The outer membrane (OM) of Gram-negative bacteria is a unique architecture with an asymmetric bilayer, which consists of phospholipids in the inner leaflet and lipopolysaccharide (LPS) in the outer leaflet [1]. In addition to phospholipids and LPS, there are a variety of bilayer-anchored lipoproteins and outer membrane β-barrel proteins (OMPs) in the OM [2]. Lipoproteins play an important role in several essential and non-essential functions, including the linkage between the OM and peptidoglycan [3], peptidoglycan synthesis [4, 5], pili and flagella assembly [6, 7], and protein and polysaccharide secretion [8, 9]. Porin, a major OMP of the OM, forms specific and non-specific channels that regulate the transport of hydrophilic molecules across the OM [10, 11].

The OM functions as an additional barrier that inhibits the transport of toxic molecules such as antimicrobial compounds. Therefore, compounds with a molecular weight (MW) of more than 600 Da cannot pass through the Gram-negative bacteria cell envelope. For example, vancomycin and daptomycin which have a MW of more than 1,400 Da are not able to penetrate the OM of Gram-negative bacteria [12]. This feature of Gram-negative bacteria is one of the major obstacles in novel antibiotic discovery targeting clinically important Gram-negative pathogens, such as Klebsiella pneumoniae, Pseudomonas aeruginosa, and Acinetobacter baumannii [13, 14]. Therefore, screening for inhibitors that block OM biogenesis is a good strategy for the development of novel antimicrobial agents. In this review, we will summarize recent findings about the transport and assembly of OM components and analyze trends in the development of antimicrobial compounds inhibiting OM assembly.

Lipoproteins and OMPs

Lipoproteins containing a so-called lipobox, a consensus sequence [LVI][ASTVI][GAS]C, are translocated across the inner membrane (IM) by the Sec or Tat system [15, 16]. The maturation of lipoprotein precursors occurs on the outer leaflet of the IM by sequential functions of phosphatidylglycerol/prolipoprotein diacylglyceryl transferase (Lgt), lipoprotein signal peptidase (LspA), and phospholipid/apolipoprotein transacylase (Lnt) [15]. Most mature lipoproteins are then transported to the OM by the LolABCDE pathway [3, 15, 16], and few lipoproteins with a so-called Lol avoidance signal (i.e. an Asp residue at position +2) remain in the IM [17].

The LolCD₂E complex belonging to the ABC transporter superfamily extracts lipoproteins from the IM and transfers extracted lipoproteins to the periplasmic chaperone LolA, resulting in the formation of a soluble lipoprotein-LolA complex (Fig. 1) [16, 18]. This step depends on ATP hydrolysis of the cytoplasmic ATPase LolD [18, 19]. LolE and LolC, which have a significantly structural similarity, seem to play distinct roles, recognition and binding of lipoproteins [20] and the transfer of lipoproteins to LolA [21], respectively.

Figure 1. The outer membrane assembly machines in E. coli. (A) The Lol pathway. After IM insertion of lipoprotein precursors by the Sec system, lipoprotein precursors are processed to lipoproteins. The lipoproteins extracted by the LolCD2E complex are transferred to LolB in the OM through the periplasmic protein LolA. Lipoproteins of LolB are released in the OM. (B) The Bam pathway. OMPs secreted by the Sec system are inserted into the OM by the Bam assembly machine. (C) The Lpt pathway. LPS extracted by the LptB2CFG complex are transferred to the LptDE complex in the OM through the transenvelope bridge of LptA that spans the entire periplasmic region. The LptDE complex inserts LPS into the outer leaflet of the OM. (D) Phospholipid transport systems. PbgA transports CL in the IM to the OM. The Mla pathway transports PG and PE in the OM to the IM. E. coli has two systems (the Pqi and Yeb pathway) with a central channel capable of mediating phospholipid transport, but it has not been determined yet whether the Pqi and Yeb pathway mediate only retrograde transport of phospholipids or bidirectional migration of phospholipids.

