On the role of prolyl oligopeptidase in health and disease

1. Introduction 

One of the main functions of proteases is their participation in posttranslational modification. Virtually all bioactive neuropeptides are synthesized by proteolytic cleavage of inactive precursors. Moreover, the peptidergic transmitters are proteolysed to terminate their action. In this respect, this process differs from the main mechanism involved in terminating the action of classical transmitters, which is based on their retro-transport from the presynaptic space back into the cytosol by specific transport systems.

Most peptidases, even those lacking specificity, are unable to hydrolyse peptide bonds formed by proline residues due to the unique structure of this imino-acid and the special constraints it exerts on the peptide backbone structure. In fact, proline residues appear near the amino terminus of many biologically active peptides, and in that way protects the peptide against degradation. There are however, specific peptidases, which recognize the proline ring. Many of these enzymes have important physiological roles and are currently targets of the pharmaceutical industry (Cunningham and O’Connor, 1997b, Rosenblum and Kozarich, 2003).

The prolyl oligopeptidase family of serine proteases, classified as S9 and belonging to the SC clan, includes prolyl oligopeptidase (POP, EC 3.4.21.26); dipeptidyl peptidase IV (DPPIV, EC 3.4.14.5), acylaminoacyl peptidase (ACPH, EC 3.4.19.1) and oligopeptidase B (OB, EC 3.4.21.83) (Rawlings et al., 1991). The POP family is different from the classical serine proteases, trypsin and subtilisin, based on their specificity for peptide substrates and their different catalytic triad topology. An amino acid sequence comparison between POP family enzymes from different species illustrates the evolutionary relationships. The phylogenetic trees indicate that POP family enzymes evolved even before the archaea, prokaryota and eukaryota diverged along their evolutionary lines between 2000 and 4000 million years ago. This suggests that the members of POP family were present in the last universal common ancestor of all life forms (Venäläinen et al., 2004).

Much of our understanding of POP is based on in vitro peptidase assays using standard peptides in non-physiological conditions. These assays indicate that POP cleaves nearly all naturally occurring proline containing peptides, suggesting that it may have a physiological role in modulating the levels of all neuronal peptides and hormones containing this residue. As these peptides regulate a large number of signalling pathways, POP potentially has the capacity to modulate a variety of cellular tasks. However, this seems unlikely given that many in vitro targets have not been confirmed in vivo. Moreover, posttranslational modification or limited access to the enzyme may also modulate the effective function of POP.

2. Prolyl oligopeptidase – general aspects 

POP was first found in human uterus as an oxytocin-cleaving enzyme (Walter et al., 1971), and was named post-proline cleaving enzyme (PPCE), since it cleaves peptides at the C-side of proline residue. The name was later changed to prolyl endopeptidase (PEP) and in 1992 it was recommended that the enzyme should be referred to as prolyl oligopeptidase (POP), since it does not cleave large proteins (IUBMB, 1992). There are also other names for this activity in the literature, e.g. PO, PE and post-proline endopeptidase. Eukaryotic POP cleaves short peptides, up to 30 amino acids long, and specific assays have been developed to screen for its occurrence in complex samples. These include digestion assays of model substrates (see Table 1) and/or sensitivity to Z-Pro-Prolinal (ZPP) or analogues (see Table 2). Accordingly, in most of the studies, the cleavage of these putative POP-specific substrates has previously been taken to indicate the presence of the enzyme. According to all genetic or structural information in higher eukaryotes, POP is considered to be a soluble cytoplasmic protein. There are no reports on protein modification or processing which would lead to its association with any other intracellular structures, or trafficking outside the cell. Nonetheless, POP-like activity has been described in a number of alternative locations including membranes, organelles such as mitochondria, the extracellular space, and body fluids including serum and cerebrospinal fluid. The available data favour the presence of alternative proteins in serum and in synaptosomal membranes (O’Leary et al., 1996, Cunningham and O’Connor, 1997a) but in most of the cases the identity of the proteins responsible for this POP-like activity is not known. In the light of the increasing evidence indicating that the activity measured with so-called “specific substrates”, and sensitivity to so-called “specific inhibitors”, may not be exclusively associated to soluble unmodified POP, posttranslational modification and/or alternative gene products must be considered in order to explain all the different locations and processes that have been attributed to POP. Kinetic studies of complex samples, demonstrating differences in sensitivity to inhibition, substrate specificity, or affinity changes, further point to the presence of alternative or modified enzymes. Thus, the discussion which follows considers published information concerning definite cytoplasmic POP features as well as studies where the data indicate a possible alternative POP-like activity.

Table 1. Small (<30-mer) proline-containing bioactive peptides which are susceptible to digestion by POP and the type of evidence for digestion in vitro, or in vivo

