Bacterial virulence regulation through soluble peptidoglycan fragments sensing and response: knowledge gaps and therapeutic potential

Abstract

Given the growing clinical–epidemiological threat posed by the phenomenon of antibiotic resistance, new therapeutic options are urgently needed, especially against top nosocomial pathogens such as those within the ESKAPE group. In this scenario, research is pushed to explore therapeutic alternatives and, among these, those oriented toward reducing bacterial pathogenic power could pose encouraging options. However, the first step in developing these antivirulence weapons is to find weak points in the bacterial biology to be attacked with the goal of dampening pathogenesis. In this regard, during the last decades some studies have directly/indirectly suggested that certain soluble peptidoglycan-derived fragments display virulence-regulatory capacities, likely through similar mechanisms to those followed to regulate the production of several β-lactamases: binding to specific transcriptional regulators and/or sensing/activation of two-component systems. These data suggest the existence of intra- and also intercellular peptidoglycan-derived signaling capable of impacting bacterial behavior, and hence likely exploitable from the therapeutic perspective. Using the well-known phenomenon of peptidoglycan metabolism-linked β-lactamase regulation as a starting point, we gather and integrate the studies connecting soluble peptidoglycan sensing with fitness/virulence regulation in Gram-negatives, dissecting the gaps in current knowledge that need filling to enable potential therapeutic strategy development, a topic which is also finally discussed.

This review gathers and integrates the evidence of existence of a soluble peptidoglycan-dependent signaling able to impact Gram-negatives virulence, discussing the current knowledge gaps and therapeutic potentials related to this topic.

Introduction

One of the most important threats to public health currently is the phenomenon of antibiotic resistance, which has a great impact in the context of nosocomial infection. This worrying menace has obvious clinical and also economic consequences that could eventually lead to our healthcare systems collapsing (Gandra et al. 2014, Friedman et al. 2016). This situation is even more severe in the current scenario of the SARS-CoV-2 pandemic, since it results in a significant increase in hospital and ICU admissions, with a consequent rise in the numbers of nosocomial infections and antibiotic resistance (Lai et al. 2021, Segala et al. 2021).

Antibiotic resistance is not only a great enemy per se during a given infection, but also from a more general point of view: if antibiotics lose their effectiveness, many modern medical techniques will become very risky, in essence taking us back to the preantibiotic era in which, for instance, any surgical process was a severe threat to the patients’ life. Additionally, any immunosuppressive treatment will be virtually eliminated from consideration if we run out of effective antibiotics (Teillant et al. 2015). In fact, the WHO has warned of the dramatic consequences that antibiotic resistance will entail if no additional measures are implemented to contain this phenomenon: by 2050, deaths due to antibiotic-resistant infections may reach 10 million per year. This silent pandemic is obviously more dangerous when affecting certain bacterial species that stand out because of their virulence, dissemination capacity, and morbidity–mortality rates. These pathogens are, therefore, those for which the development of new therapeutic options is more urgent, and some of them have been brought together in the ESKAPE group (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp.) (Tacconelli et al. 2018, de Oliveira et al. 2020), posing a paradigmatic example (among others that could be cited) of the antibiotic resistance phenomenon.

Since the development of new conventional antibiotics is quite scarce (with encouraging exceptions such as certain new combinations of β-lactams/β-lactamase inhibitors; Yahav et al. 2020), research has been pushed to explore other imaginative alternatives. For instance, therapies based on nanoparticles carrying different bactericidal agents, natural and modified antimicrobial peptides, phages (or their derived products such as lysins), those aimed at interfering with bacterial resistance to rehabilitate classic antibiotics, and those intended to attenuate microbial virulence, are currently under development (Theuretzbacher and Piddock 2019, Wang et al. 2020). Among the latter, called antivirulence therapies, different initiatives are being followed, and regardless of the specific target, they are intended to reduce the pathogenic power of the bacterium, thereby increasing the possibilities of clearance by the patient’s immune system, and decreasing the damages and clinical consequences caused by the infecting microbe (Maura et al. 2016, Fleitas Martínez et al. 2019, Hotinger et al. 2021).

Related to the search of targets for antivirulence therapies development, one of the bacterial elements intimately linked to pathogenesis is peptidoglycan (PGN) and its metabolism, since they are essential for cell viability and other bacterial needs. In this sense, this kind of exoskeleton acts by counteracting osmotic pressure, providing shape to the bacterium, enabling cell division, anchoring structures that are essential for pathogenesis such as flagella or toxin secretion systems, and accumulating modifications leading to resistance against certain immune attacks such as lysozyme, bile, or antimicrobial peptides (Juan et al. 2018). Obviously, if any of these PGN-related functions/features is altered, bacterial pathogenic power will be significantly impaired, making these processes potential antivirulence targets. Additionally, not the entire PGN as a structure, but the release of cleaved fragments into the extracellular medium has also been related to the inflammation-dependent pathogenesis of certain infections (such as those caused by Bordetella pertussis or Neisseria spp.; Rosenthal et al. 1987, Woodhams et al. 2013, Chan and Dillard 2016), and to other bacteria–host or bacteria/fungus interactions (Irazoki et al. 2019). Hence, PGN is not only an excellent target for classic antimicrobials but could also pose a promising virulence-linked Achilles’ heel, although it has been barely investigated in the latter sense.

Beyond the structural or inflammatory points of view, PGN displays well-known processes of degradation/remodeling and recycling that provide soluble fragments (receiving the generic denomination of muropeptides or anhydro-muropeptides depending on the context), which take part in the modulation of bacterial behavior, for instance by controlling the production of several β-lactamases (Park and Uehara 2008, Juan et al. 2017b, Dik et al. 2018). This well-known phenomenon provides a clear proof of regulatory capacity shown by certain PGN fragments, usually through binding to transcriptional regulators (mostly of the LysR family) and/or two-component systems (Juan et al. 2017b). This is one of the starting points on which we lean to make a thorough review of current knowledge dealing with interconnections between release/sensing of PGN fragments and fitness/virulence regulation in Gram-negative pathogens (main findings summarized in Table 1), since it is a relatively unknown topic that shows interesting therapeutically exploitable clues.

Table 1.

Main phenomena suggesting the existence of PGN-derived signaling impacting the fitness/virulence of Gram-negatives.

Sensor/regulator	Species	Phenomenon	Effector PGN fragment	References
AmpR (LysR-type regulator)	K. pneumoniae	The presence of AmpR regulators codified together with horizontally acquired class C β-lactamases is associated with increased capsule synthesis and virulence	Unknown	Hennequin et al. (2012), Zhang et al. (2020)
AmpR (LysR-type regulator)	P. aeruginosa	Activator role for acute virulence factors and repressor for biofilm formation	Unknown	Balasubramanian et al. (2011, 2012, 2015)
AmpR (LysR-type regulator)	S. maltophilia	Repressor for the production of the principal quorum sensing signal (DSF), impairing biofilm formation, oxidative stress resistance, and general virulence	Unknown	Alcaraz et al.(2022)
CreBC (two-component system)	P. aeruginosa	Major role contributing to bacterial fitness and biofilm development, especially through its effector inner membrane protein CreD, proportionally expressed with regards to CreBC operon	Unknown	Zamorano et al. (2014)
CreBC (two-component system)	S. maltophilia	Maintenance of cell envelope integrity, promotor of protease secretion, and repressor of motility. Repressor for CreD expression	Unknown	Huang et al. (2015a, b)
CpxAR-CpxP	Multiple species	Multiple studies demonstrating that CpxA is able to sense muropeptide alterations in the periplasm, and others that CpxR is a bona fide regulator of virulence-related genes. Depending on the study/species, the effect of CpxR seems to be a promotor vs. repressor of bacterial pathogenesis. Proposed mechanism: under certain conditions (e.g. β-lactam challenge), the soluble muropeptide pool in the periplasm allegedly changes, and then the differentially accumulated fragments could competitively displace CpxP from CpxA EBD (to which it is bound in regular conditions so as to keep the system switched off). This circumstance would enable CpxA autokinase activity and the subsequent phosphorylation of CpxR, finally promoting/repressing the expression of genes under its control	Unknown	Humphreys et al. (2004), Fleischer et al. (2007), MacRitchie et al. (2008), Mitobe et al. (2011), Liu et al. (2012), Debnath et al. (2013), Weatherspoon-Griffin et al. (2011), Tschauner et al. (2014), Bontemps-Gallo et al. (2015), Acosta et al. (2015), De la Cruz et al. (2015, 2016), Tanner et al. (2016), Gangaiah et al. (2017), Li et al. (2018), Xie et al. (2018), Subramaniam et al. (2019), Thanikkal et al. (2019), Vogt et al. (2019), Masi et al. (2020), Wu et al. (2021)
AgtR (?)	P. aeruginosa	Increased transcription of the pqs operon, responsible for the quorum sensing signal PQS, which in turn promotes the production of pyocyanin and proteases	GlcNAc alone or larger GlcNAc-containing PGN fragments, providing phosphorylated GlcNAc derivatives through NagE phosphotranferase	Korgaonkar and Whiteley (2011), Korgaonkar et al. (2013), Lee and Zhang (2015)
LuxR-type receptors (binding site for quorum sensing signals)	P. aeruginosa, C. violaceum, and E. coli	Partial inhibition of the transcription of genes controlled by quorum sensing-linked LuxR-type regulators, including important virulence-related actors such as elastases and proteases	Phosphorylated GlcNAc derivatives provided by NagE	Lee and Zhang (2015), Kimyon et al. (2016)
Unknown	E. coli	Downregulation of two extracellular adhesion factors: type 1 fimbrial adhesins and extracellular curli fibers. Decrease of the elicited inflammatory response, thus promoting dissemination	GlcNAc-6-phosphate	Sohanpal et al. (2004, 2007), Barnhart et al. (2006), Naseem and Konopka (2015)
Unknown	E. coli	Prevention of quiescence status that appears at low population density. Therefore, promotion of quorum sensing-independent growth	Tetra- or penta-peptide stem chains, both from Gram-positive or -negative bacteria	DiBiasio et al. (2020)
Unknown	E. coli, P. aeruginosa	Rescue cells from stationary phase, resuming growth even in a scenario of nutrient scarcity	1,6-anhydro-MurNAc-GlcNAc-peptides	Jõers et al. (2019)
Unknown	Salmonella enterica, R. solanacearum	Accumulation of muropeptides due to cytosolic AmpD amidase disruption leads to reduced fitness and virulence attenuation	Unknown	Tans-Kersten et al. (2000), Folkesson et al. (2005)
Unknown	X. campestris	Accumulation of muropeptides due to cytosolic NagZ glucosaminidase disruption leads to virulence attenuation	Unknown	Yang et al. (2014)

‘?’ indicates that this is the most plausible element involved in the PGN sensing/response, although not fully demonstrated. Abbreviations; DSF: diffusible signal factor; EBD: effector-binding domain; GlcNAc: N-acetyl glucosamine; PGN: peptidoglycan; and PQS: 2-heptyl-3-hydroxy-4-quinolone.

