Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2015 Apr 24.
Published in final edited form as: Cell Rep. 2014 Apr 13;7(2):339–347. doi: 10.1016/j.celrep.2014.03.034

Structure, Mechanism, and Specificity of a Eukaryal tRNA Restriction Enzyme

Anupam K Chakravarty 1, Paul Smith 1, Radhika Jalan 1, Stewart Shuman 1,✉
PMCID: PMC4121121  NIHMSID: NIHMS579647  PMID: 24726365

Summary

tRNA restriction by anticodon nucleases underlies cellular stress responses and self-nonself discrimination in a wide range of taxa. Anticodon breakage inhibits protein synthesis, which in turn results in growth arrest or cell death. The eukaryal ribotoxin PaT elaborated by Pichia acaciae inhibits growth of Saccharomyces cerevisiae via cleavage of tRNAGln(UUG). We find that recombinant PaT incises a synthetic tRNAGln(UUG) stem-loop RNA by transesterification at a single site 3′ of the wobble uridine, yielding 2′,3′-cyclic phosphate and 5′-OH ends. Incision is suppressed by replacing the wobble nucleobase with adenine or guanine. We report the crystal structure of PaT, which unveils a novel fold and active site, essential constituents of which are illuminated by mutagenesis. Pichia acaciae evades self toxicity via a distinctive intracellular immunity protein, ImmPaT, which binds PaT and effaces nuclease activity. Our results highlight the evolutionary diversity of tRNA restriction and immunity systems.

Introduction

The yeast Pichia acaciae secretes a tRNA anticodon nuclease toxin encoded by a resident cytoplasmic linear DNA “killer” plasmid (Satwika et al., 2012). The secreted form of the ribotoxin is a complex of PaOrf1 and PaOrf2 subunits that arrests growth of the non-self yeast Saccharomyces cerevisiae (Fig. 1A). The chitin-binding PaOrf1 subunit interacts with the target cell surface to effect the delivery of the PaOrf2 subunit – the Pichia acaciae toxin, PaT – into the cytoplasm of the target cell. PaT elicits toxicity in vivo via cleavage of the anticodon loop of tRNAGln(UUG) (Fig. 1B), thereby depleting the functional pool of this isoacceptor (Klassen et al. 2008). To prevent self-killing, either by autocrine uptake of secreted toxin or by residual free cytoplasmic PaT, the killer plasmid encodes an immunity factor ImmPaT (the PaOrf4 protein) that protects P. acaciae via an unknown mechanism (Paluszynski et al., 2007).

Fig. 1. PaT is toxic in vivo and incises a synthetic tRNAGln stem-loop in vitro at a unique site 3′ of the wobble uridine.

Fig. 1

Open in a new tab

(A) Secreted tRNA ribotoxins defend fungi against non-self species. Pichia acaciae harbors a killer plasmid episome that encodes a secreted heterodimeric toxin. The chitin-binding Orf1 subunit interacts with the surface of S. cerevisiae target cells and mediates delivery of the toxic PaT (Orf2) subunit into the cytoplasm, where it arrests the growth of the target cell. (B) PaT is a tRNA anticodon nuclease that incises tRNAGln(UUG). (C) Ribotoxicity assay. Schematic depiction of yeast cells harboring the pGAL-PaT plasmid (2μ LEU2) in the presence of glucose or galactose as the carbon source. In this experment, the N-terminal peptide sequence of PaT was varied as indicated on the left. The reference N-terminal sequence of the unprocessed PaOrf2 protein is shown at bottom, with the presumed signal peptide in lower case and colored blue. By convention, the asparagine is designated amino acid 2 of the intracellular PaT (Meineke et al., 2012). Aliquots of serial dilutions of the indicated S. cerevisiae pGAL-PaT strains were spotted on SD-Leu– agar containing 2% glucose or 2% galactose. (D) Purified recombinant PaT. An aliquot (6 μg) of purified gel-filtered PaT–His6 was analyzed by SDS-PAGE. The Coomassie blue-stained gel is shown. The positions and sizes (kDa) of marker polypeptides are indicated on the right. (E) Anticodon nuclease activity. RNase reaction mixtures (10 μl) containing 10 mM Tris-HCl (pH 7.5), 50 mM NaCl, 1 mM DTT, 0.1 μM 5′ 32P-labeled 27-mer stem-loop RNA (shown at right with the UUG anticodon colored magenta and the 5′-labeled nucleotide in red) and either no protein (lane –) or 0.063, 0.125, 0.25 0.5 1, or 2 μM purified PaT–His6 (increasing from left to right) were incubated at 30°C for 60 min. The products were analyzed by urea-PAGE in parallel with size marker ladders generated by: (i) partial alkaline hydrolysis of the same 5′ 32P-labeled 27-mer RNA substrate (lane OH); (ii) partial alkaline hydrolysis of an otherwise identical 27-mer in which the wobble nucleoside was U2′F and thereby alkali-resistant (OH*); and (iii) treatment of the 32P-labeled 27-mer RNA with RNase T1, which cleaves 3′ of the lone unpaired guanosine in the anitcodon loop to yield a 5′-labeled 15-mer with a 3′-phosphate end. An autoradiograph of the gel is shown; the identities of the 3′ terminal nucleotides are indicated next to the ladder. See also Figures S1 and S2.

The yeast Kluyveromyces lactis has an analogous system of self-nonself discrimination via the secreted toxin zymocin, the γ-toxin subunit of which is an anticodon nuclease that targets tRNAGlu(UUC) (Lu et al., 2005). In addition to their primary targets, PaT and γ-toxin can also cleave, albeit to a lesser extent, the two other yeast tRNAs that have an mcm5s2U wobble base: tRNAGlu and tRNALys for PaT; tRNAGln and tRNALys for γ-toxin (Lu et al., 2005; Klassen et al., 2008). Whereas γ-toxin is strictly dependent on the wobble mcm5 modification for its ribotoxicity, PaT is not. It is noteworthy that, despite their tRNA substrate overlap, PaT has no apparent primary structure similarity to K. lactis γ-toxin. Nor does the PaT primary structure resemble any of the exemplary bacterial tRNA anticodon nucleases: PrrC (Blanga-Kanfi et al., 2006), colicin E5 (Ogawa et al., 1999), colicin D (Tomita et al., 2000), or VapC (Winther and Gerdes, 2011). Indeed, PaT has no similarity to any known nucleases or phosphotransferases. Its sole retrievable homolog is a plasmid-encoded toxin of unknown target specificity from the yeast Debaryomyces robertsiae (Klassen et al., 2004).

