Purification and Characterization of the Yeast Glycosylphosphatidylinositol-anchored, Monobasic-specific Aspartyl Protease Yapsin 2 (Mkc7p)*

The Saccharomyces cerevisiae YPS2(formerly MKC7) gene product is a glycosylphosphatidylinositol-linked aspartyl protease that functions as a yeast secretase. Here, the glycosylphosphatidylinositol-linked form of yapsin 2 (Mkc7p) was purified to homogeneity from the membrane fraction of an overexpressing yeast strain. Purified yapsin 2 migrated diffusely in SDS-polyacrylamide gel electrophoresis (molecular mass ∼ 200 kDa), suggesting extensive, heterogeneous glycosylation. Studies using internally quenched fluorogenic peptide substrates revealed cleavage by the enzyme carboxyl to Lys or Arg. No cleavage was seen when both Lys and Arg were absent. No significant enhancement was seen with multiple basic residues. However, cleavage always occurred carboxyl to the most COOH-terminal basic residue.V max/K m was insensitive to P2 and P3 residues except that Pro at P2 blocked cleavage entirely. These results suggest that yapsin 2 is a monobasic amino acid-specific protease that requires a basic residue at P1 and excludes basic residues from P1′. The pH dependence ofV max/K m for a substrate containing a pro-α factor cleavage site was bell-shaped, with a maximum near pH 4.0. However,V max/K m for a substrate mimicking the α-secretase site in human β amyloid precursor protein was optimal near pH 6.0, consistent with cleavage of β amyloid precursor protein by yapsin 2 when expressed in yeast.

In Saccharomyces cerevisiae, the KEX2 gene encodes a Ca 2ϩdependent, transmembrane serine protease responsible for proprotein processing after pairs of basic residues (1)(2)(3)(4). Genes encoding two homologous membrane-associated aspartyl proteases have been isolated as multicopy suppressors of kex2 null mutant phenotypes. The MKC7 gene, now termed YPS2, 1 was isolated as a multicopy suppressor of the cold sensitivity of kex2 mutants (5). YAP3, now termed YPS1, 1 was isolated as a multicopy suppressor of the kex2 defect in pro-␣ factor processing (6). YPS1 was also identified as a multicopy suppressor of mutations that blocked processing in yeast of anglerfish proso-matostatin at a single Arg-type cleavage site (7).
YPS1 and YPS2 encode aspartyl proteases, yapsin 1 and yapsin 2, that are structurally and functionally similar to each other and are associated with the membrane through glycosylphosphatidylinositol (GPI) 2 anchors (5,8). Interestingly, disruptions of the KEX2, YPS1, and YPS2 genes resulted in synergistic effects on cell growth at low and high temperatures, suggesting partial overlap in the specificity and/or physiological function of these three enzymes (5). Yapsin 1 has been shown to cleave at clusters of basic amino acid residues in vitro, indicating that the enzyme has specificity for basic amino acid residues (9,10), and a structural model suggests that the enzyme may interact with substrates residues from P 6 to P 6 Ј (11). 3 However, studies of the specificity of these aspartyl proteases have yet to be carried out with systematic sets of model substrates.
Recently, we demonstrated that the majority of yapsin 1 and yapsin 2 activity is located at the cell surface (12), and we and others have shown that these enzymes are required for cleavage of the Alzheimer's disease-related human ␤-amyloid precursor protein (␤APP) at the ␣-secretase cleavage site when ␤APP is expressed in yeast (12)(13)(14)(15). The identification of additional genes homologous to YPS1 and YPS2 in the S. cerevisiae genome 1 suggests the existence of a family of GPI-anchored, cell surface aspartyl proteases that function as "secretase" or "sheddase" activities in yeast (see Ref. 16 for review of such activities). Just as Kex2 protease was the prototype for a family of homologous mammalian proprotein processing enzymes (4,17), yapsin 1 and yapsin 2 are potential prototypes of a family of eukaryotic secretase enzymes distinct from TACE and related metalloproteases (18,19).
