Resuscitation-promoting factor (Rpf) is a muralytic enzyme that increases the culturability of dormant bacteria. Recently, considerable progress has been made in understanding the structure, function and physiological role of Rpfs in different organisms, most notably the major human pathogen, Mycobacterium tuberculosis, which encodes multiple rpf-like genes. A key unresolved question, however, concerns the relationship between the predicted biochemical activity of Rpfs — cleavage of the β-1,4 glycosidic bond in the glycan backbone of peptidoglycan — and their effect on culturability. In M. tuberculosis, the interaction between RpfB and the d,l-endopeptidase, Rpf interacting protein A (RipA), enables these proteins to synergistically degrade peptidoglycan to facilitate growth. Furthermore, the combined action of Rpfs with RipA and other peptidoglycan hydrolases might produce muropeptides that could exert diverse biological effects through host and/or bacterial signaling, the latter involving serine/threonine protein kinases. Here, we explore these possibilities in the context of the structure and composition of mycobacterial peptidoglycan. Clearly, a deeper understanding of the role of Rpfs and associated peptidoglycan remodeling enzymes in bacterial growth and culturability is necessary to establish the significance of dormancy and resuscitation in diseases such as tuberculosis, which are associated with long-term persistence of viable bacterial populations recalcitrant to antibiotic and immune clearance.
The success of bacteria is evident in their enormous diversity and ubiquitous distribution, which derives primarily from their ability to adapt to and even thrive in varying and often hostile environments. This notion is exemplified by the fact that many bacterial species are able to enter into a state of dormancy, defined as ‘a reversible state of low metabolic activity in which cells can persist for extended periods without division’ (Kell & Young, 2000), under nutrient-poor conditions. The reversibility of this state enables dormant organisms, which are characterized by impaired culturability, to ‘reawaken’ or resuscitate under conditions permissive to regrowth. This impaired culturability is defined by the inability to score growth on media normally conducive to optimal replication. Critically, the concepts of culturability and dormancy both imply that the organism retains the potential for future growth. Included among the organisms that are able to enter into dormancy is the major human pathogen, Mycobacterium tuberculosis (Shleeva et al., 2002), the causative agent of tuberculosis. One-third of the world's population is estimated to be infected with M. tuberculosis (Dye et al., 2005). A small proportion of those infected progress to primary disease, whereas the majority develop clinically asymptomatic latent tuberculosis infection (LTBI) (Stewart et al., 2003). Immunocompetent individuals who are latently infected with M. tuberculosis are estimated to have a 10% lifetime risk of developing reactivation disease, which can occur many years after the primary infection (Lillebaek et al., 2002). This risk increases significantly in immunocompromised individuals (Corbett et al., 2003), underscoring the central role of the immune system in controlling latency and reactivation. Although little is currently known about the physiology of M. tuberculosis in LTBI, understanding the nature of the persistent state of the organism during latency, and the mechanisms by which this state is controlled, represents an important area of tuberculosis research today (Young et al., 2009). In this context, an intriguing question posed in recent years is whether — and to what extent — the clinically defined phenomena of latency and reactivation might be related to the microbiological phenomena of dormancy and resuscitation (Young et al., 2005).
The view of bacterial dormancy and resuscitation was revolutionized by the discovery, in 1998, of the resuscitation-promoting factor (Rpf) from Micrococcus luteus by Mukamolova (1998). Micrococcus luteus Rpf is the founder member of a family of secreted proteins found throughout the Actinobacteria and in Firmicute species, with many organisms, including mycobacteria, possessing multiple Rpf homologues (Kell & Young, 2000; Ravagnani et al., 2005; Schroeckh & Martin, 2006). In this review, we summarize the key advances that have been made in understanding the biochemistry of Rpfs from various organisms and the physiological roles of these proteins, specifically in the context of the cell wall remodeling function implied by their muralytic activity (Cohen-Gonsaud et al., 2005; Mukamolova et al., 2006). Much of the earlier research on Rpfs is covered here only briefly as it has been reviewed previously in a number of excellent articles to which the reader is referred (Kell & Young, 2000; Young et al., 2005; Keep et al., 2006a, b; Hett & Rubin, 2008).
