**1. Introduction**

20 DNA Repair

[98] Ivancic-Bace I, Peharec P, Moslavac S, Skrobot N, Salaj-Smic E, Brcic-Kostic K. RecFOR

[99] Bentchikou E, Servant P, Coste G, Sommer S. A major role of the RecFOR pathway in

[100] Imlay JA, Linn S. Mutagenesis and stress responses induced in Escherichia coli by

[101] Galitski T, Roth JR. Pathways for homologous recombination between chromosomal direct repeats in Salmonella typhimurium. Genetics 1997;146(3):751-67. [102] Cox MM. Historical overview: searching for replication help in all of the rec places.

recB mutant of Escherichia coli. Genetics 2003;163(2):485-94.

hydrogen peroxide. J Bacteriol 1987;169(7):2967-76.

Proc Natl Acad Sci U S A 2001;98(15):8173-80.

PLoS Genet;6(1):e1000774.

function is required for DNA repair and recombination in a RecA loading-deficient

DNA double-strand-break repair through ESDSA in Deinococcus radiodurans.

Bacteria respond to DNA damage by inducing the expression of numerous proteins involved in DNA repair and the reversible arrests of DNA replication and the cell division cycle (Fernandez De Henestrosa *et al*, 2000). This general rule may be violated by a conserved bacterial protein termed RloC (Davidov & Kaufmann, 2008). RloC combines structural-functional properties of two unrelated proteins (i) the universal DNA-damageresponsive/DNA-repair protein Rad50/SbcC (Williams *et al*, 2007) and (ii) the translationdisabling, phage-excluding anticodon nuclease (ACNase) PrrC (Blanga-Kanfi *et al*, 2006). These seemingly conflicting features may be reconciled in a model where RloC is mobilized as an antiviral back-up function during recovery from DNA damage (Davidov & Kaufmann, 2008), when DNA restriction, the cell's primary immune system is temporarily shut-off (Thoms & Wackernagel, 1984). Another intriguing feature of RloC is its ability to excise its substrate's wobble nucleotide (Davidov & Kaufmann, 2008). This harsh lesion is expected to encumber reversal by phage enzymes that repair the tRNA nicked by PrrC (Amitsur *et al*, 1987). Evaluating RloC's salient features and purported role requires prior description of its more familiar distant homolog PrrC and a DNA-damage-sensing device RloC shares with Rad50/SbcC. We conclude with an account of cellular RNA and DNA repair tools related to the phage tRNA repair mechanism that counteracts PrrC and may be frustrated by RloC.

#### **2. PrrC – A potential phage-excluding tool counteracted by tRNA repair enzymes**

#### **2.1 A host-phage survival cascade yields an RNA repair pathway**

RNA repair may seem unnecessary because damaged RNA molecules can be readily replenished by re-synthesis. Yet, there exist situations where RNA repair could be the preferred or only possible option. A case in point is presented by an RNA repair pathway triggered by the ACNase PrrC. This conserved bacterial protein was detected in quest of roles of two phage T4-encoded enzymes: 3'-phosphatase/5'-polynucleotide kinase (PseT/Pnk,

<sup>\*</sup> Elena Davidov, Emmanuelle Steinfels-Kohn, Ekaterina Krutkina, Daniel Klaiman, Tamar Margalit, Michal Chai-Danino and Alexander Kotlyar

*Tel Aviv University, Israel* 

RloC: A Translation-Disabling tRNase Implicated in

linked R-M system is compromised.

could be one.

Phage Exclusion During Recovery from DNA Damage 23

existence of a "smarter" ACNase able to encumber phage reversal. Later we ask if RloC

Fig. 1. A host-phage survival cascade gives rise to an RNA repair pathway. A. The optional host locus *prr* comprises the core ACNase gene *prrC* and flanking genes encoding the type Ic DNA R-M protein EcoprrI that silences PrrC's ACNase activity. Arrows mark transcription start sites. B. Cleavage-ligation of tRNALys in phage T4 infected *E. coli prr+*. T4's anti-DNA restriction factor Stp inhibits EcoprrI and activates the latent ACNase. The resultant disruption of tRNALys is reversed by the T4's tRNA repair enzymes Pnk and Rnl1.

Nested *prr* loci where *prrC* intervenes a type Ic *hsd* locus (Fig 1A) appear sporadically in distantly related bacteria. They are present in some strains of a given species but not in others, as would a niche-function (Blanga-Kanfi *et al*, 2006). They abound among *Proteobacteria*, are less frequent in *Bacteroidetes* and *Firmicutes*, rare in *Actinobacteria* and apparently absent from *Cyanobacteria*. PrrC's phylogenic tree does not match the bacterial, unlike the associated type Ic R-M protein, which only rarely teams with PrrC. In contrast, a stand-alone *prrC* gene has not been detected so far. These facts hint that PrrC can be readily transmitted by horizontal gene transfer (HGT), possibly from a *prr* donor to an *hsd* acceptor. The dependence of PrrC's function on its detoxifying partner, the linked R-M system is indicated also by their coincident inactivation in a *Neisseria meningitidis* strain (Meineke and Shuman, pers. comm.). This addiction and the similar ACNase activities of various PrrC orthologs examined (Davidov & Kaufmann, 2008;Meineke *et al*, 2010) further suggest that PrrC acts in general as a translation-disabling, antiviral contingency mobilized when the

