Three Proteolytic Systems in the Yeast Saccharomyces cerevisiae”

The proteases of yeast that were first identified and character- ized were detected biochemically (1-4) and proved to be of vacuolar (lysosomal) origin (5, 6). An additional nonvacuolar set was defined genetically by mutations that caused incomplete proteo- lytic processing of precursors to killer toxin and the pheromone a-factor (7-13). Once the dominant vacuolar proteases could be eliminated by mutation, a substantial number of new enzyme activities, including additional endoproteinases, carboxypepti-dases, aminopeptidases, and dipeptidyl aminopeptidases, were detected biochemically (14-16). I will concentrate here on three groups of proteases: the cytosolic proteasome, vacuolar proteases, and proteases located within the secretory pathway. Not only is a deal known about the functions of these enzymes, but also, where zymogens are involved, several of the maturation pathways have been elucidated. The compartments and their contained cleavages occur during maturation of Prclp. After signal peptidase cleavage (46), an amino-terminal peptide is removed in the late Golgi or the vacuole (47-49) in a reaction catalyzed by PrA (48). In vitro data suggest that a third cleavage is catalyzed by PrB (50), a reaction that can bypass the PrA-catalyzed cleavage and is presumed to account for phenotypic lag (42,43).

a Gene names are eiven accordine to eenetic convention.' The primary name protease I11 predict yes is iven first; assignGents after theynits1 discovery are given in parentheses. 'DPAP-A shows marked homology to DPAP-B (56). Since DPAP-B is inhibited bv Dhenvlmethvlsulfonvl fluoride. I infer DPAP-A will he as well, although eGidence-is lackhg.

Proteinase A
PrA, encoded by the PEP4 gene, is an aspartyl proteinase with similarity to the two-domain class of aspartyl proteinases that includes pepsin, renin, cathepsin D, and penicillopepsin (27,28). This glycoprotein of 42 kDa carries two asparagine-linked glycosyl side chains; its four cysteines are thought to form disulfide bonds (29). Three proteolytic cleavages occur during the posttranslational maturation of Pep4p. After removal of the signal sequence by signal peptidase (30), another 47 amino acids are removed from the NH, terminus late in the Golgi or in the vacuole (31). This intramolecular reaction is autocatalytic, since mutational change of either aspartate residue of the active site (Asp -+ Asn) results in nearly complete failure of processing.' The final cleavage removes 7 amino acids; it is catalyzed by PrB (32).3 The final "intermediate" has not been detected kinetically; it accumulates in a p r b l A mutant (32).

Proteinase B
PrB, encoded by the P R B l gene, is a member of the subtilisin family of serine proteases, which includes proteinase K (33). Despite the presence of three potential tripeptide acceptors for Asn-linked glycosyl chains within the catalytic region (five in the precursor), the 31-kDa enzyme carries no Asn-linked glycosyl side chains (34); since it is a glycoprotein we presume it carries 0-linked mannose (35).
Four proteolytic cleavages occur during maturation of Prblp ( Fig. 1). After removal of the signal sequence by signal peptida~e,~ about 260 additional amino acids are removed from the NH, terminus in the endoplasmic reticulum (36,37). This reaction is intramolecular and autocatalytic, for it does not occur if the active site serine (Ser + Ala) or aspartate (Asp + Asn) is changed by mutation (38).4 The third cleavage, which converts the 40-kDa product of autocatalysis to a 37-kDa intermediate, is catalyzed by PrA (36). The final cleavages occur late, in either the late Golgi or, more likely, the vacuole (36). We infer that the fourth and final cleavage, which converts the 37-kDa species to the mature 31-kDa mature enzyme, is autocatalytic in nature, because a mutation, prbl-628, profoundly reduces the rate of production auto, autocatalytic; KexZp, Kex2 endoproteinase; *, active in intramolecular autocatalysis; **, proteolytically active on other molecules. PrB is thought to catalyze two cleavages that bypass the requirement for PrA, giving rise to the phenomenon of phenotypic lag (42,43,50). The two suspected cleavages are signified by (PrB) adjacent to curued reaction arrows. Not to scale.
of 31-kDa PrB from the 37-kDa species (36). The prbl-628 mutation results in a change of Ala'52 to Thr15',4 which lies close to the essential Asn'" of the oxyanion hole (39, 40) and within 10 A of the active site serine (41). The defect is in active site function, not substrate properties, since the prbl-628 mutation is completely recessive and all the antigen produced in the heterozygote is of the mature form.4 We presume that protease B itself can also catalyze this cleavage in trans, albeit less efficiently, and that this is the molecular explanation for the phenomenon called phenotypic lag (42,43).

