key: cord-0007982-uwpo6zvd
authors: Dalle Ore, Florence; Ajandouz, El Hassan; Giardina, Thierry; Puigserver, Antoine
title: The membrane-bound basic carboxypeptidase from hog intestinal mucosa
date: 1999-10-15
journal: Biochim Biophys Acta Biomembr
DOI: 10.1016/s0005-2736(99)00122-4
sha: e9b16f89ed2b21d008719bfd2bdb3be8a9b4dda1
doc_id: 7982
cord_uid: uwpo6zvd

The carboxypeptidase activity occurring in hog intestinal mucosa is apparently due to two distinct enzymes which may be responsible for the release of basic COOH-terminal amino acids from short peptides. The plasma membrane-bound carboxypeptidase activity which occurs at neutral optimum pH levels was found to be enhanced by CoCl(2) and inhibited by guanidinoethylmercaptosuccinic acid, o-phenanthroline, ethylenediamine tetraacetic acid and cadmium acetate; whereas the soluble carboxypeptidase activity which occurs at an optimum pH level of 5.0 was not activated by CoCl(2) and only slightly inhibited by o-phenanthroline, ethylenediamine tetraacetic acid, NiCl(2) and CdCl(2). The latter activity was presumably due to lysosomal cathepsin B, which is known to be present in the soluble fraction of hog intestinal mucosa. Although the membrane-bound enzyme was evenly distributed along the small intestine, it was not anchored in the phospholipidic bilayer via a glycosyl-phosphatidylinositol moiety, as carboxypeptidase M from human placenta is. The enzyme was not solubilized by phosphatidylinositol-specific phospholipase C, but was solubilized to practically the same extent by several detergents. The purified trypsin-solubilized form is a glycoprotein with a molecular mass of 200 kDa, as determined by performing SDS–PAGE and gel filtration, which differs considerably from the molecular mass of human placental carboxypeptidase M (62 kDa). It was found to cleave lysyl bonds more rapidly than arginyl bonds, which is not so in the case of carboxypeptidase M, and immunoblotting analysis provided further evidence that hog intestinal and human placental membrane-bound carboxypeptidases do not bear much resemblance to each other. Since the latter enzyme has been called carboxypeptidase M, it is suggested that the former might be carboxypeptidase D, the recently described new member of the carboxypeptide B-type family.

The membrane-bound basic carboxypeptidase from hog intestinal mucosa 1

Basic carboxypeptidases are metallo-enzymes which cleave COOH-terminal arginine and lysine from peptides and proteins. They occur widely among mammals, and the secretory pancreatic carboxypeptidase B (CPB) is generally taken to be the most representative member of this class of enzymes [1] . Although pancreatic CPB is responsible for the intestinal digestion of food proteins and peptides, the other basic carboxypeptidases are not digestive enzymes but have regulatory e¡ects on the activity of biological peptides such as anaphylatoxins, enkephalins and bradykinins [1] . Since all the mammalian basic carboxypeptidases have similar enzymatic activities, they may play di¡erent roles in vivo, depending on their localization and physical properties.

Human plasma CPN (which is also known as kininase I) is synthesized in the liver and released into the bloodstream in the form of a large tetrameric complex (280 kDa) containing two active subunits (48^55 kDa) and two inactive glycosylated subunits (83 kDa), the latter of which stabilize the active ones and keep them in the solubilized form [2, 3] . Since CPN has mainly been found in circulating blood, its most likely function seems to be to protect the organism from the action of potent vasoactive and in£ammatory peptides such as kinins and/or anaphylatoxins [1, 4] . A plasma pro-CPB of hepatic origin which is structurally similar to pancreatic pro-CPB has recently been identi¢ed as a circulating protein, isolated from the blood, and found to form a non-covalent association with other plasma proteins [5, 6] . Unlike pancreatic CPB, which has a purely digestive function, plasma CPB might be involved in ¢brinolysis and has been de¢ned as a thrombin activatable ¢brinolysis inhibitor, or TAFI [7, 8] .

CPH, which is also known as CPE or enkephalin convertase, is located in the secretory granules of pancreatic islets, as well as in the adrenal and pituitary glands and brain [9^13] , where it is present in both the soluble and membrane-bound forms [14, 15] . This enzyme di¡ers from the other basic carboxypeptidases in that it has an acidic optimum pH, which is consistent with the acidic interior of the granules [16, 17] . CPH is probably involved in the removal of residual COOH-terminal arginine or lysine residues from peptidic prohormones during their processing in secretory granules [1, 18] . Another recently discovered enzyme, CPD, which has similar characteristics but a di¡erent molecular mass, may also be involved in the processing of the peptides and proteins that transit via the secretory pathway [19, 20] .

CPM is a widely distributed membrane-bound basic carboxypeptidase which has been puri¢ed to homogeneity, cloned and sequenced from human placenta [21, 22] , where it is attached to the plasma membrane via a phosphatidylinositol glycan tail [23] . In view of its presence in plasma membranes and the fact that its optimal activity occurs at neutral pH levels, CPM might well participate in the local control of peptide hormone activity. Its proximity to plasma membrane receptors suggests that it may either inactivate or a¡ect in some way the receptor speci¢city of peptide hormones released at local tissue sites [1] .

