Characterization of membrane-bound prolyl endopeptidase from brain Jofre Tenorio-Laranga 1 , Jarkko I. Vena ¨ la ¨ inen 2 , Pekka T. Ma ¨ nnisto ¨ 3 and J. A. Garcı ´a-Horsman 1,3 1 Centro de Investigacio ´ n Prı ´ ncipe Felipe, Valencia, Spain 2 Department of Pharmacology and Toxicology, University of Kuopio, Finland 3 Division of Pharmacology and Toxicology, University of Helsinki, Finland Prolyl oligopeptidase (POP; EC 3.4.21.26) is a serine peptidase with prolyl endopeptidase (PE) activity, cleaving short peptides at the C-terminal side of pro- line residues, and is highly expressed in brain. Given that several neuropeptides, such as substance P, argi- nine–vasopressin, thyroliberin and gonadoliberin, are putative POP substrates, the importance of this prote- ase in several brain processes has been suggested [1]. However, the precise role of POP in the brain has yet to be defined. Specific inhibitors of POP increase the levels of these neuropeptides in the brain, exert anti- amnesia effects, and reverse memory and learning defi- cits produced by certain lesions [2]. Mammalian POP, encoded by the gene Prep, has been purified and crys- tallized, and its structure has been solved; it has been considered to be soluble cytoplasmic enzyme. There has not been any structural or sequence-derived infor- mation that would suggest that the Prep gene product Keywords neuropeptides; neurotransmission; peptide metabolism; prolyl endopeptidase; prolyl oligopeptidase Correspondence J. A. Garcı ´ a-Horsman, Division of Pharmacology and Toxicology, University of Helsinki, Viikinkaari 5E, 00014 Helsinki, Finland Fax: +358 9 191 59471 Tel: +358 9 191 59459 E-mail: arturo.garcia@helsinki.fi (Received 7 March 2008, revised 3 July 2008, accepted 4 July 2008) doi:10.1111/j.1742-4658.2008.06587.x Prolyl oligopeptidase (POP) is a serine protease that cleaves small peptides at the carboxyl side of an internal proline residue. Substance P, arginine– vasopressin, thyroliberin and gonadoliberin are proposed physiological substrates of this protease. POP has been implicated in a variety of brain processes, including learning, memory, and mood regulation, as well as in pathologies such as neurodegeneration, hypertension, and psychiatric disor- ders. Although POP has been considered to be a soluble cytoplasmic pepti- dase, significant levels of activity have been detected in membranes and in extracellular fluids such as serum, cerebrospinal fluid, seminal fluid, and urine, suggesting the existence of noncytoplasmic forms. Furthermore, a closely associated membrane prolyl endopeptidase (PE) activity has been previously detected in synaptosomes and shown to be different from the cytoplasmic POP activity. Here we isolated, purified and characterized this membrane-bound PE, herein referred to as mPOP. Although, when attached to membranes, mPOP presents certain features that distinguish it from the classical POP, our results indicate that this protein has the same amino acid sequence as POP except for the possible addition of a hydro- phobic membrane anchor. The kinetic properties of detergent-soluble mPOP are fully comparable to those of POP; however, when attached to the membranes in its natural conformation, mPOP is significantly less active and, moreover, it migrates anomalously in SDS ⁄ PAGE. Our results are the first to show that membrane-bound and cytoplasmic POP are encoded by variants of the same gene. Abbreviations AMC, amido-4-methylcoumarin; cPOP, cytoplasmic prolyl oligopeptidase; ER, endoplasmic reticulum; HA, hydroxylapatite; mPOP, membrane-bound prolyl oligopeptidase; PE, prolyl endopeptidase; POP, prolyl oligopeptidase; PPP, pure pig recombinant prolyl oligopeptidase; Z-Gly-Pro-AMC, N-carbobenzoxy-glycyl-prolyl-7-amido-4-methyl-coumarin; ZPP, N-carbobenzoxy-prolyl-prolinal. FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS 4415 is present in any locations inside or outside the cells other than the cytoplasm, and no variants have been predicted or reported. This has been considered para- doxical, due to the extracellular location of the puta- tive POP substrates [3]. Nevertheless, PE activity has been detected in all biological fluids and in membranes from most of the tissues studied, especially the brain. Different extracellular proteins with PE activity have been described. In serum, a PE activity, insensitive to specific POP inhibitors, has been identified as fibro- blast activation protein or seprase, but important levels of PE activity itself, sensitive to POP-specific inhibi- tors, have also been detected [4,5], although confirma- tion of this enzyme’s identity, by direct sequencing or by antibody binding, has not been provided. Membrane-bound PE activity has been detected and measured, and has been considered by several authors to be POP activity [6–8]. However, isolated prepara- tions have been analyzed and regarded as a different peptidase, as these preparations have shown some physical and enzymatic features that are different from those of its classical cytoplasmic counterpart [9,10]. Recently, we have detected binding of antibody against POP in internal membranes in immunohistochemistry studies in rat brain [11]. However, no clear identifica- tion of the protein responsible for this activity has been provided. Here, we report the purification and identifi- cation of the membrane-bound PE (herein referred to as membrane-bound prolyl oligopeptidase, mPOP) from pig brain and characterization of the enzyme’s properties in comparison to POP. The nature of mPOP association with membranes was also studied. Results Membrane-associated PE activity is tightly bound preferentially to synaptosomes and endoplasmic reticulum (ER) Initial whole membrane fractionation of pig brain