**3.1 Functional organization**

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

RloC: A Translation-Disabling tRNase Implicated in

be described in the following sections.

**3.3 RloC wobble-nucleotide-excising activity** 

overwhelm the natural (Meidler *et al*, 1999).

**3.4 RloC may frustrate phage reversal** 

Phage Exclusion During Recovery from DNA Damage 29

The role ascribed to the R-M proteins in RloC's ACNase regulation need not contradict the existence of additional or alternative switches provided by the coiled-coil/zinc-hook insert. For example, silencing of the ACNase function by the latter device could be advantageous when RloC is introduced by HGT into a new host. Namely, silencing by a pre-existing R-M system could require a highly promiscuous interaction between the two partners. The possibility that RloC is endowed with an internal ACNase silencing mechanism agrees with properties of the ortholog encoded by the thermophile *Geobacillus kaustophilus* (*Gka*RloC) to

Due to its potential toxicity, RloC's ACNase activity was expected to be as unstable as PrrC's (Blanga-Kanfi *et al*, 2006). Indeed, among several RloC orthologs investigated, only *Gka*RloC proved sufficiently stable to warrant its *in vitro* characterization (Davidov & Kaufmann, 2008). Yet, even *Gka*RloC's ACNase is intrinsically unstable. Its *in vitro* activity is highest at 25°C and undetectable at 45°C (our unpublished results) although *G. kaustophilus* grows optimally at 65°C (Takami *et al*, 2004). When expressed in *E. coli Gka*RloC preferentially cleaved tRNAGlu. However, identifying RloC's natural substrate must await physiological studies. This reservation is based on the experience gained with PrrC, the over-expression of which results in cleavages of secondary substrates that

A more striking difference between RloC and PrrC is the ability of the former to cleave its tRNA substrates successively, first 3' and then 5' to the wobble position (Davidov & Kaufmann, 2008). Such an excision reaction using as a substrate yeast tRNAGlu radiolabeled 3' to the wobble base is shown in Fig. 5. The incision of this substrate 3' to the wobble base yields a labeled 5' fragment containing residues 1-34. This intermediate is further cleaved immediately upstream, yielding the labeled wobble-nucleotide. Under these *in* vitro conditions *Gka*RloC inadvertently incises the substrate also 5' to the wobble base but this reaction yields a dead-end product that is not further cleaved. This is indicated by the accumulation of this product when the overall reaction declines; and of RloC to cleave it when generated by PrrC, which normally cleaves its substrates 5' to the wobble position. Such a 5' incision product of GkaRloC is not detected *in vivo* and, therefore, is considered an *in vitro* artifact. The excision of the wobble nucleotide has been observed with different tRNA and anticodon-stem-loop substrates and was catalyzed also by a mesophilic RloC

The harsh lesion inflicted by *Gka*RloC could render this ACNase a more potent antiviral device than PrrC. Namely, RloC could perform the successive cleavages of its substrate in a processive manner, i.e., without releasing the incision intermediate. The phage tRNA repair enzymes would in that case process and ligate back the fragments lacking the wobble nucleotide and yield a defective product. Conversely, if *Gka*RloC's incision intermediate were accessible, the repair enzymes would faithfully restore the original tRNA substrate. Simulated *in vitro* encounters between *Gka*RloC and T4 Pnk or both tRNA repair enzymes indicated that a sizable fraction of its incision intermediate was occluded from the repair enzymes (Davidov & Kaufmann, 2008; and unpublished data). It is possible that under physiological conditions RloC's would more effectively occlude its incision intermediate.

species of *E. coli* APECO1 (Davidov & Kaufmann, 2008; unpublished data).

distant DNA molecules (Moreno-Herrero *et al*, 2005). Other proteins belonging to the SMC (Structure Maintenance of Chromosomes) super-family exhibit similar DNA bridging activity but link their coiled-coil protrusions via apical hydrophobic domains (Hirano, 2005). RloC is the only known protein other than Rad50/SbcC with a coiled-coil/zinc-hook containing ABC-ATPase domain. Therefore, cellular functions imparted by Rad50/SbcC may provide clues to RloC's.

Fig. 4. RloC and PrrC share the same functional organization. The alignment of *Gka*RloC and *Eco*PrrC sequences reveals shared ABC ATPase and ACNase motifs and presence in RloC's N-domain of a large coiled-coil (CC) stretch interrupted by a loop containing the zinc hook motif CXXC (adapted from ref. 4).

#### **3.2 RloC's occurrence and genomic attributes**

RloC genes appear in major bacterial phyla except for *Cyanobacteria*. They are often encased within a cryptic mobile element as a single cargo gene. This pattern and a phylogenic tree not matching the bacterial suggest that RloC is readily transmitted by HGT, like PrrC. RloC's genes are also sporadically distributed but they occur ~3-fold more frequently than PrrC's. These facts suggest that the niche function RloC provides is more beneficial to its bacterial host.

RloC was originally identified as one of various open reading frames that intervene type Ia *hsd* loci in different *Campylobacter jejuni* strains (Restriction Linked Orf, Miller *et al*, 2005). This fact and the overall resemblance to PrrC could be taken to indicate that RloC is a related ACNase also silenced by an associated Hsd protein (Davidov & Kaufmann, 2008). Yet, only ~10% of the identified RloC orthologs turned out to be linked to type Ia or the distantly related type III DNA R-M system. Nonetheless, other genomic attributes suggested that the majority of the non-linked RloC orthologs team with an R-M system *in trans*. First, most bacteria encoding them encode also a suitable R-M system while in those lacking it RloC often features poor ATPase or ACNase motifs, as if inactivated. Second, some *rloC* genes are flanked by a cryptic *hsd* locus, a full-fledged homologue of which exists elsewhere in the genome, hinting that a past Hsd-RloC interaction *in cis* was superseded by one *in trans*. Third, RloC is occasionally linked to an ArdC-like anti-DNA restriction factor (Belogurov *et al*, 2000) with or without an adjacent R-M system, suggesting its possible regulation by an R-M system in either case. Fourth, non-linked *rloC* and *hsd* genes of one species, but not their respective flanking genes can be missing both from related, syntenic species [e.g., *Acinetobacter* sp. ADP1 *rloC* and *hsd* (ACIAD0152, ACIAD3430-2) but not flanking genes are missing from various *A. baumannii* strains] (http://www.cns.fr/agc/ microscope/mage/viewer.php?S\_id=36&wwwpkgdb =aa12fda27bb61b62ac34913acfd35916.)

distant DNA molecules (Moreno-Herrero *et al*, 2005). Other proteins belonging to the SMC (Structure Maintenance of Chromosomes) super-family exhibit similar DNA bridging activity but link their coiled-coil protrusions via apical hydrophobic domains (Hirano, 2005). RloC is the only known protein other than Rad50/SbcC with a coiled-coil/zinc-hook containing ABC-ATPase domain. Therefore, cellular functions imparted by Rad50/SbcC