The host-phage survival cascade depicted in Fig. 1B entails some caveats. Namely, the DNA of T4 and related phages incorporates 5-hydromethylcytosine (5-HmC) instead of cytosine and 5-HmC is further glucosylated at the DNA level (Morera *et al*, 1999). Due to this hypermodification the phage DNA is refractory to many DNA restriction nucleases (Miller *et al*, 2003b) including EcoprrI and, hence, need not be protected from them by Stp. Moreover, a T4 mutant with unmodified cytosine in its DNA succumbs to EcoprrI's restriction, notwithstanding Stp's presence. The failure of Stp to protect this EcoprrI-sensitive mutant can be accounted for by the delayed-early schedule of its expression, a few minutes after the onset of the infection (Jabbar & Snyder, 1984;David *et al*, 1982). Due to these reasons EcoprrI's DNA restriction and Stp's anti-restriction activities were investigated using surrogate lambdoid phages (Jabbar & Snyder, 1984;Penner *et al*, 1995). Yet, the conservation of Stp's sequence among T4-like phages (Penner *et al*, 1995) http://phage.ggc.edu/,

henceforth Pnk) (Richardson, 1965;Becker & Hurwitz, 1967;Cameron & Uhlenbeck, 1977) and RNA ligase 1 (Rnl1, Silber *et al*, 1972;Ho & Shuman, 2002). The combined activities of Pnk and Rnl1 seemed tailored to fix RNA nicks, converting 3'-phosphoryl or 2',3'-cyclic phosphate and 5'-OH cleavage ends into 3'5' phosphodiester linkages (Kaufmann & Kallenbach, 1975;Amitsur *et al*, 1987). Suggested alternative roles in DNA metabolism (Novogrodsky *et al*, 1966;Depew & Cozzarelli, 1974) were assigned in later years to a related eukaryal DNA kinasephosphatase essential for genome stability and a possible therapeutic target in cancer cells rendered resistant to genotoxic drugs (Weinfeld *et al*, 2011).

Pnk and Rnl1 are dispensable for T4 growth on common *E. coli* laboratory strains but required on a rare host encoding the optional locus *prr* (*pnk* and *rnl1* restriction) (Depew & Cozzarelli, 1974; Sirotkin *et al*, 1978; Runnels *et al*, 1982; Jabbar & Snyder, 1984). Mutating a minuscule T4 orf termed *stp* (*s*uppressor of *t*hree-prime *p*hosphatase) abrogates *prr* restriction (Depew & Cozzarelli, 1974;Depew *et al*, 1975;Chapman *et al*, 1988;Penner *et al*, 1995). These facts reinforced the notion that Pnk and Rnl1 cooperate in RNA nick repair. They also led to the detection of the *prr*-encoded latent ACNase comprising the core ACNase PrrC and PrrC's silencing partner, the associated type Ic DNA restriction-modification (R-M) system EcoprrI (Levitz *et al*, 1990;Linder *et al*, 1990;Amitsur *et al*, 1992;Tyndall *et al*, 1994). EcoprrI and PrrC are also genetically linked, the ACNase core gene *prrC* is flanked by the genes encoding the three R-M subunit types *hsdMSR*/*prrABD* (Fig. 1A).

Type I R-M systems to which EcoprrI belongs recognize with their HsdS subunit a bipartite target containing a variable 6-8nt long spacer such as EcoprrI's CCAN7RTGC (Tyndall *et al*, 1994). HsdS associates with two HsdM protomers to form a site-specific DNA methylase (HsdM2S). Further attachment of two HsdR protomers yields a full-fledged R-M protein (HsdR2M2S). The R-M protein ignores a fully methylated target and readily methylates a hemi-methylated one. A fully unmodified target, usually of foreign DNA, induces the helicase domains of the HsdR protomers to pump-in DNA flanking the target sequence at the expense of ATP hydrolysis. This translocation and consequent DNA looping go on until an obstacle is encountered and cleavage occurs, usually far away from the specific recognition site. The type I R-M proteins are divided into families by antigenic crossreactivity, subunit interchangeability and sequence similarity. PrrC is invariably linked to type Ic family members while RloC may interact with type Ia or the distantly related type III R-M proteins. For detailed coverage of DNA restriction and anti-restriction the readers are encouraged to consult relevant reviews (Murray, 2000;Dryden *et al*, 2001;Youell & Firman, 2008;Janscak *et al*, 2001).

