Structure-function relations in the NTPase domain of the antiviral tRNA ribotoxin Escherichia coli PrrC

Abstract

Breakage of tRNA by Escherichia coli anticodon nuclease PrrC (EcoPrrC) underlies a host antiviral response to phage T4 infection. Expression of EcoPrrC is cytocidal in yeast, signifying that PrrC ribotoxicity crosses phylogenetic domain boundaries. EcoPrrC consists of an N-terminal NTPase module that resembles ABC transporters and a C-terminal nuclease module that is sui generis. PrrC homologs are prevalent in many other bacteria. Here we report that Haemophilus influenzae PrrC is toxic in E. coli and yeast. To illuminate structure-activity relations, we conducted a new round of mutational analysis of EcoPrrC guided by primary structure conservation among toxic PrrC homologs. We indentify 17 candidate active site residues in the NTPase module that are essential for toxicity in yeast when EcoPrrC is expressed at high gene dosage. Their functions could be educed by integrating mutational data with the atomic structure of the transition-state complex of a homologous ABC protein.

INTRODUCTION

The Escherichia coli tRNA anticodon nuclease PrrC (EcoPrrC) mediates an RNA-damaging innate immune response to bacteriophage T4 infection (Kaufmann 2000). The “ribotoxin” activity of EcoPrrC is normally suppressed by its association with its cognate “antitoxin”, a type I DNA restriction-modification enzyme (EcoprrI) encoded by neighboring ORFs in the prr operon (Levitz et al. 1990; Tyndall et al. 1994). The latent EcoPrrC nuclease is activated by the virus-encoded Stp peptide synthesized early during T4 infection (Amitsur et al. 1989, 1992; Penner et al. 1995). The activated form of EcoPrrC incises the tRNA^Lys(UUU) anticodon loop at a single site 5′ of the wobble uridine, leaving 2′,3′ cyclic phosphate and 5′-OH ends at the break (Amitsur, 1987). Depletion of the tRNA^Lys pool inhibits the synthesis of T4 late proteins and blunts spread of the virus through the bacterial population. In effect, the initially infected host bacteria altruistically commit suicide to protect the community. However, phage T4 thwarts the host defense strategy by encoding a tRNA repair system that restores late viral protein synthesis (Sirotkin et al. 1978; Amitsur et al. 1987).

EcoPrrC consists of two domains: an N-terminal nucleoside triphosphate phosphohydrolase (NTPase) module (aa 1–264) related to the ABC transporter NTPase family and a distinctive C-terminal ribonuclease module (aa 265–396) (Kaufmann 2000; Blanga-Kanfi et al. 2006) (Fig. 1). The NTPase and nuclease domains must be linked in cis for EcoPrrC to exert its toxicity (Meineke et al. 2011). The NTPase domain of EcoPrrC activates the latent anticodon nuclease via a mechanism entailing both GTP hydrolysis and the binding of dTTP, a putative allosteric effector (Amitsur et al. 2003; Klaiman and Kaufmann 2011). Physical and genetic studies, and analogy to ABC family proteins, support a model in which EcoPrrC toxicity requires head-to-tail dimerization of the NTPase modules to form two composite sites for NTP binding and hydrolysis (Blanga-Kanfi et al. 2006; Klaiman et al. 2007; Meineke et al. 2011).

Fig. 1. Homology-guided mutagenesis of EcoPrrC.

The amino acid sequence of EcoPrrC is aligned to sequences of the NmePrrC, SmuPrrC, and HinPrrC proteins. Positions of side chain identify/similarity in all four proteins are indicated by ● above the alignment. The conserved peptide motifs of the N-terminal ABC-type NTPase domain (labelled A, C, B, D and H) are demarcated by brackets. The PrrC-box of the NTPase domain and the lysine anticodon recognition peptide (“LARP”) motif of the nuclease domain are labelled and demarcated by brackets. EcoPrrC residues defined as important for yeast toxicity by alanine mutagenesis of the CEN-based prrC gene are shaded green; nonessential residues are shaded yellow. The new residues mutated in the present study are denoted by ▼ above the alignment.

PrrC homologs are present in the proteomes of many other bacteria (Davidov and Kaufmann 2008), though their biological activities in their native environments are uncharted. The ribotoxicity of bacterial PrrC is portable to eukarya. Induced expression of EcoPrrC or Streptococcus mutants (Smu) PrrC in budding yeast cells is fungicidal, signifying that PrrC can be toxic in a eukaryon in the absence of any other bacterial or viral proteins (Meineke et al. 2011). The apparent lack of toxicity of Neisseria meningitidis (Nme) PrrC was attributed to a nuclease-inactivating arginine substitution for an essential tryptophan (highlighted in red in Fig. 1); a single back-mutation of this arginine to tryptophan sufficed to make NmePrrC toxic when expressed in yeast or E. coli (Meineke and Shuman 2012).

