buffer, pH 2.8, containing 0.1% Triton X-100. The eluate was then passed through an fast-protein liquid chromatography column (Superdex 75 gel filtration column (Amersham Biosciences)), and the fractions containing the carboxyl-terminal peptide of ACE were pooled and concentrated. The pooled fraction was further purified by a C4 reverse phase high-performance liquid chromatography column (Vydac) and eluted with a gradient of acetonitrile in 0.1% trifluoroacetic acid. The single fraction containing the carboxyl-terminal fragment was used for Mass specrometric analysis and aminoterminal sequencing.
Mass Spectrometric Analysis of the Purified Cell-bound Cytoplasmic Fragment-An in-gel tryptic digest of the 8-kDa Coomassie Bluestained band was analyzed using capillary column LC-tandem mass spectrometry. Peptide mapping and phosphopeptide detection were carried out using an LCQ Deca ion trap mass spectrometer system (ThermoFinnigan, San Jose, CA) equipped with a nanospray ionization source (Protana, Odense, Denmark). The source was operated under microspray conditions at a flow rate of 200 nl/min. The digests were analyzed by reversed-phase capillary high-performance liquid chromatography using a 50-m i.d. column with a 15-m i.d. tip (New Objective Corp., Woburn, MA). The column was packed with ϳ6 cm of C18 packing material (Phenomenex, Torrence, CA) and eluted using a 45-min gradient of increasing acetonitrile (2-70%) in 50 mM acetic acid. The data were acquired in the data-dependent mode, recording a mass spectrum and three collision-induced dissociation (CID) spectra in repetitive cycles (33). The program Sequest was used to compare all CID spectra recorded to the sequence of rabbit ACE (NCBI accession number 113043) and considering the appropriate changes in the serine residue mass of ϩ80 Da. The data were also searched in the chromatographic mode for neutral loss of 33 and 49 Da, corresponding the loss of H 3 PO 4 from triply and doubly charged ions (respectively), using the instrument data system program Xcaliber. The molecular weight of the intact cleaved carboxyl-terminal fragment was also measured with this system using the same chromatographic conditions but recording only mass spectra. The elution of the fragment protein was recognized by plotting a base peak chromatogram, with the resultant mass spectrum produced by averaging the spectra across that chromatographic peak.
Amino-terminal Sequencing-The amino-terminal sequencing of the purified cell-bound carboxyl-terminal peptide of ACE, blotted on a polyvinylidene difluoride membrane, was carried out by Edman degradation (34,35), using a Procise, Model 492, protein sequencer (Applied Biosystems, Foster City, CA) fitted with a 140c microgradient system, 785A programmable absorbance detector, and a 610A (version 2.1) data analysis system.
Mutation of Serine Residues in Intracellular Domain of ACE-Sitedirected mutagenesis by the mega primer PCR method (36) was used for mutating Ser 713 , Ser 718 , Ser 730 , and Ser 737 of ACE cytoplasmic domain.

CaM Inhibitors Stimulate Cleavage Secretion of ACE-In ACE89
cells, ACE is constitutively cleaved and secreted at a slow basal rate. Upon addition of trifluoperazine, an inhibitor of calmodulin (CaMI), to the culture medium, secretion of ACE, as assessed by measuring the enzyme activity of cleaved ACE in the medium, increased linearly up to an inhibitor concentration of 20 M (Fig. 1A). To confirm the stimulatory effect of trifluoperazine on ACE cleavage secretion, ACE89 cells were pulse-labeled with [ 35 S]methionine, and the biosynthesis, processing, and cleavage secretion of newly synthesized labeled ACE molecules were followed in the presence of the inhibitor. ACE-related proteins were immunoprecipitated from the cell extract and the culture media by anti-ACE antibody and analyzed by SDS-PAGE. ACE cleavage secretion, as quantitated by phosphorimaging analysis of dried gels, increased in the presence of 10 and 20 M concentration of the CaMI, from a basal level of 13 Ϯ 2% to 42 Ϯ 2% and 53 Ϯ 3%, respectively (Fig. 1B). Compound 3, a known inhibitor of ACE-secretase blocked the CaMI-induced secretion as well, indicating that the effect of CaMI was mediated by the ACE-secretase. To examine whether CaM could also regulate ACE secretion from natural ACE-producing cells, we used primary HUVEC cultures. When CaMI was added to the HUVEC culture media, the secretion of sACE was enhanced, as detected by the increased ACE activity in the medium (Fig. 1C). When Compound 3, an inhibitor of ACE-secretase, was present in the media, the enhanced secretion of ACE by CaMI was inhibited.