	Substrate	Sequence	Type of evidencea
	Substance P	R-P-K-P-Q-Q-F-F-G-L-M	A, B, C, D(t), E, F
	TRH	pQ-H-P–NH2	A, D(t)
	GnRH	pQ-H-W-S-Y-G-L-R-P-G-NH2	A, B
	Arginine-vasopressin	C-Y-F-Q-N–C-P-R-G	A, D, E, F(t)
	Angiotensin I	D-R-V-Y-I-H-P–F-H-L	A
	Angiotensin II	D-R-V-Y-I-H-P–F	A, B(t), C, D
	Angiotensin III	R-V-Y-I-H-P–F	A
	Angiotensin IV	V-Y-I-H-P–F	A
	Bradykinin	R-P–P-G-F-S-P–F-R	A, B(t), G
	Oxytocin	C-Y-I-Q-N–C-P-L-G-NH2	A
	β-Endorphin	Y-G-G-F-M-T-S-E-K-S-Q-T-P-L-V-T-L-F-K-N-A-I-I-K-N-A-Y-K-K-G-E	A
	Neurotensin	pQ-L-Y-E-N-K-P-R-R-P-Y-I-L	A
	α-MSH	S-Y-S-M-E-H–F-R-W-G-K-P-V-NH2	D(t)
	β-Casomorphin	Y-P–F-P-G-P-I	A
	LVV-hemorphin-7	L-V–V-Y-P-W-T-Q-R-F	C
	Morphiceptin	Y-P–F-P–NH2	A
	Urotensin II	D-T-P-D-C–F-W-K-Y-C-V	A
	Octadecaneuropeptide ODN	Q-A-T-V-G-D-V-N-T-D-R-P-G-L-L-D-L-Kb
	Humanin	M-A-P-R-G-F-S-C-L-L-L-L-T-S-E-I-D-L-P-V-K-R-R-Ab	A
	Endomorphin-1	Y-P-W-F-NH2	n.f.
	Endomorphin-2	Y-P–F-F-NH2	n.f.
	Enkelytin	F-A-E-P-L-P–S-E-E-E-G-E-S-Y-S-K-E-V-P-E-M-E-K-R-Y-G-G-F-M-OH	n.f.
	Tyr-MIF-1	Y-P-L-G-NH2	n.f.
	Tyr-W-MIF-1	Y-P-W-G-NH2	n.f.
	Dynorphin A	Y-G-G-F-L-R-R-I-R-P-K-L-K-W-D-N-Q	n.f.
	Galanin	G-W-T-L-N-S-A-G-Y-L-L-G-P-H-A-V-G-N–H-R-S-F-S-D-K-N-G-L-T-S	n.f.
	Neuromedin N	K-I-P-Y-I-L	n.f.
	Neuromedin U	Y-K-V-N-E-Y-Q-G-P-V-A-P-S-G-G-F-F-L-F-R-P-R-N–NH2	n.f.
	Kinetensin	I-A-R-H-P-Y-F-L	n.f.
	Orexin	R-S-G-P-G-L-Q-G-R-L-Q-R-L-L-Q-A-S-G-N–H-A-A-G-I-L-T-M-NH2	n.f.
	Somatostatin	S-A-N-S-N-P-A-M-A-P-R-E-R-K-A-G-C-K-N-F-F-W-K-T-F-T-S–C	n.f.
	P55-Tumor necrosis factor	L-P-Q-I-E-N-V-K-G-T-E-E	n.f.
	P75-Tumor necrosis factor	S-M-A-P-G-A-V-H-L-P-Q-P	n.f.
	β-MSH	D-F-D-M-L-R-C-M-L-G-R-V-Y-R-P–C-W-Q-V	n.f.
	Melanin concentrating hormone	D-F-K-K-D-E-G-P-Y-R-M-E-H–F-R-Y-G-S-P-P-K-D	n.f.
	Tuftsin	T-K-P-R	n.f.
	TRH-precursor peptide	K-R-Q-H-P-G-K-R	n.f.
	Gastrin	pE-G-P-W-L-E-E-E-E-E-A-Y-G-W-M-D-F-NH2	N
		Model compounds for activity assay
		Suc-Gly-Pro-AMC
		Suc-Gly-Pro-Leu-Gly-Pro-AMC
		Suc-Ala-Pro-pNA
		Z-Ala-Pro-4MbNA
		Z-Gly-Pro-AMC
		Z-Gly-Pro-bNA
		Z-Gly-Pro-pNA
		Z-Gly-Pro-Ala-OH
		Z-Gly-Pro-Leu-Gly-Pro-OH
		Z-Lys-Pro-4MbNA
		Pyr-His-Pro-bNA

See text for discussion. The cleavage bonds are shown in bold.

a A, Digestion by purified POP; B, digestion by tissue crude preparations; C, digestion sensitive to POP inhibitors by crude tissue preparations; D, modulation of peptide levels by POP inhibitors in vivo; E, peptide effect potentiated by POP inhibitors in vitro; F, peptide effect potentiated by POP inhibitors in vivo; G, genetic modulation of POP by the peptide or analogues; (t) tissue or cell-type dependent; N, peptide insensitive to POP; n.f. no records found. b Unusual cleavage site(s).

Table 2. Inhibitors of prolyl oligopeptidase

	Compound	Potency	Reference
	Z-Pro-Prolinal	Ki=0.35nM (mouse POP)	Bakker et al. (1990)
	JTP-4819	IC50=0.83nM (rat POP)	Toide et al., 1995a, Toide et al., 1995b
	ONO-1603	Ki=12nM (rat POP)	Katsube et al. (1996)
	SUAM-1221	IC50=7.2nM (rat POP)	Atack et al. (1991)
	S 17092	Ki=1.5nM (human POP)	Barelli et al. (1999)

3. Structure and catalytic mechanism 

POP has been purified from several eukaryotic and prokaryotic sources. The 710 amino acid long porcine POP and its complex with benzyloxycarbonyl-Pro-Prolinal (Z-Pro-Prolinal or ZPP) have been crystallized and the structure resolved to high resolution (Fülöp et al., 1998) (Protein Data Bank codes 1qfm and 1qfs). POP has a cylindrical shape with a height of 60Å and a diameter of 50Å, and consists of two domains (Fig. 1). The peptidase domain is formed by the N- and C-termini (residues 1–72 and 428–710) containing the catalytic triad (Ser554, Asp641 and His680), and is arranged in a classical α/β hydrolase-fold. The seven-bladed β-propeller domain (residues 73–427) is radially arranged around the central tunnel embedded within the cylinder, where the narrow active site is located. The first and the last blades are stabilised together by relatively weak hydrophobic interactions, contrary to the situation in the classical β-propellers found in other enzymes where hydrogen bonds and disulfide bridges form a “velcro” lock (Fülöp et al., 1998).

View Large Image Download to PowerPoint

Fig. 1. The structure of prolyl oligopeptidase with the inhibitor Z-Pro-Prolinal bound at the active site (Protein Data Bank code 1qfs).

POP cleaves peptides at the carboxyl side of an internal proline (-Pro-Xaa-; where Xaa≠Pro), and it will not cleave an N-blocked peptide where Pro is the second amino acid (Cunningham and O’Connor, 1997a, Cunningham and O’Connor, 1997b, Rosenblum and Kozarich, 2003). The enzyme interacts maximally with six amino acid residues of the substrate peptide: those in positions P4, P3 and P2 from the N-side, and those in positions P1′ and P2′ from the C-side of the proline which occupies the P1 position (Fülöp et al., 2001). The highest reaction rates are obtained when P1′ is a hydrophobic residue (Cunningham and O’Connor, 1997b, Rosenblum and Kozarich, 2003).

POP also cleaves – Ala-X bonds but at a much lower reaction rate (Polgár, 1992). Recently, it has been reported that a bacterial POP is able to cleave Val-X bonds even slower than Ala-X bonds, but the physiological relevance of this fact is claimed to be negligible (Leprince et al., 2006). A different POP specificity has also been reported recently where the cellular rescue factor humanin is digested in vitro after cysteine (Bar et al., 2006), however this substrate affinity has not been determined. POP catalytic efficiency (kcat/Km) for Ala-Ala-Cys-AMC was reported to be in the same order of magnitude as for most standard substrates (Table 1) but the Cys-containing substrate Km is 2–3 orders of magnitude higher than the Pro-containing counterparts (see for example Szeltner et al., 2000, Szeltner et al., 2002).

The S1 binding pocket has more features contributing to specificity than the other sub-sites, involving residues that provide a hydrophobic environment and a snug fit for the proline residue of the substrate. The S3 subsite favours the interaction with hydrophobic residues in P3 position. It has been shown that the S1 subsite of POP has remained virtually unchanged throughout the evolution whereas S3 subsite exhibits substantial variation between different species (Venäläinen et al., 2004).

