The LiaFSR-LiaX System Mediates Resistance of Enterococcus faecium to Peptide Antibiotics and to Aureocin A53- and Enterocin L50-Like Bacteriocins

Enterococcus faecalis and Enterococcus faecium are common commensal bacteria of the human gastrointestinal tract, but they are also opportunistic pathogens that have emerged as a major cause of urinary tract infections, surgical wounds, endocarditis, and nosocomial bacteremia. Moreover, they have become increasingly multidrug resistant (MDR) and resistant to high concentrations of aminoglycosides such as streptomycin, kanamycin (KAN), and gentamicin (GEN), β-lactams such as ampicillin, and the last-resort antibiotic vancomycin (vancomycin-resistant enterococci [VRE]) (1, 2). Notably, they can transfer resistance genes to other medically important microorganisms, as has been shown for vancomycin-resistant Staphylococcus aureus (3). Thus, finding new antimicrobial agents against the enterococci is critically important.

The cell envelope-targeting mechanism of action is particularly promising for the development of such next-generation antimicrobials. The cytoplasmic membrane and cell wall are essential and highly conserved structures, so the development of resistance to agents targeting them is difficult (4). Daptomycin (DAP) and gramicidin (GRA) are U.S. Food and Drug Administration (FDA)-approved cell envelope-active antibiotics from the group of nonribosomally synthesized antimicrobial peptides (AMPs) (5, 6). Gramicidin forms dimeric channels in the membrane mediating ion conduction (6). DAP-Ca²⁺ oligomerizes and forms tripartite complexes with phosphatidylglycerol and undecaprenyl-coupled cell envelope precursors, such as lipid II, blocking cell wall synthesis. Complex formation triggers delocalization of several membrane-associated enzymes involved in peptidoglycan biosynthesis and drastic rearrangement of the cytoplasmic membrane (7, 8). DAP is used to treat infections caused by VRE, but despite its membrane-directed mechanism of action, DAP resistance is increasingly observed in Enterococcus spp. (9).

Bacteriocins, which are ribosomally synthesized AMPs of bacterial origin, are divided into posttranslationally modified lantibiotics (class I) and unmodified nonlantibiotics (class II). Nisin is the best-characterized lantibiotic with broad-spectrum and antienterococcal activity, whose dual mode of action includes inhibition of peptidoglycan synthesis through interaction with lipid II, followed by pore formation by penetrating the cell membrane (10). Among nonlantibiotics, two families of aureocin A53 (AurA53)- and enterocin L50 (EntL50)-like bacteriocins with broad-spectrum activity against enterococci are distinguished by sharing a similar structural saposin-like fold (11, 12). It is composed of three or four amphiphilic α-helices, of which the cationic, hydrophilic residues are exposed on the surface and the hydrophobic ones are packed inside to form a hydrophobic core (13, 14). The saposin-like fold is thought to enable these AMPs to penetrate the cell membrane in the absence of specific receptors and act either by pore formation or generalized displacement of membrane lipids and disruption of membrane structure (11, 14).