Most of the intracellular effects of CaM are mediated by protein phosphorylation catalyzed by CaM-dependent protein kinases (37). CaM kinase II, a multifunctional protein, which has been studied extensively, is known to mediate many of the biological effects of CaM. To determine if CaM kinase II plays a role in ACE cleavage secretion, we investigated the effect of KN-93, a specific inhibitor of CaM kinase II, on ACE cleavage secretion. As illustrated in Fig. 2A, KN-93 added to ACE89 cells, increased the secretion of ACE in a concentration-dependent manner. Again, pulse-chase experiments, in the presence of KN-93, confirmed these observations. Fig. 2B shows that 14 Ϯ 1% of ACE was secreted out from the untreated cells in 2 h, whereas secretion increased to 21 Ϯ 2% when cells were treated with 20 M KN-93. The observed increase of cleavage secretion induced by inhibitors of both CaM ( Fig.  1), as well as CaM kinase II (Fig. 2), strongly suggests that CaM influences the cleavage secretion process.
Residues 709 -719 in the Cytoplasmic Domain of ACE Are Required-We used a transient transfection system of ACE expression in HeLa cells for delineating the region of ACE required for the enhancement of cleavage secretion by CaMI. We have shown previously that in this system ACE is synthesized and cleavage-secreted in a manner similar to that of ACE89 cells (29). When we expressed a mutant ACE, from which the carboxyl-terminal cytoplasmic domain had been deleted completely (gACE⌬709 -737), the cells secreted the mutant ACE more efficiently than the wild type ACE (TABLE ONE; see also Ref. 25). A partial deletion of this domain (gACE⌬719 -737) did not enhance secretion significantly, indicating that residues 709 -719 might slow down the process of cleavage secretion. Trifluoperazine increased cleavage secretion of Wt ACE (14 Ϯ 1% to 34 Ϯ 1%) and the partial carboxyl-terminal-tail deleted mutant to a similar extent (TABLE ONE). On the contrary, secretion of the mutant gACE⌬709 -737, from which the entire cytoplasmic domain had been deleted, was not affected significantly (25-28%) by CaMI treatment. Another mutant, from which only the residues 709 -719 had been deleted, was cleavage-secreted faster than Wt ACE (14 Ϯ 1% to 29 Ϯ 3%) (TABLE ONE), and CaMI treatment did not induce its secretion significantly (29 Ϯ 3% to 34 Ϯ 4%). These results demonstrated that residues 709 -719 are required for CaMI-mediated enhanced secretion of ACE. Similar results were obtained when ACE was allowed to secrete for 4 h (data not shown).
Role of MEK and Protein Kinase p38 in CaMI-induced Cleavage Secretion of ACE-Because various signaling pathways have been implicated in the regulation of cleavage secretion of different ectoproteins we tested the roles of specific protein kinases, in the CaMI-induced cleavage secretion of ACE. ACE89 cells were pre-treated with inhibitors of different protein kinases, labeled with [ 35 S]methionine for 30 min and chased for 2 h in the presence of CaMI, in the presence of the inhibitors throughout the experiment. Inhibitors of phosphoinositide 3-kinase, Janus Kinase 2, p70S6 kinase, or phospholipase C␥, did not show any significant effect (data not shown). As shown in Fig. 3A, U0126 and SB203580, inhibitors of MEK and protein kinase p38 respectively, blocked the CaMI-induced cleavage secretion of ACE. Our observation that CaMI-induced cleavage secretion of ACE is blocked by inhibitors of MEK and protein kinase p38, suggested that constitutive activation of these signaling pathways would induce increased cleavage secretion of ACE. To test this hypothesis, we co-transfected HeLa cells with expression vectors of ACE and constitutively active (ca) MEK1 or MKK6 and observed enhanced cleavage secretion of ACE (Fig. 3B). When dominant-negative (dn) forms of MEK1 or protein kinase p38 were expressed, the CaMI-induced cleavage secretion of ACE was blocked (Fig. 3B). Taken together, these results indicate the involvement of MEK and p38 protein kinase in CaMI-induced cleavage secretion of ACE.
