Expression of the SMK1 gene which encodes the yeast killer toxin SMKT is lethal in Saccharomyces cerevisiae. Effects of deletion and site-directed mutagenesis of SMK1 on the lethality and the secretion of the gene products were examined. Deletion of the interstitial γ peptide or the C-terminal loop from Ala208 to the C-terminal Asp222 had no effect on the lethality. Those SMK1 products that lacked either the γ peptide or the C-terminal loop were expressed in the cells but were not secreted into the culture medium, suggesting that these peptides may have a role in secretion or in protein stability. On the other hand, deletion of the signal sequence resulted in complete loss of the lethal activity. Entering the secretory pathway may be critical for the lethality. Further, deletion of the region from the C-terminus to Leu207 resulted in loss of the lethal activity. Leu207 is located at the C-terminus of the central strand of the β-sheet structure of SMKT and its side chain is thrust into a hydrophobic environment between the β-sheet and the α-helices. The result obtained upon substitutions of Ala, Ser or Glu for Leu207 suggested that the side chain of Leu207 stabilizes the hydrophobic environment that contributes to the overall structure of the SMK1 product.
SMKT is a killer toxin produced by the halotolerant yeast Pichia farinosa KK1. The toxin exhibits maximum activity against Saccharomyces cerevisiae, Zygosaccharomyces rouxii, etc., in the presence of 2 M NaCl and was named salt-mediated killer toxin (SMKT) (Suzuki and Nikkuni, 1989). The killer gene SMK1 is an endogenous P.farinosa chromosomal gene. SMK1 encodes a preprotoxin consisting of 222 amino acid residues, comprising a typical signal sequence, a hydrophobic α subunit (63 amino acids), an interstitial γ polypeptide (62 amino acids) with a putative glycosylation site and a hydrophilic β subunit (77 amino acids) (Suzuki and Nikkuni, 1994). The mature toxin, formed through processing of the preprotoxin, is a heterodimer consisting of α and β subunits. Investigation of the crystal structure of SMKT revealed that the α and β subunits are folded together in a single ellipsoidal domain (Kashiwagi et al., 1997). SMKT is active only below pH 5, whereas the two subunits dissociate above pH 5, resulting in loss of the killer activity (Suzuki and Nikkuni, 1994; Suzuki et al., 1997). It seems that four pairs of interactions of the carboxyl groups are responsible for these characteristics. Under neutral or basic conditions, these interactions would become repulsive and destabilize the structure (Kashiwagi et al., 1997).
A variant SMKT with higher stability and at the same time showing higher killer activity is now being sought in view of the need for further elucidation of the structure–function relationship of SMKT and the potential use of the toxin in a variety of fields. To accomplish this goal, a suitable system for expression of SMK1 is indispensable. However, despite several attempts to establish an SMK1 expression system, expression of active SMKT has not yet been achieved.
In the case of the K1 toxin of S.cerevisiae, the immunity is determined by the toxin precursor molecule, and expression of the K1 killer gene confers immunity (Boone et al., 1986; Zhu and Bussey, 1991). On the other hand, expression of the γ subunit of Kluyveromyces lactis toxin results in death of the host cells. However, this lethal effect was found to be prevented by expression of the killer immunity gene, which is specified as ORF3 of the linear plasmid pGKL1 of K.lactis (Tokunaga et al., 1989). By analysis of resistant mutants, it was suggested that intracellular expression of the γ subunit mimics treatment with exogenous toxin (Butler et al., 1991).
In this study, we show that expression of SMK1 under the control of a galactose-inducible promoter is lethal in S.cerevisiae. We also determined the SMK1 regions required for the lethality in S.cerevisiae. The roles of each polypeptide module defined by the crystal structure are discussed in terms of their lethality in S.cerevisiae.
Materials and methods
Yeast strain and media
Saccharomyces cerevisiae W303-1A (MATa, ade2, his3, leu2, trp1, ura3, can1) and CS701C (spf1::LEU2 in W303-1A background) were used throughout this study. YPD contained 1% yeast extract, 2% peptone and 2% glucose and YPD/MB was YPD containing 0.003% methylene blue and 2% agar. Selective media contained 0.67% yeast nitrogen base (Difco, USA), 2% glucose and appropriate supplements as needed. YPGal contained 1% yeast extract, 2% peptone, 2% galactose and 25 mM citrate-phosphate buffer (pH 3.5). YPGal/MB plates contained 1% yeast extract, 2% peptone, 2% galactose, 0.003% methylene blue and 2% agar.