It has been demonstrated that POP catalysis is controlled by a gating filter mechanism that only allows small peptides to gain access to the active site; in this way, larger structural peptides and proteins are protected from proteolysis (Fülöp et al., 2000). The 3D structure of POP has helped to understand the mechanism for the specificity for small peptides and describe how the substrate-induced conformational change is the rate limiting step during catalysis (Polgár, 1991, Polgár, 1992, Polgár et al., 1993, Fülöp et al., 1998, Fülöp et al., 2000, Fülöp et al., 2001, Szeltner et al., 2004). The enzyme active site is located in a large cavity at the interface of the two domains, and it has been proposed that there is a narrow hole at the centre of the β-propeller at the bottom of the enzyme, through which the substrate enters. This narrow entrance could be widened by movements of residues with flexible bonds covering the central tunnel, thus this entrance would act as filter, permitting only small peptides to enter into the active site (Fülöp et al., 1998, Fülöp et al., 2000). However, the apparent rigid crystal structure of the propeller does not explain how the substrate can approach the active site. The 3D structure and kinetic analysis of mutated variants of POP (Szeltner et al., 2004), along with molecular dynamic studies (Fuxreiter et al., 2005) have strongly suggested that there is an alternative route for substrate. The studies indicate that the substrate induces an opening at the interface of the peptidase and the β-propeller domains while entering into the active site and that concerted movements of the propeller and peptidase domains are required for enzyme action. This mechanism was verified when the structure of Sphingomonas capsulata POP in the open configuration was resolved (Shan et al., 2005).

The catalytic mechanism of POP conforms to the general serine protease reaction mechanism via base-assisted catalysis by histidine, which occurs through a tetrahedral intermediate, generating a covalent acyl enzyme complex that is subsequently hydrolyzed. The hydrolysis involves a second tetrahedral intermediate which breaks down via acid-assisted catalysis by histidine (Leung et al., 2000). The cycle has been confirmed by elegant X-ray and NMR experiments (Kahyaoglu et al., 1997, Fülöp et al., 1998).

4. Gene expression and distribution 

The gene of the mouse cytoplasmic POP, Prep, maps to chromosome 10B2-B3, it is spread in a region of about 92kb in size, and it contains 15 exons (Kimura et al., 1999). The catalytic domain of the enzyme is coded by exons 1–3 and 10–15 and the propeller domain is coded by exons 3–10. Two of the catalytic triad residues (Asp641 and His680) coding regions are located within exon 15 and the active site serine residue (Ser554) is encoded by exon 13. An National Centre for Biotechnology Information (NCBI) gene databank search reveals only one POP gene in humans with a similar structure of 15 exons. This gene is located at chromosome 6q22. Analysis of mice chromosomal sequence up-streams to the POP start codon has been interpreted to indicate that the gene codes for a house-keeping enzyme (Kimura et al., 1999). However, this analysis was limited to a relatively short region. More detailed studies in this matter are currently lacking.

POP cDNA has been sequenced from a large variety of organisms including bacteria, protozoa, plants, and animals (Yoshimoto et al., 1988, Rennex et al., 1991, Yoshimoto et al., 1991, Chevallier et al., 1992, Shirasawa et al., 1994, Vanhoof et al., 1994, Robinson et al., 1995, Yoshimoto et al., 1997, Ishino et al., 1998, Kabashima et al., 1998) but not from yeasts, where no homologous gene has been found. Mammalian POP has been more widely studied, but also the bacterial version has been focus of attention. Parasitic protozoan POP has been also studied as it might be implicated in the infection process. Despite the close sequence similarity between bacterial, and probably protozoan POP with the mammalian counterpart, they do not share the same physiological roles. Differences between these different kingdom enzymes indicate that their specificity would not be fully comparable, and kinetic features should probably not be confused. Several studies have reported substrate and/or inhibitor kinetic features in bacterial enzymes and extrapolate their implications to the mammalian context. However, this has only a weak basis. All of the mammalian enzymes for which the sequence is available are 710 amino acids long (approximately 80kDa), globular and cytoplasmic (Venäläinen et al., 2004), but some mature bacterial POP are membrane bound and active extracellularly (Xie et al., 2004), as are most of the protozoan versions (Grellier et al., 2001, Yang et al., 2006). Bacterial enzymes are able to digest longer peptides, even proteins and some are capable of autoproteolysis (Harwood et al., 1997, Harwood and Schreier, 2001), in contrast to the mammalian enzymes for which a no peptide substrate longer than 30-mer has been found, and there is structural, biochemical, and molecular dynamics evidence that it is indeed sterically impossible for these enzymes to accommodate larger proteins (Fülöp et al., 1998, Fülöp et al., 2000, Szeltner et al., 2004, Fuxreiter et al., 2005).

In mammals, POP activity has been detected in all organs and tissues studied. In rats, some studies indicate that the highest enzyme activity is found in the brain, with decreasing amounts in liver, heart, kidney, spleen, and lung (Yoshimoto et al., 1979, Taylor et al., 1980). The activity distribution in rabbit tissues has been shown to be similar to that of rat (Orlowski et al., 1979).

In humans, POP activity measured in body fluids (blood, seminal fluid, prostate fluid) is much lower than the activity assayed in tissues (Orlowski et al., 1979, Yoshimoto et al., 1979, Taylor et al., 1980, Goossens et al., 1996b). Proliferating cells exhibited enhanced activity and was shown to be maximal in fibroblasts, decreasing in epithelial cells, renal cortex, testis, heart endothelial cells, lymphocyte, platelet, rectum, spleen and lung, respectively. One interesting finding is that the activity in prostate, lung and sigmoid tumours is significantly higher than in healthy tissues (Taylor and Dixon, 1980).

There have been few studies done on the distribution of POP within the central nervous system. In humans, the highest total POP activity was found in the brain cortices with the cerebellum being three times lower (Kato et al., 1980b, Irazusta et al., 2002). Moderate activity was measured in the striatum, hypothalamus, hippocampus and amygdala. The regional distribution of POP activity within the rat brain differs from that found in humans, and in mouse (Dauch et al., 1993, Agirregoitia et al., 2003a, Agirregoitia et al., 2003b). In the rat, the distribution is rather homogeneous in most of the brain areas, with the lowest activities being found in the cerebellum and the pituitary gland (Dauch et al., 1993, Irazusta et al., 2002, Agirregoitia et al., 2003a, Agirregoitia et al., 2003b). In another study, POP mRNA levels in the rat brain and pituitary were well correlated with total enzymatic activity measurements (Bellemere et al., 2004) but important differences were found, as is schematised in Fig. 3. Maximal POP-like activity was detected in rat cerebral cortex, where the mRNA was only at medium levels of expression. The lowest activity was detected in the cerebellum where the highest mRNA levels were found (Fig. 3). The differences between mRNA and protein expression can be attributed to several reasons. First, the site of enzymatic action does not have to coincide with the site at which POP is translated, i.e., the protein could be synthesized in one location and be transported to another location where it is active. In fact, Schulz et al. (2005) have presented evidence that POP is associated to perinuclear cytoskeletal proteins. This in turn may indicate involvement of POP in axonal transport, and indeed in an intracrine system, in which active intracellular peptidases are regulating the peptide balance (see also VI). Although this is just an hypothetical situation, it would be worth to test. Second, since the activity has been measured in complex crude samples, the possibility of detecting similar activity attributable to some uncharacterised peptidase, should not be ruled out, especially when the assay has been undertaken on particulate preparations (see membrane POP section). In any event, it is interesting to note that POP distribution studies have indicated that the enzyme is actively expressed in brain areas containing various neuropeptide receptors, such as SP receptors NK1 and NK3, AVP receptors V1a and V1b and TRH receptors TRH-R1 and TRH-R2 (Bellemere et al., 2004).