LolA interacts with lipoproteins through interaction between the large hydrophobic cavity of LolA and the acyl chains of lipoproteins, which shields the highly hydrophobic region of lipoprotein from the hydrophilic periplasm [22]. LolA transports lipoproteins to the OM lipoprotein LolB, which shares remarkably structural similarity with LolA [15]. The acyl chain transfer between LolA and LolB seems to be achieved by the directional mouth-to-mouth delivery driven by an affinity difference between the hydrophobic cavities of LolA and LolB for the acyl chain [21]. Lipoproteins of LolB are released in the OM, but its precise mechanism remains poorly understood [3]. The protruding loop of LolB seems to be important for lipoprotein insertion, as a LolB variant with a point mutation within this loop can receive lipoproteins from LolA, but is not able to mediate the OM insertion of lipoproteins [23]. All lol genes are essential in Escherichia coli[15], and although the lolA and lolDgenes are conserved in all proteobacteria, conservation of other genes varies [15]. For example, LolB is absent in α-, δ-, and ε-proteobacteria [3]. A recent study suggests that E. coli may have an alternate transport pathway independent of LolA and LolB proteins. Perhaps species without LolB seem to use a LolAB-independent alternate pathway as the main system for lipoprotein transport.

OMPs

OMPs are OM transmembrane proteins with a β-barrel structure consisting of antiparallel β-sheets [24]. OMPs synthesized in the cytoplasm are transported to the periplasm by the Sec system. The nascent OMPs in the periplasm are guided by several periplasmic chaperones, such as SurA and Skp, which deliver OMPs to the β-barrel assembly machinery (BAM) complex in the OM (Fig. 1)[25]. The BAM complex is an insertase machinery consisting of five proteins (BamABCDF). BamA, an OMP itself, has a highly conserved D15 domain forming a 16-strand β-barrel, which is essential for the insertase activity of the BAM complex [26], and five periplasmic polypeptide transport associated domains (POTRA domains) in the N-terminal region, which function as protein-protein interaction domains [27]. BamA forms a tight complex with four lipoproteins (BamBCDE). Based on structural and functional studies, there are two working models: the assistance model of BamA for spontaneous insertion of OMPs (BamA-assisted model) and the conformational change model of OMPs within BamA followed by insertion into the OM via lateral opening of BamA (BamA-budding model) [25], but their precise mechanism remains unclear. Because BamA is itself an OMP, there is a “chicken-or-the-egg” question about BamA assembly. Recent studies suggest that BamB and BamD assist in the early assembly of BamA, and then the BAM complex containing all components may complete the BamA biogenesis [28, 29].

