Involvement of Cell Surface Glycosyl-phosphatidylinositol-linked Aspartyl Proteases in α-Secretase-type Cleavage and Ectodomain Solubilization of Human Alzheimer β-Amyloid Precursor Protein in Yeast*

Human β-amyloid precursor protein (APP) introduced into yeast undergoes α-secretase-type cleavage, suggesting that yeast have α-secretase-like protease(s). Here we report that two structurally and functionally related glycosyl-phosphatidylinositol-linked yeast aspartyl proteases, Mkc7p and Yap3p (collectively termed yapsin), are responsible for α-secretase-type cleavage of APP expressed in yeast, resulting in release of soluble APP into the extracellular space. Disruption ofMKC7 and YAP3 in a vacuolar protease-deficient strain abolished this APP cleavage/release, and APP cleavage/release could be restored by introduction of MKC7 orYAP3 on a single copy plasmid. Purified Mkc7p cleaved an internally quenched fluorogenic APP peptide substrate at the α-secretase cleavage site. Measurement of proteolytic activity either in yeast homogenates or on the yeast cell surface revealed that most Mkc7p and Yap3p activities were localized at the cell surface. These results establish a molecular basis for α-secretase-type cleavage in yeast and support the generally held concept that α-secretase cleavage of APP occurs at the cell surface.

brane protein, can undergo at least two alternative pathways of proteolytic processing. ␤-Amyloid peptide (A␤), the component of cerebral plaques associated with Alzheimer's disease, is produced by proteolytic processing of APP and includes 28 residues of the N-terminal lumenal domain and the first 12-14 residues of the transmembrane domain (1). Alternatively, cleavage of a Lys-Leu bond the lumenal portion of the A␤ region of APP by a membrane protein-solubilizing proteolytic activity, termed "␣-secretase," releases the lumenal portion of APP as a soluble protein and precludes the formation of A␤ (1). There has been intensive interest in identifying the enzymes that cleave APP to determine their relationship to the etiology of Alzheimer's disease. However, the responsible molecules remain unknown. In particular, molecular identification of ␣-secretase could provide insight into the general mechanism of integral membrane protein cleavage (2) and clarify the role of this enzyme in the pathogenesis of Alzheimer's disease.
We have reported that the yeast Saccharomyces cerevisiae has ␣-secretase-like activities: human APP expressed in S. cerevisiae undergoes proteolytic cleavage at the ␣-secretase site (3,4). Existence of an ␣-secretase-like activity was also observed in insect cells (5), suggesting that the enzymes are widely conserved among metazoans. Therefore, identification of yeast secretases could lead to the discovery of counterparts among other phyla. Here we demonstrate that homologous glycosyl-phosphatidylinositol (GPI)-linked aspartyl proteases, Mkc7p (6) and Yap3p (7, 10 -12), termed "yapsin" (33), are responsible for ␣-secretase cleavage of human APP in yeast.
Plasmids-DNA manipulations followed standard methods (15). Plasmids pMF-APP751 and pMF were as described (3). pMF-APP751 consists of the prepro-segment of prepro-␣-factor that includes processing sites for Kex2 (Lys-Arg) followed by residues 19 -751 of human APP751, under control of the GAL1 promoter on the CEN4 ARS1 URA3 plasmid pBM258 (16). pMF consists of the prepro-␣-factor structural gene MF␣1, deleted of three of the four ␣-factor repeats, under control of the GAL1 promoter on the plasmid pBM258. pRS314 is a CEN plasmid harboring TRP1 (17). pRS4MKC7 is a SpeI-SmaI fragment containing the MKC7 gene inserted into plasmid pRS314. pRS4YAP3 is a SalI-XbaI fragment containing the YAP3 gene inserted into plasmid pRS 314. pRS316 is a CEN plasmid harboring URA3 (17). pRS6MKC7 is a SpeI-SmaI fragment containing the MKC7 gene inserted into plasmid pRS316. pYO324 is pRS304 carrying the 2-m replication origin (18) and was a gift from Dr. Ohya (University of Tokyo, Tokyo, Japan). pYO4MKC7 is a SpeI-SmaI fragment containing the MKC7 gene inserted into plasmid pYO324. For overexpression of MKC7 or YAP3, a fragment containing MKC7 or YAP3 structural gene was placed under the control of the strong promoter, TDH3, by insertion into multicopy vector pG5 (19), creating pG5MKC7 for MKC7 and pG5YAP3 for YAP3, respectively. Construction of pG5MKC7 will be described elsewhere. 2 * This work was supported by National Institutes of Health Grants AG11508, AG09464, AG10491, AG13780, AG05689, and GM39367. 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. § Present address: National Inst. for Longevity Sciences, 36-3 Gengo, Morioka, Obu, Aichi 474, Japan.