View Large Image Download to PowerPoint

Fig. 2. Inositol recycling. Stimulus activates phospholipase C (PLC), which cleaves phosphatidylinositol bisphosphate [PI(4,5)P2] to inositol-1,4,5-trisphosphate (IP3) and diacylglycerol (1,2-DAG). Lithium blocks inositol polyphosphate 1-phosphatase (IPPase) and inositol monophosphatase (IMPase), which inhibits recycling of inositol and synthesis from glucose-6-phosphate. IP3 is also released from high order inositol phosphates. Multiple inositol polyphosphate phosphatase (MInsPP) dephosphorylates IP6 to IP3 and is regulated by POP (Harwood, 2003, Williams and Harwood, 2000).

View Large Image Download to PowerPoint

Fig. 3. Comparison of PEP mRNA content and specific activity in main rat brain areas. Data taken from Bellemere et al. (2004), mRNA content (white bars), and from Irazusta et al. (2002), enzymatic activity (black bars). ∗mRNA content determined in caudate putamen. 1 Total activity.

Recent imunohistochemical experiments have revealed that, in POP transfected neuronal and glial cells in culture, the enzyme is localized fibrillary and perinuclearly (Schulz et al., 2005), but in mouse and human frontal cortex slices, the peptidase seems to be only localized in neurons, since POP immunoreactivity was not observed in activated glial cells in the proximity of β-amyloid plaques either in human patients with Alzheimer’s disease (AD) or Tg2576 mice brain samples (Rossner et al., 2005).

5. Inhibition 

The discovery that POP inhibitors can revert memory loss caused by amnesic compounds like scopolamine (Yoshimoto et al., 1987) initiated an intense search for specific inhibitors. Hundreds of compounds either synthetic or natural in origin have been tested for their POP inhibitory activity (for recent reviews see for example De Nanteuil et al., 1998, Wallén, 2003).

The vast majority of the reported POP inhibitors are substrate-like compounds. The strong preference of POP to cleave peptides specifically at the C-terminal of proline residue has guided the design of POP inhibitors and most of the compounds contain a proline or proline analogue residue. Z-Pro-Prolinal, JTP-4819, ONO-1603, SUAM-1221 and S 17092 are probably the most widely studied POP inhibitors (Table 2). These are all substrate-like compounds with majority of them having a substituted or unsubstituted pyrrolidinyl group at the P1 site. The only exception is S 17092, in which a thiazolidine has been successfully incorporated at the P1 position. The nature of the substitution group on the P1 site pyrrolidinyl ring has been shown to be crucial for the inhibitor binding into the enzyme active site. An aldehyde group at the P1 site of Z-Pro-Prolinal forms a hemiacetal adduct with the nucleophilic hydroxyl group of the active site Ser554 residue. This covalent adduct mimics the tetrahedral transition state of the enzyme-catalyzed reaction and improves the affinity of the inhibitor for the enzyme (Wilk and Orlowski, 1983, Fülöp et al., 1998). Also other electrophilic P1 site substituents, cyano- and hydroxyacetyl groups, have been successfully used to improve the affinity of the inhibitors. It is probable that the cyano and hydroxyacetyl groups react similar to the aldehyde group, forming imino ether and hemiketal adducts with the active site Ser554 residue (Wallén et al., 2002a).

More detailed analysis of the inhibition by compounds with electronegative substituents has shown that the addition of aldehyde (Wallén et al., 2002b), hydroxyacetyl and cyano group (Bakker et al., 1990) at the P1 site results in slow-binding type of inhibition. The slow binding inhibition is a phenomenon in which the inhibition of enzyme activity occurs relatively slowly and analysis of slow-binding inhibition allows the calculation of enzyme–inhibitor association and dissociation rates. Although electronegative P1 site substituents decrease to some degree the onset of inhibition, more striking effects are seen in the dissociation rates of enzyme–inhibitor complex as cyano and hydroxyacetyl groups at P1 site decreased the dissociation rate to 1/100 or lower. This difference could also be seen in the duration of POP inhibition in the rat tissues after administration of inhibitors – those compounds with electronegative P1 site substituents produced much longer lasting inhibition in the tissues than their unsubstituted counterparts (Venäläinen et al., 2006).

L-proline is the preferred residue also at the P2 position of POP inhibitors and replacement of the L-proline residue has been difficult without decreasing the inhibitor potency. However, a few successful replacements have been introduced, using L-thioproline (Saito et al., 1991), bicyclic L-proline analogue (Barelli et al., 1999), bulky 5-substituents of L-proline (Wallén et al., 2003) or cyclopent-2-enecarbonyl (Jarho et al., 2004) groups at the P2 site.

The substrate binding subsite S3 of POP is lined with non-polar residues (Fülöp et al., 1998) and therefore lipophilic aliphatic structures are favoured at the P3 position of the POP inhibitors (Jarho et al., 2005). The area beyond the S3 subsite is large and the interactions between POP and the inhibitors are weak. Therefore, this cavity can accommodate several different inhibitor structures without affecting the potency and this can be used as a modification site for the ADME properties optimisation (Jarho et al., 2005).

POP inhibition in a biological context shall be discussed in following sections.

6. Substrates 

In principle, all natural short peptides with an internal proline residue are potential POP substrates. However, due to cellular location of peptides and also to secondary modifications (see below), some of these peptides would theoretically not be accessible to this peptidase which resides in the cytosol. In fact, the final steps of neuropeptide processing occur in synaptic vesicles and the degradation of most of these peptides is believed to occur in the synaptic cleft. On the other hand, there is some recent evidence that several POP substrates, like thyrotropin releasing hormone (TRH), gonadotropin releasing hormone (GnRH), angiotensins, oxytocin and dynorphin, are involved in intracrine systems (for a review, see Re and Cook, 2006 and references therein). Some substance P receptors are also expressed intracellularly (Levesque et al., 2006). Therefore, the intracellular termination of the intracrine effect of peptides would have sense in this context.

Table 1 summarizes the known bioactive peptides which have at least one internal proline and thus, are susceptible to degradation by POP. For the majority of these peptides, there exists some in vitro evidence that they are indeed POP substrates as has been demonstrated by assays with crude tissue preparation or with purified tissue or recombinant enzymes. The in vivo involvement of POP in the metabolism of those substrates is all indirect, and no conclusive evidence has been reported mainly because crude preparations have been used. However, there are a few cases in which the evidence indicates that POP is indeed controlling peptide levels, and several examples will be discussed below.