CaM Binds to the Cytoplasmic Domain of ACE-While studying the biosynthesis of ACE in ACE89 cells metabolically labeled with [ 35 S]methionine, we often detected a 17-kDa protein band in the anti-ACE immunoprecipitate. The 17-kDa protein was not detected when C127 cells, the parent cell line that does not express ACE, was used, indicating the specificity of interaction between the 17-kDa protein and ACE (data not shown). In addition, the inability of the anti-ACE antibody to recognize the 17-kDa protein by Western blotting (data not shown) confirmed that it was not a cleavage product of ACE. To determine if the 17-kDa protein present in the ACE-immunoprecipitate was CaM, the SDS-PAGE-analyzed immunoprecipitate was immunoblotted with anti-CaM antibody. Indeed, the CaM-antibody recognized the 17-kDa protein which co-migrated with purified CaM (Fig. 4A, control and CaM, respectively). When ACE89 cells were treated with CaMI, the 17-kDa CaM was not present in the immunoprecipitate (Fig. 4A). Similar results were obtained when calmidazolium and W7 (N-(-6-aminohexyl)-5-chloro-1-napthalenesulfonamide), two other inhibitors of CaM, were used (data not shown). These results demonstrated that CaM is bound to ACE, and this binding is disrupted by inhibitors of CaM. Similar to ACE89 cells, CaM co-immunoprecipitated with ACE from HeLa cells (Fig. 4B, lane 2

ACE Is Phosphorylated in Its Cytoplasmic Domain-Because
CaM kinase II, a serine/threonine kinase, is involved in the regulation of ACE cleavage secretion, we wondered whether ACE itself is phosphorylated. To examine this possibility, ACE89 cells were labeled either with [ 35 S]methionine or [ 32 P]orthophosphate. The labeled ACE was immunoprecipitated and subjected to SDS-PAGE analysis followed by autoradiography. As expected, the 35 S-labeled, mature, 116-kDa ACE as well as its underglycosylated, 95-kDa biosynthetic precursor were present in the cell extract, as was the secreted ACE in the immunoprecipitated media (Fig. 5, lanes 1 and 2). The 32 P-labeled cells, analyzed in a similar fashion, showed that the mature 116-kDa ACE was also labeled with 32 P (lane 3). However, neither the underglycosylated precursor nor the secreted ACE was 32 P-labeled (lanes 3 and 4). A Western blot of lanes 3 and 4 with ACE antibody confirmed the presence of all three ACE species in the 32 P-labeled immunoprecipitate (data not shown).
Because cell-bound mature ACE, but not secreted ACE, was phosphorylated (Fig. 5, lane 4), it was likely that the phosphorylated residue was present carboxyl-terminal to the cleavage site (residue 663), i.e. within the juxtamembrane, the transmembrane, or the cytoplasmic domain. The possibility that the cytoplasmic domain could be the site of phosphorylation was tested by expressing the cytoplasmic domain-deleted mutant of ACE. A 32 P radioautogram of the immunoprecipitated Wt ACE and the mutant ACE demonstrated that only the Wt ACE was phosphorylated (Fig. 5B, lanes 2 and 3), although similar amounts of both proteins were detected in the Western blot (data not shown). These results conclusively identified the cytoplasmic domain of ACE as the site of phosphorylation.
