The exoribonuclease Polynucleotide Phosphorylase influences the virulence and stress responses of yersiniae and many other pathogens

Microbes are incessantly challenged by both biotic and abiotic stressors threatening their existence. Therefore, bacterial pathogens must possess mechanisms to successfully subvert host immune defenses as well as overcome the stress associated with host-cell encounters. To achieve this, bacterial pathogens typically experience a genetic re-programming whereby anti-host/stress factors become expressed and eventually translated into effector proteins. In that vein, the bacterial host-cell induced stress-response is similar to any other abiotic stress to which bacteria respond by up-regulating specific stress-responsive genes. Following the stress encounter, bacteria must degrade unnecessary stress responsive transcripts through RNA decay mechanisms. The three pathogenic yersiniae (Yersinia pestis, Y. pseudo-tuberculosis, and Y. enterocolitica) are all psychrotropic bacteria capable of growth at 4°C; however, cold growth is dependent on the presence of an exoribonuclease, polynucleotide phosphorylase (PNPase). PNPase has also been implicated as a virulence factor in several notable pathogens including the salmonellae, Helicobacter pylori, and the yersiniae [where it typically influences the type three secretion system (TTSS)]. Further, PNPase has been shown to associate with ribonuclease E (endoribonuclease), RhlB (RNA helicase), and enolase (glycolytic enzyme) in several Gram-negative bacteria forming a large, multi-protein complex known as the RNA degradosome. This review will highlight studies demonstrating the influence of PNPase on the virulence potentials and stress responses of various bacterial pathogens as well as focusing on the degradosome-dependent and -independent roles played by PNPase in yersiniae stress responses.

Despite physical evidence of degradosome assembly observed in a number of organisms, the precise physiological functions carried out by these large multiprotein complexes/hyperstructures have yet to be fully elucidated (Norris et al., 2012). Using bacterial two-hybrid analysis as well as co-immunoprecipitation techniques, we recently determined that a Yersinia pseudotuberculosis degradosome, comprised of RNase E, PNPase, and RhlB helicase associates (Henry et al., 2012), extending an earlier report of physical interaction between only PNPase and RNase E in the aforementioned organism (Yang et al., 2008).

PNPase
Originally identified in E. coli, PNPase, encoded by the pnp gene, is an ∼80 kDa homo-trimeric exoribonuclease that poorly degrades structured RNA molecules (Guarneros and Portier, 1991;Symmons et al., 2000). The Yersinia pestis bv. Antique's highly conserved PNPase (UniProt Entry: Q1C3L8) is a slightly smaller enzyme (∼78 kDa) but shares 89% sequence similarity with that of its E. coli counterpart. Additionally, PNPase acts as a polymerase when the concentration of nucleotide di-phosphates is greater than inorganic phosphate, as shown in E. coli (Deutscher and Li, 2001). Historically, PNPase played a large part in determining the genetic code (Grunberg-Manago et al., 1955), winning Dr. Severo Ochoa the 1959 Nobel Prize in Physiology or Medicine. As mentioned above, PNPase is also a degradosome constituent in E. coli and Y. pseudotuberculosis, but exactly what percentage of PNPase's degradative effort is spent while associated with the degradosome has yet to be determined (Lawal et al., 2011). To date, the synthetic/polymerase activity of PNPase has not been demonstrated in the yersiniae. PNPase consists of two catalytic domains as well as both structurally conserved KH and S1 RNA-binding domains in its extreme carboxy-terminus.