EcoprrI normally silences PrrC's ACNase activity in the uninfected cell (Fig. 1B). The significance of this masking interaction is indicated by the "double-edged" nature of the T4 encoded peptide Stp, mutations in which suppress *prr* restriction. Thus, Stp inhibits EcoprrI's DNA restriction, probably its intended function; and activates the latent ACNase, its host co-opted task (Penner *et al*, 1995). Once activated PrrC nicks cellular tRNALys 5' to the wobble base, yielding 2', 3'-cyclic phosphate and 5'-OH termini. Since T4 shuts-off host transcription (Mathews, 1994) and does not encode tRNALys (Schmidt & Apirion, 1983) the lesion inflicted by PrrC could disable T4 late translation and contain the infection (Sirotkin *et al*, 1978). However, T4 overcomes also this hurdle by using Pnk and Rnl1 to resuscitate the damaged tRNALys. Pnk heals the cleavage termini, converting them into a 3'-OH and 5'-P pair that Rnl1 seals (Amitsur *et al*, 1987)(Fig. 1B). In other words, this host-phage survival cascade gave rise to an RNA repair pathway. The ability of the *prr*encoded latent ACNase to restrict only tRNA repair-deficient phage invokes the possible

henceforth Pnk) (Richardson, 1965;Becker & Hurwitz, 1967;Cameron & Uhlenbeck, 1977) and RNA ligase 1 (Rnl1, Silber *et al*, 1972;Ho & Shuman, 2002). The combined activities of Pnk and Rnl1 seemed tailored to fix RNA nicks, converting 3'-phosphoryl or 2',3'-cyclic phosphate and 5'-OH cleavage ends into 3'5' phosphodiester linkages (Kaufmann & Kallenbach, 1975;Amitsur *et al*, 1987). Suggested alternative roles in DNA metabolism (Novogrodsky *et al*, 1966;Depew & Cozzarelli, 1974) were assigned in later years to a related eukaryal DNA kinasephosphatase essential for genome stability and a possible therapeutic target in cancer cells

Pnk and Rnl1 are dispensable for T4 growth on common *E. coli* laboratory strains but required on a rare host encoding the optional locus *prr* (*pnk* and *rnl1* restriction) (Depew & Cozzarelli, 1974; Sirotkin *et al*, 1978; Runnels *et al*, 1982; Jabbar & Snyder, 1984). Mutating a minuscule T4 orf termed *stp* (*s*uppressor of *t*hree-prime *p*hosphatase) abrogates *prr* restriction (Depew & Cozzarelli, 1974;Depew *et al*, 1975;Chapman *et al*, 1988;Penner *et al*, 1995). These facts reinforced the notion that Pnk and Rnl1 cooperate in RNA nick repair. They also led to the detection of the *prr*-encoded latent ACNase comprising the core ACNase PrrC and PrrC's silencing partner, the associated type Ic DNA restriction-modification (R-M) system EcoprrI (Levitz *et al*, 1990;Linder *et al*, 1990;Amitsur *et al*, 1992;Tyndall *et al*, 1994). EcoprrI and PrrC are also genetically linked, the ACNase core gene *prrC* is flanked by the genes encoding the three

Type I R-M systems to which EcoprrI belongs recognize with their HsdS subunit a bipartite target containing a variable 6-8nt long spacer such as EcoprrI's CCAN7RTGC (Tyndall *et al*, 1994). HsdS associates with two HsdM protomers to form a site-specific DNA methylase (HsdM2S). Further attachment of two HsdR protomers yields a full-fledged R-M protein (HsdR2M2S). The R-M protein ignores a fully methylated target and readily methylates a hemi-methylated one. A fully unmodified target, usually of foreign DNA, induces the helicase domains of the HsdR protomers to pump-in DNA flanking the target sequence at the expense of ATP hydrolysis. This translocation and consequent DNA looping go on until an obstacle is encountered and cleavage occurs, usually far away from the specific recognition site. The type I R-M proteins are divided into families by antigenic crossreactivity, subunit interchangeability and sequence similarity. PrrC is invariably linked to type Ic family members while RloC may interact with type Ia or the distantly related type III R-M proteins. For detailed coverage of DNA restriction and anti-restriction the readers are encouraged to consult relevant reviews (Murray, 2000;Dryden *et al*, 2001;Youell & Firman,

EcoprrI normally silences PrrC's ACNase activity in the uninfected cell (Fig. 1B). The significance of this masking interaction is indicated by the "double-edged" nature of the T4 encoded peptide Stp, mutations in which suppress *prr* restriction. Thus, Stp inhibits EcoprrI's DNA restriction, probably its intended function; and activates the latent ACNase, its host co-opted task (Penner *et al*, 1995). Once activated PrrC nicks cellular tRNALys 5' to the wobble base, yielding 2', 3'-cyclic phosphate and 5'-OH termini. Since T4 shuts-off host transcription (Mathews, 1994) and does not encode tRNALys (Schmidt & Apirion, 1983) the lesion inflicted by PrrC could disable T4 late translation and contain the infection (Sirotkin *et al*, 1978). However, T4 overcomes also this hurdle by using Pnk and Rnl1 to resuscitate the damaged tRNALys. Pnk heals the cleavage termini, converting them into a 3'-OH and 5'-P pair that Rnl1 seals (Amitsur *et al*, 1987)(Fig. 1B). In other words, this host-phage survival cascade gave rise to an RNA repair pathway. The ability of the *prr*encoded latent ACNase to restrict only tRNA repair-deficient phage invokes the possible

rendered resistant to genotoxic drugs (Weinfeld *et al*, 2011).

R-M subunit types *hsdMSR*/*prrABD* (Fig. 1A).

2008;Janscak *et al*, 2001).

existence of a "smarter" ACNase able to encumber phage reversal. Later we ask if RloC could be one.

