Neprilysin degrades both amyloid beta peptides 1-40 and 1-42 most rapidly and efficiently among thiorphan- and phosphoramidon-sensitive endopeptidases.

To identify the amyloid beta peptide (Abeta) 1-42-degrading enzyme whose activity is inhibited by thiorphan and phosphoramidon in vivo, we searched for neprilysin (NEP) homologues and cloned neprilysin-like peptidase (NEPLP) alpha, NEPLP beta, and NEPLP gamma cDNAs. We expressed NEP, phosphate-regulating gene with homologies to endopeptidases on the X chromosome (PEX), NEPLPs, and damage-induced neuronal endopeptidase (DINE) in 293 cells as 95- to 125-kDa proteins and found that the enzymatic activities of PEX, NEPLP alpha, and NEPLP beta, as well as those of NEP and DINE, were sensitive to thiorphan and phosphoramidon. Among the peptidases tested, NEP degraded both synthetic and cell-secreted Abeta1-40 and Abeta1-42 most rapidly and efficiently. PEX degraded cold Abeta1-40 and NEPLP alpha degraded both cold Abeta1-40 and Abeta1-42, although the rates and the extents of the digestion were slower and less efficient than those exhibited by NEP. These data suggest that, among the endopeptidases whose activities are sensitive to thiorphan and phosphoramidon, NEP is the most potent Abeta-degrading enzyme in vivo. Therefore, manipulating the activity of NEP would be a useful approach in regulating Abeta levels in the brain.

Alzheimer's disease (AD) 1 is characterized by the accumula-tion of amyloid ␤ peptide (A␤) in the brain. A␤ is composed of 39 -43 amino acids and is constitutively produced by proteolysis of the ␤-amyloid precursor protein (APP). Alterations in either synthesis or clearance of A␤ may potentially contribute to increased levels of A␤ and amyloid deposits. Although much attention has been focused on the production of A␤, little is known about how A␤ is degraded and cleared, especially in the brain. Identification of the peptidases involved in A␤ catabolism in vivo is important for the development of therapeutics designed to prevent or treat AD.
Proteases, including cathepsin D (1), serine protease-␣ 2 -macroglobulin complex (2) and insulin-degrading enzyme (3), were purified and identified as A␤-degrading enzymes in vitro. Several other recombinant or purified peptidases have also been shown to degrade A␤ in vitro (4 -14). We focused on the in vivo catabolism of A␤42, because this specific form, rather than A␤40, is considered to be the primary pathogenic agent in AD. Recently, we demonstrated that an endopeptidase(s) similar or identical to neprilysin (NEP), whose activity is sensitive to thiorphan and phosphoramidon, is involved in the catabolism of A␤1-42 in vivo (15,16).
NEP is a type II membrane protein on the cell surface and is classified as a member of the M13 family. NEP hydrolyzes and inactivates several circulating peptides, such as enkephalin, atrial natriuretic peptide, endothelin, and substance P, and has wide tissue distribution and substrate specificity (17). The M13 family comprises six zinc-dependent metalloproteases, NEP, endothelin-converting enzyme (ECE-1) (18), ECE-2 (19), KELL antigen (20), phosphate regulating gene with homologies to endopeptidases on the X chromosome (PEX) (21), and the recently identified damage-induced neuronal endopeptidase (DINE)/X-converting enzyme (XCE) (22,23). Among these, only NEP and DINE (22) have been shown to be sensitive to both thiorphan and phosphoramidon. Because ECE-1 and ECE-2 are not inhibited by thiorphan (18,19), they can be excluded as candidates for A␤1-42-degrading enzymes in vivo. KELL is only partially inhibited by phosphoramidon (24) and is mainly present in erythroid tissues. Although PEX degrades parathyroid hormone-derived peptides (25), its sensitivity to thiorphan or phosphoramidon is unknown. These facts indicate that NEP, DINE, and PEX are candidates for A␤1-42-degrading enzymes in vivo. We also need to consider the possible presence of an unidentified protease(s), whose activity is sensitive to thiorphan and phosphoramidon and responsible for A␤1-42 degradation.