PNPase, NON-CODING sRNAs, AND Hfq (RNA-BINDING PROTEIN)
With the discovery of functional, small non-coding RNA molecules (sRNAs) involved in regulating gene expression, the entire paradigm associated with "what constitutes a gene" was shifted. sRNAs became increasingly studied when they were found to influence bacterial physiology and even virulence (Gottesman et al., 2001;Shimoni et al., 2007;Repoila and Darfeuille, 2009;Silva et al., 2011;Storz et al., 2011). More specifically, in E. coli, PNPase was found to stabilize several regulatory sRNAs including: RyhB, SgrS, and CyaR. Further, RNase E appeared responsible for their increased turnover/decay in the absence of PNPase (De Lay and Gottessman, 2011). These findings strongly suggest that the degradosome assembly (which includes the presence of PNPase) is required for stabilizing some regulatory sRNAs in E. coli. Perhaps the degradosome is protecting virulence-regulating sRNAs in bacterial pathogens as well?
In Y. pseudotuberculosis, 150 sRNA species (Koo et al., 2011) have been identified, many of which are involved in Yersinia virulence (Koo and Lathem, 2012;Lathem, 2012). Upon proteome profiling, the SraG sRNA was found to regulate expression of 16 different genes as evidenced by differential protein expression in Y. pseudotuberculosis. Interestingly, SraG appeared to positively regulate levels of PNPase, one of the 16 differentially expressed proteins (Lu et al., 2012). These powerful sRNAs used for modulating gene expression complement other post-transcriptional regulating factors including: ribonucleases (e.g., PNPase), RNAbinding proteins, and thermal riboswitches (e.g., the low calcium response at 37 • C) (Schiano and Lathem, 2012).
In many instances, Hfq (an RNA-binding protein) will bind to sRNAs either promoting their stabilization or degradation (Silva et al., 2011). In the case of Yersinia pestis, 31 unique sRNAs were discovered, and the majority of those required the presence of Hfq for their expression (Beauregard et al., 2013). A separate global analysis of sRNAs revealed as many as 43 species in Y. pestis (Qu et al., 2012). Further, of the two Y. pestis iron-responsive sRNA RyhB homologs, only RhyB1 required Hfq for stabilization while RyhB2 s stability was Hfq-independent . Further, PNPase has been shown to play a crucial role in the degradation of sRNAs that do not associate with Hfq .
However, in E. coli, Hfq was shown to be necessary for PNPasemediated degradation of the Cold shock Protein A encoding transcript (cspA) during cold shock (Hankins et al., 2010). Hfq also influences the virulence of a number of pathogenic bacteria (Chao and Vogel, 2010) including Y. pestis and Y. pseudotuberculosis (Geng et al., 2009;Schiano et al., 2010), while also influencing the growth of the latter (Bai et al., 2010). Perhaps Hfq is the link between yersiniae PNPase, sRNAs, and bacterial virulence?

PNPase AND COLD SHOCK/COLD GROWTH
Bacterial responses to low temperature stress are generally referred to as "cold shock." PNPase confers cold growth capabilities to a number of bacteria (including several pathogens) including: E. coli, the yersiniae, Salmonella enterica, Campylobacter jejuni, S. aureus, and B. subtilis (Jones et al., 1987;Wang and Bechhofer, 1996;Goverde et al., 1998;Clements et al., 2002;Rosenzweig et al., 2005Rosenzweig et al., , 2007Anderson and Dunman, 2009;Haddad et al., 2009Haddad et al., , 2012. Considering that C. jejuni and several pathogenic yersiniae are contracted via the fecal oral route, PNPase could serve to promote bacterial persistence on processed meats that are shipped at refrigerated temperatures (thereby facilitating foodborne illness).
During the cold shock response in E. coli, cold inducible genes, including six cold shock protein genes (csp's), become upregulated (Jones et al., 1987;Goldstein et al., 1990). Additionally, cold-inducible pnp and rnr genes, encoding PNPase and RNase R respectively, were also found to be cold-inducible in E. coli (Jones et al., 1987;Cairrão et al., 2003). Interestingly, RNase R was required for specifically degrading ompA transcript (encoding outer membrane protein A-OmpA) during stationary phase growth of E. coli (Andrade et al., 2006), strongly suggesting that RNase R is not only cold-inducible but also growth-phase inducible.