An extensive survey of the effects of alanine and conservative mutations on EcoPrrC toxicity in yeast has so far identified 22 functionally important residues in the NTPase domain and 15 in the nuclease domain (Meineke et al. 2011; Meineke and Shuman, 2012) (see Fig. 1; positions highlighted in green without a ▼symbol). Replacing these important side chains by alanine resulted in loss of toxicity in yeast when the prrC gene was expressed from a CEN plasmid. Nineteen residues were defined by the alanine scan as inessential for EcoPrrC toxicity; i.e., alanine changes did not affect toxicity when the mutants were expressed from a CEN plasmid (Fig. 1; positions highlighted in yellow without a ▼symbol). As one might expect, many (though not all) of the important residues in EcoPrrC are conserved in the toxic SmuPrrC and latently toxic NmePrrC proteins (Fig. 1). As noted previously (Meineke et al. 2011), and discussed in detail below, the roles of some of the NTPase domain residues can be surmised from their location within the defining peptide motifs of structurally characterized ABC proteins (Fig. 1) (Oldham and Chen 2011). By contrast, the nuclease domain has no discernible primary structure similarity to any known ribonucleases or tRNA-binding proteins, which makes it difficult to guess which essential residues might be directly involved in catalysis versus substrate recognition versus PrrC folding/stability. By testing the 15 nontoxic alanine mutants in the EcoPrrC nuclease module for restoration of toxicity by increasing gene dosage, we could distinguish between hypomorphs and potential nuclease nulls (Meineke and Shuman 2012). This criterion highlighted eight essential residues (Asp276, Asp287, His295, His315, Arg320, Glu324, Arg349, and His356) as candidate active site constituents. Among these, Arg320, Glu324 and His356 had been proposed by the Kaufmann lab to comprise a catalytic triad for transesterification chemistry (Blanga-Kanfi et al. 2006).

Here, we extend our studies of bacterial PrrC on two fronts. First, we interrogated the in vivo activity of the PrrC homolog encoded by Haemophilus influenzae strain 86-028NP (Harrison et al. 2005) and thereby found that HinPrrC was toxic in yeast and growth suppressive in E. coli. Second, we used the primary structure conservation among confirmed ribotoxic PrrC homologs to guide a further round of mutagenesis of EcoPrrC (Fig. 1). We thereby identified 9 new side chains in the NTPase module and two in the nuclease domain that are important for toxicity in yeast. We interpret the sum of mutational data for the NTPase domain in light of recently reported atomic structures of an exemplary ABC protein that illuminate the Michaelis complex and the transition state of the phosphohydrolase reaction (Oldham and Chen 2011).

RESULTS AND DISCUSSION

Haemophilus influenzae PrrC is toxic in vivo

The 384-aa HinPrrC polypeptide (Genbank accession YP_247935) has 274 positions of side-chain identity/similarity to EcoPrrC in a pairwise alignment. Whereas most of the essential residues in EcoPrrC are conserved in HinPrrC, there is a conspicuous exception whereby the Asp287 side chain of EcoPrrC is replaced by a glycine in HinPrrC (Fig. 1). We noted previously that the D287A mutation abolished EcoPrrC toxicity in yeast, even at high gene dosage (Meineke and Shuman, 2012). Asp287 is located within the lysine anticodon recognizing peptide (LARP) motif, ²⁸⁴KYGDSNKSFSY²⁹⁴ (Klaiman et al. 2007) (Fig. 1). The Kaufmann lab has shown that missense mutations at Asp287 alter the tRNA cleavage preferences of EcoPrrC in vivo when expressed in E. coli and/or in vitro, e.g., such that particular Asp287 mutants are either more or less fastidious regarding the impact of wobble uridine base modifications (Jiang et al. 2001; 2002). The presence of glycine at this position in HinPrrC raised the prospects that: (i) HinPrrC (like NmePrrC) is nontoxic because of an inactivating missense change in the nuclease module; or (ii) HinPrrC is toxic despite this change in the LARP motif, possibly reflecting a distinct target specificity.