Thus, after a general review of the Gram-negative PGN metabolism and the main transcriptional regulators and two-component systems showing evidence of their capacity for muropeptide responsiveness (regardless of their linkage or not with β-lactamase regulation), we gather and integrate other studies suggesting the existence of soluble PGN-related signaling leading to modulation of fitness/virulence. We take into account the phenomena involving muropeptides released from the bacterium’s own cell wall and also those related to PGN fragments from the external medium, thus providing the double perspective of intra- and intercellular PGN-derived signaling. We finally discuss the possibilities of therapeutic development based on all the evidence reviewed, emphasizing the gaps in our knowledge that need filling in order to achieve the goal of antivirulence therapies development.

PGN metabolism as a source of signaling molecules to modulate bacterial behavior

An overview of the Gram-negative PGN structure together with some of its most important building/turnover reactions is shown in the Fig. 1 to facilitate understanding of this manuscript. Nonetheless, for further information regarding PGN structure and metabolism the reader may turn to excellent reviews (Park and Uehara 2008, van Heijenoort 2011, Johnson et al. 2013, Shaku et al. 2020).

Figure 1.

General structure of Gram-negative PGN, with the main periplasmic hydrolase cleavage sites. The sugar backbone of PGN is composed of repeated MurNAc–GlcNAc dissacharides, constituting large glycan chains. These are elongated through the incorporation of new PGN units (disaccharide–pentapeptides) into the sacculus, performed by PBPs with a glycosyl transferase domain (yellow star, formation of β(1–4) glycosidic bonds]). Transpeptidation performed by PBPs enables the cross-linking (green horizontal bar) of lateral peptides (in turn linked to MurNAc), usually through a bond between the DAP of one peptide and the fourth d-Ala of the other (although cross-links between two DAPs, for instance, may also appear). The blue lightning bolt represents carboxypeptidase activity, usually performed by the same PBPs during the transpeptidation process, whereas the green lightning bolts represent different endopeptidase activity variants. The yellow lightning bolt represents the lytic tranglycosylase cleavage site, whereas the orange one represents N-acetylmuramyl-l-alanine amidase activity. Abbreviations: Ala: alanine; DAP: meso-di-aminopimelic acid; Glu: glutamic acid; GlcNAc: N-acetyl-glucosamine; MurNAc: N-acetyl-muramic acid; PBP: penicillin-binding protein; and PGN: peptidoglycan.

Bacterial PGN, also known as murein sacculus, is a semirigid structure that, although it may only represent 2% of the cell mass in Gram-negatives, is essential for cell viability and virulence because of the abovementioned reasons (Park and Uehara 2008, Juan et al. 2018). It is a mesh like-structure composed of glycan chains [repetitions of N-acetyl-muramic acid (MurNAc) and N-acetyl-glucosamine (GlcNAc) disaccharides] connected through cross-linked lateral stem peptides in turn bound to MurNAc residues. These lateral peptides are initially made up of 5 amino acids [pentapeptides (P5): l-alanine-d-glutamic acid-meso-diaminopimelic acid (DAP)-d-alanine-d-alanine in most Gram-negatives] when the new building units are incorporated into the nascent PGN. However, through transpeptidation (the process that allows the cross-linking between peptide chains, performed by penicillin-binding proteins; PBPs) and other fine-tuning reactions, distal d-alanine(s) are usually cleaved to provide lateral tri- (P3) or tetra-peptides (P4) in the mature sacculus (Park and Uehara 2008, Dik et al. 2018). Meanwhile, glycan chains are elongated due to the glycosyl transferase domain of certain PBPs, which enables the incorporation of new PGN monomers (GlcNAc–MurNAc–pentapeptide units) provided by lipid II (undecaprenyl–pyrophosphate–GlcNAc–MurNAc–pentapeptide). This PGN precursor, embedded in the inner membrane, is the last step of the cytosol PGN anabolic reactions that provide material to be incorporated into the nascent sacculus. For more information regarding the final steps of PGN building, some reviews may be resourced (Egan et al. 2015, Malin and de Leeuw 2019) (Fig. 1).

PGN is not a static element but rather the opposite. In fact, bacteria continuously degrade their PGNs through tightly controlled enzymes, generically known as PGN hydrolases to allow cell size increase, septation and division, and anchorage of different structures. Among the activities performed by these periplasmic enzymes, which are ultimately the driving force to provide PGN-derived soluble signals, it is worth highlighting (Fig. 1): (i) lytic transglycosylases (cleave the bond between GlcNAc and MurNAc, resulting in the formation of 1,6-anhydro-MurNAc products, called anhydromuropeptides); (ii) endopeptidases (break the cross-linkages between two stem peptides or act within a unique peptide chain depending on the enzyme variant); (iii) carboxypeptidases (remove the C-terminal amino acids in stem peptides); and (iv) N-acetylmuramyl-l-alanine amidases (cleave the bond between stem peptides and the MurNAc units in the sacculus). All these enzymes enable a controlled degradation of PGN, which in Escherichia coli for instance reaches ca. 50% in each generation (Park and Uehara 2008, van Heijenoort 2011, Johnson et al. 2013). However, it has been estimated that only ca. 6%–8% of the degraded fragments are lost to the external medium. Thus, more than 90% of cleaved PGN is efficiently recovered through so-called recycling pathways in the cytosol, which enable the reincorporation of released fragments into anabolic routes to reconstruct the sacculus (Park and Uehara 2008, Irazoki et al. 2019, Shaku et al. 2020). Although the recycling of PGN has been interpreted as an energy saving strategy, its real function is not completely clear since, for instance, it is not essential under laboratory experimental conditions. Thus, although internalized PGN fragments could theoretically work as carbon and energy sources allowing growth under nutrient-limiting conditions, there is no strong evidence supporting this idea, at least in Gram-negatives (Irazoki et al. 2019). Hence, the biological reasons for the aforementioned high level of PGN turnover could go beyond purely structural and/or energy issues, having a meaning related to the internal/external signaling and modulation of bacterial behavior depending on the conditions (Shaku et al. 2020), an idea that fits perfectly with the topic of this review.

Although species-specific particularities may appear, most of the fragments released through the abovementioned PGN hydrolases-mediated degradation are 1,6-anhydro-MurNAc–GlcNAc disaccharides bound to variable length stem peptides (mostly tetrapeptides) that reach the cytosol through specific permeases such as the archetypical AmpG (Park and Uehara 2008, Torrens et al. 2019a). Additionally, free stem peptides derived from the action of periplasmic amidases are also believed to reach the cytosol through the oligo-peptide permease Opp, although quantitatively speaking, it was estimated that this pipeline of material for recycling is minor (Park 1993). Once inside the cytosol, some relevant enzymes that metabolize anhydromuropeptides are: (i) the 1,6-anhydro-MurNAc-l-alanine amidase AmpD, which cleaves the bond between the stem peptide and 1,6-anhydro-MurNAc; (ii) the N-acetyl-glucosaminidase NagZ, which releases GlcNAc from the 1,6-anhydro-MurNAc–GlcNAc disaccharides, and (iii) the l, d-carboxypeptidase LdcA, that cleaves the distal d-Alanine of stem tetrapeptides. Therefore, after the performance of these enzymes, free units of 1,6-anhydro-MurNAc, GlcNAc, and tripeptides are mostly obtained in the cytosol and enter different anabolic reactions that enable the synthesis of new PGN building units. Among these reactions, for instance the pathways for the reutilization of free 1,6-anhydro-MurNAc providing uridine diphosphate (UDP)–MurNAc, or that mediated by the murein peptide ligase Mpl, mediating the generation of UDP–MurNAc–tripeptides, have been shown to be crucial for the correct synthesis of the final lipid II PGN precursor (Park 1993, Templin et al. 1999, Park and Uehara 2008, Dik et al. 2018, Fisher and Mobashery 2020). Interestingly, some of the PGN intermediaries accumulated into cytosol, proceeding from the abovementioned catabolic/anabolic reactions display intracellular signaling tasks mostly in the context of β-lactamase regulation as shown below, reinforcing the idea of a soluble PGN sensing/response as a very relevant mechanism for bacterial behavior regulation.

LysR-type regulators: well-known systems of soluble PGN sensing-response in the context of β-lactamase regulation and beyond

As seen throughout this review, the information available regarding PGN metabolism-linked β-lactamase regulation is more abundant than that connecting potential PGN-derived signaling with fitness/virulence modulation, evidencing that this is an insufficiently investigated field despite the interesting clues published. A perfect example of this circumstance is the well-known role that specific LysR family transcriptional regulators play in modulating the production of some specific Gram-negative β-lactamases, although they are the most abundant type of transcriptional regulators in prokaryotes and control an almost endless array of genes/functions beyond β-lactam resistance (Schell 1993, Balcewich et al. 2010).

Returning to the β-lactamase context, the mechanism exerted by specific representatives of these LysR regulators could be considered the classic model under which Ambler’s class C intrinsic enzymes from different Enterobacteriaceae species are controlled, and whose conservation in some other relevant Gram-negative pathogens such as P. aeruginosa was later demonstrated (Ambler 1980, Lindberg et al. 1985, 1987, Lindquist et al. 1989, Jacobs et al. 1994, 1997, Dietz et al. 1996, 1997, Langaee et al. 1998). LysR-type regulators also modulate the production of many other enzymes in Gram-negatives (although with certain particularities in some cases), such as the intrinsic class A β-lactamases of Burkholderia cepacia complex (BCC), Proteus vulgaris, Rhodopseudomonas capsulatus, and Citrobacter diversus, as well as the intrinsic blaL1 and blaL2 from Stenotrophomonas maltophilia (class B and A, respectively) (Juan et al. 2017b). What is more, some horizontally acquired β-lactamases, such as some variants of blaACT, blaCMY, blaSFO, blaSME, blaNMC, blaDHA, and blaIMI among others, are often codified in mobile elements together with their corresponding LysR-type regulators (Datz et al. 1994, Fernández et al. 2012, Seral et al. 2012, Juan et al. 2017b) and would, therefore, be under the control of the inducible mechanism explained below.