Fig. 1. A host-phage survival cascade gives rise to an RNA repair pathway. A. The optional host locus *prr* comprises the core ACNase gene *prrC* and flanking genes encoding the type Ic DNA R-M protein EcoprrI that silences PrrC's ACNase activity. Arrows mark transcription start sites. B. Cleavage-ligation of tRNALys in phage T4 infected *E. coli prr+*. T4's anti-DNA restriction factor Stp inhibits EcoprrI and activates the latent ACNase. The resultant disruption of tRNALys is reversed by the T4's tRNA repair enzymes Pnk and Rnl1.

Nested *prr* loci where *prrC* intervenes a type Ic *hsd* locus (Fig 1A) appear sporadically in distantly related bacteria. They are present in some strains of a given species but not in others, as would a niche-function (Blanga-Kanfi *et al*, 2006). They abound among *Proteobacteria*, are less frequent in *Bacteroidetes* and *Firmicutes*, rare in *Actinobacteria* and apparently absent from *Cyanobacteria*. PrrC's phylogenic tree does not match the bacterial, unlike the associated type Ic R-M protein, which only rarely teams with PrrC. In contrast, a stand-alone *prrC* gene has not been detected so far. These facts hint that PrrC can be readily transmitted by horizontal gene transfer (HGT), possibly from a *prr* donor to an *hsd* acceptor. The dependence of PrrC's function on its detoxifying partner, the linked R-M system is indicated also by their coincident inactivation in a *Neisseria meningitidis* strain (Meineke and Shuman, pers. comm.). This addiction and the similar ACNase activities of various PrrC orthologs examined (Davidov & Kaufmann, 2008;Meineke *et al*, 2010) further suggest that PrrC acts in general as a translation-disabling, antiviral contingency mobilized when the linked R-M system is compromised.

The host-phage survival cascade depicted in Fig. 1B entails some caveats. Namely, the DNA of T4 and related phages incorporates 5-hydromethylcytosine (5-HmC) instead of cytosine and 5-HmC is further glucosylated at the DNA level (Morera *et al*, 1999). Due to this hypermodification the phage DNA is refractory to many DNA restriction nucleases (Miller *et al*, 2003b) including EcoprrI and, hence, need not be protected from them by Stp. Moreover, a T4 mutant with unmodified cytosine in its DNA succumbs to EcoprrI's restriction, notwithstanding Stp's presence. The failure of Stp to protect this EcoprrI-sensitive mutant can be accounted for by the delayed-early schedule of its expression, a few minutes after the onset of the infection (Jabbar & Snyder, 1984;David *et al*, 1982). Due to these reasons EcoprrI's DNA restriction and Stp's anti-restriction activities were investigated using surrogate lambdoid phages (Jabbar & Snyder, 1984;Penner *et al*, 1995). Yet, the conservation of Stp's sequence among T4-like phages (Penner *et al*, 1995) http://phage.ggc.edu/,

RloC: A Translation-Disabling tRNase Implicated in

(Schirmer & Evans, 1990) (Fig. 2B).

binding site.

**2.3 Players in PrrC's silencing and activation** 

discussed later in this section.

Phage Exclusion During Recovery from DNA Damage 25

Accordingly, the PrrC dimer of dimers assumes a phosphofructokinase-like topology

Fig. 2. Functional structure and possible quaternary organization of PrrC. **A**. PrrC's Nproximal ABC-ATPase domain features motifs involved in binding and hydrolysis of the nucleotide's triphosphate moiety (Walker A, Q-loop, ABC signature (ABC), Walker B, Dloop and linchpin Switch region (SW) but not the nucleobase recognizing Y-loop motif. The unique PrrC Box motif shown in WebLogo format, a putative functional substitute of the Y-loop, could confer the unusual GTP/dTTP specificity of PrrC. **B.** Antiparallel dimerization of the N-domains (Moody & Thomas, 2005) and anticipated parallel dimerization of the C-domains (Klaiman *et al*, 2007) suggest that PrrC assumes a

phosphofructokinase-like quaternary topology (Schirmer & Evans, 1990). NBS – nucleotide

As mentioned, PrrC's toxic activity is normally silenced, being unleashed only during phage infection. The requisite switches are provided in the case of *Eco*PrrC by its silencing partner EcoprrI, the phage T4-encoded anti-DNA restriction factor Stp and the motor domains of the ACNase protein itself. Insights into the underlying mechanisms were provided by the discrepant behaviors of the latent ACNase holoenzyme and the core ACNase activity of the unassociated PrrC. Thus, *in vitro* activation of the latent ACNase requires besides the Stp peptide, the DNA tethered to EcoprrI, GTP hydrolysis and the presence of dTTP. In contrast, the overt activity of the core ACNase is refractory to Stp, DNA and GTP but rapidly decays without dTTP (Amitsur *et al*, 2003;Blanga-Kanfi *et al*, 2006). These differences have been taken to indicate that Stp triggers the activation of the latent ACNase, GTP hydrolysis drives conformational changes needed to turn it on while the binding of dTTP stabilizes the ACNase once activated. The possible role of EcoprrI's DNA ligand is

GTP and dTTP probably exert their respective ACNase activating and stabilizing functions by interacting with PrrC's N-domains. This is suggested by their binding to

indicates that this anti-DNA restriction factor provides selective advantage, e.g., preventing nucleases related to EcoprrI from cleaving nascent, not yet glucosylated progeny DNA. The importance of Pnk and Rnl1 as PrrC's countermeasures is suggested by the following observations. First, docking tRNA on the crystal structure of T4 Pnk or Rnl1 places the anticodon loop at their respective active sites. These outcomes have been taken to indicate that both Pnk and Rnl1 evolved to repair a disrupted anticodon loop (Galburt *et al*, 2002;El Omari K. *et al*, 2006). Second, T4-related phages expected to infect *prr*-encoding bacteria feature both Pnk and Rnl1 (Miller *et al*, 2003a;Blondal *et al*, 2005;Blondal *et al*, 2003) whereas T4-related cyanophages, which are less likely to encounter *prr*, lack these tRNA repair proteins (http://phage.ggc.edu/).