SMK1 and SMK1 derivatives were inserted into pYES2 (Invitrogen, USA) containing the Gal1 promoter (pGal) and the terminator for expression, and URA3 and 2μ sequences for selection and replication in yeast. Yeast transformation was performed by electroporation (Becker and Guarente, 1991). Standard molecular manipulations were as described by Sambrook et al. (1989). Site-directed mutagenesis was performed using the QuickChange™ mutagenesis kit (Stratagene, USA) according to the manufacturer's instructions. The positive clones were verified by DNA sequencing using a BigDye terminator sequence kit (Perkin Elmer) and specific primers. In analysis of the expressed SMK1 product, transformants were cultured in 10 ml selective medium overnight. The cells were collected by centrifugation, resuspended in 2 ml YPGal and cultured overnight. Preparation of cell lysates, precipitation of secreted proteins by TCA, electrophoresis and immunoblotting were performed as described previously (Suzuki, 1999).
Results and discussion
An expression vector, pCS224, was constructed for S.cerevisiae in which SMK1 was inserted under the control of a galactose-inducible promoter, pGal1 (Figure 1A). When SMK1 was induced by galactose, the colonies of transformants showed a blue color in the presence of methylene blue, which is an oxidation–reduction indicator, employed to stain dead cells. On the other hand, the color of the colonies on YPD/MB plates did not change. We also examined the toxicity of the expression of SMK1 by plasmid curing. After a 2 day culture period in YPGal medium, 100% of the living cells failed to retain pCS224, whereas only 20% of the living cells lost the vector pYES2 (data not shown). These results suggest that expression of SMK1 is lethal in S.cerevisiae and that the SMKT immunity system is different from that of the K1 toxin in respect of the expression of the preprotoxin.
SPF1 is a gene coding for a yeast P-type ATPase that is required for sensitivity to SMKT. The spf1 disruptants are highly resistant to exogenous SMKT (Suzuki and Shimma, 1999). To determine whether the spf1 disruptants show resistance to endogenously expressed SMK1, pCS224 was introduced into the spf1 disruptant strain CS701C. The colonies of the transformants on YPGal/MB plates did not show a blue color. This resistant strain concurrently acquired resistance to endogenously expressed SMK1, suggesting that the spf1 disruptants have a modified toxin target site. Whether the interaction between the SPF1 product and the SMK1 product is responsible for the lethal effect, and whether the ATPase is involved in the synthesis of the target site should be investigated.
Using the methylene blue plate assay, the lethal effect of modified SMK1 was examined (Figure 1A). Substitution of the α-factor signal sequence for the original signal sequence of SMK1 (pCS204), or deletion of the γ peptide region (pΔγ), did not affect the lethality. However, deletion of the signal peptide region (pΔSig) resulted in the loss of lethality, suggesting that entering the secretory pathway is important for the toxicity. In the case of the K.lactis toxin, expression of the γ subunit without a signal sequence is also lethal for the host cells (Tokunaga et al., 1989). Therefore, the mechanism responsible for the toxicity of SMKT may be different from that in the case of the K.lactis toxin.
Deletion of the 3′ region corresponding to 21 amino acid residues at the C-terminus (del39) also resulted in the loss of lethality. In order to determine the amino acid residues in the C-terminal region that are involved in the lethality, a stop codon or an amino acid substitution was introduced into the 3′ region of SMK1 by site-directed mutagenesis. The stop codons inserted between Ala208 and the C-terminus (pMK207–pMK221) constituting a C-terminal loop had no effect on the lethality. On the other hand, the stop codon insertion at Leu207 (pMK206) resulted in the loss of activity (Figure 1B). To investigate the role of the Leu207 residue, Leu207 was replaced by Ala (pMK207A), Ser (pMK207S) and Glu (pMK207E). Cells expressing the plasmid-encoding genes with the Ala or Ser substitution showed a lighter blue color than those harboring pCS224, whereas the Glu substitution resulted in complete loss of lethality (Figure 1B). As shown in Figure 2, Leu207 is located at the C-terminus of the central strand of the β-sheet structure of SMKT and its side chain is thrust into a hydrophobic environment between the β-sheet and the α-helices. The results obtained from the substitutions of Ala, Ser or Glu for Leu207 indicated that the nonpolar side chain of Ala or Ser fits the hydrophobic pocket, but the carboxyl group of Glu destabilizes the structure critical for the lethal effect of the SMK1 product. Thus, the side chain of Leu207 may play a key role in stabilizing the overall structure of the SMK1 product.
