Prodomain Processing of Asp1 (BACE2) Is Autocatalytic*

Generation of the amyloid peptide through proteolytic processing of the amyloid precursor protein by β- and γ-secretases is central to the etiology of Alzheimer's disease. The highly elusive β-secretase was recently identified as a transmembrane aspartic proteinase, Asp2 (BACE). The Asp2 homolog Asp1 (BACE2/DRAP) has also been reported to exhibit β-secretase cleavage of amyloid precursor protein. Most aspartic proteinases are generated as inactive proenzymes, requiring removal of the prodomain to generate active proteinase. Here we show that prodomain processing of Asp1 occurs between Leu62 and Ala63 and is autocatalytic. Asp1 cleaved a maltose-binding protein-Asp1 prodomain fusion protein and a synthetic peptide at this site. Mutation of one of the conserved catalytic aspartic acid residues in the active site of Asp1 to asparagine (D110N) abolished this cleavage. Mutation of P1′ and P2′ residues in the substrate to phenylalanine reduced cleavage at this site. Asp1 expressed in cells was the mature form, and prodomain processing occurred intramolecularly within the endoplasmic reticulum/early Golgi. Interestingly, a proportion of mature Asp1 was expressed on the cell surface. When full-length Asp1(D110N) was expressed in COS-7 cells, it was not processed, suggesting that no other proteinase can activate Asp1 in these cells.

Generation of the amyloid peptide through proteolytic processing of the amyloid precursor protein by ␤and ␥-secretases is central to the etiology of Alzheimer's disease. The highly elusive ␤-secretase was recently identified as a transmembrane aspartic proteinase, Asp2 (BACE). The Asp2 homolog Asp1 (BACE2/DRAP) has also been reported to exhibit ␤-secretase cleavage of amyloid precursor protein. Most aspartic proteinases are generated as inactive proenzymes, requiring removal of the prodomain to generate active proteinase. Here we show that prodomain processing of Asp1 occurs between Leu 62 and Ala 63 and is autocatalytic. Asp1 cleaved a maltose-binding protein-Asp1 prodomain fusion protein and a synthetic peptide at this site. Mutation of one of the conserved catalytic aspartic acid residues in the active site of Asp1 to asparagine (D110N) abolished this cleavage. Mutation of P 1 and P 2 residues in the substrate to phenylalanine reduced cleavage at this site. Asp1 expressed in cells was the mature form, and prodomain processing occurred intramolecularly within the endoplasmic reticulum/early Golgi. Interestingly, a proportion of mature Asp1 was expressed on the cell surface. When full-length Asp1(D110N) was expressed in COS-7 cells, it was not processed, suggesting that no other proteinase can activate Asp1 in these cells.
Alzheimer's disease is a neurodegenerative disorder characterized by the deposition of the amyloid ␤-peptide in the brain as amyloid plaques (1,2). The amyloid ␤-peptide is generated through sequential proteolytic processing of the amyloid precursor protein (APP) 1 by ␤and ␥-secretases (3). Until recently, the identity of both ␤and ␥-secretases had eluded researchers for over a decade. We (4) and others (5)(6)(7)(8) identified ␤-secretase as a novel aspartic proteinase, Asp2 (BACE/Memapsin-2). Asp2 fulfills many of the key characteristics of ␤-secretase (9). Asp2 mRNA is abundant in the brain, and overexpression of Asp2 in cultured cells results in an increase in the level of ␤-secretasederived soluble APP and intracellular C-terminal fragments. An increase in the level of secreted amyloid ␤-peptide has also been observed in cells overexpressing Asp2. In addition, Asp2 colocalizes with APP within intracellular Golgi compartments, where ␤-secretase cleavage is known to occur (10). The homolog of Asp2, Asp1 (BACE2), has also been identified (11,12). Like Asp2, Asp1 also cleaves APP at the ␤-site, resulting in an increase in the level of ␤-secretase-derived soluble APP and intracellular C-terminal fragments (13,14). Thus, Asp1 represents a second ␤-secretase candidate. Recent reports suggest that the presenilins are novel aspartic proteinases that either exhibit ␥-secretase activity (15)(16)(17)(18) or form part of a large multimeric complex that includes ␥-secretase activity (19).
