Attention Seeker: Production, Modification, and Release of Inflammatory Peptidoglycan Fragments in Neisseria Species

ABSTRACT

Maintenance of the structural macromolecule peptidoglycan (PG), which involves regulated cycles of PG synthesis and PG degradation, is pivotal for cellular integrity and survival. PG fragments generated from the degradation process are usually efficiently recycled by Gram-negative bacteria. However, Neisseria gonorrhoeae and a limited number of Gram-negative bacteria release PG fragments in amounts sufficient to induce host tissue inflammation and damage during an infection. Due to limited redundancy in PG-modifying machineries and genetic tractability, N. gonorrhoeae serves as a great model organism for the study of biological processes related to PG. This review summarizes the generation, modification, and release of inflammatory PG molecules by N. gonorrhoeae and related species and discusses these findings in the context of understanding bacterial physiology and pathogenesis.

INTRODUCTION

Peptidoglycan (PG) forms a meshlike sacculus surrounding the cell membrane of a bacterial cell, providing cell shape and protection against turgor pressure. Expansion and reshaping of the sacculus during cell growth and cell separation is a dynamic process that requires spatiotemporally coordinated degradation and synthesis of PG. Small PG fragments generated when cross-links are broken and PG strands are degraded; these PG fragments are either taken up into the cytoplasm for recycling or are released from the cell. Neisseria gonorrhoeae is unusual among Gram-negative bacteria in that it releases significant amounts of soluble PG fragments during growth (1).

The release of PG fragments by N. gonorrhoeae has been of particular interest due to the effects of these molecules in gonococcal infections. Host tissue damage during gonococcal infections is partly facilitated by host inflammation and cytokine production on recognition of certain PG fragments (2, 97). PG fragments from N. gonorrhoeae were found to induce the death and sloughing of ciliated cells in human Fallopian tube (FT) organ culture, reproducing the tissue damage seen in patients with gonococcal pelvic inflammatory disease (1, 3). The PG fragments that damage FT tissue are monomeric subunits of PG comprised of disaccharides with either a 3- or 4-amino-acid chain attached to the N-acetylmuramic acid, referred to as PG monomers (Fig. 1A, compound III). A few other bacterial species have been shown to use PG fragments in a similar way. Bordetella pertussis releases PG monomer with a 4-amino-acid chain, and this molecule induces the death and sloughing of ciliated cells in tracheal tissue (4, 5). Vibrio fischeri uses the same PG monomer to induce tissue regression and ciliated cell death in the Hawaiian bobtail squid during establishment of symbiosis (6, 7).

FIG 1.

Small PG fragments released by Neisseria. (A) Chemical structures of PG fragments released by Neisseria: I, glycosidically linked PG dimers; I*, tetrasaccharide peptide; II, cross-linked PG dimers; III, PG monomers; IV, free peptide; V, free disaccharide; VI, free anhydro-MurNAc. Lighter shading indicates portions of peptides present in some but not all PG fragments. Note that all muropeptides have anhydro ends on the MurNAc sugar. (B to D) PG fragments labeled with [³H]glucosamine (GlcNH₂) (B,C) or [³H]DAP (D) released by N. gonorrhoeae (B to D), N. meningitidis (B, D), N. sicca (C), and N. mucosa (C) separated using size exclusion chromatography. Peaks correspond to the structures in panel A. Peaks a and b have not been identified by mass spectrometry.

The lack of extensive redundancy in peptidoglycanases, genetic tractability of the organism, and relevance of PG fragments to infections make N. gonorrhoeae an attractive organism for the study of PG metabolism and fragment release. This review focuses on how PG fragments are generated, which PG fragments get released by N. gonorrhoeae and related species of Neisseria, and the consequences of PG fragment release in different niches.