#### **2.2 PrrC's functional organization**

PrrC comprises a regulatory motor domain occupying the N-proximal two thirds of its 396aa polypeptide (*Eco*PrrC). The remaining part constitutes the ACNase domain (Fig. 2A). The N-domain resembles ATP Binding Cassette (ABC) ATPases. These are universal motor components found in membrane-spanning transporters and in soluble proteins engaged in DNA repair, translation and related functions (Hopfner & Tainer, 2003). PrrC's N-domain differs from typical ABC ATPases in certain sequence attributes and in its unusual nucleotide specificity. The ABC ATPase motifs found in it partake in binding and hydrolysis of the nucleotide triphosphate moiety (Chen *et al*, 2003). However, the nucleobase recognizing motif of many transporter ABC ATPases termed A- or Y-loop (Ambudkar *et al*, 2006) is missing from PrrC. On the other hand, PrrC contains between its Walker A and Q-loop motifs a unique 16-residue motif rich in aromatic, acidic and other hydrophilic residues (Fig. 2A). This PrrC Box motif is highly degenerate (or rudimental) in RloC and is missing from other ABC ATPases and any other protein in the public database (Amitsur *et al*, 2003;Blanga-Kanfi *et al*, 2006). The PrrC Box candidates as a Y-loop substitute, responsible perhaps for PrrC's unusual specificity, the ability to simultaneously interact with its two different effector nucleotides GTP and dTTP (Blanga-Kanfi *et al*, 2006; unpublished data).

PrrC's ACNase domain harbors a catalytic ACNase triad (Arg320-Glu324-His356 in *Eco*PrrC) shared also by most RloC's orthologs except for a few cases where Glu is replaced by Asp. By analogy with the catalytic triad of RNase T1 (Gerlt, 1993;Steyaert, 1997), in the PrrC/RloC triad Glu and His could function as respective general base and acid catalysts while Arg could stabilize the pentameric transition state phosphate. The ACNase domain contains also residues implicated in recognition of the substrate's anticodon. Mutating one of them, *Eco*PrrC's Asp287 impairs the reactivity of the natural substrate and enhances that of analogs with a hypomodified or heterologous wobble base. These compensations hint that Asp287 interacts with the wobble base modifying side chain (Meidler *et al*, 1999;Jiang *et al*, 2001;Jiang *et al*, 2002).

When PrrC is expressed by itself it exhibits overt (core) ACNase activity. This core activity purifies with an oligomeric PrrC form, possibly a dimer of dimers. The N-domains of each dimer are expected to create two nucleotide binding sites (NBS) at their anti-parallel dimerization interfaces, as do typical ABC ATPases (Hopfner *et al*, 2000;Chen *et al*, 2003). In contrast, the ACNase C-domains are thought to dimerize in parallel, judged from the (i) behavior of a peptide mimic of a PrrC region implicated in the recognition of the tRNA substrate and (ii) ability of single to-Cys replacements in an overlapping PrrC region to induce disulphide-bond-dependent subunit dimerization (Klaiman *et al*, 2007).

indicates that this anti-DNA restriction factor provides selective advantage, e.g., preventing nucleases related to EcoprrI from cleaving nascent, not yet glucosylated progeny DNA. The importance of Pnk and Rnl1 as PrrC's countermeasures is suggested by the following observations. First, docking tRNA on the crystal structure of T4 Pnk or Rnl1 places the anticodon loop at their respective active sites. These outcomes have been taken to indicate that both Pnk and Rnl1 evolved to repair a disrupted anticodon loop (Galburt *et al*, 2002;El Omari K. *et al*, 2006). Second, T4-related phages expected to infect *prr*-encoding bacteria feature both Pnk and Rnl1 (Miller *et al*, 2003a;Blondal *et al*, 2005;Blondal *et al*, 2003) whereas T4-related cyanophages, which are less likely to encounter *prr*, lack these tRNA repair

PrrC comprises a regulatory motor domain occupying the N-proximal two thirds of its 396aa polypeptide (*Eco*PrrC). The remaining part constitutes the ACNase domain (Fig. 2A). The N-domain resembles ATP Binding Cassette (ABC) ATPases. These are universal motor components found in membrane-spanning transporters and in soluble proteins engaged in DNA repair, translation and related functions (Hopfner & Tainer, 2003). PrrC's N-domain differs from typical ABC ATPases in certain sequence attributes and in its unusual nucleotide specificity. The ABC ATPase motifs found in it partake in binding and hydrolysis of the nucleotide triphosphate moiety (Chen *et al*, 2003). However, the nucleobase recognizing motif of many transporter ABC ATPases termed A- or Y-loop (Ambudkar *et al*, 2006) is missing from PrrC. On the other hand, PrrC contains between its Walker A and Q-loop motifs a unique 16-residue motif rich in aromatic, acidic and other hydrophilic residues (Fig. 2A). This PrrC Box motif is highly degenerate (or rudimental) in RloC and is missing from other ABC ATPases and any other protein in the public database (Amitsur *et al*, 2003;Blanga-Kanfi *et al*, 2006). The PrrC Box candidates as a Y-loop substitute, responsible perhaps for PrrC's unusual specificity, the ability to simultaneously interact with its two different effector nucleotides GTP and dTTP (Blanga-