The above observations with gACE phosphorylation in tissue-culture cells prompted us to examine the in vivo phosphorylation status of sACE, the other isozyme of ACE synthesized by vascular endothelium. For this purpose we used extracts of rabbit lung, a rich source of sACE, for immunoprecipitation with either anti-ACE or anti-phospho-Ser/ Thr antiserum. The immunoprecipitates were Western blotted with anti-ACE, and as shown in Fig. 6, the antibody recognized the 178-kDa sACE in both immunoprecipitates indicating that sACE was Ser/Thr phosphorylated in rabbit lung.
CaMI Causes Dephosphorylation of ACE-Because CaM and CaMK mediate protein phosphorylation we wondered whether their inhibitors, which enhance secretion, affect the phosphorylation status of ACE. For this purpose, ACE89 cells were cultured in the presence of [ 32 P]orthophosphate to radiolabel ACE and then treated with either CaMI or PMA (a known enhancer of ACE secretion). As shown in Fig. 7, both caused significant dephosphorylation of ACE as compared with FIGURE 6. Somatic ACE is Ser-phosphorylated. Somatic ACE was extracted from rabbit lung (22) and immunoprecipitated with either anti-ACE antibody or anti-phospho-Ser/ Thr antibody. All immunoprecipitates were analyzed by SDS-PAGE and Western blotted with anti-ACE antibody or pre-immune serum as indicated.   untreated cells (Fig. 7B). These results indicate that CaMI and PMA probably block ACE Ser phosphorylation or promote its dephosphorylation. As expected, both enhanced ACE secretion (Fig. 7A).
Identification of the Phosphorylation Target-The cytoplasmic domain of gACE has four serine residues. To determine which of these residues is phosphorylated, the 8-kDa cell-bound carboxyl-terminal peptide of ACE, which is left behind after the ectodomain is cleaved, was purified from ACE89 cells, in-gel digested with trypsin and analyzed by LC-tandem Mass Spectrometry (Fig. 8). The data were analyzed using all the CID spectra that were acquired to search the NCBI non-redundant data base. The major phosphorylation site was subsequently established from the presence of the phosphopeptide QPHHGPQFG-pSEVELR. The CID spectrum in Fig. 8 shows the presence of a doubly charged base peak (m/z 842.7), generated by the loss of H 3 PO 4 moiety (molecular mass of 80) from the phosphopeptide containing Ser 730 in the carboxyl-terminal fragment of ACE. Thus, Ser 730 was unequivocally shown to be phosphorylated.
Both Phosphorylated and Unphosphorylated ACE Proteins Are Cleaved-To further investigate the relationship between ACE phosphorylation and its cleavage secretion, we examined whether both phos-phorylated and unphosphorylated forms of the protein are cleaved. For this purpose, we analyzed the 8-kDa carboxyl-terminal fragment that is retained in the cells, as the ectodomain is cleaved off. To examine this fragment in more detail, membranes were isolated from ACE89 cells, solubilized with detergent, and analyzed by gradient SDS-PAGE followed by Western blotting. In the membrane extract prepared from untreated cells, the anti-carboxyl-terminal peptide antibody recognized two distinct peptides of similar intensities (Fig. 9A). When the cells were treated with PMA or CaMI, the same peptides were present, but the lower band of the doublet was more abundant. The presence of two different peptides, both of which cross-reacted with the anti-carboxylterminal antibody, raised the possibility that the cleavage had occurred at two different sites. If that were the case, the amino-terminal residues of the peptides would be different. However, when the amino-terminal sequences were determined by Edman degradation, the same sequences were obtained for peptides purified from untreated or PMA-treated cells: the sequence of 13 and 15 amino-terminal residues of the purified carboxyl-terminal peptides obtained from untreated and PMA-treated cells matched perfectly with that of ACE residues 664 -676 and 664 -678, respectively (22). Although, both purified peptide samples, as expected, contained both peptides of the doublet, no other major sequence was detected. These results clearly established that, a single peptide bond, between Arg 663 and Ser 664 of ACE (also (22)), is cleaved during constitutive or stimulated secretion.
In a complementary experiment, we determined the carboxyl-terminal sequence of purified secreted ACE from the culture media of untreated or PMA-treated ACE89 cells. This was done by identifying the amino acid released by digestion with carboxypeptidase Y. The result agreed with the above conclusion, because the amino acids released in both cases were TPNSAR, the exact residues that are located at 658 -663 at the carboxyl-terminal of secreted ACE. Hence, the difference in molecular size of the two peptides in the doublet was not due to cleavage occurring at two different sites.