Fig. 4. RloC and PrrC share the same functional organization. The alignment of *Gka*RloC and *Eco*PrrC sequences reveals shared ABC ATPase and ACNase motifs and presence in RloC's N-domain of a large coiled-coil (CC) stretch interrupted by a loop containing the zinc hook

RloC genes appear in major bacterial phyla except for *Cyanobacteria*. They are often encased within a cryptic mobile element as a single cargo gene. This pattern and a phylogenic tree not matching the bacterial suggest that RloC is readily transmitted by HGT, like PrrC. RloC's genes are also sporadically distributed but they occur ~3-fold more frequently than PrrC's. These facts suggest that the niche function RloC provides is more beneficial to its

RloC was originally identified as one of various open reading frames that intervene type Ia *hsd* loci in different *Campylobacter jejuni* strains (Restriction Linked Orf, Miller *et al*, 2005). This fact and the overall resemblance to PrrC could be taken to indicate that RloC is a related ACNase also silenced by an associated Hsd protein (Davidov & Kaufmann, 2008). Yet, only ~10% of the identified RloC orthologs turned out to be linked to type Ia or the distantly related type III DNA R-M system. Nonetheless, other genomic attributes suggested that the majority of the non-linked RloC orthologs team with an R-M system *in trans*. First, most bacteria encoding them encode also a suitable R-M system while in those lacking it RloC often features poor ATPase or ACNase motifs, as if inactivated. Second, some *rloC* genes are flanked by a cryptic *hsd* locus, a full-fledged homologue of which exists elsewhere in the genome, hinting that a past Hsd-RloC interaction *in cis* was superseded by one *in trans*. Third, RloC is occasionally linked to an ArdC-like anti-DNA restriction factor (Belogurov *et al*, 2000) with or without an adjacent R-M system, suggesting its possible regulation by an R-M system in either case. Fourth, non-linked *rloC* and *hsd* genes of one species, but not their respective flanking genes can be missing both from related, syntenic species [e.g., *Acinetobacter* sp. ADP1 *rloC* and *hsd* (ACIAD0152, ACIAD3430-2) but not flanking genes are missing from various *A. baumannii* strains] (http://www.cns.fr/agc/ microscope/mage/viewer.php?S\_id=36&wwwpkgdb =aa12fda27bb61b62ac34913acfd35916.)

may provide clues to RloC's.

motif CXXC (adapted from ref. 4).

bacterial host.

**3.2 RloC's occurrence and genomic attributes** 

The role ascribed to the R-M proteins in RloC's ACNase regulation need not contradict the existence of additional or alternative switches provided by the coiled-coil/zinc-hook insert. For example, silencing of the ACNase function by the latter device could be advantageous when RloC is introduced by HGT into a new host. Namely, silencing by a pre-existing R-M system could require a highly promiscuous interaction between the two partners. The possibility that RloC is endowed with an internal ACNase silencing mechanism agrees with properties of the ortholog encoded by the thermophile *Geobacillus kaustophilus* (*Gka*RloC) to be described in the following sections.

### **3.3 RloC wobble-nucleotide-excising activity**

Due to its potential toxicity, RloC's ACNase activity was expected to be as unstable as PrrC's (Blanga-Kanfi *et al*, 2006). Indeed, among several RloC orthologs investigated, only *Gka*RloC proved sufficiently stable to warrant its *in vitro* characterization (Davidov & Kaufmann, 2008). Yet, even *Gka*RloC's ACNase is intrinsically unstable. Its *in vitro* activity is highest at 25°C and undetectable at 45°C (our unpublished results) although *G. kaustophilus* grows optimally at 65°C (Takami *et al*, 2004). When expressed in *E. coli Gka*RloC preferentially cleaved tRNAGlu. However, identifying RloC's natural substrate must await physiological studies. This reservation is based on the experience gained with PrrC, the over-expression of which results in cleavages of secondary substrates that overwhelm the natural (Meidler *et al*, 1999).

A more striking difference between RloC and PrrC is the ability of the former to cleave its tRNA substrates successively, first 3' and then 5' to the wobble position (Davidov & Kaufmann, 2008). Such an excision reaction using as a substrate yeast tRNAGlu radiolabeled 3' to the wobble base is shown in Fig. 5. The incision of this substrate 3' to the wobble base yields a labeled 5' fragment containing residues 1-34. This intermediate is further cleaved immediately upstream, yielding the labeled wobble-nucleotide. Under these *in* vitro conditions *Gka*RloC inadvertently incises the substrate also 5' to the wobble base but this reaction yields a dead-end product that is not further cleaved. This is indicated by the accumulation of this product when the overall reaction declines; and of RloC to cleave it when generated by PrrC, which normally cleaves its substrates 5' to the wobble position. Such a 5' incision product of GkaRloC is not detected *in vivo* and, therefore, is considered an *in vitro* artifact. The excision of the wobble nucleotide has been observed with different tRNA and anticodon-stem-loop substrates and was catalyzed also by a mesophilic RloC species of *E. coli* APECO1 (Davidov & Kaufmann, 2008; unpublished data).

#### **3.4 RloC may frustrate phage reversal**

The harsh lesion inflicted by *Gka*RloC could render this ACNase a more potent antiviral device than PrrC. Namely, RloC could perform the successive cleavages of its substrate in a processive manner, i.e., without releasing the incision intermediate. The phage tRNA repair enzymes would in that case process and ligate back the fragments lacking the wobble nucleotide and yield a defective product. Conversely, if *Gka*RloC's incision intermediate were accessible, the repair enzymes would faithfully restore the original tRNA substrate. Simulated *in vitro* encounters between *Gka*RloC and T4 Pnk or both tRNA repair enzymes indicated that a sizable fraction of its incision intermediate was occluded from the repair enzymes (Davidov & Kaufmann, 2008; and unpublished data). It is possible that under physiological conditions RloC's would more effectively occlude its incision intermediate.

RloC: A Translation-Disabling tRNase Implicated in

domain that harbors this structure.

proteins.

Phage Exclusion During Recovery from DNA Damage 31

The ability to activate *Gka*RloC's ACNase by ATP hydrolysis in the presence of tethered DNA is in stark contrast with the behavior of PrrC's ACNase. As mentioned, PrrC's ACNase is activated by nucleotide hydrolysis only when associated with its silencing partner EcoprrI. However, its unassociated form exhibits overt ACNase activity refractory to nucleotide hydrolysis. This discrepancy raises the possibility that RloC's ACNase can be regulated by the internal device of the protein, the coiled-coil/zinc-hook and the ATPase

Fig. 6. *Gka*RloC aggregates DNA in a zinc-hook-dependent manner. A 485bp DNA fragment was incubated with increasing levels of GkaRloC's ACNase-null mutant E696A (lanes 2-6) or with its ZH mutant derivative E696A-C291G (lane 9). Lanes 1 and 8 contain only DNA, 7,10 only the indicated protein. The cartoons depict the assumed bridged DNA aggregate formed by E696A (right) and the simpler complex formed by E696A-C291G (left). The ACNase-null mutation allows high level expression and facilitates the isolation of the RloC

Fig. 7. AFM images of plasmid pUC19 (DNA) and its complex with RloC-E696A (DNA and RloC). Blue lines stretch over pure DNA regions, green lines also over regions containing the bound protein. Regions transected by the green line feature virtual heights both of the

DNA alone (~1.5nm) and of the presumptive RloC-DNA complexes (~4.5nm).