A novel cDNA encoding CPZ has recently been identi¢ed based on its homology with known metallocarboxypeptidases [24] . CPZ was expressed in the baculovirus system and found to be more active at neutral pH levels than at pH 5.5 toward substrates with COOH-terminal basic amino acids [24] . Lastly, AEBP1 was identi¢ed as a transcription repressor showing some homologies with metallocarboxypeptidases. Although it lacks the residues on which carboxypeptidase activity is thought to depend, this enzymatic protein has been reported to cleave peptides containing a COOH-terminal arginine residue [25] .

The increasing number of data becoming available nowadays on biologically active peptides has led to considerable attention being paid to mammalian basic carboxypeptidases. Little is known so far, however, about these non-digestive basic carboxypeptidases located in the intestinal tract, although it has been reported that most of the CPH activity in the rat intestine is membrane-bound, based on the ¢nding that it is stimulated by CoCl 2 and inhibited by low concentrations of guanidinoethylmercaptosuccinic acid (GEMSA) at low pH levels [26] . CPM has the same properties as CPH [27] , and the presence of both enzymes has been brie£y described in the dog small intestinal wall [28] , but no further characterization has been performed. Here, we describe the puri¢cation procedure used and give some of the characteristics of the membrane-bound basic carboxypeptidase from hog intestine mucosa, an enzyme which de¢nitely di¡ers from the other known carboxypeptidases but which is similar in some respects to the newly described CPD, the mammalian homologous form of duck gp180 protein, a 180 kDa hepatitis B virus-binding glycoprotein.

Porcine intestines were obtained from the Marseilles slaughterhouse. Hippuryl-L-arginine, trypsin, papain, subtilisin, as well as all the activators, inhibitors and detergents used here were purchased from Sigma Chemicals (St. Louis, MO, USA). Phospholipase from Bacillus cereus was from Boehringer (Mannheim, Germany). Arginine-Sepharose, phenyl-Sepharose, Sephacryl S200 HR were from Pharmacia Biotech (Uppsala, Sweden). The electrophoretic molecular mass markers were obtained from Bio-Rad Laboratories (Richmond, CA, USA). HCl, PITC and the amino acid analysis standards were purchased from Pierce (Rockford, IL, USA), while the polyvinylidene di£uoride membrane was from Millipore (Bedford, MA, USA) and the nitrocellulose sheets (0.2 Wm) from Schleicher and Schuell (Dassel, Germany). The rabbit polyclonal antiserum directed against human placental CPM and the IgG fraction of goat anti-rabbit serum conjugated with horseradish peroxidase were purchased from Tebu (Le Perray-en-Yvelines, France) and Sigma, respectively. Solvents (pure grade) were supplied by SDS (Peypin, France). All the other chemicals used were of reagent grade.

All the enzyme activities were measured at 37³C. Basic carboxypeptidase (CP) activity was measured with hippuryl-L-Arg as the substrate. The activity was ¢rst quanti¢ed by performing an HPLC assay at 254 nm of the hippuric acid released. The enzyme was incubated with 1 mM hippuryl-L-Arg in a 0.1 M Tris^HCl bu¡er (pH 7.5) containing 0.14 M NaCl. The reaction was stopped by adding 20 Wl of acetic acid to 200 Wl of reaction mixture. After extraction with 200 Wl of Cl 3 CH, the aqueous phase was removed and ¢ltered through Millipore ¢lters (0.45 Wm), and 15-Wl samples were injected into the Vydac C18-RP column (0.4U25 cm, 5 Wm) and separated at a £ow rate of 1 ml/min on a 10-min linear gradient from 10% to 25% acetonitrile in 0.01% TFA containing H 2 O. The basic CP activity was routinely determined spectrophotometrically during the puri¢cation steps with the same substrate, as previously described by Koheil and Forstner [29] , and Skidgel and Erdo « s [30] . One unit of CP activity was de¢ned as 1 Wmol of hippuryl-L-Arg hydrolysed per minute.

The activity of aminopeptidase N and alkaline phosphatase, two microvillus membrane markers, was assayed as described by Louvard et al. [31] , whereas (Na ,K )-ATPase, which was used as the basolateral plasma membrane marker, was assayed using the method developed by Murer et al. [32] . The activity of NADPH-cytochrome c reductase, a microsomal contamination marker [33] , was assayed as described by Sottocasa et al. [34] , while cytochrome c oxidase, which was used as mitochondrial marker, was assayed using Cooperstein and Lazarow's method [35] . Acid phosphatase, a lysosome marker, was assayed according to Murer et al. [32] , and cathepsin B activity was determined using the speci¢c substrate Z-Arg-Arg-AMC [36] as described by Gabrijelcic et al. [37] . Proteins were determined as described by Bradford [38] with bovine serum albumin as the standard.