The genes of these examples of β-lactamases are located in so-called divergons, i.e. regions in which the LysR-type regulator—sometimes named ampR (penR in BCC)—and β-lactamase are divergently codified, having in-between a DNA segment of variable length (often below 150 nucleotides), with overlapping promoters. This intergenic region usually contains a notably conserved high affinity-binding site for the LysR-type regulator (so-called T-N₁₁-A motif, mostly palindromic), together with other lower affinity and less conserved site(s) (Schell 1993, Maddocks and Oyston 2008). The regulator interacts with these sites through a conserved N-terminal DNA-binding domain that forms a helix-turn-helix motif, whereas an effector-binding domain (EBD) is present at the C-terminal end of the protein. This EBD displays a cleft-like structure to which different soluble molecules, called effectors or coactivators, can eventually bind to impact the role of the regulator. Thus, depending on the ligand that the regulator acquires, its tridimensional conformation changes, even enabling the formation of oligomers once the regulator is already interacting with the DNA. These complexes are then capable of repressing/promoting β-lactamase gene transcription through bending/relaxation of DNA and derived differential interactions with RNA polymerases (Schell 1993, Maddocks and Oyston 2008, Balcewich et al. 2010). Obviously, in relation to the topic of this review, AmpR-like regulators are believed to acquire these different roles depending on the soluble PGN fragment bound to the EBD (Vadlamani et al. 2015, Dik et al. 2017, 2020). Various reviews may be resourced for further details about these LysR regulators and their specific relationship with β-lactamase regulation and PGN metabolism (Schell 1993, Maddocks and Oyston 2008, Balcewich et al. 2010, Juan et al. 2017b, Dik et al. 2018), but a short summary is developed in the following paragraphs and graphically displayed in Fig. 2.

Figure 2.

General model for the regulation of β-lactamases under the control of LysR-type regulators, applicable for the intrinsic enzymes of P. aeruginosa, Enterobacteriaceae, BCC,S. maltophilia, and others. On the left side of the figure, the linkage between the LysR regulator (here called AmpR) function and PGN-derived fragments (muropeptides) is shown, including a basal situation, as well as induction and mutation-driven β-lactamase hyperproduction scenarios. The regularly generated muropeptides proceeding from a basal PGN turnover are represented as clear blue cubes. Those muropeptides differently appearing and accumulating in qualitative/quantitative terms during induction or in a mutational hyperproduction pathway are represented as white cubes. On the right side of the figure, the concomitant activation of the CreBC system contributing to AmpC-dependent resistance output is shown, displaying the particularities of S. maltophiliavs.P. aeruginosa. Abbreviations. OM: outer membrane; PGN: peptidoglycan; and IM: inner membrane.

As mentioned above, in regular conditions bacteria perform a continuous turnover of their PGNs, by a tightly regulated process of degradation, recycling, and reconstruction. Once in the cytoplasm, the anhydro-muropeptides proceeding from PGN cleavage are incorporated into so-called recycling pathways and reused for anabolic reactions, leading to PGN reconstruction together with purely de novo synthesis (Park 1993, Dik et al. 2018, Fisher and Mobashery 2020). Some of the PGN precursors synthesized following these routes, such as UDP–MurNAc–pentapeptides (UDP–MurNac–P5), besides being incorporated into the final nascent PGN routes (see above) are believed to bind the LysR-type regulator, which then takes on the abovementioned repressor role (Jacobs et al. 1994, 1997, Vadlamani et al. 2015). Therefore, in these conditions, the expression of intrinsic β-lactamase, called ampC in various Enterobacteriaceae species and P. aeruginosa for instance, is reduced to minimum levels. Meanwhile, certain β-lactams (so-called inducers, such as cefoxitin) cause an inhibition of low molecular weight PBPs entailing different alterations in the PGN sacculus, which end up influencing the amount/nature of released muropeptides (Torrens et al. 2019a). Thus, when bacteria face this aggression, these differential fragments reach the cytoplasm, cause the saturation of the anhydro-muramyl-l-alanine amidase AmpD and then overaccumulate displacing the repressor fragments such as UDP–MurNac–P5 from binding with AmpR. AmpD is key in this context since its activity (cleavage of stem peptides from 1,6-anhydro-MurNAc or from 1,6-anhydro-MurNAc–GlcNAc) is essential for the subsequent recycling pathways and, therefore, for the correct synthesis of new PGN precursors (including AmpR repressor signals). Returning to the AmpR element bound to the inducer muropeptide, this event causes the acquisition of an activator configuration of this regulator, transiently promoting the expression of β-lactamase until the β-lactam challenge disappears. In other words, when the inducer drug is hydrolyzed by the hyperproduced enzyme, PGN recovers its homeostasis, regular turnover, and basal generation of released muropeptides, and then the overaccumulated activator signals are gradually replaced by repressor PGN precursors, returning the transcription to basal levels (Fisher and Mobashery 2014, 2020, Dik et al. 2017, 2018, 2020).

Obviously, the chemical natures of the PGN fragments leading to a repressor or activator role of AmpR are different, and whereas there is a rather unanimous consensus regarding the repressor fragments (UDP–MurNac–P5), activator muropeptides are apparently variable depending on the species, and also on the pathway of hyperproduction (Uehara and Park 2002, Juan et al. 2017b). In the latter sense, besides the reversible induction process, it is usual in the clinical setting for treatment with other β-lactams (not inducers, such as ceftazidime or piperacillin, for instance) to select mutations that mimic the aforementioned situation of repressor fragment displacement by activator muropeptides in AmpR binding, but in a permanent way. In other words, the mutations selected due to the treatment pressure lead to a constitutive hyperproduction of β-lactamase that causes resistance to different β-lactams, and not only to those that are hydrolyzable by the enzyme when acting as inducers. Many different mutations have been described in this sense, not only in P. aeruginosa but also in other species of Enterobacteriaceae or BCC (Lindberg et al. 1987, Hwang and Kim 2015, Juan et al. 2017b), such as inactivating mutations in the abovementioned indirect ampC repressor ampD, in certain PBPs such as the low mass PBP4 (dacB), and amino acid changes providing a constitutive activator conformation of AmpR among others, eventually even affecting horizontally acquired β-lactamases (Cabot et al. 2012, Juan et al. 2017b, Jones et al. 2018) (Fig. 2).

Pseudomonas aeruginosa is probably the species in which the regulation of AmpC has been most deeply characterized, showing a differential accumulation of specific muropeptides acting as ampC activators when comparing the induction process vs the mutational ampD pathway (one of the most usually detected in clinical strains) or vs. other mutation-driven pathways of high level AmpC-dependent resistance (Lee et al. 2016, Juan et al. 2017b, Torrens et al. 2019a). Moreover, there are other qualitative particularities regarding PGN fragments allegedly activating the expression of intrinsic β-lactamases depending on the species besides P. aeruginosa, e.g. Citrobacter freundii, Enterobacter cloacae, Aeromonas hydrophila, and S. maltophilia, among which the length of the lateral stem peptide (P3, P4, or P5) and the presence/absence of GlcNAc bound to 1,6-anhydro-MurNAc are the most differential traits (Jacobs et al. 1994, 1997, Dietz et al. 1996, 1997, Tayler et al. 2010, Huang et al. 2017a). All these facts are related to the binding capacities of the AmpR regulators’ EBDs. For instance, muropeptides containing a terminal d-Ala-d-Ala motif (1,6-anhydro–MurNac–P5) seem to be much more effective than fragments with shorter lateral peptides for an activator binding to AmpR, although with some controversial results depending on the study (Jacobs et al. 1994, 1997, Dietz et al. 1996, 1997, Hanson and Sanders 1999, Vadlamani et al. 2015, Dik et al. 2018, Torrens et al. 2019a). In this sense, the AmpR regulators of C. freundii and P. aeruginosa are those that have been more deeply characterized: whereas the former apparently works as a cytosolic tetramer, the latter is proposed to be a dimer bound to the inner membrane (Balcewich et al. 2010, Caille et al. 2014, Vadlamani et al. 2015, Dik et al. 2018).

Returning to the PGN-derived signaling capable of modulating bacterial behavior in a more general sense, P. aeruginosa AmpR began to be dissected from a perspective beyond β-lactamase modulation almost 20 years ago, and since then it has been considered a bona fide global regulator (Kong et al. 2005, Balasubramanian et al. 2011, 2015). The extradivergon genetic repertoire apparently regulated by AmpR is very broad, consisting of hundreds of genes, turning out to be significantly different when comparing situations of presence vs absence of β-lactam challenge (Balasubramanian et al. 2012). This circumstance suggests that the genes regulated after challenge with β-lactams are more closely related to PGN fragment sensing since, as mentioned, these drugs cause an alteration of the sacculus composition through PBP inhibition and, therefore, of released muropeptides. However, it cannot be ruled out that the genes under the control of AmpR in the absence of β-lactams could also be responding to the generation and sensing of PGN fragments, which may be variable depending on other stimuli, growth phase, and so on. In fact, supporting this idea, it has been shown that depending on nutrient availability or under certain physical stress, PGN degradation/synthesis can be modulated in E. coli (through transcriptional regulators not related to AmpR, i.e. not naturally present in this species), which could provide differential amounts/natures of muropeptides with signaling and regulatory potentials (Shimada et al. 2013). Moreover, although noninducer β-lactams apparently do not affect AmpR to influence β-lactamase production, it cannot be ruled out that their action also brings about different alterations in the soluble muropeptide pool detected by AmpR EBD, which would respond by governing other sets of genes. Either way, AmpR was initially shown to affect P. aeruginosa pathogenic behavior due to its capacity for downregulating the production of pyocyanin and LasA protease and for promoting the expression of LasB elastase (all of them well-known virulence-related factors; Juan et al. 2017a) through the influence on the quorum sensing systems governed by RhlR and LasR regulators (Kong et al. 2005, Lee and Zhang 2015). Later, AmpR was also shown to affect other transcriptional regulators (including CreB for instance) and sigma factors (RpoS and AlgT/U), thus accounting for an extensive repertoire of directly/indirectly AmpR-controlled processes and suggesting a very complex interplay between potential soluble PGN sensing and modulation of the final bacterial phenotype (Balasubramanian et al. 2012, 2013, 2014). The later analysis of the proteome affected by the AmpR regulon as well as the demonstrated connection with small regulatory RNA networks and the modulation of several cyclic di-GMP phosphodiesterases (which in turn affect the accumulation of the virulence-linked signaling molecule cyclic di-GMP) added even more complexity to the actual role of this LysR element in P. aeruginosa, but also enhanced its interest as an antivirulence therapeutic target (Balasubramanian et al. 2013, 2014, 2015, Kumari et al. 2014, Hall and Lee 2018). Thus, in few words, P. aeruginosa AmpR has been described to work as a positive regulator of acute virulence while acting as a repressor of biofilm formation, one of the hallmarks of chronic infection phenotype (Balasubramanian et al. 2012, 2013, 2014, 2015), although whether or not these roles are dependent on muropeptide binding to the EBD remains elusive.