To address this issue, we queried the effect of induced HinPrrC expression of the growth of Saccharomyces cerevisiae (Fig. 2). The H. influenzae 86-028NP prrC gene was introduced into yeast on a CEN plasmid under the control of a glucose-repressed/galactose-inducible GAL1 promoter. HinPrrC induction inhibited yeast growth on agar medium containing galactose (Fig. 2). The growth inhibition was abolished by an alanine substitution for His336 in the nuclease domain of HinPrrC (denoted by ▲ in Fig. 1), which is the counterpart of the essential EcoPrrC residue His356. We attempted to identify genetically a dominant RNA target of HinPrrC toxicity in yeast, by asking whether overexpression of individual yeast tRNAs could blunt the effects of HinPrrC on yeast growth. This proved unsuccessful, insofar as 2μ plasmids expressing either tRNA^Lys(UUU), tRNA^Lys(CUU), tRNA^Glu(UUC), tRNA^Gln(UUG), tRNA^Arg(UCU), or tRNA^Leu(UAA) did not ameliorate HinPrrC toxicity, nor did simultaneous overexpression of tRNA^Lys(UUU), tRNA^Glu(UUC), and tRNA^Gln(UUG) (not shown).

Fig. 2. HinPrrC is toxic in yeast.

Wild-type HinPrrC and the H336A mutant were tested for their effects on the growth of S. cerevisiae. Serial 5-fold dilutions of yeast cells bearing a CEN plasmid encoding the indicated galactose-regulated prrC gene or an empty CEN vector (−) were spotted on agar plates (−Leu) containing glucose or galactose as specified.

The conclusion that HinPrrC is a toxin was fortified by testing the effects of induced HinPrrC expression in E. coli (Fig. 3). The Eco and Hin prrC genes were introduced into E. coli Top10 cells on pBAD plasmids under the control an arabinose-inducible promoter. Serial dilutions of E. coli pBAD-prrC cultures grown in LB medium were plated on LB agar (prrC expression repressed) or LB agar with 0.2% arabinose (prrC expression induced). The results showed that EcoPrrC and HinPrrC were comparably toxic to E. coli (Fig. 3).

Fig. 3. HinPrrC toxicity in E. coli.

Serial dilutions of E. coli cells bearing a pBAD plasmid encoding the indicated arabinose-regulated prrC gene or an empty vector (−) were spotted on LB-ampicillin agar plates with or without arabinose. When grown on control medium lacking arabinose, prrC expression is switched off and the bacteria grow normally. When grown on medium containing 0.2% arabinose, the prrC expression is turned on and, if the PrrC ribotoxin is active, bacterial growth is slowed (as evinced by the formation of tiny colonies).

Further mutational analysis of EcoPrrC

The HinPrrC experiments above bring to four the number of PrrC homologs with verified toxicity, these being EcoPrrC, NmePrrC-(R316W), SmuPrrC, and HinPrrC. An alignment of their primary structures (Fig. 1) highlights patches of strong conservation separated by segments of low homology. Within the N-terminal NTPase domain, there are 136 positions of side chain identity/similarity (denoted by ● in Fig. 1); the C-terminal nuclease domain includes 46 positions of side chain identity/similarity. Here we used the sequence alignment to guide a new round of mutagenesis of EcoPrrC, focusing primarily on the NTPase module, in which we tested the effects of alanine substitutions at 15 conserved positions (Fig. 1, denoted by ▼). In addition, we mutated two conserved tyrosines in the nuclease domain (Tyr294 and Tyr314) that are situated adjacent to two of the essential histidines (His295 and His315). Histidines are classical constituents of the active sites of transesterifying endoribonucleases, wherein they function as general base and/or general acid catalysts that abstract a proton from the ribose O2 nucleophile and donate a proton to the ribose O5′ leaving group, respectively (Raines 1998). We chose to target the tyrosines in light of the structure of tRNA splicing endonuclease, which has a tyrosine in the active site that likely serves as a general acid catalyst of RNA transesterification (Xue et al. 2006).

The EcoPrrC-Ala alleles were inserted into CEN plasmids under GAL-control. Tests of yeast growth on glucose and galactose indentified 5 nonessential side chains in the NTPase module (Lys17, Thr44, Tyr82, Glu174, Asn261) at which alanine substitution still allowed for virtually complete growth inhibition on galactose-containing medium (comparable to that seen for wild-type EcoPrrC and scored as ++ toxicity in Table 1). Alanine changes at nine positions in the NTPase module abolished EcoPrrC toxicity in yeast, such that the strains expressing these mutants grew as well on galactose-medium as the negative control strain transformed with the “empty” CEN plasmid (scored as – toxicity in Table 1). The newly defined important NTPase residues were: Tyr39, Asn42, Lys54, Tyr83, Asn84, Gln108, Ser170, Glu173 and Ser180 (Table 1; Fig. 1). Changing Gln134 to alanine elicited a partial loss of function, whereby the Q134A yeast strain was viable on galactose medium, but formed smaller colonies than the vector control (scored as + toxicity in Table 1). Tyr294 and Tyr314 in the nuclease module were also identified as important via the alanine scan.

Table 1.