The fungal ribotoxins are a fertile area for exploring the mechanisms, structures, and evolution of eukaryal tRNA restriction enzymes, which are of heightened interest in light of recent discoveries of tRNA restriction as a general eukaryal response to cellular stress (Thompson and Parker, 2009; Saikia et al., 2012). γ-toxin has been studied genetically and biochemically. It has been mutagenized extensively and its cleavage mechanism and specificity have been defined using native tRNAs and synthetic RNA substrates that mimic the anticodon stem-loop of tRNAGlu (Lu et al., 2005; 2008; Keppetipola et al., 2009, Jain et al., 2011). Yet, γ-toxin has so far eluded a structure determination. By comparison, knowledge of PaT is sparse. Here we purify biologically active homogenous PaT, characterize its anticodon nuclease activity, determine its atomic structure by X-ray crystallography, and illuminate structure-activity relationships. We also purify ImmPaT and determine how it interdicts self-killing.

Results

Purification of bioactive PaT

The bioactivity of PaT can be assayed by galactose-induced intracellular expression in S. cerevisiae of a version of the PaT protein that lacks the predicted N-terminal signal peptide and instead has the N-terminal sequence MNPTTCLNE in which a new initiating methionine is appended to the native asparagine (Klassen et al., 2004) (Fig. 1C). Here we found that toxicity was unaffected when the translation start was shifted one residue to the left, by encoding the N-terminus MGNPTTCLNE in which the glycine derives from the native PaT (Fig. 1C). In this context, the N-terminal methionine will be removed by yeast methionine aminopeptidase (Moerschell et al., 1990). By phasing the start site to the right in single amino acid increments, we determined that replacing Asn2 with methionine did not affect the toxicity of PaT, as gauged by inhibition of cell growth on medium containing galactose (Fig. 1C). By contrast, deleting Asn2 and replacing Pro3 with methionine, or deleting Asn2-Pro3 and replacing Thr4 with methionine abolished PaT toxicity and allowed growth on galactose (Fig. 1C).

Based on these findings, we elected to purify and characterize the biologically active version of PaT with the encoded N-terminal sequence MPTTCLNE. The PaT protein was produced in E. coli with a C-terminal His6 tag and purified by sequential Ni-affinity, cation exchange and gelfiltration chromatography steps. The N-terminal sequence of the apparently homogeneous recombinant PaT–His6 (Fig. 1D) was determined by automated Edman microchemistry to be PTTCLNEGAI, signifying that the encoded methionine was removed by E. coli methionine aminopeptidase.

Anticodon nuclease activity: transesterification chemistry and site specificity

To gauge anticodon nuclease activity, we incubated PaT with a 5′ 32P-labeled 27-mer RNA oligonucleotide substrate that comprises the anticodon stem-loop of yeast tRNAGln(UUG) plus five extra base-pairs at the top of the stem (Fig. 1E). Denaturing PAGE analysis in parallel with a partial alkaline hydrolysate of the 27-mer showed that PaT incised the anticodon loop at the phosphodiester 3′ of the wobble uridine to yield a single 5′ 32P-labeled cleavage product that migrated exactly on the alkaline ladder (Fig. 1E), signifying that the PaT cleavage reaction yielded a 5′-OH end on the distal (unlabeled) fragment and either a 2′,3′-cyclic phosphate or monophosphate end on the proximal (labeled) fragment. Definitive assignment of the cleavage site 3′ of the wobble uridine was made by: (i) comparison to a partial alkaline hydrolysate of an otherwise identical 27-mer RNA with a 2′-fluorine in lieu of the 2′-OH of the wobble uridine sugar, resulting in a skipped hydrolysis site in the alkaline ladder at the same site incised by PaT; and (ii) comparison to a digest of the 27-mer with RNase T1, which cleaved 3′ of the unpaired guanosine in the anticodon loop (Fig. 1E). The extent of incision of 100 nM RNA substrate by PaT was proportional to the protein concentration from 63 to 500 nM and attained an endpoint of ∼90% cleavage at 2 μM PaT (Figs. 1E and S1A). The apparent rate constant for anticodon incision under single-turnover conditions (2 μM PaT; 0.1 μM RNA) was 0.12 min−1 (Fig. S1B).

Definitive assignment of the terminus of the proximal cleavage fragment as a 2′,3′-cyclic phosphodiester was made by reacting PaT with a shorter 5′ 32P-labeled RNA substrate that would allow us to discriminate cleavage products with cyclic phosphodiester versus phosphomonoester ends. For this purpose, we employed a 17-mer exact mimetic of the tRNAGln(UUG) anticodon stem-loop (Fig. S2A) and found that the 5′ 32P-labeled 8-mer product of PaT cleavage 3′ of the wobble uridine co-migrated with the upper band of a radiolabeled doublet in the alkaline hydrolysis ladder. The split doublet bands reflect resolution by the PAGE system of otherwise identical short RNAs according to whether they have a 2′,3′-cyclic phosphate end (the upper species) or a 2′- or 3′- monophosphate end (the lower species) (Das and Shuman, 2013).

These results illuminated the chemical mechanism of the PaT anticodon nuclease, which catalyzes a single transesterification in which the O2′ atom of the wobble nucleoside sugar makes a nucleophilic attack on the 3′-5′ phosphodiester to form an RNA 2′,3-cyclic phosphodiester and expel a 5′-OH RNA leaving strand (Fig. S2B). Consistent with this mechanism, we find that anticodon cleavage by PaT was abolished by replacing the ribose 2′-OH of the wobble uridine in the 27-mer stem-loop substrate with a 2′-fluorine (Fig. 2A). Note that prevention of RNA transesterification chemistry at the “normal” wobble uridine site did not precipitate cleavage at alternative sites within the anticodon loop.