The resistance to membrane- or cell wall-active AMPs is often shaped by two-component stress-response regulatory systems (TCSs). LiaSR is the eponymous TCS first identified and characterized in Bacillus subtilis (15). LiaSR has been shown to respond strongly to lipid II-targeting DAP, vancomycin, or ramoplanin (RAM), as well as to many other cell envelope stress factors, such as bacitracin (BAC), nisin, cationic AMP LL-37, alkaline shock, detergents, organic solvents, secretion stress, or filamentous phage infection (16). The LiaSR system is comprised of LiaS, a bifunctional histidine kinase that acts as a kinase/phosphatase on the LiaR response regulator (RR) in the presence/absence of stressors (17). This system is genetically and functionally associated with the membrane-localized accessory protein LiaF, and together they form the three-component system LiaFSR. In the absence of a stressor, LiaF inhibits the LiaSR-dependent signal transduction (18), whereas in its presence, the phosphorylated RR induces the liaIH-liaGFSR locus, with the strongest activation of the liaIH genes encoding, respectively, a membrane anchor for LiaH and a homolog of the cytosolic phage shock protein A (PspA) of Escherichia coli (19, 20). Homologs of the LiaFSR systems are widespread among species belonging to the Firmicutes, with the number and nature of stressors, as well as the genes regulated by these homologs, varying widely (16). The enterococcal LiaFSR system has been studied extensively recently, mainly in the context of DAP resistance in E. faecium and E. faecalis. However, it has also been shown to be involved in resistance to other AMPs, such as lantibiotics (nisin, gallidermin, mersacidin), the antibiotic friulimicin, human beta-defensin-3 (HBD3) and cathelicidin LL-37, and synthetic RP-1 (21, 22). In E. faecalis, in addition to its own locus, LiaFSR also regulates the liaXYZ operon and genes for cell wall synthesis, cell division, transmembrane proteins, and cell envelope stress response (22). It has been proposed that a major modulator in this process is the surface protein LiaX, which, in the absence of DAP, acts as an inhibitor of the LiaFSR system, potentially by interacting with LiaF and LiaS, whereas the binding of DAP by LiaX results in the release of LiaFS and subsequent activation of this system (22). The increased activity of the LiaFSR system causes remodeling of the cell membrane (redistribution of anionic phospholipids away from the division septum, an increase in the content of phosphatidylglycerols, and a decrease of cardiolipin), protecting against the action of AMPs (22).

Here, we investigated the genetic basis of resistance to the AurA53- and EntL50-like bacteriocins and cross-resistance with a wide range of antibiotics in E. faecium. We isolated spontaneous resistant mutants of the sensitive E. faecium strain LMGT 2783 by exposing the cells to increasing concentrations of bacteriocin BHT-B. Single, nonsynonymous mutations in genes encoding the LiaFSR system or the LiaX protein were found in the resistant strains. Notably, these adaptive mutations caused cross-resistance to other AurA53- and EntL50-like bacteriocins, as well as to some antibiotics such as the lipid II-targeting DAP and RAM, the membrane-targeting GRA, and the positively charged KAN and GEN. The mutation in liaR increased the expression of liaFSR, liaX, a cell wall remodeling (sgtB) gene, and several hypothetical (liaY, liaZ, xpaC, ef0798, ef0932, ef1533) genes. Finally, we conducted a series of resistance assays for strains with the liaFSR operon or the liaX gene deleted or complemented. The results indicated that the resistance of the adaptive mutants is likely due to changes in the activity of individual components of the LiaFSR-LiaX (LiaFSR-X) system, resulting in the obligatory activation of LiaR, which then leads to an LiaR-dependent upregulation of the expression of genes conditioning cell envelope remodeling.

Enterococci are a large group of lactic acid bacteria that, owing to their tolerance to salts, acids, and activity against Listeria monocytogenes (33), have long been used in the production of fermented dairy and meat products (34). On the other hand, enterococci also include pathogenic strains carrying virulence factors and a considerable resistome, which makes them one of the most serious and steadily increasing causes of hospital epidemiological risks (35). The vast majority of enterococcal infections are caused by E. faecalis, which is more pathogenic than E. faecium. However, E. faecium is intrinsically more resistant to antibiotics and is the leading cause of VRE infections (23). Accordingly, the Infectious Diseases Society of America has classified E. faecium into a top-priority ESKAPE group of six highly virulent and MDR bacterial pathogens (Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter species) for which novel antimicrobial agents urgently need to be developed (36). Alternative measures, such as the use of bacteriocins, could be a potential solution, especially since several international health agencies recommend and support the development of such interventions (37). Due to their compelling advantages (broad or narrow spectrum of activity, high potency, lack of or low cytotoxicity, and possibility of bioengineering), bacteriocins could be used in preventive strategies separately or in combination with other antimicrobial agents (e.g., antibiotics) (38). In fact, several bacteriocins have already been subjected to preclinical and clinical trials, and thiostrepton (class I bacteriocin) is already used in veterinary medicine (39, 40). Nevertheless, in order to more fully exploit their efficacy in an era of rapidly developing multidrug resistance among pathogens, more basic research on the mechanisms of cell killing and the development of resistance to bacteriocins is needed. The present study fills some of the knowledge gaps regarding the genetic basis of the development of resistance to potent antienterococcal bacteriocins, pointing out the common and divergent features regarding cross-resistance to antibiotics.

