Growth inhibition of Saccharomyces cerevisiae by the plasmid-encoded trimeric (αβγ) zymocin toxin from dairy yeast, Kluyveromyces lactis, depends on a multistep response pathway in budding yeast. Following early processes that mediate cell-surface contact by the chitinase α-subunit of zymocin, later steps enable import of the γ-toxin tRNase subunit and cleavage of target tRNAs that carry modified U34 (wobble uridine) bases. With the emergence of zymocin-like toxins, continued zymocin research is expected to yield new insights into the evolution of yeast pathosystems and their lethal modes of action.
- Elongator complex
- Kluyveromyces lactis
- tRNA methylase Trm9
- tRNA wobble uridine (U34) modification
Secretion of lethal toxins is a common strategy for prokaryotic and eukaryotic micro-organisms to gain selective growth advantages over microbial competitors. In yeast, such protein toxins often associate with extrachromosomal dsRNAs (double-stranded RNAs) or linear dsDNA plasmids and can be distinguished on the basis of subunit composition, cell-surface specificity, membrane translocation, toxin import and cell death induction . One such pathosystem that mediates lethal interaction between killer strains of dairy yeast, Kluyveromyces lactis, and baker's yeast, Saccharomyces cerevisiae, relies on killer plasmids that encode the zymocin trimer (αβγ) and a hitherto unidentified immunity factor for self-protection of the toxin producer [2,3]. Zymocin induces a G1 cell cycle arrest that prevents sensitive yeasts from budding . Conditional expression studies revealed that toxicity resides in zymocin's γ-subunit (γ-toxin) and that toxin-resistant kti (killer toxin insensitive) and tot (γ-toxin target) S. cerevisiae mutants fall into two classes: binding/uptake defects (class I) protecting against exozymocin, and toxin-target defects (class II) causing resistance to exozymocin and γ-toxin [5–8] (Table 1).
The early zymocin response pathway
Analysis of class I scenarios (Table 1) has revealed an early zymocin response pathway in S. cerevisiae. Here, key players promote cell-surface contact and plasma membrane association with zymocin's α- and β-subunits in order to potentiate the subsequent import of γ-toxin [9–13]. Consistent with such vehicle roles for αβ, exogenously applied γ-toxin alone is biologically inert  and the α-subunit is an exochitinase . Together with purification of holo-zymocin by chitin-affinity chromatography and the predicted hydrophobicity of the β-subunit, to which the γ-toxin is disulfide-bonded, it is likely that holo-zymocin initially docks on to cell wall chitin [2,10]. Potentially, this is followed by local chitinolysis and subsequent association of the trimer (or parts thereof) with the plasma membrane [9,13]. In line with this, excess chitin suppresses zymocin action and deletions in chitin biosynthetic genes (Table 1), including the class I chitin synthase III locus KTI2/CHS3, cause zymocin protection . Moreover, UGP1 and ISR1 encoding respectively UGPase (UDP-glucose pyrophosphorylase) and a Pkc1 (protein kinase C1)-related putative kinase (Isr1) are class I dosage suppressors of zymocin  (Table 1). Suppression depends on the CDK (cyclin-dependent kinase) activity of Pho85, which phosphorylates UGPase and Isr1 in vitro [11,14]. Given the functional overlaps between this CDK and the Pkc1–MAPK (mitogen-activated protein kinase) pathway, UGPase and Isr1 may function in cell wall-related processes , and inappropriate β-glucan synthesis due to UGPase overexpression may prevent zymocin from accessing chitin . Consistent with a potential role for Isr1 as part of the Pkc1 pathway in the chitin emergency response, lack of chitin deposits at incipient bud sites and bud neck regions in high-copy ISR1 cells accounts for zymocin suppression . Class I kti6/ipt1 mutants lack the sphingolipid M(IP)2C (mannosyl-di-inositol phosphoceramide), which populates membrane lipid rafts, and kti10/pma1 mutants have a defective H+-ATPase Pma1 whose proton pump activity generates plasma membrane potentials [12,13] (Table 1). Although post-docking steps in the zymocin response are hardly understood, current models favour a role for M(IP)2C in γ-toxin import, while Pma1 might be hijacked for γ-toxin activation [12,13]. Consistently, γ-toxin is not taken up by kti6 cells and low pH bypasses the proton pump defects and the class I resistance of kti10 mutants [12,13]. In addition to the G1 arrest, zymocin and a related toxin from Pichia acaciae inhibit yeast matings . This may hinder target haploid cells to enter the more robust and natural diploid life cycle stage, a notion supported by articulate toxin protection of diploids or pseudodiploid yeast cells that carry an extra copy of the mating-type locus MAT [7,15] (Table 1).