Here we report the purification and characterization of yapsin 2. Experiments using a series of internally quenched fluorogenic peptide substrates (IQ substrates) show that yapsin 2 cleaves carboxyl to single basic amino acid residues (Lys or Arg). In substrates with doublets and triplets of basic residues, the enzyme cleaves carboxyl to the most COOH-terminal basic residue. These results indicate specificity for a basic residue at P 1 , but discrimination against basic residues at the P 1 Ј position. This specificity explains how overexpression of this monobasic-specific enzyme can suppress the loss of the dibasic-specific enzyme, Kex2p. The pH profile of yapsin 2 activity showed a marked substrate dependence, suggesting that the enzyme may have significant activity at near neutral pH on certain substrates.
Materials and Reagents-Fast-flow Q-Sepharose, chelating Sepharose 6B, and Sephacryl S-300 were purchased from Amersham Pharmacia Biotech. Chelating Sepharose 6B was charged with Cu 2ϩ by passing a solution of 25 mM CuSO 4 through the column. IQ substrates were synthesized as described (22)(23)(24). Oligonucleotides were from Operon Technologies, Inc.
Plasmids-DNA manipulations followed standard methods (25). For overexpression of YPS2, a fragment containing the YPS2 structural gene (5) was placed under control of the TDH3 promoter by insertion into the multicopy (2 m) vector pG5 (26). A BglII site was introduced upstream of the HindIII site and an XhoI site was introduced downstream of the ApaI site in YPS2 by the polymerase chain reaction (primer sequences were 5Ј to 3Ј: GGAAGATCTAAGCTTTTGTCGTTAT-TCGC and GGAAGATCTCGAGGGCCCAGTACCGTTGGCAA). The product of this reaction was subcloned into pBS SK(Ϫ) (Stratagene) with a BglII linker inserted at the SmaI site in the polylinker. In the resulting plasmid, the HindIII-ApaI fragment was replaced with the corresponding fragment from the original YPS2 clone. The resulting plasmid was digested with BglII and XhoI and the YPS2 fragment ligated to pG5 which had been cut with BglII and SalI, placing the YPS2 structural gene under control of the TDH3 promoter. This final expression plasmid was called pG-MKC7.
Purification Procedure-Strain HKY26 containing plasmid pG-MKC7 was grown to a density of 6 ϫ 10 7 cells/ml in 3 liters of synthetic complete minus uracil. Cells were harvested and washed by centrifugation at 3,000 ϫ g. 15 ml of packed cells were resuspended with 24 ml of Buffer A (50 mM HEPES, pH 7.5, 10 mM EDTA, 0.5 mM phenylmethylsulfonyl fluoride, 0.1 mM TPCK, 0.1 mM TLCK, 1 mM benzamidine hydrochloride), mixed with 15 ml of glass beads (0.5 mm), and broken by vortexing 60 ϫ 20 s at 4°C (all subsequent steps were carried out at 4°C). Broken cells were centrifuged for 3 min at 500 ϫ g to remove cell debris (clear lysate). The supernatant fraction was centrifuged for 1 h at 100,000 ϫ g, and the resulting pellet was washed with 50 ml of Buffer A containing 0.5 M NaCl by recentrifugation at 100,000 ϫ g for 30 min. The final pellet (crude membrane fraction) was solubilized in 50 ml of Buffer A containing 1% Triton X-100, incubated for 30 min on ice, and centrifuged at 100,000 ϫ g for 1 h. The pellet was back-extracted with 30 ml of Buffer A and the two supernatant fractions were pooled (solubilized membrane fraction) and used for further purification.