Like all aspartic proteinases, Asp1 and Asp2 are generated as proenzymes; and in most cases, removal of the prodomain is required to produce mature active enzyme. Therefore, an analysis of the mechanism of prodomain cleavage and the identification of the protease(s) that catalyze this event are essential for further understanding of the amyloidogenic cascade. Clearly, any upstream activating proteinases could represent novel therapeutic targets through which APP processing might be modulated. Recently, it has been reported that activation of Asp2 is mediated by furin or a furin-like enzyme that recognizes the RXXR motif immediately upstream of the prodomain cleavage site (20 -22). Having previously shown that prodomain processing in Asp1 occurs between residues Leu 62 and Ala 63 (13), we aimed in this study to determine whether this is an intra-or intermolecular event and to identify any enzyme(s) that could mediate this cleavage.

EXPERIMENTAL PROCEDURES
cDNA Constructs-Asp1 (base pairs 61-276) and Asp2 (base pairs 64 -225) fragments spanning the prodomain cleavage site (Ala 21 -Tyr 91 ) were amplified by polymerase chain reaction, in which the reverse oligonucleotide encoded a His 6 tag followed by a termination codon at the 3Ј-end. The fragments were cloned in frame to the maltose-binding protein (MBP) sequence at the BamHI/HindIII sites in pMal2c (New England Biolabs, Inc.) to generate MBP-Asp1pro-His 6 and MBP-Asp2pro-His 6 . Sequences were confirmed by dideoxy nucleotide sequencing. Generation of the Asp1-Fc/Signal pIgPlus construct has been described previously (13). The Asp2-Fc construct was generated by subcloning the Asp2 cDNA (base pairs 61-1359) into the HindIII/XhoI sites in the Signal pIgPlus vector (R&D Systems). Mutant constructs and the MBP-Asp1 prodomain construct terminating at the prodomain cleavage site were generated by QuikChange TM site-directed mutagenesis (Stratagene) and sequenced.
Expression and Purification of Fusion Proteins-Asp1-Fc and Asp2-Fc were expressed in COS-7 cells and purified from the culture medium as described previously (13). The MBP fusion proteins were expressed in Escherichia coli and purified. Briefly, protein expression was induced in 800 ml of LB medium with 0.3 mM isopropyl-␤-Dthiogalactopyranoside for 3 h at 37°C. The cells were harvested by centrifugation at 6000 ϫ g for 10 min and lysed by sonication in 50 ml of 20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM EDTA, 1% Nonidet P-40, 10 mg of lysozyme, and 0.1 mM phenylmethylsulfonyl fluoride at 4°C. The cell lysate was centrifuged at 10,000 ϫ g for 20 min, and the supernatant was incubated with amylose resin (5 ml) at 4°C for 1 h. The amylose resin was packed into a disposable column and washed twice with 10 ml of 20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM EDTA, and 0.1 mM phenylmethylsulfonyl fluoride. The protein was eluted with 10 mM maltose, 20 mM Tris-HCl (pH 7.4), 200 mM NaCl, 1 mM EDTA, and 0.1 mM phenylmethylsulfonyl fluoride. Fractions (1 ml) containing protein were pooled and dialyzed against 10 mM Tris-HCl (pH 6.8), 150 mM NaCl, 1 mM EDTA, 1 mM dithiothreitol, and 0.1 mM phenylmethylsulfonyl fluoride. * 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.
Cleavage Assays-MBP fusion proteins (2.5 g) were incubated with Asp1-Fc or Asp2-Fc (25 nM) in 100 mM sodium acetate (pH 4.5-6.5) in a final volume of 60 l at 37°C, and an aliquot (12 l) was removed at the indicated times. An equal volume of gel sample buffer was added to terminate the reaction, and samples were analyzed on 8% Tris/glycine gels (Novex). Proteins were visualized by staining with Coomassie Brilliant Blue.