7. Regulation 

In principle, POP regulation might be operative at different levels: genetic, posttranslational, hormonal and kinetic, but the mechanisms involved, and the relative importance of these putative regulative strategies are not known.

At the gene level, there are hints of multiple initiation sites (Kimura et al., 1999), although there are no reports which would have identified any protein variants. Some of the evidence for genetic regulation is indirect. POP expression has been shown to be up-regulated by deletion of math5, a transcription factor involved in neural development of embryonic retinal cells (Mu et al., 2005). Early studies in rat have indicated that POP is developmentally regulated since activity in brain lung and kidney was increased until the second week after birth, then decreasing during maturation (Kato et al., 1980a, Fuse et al., 1990). More recent tests have revealed that brain POP activity is already high in embryonic stages. It is only slightly increased in striatal tissue after first few days after birth, and then decreased until adulthood (Agirregoitia et al., 2003a). No further changes were recorded during aging (from postnatal month 3 to 22), except in the lungs, where a significant decreased was observed (Agirregoitia et al., 2003b), and in the brain stem, where an increase of particulate POP was detected at postnatal month 22 (Agirregoitia et al., 2003a). On the other hand, reports have highlighted variations on POP expression according to age or environmental conditions. In 22-month-old mice, POP mRNA levels have been reported to be increased 11-fold in the hypothalamus and 2.7-fold in the cortex compared to those levels in 2 month old mice (Jiang et al., 2001). This increase was the highest among 11,000 genes studied. This study is in line with earlier report where POP-like immunoreactivity in mice has been found to increase with aging (Fukunari et al., 1994). Similarly, mice subjected to an enriched environment showed POP gene expression reduction to one third in the cortex, a decrease comparable with that seen for several caspase genes (Rampon et al., 2000).

It is known that POP activity is sensitive to redox conditions (Polgár, 2002) and indications of modifications via disulfide bonds have been suggested. Furthermore, intermolecular disulfide bonds in POP were found to be formed during oxidative stress (Cumming et al., 2004). These findings seem to suggest that redox sensitive modification could be operative for POP.

Ohta et al. (1992) showed that total POP activities in uterus and ovary of intact female mice vary during the estrous cycle, and estradiol-17b or progesterone treatments could significantly increase this activity in the uterus of ovariectomized mice. Administration of cortisol has also been reported to increase POP activity in cultures of placental tissues (Yasuda et al., 1992). Anti-androgen treatment of patients suffering from prostate carcinoma increased serum POP activity (Goossens et al., 1996a). These studies indicate that total POP activity is at least partly modulated by steroids. However, it is not known whether the increase in enzyme activity results from a specific increased POP synthesis or a modulation of enzyme activity.

Yoshimoto et al. (1982) found an endogenous peptide that inhibited POP activity in a competitive manner. This peptide was highly specific for POP and it was widely distributed in rat and porcine organs. A few years later this 7000Da peptide was purified from rat brain and an inhibition constant (Ki) value of 0.67mM was reported (Soeda et al., 1985). Some experiments have shown that certain polyamines can regulate enzyme activity by reversing the POP inhibition by this peptide. Polyamines did not act directly on POP, i.e. they were unable to alter the activity of the purified enzyme (Soeda et al., 1986). More evidence for a concerted effect of the endogenous POP inhibitor and polyamines was observed as parallel changes of total POP activity and polyamine levels. Inverse changes of polyamine and POP inhibitor levels, were detected following the liver regeneration which occurs after hepatectomy in rats (Yamakawa et al., 1994). That study provided convincing evidence that the activity is regulated both by endogenous inhibitor levels and by intracellular cationic charges such as those present in polyamines.

It has also been shown that partial digestion of pig POP by trypsin increased the catalytic activity of the enzyme (Polgár and Patthy, 1992, Szeltner et al., 2004). It was shown that low levels of trypsin specifically cleave the bond Lys196-Ser197, and apparently the two proteolytic fragments are held together in a catalytically competent complex. It is not known if this trypsin-induced activation has any physiological significance.

POP activity is sensitive to the phosphorylation state of the substrate. It has been shown that the reaction rate can be increased if the P1′ residue is phosphorylated – in fact the enzyme even failed to cleave the Ala-Ser bond if the serine was unphosphorylated (Rosen et al., 1991). This is interesting since it suggests that peptide phosphorylation could control the processing of POP substrates (Kaspari et al., 1996).

8. Physiological roles 

The wide distribution of POP activity, its in vitro specific activity towards several bioactive neuropeptides and the changes in its activity under certain conditions have led to speculation that this enzyme is involved in several important physiological functions. These include cell division and differentiation, learning and memory, signal transduction and psychiatric disorders. However, a precise understanding of the actual relevance of POP is still largely missing.

Total POP activity in the rat liver during postnatal development has been found to be highly correlated with the extent of proliferation and differentiation of liver cells (Matsubara et al., 1998). In female pigs, the mRNA expression and activity of POP were higher in small ovarian follicles than in larger follicles, suggesting that POP is involved in the early stages of follicular development (Kimura et al., 1998). The changes in the POP mRNA distribution in the mouse testis during sexual maturation also indicate that this enzyme might be implicated in meiosis and differentiation of the testicular cells (Kimura et al., 2002).

Studies with POP inhibitors have revealed more detailed information about the link between POP activity and cell division. Addition of the POP inhibitor, N-benzyl-oxycarbonyl-thioprolyl-thioprolynal-dimethylacetal (ZTTA), to cultured Sarcophaga leg imaginal discs, prevented disc differentiation from entering into the elongation stage (Ohtsuki et al., 1994). Furthermore, in an embryonic Sarcophaca NIH-Sape-4 cell line, ZTTA inhibited DNA synthesis, and therefore cell proliferation, but had no effect on protein synthesis (Ohtuski et al., 1997). In addition, POP has been shown to be involved in the proliferation and DNA replication in a mouse Swiss 3T3 cell line (Rennex et al., 1991). Parallel observations have been made also in animal studies since repeated administration of the POP inhibitor Z-Gly-Pro-CH2 to partially hepatectomized rats significantly suppressed the rate of liver regeneration, and the suppressed liver growth was reversed by terminating the inhibitor treatment (Yamakawa et al., 1994).

The mechanisms by which POP is involved in the cell proliferation and differentiation are unclear at the moment. The localization of POP in the nuclei of growing but not adult NIH-Sape-4 cells and the partial nuclear localization in Swiss 3T3 cells suggest that this peptidase could have a role in the degradation of some nuclear protein kinases, or their inhibitors, which are involved in cell division (Ishino et al., 1998).

POP has been suggested to be involved in memory-related behaviour since several putative POP substrates, such as SP, AVP and TRH are known to enhance learning and memory (de Wied et al., 1984, Griffiths, 1987, Kovacs and De Wied, 1994, Huston and Hasenohrl, 1995, Hasenohrl et al., 2000, Morain et al., 2002). The expression of POP in the rat brain also strongly overlaps with the distribution of receptors for SP, AVP and TRH (Rossner et al., 2005). In the human brain, the highest POP activities are found in areas that are involved in learning and memory processes (Irazusta et al., 2002). As emphasized above, there is evidence of differential expression of POP gene according to age and environmental conditions, suggesting that it may play a role in learning and memory (Miura et al., 1997).