Rpfs as growth-stimulatory factors
Micrococcus luteus Rpf is an essential, secreted protein with a number of key properties: it increases the number of CFUs when added to dormant cultures, stimulates the growth of extensively washed cells, decreases the growth lag time and allows growth in media not normally conducive to culturing of this organism (Mukamolova et al., 1998, 1999, 2002a). Micrococcus luteus Rpf exerts its growth-stimulatory effects from an extracellular location because it associates with the cell wall, most likely through a LysM-type domain (Mukamolova et al., 2002a). In a key advance, this protein was shown to possess muralytic activity that correlates with its resuscitation-promoting activity (Cohen-Gonsaud et al., 2005; Mukamolova et al., 2006). This finding was consistent with early structural predictions (Cohen-Gonsaud et al., 2004) and implicated Rpf directly in cell wall remodeling through hydrolysis of the glycan backbone of peptidoglycan. Importantly, the growth-stimulatory properties of recombinant forms of the five Rpfs found in M. tuberculosis (RpfA–E) (Cole et al., 1998) suggest a conservation in Rpf function among high GC content Gram-positive bacteria (Mukamolova et al., 2002b). This conclusion was supported by the observation that Micrococcus luteus Rpf is able to stimulate the growth of an aged culture of the Academia strain of M. tuberculosis (Shleeva et al., 2002) and recovery of M. tuberculosis grown in murine macrophages (Biketov et al., 2000) and in BACTEC cultures (Wu et al., 2008). Cross-species activity has also been observed for Rpfs from other Actinobacteria (Schroeckh & Martin, 2006), providing further evidence for conservation of function within this protein family. Moreover, in accordance with a (collective) role for these proteins in bacterial resuscitation, progressive rpf-like gene loss was found to impair the ability of M. tuberculosis to resuscitate spontaneously from a nonculturable state (Downing et al., 2005; Kana et al., 2008).
Not surprisingly, the Rpfs of M. tuberculosis have been the subject of particular interest and attention; however, their multiplicity has presented significant challenges for dissecting individual/specialized vs. collective functions in growth, resuscitation and virulence (Downing et al., 2004, 2005; Tufariello et al., 2004, 2006; Kana et al., 2008; Russell-Goldman et al., 2008). Initial studies that reported the deletion of individual rpf genes with no significant phenotypic consequence led to the suggestion of a functional redundancy within this protein family (Downing et al., 2004; Tufariello et al., 2004). Sequential deletion mutagenesis subsequently revealed that the entire RpfA–E family is dispensable for the growth of M. tuberculosis in an axenic culture (Kana et al., 2008). However, the observation that deletion of three rpf genes in varying combinations conferred differential growth defects in vivo provided the first evidence that the Rpfs may serve discrete functions in this organism (Downing et al., 2005). Subsequent observations supported this view: deletion of rpfB in M. tuberculosis Erdman was found to result in delayed reactivation from chronic infection in mice following immune suppression (Tufariello et al., 2006). Furthermore, deletion of two rpf genes in various combinations resulted in no significant growth defects of M. tuberculosis Erdman in vitro and in vivo, with the exception of strains that lacked rpfB, which displayed delayed reactivation kinetics in mice following immune suppression, with the defect in a ΔrpfAΔrpfB double mutant being more profound than in a ΔrpfBΔrpfD mutant (Russell-Goldman et al., 2008). The ΔrpfAΔrpfB mutant of M. tuberculosis Erdman was also defective for survival during chronic infection in mice, displayed altered colony morphology and induced a differential cytokine response in infected macrophages (Russell-Goldman et al., 2008), consistent with the reduced virulence of a corresponding ΔrpfAΔrpfB double mutant of M. tuberculosis H37Rv following an intraperitoneal injection in a murine model of infection (Biketov et al., 2007). In a further development, the differential in vivo growth and survival phenotypes of quadruple mutants of M. tuberculosis H37Rv, combined with the variability in the kinetics of colony formation and the sensitivity of these strains to cell wall stress, revealed a functional hierarchy within the rpf gene family in M. tuberculosis, with rpfB and rpfE ranking above rpfD in terms of the phenotypes assessed (Kana et al., 2008). More recently, rpfE was the only single rpf-like gene, of five, in Mycobacterium bovis that was identified as being essential for the transition from slow to fast growth in a chemostat (Beste et al., 2009), further suggesting some degree of functional divergence among these proteins.