Fig. 5. RloC excises the wobble nucleotide. Yeast tRNAGlu 32P-labeled 3' to the wobble base was incubated with *Gka*RloC. The 34mer resulting from incision 3' to the wobble base is further cleaved, yielding the wobble nucleotide. The 43mer resulting from incision 5' to the wobble base is a dead-end product that is not further cleaved. It is considered an *in vitro* artifact, as explained in the text. In the cartoon depicting these reactions the substrate is schematically represented by the anticodon stem loop outline. marks the labeled phosphate. U9 is the modified wobble base 5-methoxycarbonylmethyl-2-thiouridine. (mcm5s2U).

Moreover, repeated cleavage-ligation cycles would diminish the proportion of any incision intermediate ligated back by phage enzymes. On the other hand, the existence of tRNA repair enzymes that more efficiently extract RloC's incision intermediate and generate perhaps repair products immune to re-cleavage (Chan *et al*, 2009b) cannot be excluded. Clearly, whether RloC does frustrate phage reversal remains to be examined in situations closer to the natural.

#### **3.5 RloC's DNA bridging domain regulates its ACNase**

RloC's second striking feature is the coiled-coil/zinc-hook insert in its ABC ATPase headdomain. The presence of this structure raised the possibility that RloC is endowed with Rad50-like DNA bridging activity and uses such a faculty to respond to DNA damage cues by turning on its ACNase. That RloC is in fact endowed with DNA bridging activity is indicated by an electrophoresis mobility shift experiment and by scanning force microscopy (AFM) imaging. In the first experiment we compared *Gka*RloC constructs with an intact or mutated zinc-hook. The first protein aggregated a dsDNA probe that the second only bound (Fig. 6). Their discrepant behavior suggests that the aggregation was due to the formation of zinc-hookdependent DNA bridges. Preliminary AFM imaging data reinforce this assumption (Fig. 7).

That RloC's ACNase is regulated by the protein's coiled-coil/zinc-hook and ATPase headdomain is indicated by several observations. First, mutating RloC's zinc-hook dramatically enhances its ACNase activity *in vivo* and *in vitro* (Davidov & Kaufmann, 2008). Second, *Gka*RloC's ACNase activity is modestly enhanced by ATP and further stimulated when the protein is also tethered to DNA (Fig. 8). In contrast, DNA alone has no effect on the ACNase and the residual ACNase activity seen without added ATP is abolished by the non-hydrolyzable analog AMP-PNP. Presumably, RloC's interaction with DNA turns on its ATPase to drive conformational changes that activate the ACNase. Interestingly, mutating the zinc-hook renders the ACNase refractory to these various agents, uncoupling the ACNase from the protein's internal controls (not shown). Together, these facts suggest that RloC's mode of interaction with DNA, which is sensed by its coiled-coil/zinc-hook monitoring device and relayed by the ATPase (Fig. 9), determines if the protein's ACNase will be silenced or turned on.

Fig. 5. RloC excises the wobble nucleotide. Yeast tRNAGlu 32P-labeled 3' to the wobble base was incubated with *Gka*RloC. The 34mer resulting from incision 3' to the wobble base is further cleaved, yielding the wobble nucleotide. The 43mer resulting from incision 5' to the wobble base is a dead-end product that is not further cleaved. It is considered an *in vitro* artifact, as explained in the text. In the cartoon depicting these reactions the substrate is schematically represented by the anticodon stem loop outline. marks the labeled phosphate. U9 is the

Moreover, repeated cleavage-ligation cycles would diminish the proportion of any incision intermediate ligated back by phage enzymes. On the other hand, the existence of tRNA repair enzymes that more efficiently extract RloC's incision intermediate and generate perhaps repair products immune to re-cleavage (Chan *et al*, 2009b) cannot be excluded. Clearly, whether RloC does frustrate phage reversal remains to be examined in situations closer to the natural.

RloC's second striking feature is the coiled-coil/zinc-hook insert in its ABC ATPase headdomain. The presence of this structure raised the possibility that RloC is endowed with Rad50-like DNA bridging activity and uses such a faculty to respond to DNA damage cues by turning on its ACNase. That RloC is in fact endowed with DNA bridging activity is indicated by an electrophoresis mobility shift experiment and by scanning force microscopy (AFM) imaging. In the first experiment we compared *Gka*RloC constructs with an intact or mutated zinc-hook. The first protein aggregated a dsDNA probe that the second only bound (Fig. 6). Their discrepant behavior suggests that the aggregation was due to the formation of zinc-hookdependent DNA bridges. Preliminary AFM imaging data reinforce this assumption (Fig. 7). That RloC's ACNase is regulated by the protein's coiled-coil/zinc-hook and ATPase headdomain is indicated by several observations. First, mutating RloC's zinc-hook dramatically enhances its ACNase activity *in vivo* and *in vitro* (Davidov & Kaufmann, 2008). Second, *Gka*RloC's ACNase activity is modestly enhanced by ATP and further stimulated when the protein is also tethered to DNA (Fig. 8). In contrast, DNA alone has no effect on the ACNase and the residual ACNase activity seen without added ATP is abolished by the non-hydrolyzable analog AMP-PNP. Presumably, RloC's interaction with DNA turns on its ATPase to drive conformational changes that activate the ACNase. Interestingly, mutating the zinc-hook renders the ACNase refractory to these various agents, uncoupling the ACNase from the protein's internal controls (not shown). Together, these facts suggest that RloC's mode of interaction with DNA, which is sensed by its coiled-coil/zinc-hook monitoring device and relayed by the ATPase (Fig. 9),

modified wobble base 5-methoxycarbonylmethyl-2-thiouridine. (mcm5s2U).

**3.5 RloC's DNA bridging domain regulates its ACNase** 

determines if the protein's ACNase will be silenced or turned on.

The ability to activate *Gka*RloC's ACNase by ATP hydrolysis in the presence of tethered DNA is in stark contrast with the behavior of PrrC's ACNase. As mentioned, PrrC's ACNase is activated by nucleotide hydrolysis only when associated with its silencing partner EcoprrI. However, its unassociated form exhibits overt ACNase activity refractory to nucleotide hydrolysis. This discrepancy raises the possibility that RloC's ACNase can be regulated by the internal device of the protein, the coiled-coil/zinc-hook and the ATPase domain that harbors this structure.

Fig. 6. *Gka*RloC aggregates DNA in a zinc-hook-dependent manner. A 485bp DNA fragment was incubated with increasing levels of GkaRloC's ACNase-null mutant E696A (lanes 2-6) or with its ZH mutant derivative E696A-C291G (lane 9). Lanes 1 and 8 contain only DNA, 7,10 only the indicated protein. The cartoons depict the assumed bridged DNA aggregate formed by E696A (right) and the simpler complex formed by E696A-C291G (left). The ACNase-null mutation allows high level expression and facilitates the isolation of the RloC proteins.

