Quantitative assessment of enzyme specificity in vivo: P₂ recognition by Kex2 protease defined in a genetic system

Abstract

The specificity of the yeast proprotein-processing Kex2 protease was examined in vivo by using a sensitive, quantitative assay. A truncated prepro-α-factor gene encoding an α-factor precursor with a single α-factor repeat was constructed with restriction sites for cassette mutagenesis flanking the single Kex2 cleavage site (-SLDKR↓EAEA-). All of the 19 substitutions for the Lys (P₂) residue in the cleavage site were made. The wild-type and mutant precursors were expressed in a yeast strain lacking the chromosomal genes encoding Kex2 and prepro-α-factor. Cleavage of the 20 sites by Kex2, expressed at the wild-type level, was assessed by using a quantitative-mating assay with an effective range greater than six orders of magnitude. All substitutions for Lys at P₂ decreased mating, from 2-fold for Arg to >10⁶-fold for Trp. Eviction of the Kex2-encoding plasmid indicated that cleavage of mutant sites by other cellular proteases was not a complicating factor. Mating efficiencies of strains expressing the mutant precursors correlated well with the specificity (k_cat/K_M) of purified Kex2 for comparable model peptide substrates, validating the in vivo approach as a quantitative method. The results support the conclusion that K_M, which is heavily influenced by the nature of the P₂ residue, is a major determinant of cleavage efficiency in vivo. P₂ preference followed the rank order: Lys > Arg > Thr > Pro > Glu > Ile > Ser > Ala > Asn > Val > Cys > AsP > Gln > Gly > His > Met > Leu > Tyr > Phe > Trp.

Substrate discrimination is central to the physiological function of enzymes (1). Much insight into the energetics of enzyme-substrate recognition has come from studies of serine proteases that mediate key pathways, including protein degradation, coagulation, and proprotein processing (2, 3). The substrate specificity of Kex2 protease (“kexin,” EC No. 3.4.21.61) from the yeast Saccharomyces cerevisiae is of particular interest because it is the prototype of the eukaryotic subtilisin-related proprotein processing enzymes that cleave sites consisting of pairs or clusters of basic residues (ref. 4 and see refs. 5 and 6 for reviews).

Although three-dimensional structures of Kex2 family enzymes are not available, kinetic studies of cleavage of model substrates (7–11), mutational analysis of substrate cleavage sites (12, 13), and modeling studies (14) suggest that like degradative subtilisins, Kex2 family enzymes interact with substrate P₁, P₂, and P₄ side chains¶ but are vastly more discriminating at each position. Studies of secreted, soluble Kex2 protease using peptidyl methylcoumarinamide (peptidyl-MCA) substrates revealed a highly efficient enzyme, with k_cat/K_M values for cleaving the best Lys-Arg substrates in the range of 1–5 × 10⁷ s⁻¹ M⁻¹ (7, 9, 10). These same studies suggested rules for recognition of P₁ and P₂ residues (7, 9). Kex2 prefers Arg over Lys at P₁ by up to a factor of 100- to 10,000-fold in k_cat/K_M (7, 9). A 130-fold decrease in k_cat with Lys at P₁ suggested that Arg is required at P₁ for achieving a high catalytic rate (7). In contrast, substitutions for Lys at P₂ resulted in increases in K_M without significantly altering k_cat (7). More recent kinetic studies have extended our understanding of recognition of the P₁ and P₂ residues and demonstrated the existence of an S₄ subsite in Kex2 (9, 10).

Until now, comparable detailed information about specificity in vivo has been lacking. Here, we introduce a method for analyzing Kex2 specificity in vivo that relies on the essential role of the enzyme in yeast mating. Localized to a late Golgi compartment (16, 17), Kex2 cleaves the α-mating pheromone precursor, pro-α-factor, carboxyl to Lys-Arg sites (Fig. 1A) (19, 22). Kex2 cleavage is essential for production of mature α-factor and therefore is essential for mating of α haploid cells with haploid cells of the a-mating type. A sterile α strain was constructed that lacks both the chromosomal genes encoding the α-factor precursors (23, 24) and the KEX2 gene. Mating competence can be restored by introducing plasmids encoding Kex2 and prepro-α-factor. The wild-type KEX2 gene is reintroduced on a single-copy plasmid under the control of its own promoter to preserve a normal level of expression. A second single-copy plasmid contains a modified MFα1 gene, MFα1–100, that encodes a truncated α-factor precursor (T-pαf) with a single α-factor repeat unit (Fig. 1B). Mating of the strain depends on cleavage of T-pαf by Kex2 at a single site that can be modified by cassette mutagenesis (Fig. 1C). The effects of alterations of the cleavage site on the efficiency of cleavage by Kex2 protease are determined by using a quantitative-mating assay that has an effective range of greater than six orders of magnitude (25). This permits a sensitive, quantitative assessment of the interaction of Kex2 protease with wild-type and mutant cleavage sites in vivo.