homogenate resulted in the partitioning of total PE activity between membrane-bound and soluble frac- tions with a 40 : 60 ratio. As reported previously [10], high-salt washes and a hypotonic treatment were required to detach loosely bound PE activity from the membranes. Accordingly, we found that a considerable amount of PE activity bound to the membranes was released upon a 0.5 m NaCl wash of total membrane preparation (Table 1). A hypotonic wash and two fur- ther salt washes were necessary to ensure that all the loosely bound POP was released. Further washes released no detectable activity from the membranes, but detectable levels were tightly attached to them (Table 1), and those were sensitive to specific POP inhibitors (see below). Following this series of washes, the membranes were further fractionated by centrifugation on a sucrose gra- dient to determine which types of membrane contained PE activity. After a three-cushion gradient (0.8, 1.0 and 1.2 m sucrose), we were able to separate three dif- ferent membrane fractions, low density (on top of the 0.8 m layer), medium density (0.8 and 1.0 m interface) and high density (1.0 and 1.2 m interface). With the use of specific enzyme marker assays (Fig. 1), we iden- tified the heavy membranes as ER, whereas the mem- branes of intermediate density were mainly composed of synaptosomal and mitochondrial membranes. The light fraction contained myelin membranes, as described previously [10]. Although we detected the presence of PE activity in all membrane fractions, this activity was maximal in the synaptosomal fraction (Table 2), similar to the observations reported by O’Leary & O’Connor [9]. As the activity detected in the various membrane fractions could be attributable to other peptidases, we applied the purification proto- col (detailed in Experimental procedures) to both ER and synaptosomal membranes. All elution profiles resulting from this purification scheme were identical, regardless of the membrane fraction origin. Moreover, analysis of these preparations revealed that the kinetic properties of both fractions were also identical (data not shown). Thus, we decided to employ the purifica- tion protocol using whole membrane preparation as starting material, as the yield was considerably higher. A multistep protocol enriches membrane PE activity > 2000-fold PE activity solubilization from membranes was only achieved by extraction with detergents, such as Triton X-100 at 0.4%. This detergent treatment was sufficient to solubilize all activity associated with Table 1. POP activity partitioning during membrane preparation from pig brain crude extract and membrane wash effects on recov- ered activity. Sample Volume (mL) Protein (mg) Specific activity (nmolÆmin )1 Æmg )1 ) Total activity (nmolÆmin )1 ) Crude extract 1080 8704 1.0 8645 Unwashed membranes 500 2354 1.4 3384 0.5 M NaCl wash 450 305 9.0 2760 Water wash 400 107 13.8 1480 4 M NaCl wash 300 92 4.8 440 Washed membranes 90 1005 0.2 161 Membrane-bound prolyl oligopeptidase: mPOP J. Tenorio-Laranga et al. 4416 FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS membranes, and also produced a three-fold activation of enzymatic activity (Table 3). This increase was not due to assay conditions, such as the presence of reduc- tants. Solubilized membrane PE activity was sub- sequently purified by several chromatography steps. The use of a DEAE column as a first step eliminated all cationic and most weak anionic protein contami- nants, which altogether constituted more than half of the total protein (Table 3). Substantial further purifica- tion was achieved with phenyl–Sepharose and hydroxyl- apatite (HA) columns (see Fig. S1). Although this step yielded more than 300-fold PE activity purification (Table 3) with respect to total membrane, SDS ⁄ PAGE revealed the presence of several protein bands (data not shown). Thus, to further purify PE activity, we dialyzed the HA pool to decrease the salt concentra- tion and reapplied it to a DEAE column (see Fig. S1 for column chromatogram). This procedure enriched activity 1700-fold, but SDS ⁄ PAGE analysis still revealed several contaminating proteins. Consequently, native gel electrophoresis was utilized to improve puri- fication. Under these conditions, a single band was revealed by silver staining that coincided with the PE activity profile of the gel lane (Fig. 2). It is important to note that once PE activity was solubilized, neither stability nor activity was modified by detergent concen- tration. Extensive removal of Triton X-100, by series of dilutions and ultrafiltrations where the detergent was undetectable (< 0.0001%), did not produce pro- tein precipitation or loss of activity. The gel filtration profile of solubilized membrane PE activity varies with ionic strength, similarly to that of cytoplasmic POP (cPOP) It has been suggested previously that the enzyme responsible for membrane PE activity is distinct from POP, on the basis of the difference between their molecular masses (87 kDa versus 65 kDa) obtained in gel filtration experiments [9,10]. The theoretical molecular mass of POP is 80 kDa, which agrees with estimates from SDS ⁄ PAGE [1]. Initially, we also thought that membrane PE was heavier than POP, as under our conditions (20 mm potassium phos- phate) it eluted with a molecular mass of 95 kDa by gel filtration. However, when higher salt concentra- tions were used, membrane PE activity also eluted with a molecular mass of 65 kDa (Fig. 3). This behavior did not depend on the Triton X-100 con- centration, and was very similar to that of POP when run in the same conditions. Identification of the protein responsible for membrane PE activity Peptides produced by trypsin digestion of purified membrane PE were analyzed