Fig. 7. AFM images of plasmid pUC19 (DNA) and its complex with RloC-E696A (DNA and RloC). Blue lines stretch over pure DNA regions, green lines also over regions containing the bound protein. Regions transected by the green line feature virtual heights both of the DNA alone (~1.5nm) and of the presumptive RloC-DNA complexes (~4.5nm).

RloC: A Translation-Disabling tRNase Implicated in

2010;Paull, 2010) as well as relevant chapters in this book.

which positions the DNase to resect DSB ends (Williams *et al*, 2011).

Darmon *et al*, 2007).

Phage Exclusion During Recovery from DNA Damage 33

useful guiding hypotheses. Here it will suffice to briefly summarize pertinent features of this universal DNA-damage-responsive, DNA-repair protein. For comprehensive coverage several recent reviews are suggested (Hirano, 2006;Stracker & Petrini, 2011;Williams *et al*,

Archaeal Rad50 and the bacterial SbcC counterparts associate with the respective dimeric DNases Mre11 or SbcD. The eukaryal Rad50-Mre11 complex (MR) further associates with an adapter protein termed Nbs1 (Xrs2 in yeast), which links the ternary complex to key DNA damage checkpoints. The ternary MRN complex controls key sensing, signaling, regulating, and effecter responses triggered by DNA double-strand breaks (DSB). These responses include the activation of master regulators such as ATM as well as roles in homologous recombinational repair (HRR), microhomology-mediated end joining (MMEJ) and, occasionally, non-homologous end-joining (NHEJ). Rad50 figures in these transactions as a DNA-bridging SMC protein, using its coiled-coil/zinc-hook and ATPase to properly orient the DNA molecules it bridges and its associated protein partners (Hirano, 2005;Stracker & Petrini, 2011;Williams *et al*, 2010;Paull, 2010;Stracker & Petrini, 2011). As mentioned, Rad50's coiled-coils bend when the ATPase domains are free and stretches when tethered to DNA (van *et al*, 2003;Moreno-Herrero *et al*, 2005). This flexibility also allows the linked ATPase domain to communicate nucleotide binding and DNA ligand signals across distances and between components of the complex. These transmissions depend, among others, on the binding of Mre11 to the coiled-coil portion closest to the ATPase domain,

Rad50's bacterial homologue SbcC may likewise exert its function as a DNA bridging protein, directing SbcD to cleave hairpin structures that impede DNA replication and initiate DSB that drive HRR (Darmon *et al*, 2010;Storvik & Foster, 2011;White *et al*, 2008). Interestingly, over-expressed in *E. coli, SbcC* co-localizes with the replication factory whereas SbcD is dispersed throughout the cytoplasm. Their discrepant behaviors underlie the proposal that at its low, natural level SbcC constantly checks the replication fork for misfolded DNA, recruiting SbcD only when repair is required. A different distribution in *B. subtilis* suggests that in this organism SbcCD partakes also in NHEJ (Mascarenhas *et al*, 2006;

*Gka*RloC could use its DNA bridging activity (Figs. 5, 6) to monitor the status of cellular DNA molecules like Rad50 and SbcC. However, there is no evidence that RloC associates with a DNase corresponding to Mre11 or SbcD. On the other hand, RloC's regulatory domain, Rad50/SbcC's counterpart is uniquely appended to the translation-disabling ACNase domain. It is tempting to speculate therefore that the ACNase C-domain interacts with the regulatory N-domain in a manner analogous to Mre11's, i.e., tethers to the proximal portion of the coiled-coil fiber emerging from the ATPase head-domain. Such a contact could help transduce DNA damage signals sensed by RloC's DNA monitoring device and relayed by the ATPase to the ACNase effecter domain. The existence of such a signal transduction pathway agrees with the effects of RloC's zinc-hook mutations, ATPase and

The suggestions that RloC's ACNase is activated in response to DNA damage and, consequently, arrests translation may seem self-contradictory. After all, bacteria normally respond to DNA insults by enhancing the synthesis of DNA repair and other stress responsive proteins (Fernandez De Henestrosa *et al*, 2000). This apparent contradiction may be reconciled by considering the phenomenon of DNA restriction alleviation (RA) (Thoms & Wackernagel, 1984). RA is enacted in response to genotoxic stress as a protective measure

tethered DNA on its ACNase function (Davidov & Kaufmann, 2008) (Fig 8).

Fig. 8. *Gka*RloC's ATPase and tethered DNA cooperatively regulate its ACNase function. *Gka*RloC's ACNase activity was assayed using as a substrate a 5'-32P labeled anticodon-stemloop analog corresponding to mammalian tRNALys3 (ASL). The reaction was performed in the absence or presence of 2mM of ATP and/or 10ng/μl of BstE II digested λ DNA, or in the presence of the non-hydrolyzable ATP analog AMP-PNP. The 7mer is a radiolabeled fragment resulting from the final excision reaction

Fig. 9. RloC's anticipated DNA bridging activity. By analogy with Rad50, RloC bridges DNA through Zn++ (orange circles) coordinated at zinc-hook (ZH) dimerization interfaces (yellow circles) at the apical tips of the coiled-coils protruding from the DNA-borne ATPase head domains (pink circles). The status of the bound DNA sensed by RloC determines if its ATPase will be activated and drive structural changes needed to switch on the ACNase domains (split green ovals) toward tRNA cleavage.

#### **3.6 Is RloC a suicidal DNA-damage-responsive device?**

If RloC can be regulated by its internal devices, what role plays the anticipated interaction of RloC with a DNA R-M protein? Do these external and internal devices cooperate or act separately, responding to the same or different environmental cues? The present state of RloC's research does not permit us to distinguish between these possibilities, let alone assign to this protein specific biological functions. However, cues provided by Rad50/SbsC, the only other known coiled-coil/zinc-hook containing entity, may facilitate the formulation of

Fig. 8. *Gka*RloC's ATPase and tethered DNA cooperatively regulate its ACNase function. *Gka*RloC's ACNase activity was assayed using as a substrate a 5'-32P labeled anticodon-stemloop analog corresponding to mammalian tRNALys3 (ASL). The reaction was performed in the absence or presence of 2mM of ATP and/or 10ng/μl of BstE II digested λ DNA, or in the presence of the non-hydrolyzable ATP analog AMP-PNP. The 7mer is a radiolabeled

Fig. 9. RloC's anticipated DNA bridging activity. By analogy with Rad50, RloC bridges DNA through Zn++ (orange circles) coordinated at zinc-hook (ZH) dimerization interfaces (yellow circles) at the apical tips of the coiled-coils protruding from the DNA-borne ATPase head domains (pink circles). The status of the bound DNA sensed by RloC determines if its ATPase will be activated and drive structural changes needed to switch on the ACNase

If RloC can be regulated by its internal devices, what role plays the anticipated interaction of RloC with a DNA R-M protein? Do these external and internal devices cooperate or act separately, responding to the same or different environmental cues? The present state of RloC's research does not permit us to distinguish between these possibilities, let alone assign to this protein specific biological functions. However, cues provided by Rad50/SbsC, the only other known coiled-coil/zinc-hook containing entity, may facilitate the formulation of

fragment resulting from the final excision reaction

domains (split green ovals) toward tRNA cleavage.