PrrC's ACNase domain harbors a catalytic ACNase triad (Arg320-Glu324-His356 in *Eco*PrrC) shared also by most RloC's orthologs except for a few cases where Glu is replaced by Asp. By analogy with the catalytic triad of RNase T1 (Gerlt, 1993;Steyaert, 1997), in the PrrC/RloC triad Glu and His could function as respective general base and acid catalysts while Arg could stabilize the pentameric transition state phosphate. The ACNase domain contains also residues implicated in recognition of the substrate's anticodon. Mutating one of them, *Eco*PrrC's Asp287 impairs the reactivity of the natural substrate and enhances that of analogs with a hypomodified or heterologous wobble base. These compensations hint that Asp287 interacts with the wobble base modifying side chain (Meidler *et al*, 1999;Jiang *et al*,

When PrrC is expressed by itself it exhibits overt (core) ACNase activity. This core activity purifies with an oligomeric PrrC form, possibly a dimer of dimers. The N-domains of each dimer are expected to create two nucleotide binding sites (NBS) at their anti-parallel dimerization interfaces, as do typical ABC ATPases (Hopfner *et al*, 2000;Chen *et al*, 2003). In contrast, the ACNase C-domains are thought to dimerize in parallel, judged from the (i) behavior of a peptide mimic of a PrrC region implicated in the recognition of the tRNA substrate and (ii) ability of single to-Cys replacements in an overlapping PrrC region to induce disulphide-bond-dependent subunit dimerization (Klaiman *et al*, 2007).

proteins (http://phage.ggc.edu/).

**2.2 PrrC's functional organization** 

Kanfi *et al*, 2006; unpublished data).

2001;Jiang *et al*, 2002).

Accordingly, the PrrC dimer of dimers assumes a phosphofructokinase-like topology (Schirmer & Evans, 1990) (Fig. 2B).

Fig. 2. Functional structure and possible quaternary organization of PrrC. **A**. PrrC's Nproximal ABC-ATPase domain features motifs involved in binding and hydrolysis of the nucleotide's triphosphate moiety (Walker A, Q-loop, ABC signature (ABC), Walker B, Dloop and linchpin Switch region (SW) but not the nucleobase recognizing Y-loop motif. The unique PrrC Box motif shown in WebLogo format, a putative functional substitute of the Y-loop, could confer the unusual GTP/dTTP specificity of PrrC. **B.** Antiparallel dimerization of the N-domains (Moody & Thomas, 2005) and anticipated parallel dimerization of the C-domains (Klaiman *et al*, 2007) suggest that PrrC assumes a phosphofructokinase-like quaternary topology (Schirmer & Evans, 1990). NBS – nucleotide binding site.

#### **2.3 Players in PrrC's silencing and activation**

As mentioned, PrrC's toxic activity is normally silenced, being unleashed only during phage infection. The requisite switches are provided in the case of *Eco*PrrC by its silencing partner EcoprrI, the phage T4-encoded anti-DNA restriction factor Stp and the motor domains of the ACNase protein itself. Insights into the underlying mechanisms were provided by the discrepant behaviors of the latent ACNase holoenzyme and the core ACNase activity of the unassociated PrrC. Thus, *in vitro* activation of the latent ACNase requires besides the Stp peptide, the DNA tethered to EcoprrI, GTP hydrolysis and the presence of dTTP. In contrast, the overt activity of the core ACNase is refractory to Stp, DNA and GTP but rapidly decays without dTTP (Amitsur *et al*, 2003;Blanga-Kanfi *et al*, 2006). These differences have been taken to indicate that Stp triggers the activation of the latent ACNase, GTP hydrolysis drives conformational changes needed to turn it on while the binding of dTTP stabilizes the ACNase once activated. The possible role of EcoprrI's DNA ligand is discussed later in this section.

GTP and dTTP probably exert their respective ACNase activating and stabilizing functions by interacting with PrrC's N-domains. This is suggested by their binding to

RloC: A Translation-Disabling tRNase Implicated in

**3.1 Functional organization** 

Phage Exclusion During Recovery from DNA Damage 27

Fig. 3. DNA tethered to EcoprrI could figure in PrrC's regulation. A. ssDNA inhibits PrrC ACNase. PrrC ACNase was assayed using a 5'-32P labeled anticodon stem loop substrate and increasing levels of a nonspecific 17nt PCR primer. B. PrrC contacts DNA regions flanking EcoprrI's target. A 249bp DNA fragment with a near-central EcoprrI site was singly