STRUCTURE AND COMPOSITION OF PG IN N. GONORRHOEAE

N. gonorrhoeae PG has the same basic structure as that of most Gram-negative bacteria (8). The glycan backbone is made of repeating subunits of N-acetylglucosamine (GlcNAc) and N-acetylmuramic acid (MurNAc) linked by β-1,4-glycosidic bonds. The peptide stem consists of 2 to 5 amino acids, in the order l-alanine, d-glutamate, meso-diaminopimelic acid (DAP), d-alanine, and d-alanine, in which l-alanine forms an amide bond with MurNAc. Approximately 70% to 80% of the peptide stems in the gonococcal sacculus are tetrapeptides, 20% to 30% are tripeptides, and a small percentage are pentapeptides or dipeptides (9–11). N. gonorrhoeae has high levels (∼40%) of PG cross-linking, primarily in the form of amide bonds between a DAP moiety of one peptide stem and d-alanine of another, with fewer DAP-DAP bonds (9). Some Gram-negative bacteria anchor the PG layer and outer membrane via the covalently bound Braun's lipoprotein; however, gonococci lack Braun's lipoprotein (12). Instead, Neisseria species express OmpA-domain outer membrane proteins, such as gonococcal protein III (known as RmpM in Neisseria meningitidis), that bind PG noncovalently. These proteins act to stabilize the outer membrane and may function to anchor the outer membrane to the PG layer (13, 14). Unlike Escherichia coli but similar to many other Gram-negative bacteria, N. gonorrhoeae acetylates PG at the C-6 position of MurNAc (O-acetylation) (15–18). Deacetylation of PG is required for degradation of the PG strands by gonococcal lytic transglycosylases and the generation of most of the soluble PG fragments (19).

PG FRAGMENTS RELEASED BY NEISSERIA SPECIES

Although N. gonorrhoeae releases some large fragments of PG by autolysis, the majority of PG fragments released are small fragments liberated during normal PG turnover and cell growth, without lysis (10, 20). PG fragments released by Neisseria species during growth have 1,6-anhydro bonds on the MurNAc sugar due to the action of lytic transglycosylases (10). This is in contrast to PG fragments liberated by lysozyme treatment, which produces reducing ends. Release of these PG fragments can be tracked by metabolic labeling of PG, and the types and amounts of PG fragments released can be analyzed by size exclusion chromatography (21). Gonococci release PG monomers, PG dimers, tetrasaccharide peptides, disaccharide, anhydro-MurNAc, and free peptide stems (Fig. 1A and B). PG dimers are two monomeric units linked by a glycosidic bond between the two disaccharide units (glycosidically linked dimer) or by cross-linking the two peptide stems (peptide-linked dimer). Tetrasaccharide peptides have two glycosidically linked disaccharide units with only one peptide stem attached to the anhydro-MurNAc sugar (11).

Most of the PG fragments released by N. gonorrhoeae are in the form of PG monomers with tripeptide or tetrapeptide stems (Fig. 1B). Tetrapeptide monomer is also commonly known as tracheal cytotoxin (TCT) (22). Gonococci release ∼72% to 80% tripeptide monomer and 20% to 28% tetrapeptide monomer (2, 23). This is in contrast to the amount of tripeptide versus tetrapeptide stems in the sacculi (20% to 30% versus 70% to 80%, respectively), suggesting that an l,d-carboxypeptidase in the periplasm trims the peptide stem before release (9, 10). In contrast, B. pertussis releases exclusively tetrapeptide monomer (4, 22). Escherichia coli, which is known to not release much PG monomer, releases 6% to 8% of free peptides per generation, the majority being DAP–d-alanine dipeptides (24).