In this study, considering the redundancy of peptidases, we investigated the presence of unidentified peptidase(s), which is * This work was supported in part by research grants from RIKEN Brain Science Institute, Special Coordination Funds for promoting Science and Technology of Science and Technology Agency, Ministry of Health and Welfare, Ministry of Education, and Takeda Chemical Industries. 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AF302075, AF302076, and AF302077.
A similar procedure was performed to obtain cDNA homologous to NEP and DINE/XCE. Briefly, two oligonucleotides, 5Ј-AA ( (ENIADNG) of mouse NEP, respectively, were prepared, and PCR was performed under the same conditions described above. Products ϳ300 bp in length were sequenced.
Human NEP and PEX cDNAs were amplified from human hippocampus Marathon Ready cDNA (CLONTECH). Rat DINE cDNA was obtained as described previously (22). FLAG sequences (MDYKDDDDK) were introduced at the C-terminal end of each cDNA by PCR using appropriate primers containing FLAG nucleotides. All the cDNAs were subcloned into the mammalian expression vector pcDNA3.1 (ϩ) (Invitrogen).
Cell Culture and Transfection-293 human embryonic kidney cell lines were cultured in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% fetal bovine serum and 1% penicillin/streptomycin (Life Technologies, Inc.). The 293 cells were transfected with either an empty vector, NEP, PEX, NEPLP ␣, NEPLP ␤, NEPLP ␥, or DINE cDNAs using a calcium phosphate method. Stable 293 cell lines were obtained by selection with 600 g/ml G418. Stable mouse neuroblastoma N2a cells expressing both human APP695 with the Swedish mutation and presenilin 2 (PS2) with the N141I mutation were obtained by selection with 150 g/ml hygromycin and 500 g/ml G418, respectively.
Antibodies-The rabbit polyclonal anti-NEPLPC (named as 065P and M97P) and anti-PEXC (M96G) antibodies were raised against a synthetic peptide, CCPRGSPMHPMKRCRIW corresponding to the Cterminal amino acid residues of the NEPLP, glutathione S-transferase fusion protein encompassing residues 534 -646 of the NEPLP ␣, and glutathione S-transferase fusion protein encompassing residues 700 -749 of human PEX, respectively. 56C6 (Novocastra Laboratory, Tyne, UK) and M2 (Stratagene, La Jolla, CA) are mouse monoclonal antibodies against the extracellular domain of human NEP and FLAG, respectively. DINE antibody was characterized previously (22).
Western Blot Analysis-Cells were homogenized in solution A (0.1 M Tris-HCl, pH 8.0, 0.15 M NaCl, 1 g/ml leupeptin, and 1 g/ml pepstatin A) and centrifuged at 500 ϫ g for 5 min. Membranes were prepared by precipitation of the postnuclear supernatant at 100,000 ϫ g for 30 min. The supernatant was used as a cytosolic fraction. The resulting pellet was lysed in solution A containing 1% Triton X-100 for 1 h at 4°C and centrifuged at 100,000 ϫ g for 15 min. The supernatant was used as a membrane fraction. Aliquots of the membrane fraction (3 g) or 10 l of culture medium (CM) were separated on 7.5% SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membrane. The blotted membrane was blocked with 5% skim milk in a buffer containing 10 mM Tris-HCl, pH 8.0, and 0.15 M NaCl and sequentially incubated with the primary antibody in the buffer with 0.05% Tween 20 and with horseradish peroxidase-goat anti-rabbit or anti-mouse IgG (Amersham Pharmacia Biotech, Buckinghamshire, UK) and visualized using an ECLplus kit (Amersham Pharmacia Biotech) according to the manufacturer's directions. Cell lysates were also prepared using the solution A containing 1% Triton X-100.