The solubilized membrane fraction was applied to a 40-ml (3 ϫ 8 cm) Fast Flow Q-Sepharose column equilibrated with Buffer B1 (50 mM HEPES, pH 7.0, 0.5 mM EDTA, 10 mM NaCl, 0.5% Triton X-100). The column was washed with 2 column volumes of Buffer B1 and 8 column volumes of Buffer B2 (Buffer B1 minus EDTA) and eluted with 200 ml of Buffer C (Buffer B2 plus 0.2 M NaCl). 2 l of each fraction (10 ml) were diluted 10-fold and assayed, and fractions containing activity were pooled. The pool was adjusted to a final pH of 7.5 and a final concentration of 0.3 M NaCl by adding one-half volume of 50 mM HEPES (pH 9.0) containing 0.5 M NaCl and applied to a Cu 2ϩ -charged chelating Sepharose column (1.8 ϫ 6 cm) equilibrated with Buffer D (50 mM HEPES, pH 7.5, 0.1% Triton X-100, 0.3 M NaCl), washed with 50 ml of buffer D and 50 ml of Buffer E (50 mM HEPES (pH 7.5), 0.7% CHAPS, 0.3 M NaCl), and eluted with 50 ml of Buffer F (50 mM sodium acetate, pH 5.5, 0.7% CHAPS, 0.3 M NaCl). 2 l of each fraction (2.5 ml) were diluted 10-fold and assayed. Fractions containing activity were pooled (ϳ30 ml) and concentrated with a Centriprep-100 cartridge (Amicon, Inc.) to a final volume of ϳ400 l. The concentrated sample was split in half, and each half was applied to a Sephacryl S-300 (1 ϫ 49 cm) gel filtration column equilibrated with Buffer E. 2 l of each fraction (860 l) were diluted 10-fold and assayed. Fractions containing activity (see legend to Fig. 2B) were pooled and concentrated using a Centriprep-100 cartridge to produce the final fraction. Protein was measured using the BCA assay (Pierce). SDS-PAGE was carried out by the method of Laemmli (27).
Sequencing of Cyanogen Bromide-cleaved Yapsin 2-Yapsin 2 (8.7 g, 120 pmol) was cleaved by overnight incubation in the dark at room temperature in 70% formic acid containing 1% CNBr, and subjected to 10 -20% Tricine gradient gel electrophoresis. After electroblotting to ProBlott membrane (Applied Biosystems), one of two major bands was subjected to automated Edman degradation by the Protein and Carbohydrate Core Facility at the University of Michigan.
Determination of Cleavage Sites in IQ Substrates-0.04 g/ml yapsin 2 and 200 ϳ 400 pmol of IQ substrate were mixed and incubated in standard reaction buffer for 60 min. The reaction was stopped by addition of 0.1 N NaOH and subjected to automated Edman degradation by the Protein and Carbohydrate Core Facility at the University of Michigan.
Determination of V max /K m for IQ Substrates-V max /K m was determined using pseudo-first order kinetics. At substrate concentrations much lower than K m , product formation follows a simple first order curve (28). Reactions (600 l) were started by addition of yapsin 2 (final concentration, 0.02 to 1 g/ml). 100-l samples taken at times between 0 and 40 min were quenched with 600 l of 1 M Tris-HCl (pH 10). Raw fluorescence data were fitted to an exponential curve with floating end points using Kaleidagraph version 3.0. Substrate concentrations were between 1 and 2 M. All progress curves had r Ͼ 0.99 (Pearson's r value).
The pH dependence of V max /K m was measured in reaction buffer containing 20% sucrose, 10 mM CaCl 2 , and 0.007% CHAPS to prevent the loss of the activity during incubation at each pH (see "Results"). Theoretical curves for bell-shaped pH V max /K m profiles were calculated by nonlinear regression using Kaleidagraph as previously published (29).