CONSEQUENCES OF PG FRAGMENT RELEASE

PG fragments have known arthropathic, somnogenic, pyrogenic, cytotoxic, and appetite suppression effects (3, 4, 25–28). Gonococcus-derived PG fragments as small as PG monomers stimulate arthritis when injected into rats (25). Treatment of FT explants with PG monomers or whole bacteria results in the death and sloughing of ciliated FT cells, mimicking the pathophysiology of pelvic inflammatory disease (Fig. 2) (3). The mechanisms involved in the cytotoxic effects of PG monomers are still being worked out. In studies of B. pertussis, treatment of hamster tracheal tissue explants with tetrapeptide monomer (TCT) resulted in interleukin 1 (IL-1) and nitric oxide production in nonciliated epithelial cells, and the nitric oxide caused ciliated cell death (4). It is not clear whether this model is applicable to the death of ciliated FT cells during gonococcal infection, because inhibition of nitric oxide synthesis did not reduce cell death in FT during gonococcal infection (29). Furthermore, McGee et al. (30) demonstrated a correlation between tumor necrosis factor α (TNF-α) and FT damage. Additionally, not all tissue and cell types respond to PG fragments equally due to heterogeneity in PG receptor expression levels (31).

FIG 2.

PG released by N. gonorrhoeae kills and causes extrusion of ciliated FT cells. Scanning electron microscopy of uninfected (A) and gonococcus-infected (B) FT explants. The arrow points to a gonococcal cell. Scale bars = 2 μm.

Nevertheless, recognition of small PG fragments by the intracellular receptors NOD1 and NOD2 elicits a proinflammatory immune response in humans and other mammals (32). Human NOD1 senses free tripeptide and tripeptide monomer and is expressed by many different cell types in humans (33–35). Human NOD2 binds to muramyl dipeptide (MurNAc l-alanine–d-glutamate) and PG monomers with a reducing end and is expressed by a limited set of cell types, including leukocytes and epithelial cells (36–40). Curiously, murine NOD1 recognizes tetrapeptide monomer instead of tripeptide monomer (41). The human receptor that recognizes tetrapeptide monomer has not been definitively identified, although some evidence suggests that NOD1 can recognize tetrapeptide monomers at micromolar concentrations (31, 41). As such, studies examining PG-related effects on pathogenesis in mice must be interpreted with caution when considering their relation to human diseases. Regardless, in mice and humans, recognition of PG fragments by NOD receptors culminates in a local NF-κB-dependent proinflammatory innate immune response, typically mediated by the production of cytokines such as TNF-α, IL-6, IL-8, and IL-1β (30, 33, 36, 42, 43).

One consequence of proinflammatory cytokine release is the recruitment of neutrophils to the infection site. Although neutrophils are capable of killing N. gonorrhoeae, ∼50% of N. gonorrhoeae cells survive and replicate intracellularly in neutrophils (44–46). Neutrophils do not appear to express NOD1 (47). Infection of neutrophils with a gonococcal strain that produced no PG monomers but instead produced more glycosidically linked PG dimers resulted in increased killing by neutrophils (48). Also, neutrophils infected with the mutant showed increased primary and secondary granule exocytosis and increased secondary granule fusion with the phagosome (48). One explanation for this result is that the PG dimers produced by the mutant were degraded by neutrophil lysozyme, creating reducing monomers capable of stimulating NOD2 and causing enhanced killing of the bacteria (31, 40). Thus, it appears that N. gonorrhoeae diminishes a NOD2 response in neutrophils by degrading PG fragments to 1,6-anhydro-containing PG monomers (40).

GENERATION AND RELEASE OF PG FRAGMENTS BY N. MENINGITIDIS AND NONPATHOGENIC NEISSERIA SPECIES

N. meningitidis has the same set of PG-degrading enzymes and releases different amounts of the same types of PG fragments as N. gonorrhoeae (Fig. 1B and D). N. meningitidis releases 3-fold fewer of the proinflammatory PG monomers than N. gonorrhoeae and a smaller percentage of the tripeptide monomers that are NOD1 agonists (2). N. meningitidis greatly increases the release of broken-down PG fragments, including free peptides and anhydro-MurNAc. Comparing overall release of PG by the two pathogenic Neisseria species, we found that N. meningitidis releases 4% of PG fragments generated from cell wall breakdown, whereas N. gonorrhoeae releases 15%. Thus, it is clear that N. meningitidis releases less PG overall and breaks down more fragments into their constituents. Although N. meningitidis is often recognized for the devastating invasive diseases meningitis and meningococcemia, it usually colonizes the nasopharynx of 15% of humans without causing symptoms or disease (49). Whether PG fragment release adds to the inflammatory response in invasive meningococcal disease is unclear, but the smaller amounts of PG fragments released by N. meningitidis may allow the bacteria to adopt their usual asymptomatic lifestyle.