Although much less studied than in P. aeruginosa, AmpR has also been shown to be involved in the regulation of virulence in some other species that harbor this regulator. However, whether this regulation appears as a response following soluble PGN sensing by the EBD has not been investigated to date. Interestingly, in S. maltophilia AmpR has been shown to negatively influence the production of the principal quorum sensing signal (namely diffusible signal factor, DSF), which impairs biofilm formation, oxidative stress resistance, and virulence in invertebrate model (Alcaraz et al. 2022). Conversely, a decade ago it was reported in K. pneumoniae that the AmpR regulator coacquired with a plasmid-encoded blaDHA-1 enzyme promoted capsule synthesis and derived resistance to killing by the complement system. In this same paper, the capacity of AmpR to regulate biofilm formation and type 3 fimbrial gene expression was also demonstrated, with obvious impacts on adhesion to host tissues. More recently, the presence of AmpR in the genome of certain strains of carbapenem-resistant K. pneumoniae was reported to be associated with increased production of capsule and virulence too (Hennequin et al. 2012, Zhang et al. 2020).

PGN fragments-responsive two-component systems and modulation of virulence-related features

Similar to what has been explained for LysR-type regulators, two-component systems pose very well-known mechanisms to sense certain stimuli and respond with adequate adaptation to external conditions (Francis et al. 2017, Sultan et al. 2021, Wang et al. 2021). These systems, usually codified as operons, are very widely distributed among bacteria and consist of a membrane sensor kinase that detects certain stimuli (pH changes, redox potential, metabolites, pressure, altered proteins, and so on) and reacts by autophosphorylation, plus a cytosolic transcriptional regulator, i.e. activated through the phosphorylation executed by the cited sensor. Once phosphorylated, the transcriptional regulator impacts the expression of the genes under its control, whose promoters usually contain conserved regions where the regulator binds, finally providing an appropriate response for the sensed stimulus (Padilla-Vaca et al. 2017, Tierney and Rather 2019). Some representatives of these systems have been proposed to be PGN fragment-responsive in the context of β-lactamase regulation and resistance, although evidence of their capacity to modulate features closely related with fitness/virulence is also available and delved into in this section.

CreBC system

To start with, a mechanism complementing the AmpR-dependent regulation of intrinsic β-lactamases in certain species is the concomitant activation of the two-component system CreBC happening under specific circumstances. This is very worthy of being dissected here because of its likely role in connecting muropeptide sensing and phenotype regulation. This system displays significant implications beyond β-lactam resistance and has been deeply studied in E. coli, where it mostly works as a metabolic modulator in response to changes in the medium nutrients (which gives its denomination, from carbon source responsive). In this case, CreC is the inner membrane sensor, capable of responding to periplasmic stimuli, whereas CreB is the transcriptional regulator (Avison et al. 2001). In P. aeruginosa, CreBC is activated during β-lactam induction but more importantly after dacB (PBP4) mutational inactivation, resulting in a complex metabolic response that enhances resistance level—illustrated by an increase in minimum inhibitory concentrations for antipseudomonal β-lactams—derived from a given amount of produced AmpC, in turn not controlled by CreB but only by AmpR (Moya et al. 2009) (Fig. 2). It has been proposed that PBP4 thus works as a sentinel for PGN damage (i.e. the activity of an inducer β-lactam): its inhibition by the drug leads to changes in PGN that are somehow sensed by CreC, activating the subsequent responses of transcriptome modulation governed by CreB parallel to AmpC hyperproduction (Moya et al. 2009). In fact, the mutational inactivation of dacB is probably the most frequent mechanism of constitutive AmpC hyperproduction and resistance in P. aeruginosa clinical strains, and is mediated by the same activator muropeptides (1,6-anhydro–MurNac–P5) enabling induction (Juan et al. 2017b, Torrens et al. 2019a) (Fig. 2).

Nevertheless, the details as to how CreC can detect the cited alterations in PGN composition have not yet been characterized, although they probably have a lot to do with differential generation of periplasmic muropeptides, since in regular conditions PBP4 cleaves the terminal d-Ala of pentapeptides within mature PGN among other roles (Ropy et al. 2015). Thus, if some alteration in the sacculus happens because of the absence/inhibition of PBP4, this should be reflected in the muropeptides cleaved and released, and this circumstance could then be sensed by a potential EBD located in the inner membrane sensor CreC. However, the existence of a CreC EBD capable of binding muropeptides has not been specifically demonstrated, although the data available strongly suggest this possibility. The idea of a muropeptide-binding cleft in CreC could be additionally supported by the fact that the ampD inactivation-mediated route of AmpC hyperproduction does not activate CreBC, likely because its derived muropeptide accumulation alteration only affects the cytosol (where AmpD works) and not the periplasm (where the sensor CreC should detect soluble PGN fragment alterations) (Moya et al. 2009). Consequently, the fact that CreBC is not activated in ampD-defective mutants provides a lower resistance output for a similar amount of produced AmpC compared with the dacB mutational pathway (Moya et al. 2009, Torrens et al. 2019a).

As expected, in P. aeruginosa it has been shown that CreB can affect the expression of several genes (related to fitness and biofilm formation among other features) participating in the complex metabolic response that improves AmpC-dependent resistance. This response has been described to be intimately linked to the effector inner membrane protein CreD (whose expression is used as a measure of CreBC activation, although its actual role is unknown) and when the bacterium faces subinhibitory concentrations of β-lactams. This circumstance could once more be indicative of the CreC capacity for sensing differentially accumulated muropeptides in the periplasm and activating CreB in response to the alterations of PGN caused by the drug, or even other stimuli (Zamorano et al. 2014). Moreover, a certain level of control capacity over genes related to virulence, such as those of the exoenzyme S, the fimbrial precursor PilA, or various proteins related to the synthesis of phenazines (aromatic metabolites among which the cytotoxic and inflammatory pyocyanin stands out; Hall et al. 2016, Higgins et al. 2018) has also been demonstrated. CreBC influence on genes of the nitrogen regulatory pathway such as the NAR operon (key for the development of biofilms), as well as the NIR and NOS clusters has also been reported, suggesting a capacity to impact motility, biofilm formation, and virulence (Palmer et al. 2007, Van Alst et al. 2007, Zamorano et al. 2014).

In the case of S. maltophilia, in which its two intrinsic β-lactamases blaL1 and blaL2 are under the control of a unique AmpR-type regulator (although blaL2 is the only one codified in the divergon, and therefore, apparently more responsive to the regulator than blaL1), some particularities have been found. For instance, (i) the fact that all β-lactams act as inducers, and therefore, trigger a virtually constitutive β-lactamase hyperproduction and resistance to these drugs; and (ii) activation of the CreBC response seemingly impacts the expression of several genes, in particular differentially promoting the expression of certain periplasmic PGN hydrolases (MltB1 and MltD2), which in turn cause an increased turnover and accumulation of AmpR-activating muropeptides (Huang et al. 2015a, Juan et al. 2017b). Obviously, this circumstance enhances the production of the enzymes blaL1 and blaL2 to an even greater extent, with this phenomenon apparently being exclusive to this species (Fig. 2).

Still in S. maltophilia, the works regarding the CreBC system have been intimately linked to the study of the inner membrane effector CreD, which displays very interesting differences compared to its homologues from other species. In this sense, in P. aeruginosa and E. coli the expression of creD is positively regulated by the activated CreBC system, and the same happens for blrD, the homologue of creD in Aeromonas spp., promoted through the BlrAB response (a system homologue of CreBC, see below). Conversely, in S. maltophilia, although the genetic organization is very similar to that of E. coli and P. aeruginosa (creD located a few nucleotides downstream of creBC), the expression of creD is not proportional to the CreBC response, but rather the opposite: i.e. CreBC acts as a repressor for creD expression. Moreover, unlike CreD in P. aeruginosa or blrD in Aeromonas spp., expression of this inner membrane protein is not activated in the presence of inducer β-lactams, but rather is positively regulated by bacterial density. Furthermore, specifically in S. maltophilia, CreD has been shown to have an essential role for the maintenance of cell envelope integrity whereas the CreBC system seemingly works together with an unidentified additional response regulator as a kind of promotor of protease secretion and repressor of motility, which has obvious virulence-related implications (Huang et al. 2015a, b).

Contrary to P. aeruginosa and S. maltophilia, a potential role of CreBC in other Gram-negatives on β-lactamase-dependent resistance has not yet been studied, although this system is codified in the genome of many representatives of Enterobacteriaceae or in BCC for instance and, as such, a potential interaction with muropeptides and capacity for regulating bacterial behavior, fitness, and virulence cannot be ruled out (Zeng and Lin 2013, Juan et al. 2017b).