**3.6 Is RloC a suicidal DNA-damage-responsive device?** 

useful guiding hypotheses. Here it will suffice to briefly summarize pertinent features of this universal DNA-damage-responsive, DNA-repair protein. For comprehensive coverage several recent reviews are suggested (Hirano, 2006;Stracker & Petrini, 2011;Williams *et al*, 2010;Paull, 2010) as well as relevant chapters in this book.

Archaeal Rad50 and the bacterial SbcC counterparts associate with the respective dimeric DNases Mre11 or SbcD. The eukaryal Rad50-Mre11 complex (MR) further associates with an adapter protein termed Nbs1 (Xrs2 in yeast), which links the ternary complex to key DNA damage checkpoints. The ternary MRN complex controls key sensing, signaling, regulating, and effecter responses triggered by DNA double-strand breaks (DSB). These responses include the activation of master regulators such as ATM as well as roles in homologous recombinational repair (HRR), microhomology-mediated end joining (MMEJ) and, occasionally, non-homologous end-joining (NHEJ). Rad50 figures in these transactions as a DNA-bridging SMC protein, using its coiled-coil/zinc-hook and ATPase to properly orient the DNA molecules it bridges and its associated protein partners (Hirano, 2005;Stracker & Petrini, 2011;Williams *et al*, 2010;Paull, 2010;Stracker & Petrini, 2011). As mentioned, Rad50's coiled-coils bend when the ATPase domains are free and stretches when tethered to DNA (van *et al*, 2003;Moreno-Herrero *et al*, 2005). This flexibility also allows the linked ATPase domain to communicate nucleotide binding and DNA ligand signals across distances and between components of the complex. These transmissions depend, among others, on the binding of Mre11 to the coiled-coil portion closest to the ATPase domain, which positions the DNase to resect DSB ends (Williams *et al*, 2011).

Rad50's bacterial homologue SbcC may likewise exert its function as a DNA bridging protein, directing SbcD to cleave hairpin structures that impede DNA replication and initiate DSB that drive HRR (Darmon *et al*, 2010;Storvik & Foster, 2011;White *et al*, 2008). Interestingly, over-expressed in *E. coli, SbcC* co-localizes with the replication factory whereas SbcD is dispersed throughout the cytoplasm. Their discrepant behaviors underlie the proposal that at its low, natural level SbcC constantly checks the replication fork for misfolded DNA, recruiting SbcD only when repair is required. A different distribution in *B. subtilis* suggests that in this organism SbcCD partakes also in NHEJ (Mascarenhas *et al*, 2006; Darmon *et al*, 2007).

*Gka*RloC could use its DNA bridging activity (Figs. 5, 6) to monitor the status of cellular DNA molecules like Rad50 and SbcC. However, there is no evidence that RloC associates with a DNase corresponding to Mre11 or SbcD. On the other hand, RloC's regulatory domain, Rad50/SbcC's counterpart is uniquely appended to the translation-disabling ACNase domain. It is tempting to speculate therefore that the ACNase C-domain interacts with the regulatory N-domain in a manner analogous to Mre11's, i.e., tethers to the proximal portion of the coiled-coil fiber emerging from the ATPase head-domain. Such a contact could help transduce DNA damage signals sensed by RloC's DNA monitoring device and relayed by the ATPase to the ACNase effecter domain. The existence of such a signal transduction pathway agrees with the effects of RloC's zinc-hook mutations, ATPase and tethered DNA on its ACNase function (Davidov & Kaufmann, 2008) (Fig 8).

The suggestions that RloC's ACNase is activated in response to DNA damage and, consequently, arrests translation may seem self-contradictory. After all, bacteria normally respond to DNA insults by enhancing the synthesis of DNA repair and other stress responsive proteins (Fernandez De Henestrosa *et al*, 2000). This apparent contradiction may be reconciled by considering the phenomenon of DNA restriction alleviation (RA) (Thoms & Wackernagel, 1984). RA is enacted in response to genotoxic stress as a protective measure

RloC: A Translation-Disabling tRNase Implicated in

perhaps by RloC.

**4.2 Cellular RNA nick repair systems** 

their cooperation could be rather widespread.

Phage Exclusion During Recovery from DNA Damage 35

of phage T4 on its tRNA repair proteins as a means to overcome the disruption of tRNALys by the host's ACNase PrrC (section 2.1). Another relevant example is the AlkB RNA demethylase of certain single stranded plant RNA viruses. The intended role of this demethylase is probably the removal of toxic methyl groups from the viral genomic RNA (van den *et al*, 2008). Homologous bacterial and human RNA-specific AlkB methylases could save the resources and/or time needed to re-synthesize damaged RNAs. In fact, these enzymes have been found able to resuscitate damaged RNA models while distinguishing between natural base modifications and toxic ones. However, the biological relevance of these findings remains uncertain (Aas *et al*, 2003;Ougland *et al*, 2004). Another DNA repair protein with possible roots in RNA metabolism is the abasic DNA endonuclease APE1 (Tell *et al*, 2010). Below we focus on recently discovered cellular tools able to repair nicked RNA, as do the phage T4-encoded proteins Pnk and Rnl1 that counteract PrrC and are frustrated

The RNA phosphodiester linkage is vulnerable to nucleophilic attack. Deprotonation of its adjacent 2' oxygen, subsequent formation of a pentameric phosphate intermediate and 5'-O protonation disrupt it, yielding 2', 3' cyclic phosphate and 5'-OH cleavage ends. This reaction occurs spontaneously and nonspecifically under physiological conditions but is also catalyzed at critical target sites by stress-responsive tRNases (Thompson & Parker, 2009) and secreted ribotoxins (Wool *et al*, 1992;Masaki & Ogawa, 2002;Lu *et al*, 2005;Jablonowski *et al*, 2006;Klassen *et al*, 2008). Some of the small self-cleaving ribozymes that catalyze it also catalyze the reverse reaction, converting 2',3'-cyclic-P and 5'-OH ends into a 3'-5' phosphodiester linkage (Ferre-D'Amare & Scott, 2010). A similar RNA ligase activity involved in tRNA splicing was detected early on in HeLa cell extracts (Filipowicz & Shatkin, 1983) and later in an archaeon (Gomes & Gupta, 1997). The protein catalyzing it termed RtcB has been recently identified in an archaeon, human cells and bacteria (Englert *et al*, 2011;Popow *et al*, 2011;Tanaka & Shuman, 2011). The archaeal and human proteins join 5' and 3' exons of tRNAs and the human possibly also those of the mRNA of an unfoldedprotein-response factor (Englert *et al*, 2011;Popow *et al*, 2011). A role for the bacterial RtcB has not been assigned yet. However, its possible participation in an RNA-nick-repair pathway is suggested by the operon RtcB shares with the RNA 3'-P cyclase RtcA. RtcA turns the 3'-P end into 2', 3'-P> through an adenylated intermediate, analogous to the manner in which RNA and DNA ligases activate 5'-P termini (Genschik *et al*, 1998). Thus, combined, RtcAB could convert a 3'-P and 5'-OH pair into a 3'-5' phosphodiester linkage. Unlike RtcA, the RtcB mediated transesterification reaction does not require an energy source although it may be allosterically directed by bound GTP (Tanaka & Shuman, 2011). Given their ability to repair such RNA nicks, RtcAB or RtcB alone could mend accidentally broken RNAs, restore RNAs temporarily inactivated by stress-responsive RNases (Neubauer *et al*, 2009;Zhang *et al*, 2005) or counteract ribotoxins secreted by rival cells (Masaki & Ogawa, 2002). Moreover, the existence of both RtcA and RtcB in all three domains of life (Tanaka & Shuman, 2011;Englert *et al*, 2011;Popow *et al*, 2011) suggests that