LPS

LPSs, which are composed of lipid A, core oligosaccharide, and O-antigen polysaccharide chains, are transported by the lipopolysaccharide transport (Lpt) pathway consisting of seven essential proteins (LptABCDEFG) (Fig. 1) [30]. The Lpt pathway is divided into two subassemblies, the nonconventional ABC transporter LptB₂CFG located at the IM and LptDE located at the OM. Two sub-complexes are connected by the periplasmic chaperone LptA, which oligomerizes to form a transenvelope bridge that spans the entire periplasmic region from the IM to the OM [31]. LptA interacts with each other in a head-to-tail orientation [32] and their N-terminal and C-terminal regions interact with the C-terminal regions of LptC in the IM and the N-terminal regions of LptD in the OM, respectively [33]. Similar to the LolCD₂E complex, the LptB₂CFG complex extracts LPS from the outer leaflet of the IM using ATP hydrolysis and transfers it to LptC, an IM protein containing a single N-terminal transmembrane helix and a large soluble periplasmic domain [34]. LptC interacts with LptB₂CFG complex and delivers LPS to LptA [34]. LptA has a highly structural similarity (a so-called Lpt-fold) with the periplasmic domain of LptC [35] and the Lpt-fold has a hydrophobic pocket for binding to the lipid moiety of LPS [34]. Similar to the transfer of lipoproteins between LolA and LolB [21], the transfer of LPS from LptC to LptA seems to be achieved by the spontaneous delivery driven by an affinity difference between LptC and LptA for the lipid moiety [35]. Although the exact delivery mechanism of LPS to the OM by LptA and the number of LptA present in the transenvelope bridge are not fully determined, several studies show that LptA oligomerization is necessary for the interaction with LPS and the delivery through periplasm [36, 37]. The LptDE complex, comprising the β-barrel protein LptD and the lipoprotein LptE, forms a so-called plug-and-barrel structure with the entire insertion of LptE inserted into the β-barrel of LptD [38]. The N-terminal periplasmic domain of LptD has an Lpt-fold for interaction with LPS and seems to be involved in the delivery of LPS from LptA to the LptDE complex [30]. Several studies suggest that LPS may laterally pass through the β-barrel lumen of LptsD and may be directly inserted into the outer leaflet of the OM [39-41]. LptE seems to play various important roles in controlling the correct maturation of the LptD disulfide bonds [42], plugging the too large lumen of LptD [30, 43, 44], and disaggregating LPS during transport through interaction with negative charges of LPS [45].

Phospholipids

The envelope of E. coli is composed of three major phospholipids, the phophatidylethanolamine (PE, approxi-mately 70% of the total lipid), the phosphatidylglycerol (PG, approximately 20%), and cardiolipin (CL, approximately 10%) [2]. Unlike other OM components, very little is known about the transport of phospholipids from the IM to the OM. Several systems involved in maintaining OM asymmetry under stress conditions have been reported [2, 46, 47]. The outer leaflet of the OM consists of LPS and the inner leaflet is composed of phospholipids. The unique asymmetry of the OM seems to be important for its barrier function [48]. The phospholipids mislocalized in the outer leaflet of the OM may make a patch in the OM which is permeable to hydrophobic molecules, resulting in sensitivity to detergents and bile salts [48]. Several systems to restore the lipid asymmetry of the OM were identified: the phospholipase PldA degrades phospholipids in the outer leaflet of the OM [49] and the palmitoyltransferase PagP transfers the acyl chain (palmitate) of surface-exposed phospholipids to lipid A of LPS (hepta-acylation of LPS) [50]. The maintenance of lipid asymmetry (Mla) pathway mediates retrograde transport of phospholipids mislocalized in the outer leaflet of the OM (Fig. 1) [46]. This pathway comprises the MlaA-OmpC complex in the OM, the periplasmic protein MlaC, and the MlaB₂E₂DF₂ complex in the IM [2]. The mla genes are not essential, but their deletion mutants show phenotypes of OM permeability defect and an increased accumulation of phospholipids in the outer leaflet of the OM [46]. MlaA interacts with the OM porin OmpC [51]. The deletion of the ompC gene results in a phenotype similar to the mlaA deletion mutant, suggesting that OmpC may mediate the extraction of mislocalized phospholipids, although its mechanism has not been determined. E. coli has two additional systems involved in the maintenance of lipid asymmetry, the PqiABC complex and the YebST complex (Fig. 1) [47]. TLC analysis using purified proteins showed that members (MlaD, PqiB, and YebT) of the mammalian cell entry (MCE) protein family bind to phospholipids [47]. The EM structures of three MCE proteins suggest that PqiB and YebT may form a syringe-like architecture and an elongated tube, respectively, which span the entire periplasmic region from the IM to the OM [47]. It has not been determined yet whether PqiABC and YebST mediate only retrograde transport of phospholipids or bidirectional migration of phospholipids. Notably, proteins involved in the bulk transport of phospholipids from the IM to the OM have not been identified yet. Recently, the IM protein PbgA was identified as the CL transporter in Salmonella [52], but it remains to be evaluated whether this protein is a general transporter of CL or is involved in CL transport under stress conditions.