by liquid chromatogra- phy–MS ⁄ MS on Qstar (HPLC ⁄ Q ⁄ TOF) or MALDI- TOF ⁄ TOF MS, which revealed that these fragments correspond to the sequence of mammalian POP 0.075 0.050 0.025 0.000 TM HM Suc. dehydrog. (units·mg –1 ) TM MM HM LM PSD-95 Calnexin ER-60 90 kDa 50 kDa 90 kDa LM MM Fig. 1. Identification of the different membrane fractions where PE activity is bound. Various membrane fractions were obtained fol- lowing application of washed total pig brain membrane preparation to sucrose gradients. The mitochondrial marker, succinate dehydro- genase, was measured in every fraction (upper panel), as described in Experimental procedures. PSD-95 (synaptosomal marker), calnexin and ER-60 (ER markers) were assayed by western blotting (lower panels). TM, total membranes; LM, low-density membranes; MM, medium-density membranes; HM, high-density membranes. The amounts of protein loaded onto gels were as follows: for the PSD- 95 blot, 70 lg of TM, 96 lg of LM, 89 lg of MM, and 93 lgof HM; for the calnexin blot, 70 lg of TM, 96 lg of LM, 89 lg of MM, and 93 lg of HM; and for the ER-60 blot, 8 lgofMM and 5 lg of HM. Table 2. Membrane-bound POP activity of different-density mem- brane fractions obtained by sucrose gradient centrifugation. Sample Total volume (mL) Specific activity (nmolÆmin )1 Æmg )1 ) Total activity (nmolÆmin )1 ) Washed membranes 60 0.50 145 Low-density membranes 2.3 0.61 9 Medium-density membranes 4.5 0.45 24 High-density membranes 10 0.17 21 J. Tenorio-Laranga et al. Membrane-bound prolyl oligopeptidase: mPOP FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS 4417 (EC 3.4.21.26) (Fig. 4). To further confirm this, wes- tern blots were performed using an antibody specific against POP. All active fractions obtained during the membrane PE purification process reacted with the antibody against POP, thereby confirming the sequence similarity between these two variants (Fig. 5A). Analy- sis of the membrane PE tryptic peptides by MS did not provide any insights into the membrane associa- tion mechanism, such as sugar or lipid attachments (not shown). Within the membranes, mPOP shows anomalous SDS ⁄ PAGE migration Prior to Triton X-100 solubilization, different brain membrane preparations were analyzed by western blot- ting, using POP as a control. A significant fraction of mPOP migrated faster than POP when a whole mem- brane preparation was subjected to SDS ⁄ PAGE, and anti-POP reactive bands were detected by western blot- ting (Fig. 5B). The proportion of this ‘lighter’ form, relative to that which corresponds to the cytoplasmic purified POP control, varied with the type of mem- brane fraction analyzed. As can be seen in Fig. 5B, the myelin fraction showed a band in the western blot at the same size as the pure pig soluble recombinant POP (PPP), but the synaptosomal fraction showed two anti- POP reactive bands in a ratio of approximately 50%. The heavier ER membrane fraction contained almost Fig. 2. Membrane PE purification by native electrophoresis from the concentrated HA active fractions. Three adjacent lanes of a native Triton–PAGE gel were loaded with 10 lg each of protein. After electrophoresis (see Experimental procedures), the central lane was excised in 5 mm pieces along the lane vertical axes. Hori- zontal blade cuts were around 8 mm long such that small incisions at the edge of adjacent lanes were produced to find the corre- sponding pieces in the western blot, made with the first lane, and the protein stain (silver), made with the third lane. Fig. 3. Membrane PE gel filtration on Superdex-200 in 100 mM phosphate buffer ( ) and in 20 mM potassium phosphate buffer ( ), performed as described in Experimental procedures. The positions of elution of molecular mass standards are indicated: ribonuclease (Rib), 13.7 kDa; chymotrypsinogen (Chym), 25 kDa; ovoalbumin (Ovo), 43 kDa; BSA, 37 kDa; aldolase (Ald), 158 kDa; catalase (Cat), 232 kDa; ferritin (Fer), 440 kDa; and thyroglobulin (Thyr), 669 kDa. Table 3. Purification of mPOP from a total membrane preparation. POP was assayed as described in Experimental procedures. Total activity is specific activity multiplied by total protein in milligrams; after Triton X-100 extraction, there is a > 3-fold activation of activity, which is reflected by an increase in total activity. Yield percentage refers to the activity in detergent extract, which is shown in parentheses. Sample Total volume (mL) Total protein (mg) Specific activity (nmolÆmin )1 Æmg )1 ) Total activity (nmolÆmin )1 ) Yield (%) Fold purification Total membranes 150 855 0.5 453 100 1 Triton X-100 extraction 1000 533 3.3 1759 388 (100) 6.6 DEAE 680 232 7.4 1717 379 (98) 14.8 Phenyl–Sepharose 160 43 39 1679 371 (95) 78 HA 25 9.5 169 1595 352 (90) 338 Second DEAE 6.5 0.9 857 771 170 (43) 1714 Membrane-bound prolyl oligopeptidase: mPOP J. Tenorio-Laranga et al. 4418 FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS exclusively the light band. This anomalous behavior was reproducible, and was consistently eliminated by membrane solubilization with Triton X-100 (Fig. 5C). In these cases, the anti-POP reactive band always ran at the same molecular mass as the band that corre- sponded to cPOP, or to the pure POP control, regard- less of the membrane fraction that the sample was prepared from. Kinetic properties of purified soluble membrane PE We next attempted to differentiate membrane PE from POP on the basis of kinetic properties. This compara- tive analysis included semipure pig brain POP, PPP, and membrane PE purified by HA chromatography. Accordingly, different inhibitors and substrates were tested, and the kinetic behavior of membrane PE, in comparison to that of POP, was evaluated. It is known that POP is inhibited by some divalent cations [12,13], including the heavy metals Mn 2+ , Cu 2+ ,Ni 2+ , and Zn 2+ [14]. Our results demonstrate that there were no significant differences between membrane PE and POP regarding their sensitivity to these metals (Table 4). In addition, we compared the effects of some general serine, cysteine and metallo- protease inhibitors, specific POP inhibitors, such as N-carbobenzoxy-prolyl-prolinal (ZPP) and JTP-4819, the proteasome inhibitor N-carbobenzoxy-leucyl-leucyl- leucyl-COH, the specific dipeptidyl peptidase IV inhibitor HIV-1 tat (1–9) fragment with sequence H-Met-Asp-Pro-Val-Asp-Pro-Asn-Ile-Glu-OH, and the POP inhibitor a 2 -gliadin 33-mer peptide [15] (see A B C Fig. 5. Western blots of different membrane PE preparations obtained using an antibody against POP. (A) Cross-reactivity against membrane PE and recombinant POP of antibody against POP. Membrane PE, 8 lg of protein; PPP, 120 ng of protein. (B, C) Blot on total membranes (TM), light-density membranes (LM), medium- density membranes (MM) and high-density membranes (HM) from pig brain before (B) or after (C) Triton X-100 solubilization and com- pared with PPP. Protein amounts were as follows: (B) PPP 0.15 lg, TM 30 lg, LM and HM 60 lg, and HM 120 lg. (C) PPP 0.15 lg, TM 12 lg, LM 23 lg, MM 23 lg, and HM 30 lg. Fig. 4. POP amino acid sequence from Sus scrofa (accession num- ber NP_001004050 Ver NP_001004050.1 GI:51592147). Peptides from tryptic digestion of purified membrane PE are highlighted (see Doc. S1 in Supporting information also). Table 4. Effects of divalent cations, ionic strength and chelators on POP activity of membrane PE (A) and on cPOP (B). 0 l M 1 lM 10 lM 100 lM 1mM 10 mM (A) Membrane PE Mg 2+ 100 100 101 104 101 98 Ca 2+ 100 97 99 103 101 97 Mn 2+ 100 98 96 86 80 28 Cu 2+ 100 102 98 90 11 8 Ni 2+ 100 96 94 78 8 8 Zn 2+ 100 93 94 13 8 8 0m M 100 mM 200 mM 400 mM 600 mM 800 mM NaCl 100 124 127 153 138 133 0m M 2.5 mM 5mM 10 mM 25 mM 125 mM EDTA 100 89 78 62 42 26 (B) cPOP Mg 2+ 100 101 99 96 99 98 Ca 2+ 100 96 98 104 93 95 Zn 2+ 100 96 86 4 3 3 0m M 2.5 mM 5mM 10 mM 25 mM 125 mM EDTA100 86 85 64 46 24 J. Tenorio-Laranga et al. Membrane-bound prolyl oligopeptidase: mPOP FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS 4419 Table S1). As expected, both membrane PE and POP were very sensitive to the specific POP inhibitors. Fur- thermore, the kinetic constants produced by careful titration of ZPP, a very specific and potent POP inhib- itor, were equivalent for both preparations (Fig. 6). Both membrane PE and POP were partially resistant to phenylmethanesulfonyl fluoride, a generic serine protease inhibitor that has a low efficiency in inhibit- ing POP [1]. We also found that membrane PE was inhibited by SH-reactive compounds such as malei- mides, similarly to POP, and that both were similarly resistant to dipeptidyl peptidase IV and proteasome inhibitors. In addition, membrane PE showed the same sensitivity to the 33-mer peptide (see Table S1) as has been reported for POP [15]. Previous reports suggest that membrane PE and POP display similar specificity for several proline- containing neuropeptides, including bradykinin, angio- tensin II, neurotensin, substance P, and gonadoliberin [10]. In addition to these substrates, we assayed several other peptides in an effort to define functional differ- ences between membrane PE and POP. However, as shown in Table 5, no differences were observed for any of the substrates tested. Membrane PE probably associates with the membrane by nonprotein hydrophobic anchoring The nature of the association of membrane PE with membranes is not known. We tried to address this question in several ways. Analysis of the hydrophobic- ity profile of the POP protein sequence revealed that the presence of membrane-spanning segments is highly improbable (see Doc. S2 in Supporting information). Our experiments with membrane PE extracted with Triton X-114 demonstrated that this enzyme activity is quantitatively partitioned in the hydrophilic phase, arguing against any important hydrophobic protein domain that would link the protein to the intermem- brane milieu. To further confirm this, we ran a temperature dependence assay with membrane PE-containing mem- branes (Fig. 7). Intrinsic membrane proteins display a break in the Arrhenius plot due to membrane phase transition, and delipidation of these particulate enzymes eliminates the discontinuity in the plots [16]. We did not observe any break in the membrane PE Arrhenius plot; however, there was a significant change in the slope of the temperature dependency curve when compared with that obtained for POP (Fig. 7). Fur- thermore, the membrane PE Arrhenius plot resembled that of POP when the experiment was performed after detergent solubilization of membrane PE. POP has been implicated in axonal transport [17]. Soluble protein elements in these processes are recruited to membranes, through interaction with other proteins and SH-bonding and ⁄ or divalent cation (Ca 2+ )-depen- dent mechanisms. To test whether a similar process would mediate the membrane PE association with membranes, we evaluated the expression levels of PE in membranes after homogenization or washes with N-ethylmaleimide or EDTA. The presence of N-ethyl- Fig. 6. Inhibitory effect of ZPP (Z-Pro-Prolinal) on POP (s) or mem- brane PE ( ) activity. The assay was performed as described in Experimental procedures in the presence of the corresponding inhibitor concentrations during preincubation. The estimated IC50 values for membrane PE (continuous line) and POP (broken line) were 0.48 n M (± 0.005 nM) and 0.52 nM (± 0.005 nM) respectively. Table 5. Substrate specificity studies on membrane PE as compared with POP. +, ZPP-sensitive cleavage of the peptide occurred; ), cleavage