Immediately after the hogs were killed, their small intestines were removed, washed with a 50 mM TrisĤ Cl bu¡er (pH 8.0) and placed at 4³C as soon as possible. The mucosa was scraped o¡ with a microscope slide and kept at 320³C until being further processed. Crude brush-border membrane preparations were obtained as previously described by Schmitz et al. [39] and modi¢ed by Maury et al. [40] , with additional slight modi¢cations. Hog intestinal mucosal scrapings (200 g) were homogenized in four times their mass of a 5 mM Tris^HCl bu¡er (pH 7.3) containing 0.25 M sucrose, 10 mM KCl and 1 mM MgCl 2 in a motor-driven Te£on^glass homogenizer, and ¢ltered through a gauze. The resulting homogenate was submitted to di¡erential centrifugation, and the ¢nal brush-border membrane-enriched pellet was resuspended in 100 ml of a 20 mM Tris^HCl bu¡er (pH 7.3) containing 0.14 M NaCl, 10 mM KCl and 1 mM MgCl 2 . The other pellets were resuspended in 200 ml of the same bu¡er.

The pH^activity pro¢les were determined at 37³C with hippuryl-L-Arg as the substrate in the following two bu¡ers: 0.1 M sodium acetate (pH 3.0 to 7.0) and 0.1 M Tris^HCl (pH 7.0 to 9.0) containing 0.14 M NaCl. The e¡ects of a number of reagents on basic CP activities were investigated after a 15-min incubation period at 37³C in the bu¡ers used for the enzyme activity determinations described above. The reaction was started by adding the hippuryl-L-Arg substrate and the relative activity was expressed as a percentage of the control value.

Small intestines, stomachs and colons were removed immediately after the hogs were killed and washed with a 50 mM Tris^HCl bu¡er (pH 8.0) and stored at 4³C. The intestines were further cut into 1-m pieces, their mucosa were scraped o¡ with a microscope slide and crude brush-border membrane fractions were prepared as described above.

Crude brush-border membrane preparations were washed ¢ve times with a 20 mM Tris^HCl bu¡er (pH 7.3) containing 2 M NaCl, 10 mM KCl and 1 mM MgCl 2 . The ¢nal pellets were resuspended in the same bu¡er containing only 0.14 M NaCl, and the resulting suspensions (about 2 mg/ml of proteins) were incubated with various detergents under gentle shaking at 4³C for 60 min. Suspensions were then centrifuged at 105 000Ug for 30 min and the enzyme activities in the pellets and supernatants were measured. Solubilized activities were expressed as percentages of total activities.

Prior to the solubilization experiments with PI-PLC from Bacillus cereus, the ¢nal pellets were resuspended in a 0.1 M HEPES bu¡er (pH 8.0) containing 1 mM PMSF, 0.1 mM PCMS, 0.1 mM TLCK and 1 mM leupeptine in order to inhibit the action of any brush-border membrane proteases and proteases possibly contaminating the commercial PI-PLC preparation used. The resulting suspensions (about 2 mg/ml of proteins) were incubated at 37³C for 2 h with 1 U/ml of PI-PLC from Bacillus cereus (1 unit of PI-PLC cleaves 1 nmol of phosphatidylinositol per minute). We then proceeded as described in the case of detergents.

As regards the solubilization process used with proteases, the ¢nal pellets were resuspended in a 20 mM Tris^HCl bu¡er (pH 7.5) containing 0.25 M sucrose to obtain a 0.5 mg/ml protein solution. After the diluted membrane fractions had been incubated under gentle shaking at 4³C with increasing amounts of trypsin, papain and subtilisin, they were then processed as described in the case of detergents.

Once the brush-border membranes had been incubated with 10% trypsin (w/w) at 4³C for 24 h, trypsin was inhibited with 4 mM PMSF. The suspension was then centrifuged at 105 000Ug for 30 min and the resulting supernatant concentrated and dialyzed against a 20 mM Tris^HCl bu¡er (pH 7.5) in an Amicon Concentrator with a PM-10 membrane. Solubilized CP was separated from the other proteins by performing chromatography using a Waters 650 E FPLC system, and 5-ml fractions were collected. The solubilized protein concentrate (about 50 ml) was ¢rst loaded on an arginine-Sepharose column (2.2U10 cm) pre-equilibrated with the dialysis bu¡er at a £ow rate of 1 ml/min. After washing the column with the equilibration bu¡er, the proteins were eluted with a 90-min linear gradient from 0 to 0.15 M NaCl followed by a 30-min linear gradient from 0.15 to 0.5 M NaCl in the same bu¡er. The active fractions were pooled and dialyzed overnight against 5 l of a 20 mM Tris^HCl bu¡er (pH 7.5) containing 1 M (NH 4 ) 2 SO 4 . The dialyzed solution was then applied, at a £ow rate of 1 ml/min, to a phenyl-Sepharose column (2.2U10 cm) pre-equilibrated with the dialysis bu¡er, which was subsequently washed with the equilibration bu¡er. The proteins were eluted with a 120-min linear gradient from 1 to 0 M (NH 4 ) 2 SO 4 in the same bu¡er. The active fractions were pooled, dialyzed overnight against 10 l of a 20 mM TrisĤ Cl bu¡er (pH 7.5) containing 0.35 M NaCl and concentrated to 1 ml in an Amicon Concentrator equipped with a PM-10 membrane. The concentrate was further applied to a Sephacryl-S200 HR column (2.6U60 cm) pre-equilibrated with the dialysis bu¡er, at a £ow rate of 1 ml/min. After eluting the active fractions with the same bu¡er, they were pooled, dialyzed overnight against 5 l of a 20 mM TrisĤ Cl bu¡er (pH 7.5) containing 0.14 M NaCl, and the protein material was concentrated to 2 ml in an Amicon Concentrator with a PM-10 membrane.