Fig. 2. Effect of wobble base substitutions and ribose modification.

Fig. 2

Open in a new tab

(A) Reaction mixtures (10 μl) containing 0.1 μM 5′ 32P-labeled 27-mer stem-loop RNA with the wobble nucleoside as specified (U, U2′F, A, C or G) and either 2 μM PaT (lanes +) or no enzyme (lanes –) were incubated at 30°C for 60 min. The products were analyzed by urea-PAGE in parallel with partial alkaline hydrolysates of each of the 32P-labeled 27-mer RNAs (lanes OH). The species in each alkaline ladder corresponding to the 5′-fragment cleaved 3′ of the wobble nucleoside is denoted by ●. (B) The extents and sites of incision by PaT of the 27-mer RNAs with U, C, A or G wobble nucleosides are plotted in bar graph format, where the anticodon loop sequences are arrayed on the x-axes, with the variable wobble nucleobases highlighted in the gray boxes. Each datum in the graphs is the average of three separate experiments ±SEM. Note that the y-axis scales are different in each graph.

The anticodon nuclease activity of PaT did not require a divalent cation cofactor. Inclusion of 10 mM EDTA in the reaction mixtures was without effect on the extent or site of cleavage of the 27-mer stem-loop RNA; by contrast, inclusion of 10 mM MgCl2 inhibited cleavage (not shown). Metal-independence (and inhibition by Mg2+) is a property shared with K. lactis γ-toxin (Keppetipola et al., 2009), which, like PaT, cleaves 3′ of a wobble uridine via a transesterification mechanism. γ-toxin, which requires the mcm5 wobble uridine modification for its toxicity, is unable to cleave synthetic oligonucleotide substrates that mimic the unmodified tRNAGlu(UUC) anticodon stem-loop unless the reactions are supplemented with high concentrations (1 to 2 M) of the osmolyte trimethylamine oxide (TMAO) (Lu et al., 2008; Keppetipola et al., 2009). By contrast, PaT efficiently cleaved an unmodified tRNAGln(UUG) anticodon stem-loop substrate in the absence of any additive, consistent with the toxicity of PaT to a yeast elp3Δ strain that cannot synthesize the mcm5 moiety (Klassen et al., 2008). Nonetheless, we found that 2 M TMAO stimulated PaT anticodon nuclease specific activity fourfold (Fig. S1A) and increased the single-turnover cleavage rate constant four-fold (Fig. S1B), without affecting the site specificity of cleavage 3′ of the wobble uridine (Fig. S2A and data not shown). These experiments unambiguously define the wobble 3′ phosphodiester as the target for PaT anticodon incision and they cast doubt on earlier interpretations of PaT specificity (Klassen et al., 2008), based on Northern blot analysis of the products of tRNA cleavage by a crude PaT-His6 preparation, imputing that: (i) PaT cleaves tRNAGln(UUG) at two sites, one 3′ of the wobble uridine and one two nucleotides upstream; (ii) PaT cleaves tRNAGln(UUG) lacking the mcm5U modification only at the site two nucleotides upstream of the wobble nucleoside; and (iii) mcm5 is required for cleavage at the wobble uridine.

Anticodon wobble base recognition

To probe the role of the wobble nucleobase, we presented PaT with a set of otherwise identical 27-mer stem-loop substrates in which the wobble position was either U, C, G or A and analyzed the reaction products by urea PAGE (Fig. 2A). The extents of cleavage at nucleotides within the anticodon loop sequence are plotted in bar graph format for each substrate in Fig. 2B. (Note that the y-axis scales are different in each graph.) The salient findings were as follows: (i) the wobble cytosine reduced cleavage efficiency (to 34%) without affecting cleavage site choice; (ii) the wobble adenine effaced cleavage at the wobble nucleoside and shifted the incision site by one nucleotide to the upstream uridine, which was cleaved less effectively (23%) than the normal wobble U; and (iii) the wobble guanine strongly inhibited cleavage at the wobble position (2.2%) and elicited only feeble cleavage at the flanking 5′ uridine (2.7%) and 3′ uridine (1.4%) positions. We conclude that PaT “sees” the wobble pyrimidine nucleobase in tRNAGln as a key specificity determinant, with a preference for U over C. These biochemical properties of pure PaT are consistent with genetic evidence that the tRNAGln(CUC) isoacceptor is a secondary target that contributes to PaT toxicity in vivo (Klassen et al., 2008). Note that none of the wobble base substitutions elicited cleavage at the uridine two nucleotides upstream of the wobble nucleoside.

Crystal structure of Pichia acaciae toxin

We produced two versions of recombinant PaT for crystallization trials: N(Ser)•PaT and PaT–His6. The N(Ser)•PaT protein was derived from a His10Smt3•PaT fusion produced at high levels in bacteria and recovered by Ni-affinity chromatography. Cleavage of the His10Smt3 domain by the Smt3-protease Ulp1 yielded a version of PaT with N-terminal sequence SMNPTTCLNE, in which the terminal Ser residue originates from the Smt3 fusion junction. Ulp1 cleavage of His10Smt3•PaT was inefficient (compared to many other Smt3 fusions we have worked with), requiring prolonged incubation with high levels of Ulp1, after which there was a significant fraction of residual intact His10Smt3•PaT. Although this was not an obstacle to obtaining large amounts of pure N(Ser)•PaT, it suggested that the N-terminus of PaT at the Smt3•PaT junction was relatively inaccessible. We grew crystals of native N(Ser)•PaT by hanging drop vapor diffusion against a precipitant solution containing 100 mM sodium citrate (pH 5.6), 3.0-3.5 M (NH4)2SO4, 100 mM magnesium acetate. The native N(Ser)•PaT crystal diffracted X-rays to 1.8 Å resolution and belonged to space group P212121. The structure was solved by using SAD phases obtained for an isomorphous crystal of SeMet-substituted N(Ser)•PaT that diffracted X-rays to 2.1 Å resolution. The refined model of native N(Ser)•PaT at 1.8 Å resolution (Rwork/Rfree = 0.160/0.196; Table S1) consisted of two protomers in the asymmetric unit. The A protomer model extended from the N-terminal junction-derived serine to the native C-terminal Val319, with a two amino acid gap at residues 307-308. The B protomer model extended from N(Ser) to Lys318. The A and B protomers of N(Ser)•PaT superimposed with an rmsd of 0.75 Å at 315 Cα positions. The N(Ser)•PaT protomers were monomeric in the crystal, consistent with the elution of N(Ser)•PaT as a monomer during preparative gel filtration.