of the peptide did not occur. Peptide Sequence Membrane PE POP 12-mer H 2 N-QLQPFPQPQLPY-OH ++ 33-mer LQLQPFPQPQLPYPQPQLPYPQPQLPYPQPQPF )) HIV H-MDPVDPNIE-OH ++ PEP-3 YGRKKRRQRRRG-NH 2 )) PEP-26 RGTICKKTMLDGLNNYCTGVGR-NH 2 )) PEP-48_2 Ac-LINEEEFFDAVEAALDRQ-NH 2 )) PEP-50 Ac-PYSRSSSMSSIDLVSASDDVHRFSSQ-NH 2 )) PEP-52 Ac-CDPGYIGSR-NH 2 ++ Membrane-bound prolyl oligopeptidase: mPOP J. Tenorio-Laranga et al. 4420 FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS maleimide or EDTA, or both, during homogenization or membrane washes did not alter the amount of PE detected in the membrane fraction (data not shown). An alternative explanation is that membrane PE is attached to membranes through a hydrophobic anchor that has been added to the protein post-translationally. A web-based (expasy) analysis of the POP sequence for post-translational modifications related to glycosyl- ation, myristoylation, prenylation and glycosylphos- phatidyl inositol anchoring returned very low scores (see Doc. S2 in Supporting information). Furthermore, POP lacks a signal sequence required for some of these modifications. The only possibility that was found is palmitoylation of the Cys563 within the sequence GGLLVATCANQRPDL(556–570), which, according to the css-palm server (http://bioinformatics.lcd-ust- c.org/css_palm/), has a relatively good score for modifi- cation (see Doc. S2 in Supporting information and Fig. 8). Using [ 3 H]palmitate, or [ 3 H]palmitoyl-CoA, we have tried to measure in vitro or metabolic palmitoylation, but our attempts have been unsuccessful, in part because of the relatively low expression levels of endogenous membrane PE. These results, however, do not rule out this possibility. Discussion This article reports the identification of the protein responsible for PE activity in membranes isolated from pig brain, which we now call mPOP. For more than 20 years, it has been known that some PE activity it associated with membranes from almost all tissues and especially from brain [18–20]. Further- more, mPOP activity has been found to change with age [6,21]. Purification of mPOP from bovine brain has been attempted previously [9,10], and on the basis of those studies, it was concluded that this peptidase is expressed mainly in the synaptosomal fraction and has a heavier mass (87 kDa) than POP (65 kDa). Furthermore, on the basis of sensitivity to thiol-reactive inhibitors, mPOP was thought to be a thiol-dependent metallopeptidase [10], but the basic problem was that this enzyme has never been readily identified before. In an attempt to clarify the identity of mPOP, we undertook the task of purifying and characterizing particulate POP from pig brain. We have confirmed that a significant amount of PE activity can be mea- sured in the particulate fraction from crude pig brain homogenates. In our preparations, this activity accounted for around 40% of the total homogenate activity, similar to the 50% reported previously for the corresponding fraction from bovine brain [9]. However, after osmotic shock and high-salt treat- ment, our total membrane preparation contained less than 5% of total activity, as compared with 20% recovery reported earlier for washed synaptosomal membranes from bovine brain [9,10]. This discrep- ancy may be species-related, as the preparation con- ditions were essentially the same. The PE activity present in our washed membrane preparation was very tightly bound, and was solubilized only after detergent treatment. Upon sucrose gradient fraction- ation, some PE activity was present in the heavier membrane fraction containing ER, but the majority was detected within the synaptosomal membrane fraction, consistent with a role for mPOP in synapse function. This is consistent with the finding of bind- ing of antibody against POP to internal neuronal membranes in rat brain slices [11]. Kinetic experiments demonstrated that the substrate preference and the inhibitor sensitivity of purified mPOP are identical to those of POP (see Table 4 and Table S1). Furthermore, POP-specific antibodies cross-reacted with mPOP on western blots. Analysis of peptide fragments generated by trypsin digestion identified mPOP protein as POP. Thus, we found only two features that distinguish the two forms, and those are attributed to the membrane milieu where mPOP resides. One was the tight membrane associa- tion, which could only be disrupted by detergent solu- bilization. The second feature was the different Fig. 7. Arrhenius plots of membrane PE ( ), POP ( ) and solubi- lized membrane PE (d). Activity assays were performed in dupli- cate as described in Experimental procedures, but reaction tubes were preincubated at every temperature (every 2 °C between 10 and 40 °C) and the reaction was incubated at the corresponding temperature for 40 min. Solid lines represent the linear regres- sions. J. Tenorio-Laranga et al. Membrane-bound prolyl oligopeptidase: mPOP FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS 4421 temperature dependence between mPOP activity, when is bound to membranes, and POP activity. For transmembrane enzymes, Arrhenius plots generally show a break that corresponds to the membrane melting point. At temperatures above this point, the more fluid environment produces a decrease of activa- tion energy, producing a convex Arrhenius plot [22]. However, we found that the Arrhenius plot of mPOP did not show any break but did reflect an increase of activation energy in relation to POP or Triton-solubi- lized mPOP. Interestingly, these two latter forms were very similar to each other (Fig. 6). We interpret this as an inhibitory effect of the membranous milieu, or of the putative anchor, or both, on mPOP activity, as the Arrhenius profile switched to a POP profile