SDS^PAGE was performed using Laemmli's method [41] with 10% slab gels. Native PAGE was also carried out in 10% slab gels as described by Gabriel [42] . Proteins were detected by performing Coomassie blue or silver staining, while glycosylated proteins were detected with Schi¡ reagent according to Zacharius et al. [43] with bovine serum albumin and fetuin from fetal calf serum as the standards.

The molecular mass of the native enzyme was estimated by performing gel ¢ltration using a Sephacryl-S200 HR column (2.6U60 cm) under the same ex-perimental conditions as for the puri¢cation procedure.

Proteins were electrotransferred from slab gels to PVDF membranes as described by Matsudaira [44] , stained with Ponceau red and excised from the membrane for analysis. Amino acid compositions were determined using the Picotag procedure [45] , after hydrolyzing the proteins with 6 N HCl at 110³C for 24 h and derivatizing the resulting amino acids with PITC before injecting them into the Picotag Waters column (0.39U30 cm). Edman degradation was carried out on an Applied Biosystems Sequencer Model 470 according to Hewik et al. [46] . Phenylthiohydantoin (PTH) identi¢cation was carried out by means of an Applied PTH column (0.21U22 cm, 5 Wm).

Proteins were electro-transferred from polyacrylamide gels to nitrocellulose as described by Burnette [47] . After the transfer, Ponceau red protein staining and immunoprinting were carried out according to Coudrier et al. [48] as described by Gorvel et al. [49] .

Basic CP activity was detected in homogenates from hog intestinal mucosa assayed with hippuryl- At each step in the subcellular fractionation procedure, the enzyme activities were measured in the pellet and the supernatant and expressed as a percentage of the total activity. Results are means based on three separate subcellular fractionations.

L-Arg at pH 7.5. Subcellular fractionation yielded a microvillar membrane fraction containing about 40% of the total basic CP activity and a soluble fraction containing the remaining 60% (Table 1) . When the microvillar membrane fraction was washed with 2 M NaCl, no signi¢cant release of basic CP activity into the ¢nal supernatant was observed, which indicates that the enzyme was probably tightly bound to the intestinal cell membranes. Two di¡erent pH activity pro¢les were obtained when the membrane-bound and soluble CP activities were tested with hippuryl-L-Arg at pH values ranging from 3.0 to 9.0 (Fig. 1) . In the microvillar membrane fractions, the activity was highest at pH 7.5, and the enzyme was still 72% active at pH 9.0, whereas the activity dropped sharply at acidic pH levels and was practically nil below pH 4.5. These ¢ndings are in agreement with those obtained on CPM from human placenta and from MDCK cells [21, 50] . Hog intestinal basic CP showed the highest level of activity at pH 5.0 with the soluble enzyme, and was still 57% active at pH 3.0, but its activity dropped sharply at basic pH levels. All these data suggested either that CPH may have been present, or that the lysosome did not remain intact during the subcellular fractionation, which would result in the soluble fractions being contaminated with lysosomal enzymes, which was apparently the case (Table 1) . It was therefore concluded that the soluble CP activity showing an acidic optimum pH might be that of cathepsin B (also named lysosomal carboxypeptidase), which is not a metallocarboxypeptidase, but a cysteine proteinase [51] . This hypothesis was further con¢rmed upon testing the e¡ects of some inhibitors and activators on the enzyme activity at acidic pH levels.

As shown in Table 2 , the membrane-bound CP activity was inhibited by o-phenanthroline and Chemicals were incubated at 37³C for 15 min at pH 7.5 and pH 5.0 with hog intestinal microvillar membrane fractions and soluble fractions, respectively, before adding the substrate hippuryl-L-Arg. The activities are expressed as percentages of the control values, and the results are means of duplicate assays on two separate subcellular fractionations. a On human placental CPM [21] b On bovine adrenal enkephalin convertase (CPH) [9] . c On lysosomal carboxypeptidase (cathepsin B) [9] .

EDTA, which strongly suggests that zinc or some other cation was required for the enzyme activity to occur, whereas the serine protease inhibitor PMSF, as well as PCMS, the sulfhydryl protease inhibitor which is known to be active on CPH, had no appreciable e¡ects. The fact that no SH group was involved in the enzymatic activity was con¢rmed by the lack of e¡ect of HgCl 2 . As in the case of other mammalian basic CP, CoCl 2 increased the membrane-bound CP peptidase activity, while cadmium acetate was slightly inhibitory. As regards the soluble basic CP activity, no activation was observed with CoCl 2 , the speci¢c activator of basic carboxypeptidases, and only 50% inhibition occurred with o-phenanthroline, whereas EDTA and the two metal cations Ni 2 and Cd 2 were almost ine¡ective, in sharp contrast with what occurs in the case of CPH [9] . The subcellular localization, optimum pH and spe-ci¢c activation or inhibition by both CPM and hog intestinal membrane-bound CP reagents were very similar indeed (Table 2) . It seems very likely that the soluble CP activity was mainly due to a lysosomal cathepsin, since this enzyme was also detected in the soluble fractions (data not shown) in the assay on cathepsin B activity with the speci¢c substrate Z-Arg-Arg-AMC. However, the possibility that a soluble form of CPH or any other basic CP may have been present cannot be de¢nitely ruled out.