N(Ser)•PaT was obtained in 40-fold higher yield than PaT–His6 when expressed in E. coli, a finding that was puzzling until we assayed N(Ser)•PaT for RNA cleavage and found it to be inert. We suspect that production of catalytically active PaT–His6 protein in E. coli was self-limited by PaT's inhibition of protein synthesis. To ensure that we captured a structure of an active form of PaT, we grew crystals of PaT–His6, which were isomorphous to those of N(Ser)•PaT, and then solved the structure of PaT–His6 at 2.9 Å resolution (Rwork/Rfree = 0.188/0.247) by refinement of the data against the N(Ser)•PaT model (Table S1). The refined model of PaT–His6 extended continuously from Pro3 (the N-terminus of the protein; Fig. 1D) to Lys318 in the A protomer, and from Pro3 to Lys318 with a two amino acid gap at residues 308-309 in the B protomer.

Except for the extra N-terminal Ser-Met-Asn peptide in N(Ser)•PaT, the tertiary structures of inactive N(Ser)•PaT and active PaT–His6 were virtually identical. The two A protomers aligned with 0.45 Å rmsd at 312 Cα positions and the two B protomers with 0.54 Å rmsd at 313 Cα positions. A superposition of the A protomer structures (Fig. S3) showed that the N-terminus of PaT is located within a recessed pocket (explaining the relative resistance of the His10Smt3•PaT fusion protein to cleavage by Ulp1) and that the extra N-terminal peptide occluded the ribonuclease active site (discussed below).

Overview of the PaT tertiary structure

PaT is a single domain protein composed of a helical core ‘book-ended’ between two four-strand β-sheets (Fig. 3A). The core comprises fourteen α-helices and two 310 helices. The secondary structure elements are displayed above the PaT amino acid sequence in Fig. 3C. Each of the β-sheets is made up of a pair of β-hairpins. The topology of the sheet at the right of the image in Fig. 3A is β2↓•β1↑•β8↑•β7↓. The topology of the sheet at left is β4↓•β3↑•β5↓•β6↑.

Fig.3. Crystal structure of PaT reveals a unique fold.

Fig.3

Open in a new tab

(A) The tertiary structure of PaT–His6 is shown as a ribbon trace with magenta β strands, cyan α helices, blue 310 helices, and beige intervening loops and turns. The N and C termini of the polypeptide are indicated; selected α helices and all β strands are numbered according to their order in the primary structure. (B) The surface model of PaT in the same orientation as panel A and its vacuum electrostatics were generated in Pymol. Two sulfate anions coordinated on the protein surface are shown as stick models. A chloride anion is shown as a green sphere. (C) The aligned primary structures of the Pichia acaciae (Pac) PaT polypeptide and the homologous toxin from Debaryomyces robertsiae (Dro) are shown. Positions of side chain identity/similarity are denoted by ● above the sequence. Gaps in the alignment are indicated by dashes. The secondary structure elements of PaT are shown above the amino acid sequence, with β strands rendered as arrows and helices as cylinders, colored as in panel A. The amino acids in PaT that were subjected to alanine scanning in the present study are highlighted in shaded boxes: in green for the essential residues and gray for the nonessential residues. See also Figure S4.

A DALI search of the protein database revealed that PaT has a unique fold; no “hits” were recovered with a Z score >3.5. That PaT is sui generis in its tertiary structure accords with its lack of primary structure similarity to any polypeptide except the Debaryomyces toxin (Fig. 3C). The salient point is that the PaT structure is wholly unlike the known structures of any other transesterifying ribonucleases, e.g., tRNA ribotoxins colicin E5 and colicin D, tRNA splicing endonuclease, mRNA ribotoxin MazF, rRNA ribotoxin restrictocin, or generic ribonucleases barnase, RNase A and RNase T1 (see comparisons in Fig. S4).

A surface electrostatic image of PaT (shown in Fig. 3B in the same orientation as the ribbon tertiary fold image in Fig. 3A) highlights an extended area of positive potential (colored blue), suggestive of an RNA-binding surface, on which two lateral sulfate anions and a central chloride anion are docked (Fig. 3B).

The PaT active site

A stereo view of the PaT–His6 anion-binding pocket, with a sulfate and chloride engaged, is shown in Fig. 4A. The chloride anion is surrounded by Pro3, Thr4, and Thr5. We speculate that the chloride overlaps the position of the scissile phosphate, which would account for the lack of in vivo toxicity of the PaT variants with alternative translation start sites that replace Pro3 and Thr4 with Met (Fig. 1C). Three basic amino acid side chains are in the vicinity of the chloride: Arg172, Lys175, and His287. Arg172 is a central component of a hydrogen-bonding network that includes Thr5, Glu9, and His138 (Fig. 4A). Lys175 contacts the main-chain carbonyls of Tyr83 and Gly83 and makes a cation–π interaction with the Tyr83 aromatic ring. The nearby sulfate anion is coordinated by His138 and the main-chain amide of Leu124; Lys125 is also proximal to the sulfate (Fig. 4A).

Fig. 4. Distinctive active site of PaT.