BlrAB system

In the opportunistic pathogen Aeromonas spp. there is a typical two-component system highly homologous (60%–70%) to CreBC, i.e. also involved in resistance (Juan et al. 2017b). This system is called BlrAB (denomination coming from “β-lactam resistance”) and controls the expression of various intrinsic β-lactamases in this genus. However, unlike what happens with the previously explained enzymes, in Aeromonas spp. there is no LysR-type element, and thus the control of β-lactamase gene transcription is directly performed by the regulator BlrA (Fig. 3) (Tayler et al. 2010, Tierney and Rather 2019). Although different names have been assigned to the intrinsic enzymes of Aeromonas spp. and although some species can display certain particularities, the general trend is the simultaneous presence of a class B carbapenemase (called CphA or Imi), a class C cephalosporinase (CepH/S or CAV-1) and a class D oxacillinase (AmpH or AmpS). Whereas the class D enzyme is physically linked to the BlrAB regulon, the other two enzymes are codified in different parts of the genome. However, they are also controlled by the BlrA regulator because they display so-called Blr tags, which are short conserved sequences located upstream of the genes controlled by this system: the higher the number of copies of these tags upstream of a gene, the greater the influence BlrA can exert over its expression. A general rule for these β-lactamases is their BlrAB-linked coordinate regulation and typical inducibility through inducer β-lactams such as cefoxitin or imipenem (Juan et al. 2017b).

Figure 3.

General model for the BlrAB-dependent regulation of intrinsic β-lactamases in Aeromonas spp. Disaccharide-P5 stands for 1,6-anhydro–MurNAc–GlcNAc. Abbreviations. OM: outer membrane; PGN: peptidoglycan; IM: inner membrane; MurNAc: N-acetyl-muramic acid; and GlcNAc: N-acetyl-glucosamine.

Another interesting difference between this regulation system and that linked to LysR-type elements, is that the latter needs muropeptides to reach the cytosol in order to bind the EBD of the regulator and enable its activator/repressor role. Conversely, in Aeromonas spp., the differentially accumulated soluble PGN fragments theoretically appearing during induction are allegedly detected by the inner membrane BlrB sensor when they are in the periplasm and, afterwards, this kinase activates cytosolic BlrA through a phosphoryl group transference. Therefore, an internalization of muropeptides into the cytosol would not be necessary to regulate the expression of Aeromonas spp. β-lactamases (Fig. 3). However, the potential existence of a periplasmic intermediary actor between activator muropeptides (1,6-anhydro–MurNAc–GlcNAc–P5 in this genus) and BlrB has not been ruled out. Consequently, neither has the potential muropeptide-binding EBD been characterized in BlrB, although all the evidence suggests that this is a strong possibility. Similar to what is explained above, in Aeromonas spp. mutations that constitutively activate the expression of its intrinsic β-lactamases eventually occur, thereby conferring resistance to several β-lactams and not only to hydrolysable inducers. Some examples of these mutations are: (i) inactivation of dacB, with effects presumably similar to those in P. aeruginosa; (ii) disruption of the periplasmic D-D carboxypeptidase BlrY (which shares this activity with PBP4, therefore, suggesting similar consequences); and (iii) a constitutive activation of BlrB through specific amino acid changes that promote its autophosphorylation and enable the constant activation of BlrA without muropeptide binding (Tayler et al. 2010, Juan et al. 2017b). Like what has been explained for P. aeruginosa or S. maltophilia CreBC, the activation of BlrAB apparently influences the expression of several genes presumably contributing to the resistance output derived from β-lactamase hyperproduction, including among them the inner membrane protein BlrD (equivalent to CreD), although its actual role has not been characterized (Fig. 3). In any event, these data in Aeromonas spp. reinforce the idea of a PGN-derived signaling capable of modulating its transcriptome and behavior.

CpxAR system

To conclude this section, it is interesting to provide some information regarding an additional two-component system that displays recent robust evidence of potentially connecting soluble PGN sensing with modulation of fitness/virulence, as is CpxAR. Moreover, unlike CreBC and BlrAB, the CpxAR system (CpxA, inner membrane sensor and CpxR, cytosolic transcriptional regulator) seems to work in a β-lactamase regulation-independent context. This system was initially associated with the capacity for sensing protein misfolding in the periplasm, but was later related to the sensing/responses intended to ensure the maintenance of bacterial envelope homeostasis in the presence of other stress/aggressions (Sugawara et al. 2021). In fact, a proper degree of activation of this system has been shown to protect Enterobacteriaceae against β-lactam challenge, through the modulation of the expression of periplasmic enzymes that alter the profile of PGN linking, for instance increasing the abundance of DAP–DAP cross-links. Meanwhile, its excessive activation has been shown to cause aberrant morphology, division and growth defects, and increased susceptibility to lysis (Raivio et al. 2013, Bernal-Cabas et al. 2015, Delhaye et al. 2016). In fact, the overactivation of this system has been demonstrated to be the initial mechanism through which the innate immunity’s PGN Recognition Proteins (PGLYRPs) are able to kill susceptible Gram-negative species (Torrens et al. 2020, Escobar-Salom et al. 2022). More specifically, the overactivation of CpxAR caused by PGLYRPs derives in the induction of multiple stresses that result in membrane depolarization, blockade of cytosolic PGN, protein and nucleic acid synthesis, and massive production of hydroxyl radicals that are finally responsible for bacterial death (Kashyap et al. 2011, 2014, 2017). Interestingly as a differential trait of CpxAR, an additional actor, namely the periplasmic protein CpxP, has been proposed to be essential to maintain a finely tuned low level of activation of the system. More specifically, in regular conditions CpxP would work as a blocker for the autokinase activity of CpxA by binding to this sensor’s EBD, therefore, disabling a subsequent phosphotransferase activity over CpxR. Conversely, during β-lactam challenge (and likely when bacteria face other attacks to their PGN), the soluble muropeptide pool in the periplasm allegedly changes, and then the differentially accumulated fragments could competitively displace CpxP from CpxA EBD, which would enable its autokinase activity and the subsequent phosphorylation of CpxR, finally promoting the expression of genes under its control (Fleischer et al. 2007, Tschauner et al. 2014, Masi et al. 2020). Interestingly, among the array of these genes, other periplasmic enzymes can be found, such as those codifying the amidases AmiA and AmiC (responsible for the cleavage of PGN stem peptides within the periplasm), whose increased expression entailed improved resistance to certain antimicrobial peptides in Enterobacteriaceae, obviously impacting virulence (Weatherspoon-Griffin et al. 2011). These data suggest the capacity of CpxAR, at least in Enterobacteriaceae, to sense damage/alterations in PGN, likely through interaction with differential soluble fragments, and to respond with a differential production of enzymes related to the periplasmic metabolism of PGN in order to maintain cell wall homeostasis.

On the other hand, there are many papers that demonstrate a great capacity of CpxAR for virulence-related gene modulation (Humphreys et al. 2004, MacRitchie et al. 2008). This is applicable not only to Enterobacteriaceae but to other clinically relevant Gram-negatives such as S. maltophilia and Aeromonas spp. and other species (Mitobe et al. 2011, Tanner et al. 2016, Gangaiah et al. 2017, Li et al. 2018, Xie et al. 2018, Subramaniam et al. 2019, Vogt et al. 2019, Wu et al. 2021). The regulation capacity of CpxAR affects different virulence-related features, mostly depending on the species, including adherence to host tissues, resistance to serum, expression of toxin secretion systems, cellular invasion capacity, biofilm formation, and so on. But paradoxically, the CpxAR response seemingly exerts an activator effect on virulence factors in some species (Debnath et al. 2013, Bontemps et al. 2015, Li et al. 2018, Xie et al. 2018), whereas its role is apparently repressor in others (Liu et al. 2012, Acosta et al. 2015, De la Cruz et al. 2015, 2016, Thanikkal et al. 2019, Vogt et al. 2019). These interesting particularities could be due to the ligand(s) that the CpxA sensor may bind, that could change depending on the scenario in which bacteria find themselves at each moment. This differential signal binding would finally impact the level of activation of CpxAR, and therefore, its control exerted over virulence-related genes. However, determining whether this possibility actually exists, and the exact nature of these potential signals initiating the response, poses a field that remains quite unknown. Regardless, CpxAR has been proposed as a promising therapeutic target for uropathogenic E. coli infections, in this case specifically through a provoked overactivation of CpxR, which entails notable virulence attenuation (Dbeibo et al. 2018). Strikingly, the CpxAR system has been barely studied in other relevant Gram-negative pathogens such as P. aeruginosa or A. baumannii, and therefore, much work is needed to understand whether this system could be an antivirulence target also in these and other species (Zhao et al. 2021).

In conclusion, there is a great body of evidence demonstrating that CpxAR is responsive to PGN alterations and that among the repertoire of controlled genes there are representatives codifying for PGN metabolism-related enzymes and for virulence factors. Thus, the final missing piece to close the circle that this review searches for, is to clearly demonstrate that this virulence regulation is directly due to CpxA sensing of differentially accumulated muropeptides, and to identify these signals, the stimuli triggering their release, and their specific effects on CpxR, and therefore, on virulence. If we manage to do so in the future we will be a step closer to the development of antivirulence therapies targeting this system.

The exogenous source of PGN fragments: not only intra- but also intercellular signaling?

As mentioned above, GlcNAc is one of the sugars composing the glycan chains of bacterial PGN, although it is also found in the eukaryotic context as a component of mammalian serum, mucin, and even polymers such as chitin. Further, this sugar has been shown to be sensed by different species of Gram-negatives that respond to its detection with various metabolic and/or pathogenesis-related behaviors. In fact, GlcNAc and other PGN fragments have been described not only to work as a PGN-derived intra- and intercellular signal for a given species, but also to be a very important interspecies and interkingdom messenger, enabling different responses based on transcriptome modulation (Konopka 2012, Dworkin 2014, Irazoki et al. 2019, Crump et al. 2020). In this regard, a decade ago P. aeruginosa was shown to interpret the presence of GlcNAc, shed in large amounts from the cell wall of Gram-positive bacteria (but often not abundantly enough in the host in absence of these microbes), as a signal to stimulate the production of extracellular factors such as pyocyanin and the proteases LasA and LasB (Korgaonkar and Whiteley 2011, Korgaonkar et al. 2013). These compounds increase interspecies competitiveness—phenazines and LasA are known for their broad antimicrobial activity—and general virulence of P. aeruginosa. Actually, GlcNAc was demonstrated to be the molecule required to trigger these responses, which were thus also activated by GlcNAc-containing polymer PGN, but not by fragments devoid of this sugar. This phenomenon, for which the inner membrane GlcNAc phosphotransferase transporter NagE (Fig. 4) was proven to be indispensable, is likely very important for instance in the lungs of cystic fibrosis patients, in which P. aeruginosa must compete with other Gram-positives such as S. aureus. Moreover, this could also be applicable to other types of infections, in which the presence Gram-positive colonizers could act as a synergistic factor for P. aeruginosa pathogenesis, aggravating the clinical outcome (Korgaonkar and Whiteley 2011, Naseem and Konopka 2015).