As shown in Fig. 2 , basic CP activity was detected especially in the brush-border membrane preparations from hog small intestine. No signi¢cant di¡erence between the duodenum, jejunum and ileum was observed, in agreement with the results obtained by Lynch et al. [26] on the localization of the membrane-bound form of CPH in the rat gastrointestinal tract. Since the methods used in the latter case were not speci¢c to CPH, but can be used to detect all the basic CPs, the enzyme activity of the rat gastrointestinal tract may also involve CPM. Here we observed that, in the rat intestine, the speci¢c activity of the microvillar basic CP (about 22.5 nmol/min.mg) does not di¡er signi¢cantly between the duodenum, jejunum and ileum (data not shown).

Since hog intestinal basic CP is apparently ¢rmly attached to the microvillar plasma membrane, we compared its pattern of solubilization by detergents, phospholipase C and proteases in order to gain insights into the way in which the CP is anchored to the membrane. Fig. 3 clearly shows that the pattern of solubilization of the basic CP activity by a series of six detergents was comparable to that of aminopeptidase N, but di¡ered greatly from that of alkaline phosphatase. The latter two enzymes are generally thought to be suitable markers for labeling proteins anchored by a hydrophobic amino acid residue sequence and by a GPI moiety, respectively, as pointed out by Hooper and Turner [52] in their study on pig kidney microvillar ectoenzymes. It is worth mentioning that only octyl-glucoside and CHAPS, two detergents with high CMC values, e¡ectively released substantial amounts ( s 60%) of GPI-anchored alkaline phosphatase activity, whereas all the detergents used solubilized 60^80% of the membrane-bound APN and basic CP activities.

When hog intestinal microvillar membrane fractions were incubated with PI-PLC from Bacillus cereus, a speci¢c release ( s 50%) of alkaline phosphatase was observed, but neither aminopeptidase N nor basic CP were solubilized (data not shown). The presence of a mixture of protease inhibitors during the incubation of membrane fractions with PI-PLC, as indicated in Section 2, rules out the possibility that contaminating proteases may have been involved in the release of the GPI-anchored enzyme. This result therefore con¢rms the above conclusion that hog intestinal microvillar basic CP may not be bound to the membrane via a GPI tail like human placental CPM [23] , but via a hydrophobic peptidic anchor. The same conclusion was previously reached by Hooper and Turner [53] in their study on microvillar basic CP from pig and human kidney.

As shown in Fig. 4 , when hog intestinal microvillar membrane fractions were incubated at 4³C with 5% trypsin (w/w), as much as 40% of the total basic CP activity was released after a 24-h treatment, and more than 80% with 10% trypsin (w/w). Aminopeptidase N amounted to 10% and 50%, respectively, whereas no alkaline phosphatase was released at all. Similar results were obtained with papain and subtilisin (data not shown). After tryptic release of the soluble form of basic CP from its hydrophobic peptide anchor, the enzymatic activity remained intact. In addition, trypsin can be expected to cleave o¡ only a relatively short peptide (about 20^30 amino acids) from the whole protein as observed in the case of CPM [50] . The tryptic release of a hydrophilic form of the basic CP from the microvillar plasma membrane was therefore taken as the starting point for the enzyme puri¢cation procedure. 

Basic CP was puri¢ed from hog intestinal mucosa using a 6-step procedure including the preparation of microvillar plasmic membrane fraction, washing it with 2 M NaCl and treating it with trypsin to solubilize the enzyme, on which an a¤nity chromatography was subsequently performed on arginine-Sepharose, prior to hydrophobic chromatography on phenyl-Sepharose, and ¢nally gel ¢ltration on Sephacryl-S200 HR. As summarized in Table 3 , in a typical experiment starting with 200 g of intestinal mucosa, the isolation procedure yielded about 2 mg of protein with a speci¢c activity of 100 nmol/min per mg. The overall enzyme activity yield was 11% and a 500-fold puri¢cation of the enzyme was obtained. It is worth mentioning that the arginine-Sepharose-bound basic CP activity could not be speci¢cally eluted from the column using the basic CP speci¢c inhibitor GEMSA, which nevertheless completely inhibited the enzyme (data not shown). This might be due to the molecular mass of the protein which is about four times that of the other basic carboxypeptidases.

As shown in Fig. 5 , the protein thus puri¢ed gave a single PAGE band under non-denaturing conditions, which indicates that the isolated protein might be homogeneous, but unfortunately two bands were observed in the presence of SDS. No further improvement of the puri¢cation level could be achieved although a number of other chromatographic procedures were tested.

Equal volumes of native enzyme and sweet potato L-amylase, the molecular mass of which is known to be about 200 kDa, were eluted from the Sephacryl-S200 HR column (data not shown). The molecular masses of the two protein bands separated by SDST PAGE under reducing conditions (Fig. 5 , lane 2) were found to be 200 kDa and 120 kDa, respectively. When the SDS^PAGE was performed under non-reducing conditions, in the absence of DTT, and without any heating of the protein sample (Fig. 5, lane 3) , a single band at 200 kDa was obtained on the contrary, which strongly suggests that the two protein units might be linked by one or more disul¢de bonds. However, when the SDS^PAGE was carried out in the absence of DTT and heating the protein sample before running the electrophoresis (Fig. 5, lane 4) , the 120 kDa protein was separated from the 200 kDa protein in the same way as under reducing and heating conditions. It is possible that the 120 kDa band may re£ect the presence of a proteolytic fragment from the 200 kDa protein.