Figure 1.

Schematic structures of α-factor precursors and overview of the genetic system. (A) Signal peptidase cleavage of the major form of prepro-α-factor (165 residues), which is encoded by the MFα1 gene, results in production of 146 residue pro-α-factor (18). Kex2 protease cleaves pro-α-factor between the glycosylated “pro” domain and the first α-factor repeat carboxyl to the sequence Ser-Leu-Asp-Lys-Arg↓ and at three sites between the α-factor repeats carboxyl to the sequence Pro-Met-Tyr-Lys-Arg↓ (19). Production of mature α-factor requires exoproteolysis by Kex1 carboxypeptidase to remove C-terminal Lys and Arg residues from the internal repeats (20) and by Ste13 dipeptidyl aminopeptidase to remove N-terminal GluAla and AspAla dipeptides from each repeat (21). (B) The product of the MFα1–100 gene contains a single α-factor repeat. Production of mature α-factor requires cleavage by Kex2 at a single site. (C) Schematic of yeast strain ABY08 carrying plasmids YCpG-MFα1–100 and pAL5-KEX2. The Xba I and HindIII sites in YCpG-MFα1–100 were used for cassette mutagenesis of the P₂ Lys (Lys⁸⁴). An important feature of this system is that cleavage of both wild-type and mutant sites must occur in the correct “phase,” i.e., immediately following the P₁ Arg. This is because production of mature α-factor requires removal of two Glu-Ala dipeptides by the Ste13 peptidase.

MATERIALS AND METHODS

Plasmid Constructions.

Routine molecular biology techniques were performed as described (26). The backbone of pAL5-KEX2 was constructed from the plasmid pRS314 (27). Silent mutations were introduced into the unique XbaI and HindIII sites in the TRP1 gene by one-step, site-directed mutagenesis (8) by using a 4.3-kbp PCR fragment generated by mutagenic primers 5′-acagaggccgcagaatgtgcCctagattccg-3′ and 5′-gctaacataaaatgtaagctCtcggggctct-3′. A ≈3.9-kbp BamHI fragment of plasmid mYCpN3-KEX2 (17) containing the KEX2 promoter, ORF and 3′-untranslated sequences was inserted anticlockwise into a BamHI linker (New England Biolabs), replacing the existing polylinker. The KEX2-coding sequence contained a unique SacII site introduced at the Pro¹⁰²Arg¹⁰³ codons of the KEX2-coding sequence with the primer 5′-ccatgatttattcccgcgGaacgaccta-3′ by the method of Kunkel (28).

KEX2 knockout plasmid pAL2-kex2Δ2∷URA3 was constructed by inserting the 3,843-bp BglII–BamHI fragment of pNKY51, containing the URA3 gene flanked by direct repeats of Salmonella typhimurium hisG-sequences (29), into the BglII site of pKX11 (R.S.F., unpublished observations), replacing the entire KEX2-coding sequence.

Plasmid pBM258 is a derivative of YCp50 (30) that contains a 685-bp fragment with the bidirectional GAL1,10 promoter (31). A 530-bp BamHI to SalI fragment of pAB111 (gift of A. Brake, Chiron Corp.) containing the MFα1 ORF (23) was inserted into pBM258 generating plasmid pBM258-MFα1 in which the MFα1 structural gene is placed under GAL1-promoter control. Cleavage with HindIII and religation generated plasmid pBM258-MFα1-ΔHIII in which the first cleavage site was fused to the fourth mature α-factor repeat. A silent XbaI restriction site 5′ of the DNA encoding the Kex2 cleavage site was introduced (as above) with the oligonucleotide 5′-ctctcttttatctagaGaTaccccttc-3′ to generate plasmid YCpG-MFα1–100.