During acclimation, immediately following cold-shock, PNPase is required for degradation of unnecessary csp (that likely trap ribosomes) and other cold-inducible transcripts in both E. coli and Yersinia enterocolitica (Goverde et al., 1998;Neuhaus et al., 2000;Yamanaka and Inouye, 2001;Polissi et al., 2003). More specifically, in Y. enterocolitica PNPase, together with RNase E, influences cold growth adaptation through its recognizing and cleaving AGUAA motifs (termed cold-shock-cut boxes) in csp transcripts (Neuhaus et al., 2003). Furthermore, the ability of PNPase to promote cold growth in the yersiniae absolutely required the catalytic activity of PNPase and to a lesser extent its RNA-binding domains, the S1 RNA-binding domain more than the KH RNA-binding domain (Rosenzweig et al., 2005). Surprisingly, we found that PNPase was not required for chemical stress responses associated with antibiotic exposures that target protein translational machinery, the cellular membrane, and the yersiniae cell wall (Henry et al., 2012), indicating that PNPase might not be a global stress regulator but rather a regulator of specific encountered stresses in yersiniae, like the cold temperature stress.

PNPase AND BACTERIAL VIRULENCE
PNPase influences the virulence of number of human pathogens as well as one animal pathogen. In the Gram-negative Dichelobacter nodosus (the etiological agent of foot root in sheep), PNPase was found to negatively regulate virulence/twitching motility, primarily through its S1 RNA-binding domain (Palanisamy et al., 2010). Whether animal or human pathogens, when Gram-negative pathogens, including the yersiniae and salmonellae, are confronted by threatening innate immune cells, a rapid genetic re-programming occurs resulting in anti-host type three secretion system (TTSS) genes becoming upregulated (Cornelis and Van Gijsegem, 2000;Cornelis, 2002;Viboud and Bliska, 2005). It has now been firmly established that a number of ribonucleases (e.g., PNPase, RNase R, RNase E, RNase III) influence bacterial virulence in a wide range of Gram-negative and -positive pathogens (Anderson and Dunman, 2009;Lawal et al., 2011;Jester et al., 2012;Matos et al., 2012). Specifically, PNPase was required for the optimal virulence of several bacterial human pathogens (Table 1), in several instances by targeting TTSS targets.
In S. enterica, PNPase is required for optimal virulence in both mouse and cell culture infection models (the latter causing acute infections). However, in a seemingly counter-intuitive observation, PNPase works to negatively regulate macrophage invasion and expression of type TTSS genes involved in macrophage invasion (Clements et al., 2002). Additionally, PNPase works to also negatively regulate Salmonella plasmid virulence genes (spv) transcript levels by acting through the spv activator/regulator spvR, thereby reducing intracellular survival of S. enterica (Ygberg et al., 2006). In Salmonella, PNPase might be acting as a molecular switch, whereby a persistent infection is caused unless active PNPase promotes acute infections. As was the case for the cold shock response, following host immune cell stress, the yersiniae must resume pre-stress gene expression profiles through the removal of unwanted and unnecessary TSSS-associated transcripts. Is PNPase, and, by extension, the degradosome involved in this post-TTSS acclimation following the host-cell encounter. This question becomes even more interesting when considering that both the RNA degradosome and the T3SS are large multi-protein hyperstructure complexes that could be interacting (Norris et al., 2012). Of the 11 known Gram-negative Yersinia spp., only Y. enterocolitica, Y. pseudotuberculosis, and Y. pestis are human pathogens. While fecal-oral transmission of Y. pseudotuberculosis and Y. enterocolitica causes self-limiting gastroenteritis (Galindo et al., 2011), Y. pestis (transmitted by the bite of an infected flea) has caused three major human pandemics as well as the great plagues of London in the 1660s (Inglesby et al., 2000). In both Y. pestis and Y. pseudotuberculosis, PNPase was required for optimal virulence in cell culture infections (as measured by bacterial proliferation vis a vis mouse macrophage-like cells and degree of induced HeLa cell rounding, an indirect measure of TTSS function) (Rosenzweig et al., 2005) as well as in murine models of infection (Rosenzweig et al., 2007;Lawal et al., 2011). Interestingly, there was no evidence of PNPase reducing TTSS promoter activity in Y. pseudotuberculosis wild type and pnp deletion mutants that harbored a yopE::gfp (Rosenzweig et al., 2005). In contrast, PNPase was required for robust Yop D and E secretion during an early kinetic TTSS induction window (between 0 and 20 min), but not for optimal Yop production levels (Rosenzweig et al., 2005). Counter-intuitively, PNPase promotes robust Yop secretion during initial TTSS induction despite negatively regulating T3SS transcript and protein production levels (Rosenzweig et al., 2007).