A more intricate RNA-nick-repair pathway is catalyzed by the bacterial proteins PnkP and Hen1. PnkP and Hen1 share the same operon and form a tetrameric P2H2 complex (Martins & Shuman, 2005;Chan *et al*, 2009b). The reactions catalyzed by the PnkP component of the complex resemble those mediated by phage T4 Pnk and Rnl1 (section 1) and the yeast and

intended to prevent degradation of self DNA. In the best documented RA case, the restriction subunit HsdR of the type Ia R-M protein EcoKI is degraded by the protease ClpXP (Makovets *et al*, 2004). In the case of the type Ic protein EcoR124I, RA may entail dissociation or functional occlusion of the HsdR subunit (Youell & Firman, 2008). RA prevents the degradation of fully unmodified portions of the cellular DNA synthesized during the recovery from DNA damage, mainly by HRR. In fact, exposure of an RAdeficient mutant to DNA damage causes DSB and eventual cell death (Cromie *et al*, 2001;Makovets *et al*, 2004;Blakely & Murray, 2006).

RA exacts also a price. Namely, inactivation of the cell's primary immune system renders it highly vulnerable to phage infection (Yamagami & Endo, 1969;Blakely & Murray, 2006). In theory, RloC could benefit its host in this situation by acting as an antiviral back-up device, mobilized when the cell is infected by a phage during recovery from DNA damage. The activation of RloC under these circumstances would prevent the spread of the phage to other members of the vulnerable bacterial population. In this regard RloC could resemble PrrC, which fails to rescue the cell in which it is turned on but can contain the infection. However, the proposed mode of RloC's activation calls for combined inputs of DNA damage and phage infection. Namely, phage infection alone would be offset by the functional DNA restriction nuclease while DNA damage alone would be effectively dealt with by the SOS response (Friedberg *et al*, 2006). It is noteworthy that exposure of an RloC encoding species to mytomycin C did not induce detectable ACNase activity (unpublished results).

Clearly, the above model raises more questions than it attempts to answer. For example, how does the anticipated RloC-Hsd interaction fit in this scheme? Do the genotoxic and viral stress signals cooperate or act separately? Can RloC frustrate phage encoded tRNA repair? To address these issues it will be necessary to employ experimental systems based on natural RloC-encoding hosts and cognate T4-like phages that activate RloC and encode a tRNA repair system.

#### **4. RNA damage repair**

#### **4.1 Why repair damaged RNA?**

The emergence of an RNA cleavage-ligation pathway in the wake of a host-parasite encounter (Fig. 1) brought to the fore the rather overlooked subject of RNA damage repair. RNA is susceptible to the same agents that threaten DNA. Radiation and chemicals that break the DNA backbone and modify its bases have similar effects on RNA and its precursors. RNA is also attacked by stress responsive RNases (Thompson & Parker, 2009) and various secreted ribotoxins (Wool *et al*, 1992;Masaki & Ogawa, 2002;Lu *et al*, 2005). What is more, its backbone is more sensitive to spontaneous hydrolysis than DNA's. Yet, the repair of damaged RNA seems necessary only in cases where its replenishment by resynthesis is not possible, e.g., when a DNA template to transcribe from is missing.

Thus, it is conceivable that RNA repair tools played a critical role in sustaining the genomes of the hypothetical RNA and RNA/Protein Worlds (Cech, 2009). One may further speculate that some of these tools could have evolved into extant devices with similar RNA repair tasks or expatiated roles in other RNA transactions (Abelson *et al*, 1998;Sidrauski *et al*, 1996) or even in DNA repair (Aas *et al*, 2003;Tell *et al*, 2010).

RNA repair can be the only option also in extant situations, especially when a DNA template to transcribe from is missing. A relevant example already given here is the reliance

intended to prevent degradation of self DNA. In the best documented RA case, the restriction subunit HsdR of the type Ia R-M protein EcoKI is degraded by the protease ClpXP (Makovets *et al*, 2004). In the case of the type Ic protein EcoR124I, RA may entail dissociation or functional occlusion of the HsdR subunit (Youell & Firman, 2008). RA prevents the degradation of fully unmodified portions of the cellular DNA synthesized during the recovery from DNA damage, mainly by HRR. In fact, exposure of an RAdeficient mutant to DNA damage causes DSB and eventual cell death (Cromie *et al*,

RA exacts also a price. Namely, inactivation of the cell's primary immune system renders it highly vulnerable to phage infection (Yamagami & Endo, 1969;Blakely & Murray, 2006). In theory, RloC could benefit its host in this situation by acting as an antiviral back-up device, mobilized when the cell is infected by a phage during recovery from DNA damage. The activation of RloC under these circumstances would prevent the spread of the phage to other members of the vulnerable bacterial population. In this regard RloC could resemble PrrC, which fails to rescue the cell in which it is turned on but can contain the infection. However, the proposed mode of RloC's activation calls for combined inputs of DNA damage and phage infection. Namely, phage infection alone would be offset by the functional DNA restriction nuclease while DNA damage alone would be effectively dealt with by the SOS response (Friedberg *et al*, 2006). It is noteworthy that exposure of an RloC encoding species to mytomycin C did not induce detectable ACNase activity (unpublished

Clearly, the above model raises more questions than it attempts to answer. For example, how does the anticipated RloC-Hsd interaction fit in this scheme? Do the genotoxic and viral stress signals cooperate or act separately? Can RloC frustrate phage encoded tRNA repair? To address these issues it will be necessary to employ experimental systems based on natural RloC-encoding hosts and cognate T4-like phages that activate RloC and encode a

The emergence of an RNA cleavage-ligation pathway in the wake of a host-parasite encounter (Fig. 1) brought to the fore the rather overlooked subject of RNA damage repair. RNA is susceptible to the same agents that threaten DNA. Radiation and chemicals that break the DNA backbone and modify its bases have similar effects on RNA and its precursors. RNA is also attacked by stress responsive RNases (Thompson & Parker, 2009) and various secreted ribotoxins (Wool *et al*, 1992;Masaki & Ogawa, 2002;Lu *et al*, 2005). What is more, its backbone is more sensitive to spontaneous hydrolysis than DNA's. Yet, the repair of damaged RNA seems necessary only in cases where its replenishment by re-

Thus, it is conceivable that RNA repair tools played a critical role in sustaining the genomes of the hypothetical RNA and RNA/Protein Worlds (Cech, 2009). One may further speculate that some of these tools could have evolved into extant devices with similar RNA repair tasks or expatiated roles in other RNA transactions (Abelson *et al*, 1998;Sidrauski *et al*, 1996)

RNA repair can be the only option also in extant situations, especially when a DNA template to transcribe from is missing. A relevant example already given here is the reliance

synthesis is not possible, e.g., when a DNA template to transcribe from is missing.

or even in DNA repair (Aas *et al*, 2003;Tell *et al*, 2010).