RESULTS
Purification of Yapsin 2-Yapsin 2 was purified from an overexpressing strain lacking the YPS1 gene (see "Experimental Procedures"). The major proteolytic activity (70% of total activity) was recovered in the membrane fraction (100,000 ϫ g pellet) and solubilized by 1% Triton X-100 (5). Detergent extraction resulted in approximately 10-fold purification relative to the cleared lysate (data not shown). Further purification of yapsin 2 from the solubilized membrane fraction is described in detail under "Experimental Procedures." Although pepstatin A affinity column chromatography has been used to purify several aspartyl proteases, including yapsin 1 (30,31), less than 10% of yapsin 2 activity bound to the column and the yield of enzyme activity upon elution was extremely low. Therefore, we purified yapsin 2 using conventional column chromatography. Following anion exchange separation on Q-Sepharose, step elution from Cu 2ϩ -charged chelating Sepharose resulted in a major, diffuse band migrating at ϳ200 kDa that was visualized by SDS-PAGE (Fig. 1A, lane 4) and coincided with the peak of activity (data not shown). Note that although equal amounts of activity (10 5 units, equivalent to 6 g of yapsin 2) were loaded in lanes 2 (detergent-solubilized membranes), 3 (Q-Sepharose pool), and 4 (chelating Sepharose 6B peak) in Fig. 1A, the heterogeneity and diffuse migration of the yapsin 2 protein prevented the band from being obvious until elution from Q-Sepharose. The chelating Sepharose 6B peak was subjected to Sepharose S-300 gel filtration (Fig. 1A, lane 5), and the peak of yapsin 2 activity coincided with the appearance of this diffuse band which was separated from other, lower molecular weight polypeptides that eluted later (Fig. 2, A and B). The peak fractions were pooled to avoid the lower molecular weight polypeptides and the pooled enzyme was subjected to both SDS-7.5% PAGE and SDS-15% PAGE (Fig. 1, A, lane 5, and B). The protein with ϳ200 kDa molecular mass in the pooled fraction was judged to be Ͼ90% pure by Coomassie Blue staining ( Fig. 1 A, lane 5, and B). As NH 2 -terminal sequence of this polypeptide could not be detected, we determined the sequence of a fragment generated by cyanogen bromide cleavage. The sequence obtained corresponded to yapsin 2 residues 269 -275 after Met 268 (data not shown) (5), demonstrating that the purified polypeptide migrating diffusely at ϳ200 kDa was encoded by the YPS2 gene. 450 g of yapsin 2 protein were obtained from 120 ml of Triton X-100-solubilized membranes, representing a lysate of 20 ml of packed cells from a 4-liter culture (see "Experimental Procedures"). The final specific activity was 2 ϫ 10 7 units/mg, corresponding to a 22-fold purification from the Triton X-100-solubilized membranes of the overexpressing strain (Table I), and equivalent to approximately 200-fold purification from the initial cell lysate with an overall yield of 9.9% (data not shown). Consideration of the fact that yapsin 2 activity is already overexpressed approximately 150 -300-fold in HKY26[pG-MKC7] (12) leads to the conclusion that yapsin 2 is relatively inabundant in wild-type cells.
pH Activity Profile and Pepstatin A Sensitivity-Yapsin 2 contains two potential catalytic aspartic amino acids found in aspartyl proteases (5,32). In aspartyl proteases, titration of these residues is responsible for the bell-shaped pH dependence of the specificity constant (k cat /K m or V max /K m ) exhibited by aspartyl proteases (29,33). Aspartyl proteases are also characterized by sensitivity to pepstatin A (32). Therefore, we examined whether purified yapsin 2 exhibited the characteristic pH dependence and pepstatin A sensitivity of aspartyl proteases.
Prior to examining the pH dependence of the proteolytic activity, we assessed the effect of pH on yapsin 2 stability. Yapsin 2 lost activity rapidly at pH values Ͼ5 or Ͻ4 during incubation at 37°C. The t1 ⁄2 values for loss of activity at 37°C were 2 min at pH 6.0, 1.5 min at pH 7.0, 1.2 min at pH 8.0, and 4 min at pH 3 (data not shown). Addition of 20% sucrose and 10 mM CaCl 2 prevented this loss of activity during incubation at 37°C. Under these conditions, no loss of activity was observed at pH 4.0, 5.0, 6.0, or 7.0 during a 30-min incubation, although the enzyme still exhibited significant instability outside this pH range (t1 ⁄2 ϭ 27 min at pH 3.5 and 15 min at pH 8.0, data not shown). The pH dependence of V max /K m was measured under the conditions and over a pH range in which yapsin 2 was stable (Fig. 3). Yapsin 2 exhibited a bell-shaped pH response with an acidic pH optimum similar to that of other aspartyl proteases when IQ substrate 2, with the sequence SLDKRE-AEA, was used (Fig. 3A). Data were fitted to an equation for a bell-shaped curve using nonlinear regression, resulting in pK a values of 3.8 and 4.3 for the two ionizing groups with a pH optimum at approximately 4.0.