Mass Spectrometric Analysis of MBP-Asp1 Fusion Proteins Incubated with Asp1-Fc-Cleavage assays were set up as described above. 10 -30 l of the incubation solutions were purified using 10-l pipette tips loaded with a C 18 resin (ZipTip, Millipore Corp.) employing the protocol supplied by the manufacturer. For MALDI-MS analysis, peptides were directly eluted from the ZipTip on to the MALDI plate with matrix solution (␣-cyano-4-hydroxycinnamic acid in water/acetonitrile (2:1) containing 0.1% trifluoroacetic acid). For Nanospray MS analysis, peptides were eluted from the ZipTip using water/acetonitrile (1:1) containing 5% formic acid. The eluate was immediately loaded into a Nanospray microcapillary (type N, Protana, Odense, Denmark). MALDI-time-of-flight-MS was performed on a TofSpec SE instrument (Micromass, Manchester, United Kingdom). Nanospray MS and MS/MS experiments were performed on an orthogonal acceleration quadrupole time-of-flight mass spectrometer (Q-Tof, Micromass) equipped with a Z-spray ion source for Nanospray analysis.
Peptide Cleavage Studies-Peptide ADGLALALEPALAK(Ahxdnp)G (50 M) was incubated with Asp1-Fc or Asp2-Fc (50 nM) in 50 mM sodium acetate and 20 mM NaCl (pH 4.5) for 2 h in a final assay volume of 30 l. The reaction was terminated by the addition of 4 volumes of 5% trifluoroacetic acid. The assay components were loaded onto a Poros R1 column (Applied Biosystems) in 0.08% trifluoroacetic acid and eluted with a linear gradient of 0.08% trifluoroacetic acid in acetonitrile (linear gradient of 5-50% acetonitrile over 15 min). The chromogenic dnp group on the full-length peptide and carboxyl terminus-derived cleavage products was followed by monitoring at 360 nm. Amino terminus-derived products are not detected at 360 nm due to loss of the dnp group. To determine the K m , reactions contained 10 nM enzyme and increasing substrate concentrations. Reactions were incubated at 37°C for 30 min and terminated by the addition of trifluoroacetic acid, and the amount of peptide cleaved was determined by HPLC as described above. Data were fitted to the Michaelis-Menten equation using GraFit Version 4.09. Incubations with inhibitory peptides were carried out as described above following preincubation of Asp1-Fc with 10 M MBP-tagged Asp1 prodomain peptide or 1 M APP ␤-secretase inhibitor H-Lys-Thr-Glu-Ile-Ser-Glu-Val-Asn-Stat-Val-Ala-Glu-Phe-OH (Calbiochem) at 37°C for 30 min.
Cell Transfections and Immunodetection-COS-7 cells overexpressing APP751 carrying the Swedish mutation (K651N/M652L) were transiently transfected with Asp1mycHis 6 /pcDNA3.1 or mutant Asp1(D110N)MycHis 6 /pcDNA3.1 using LipofectAMINE Plus reagent (Life Technologies, Inc.). 24 h post-transfection, cells were grown at the indicated temperatures or in the presence of brefeldin A (10 g/ml) or monensin (20 M) for 16 h, after which time cell lysates were generated as described previously (4). Proteins in the cell lysates were deglycosylated by treatment with N-glycosidase F (Roche Molecular Biochemi-cals) as described by the manufacturer. Cell lysates (30 g of protein) were resolved on Tris/glycine-SDS-polyacrylamide gels (Novex) for Western blot analysis using an anti-His 6 antibody (Roche Molecular Biochemicals) as described previously (4) or an anti-Asp1 C terminus antibody (ProSci Inc.).
Cell-surface Biotinylation Reactions-COS-7 cells overexpressing APP751 carrying the Swedish mutation were transiently transfected with Asp1mycHis 6 /pcDNA3.1. 48 h post-transfection, cells were washed extensively with phosphate-buffered saline prior to incubation with EZ-link TM sulfosuccinimidobiotin (Perbio Science UK Ltd.) at 0.5 mg/ml in phosphate-buffered saline for 1 h at room temperature. Cells were washed extensively with phosphate-buffered saline containing 20 mM glycine and harvested for lysis as described previously (4). Cell-surface biotinylated proteins were purified using streptavidin-agarose and eluted in gel sample buffer. Samples were resolved on Tris/glycine-SDSpolyacrylamide gels, followed by Western blot analysis with an anti-His 6 antibody, antibody WO2 raised against amino acids 1-16 of the amyloid ␤-peptide domain of APP (4), or with anti-KDEL antibody (Stressgen Biotech Corp.).