Fig. 4

Open in a new tab

(A) Stereo view of the putative active site of PaT–His6 with selected amino acid side chains and main chain atoms depicted as stick models with beige carbons. A sulfate anion is rendered as a stick model; a chloride anion is depicted as a green sphere. Atomic contacts are indicated by dashed lines. Site side chains found, by alanine scanning, to be essential for toxicity are denoted by red labels. (B) Effects of alanine mutations on in vivo toxicity. Aliquots of serial dilutions of the S. cerevisiae pGAL-PaT strains, expressing wild-type PaT or the indicated alanine mutants, were spotted on SD-Leu– agar containing 2% glucose or 2% galactose. Cells carrying the empty 2μ vector were tested in parallel as controls. Essential residues were those at which alanine substitution permitted growth on galactose. (C) Effects of selected PaT mutations on anticodon nuclease activity in vitro. Aliquots (5 μg) of the SP-Sepharose preparations of wild type PaT–His6 and the R172A, K175A, S283A and H287A mutants were analyzed by SDS-PAGE. The Coomassie blue-stained gel is shown in the left panel. The PaT–His6 proteins (2 μM) were reacted with the 5′ 32P-labeled 27-mer stem-loop RNA (0.1 μM) and 2 M TMAO for 60 min at 30°C. The extents of anticodon cleavage are plotted in the bar graph in the right panel; each datum is the average of three separate experiments ±SEM. See also Figure S3.

To test the relevance of candidate active site functional groups to PaT bioactivity, we replaced Thr5, Glu9, His138, Arg172, Lys175, and His287 individually with alanine. We also introduced alanine in lieu of nearby residues Ser283 and Asp286, and basic residues Lys99 and Lys240 located on the positively charged surface of PaT, albeit at a distance from the active site. The PaT-Ala mutants were tested for galactose-dependent inhibition of S. cerevisiae growth (Fig. 4B). The alanine scan revealed that Thr5, Glu9, His138, Arg172, and Lys175 were essential for PaT bioactivity, as judged by the fact that yeast cells expressing the respective alanine mutants were able to grow on galactose medium (Fig, 4B). By contrast, PaT retained its toxicity when Lys99, Lys125, Lys240, Ser283 and Asp286 were mutated to alanine, signifying that these side chains are inessential in vivo.

Selected alanine mutations were introduced into the PaT-His6 context and the PaT-Ala proteins were purified from E. coli (Fig. 4C) and tested for anticodon nuclease activity. The R172A, K175A and H287A mutants, which were non-toxic in vivo, were inactive as anticodon nucleases in vitro (Fig. 4C). These inactive PaT-His6 mutants were produced and purified in higher amounts (4 to 5-fold) than the wild-type PaT-His6, consistent with PaT nuclease activity exerting a brake on PaT expression. The bioactive S283A variant retained anticodon nuclease activity in vitro, incising the 27-mer stem-loop substrate 3′ of the wobble uridine with 52% yield (versus 90% for wild-type). These mutational data are consistent with our assignment of the nuclease active site to the pocket shown in Fig. 4A.

The constellation and spatial arrangement of amino acids in the PaT active site is distinct from other ribonucleases. The classic mechanism of RNA cleavage by metal-independent transesterification relies on general base catalysts to abstract a proton from the ribose O2′ nucleophile, general acid catalysts to donate a proton to the O5′ leaving strand, and basic residues or other hydrogen bond donors to stabilize the pentacoordinate transition state of the scissile phosphate. The structure and structure-guided mutational data recommend Glu9, His287 and/or His138 as potential acid-base catalysts and Arg172 in transition state stabilization. It is noteworthy that the superposition of the inactive N(Ser)•PaT structure on active PaT–His6 structure highlights perturbations of the active site, particularly by the non-native N-terminal Sertag residue, which is tethered in place via hydrogen bonds to Sertag-Oγ and the main-chain carbonyl from the essential Arg172 side chain and from the Sertag-Oγ and N-terminal amino nitrogen to Asp286. With the extra N-peptide docked in this conformation, the Asn2 side chain occupies the same position as the chloride ion the PaT-His6 active site. Thus, the extra peptide occludes the active site and likely accounts for why the N(Ser)•PaT is inert as an anticodon nuclease.

Mechanism of ImmPaT protection from self-killing by PaT

The P. acaciae cytoplasmic plasmid encodes an immunity factor ImmPaT that protects the yeast from being killed by uptake of extracellular toxin. The 363-amino acid ImmPaT protein has no primary structure similarity to the γ-toxin immunity factor and no resemblance to any protein in the NCBI database other than the homologous immunity factor from the yeast Debaryomyces robertsiae. ImmPaT acts intracellularly to neutralize intracellular PaT by an unknown mechanism (Paluszynski et al., 2007). In addition to protecting the toxin-producing yeast, ImmPaT exerts strong selection pressure to maintain the killer plasmid during vertical transmission, because daughter cells of mitotic division that lose the plasmid will be killed by their toxin-secreting sisters.

In order to query the mechanism of immunity, we produced ImmPaT in bacteria and purified the recombinant protein (Fig. 5A). The 44 kDa ImmPaT protein gel-filtered as a monomer (Fig. 5C) and had no ribonuclease activity itself when reacted in 40-fold molar excess with the anticodon stem loop substrate (Fig. 5B). We found that simply mixing ImmPaT with PaT–His6 at molar ratios of 0.5:1, 1:1, and 2:1 elicited progressive inhibition of the anticodon nuclease activity of PaT, with complete effacement of the nuclease at a 2:1 ratio of ImmPaT:PaT (Fig. 5B). This was a specific effect of ImmPaT, insofar as a two-fold excess of BSA had no effect on PaT (Fig. 5B).

Fig. 5. Pichia acaciae toxin immunity protein (ImmPaT) binds PaT in vitro and effaces its endoribonuclease activity.

