Purification and Characterization of a Novel Neurotensin-degrading Peptidase from Rat Brain Synaptic Membranes*

A peptidase that cleaved neurotensin at the Pro'O- Tyrll peptide bond, leading to the formation of neuro-tensin-(1-10) and neurotensin-(11-13), was purified nearly to homogeneity from rat brain synaptic membranes. The enzyme appeared to be monomeric with a molecular weight of about 70,000-75,000 as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and high pressure liquid chromatography filtration. Isoelectrofocusing indicated a PI of 5.9-6. The purified peptidase could be classified as a neutral metallopeptidase with respect to its sensitivity to pH and metal chelators. Thiol-blocking agents and acidic and serine protease inhibitors had no effect. Studies with specific peptidase inhibitors clearly indicated that the purified enzyme was distinct from en- zymes capable of cleaving neurotensin at the Pro'O-Tyr" bond such as proline endopeptidase and endopep- tidase 24-11. The enzyme was also distinct from other neurotensin-degrading peptidases such as angiotensin- converting enzyme and a recently purified rat brain soluble metalloendopeptidase. The peptidase displayed a high affinity for neurotensin (K, = 2.6 MM). Studies on its specificity revealed that neurotensin-(9-13) was the shortest obtained with two distinct preparations of posthydroxylapatite activity. a-MSH, a-melanotropin-stimulating hormone; TRH, thyrotropin releasing hormone; LHRH, luteinizing hormone releasing hormone; VIP; vasoactive intestinal peptide; GIP, gastric inhibitory peptide.

A peptidase that cleaved neurotensin at the Pro'O-Tyrll peptide bond, leading to the formation of neurotensin-(1-10) and neurotensin-(11-13), was purified nearly to homogeneity from rat brain synaptic membranes. The enzyme appeared to be monomeric with a molecular weight of about 70,000-75,000 as determined by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and high pressure liquid chromatography filtration. Isoelectrofocusing indicated a PI of 5.9-6. The purified peptidase could be classified as a neutral metallopeptidase with respect to its sensitivity to pH and metal chelators. Thiol-blocking agents and acidic and serine protease inhibitors had no effect. Studies with specific peptidase inhibitors clearly indicated that the purified enzyme was distinct from enzymes capable of cleaving neurotensin at the Pro'O-Tyr" bond such as proline endopeptidase and endopeptidase 24-11. The enzyme was also distinct from other neurotensin-degrading peptidases such as angiotensinconverting enzyme and a recently purified rat brain soluble metalloendopeptidase. The peptidase displayed a high affinity for neurotensin ( K , = 2.6 MM). Studies on its specificity revealed that neurotensin-(9-13) was the shortest neurotensin partial sequence that was able to fully inhibit [3H]neurotensin degradation. Shortening the C-terminal end of the neurotensin molecule as well as substitutions in positions 8 , 9, and 11 by Damino acids strongly decreased the inhibitory potency of neurotensin. Among 20 natural peptides, only angiotensin I and the neurotensin-related peptides (xenopsin and neuromedin N) were found as potent as unlabeled neurotensin.
Neurotensin is a tridecapeptide (<Glu'-Leu2-Tyr3-Glu4-Asn5-Lys6-Pro7-Arp8-Arg9-Pro10-Tyr11-Ile12-Leu'3-OH, where <Glu is pyroglutamic acid) that fulfills several criteria suggesting a role of neurotransmitter or neuromodulator in the central nervous system (1,2). In particular, specific binding sites for neurotensin have been characterized in rat brain (3)(4)(5). In this context, it was of interest to investigate the expected mechanism by which neurotensin was readily inactivated after the interaction with its receptors in the brain. Several authors have shown that neurotensin was destroyed after exposure to brain tissues (6) or homogenates (7). We have recently proposed a model for the inactivation of neurotensin by highly purified rat brain synaptic membranes (8).
In this preparation, we have shown (8-10) that primary * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$ To whom all correspondence should be addressed.
inactivating cleavages occurred at the ArgS-Arg', Pro"-Tyr", and Tyr"-Ile'2 bonds, leading to the formation of the biologically inactive fragments neurotensin-(1-8), neurotensin-( 1lo), neurotensin-(1-11), neurotensin-(ll-13), and neuroten-sin49-13). We have clearly demonstrated that a recently purified rat brain soluble metalloendopeptidase (11) and endopeptidase 24-11 were totally responsible for the primary cleavages at the ArgS-Arg9 and Tyr"-Ile'* peptidyl bonds, respectively (8,9). Although endopeptidase 24-11 was also shown to participate to the cleavage at the Pro''-Tyr" bond, we established that the major contribution at this site was due to a peptidase not identified but clearly distinct from proline endopeptidase (10). Several data suggest that cleavage of the Pro''-Tyr" bond is a key event in the physiological process that leads to neurotensin inactivation. Thus, several authors have described the enhanced potency in the central nervous system, in vivo, of neurotensin analogues in which the tyrosyl residue in position 11 was substituted by a Damino acid (12)(13)(14). Furthermore, we demonstrated that these modified analogues were totally resistant to degradation to brain tissues in vitro and in vivo (15). Here we report on the purification and characterization from rat brain synaptic membranes of the as yet unidentified peptidase activity that cleaves neurotensin at the Pro''-Tyr" bond.
EXPERIMENTAL PROCEDURES'