Mutational effects on EcoPrrC toxicity in yeast

PrrC	Toxicity on galactose
K17A	++
Y39A	−
Y39F	++
N42A	−
N42D	−
N42Q	−
N42S	−
T44A	++
K54A	−
K54R	++
K54Q	++
Y82A	++
Y83A	−
Y83F	−
N84A	−
N84D	−
N84Q	−
Q108A	−
Q108E	−
Q108K	−
Q108N	−
Q134A	+
Q134N	++
Q134E	+
S170A	−
S170C	++
S170T	−
E173A	−
E173D	−
E173Q	−
E174A	++
S180A	−
S180C	++
S180T	++
N261A	++
Y294A	−
Y294F	−
Y314A	−
Y314F	++

Structure-activity relations at Tyr294 and Tyr314

Toxicity was restored when phenylalanine replaced Tyr314, signifying that the phenyl ring sufficed for function and excluding Tyr314 as a proton donor or acceptor in catalysis and as a hydrogen-bond donor in RNA recognition or protein folding. By contrast, there was no gain of function when Tyr294 was replaced with phenylalanine, signifying that the hydroxyl at this position is essential for toxicity when EcoPrrC is expressed from a CEN plasmid. To evaluate whether the requirements for Tyr294 and Tyr314 might be circumvented by increasing gene dosage, we transferred the _GAL-PrrC-Y294A and _GAL-PrrC-Y314A expression cassettes to 2μ plasmids, introduced them into S. cerevisiae, and tested the transformants for galactose-dependent toxicity. The instructive findings were that 2μ Y314A was fully toxic on galactose (++) and 2μ Y294A was partially toxic (+) (Table 2). These results weigh against either of the tyrosines being essential for catalysis of transesterification.

Table 2.

High gene dosage can restore toxicity of certain defective EcoPrrC mutants

2μ PrrC	Toxicity on galactose
Y39A	−
N42A	−
K46A	−
T47A	−
R48A	−
K54A	+
Y83A	−
N84A	−
E88A	++
W93A	−
D94A	+
N95A	+
Q108A	++
K168A	−
S170A	++
K171A	++
E173A	−
S180A	++
D215A	−
D216A	−
S219A	−
S220A	−
D222A	−
D222E	++
D222N	−
D223A	++
N224A	++
H225A	++
K245A	++
T250A	−
H251A	−
N257A	+
E262A	+
Y294A	+
Y314A	++

Structure-activity relations in the NTPase domain

We determined structure-activity relationships for each of the nine new essential residues in the NTPase domain, plus Gln108, by testing the effects of 20 conservative substitutions. The conservatively mutated alleles were inserted into CEN plasmids under GAL-control and tested for toxicity on galactose medium. The results are summarized in Table 1. In addition, we transferred the present collection of nontoxic NTPase domain mutants, plus many of the defective NTPase domain mutants identified previously (Meineke et al. 2011), to 2μ vectors and tested whether increased gene dosage restored toxicity in yeast. Those results are summarized in Table 2. The findings are discussed below, wherever possible extracting mechanistic insights by reference to the 2.4 Å crystal structure of the homologous NTPase domain of the E. coli maltose transporter (an exemplary ABC protein) in a complex with ADP-vanadate and magnesium that mimics the transition-state of the NTP phosphohydrolase reaction (Oldham and Chen 2011) (Fig. 4).

Fig. 4. Composite active site of an ABC NTPase homodimer.

Stereo view of the NTPase active site in the crystal structure of E. coli maltose transporter in complex with Mg²⁺ (magenta sphere) and ADP-vandate (stick model) (pdb id: 3PUV). The phosphohydrolase active site of the homodimeric enzyme is a composite of structural elements derived from a cis protomer (colored green) that provides the A-box, B-box, Q-loop and H-loop motifs and a trans protomer (colored beige) that provides the D-loop and C-loop motifs. Ionic and hydrogen bonding interactions are denoted by dashed lines. Two waters in the octahedral metal coordination complex are rendered as red spheres. The amino acid numbering refers to the equivalent positions in the NTPase domain of EcoPrrC.

The maltose transporter NTPase domain is a homodimer, arranged head-to-tail, with two composite NTPase active sites formed by motifs derived mainly from the cis protomer (colored green in Fig. 4), which provides the A-box, B-box and H-loop motifs (indicated by brackets over the PrrC sequences in Fig. 1) that bind the NTP and the metal cofactor. The trans protomer (colored beige in Fig. 4) contributes the C-loop and D-loop motifs of the active site, that respectively stabilize the pentacoordinate phosphorane transition state of the NTP γ-phosphate and make cross-protomer contacts to the A-box and B-box.