Many bacteriocins act by targeting a specific receptor. Such receptor-dependent activity usually narrows down the spectrum of susceptible bacteria and additionally, through receptor mutations, may lead to the rapid development of resistance, as has been observed, for example, in the case of the mannose phosphotransferase (Man-PTS) receptor, whose inactivation induced by the presence of a bacteriocin led to full resistance of the mutants (41, 42). Therefore, the receptor-independent, membrane-directed AurA53- and EntL50-like bacteriocins, with their broad spectrum of activity against Gram-positive bacteria, may be very promising in treating or aiding in the treatment of certain bacterial infections, including those caused by pathogenic enterococci (11, 14, 26). The cytoplasmic membrane is a critical and highly conserved structure, and one would expect difficulty in the development of resistance to membrane-targeting antimicrobial compounds. However, we did observe here such a phenomenon induced by a membrane-active antimicrobial tested against E. faecium. Moreover, the mild resistance, in the range between 2- and 16-fold, triggered by a bacteriocin from the aureocin A53-like group, BHT-B, conferred a cross-resistance of the mutants to other antimicrobial agents, including not only BHT-B-related bacteriocins but also clinically relevant peptide antibiotics targeting the membrane (GRA) or cell wall (RAM and DAP) and also two aminoglycosides (KAN and GEN). Bacteriocin resistance did not change the fitness of mutants with regard to their efficiency of multiplication compared to that of WT cells, their ability to utilize different carbon sources, and their resistance to NaCl or SDS. The absence of biological costs of acquiring resistance may suggest that mechanisms other than those used to protect against AurA53- and EntL50-like bacteriocins are involved in protection against salinity and detergent stress.

Whole-genome sequencing of the resistant E. faecium mutants identified adaptive mutations in the liaFSR and liaXYZ operons. LiaFSR is a well-known regulatory system engaged in the response to cell membrane stress in Gram-positive bacteria, while the role of LiaXYZ remains poorly understood with the exception of the LiaX function as the main modulator of LiaFSR in E. faecalis (22). Previous studies have shown that the development of enterococcal resistance to DAP most commonly involves mutations in the liaFSR genes leading to MICs ranging from 3 μg/mL to as high as 48 μg/mL (43). The DAP MIC for the WT E. faecium strain LMGT 2783 used here was 3 μg/mL, within the 2- to 4-μg/mL range typical for most of the several thousand strains of this species isolated from European and U.S. medical centers (44). In contrast, the BHT-B-resistant mutants obtained exhibited slightly higher DAP MIC values, up to ~8 μg/mL; they were sufficiently high to exceed the susceptible dose-dependent or even resistant clinical MIC breakpoints set for DAP and E. faecium by the Clinical and Laboratory Standards Institute (CLSI) (≤4 and ≥8 μg/mL, respectively) (45). Although the occurrence of LiaFSR adaptive mutations has already been reported as a factor in the development of resistance to DAP and some other antimicrobial compounds across enterococci (21, 22, 43, 46–49), it has not been linked to the resistance to cell envelope-active peptide antibiotics or AurA53- and EntL50-group bacteriocins studied in this work, nor to aminoglycosides, which are chemically different from the above-mentioned compounds and use a wholly distinct mechanism of action.