As a phosphoserine was identified in the cytoplasmic domain of ACE, we examined the phosphorylation status of the cleaved cell-bound carboxyl-terminal fragment in more detail. For this purpose a LC-tandem mass spectrometric analysis was carried out for determining the exact mass of the purified fragment. The spectrum (Fig. 9B) shows the pres-  for 1 h), or without any added agent (C). The cleavage secretion (%) was determined as described before and presented as a bar graph. *, p value, compared with untreated cells, Ͻ0.05. B, ACE89 cells were metabolically labeled with [ 32 P]orthophosphate for 16 h, the incubation was continued in the absence or presence of CaMI for 2 h or PMA for 1 h. ACE was immunoprecipitated from the cell lysates and analyzed by SDS-PAGE and 32 P autoradiography. FIGURE 8. Identification of Ser 730 as the target of phosphorylation by mass spectrometry. The cell-bound cleaved carboxyl-terminal fragment of ACE was purified from ACE89 cells. An in-gel tryptic digest of the 8-kDa Coomassie Brilliant Bluestained purified peptide was analyzed using capillary column LC-tandem mass spectrometry. The data were analyzed using CID spectra to search the NCBI non-redundant data base. The CID spectrum of the phosphopeptide containing Ser 730 shows the facile loss of H 3 PO 4 to produce the doubly charged base peak in the spectrum (m/z 842.7). The peptide sequence and the site of phosphorylation (QPHHGPQFGpSEVELR) were subsequently established by interpretation of the low abundance fragment ions. ence of overlapping mass spectra of the phosphorylated and dephosphorylated forms of the fragment. The charge state of each set of ions was used to calculate the molecular masses of each protein fragment of the overlapping spectra, which were 8223.4 and 8303.8 Da. Thus the difference in the molecular masses of the two components of the overlapping spectra resembled that of a phosphate ion of molecular mass 80 Da. These results clearly indicate that the cell-bound carboxyl-terminal fragment of ACE was present both as a phosphorylated as well as an unphosphorylated peptide. The phosphopeptide possibly migrated slower in the gel, giving rise to the observed doublet.
Phosphorylation of ACE Does Not Play a Role in Cleavage Secretion-To investigate the possible role of phosphorylation in cleavage secretion of ACE, we resorted to mutational analyses. To generate a mutant that was not phosphorylated, the four Ser residues present in the cytoplasmic domain of ACE were individually mutated to alanine or glutamic acid. Cells expressing Wt or mutant ACE were metabolically labeled with [ 32 P]orthophosphate, and the anti-ACE immunoprecipitates were analyzed by autoradiography (Fig. 10A). Three of the ACE mutants were radiolabeled as was the Wt ACE, indicating that Ser 713 , Ser 718 , or Ser 737 were not phosphorylated. On the contrary, the mutant ACE S730A was not radiolabeled (overexposure of the film did not show any labeling, data not shown). Western blotting with anti-ACE antibody indicated that all mutants were expressed at approximately similar amounts (data not shown). A different set of ACE mutants, in which the serine residues were mutated to glutamic acids, showed essentially similar results (data not shown). These mutational studies supported the mass spectroscopic observation that Ser 730 was phosphorylated and demonstrated that only Ser 730 , and no other serine residue, was phosphorylated.
To address a putative role of phosphorylation in determining the rate of cleavage secretion, we compared the Wt protein with the S730A mutant protein: the rates of their cleavage secretion were indistinguishable ( Fig 10B). Similar to the Wt protein, this mutant protein and other Ser to Ala mutant proteins co-immunoprecipitated with CaM (data not shown). Moreover, CaMI stimulated cleavage secretion of all mutants equally well (data not shown). Finally, another mutant, S730D, in which the target Ser residue had been replaced by the phosphoserine mimetic Asp, was cleaved at the same rate as the Wt and the S730A mutant proteins, and both PMA and CaMI enhanced the cleavage of all three proteins (TABLE TWO). As expected, the S730D mutant protein was not phosphorylated (data not shown). When taken together, these data indicate that Ser 730 phosphorylation has no bearing on the rate of cleavage of ACE but binding of proteins, such as CaM, regulates the process.