Yeast Strain Construction.

Yeast media and methods were essentially as described (32). SD is synthetic minimal medium containing 2% (wt/vol) glucose, SDC is synthetic complete medium with 2% (wt/vol) glucose, and SGC is synthetic complete medium containing 2% (wt/vol) galactose. Omissions from complete media formulations to create “drop-out” media are indicated as “-Ura,” etc. Strains are described in Table 1. Strains SEY6210 and SEY6211 (33) were mated and sporulated and tetrads dissected to generate ABY02. Strain JE104 containing plasmid pBM258-MFα1 was crossed with ABY02 and the resulting diploid was cured of pBM258-MFα1 and transformed with a 5,124-bp BamHI fragment of pAL2-kex2Δ2∷URA3 to delete the chromosomal KEX2 gene. The diploid was sporulated, and sterile kex2Δ2∷URA3 his3 HIS4 segregants were identified by complementation and cosegregation of Ura⁺ and cold sensitivity (34, 35). A MATα candidate was plated on 5-fluoroorotic acid medium (36) to select to generate Ura⁻ strain ABY06. Strain ABY08 was derived from ABY06 by selection on α-aminoadipic acid (37) to isolate a Lys⁻ auxotroph shown subsequently to be lys2 LYS5 by complementation.

Table 1.

Yeast strains used in this study

Strain	Genotype	Source
SEY6210	MATα leu2-3,112 ura3-52 his3-Δ200 trpl-Δ900 lys2-810 suc2-Δ9 Gal+	S. D. Emr
SEY6211	MATa leu2-3,112 ura3-52 his3-Δ200 trp1-Δ900 ade2-101 suc2-Δ9 Gal+	S. D. Emr
ABY02	MATa leu2-3,112 ura3-52 his3-Δ200 trp1-Δ900 ade2-101 lys2-810 suc2-Δ9 Gal+	This work
JE104	MATα mfα1Δ∷LEU2 mfα2Δ∷LYS2 ura3-52 his4 lys2 suc2-Δ9 Gal+	J. Thorner
DC14	MATa his1	J. Thorner
DC17	MATα his1	J. Thorner
CRY2	MATα his3-11,15 leu2-3,112 ura3-1 trp1-1 ade2-1 can1-100	Fuller lab
ABY06	MATα mfa1Δ∷LEU2 mfa2Δ∷LYS2 kex2Δ∷hisg ura3-52 his3-Δ200 trp1-Δ900 ade2-101 suc2-Δ9 Gal+	This work
ABY08	MATα mfα1Δ∷LEU2 mfα2Δ∷lys2 kex2Δ2∷hisg ura3-52 his3-Δ200 trp1-Δ900 ade2-101 suc2-Δ9 Gal+	This work

S. D. Emr, University of California, San Diego; J. Thorner, University of California, Berkeley. 

Mutagenesis of MFα1–100 and Mating Assays.

Plasmid YCpG-MFα1–100 was cleaved with XbaI and HindIII, and the digest was ligated to a 100-fold excess of 20 mer oligonucleotides 5′-ctagatNNNagagaggctga-3′and 5′-agcttcagcctctctNNNat-3′. Mutations were confirmed by sequencing alkali-denatured, double-stranded plasmid DNA. MFα1–100 alleles encoding substitutions for the P₂ Lys residue at position 84 were designated as shown in Table 2. To perform quantitative-mating assays, mutant plasmids were introduced into ABY08 with and without pAL5-KEX2. In the qualitative assay, patches of ABY08 transformants were grown on YPAGal plates for 12 hr at 30°C and then replica-plated onto lawns of DC14 cells on SD plates. In the quantitative assay (25), a fixed number of DC14 cells (1–1.5 × 10⁷) were mixed with exponentially growing cultures of α-cells (2–5 × 10⁶), and each mixture was collected on a nitrocellulose filter (HATF, Millipore). Filters were incubated on YPAGal plates at 30°C for 12 hr to allow mating. Cells were resuspended in distilled water and dilutions of the mating reaction were plated on SD plates to select prototrophic diploids. Mating frequency was expressed as the number of prototrophic colonies on minimal (SD) plates per input α cell. Typically, the mating frequency of ABY08 [pAL5-KEX2, YCpG-MFα1–100] was ≈0.2. Input α cells were measured by titering serial dilutions on permissive medium. Mating efficiency was the mating frequency of experimental samples divided by that for control strain ABY08 [pAL5-KEX2, YCpG-MFα1–100]. The mating efficiency of strain ABY08 [pAL5-KEX2, YCpG-MFα1–100] is defined as 1.0. Mating with Glu at P₂ was unexpectedly high. To establish definitively that this was due to cleavage after Glu-Arg, the plasmid encoding the putative MFα1–104 was recovered, sequenced, and reintroduced into ABY08. The mating behavior of this strain was identical to that previously observed.