Carboxypeptidase Y
CpY, encoded by the PRCl gene, is a serine protease. This 61-kDa glycoprotein shows broad substrate specificity and carries four Asn-linked glycosyl side chains (44,45). Three proteolytic cleavages occur during maturation of Prclp. After signal peptidase cleavage (46), an amino-terminal peptide is removed in the late Golgi or the vacuole (47)(48)(49) in a reaction catalyzed by PrA (48). In vitro data suggest that a third cleavage is catalyzed by PrB (50), a reaction that can bypass the PrA-catalyzed cleavage and is presumed to account for phenotypic lag (42,43).

Carboxypeptidase S
CpS, encoded by the CPSl gene, is a metal ion-dependent carboxypeptidase. We know little of its synthesis or possible precursor(s). However, its synthesis is not dependent on PrA activity (51).

Aminopeptidase I
ApI, encoded by the LAP4 gene, is a metalloexopeptidase. It is a glycoprotein of 640 kDa and contains 12 subunits (52). Because pep4 mutations greatly reduce the activity of ApI (53), we expect a zymogen precursor to ApI, with production of active enzyme catalyzed by PrA. Whether the large antigenic species reported for cells in steady state (54) corresponds to such a precursor is not yet known. The LAP4 gene does not appear to encode a signal sequence of the type normally responsible for entry into the endoplasmic reticulum (54).

Aminopeptidase Co
ApCo is a 100-kDa metalloexopeptidase that requires Co2+ (55). Nothing has been reported about its gene or its synthesis.

Dipeptidyl Aminopeptidase B
DPAP-B, encoded by the DAP2 gene, is a 120-kDa integral membrane glycoprotein of the yeast vacuole (56). An apparently typical hydrophobic helical domain near the NH2 terminus is postulated to function as an internal signal sequence and transmembrane anchor for this presumed type I1 integral membrane glycoprotein. Production of active DPAP-B does not require PrA activity (56).

Maturation Information
Ignoring the contributions of signal peptidase, one can summarize what is known about production of these vacuolar proteases as follows. Some (PrA, PrB, CpY, ApI) but not all (CpS, DPAP-B) of the vacuolar proteases are synthesized as inactive precursors. All of the proteolytic cleavages of the maturation pathways are self-catalyzed or catalyzed by PrA or PrB. Thus, all of the information required for the maturation cleavages and the ultimate activities of all of these proteases is self-contained within this set of protease precursors. One can extend this concept of self-sufficiency of 'maturation information to include other vacuolar hydrolases, since production of activity of the vacuolar species of alkaline phosphatase (51), trehalase (57), and RNase(s) (51) is also dependent on PrA. The roles of PrA and PrB are not equally important in these maturation pathways. PrA plays an essential role in the production of itself, PrB, CpY, and ApI, among proteases, and alkaline phosphatase, trehalase, and RNase(s), among other hydrolases. The requirement of PrB for production of activity is limited to production of PrB activity itself. In the other cases examined, PrB seems to function to "trim" the ends of enzymatically active species to give enzyme of "mature" size. The functional significance of such trimming is not known.