DISCUSSION
ACE is a type-I membrane protein that undergoes ectodomain shedding at membrane proximal region to generate an enzymatically active soluble form, which is found in the body fluids (21,22). ACE plays a central role in blood pressure regulation. Transgenic studies have shown that secreted ACE alone could not maintain a normal blood pressure; expression of ACE in vascular endothelial cells is absolutely necessary (20). A suitable balance between the levels of ACE on the cell surface and in the serum is critical for maintaining a normal blood pressure. Therefore, the regulation of ACE cleavage secretion might be necessary for maintaining this balance. Our previous studies have demonstrated that ACE shedding is a regulated process, which could be stimulated by PMA, an activator of protein kinase C (24), or pervanadate, an inhibitor of tyrosine phosphatases (28), and inhibited by inhibitors of specific class of metalloproteases (22). Specific cellular proteins are associated with ACE molecules and regulate the ACE shedding process (27).
Results presented here show that, in addition to PMA, CaM inhibitors stimulate ectodomain shedding of ACE. Cleavage of ACE induced by CaMI, as well as PMA, is highly sensitive to the metalloprotease inhibitor Compound 3, suggesting that CaMI-and PMA-mediated ACE shedding machinery is converged to the same metalloprotease-dependent step. An inhibitor of CaMKII, KN-93, also stimulates the ACEectodomain shedding in a dose-dependent fashion, indicating the roles of Ca 2ϩ -regulatory protein in the shedding process. While analyzing the proteins associated with ACE molecule, we determined that CaM is associated to ACE. The association of CaM with ACE could be disrupted by CaMI, as well as PMA (data not shown). Although CaMI and PMA share these common characteristics in ACE shedding, differences exist in their modes of action. PMA-stimulated ACE shedding does not require the cytoplasmic domain of ACE, whereas the action of CaMI does. A small region of the cytoplasmic domain (709 -719) was identified as the region necessary for CaMI-induced ACE shedding, as well as CaM-binding to ACE. Thus, dissociation of CaM from the cytoplasmic domain of ACE, which happens in response to either CaMI or PMA, cannot be the only cause of enhanced cleavage secretion. However, it is possible that both agents use this mechanism but PMA uses others in addition, which could be the reason why the effect of PMA is more pronounced than that of CaMI.
In a previous study (25), we reported that only the extracellular domain of ACE is necessary for its cleavage secretion. This domain, when attached to a non-secreted membrane protein, could cause its cleavage secretion. But in the same study, we also observed that a cytoplasmic tail-less mutant of ACE was secreted more efficiently than the Wt protein (see Fig. 1 in Ref. 25), indicating a negative regulatory role of the cytoplasmic tail. In this study we observed that CaM binds to the cytoplasmic domain of ACE, and this binding is responsible for downregulating ACE secretion. When CaM is dissociated from the ACE cytoplasmic domain by agents such as CaMI, or when it cannot bind to ACE because of a deletion of its binding site, the cleavage secretion is stimulated. Thus we have uncovered a physiological negative regulation of ACE shedding that is mediated by the cytoplasmic domain, a domain that is not required for recognition by the ACE-secretase, but used to negatively regulate the process.