Elongator, a key zymocin effector complex
As for the intracellular toxin target, ten class II KTI genes suggested a biochemical pathway or a multiprotein effector complex of zymocin  (Table 1). Indeed, KTI and TOT (toxin target effector) genes identified a toxin effector role for the six-subunit Elongator complex from S. cerevisiae, and Elongator inactivation or defects in several Elongator-related factors nullified zymocin [8,16–25] (Table 1). Elongator-minus cells have defects in cell cycle progression and they die in the presence of various thermal and chemical stresses, suggesting that the Elongator complex is particularly important for cell viability under stress conditions [8,16–25]. In line with this, the Elongator complex is involved in processes as distinct as RNA polymerase II transcription [16,17,26,27], exocytosis  and tRNA modification . Whether these all represent individual Elongator roles or lie downstream of a unique Elongator function are open questions . In accordance with a transcriptional role, Elongator partners with the RNA polymerase II holoenzyme and assists the transcription machinery by virtue of its HAT (histone acetylase) subunit Elp3 [17,26,31]. Markedly, holo-Elongator was found to associate with nascent, unspliced mRNAs during transcription in the nucleus . Consistent with cytoplasmic Elongator processes, substantial pools of Elongator are cytosolic , and Elongator interacts with Sec2, a GTP/GDP exchanger for GTPase Sec4 which regulates post-Golgi trafficking and polarized exocytosis .
Elongator and tRNA U34 (wobble uridine) modification
A novel Elongator process that has recently surfaced is tRNA U34 modification . In addition to holo-Elongator, most if not all, class II zymocin- and Elongator-related proteins (Table 1) partake directly or indirectly in the U34 pathway . The U34 modifications include ncm5U (5-carbamoyl-methyluridine), cm5U (5-carboxymethyluridine), mcm5U (5-methoxycarbonylmethyluridine) and mcm5s2U (5-methoxycarbonylmethyl-2-thiouridine), and in Elongator mutants, these 5-substituted U34 bases are not formed  (Figure 1). Abolished mcm5 side chain formation eliminates the performance of U34-containing tRNA suppressors [29,33]. Thus nonsense read-through of ade2-1 and can1-100 by tRNA suppressor SUP4 requires mcm5U34  and missense suppression of cdc8-1 by tRNA suppressor SOE1 depends on mcm5s2U34 . These findings imply that by influencing anticodon–codon interactions, Elongator-dependent U34 modifications may affect tRNA functions during mRNA translation. So the pleiotropic, Elongator-minus traits (see above) may result from tRNA modification defects . In line with this notion, many Elongator defects can be bypassed by excess levels of two U34-containing tRNAs , suggesting that the bona fide role (see above) for holo-Elongator may be tRNA modification. In addition to Elongator, the class II KTI1/TRM9 gene product operates in the U34 pathway  (Table 1; Figure 1). Trm9 is a tRNA methyltransferase that acts downstream of Elongator, and trm9Δ mutants lack the methylation in the mcm5 side chain . Remarkably, the methylation defect of trm9Δ cells protects as efficiently against zymocin as lack of the entire mcm5 chain in Elongator mutants [33,36]. This suggests that Elongator's toxin effector role is to promote U34 tRNA methylation by Trm9, which eventually provides the key pathological determinant for zymocin [36,40] (Figure 1). Data that excess levels of Elongator- and Trm9-dependent tRNAs suppress zymocin support this notion [7,33,36]. Together with the observation that suppression is abolished by TRM9 overexpression, hypomethylated tRNAs can bypass zymocin, and excessive Trm9 methylation overcomes this bypass to restore zymocin lethality .