The view that has emerged from mouse infection studies with rpf-deficient strains of M. tuberculosis is that the Rpfs play an important role in pathogenesis and bacterial recrudescence from chronic infection. The rpfA–E genes are expressed in varying abundance in a growth-phase-dependent fashion in vitro (Mukamolova et al., 2002b; Tufariello et al., 2004) and in mouse lung (Tufariello et al., 2004). This, together with the observation that rpf transcripts were detected in tissue from humans infected with M. tuberculosis (Rachman et al., 2006; Davies et al., 2008), and during tuberculosis infection in rabbits (Kesavan et al., 2009), underscores their potential importance in pathogenesis. Interestingly, all of the M. tuberculosis Rpfs other than RpfC are immunogenic because they elicit the production of antibodies in mice and stimulate the proliferation of lymph node T cells, suggesting that they modulate the immune response during infection and may serve as useful vaccine candidates (Yeremeev et al., 2003; Zvi et al., 2008). This conclusion was corroborated by the observation that immunization of mice with M. tuberculosis RpfE or Micrococcus luteus Rpf confers some measure of protection against a subsequent challenge (Yeremeev et al., 2003). With regard to a role for Rpfs in recrudescence from chronic infection in mice, it is important to note a recent study demonstrating that the apparent stability of the bacillary count observed in chronic infection belies the fact that bacterial replication continues to occur throughout this stage of infection, that is, a dynamic population of growing and dying organisms is kept in check by the host immune system (Gill et al., 2009). Consequently, the inability of rpf-deficient mutants to initiate growth upon immune suppression in mice may instead reflect a role for Rpfs in the recovery of bacilli that have sustained an immune-mediated assault rather than in the regrowth of organisms from a strictly nonreplicating or a metabolically quiescent state.
In addition to M. tuberculosis and M. bovis Bacillus–Calmette–Guerin (BCG), the genomes of other sequenced mycobacterial species contain multiple rpf homologues (Table 1). Notably, the genome of Mycobacterium leprae has rpfB and rpfC homologues, and the 3′-terminal region of rpfA that encodes the Rpf domain and may thus be functionally active. Since M. leprae has undergone extensive reductive evolution and is considered to possess the ‘minimal’ genome compatible with mycobacterial survival (Cole et al., 2001), the presence of multiple rpf homologues in this organism suggests that a multiplicity of rpf genes is essential in pathogenic mycobacteria. A similar trend is observed in Mycobacterium ulcerans, which has also undergone reductive evolution, but retains homologues of rpfA, rpfB, rpfC and rpfE (Stinear et al., 2007), as observed in Mycobacterium avium and Mycobacterium marinum. The saprophyte, Mycobacterium smegmatis, encodes four rpf genes, two of which are located in the same genomic region and are homologous to M. tuberculosis rpfC and rpfE.
|rpf gene||M. tuberculosis H37Rv||M. tuberculosis CDC1551||M. bovis BCG Pasteur 1173P2||M. bovis AF2122/97||M. leprae TN||M. ulcerans Agy99||M. avium ssp. paraTB K-10||M.avium 104||M. marinum M||M. smegmatis mc2155|
|rpf gene||M. tuberculosis H37Rv||M. tuberculosis CDC1551||M. bovis BCG Pasteur 1173P2||M. bovis AF2122/97||M. leprae TN||M. ulcerans Agy99||M. avium ssp. paraTB K-10||M.avium 104||M. marinum M||M. smegmatis mc2155|
All annotations were confirmed from at least two data sources. GenoList houses information for Mycobacterium tuberculosis (strain H37Rv), Mycobacterium leprae TN, Mycobacterium bovis AF2122/97, M. bovis BCG Pasteur 1173P2, Mycobacterium marinum M and Mycobacterium ulcerans Agy99 (http://genolist.pasteur.fr). The J. Craig Venter Institute (JCVI) houses information for M. tuberculosis (strain CDC1551), M. leprae TN, M. bovis AF2122/97, Mycobacterium smegmatis mc2155, Mycobacterium avium 104 and M. avium ssp. paratuberculosis K-10 (). Direct blast analysis at NCBI was possible for all of these strains, except M. marinum M and M. ulcerans Agy99 (). WebACT analysis was used in conjunction with blast analysis and GenoList annotation ().
RpfA in M. bovis has an internal 78 amino acid deletion in the proline–alanine-rich region, when compared with M. tuberculosis and M. bovis BCG, but still retains the catalytic glutamate.
Annotated by analyzing gene synteny in comparison with the corresponding region in M. tuberculosis.