of lower activation energy upon detergent solubilization. It is remarkable that Triton X-100 activated mPOP activity at least by a factor of 3 (Table 3). It was also observed that the SDS ⁄ PAGE migration of mPOP, when still attached to the native membranes, was atypical. Consistently, different membrane fractions showed a reactive band for antibody against POP that migrated at a slightly, but perceptibly, lower molecular mass, present at different proportions for the different fractions (Fig. 5B). In total membranes, this lighter band was a major component, becoming almost exclusive in the heavy membrane fraction. Medium-density membranes presented two bands: the light band and the one that matched the soluble POP control. On the other hand, in the light membrane fraction, the latter band was the only one appearing. It was remarkable that this lower molecular mass band completely disappeared from all membrane frac- tions when the very same samples were solubilized with Triton X-100. After this treatment, only one band around 80 kDa, as with the cPOP control, was detected in all cases. It can be argued that the mem- brane-associated state induces a more compact and tighter conformation in which the anchor site is less accessible to reduction, thereby resulting in faster migration. After solubilization and membrane disrup- tion, the protein anchor might be more accessible to reducing agents, and thus could be readily dissociated to yield the soluble form with normal migration in SDS ⁄ PAGE (Fig. 5B,C). The fact that both normally migrating and abnormally migrating bands, from the different membrane fractions, appear in different proportions could be explained by bringing into play Fig. 8. Localization of Cys563 on the 3D model of pig POP. Peptidase_S9 domain residues are shown in magenta; propeller domain chains are represented in navy, brown, green, gray and orange; Cys563 is shown in yellow. Modeled from file mmd- bId:21074, for POP from porcine brain, from the NCBI’s Entrez Structure database, and handled by CN3D v. 4.1 (NCBI) software. Membrane-bound prolyl oligopeptidase: mPOP J. Tenorio-Laranga et al. 4422 FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS the different physical and chemical properties of the corresponding membranes, which would obviously affect the interactions with mPOP. Several studies indicate that POP interacts with cyto- skeleton proteins and that it is probably involved in axonal transport [17]. Components of neuronal membrane trafficking are tightly associated with membrane-bound organelles in a chaperone-mediated or chaperone-sensitive way, and are therefore resistant to in vitro treatments, including high salt, which release most peripheral membrane proteins [23]. Homogeni- zation with buffers containing millimolar levels of SH-modifying agents, such as N-ethylmaleimide, or chelators of divalent cations, such as EDTA, is known to significantly release these proteins from membranes. Proteins such as annexins associate with membranes in aCa 2+ -dependent way, and EDTA disrupts this asso- ciation and solubilizes the proteins [24]. We did not note any effects of N-ethylmaleimide or EDTA during homogenization or membrane washing on the mem- brane PE levels (not shown). These data indicate that the mechanism of mPOP membrane association is most probably not mediated by SH interaction or by divalent cations. On the whole, our results may suggest that POP undergoes a post-translational modification in which a membrane anchor is added to the protein, attaching it to membranes. The nature of this putative anchor remains to be determined, but it is most likely a hydro- phobic chain, as POP does not contain any substantial hydrophobic domain that could be used to embed it in cellular membranes (see Doc. S2 in Supporting infor- mation). Our analysis of the POP amino acid sequence suggests palmitoylation as a possible post-translational modification; the well-conserved sequence of pig POP, GGLLVATCANQRPDL(556–570), yields a very good score for this kind of modification according to css- palm software [25]. Examination of the 3D structure of pig POP revealed that a critical cysteine is situated very near the surface of the enzyme, on the bottom side of the catalytic domain, making it a good candi- date for the putative palmitoylation (Fig. 8). In fact, this cysteine residue is totally conserved among all eukaryotic POP genes sequenced [26]. Although S-palmitoylation is theoretically sensitive to SH group- reducing agents, this site could be buried when mPOP is anchored to the membrane. Although our attempts to measure in vitro or metabolic palmitoylation have not succeeded, the possibility of this post-translational modification is still open. It is also important to note that we did not use any inhibitors for lytic enzymes during mPOP preparation, and it is possible that the putative membrane anchor was removed during the process of isolation and purifi- cation, or during trypsinization and peptide extraction, preventing its identification by MS. Additionally, membrane-bound proteins, which interact directly with the hydrophobic inner membrane phase, require the presence of an amphiphilic compound, such as a deter- gent or a phospholipid, for stability and activity. We found that once mPOP is solubilized from membranes, the concentration of Triton X-100 can be considerably decreased without any protein precipitation or loss of activity. This strongly points to the intrinsically soluble nature of mPOP, as strongly hydrophobic proteins tend to aggregate and lose activity during lipid or detergent removal. The cytoplasmic location of POP is paradoxical when considering its major role in the metabolism of extracellular neuropeptides [3]. Recently, other