Functions o f Vacuolar Proteases
Much of our understanding of the functions of these enzymes stems from analyses of mutant strains. Structural gene mutations for CpY, CpS, or PrB (prcl, cpsl orprbl, respectively) eliminate only the activity of the enzyme encoded by the affected gene, making analysis straightforward. Mutations in PEP4, the structural gene for PrA, can give different phenotypes, depending upon the severity of the mutation. Insertion or nonsense mutations (e.g. pep4-3) completely prevent production of active PrA. Its absence ensures failure of the cell to produce all enzyme activities in whose maturation pathways PrA participates, including PrB, CpY, ApI, alkaline phosphatase, RNase(s), and vacuolar trehalase (32,48,51,53,57,58). Missense mutations in PEP4 can cause a greater range of phenotypes, ranging from completely pleiotropic (pep4-11) to an effect only on PrA activity (pep4-625) (32). Mature sized (presumably autoprocessed) PrA antigen that failed to hydrolyze hemoglobin was found in the pep4-625 mutant. All other vacuolar hydrolase activities in this mutant were at wildtype levels, implying that PrA functioned properly in the maturation pathways (32,59). On the surface, the pep4-625 mutation allows one to assess the consequence of loss of activity of PrA alone. This seems unwarranted, however, since all in uiuo assays of PrA function are positive and only the in vitro test of PrA activity is negative. Thus caution must be exercised in evaluating analyses employingpep4-625 and, presumably, pral-1, which also appears to lack only PrA activity (60) (prul-1 is apep4 allele and should be so called by genetic convention; the sequence change has not been reported).
One can with some confidence assign certain functions to the vacuolar proteases. None of them is essential for cell viability. The two endoproteinases, PrA and PrB, but principally PrA, catalyze cleavages that result in maturation of precursors and activation of zymogens of other vacuolar proteases and hydrolases (see above). Since the enzymes show little specificity in cleavage of denatured peptides (61, 62), the specificity manifested in the maturation pathways is presumed to reflect conformational constraints.
All of the proteases probably participate in degradation of some proteins and peptides that turn over as a normal part of the life cycle (Kex2 protease (63); a-factor (64)) but are not involved in the initial steps in turnover of nonsense fragments (65), missense proteins (65, 66), or unassembled ribosomal proteins (67) or enzyme subunits (66). It is very clear that PrA and PrB activities are required for degradation of analog-containing proteins, for starvation-induced vegetative protein degradation, and for the ability to survive nitrogen starvation (PrA is particularly important for the last effect) (66).
The vacuolar proteases account for the bulk of the protein degradation that takes place when cells are starved for nitrogen in sporulation medium (66, 68). In the absence of PrB activity, the frequency of ascospore formation can range from nil to high, depending on conditions of preculture and background genotype (66, 68). Pleiotropic pep4 mutations eliminate ascospore formation (66, 68); whether this is caused by the low rates of protein degradation or the loss of other enzyme activities (RNase(s), trehalases, etc.) cannot be assessed. pral-1 homozygotes sporulate poorly (66), while pep4-625 homozygotes sporulate rather well5; these latter results are difficult to evaluate, as discussed above.
It is unclear whether vacuolar proteases contribute to degradation of fructose 1,6-bisphosphatase during catabolite inactivation, since there are reports on both sides of the issue (66, 69, 70). Vacuolar proteases appear not to be involved in other cases of catabolite inactivation that have been tested (71, 72).
In sum, in growing cells the vacuolar proteases catalyze maturation of precursors and participate in a limited amount of protein turnover. In response to stress, they catalyze massive amounts of protein degradation that facilitates cellular restructuring, particularly during sporulation.