MSMEG_4640 and MSMEG_4643 both show similarity to rpfC and rpfE from M. tuberculosis. This genomic region in M. smegmatis is more similar to that of rpfE in M. tuberculosis.
Transcriptional regulation and post-translational modification of Rpfs
Significant progress has been made in defining the transcriptional mechanisms that govern the expression of the Rpfs of M. tuberculosis. As mentioned previously, the rpfA–E genes are expressed at varying levels in a growth-phase-dependent fashion in vitro and in vivo (Mukamolova et al., 2002b; Tufariello et al., 2004). Importantly, although detected in actively growing cells, transcript was not detected for any of the rpf genes in M. bovis BCG grown to the stationary phase or following prolonged starvation (Mukamolova et al., 2002b). In M. tuberculosis, deletion of rpfB, rpfC, rpfD and rpfE individually resulted in a small, but significant upregulation of some or all of the remaining rpf genes during logarithmic-phase growth in axenic culture, which was suggestive of a degree of regulatory cross-talk between the genes (Downing et al., 2004). However, a subsequent study reported a reversal of these effects upon deletion of further rpf genes, arguing against the notion of regulatory cross-talk within the rpf gene family (Kana et al., 2008). Instead, there is growing evidence to suggest that the five rpf genes in M. tuberculosis are regulated independent of one another by distinct mechanisms. For example, rpfA is negatively regulated by a homologue of the cAMP receptor protein that binds the promoter region and represses gene expression (Rickman et al., 2005). On the other hand, rpfC appears to be positively regulated by the alternate sigma factor, SigD (Raman et al., 2004), and by a homologue of the site-two protease (Makinoshima & Glickman, 2005). Expression analysis has also shown that rpfC may be negatively regulated by the MprAB two-component regulatory system that affects many genes in the SigD regulon (Pang et al., 2007). A recent study has further underscored the differences in the transcriptional responsiveness of the M. tuberculosis rpf genes, with rpfE alone being significantly downregulated in response to vancomycin treatment (Provvedi et al., 2009).
The genome of Corynebacterium glutamicum encodes two rpf-like genes, rpf1 and rpf2, which are similar to rpfA and rpfB from M. tuberculosis, respectively (Hartmann et al., 2004); the latter seems to be controlled by an elaborate network of regulators including RamA, RamB and GlxR (Jungwirth et al., 2008), all of which regulate gene expression in response to growth on different carbon sources. Furthermore, rpf1 is also under the direct control of GlxR, suggesting a broader, possibly overlapping, pathway that regulates both rpfs in this organism (Kohl et al., 2008). Interestingly, it has been shown that Rpf2 in C. glutamicum is post-translationally modified via glycosylation through the addition of a carbohydrate moiety that contains mannose and galactose (Hartmann et al., 2004), providing the first evidence for (differential) post-translational modification of Rpfs. The glycosyltransferase responsible for the addition of sugar groups is encoded by the pmt gene, which bears strong similarity to other protein-O-mannosyltransferases including Rv1002c from M. tuberculosis, suggesting that RpfB may also be glycosylated in this organism (Mahne et al., 2006). However, the relevance, if any, of glycosylation to the biological activity of this and other Rpfs is yet to be established.