9. POP and disease 

Different POP inhibitors have effects on memory and learning tasks in rodents under a variety of experimental conditions or during senescence (Morain et al., 2002). The memory enhancing effects of POP inhibition has also been observed in primates; S 17092 significantly improved cognitive performance of a monkey treated with 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP), a dopaminergic toxin (Schneider et al., 2002).

A number of studies have shown that the levels of several neuropeptides are altered in the brains of patients suffering from the neurodegenerative diseases which evoke memory loss, such as AD. For example, post-mortem levels of SP, AVP and NT have been reported to be decreased in several brain areas of AD patients (Mazurek et al., 1986, Beal et al., 1987, Beal and Mazurek, 1987, Nemeroff et al., 1989, Benzing et al., 1990). Since these neuropeptides are putative POP substrates, it has been speculated that also this activity might be altered in AD (Ishiura et al., 1990). Several studies have explored this possibility but the results have been conflicting or even contradictory. Different studies have reported either increases (Aoyagi et al., 1990a, Aoyagi et al., 1990b) or decreases (Gibson et al., 1991, Ichai et al., 1994, Terwel et al., 1998) of POP activities in the post-mortem brains of AD patients. Decreased POP activity in AD patients has been linked to antemortem hypoxia (Terwel et al., 1998) since cytoplasmic POP is sensitive to the intracellular redox state, which is inhibited by the oxidized glutathione form (GSSG) that accumulates under stress conditions (Tsukahara et al., 1990). Oxidized glutathione most likely inhibits POP by reacting with Cys255 and thus sterically blocking the binding of the substrate to the enzyme active site (Fülöp et al., 2000). Therefore, the POP activities measured from AD patients may reflect the conditions present during antemortem hypoxia, such as oxidative stress (Terwel et al., 1998). This may, in part, explain the conflicting results from assays of POP activities in the brains of AD patients.

In addition to AD, POP activity has been claimed to be altered in other neurodegenerative diseases such as Lewy body dementia, Parkinson’s disease (PD), and Huntington disease (HD) (Mantle et al., 1996). On the other hand, delayed cell death in brain ischemia can be reverted by POP inhibitors or by AVP or TRH administration (Shishido et al., 1999a). High POP expression has been reported in murine T-cells upon activation conducive to cell death (Odaka et al., 2002), and POP inhibitors can offer protection against this activation-induced cell death. It has been shown in cultured cells, that after apoptosis induction, GAPDH is over expressed and cell death is followed by translocation of this enzymes to the nucleus (Ishitani et al., 1998). In that study, it was proposed that deprenyl derivatives (PD drugs) may inhibit GAPDH translocation, while AD drugs such as tacrine and ONO-1603, which is a POP inhibitor, can block GAPDH overexpression and translocation. In fact, it has been shown that POP inhibitors are able to counteract oxidative stress factors (ROS) in certain types of cells in culture, and inhibit GAPDH nuclear translocation induced by apoptotic toxicants (Puttonen et al., 2006). As massive cell loss is implicated during neurodegeneration and there is increasing evidence that apoptosis is involved in that cell death (Berry and Boulton, 2003) it could be speculated that POP would have a role in neurodegeneration.

Changes in total serum POP activity (see also discussion on serum POP in Section 9) have been linked with several psychiatric disorders, such as depression, mania, schizophrenia, eating disorders and stress induced anxiety. However, no specific mechanisms have been discovered to explain the involvement of POP in these pathologies. Total serum POP-like activity have been found to be significantly lower in the plasma of patients with depression compared to normal individuals (Maes et al., 1994, Maes et al., 1998b). Serum POP activity also negatively correlated with the severity of the depression; patients with major depression had lower POP activities than those with minor depression. Melancholic depressive patients showed the lowest POP activities (Maes et al., 1994). Treatment with antidepressant drugs, such as fluoxetine, significantly increased the serum total POP activity in depressed subjects. In contrast to depressive patients, manic and schizophrenic subjects exhibited increased serum total POP activity. Treatment of manic patients with valproate restored the POP activity to around control levels. Treatment with neuroleptics had no effect on POP activity in schizophrenic patients (Maes et al., 1995).

Interestingly, higher levels of POP activity have been detected in patients with post-traumatic stress disorder (Maes et al., 1999b) and in subjects that are anxiety responders to psychological stress (Maes et al., 1998a). Low serum POP activity has been related to anorexia and bulimia nervosa (Maes et al., 2001) and also to alcohol dependence (Maes et al., 1999a). In humans, ZPP sensitive activity makes a lower contribution to the serum total POP activity levels (Breen et al., 2004) but both activities have been found to be decreased in patients treated for bipolar disease, and no difference was found for schizophrenic patients. However, no distinction was made between ZPP sensitive and resistant POP activity in most of these studies and thus, activity changes will have to be re-evaluated in the light of the recent findings.

At the moment, there are two broad theories that explain how POP activity may influence the psychological state. Several substrates of POP, such as TRH and AVP, are known to be involved in the pathophysiology of depression and significant alterations in SP, neurotensin and AVP levels have been observed in schizophrenic patients. Therefore, altered POP activity, could modulate emotion and response to stress by changing the concentrations of these behaviourally active neuropeptides (Maes et al., 2004).

The other potential mechanism was proposed when it was discovered that cytoplasmic POP is linked to inositol-1,4,5-triphosphate (IP3) formation and that POP inhibition may counteract the effects of lithium therapy, a drug used in the therapy of manic-depression (Berridge et al., 1989, Williams et al., 1999). One of the cellular effects of lithium is the inhibition of inositol monophosphatase (IMPase) and inositol polyphosphate 1-phosphatase (IPPase), which are the key enzymes in the synthesis and recycling of inositol. As a result, the concentration of the second messenger IP3 is reduced and this, in part, has been claimed to explain at least some of the anti-manic effects of lithium (Berridge et al., 1989). In Dictyostelium, it was found that a mutant deficient for the dpoA gene (POP coding gene) was resistant to the effects of lithium. The same effect was seen in wild-type cells treated with a POP inhibitor and it was found that the loss of POP activity evoked a 4-fold increase in the IP3 concentration, effectively counteracting the lithium effect (Williams et al., 1999). POP has been shown to control the IP3 levels also in a mammalian cell line (Schulz et al., 2002). POP controls the IP3 synthesis by negatively regulating multiple inositol polyphosphate polyphosphatase (MInsPP), which generates IP3 from IP5/6 (Fig. 2). IP5/6 is very abundant in the cell and it may act as a cellular store for inositol and phosphate. Thus, POP inhibition activates MInsPP and increases the cellular pool of inositol by elevating the formation of IP3 from higher order inositol phosphates. POP inhibitors also reverse the effect of two other mood-stabilizing drugs, carbamazepine and valproic acid (VPA) (Williams et al., 2002). It had been suggested that these mood stabilizing drugs have a common effect on inositol signalling and that the disruptions in the inositol metabolism may represent the cellular basis for mood disorders (Harwood, 2003, Harwood and Agam, 2003). Recently, it has been found that VPA is itself a POP inhibitor (Cheng et al., 2005), explaining the antagonism between these two. However, this still does not explain the interaction between lithium and carbamazepine.