Asp1 Cleaves MBP-Asp1pro-His 6 at the Prodomain Cleavage
Site-To investigate prodomain processing of Asp1, we generated a recombinant MBP-Asp1 prodomain fusion protein that encodes MBP fused to Asp1 (amino acids 21-91) encompassing the prodomain cleavage site, with a His 6 tag at the C terminus (MBP-Asp1pro-His 6 ) (Fig. 1a). Modeling studies using the published sequence of Asp1 (BACE2) and sequence alignments were used to define the limits of the prodomain. The MBP-Asp1pro-His 6 protein was expressed in E. coli and purified using amylose resin. Purified MBP-Asp1pro-His 6 migrated as a doublet of ϳ51 kDa on gels (Fig. 1b) and was immunoreactive with both an anti-MBP and an anti-His 6 antibody (Fig. 1c). We have shown by N-terminal sequencing that both bands have the same MBP N terminus; hence, the difference in the two bands is most likely to be in the number of histidine residues at the C terminus.
To confirm this, in-gel digestion of these two bands followed by peptide mass fingerprinting employing MALDI-time-offlight-MS was performed. The results indicated that both bands correspond to the MBP-Asp1pro-His 6 fusion protein.
However, the N-terminal peptides as well as the His 6 -containing C-terminal peptides were not observed directly in the MALDI mass spectra. The failure to detect these peptides may have been caused by suppression effects known to occur in the MALDI analysis of digest mixtures (23). A minor band of lower molecular mass (ϳ40 kDa) was also observed that was immunoreactive with the anti-MBP antibody, but not with the anti-FIG. 1. Generation of MBP-Asp1pro-His 6 . a, schematic diagram showing the structure of MBP-Asp1pro-His 6 . Asp1 (amino acids 21-91; sequence shown below) was tagged at the N terminus with MBP and at the C terminus with a sixhistidine tag. The prodomain cleavage site is indicated (arrow). b, Coomassie Blue-stained gel of MBP-Asp1pro-His 6 , which migrated as a doublet. c, Western blot analysis of MBP-Asp1pro-His 6 with the anti-MBP or anti-His 6 antibody. d, Coomassie Blue-stained gel of purified Asp1-Fc and Asp2-Fc fusion proteins. His 6 antibody, suggesting that it represents a truncated form of the substrate or free MBP.
Cleavage of MBP-Asp1pro-His 6 at the prodomain cleavage site (Leu2Ala) is expected to generate two cleavage products, a large 47-kDa amino-terminal fragment and a small 4-kDa carboxyl-terminal fragment. To determine if Asp1 or Asp2 can mediate cleavage at this site, MBP-Asp1pro-His 6 was incubated with pure catalytically active Asp1 or Asp2, both expressed as soluble Fc fusion proteins (Fig. 1d). Incubation of MBP-Asp1pro-His 6 with Asp1-Fc resulted in a clear shift in the molecular mass of the prodomain substrate from ϳ51 to ϳ47 kDa (Fig. 2a), consistent with cleavage at the predicted site and loss of the 4-kDa carboxyl-terminal fragment, which was not detected. Western blot analysis showed that whereas MBP-Asp1pro-His 6 was immunoreactive with both anti-MBP and anti-His 6 antibodies, the 47-kDa cleavage product was immunoreactive only with the anti-MBP antibody due to loss of the C-terminal His 6 tag following cleavage (Fig. 2c). MS analysis of MBP-Asp1pro-His 6 incubated with Asp1-Fc was undertaken to detect the 4-kDa carboxyl-terminal fragment. MALDI-MS revealed the presence of one major and two minor carboxylterminal peptides in the samples incubated Asp1-Fc that were absent from control samples. To define the cleavage sites, Nanospray MS/MS analysis of the three cleavage peptides was performed. MS/MS sequence data showed clearly that the bulk of the processing of MBP-Asp1pro-His 6 occurred at the expected site, yielding the carboxyl-terminal peptide starting with ALEP (Fig. 2d). In addition, MS/MS sequencing of the two minor peptides revealed cleavage of MBP-Asp1pro-His 6 to yield carboxyl-terminal peptides beginning with LAMVD and AMVD (Fig. 2d). The ability of Asp1 to mediate prodomain cleavage at the correct site in the context of a recombinant fusion protein suggests that in vivo, this proteinase may autoactivate, removing its prodomain to generate the mature enzyme. In contrast, cleavage of MBP-Asp1pro-His 6 by Asp2 is very inefficient (Fig.  2a), suggesting that it is unlikely to activate Asp1 in vivo.