In addition to the pathogenic Neisseria species, humans also host eight species of nonpathogenic Neisseria (50). These species live in oral, nasopharyngeal, or vaginal niches and cause disease under rare circumstances in immunocompromised people. Analysis of PG fragment release has been done for a few of these species. These experiments showed that PG fragment release is lower in overall amounts and that release of PG monomers is significantly lower than for N. gonorrhoeae (51). Neisseria mucosa and Neisseria sicca were each found to release 5% of the PG monomers removed from their cell walls. The types of fragments released were similar to those released by N. meningitidis; however, N. mucosa and N. sicca did not release PG dimers (Fig. 1C). This difference in dimer release by pathogenic and nonpathogenic Neisseria may have immunological implications, because glycosidically linked dimers stimulate NOD2 (40). Thus, the absence of PG dimer release may help these nonpathogens avoid immune stimulation (40).

GENERATION OF PG DIMERS AND MONOMERS BY LYTIC TRANSGLYCOSYLASES

There are five characterized and two putative lytic transglycosylases (LTs) encoded on the gonococcal chromosome, of which two, LtgA and LtgD, play major roles in the generation of PG monomers (Fig. 3) (54). LtgA is a homolog of the E. coli-soluble LT Slt70, and LtgD is a homolog of E. coli MltB/Slt35, which has membrane-bound and soluble forms, respectively (52, 53). In contrast to the E. coli proteins, neisserial LtgA and LtgD are found exclusively as lipid-anchored outer membrane proteins (19). Deletion of ltgA results in a 38% reduction in the amount of PG monomers released, whereas deletion of ltgD results in a 62% reduction (54–56). The ltgA and ltgD mutants release fewer disaccharide and more PG dimers and multimers than wild-type strains (55, 56). Deletion of ltgA and ltgD abolishes release of PG monomers and disaccharide and increases the release of glycosidically linked dimers and larger PG fragments (56). Interestingly, deletion of ltgA in a mutant defective for PG recycling results in a more drastic effect on PG monomer release than deletion of ltgD in the same background, indicating that LtgA generates more PG fragments in the periplasm than generated by LtgD (19).

FIG 3.

Sites of enzyme activity on PG strands. Lytic transglycosylases (LT) act to digest the bond between N-acetylmuramic acid and N-acetylglucosamine, while amidases sever the linkage between the peptide chain and the sugar backbone. Endopeptidases (EP) cleave the cross-links between two peptide chains, while d,d- and l,d-carboxypeptidases (DD-CP and LD-CP) trim the peptide stems. PG de-O-acetylases (also known as acetyl-PG esterases) remove the O-acetyl group on the C6 hydroxyl of MurNAc.

While both LtgA or LtgD can digest fragments as small as a synthetic PG dimer in vitro, LtgA liberates more species of PG fragments from whole sacculi than liberated by LtgD (19). Sequential digestion of sacculi with LtgA and then LtgD or vice versa liberates more PG fragments than sequential digestion with the same enzyme (19). In addition, LtgA is localized to the septum, whereas LtgD is distributed around the gonococcal cell (Fig. 4) (19). Collectively, these results indicate that, whereas LtgA and LtgD may largely digest the same substrates, they each have specificities for substrates not acted on by the other enzyme, with LtgA having a wider substrate range than LtgD. Furthermore, LtgA produces more PG fragments than LtgD, although PG fragments generated by LtgA are more efficiently recycled by AmpG than those generated by LtgD. In contrast, a larger amount of LtgD-generated PG fragments is released.