Figure 4.

Representation of the different N-acetyl-glucosamine-dependent phenomena of bacterial virulence modulation in P. aeruginosa and E. coli, as an example of an external source of PGN fragments acting as regulatory signals. The red hexagons represent N-acetyl-glucosamine, whereas the green ones represent 1,6-anhydro-N-acetyl-muramic acid groups. The 6P or α1P tags represent the different N-acetyl-glucosamine phosphorylated derivatives. Abbreviations: SP: stem peptide linked to 1,6-anhydro-N-acetyl-muramic acid; GlcNAc: N-acetyl-glucosamine; OM: outer membrane; PGN: peptidoglycan; IM: inner membrane; and QS: quorum sensing.

A few years later, the pathway through which P. aeruginosa responds to exogenous GlcNAc mostly proceeding from Gram-positive PGN was partially deciphered, demonstrating the essential implication of a putative two-component system cytosolic regulator, namely AgtR (PA0601 gene from PAO1 reference strain) (Korgaonkar et al. 2013, Naseem and Konopka 2015). This regulator was required to enable increased transcription of the pqs operon, responsible for the consequent higher release of the quorum sensing signaling molecule 2-heptyl-3-hydroxy-4-quinolone (referred to as Pseudomonas quinolone signal, PQS), which in turn promoted the production of pyocyanin and proteases in the presence of GlcNAc alone or larger GlcNAc-containing PGN fragments (Korgaonkar et al. 2013, Lee and Zhang 2015). However, this study did not elucidate the exact mechanism for GlcNAc sensing, although it could be argued that AgtS (PA0600), which is the inner membrane sensor of the aforementioned putative two-component system, would be the most likely candidate. Nevertheless, although in another paper a defective mutant in AgtS was shown to be less virulent in a Caenorhabditis elegans model (Lewenza et al. 2014), this sensor was seemingly not essential for GlcNAc responsiveness, which would suggest that AgtR could act both as a sensor and transcriptional regulator (Korgaonkar et al. 2013). This would be in accordance with the previously demonstrated need for NagE-dependent transport of GlcNAc into the cytoplasm, providing GlcNAc-6-P molecules, in order to trigger the abovementioned virulence potentiation (Korgaonkar and Whiteley 2011, Korgaonkar et al. 2013, Naseem and Konopka 2015) (Fig. 4). In any case, there are still some points that need deciphering; for instance, the exact mechanism that P. aeruginosa follows to internalize exogenous larger PGN fragments and obtain from them GlcNAc (perhaps through endogenous PGN hydrolases in the periplasm) and/or derivative molecules that ultimately work as virulence activators. Altogether, these phenomena suggest that targeting Gram-positive colonizers could be an interesting approach for treating P. aeruginosa-dominated polymicrobial infections, e.g. chronic wounds, thereby representing an imaginative strategy to hamper the virulence of this pathogen (Korgaonkar and Whiteley 2011, Korgaonkar et al. 2013, Naseem and Konopka 2015).

Paradoxically, in a more recent study, GlcNAc (and more specifically its phosphorylated cytoplasmic derivatives) were proposed to partially inhibit the transcription of genes controlled by LuxR-type (LasR in P. aeruginosa) quorum sensing-linked regulators in Chromobacterium violaceum, E. coli, and also P. aeruginosa, among which we find several important virulence-related actors such as elastases and proteases (Lee and Zhang 2015, Kimyon et al. 2016). The underlying mechanism proposed was a competitive inhibition of the LuxR receptor-binding site for the quorum sensing signal (N-acyl homoserine lactone) by phosphorylated GlcNAc molecules (-α-1-P and -6-P), as determined using in silico docking studies. However, in E. coli and P. aeruginosa only one reporter system was used to determine LuxR-type regulator activity in response to GlcNAc derivatives, and therefore, a detailed analysis of the resulting virulence phenotypes is still missing (Fig. 4) (Kimyon et al. 2016). Regardless, other studies had previously demonstrated that in E. coli, GlcNAc proceeding from the host could play a repressor role for certain virulence-related properties, which could paradoxically benefit the bacterium for infection. In this sense, GlcNAc accumulated within the cytosol as GlcNAc-6-P was shown to downregulate the expression of two extracellular adhesion factors (type 1 fimbrial adhesins and extracellular Curli fibers) in E. coli. This was interpreted as a strategy to decrease the levels of these proinflammatory structures—making the bacterium less detectable by the immune system—and therefore, promoting dissemination within the host (Sohanpal et al. 2004, 2007, Barnhart et al. 2006, Naseem and Konopka 2015). The mechanisms through which this phenotype was achieved were certainly intricate, involving the GlcNAc-6P-responsive repressor NagC, which acts on the nag regulon involved in GlcNAc catabolism, but also involving phase-variation mechanisms affecting the fim genes responsible for type 1 fimbriae expression (Sohanpal et al. 2004, 2007, Barnhart et al. 2006). In any event, more recently it has been shown that GlcNAc present in the intestinal mucin of the host could have similar effects on E. coli adhesion factors that would ultimately influence its biofilm formation and colonization capacities in the gut (Le Bihan et al. 2017, Sicard et al. 2018). It remains to be elucidated whether these results in E. coli could also be reproduced with Gram-positive PGN as a stimulus, as demonstrated in P. aeruginosa.

Some interesting unsolved questions arising from all these data are: (i) whether bacteria can discriminate between exogenously acquired GlcNAc and that proceeding from their own PGN turnover and recycling; (ii) whether the latter could also perform fitness/virulence signaling tasks, since it may be accumulated in differential amounts depending on PGN turnover-degradation and muropeptide release situation (e.g. β-lactam challenge); and (iii) what is the actual signaling molecule that modulates virulence, GlcNAc per se, or one (or various) of its cytosolic phosphorylated derivatives? (Fig. 4). In this regard, exogenous isolated GlcNAc is allegedly internalized into the cytosol exclusively through inner membrane NagE phosphotranferase, which necessarily provides GlcNAc-6-P molecules. Subsequently, through different reactions, GlcNAc-α-1-P is obtained, which will be used for anabolic reactions leading to PGN reconstruction. On the other hand, dealing with the bacterium’s own PGN, turnover-derived GlcNAc proceeding from the cytosolic action of NagZ on the anhydro-muropeptides released, can also be incorporated into these routes owing to AmgK in P. aeruginosa or NagK in E. coli, thereby finally providing GlcNAc-α-1-P (Fig. 4) (Park and Uehara 2008, Plumbridge 2009, Borisova et al. 2014, 2017, Fumeaux and Bernhardt 2017, Acebrón et al. 2017, Dik et al. 2018). Thus, exogenous GlcNAc could work as a signal for a potential sensor located in the inner membrane, but not in the cytosol, where it is in phosphorylated forms due to NagE internalization and subsequent reactions. Conversely, GlcNAc proceeding from the own PGN is cleaved from anhydro-muropeptides in the cytosol, where it could momentarily act as a signal before entering the phosphorylation pathways. Therefore, in both cases (exogenous and endogenous origin), phosphorylated derivatives could work as a signal to be bound by a regulator but only in the cytosol (Fig. 4). Either way, all these possibilities, which may not even be exclusive, remain to be ascertained to fully understand the possibilities of interfering in GlcNAc-dependent signaling as a strategy to attenuate virulence.

Beyond GlcNAc, other PGN-derived fragments have been reported to display a clear capacity for bacterial behavior modulation. This would be the case of the stem tetra- and penta-peptides (P4 and P5) cleaved from the glycan chains of PGN, which have been recently demonstrated to prevent the quiescent nonproliferative status that appears at low population density in uropathogenic E. coli (DiBiasio et al. 2020). Interestingly, this metabolic response is obtained when providing these fragments exogenously, both obtained from E. coli and from Gram-positive PGN, indicating that the nature of the third amino acid in the chain (DAP vs. l-lysine, respectively) is not important to activate this quorum sensing-independent growth. Meanwhile, the known mechanism for the uptake of these peptides into the cytosol, namely the oligo-peptide permease system Opp, seems dispensable for the response, from which it is deduced that there exists either an additional unknown transporter for the aforementioned P4 and P5, or a sensor responsible for their detection in the periplasm. Moreover, shorter stem peptides or larger PGN fragments including sugar moieties did not exert the abovementioned effects on the bacterium, suggesting quite a tight array of PGN fragments to which the unidentified sensor/regulator could respond (DiBiasio et al. 2020). Identification of this sensor/regulator as well as the downstream signaling events causing the prevention of quiescence is a very interesting topic worth delving into, because of the clear bacterial fitness-related implications.

Moreover, in E. coli it has been demonstrated that other different exogenously provided PGN fragments are able to modulate growth in this case rescuing cells from the stationary phase, a phenomenon partially demonstrated in P. aeruginosa as well (Jõers et al. 2019). Similar facts were previously seen in Gram-positives such as Mycobacterium tuberculosis and Bacillus subtilis through pathways involving eukaryotic-like serine–threonine kinases absent in Gram-negatives (Shah et al. 2008, Mir et al. 2011). In any case, both the abovementioned data regarding quorum sensing-independent growth and also these findings dealing with the capacity of certain muropeptides to stimulate bacterial growth when nutrients become scarce have very interesting implications for pathogenesis, since this is inextricably linked to bacterial viability, metabolic activity, and population increase. Returning to the recent work of Jõers et al.(2019) the molecules shown to be more capable of resuming bacterial growth are purified anhydro-muropeptides, i.e. 1,6-anhydro–MurNAc–GlcNAc disaccharides bound to a variable length stem peptide, which in fact are the ones mostly released due to PGN hydrolases action. In contrast, neither purified but not digested PGN nor MurNAc and tripeptide as separate molecules have shown the aforementioned growth resumption-stimulating activity. Although it could be argued that this resumption would be due to the use of these molecules as nutrients, the authors ruled out this possibility by using KO mutants in genes essential for the use of these PGN fragments as a carbon source, which behave as wildtype strains in terms of their responsiveness to muropeptides. This circumstance suggests the existence of a specific receptor for the abovementioned PGN fragments located either on the cell surface or in the periplasmic space that would somehow trigger the signaling (Jõers et al. 2019). Obviously, the actual sensor and also the subsequent events/actors allowing the bacterium to exit from dormancy are still unknown, posing a very interesting field worth delving into.