The development of antimicrobial agents inhibiting the transport of OM components, including lipoproteins, OMPs, LPS, and phospholipids, would provide an entirely new class of antimicrobial agents. We analyzed all studies on the development of inhibitors targeting OM assembly machines.

Inhibitors of Lipoprotein Transport

The identification of chemical compounds that inhibit the Lol pathway has been extensively researched (Table 1). The first identified compounds, CCT-00431 and CCT-00432, were found by an inhibition assay of the LolA-dependent release of L10P, an Lpp derivative, from the LolCDE-L10P complex [53]. CCT-00431 and CCT-00432 inhibited the formation of the LolA-L10P complex by 30% and 60%, respectively [53]. The minimal inhibitory concentrations (MICs) of CCT-00431 and CCT-00432 for E. coli strains were around 16 μg/ml, and their MICs for the lpp-deleted mutant increased up to 128 μg/ml [53], suggesting that the target of these compounds is the Lol pathway and that the screening assay is useful to identify inhibitors targeting the Lol pathway. In 2009, a novel chemical compound, MAC13243, targeting the LolA protein was identified through a high-throughput screen for suppressors of growth inhibitory compounds by the high expression of the essential genes [54]. This assay is based on the assumption that the high expression of an essential target of an antimicrobial agent may suppress the growth inhibitory effect by the antimicrobial agent. The accuracy of the assay was proven by the identification of targets of six well- known antimicrobial agents as high-copy suppressors of lethality [54]. For example, the folC gene involved in the folate biosynthesis was found to be a suppressor of chemical lethality of sulfamethoxazole, an antimicrobial agent that inhibits the folate synthesis. This screen assay identified a novel inhibitor of LolA, MAC13243. MAC13243 inhibited the function of LolA through direct interaction and the dissociation constant (KD) between MAC13243 and LolA was 7.5 μM [54]. The MICs of MAC13243 for E. coli and P. aeruginosa were 16 and 8 μg/ml, respectively, and the MIC of MAC13243 was not affected by the overexpression of the acrB gene encoding an efflux pump [54]. Because the MICs of MAC13243 for Gram-positive bacteria, such as Bacillus subtilis and Staphylococcus aureus, were significantly higher (>256 μg/ml) than those of Gram-negative bacteria, MAC13243 is an antimicrobial lead compound against Gram-negative bacteria selectively. In an aqueous solution, MAC13243 can be degraded into 3,4-dimethoxyphenethylamine and S-(4-chlorobenzyl)isothiourea [55]. S-(4-chlorobenzyl) isothiourea is an analog of A22 [S-(3,4-dichlorobenzyl) isothiourea], a well-known molecule inhibiting the actin- like bacterial cell shape determining protein MreB [56]. Notably, like MAC13243, the antimicrobial activities of S- (4-chlorobenzyl)isothiourea and A22 were suppressed by LolA overproduction, while their activities were sensitized by LolA depletion [55]. These results suggest that S-(4- chlorobenzyl)isothiourea and A22 have a mode of action similar to MAC13243. This assumption was confirmed by the interaction of these compounds with LolA. Therefore, S-(4-chlorobenzyl)isothiourea and A22 seem to have a similar mode of action that includes inhibition of LolA as well as MreB.