Rpf structure and biochemical function
The presence of N-terminal signal sequences suggests that Rpfs are exported to the cell wall and translocated to an extracellular location, as demonstrated for Micrococcus luteus, M. bovis and M. tuberculosis (Mukamolova et al., 2002a, b). In addition to the conserved Rpf domain, members of the Rpf family contain amino- and carboxy-terminal sequences that distinguish the proteins from one another (Mukamolova et al., 2002b). For example, M. tuberculosis RpfA carries an extensive Pro–Ala-rich sequence C-terminal to the Rpf domain, whereas RpfB has an N-terminal extension carrying a lipoprotein lipid attachment site. The functional impact of these distinctive structural elements has not been investigated, but it is tempting to speculate that they may be associated with modulating Rpf function, substrate specificity/selectivity and/or interaction with other proteins. Structural modeling of the Rpf domain suggested that it would adopt a fold similar to that of c-type lysozymes (Cohen-Gonsaud et al., 2004). This prediction was later confirmed by the nuclear magnetic resonance structure of the catalytic domain of M. tuberculosis RpfB, that revealed significant structural similarity to soluble lytic transglycosylases (SLTs) with an invariant, catalytic glutamate, and associated peptidoglycan-binding residues being conserved (Cohen-Gonsaud et al., 2005). The recently determined crystal structure of a fragment of RpfB corroborated these findings and further showed that the structure of the Rpf and associated G5 domain adopt a comma-like fold (Ruggiero et al., 2009). The structure of the catalytic domain is mini-lysozyme-like because it contains the most conserved helices from previously determined c-type lysozyme structures; as such, it is more similar to g-type lysozymes and SLTs that have less complex structures (Ruggiero et al., 2009). The crystal structure of the G5 domain in RpfB revealed a novel super secondary structure, a β-triple helix-β motif, which exposes a large number of hydrogen bond donors that confer adhesive properties on the protein. This, combined with the observation that the G5 domain is present in proteins that bind the cell wall or are involved in cell–cell adhesion and biofilm formation, supports the notion that RpfB is cell wall-associated, as predicted previously (Rezwan et al., 2007). With respect to the mechanism of action, the structure suggests that the G5 domain anchors RpfB in the cell wall and correctly positions the catalytic domain for cleavage of the β-1,4 glycosidic bond in the glycan strand of peptidoglycan (Ruggiero et al., 2009).
The muralytic activity of Rpfs was established by demonstrating that Micrococcus luteus Rpf is able to degrade cell wall material and to bind to, and cleave, a synthetic tri-N-acetylglucosamine (NAG) cell wall mimetic as well as other Gram-positive cell wall-like substrates (Mukamolova et al., 2006; Telkov et al., 2006). Truncation of the protein or mutation of the catalytic glutamate diminished this activity significantly and simultaneously abolished the ability of Rpf to stimulate growth, suggesting that cell wall lysis is directly related to growth stimulation by Rpfs (Cohen-Gonsaud et al., 2005; Mukamolova et al., 2006). The structural findings suggest that Rpfs cleave the β-1,4 glycosidic bond between N-acetylmuramic acid (NAM) and NAG (Fig. 1), although no experimental evidence exists to confirm this hypothesis. It is also not known whether the enzymatic activity of Rpf is that of a lysozyme (i.e. a hydrolase), or a lytic transglycosylase (LT), which shares the same substrate specificity as lysozyme, but, rather than catalyzing a hydrolysis reaction, cleaves peptidoglycan with the concomitant formation of an intramolecular 1,6-anhydromuramoyl product (Scheurwater et al., 2008; Vollmer et al., 2008). LTs cleave the glycan backbone to create space for the insertion of new peptidoglycan units during growth (Vollmer et al., 2008) and are also involved in the cell wall remodeling required for the insertion of structures such as secretion systems that span the envelope (Scheurwater et al., 2008).
The collective dispensability of the rpfA–E genes for the growth of M. tuberculosis in an axenic culture (Kana et al., 2008) suggests that, under these conditions, the lytic function(s) of the Rpf family can be served by (an)other enzyme(s) and that the Rpfs rather play an auxiliary role in some essential cellular function such as septation (Hett et al., 2007). However, M. tuberculosis strains lacking multiple rpf genes are significantly attenuated in the murine model of tuberculosis infection, suggesting that the in vivo environment unmasks a collectively essential role for Rpfs in bacterial growth and survival (Downing et al., 2005; Kana et al., 2008). Stress imposed by the host immune system results in significant cell wall damage, which may require Rpf activity to restore culturability. This notion is supported by the fact that Rpf deficiency increases the sensitivity of M. tuberculosis to cell wall-damaging agents (Kana et al., 2008) and that the culturability of this organism is reduced by residence in macrophages (Biketov et al., 2000). Rpf deficiency may therefore increase the sensitivity of M. tuberculosis to antibiotics that target peptidoglycan biosynthesis or some other cellular component involved in maintaining cell wall integrity. Alternatively, Rpf function may be required for the release of (anhydro)muropeptide fragments that can be recycled and serve in bacterial (Jacobs et al., 1994, 1997) and/or host signaling (Boneca, 2005) (Fig. 1). This notion is explored in greater detail below.