A radically different pH dependence of V max /K m was observed when IQ substrate 21 was used (Fig. 3B). This substrate, with the sequence HHQKLVF, corresponds to the sequence of the ␣-secretase site in human ␤APP (22) and exhibits a pH rate profile with apparent pK a values of 5.7 and 6.7 with  a pH optimum of 6.0. Additionally, the pH dependence observed with this substrate may be more complicated than a simple bell-shaped curve generated by two ionizable groups, because the data diverged from the model at high pH. Cleavage of substrate 21 by purified yapsin 2 was inhibited by high concentrations of pepstatin A at pH 6.0. Therefore, efficient cleavage of the ␣-secretase substrate at high pH is likely to depend on yapsin 2 and not on some contaminating protease in the preparation. Furthermore, substrate 21 was actually a better substrate than substrate 2 at pH 6.0, indicating that yapsin 2 does not simply assume a partially active, nonspecific conformation at this pH. Instead, the optimal pH for cleavage by yapsin 2 was highly sensitive to substrate sequence in addition to the ionization state of the two catalytic aspartic acids, an observation also made with human immunodeficiency virus-1 protease and certain substrates (29).
Pepstatin A inhibited yapsin 2 with a relatively high apparent K i , 4.6 M, that may reflect the preference of the enzyme for hydrophilic residues in substrate sequences (Fig. 4, see below). Leupeptin (Fig. 4), TLCK, TPCK, phenylmethylsulfonyl fluoride, and E64 (data not shown) did not inhibit yapsin 2 activity. Addition of EDTA, EGTA, Cu 2ϩ , Mg 2ϩ , Ca 2ϩ , or Mn 2ϩ had no effect on the activity (data not shown), despite the fact that yapsin 2 bound to Cu 2ϩ -charged chelating Sepharose.
Substrate Specificity-The natural substrates of yapsin 2 and yapsin 1 are still unknown. However, overexpression of either YPS1 or YPS2 suppressed the sterility of a kex2 null mutant (5,6), suggesting that these enzymes can cleave pro-␣ factor at the Kex2 cleavage sites. In addition, combined deletion of the YPS1 and YPS2 genes blocked cleavage of human ␤APP at the ␣-secretase site, suggesting that human ␤APP is a substrate of yapsin 2 (12,15). Therefore, the substrate specificity of yapsin 2 was tested using a series of IQ substrates based on the pro-␣ factor sequences (set A in Table II) and human ␤APP sequences (set B in Table II). Cleavage sites were determined by automated Edman degradation (Table III, "Experimental Procedures").