roles have been suggested for POP, such as axonal transport and ⁄ or modulation of intracrine peptide regulation [1]. The existence of an alternative particulate form with PE activity, a different enzyme but with the same func- tional properties as POP, that is responsible for the degradation of neuropeptides in the synaptosomal cleft, would solve the localization problem. However, attempts to identify this enzyme have been unsuccess- ful [9,10]. Contrary to expectations, we found several pieces of evidence suggesting that the particulate form is, in fact, a variant of POP (EC 3.4.21.26); mPOP has the same gel filtration, immunological, activity and inhibitory properties as soluble POP, and even MS data provided high confidence in the identification. Our results additionally suggest that a post-transla- tional modification is necessary for POP to be associ- ated with membranes. Furthermore, the data reported here also suggest that this modification is sensitive to the reducing state of the environment, and this may indicate the existence of specific cell machinery that controls the association–dissociation event. One funda- mental question is the orientation of mPOP in the synaptosomes. Fast and effective neuropeptide degra- dation in the synaptosomal cleft would only be possi- ble if mPOP was facing that side. Another interesting aspect invoked is the possibility that the membrane association of POP is connected to its transport out of the cell, the membrane-bound form being only a tran- sitory stage. Experimental procedures Total membrane preparation Eighty grams of freshly isolated pig brain were homoge- nized in 320 mL of ice-cold 0.32 m sucrose in 100 mm J. Tenorio-Laranga et al. Membrane-bound prolyl oligopeptidase: mPOP FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS 4423 potassium phosphate buffer (pH 7.4), sonicated, and centri- fuged at 1000 g for 10 min. The resulting supernatant was sonicated again, and centrifuged at 155 100 g for 30 min. The pellet was washed first with 0.5 m NaCl, once with dis- tilled water, twice with 4 m NaCl, and finally with distilled water. The resulting membranes were resuspended in 100 mm potassium phosphate buffer (pH 7.4). Membrane fractionation An aliquot of the total membrane preparation was brought to 0.32 m sucrose and 100 mm potassium phosphate buffer (pH 7.4), and layered on a discontinuous sucrose gradient (1.2, 1 and 0.8 m sucrose). The tubes were centrifuged at 80 000 g for 90 min, in a swing rotor. After centrifugation, the membrane layers were carefully collected by pipetting. Each fraction was diluted to 0.32 m sucrose, centrifuged at 155 100 g, and resuspended in 100 mm potassium phosphate buffer (pH 7.4). Enzymatic assays PE activity was assayed by measuring the fluorescence released from the substrate N-carbobenzoxy-glycyl-prolyl-7- amido-4-methyl-coumarin (Z-Gly-Pro-AMC) (200 lm), as previously reported [27], by incubating protein samples in 100 mm sodium phosphate buffer (pH 7.0). The assay was stopped by the addition of 1 m sodium acetate buffer (pH 4.2). A succinate dehydrogenase assay was performed in 0.3 mL containing 0.01 m sodium succinate, 0.05 m phos- phate (pH 7.5), and 0.4 mg (in 50 lL) of each membrane fraction. The mixtures were incubated at 37 °C for 15 min, 0.1 mL of 2.5 mgÆmL )1 p-iodotetrazolium violet was added, and incubation was continued for a further 10 min. The reaction was stopped with 1 mL of ethyl acetate ⁄ ethanol ⁄ trichloroacetic acid (5 : 5: 1), and centrifuged at 16 000 g for 1 min; absorbance was measured at 490 nm. Protein determination Protein was determined by the Bradford method (Bio-Rad, Hercules, CA, USA) using BSA (Sigma-Aldrich, St Louis, MO, USA) as standard, and in the presence of 0.1% Triton X-100 when required. SDS ⁄ PAGE and western blotting Samples were diluted 1 : 1 with loading buffer (100 mm Tris ⁄ HCl, pH 6.8, 70% glycerol, 2% SDS, 0.005% bromo- phenol blue, 10 mm b-mercaptoethanol) and separated on 8% or 10% polyacrylamide ⁄ bis-acrylamide Tris ⁄ HCl dis- continuous gels. Gels were stained for protein or trans- ferred to nitrocellulose for blotting. For protein staining, gels were fixed with methanol ⁄ acetic acid ⁄ water (4 : 5 : 4) for 30 min, washed with water for 30 min (two changes), sensitized with 0.02% sodium thiosulfate for 1–2 min, and incubated with 0.1% silver nitrate for 20 min at room tem- perature. Gels were then rinsed twice with distilled water, and bands were visualized by incubating for 1–2 min with 2% sodium carbonate and 0.04% formaldehyde. Western blotting was performed under standard conditions using primary antibody [28] diluted 1 : 5000 (with 0.5 m NaCl, 20 mm Tris-HCL pH 7.5 and 0.05% Tween 20), and the anti-(chicken horseradish peroxidase) complex diluted 1 : 50 000 (Pierce, Rockford, IL, USA). Protein visualiza- tion was performed using an ECL kit (Amersham-Biosci- ence, Little Chalfont, UK), following the manufacturer’s instructions. Purification of mPOP The total membrane preparation (see above) was extracted with 0.4% Triton X-100 (at 1 mg of Triton X-100 per mg of protein) in 20 mm buffer and 100 mm NaCl on ice for 1 h. After the extraction, the samples were centrifuged at 155 100 g for 30 min and the pellet was discarded. Buffer was exchanged by dilution and ultrafiltration (CentriPrep 50; Amicon, Millipore Corp. Billerica, MA, USA), with DEAE equilibration buffer [50 mm Tris, pH 7.4, 1 mm EDTA, 5 mm dithiothreitol, 0.05% Triton X-100], bound to an equilibrated DEAE– Sepharose Fast Flow column (1.6 · 10 cm; Amersham, Uppsala, Sweden), washed with DEAE equilibration buf- fer, and step-eluted with 500 mm NaCl in the same buf- fer. The eluted pool was concentrated and the buffer was exchanged (as stated above) for