Many enzymes involved in peptidoglycan synthesis and degradation either function in complexes with other enzymes or are able to interact transiently with other proteins in order to exert their effects. A suggestion that this may apply in the case of Rpfs stems from the fact that one of the Rpfs in Streptomyces coelicolor and Streptomyces avermitilis occurs as a fusion with a putative M23 Gly–Gly peptidase domain (Keep et al., 2006a). A yeast two-hybrid screen for interacting partners for the M. tuberculosis Rpfs revealed that RpfB and RpfE interact with the NlpC/p60-type endopeptidase, Rpf interacting protein A (RipA) (Hett et al., 2007). This enzyme is predicted to hydrolyze the bond between d-iso-glutamine (d-iGln) and meso-diaminopimelic acid (m-DAP) in the peptidoglycan stem peptide (Boneca, 2005; Hett et al., 2007) (Fig. 1). RipA is secreted, cell wall associated and synergizes with RpfB in peptidoglycan hydrolysis (Hett et al., 2008). The colocalization of RipA and RpfB to the septa of dividing cells suggests a concerted role for these enzymes in cell division (Hett et al., 2007). Interestingly, RipA is essential for normal cell division in M. smegmatis (Hett & Rubin, 2008), and also in M. tuberculosis (Sassetti et al., 2003), even though both these organisms contain other ripA orthologues. RipA depletion in M. smegmatis resulted in growth arrest and the formation of long-branched chains owing to incomplete resolution of septa, and also sensitized the cells to β-lactam antibiotics (Hett et al., 2008). These results suggest that RipA plays an essential role in the final stages of cell division and that its activity may be modulated or enhanced by RpfB. An analogous synergy between peptidoglycan-degrading enzymes is seen in Listeria monocytogenes, where the combined activities of a muramidase (NamA) and the d,l-endopeptidase, P60, contribute to virulence (Lenz et al., 2003), suggesting an important role for these peptidoglycan hydrolytic enzymes in pathogenesis (Boneca, 2005). In the case of M. tuberculosis, the role of RpfB appears to be auxiliary in cell division, but nonredundant in pathogenesis (Tufariello et al., 2006; Russell-Goldman et al., 2008).
Mycobacterial peptidoglycan and its structural remodeling
Modulation of peptidoglycan biosynthesis and composition occurs during normal cellular growth and is required to maintain cell size, shape, cytoplasmic turgor and effect correct septation during cell division. The peptidoglycan structure also plays an important role in the maintenance of dormancy and heat resistance in spores of Bacillus species, which display a lower degree of peptidoglycan cross-linking and substitution of a significant percentage of muramic acid with muramic δ-lactam residues, a replacement that is required for hydrolysis of the cortex during germination (Atrih & Foster, 1999).
Because the Rpfs function in cell wall lysis, their physiological role(s) in mycobacterial growth, culturability and virulence is likely to be linked to the structure of peptidoglycan and the changes in its composition and structure that may occur as a function of the growth state of the bacterium. Mycobacterial peptidoglycan has certain characteristics that are thought to play a key role in enabling pathogenic species to survive in the host (Fig. 1). The peptidoglycan is contained within a mycolyl–arabinogalactan–peptidoglycan complex, which constitutes the major structural component of the cell envelope (Crick et al., 2001). Modifications of the free carboxylates of m-DAP and d-iGln are also observed (Lavollay et al., 2008). In addition, the muramic acid moieties in the peptidoglycan are both N-glycolylated and N-acetylated (Mahapatra et al., 2005; Raymond et al., 2005), a feature shared with very few other genera of bacteria. Although dispensable in M. smegmatis, N-glycolylation of muramic acid is thought to play a role in lysozyme resistance of the peptidoglycan and contribute to stability of the mycobacterial cell wall (Raymond et al., 2005). Finally, mycobacterial peptidoglycan contains direct 3→3 cross-links between (modified) m-DAP–m-DAP moieties. A significant shift from 4→3 to 3→3 cross-linking occurs during stationary-phase adaptation in M. tuberculosis (Lavollay et al., 2008). The 3→3 cross-links are generated through l,d-transpeptidation by Rv0116c, a member of the active-site cysteine peptidase family that is highly induced under nutrient starvation (Betts et al., 2002) and irreversibly inactivated by carbapenems (Lavollay et al., 2008). The unusually high content of 3→3 cross-links in the stationary phase-adapted peptidoglycan is thought to render it resistant to hydrolysis by endopeptidases.