Interestingly, different PNPase determinants were required for various yersiniae stress responses ( Table 2). Whereas PNPase's catalytic activity was dispensable for inducing HeLa cell rounding, it was absolutely required during cold stress (Rosenzweig et al., 2005). However, it was PNPase's S1 RNA-binding domain that proved necessary for inducing optimal HeLa cell rounding. In fact, S1 RNA-binding domains (derived from exoribonucleases PNPase, RNase II, and RNase R) alone were sufficient in complementing the yersiniae pnp deletion mutants when expressed in trans (Rosenzweig et al., 2005). Furthermore, the minimal S1 domain region required for pnp complementation in TTSS function and HeLa cell cytotoxicity was later determined to be amino acid residues 65-125 (Rosenzweig et al., 2007).
Within the required 65-125 amino acid S1 domain region, we were able to identify the specific S1 domain residue (F638), which proved necessary for not just TTSS function (as measured by HeLa cell rounding) but also cold growth (Rosenzweig et al., 2007). The yersiniae PNPase F638 is highly conserved and present not just in the S1 domain of E. coli PNPase but also in the S1 domain of E. coli RNase E at reside F57 (Diwa et al., 2002). Further, a co-crystal structural determination revealed that within the catalytic core of E. coli's RNase E that includes its S1 domain, the F57 residue (conserved with PNPase's F638) might play a facilitating role in binding RNA while residues F67 and K112 actually contact the RNA molecule (Callaghan et al., 2005). Since our previous studies revealed that both the cold-stress response as well as the HCISR require the S1 RNA-binding domain of PNPase (Rosenzweig et al., 2005(Rosenzweig et al., , 2007Henry et al., 2012), perhaps an as of yet unidentified upstream regulator (or novel factor) will converge on the PNPase S1 domain's RNA-binding property?
When considering that PNPase confers cold growth capabilities to a number of bacterial pathogens including Y. pestis, Y. enterocolitica, Y. pseudotuberculosis, S. enterica, C. jejuni, and S. aureus (Goverde et al., 1998;Clements et al., 2002;Rosenzweig et al., 2005Rosenzweig et al., , 2007Anderson and Dunman, 2009;Haddad et al., 2009Haddad et al., , 2012, PNPase acts as a "virulence empowering factor" by enhancing the pathogens' survival during adverse/hostile environmental conditions. In another striking example of this, PNPase becomes upregulated ∼4-fold in the dental pathogen, Streptococcus mutans during acid-shock/encounters with low pH (Len et al., 2004). Perhaps, in the same way PNPase serves as a "virulence empowering factor" during low temperature exposure, it may be enhancing the survival of an oral pathogen when local environments on tooth surfaces become acidic during decay? In another Gram-positive pathogen, PNPase appeared to play a role in S. pyogenes sag A (a virulence transcript encoding a streptolysin) and sda (encoding an anti-neutrophil DNase) degradation during late-exponential growth phase (Barnett et al., 2007). Taken together, PNPase appears to influence the virulence potential and/or survival of a broad range of pathogens, including the pathogenic yersiniae. Catalytic activity plays an essential role in cold growth Rosenzweig et al., 2005 ? Required for virulence in a catalytic-activity-independent manner Rosenzweig et al., 2005 ? S1 domain sufficient for mediating virulence Rosenzweig et al., 2005 ?