FIG 4.

Model of PG fragment generation in N. gonorrhoeae. Model for PG fragment generation, release, and recycling by LtgA and LtgD (top right) and by the septal apparatus during cell separation (bottom right). Storm images (bottom left panels) demonstrate subcellular localization of LtgA and LtgD in N. gonorrhoeae. Scale bars = 1 μm.

GENERATION OF PG-DERIVED PEPTIDES AND DISACCHARIDE

N. gonorrhoeae has only one periplasmic N-acetylmuramyl-l-alanine amidase, AmiC, which cleaves septal PG between two daughter cells and therefore plays an important role in cell separation. In contrast, E. coli encodes four periplasmic amidases, in which three of the four have substantially redundant functions (57, 58). Such amidases cleave the bond between MurNAc and l-alanine, thereby severing the peptide stem from the glycan chain (Fig. 3). In E. coli and N. gonorrhoeae, AmiC activity is enhanced by the regulatory protein NlpD (11, 59–61). NlpD is proposed to bind to and displace an inhibitory loop to expose the AmiC active site, as has been described for the E. coli activator protein EnvC and amidases AmiA and AmiB (11, 60–62). Deletion of gonococcal amiC or nlpD results in a cell separation defect and increased rate of cell death, potentially due to a loss of outer membrane integrity in large bacterial clumps sharing an outer membrane (11, 20, 61).

In terms of PG fragment release, deletion of amiC in N. gonorrhoeae results in an increase in PG dimer release and abolishment of tetrasaccharide peptide, disaccharide, and free peptide release (11, 20). Purified gonococcal AmiC can digest whole sacculi, generating free peptides and cross-linked peptides, indicating that AmiC can act on cross-linked, multistranded PG, such as the thick septal PG (11). This result is consistent with literature for E. coli, where amidases have been demonstrated to localize to and cleave peptides at the septum during cell separation (63, 64). Purified gonococcal AmiC can also cleave a synthetic dimer to generate tetrasaccharide peptide and a free peptide, consistent with in vivo results that tetrasaccharide peptide release is abrogated in an amiC mutant, suggesting that AmiC may bind to at least one muropeptide moiety to cleave the peptide off an adjacent unit (11).

LtgC (E. coli homolog MltA) is an LT that plays a minor role in generating PG fragments and is necessary for cell separation. Deletion of ltgC results in large clumps of cells with irregular or thickened septa that are more sensitive to autolysis in N. gonorrhoeae and the closely related species N. meningitidis (65–67). In an ltgC mutant, disaccharide release is abolished, suggesting that LtgC may preferentially digest glycan strands that have been acted on by AmiC (54, 65). Mutants defective in ltgC release little to no tetrasaccharide peptide (65). The E. coli LtgC homolog can also digest naked glycan strands to generate free disaccharide (68). Collectively, these results suggest an order of operation for some of the PG degradation reactions in N. gonorrhoeae. AmiC removes stem peptides from PG strands, allowing LtgC to degrade the mostly naked glycan strands into disaccharides and tetrasaccharide peptide. LtgA is also localized to the septum (Fig. 4), but since it is not required for cell separation, it is unlikely to be directly involved with the septum-splitting proteins (19, 56).