Ectodomain shedding of several other cell-surface molecules, includ-ing ␤-amyloid precursor protein, transforming growth factor-␣, receptor tyrosine kinase, Trk A, is stimulated by CaM inhibitors (38). The CaMI-induced proteolytic processing of these proteins is also mediated by metalloprotease-dependent pathways. Ectodomain shedding of CD44, an adhesion molecule for hyaluronic acid, has been reported to be stimulated by CaMI, and the CaMI-induced shedding of CD44 is mediated by the protease, ADAM10 (39). PMA also stimulates CD44 shedding; however, this is a ADAM17-mediated event. Thus, stimulations of secretion by CaMI and PMA can be mediated by different pathways. Calmodulin has been shown to interact with L-selectin, a leukocyte cell-surface adhesion molecule, and dissociation of CaM from L-selectin by CaMI induced its proteolytic release from the cell surface (40). Another study revealed that CaM binding occurs at the membrane proximal region of the L-selectin cytoplasmic domain, and mutations within this region dissociate CaM and induce shedding of L-selectin (41). Similar to these observations, it has recently been reported that CaM is bound to glycoprotein VI, a platelet membrane glycoprotein and is dissociated by CaMI, W7 (42). Dissociation of CaM from glycoprotein VI resulted in a time-dependent loss of glycoprotein VI from the platelet-surface. In another study (43), it has been demonstrated that CaMIs trigger intracellular signaling events that led to cell-surface proteolytic processing of MT1-MMP but do not directly associate with MT1-MMP to induce its shedding process. This indicates that CaMI induces proteolytic processing of MT1-MMP by mechanisms independent of the association of CaM with its cytoplasmic domain. The above examples illustrate that, although shedding of many ectoproteins is enhanced by CaMI, the underlying mechanisms could be quite different. Most of our mechanistic study was performed with germinal (g) isozyme of ACE, but cleavage of the somatic (s) isozyme is expected to be regulated similarly, because the C-terminal half of sACE, which encompasses all of the cytoplasmic and transmembrane domains and much of the ectodomain, is identical to gACE. Moreover, the cleavage sites on the two proteins are the same, and secretion of both is inhibited by Compound 3 (21,22). In line with these facts, we observed that secretion of sACE by human endothelial cells was also stimulated by CaMI (Fig. 1C).
In the present study we also showed that gACE was phosphorylated on a specific serine residue (Ser 730 ) in the cytoplasmic domain and no other phosphoserine residue could be detected in the ACE molecule. The phosphorylation of ACE was inhibited by CaMI, as well as PMA leading to the idea that unphosphorylated ACE might be cleaved more efficiently. But a series of experiments disproved this idea. First, both phosphorylated and unphosphorylated Wt ACE proteins were shown to be cleaved. Second, CaM was bound to both Wt and S730A mutant (data not shown). Third, the unphosphorylated mutant of ACE (S730A) was shed at a rate similar to that of Wt ACE, and its shedding could be stimulated by CaMI and PMA as well. Fourth, the S730D mutant, which mimics the completely phosphorylated Wt protein, was not cleaved more efficiently, and its basal rate of cleavage was enhanced by CaMI or PMA as well. It is, therefore, safe to conclude that ACE serine-phosphorylation does not directly regulate ACE shedding.
With regards to the possible role of phosphorylation in the regulation of secretion, it is worth comparing our observations with those of Kohlstedt et al. (44). These authors reported that human sACE is phosphorylated at Ser 1270 , which corresponds to gACE 730, the exact site of phosphorylation reported here. But, these authors claimed that this phosphorylation is mediated by casein kinase 2, and it inhibits cleavage secretion of human sACE. As elaborated above, in our system we could not confirm these observations. In mouse ACE89 cells, the casein kinase 2 inhibitor DRB had no effect on gACE cleavage secretion, whereas in

CaMI-and PMA-induced cleavage secretion of ACE is not dependent on the phosphorylation status of ACE
HeLa cells were transfected with wild type or unphosphorylated mutants of ACE (S730A and S730D) then pulse-labeled with ͓ 35 S͔methionine, and the label was chased in the presence (ϩ) or absence (Ϫ) of PMA for 1 h. For CaMI-induced cleavage, the label was chased for 3 h. Cleavage secretion was quantified as described in Fig. 1. human HeLa cells it did enhance the process. But the latter effect is unlikely to be mediated by Ser 730 dephosphorylation, because the S730A mutant was equally affected. The apparent differences between the two groups' results could be attributed to several possible reasons. Although human ACE and rabbit ACE have strong sequence conservations, the cytoplasmic tails are not identical; for example, there are four Ser residues in the rabbit protein but five in the human protein.