Peptidoglycan and muropeptide sensing and signaling in bacteria
A key unresolved question is whether the function of Rpfs in cell wall remodeling is simply to facilitate diffusion across the bacterial cell wall, or whether these enzymes mediate their effects through the production of muropeptide fragments that act as signaling molecules (Keep et al., 2006a). Bacteria bind and respond to peptidoglycan through the action of eukaryotic-like serine/threonine protein kinases (STPKs) and associated serine/threonine protein phosphatase systems (Shi et al., 1998). STPKs are single, membrane-spanning polypeptides that have an external sensor domain and an internal cytoplasmic protein kinase domain that phosphorylates target proteins upon detection of a cognate signal, resulting in a coordinated cellular response through a signal cascade. The Bacillus subtilis STPK, PrkC, has been implicated in growth and developmental processes by modulating sporulation and biofilm formation (Madec et al., 2002). PrkC has an external domain consisting of three repeat PASTA (penicillin-binding protein and serine/threonine kinase associated) domain motifs and is able to form membrane-associated dimers. Like other STPKs, PrkC is activated by autophosphorylation (Madec et al., 2003) and is regulated by a PPM family protein phosphatase, PrpC, which dephosphorylates the active form of PrkC (Obuchowski et al., 2000) and its substrates (Gaidenko et al., 2002). A PASTA domain-containing homologue of PrkC in Enterococcus faecalis has similarly been implicated in growth, sensitivity to cephalosporins, vancomycin and detergents that target cell wall biogenesis, and in maintenance of cell shape (Kristich et al., 2007). In an exciting new development, a recent study has shown that m-DAP-containing muropeptides released during active growth are powerful germinants of B. subtilis spores (Shah et al., 2008). Furthermore, muropeptide-dependent germination required PrkC, which was shown to bind peptidoglycan and effect a signal cascade that leads to spore germination (Shah et al., 2008). Together, these data define a novel pathway for bacterial resuscitation from a nongrowing state via the sensing of specific peptidoglycan fragments by an STPK. Given the biochemical function of Rpfs in cell wall remodeling, it is plausible that these factors play a key role in producing the signaling molecule(s) that is able to bind to the PASTA domain of PrkC homologues (Fig. 1).
The homologue of PrkC in mycobacteria is the essential STPK, PknB (Fernandez et al., 2006). The pknB gene is located in a cluster that includes pstP that encodes its cognate phosphatase (Boitel et al., 2003), the STPK-encoding pknA that is associated with cell morphology (Chaba et al., 2002) and other genes involved in cell growth such as rodA and pbpA (Cole et al., 1998). Both pknA and pknB are markedly downregulated during carbon starvation, implicating their encoded kinases in growth and cell division (Betts et al., 2002). The pknB-containing gene cluster is conserved in other mycobacteria, including M. leprae, suggesting that it plays an essential role in cell growth and survival. PknB shares a high degree of structural similarity to other eukaryotic-like STPKs, with an intracellular kinase domain that undergoes auto-phosphorylation as well as an extracellular sensing domain containing four PASTA motifs, suggesting that this protein is also able to bind peptidoglycan (Ortiz-Lombardia et al., 2003; Young et al., 2003). PknB substrates identified to date include proteins involved in cell division such as Wag31(activated directly by PknA and indirectly by PknB) and PpbA, both of which localize to the cell poles during cell division and whose loss or depletion results in significant morphological defects (Kang et al., 2005, 2008). PknB can also phosphorylate and reduce the acetyl-transferase activity of GlmU, an enzyme involved in the formation of the peptidoglycan biosynthetic precursor, UDP-N-acetylglucosamine, suggesting that it may modulate bacterial growth by regulating the early stages of peptidoglycan synthesis (Parikh et al., 2009).
It is tempting to speculate that the combined action of Rpfs and other enzymes, such as l,d-carboxypeptidase, generates m-DAP-terminating muropeptides such as the disaccharide tripeptide shown in Fig. 1, which can bind to PknB and trigger resuscitation in a manner analogous to spore germination in B. subtilis (Shah et al., 2008). The collective importance of the Rpfs for resuscitation of M. tuberculosis from a nonculturable state (Downing et al., 2005; Kana et al., 2008) suggests that the peptidoglycan cleavage function of the Rpfs in this particular case cannot be functionally substituted by other enzymes. A key factor in this regard might be the extensively 3→3 cross-linked structure of the peptidoglycan in nondividing, aged cells that may restrict cleavage of the glycan strand to the Rpfs. The insights gained recently from comparative genomic analyses of mycobacteriophages are particularly informative for this hypothesis (Pedulla et al., 2003). A motif related to Rpf has been found to be embedded in the tape measure protein (Tmp) of a number of mycobacteriophages, together with two other motifs that are also associated with peptidoglycan-degrading activity. The peptidoglycan-degrading activity of one of the motifs was found to facilitate phage infection and DNA injection into stationary-phase cells, suggesting that inclusion of peptidoglycan-degrading activity in the Tmp is a strategy that mycobacteriophages use to deal with growth phase-dependent alterations in the peptidoglycan structure of their hosts (Piuri & Hatfull, 2006).