A-box mutations

The A-box motif GxGK(T/S) is situated between a β-strand and α-helix of the NTPase module and forms a classical P-loop structure in which the main-chain amide nitrogens and the signature lysine side chain (Lys46 in EcoPrrC) coordinate the NTP phosphate oxygens (Fig. 4). The signature threonine/serine side chain vicinal to the lysine (Thr47 in PrrC) coordinates the divalent cation cofactor that bridges the β and γ phosphates (Fig. 4). Lys46 and Thr47 were both essential for toxicity in yeast when EcoPrrC was expressed from a CEN plasmid (Meineke et al. 2011). Consistent with a direct role of Lys46 and Thr47 in NTP utilization by EcoPrrC, we found that the K46A and T47A mutants also failed to inhibit cell growth when expressed from a 2μ plasmid (Meineke et al. 2011) (Table 2). The A-box residue Asn42, identified presently as important for EcoPrrC activity, is located in the P-loop; its equivalent in the maltose transporter is a serine that donates a hydrogen bond from Oγ to one of the trigonal planar γ-phosphate oxygens (Fig. 4). We presume that EcoPrrC Asn42 Nδ plays a similar role in transition-state stabilization. Conservative replacement of Asn42 with aspartate failed to revive activity in yeast (Table 1), likely because of the attendant electrostatic repulsion between the carboxylate and the NTP phosphate. That the N42Q mutant was also inactive, attests to the steric constraint on the distance from the main-chain to the amide. It was noteworthy that changing Asn42 to serine (the side chain in the maltose transporter) also failed to revive toxicity (Table 1), suggesting that either the Asn42 Oδ makes important contact in EcoPrrC or the imputed contacts to the NTP γ-phosphate require a longer distance from the main chain to the H-bond donor atom in EcoPrrC than in the maltose transporter. Expressing EcoPrrC-N42A from a 2μ plasmid did not elicit a gain of toxicity (Table 2).

The Tyr39 residue identified presently as important in EcoPrrC is situated in the β-strand preceding the P-loop; the equivalent residue in maltose transporter is a valine that projects into the hydrophobic core of the protein. Whereas replacing Tyr39 with phenylalanine restored toxicity when the Y39F mutant was expressed from a CEN plasmid (Table 1), the Y39A mutant remained nontoxic when expressed from a 2μ plasmid (Table 2). We surmise that Tyr39 plays a structural role in EcoPrrC. The Lys54 side chain identified as important in the alanine scan is located in the distal part of the A-box α-helix, away from the active site. Full toxicity was restored when Lys54 was substituted with either arginine or glutamine (Table 1), implicating hydrogen bonding as the key property of this side chain. Overexpression of the K54A mutant partially restored toxicity (Table 2), suggesting structural role for this conserved PrrC residue.

Arg48, a conserved component of the PrrC A-box (Fig. 1), was shown previously to be essential for CEN-driven toxicity of EcoPrrC in yeast, i.e., R48A, R48K and R48Q mutants were inactive (Meineke et al. 2011). Overexpressing the R48A mutant on a 2μ plasmid did not revive toxicity (Table 2). The equivalent residue in maltose transporter is a threonine that is oriented in the direction of the nucleobase of the NTP substrate. We can speculate that an arginine side chain in this position of EcoPrrC might make interact with the base of the NTP substrate, possibly via a cation-π stack.

B-box and D-loop mutants

The EcoPrrC peptide segment ²¹⁵DDPVSSLDDNH²²⁵, which embraces 8 functionally important side chains (Fig. 1), is composed of two distinct ABC motifs that form the active site: a B-box (YVFIDD²¹⁶) derived from the cis protomer and a D-loop (SSLD²²² in EcoPrrC) derived from the trans protomer (Fig. 4). The B-box provides two key carboxylates to the phosphohydrolase active site. The proximal aspartate (Asp215 in EcoPrrC) is a component of the octahedral metal coordination complex. The vicinal Asp216 (a glutamate in the maltose transporter B-box) coordinates the water nucleophile at the apical position of the pentacoordinate transition state depicted in Fig. 4. The Asp215 and Asp216 residues in EcoPrrC were both essential for CEN-driven toxicity and neither could be functionally substituted with asparagine or glutamate, signifying that a carboxylate is critical at both positions and that the PrrC active site cannot accommodate the longer main-chain to carboxylate linker of Glu versus Asp (Meineke et al. 2011). Here we found that expressing EcoPrrC-D215A or EcoPrrC-D216A on a 2μ plasmid did not restore toxicity (Table 2).