Table 2.

Effects of P₂ substitutions on processing of T-pαf in vivo measured by quantitative mating^*

Genotype†			Plasmids‡		(-allele)	P2	P1↓	Relative mating efficiency	± % Error§
KEX2	MFα1	MFα2	KEX2	MFα1
+	+	+	—	—		LYS	-Arg	6.4	4
Δ	Δ	Δ	+	+	-wt	LYS	-Arg	4.5	10
Δ	Δ	Δ	+	+	-100	LYS	-Arg	1.0	13
Δ	Δ	Δ	+	+	-114	ARG	-Arg	0.51	11
Δ	Δ	Δ	+	+	-116	THR	-Arg	0.13	26
Δ	Δ	Δ	+	+	-112	PRO	-Arg	0.095	8
Δ	Δ	Δ	+	+	-104	GLU	-Arg	0.088	10
Δ	Δ	Δ	+	+	-108	ILE	-Arg	0.054	7
Δ	Δ	Δ	+	+	-115	SER	-Arg	0.043	12
Δ	Δ	Δ	+	+	-101	ALA	-Arg	0.032	16
Δ	Δ	Δ	+	+	-111	ASN	-Arg	0.029	2
Δ	Δ	Δ	+	+	-117	VAL	-Arg	0.014	15
Δ	Δ	Δ	+	+	-102	CYS	-Arg	8.5 × 10−3	7
Δ	Δ	Δ	+	+	-103	ASP	-Arg	2.7 × 10−3	5
Δ	Δ	Δ	+	+	-113	GLN	-Arg	2.4 × 10−3	14
Δ	Δ	Δ	+	+	-106	GLY	-Arg	2.3 × 10−3	4
Δ	Δ	Δ	+	+	-107	HIS	-Arg	7.2 × 10−4	1
Δ	Δ	Δ	+	+	-110	MET	-Arg	6.8 × 10−4	14
Δ	Δ	Δ	+	+	-109	LEU	-Arg	4.3 × 10−4	17
Δ	Δ	Δ	+	+	-119	TYR	-Arg	1.7 × 10−4	4
Δ	Δ	Δ	+	+	-105	PHE	-Arg	6.3 × 10−6	7
Δ	Δ	Δ	+	+	-118	TRP	-Arg	<1.0 × 10−6	NA
Δ	Δ	Δ	+	+	-120	amb	-Arg	<1.0 × 10−6	NA
Δ	Δ	Δ	+	−		- - - - - - - - - -		<1.0 × 10−6	NA
Δ	Δ	Δ	−	+	-100	LYS	-Arg	<1.0 × 10−6	NA

^*Assays were performed three to eight times for each amino acid substitution at P₂ but only twice for nonmating controls (e.g., amber codon at P₂ or complete absence of KEX2 or MFα1 gene). 

^†“+ + +” strain was CRY2; “Δ Δ Δ” was ABY08. 

^‡The KEX2 plasmid was pAL5-KEX2, the MFα1-WT plasmid was pBM258-MFα1, and the rest of the plasmids were YCpG-MFα1-100 or derivatives with the indicated P₂ substitution. 

^§“% Error” is SEM expressed as a percentage. 

Detection of Pro-α-Factor by Immunoblotting.