phenyl–Sepharose equili- bration buffer [900 mm (NH 4 ) 2 SO 4 ,50mm Tris ⁄ HCl, pH 7.4, 5 mm dithiothreitol, 1 mm EDTA], and loaded onto a phenyl–Sepharose High Performance column (0.7 · 2.5 cm; Amersham). Activity eluted within the flow-through. Most of the contaminating protein was retained in the column and eluted by a wash without (NH 4 ) 2 SO 4 . Peak fractions were pooled, the buffer was exchanged for HA equilibration buffer (10 mm potassium phosphate, pH 7.4, 5 mm dithiothreitol, 1 mm EDTA, 0.05% Triton X-100), and the sample was loaded onto a 0.59 · 3.6 cm HA column (Bio-Rad). Activity was eluted at around 250 mm potassium phosphate over a 10– 500 mm linear gradient (see Fig. S1). The buffer of the HA pool was exchanged with DEAE buffer 2 (EDTA 1mm,5mm dithiothreitol, 50 mm Tris ⁄ HCl, pH 6.6), loaded onto an equilibrated DEAE HiTrap Fast Flow column (0.7 · 2.5 cm; Amersham), and eluted with a 0–500 mm NaCl gradient. Activity eluted at around 200 mm salt. All chromatographic steps were accom- plished using an A ¨ KTA prime system and monitored with primeview plus software (both from Amersham). All chromatography profiles are shown in Fig. S1. Membrane-bound prolyl oligopeptidase: mPOP J. Tenorio-Laranga et al. 4424 FEBS Journal 275 (2008) 4415–4427 ª 2008 The Authors Journal compilation ª 2008 FEBS [...]... Ontogeny of soluble and particulate prolyl endopeptidase activity in several areas of the rat brain and in the pituitary gland Dev Neurosci 25, 316–323 7 Irazusta J, Larrinaga G, Gonzalez-Maeso J, Gil J, Meana JJ & Casis L (2002) Distribution of prolyl endopeptidase activities in rat and human brain Neurochem Int 40, 337–345 8 Agirregoitia N, Casis L, Gil J, Ruiz F & Irazusta J (2007) Ontogeny of prolyl endopeptidase. .. 9 O’Leary RM & O’Connor B (1995) Identification and localisation of a synaptosomal membrane prolyl endopeptidase from bovine brain Eur J Biochem 227, 277– 283 10 O’Leary RM, Gallagher SP & O’Connor B (1996) Purification and characterization of a novel membrane- 4426 15 16 17 18 19 20 21 22 23 bound form of prolyl endopeptidase from bovine brain Int J Biochem Cell Biol 28, 441–449 Myohanen TT, Venalainen... distribution of rat brain prolyl oligopeptidase and its association with specific neuronal neurotransmitters J Comp Neurol 507, 1694–1708 Bausback HH & Ward PE (1986) Vascular, post proline cleaving enzyme: metabolism of vasoactive peptides Adv Exp Med Biol A 198, 397–404 Kato T, Nakano T, Kojima K, Nagatsu T & Sakakibara S (1980) Changes in prolyl endopeptidase during maturation of rat brain and hydrolysis of. .. Miettinen R & Mannisto PT (2007) ¨ ¨ Distribution of immunoreactive prolyl oligopeptidase in human and rat brain Neurochem Res 32, 1365– 1374 Venalainen JI, Juvonen RO, Forsberg MM, Garcı´ a¨ ¨ Horsman A, Poso A, Wallen EA, Gynther J & Man¨ nisto PT (2002) Substrate-dependent, non-hyperbolic ¨ kinetics of pig brain prolyl oligopeptidase and its tight Membrane-bound prolyl oligopeptidase: mPOP binding inhibition... P (1984) Inactivation of neurotensin by rat brain synaptic membranes Cleavage at the Pro10-Tyr11 bond by endopeptidase 24.11 (enkephalinase) and a peptidase different from proline -endopeptidase J Neurochem 43, 1295–1301 Dalmaz C, Netto CA, Volkmer N, Dias RD & Izquierdo I (1986) Distribution of proline endopeptidase activity in sub-synaptosomal fractions of rat hypothalamus Braz J Med Biol Res 19,... O’Connor B (2003) Characterisation of the active site of a newly-discovered and potentially significant post-proline cleaving endopeptidase called ZIP using LC-UV-MS Analyst 128, 670– 675 5 Cunningham DF & O’Connor B (1997) Identification and initial characterisation of a N-benzyloxycarbonylprolyl-prolinal (Z-Pro-prolinal)-insensitive 7-(N-benzyloxycarbonyl-glycyl -prolyl- amido)-4-methylcoumarin (Z-Gly-Pro-NH-Mec)-hydrolysing... cPOP Purification of PPP PPP was expressed in Escherichia coli and purified as described previously [29] Peptide digestion assay The assay mixture (140 lL) was composed of 50 mm Tris ⁄ HCl (pH 7.0) and cPOP, or mPOP, to 4 nmolÆmin)1 of activity Each peptide was added (prewarmed) at a final concentration of 140 lm The reaction was carried out at 30 °C for 60 min and stopped by the addition of trifluoroacetic... Juvonen RO & Mannisto PT (2004) ¨ ¨ ¨ ¨ Evolutionary relationships of the prolyl oligopeptidase family enzymes Eur J Biochem 271, 2705–2715 Venalainen JI, Garcı´ a-Horsman JA, Forsberg MM, ¨ ¨ Jalkanen A, Wallen EA, Jarho EM, Christiaans JA, Gynther J & Mannisto PT (2006) Binding kinetics and ¨ ¨ duration of in vivo action of novel prolyl oligopeptidase inhibitors Biochem Pharmacol 71, 683–692 Myohanen... Venalainen JI ¨ ¨ ¨ ¨ (2007) On the role of prolyl oligopeptidase in health and disease Neuropeptides 41, 1–24 2 Mannisto PT, Venalainen J, Jalkanen A & Garcı´ a¨ ¨ ¨ ¨ Horsman JA (2007) Prolyl oligopeptidase: a potential target for the treatment of cognitive disorders Drug News Perspect 20, 293–305 3 Brandt I, Scharpe S & Lambeir AM (2007) Suggested functions for prolyl oligopeptidase: a puzzling paradox... J Neurochem 35, 527–535 Kato T, Okada M & Nagatsu T (1980) Distribution of post-proline cleaving enzyme in human brain and the peripheral tissues Mol Cell Biochem 32, 117–121 Garcı´ a-Horsman JA, Venalainen JI, Lohi O, Auriola ¨ ¨ IS, Korponay-Szabo IR, Kaukinen K, Maki M & ¨ Mannisto PT (2007) Deficient activity of mammalian ¨ ¨ prolyl oligopeptidase on the immunoactive peptide digestion in coeliac . Characterization of membrane-bound prolyl endopeptidase from brain Jofre Tenorio-Laranga 1 , Jarkko I. Vena ¨ la ¨ inen 2 ,. identifi- cation of the membrane-bound PE (herein referred to as membrane-bound prolyl oligopeptidase, mPOP) from pig brain and characterization of the enzyme’s properties