The eponymous aspartate side chain of the D-loop (Asp222 in EcoPrrC) caps an α-helix and makes a cross-protomer hydrogen bond with a phosphate-binding main chain amide of the A-box (Fig. 4). This contact would be to the Asn42 main chain amide in EcoPrrC. The D-loop Asp222 residue of EcoPrrC was essential for CEN-based toxicity and irreplaceable by asparagine or glutamate (Meineke et al. 2011). When the same alleles were tested on 2μ vectors, we found that D222A and D222N were inactive, whereas D222E was toxic (Table 2), thereby attesting to the importance of the carboxylate functional group in assembling the composite active site. The Ser219 and Ser220 residues preceding the D-loop aspartate were both essential for CEN-based PrrC toxicity (Meineke et al. 2011) and we show here that mutants S219A and S220A remained nontoxic when expressed on 2μ plasmids (Table 2). We reported previously that CEN-based toxicity was restored when Ser219 was replaced by threonine or cysteine, signifying that hydrogen-bonding of the Oγ atom is the functionally relevant property. Indeed, the equivalent D-loop serine in the maltose transporter makes intra- and cross-promoter hydrogen bonds to neighboring D-loop side asparagine chain (equivalent to Ser220 in the PrrC D-box) (Fig. 4). At Ser220 of EcoPrrC, the S220T and S220C mutants were inactive in vivo when expressed from CEN plasmids (Meineke et al. 2011), which suggests a tight steric constraint on the Ser220 Oγ. The corresponding aparagine of the maltose transporter D-loop donates a hydrogen bond to the carboxylate of the B-box glutamate, equivalent to EcoPrrC Asp216 (Fig. 4).

The Asp223, Asn224, and His225 residues flanking the EcoPrrC D-loop aspartate were found to be important for CEN-based toxicity in a previous alanine scan (Meineke et al. 2011), yet they are not conserved in the maltose transporter, wherein the corresponding residues are Ala, Ala, and Leu. Here we find that full toxicity was revived when the D223A, N224A and H225A alleles were expressed from 2μ plasmids (Table 2), which we take to mean that these three D-loop mutants are hypomorphs.

C-loop mutants

The C-loop (also called the ABC-signature motif) of EcoPrrC (KLSKGE¹⁷³) includes five residues important for CEN-based toxicity (Fig. 1). In the maltose transporter structure, the C-loop of the trans protomer packs against the nucleoside and γ-phosphate of the NTP substrate, so that the C-loop and the A box P-loop of the cis protomer together form an oxyanion hole for the γ-phosphate (Fig. 4). The EcoPrrC C-loop residue Ser170, identified as important by the present alanine scan, is conserved as serine in the maltose transporter, where it donates a hydrogen bond from Oγ to one of the trigonal planar γ-phosphate oxygens (Fig. 4). Here we found that CEN-based toxicity of EcoPrrC was restored by conservative replacement of Ser170 with cysteine, whereas threonine was inactive (Table 1), thereby underscoring the relevance of hydrogen-bonding and the steric constraints (evident in the maltose transporter active site structure) that would not accommodate an extra methyl group on threonine. Yet, the fact that toxicity of the S170A allele was restored by expression from a 2μ plasmid (Table 2) militates against Ser170 being essential for a productive NTP-bound state, perhaps because its contact to a planar γ-phosphate oxygen is functionally redundant with the hydrogen bond to the same oxygen donated by the A-box P-loop side chain equivalent to EcoPrrC Asn42 (Fig. 4).

The function of the C-loop Glu173 of EcoPrrC could not be fulfilled by aspartate or glutamine (Table 1), nor was activity regained by expression of E173A from a 2μ plasmid (Table 2). The equivalent C-loop residue in maltose transporter is a glutamine, which make hydrogen-bonds to both ribose hydroxyls of the NTP substrate and to the main chain amide of the C-loop serine (Fig. 4). These same contacts could be fulfilled by glutamate in EcoPrrC, though it is not clear why glutamine is ineffective in the PrrC context.

EcoPrrC residues Lys168 and Lys171 were implicated previously as important for CEN-based toxicity (Meineke et al. 2011), though neither is conserved in maltose transporter, where the corresponding side chains are Ala and Gly, respectively. Toxicity of the K171A allele was restored when expressed from a 2μ plasmid (Table 2). However, no gain of function was seen when K168A was overexpressed (Table 2). In the maltose transporter, the Cβ atom of the alanine corresponding to EcoPrrC Lys168 is located 3.9 Å away from the N6 atom of the adenine base of ATP. We can speculate that Lys171 of EcoPrrC, which is conserved in other PrrC homologs (Fig. 1) and is not functionally replaceable by arginine or glutamine (Meineke et al. 2011), may be a determinant of NTP specificity, by donating a hydrogen bond from Nζ to the nucleobase.