To determine whether the Asp2-Fc or Asp1-Fc fusion protein can process the Asp2 prodomain, we generated an MBP-Asp2pro-His 6 fusion protein in which amino acids 22-66 of Asp2 were fused to MBP with a C-terminal His 6 tag. This protein was also expressed in E. coli and purified using amylose resin. Incubation of this protein (ϳ49 kDa) with Asp1-Fc or Asp2-Fc did not result in a detectable shift in molecular mass ( Fig. 2b). Hence, in contrast to Asp1, prodomain cleavage of Asp2 is unlikely to be autocatalytic, and it does not appear to be mediated by Asp1. This result is consistent with reports suggesting that prodomain processing of Asp2 is mediated by another proteinase (20 -22).
Asp1 Prodomain Processing Is Optimal at pH 4.5-To determine the optimum pH profile for cleavage, MBP-Asp1pro-His 6 was incubated with Asp1-Fc at pH 4.5, 5.5, or 6.5. Of the pH values tested, Asp1-Fc cleaved MBP-Asp1pro-His 6 efficiently at pH 4.5, with 80% of the substrate converted into product within 30 min (Fig. 3b). At pH 5.5, cleavage was still evident, although only 24% of the substrate was converted to product within 30 min. At pH 6.5, cleavage was less efficient, with only 22% turnover of substrate within 30 min.
The Asp1 Active-site Mutant D110N Cannot Autoactivate or Cleave MBP-Asp1pro-His 6 -The catalytic activity of aspartic proteinases is dependent on the presence of two aspartic acid residues in the active site. Hence, processing of MBP-Asp1pro-His 6 should be abolished if either of the catalytic aspartic acid residues in Asp1 is mutated. The first catalytic aspartic acid residue in Asp1-Fc was mutated to asparagine (D110N), and the protein was expressed in COS-7 cells and purified from the medium using protein A-Sepharose. In the cleavage assays, Asp1(D110N)-Fc was unable to cleave MBP-Asp1pro-His 6 at the prodomain cleavage site since no shift in molecular mass was apparent (Fig. 3a). N-terminal sequencing of mutant Asp1-Fc revealed that 100% of the protein still possessed its prodomain, whereas wild-type Asp1-Fc expressed in COS-7 cells was fully processed and had the amino-terminal sequence A 63 LEP (13). Thus, it appears that the mutant Asp1(D110N)-Fc is not processed in COS-7 cells due to its inability to autoactivate.
Mutations in the Prodomain Cleavage Site Affect Cleavage Site Selection-Phenylalanine scanning mutagenesis was conducted through the Asp1 prodomain cleavage site to determine the effect on prodomain processing. Residues P 2 -P 2 Ј (ALAL) in MBP-Asp1pro-His 6 were mutated individually to phenylalanine, and the purified mutant proteins were incubated with Asp1-Fc. Cleavage of the mutant MBP-Asp1pro-His 6 substrates appeared to be as efficient as that of the wild-type substrate when expressed as the proportion of substrate cleaved (Fig. 3c). However, MALDI-MS analysis of the 4-kDa cleavage product revealed that mutation of the P 1 Ј and P 2 Ј 6 and MBP-Asp2pro-His 6 , respectively, were incubated with either Asp1-Fc (25 nM) or Asp2-Fc (25 nM) in 100 mM sodium acetate (pH 4.5) at 37°C for the indicated times. Reactions were resolved on 8% Tris/glycine-SDS-polyacrylamide gels for staining with Coomassie Brilliant Blue. Cleavage of MBP-Asp1pro-His 6 at the predicted site resulted in the generation of a 47-kDa cleavage product. c, MBP-A-sp1pro-His 6 was incubated with Asp1-Fc at 37°C for the indicated times, followed by Western blot analysis with the anti-MBP or anti-His 6 antibody. d, shown is the sequence of Asp1 around the prodomain cleavage site, with the cleavages observed highlighted. Cleavage at the predicted site is indicated with a solid arrow, and the other cleavages are indicated with broken arrows.