The liaXYZ operon identified in this work as another target for adaptive mutations is much less well understood and, until recently, was described in the literature as encoding proteins with unknown functions (previously designated YvlB-PspC-YvlD). However, following the identification of mutations in liaX in DAP-resistant E. faecalis mutants, its potential role in protection against this antibiotic was suggested (46, 50, 51). The final resolution of its pivotal function in E. faecalis was achieved in the work of Khan et al., in 2019, who showed that LiaX binds to DAP and thereby leads to activation of the downstream LiaFSR response pathway (Fig. 4A and B) (22). Since LiaX functions as an intrinsic modulator of this system’s response to cellular stress, it is tempting to propose that LiaFSR, usually referred to as a three-component system, is in fact more complex and together with LiaX could actually constitute a four-component LiaFSR-X system.

Using in silico analysis, we identified different effects of the adaptive mutations in the liaF and liaX genes encoding two inhibitors of the LiaFSR-X system (truncation of the C-terminal domains due to frameshift or nonsense mutations, e.g., LiaF^Gln115X, LiaF^Ile64IlefsX4, LiaX^{Leu461TrpfsX20}, or LiaX^Gln507X) in comparison with those in liaS and liaR encoding two signaling transducers (only amino acid substitutions such as LiaR^Thr176Lys or LiaS^Gln184Arg in the active sites of the respective enzymes). This suggests that the mutations in liaF and liaX could be classified as loss-of-function mutations resulting in the inactivation of the LiaF and LiaX inhibitors. In contrast, since there were no frameshift or nonsense mutations in the LiaSR signal transducers that would disable the pathway, the missense mutations in liaS and liaR could be classified as gain-of-function mutations that likely resulted in activation of LiaS and LiaR. Remarkably, although all these adaptive mutations were of a different nature (missense, nonsense, frameshift) and impact on the particular component of the LiaFSR-X system (activation or inactivation), and occurred in genes with dissimilar functions (activators or inhibitors of the LiaFSR-X response pathway), the levels of resistance to a particular antimicrobial compound were comparable among all the mutants, suggesting similar impacts of the individual components of the LiaFSR-X system on the development of resistance to this compound. Given the response cascade in the LiaFSR-X pathway, which is activated by antimicrobial peptides abolishing the inhibitory effect of LiaXF over LiaSR (Fig. 4B), it seems quite reasonable that the outcome of the adaptive mutations generated here is to decouple this system from the inducing action of the antibacterial compounds.

Thus, we propose that in the MUT_137, MUT_138, and MUT_139 strains, the effect of the loss-of-function mutations in liaX was an abolishment of the LiaXF-dependent inhibition of LiaS and a consequent activation of LiaR as a result of its phosphorylation by LiaS. Similarly, the loss-of-function mutations in liaF carried by MUT_130 and MUT_136 strains released LiaS from the LiaF-dependent inhibition, which in turn led to LiaR activation (Fig. 4C). Remarkably, inactivation of LiaX alone was sufficient to release the histidine kinase LiaS from the inhibitory effect of LiaF, which is known to be a potent inhibitor of LiaS in the absence of a stressor. This suggests that in fact, a complex of interacting LiaX and LiaF proteins is required for the LiaS inhibition and the absence of either component abolishes this inhibition. In contrast, the gain-of-function mutations in liaS and liaR, which appeared in the MUT_137 and MUT_132 mutants, respectively, led to direct activation of the encoded proteins despite the unaffected inhibitory action of the LiaX-LiaF complex (Fig. 4C). The MUT_137 strain is unusual because of its double mutation of the loss-of-function (in liaX) and gain-of-function (in liaS) types (Fig. 4C), but its resistance level is similar to that of the single mutants. This lack of additivity is consistent with the proposed mode of action of the two types of mutations.