Zymocin, a eukaryotic tRNase toxin
Consistent with the latter observations, zymocin's lethal γ-toxin subunit has been shown to be an ACNase (anticodon nuclease) that targets Elongator- and Trm9-dependent tRNAs . Hence, Elongator and trm9 mutants protect against tRNA restriction by zymocin, a finding consistent with their pronounced class II resistance [8,33,36]. Among roughly a dozen Elongator-dependent tRNAs , the spectrum targeted by zymocin is restricted to tRNAGlu, tRNALys and tRNAGln, species that all carry mcm5s2U34 [33,36]. Consistently, overexpression of these tRNAs alone or in combination suppresses zymocin, while overexpression of tRNAArg, which carries mcm5U34, cannot [8,33,36]. This suggests that U34 thiolation in position 2, a modification absent from tRNAArg, contributes to zymocin specificity, too (Figure 1). Based on the findings that U34 thiolation is intact in Elongator mutants , it will be intriguing to identify and analyse components for the U34 thiolation pathway including Ncs2  by using zymocin as a molecular diagnostic tool (Figure 1). It is unclear how Elongator and Elongator-related factors promote U34 modification . Preliminary analyses indicate that Elongator precipitates the U34-containing tRNAGlu, but not tRNAMet, and that the HAT subunit Elp3 carries an iron–sulfur cluster able to co-ordinate SAM (S-adenosylmethionine) [29,31,37]. Together with evidence that archaeal Elp3 cleaves SAM, a feature typical of radical SAM enzymes, Elongator may have a second catalytic function that relates to U34 modification . Cross-complementation studies between S. cerevisiae and Arabidopsis thaliana suggest that Elongator's role in tRNA modification appears to be conserved between yeast and plants . Thus zymocin resistance of a yeast mutant lacking Elongator subunit Elp1 was complemented in trans by the homologous Elongator plant gene ELO2, thereby restoring zymocin susceptibility . In vitro, Trm9 alone catalyses U34 methylation reactions that lead to modified nucleosides . However, this does not preclude the involvement of auxiliary factors and Trm9 partners with Trm112 in vivo . Notably, Trm112 is also found in complex with the methylases Mtq2 and Trm11  that methylate (i) translation release factor Sup45 when complexed with Sup35 , and (ii) position G10 of tRNA . Together with other dimeric tRNA methylases (Trm6·Trm61 and Trm8·Trm82) in yeast, U34 methylation may thus require a Trm9·Trm112 complex (Figure 1) rather than the Trm9 polypeptide alone.
tRNA depletion, a lethal strategy by microbial ACNases and tRNase toxins
Zymocin's tRNase subunit cleaves anticodons between positions 34 and 35 (Figure 1) and generates a 2′,3′-cyclic phosphate and a 5′-hydroxy group . Although the cleavage site differs from the one generated during tRNA intron removal by tRNA splicing endonuclease, both ACNases are similar and relate to the bacterial tRNases PrrC and colicins D and E5 [36,42–45] (Figure 2). Similarly to zymocin, the colicins are plasmid-encoded tRNase killer toxins and cleave respectively tRNAArg iso-acceptors or tRNATyr, tRNAHis, tRNAAsn and tRNAAsp [44,45] (Figure 2). Moreover, zymocin's cleavage requirement for U34 modification resembles that of PrrC . Data that colicin cleavage of tRNAs inactivates tRNA aminoacylation and that growth inhibition by the antibiotic neomycin B involves cleavage of tRNAPhe suggest that tRNA restriction causes cell death through tRNA depletion [45,47] (Figure 2). So, granted that tRNA depletion by zymocin affected the translational tRNA functions in yeast, the observed G1 cell cycle arrest  by zymocin may be due to a block in mRNA translation. Consistently, conditional inactivation of essential CDC (cell division cycle) genes coding for translationally relevant proteins causes cell cycle blocks in yeast [48–50]. With zymocin's tRNase toxin hijacking the U34 modification pathway in yeast, it will be interesting to see whether this lethal strategy is unique. The emergence of a few zymocin-like killer plasmid systems from Debaryomyces, Pichia and Wingea  implies that the phenomenon is spread more among yeast or fungal genera than previously anticipated.
We thank C. Bär, H. Berndt, L. Fichtner, F. Frohloff, C. Mehlgarten, J.-E. Täubert, S. Schewtschik, P. Studte, E. van der Zalm, R. Zabel and S. Zink for contribution of data. R.S. was supported by the DFG (Deutsche Forschungsgemeinschaft; Scha750/2-1/4 and SFB648), the ‘Fonds der Chemischen Industrie’ and the ‘Alexander von Humboldt-Stiftung’. In addition, FEMS (Federation of European Microbiological Societies) and FEBS (Federation of European Biochemical Societies) fellowships to C. Bär, D.J., L. Fichtner, C. Mehlgarten and S. Zink and support to C. Bär, R. Zabel and E. van der Zalm by the Land Sachsen-Anhalt ‘Graduiertenförderung’ and Graduate School ‘Plant Protein Complexes – Structure, Function and Evolution’ are greatly acknowledged.
British Yeast Group Meeting 2007: Independent Meeting held at the Paramount Palace Hotel, Buxton, U.K., 26–28 March 2007. Organized and Edited by A. Goldman (Sheffield, U.K.).
Abbreviations: ACNase, anticodon nuclease; CDK, cyclin-dependent kinase; dsRNA, double-stranded RNA; HAT, histone acetylase; kti, killer toxin insensitive; mcm5s2U, 5-methoxycarbonylmethyl-2-thiouridine; mcm5U, 5-methoxycarbonylmethyluridine; M(IP)2C, mannosyl-di-inositol phosphoceramide; SAM, S-adenosylmethionine; TOT, toxin target effector; U34, wobble uridine; UGPase, UDP-glucose pyrophosphorylase
- © The Authors Journal compilation © 2007 Biochemical Society