Proteases of the Secretory Pathway
Signal Peptidase-Signal peptidase is an integral membrane protein that contains at least four subunits, one of which is glycosylated (73). The 18-kDa subunit is the product of the SECl1 gene, a not unexpected finding, since signal peptides of secreted proteins are not removed in the sed1 mutant at high temperature (74) and the predicted sequence of Secllp shows marked similarity to that of one subunit of canine signal peptidase (75).
Ken2 Endoprotease-Kex2 protease, encoded by the KEX2 gene, is an endoproteinase that cleaves on the COOH-terminal side of Lys-Arg or Arg-Arg paired basic residues (9, 76). It is a serine protease of the subtilisin class (77) and requires Ca2+ for activity. This glycoprotein carries both N-linked and 0-linked sugars. Kex2 protease is an integral membrane protein, anchored by a sub-COOH-terminal transmembrane domain in a compartment thought to be the late Golgi (78, 79). The KEX2 gene was first identified by mutations that prevented production of killer toxin (killer zpression) and caused sterility in cells of a mating type because of failure to produce the pheromone a-factor (7,8). The a-factor precursor contains four copies of the pheromone peptide sequence separated by spacers (Fig. 2). The cleavage catalyzed by Kex2 protease is shown in Fig. 2.
Two proteolytic cleavages occur during production of Kex2 protease. After removal of the signal sequence by signal peptidase (63), additional amino acids are removed from the NH, terminus (63) before the protein leaves the endoplasmic reticulum. The processing is thought to be intramolecular and autocatalytic; Lys-Arg sequences are found at residues 79-80 and 108-109, in a region that precedes the subtilisin homology, consistent with an autocatalytic cleavage mechanism (63).
Kexl Carboxypeptidase-Kexl carboxypeptidase, encoded by the KEXl gene, is a serine protease (12, 13). It appears to be specific for basic residues (13). This glycoprotein contains Nlinked sugars (13). Kexl protease is an integral membrane protein. There are no indications of a zymogen form for this enzyme. Preliminary results suggest a late Golgi location (80).
The KEXl gene was first identified by mutations that prevented production of killer toxin (7, 8). Unlike kex2 mutations, however, kexl mutations have no effect on fertility of cells of a mating type. The cleavages thought to be catalyzed by Kexl protease are shown in Fig. 2. The fertility of kexl mutants of a mating type depends on a-factor production solely from the COOH-terminal a-factor repeat (12), proving that Kexl protease, in keeping with its substrate specificity (13), catalyzes removal of the COOH-terminal Lys and Arg residues.
Dipeptidyl Aminopeptidase A-DPAP-A, encoded by the STE13 gene ( l l ) , is a membrane protein (11). The gene has been sequenced.6 The predicted polypeptide shows marked similarity to that of DPAP-B both in primary structure and in its gross topology (56). It is expected to be a type I1 glycoprotein, anchored in the Golgi by its sub-NH,-terminal internal signal sequence, 'E. Jones, unpublished data. C. A. Flanagan and J. Thorner, personal communication.

FIG. 2. Proteolytic cleavages in maturation of the a-factor (aF)
precursor. Cleavages occur in the Golgi in the order Kex2 endoproteinase, Kexl Carboxypeptidase, DPAP-A (86). Arrows identify the bonds cleaved within one spacer unit that lies between the a-factor COOH terminus, Met-Tyr, and the next NH, terminus, Trp-His.
with a large lumenal domain and a small cytosolic domain. By analogy with what is known about production of DPAP-B, no zymogen form is expected. DPAP-A catalyzes cleavage of Glu-Ala or Asp-Ala dipeptides from the a-factor precursor (Fig. 2). Yeast Aspartyl Protease III (Yap3 Protease)-A gene, YAP3, was described whose product, when overproduced, catalyzed cleavage of the a-factor precursor on the COOH-terminal side of paired basic residues when Kex2 protease activity was absent (81). Conceptual translation of YAP3 yields a protein with sequence similarity to the two-domain group of aspartyl proteases. Its topological features resemble those of Kex2 protease. It is predicted to have an NH,-terminal signal sequence and a sub-COOH-terminal transmembrane anchor; most of the protein should be within the compartmental lumen. In alignment with other aspartyl proteases, Yap3 protease has an extra -45 amino acids at its NH2 terminus (excluding the predicted signal sequence). Interestingly, there are two Lys-Arg pairs in this 45amino acid stretch. The second pair is at residues 66-67, within one amino acid of the NH, terminus (at 68 or 69) predicted by alignment. I predict that Yap3 protease will have a zymogen form that will undergo autocatalytic activation, probably in the endoplasmic reticulum. Neither the synthesis nor localization of Yap3 protease has been studied.