The sugar-speci¢c pattern of slab gel staining obtained with periodate-Schi¡'s reagent clearly indicated that the two protein bands were glycosylated. Therefore, the native enzyme might be a 200 kDa glycoprotein, as compared with the molecular mass of CPM, which is 60 kDa [21] . The only basic carboxypeptidase which is known to be a single-chain glycoprotein with a molecular mass of 180 kDa is actually CPD, a carboxypeptidase which was recently characterized and found to have similar enzymatic properties to those of CPH [19] . Upon being subjected to SDS^PAGE, the puri¢ed membrane-bound and soluble forms of CPD from bovine pituitary gland showed a single 180 kDa band in the former case, and a 170 kDa and a 135 kDa band in the latter case [20] . The two polypeptide chains observed upon performing SDS^PAGE on the solubilized form of the basic CP puri¢ed from hog intestine were similar to those observed in the soluble form of CPD, and thus suggested that a membrane-bound CPD may also exist in the intestine.

The basic CP puri¢ed in this study was able to cleave hippuryl-L-Lys about 1.5-fold faster than hippuryl-L-Arg, contrary to what has been found to occur in the case of CPM [21] , and its optimum pH was about 6.5 with the latter substrate. Since the enzyme was not detected when immunoblotting was performed with a polyclonal antiserum against human placental CPM, intestinal CP can de¢nitely be said to be a structurally distinct protein from human placen- Results are expressed as a percentage of total residues determined. n.d., not determined. a Mean value based on two analyses. b Deduced from the cDNA nucleotide sequence of CPM [22] . c Deduced from the cDNA nucleotide sequence of CPD [54] .

tal CPM-type enzyme. Since no NH 2 -terminal residue was detected using the Applied Biosystems Automatic Sequencer, the protein N-terminus was taken to be blocked. The amino acid composition of the two solubilized forms of CP from intestine was determined and compared with those of human CPM and CPD (Table 4 ). The two proteins separated by SDS^PAGE showed similar amino acid contents, except for Asx and Glx, which suggests that the lower molecular mass form might be due to some proteolysis of the higher molecular mass form. The 200 kDa glycoprotein had roughly the same amino acid distribution as CPM and CPD, but its lysine content was not so high. However, although the amino acid compositions of hog intestinal CP, human CPM, and human CPD are comparable, the exact identity of the intestinal enzyme still needs to be con¢rmed.

The COOH-terminal arginine and lysine from peptides and proteins may be cleaved by several carboxypeptidases, which show the same pattern of carboxypeptidase B activity, but have di¡erent molecular and enzymatic characteristics [1] .

The results of the present study show that hog intestinal mucosa contains at least a membranebound carboxypeptidase capable of releasing basic COOH-terminal amino acids from short peptides. The trypsin-solubilized form of CP was isolated from the hog intestine, with a speci¢c activity of 100 nmol/min per mg using hippuryl-L-Arg as the substrate. According to its subcellular localization and optimum pH, the enzyme activity greatly resembles that of CPM in the human placenta as well as that of MDCK cells [21, 50] . The basic CP activity was equally distributed between the duodenum, jejunum and ileum, in agreement with the pattern of distribution of CPH in the rat gastrointestinal tract [26] . However, since the methods used by the latter authors were not really speci¢c to CPH activity [27] , the enzyme involved might have been CPM, and this might explain why CPH was sometimes found in areas apparently devoid of any endocrine function in the rat gastrointestinal tract. Although we observed that the ratio between intestinal membranebound and soluble CP activities was the same as that between membrane-bound and soluble CPH [26] , it is worth noting that two distinct enzymes with di¡erent optimum pH values and behavior towards activators and inhibitors were found to exist in hog intestine. The soluble CP activity was probably due to cathepsin B, a cysteine enzyme of lysosomal origin with several endo-and exopeptidase activities [51, 55] usually found in the gastroduodenal mucosa [56, 57] . This enzyme, which is not a metallocarboxypeptidase, was not further characterized in this study.

Hog intestinal basic CP did not appear to have the characteristics of a GPI-anchored enzyme, judging from its pattern of solubilization by a range of detergents, in addition to the fact that it was not released from microvillar membrane preparations in response to PI-PLC. Consequently, this cell-surface peptidase is probably anchored by a sequence of hydrophobic amino acid residues to the lipid bilayer. The basic carboxypeptidases present in human and hog kidney microvillar membranes are known to have the same pattern of solubilization by detergents and to resist solubilization by PI-PLC [53] , whereas CPM from human placenta and that from MDCK cells are anchored via a GPI tail [23, 50] . The discovery in human placenta of a membrane-bound carboxypeptidase which was not GPI-anchored and had di¡erent characteristics from those of CPM [23] may support the idea that membrane-bound CPM is not the sole microvillar carboxypeptidase. Since Hooper and Turner [53] described the solubilization of basic CP from the rat kidney without further characterizing the enzyme, it is therefore possible that the microvillar membrane-bound carboxypeptidase is another member of the B-type carboxypeptidase family. A placental enzyme distinct from CPM was recently partially puri¢ed and denoted CPD [58] , a basic carboxypeptidase which was still unknown when the present work was carried out.