Set A tested the ability of yapsin 2 to cleave sequences containing 1, 2, or 3 basic amino acid residues. Yapsin 2 cleaved all sequences in set A, except for the one containing a Pro-Arg site, with similar V max /K m values, corresponding to a k cat /K m of ϳ10 7 s Ϫ1 M Ϫ1 , assuming a molecular weight of 64,000 and that all yapsin 2 molecules were active. In every case where cleavage was observed, the cleavage site was carboxyl to the most COOH-terminal basic residue, Arg, Lys, or ornithine, revealing that yapsin 2 prefers a basic residue at P 1 but fails to discriminate between structurally dissimilar basic residues at P 1 (Table III, substrates 3, 5, 6, 8, 10, 11, and 13). In addition, these results demonstrate that yapsin 2 exhibits essentially no selectivity at P 2 with the exception of excluding Pro at P 2 . The further observation that cleavage was always seen carboxyl to the most COOH-terminal basic residue in sequences with multiple basic residues suggests strongly that basic residues are excluded from P 1 Ј by yapsin 2 (Table III,  In the case of set B (Table II), six IQ substrates based on human ␤APP cleavage sites were tested. Significant cleavage was observed only in substrates 15, 17, and 21, which contained a Lys residue, consistent with a requirement by yapsin 2 for a basic residue at P 1 . Cleavage in each case tested occurred carboxyl to the Lys residue (Table III, substrates 15 and  21). At pH 4.0, V max /K m for these substrates was 2 orders of magnitude less than for the substrates based on pro-␣ factorbased sequences, although as pointed out previously (see Fig.  3), V max /K m for substrate 21 increased by ϳ20-fold at pH 6.0. The lower cleavage efficiency for set B versus set A substrates was probably due to the effect of sequence context rather than FIG. 3. pH dependence of yapsin 2 activity. A, pH dependence of yapsin 2 cleavage of IQ substrate 2 (A) and IQ substrate 21 (B). Assay mixtures contained 20% sucrose, 10 mM CaCl 2, 0.007% CHAPS, 50 mM sodium citrate (pH 3.5, 4.0, and 4.5), sodium Mes (pH 5.0, 5.5, 6.0, and 6.5), or sodium HEPES (pH 7.0). Incubation times for reactions at pH 4.0 to 7.0 were 30 min and 10 min for reactions at pH 3.5 and were chosen so that the enzyme was stable during the reaction (Յ5% loss of activity for pH 4.0 -7.0, ϳ20% loss of activity for pH 3.5; see text for details). Reactions were run under pseudo-first order conditions (substrate concentration below 2 M). Theoretical curves for bell-shaped pH V max /K m profiles were calculated by nonlinear regression using Kaleidagraph as described under "Experimental Procedures." V max /K m values were normalized to 1 g/ml yapsin 2. to a preference for Arg over Lys at P 1 , because yapsin 2 cleaved Lys-Arg (substrates 1, 2, 3, and 5), Arg-Arg (substrate 8), Lys-Lys (substrate 6), Arg-Lys (substrate 10), and Lys-Orn (substrate 7) sites in set A with comparable efficiencies (Table  II). Yapsin 2 was capable of cleaving at the human ␤APP ␣-secretase cleavage site, consistent with the observed cleavage of ␤APP when expressed in yeast (12)(13)(14)(15). Yapsin 2 did not cleave the ␤-secretase substrate, substrate 15, correctly (i.e. carboxyl to the Met residue) and did not cleave the ␥-secretase substrate, substrate 20, at all (Table II).

DISCUSSION
The cell-surface aspartyl protease yapsin 2 was purified approximately 200-fold in its native GPI-anchored form from the initial clear lysate made from the overexpressing strain, demonstrating that it is not necessary to truncate such an enzyme in order to purify it, as was done in the case of yapsin 1 (30). The purified enzyme migrated diffusely with a mean molecular mass of ϳ200,000 on SDS-PAGE, although the molecular mass predicted from the sequence of the YPS2 gene is only 64 kDa (5). The high apparent molecular weight and the presence of nine potential Asn-linked glycosylation sites in the yapsin 2 sequence (5) suggested that yapsin 2 was highly glycosylated. Consistent with this expectation, yapsin 2 was found to bind tightly to a concanavalin A column (data not shown). Digestion of yapsin 2 with N-glycanase, which removes Asn-linked oligosaccharides (34), resulted in an increased mobility in SDS-PAGE, but not to the predicted molecular weight. Furthermore, the band was still diffuse (data not shown), suggesting that yapsin 2 also contains heterogeneous Ser/Thr-linked oligosaccharides. Yapsin 2 eluting late in Sephacryl S-300 gel filtration (Fig. 2B) exhibited higher mobility in SDS-PAGE than early eluting enzyme, also consistent with extensive, heterogeneous glycosylation. A high degree of heterogeneous glycosylation is characteristic of yeast periplasmic and cell wall proteins and is a feature of yapsin 1 as well (8,35).