RESULTS
Purification of the Peptidase- Table I shows that the purification procedure led to a 541-fold enrichment of the peptidase with an overall yield of 15%. Chromatography of the Triton X-100-solubilized material on DEAE-Trisacryl resin indicated that the peptidase was quantitatively retained on the column and eluted as a single peak of neurotensin-(1-10)generating activity by increasing the ionic strength ( Fig. 1). At this step, the pooled fractions kept for further purification were totally resolved from aminopeptidase M, proline endopeptidase, and post-proline dipeptidylaminopeptidase as illustrated in Fig. 1 by the profiles of L~u -~A M C -, Z-Gly-Pro-7AMC-, and Gly-Pro-7AMC-hydrolyzing2 activities, respect Portions of this paper (including "Experimental Procedures") are presented in miniprint at the end of this paper. Miniprint is easily read with the aid of a standard magnifying glass. Full size photocopies are available from the Journal of Biological Chemistry, 9650 Rockville Pike, Bethesda, MD 20814. Request Document No. 86M-604, cite the authors, and include a check or money order for $2.00 per set of photocopies. Full size photocopies are also included in the microfilm edition of the Journal that is available from Waverly Press.
tively. However, the pool was contaminated by a basic aminopeptidase (Arg-7AMC profile (21)) and a neurotensin-(1-8)-generating activity. The latter enzyme was identified as the recently purified "rat brain soluble metalloendopeptidase" (11) by the use of its specific inhibitor CPE-Ala-Ala-Phe-pAB (22). As shown in Fig. 2, the two contaminating enzymes were totally separated from the neurotensin-( 1-10)-generating activity after chromatography on hydroxylapatite. Fig. 3 illustrates the degradative pattern of neurotensin after incubation with an aliquot of the hydroxylapatite-pooled fractions. The only two degradation products detected after HPLC were neurotensin-(1-10) and neurotensin-(ll-13), which confirmed that the neurotensin-( 1-10)-generating activity was totally free of other contaminating neurotensin-degrading peptidases.

Molecular Weight, Purity, and Functional Properties-
HPLC filtration of the purified peptidase allowed us to determine a molecular weight of about 75,000 for the native enzyme ( Fig. 4). This value was in good agreement with that deduced from one-dimensional polyacrylamide gel electrophoresis (PAGE) in dissociating conditions that revealed a single band with an apparent M, of 72,000 (Fig. 5 ) . Fig. 6 shows a twodimensional SDS-PAGE analysis of a higher amount of posthydroxylapatite proteins. The pattern confirmed a major spot characterized by a PI of 5.9-6 and a M, of 72,000, but also revealed a few minor faint spots of lower M, and slightly different PI (Fig. 6).
Characterization of the Peptidase-The routine assay described under "Experimental Procedures" was used to further characterize the purified peptidase. The enzyme activity was maximal between pH 7 and 8 (not shown). An Eadie plot for the hydrolysis of [3H]neurotensin by the peptidase is presented in Fig. 7. This plot was linear as expected for a pure enzyme and gave a K,,, value of 2.6 p~.
The effect of various inhibitors is summarized in Table 11. The peptidase was very sensitive to metal chelators and was fully inhibited by 1 mM o-phenanthroline. The serine protease inhibitors diisopropyl fluorophosphate, phenylmethylsulfonyl fluoride, and benzamidine were ineffective as well as the thiolblocking agent iodoacetamide and the acidic protease inhibitor pepstatin. Furthermore, @-mercaptoethanol had no effect on [3H]neurotensin-degrading activity. Several specific inhibitors of purified peptidases were tested against the enzyme. Captopril, the potent angiotensin-converting enzyme inhibitor (23), and thiorphan, the endopeptidase 24-11 inhibitor (24), were found totally inactive at micromolar concentrations. Furthermore, the activity was totally insensitive to 10 PM of the aminopeptidase inhibitor bestatin (25) and was not ylapatite. An aliquot of posthydroxylapatite pool was concentrated on ultrafiltration cell, injected onto two connected columns, and eluted as described under "Experimental Procedures." All fractions were assayed for neurotensin degradation. Total (VT) and void (V,) volumes were determined by injecting 10 nmol of neurotensin-(11-13) and dextran blue, respectively. Molecular weight of the native enzyme was deduced from a standard curve established from the retention times of various markers of known molecular weights.