GENERATION OF PEPTIDE STEM VARIATIONS IN PG FRAGMENTS

Trimming of peptide stems and cleavage of peptide cross-links are done by carboxypeptidases and endopeptidases, respectively (Fig. 3). N. gonorrhoeae encodes several peptidoglycanases that can perform either or both functions. Two low-molecular-weight penicillin binding proteins, PBP3 and PBP4, have d,d-carboxypeptidase activity and preferentially digest cross-linked peptides (69–71). PBP3 also has d,d-endopeptidase activity (69). N. gonorrhoeae that lacks PBP3 releases more PG dimers and more pentapeptide monomer and has more pentapeptide stems in the sacculi (72; unpublished observation). N. gonorrhoeae that lacks PBP3 and PBP4 releases fewer disaccharide and PG monomers, more PG multimers, and multiple unidentified PG fragment peaks, nearly all of which are larger than PG monomers (72). The double mutant also exhibits reduced viability and more irregularly sized (smaller or larger) cells (69, 72). Cumulatively, these data suggest that PBP3 and PBP4 perform overlapping but important functions in symmetric cell division and PG fragment release.

Zarantonelli et al. (73) examined a penicillin-insensitive strain of N. meningitidis and found that it had reduced d,d-carboxypeptidase activity, even though the genetic alteration to make it penicillin insensitive was in the gene for the biosynthetic transpeptidase PBP2. The sacculus composition showed decreased amounts of tetrapeptide stems and increased amounts of pentapeptide stems, as well as increased tetra-penta and tetra-tetra-penta cross-links (73). The penicillin-insensitive strain stimulated a smaller NOD1 response in human epithelial cells than the wild-type strain, suggesting that it released fewer PG monomers (73). To explain the decreased d,d-carboxypeptidase activity, PBP2 was investigated for a physical interaction with PBP3, and binding was detected (73). Overall, these results suggest that PBP3 activity is necessary for wild-type levels of proinflammatory PG fragment release.

Generation of tripeptide stems in gonococcal and meningococcal liberated PG fragments occurs through the action of the l,d-carboxypeptidase LdcA (74, 75). Mutation of ldcA in N. gonorrhoeae or N. meningitidis leads to an increase in the release of PG dimers and multimers (76, 77). PG monomers and free peptides released by an ldcA mutant are exclusively comprised of four-amino-acid versions (77). The increase in PG dimers and multimers released by the ldcA mutants compared with wild-type strains demonstrates the endopeptidase activity of LdcA. LdcA encoded by Neisseria species and closely related genera contains a Tat signal sequence and a lipidation site and functions as an outer membrane lipoprotein (77, 78). In E. coli and most other Gram-negative bacteria, LdcA is a cytoplasmic protein that plays a role in PG recycling by digesting tetrapeptides into tripeptides, which can then be directly attached to UDP-MurNAc to generate the PG precursor UDP-MurNAc-tripeptide (79).

RECYCLING OF PG FRAGMENTS AND IMPACT ON PG RELEASE

In Gram-negative bacteria, PG fragments liberated from the sacculi are usually efficiently transported back into the cytoplasm by the permease AmpG, broken down into smaller constituents, and utilized to synthesize new PG strands or for general metabolism (80). Deletion of ampG in N. gonorrhoeae, N. meningitidis, N. sicca, and N. mucosa leads to a large increase in the amount of PG monomers, an increase in the amount of disaccharide, but no obvious differences in the amount of PG dimers released compared to wild-type strains (2, 51, 74). It is not known whether AmpG can transport PG dimers, although our findings suggest that AmpG does not or is inefficient at transporting PG dimers. The role of AmpG in modulating the amount of PG fragments released is not limited to Neisseria species. Mutation of ampG results in 24- and 100-fold increases in the amount of TCT released by B. pertussis and V. fischeri (81, 82). In fact, the release of TCT by B. pertussis is a direct effect of the presence of an insertion sequence 85 bp upstream of the B. pertussis ampG coding sequence, reducing ampG expression (81). Deletion of ampG in E. coli results in release of a large amount of disaccharide with a small amount of monomer, suggesting that E. coli digests PG fragments extensively into smaller pieces in the periplasm (21). This result is reminiscent of the PG fragment release pattern of N. meningitidis, which releases fewer PG monomers and more anhydro-MurNAc than N. gonorrhoeae (2, 51).