Yapsin 2 exhibits a bell-shaped pH dependence of V max /K m and is sensitive to pepstatin A as found for other aspartyl proteases (29,32,33). The bell-shaped pH profile of aspartyl proteases has been shown to depend on the pK a values of two conserved catalytic aspartyl residues; one of these is deprotonated with pK a between 1 and 3, and the other protonated with a higher pK a of approximately 4.5 (29,32). The pK a values measured for cleavage by yapsin 2 of IQ substrate 2, whose sequence is based on one of two Kex2p cleavage sites in pro-␣ factor, were 3.8 and 4.3, resulting in a very narrow pH optimum of approximately 4.0 (Fig. 3A). Because care was taken to measure the pH dependence under conditions in which the enzyme was stable, the pH profile most likely reflects the intrinsic pH dependence of the reaction between free enzyme and substrate as opposed to an effect of pH on folding. It is possible that the high value of the lower pK a observed with IQ substrate 2 may reflect titration of an acidic side chain in the substrate itself. Indeed, the quite different pH dependence of cleavage of IQ substrate 21 (Fig. 3B), presumably due to titration of His residues in the substrate, demonstrates the existence of dramatic substrate-dependent effects on catalysis by yapsin 2. Human immunodeficiency virus-I protease has been shown to exhibit different pH dependence of k cat and K m for substrates that differ by a single acidic side chain at P 2 Ј, although the shift in pH optimum for k cat /K m was not nearly as large (29). The shift of the pH optimum of yapsin 2 to near neutrality with substrate 21, which represents the ␣-secretase cleavage site in human ␤APP, is likely a relevant factor in the efficiency of cleavage of human ␤APP by yapsin 2 in yeast in vivo (12). Furthermore, the ability of yapsin 2 to cleave certain substrates effectively in a higher pH range may be relevant to the physiological function of yapsin 2 as a processing enzyme in the secretory pathway or at the cell surface in yeast. Further study will be needed to determine how substrate sequence and pH affect substrate binding and catalysis by yapsin 2.
Summarizing the analysis of our specificity studies, yapsin 2 appears to require a basic residue at P 1 . This requirement seems to be based on the charge rather than on the structure of the side chain, because structurally dissimilar basic side chains gave comparable V max /K m values. Yapsin 2 does not appear to cleave between basic residues, however, suggesting that the enzyme excludes basic residues at P 1 Ј. This distinguishes yapsin 2 from yapsin 1, which has been shown to cleave between basic residues in certain substrates (9). However, analysis of strains lacking YPS1, YPS2, or both genes indicates that yapsin 2 and yapsin 1 are at least partially physiologically redundant in yeast and therefore must cleave at least an overlapping pool of substrates (7). In addition, of 25 residues predicted on the basis of a structural model of yapsin 1 to interact with substrate side chains from P 6 to P 6 Ј (11), 20 are conserved in yapsin 2 (5). Direct comparison of the two enzymes using identical substrates should lead to a better understanding of the relative specificities of the two enzymes.