Anti-(pro-α-factor) antibodies were obtained by immunizing two virgin New Zealand white rabbits with a purified, E. coli-expressed β-galactosidase/pro-α-factor fusion protein (38) followed by four additional boosts with purified E. coli-expressed pro-α-factor (A.B. and R.S.F., unpublished experiments). Antibodies were affinity purified (16). Yeast transformants expressing T-pαf were grown in SGC-Ura medium to an OD of 2.0 at 600 nm, cells were removed by centrifugation, BSA (Pierce) was added to culture medium (3 μg/ml), and protein was precipitated by adding 0.1 volume of 0.15% (wt/vol) sodium deoxycholate and 90% (wt/vol) trichloroacetic acid at 0°C. After centrifugation, pellets were washed three times with 1 ml of ice-cold acetone and resuspended in 20 μl of 50 mM sodium phosphate, pH 5.5, 0.02% (wt/vol) sodium dodecylsulfate, 0.5 mM phenylmethylsulfonyl fluoride, and 10 mM NaN₃ and heated to 97°C for 3 min. Protein samples were deglycosylated by digestion with 0.5 munit of Endoglycosidase H (Boeringer Mannheim) for 16 hr at 37°C. Samples were subjected to SDS/PAGE (15% acrylamide), electroblotted to nitrocellulose (Millipore), and probed with a 1:5000 dilution of anti-(pro-α-factor) antibody. Immunoblots were probed with goat anti-rabbit horseradish peroxidase IgG and developed with enhanced chemiluminescence (Amersham) according to the manufacturer’s recommendations.

RESULTS

Sensitivity and Range of the Assay System.

Several controls were used to evaluate the genetic system. In vivo, wild-type pro-α-factor is processed to completion by Kex2, resulting in high levels of α-factor secretion. To determine whether the quantitative-mating assay was sensitive to slight reductions in α-factor production from the wild-type levels, the mating ability of strains expressing plasmid-encoded α-factor precursors with one or four α-factor repeats was examined. The mating efficiency of strain ABY08 containing the wild-type KEX2 gene on plasmid pAL5-KEX2 and expressing the wild-type MFα1 gene under GAL1 promoter control was quite similar to that of the wild-type MATα strain, CRY2 (Table 2). In contrast, the mating efficiency of ABY08 containing pAL5-KEX2 and expressing the truncated pro-α-factor (T-pαf) encoded by the Mfα1–100 gene under GAL1-promoter control was 4.5-fold lower (Table 2). Thus the mating assay responded to differences in levels of α-factor production even under conditions of efficient processing. Mating was undetectable in transformants lacking either the KEX2 or prepro-α-factor gene. Mating of ABY08 with MFα1–100 and pAL5-KEX2 was at least 10⁶-fold greater (Table 2). Thus, the quantitative-mating assay provided a sensitivity range of six orders of magnitude with which to assess the effects of substitutions at the P₂ position in T-pαf on the efficiency of cleavage by Kex2 protease in vivo. The precision of the assay was such that 2-fold differences in mating efficiency were reproducible.

Effects of Substitutions at P₂.

As shown in Table 2, all substitutions for Lys at P₂ reduced mating efficiency, with the effects spread over the entire sensitivity range of the assay. As expected, the best cleavage sites contained Lys or Arg at P₂, although substitution of Arg at P₂ led reproducibly to a 2-fold lower mating efficiency. Analysis of the data identified several general features as well as certain side-chain specific effects. With three clear exceptions, there was an approximate, inverse correlation between the total solvent accessible surface area of the P₂ side chain and the log of the mating efficiency (Fig. 2A). The exceptions were Lys, Arg, and Gly. In the case of Lys and Arg, efficient cleavage is likely due to favorable electrostatic interactions with the enzyme contacts. Studies of purified Kex2 appear to rule out contributions to substrate binding or catalysis from interactions between the methylene groups of P₂ side chains and the enzyme (9) Moreover, the Met side chain is nearly isosteric with the aliphatic portion of the Lys side chain with similar packing volume and hydrophobic surface area. Substitution of Met for Lys at P₂ resulted in a decrease of ≈1,500-fold in mating efficiency, whereas substitution of Ala for Lys resulted in only a ≈30-fold decrease. In the case of Gly at P₂, the most likely explanation for poor mating is poor cleavage due to high conformational flexibility. Substitution of large, aromatic side chains at the P₂ position had the most drastic effects. Introduction of Phe at P₂ resulted in barely detectable mating, with the efficiency reduced 160,000-fold. Introduction of Trp at P₂ resulted in apparent sterility. This was equivalent to the reduction of mating efficiency seen with strain ABY08 lacking either KEX2 or an MFα gene, a level at least 10⁶-fold below wild-type mating, the same as the level of mating observed with insertion of an amber codon at the P₂ codon position.