FIG. 2. Asp1-Fc cleaves MBP-A-sp1pro-His 6 at the prodomain cleavage site. a and b, MBP-Asp1pro-His
residues reduced the level of the peptide detected with the N-terminal sequence ALEP and increased the level of peptide detected with the N-terminal sequence LAMVD. In contrast, mutations at the P 1 and P 2 positions had no effect on cleavage. This indicates that the active site of Asp1 is unable to accommodate large bulky residues at either the P 1 Ј or P 2 Ј position.
Asp1 Cleaves Synthetic Peptides at the Prodomain Cleavage Site-We went on to determine whether Asp1 can cleave a synthetic peptide spanning the prodomain cleavage site of Asp1. Incubation of Asp1-Fc with peptide ADGLALALEPALA-K(Ahx-dnp)G resulted in cleavage to give a single dnp-labeled product (Fig. 4, a and b). MS analysis showed that this cleavage occurred at the predicted prodomain cleavage site, between Leu 6 and Ala 7 . Asp2-Fc also cleaved the Asp1 prodomain peptide at this site, albeit poorly in comparison with Asp1-Fc (data not shown). An initial velocity plot of Asp1-Fc-mediated cleavage of the synthetic peptide with respect to substrate concentration is shown in Fig. 4c, giving a K m at pH 4.5 of 197 Ϯ 30 M. To determine if the Asp1 prodomain has any inhibitory activity, assays were carried out in the presence of an MBPtagged Asp1 prodomain peptide (residues 21-62). Preincubation of active Asp1-Fc (50 nM) with a 10 M concentration of this peptide resulted in ϳ16% inhibition of cleavage of synthetic Asp1 cleavage peptide under standard conditions (see "Experimental Procedures"). This contrasts with Ͼ50% inhibition of cleavage by 1 M APP ␤-secretase inhibitor, indicating that the prodomain is a very weak inhibitor of Asp1.
Prodomain Processing of Asp1 in Vivo-To determine if Asp1 autoactivates in vivo and to identify the intracellular compartment in which prodomain processing occurs, COS-7 cells overexpressing APP751 carrying the Swedish mutation (K651M/ N652N) were transiently transfected with Asp1MycHis 6 or mutant Asp1(D110N)MycHis 6 . To detect Asp1 proenzyme, transfected cells were grown at reduced temperatures, which have been shown to lead to the retention of proteins in different compartments of the secretory pathway (24,25). Cell lysates were generated and subjected to deglycosylation with N-glycosidase F prior to Western blot analysis with the anti-His 6 antibody. Untreated Asp1 migrated as a protein of ϳ60 kDa (Fig. 5a, first lane) and, upon deglycosylation, exhibited a shift in mobility to a molecular mass of ϳ50 kDa due to removal of the N-linked glycans (second lane). Deglycosylation and immunodetection of Asp1 in the cell lysate from cells grown at 15°C, which inhibits the transit of proteins from the endoplasmic reticulum (ER) to the Golgi, revealed two closely migrating bands of ϳ50 and ϳ54 kDa (Fig. 5a, third lane). The 54-kDa protein corresponds to the prodomain-containing form of Asp1, whereas the 50-kDa protein is the mature proteinase that lacks the prodomain. Hence, at low temperature, prodomain processing of Asp1 is reduced, resulting in the accumulation of the proenzyme. In contrast to wild-type Asp1, when Asp1(D110N)-MycHis 6 was deglycosylated, it migrated exclusively as the proenzyme of ϳ54 kDa following growth at 37 and 15°C (Fig.  5a, fourth and fifth lanes). Thus, processing is abolished in the mutant enzyme. A further implication of this result (see below) is that, in these cells, there is no other enzyme that is capable of processing Asp1.
The detection of mature as well as pro-Asp1 following growth at 15°C (Fig. 5, a and b) suggests that processing can occur early in the secretory pathway. When cells were grown at 20°C, which inhibits the transit of protein from the trans-Golgi network to the cell surface (25), very little proenzyme accumulated (Fig. 5b), implying that very little prodomain processing occurs in post-Golgi compartments. To further confirm the identity of the intracellular site where prodomain processing of Asp1 occurs, cells were grown in the presence of drugs known to interfere with the trafficking of proteins (Fig. 5b). Growth in the presence of the fungal metabolite brefeldin A, which causes a redistribution of proteins that normally reside in the Golgi into the ER (26), revealed the accumulation of both pro and mature forms of Asp1. In contrast, growth in the presence of the monovalent ionophore monensin, which disrupts late Golgi and endosomal functions (27), caused the accumulation of mature Asp1, but little proenzyme. Hence, the majority of prodomain processing of Asp1 does not occur in late Golgi/endosomal compartments, but rather in the ER/early Golgi compartments.