Although the basic CP from hog intestinal mucosa has some properties in common with CPM, it is definitely a distinct enzyme, since CPM is a GPI-anchored single-chain 60 kDa glycoprotein, whereas hog intestinal CP is apparently a single-chain 200 kDa glycoprotein which is anchored to the membrane by a number of hydrophobic amino acid residues. Moreover, intestinal CP cleaved lysyl bonds faster than arginyl bonds, contrary to CPM, and immunoblotting analysis provided further evidence that hog intestinal CP was di¡erent from CPM. The molecular mass of the intestinal CP (200 kDa) was not very di¡erent from that of the membranebound CPD (180 kDa), which was ¢rst described in the bovine pituitary gland as a CPH-like enzyme [19, 20] . CPD is a widely distributed enzyme [20, 58] which has recently been partly puri¢ed from human placental microvilli [58] , indicating that CPM is not the sole microvillar carboxypeptidase of the B-type. The optimum pH of the puri¢ed intestinal CP was actually between 6.0 and 7.0, which is similar to that of human placental CPD (5.5^6.5).

The sole membrane-bound CP which was isolated and characterized from hog intestine in the present study was identi¢ed as a new member of the B-type metallocarboxypeptidases, and is consequently di¡erent from CPB, CPN, pCPB, CPH and CPM. It appeared to be similar to the newly described CPD, which is the mammalian homologue of duck gp180, a 180 kDa hepatitis B virus-binding glycoprotein which was recently cloned and sequenced [59] . It is worth noting that unlike bovine CPD, the N-terminus of which was determined chemically [19, 20] , no NH 2 -terminal residue could be detected when NH 2terminal Edman degradation was performed on the gp180 protein puri¢ed from duck liver [59] . Although CPD occurs in a wide range of tissues, this is the ¢rst time a CPD-like enzyme is puri¢ed from intestinal tissue. Further characterization of the structural, catalytic and immunological properties of hog intestinal CP is still necessary to de¢nitely con¢rm that this enzyme is a CPD.

As the various carboxypeptidases described so far have the same catalytic properties, their speci¢c biological function must depend mainly on their site of action, which is generally directly related to their physical state (e.g., solubilized vs membrane-bound). It is therefore widely recognized that the pancreatic CPB present in the intestine in the form of a soluble enzyme contributes along with other digestive enzymes to degrading food proteins and peptides, whereas CPH in the secretory granules of neuroendocrine tissues might be involved in the maturation and sorting of peptidic hormones [60, 61] . Two other carboxypeptidases present in most mammalian tissues and organs, namely CPN, which circulates in the blood, and CPM, which is bound to the plasma membrane of several cells, play key physiological roles, consisting of regulating the activity of various peptides. These two enzymes in particular have been described as kininases [62] , although their exact function has not yet been established. The function of CPD is still unknown, but the recent ¢nding that it is located mainly in the trans-Golgi network is consistent with the idea that it is involved in the processing of a broad range of proteins that transit along the secretory pathway, playing a similar role to that of CPH in prohormone processing in the regulated secretory pathway [63] . In addition, the ¢nding that CPD is recycled from the cell surface to the trans-Golgi network constitutes important evidence supporting the suggestion that it may act as a receptor for hepatitis B virus [63] . Another metal containing peptidase, aminopeptidase N, has been found to be a coronavirus receptor [64, 65] . The question as to whether the intestinal membrane-bound basic CP has a speci¢c and exclusive degradative role completing that of pancreatic CPB, or is involved independently or concomitantly in the regulation of some of the active biological peptides present in the luminal secretions of the intestine, as well as in the immunological response of the intestinal mucosa to the presence of exogenous peptidic antigens, still remains to be elucidated.

Basic carboxypeptidases: regulators of peptide hormone activity

Human plasma carboxypeptidase N. Isolation and characterization

Erdo « s, Isolation and characterization of the subunits of human plasma carboxypeptidase N (kininase I)

Potentiation of the anaphylatoxins in vivo using an inhibitor of serum carboxypeptidase N (SCPN). I. Lethality and pathologic e¡ects on pulmonary tissue

Isolation, molecular cloning, and partial characterization of a novel carboxypeptidase B from human plasma

Activation and characterization of procarboxypeptidase B from human plasma

Puri¢cation and characterization of TAFI, a thrombin-activable ¢brinolysis inhibitor

TAFI, or plasma procarboxypeptidase B, couples the coagulation and ¢brinolytic cascades through the thrombin-thrombomodulin complex

Enkephalin convertase: puri¢ca-tion and characterization of a speci¢c enkephalin-synthesizing carboxypeptidase localized to adrenal chroma¤n granules

Enkephalin convertase: a speci¢c enkephalin synthesizing carboxypeptidase in adrenal chroma¤n granules, brain, and pituitary gland

Puri¢cation and characterization of enkephalin convertase, an enkephalin-synthesizing carboxypeptidase

Carboxypeptidase B-like converting enzyme activity in secretory granules of rat pituitary

The insulin-secretory-granule carboxypeptidase H. Puri¢cation and demonstration of involvement in proinsulin processing