The EcoPrrC Ser180 residue located just downstream of the C-loop is conserved in other PrrC homologs (Fig. 1). Whereas S180A was nontoxic, installation of either cysteine or threonine restored CEN-based activity in yeast (Table 1), suggesting the relevance of hydrogen-bonding. However, increasing the gene dosage of S180A also restored toxicity (Table 2), signifying that this is a hypomorphic mutation rather than an NTPase null.

H-loop mutations

The KFIITTH²⁵¹ peptide of EcoPrrC that spans three important side chains (Fig. 1) is the counterpart of the conserved H-loop motif of ABC-type NTPases (also called the “switch” motif). In the maltose transporter, the H-loop connects a β-strand to an α-helix. The signature histidine (His251 in EcoPrrC) donates a hydrogen bond to one of the trigonal planar γ-phosphate oxygens in the transition state (Fig. 4). His251 of EcoPrrC is strictly essential for its CEN-driven activity in yeast, insofar as the H251A, H251N and H251Q mutants were nontoxic (Meineke et al. 2011). Increasing the gene dosage of H251A did not revive toxicity in yeast (Table 2). The vicinal threonine-Oγ (Thr250 in EcoPrrC) makes a hydrogen bond to the main-chain amide of the H loop residue on the carboxyl side of the histidine of the maltose transporter and thereby stabilizes the H-loop conformation. Overexpressing EcoPrrC-T250A did not elicit a gain of toxicity (Table 2). The upstream Lys245 is important for EcoPrrC and conserved in other PrrC homologs. Positive charge appeared to be pertinent, insofar as K245R was active in yeast in single copy, whereas K245Q was not (Meneke et al. 2011). The corresponding position in maltose transporter is a threonine located in the β-strand preceding the H-loop, far away from the active site. Our finding that overexpressing K245A restored full toxicity in yeast (Table 2), suggests a structural role for Lys245. Similar inferences may be drawn for Asn257 and Glu262, which are located downstream of the H-loop, conserved among PrrCs, but not in maltose transporter. Whereas N257A and E262A mutations were found previously to abolish CEN-based toxicity (Meineke et al. 2011), we now find that overexpression of N257A and E262A from 2μ plasmids partially restored function (Table 2).

The PrrC-box

There are six essential residues within the ⁸²YYNAFYEDLFYWDND⁹⁶ segment of the EcoPrrC NTPase domain that has been dubbed the “PrrC-box” by the Kaufmann lab (Blanga-Kanfi et al. 2006) in light of its strong conservation among PrrC homologs (Fig. 1). Tyr83 and Asn84 were newly defined above by alanine scanning as important for CEN-based toxicity. Overexpression of Y83A or N84A failed to revive activity (Table 2). Replacing Tyr83 with phenylalanine had no salutary effect (Table 1), signifying that that the tyrosine hydroxyl is needed. Installing aspartate or glutamine in lieu of Asn84 was similarly ineffective (Table 1), indicating that the amide functional group at position 84 is critical and implying a steric constraint on the length of main-chain to amide linker. Glu88, Trp93, Asp94 and Asn95 were deemed important for CEN-based toxicity in an earlier study (Meineke et al. 2011). Here we found that overexpression of E88A restored full toxicity, whereas overexpressing D94A and N95A elicited a partial gain of function (Table 2). The 2μ W93A allele remained nontoxic (Table 2).

There is little primary structure similarity between the PrrC box and the corresponding segments of ABC proteins, including maltose transporter, though it is possible that the PrrC box is a divergent analog of the ABC Q-loop motif, which is located between the A-box and the C-loop in the ABC primary structure. The Q loop is a mobile hinge that is sensitive to the presence of NTP and metal ligands. The eponymous glutamine side chain of the maltose transporter Q loop makes direct contacts with metal and one of the trigonal planar γ-phosphate oxygens in the transition state (Fig. 4). If the PrrC-box is analogous to the Q-loop, then the mutational data for EcoPrrC point to Asn84 as the most likely candidate for a direct role in metal-binding and engagement of an NTP γ-phosphate.

We did evaluate two glutamines in EcoPrrC, located between the A-box and C-loop, as Q-loop candidates. We rejected Gln134 as a catalytic Q-loop residue, based on the findings that the Q134A and Q134E alleles were both partially toxic when expressed on a CEN plasmid (Table 1) and the lack of conservation of this residue in HinPrrC, where it is a histidine (Fig. 1). On the other hand, Gln108 is conserved in all PrrC homologs shown in Fig. 1, and the mutant alleles Q108A, Q108E, Q108K and Q108N alleles were all nontoxic when expressed from CEN plasmids (Table 1). Yet, the finding that expression of Q108A from a 2μ plasmid was toxic (Table 2) appeared to rule out Gln108 as the catalytic Q-loop residue.