The results described above for the processing of the activesite mutant enzyme indicate that no other intracellular proteinase is capable of cleaving the prodomain of Asp1 and that prodomain cleavage is an exclusively intramolecular event. To confirm this finding, Asp1(D110N)MycHis 6 was coexpressed with untagged wild-type Asp1. Western blot analysis of the cell lysates with the anti-Asp1 C terminus antibody showed that expression of both Asp1 proteins was successful (Fig. 5c, upper  panel). If the cotransfected untagged Asp1 is able to cleave the prodomain of Asp1(D110N)MycHis 6 intermolecularly, the mature form of this protein should now be detected with the antimyc or anti-His 6 antibody. However, Western blot analysis with the anti-His 6 antibody revealed that Asp1(D110N)-Myc␤␤His 6 still migrated as the proenzyme in these cells FIG. 3. Properties of Asp1 prodomain processing in vitro. a, MBP-Asp1pro-His 6 was incubated with Asp1-Fc or Asp1(D110N)-Fc at 37°C for the indicated times. Reactions were resolved on 8% Tris/glycine-SDS-polyacrylamide gels and stained for protein with Coomassie Brilliant Blue. b, prodomain processing of Asp1 was optimal at pH 4.5. MBP-Asp1pro-His 6 was incubated with Asp1-Fc in 100 mM sodium acetate (pH 4.5, 5.5, or 6.5) at 37°C for the indicated times. Reactions were analyzed on SDS-polyacrylamide gels, followed by densitometric analysis to determine the amount of substrate remaining at each time point. c, mutant MBP-Asp1pro-His 6 substrates were incubated with Asp1-Fc at 37°C for the indicated times. Reactions were analyzed on SDS-polyacrylamide gels, followed by densitometric analysis to determine the amount of substrate remaining at each time point. WT, wild-type. (Fig. 5c, lower panel). Hence, prodomain processing of Asp1 occurs exclusively through an intramolecular reaction.
Mature Asp1 Is Present at the Cell Surface-Proportions of cellular APP and Asp2 have been shown to be present on the cell surface (28 -30). To determine whether Asp1 is also expressed at the cell surface, we performed biotinylation experiments on COS-7 cells overexpressing APP751 carrying the Swedish mutation and transiently transfected with Asp1. FIG. 5. Prodomain processing of Asp1 in vivo. COS-7 cells expressing APP751 carrying the Swedish mutation were transiently transfected with the following. a, Asp1mycHis 6 or Asp1(D110N)MycHis 6 , followed by growth at the indicated temperatures. Cell lysates were generated, and proteins were deglycosylated by treatment with N-glycosidase F prior to Western blot analysis with the anti-His 6 antibody. b, Asp1mycHis 6 , followed by growth at reduced temperature or in the presence of brefeldin A (BFA) or monensin (MON). Cell lysates were generated, and proteins were deglycosylated by treatment with N-glycosidase F prior to Western blot analysis with the anti-His 6 antibody. c, untagged wild-type Asp1, Asp1(D110N)MycHis 6 , or both. Cell lysates were generated, and proteins were deglycosylated by treatment with N-glycosidase F prior to Western blot analysis with the anti-Asp1 C terminus antibody (upper panel) or the anti-His 6 antibody (lower panel). d, Asp1MycHis 6 , followed by cell-surface biotinylation as described under "Experimental Procedures. Transfected cells were incubated with sulfosuccinimidobiotin to biotinylate the cell-surface proteins. Cell lysates were generated, and the biotinylated cell-surface proteins were immunoprecipitated with streptavidin-agarose. A significant level of total Asp1 was detected at the cell surface (Fig. 5d, upper  panel, lane 1). As a control, the cell-surface and intracellular cell extracts were also probed for APP, which is known to reside on the cell surface (28,29), and for the endoplasmic reticulumlocalized chaperones Grp94 and Grp78 (31). Consistent with published data, we found that APP was present at the cell surface and intracellularly (Fig. 5d, middle panel, lanes 1 and  2), whereas the endoplasmic reticulum-resident proteins were detected only within the cell (lower panel, lane 2). To determine if Asp1 residing at the cell surface is the mature proteinase lacking the prodomain, the biotinylated cell extracts were deglycosylated with N-glycosidase F prior to incubation with streptavidin-agarose. Following deglycosylation, the biotinylated cell-surface Asp1 migrated at a molecular mass consistent with the mature form of the protein (Fig. 5e, lane 1). Thus, prodomain processing of Asp1 occurs during its transit through the secretory pathway en route to the cell surface, and the protein expressed on the cell surface is the processed, potentially active form of the enzyme. However, prodomain processing is not a prerequisite for transit to the cell surface, as Asp1(D110N)mycHis 6 , which still possesses its prodomain, was also expressed at the cell surface (data not shown).