Puri¢cation and characterization of a membrane-bound enkephalin-forming carboxypeptidase,`enkephalin convertase

Identi¢cation of the pH-dependent membrane anchor of carboxypeptidase E (EC 3.4.17.10)

Internal pH and state of ATP in adrenergic chroma¤n granules determined by 31 P nuclear magnetic resonance spectroscopy

The internal pH and membrane potential of the insulin-secretory granule

Puri¢cation and characterization of carboxypeptidase D, a novel carboxypeptidase E-like enzyme, from bovine pituitary

Tissue distribution and characterization of soluble and membrane-bound forms of metallocarboxypeptidase D

Human carboxypeptidase, Puri¢cation and characterization of a membranebound carboxypeptidase that cleaves peptide hormones

Skidgel, Molecular cloning and sequencing of the cDNA for human membrane-bound carboxypeptidase M. Comparison with carboxypeptidases A, B, H, and N

Human placental carboxypeptidase M is anchored by a glycosyl-phosphatidylinositol moiety

Cloning and expression of human carboxypeptidase Z, a novel metallocarboxypeptidase

A eukaryotic transcriptional repressor with carboxypeptidase activity

Enkephalin convertase in the gastrointestinal tract and associated organs characterized and localized with

Erdo « s, Enhanced Co 2 activation and inhibitor binding of carboxypeptidase M at low pH. Similarity to carboxypeptidase H (enkephalin convertase)

Erdo « s, Carboxypeptidase (CP) M and H in small intestine

Isoelectric focusing of carboxypeptidase N

Erdo « s, Carboxypeptidase N (arginine carboxypeptidase

On the preparation and some properties of closed membrane vesicles from hog duodenal and jejunal brush border

The surface membrane of the small intestinal epithelial cell. I. Localization of adenyl cyclase

Enzymatic distinction of rat intestinal cell brush border and endoplasmic reticular membranes

An electron-transport system associated with the outer membrane of liver mitochondria. A biochemical and morphological study

A microspectrophotometric method for the determination of cytochrome oxidase

Fluorimetric assays for cathepsin B and cathepsin H with methylcoumarylamide substrates

Cathepsins B, H, and L in human breast carcinoma

A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding

Puri¢cation of the human intestinal brush border membrane

The ¢lamentous brush border glycocalyx, a mucin-like marker of enterocyte hyper-polarization

Cleavage of structural proteins during the assembly of the head of bacteriophage T4

Analytical disc gel electrophoresis

Glycoprotein staining following electrophoresis on acrylamide gels

Sequence from picomole quantities of proteins electroblotted onto polyvinylidene di£uoride membranes

Rapid analysis of amino acids using pre-column derivatization

A gas-liquid solid phase peptide and protein sequanator

Western blotting': electrophoretic transfer of proteins from dodecyl sulfate^polyacrylamide gels to un-modi¢ed nitrocellulose and radiographic detection with antibody and radioiodinated protein A

Characterization of an integral membrane glycoprotein associated with the micro-¢laments of pig intestinal microvilli

Aminopeptidase N-and human blood group A-antigenicity along the digestive tract and associated glands in the rabbit

Erdo « s, Carboxypeptidase M in Madin-Darby canine kidney cells. Evidence that carboxypeptidase M has a phosphatidylinositol glycan anchor

Ectoenzymes of the kidney microvillar membrane. Di¡erential solubilization by detergents can predict a glycosyl-phosphatidylinositol membrane anchor

Ectoenzymes of the kidney microvillar membrane. Aminopeptidase P is anchored by a glycosyl-phosphatidylinositol moiety

Sequence of human carboxypeptidase D reveals it to be a member of the regulatory carboxypeptidase family with three tandem active site domains

Multiple proteolytic action of rat liver cathepsin B: speci¢c-ities and pH-dependencies of the endo-and exopeptidase activities

Westro « m, Cathepsin B and D activities in intestinal mucosa during postnatal development in pigs. Relation to intestinal uptake and transmission of macromolecules

Immunocytochemical localization of cathepsins B, H, and L in the rat gastro-duodenal mucosa

Identi¢cation of a membrane-bound carboxypeptidase as the mammalian homolog of duck gp180, a hepatitis B virusbinding protein

Ganem, gp180, a host cell glycoprotein that binds duck hepatitis B virus particles, is encoded by a member of the carboxypeptidase gene family

Carboxypeptidase E is a regulated secretory pathway sorting receptor: genetic obliteration leads to endocrine disorders in Cpe fat mice

Intracellular misrouting and abnormal secretion of adrenocorticotropin and growth hormone in Cpe fat mice associated with a carboxypeptidase E mutation

Bioregulation of kinins: kallikreins, kininogens, and kininases

Intracellular tra¤cking of metallocarboxypeptidase D in AtT-20 cells: localization to the trans-Golgi network and recycling from the cell surface

Aminopeptidase N is a major receptor for the enteropathogenic coronavirus TGEV

Human aminopeptidase N is a receptor for human coronavirus 229E

We are grateful to Mr. Claude Villard for performing the amino acid compositions, Mr. Jacques Bonicel for performing the sequence determination, Prof. Vito Turk for his help with the cathepsin B assay and Dr. Jessica Blanc for revising the English manuscript.