Peptidase Assay-Membrane fractions (0.5 g) from the 293 cells transfected with the various cDNAs were incubated with 50 M Z-Ala-Ala-Leu-p-nitroanilide (ZAAL-pNA) (Peptide Institute, Osaka, Japan) in 100 l of 100 mM MES (pH 6.5) or 50 mM HEPES (pH 7.2) for 1 h at 37°C. The reaction mixture was added with 0.4 milliunit of leucine aminopeptidase (Sigma Chemical Co., St. Louis, MO), further incubated for 20 min at 37°C and measured at 405-nm absorption. For the inhibition study the membrane fractions were preincubated with 10 M thiorphan or 10 M phosphoramidon for 5 min before the addition of ZAAL-pNA.
Digestion Assay of Cold A␤1-40 and A␤1-42-A␤ (Bachem, Torrance, CA) was dissolved in dimethyl sulfoxide at 5 mg/ml as a stock solution. 2 g of A␤1-40 (4.6 M) or 5 g of A␤1-42 (11 M) was incubated with 10 g of the membrane fractions in 100 l of 50 mM HEPES (pH 7.2) containing 1 g/ml leupeptin and 1 g/ml pepstatin A at 37°C for the times indicated. The reaction was stopped by adding 200 l of buffer A (0.1% trifluoroacetic acid), and the solution was injected into a reverse-phase HPLC (CAPCELL PAK C18 UG 120, particle size 5 m, column dimensions 4.6 ϫ 250 mm) (Shiseido, Tokyo, Japan) equilibrated with the buffer A. The HPLC protocol used to analyze 3 H/ 14 C-radiolabeled A␤1-42 was not suitable for analyzing cold A␤ peptides, because it produced high background noise when analyzed at the absorbance of 210 nm. A␤ and its proteolytic products were eluted with a linear solvent gradient of 0 -70% buffer B (0.1% trifluoroacetic acid in acetonitrile) over 10 -40 min at a flow rate of 0.5 ml/min at 50°C. Peptides were detected by absorbance at 210 nm. 10 M thiorphan was preincubated for the inhibition.
Mass Spectrometric Analysis and Protein Sequencing-HPLC peaks in digestion assays of cold A␤ peptides were collected, dried under vacuum, and analyzed by a mass spectrometer and a protein sequencer as described previously (15).
Kinetic Analysis-Various amounts of cold A␤ peptides were incubated with membrane fractions of pcDNA-transfected cells or NEPtransfected cells, and the remaining peaks of the A␤ peptides on the HPLC chart were used to determine the initial A␤ concentrations (S 0 ) or A␤ concentrations at time t (S t ), respectively. Note that membrane fractions of pcDNA-transfected cells showed no detectable degradation to A␤. K m and V max were determined according to the following first degree equation (26), Digestion Assay of Secreted A␤ from N2a Cells-CM from the N2a cells in the absence of fetal bovine serum was collected, diluted 5-fold with medium, and used as a substrate. The CM was incubated with 4 g of membrane fractions in 100 l of 50 mM HEPES (pH 7.2) at 37°C for 16 h. The reaction was stopped by boiling and the samples were subjected to BAN50/BA27 and BAN50/BC05 sandwich ELISAs to quantify A␤1-40 and A␤1-42, respectively (27).