DISCUSSION
All known aspartic proteinases are generated as proenzymes, requiring removal of the prodomain to generate mature proteinase (32). Although most aspartic proteinases such as pepsin and cathepsin E autoactivate upon acidification, others such as renin require a separate proteinase activity for prodomain cleavage. It has recently been reported that Asp2 does not autoactivate and that removal of its prodomain is mediated by furin or a furin-like enzyme (20 -22). We report that in contrast to Asp2, Asp1 is capable of autocatalytic activation, cleaving its prodomain to generate mature proteinase. In vitro, Asp1 cleaved both an MBP-Asp1 prodomain fusion protein and a synthetic peptide at the predicted prodomain cleavage site (Leu 62 2Ala 63 in full-length Asp1). This cleavage was abolished when one of the conserved aspartic acid residues in the active site of Asp1-Fc was mutated to asparagine. In addition, mutation of either the P 1 Ј (Ala) or P 2 Ј (Leu) residue in the MBP-Asp1 prodomain substrate to phenylalanine reduced cleavage at this site. The other cleavages observed when MBP-Asp1pro-His 6 was incubated with Asp1-Fc are unlikely to be physiologically relevant, as Asp1-Fc secreted from cells had only one N-terminal sequence, A 63 LEP. In vivo, the majority of wild-type Asp1 expressed in cells was the mature form lacking the prodomain. However, when Asp1(D110N) was expressed in cells, the proenzyme accumulated, presumably due to its inability to autoactivate. Similarly, soluble Asp1(D110N)-Fc secreted from COS-7 cells still possessed the prodomain, whereas wild-type Asp1-Fc was fully processed. Prodomain processing of Asp1 occurred intramolecularly, as coexpression of Asp1(D110N) with fulllength wild-type Asp1 did not result in prodomain processing of mutant Asp1. These observations indicate not only that prodomain cleavage is autocatalytic, but also that no other cellular activity can compensate for the lack of prodomain processing of the mutant Asp1(D110N) in these cells. Although we showed in vitro that Asp2 could cleave both MBP-Asp1pro-His 6 and a synthetic peptide, cleavage was inefficient compared with Asp1, suggesting that Asp2 is unlikely to be a major Asp1processing enzyme in vivo.
In addition, we showed in cultured cells that autoactivation of Asp1 occurred in the ER/early Golgi. Mature Asp1 was observed in cells following growth at 15°C or in the presence of brefeldin A. This is in contrast to prodomain processing of Asp2, which is dependent upon an enzyme that resides in the trans-Golgi network (20). Autocatalytic activity of aspartic proteinases is known to be induced by acidic conditions (32). Although the pH of the ER/early Golgi compartment is only mildly acidic at pH ϳ6.5 (33,34), we have demonstrated in vitro that Asp1 exhibited activity at this pH, which may be sufficient to allow prodomain processing in vivo. The prodomain of Asp1 did not exhibit any significant inhibitory activity in vitro, suggesting that the prodomain may be required for proper folding of the protein, as has been reported for Asp2 (35). To conclude, prodomain processing of Asp1 and Asp2 is clearly very different; Asp1 autoactivates upon acidification, whereas Asp2 requires a separate enzymatic activity to cleave its prodomain. The autocatalytic nature of Asp1 suggests a less stringent regulation of its activity compared with Asp2. However, it is feasible that other intracellular factors exist that influence the activity of Asp1.