Fig. 5

Open in a new tab

(A) Recombinant ImmPaT. An aliquot (6 μg) of purified ImmPaT was analyzed by SDS-PAGE. The Coomassie blue-stained gel is shown. The ImmPaT polypeptide is denoted by ◄. The positions and sizes (kDa) of marker polypeptides are indicated on the left. (B) ImmPaT inhibits PaT anticodon nuclease activity. PaT–His6 and either ImmPaT or BSA in amounts specified were preincubated for 5 min at 30°C, then reacted with 0.1 μM (1 pmol) of 5′ 32P-labeled 27-mer stem-loop RNA for 60 min at 30°C. The products were analyzed by urea-PAGE and visualized by autoradiography. (C and D) PaT•ImmPaT complex formation. PaT–His6 alone, ImmPaT alone, and a mixture of PaT–His6 and ImmPaT (1:1.5 molar ratio; preincubated for 5 min at 30°C) were gel-filtered through a 25-ml Superdex 200 column in 10 mM Tris-HCl, pH 7.5, 50 mM NaCl, 1 mM DTT. The superimposed chromatographic profiles (monitored by A280 as a function of elution volume) are shown as solid lines in panel C. The dotted line in panel C depicts the elution profile of a mixture of ferritin, aldolase, conalbumin, ovalbumin, and carbonic anhydrase, with native molecular weights (kDa) indicated by arrows, that was used to calibrate the column. Panel D shows the polypeptide composition of the serial column fractions from the gel-filtration profile of the mixture of PaT–His6 and ImmPaT. The fractions corresponding to the PaT•ImmPaT complex are indicated by the bracket. (E) The PaT•ImmPaT complex has no anticodon nuclease activity. 5′ 32P-labeled 27-mer stem-loop RNA (20 nM) was reacted with either 0.4 μM PaT–His6 or PaT•ImmPaT complex (from the peak Superdex fraction denoted by ● in panel D), for 60 min at 30°C. The products were analyzed by urea-PAGE and visualized by autoradiography.

To test whether PaT and ImmPaT interact physically, we mixed the recombinant PaT–His6 and ImmPaT proteins (1:1.5 molar ratio) and subjected the mixture to gel filtration, which revealed the formation of an ∼80 kDa protein complex that was not present when PaT or ImmPaT alone were gel filtered (Fig. 5C). SDS-PAGE analysis of the eluate fractions showed that the ∼80 kDa peak comprised a PaT•ImmPaT heterodimer (Fig. 5D). The isolated PaT•ImmPaT complex was devoid of anticodon nuclease activity (Fig. 5E). Thus, the immunity mechanism of the fungal ribotoxin system entails direct binding of the immunity factor to the toxin and masking of its nuclease activity, presumably via steric occlusion of the ribotoxin's RNA binding site by the antitoxin, as has been seen with other ribotoxin•antitoxin pairs (Luna-Chavez et al., 2006; Yajima et al., 2006; Graille et al.,2004; Simanshu et al., 2013).

Discussion

tRNA restriction via incision of the anticodon loop is a deeply rooted biological response to diverse stresses, including virus infection, starvation, and reactive oxygen species (Thompson and Parker, 2009). tRNA restriction also plays a prominent role in self-nonself distinction, especially among microbial taxa. Anticodon nucleases elaborated by bacteria and fungi have extraordinary specificity for their tRNA targets, often dictated by recognition of a chemically modified nucleobase, e.g., wobble queosine in the case of colicin E5 (Yajima et al., 2006), wobble mcm5s2U in the case of γ-toxin (Lu et al., 2005), and wobble mnm5s2U plus t6A37 in the case of PrrC (Jiang et al., 2002). Among tRNA restriction enzymes that recognize the wobble base, the site of cleavage can either be at the wobble 3′-phosphodiester (colicin E5 and γ-toxin) or at the wobble 5′-phosphodiester (PrrC).

Creating tRNA breaks with a bulky modified base at one of the broken ends may exploit an Achilles heel in the target cell's capacity to repair the tRNA, which in S. cerevisiae can be impeded by the modified base (Nandakumar et al., 2008; Meineke and Shuman, 2012). The bacterial tRNA restriction enzyme RloC (a distant cousin of PrrC) circumvents the potential for tRNA repair that thwarts PrrC (Amitsur et al., 1987), by incising the anticodon loop at two sites, 3′ and 5′ of the wobble nucleoside, resulting in excision of the intervening wobble mononucleotide (Davidov and Kaufmann, 2008; Klaiman et al., 2012), so that repair would result in a non-functional tRNA with an internal deletion. It was suggested that PaT also makes two breaks in the anticodon loop of its targets – 3′ of the wobble uridine and two nucleotides upstream – based on the observation of a doublet cleavage product by Northern blotting of the products of a reaction of partially purified PaT with total yeast tRNA (Klassen et al., 2008; Meineke et al., 2012). However, this interpretation was not bolstered by direct identification of the positions and chemical structures of the broken RNA ends (Klassen et al., 2008), nor was it demonstrated that the predicted dinucleotide excision product was formed. When the mcm5U wobble modification was missing, only the lower band of the doublet was observed, which was interpreted to indicate that PaT makes only one incision at a site two nucleotides upstream of the unmodified wobble U (Klassen et al., 2008).

In the present study, we show that PaT cleaves a model tRNAGln(UUG) anticodon stem-loop at a single site 3′ of the unmodified wobble uridine, via a metal-independent transesterification mechanism that yields 2′,3′-cyclic phosphate and 5′-OH RNA ends. PaT never cleaves at the uridine two nucleotides upstream of the wobble position: not when the wobble cleavage is suppressed chemically by the wobble 2′-F; not when the usual cleavage is abolished by installing an adenine wobble base and the cleavage site shifts to the uridine one nucleotide upstream. Therefore, we conclude that PaT is an incisional tRNA restriction enzyme, with inherent specificity for a wobble pyrimidine, and limited plasticity with respect to phasing of the scissile uridine-3′-phosphodiester within the anticodon loop.