**3. RloC - A translation-disabling and potential DNA-damage-sensing protein** 

RloC is a conserved bacterial protein that shares PrrC's overall organization into a motor Ndomain and ACNase C-domain (Fig. 4) (Davidov & Kaufmann, 2008). However, RloC is about twice as large, its orthologs ranging in size between 650 to 900 residues compared to 350-420 with PrrC. This increase is mainly due to a long coiled-coil forming sequence inserted between RloC's Walker A and ABC signature motifs. This coiled-coil sequence contains near its center a loop featuring the conserved zinc-hook motif CXXC. A similar coiled-coil insert in an ABC ATPase head-domain characterizes the universal DNA-damagecheckpoint/DNA-repair protein Rad50/SbcC (Hopfner *et al*, 2002;Connelly *et al*, 1998). Rad50's insert protrudes from the ATPase head-domain as an antiparallel coiled-coil presenting the zinc-hook motif at its apex. The apical ends of two such protrusions dimerize by coordinating Zn++ to their four cysteines. This zinc-hook linkage can arise intramolecularly, connecting the two coiled-coil protrusions of the same Rad50 dimer. Alternatively, when Rad50's ATPase head-domains are bound to DNA the two protrusions straighten. In this form they can dimerize only inter-molecularly, bridging in this manner

32P-labeled at specific sites and tethered to the EcoprrI-PrrC complex. Following UVirradiation, DNase I digestion, the photo-labeled PrrC was immunoprecipitated, separated by SDS-PAGE and monitored by autoradiography. Brown and blue asterisks indicate sites PrrC did or did not crosslink to, respectively. C. In this model DNA unwound by EcoprrI silences PrrC and its rewinding due to Stp's interaction with EcoprrI unleashes the ACNase.

full-sized PrrC protein or PrrC's isolated N-domains with vastly differing affinities (mMand μM-range, respectively) and without displacing each other (Amitsur *et al*, 2003;Blanga-Kanfi *et al*, 2006; and unpublished data). This unusual specificity distinguishes PrrC from its distant homolog RloC and other ABC ATPase-containing proteins, which bind and hydrolyze ATP or GTP (Guo *et al*, 2006) and are not expected to avidly bind dTTP (our unpublished data).

The biological significance of PrrC's idiosyncratic interaction with dTTP has been hinted at by the dramatic increase in the cellular level of dTTP early in phage T4 infection, when the ACNase is induced (Amitsur *et al*, 2003;Blanga-Kanfi *et al*, 2006). The increased level of dTTP benefits the phage by safeguarding effective and faithful replication of its AT-rich DNA. In fact, delaying dTTP's accretion by mutating T4's dCMP deaminase (Cd) elicits a mutator phenotype indicated by increased frequency of ATGC transitions (Sargent & Mathews, 1987). The Cd deficiency, and, by implication, the consequent delay in dTTP's accretion, also reduce 2-3 fold the extent of the PrrC-mediated cleavage of tRNALys. This partial inhibition does not suffice to suppress *prr* restriction but is synthetically suppressive with a leaky *stp* mutation that also fails to suppress *prr* restriction by itself (Klaiman & Kaufmann, 2011). Thus, dTTP's accretion is another T4 contraption expatiated by the bacterial host, in that case to stabilize the activated ACNase.

PrrC's ability to "gauge" changes in dTTP's level could benefit its host also by precluding the toxicity of any free PrrC molecules that could arise in the uninfected cell due to their translation in excess over EcoprrI or dissociation from the latent holoenzyme. Their excessive translation may be stochastic or programmed to saturate the silencing partner. PrrC's dissociation from the latent holoenzyme may be accidental or due to EcoprrI's disruption in response to DNA damage (Restriction Alleviation, RA) (Makovets *et al*, 2004) (see also section 3.6). Free PrrC's cytotoxicity has been indicated by the coincident inactivation of *prrC* and linked *hsd* genes, by the self-limiting expression of free PrrC (Meidler *et al*, 1999;Blanga-Kanfi *et al*, 2006) and the rapid *in vivo* inactivation of the core ACNase (Amitsur *et al*, 2003). The ACNase enhancing effects of dTTP's accretion during phage T4 infection (Klaiman & Kaufmann, 2011) and *in vitro* stabilization of the core ACNase by dTTP (Amitsur *et al*, 2003) suggest that the *in vivo* instability of the core ACNase owes to the relatively low dTTP level in the uninfected cell. Although this level far exceeds that needed to stabilize the core ACNase *in vitro*, the actual level availed to PrrC in the cell could be prohibitively low due to localization of the nucleotide pools (Wheeler *et al*, 1996). In sum, we propose that PrrC's ability to gauge dTTP's level not only stabilizes its activated form but also confines the toxicity of this ACNase to the viral target.