In conclusion, the studies gathered in this and previous sections clearly demonstrate that the PGN-derived signaling affecting bacterial phenotype is probably a much more widespread phenomenon than initially thought, applicable both to intra- and intercellular contexts. This is quite logical, since PGN is one of the bacterial elements, i.e. probably more responsive to exogenous attacks and more susceptible to undergoing changes depending on stressful conditions, growth phase, and other circumstances (Cigana et al. 2009, Torrens et al. 2019b, Anderson et al. 2022), making of it an excellent resource to work as a messenger by releasing fragments that could work not only as intracellular signals but also between cells, posing a kind of quorum sensing-like communication responsible for the modulation of behavior depending on the conditions. Altogether, these facts suggest that PGN-derived signaling is likely exploitable from the therapeutic perspective, even in the sense of developing modified muropeptide derivatives that once inside the cell could work as Trojan horses by interfering with regular signaling to dampen bacterial fitness and/or pathogenesis.

Other clues suggesting additional mechanisms of virulence regulation through soluble PGN sensing-response

Although with less strong evidence than that reviewed in the previous sections, there are other studies that suggest the existence of additional mechanisms of virulence regulation after sensing/response to PGN fragments. For instance, Folkesson et al. (2005) showed that the cytosolic accumulation of anhydro-muropeptides caused by ampD inactivation in Salmonella enterica brought about a dramatic virulence attenuation in this pathogen, illustrated by reduced invasion and intracellular growth toward macrophages and impaired fitness in a murine model. The authors demonstrated that PGN recycling impairment per se was not the cause of the attenuated phenotype, since a mutant defective in AmpG behaved like a wildtype in terms of virulence. Therefore, it was proposed that LysR-family regulators such as SinR or SpvR present in this species could be responsible for virulence attenuation: after binding with the differentially accumulated muropeptides derived from AmpD disruption, they would finally downregulate the expression of genes related to the aforementioned features (Folkesson et al. 2005). Similar results had been previously obtained in the plant pathogens Ralstonia solanacearum and Xanthomonas campestris, in which the deletion of two enzymes repeatedly mentioned throughout this paper, namely AmpD and NagZ respectively, could lead to the differential accumulation of PGN fragments acting as virulence-repression signals, through binding to as-yet-unknown regulators (Tans-Kersten et al. 2000, Yang et al. 2014). Instead of this, the opposite idea could also be argued: AmpD or NagZ absence would impair the production of certain PGN-derived signals that could be needed, after binding with the appropriate regulator, to activate the regular expression of virulence genes. In any event, these are equally interesting possibilities that need to be elucidated in the future, and whatever the real option is, it would pose a clever adaptive strategy: by sensing the differential accumulation of muropeptides, bacteria could interpret whether the scenario is more or less appropriate for the expression of virulence genes, in order to have more success in each specific situation.

Returning to Salmonella, it has been recently demonstrated the existence of a periplasmic protein—namely ScwA—that ultimately determines the level of activation of certain PGN hydrolases (Cestero et al. 2021). Obviously, depending on the lower/higher degree of action of these enzymes, the release of PGN fragments into the extracellular milieu could vary, which would impact the inflammation-related pathogenic mechanisms of this species. Moreover, this differential PGN cleavage and fragment generation could also entail potential intracellular signaling additionally impacting the expression of virulence factors through unknown regulators, although this is a possibility not yet investigated. In fact, the stimuli and pathways that in turn define the level of action of ScwA have not been elucidated, although they could eventually turn out to be similar to the abovementioned S. maltophilia model for CreBC and activation of PGN hydrolases in response to differential muropeptide accumulation (Huang et al. 2015a, b, 2017a,b, Cestero et al. 2021).

In P. aeruginosa, simultaneous combination of PGN recycling blockade plus AmpC hyperproduction or OXA-type β-lactamases expression (with interesting differences between enzyme variants) has been shown to cause a dramatic attenuation of virulence in an invertebrate model (Pérez-Gallego et al. 2016, Barceló et al. 2022). Since certain β-lactamases likely retain residual PGN-ase activity proceeding from a common ancestral origin with PBPs (Juan et al. 2018), the authors proposed that the production of these enzymes in a background of insufficient PGN precursors synthesis (due to recycling blockade) could have a greater impact on the sacculus, making it weaker and more susceptible to lysis. This would obviously impact cell viability, fitness, and virulence. However, the possibility of this residual activity not only affecting the sacculus, but also soluble muropeptides, has not been ruled out. In this sense, the specific β-lactamase produced in a background of impaired production of PGN precursors could, therefore, alter the soluble PGN fragments accumulated, by preferentially cleaving certain ones. Then, if there existed one or more regulator(s) with EBDs capable of sensing muropeptides, the altered muropeptide pool would differentially affect the activity of the regulator, finally impacting the expression of virulence-related genes. This idea could be supported by the work of Pérez-Gallego et al. (2016), in which certain genes essential for P. aeruginosa virulence, such as lasA, plcB (phospholipase C), exoS (exotoxin delivered by the type III secretion system), and so on, were shown to be downregulated in the scenario of PGN recycling disruption plus AmpC hyperproduction. However, future work will be needed to decipher whether such a signaling pathway leading to virulence attenuation actually exists in P. aeruginosa.

Finally, there are some other examples that suggest additional mechanisms of soluble PGN sensing and response, again in the context of β-lactamase regulation, which may have implications beyond antibiotic resistance. This could be the case of the intrinsic blaOXA-114 β-lactamase from the opportunistic pathogen Achromobacter xylosoxidans, which is allegedly capable of being constitutively hyperproduced under certain circumstances. In this sense, it has been demonstrated that deletion of a BlaI-family transcriptional repressor and the downwards-codified DD-endopeptidase (which takes part in PGN turnover in regular conditions), both located immediately upstream of the β-lactamase gene, is associated with an increase in β-lactam resistance. This could pose a clue suggesting an unknown system to detect the alteration in PGN caused by the absent endopeptidase, which would end up modulating production of the enzyme and perhaps other sets of genes (Ridderberg et al. 2015). More recently, in the same genus (Achromobacter ruhlandii), some mutations in the abovementioned protein taking part in PGN recycling, namely the ligase Mpl, have also been associated with increased β-lactam resistance, supporting the aforementioned possibility (Andersen et al. 2022). In fact, this mutational target has also been described as a cause of stable AmpC hyperproduction in P. aerginosa (Tsutsumi et al. 2013, Cabot et al. 2018).

Meanwhile, intrinsic class B enzymes of the pathogen Elizabethkingia meningoseptica have been shown to display expression changes depending on growth phase. Although the regulatory mechanism remains elusive for the time being, it could be related to a cell-wall fragment sensing-response, since it is well-known that PGN accumulates different qualitative and quantitative features when comparing exponential vs. stationary states (González and Vila 2012, Torrens et al. 2019b). Besides, in the case of the two intrinsic class D OXA-type enzymes codified in the genome of the emerging opportunistic pathogen Ralstonia picketii (OXA-22 and OXA-60), a very particular regulator has been proposed, namely ORF-RP3. The structuration of this alleged regulator, shown to be indispensable for the induction of β-lactamases in this species, resembles that of the LysR-type elements, since it is arranged in a divergon with a fragment ca. 200 nucleotides between the regulator and the divergently codified blaOXA-60 gene. However, at amino acid level, the ORF-RP3 element has no significant homology with any other known transcriptional modulator. Nonetheless, the implication of this element in features beyond β-lactamase regulation (resistance to pH/osmolarity alterations and survival in the stationary phase) was also demonstrated, supporting the idea of its potential activity as a global regulator. Obviously, the fact that ORF-RP3 activates β-lactamase expression during challenge with an inducer β-lactam (with the PGN alteration implications that this entails) is the final clue that suggests plausible interplay between PGN sensing and behavior modulation in this species as well (Girlich et al. 2009, Juan et al. 2017b).

Finally, in the environmental Shewanella oneidensis, different PGN-related mutational targets leading to hyperproduction of its intrinsic class D enzyme have been described, although no data regarding the mechanism of regulation are available to date. Among these mutations, the inactivation of PBP1a itself, or the LpoA lipoprotein that acts as an essential cofactor for the correct activity of this same PBP, have been described. Moreover, in contrast with β-lactamase-linked LysR regulators (absent in S. oneidensis), for which the presence of functional AmpG permease and NagZ N-acetyl-glucosaminidase are essential to enable mutational hyperproduction, in S. oneidensis, disruption of these proteins is associated with higher β-lactam resistance. More recently, disruption of three PGN hydrolases (SltY, MltB, and MltB2) has been shown to further increase β-lactam resistance in this species. All these perhaps paradoxical circumstances undoubtedly suggest a close and complex relationship of this β-lactamase control system with PGN metabolism, but future work is needed to fully understand the mechanism and the potential regulatory implications affecting wider sets of genes (Korfman and Sanders 1989, Chahboune et al. 2005, Zamorano et al. 2010, 2011, Yin et al. 2014, 2015, 2018).