RESULTS
cDNA Cloning of NEPLPs-To clone peptidases homologous to NEP and PEX, we designed degenerate oligonucleotide primers. With mouse brain cDNA as a template, the PCR resulted in the amplification of DNA with the predicted size (150 bp). After subcloning the PCR products and sequencing individual clones, we found that 11 of the 42 clones had a novel identical sequence that was similar to but distinct from members of the M13 family. The other 16 clones were identical to ECE-1, 11 clones were NEP, 2 clones were ECE-2, and 2 clones were PEX. During sequencing the full-length of the novel gene, we noted the presence of three isoforms, which probably resulted from alternative splicing. We termed them NEPLP ␣, NEPLP ␤, and NEPLP ␥, and their structural characteristics are outlined in Fig. 1. Of 42 isolated full-length clones, 21 were NEPLP ␣, 20 were NEPLP ␤, and 1 was NEPLP ␥. Compared with NEPLP ␣, NEPLP ␤ and NEPLP ␥ had an insertion of 23 amino acids immediately following the putative transmembrane region and 37 amino acids near the center of the protein, respectively. All the isoforms were predicted to be type II transmembrane proteins with a putative zinc-binding domain (HEXXH) in the extracellular portion. NEPLPs showed the highest homology (ϳ54%) to NEP among the members of the M13 family. The degenerate primers, which were designed to amplify the peptidases homologous to NEP and DINE/XCE, did not amplify a novel gene. Recently, three research groups independently reported a novel member of the M13 family (28 -30). NEPLP ␣ and NEPLP ␤ were identical to SEP ⌬ /splice 1 and SEP/NL1/ NEPII, respectively. NEPLP ␥ was a novel isoform.
Expression of NEPLPs, NEP, PEX, and DINE in the 293 Cell Line-We stably transfected NEPLP cDNAs in 293 cells and determined the expression using the 065P antibody, which was raised against the C-terminal portion of NEPLP. In the 293 cells transfected with pcDNA vector alone (referred to as pcDNA cells), no endogenous NEPLP protein was detected ( Fig.  2A, lane 5). In membrane fractions of 293 cells stably transfected with NEPLP ␣ (NEPLP ␣ cells), 100-and 120-kDa broad bands were detected ( Fig. 2A, lane 6). Because NEPLP ␣ was expressed as a single 100-kDa band in COS cells (not shown) and Chinese hamster ovary cells (28), the 120-kDa band probably represented a highly glycosylated or modified protein specific for the 293 cells. NEPLP ␤ and NEPLP ␥ proteins were expressed in membrane fractions as 110-and 115-kDa bands, respectively ( Fig. 2A, lanes 7 and 8). This corresponded to the increasing sizes of amino acids, although proteins with higher molecular mass did not accumulate as seen in NEPLP ␣ cells. The amounts of NEPLP ␤ and NEPLP ␥ proteins were lower than that of NEPLP ␣. NEPLP ␣, NEPLP ␤, and NEPLP ␥ were not detected in the cytosolic fractions (not shown). In the cul-ture medium, we could detect only NEPLP ␤ (Fig. 2C, lanes 13) with a molecular mass of 125 kDa, which was larger than that of membrane-associated NEPLP ␤. This result suggested that the 23-amino acid sequences inserted in NEPLP ␤ possessed a secretion signal and that NEPLP ␤ is much more glycosylated or modified in the secretory pathway than in the membrane. Indeed Ikeda et al. (28) reported that the molecular mass of membrane-associated and secreted NEPLP ␤/SEP were reduced to the same mass (89 kDa) by peptide-N-glycanase F treatment.