The first 15 aa residues of the S1 domain are dispensable for virulence; however, the first 30 aa residues are required for virulence Rosenzweig et al., 2007 ? Amino acid residue F638 within the PNPase S1 domain is required for virulence and cold growth Rosenzweig et al., 2007 Yersinia pseudotuberculosis No S1 and KH domains also contribute to cold growth. Rosenzweig et al., 2005 No Degradosome-independent role in cold growth Henry et al., 2012 Yes Degradosome assembly required for virulence Yang et al., 2008 Yes

DO RIBONUCLEASES OTHER THAN PNPase CONTRIBUTE TO BACTERIAL STRESS RESPONSES?
PNPase is not the sole ribonuclease involved in bacterial stress responses. As mentioned earlier, RNase R was shown to be upregulated in E. coli as well as being necessary for optimal cold growth (Cairrão et al., 2003). However, in Y. pestis neither RNase R nor RNase II was necessary for cold growth in the yersiniae (Lawal et al., 2013). That does not, however, necessarily mean that RNase R and/or RNase II are not also up-regulated in the yersiniae during cold stress, something that has yet to be determined. Interestingly, RNase R has been shown to play a role in the HCISR of S. flexnari (Tobe et al., 1992), Aeromonas hydrophila (Erova et al., 2008), and H. pylori (Tsao et al., 2009). In contrast, RNase R did not appear necessary for Y. pestis virulence (Rosenzweig et al., 2005). With regards to the yersiniae, it seems that PNPase specifically influences cold (Rosenzweig et al., 2005(Rosenzweig et al., , 2007Henry et al., 2012), oxidative (Henry et al., 2012), and host-cell induced stress responses (Rosenzweig et al., 2005(Rosenzweig et al., , 2007.

DEGRADOSOME DEPENDENCE FOR YERSINIAE VIRULENCE AND DURING OXIDATIVE STRESS BUT NOT DURING COLD STRESS
Using bacterial two-hybrid analysis as well as coimmunoprecipitation techniques, the Y. pseudotuberculosis degradosome was determined to contain RNase E, PNPase, and RhlB helicase (Henry et al., 2012), extending an earlier report of physical interaction between only PNPase and RNase E in the aforementioned organism (Yang et al., 2008). To create a degradosome assembly-incompetent Y. pseudotuberculosis strain, a dominant-negative, truncated RNase E variant (residues 1-465), containing only its amino terminal catalytic and S1 domains, was ectopically expressed (Briegel et al., 2006;Yang et al., 2008). RNase E autoregulates itself post-transcriptionally; however, when the aforementioned dominant-negative variant was expressed, its autoreagulatory activity was inhibited (Briegel et al., 2006;Yang et al., 2008). Since the carboxy-terminal degradosome scaffolding region was absent when the above mentioned dominant-negative variant was expressed, the yersiniae degradosome could not assemble, as was seen in Y. pseudotuberculosis (Yang et al., 2008).