Concluding remarks

Throughout this review we provide strong evidence supporting the idea of PGN-derived signaling with regulatory effects on bacterial virulence existing in different Gram-negative species. Some of the studies gathered display clear proof of the existence of this signaling, which has different mechanistic variants and particularities depending on the species, whereas others pose only plausible clues. In any case, even in the papers that unequivocally demonstrate the existence of a bacterial capacity for soluble PGN/sensing and transcriptomic response impacting pathogenesis, there are still many questions remaining in order to fully understand the pathways involved. Another idea that can be extracted from the information compiled is that, although there are some models common to various species, there are others that are very specific to a particular microbe, or even some cases in which an allegedly similar mechanism works in quite different ways, e.g. CreBCD from P. aeruginosavs.S. maltophilia. This would also be applicable for AmpR in the same two species, described to be an acute virulence activator in the former, but a repressor in the latter (Huang et al. 2015a, b, 2017a, Alcaraz et al. 2022). Some other contradictory effects on virulence when considering a given regulator have been pointed out in this review, for instance regarding CpxAR (activator vs. repressor of virulence depending on the species/study), revealing a very complex interplay between potential PGN sensing and response that needs clarifying before the development of therapeutics can be considered. Even some cases of the same regulator having a double-edged sword effect by promoting some virulence factors and inhibiting others in a same scenario and species have been reported (e.g. CreBC control over protease secretion vs. motility in S. maltophilia), whereby more complexity is added to the topic (Huang et al. 2015a, b, 2017a).

Moreover, although the muropeptide-binding capacity of some AmpR EBDs has been deeply studied in the context of β-lactamase regulation (Balcewich et al. 2010, Caille et al. 2014, Vadlamani et al. 2015, Dik et al. 2018), the potential PGN-binding clefts of other elements likely linked to PGN fragment sensing/response remain largely uncharacterized, such as those of the inner membrane sensors of the two-component systems BlrAB and CreBC. In this regard, there is also a clear lack of knowledge concerning the repertoire of controlled genes by some potential PGN-responsive regulators, such as those under the control of CreBC in S. maltophilia or BlrAB in Aeromonas spp.; i.e. a deep study of Cre/Blr tags in the genomes of these species has barely been approached.

Another key question that remains poorly understood is whether the influence on bacterial behavior that some of the mechanisms reviewed show could appear not only under after β-lactam challenge but also in regular situations and/or in response to other stimuli (e.g. growth phase, nutrient availability, PGN-targeting immune attacks, and so on), which may also influence the pool of soluble muropeptides. This would entail continuous sensing of released PGN-fragments by certain regulatory mechanisms, which would enable a differential fitness/virulence-related transcriptome depending on the situation, as happens with the typical AmpC–AmpR systems regarding β-lactam resistance context (Juan et al. 2017b). In this sense, once a more solid knowledge on this field has been obtained, the identification of which PGN fragments potentially exert a positive/repressive effect on the pathogenesis of each species/specific regulator will also be necessary. For this goal, several techniques are available now (ultra high performance liquid chromatography–mass spectrometry, crystallization, small angle X-ray scattering, and so on) that could identify and quantify the differential accumulation of PGN fragments associated to each virulence phenotype, and also determine the interaction of each muropeptide with a given EBD, as demonstrated by recent studies in the field (Vadlamani et al. 2015, Dik et al. 2017, Torrens et al. 2019a, Hernández et al. 2020).

There are also other specific topics that need further delving into, such as the potential mechanisms very specific to certain microbes, or ascertaining whether certain clues suggesting a PGN-related regulation of virulence are real. In the two cases, a great amount of work is yet to be carried out, because mostly, only phenotypic effects have been observed in these studies but there is still no clue as to the nature of the sensor/regulators and PGN-derived signals involved. Another large field of related knowledge to expand is the study of dozens of poorly characterized two-component systems and LysR-type regulators codified in Gram-negative genomes. In fact, many of these are expected to act as global regulators, therefore, displaying plenty of virulence-related genes included within their controlled repertoire, as happens with other better studied elements such as AmpR, CreBC, BlrAB, and CpxAR. As this kind of regulatory mechanisms have an ability to sense/respond to different stimuli, and PGN is one of the more dynamic bacterial structures as well as being more affected by external aggressions or scenarios (PGN-targeting immune attacks such as lysozyme, β-lactams, growth phase, and so on) thereby providing plenty of soluble derived fragments, it is highly expected that several of these systems may interpret the differential accumulation of muropeptides as signals. In fact, the particular release of muropeptides during β-lactam challenge is a situation that could resemble that of an attack of a PGN-targeting immune element, and therefore, it seems likely that bacteria might also display responsiveness by modulating certain features that ameliorate survival in this harmful scenario. This would have obvious virulence-related implications, since one of the slopes of microbial pathogenesis is to develop systems to circumvent the action of immunity. In this context, P. aeruginosa encodes over 60 two-component systems, whereas in E. coli there are approximately 30 of them in the chromosome. Many of these are yet to be dissected, although others have been clearly shown to strongly influence the expression of virulence-related features (Francis et al. 2017, Sultan et al. 2021, Wang et al. 2021). However, the stimuli to which these regulatory pathways react are also largely unknown. Thus, ascertaining whether and what type of PGN fragments could stimulate these systems and promote transcriptome changes poses a very interesting, broad field to develop. Similar reasoning could be made for LysR-type regulators; for instance, 113 genes are annotated as encoding LysR-type transcriptional regulators in P. aeruginosa PAO1 strain, but the functions and potentials for PGN fragment binding of many of them remain largely unknown (Modrzejewska et al. 2021).

All in all, if any of the soluble PGN-based systems reviewed herein come to be deeply characterized and understood, we will be able to identify related weak points susceptible to being used as antivirulence targets. In this regard, the final strategies could be varied; an initial approach could be to directly inhibit the inner membrane sensor (in a given two-component system) or the cytosolic regulator that acts as a promoter of virulence after PGN sensing. Another strategy could be to block the entrance of PGN-derived virulence activator signals into the cytosol, if they exist, by targeting AmpG for instance. In fact, this strategy has already been explored in P. aeruginosa, and shown not only to dampen virulence, but also to obviously disable AmpC-mediated resistance (Torrens et al. 2019c). Similarly, targeting enzymes that are essential for the elimination of PGN-derived virulence inhibitory signals,—or what would be practically the same, enzymes needed for the generation of pathogenesis-activating signals—could be a very interesting idea. In this sense, although a very wide array of enzymes take part in PGN metabolism, NagZ could be a good candidate, since it has been shown to be essential for the virulence of X. campestris, and also a factor partially contributing to P. aeruginosa fitness within the host and mostly essential for AmpC-related resistance (Torrens et al. 2019c). Actually, different initiatives to develop NagZ inhibitors (with the final goal of disabling AmpC hyperproduction) are nowadays ongoing, with encouraging results (Stubbs et al. 2007, Ho et al. 2018, Bouquet et al. 2022). The fact that these studies have the goal of blocking AmpC-dependent resistance is not a drawback but rather the opposite, since with a single drug we could be both limiting bacterial resistance (therefore, rehabilitating β-lactam’s effectiveness), and also attenuating the virulence of the microbe owing to interference with PGN signaling. Thus, combined treatments of β-lactams plus PGN-signaling inhibitors could be an excellent option in this scenario. In any case, a necessary first step to develop this kind of drugs, would be to reliably identify the enzymes (such as NagZ) that are essential for the generation/elimination of the aforementioned PGN-related signals.

As opposed to using chemically designed inhibitors for specific enzymes, another interesting strategy could be the use of muropeptide derivatives mimicking those that naturally bind to sensor/regulators and activate virulence, but with added modifications that would disable the process. In other words, competitive molecules for binding with the EBD that would displace the naturally released PGN fragments, but which, once bound, would not promote the regulator’s conformational change that promotes the controlled genes’ expression. However, prior to the development of this kind of muropeptide homologues, a rigorous characterization of the binding cleft of the chosen regulator should be carried out in order to ascertain the real possibilities of inhibition. In fact, similar interference strategies have been followed with synthetic molecules that mimic quorum sensing activators, and that therefore, block the responses controlled by this mechanism (Ó Muimhneacháin et al. 2018, Tung and Quoc 2021, West et al. 2022). As mentioned above, competitive inhibition of quorum sensing signal-binding sites in the corresponding receptor by phosphorylated GlcNAc derivatives was proposed as a mechanism leading to virulence attenuation, posing an interesting hybrid strategy combining PGN-derived signaling and quorum sensing interference (Lee and Zhang 2015, Kimyon et al. 2016). Similarly, if some natural PGN-derived signals potentially acting as virulence inhibitors (such as for instance those starring the mentioned phenomena linked to AmpD deletion in Salmonella; Folkesson et al. 2005) are reliably identified, a clever strategy could be to administer them at high doses with the goal of being internalized by the bacterium and trigger the pathogenic attenuation. Either way, a very important point to take into account when developing any of these strategies is accessibility to the target; in this sense, the ability of the drug to get to periplasmic enzymes/sensors is apparently easier than reaching a cytosolic regulator, which is additionally protected by the inner membrane. Moreover, the use of small muropeptide mimetics could be a more likely successful idea, since bacteria have been shown to naturally incorporate large PGN fragments proceeding from the external milieu, as explained above (Korgaonkar and Whiteley 2011, Korgaonkar et al. 2013, Naseem and Konopka 2015).

Therefore, although we are still far from the therapeutic application of antivirulence options developed in the context of PGN signaling-pathogenesis interplay, it is a field that offers the encouraging clues displayed throughout this review. Moreover, the potential of antivirulence is reinforced by the fact that, in contrast to conventional antibiotics, antivirulence drugs are believed not to impose a high selective pressure on bacterial populations, and would therefore, not promote the dissemination of antibiotic resistance and virulence genes (Ogawara 2021). In conclusion, the PGN fragment-dependent sensing/response connected to virulence regulation seems to be an insufficiently studied phenomenon in Gram-negatives, probably occurring as an adaptive strategy in a much more generalized way than expected. Hence, we need to delve into its study in order to exploit this interplay as an effective antivirulence strategy, which appears as a promising option in the current scenario of a growing shortage of effective antibiotics.

Conflict of interest

The authors declare that this work was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Funding

This work was supported by the Balearic Islands Government grant FPI/2206/2019 and the Spanish Network for Research in Infectious Diseases (REIPI, RD16/0016/0004) and grants IJC2019-038836-I (Ministerio de Ciencia e Innovación, Spain), PI18/00681, PI21/00753, PI21/00017, and FI19/00004 from the Instituto de Salud Carlos III (Spain) cofinanced by the European Regional Development Fund “A way to achieve Europe.”