2 H. Komano, N. Rockwell, G. T. Wang, G. A. Krafft, and R. S. Fuller, submitted for publication. In this study, the IQ peptide containing the APP ␣-secretase site was shown to be a substrate for purified Mkc7p (yapsin 2). The peptide fragments were analyzed by mass spectrometry To construct pG5YAP3, the BamHI and SalI sites were introduced upstream of NdeI site and downstream of the BstEII site of YAP3, respectively, by PCR. Then the NdeI-BstEII fragment from the PCR product was replaced with the original fragment from YAP3. This was digested with BamHI and SalI and ligated to pG5, which had been cut with BglII and SalI.
Radiolabeling and Immunoprecipitation-Radiolabeling was performed as described (14,20). Yeast strains containing pMF-APP751 or pMF were grown at 30°C in low sulfate medium containing 100 mM ammonium sulfate and 2% galactose (14). Cultures were harvested at a density of ϳ10 7 cells/ml and subjected to sulfate depletion for 30 min by resuspension in low sulfate medium containing 20 M ammonium sulfate and 2% galactose, following which cells were labeled for 30 min with 150 Ci of [ 35 S]methionine/ml (NEN Life Science Products). Labeled cells (1 ml) were chilled on ice after addition of 10 mM sodium azide, harvested by centrifugation, and washed once with 10 mM HEPES, pH 7.0, containing protease inhibitors (10 mM EDTA, 1 mM phenylmethylsulfonyl fluoride, 100 M N-tosyl-L-phenylalanine chloromethyl ketone, 100 M N ␣ -(p-tosyl) lysine chloromethyl ketone, 1 mM benzamidine-HCl, 25 M pepstatin A). Cells were stored at Ϫ80°C prior to lysis. Cell lysis was performed as described (14). To immunoprecipitate the C-terminal fragment of human APP, 2 l of affinity-purified rabbit antibody 369 against the C-terminal 645-694 residues of APP695 (21) and 30 l of 50% Pansorbin slurry (Calbiochem) were added to the lysate made from 1 ml of labeled cells. Immunoprecipitates were washed and then solubilized in SDS sample buffer (50 mM Tris-HCl, pH 6.8, 2% SDS, 30%, v/v glycerol, 5% 2-mercaptoethanol, 5 mM EDTA) at 97-98°C for 3 min, as described previously (14) and subjected to SDS-polyacrylamide gel electrophoresis and fluorography.
Immunoprecipitation of released soluble APP (sAPP) was performed as described (4), because most sAPP is bound to the surface of yeast cells (4). A 1-ml aliquot of labeled cells was washed once with 50 mM Tris-HCl, pH 10.5, containing 5 mM dithiothreitol, resuspended in 1 ml of the same buffer, and then incubated for 30 min at 30°C. After centrifugation at 40,000 ϫ g for 15 min to pellet the cells, the clear supernatant was mixed with 10 l of mouse monoclonal antibody 22C11 (22) against the N-terminal intralumenal domain of APP, 5 l of rabbit anti-mouse IgG, and 25 l of 50% protein A-Sepharose (Amersham Pharmacia Biotech) slurry, to immunoprecipitate the sAPP. The immunoprecipitates were solubilized in SDS sample buffer and subjected to SDS-polyacrylamide gel electrophoresis and fluorography.
Peptide Sequencing-0.04 g/ml of Mkc7p and 200 pmol of IQ substrate, AcArg-Glu(EDANS)-Val-His-His-Gln-Lys-Leu-Val-Phe-Lys(D-ABCYL)Arg, were incubated for 60 min under standard reaction conditions. 2 The reaction was terminated by addition of 0.1 N NaOH (final concentration, 50 mM). N-terminal sequencing and mass spectrometry (MALDI-TOF-MS) of cleavage products were carried out by The Biomedical Research Core Facilities at the University of Michigan.
Assay of Cell Surface Proteolytic Activity-Assay of cell surface proteolytic activity was performed essentially as described (25). Cells were grown to a density of  total activity, cells were lysed by three cycles of freezing on dry ice and thawing at 0°C in reaction buffer containing protease inhibitors and 1% Triton X-100.

␣-Secretase Cleavage of Human APP Expressed in Yeast:
Blockade by Double Disruption of MKC7 and YAP3-Double disruption of MKC7 and YAP3 greatly diminished ␣-secretase site cleavage of APP, as evidenced by the levels of cell-associated APP C-terminal fragment (Fig. 1A, lane 1 versus lane 3), and ␣-secretase-type APP cleavage was completely abolished when the double disruption was created in a vacuole proteasedeficient strain (Fig. 1A, lane 5 versus lane 7). Expression of MKC7 (Fig. 1B, lane 2 versus lane 1) or YAP3 (Fig. 1B, lane 4  versus lane 1) under a wild-type promoter on a single copy yeast centromeric plasmid restored this cleavage. The YAP3-transformed cells showed greater APP ␣-secretase activity than did the MKC7-transformed cells (Fig. 1B, lanes 2 and 4). Overexpression of MKC7 greatly enhanced ␣-secretase-type APP cleavage (Fig. 1B, lane 6). These results indicate that both Mkc7p and Yap3p can catalyze ␣-secretase site APP cleavage in intact cells, although some of this cleavage can involve vacuolar proteases. Of note, it was frequently possible to detect a slowly migrating APP C-terminal fragment (Fig. 1B) potentially consistent with a "␤-C-terminal-fragment" or "C100-fragment" bearing A␤ at its N terminus (26) and thus constituting the immediate precursor of A␤. Immunochemical and radiochemical characterization of this APP C-terminal fragment is underway.