Figure 2.

(A) Plot of relative mating efficiency vs. surface area of the P₂ side chain. Total solvent-accessible side chain surface areas for all residues except Gly were as described by Creighton (39). For Gly, total solvent-accessible side chain surface was estimated as half the surface area of a hydrogen atom. (B) Plot of relative mating efficiency vs. accessible nonpolar surface area. Total solvent-accessible nonpolar side chain surface areas for all residues except Gly were as described by Creighton (39). For Gly, total solvent accessible nonpolar side chain surface was estimated as one-half the surface area of a hydrogen atom.

Correlation between mating efficiency and solvent-accessible nonpolar surface area of P₂ side chains also was examined (Fig. 2B). With the notable exception of Ile (see below), increasing solvent-accessible nonpolar surface area of hydrophobic residues (Pro, Val, Met, Leu, Tyr, Phe, and Trp) exhibited a better inverse correlation with mating efficiency than did total solvent accessible area, consistent with the general conclusion that the P₂-binding site is relatively hydrophilic. However, the mating efficiency in the case of residues with small and/or hydrophilic side chains (Asp, Gln, Asn, Ser, Glu, Ala, Thr) exhibited no obvious correlation in this plot.

The effects of certain substitutions were unexpected and will require detailed information about the three-dimensional structure of the active site to be explained. Introduction of Glu at P₂ caused only an 11-fold reduction in mating, despite the introduction of negative charge. Introduction of Asp resulted in a much greater decrease in mating (370-fold). Interestingly, mating with Gln was reduced ≈420-fold, whereas mating with Asn was only reduced 34-fold. The P₂-binding site appears to favor β-branched residues. Ile at P₂ resulted in a 19-fold reduction in mating whereas with Leu, mating was reduced 2,300-fold. Thr was the third best residue at P₂, reducing mating only 7-fold relative to Lys whereas Ser reduced mating efficiency 23-fold.

Observed Cleavage Is Due to Kex2 Protease.

One potentially complicating factor in the assay was the possibility that some of the P₂ mutant precursors were being cleaved by a protease other than Kex2. To examine this possibility, plasmid pAL5-KEX2 was evicted from each of the 20 strains expressing a form of T-pαf with either the wild-type or one of the mutant cleavage sites. In each case, loss of the KEX2 plasmid resulted in negligible mating, indicating that mating depended on cleavage by Kex2 protease (data not shown).

Mutations Do Not Affect T-pαf Stability or Secretion.

Reduced mating associated with any or all of the mutant precursors might have been caused by instability or improper localization rather than by reduced efficiency of cleavage by Kex2 protease. Culture media of ABY08 transformants expressing MFα1–100 or one of the P₂ substitution alleles, but lacking the KEX2 gene, were concentrated and digested with Endo-glucosaminidase H to remove N-linked oligosaccharides. Affinity-purified antibody raised against pro-α-factor was used to characterize the deglycosylated, secreted α-factor precursors in these media fractions by immunoblotting (Fig. 3). Polypeptides similar in mobility and abundance to the MFα1–100 protein were detected in the culture media of all but three transformants. As expected, the strain containing the vector, pBM258, without an insert produced no cross-reacting material. Likewise, the medium from ABY08 expression MFα1–120, which contained a nonsense codon in the P₂ position, showed no detectable antigen. The strain expressing MFα1–102, which encodes a precursor with a substitution of Cys for the P₂ Lys⁸⁴, produced a precursor-related antigen that migrated much more slowly than the others. This precursor appeared to dimerize through an interchain disulfide because (i) Cys⁸⁴ was the only Cys residue in the polypeptide and (ii) this precursor could only be deglycosylated if first treated with β-mercaptoethanol. Thus with the exception of Mfα1–102p, all mutant precursors containing an amino acid substitution at the P₂ position were produced in normal amounts and transported normally through the secretory pathway.

Figure 3.

Expression of mutant α-factor precursors. Yeast strain ABY08 was transformed with pBM258 (control plasmid or “−”), YCpG-MFα1–100 (K), and with plasmids carrying P₂-substituted mutant alleles of MFα1–100 indicated using the single letter amino acid code. The amber termination codon in place of Lys84 is labeled “amb.” Enzymatically deglycosylated proteins from culture media were separated by SDS/PAGE, blotted, and probed with anti-(pro-α-factor) antiserum as described in Materials and Methods. Mfα1–120p with Cys at P₂ formed an intermolecular disulfide and was only partially digested (shown) with Endoglycosidase H unless reduced with β-mercaptoethanol before deglycosylation (data not shown).