5
at the Ar$-Ar? peptide bond produced a 30% inhibition at 10 FM, a concentration 10-fold higher than the Kr for its enzyme.
In order to assess the specificity of the peptidase for neurotensin, a series of analogues with structural or conformational modifications were tested as competitors of [3H]neurotensin degradation. Table IV indicates that tyrosine in position 11 could be replaced by an aromatic residue (Trp, Phe) without alteration of the IC,, (compare 1, 11, and 12), whereas substitution by the corresponding D-amino acid strongly affected the inhibitory potencies (compare analogues 1, 11, and 12 to 13-16). Investigations at the sites 8 and 9 clearly indicated that the charge of the amino acid was not an important requirement for the expression of the activity  [3H]Neurotensin was incubated as described under "Experimental Procedures" with 90 ng of peptidase, for 30 min at 37 "C, in the absence (control) or in the presence of 7 increasing concentrations of various neurotensin partial sequences. Activity was monitored by the two-step chromatography on SPC25 Sephadex (see "Experimental Procedures"). ICs0 values correspond to the concentration of peptide that half-inhibited control [3H]neurotensin degradation. Data are the mean values of two or three separate determinations obtained with two distinct preparations of posthydroxylapatite activity.
No. Sequence   1 neurotensin-related peptides, neuromedin N (27) and xenopsin (28), were the only natural peptides with full inhibitory potency. The chicken neurotensin variant LysS-Amg-neurotensin-(8-13) (29) and substance P were, respectively, 4-and 10-fold less potent than neurotensin. HPLC of the peptides 2-6 after prolonged exposure to the enzyme clearly showed a marked decrease of intact peptide and, therefore, indicated that they were substrates of the peptidase (not shown). The sites of cleavages of the neurotensin-related peptides were deduced from the retention times of the degradation products after HPLC separation. Table VI indicates that hydrolysis by the peptidase always occurred at the Pro-Tyr (neurotensin, neuromedin N, and Lyss-Amg-neurotensin-(8-13)), or Pro-Trp (xenopsin) peptidyl bonds. The other natural peptides (7-20) displayed very little, if any, ability to inhibit [3H] neurotensin degradation. None of them was found to behave as a substrate of the enzyme (not shown).

DISCUSSION
The present paper reports on the purification and characterization from rat brain synaptic membranes of a neurotensin-degrading metallopeptidase that cleaves neurotensin a t the Pro''-Tyr" peptidyl bond, yielding the biologically inactive fragments neurotensin-( 1-10) and neurotensin-( 11-13).
The purification procedure leads to a 541-fold enrichment of the peptidase with a yield of 15% and affords a nearly homogeneous protein as judged by one-and two-dimensional SDS-PAGE. The apparent molecular weight of the native enzyme as determined by HPLC filtration is 75,000. This value is in close agreement with that deduced from SDS-PAGE analysis (72,000). These results indicate that the peptidase is a monomer. In order to facilitate the characterization of the purified peptidase, it was necessary to set up a suitable routine assay

TABLE V Effect of natural peptides on PHIneurotensin degradation
[3H]Neurotensin was incubated as described under "Experimental Procedures" with 90 ng of peptidase, for 30 min at 37 "C, in the absence (control) or in the presence of 7-9 concentrations of various natural peptides. Activity was monitored by the two-step chromatography on SPC25 Sephadex (see "Experimental Procedures"). ICSO values correspond to the concentration of peptide that half-inhibited control (3H]neurotensin degradation. Data are the mean values of two or three determinations obtained with two distinct preparations of posthydroxylapatite activity. a-MSH, a-melanotropin-stimulating hormone; TRH, thyrotropin releasing hormone; LHRH, luteinizing hormone releasing hormone; VIP; vasoactive intestinal peptide; GIP, gastric inhibitory peptide.