When we expressed other members of the M13 family (NEP, PEX, and DINE) in the 293 cells, these proteins were detected as 110-, 110-, and 95-kDa bands in membrane fractions ( Fig.  2A, lanes 2, 4, and 10), whereas they were not detected in the pcDNA cells ( Fig. 2A, lanes 1, 3, and 9). NEP, PEX, and DINE were not detected in cytosolic fractions or CM (not shown). To estimate the relative amounts of the peptidases, we expressed each cDNA to which FLAG sequence was added at its C-terminal end. Because nearly identical intensities of bands were detected with the anti-FLAG antibody (Fig. 2D), equal amounts of peptidases with FLAG were loaded onto the gels. Fig. 2E shows the results of loading the same amounts of FLAG-tagged peptidases as those in Fig. 2D. The gel shown in Fig. 2E was also loaded with 0.5 g of lysates from NEP, PEX, and NEPLP ␣ cells and 1.5 g from DINE cells and was then stained with  1 and 2), M96G (lanes 3 and 4), and 065P (lanes 5-8). B, membrane fractions of 293 cells transiently transfected with pcDNA and DINE (lanes 9 and 10, respectively) were subjected to Western analysis using the DINE antibody. C, culture medium (CM) of 293 cells stably transfected with pcDNA, NEPLP ␣, NEPLP ␤, and NEPLP ␥ were subjected to Western analysis using 065P (lanes 11-14). The 66-kDa bands indicated with an asterisk may be nonspecific staining of albumin derived from the serum. D and E, estimation of relative amounts of each protein expressed in 293 cells. Cell lysates of 293 cells transiently transfected with cDNAs tagged with FLAG were subjected to Western analysis using the anti-FLAG antibody M2 (D). The amounts of lysates transfected with NEP, PEX, NEPLP ␣, and DINE without FLAG were 0.5, 0.5, 0.5, and 1.5 g, respectively (E). Each protein was blotted with the same antibodies as in A and B with the exception of NEPLP ␣. NEPLP ␣ was blotted with the M97P antibody. Note that almost equal amounts of peptidases were loaded. the corresponding antibodies. Almost identical band intensities resulted regardless of whether or not the peptidases contained FLAG. We estimated from the results that the relative amounts of peptidases were 1:1:1:1/3 for NEP:PEX:NEPLP ␣:DINE. The addition of FLAG to PEX and NEPLP ␣ reduced their molecular weight compared with their weights in the absence of FLAG (Fig. 2E), suggesting immature glycosylation or modification resulting from the FLAG sequence at the C terminus. We also observed reduced proteolytic activity of NEP-FLAG than that of NEP (not shown). Therefore, we used each peptidase without FLAG in the following proteolytic analyses.
Proteolytic Activity to Synthetic Peptide (ZAAL-pNA)-We investigated the peptidase activity of NEPLPs, and their sensitivities to thiorphan and phosphoramidon. Membrane fractions of NEPLPs and other peptidases expressed in the 293 cells were incubated with ZAAL-pNA at pH 6.5 (Fig. 3, A and  B). Membrane fractions of pcDNA cells showed a very low level of thiorphan-and phosphoramidon-sensitive activity due to the presence of an endogenous enzyme. A high degree of proteolytic activity was detected in the membrane fractions of the NEP cells (Fig. 3, A and B). PEX showed a low but significant proteolytic activity compared with that of the control (pcDNA) (Fig. 3A). Both NEPLP ␣ and NEPLP ␤ proteolyzed the peptide, and their activities were almost the same and one-third compared with that of NEP, respectively. In contrast, NEPLP ␥ had no proteolytic activity compared with the control, although the level of the NEPLP ␥ protein was similar to that of NEPLP ␤ (Fig. 2A). This suggested that NEPLP ␥ protein has different substrate specificity or that it is a zymogen that requires pro-teolytic activation. It is also possible that the 37-amino acid sequence inserted into NEPLP ␥ inhibits its activity by interfering with proper folding. DINE had no proteolytic activity in this system (Fig. 3B), although its endopeptidase activity and sensitivity to thiorphan and phosphoramidon were demonstrated previously in a baculovirus expression system (22). Proteolytic activities of NEP, PEX, NEPLP ␣, and NEPLP ␤ were inhibited by thiorphan and phosphoramidon (Fig. 3A), suggesting that they could all be candidates for A␤-degrading enzymes in vivo. The secreted form of NEPLP ␤ also had the proteolytic activity (not shown). We obtained identical results at pH 7.2, too.