It is not known if and how POP regulates MInsPP activity. Due to its large size, MInsPP cannot be a substrate for POP and therefore the effect must be indirect. This is supported by the finding that the IP3 concentration requires 12h of treatment with a POP inhibitor to reach its maximum level (Schulz et al., 2002). The lowered serum POP activity detected in patients with bipolar disorder is consistent with the IP3 theory; low POP activity could cause an elevation of IP3, which in turn would be reversed by lithium therapy.

Coeliac disease (CD) is an autoimmune-like systemic disorder in individuals having a genetic expression of the HLA type II molecules DQ2 or DQ8. It is triggered by the ingestion of gluten cereals wheat, rye and barley with manifestations not only in the intestine but also in other organs (Hausch et al., 2002, Hausch et al., 2003). A physiological repertoire of digestive enzymes produces several proline-rich peptides, notably so-called the 12-mer, 19-mer and 33-mer fragments that are toxic only to CD patients, not to healthy subjects (Shan et al., 2002, Pyle et al., 2005a). The 12-mer and 19-mer peptides are degraded by administration of mammalian POP but hydrolysis and loss of the antigenicity of the 33-mer peptide can be accomplished only by the supplementation with very high doses of bacterial POP, which is able to accept larger substrates. This finding has opened up the possibility that specific surplus POP’s could be used as an enzymatic therapy to detoxify proline-rich, digestion resistant gluten peptides already in the gut lumen of CD patients (Shan et al., 2004, Pyle et al., 2005b). Flavobacterium meningosepticum POP has been tested as an agent in the successful removal of most of the gut-digestion-resistant gliadin peptides. Unfortunately, high amounts of POP enzyme and a long exposure time are needed for efficient removal of the 33-mer (Matysiak-Budnik et al., 2005).

The role on the endogenous POP on the degradation of peptides and on the aetiology of the disease also has been explored. Matysiak-Budnik et al. (2003) reported that some resistant α2-giadin peptides can be fully digested intraluminally by enterocytes in healthy controls and in patients with treated CD (TCD), In contrast, patients with active CD (ACD) present incomplete degradation of the 33-mer peptide, and also protected intestinal transport of other gliadin peptides. That study also noted that POP activity in duodenal mucosa was higher in patients with CD than in controls or in patients with TCD, but it was concluded that the POP changes play only a minor role in the disease.

Gene expression differences between normal and CD patients was screened in a Dutch population (Diosdado et al., 2004). In this study, the POP gene (Prep) was identified as being the gene most differentially expressed between ACD and TCD patients. Prep was significantly downregulated in patients in complete remission, indicating that gluten may regulate POP expression and that the activity of POP may be impaired in ACD. In subsequent studies, the same group sequenced the Prep gene in 44 normal and diseased individuals and identified several coding and noncoding SNPs but no genetic linkage was found. However, irrespective of their genotypes, ACD patients displayed lower POP activity than TCD although neither CD group was statistically distinguishable from healthy subjects (Diosdado et al., 2005). No correlation was found with age, gender or genotype. These data contrast with the study of Matysiak-Budnik et al. (2003) where the highest level of POP activity was detected in TCD patients. In a further study, we determined both POP enzymatic activity and protein content in intestinal epithelium from CD patients within a Finnish population (García-Horsman et al., 2007). We did not find any significant correlation of POP with CD, but we also found that one of the major immunostimulant gliadin derived peptides, the so-called 33-mer, is a relatively strong inhibitor of endogenous POP, and of any purified mammalian POP. The relevance of this fact on the disease is an open question.

10. POP in serum 

The presence of POP-like activity in serum has been known for more than 25 years. Even the very early reports described differences in the molecular weight and inhibitor sensitivity between brain and serum POP (Kato et al., 1980b). Subsequently, several groups identified the activity in serum as a post-proline cleaving enzyme and a number of specific inhibitors were found to block the degradation of bio-active peptides (Faivre-Bauman et al., 1986, Groth et al., 1997). In the 90’s, more sensitive assay methods were developed and POP activity in serum was evaluated in a number of clinical samples from individuals with psychiatric disorders (Goossens et al., 1992).

ZPP insensitive POP (ZIP) was initially purified as a 174kDa dimer migrating as a 87kDa monomer in SDS–PAGE exhibiting physicochemical similarities with cytoplasmic POP (Birney and O’Connor, 2001), and thus, it was classified as a serine protease since the activity was inhibited by 4-(2-aminoethyl)benzenesulfonyl fluoride (AEBSF) and aprotinin. However, also important differences in the sensitivity towards some serine protease inhibitors were noted. One other interesting difference was its insensitivity of to sulfhydryl reductants such as dithiothreitol (DTT) or β-mercaptoethanol, compounds known to activate cytoplasmic POP. ZIP substrate specificity, however, appeared to be very similar to cytoplasmic POP since ZIP was capable of digesting proline-containing peptides smaller than 20 amino acids.

Recently, ZIP was purified with an alternative procedure (Collins et al., 2004). This preparation contained a 170kDa dimeric serine protease. Similar to cytoplasmic POP, this enzyme requires peptide substrates with residues at the P3 and P2 positions, and the affinity was increased for large hydrophobic residues in the position P1′. Proline is the preferred residue in the P1 site but ZIP can also digest peptides with Leu in that position, though with much less activity. The most marked difference between ZIP and POP is the well-documented resistance of ZIP to specific cytoplasmic POP inhibitors, such as ZPP and JTP-4819. Based on tryptic peptide sequencing, ZIP was identified as seprase (Collins et al., 2004) or fibroblast activation protein (FAP). FAP is a poorly understood integral membrane serine endopeptidase which is highly homologous with the POP family enzymes, especially with dipeptidyl dipeptidase IV (DPPIV). It may have a role in tissue remodelling during development and wound healing, and may contribute to the invasiveness of malignant cancers (Ariga et al., 2001, Chen and Kelly, 2003). FAP has not been found in normal adult tissues. Whether or not ZIP is product of the FAP gene has still to be determined, although all 10 tryptic peptide sequences from ZIP could be matched to the membrane bound FAP sequence. The main concern revolves around the soluble nature of ZIP. There are no identified alternative FAP-like genes, alternative splicing, or transmembrane modifications, which would produce a soluble enzyme. Thus, the possible processing of FAP so that it can give rise to the soluble enzyme ZIP remains speculative.