Muropeptide signaling in the host?
There is considerable evidence implicating peptidoglycan fragments and the enzymes involved in their production in the pathogenesis of many bacterial species (Bartoleschi et al., 2002; Boneca, 2005; Humann & Lenz, 2009). As shown in Fig. 1, the concerted action of Rpfs and RipA is predicted to generate a product(s) related to the biologically active muramyldipeptide, which acts as a Nod2 agonist, whereas peptidoglycan cleavage by Rpfs and other peptidases could generate muramyltripeptides containing a terminal m-DAP residue that may be detected by Nod1 (Girardin et al., 2003a, b). It is conceivable that pathogenic mycobacteria may utilize muropeptides of this type to modulate innate immune responses to infection (Jo, 2008). In this regard, the functional implications of distinctive structural features in mycobacterial peptidoglycan for the biological activity of resulting muropeptides provide interesting possibilities for the stimulation of different host pathways. Particularly relevant is the fact that the ɛ carboxylate of m-DAP was found to be largely amidated in the peptidoglycan from M. tuberculosis in the stationary phase (Lavollay et al., 2008); this modification is expected to preclude recognition by Nod1 (Girardin et al., 2003b). Other modifications that may affect muropeptide function include the 1,6-anhydromuramoyl form of the disaccharide product, possibly produced by Rpf action, and N-glycolylation of the muramic acid in muropeptides derived from mycobacterial peptidoglycan. With regard to the latter, it has been shown that N-glycolyl-muramyl dipeptide is a potent agonist for Nod2, resulting in more efficient stimulation of type I interferon responses, associated with M. tuberculosis infection, than the acetylated form of muramyl dipeptide (Pandey et al., 2009). Furthermore, disruption of namH in M. smegmatis, which encodes the enzyme responsible for hydroxylation of UDP-N-acetylmuramic acid to produce the N-glycolylated form, results in reduced Nod2-mediated modulation of the immune response ex vivo and during intraperitoneal stimulation in mice (Coulombe et al., 2009). Interestingly, N-deacetylation of the peptidoglycan has been shown to play a critical role in innate immune evasion by Listeria (Boneca et al., 2007), suggesting that specific peptidoglycan modifications might enable pathogenic mycobacteria to use analogous immune evasion strategies.
Since the discovery of Micrococcus luteus Rpf a decade ago, considerable progress has been made in understanding the structure, function and physiological role of Rpf-like enzymes in various organisms. However, many compelling questions remain. Foremost among these is whether Rpfs mediate their growth-stimulatory effect by simply facilitating diffusion across the bacterial cell wall or through the production of muropeptide fragments that can act as signaling and/or immunomodulatory molecules. Furthermore, does Rpf act like a lysozyme or an LT in cleaving the glycan backbone of peptidoglycan? This question is central to elucidating the function of the products of Rpf action. Does the multiplicity of Rpfs in organisms such as M. tuberculosis imply some degree of functional differentiation of the individual proteins in terms of substrate selectivity/specificity, and if so, how? If Rpf action leads to the production of biologically active muropeptides that can trigger/modulate bacterial and/or host signaling pathways, what effects do structural modifications such as N-amidation of m-DAP have on the activity of such molecules? Answers to these questions, which are fueling ongoing research in this area, will have important implications for understanding the significance of bacterial dormancy and resuscitation and the associated remodeling of peptidoglycan in diseases such as tuberculosis.
B.D.K. and V.M. are supported by grants from the Medical Research Council of South Africa and National Research Foundation and V.M. is supported by an International Research Scholar's grant from the Howard Hughes Medical Institute. We would like to thank Digby Warner for constructively reviewing the manuscript, Chris Sassetti for helpful discussions and Edith Machowski for assistance with the bioinformatics analysis.