2001;Makovets *et al*, 2004;Blakely & Murray, 2006).

results).

tRNA repair system.

**4. RNA damage repair** 

**4.1 Why repair damaged RNA?** 

of phage T4 on its tRNA repair proteins as a means to overcome the disruption of tRNALys by the host's ACNase PrrC (section 2.1). Another relevant example is the AlkB RNA demethylase of certain single stranded plant RNA viruses. The intended role of this demethylase is probably the removal of toxic methyl groups from the viral genomic RNA (van den *et al*, 2008). Homologous bacterial and human RNA-specific AlkB methylases could save the resources and/or time needed to re-synthesize damaged RNAs. In fact, these enzymes have been found able to resuscitate damaged RNA models while distinguishing between natural base modifications and toxic ones. However, the biological relevance of these findings remains uncertain (Aas *et al*, 2003;Ougland *et al*, 2004). Another DNA repair protein with possible roots in RNA metabolism is the abasic DNA endonuclease APE1 (Tell *et al*, 2010). Below we focus on recently discovered cellular tools able to repair nicked RNA, as do the phage T4-encoded proteins Pnk and Rnl1 that counteract PrrC and are frustrated perhaps by RloC.

#### **4.2 Cellular RNA nick repair systems**

The RNA phosphodiester linkage is vulnerable to nucleophilic attack. Deprotonation of its adjacent 2' oxygen, subsequent formation of a pentameric phosphate intermediate and 5'-O protonation disrupt it, yielding 2', 3' cyclic phosphate and 5'-OH cleavage ends. This reaction occurs spontaneously and nonspecifically under physiological conditions but is also catalyzed at critical target sites by stress-responsive tRNases (Thompson & Parker, 2009) and secreted ribotoxins (Wool *et al*, 1992;Masaki & Ogawa, 2002;Lu *et al*, 2005;Jablonowski *et al*, 2006;Klassen *et al*, 2008). Some of the small self-cleaving ribozymes that catalyze it also catalyze the reverse reaction, converting 2',3'-cyclic-P and 5'-OH ends into a 3'-5' phosphodiester linkage (Ferre-D'Amare & Scott, 2010). A similar RNA ligase activity involved in tRNA splicing was detected early on in HeLa cell extracts (Filipowicz & Shatkin, 1983) and later in an archaeon (Gomes & Gupta, 1997). The protein catalyzing it termed RtcB has been recently identified in an archaeon, human cells and bacteria (Englert *et al*, 2011;Popow *et al*, 2011;Tanaka & Shuman, 2011). The archaeal and human proteins join 5' and 3' exons of tRNAs and the human possibly also those of the mRNA of an unfoldedprotein-response factor (Englert *et al*, 2011;Popow *et al*, 2011). A role for the bacterial RtcB has not been assigned yet. However, its possible participation in an RNA-nick-repair pathway is suggested by the operon RtcB shares with the RNA 3'-P cyclase RtcA. RtcA turns the 3'-P end into 2', 3'-P> through an adenylated intermediate, analogous to the manner in which RNA and DNA ligases activate 5'-P termini (Genschik *et al*, 1998). Thus, combined, RtcAB could convert a 3'-P and 5'-OH pair into a 3'-5' phosphodiester linkage. Unlike RtcA, the RtcB mediated transesterification reaction does not require an energy source although it may be allosterically directed by bound GTP (Tanaka & Shuman, 2011). Given their ability to repair such RNA nicks, RtcAB or RtcB alone could mend accidentally broken RNAs, restore RNAs temporarily inactivated by stress-responsive RNases (Neubauer *et al*, 2009;Zhang *et al*, 2005) or counteract ribotoxins secreted by rival cells (Masaki & Ogawa, 2002). Moreover, the existence of both RtcA and RtcB in all three domains of life (Tanaka & Shuman, 2011;Englert *et al*, 2011;Popow *et al*, 2011) suggests that their cooperation could be rather widespread.

A more intricate RNA-nick-repair pathway is catalyzed by the bacterial proteins PnkP and Hen1. PnkP and Hen1 share the same operon and form a tetrameric P2H2 complex (Martins & Shuman, 2005;Chan *et al*, 2009b). The reactions catalyzed by the PnkP component of the complex resemble those mediated by phage T4 Pnk and Rnl1 (section 1) and the yeast and

RloC: A Translation-Disabling tRNase Implicated in

therapeutic target (Weinfeld *et al*, 2011).

~3-fold more frequent occurrence among bacteria.

tRNA repair enzymes.

**5. Conclusions** 

Phage Exclusion During Recovery from DNA Damage 37

endowed with an N-terminal FHA (Fork Head Associated) phosphopeptide binding domain that links PNKP to the scaffold proteins XRCC1 and XRCC4 (Bernstein *et al*, 2009). The latter recruit PNKP to exercise its functions in base excision repair (Hegde *et al*, 2008) or NHEJ (Lieber, 2008). PNKP's essential role in these ssDNA and DSB repair pathways is to convert 3'-P and 5'-OH DNA termini into 3'-OH and 5'-P pairs that are ligatable or fit for gap-filling by a DNA polymerase. A wide DNA binding cleft accounts for the ability of this protein to prefer nicked duplexes and recessed 5'-termini over ssDNA substrates and distinguishes it from the RNA end healing phage counterpart. The 3'-P and 5'-OH DNA termini are caused by ionizing radiation, genotoxic chemicals and enzymatic reactions. Specific examples include excision of abasic sites (Hazra *et al*, 2002), DSB generated by DNase II (Evans & Aguilera, 2003) and release of camptothecin- trapped topoisomerase I-DNA adducts by a tyrosine-DNA specific phosphodiesterase (Pouliot *et al*, 1999). Failure to repair such lesions underlies several inborn neural disorders. Conversely, PNKP can render cancer cells resistant to certain genotoxic drugs and, therefore, is considered itself a potential