In yeast, most of the ␣-secretase-cleaved N-terminal APP fragment (i.e. sAPP) associates with the exterior of the cells, and only a small proportion of this sAPP appears in the medium (4). Treatment of the cells with 5 mM dithiothreitol at pH 10.5 for 30 min releases much of this cell surface-associated sAPP (4). Disruption of MKC7 in a vacuolar protease-deficient strain abolished generation of this cell surface-associated sAPP (Fig. 2). Introduction of either MKC7 or YAP3 on a single copy
plasmid restored generation of this fragment, and introduction of multicopy MCK7 increased sAPP production. (Fig. 2). Thus, some APP undergoes Mkc7p-or Yap3p-mediated cleavage to produce an intracellular C-terminal fragment and a cell surface-associated, sAPP-like N-terminal APP fragment. Cleavage of APP Peptide at the ␣-Secretase Site by Purified Mkc7p-V max /K m for this IQ substrate by purified Mkc7p was determined, 2 and the cleavage site identified by Edman degradation and MALDI-TOF-MS indicated that Ͼ95% of the cleavage occurred at the Lys-Leu bond (the ␣-secretase site). 3 Mkc7p cleaved this substrate at pH 4.0 with a V max /K m of 0.03 min Ϫ1 , normalized to 1 g/ml Mkc7p, equivalent to a k cat /K m of 3.3 ϫ 10 4 M Ϫ1 s Ϫ1 assuming a fully active enzyme preparation. Although this reaction was approximately 2 orders of magnitude less than those reactions observed for cleavage of substrates based on pro-␣-factor cleavage sites at pH 4.0, the V max /K m for the ␣-secretase site in APP was increased ϳ20-fold when tested at pH 6.0 (k cat /K m ϳ 6.6 ϫ 10 5 M Ϫ1 s Ϫ1 ). 2 Localization of Yapsin at the Cell Surface-GPI-anchored proteins appear to be delivered to the cell surface (27). Yap3p in particular was reported to be localized to the plasma membrane by subcellular fractionation (12). Therefore, we directly measured the proteolytic activity of Mkc7p and Yap3p on the cell surface using both an APP-based IQ substrate as well as an IQ substrate based on pro-␣-mating factor ( Table I) that is a better substrate for these enzymes. 2 Evidence that Mkc7p and Yap3p are at the cell surface came from experiments using the highly efficient pro-␣-factor IQ substrate. Significant proportions of proteolytic activity were consistently detected at the cell surface (Table I). 2 The surface activity was not reduced in vacuolar protease deficient strains (Table I, experiments A-C), but surface activity was reduced to ϳ25% in double disruptants of MKC7 and YAP3 (Table I, experiment A). This reduction was rescued to the wild-type level by expression of MKC7 under a wild-type promoter on a single copy plasmid but not by expression of YAP3 (Table I, experiment B). Overexpression of YAP3 led to a significant proteolytic activity at the yeast cell surface but much less than that of overexpression of MKC7 (Table I, experiments B and D).
When the APP based IQ substrate was used, overexpression of either YAP3 or MKC7 resulted in cell surface peptidase activity (Table I, experiment D). The difference in the relative efficiencies of Yap3p and Mkc7p with the two IQ substrates indicates that these proteases have distinct substrate specificities.
In conclusion, Mkc7p and Yap3p, GPI-linked proteases associated with the cell surface, form the molecular basis for ␣-secretase-type cleavage in yeast, supporting the generally held concept that ␣-secretase cleavage of APP occurs at the cell surface (28). Because we demonstrate both localization of yapsin activity to the cell surface and a relationship between sAPP generation and expression of the yapsin genes, our data extend those recently reported by Zhang et al. (29), implicating Yap3p and Mkc7p in the intracellular metabolism of human APP to generate a cell-associated nonamyloidogenic C-terminal fragment. Those investigators did not describe recovery of an sAPP species or of a potential C100/␤-C-terminal-APP-fragment as we report here.
It is worth noting that two groups have recently discovered cell surface disintegrin family metalloproteinases that apparently solubilize the ectodomains of tumor necrosis factor-␣ (30,31) and perhaps transforming growth factor-␣ and APP (32). Because the latter two activities may be identical at the molecular level, it will now be possible to investigate the roles of aspartyl-and metalloproteinases as ectodomain solubilizing enzymes in various cases.
Beyond ␣-secretase-type APP cleavage and arguably more relevant to Alzheimer's disease pathobiology, highly sensitive methods now enable the investigation of whether ␤-, ␥ 40 -, and/or ␥ 42 -secretase activities exist in yeast, enabling a complete reconstitution of A␤ generation in S. cerevisiae. Such studies are now in progress (34,35).