Hydrolysis of neurotensin-related peptides by the purified metallopeptidase
Peptides (2 nmol) were incubated with the peptidase (90 ng) for 2% h at 37 "C, and incubations were then submitted to HPLC as described under "Experimental Procedures." In all cases, these prolonged incubations led to onlv two degradation Droducts that were identified by their retention times. <Glu, pyroglutamic acid. of neurotensin degradation. Tritiated neurotensin was used as substrate. We showed previously (9) that radioactivity was equally distributed on both tyrosyl residues (positions 3 and 11 of the neurotensin molecule). Therefore, the degradation products generated by the purified peptidase, i.e. neurotensin-(1-10) and neurotensin-(11-13), contain equal amounts of radioactivity. Based on the fact that neurotensin-(11-13) has no net positive charge whereas neurotensin-(1-10) and neurotensin are positively charged at neutral pH, it was possible to separate [3H]neurotensin-(ll-13) from [3H]neurotensin-(1-10) and [3H]neurotensin. Using this procedure, several properties of the purified peptidase were investigated. The enzyme is optimally active in the pH range between 7 and 8 and is readily inhibited by metal chelators such as EDTA and o-phenanthroline. By contrast, the peptidase is totally insensitive to thiol-blocking agents, acidic and serine protease inhibitors (Table 11), and can be classified as a neutral metallopeptidase.