Several compounds inhibiting the LolCD₂E complex were identified [57-59]. A pyrazole-containing compound was found by a high-throughput screen using an AmpC reporter assay [58]. The ampC β-lactamase gene is an inducible gene of Citrobacter freundii which can be used as a marker of inhibition of cell wall biosynthesis [60]. To find novel compounds that inhibit cell wall biosynthesis, a high-throughput phenotypic screening using the C. freundii AmpC reporter assay was performed [58]. A pyrazole- containing compound with an MIC of 8 μg/ml for E. coli was identified as an inhibitor of both LolC and LolE. Mutations in either lolC or lolE cause significantly large shifts in the MIC of this compound (>128 μg/ml), whereas the MICs of other antibiotics were not changed [58], suggesting that this compound may inhibit LolC and LolE activities. A spheroplast release assay showed that the pyrazole-containing compound actually inhibits Lpp release from spheroplasts to LolA. Two pyridineimidazole- containing compounds (compound 1 and 2) were also identified as inhibitors of LolC and LolE through a high- throughput screen for a compound that inhibits the growth of a permeabilized E. coli strain [57]. The screen identified two compounds that inhibit Lpp release from spheroplasts to LolA and the resistance against these two compounds was found in bacterial cells with mutations in the lolC or lolE gene [57]. Compound 2, a close analog of compound 1, showed greater antimicrobial activity: the MIC of compound 2 against E. coli and Haemophilus influenzae were 8-fold lower than compound 1 [57]. Recently, a novel inhibitor G0507 of the LolCDE complex was identified by a phenotypic screen of E. coli growth inhibition and additional selection of compounds inducing the extracytoplasmic σ^E stress response, which is shown to be activated by OM biogenesis defects [61, 62]. The pyrrolopyrimidinedione- containing compound G0507 interacted with the LolCDE complex and stimulated its ATPase activity [59]. Fully- processed Lpp were accumulated in the IM of E. coli cells, but not in the OM, after treatment with this compound. These results suggest that the LolCDE complex is a molecular target of G0507. Notably, the Q258K mutation of LolC did not affect the interaction between G0507 and the LolCDE complex, but abolished stimulation of ATPase activity by G0507, and cells with this mutation were resistant to G0507 [59]. These results indicate that the stimulation of ATPase activity is important for the mode of action of G0507. The MIC of G0507 against E. coli MG1655 is >64 μg/ml, but the MIC of its derivative G0793 is 16 μg/ml [59].

Table 1 . Antimicrobial inhibitors targeting the OM assembly machines..

Assembly machine	Inhibitor	Cellular target	IC50a	MIC (μg/ml)	MIC in the resistant mutant	Reference
The Lol pathway	CCT-00431 and CCT-00432	LolA	N.D.b	16 (E. coli NIHJ JC-2)8 (E. coli M101)	8 (Parent)32 or 128 (Δlpp)	[53]
	MAC13243	LolA	N.D.	16 (E. coli ATCC25922)4 (Pseudomonas aeruginosa PAO1)	N.D.	[54]
	S-(4-chlorobenzyl)iso-thiourea and S-(3,4-dichlorobenzyl)iso-thiourea (A22)	LolA	IC50=150 μM [S-(4-chlo-robenzyl)isothiourea]IC50=200 μM (A22)	2 (E. coli MG1655)	N.D.	[55]
	Pyrazole-containing compound	LolC and LolE	N.D.	8 (E. coli ATCC25922)0.125 (E. coli ATCC25922 ΔtolC)32 (Haemophilus influenza ATCC49247)>64 (Pseudomonas aeruginosa PAO1)	0.125 (Parent)>128 [LolC(G254V)]>128 [LolE(G195S)]>128 [LolE(P365C)]>128 [LolE(P367Y)]	[58]
	Pyridineimidazole-containing compound 1	LolC and LolE	N.D.	32 (E. coli ATCC25922)0.25 (E. coli ATCC25922 ΔtolC)2 (Haemophilus influenza ATCC49247)>64 (Pseudomonas aeruginosa PAO1)	0.125 (Parent)>64 [LolC(N265K)]>64 [LolE(I59N)]64 [LolE(P372L)]>64 [LolE(L371P)]	[57]
	Pyridineimidazole-containing compound 2	LolC and	N.D.	4 (E. coli ATCC25922)<0.06 (E. coli ATCC25922 ΔtolC)0.25 (Haemophilus influenza ATCC49247)>64 (Pseudomonas aeruginosa PAO1)	<0.06 (Parent)>64 [LolC(N265K)]>64 [LolE(I59N)]8 [LolE(P372L)]>64 [LolE(L371P)]	[57]
	Pyrrolopyrimidinedione compound G507	LolCDE	N.D.	>64 (E. coli MG1655)0.5 (E. coli MG1655 ΔtolC)1 (E. coli MG1655 imp4213)16 (TQ/he MIC of G0793 against E. coli MG1655)	N.D.	[59]
The Lpt pathway	Compound 1 and 2	LptB	IC50=25 μM (compound 1) IC50=17 μM	N.D.	N.D.	[63]
	4-Phenylpyrrolocarba-zole derivatives, 1a and 1b	LptB	IC50=119 μM	>100 (E. coli MC4100) 25(E. coli NR698)	N.D.	[64]
	Peptidomimetic compound L27-11	LptD	N.D.	0.01 (Pseudomonas aeruginosa ATCC27853)0.004 (Pseudomonas aeruginosa PAO1)	0.06 (Parent)>64 (Duplication of residues 210-215 of LptD)	[68]
	β-hairpin-shaped peptidomimetic compound LB-01	LptD	N.D.	0.008 (Pseudomonas aeruginosa ATCC27853)0.015 (Pseudomonas aeruginosa PAO1)	N.D.	[71]
	β-hairpin-shaped peptidomimetic compound Dab4NMe	LptD	N.D.	0.005 (Pseudomonas aeruginosa ATCC27853)0.009 (Pseudomonas aeruginosa PAO1)	N.D.	[71]
The BAM pathway	β-hairpin macrocyclic peptide JB-95	BamA and LptD	N.D.	0.25 (E. coli ATCC25922)1 (Acinetobacter baumannii ATCC17978)4 (Pseudomonas aeruginosa PAO1)2 (Staphylococcus aureus ATCC29213)	N.D.	[73]