Functions of Proteases of the Secretory Pathway
Since the secll mutation causes temperature-sensitive lethality, signal peptidase is presumed to be an essential enzyme (74). None of the other proteases of the secretory pathway are essential for viability. These proteases act to process precursors to one or more secreted peptides, with a-factor pheromone and killer toxin being known end products. The normal substrate for Yap3 protease (when expressed at normal levels) is unknown.

Comparison of Synthesis of Vacuolar and Secretory Pathway Proteases
Comparison of the routes of production of vacuolar proteases and secretory pathway proteases, both of which traverse the endoplasmic reticulum and most of the Golgi during production, surfaces the interesting fact that all of the endoproteinases studied (Kex2 protease, PrA, PrB) are synthesized as zymogens that undergo an intramolecular, autocatalytic cleavage step. For Kex2 protease, autocatalysis removes an NH,-terminal propeptide, and the product is active before the protein leaves the endoplasmic reticulum. For PrB, autocatalysis likewise removes a long NH,terminal propeptide within the endoplasmic reticulum, but a COOH-terminal 60-amino acid "silencer" peptide prevents the product from being active. It is only later in the pathway, in the late Golgi or vacuole, that removal of the "silencer" peptide is initiated and the enzyme is activated.
As has been reported for some prokaryotic proteases (82-85), the propeptide is required if active PrB is to be produced. Its presence ensures that the three Asn receptors within the catalytic sequence do not become glycosylated and that PrA-catalyzed cleavage of the 40-kDa intermediate, rather than degradation, occurs.4 In addition to any functions the pro sequences may provide in inducing conformational changes to allow further processing or activity, they in effect buy time in a spatial continuum. A substantial fraction of the enzymes that pass through the endoplasmic reticulum section of the secretory pathway has paired basic residues that could render them susceptible to Kex2 protease (PrA, PrB, ApI, DPAP-A and -B, alkaline phosphatase, invertase, Kexl protease). Yet none are cleaved by Kex2 protease. Possibly these enzymes, immediately after translocation into the lumen of the (rough) endoplasmic reticulum, assume conformations that bury the basic pairs before KexP protease activates itself and becomes active on other molecules (possibly elsewhere in the endoplasmic reticulum). The fact that the PrA-catalyzed cleavages that finally lead to active PrA and PrB occur so late in the pathway, and possibly only within the vacuole itself, ensures that the secretory pathway itself is protected from the action of fairly nonspecific proteases that can be present at high levels.
Several of the vacuolar exopeptidases (and other hydrolases) are synthesized as zymogens, whereas none of the secretory exoproteases appears to be. It is noteworthy that the exopeptidases of the secretory pathway have very restricted substrate specificities whereas vacuolar exopeptidases show broad specificities. Possibly this difference, combined with the fact that vacuolar enzymes may he expressed to high levels, necessitates the existence of zymogens for the vacuolar exopeptidase species, since premature or inopportune expression of the vacuolar activities could be prejudicial to the secretory system.

Connections
Both the cytosolic proteasome and the vacuolar proteases participate in the massive protein degradations triggered by analog-containing proteins or by nitrogen starvation, especially in connection with meiosis and sporulation. It is unknown whether this co-participation simply reflects coincidence in timing of the responses of two separate systems or is an indication of a deeper interaction or interrelationship between the systems. And it is possible that the tantalizing hints of a proteolytic system within the endoplasmic reticulum that degrades unstable pro segments and unprocessed precursors4 signify another connection to the vacuolar system.
We are beginning to get a clear picture of the roles and functions within cells of these proteolytic systems as single entities. We can expect to ferret out any connections between them and gain a better understanding of the totality in the future.