Another question is whether the assay of serum POP activity is relevant to any assessment of brain POP activity. Nonetheless, there is some evidence from studies in sheep, that POP is released from the vascular bed of the brain and this might represent even a major source of serum POP (Lawrence et al., 1992).

11. Membrane POP 

The presence of a particulate POP activity also was detected more than 20 years ago (Dresdner et al., 1982) and subsequently, the activity has been measured in almost all tissues (Dauch et al., 1993, Irazusta et al., 2001, Agirregoitia et al., 2003a, Agirregoitia et al., 2003b). Due in part to the scepticism about cytoplasmic POP being responsible for the membrane bound activity, efforts have been made to identify this enzyme, and preparations of brain membrane prolyl-specific peptidases have been characterized (O’Leary and O’Connor, 1995, O’Leary et al., 1996). It has been established that cytoplasmic POP can bind so tightly to membranes that only several washes with high salt buffers can detach the activity from these structures (O’Leary and O’Connor, 1995). Checler et al. (1986) described a rat brain proline-specific synaptosomal membrane serine peptidase which was able to cleave neurotensin. They classified the enzyme as a neutral metallopeptidase. This enzyme possessed similar properties to the cytoplasmic POP but had a greater preference for aromatic residues at the substrate P1′ position, and it was not inhibited by ZPP, which clearly differentiated it from POP. Later, the enzyme was cloned, sequenced and classified as zinc-containing metallopeptidase. It was named endopeptidase 3.4.24.16, and it was demonstrated that it was not a proline-specific peptidase (Dauch et al., 1995).

Another report (O’Leary et al., 1996) described a purified 87kDa enzyme also similar to cytoplasmic POP, with high specificity for proline-containing neuropeptides. It was suggested that it was a cysteine protease, rather than a serine protease like cytoplasmic POP, since phenylmethylsulfonyl fluoride (PMSF) and benzamidine were unable to inhibit its activity, whereas thiol protease inhibitors were very effective. On the other hand, this enzyme did exhibit high sensitivity to ZPP, and its sensitivity of metal chelators was interpreted to mean that it requires the presence of a metal for activity.

It is of interest that the endopeptidase 3.4.24.16 sequence, although unrelated to cytoplasmic POP, also shows no particular features of an intrinsic membrane-bound protein, no signal sequence nor glycosylation sites, but it has been demonstrated to be strongly associated to membranes. Furthermore, it has been claimed that the single 3.4.24.16 gene codes for three forms, a cytoplasmic form, a membrane associated form, and a mitochondrial form, with the presence of different initiation sites accounting for this versatility (Dauch et al., 1995). No signal sequences nor glycosylation sites are apparent within the cytoplasmic POP sequence, but different initiation sites have been proposed for POP (Ishino et al., 1998). The relevance of this proposal needs to be addressed in the future.

12. Other POP-like enzymes 

There is emerging evidence for the existence of POP-like enzymes, and also indications that some of them might be structurally related to cytoplasmic POP (Liu et al., 1999, Parvari et al., 2001, Okazaki et al., 2003). Parvari et al. (2001) reported a gene deletion in chromosome 2p16 associated with cystinuria, which is a disorder of the transport of cystine and dibasic amino acids through the epithelial cells of the renal tubuli and the intestinal brush border. Within the genes deleted, an earlier unidentified gene, KIAA0436, was found. This gene has been found also in monkeys, cows, mice and other species, it shares a close homology with the gene of cytoplasmic POP and has now been proposed as a POP-like protein gene (Okazaki et al., 2003). A functional protein analysis, based on activity screening, has positively identified KIAA-POP protein as a serine hydrolase that is expressed in several tissues (Liu et al., 1999). In a recent study, a more detailed analysis revealed that this gene, now called PREPL, codes for an oligopeptidase B homologue rather than a POP-like gene product, and probably it cleaves the peptides after Arg or Lys residues (Szeltner et al., 2005). In that work, several splice variants were identified and one of them, Prepl A, a 638 residue protein, has been expressed in E. coli. Unfortunately, although there are indications that the catalytic serine is the reactive residue, the protein showed negligible hydrolytic activity with all the substrates so far tested, suggesting that it may have some no-enzymatic function (Szeltner et al., 2005). The other splice variants have not been studied.

13. Concluding remarks 

In vitro, cytoplasmic POP is able to digest a large number of biologically active peptides. However, in vivo, there is only a limited evidence that POP is responsible for the metabolism of any of the proposed peptides. However, there is a strong circumstantial evidence that implicates some as more likely physiological substrates, e.g. SP, AVP, TRH, and GnRH but even for these peptides, conflicting results have been reported. In general, it seems that the role of cytoplasmic POP is different in different tissues, and even in different structures of the same tissue. POP is evidently subject to differential regulation depending on the location. The main complication is that most of the studies have been performed on crude and complex samples and, in most of the cases, detection of proteolytic products has been indirect due to the difficulties to assay the peptides directly. On the other hand, most of the peptides considered to be substrates, have indeed physiological relationships and their crosstalk adds a further dimension to the problem. Accordingly, it is hard to establish concussively the physiological relevance of POP.

One of the processes that has been best substantiated where POP is indeed involved is the inositol cycle. However, even in this case a precise mechanism by which the peptidase is able to exert its role is still unknown. On the other hand, the site of POP action on some of the putative physiological substrates would have to be extracellular but the genetic and structural information indicates an intracellular location of the peptidase. Evidently, in order to explain some of these functions, POP would have to be secreted, and the occurrence of POP-like immunoreactivity and specific inhibitor sensitive enzymatic activity in a variety of locations such as membranes, mitochondria and body fluids, like serum, CSF and seminal fluid, suggest an export process and specific targeting. These events would require in turn the existence of POP genetic variants or posttranslational modifications or both. Since there are some putative physiological POP substrates, probably being involved in intracrine systems (Re and Cook, 2006), the intracellular location of POP would be relevant for their regulation. This theory is partially supported by the findings that POP it is localized perinuclearly (Schulz et al., 2005), and that inhibitors interfere with the protein traffic to nucleus (Puttonen et al., 2006). However, direct evidence for any of these processes are currently missing.

There is emerging evidence that other different gene products are responsible for some of the POP-like activities, as other peptidases with very similar structures, activities, and sensitivities to inhibitors have been detected, e.g. FAB or prolyl carboxypeptidase. A great deal of work has, and is being done to address questions about the physiological role of cytoplasmic POP and on the discovery of variants either from the same, or from alternative genes, to try to clarify the complex role of POP in peptide processing and metabolism.

PREPL is located on chromosome 2p21, and not in chromosome 2p16, as we review here and as was originally described by Parvari et al. (2001) but corrected recently by Jaeken et al. (2006) and by Martens et al. (2006). These recent studies report more detailed description of the chromosomal deletions that showed more prevalence than anticipated, and also established that the syndrome carrying those deletions is now called hypotonia-cystinuria syndrome. These studies also provide evidence to suggest that the syndrome is caused by absence of functional KIAA0436 (PREPL gene) among other genes.