Different species of Neisseria have different PG recycling efficiencies. Despite 97% amino acid identity between the AmpG sequence of N. gonorrhoeae and that of N. meningitidis, gonococcal AmpG is less efficient than meningococcal AmpG at recycling PG fragments (51). Three residues near the C-terminal end of AmpG cooperatively control the difference in recycling efficiency seen in N. gonorrhoeae and N. meningitidis. Mutations of these three residues from the gonococcal to the meningococcal variants decrease PG monomer release by N. gonorrhoeae by half. The reverse experiment increases the amount of PG monomer released by N. meningitidis by 40%. Gonococcal AmpG is also less efficient at recycling than the AmpG variants from N. sicca and N. mucosa (51).

AmpD is a cytoplasmic amidase that cleaves the peptide stem off PG monomers and anhydro-MurNAc-peptide to generate disaccharide or anhydro-MurNAc (83, 84). AmpD does not play a direct role in the generation of free peptides in the periplasm. However, a gonococcal ampD mutant releases fewer PG monomers and minimal amounts of disaccharide (74). In the absence of AmpD, N. gonorrhoeae increases the uptake and metabolism of disaccharide, thereby altering the amount of PG fragments released (74). This result suggests that N. gonorrhoeae can regulate PG fragment release in response to different levels of PG recycling intermediates in the cytoplasm. Accumulation of the recycling intermediate anhydro-MurNAc-peptide in an ampD mutant led to constitutive expression of the AmpC β-lactamase in a number of Gram-negative species, including Pseudomonas aeruginosa, Citrobacter freundii, Enterobacter cloacae, and Stenotrophomonas maltophila, conferring resistance against β-lactam antibiotics (85–89). Perturbation to the PG biosynthesis and recycling pathway due to β -lactam antibiotic treatment is thought to lead to the same induction of AmpC expression. Although Neisseria species do not have the AmpC β-lactamase system or a homolog of the PG-binding transcriptional regulator AmpR, perturbations to the recycling pathway can lead to alteration of the amount of PG fragments released by the bacteria. Given that a gonococcal ampD mutant amasses anhydro-MurNAc-peptide in the cytoplasm, PG fragment recycling intermediates may play a role as messenger molecules for regulating gene expression of one or more PG recycling enzymes in gonococci through an uncharacterized mechanism (74).

MODEL OF PG FRAGMENT GENERATION IN N. GONORRHOEAE

The analyses of PG fragment release by different PG degradation mutants allow us to propose a model for PG fragment generation and release (Fig. 4). The PG strands are first deacetylated by the PG de-O-acetylase ApeI. LtgA and LtgD then act to generate PG monomers and PG dimers by degrading the PG strands from the GlcNAc end and toward the anhydro-MurNAc found at the end of each strand (90). Degradation of strands to PG monomers requires an endopeptidase, likely PBP3 or PBP4, to remove peptide cross-links. Before their release from the cell, some of the PG fragments are acted on by LdcA, which converts a majority of PG monomers into the tripeptide monomer form that serves as an NOD1 agonist. LdcA also degrades some of the peptide-linked dimers to PG monomers. LtgA is localized to the septum, and most of the PG fragments it produces are shuttled into the cytoplasm for recycling. It is unclear why PG fragments produced by LtgA are so efficiently recycled, but we hypothesize that the PG fragment permease AmpG might be associated with LtgA via interactions with an unknown protein (Fig. 4, hyp protein). The proteins that function in cell separation act to create the free sugars and free peptides that are released by gonococci. An endopeptidase together with AmiC acts to cut cross-links and remove peptide stems from the PG strands. LtgC degrades the mostly naked PG strands into free anhydro-disaccharide plus tetrasaccharide peptide. Some of the free peptides that are liberated are acted on by LdcA to create free tripeptides before they are released into the milieu. On digestion of existing PG strands, the PG biosynthetic machinery acts to build new strands of PG for cellular enlargement (91, 92).