Concluding remarks

The antiviral tRNAase EcoPrrC and its various homologs (Davidov and Kaufmann 2008) are the only known ribotoxins with an ABC-like NTPase domain. The present study of the EcoPrrC NTPase fortifies the conclusion that NTP utilization is essential for PrrC toxicity via: (i) comprehensively surveying the structure-activity relations at phylogenetically conserved amino acids; (ii) applying gene dosage compensation as a criterion to discriminate hypomorphic mutations from true nulls, thereby pinpointing the likely catalytic components of the active sites; and (iii) relying on the extremely informative atomic structure of the Mg²⁺•ADP-vanadate complex of the maltose transporter (Oldham and Chen 2011) as a hermeneutic guide to the mutational data. The upshot of this analysis is that the active site and catalytic mechanism of the PrrC NTPase are likely to adhere closely to that of ABC proteins, especially with respect to the contributions of the essential components of the A-box, B-box, C-loop, D-loop and H-loop motifs. We suggest that the “missing” counterpart of the ABC protein catalytic Q-loop might reside within the PrrC-box. We note that EcoPrrC activity relies on additional clade-specific NTPase amino acids not found in other ABC proteins, among which could be determinants of NTP substrate specificity or mediators of activating signals between the NTPase and nuclease domains.

Correlative biochemical and structural studies of PrrC are hindered by the self-limiting capacity for expression of active PrrC, i.e., because PrrC curtails protein synthesis in vivo. Yet, it is feasible to produce nontoxic EcoPrrC mutants in amounts amenable to structural studies. Indeed, we have grown crystals of EcoPrrC-H356A, though they diffract to no better than 8 Å resolution (B. Meineke and P. Smith, unpublished). It is conceivable that success on the structural front will hinge on cocrystallization of PrrC with a tRNA substrate. Until such time, our mutational studies of PrrC have reached their effective endpoint.

MATERIALS AND METHODS

PrrC yeast toxicity assays

Yeast CEN LEU2 plasmids contained the EcoPrrC or HinPrrC ORFs under the transcriptional control of a GAL1 promoter. To construct the HinPrrC expression plasmid, the Haemophilus influenzae 86-028NP gene encoding HinPrrC (Genbank accession YP_247935) was amplified by PCR from genomic DNA (a gift of Dr. Robert Munson, Nationwide Children’s Hospital, Columbus, Ohio) with a sense-strand primer that introduced an NdeI site at the translation start codon and an antisense-strand primer that introduced a SalI site downstream of the stop codon. The PCR product was digested with NdeI and SalI and inserted between the corresponding sites in pYC-EcoPrrC in lieu of the EcoPrrC fragment. Missense mutations were introduced into the prrC genes by two-stage overlap extension PCR. The prrC ORF was sequenced in each case to verify the intended hybrid junctions or coding change and exclude the acquisition of unwanted coding changes during amplification and cloning. EcoRI/SalI fragments containing _GAL1-prrC-Ala expression cassettes were excised from the respective CEN plasmids and inserted into multicopy yeast plasmid pRS423 (2μ HIS3). Yeast cells were transformed with PrrC plasmid DNAs and transformants were selected on appropriate minimal synthetic media on 2% (w/v) bacto agar plates. Toxicity of the plasmid-encoded PrrC proteins was gauged as described (Meineke et al. 2011). Cells derived from single transformants were grown at 30°C in liquid culture in selective media containing 2% glucose. The cultures were adjusted to A₆₀₀ of 0.1 and then diluted in water in serial 5-fold decrements. Aliquots (3 μl) of the dilutions were then spotted in parallel on selective agar plates containing either 2% glucose or 2% galactose. The plates were photographed after incubation at 30°C for 2 days (glucose) or 3 days (galactose).

PrrC bacterial toxicity assays

The EcoPrrC and HinPrrC open reading frames were amplified by PCR from their respective yeast CEN plasmids using a sense-strand primer that introduced an NheI site immediately 5′ of the translation start codon. The PCR products were digested with NheI and SalI and inserted between the corresponding restriction sites of the bacterial expression plasmid pBAD18. The prrC ORF was sequenced in each case to verify the intended coding sequence. Top10 cells (araABD⁻, Invitrogen) were transformed with pBAD-PrrC plasmids. Cells derived from single ampicillin-resistant colonies were grown in LB medium containing 200 μg/ml ampicillin for 4 h at 37°C. The culture were adjusted to attain A₆₀₀ of 0.025 and then diluted in 20-fold decrements in water. Aliquots (3 μl) of the dilutions were spotted in parallel on LB agar plates containing 100 μg/ml ampicillin with or without 0.2% L-arabinose. The plates were photographed after incubation for 24 h at 37°C.