Degradation of Radiolabeled A␤1-42-We characterized the A␤-degrading activity of the peptidases using 3 H/ 14 C-radiolabeled A␤1-42, which we had synthesized previously (15). To exclude the effects of endogenous peptidases, equal amounts of membrane fractions (0.5 g) were analyzed for their ability to proteolyze A␤. Membrane fractions of the pcDNA cells exhibited almost no proteolytic activity to the radiolabeled A␤1-42 (peak at 42 min in Fig. 4A, closed triangle) in 1 h at 37°C. NEP decreased the radioactive peak at 42 min and increased the peaks at 7 and 37 min, suggesting that NEP efficiently proteolyzed the A␤1-42 into free amino acids/small peptides and a catabolic intermediate (Fig. 4B, open triangles). This degradation by NEP was very similar to the in vivo proteolysis as described previously (15). The degradation was completely inhibited by thiorphan, suggesting that NEP directly proteolyzed the peptide (Fig. 4H). In 4 h, NEP almost completely digested the A␤1-42 (not shown). In contrast PEX, NEPLP ␣, NEPLP ␤, NEPLP ␥, and DINE did not degrade the A␤1-42, because their HPLC profiles were almost the same as that of the control (pcDNA) (Fig. 4, C-G). HPLC profiles in the 3 H mode were essentially identical to the 14 C mode (not shown), as reported previously (15), suggesting that NEP proteolyzed the A␤1-42 from both the N and C termini.
Degradation of Cold A␤1-40 and A␤1-42-We examined the degradation of cold A␤1-40 and A␤1-42 using HPLC under different conditions from those used for analyzing the degradation of 3 H/ 14 C-radiolabeled A␤1-42 (see "Experimental Procedures"). Membrane fractions of the pcDNA cells were used as a negative control (Fig. 5A, panel 1). NEP almost completely degraded the A␤1-40 peptide corresponding to the peak at 35.7 min (Fig. 5A, panel 2) in 2 h at 37°C. The degradation was accompanied by the appearance of multiple peptides between 23 and 28 min (Fig. 5A, panel 2, open triangles). NEPLP ␣ showed relatively low but significant proteolytic activity to cold A␤1-40 as discerned by the decrease of the peak at 35.7 min and the appearance of new peaks between 26 and 29 min (Fig.  5A, panel 4). Thiorphan (Fig. 5A, panels 8 and 9) and phosphoramidon (not shown) completely abolished the proteolytic activity of NEP and NEPLP ␣. PEX, NEPLP ␤, NEPLP ␥, DINE (Fig. 5A, panels 3, 5-7) and secreted form of NEPLP ␤ (not shown) showed no proteolytic activity to A␤1-40. Although we assayed 3-fold amounts of membrane fractions of NEPLP ␤, NEPLP ␥, and DINE cells to normalize the amounts of pepti-   dases, they showed no detectable proteolytic activity to cold A␤1-40 in 2 h (not shown). With a longer incubation time (Fig.  5C, 6 -12 h), NEP degraded much larger amounts of A␤1-40, although NEPLP ␣ did not further degrade A␤1-40. PEX showed very low proteolytic activity, and NEPLP ␤, NEPLP ␥, and DINE showed no detectable proteolytic activity to A␤1-40 in 6 -12 h (Fig. 5C). When cold A␤1-42 was used as a substrate, NEP degraded it in a reproducible manner as demonstrated by the decrease of the original peak at 36.4 min and the appearance of a new peak at 26.4 min (Fig. 5B, panel 2) compared with the control (Fig.  5B, panel 1) in 4 h at 37°C. In contrast, PEX, NEPLP ␣, NEPLP ␤, NEPLP ␥, and DINE (Fig. 5B, panels 3-7), and the secreted form of NEPLP ␤ (not shown) showed almost no proteolytic activity. Degradation of A␤1-42 by NEP was completely inhibited by thiorphan (Fig. 5B, panel 8) and phosphoramidon (not shown). Again, 3-fold amounts of membrane fractions of NEPLP ␤, NEPLP ␥, and DINE cells showed no detectable proteolytic activity to cold A␤1-42 in 4 h (not shown). NEP degraded greater amounts of A␤1-42 with longer incubation times (Fig. 5D, 8 -12 h). Among the other peptidases, only NEPLP ␣ degraded very small amounts of A␤1-42 in 8 -12 h (Fig. 5D). N-terminal amino acid sequence and the molecular mass of the peak at 26.4 min produced by NEP (Fig.  5B, panel 2) were DAE---and 1033.92, suggesting that the peak contained A␤1-9 (calculated mass MH ϩ ϭ 1034.02). This result is consistent with the previous study, which indicated that 3 H/ 14 C-radiolabeled A␤1-42 was cleaved at G 9 /Y 10 by thiorphan-and phosphoramidon-sensitive peptidase(s) in vivo (15). The other degrading products by NEP could not be identified probably due to further degradation.