In this chapter we addressed the possible biological role of the conserved bacterial anticodon nuclease RloC that combines two seemingly conflicting properties. One, predicted by resemblance of its regulatory region to the universal DNA-damage-checkpoint/DNA repair protein Rad50/SbcC is monitoring DNA insults. The second, predicted by its tRNase activity is disabling the translation apparatus. The co-existence of such functions in the same molecule and the regulation of one by the other suggests that RloC is designed to block translation in response to DNA damage. Such a response is suicidal since it prevents recovery from DNA damage. Hence, it must be executed only under special circumstances where cell death is advantageous. One possibility considered here is that RloC benefits its host cell by acting as an antiviral contingency during recovery from DNA damage. Under these conditions bacterial cells may shut off their primary antiviral defense, i.e., their DNA restriction activity. RloC's suicidal activity would not rescue the infected cell but would prevent the spread of the infection to other vulnerable members of the population recovering from DNA damage. Another unique property, which could make RloC particularly suited to thwart phage infection, is the ability of this ACNase to excise its substrate's wobble nucleotide. In this regard RloC differs from its distant homologue the ACNase PrrC, which only incises its tRNA substrate and is counteracted by phage tRNA repair enzymes. Therefore, it seems conceivable that the harsher lesion inflicted by RloC will encumber such phage reversal. The possibility that RloC is a more efficient antiviral device than PrrC is also hinted at by its

While these notions are supported by some demonstrated properties of RloC, testing them and identifying RloC's true call requires studying this protein under physiological conditions; ideally, using a natural host encoding it and cognate phages endowed with

The RNA repair pathway instigated by PrrC and possibly avoided by RloC brings to the fore the rather overlooked issue of RNA-damage-repair. Such repair would seem necessary only under circumstances such as the absence of a DNA template to transcribe from. Nonetheless, recent discoveries of various cellular RNA repair devices distributed in the three domains of life suggest that RNA damage repair is more prevalent, exercised perhaps also during

plant tRNA splicing ligase (Abelson *et al*, 1998). What makes this repair system unique is its ability to render the restored phosphodiester linkage immune to re-cleavage by virtue of the 2'-O methylase activity of Hen1 (Chan *et al*, 2009b). PnkP comprises an N-terminal kinase domain, a central metallophosphoesterase domain and a C-terminal ligase domain. Thus, it comprises functions similar to those of the yeast tRNA splicing ligase but differs in domain order and different origin of the phosphoesterase domain (Apostol *et al*, 1991;Martins & Shuman, 2005). Interestingly, by itself the bacterial PnkP heals 2', 3'-cyclic P and 5'-OH termini pairs and undergoes the first step in the RNA ligase reaction, its auto-adenylation, but does not proceed to activate the 5'-P end and generate the phosphodiester linkage (Martins & Shuman, 2005). This deficiency is corrected by expressing PnkP with the 2'-O methylase Hen1. Within the resultant PnkP/Hen1 complex PnkP heals and seals the cleavage termini while Hen1 2'-O methylates the dephosphorylated 3'-end prior to the ligation step. This modification renders the restored ligation junction immune to recleavage (Chan *et al*, 2009b). The bacterial Hen1 is so named because it resembles in sequence and structure the methylase domain of eukaryal miRNA methyltransferase Hen1 (Chan *et al*, 2009a). The eukaryal Hen1 protects the 3'-terminal ribose of miRNA from exonucleolytic degradation or utilization as replication primer (Chen, 2005).

As with bacterial RtcAB, the biological role of the PnkP/Hen1 is not known. Noteworthy in this regard is that PnkP/Hen1 is most abundant among *Actinobacteria.* In contrast, RtcAB is more prevalent among *Proteobacteria* and has not been detected yet in *Actinobacteria*. This coincidence raises the possibility that the two systems provide similar benefits to their respective hosts. In theory, PnkP/Hen1 complexes could defend their host cells from secreted ribotoxins more efficiently than RtcAB due to the ability to prevent re-cleavage of the susceptible RNA. It is noteworthy though that colicin-like ribotoxins that target rRNA (Bowman *et al*, 1971;Senior & Holland, 1971) or tRNA anticodon loops (Masaki & Ogawa, 2002) have not been identified yet in bacteria likely to accommodate PnkP/Hen1.

If PnkP/Hen1 were to counteract an ACNase that cleaves its substrate 3' to the wobble base like colicin E5 (Ogawa *et al*, 1999), then the repaired tRNA would contain a 2'-O methylated wobble nucleotide. Such a protective modification need not impair the tRNA's function since it exists in some natural bacterial tRNAs (Juhling *et al*, 2009). However, it cannot be excluded that PnkP/Hen1 plays additional or other roles and may be exploited differently in different bacterial hosts. One example of such a different role is hinted at by the juxtaposition of the PnkP/Hen1 and CRISPR-Cas loci of *Microscilla marina*. The CRISPR-Cas system confers adaptive immunity against foreign nucleic acids. During its antiviral interference activity specific RNA portions of the CRISPR transcript are used to target a Cas protein to cleave the invasive nucleic acid (Deveau *et al*, 2010). Hence, it may be asked if *M. marina* PnkP/Hen1 catalyze some RNA processing and/or modification steps during CRISPR RNA maturation. Finally, in a reversal of roles, one could envisage PnkP/Hen1 encoding phage able to prevent re-cleavage of a tRNA by the ACNase they counteract.

#### **4.3 An essential eukaryal DNA repair protein is related to T4 Pnk**

There are a number of examples of DNA repair devices that could have originated from RNA-specific progenitors, some of them already alluded to above. Here it will suffice to describe just one of them, related to the phage T4-encoded end healing protein Pnk. This conserved eukaryal protein termed interchangeably PNKP and Pnk1 contains 5'-kinase and 3'-phosphatase domains resembling those of T4 Pnk but arranged in the reverse order, the phosphoesterase domain preceding the kinase domain. The mammalian PNKP is also endowed with an N-terminal FHA (Fork Head Associated) phosphopeptide binding domain that links PNKP to the scaffold proteins XRCC1 and XRCC4 (Bernstein *et al*, 2009). The latter recruit PNKP to exercise its functions in base excision repair (Hegde *et al*, 2008) or NHEJ (Lieber, 2008). PNKP's essential role in these ssDNA and DSB repair pathways is to convert 3'-P and 5'-OH DNA termini into 3'-OH and 5'-P pairs that are ligatable or fit for gap-filling by a DNA polymerase. A wide DNA binding cleft accounts for the ability of this protein to prefer nicked duplexes and recessed 5'-termini over ssDNA substrates and distinguishes it from the RNA end healing phage counterpart. The 3'-P and 5'-OH DNA termini are caused by ionizing radiation, genotoxic chemicals and enzymatic reactions. Specific examples include excision of abasic sites (Hazra *et al*, 2002), DSB generated by DNase II (Evans & Aguilera, 2003) and release of camptothecin- trapped topoisomerase I-DNA adducts by a tyrosine-DNA specific phosphodiesterase (Pouliot *et al*, 1999). Failure to repair such lesions underlies several inborn neural disorders. Conversely, PNKP can render cancer cells resistant to certain genotoxic drugs and, therefore, is considered itself a potential therapeutic target (Weinfeld *et al*, 2011).