DISCUSSION AND FUTURE DIRECTIONS

Neisseria species may have been particularly well suited to evolve PG fragment release that is immunostimulatory in the human host. Neisseria species and closely related genera localize LdcA to the outer membrane and thus generate NOD1 agonists at the outer membrane rather than in the cytoplasm. Furthermore, the LTs that generate PG monomers and dimers are also localized to the outer membrane, and, at least for LtgD, this localization increases PG monomer release (19). With only a few changes to the AmpG sequence, the efficiency of the N. meningitidis AmpG permease is decreased to match that of N. gonorrhoeae (51). The resulting 3-fold increase in toxic PG monomer release is sufficient to increase NOD1 signaling and inflammatory cytokine production (2). The other changes to PG fragment breakdown that result in N. gonorrhoeae releasing mostly PG monomers and N. meningitidis or nonpathogenic Neisseria releasing more degraded PG fragments have yet to be determined.

The widely used model for PG synthesis and degradation in rod-shaped bacteria proposes that there are two sets of differentially localized enzymes, the elongasome and the divisome, which promote cell enlargement and cell division, respectively (91, 92). Conversely, cocci have been assumed to have only one septally localized machinery. Recent discoveries on PG degradation and release in Neisseria species suggest that Neisseria have septal and dispersed PG apparatuses. A septal apparatus facilitates cell separation. PG fragments generated at the septum, for the most part, are taken up into the cytoplasm for recycling. The dispersed PG apparatus is likely to play a role in cell enlargement and liberates PG fragments that are mostly released rather than recycled. Additional work has to be done to determine the components and functions of these machineries, but with the relatively few PG degradation proteins present in Neisseria species, determination of their locations and functions seems possible.

There is still much that we do not understand about the breakdown and release of small PG fragments. It appears that N. gonorrhoeae has the ability to sense PG fragments and regulate PG fragment release, but the mechanisms involved are not understood. N. gonorrhoeae that reaches the late-log to stationary phase does not show free disaccharide release, suggesting that this molecule is either consumed or not released as the bacteria experience nutrient deprivation. Similarly, ltgA ltgD mutants produce free disaccharide and some PG monomers that are present in the periplasm but not released, and ampD mutants make free disaccharide that is not released (48, 74). These results suggest that gonococci can sense the decrease in certain PG fragments in the cytoplasm in these recycling-deficient mutants and respond by increasing PG fragment uptake. This increased uptake was directly demonstrated for the ampD mutant (74). N. gonorrhoeae does not encode an obvious homolog of the PG-binding transcriptional regulator AmpR, suggesting that a different mechanism exists in these bacteria for regulating PG metabolism (93). Transcription of ltgA is positively regulated by the transcriptional regulator MtrR, suggesting that not only PG fragment recycling but also PG fragment generation may be controlled (94). If gonococci are able to regulate PG fragment breakdown and release, then changes in PG release might partly explain the differences in inflammatory responses seen in different forms of gonococcal infections.

Study of PG breakdown may lead to new antibiotic strategies. Breakdown of PG strands precedes the incorporation of new PG strands and is thus a necessary process (91). Also, changing the way the PG is broken down makes the bacteria particularly sensitive to host defenses. In studies of PG acetylation, Veyrier et al. (95) studied N. meningitidis mutants that cannot remove the acetyl groups from MurNAc and thus cannot degrade PG strands using LTs. They found that these mutants were attenuated in a mouse model of infection. Similarly, Ragland et al. (48) found that ltgA ltgD mutants of N. gonorrhoeae are highly susceptible to killing by human neutrophils. A recent report demonstrated that the natural product bulgecin A binds and inhibits LtgA (96). Furthermore, bulgecin A treatment was sufficient to make penicillin-insensitive N. gonorrhoeae strains susceptible to a number of β-lactam antibiotics. Therefore, LTs and other PG breakdown enzymes are attractive targets for further study and future antibiotic therapy.