Yet another player in PrrC's regulation is the DNA tethered to EcoprrI (Amitsur *et al*, 2003). Its possible role is suggested by three observations. First, short nonspecific ssDNA oligonucleotides avidly bind PrrC and competitively inhibit its ACNase activity (Fig. 3A and unpublished results), hinting that ssDNA encountered by PrrC in the uninfected cell helps silence the ACNase. Second, the type Ic DNA R-M protein EcoR124I unwinds short DNA stretches flanking its target sequence (van Noort *et al*, 2004;Stanley *et al*, 2006), suggesting a possible source for the putative ACNase-inhibiting ssDNA. Third, within a latent ACNase complex tethered to an EcoprrI DNA ligand PrrC was UV-crosslinked to DNA regions flanking EcoprrI's recognition site (Fig. 3B). These facts underlie a model where DNA unwound by EcoprrI helps silence PrrC and its rewinding due to Stp's interaction with EcoprrI unleashes the ACNase (Fig. 3C).

full-sized PrrC protein or PrrC's isolated N-domains with vastly differing affinities (mMand μM-range, respectively) and without displacing each other (Amitsur *et al*, 2003;Blanga-Kanfi *et al*, 2006; and unpublished data). This unusual specificity distinguishes PrrC from its distant homolog RloC and other ABC ATPase-containing proteins, which bind and hydrolyze ATP or GTP (Guo *et al*, 2006) and are not expected to

The biological significance of PrrC's idiosyncratic interaction with dTTP has been hinted at by the dramatic increase in the cellular level of dTTP early in phage T4 infection, when the ACNase is induced (Amitsur *et al*, 2003;Blanga-Kanfi *et al*, 2006). The increased level of dTTP benefits the phage by safeguarding effective and faithful replication of its AT-rich DNA. In fact, delaying dTTP's accretion by mutating T4's dCMP deaminase (Cd) elicits a mutator phenotype indicated by increased frequency of ATGC transitions (Sargent & Mathews, 1987). The Cd deficiency, and, by implication, the consequent delay in dTTP's accretion, also reduce 2-3 fold the extent of the PrrC-mediated cleavage of tRNALys. This partial inhibition does not suffice to suppress *prr* restriction but is synthetically suppressive with a leaky *stp* mutation that also fails to suppress *prr* restriction by itself (Klaiman & Kaufmann, 2011). Thus, dTTP's accretion is another T4 contraption expatiated by the

PrrC's ability to "gauge" changes in dTTP's level could benefit its host also by precluding the toxicity of any free PrrC molecules that could arise in the uninfected cell due to their translation in excess over EcoprrI or dissociation from the latent holoenzyme. Their excessive translation may be stochastic or programmed to saturate the silencing partner. PrrC's dissociation from the latent holoenzyme may be accidental or due to EcoprrI's disruption in response to DNA damage (Restriction Alleviation, RA) (Makovets *et al*, 2004) (see also section 3.6). Free PrrC's cytotoxicity has been indicated by the coincident inactivation of *prrC* and linked *hsd* genes, by the self-limiting expression of free PrrC (Meidler *et al*, 1999;Blanga-Kanfi *et al*, 2006) and the rapid *in vivo* inactivation of the core ACNase (Amitsur *et al*, 2003). The ACNase enhancing effects of dTTP's accretion during phage T4 infection (Klaiman & Kaufmann, 2011) and *in vitro* stabilization of the core ACNase by dTTP (Amitsur *et al*, 2003) suggest that the *in vivo* instability of the core ACNase owes to the relatively low dTTP level in the uninfected cell. Although this level far exceeds that needed to stabilize the core ACNase *in vitro*, the actual level availed to PrrC in the cell could be prohibitively low due to localization of the nucleotide pools (Wheeler *et al*, 1996). In sum, we propose that PrrC's ability to gauge dTTP's level not only stabilizes its activated

Yet another player in PrrC's regulation is the DNA tethered to EcoprrI (Amitsur *et al*, 2003). Its possible role is suggested by three observations. First, short nonspecific ssDNA oligonucleotides avidly bind PrrC and competitively inhibit its ACNase activity (Fig. 3A and unpublished results), hinting that ssDNA encountered by PrrC in the uninfected cell helps silence the ACNase. Second, the type Ic DNA R-M protein EcoR124I unwinds short DNA stretches flanking its target sequence (van Noort *et al*, 2004;Stanley *et al*, 2006), suggesting a possible source for the putative ACNase-inhibiting ssDNA. Third, within a latent ACNase complex tethered to an EcoprrI DNA ligand PrrC was UV-crosslinked to DNA regions flanking EcoprrI's recognition site (Fig. 3B). These facts underlie a model where DNA unwound by EcoprrI helps silence PrrC and its rewinding due to Stp's

avidly bind dTTP (our unpublished data).

bacterial host, in that case to stabilize the activated ACNase.

form but also confines the toxicity of this ACNase to the viral target.

interaction with EcoprrI unleashes the ACNase (Fig. 3C).

Fig. 3. DNA tethered to EcoprrI could figure in PrrC's regulation. A. ssDNA inhibits PrrC ACNase. PrrC ACNase was assayed using a 5'-32P labeled anticodon stem loop substrate and increasing levels of a nonspecific 17nt PCR primer. B. PrrC contacts DNA regions flanking EcoprrI's target. A 249bp DNA fragment with a near-central EcoprrI site was singly 32P-labeled at specific sites and tethered to the EcoprrI-PrrC complex. Following UVirradiation, DNase I digestion, the photo-labeled PrrC was immunoprecipitated, separated by SDS-PAGE and monitored by autoradiography. Brown and blue asterisks indicate sites PrrC did or did not crosslink to, respectively. C. In this model DNA unwound by EcoprrI silences PrrC and its rewinding due to Stp's interaction with EcoprrI unleashes the ACNase.