DISCUSSION

Most studies of the specificity of Kex2 family proteases using natural proprotein substrates, although informative, have been qualitative in nature. Studies with purified furin established requirements for P₁ and P₄ Arg residues in cleavage of the anthrax protective antigen and suggested that the nature of the P₂ and P₃ residues was less important (13). One in vivo study used a cotransfection assay to assess the ability of wild-type furin and various furin mutants to process the wild-type cleavage site (Arg-Ser-Lys-Arg↓) and three mutant cleavage sites containing single substitutions at P₁ (Gly), P₂ (Ala), or P₄ (Ala) in pro Von Willebrand factor (12). Bravo and coworkers (40) used immunoblotting to follow cleavage of wild-type and mutant forms of the human insulin proreceptor by purified secreted, soluble furin.

We have detailed a genetic system in which enzyme-substrate interactions can be quantified under physiological conditions. These studies provide a comprehensive profile for recognition of substrate P₂ residues by Kex2 in vivo, raising the question of how well these results compare with the kinetic analysis of cleavage of synthetic peptide substrates by purified Kex2. Overall, the relative k_cat/K_M values for cleavage of small peptidyl-MCA substrates (Fig. 4A) and internally quenched fluorogenic peptide substrates (Fig. 4B) by purified Kex2 protease correlate well with the relative mating efficiencies observed with comparable T-pαf variants. The general agreement between these sets of data leads to several conclusions. First, studies with model peptides have shown that with Arg at P₁, most substitutions for Lys at P₂ affected K_M without significantly altering k_cat. Thus, the correlation between k_cat/K_M values and mating efficiencies indicates that K_M is a major determinant of specificity in vivo. Second, pro-α-factor cleavage in vivo must be regulated by K_M even for very good, low K_M substrates. Even the relatively conservative substitution of Arg for Lys at P₂ in T-pαf resulted in a measurable and reproducible change in the mating assay. Studies with model peptides indicated that substitution of Arg for Lys at P₂ results in only a 4-fold increase in K_M, along with a small increase in k_cat (7). Third, only three of the six most efficiently cleaved sites reflect sequences found in natural substrates of Kex2 protease. Lys-Arg is present in the four cleavage sites of α-factor precursor (23) and is the site of intramolecular cleavage at of the N-terminal “pro” domain of Kex2 (7, 43). Kex2 protease also cleaves at a Lys-Arg, an Arg-Arg, and probably a Pro-Arg site in the M1 killer protoxin in vivo (44, 45). Hsp150p (46) and the EXG1-endcoded exoglucanase (47) also are cleaved at Lys-Arg sites. It is interesting, therefore, that the current study reveals Thr-Arg to be almost as good as Arg-Arg and slightly better than Pro-Arg as a cleavage site for Kex2 protease, suggesting Thr-Arg as a candidate physiological cleavage site for the enzyme although no such sites are known in natural substrates. This prediction is supported by analysis of furin specificity by the substrate phage display method, which suggested that in addition to Lys and Arg, Thr and Pro are acceptable P₂ residues for furin (48). Finally, the correlations observed indicate that cleavage of the T-pαf P₂ sequence variants in vivo can be accounted for by recognition of substrate sequences by Kex2 in small, unstructured peptides, suggesting that higher order structure may be relatively unimportant in utilization of pro-α-factor as a substrate by the enzyme.

Figure 4.