^aThe half maximal inhibitory concentration..

^bNot determined..

Inhibitors of LPS Transport

Chemicals inhibiting the ATPase activity of the ATPase component LptB of the LptB₂CFG complex were screened by a continuous assay that couples ATP hydrolysis to NADH oxidation [63]. Two compounds, 1 and 2, that have different structural classes, inhibited the ATPase activity of LptB with an IC50 of 25 and 17 μM, respectively [63]. The value of the MICs of these compounds was not determined. In 2013, other inhibitors of LptB were identified by the same biochemical assay, the continuous assay coupling ATP hydrolysis to NADH oxidation [64]. A compound 4-phenylpyrrolocarbazole derivative that had previously been known as a potent inhibitor of a eukaryotic kinase Wee1, which regulates mitosis entry [65], was identified as an inhibitor of LptA [64]. Several derivatives of 4- phenylpyrrolocarbazole were additionally tested for inhibition of the ATPase activity of LptB. Compounds 1a and 1b showed a weak antimicrobial activity (MIC of >100 μg/ml) against the E. coli strain MC4100, but MICs against the leaky E. coli strain NR698 were decreased (25 μg/ml in compound 1a and 12.5 μg/ml in compound 1b) [64]. Notably, the IC50 of compound 1b was 6-fold lower against LptB alone (19.6 μM) than against the LptB2FGC complex (119 μM), suggesting that compound 1b is more potent against LptB alone than the LptB2FGC complex.