The crystal structure of PaT presented here is, to our knowledge, the first of a cytotoxic eukaryal tRNA restriction enzyme. In keeping with PaT's distinctive biochemical specificity and seemingly unique amino acid sequence devoid of instructive motifs, we found that PaT has a novel tertiary structure and a distinctive active site. Thus, PaT is the founder of a new ribonuclease clade, with only one other recognizable fungal member at this point. Structures are available for a variety of metal-independent transesterifying ribonucleases with diverse specificities and biological functions (Fig. S4). Although they display distinctive folding topologies in their own right, these RNases have in common predominantly β structures and active sites formed by β strands. By contrast, PaT has a predominantly α structure, including at the active site. It will be interesting to see whether other fungal ribotoxins that are not related by amino acid sequence to PaT (e.g., γ-toxin) have any structural similarity to PaT, or whether they too are sui generis.

The unveiling of PaT as a structurally unique toxin hints that we are seeing just the tip of the iceberg of the diversity and evolutionary complexity of the ribotoxin repertoire. In the same vein, our studies of ImmPaT highlight that the underpinnings of toxin immunity are just as diverse, albeit organized around a common principle that immunity reflects a physical inhibition of toxin nuclease function.

In addition, we can envision practical applications of PaT, a eukaryal tRNA restriction enzyme, and its antagonist ImmPaT, as portable agents of tunable eukaryal ribotoxicity. For example, mammalian cell growth arrest or apoptosis in response to tRNA restriction might be triggered by: (i) simply inducing PaT expression; or (ii) co-expressing PaT and ImmPaT and then shutting off expression of the immunity protein. tRNA ribotoxicity is regarded as a potential cancer therapeutic (Ardelt et al., 2009). In bacteria, tRNA restriction was first discovered as an antiviral mechanism, one that is evaded by virus-encoded RNA repair system (Amitsur et al., 1987). The coupling of PaT expression to a eukaryal virus transcriptional program is a potential means to selectively eliminate virus-infected cells.

Experimental Procedures

Assays of PaT activity

Galactose-inducible PaT toxicity was scored as described in Fig. 1C; the details of plasmid construction and mutagenesis are provided in Supplemental Experimental Procedures. Anticodon nuclease activity was assayed as specified in Fig. 1E. Detailed methods for radiolabeling an isolating the RNA substrates and for preparation of RNA markers used to map the cleavage sites are described in Supplemental Experimental Procedures.

Recombinant proteins, crystallization and structure determination

PaT–His10, N(Ser)•PaT, SeMet-substituted N(Ser)•PaT, and ImmPaT–His6 were produced in E. coli and purified from soluble bacterial lysates by Ni-affinity, ion exchange, and gel-filtration chromatography steps as described in detail in Supplemental Experimental Procedures. Crystals of N(Ser)•PaT, SeMet-substituted N(Ser)•PaT, and PaT–His6 were grown at 22°C by hanging drop vapor diffusion against reservoirs of precipitant solutions containing 100 mM sodium citrate (pH 5.6), 3.0-3.5 M (NH4)2SO4, and either 100 mM magnesium acetate or 200 mM LiCl. Diffraction data collection and structure determinations are described in detail in Supplemental Experimental Procedures. The data and refinement statistics and model contents are summarized in Table S1.

Accession numbers

The coordinates for the refined models of N(Ser)•PaT and PaT–His6 have been deposited in the RCSB protein structure database (PDB ID codes 4087 and 4088).

Supplementary Material

01

Highlights.

  • Microbes elaborate tRNA anticodon nuclease toxins to interdict non-self species

  • Fungal ribotoxin PaT incises tRNAGln anticodon stem-loop 3′ of the wobble uridine

  • The crystal structure of PaT reveals a distinctive fold and active site

  • Toxin immunity is conferred by ImmPaT, a tRNase inhibitor

Acknowledgments

We thank Annie Heroux at NSLS for assistance with data collection. This research was supported by NIH grant GM42498 (S.S.). S.S. is an American Cancer Society Research Professor.