Kinetic Analysis of A␤ Proteolysis by NEP-We determined K m and V max values for the proteolysis by NEP using varying amounts of cold A␤ peptides (Table I). The K m values were 11.2 and 6.95 M and V max values were 158 and 21.1 nM/min for A␤1-40 and A␤1-42, respectively. It is difficult to determine the K m and V max values for the other peptidases, presumably because the K m values are too large and/or the V max values are too small to measure.
Degradation of Cell-secreted A␤1-40 and A␤1-42-We investigated the degradation of cell-secreted A␤ by the peptidases. We chose the N2a cells as A␤-producing cells because they were stably co-transfected with APP with the Swedish mutation and PS2 with the N141I mutation and secreted large amounts of both A␤1-40 and A␤1-42. We tried to examine the effect of NEP expression by co-culturing the NEP-expressing 293 cells with the N2a cells or by culturing the NEP-expressing 293 cells in CM derived from the N2a cells. However, the 293 cells used in the present study by themselves possessed a potent activity that completely removed the A␤ in CM, presumably through internalization and/or extracellular degradation employed by endogenous proteins. To avoid cell-mediated removal of A␤, the membrane fractions of the 293 cells were incubated with CM from the N2a cells at 37°C for 16 h. NEP, which was expressed either stably or transiently, reproducibly degraded both A␤1-40 and A␤1-42 compared with the control (pcDNA) (Fig. 6, A and B). PEX, NEPLP ␣, NEPLP ␤, NEPLP ␥, and DINE showed almost no proteolytic activity to the cellsecreted A␤1-40 or A␤1-42 (Fig. 6, A and B). DISCUSSION We previously reported that an endopeptidase(s) similar or identical to NEP is the most probable candidate for an A␤1-42-degrading enzyme in vivo (15). In the present study we cloned NEPLP cDNAs and compared A␤-degrading activity among NEP, PEX, NEPLPs, and DINE whose activities are sensitive to thiorphan and phosphoramidon. Using different methods (HPLC and ELISA) and different substrates (synthetic and cell-secreted A␤), we showed that among the peptidases tested NEP most rapidly and efficiently degraded not only A␤1-42 but also A␤1-40 at concentrations ranging from picomolar to micromolar in vitro. Because picomolar concentrations of A␤ peptides are present in human cerebrospinal fluids and plasmas, NEP is most likely to be an A␤-degrading enzyme in vivo.
We first demonstrated that enzymatic activities of PEX and NEPLP ␣/SEP ⌬ /splice 1 were sensitive to thiorphan and phosphoramidon and confirmed that that of NEPLP ␤/SEP/NL1/ NEPII was sensitive to these inhibitors (28,29). PEX proteolyzed synthetic A␤1-40 more slowly and to an extremely weaker extent than NEP, but it did not proteolyze synthetic A␤1-42 or cell-secreted A␤1-40 and A␤1-42. Although NEPLP ␣ had almost the same proteolytic activity to ZAAL-pNA as NEP, it degraded cold A␤1-40 and A␤1-42 more slowly and to a weaker extent than NEP and showed no detectable degrading activity to cell-secreted A␤1-40 and A␤1-42. These results suggest that PEX and NEPLP ␣ have lower affinities and/or smaller V max for A␤ than NEP, and therefore, they could not be the major degrading enzymes of endogenous A␤1-40 or A␤1-42 in vivo as compared with NEP. However, PEX and NEPLP ␣ might degrade A␤ in AD patients or APP transgenic mice brains, where A␤ is accumulated in excess. It seems unlikely that NEPLP ␤, NEPLP ␥, or DINE degrade A␤ in vivo, because they showed almost no proteolytic activity to A␤ under any of the conditions we examined.