The distinction between the specificity of yapsin 2 and that of the serine protease Kex2p is more extensive. First, Kex2p is  (24) and set B, on the human ␤APP sequence (22). In set B, substrate 15 contains the ␤-secretase cleavage site, substrates 16 -18 contain single or double substitutions corresponding to an Alzheimer's disease-causing double mutation that increases ␤-amyloid peptide production in transfected cells, substrate 20 contains the ␥-secretase site, and substrate 21 contains the ␣-secretase cleavage site (23). Assays were performed in 0.1 M sodium citrate (pH 4.0) containing 5 mM CaCl 2 , 0.07% CHAPS, and 2 M substrate. V max /K M values were measured as described under "Experimental Procedures" and normalized to 1 g/ml yapsin 2. Only progress curves with r Ͼ 0.99 (Pearson's r value) were used in determination of V max /K M values. Reported values are the mean of at least three trials, and errors are reported as standard deviations. ␤; norleucine; O; ornithine; B, EDANS conjugated to the ␥-carboxylate of glutamic acid; J, DABCYL conjugated to the ⑀-amino group of lysine; Ac-, acetyl-. No detectable cleavage: V max /K M was less than 0.0002 min Ϫ1 . Basic residues at which cleavage might be expected are shown in bold. highly specific for Arg at P 1 . Substitution of Lys for Arg at P 1 in otherwise identical substrates not only reduces k cat /K m by ϳ500 -3600-fold, it also results in a change in the rate-limiting step of the reaction from deacylation to acylation (36). Whereas Kex2p exhibits large decreases in k cat /K m when residues other than Lys or Arg are present at P 2 , yapsin 2 appears to be largely insensitive to the nature of the P 2 residue, with the exception of Pro (see below). For example, on IQ substrates 3 and 13, Kex2p exhibited k cat /K m values of 2.5 ϫ 10 7 and 5.6 ϫ 10 4 M Ϫ1 s Ϫ1 , respectively, a 450-fold difference (23). In contrast, yapsin 2 exhibited essentially identical V max /K m values (and therefore k cat /K m values) for the two substrates (Table II). Conversely, Kex2p is relatively permissive for Pro at P 2 , which reduces k cat /K m by only ϳ7-fold (24), whereas in the case of yapsin 2, V max /K m was reduced by Ͼ10 5 -fold by substitution of Pro at P 2 . The differences between yapsin 2 and Kex2p are not surprising in the sense that the enzymes represent distinct mechanistic and evolutionary classes of proteases. However, the fact that both yapsin 2 and yapsin 1 can suppress phenotypic defects of kex2 null mutant strains and that null mutations of the YPS1 and YPS2 genes result in more severe phenotypes when combined with kex2 null mutations suggests that the three enzymes cleave at least some common substrates in vivo (5). However, whereas Kex2p can be accurately categorized by its specificity for pairs of basic residues, the data presented here argue that yapsin 2 is a monobasic specific enzyme. What could be responsible for the ability of yapsin 2 to cleave Kex2 substrates accurately? The exclusion of basic residues from the P 1 Ј position in yapsin 2 most likely accounts for the ability of the enzyme to substitute for Kex2p in vivo, in that this feature directs the enzyme to cleave carboxyl to runs of basic residues. This is particularly important in the case of cleavage of pro-␣ factor, where cleavage must occur carboxyl to the Lys-Arg sites in order that the Ste13 dipeptidylaminopeptidase can complete the processing of the ␣-factor NH 2 terminus (2). Yapsin 2 also shares with Kex2p other features of proteolytic processing enzymes. Both enzymes are rare, and both exhibit high k cat /K m values, Ն10 7 M Ϫ1 S Ϫ1 , for their best substrates. These features serve to promote simultaneously high selectivity and high efficiency in processing.
A further distinction between yapsin 2 and Kex2p is their apparent localization; yapsin 2 to the plasma membrane (12) and Kex2p to the late Golgi or trans Golgi network (21,37,38). In addition to YPS1 and YPS2 there are three other related open reading frames in the yeast genome that encode potential GPI-anchored aspartyl proteases, YPS4, YPS6, and YPS7 (a fourth related gene, YPS5, contains only a partial open reading frame). 1 These enzymes appear to constitute a distinct family of aspartyl proteases, which are most likely localized at the cell surface through GPI anchors. One possible function for these enzymes is in the processing of cell wall precursors or precursors of enzymes involved in cell wall synthesis or remodeling. The two enzymes of this family that have been purified and characterized, yapsin 2 and yapsin 1, have distinguishable specificities. Therefore, the three as yet uncharacterized homologues could have yet more divergent specificities. A deeper understanding of the roles of these aspartyl proteases in protein processing will require identification of their natural substrates as well as characterization of their expression, specificity, and localization.