Footnotes

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

References

  1. Amitsur M, Levitz R, Kaufman G. Bacteriophage T4 anticodon nuclease, polynucleotide kinase, and RNA ligase reprocess the host lysine tRNA. EMBO J. 1987;6:2499–2503. doi: 10.1002/j.1460-2075.1987.tb02532.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  2. Ardelt W, Ardelt B, Darzynkiewicz Z. Ribonuclease as potential modalities in anticancer therapy. Eur J Pharmacol. 2009;625:181–189. doi: 10.1016/j.ejphar.2009.06.067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Blanga-Kanfi S, Amitsur M, Azem A, Kaufmann G. PrrC-anticodon nuclease: functional organization of a prototypical bacteria restriction RNase. Nucleic Acids Res. 2006;34:3209–3219. doi: 10.1093/nar/gkl415. [DOI] [PMC free article] [PubMed] [Google Scholar]
  4. Davidov E, Kaufmann G. RloC: a wobble nucleotide-excising and zinc-responsive bacterial tRNase. Mol Microbiol. 2008;69:1560–1574. doi: 10.1111/j.1365-2958.2008.06387.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  5. Das U, Shuman S. Mechanism of RNA 2′,3′-cyclic phosphate end-healing by T4 polynucleotide kinase-phosphatase. Nucleic Acids Res. 2013;41:355–365. doi: 10.1093/nar/gks977. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Graille M, Mora L, Buckingham RH, van Tilbeurgh H, de Zamaroczy M. Structural inhibition of the colicin D tRNase by the tRNA-mimicking immunity protein. EMBO J. 2004;23:1474–1482. doi: 10.1038/sj.emboj.7600162. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Jain R, Poulos MG, Gros J, Chakravarty AK, Shuman S. Substrate specificity and mutational analysis of Kluyveromyces lactis gamma-toxin, a eukaryal tRNA anticodon nuclease. RNA. 2011;17:1336–1343. doi: 10.1261/rna.2722711. [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Jiang Y, Blanga S, Amitsur M, Meidler R, Krivosheyev E, Sundaram M, Bajii A, Davis DR, Kaufmann G. Structural features of tRNALys favored by anticodon nuclease as inferred from reactivities of anticodon stem and loop substrate analogs. J Biol Chem. 2002;277:3826–3841. doi: 10.1074/jbc.M110072200. [DOI] [PubMed] [Google Scholar]
  9. Keppetipola N, Jain R, Meineke B, Diver M, Shuman S. Structure-activity relationships in Kluyveromyces lactis gamma-toxin, a eukaryal tRNA anticodon nuclease. RNA. 2009;15:1036–1044. doi: 10.1261/rna.1637809. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Klaiman D, Steinfels-Kohn E, Krutkina E, Davidov E, Kaufmann G. The wobble nucleotide-excision anticodon nuclease RloC is governed by the zinc-hook and DNA-dependent ATPase of its Rad50-like region. Nucleic Acids Res. 2102;40:8568–8578. doi: 10.1093/nar/gks593. [DOI] [PMC free article] [PubMed] [Google Scholar]
  11. Klassen R, Teichert S, Meinhardt F. Novel yeast killer toxins provoke S-phase arrest and DNA damage checkpoint activation. Mol Microbiol. 2004;53:263–273. doi: 10.1111/j.1365-2958.2004.04119.x. [DOI] [PubMed] [Google Scholar]
  12. Klassen R, Paluszynski J, Wemhoff S, Pfeiffer A, Fricke J, Meinhardt F. The primary target of the killer toxin from Pichia acaciae is tRNAGln. Mol Microbiol. 2008;69:681–697. doi: 10.1111/j.1365-2958.2008.06319.x. [DOI] [PubMed] [Google Scholar]
  13. Lu J, Huang B, Esberg A, Johansson MJ, Byström AS. The Kluyveromyces lactis gamma-toxin targets tRNA anticodons. RNA. 2005;11:1648–1654. doi: 10.1261/rna.2172105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Lu J, Esberg A, Huang B, Byström AS. Kluyveromyces lactis gamma-toxin, a ribonuclease that recognizes the anticodon stem loop of tRNA. Nucleic Acids Res. 2008;36:1072–1080. doi: 10.1093/nar/gkm1121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  15. Luna-Chávez C, Lin YL, Huang RH. Molecular basis of inhibition of the ribonuclease activity in colicin E5 by its cognate immunity protein. J Mol Biol. 2006;358:571–579. doi: 10.1016/j.jmb.2006.02.014. [DOI] [PubMed] [Google Scholar]
  16. Meineke B, Kast A, Schwer B, Meinhardt F, Shuman S, Klassen R. A fungal anticodon nuclease ribotoxin exploits a secondary cleavage site to evade tRNA repair. RNA. 2012;18:1716–1724. doi: 10.1261/rna.034132.112. [DOI] [PMC free article] [PubMed] [Google Scholar]
  17. Meineke B, Shuman S. Determinants of the cytotoxicity of PrrC anticodon nuclease and its amelioration by tRNA repair. RNA. 2012;18:145–154. doi: 10.1261/rna.030171.111. [DOI] [PMC free article] [PubMed] [Google Scholar]
  18. Moerschell RP, Hosokawa Y, Tsunasawa S, Sherman F. The specificities of yeast methionine aminopeptidase and acetylation of amino-terminal methionine in vivo. J Biol Chem. 1990;265:19638–19643. [PubMed] [Google Scholar]
  19. Ogawa T, Tomita K, Ueda T, Watanabe K, Uozumi T, Masaki H. A cytotoxic ribonuclease targeting specific tRNA anticodons. Science. 1999;283:2097–2100. doi: 10.1126/science.283.5410.2097. [DOI] [PubMed] [Google Scholar]
  20. Nandakumar J, Schwer B, Schaffrath R, Shuman S. RNA repair: an antidote to cytotoxic eukaryal RNA damage. Mol Cell. 2008;31:278–286. doi: 10.1016/j.molcel.2008.05.019. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Paluszynski J, Klassen R, Meinhardt F. The Pichia acaciae killer system: genetic analysis of toxin immunity. Appl Environ Microbiol. 2007;73:4373–4378. doi: 10.1128/AEM.00271-07. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Saikia M, Krokowski D, Guan BJ, Ivanov P, Parisien M, Hu GF, Anderson P, Pan T, Hatzoglou M. Genome-wide identification and quantitative analysis of cleaved tRNA fragments induced by cellular stress. J Biol Chem. 2012;287:42708–42725. doi: 10.1074/jbc.M112.371799. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Satwika D, Klassen R, Meinhardt F. Anticodon nuclease encoding virus-like elements in yeast. Appl Microbiol Biotechnol. 2012;96:345–356. doi: 10.1007/s00253-012-4349-9. [DOI] [PubMed] [Google Scholar]
  24. Simanshu DK, Yamaguchi Y, Park JH, Inouye M, Patel DJ. Structural basis of mRNA recognition and cleavage by toxin MazF and its regulation by antitoxin MazE in Bacillus subtilis. Mol Cell. 2013;52:447–458. doi: 10.1016/j.molcel.2013.09.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Thompson DM, Parker R. Stressing out over tRNA cleavage. Cell. 2009;138:215–219. doi: 10.1016/j.cell.2009.07.001. [DOI] [PubMed] [Google Scholar]
  26. Tomita K, Ogawa T, Uozumi T, Watanabe K, Masaki H. A cytotoxic ribonuclease which specifically cleaves four isoaccepting arginine tRNAs at their anticodon loops. Proc Natl Acad Sci USA. 2000;97:8278–8283. doi: 10.1073/pnas.140213797. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Winther KS, Gerdes K. Enteric virulence associated protein VapC inhibits translation by cleavage of initiator tRNA. Proc Natl Acad Sci USA. 2011;108:7403–7407. doi: 10.1073/pnas.1019587108. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Yajima S, Inoue S, Ogawa T, Nonaka T, Ohsawa K, Masaki H. Structural basis for sequence-dependent recognition of colicin E5 tRNase by mimicking the mRNA–tRNA interaction. Nucleic Acids Res. 2006;34:6074–6082. doi: 10.1093/nar/gkl729. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

01
NIHMS579647-supplement-01.pdf (3.4MB, pdf)

RESOURCES