Our results suggest that NEP directly degrades both A␤1-40 and A␤1-42 as determined by its inhibitor profiles, consistent with a previous report using purified NEP and synthetic A␤1-40 (7). Alternatively, NEP may mediate the initial cleavage of A␤ and another endogenous peptidase may further degrade A␤. Although it is possible that NEP is indirectly involved in A␤ degradation through, for instance, the proteolytic activation of another peptidase, this is unlikely because NEP is capable of proteolyzing peptides smaller than 4 -5 kDa (31). In any case, it is noted that NEP is required for A␤ degradation.
We found that NEP degraded cold A␤1-40 more rapidly than cold A␤1-42 in vitro (Fig. 5). This may be due to a higher V max for A␤1-40 than for A␤1-42, although NEP had a slightly higher affinity to A␤1-42 than to A␤1-40 (Table I). The rate of degradation by NEP was different among various A␤1-42 peptides ( Fig. 4 -6). The half-lives of radiolabeled A␤1-42, cold A␤1-42, and cell-secreted A␤1-42 were ϳ1, 8, and more than 16 h, respectively. Moreover, the degraded products of radiolabeled 1-42 and cold A␤1-42 may be different as evaluated by their retention time in HPLC. These different degrees of degradation are probably due to different conformations and concentrations of the peptides and/or different concentrations of the peptidases under the optimum conditions for each assay.
NEP is an ectoenzyme with a large extracellular domain containing a catalytic site. This indicates the direct involvement of NEP in A␤ degradation, because NEP can interact with and degrade A␤ with the correct topology on the cell surface. The fact that a soluble form of NEP is found in human plasma and cerebrospinal fluid (32,33) also supports the possible interaction between A␤ and NEP in the extracellular space. We could not determine whether NEP degraded extracellular A␤ or intracellular A␤ in this study. To address this issue a further study will be needed to measure intracellular A␤ as well as extracellular A␤ in doubly transfected cells with APP and NEP cDNAs.
In that A␤42 is preferentially accumulated in AD, it is important that NEP is capable of degrading A␤1-42. The high expression of NEP protein in the brain is restricted to striatum, olfactory tubercle, substantia nigra, choroid plexus, endopeduncular nucleus, and pontine nuclei, and moderate expression in cerebellum and low expression in hippocampus and cerebral cortex (34,35). The relatively low levels of NEP in hippocampus and cerebral cortex may explain the selective deposition of A␤42 in those areas. It is possible that altered activity of NEP in brain upon aging might lead to accumulation of A␤42. Indeed NEP mRNA and protein levels in the hippocampus and temporal gyrus were significantly lower in AD patients than those in control cases (36). It would be interesting to examine possible colocalization of NEP protein with senile plaques and neurofibrillary tangles. Genetic analysis of NEP, which is located in chromosome 3q21-27, in sporadic as well as familial AD cases may help understanding of the pathogenesis.
An intriguing finding in this study is that NEP can degrade either synthetic or cell-secreted A␤1-40 and A␤1-42 over a wide range of concentrations, from picomolar to micromolar. These results suggested not only that NEP degrades endogenous A␤ but also that NEP may be a good therapeutic target for the treatment of AD. Because there is a dynamic balance between monomer and polymer forms of A␤ (37)(38)(39), reducing the amounts of monomer A␤ will lead to a reduction of polymer A␤. It is important to search for methods that would up-regulate NEP activity or inactivate NEP inhibitors. Alternatively, targeting endogenous NEP activity to the site(s) of A␤ production or storage may reduce the A␤ levels in a specific and efficient manner without altering the total NEP activity (40). The use of NEP is advantageous in that NEP is non-destructive due to its limited ability to act only on peptides smaller than 4 -5 kDa.
Although it remains to be determined whether NEP degrades endogenous A␤ in vivo, knockout and transgenic studies of the NEP gene should elucidate this in future studies. The relative contribution of PEX and NEPLP in the catabolism of A␤ in vivo will also be clarified by such studies of each gene.