A systematic review of the published literature for reports on the genetic alterations in DAP-resistant enterococci indicates numerous mutations in the liaFSR-liaX genes. Although highly variable in terms of localization in individual genes, they are overwhelmingly of the same type as in our study, namely, loss-of-function in genes encoding inhibitors of the system (liaX and liaF) and gain-of-function in those coding for activators (liaS and liaR) (43, 46, 47, 50, 52). This indicates a common mode of resistance development via liaFSR-liaX mutations in all enterococci, as well as potentially in other species carrying these genes. Consequently, we propose that the final effect of any adaptive mutation in the liaFSR or liaX genes leading to resistance to the compounds tested always comes down to a constitutive activation of LiaR. Thus, LiaR can be activated by loss-of-function mutations in liaX or liaF, whose products in WT bacteria constitutively inhibit LiaSR activity, or by gain-of-function mutations in liaS or liaR, which in turn cause, respectively, a corresponding increase in LiaR phosphorylation by activated LiaS or an LiaS-independent activation of LiaR by, e.g., altering its oligomerization state toward the active and phosphorylation-independent DNA-binding dimer, as has already been proposed by Davlieva et al. (53).

The dominant role of LiaR in the regulation of the cellular stress response in E. faecium is confirmed by the fact that in mutants lacking the liaFSR or liaFSR-liaX genes and thus hypersusceptible to most of the antimicrobial agents tested here, an in-trans delivery of multiple copies of liaR alone was sufficient to fully reverse the effect of the deletions to produce a WT level of sensitivity or even increased resistance of the recombinant strains. We did not observe such an effect when another gene subject to gain-of-function mutation, liaS, was overexpressed. This lack of a suppressive effect is consistent with the accepted mode of action of the LiaS kinase: in the absence of its substrate, LiaR, it could not effect susceptibility of the deletion mutants. Intriguingly, we have previously noted a similar paramount role of a response regulator from the four-component stress response system YsaCB-KinG-LlrG in L. lactis. Although apparently unrelated to the LiaFSR-X system, YsaCB-KinG-LlrG is also involved in protection against AurA53- and EntL50-like bacteriocins and some peptide antibiotics (26). Similarly to LiaR in the present work, the LlrG response regulator from the YsaCB-KinG-LlrG system conferred resistance even in the absence of the other components of this pathway; this activity could be triggered by gain-of-function mutation or an upregulation of its expression (26). This implies that the relevance of response regulators as the ultimate effectors of mutations in genes of an entire pathway involved in bacterial resistance to antimicrobial agents is universal across the diverse response system functioning in different genera.

Regulation of liaFSR transcription by activated LiaR has been described for several bacterial genera, including Enterococcus (17, 22, 53–55). Also, the LiaR-dependent regulation of liaXYZ expression has been described in some aspects in both E. faecalis and E. faecium (22, 53). In contrast, other genes regulated by LiaR, including those involved in protection against external stressors, are poorly characterized, especially in E. faecium. Here, we determined in E. faecium the effect of the Thr176Lys substitution in LiaR on the expression of liaFSR, liaXYZ, and diverse other genes known to be involved in conferring resistance to peptide antibiotics (including glycopeptides, DAP, and GRA) and bacteriocins (including AurA53- and EntL50-like groups and Lcn972) in various bacterial species (12, 22, 26–32). As expected, the mutated LiaR strongly activated the expression of the liaFSR and liaXYZ operons, their mRNA levels reaching about 10 to 20 times that found in WT E. faecium. LiaY and LiaZ are transmembrane proteins of unknown function, possibly involved in the signaling cascade as interaction partners with LiaX (22). Based on LiaY homology with the E. coli PspC protein involved in maintaining cell envelope integrity under a variety of membrane stresses (19), one can speculate about its involvement in cell membrane remodeling. An almost 10-fold expression enhancement was also observed for the xpaC and sgtB genes involved, respectively, in CesSR-mediated resistance to Lcn972 in L. lactis (28, 29) and VraSR-mediated resistance to glycopeptides and β-lactam antibiotics in S. aureus (32). CesSR of lactococci and VraSR of staphylococci are equivalents of LiaSR in enterococci, so they are likely to be subject to similar regulation, although, as indicated by the data obtained here, this is not the rule, since three other lactococcal CesSR regulon genes (spxB, oatA, and spx) (27–29) and five other staphylococcal VraSR regulon genes (fmtA, murZ, pbp2, plsC, and spsA) (32) appeared not to be regulated by LiaR in E. faecium, at least under the conditions tested here. As could be predicted from species relatedness, the strongest activation by LiaR^Thr176Lys was observed for the ef0798, ef0932, and ef1533 genes, elements of the LiaR regulon in E. faecalis. These three genes are scattered throughout the chromosome and encode membrane proteins with unknown function in both species, but their regulation by LiaR suggests that they could be important for resistance development and thus they deserve further study. The EF1533 protein containing the DUF1093 domain seems particularly relevant, as it is homologous to the YxeA protein which in B. subtilis is regulated by the YxdJK TCS. This system regulates the expression of various genes involved in bacterial cell wall modifications in response to cell envelope-directed antimicrobial peptides (56–58). Similar targeting of cell envelope modifications, may be due to overexpression in the LiaR^Thr176Lys mutant of SgtB glycosyltransferase, which is involved in peptidoglycan synthesis (59, 60).