To determine whether degradosome assembly was required for yersiniae virulence, both a proliferation assay in co-culture with macrophage-like cells as well as a TTSS functional evaluation were carried out. Interestingly, Y. pseudotuberculosis proliferation was significantly reduced when the degradosome was unable to assemble (Yang et al., 2008). In fact, the degree of reduced proliferation mirrored that of the yersiniae pnp deletion mutants observed earlier (Rosenzweig et al., 2005). Furthermore, Yang et al. (2008) were able to demonstrate that degradosome assembly was also necessary for optimum TTSS function, as measured by YopE secretion. Again, this diminished TTSS function resembled that of the yersiniae pnp deletion mutants (Rosenzweig et al., 2005(Rosenzweig et al., , 2007. Taken together, these data collectively demonstrated that the yersiniae requirement for PNPase to achieve optimal virulence is degradosome-dependent. Was the PNPase contribution to yersiniae cold-growth degradosome-dependent or -independent, and did PNPase contribute to the oxidative stress response? To address this, the dominant-negative, truncated RNase E variant, residues 1-465 (Yang et al., 2008), was again employed. Interestingly, the Y. pseudotuberculosis PNPase confers cold-growth capability in a degradosome-independent manner while PNPase responds to oxidative stress, in the form of H 2 O 2 exposure, in a degradosomedependent manner (Henry et al., 2012). The yersiniae degradosome dependence for optimal virulence (Yang et al., 2008) does, however, logically align itself with our finding that the Yersinia oxidative stress response also requires PNPase in a degradosomedependent manner (Henry et al., 2012) since, during the host-cell stress, the yersiniae are likely to encounter host-cell generated reactive oxygen species.
In sharp contrast, the PNPase requirement of E. coli during oxidative stress was degradosome-independent since an RNase E truncated variant devoid of the degradosome scaffolding region did not demonstrate compromised growth when exposed to H 2 O 2 (Wu et al., 2009). In E. coli, PNPase independently prevented the accumulation of oxidatively damaged RNA (Wu et al., 2009). Curiously, however, PNPase, itself, was required for the optimal oxidative stress response in both E. coli and Y. pseudotuberculosis (Wu et al., 2009;Henry et al., 2012). Taken together, these data clearly demonstrate that PNPase contributes to various stress responses differently, at times by associating with the degradosome hyperstructure (as in the case of the yersiniae oxidative and host-cell stress response).

CONCLUSION
Throughout the literature, many examples of how ribonucleases impact bacterial virulence are apparent (Lawal et al., 2011;Matos et al., 2012;Jester et al., 2012). Whereas RNase R and RNase II have been evaluated for their impacts on bacterial virulence and various stress responses in numerous organisms, detailed studies of the aforementioned ribonucleases in the yersiniae have yet to be carried out. However, PNPase has been well-characterized in the yersiniae and found to be essential for cold growth (Goverde et al., 1998;Rosenzweig et al., 2005), required for optimal T3SS activity (Rosenzweig et al., 2005(Rosenzweig et al., , 2007, oxidative stress responses (Henry et al., 2012), and yersiniae virulence (Rosenzweig et al., 2005). In fact, the yersiniae are not alone when it comes to dependence on PNPase for optimal virulence. S. enterica and D. nodosus also require PNPase for optimal virulence, while PNPase works to enhance survival of C. jejuni, S. aureus, S. pyogenes, and S. mutans during unfavorable conditions ( Table 1). With regards to PNPase's contributions to yersiniae virulence, degradosome assembly was required, indicating that PNPase likely mediates its effect through the larger degradosome hyperstructure. Furthermore, this implicated RNase E, the chief endoribonuclease and scaffold upon which the degradosome assembles, as a virulence-associated factor (Yang et al., 2008).
Despite our furthering the aforementioned study, to determine that yersiniae degradosome assembly was also necessary for oxidative stress responses but not cold stress responses (Henry et al., 2012), still much remains unknown, specifically about the role of other ribonucleases during yersiniae stress responses. Furthermore, the precise mechanism by which PNPase (either in a degradosome-dependent or -independent manner) mediates the specific stress responses mentioned earlier also remains Frontiers in Cellular and Infection Microbiology www.frontiersin.org November 2013 | Volume 3 | Article 81 | 5 unclear. For example, whether PNPase is degrading a repressor/negative regulator or binding to (through its S1 RNA-binding domain) and protecting an activator/positive regulator remains unknown during the various stress responses. Although progress has been made on understanding the roles the ribonucleases and RNA metabolism during various bacterial stress responses, a broader survey of additional ribonucleases should be carried out in the yersiniae, possibly revealing novel chemotherapeutic targets and/or novel vaccine candidate strains.