The expression of the products from modified SMK1 was examined by immunoblotting. Transformants were grown in YPGal overnight. Cell lysates were separated into microsome fractions and soluble fractions by means of an ultracentrifuge and secreted proteins in the culture medium were precipitated with TCA. Endogenous SMK1 products were detected in the microsome fractions but not in the soluble fractions (Figure 3A). The SMK1 products corresponding to the protoxin (αγβ) and γβ produced by cleavage between the α subunit and the γ peptide were secreted into the medium, but the mature SMKT was not detected (Figure 3B). The SMK1 products from pMK207–pMK216 lacking the C-terminal loop and those from pΔγ lacking the γ peptide were detected in the microsome fractions but were not secreted. The SMK1 products from pMK207S, pMK207A and pMK208E were not secreted. Because the lethal activity was evident in the case of cells showing expression of these unsecreted SMK1 products, it is obvious that the lethal effect is not due to the toxicity of the secreted SMK1 products. However, the colonies of transformants expressing the non-secreted SMK1 product were a lighter blue color than those expressing the secreted SMK1 product. Lack of the γ peptide or the C-terminal loop, or substitution of another amino acid for the Leu207 residue may decrease the stability of the SMK1 product.
Our previous conclusion based on the results of a structural study of SMKT was that the unique pH sensitivity of SMKT is due to the interactions of four pairs of carboxyl groups (Figure 2) (Kashiwagi et al., 1997). Based on this hypothesis, Asp213 and Asp222, which presumably interact with Asp76 and Glu158, respectively, were substituted for Ala (pMK213A, pMK222A). In addition, the C-terminal Asp222 that interacts with Asp199 was removed (pMK221). Contrary to our expectations, these substitutions and the deletion did not affect either the lethality or the secretion of the products. Therefore, three interactions of carboxyl groups appear to stabilize the crystal structure by connecting the flexible C-terminal loop to the β-sheet structure but those do not contribute to the stability of the SMK1 products in solution.
In the case of the K1 toxin and the α-factor of S.cerevisiae, association of the precursors with the membrane has been observed and this has been thought to be evidence of post-translational signal cleavage. Most of the endogenous SMK1 precursors were also observed in the microsome fractions but not in the soluble fractions, suggesting that stable association with the membrane occurred. As shown in Figure 3A, a 15 kDa product of pΔγ corresponding to the α+β polypeptide was observed in the microsome fraction. A Kex2p site (Lys–Arg) exists between the glycosylated signal sequence and the α+β polypeptide. Expression of the α+β polypeptide is evidence that the precursor has reached the Golgi.
The secreted form of proSMKT (αγβ) in P.farinosa was purified and shown to have no killer activity (Suzuki, 1999). Considering the C-terminus of the α subunit and the N-terminus of the β subunit in the crystal structure of SMKT (Figure 2), the γ domain is likely to cover the upper part of SMKT. Because endogenous expression of SMK1 products with the γ domain is lethal and proper folding may be required for the lethality, it seems unlikely that the γ domain prevents the target-binding site from interacting with the target in sensitive cells. Therefore, the γ domain may mask the receptor-binding domain that is only required for the exogenous toxicity. The killer strain of P.farinosa is immune to SMKT. The immunity determinant of SMKT may prevent the protoxin from interacting with its own target. Although the lethality of SMK1 in S.cerevisiae described here is a heterologous system, it serves as a basis for elucidation of the immunity mechanism of SMKT.
We thank Dr R.Akada (Yamaguchi University) for helpful discussion and Dr G.R.Fink (Whitehead Institute for Biomedical Research) for supplying the yeast strain. This work was supported in part by a Grant-in-Aid (Protein Refolding Project) from the Ministry of Agriculture, Forestry and Fisheries.