Last, but not least, we show here that the decrease in susceptibility of E. faecium can also be induced by merely increasing liaX expression alone, without a concomitant gain-of-function mutation in LiaR, which was induced by a nucleotide change that streamlines the UP sequence in the MUT_131 mutant. The liaX transcript was elevated only 3-fold, but this was enough to induce an 8-fold decrease in the mutant's susceptibility to BHT-B. This is an ambiguous effect insofar as one would assume that the more LiaX inhibitor, the stronger the inhibition of LiaSR and thus the increase in the mutant's susceptibility, but in this case, we noted the opposite effect. However, Khan et al. observed a similar phenomenon of protection against DAP of susceptible E. faecalis strains induced by the addition of exogenous LiaX or N-terminal LiaX (22). It is suggested that the protective effect of LiaX is dependent on the presence of LiaR and that LiaX does not sequester DAP just to prevent its binding to the cell. Our study of liaFSR and liaXYZ transcripts in MUT_131 does not provide a firm conclusion regarding the effect of increased liaX expression on the participation of individual gene products in the increased susceptibility of this mutant, as the resulting mRNA levels of the individual genes were not increased at all or were increased only slightly (2- to 3-fold maximum). However, it is tempting to speculate that the 2-fold increase in liaR expression in the MUT-131 mutant is sufficient to confer resistance and that the encoded protein is the receiver of the signal from overproduced LiaX, which then passes it on to genes involved in protective remodeling of the cell envelope. In agreement with this hypothesis is the fact that in E. faecalis, the addition of DAP and exogenous LiaX or N-terminal LiaX induces a significant increase in the DAP MIC with only a small increase in liaR expression but that liaR deletion prevents the acquisition of resistance (22). In this case, further research is needed to elucidate the mechanism of LiaR activation induced by excess LiaX in Enterococcus spp.

For a long time, bacteriocins were believed not to induce the emergence of resistance among susceptible bacterial strains. However, here we show that in the presence of AurA53- and EntL50-like bacteriocins, the susceptibility of E. faecium can be reduced due to the emergence of adaptive mutations in the LiaFSR-X response system; moreover, these mutations confer resistance to certain membrane-active peptide antibiotics, but mostly not to those targeting cell wall synthesis or intracellular structures. A systematic analysis of the obtained mutations in the individual components of LiaFSR-X indicated that despite their diversity, their ultimate effect was the activation of the response regulator LiaR, which then strongly and constitutively activated genes engaged in protective cell envelope remodeling, leading to cell resistance. It seems likely that similar mechanisms can function in other species, including those using nonorthologous systems of response to extracellular stressors. Given the unique relevance of LiaR and possibly also its counterparts in the development of resistance, one should consider designing inhibitors of their activity to be administered together with membrane-acting agents to limit this deleterious phenomenon.

This work was supported by the Foundation for Polish Science (FNP).