11279
The peptidase displays a rather high affinity for neurotensin (K, = 2.6 WM, Fig. 7). From the data obtained with a series of neurotensin partial sequences, it is clear that the pentapeptide neurotensin-(9-13) is the shortest fragment with full [3H]neurotensin degradation inhibitory potency (Table 111). Shortening the C-terminal end of the neurotensin molecule (neurotensin-(l-12), neurotensin-(1-11), neurotensin-( 1-10), and neurotensin-(l-8)) gradually reduces the potency of neurotensin (Table 111). This may indicate that a Cterminal extension of at least three amino acids from the scissile Pro''-Tyr" bond is essential to maintain the sequence fully active. If this requirement is respected, the peptidase is not sensitive to the replacement of the amino acids in positions 12 (Ile) and 13 (Leu) by less hydrophobic residues (compounds 17 and 19, Table IV) as well as substitution of Tyr" by aromatic amino acids (Phe and Trp, see 11 and 12, Table IV). By contrast, the peptidase exhibits a strong stereospecificity toward Tyr" since all the analogues substituted in this position by a D-amino acid display very poor abilities to inhibit [3H]neurotensin degradation. Examination of positions 8 and 9 of the neurotensin molecule shows that the arginyl residues can be replaced by their uncharged analogues citrullines, indicating that the positive charges do not influence the apparent affinity of the peptidase for the analogue.  (Table IV).
The study concerning the specificity of the peptidase toward the amino acids of the C-terminal neurotensin hexapeptide have revealed a high specificity of the peptidase for neurotensin, illustrated by both conformational and structural requirements. This specificity is further supported by the fact that among 20 natural peptides tested as inhibitors of [3H]neurotensin degradation, only three were found as potent as neurotensin ( Table V). Two of them are the neurotensin-related peptides, neuromedin N and xenopsin. These peptides are cleaved at Pro-Tyr and Pro-Trp bonds, respectively, and respect the structural requirements discussed above (see sequences, Table VI). Angiotensin I is also as potent as neurotensin. Although the site of cleavage has not been established yet it is clear that its C-terminal end resembles that of neurotensin, especially the C-terminal tetrapeptide Pro-Phe-His-Leu. Finally, the chicken neurotensin variant is also cleaved at a Pro-Tyr peptidyl bond. If it seems clear that the apparent primary specificity of the peptidase is directed toward a Pro-X sequence (where X represents an aromatic amino acid), other results indicate that this peptidase cannot be classified as a general postproline-cleaving enzyme. Indeed, many proline-containing peptides such as thyrotropin releasing hormone, luteinizing hormone releasing hormone, oxytocin, [Ar$]vasopressin, a-melanotropin-stimulating hormone, physalaemin, and eledoisin are not substrates of the peptidase. Furthermore, in the neurotensin molecule itself, the cleavage at the Pro7-Ar$ peptidyl bond is not catalyzed by the peptidase.
Endopeptidase 24-11 (enkephalinase) and proline endopeptidase have been previously shown to hydrolyze neurotensin at the Pro''-Tyr" site. However, the peptidase described in the present study can be easily distinguished from these enzymes. Indeed,Almenoff et al. (30) have shown that purified endopeptidase 24-11 not only cleaves neurotensin at the Pro''-Tyr'l but also at the Tyr11-Ile'2 sites. Furthermore, the present study shows that the purified peptidase is insensitive to M thiorphan, a specific inhibitor displaying a high affinity (KI = 4.10-' M ) for endopeptidase 24-11 (24). Finally, Metenkephalin is not substrate for the neurotensin-degrading metallopeptidase. Proline endopeptidase was described as a serine protease hydrolyzing several biologically active peptides including neurotensin at prolyl residues (31,32). These properties do not support the hypothesis of an identity with the present peptidase that is unable to cleave neurotensin at the Pro7-Ar$ site and is not sensitive to serine protease inhibitors (diisopropyl fluorophosphate, phenylmethylsulfonyl fluoride, and benzamidine). Furthermore, a specific proline endopeptidase inhibitor (Z-Pro-Prolinal) was recently developed (22). This inhibitor does not affect the purified peptidase at M, a concentration 100-fold higher than the KI for proline endopeptidase.
Two other proline-directed peptidases have been purified. Prolylcarboxypeptidase (angiotensinase C) (EC 3.4.16.2) differs drastically from the present peptidase by its physical properties since it has been described as a lysosomal enzyme (optimal pH between 4.5 and 5.5) of higher molecular weight (115,000) and dissociated in two subunits under denaturing conditions (33,34). Postproline dipeptidyl aminopeptidase (dipeptidyl peptidase IV) (EC 3.4.14.5) has a molecular weight in the neighborhood of 200,000, and a free N-terminal amino acid is a crucial requirement for the expression of the activity (35).
Finally, three other peptidases have been isolated from mammalian brains or pituitaries and were shown to cleave neurotensin. However, their sites of cleavage on the neurotensin molecule together with their sensitivity to inhibitors make easy their distinction from the present enzyme. Indeed angiotensin-converting enzyme releases the C-terminal dipeptide of neurotensin (cleavage at the Tyr11-Ile'2 bond) (36) and is totally inhibited by a concentration of M of its specific inhibitor captopril (23), a dose totally ineffective on the neurotensin-degrading metallopeptidase. A rat brain soluble metalloendopeptidase was shown to hydrolyze neurotensin at the Ar$-Argg bond leading to the formation of neurotensin-(1-8) and neurotensin-(9-13) (11). A recently developed inhibitor of this peptidase (22) displays a weak inhibitory effect (30% of inhibition a t a concentration of M ) on the present enzyme. It is noteworthy that such a concentration totally inhibits the neurotensin-(1-8)-and neurotensin-(9-13)-generating activity detected in synaptic membranes and resolved from the neurotensin-degrading metallopeptidase by the DEAE and hydroxylapatite chromatographies (Figs. 1 and 2). Finally, a multicatalytic protease (37) called cation-sensitive neutral endopeptidase was shown to rapidly cleave neurotensin at the G1u4-Asn5, Asn5-Lys6, and Ile'*-Le~'~ and to a lesser extent at the Pro"-Tyr" peptide bonds. The enzyme also hydrolyzes some other peptides such as luteinizing hormone releasing hormone (38) that was not found to behave as a substrate for the purified peptidase described here (Table V).
With respect to its original properties including both specificity and sensitivity to inhibitors, it seems reasonable to assume that we have isolated and characterized a novel rat brain synaptic peptidase that could be named neurotensindegrading neutral metallopeptidase. The physiological importance of this enzyme in the inactivation of neurotensin remains to be established. However, it is of interest to underline the fact that the neurotensin analogues modified in position 11 by substitution with a D-amino acid ( [D-Tyr"]neurotensin, [D-Phe"]neurotensin) are resistant to degradation by the peptidase as well as the fact that they totally resist degradation in vivo, after stereotaxic intracerebroventricular injection in the rat (15). The fact that these analogues were found to be more potent than neurotensin in eliciting central effect following intracerebroventricular injection in the rat (12-14), although they are very poor agonists of neurotensin receptors in vitro (39), supports the hypothesis of a physiological neurotensin-inactivating mechanism involving the Pro''-Tyr" peptidyl bond. These data should provide a basis for a strategy consisting of the development of both neurotensin analogues resistant to degradation and inhibitors of the neurotensindegrading neutral metallopeptidase in order to assess its physiological contribution in the inactivation of neurotensin in the rat central nervous system.