In vitro vs. in vivo cleavage efficiency. (A) Peptidyl-MCA substrates. Relative efficiency of the peptidyl-MCA substrates is correlated with the in vivo-mating efficiency. Relative cleavage efficiency is defined as the ratio of k_cat/K_M for a peptidyl-MCA substrate with the indicated residue at P₂ to the k_cat/K_M for a peptidyl-MCA substrate with Lys at P₂. Data for k_cat/K_M of purified secreted, soluble Kex2 protease with peptidyl-MCA substrates are from Brenner and Fuller (7), Rockwell, et al. (9), Rockwell and Fuller (10), and N. Rockwell and R.S.F. (unpublished results). P₂ residues of substrates compared are indicated in the single letter code (β represents nor-leucine). A substrate with nor-leucine at P₂ was compared with mating efficiency with Met at P₂ in vivo (YCpG-MFα1–110). In certain cases (Arg, Thr, Ala, and Gly), two synthetic substrates were available for comparison. The dashed line represents a line fit for direct proportionality between log (k_cat/K_M) and the log of the mating efficiency (correlation coefficient, r = 0.886; slope = 0.75). (B) Internally quenched fluorogenic peptide substrates. A correlation plot of relative efficiency of cleavage of the internally quenched fluorogenic peptide substrates (k_cat/K_M) v. relative mating efficiency for corresponding P₂ substitutions. Relative cleavage efficiency is defined as the ratio of k_cat/K_M for an internally quenched fluorogenic peptide substrate with the indicated residue at P₂ to the k_cat/K_M for an internally quenched fluorogenic peptide substrate with Lys at P₂. Data for k_cat/K_M of purified secreted, soluble Kex2 protease with internally quenched fluorogenic peptide substrates are from ref. 9. P₂ residues of substrates compared are indicated in the single letter code. The internally quenched fluorogenic peptides were of the sequence Arg-GlE-norLeu-Tyr-Xaa-Arg-↓-Glu-Ala-Glu-Ala-LyD-Arg, in which “GlE” represents glutamyl-EDANS, [EDANS, 5-(2-aminoethyl-amino)-naphthalene-1-sulfonic acid], “Xaa” represents the particular P₂ substitution, and “LyD” represents lysyl-DABSYL (DABCYL, 4-(4-dimethylaminophenyl)-azobenzoyl) (41, 42). This sequence was based on the Kex2 cleavage site between the α-factor repeats in pro-α-factor. A substrate with nor-leucine (X) at P₂ was compared with mating efficiency with Met at P₂ in vivo (YCpG-MFα1–110).

The most deleterious substitutions at P₂ involved residues with bulky, nonpolar side chains, suggesting that steric exclusion poses the largest barrier to efficient cleavage, revealed most dramatically in the case of Trp at P₂. In the case of His, although the imidazole ring is polar, its bulk presumably prevents a good fit in the P₂ pocket. Comparison of Tyr and Phe reveal that although these residues have similar bulky aromatic side chains, cells expressing T-pαf with Tyr at P₂ mated 27-fold better than cells expressing T-pαf with Phe at P₂. Presumably, this results from the greater polarity of the phenolic side chain of Tyr. The poor mating with His or Tyr at P₂ indicates that neither polarity nor the possibility of positive charge can fully compensate for the unfavorable steric interactions of the large cyclic side chains.

Two exceptions to the general agreement between the in vivo and in vitro data emerged with acidic residues at P₂. As shown in Fig. 4, cells expressing T-pαf with Glu or Asp at P₂ mated significantly better than would be expected from the relative k_cat/K_M values. A possible explanation for this observation may lie in differences in sequence context. In T-pαf, the sequence spanning the cleavage site was Ser-Leu-Asp-Lys-Arg-↓-Glu-Ala-Glu-Ala. Substituting Asp or Glu at P₂ therefore introduces an acidic residue adjacent to Asp at P₃. In contrast, the peptidyl-MCA substrate with Asp at P₂ had Asn at P₃ (N. Rockwell and R.S.F., unpublished experiments). Likewise, the internally quenched substrate with Glu at P₂ was based on the sequence Met-Tyr-Lys-Arg-↓-Glu-Ala-Glu-Ala, with Tyr at P₃ (8). Repulsion between adjacent acidic side chains at P₂ and P₃ in T-pαf may result in a substrate conformation that promotes cleavage.

This work demonstrates the utility of the mating assay as a quantitative method for analyzing substrate specificity of Kex2 protease in vivo and establishes methods that form part of a genetic and biochemical to approach to understanding and engineering protease specificity (49). The combined information obtained from the physiological and biochemical studies provides a comprehensive selectivity profile of the wild-type protease that defines the range and nature of binding contacts used by the enzyme. This genetic assay also can be used in screens to identify mutant enzymes having altered substrate specificity (A.B. and R.S.F., unpublished experiments).

ABBREVIATIONS

Peptidyl-MCA

peptidyl methylcoumarin amide

T-pαf

truncated pro-α-factor