Several β-hairpin-shaped peptidomimetics (small protein- like chains designed to mimic a peptide) targeting the β- barrel OM protein LptD were developed by optimization of peptidomimetics based on the antimicrobial peptide protegrin I (PG-I) [66]. PG-I is a broad-spectrum antimicrobial peptide against Gram-negative and Gram-positive bacteria [67], but optimized peptidomimetics, L26-19 and L27-11, display a narrow-spectrum antimicrobial effect only against P. aeruginosa and other Pseudomonas spp., but not other Gram-negative and Gram-positive bacteria [68]. The MICs of L27-11 against P. aeruginosa ATCC27853 and PAO1 were 0.01 and 0.004, respectively, but all MICs against other strains, including Gram-negative bacteria (A. baumannii, K. pneumoniae, and E. coli) and Gram-positive bacteria (Enterococcus faecalis and S. aureus), were >64 μg/ml [68]. L26-19 displayed a similar antimicrobial activity. L26-19 and L27-11 displayed no cellular lytic activity but exhibited internal accumulation of membrane-like materials in cells and impairment of the OM integrity. A mutation in the lptD gene increased the MIC of L26-19 against P. aeruginosa PAO1 up to >64 μg/ml, and the interaction between L27-11 and LptD was confirmed by photoaffinity labeling analysis [68]. A recent study showed that L27-11 interacts with the periplasmic domain of LptD and its K_D is approximately 13 nM [69]. These results indicate that LptD is a cellular target of these peptidomimetics. An alanine scan experiment to examine the critical residue for antimicrobial activity of each side chain in L27-11 showed that Trp2 and Trp8 residues located on the opposite face of each side chain are critical [70]. A novel modified derivative of L27-11, called LB-01, with MICs of 0.008 and 0.015 μg/ml against P. aeuruginosa ATCC27853 and PAO1, respectively, was developed [70, 71]. An N-methyl scan experiment using N-methylation of each peptide bond showed that residues on both sides of LB-01 that do not participate in the intramolecular hydrogen bonding may play an important role in hydrogen-bonding interactions with LptD [71]. Additionally, these experiments revealed that several derivatives with N-methylation have more potent antimicrobial activity than LB-01. For example, a Dab4NMe analog of LB-01 displayed MICs of 0.005 and 0.009 μg/ml against P. aeuruginosa ATCC27853 and PAO1, respectively [71]. These results show that there is a novel avenue for optimization of peptidomimetics. An L27-11- closely-related clinical candidate Murepavadin (also called POL7080) developed by Polyphor AG has successfully completed phase-II clinical trials (clinical trials identifier NCT02096328) against pseudomonas lung infections [72], suggesting that proteins involved in the LPS transport pathway can be a potential target of antimicrobial agents.

Inhibitors of OMP Transport

One study investigated an inhibitor that targets the BAM complex. Similar to L27-11, a β-hairpin macrocyclic peptidomimetic JB-95 was identified as an inhibitor of β-barrel OM proteins BamA and LptD [73]. JB-95 induced no cellular lytic activity like L27-11, but selectively disrupted the OM integrity, but not the IM, and induced rapid depletion of β-barrel OM proteins [73]. Photolabeling experiments revealed the interaction of JB-96 with BamA and LptD. The MICs of JB-95 against E. coli ATCC25922, A. baumannii ATCC17978, P. aeruginosa ATCC27853, and S. aureus ATCC29213 were 0.25, 1, 4, and 2 μg/ml, respectively [73]. These results propose a novel avenue for developing antimicrobial agents targeting β-barrel proteins of the OM.

Upon the revelation of a more complete picture of the transport system of OM components, the possibility of developing antimicrobial agents targeting OM transport machines was examined. The most successful case is Murepavadin, an L27-11-based peptidomimetic inhibitor of LptD, which has successfully completed phase-II clinical trials [72]. Because several additional optimization procedures of L27-11 have been suggested [70, 71], there is the possibility of developing more potent antimicrobial agents. Although there have been various attempts at developing new antibiotics targeting the Lol system, an inhibitor with an MIC in the nanomolar range has not yet been developed. The high-throughput screen of inhibitors based on peptidomimetics may be useful for the development of inhibitors of the Lol system. Because the transport system of phospholipids from the IM to the OM remains undetermined, extensive studies are required in this field. The phospholipid transport system has potential to be a novel target for the development of new antimicrobial agents against Gram-negative bacteria.

This work was supported by research grants from the Advanced Biomass R&D Center (ABC) of the Global Frontier Project funded by the Ministry of Science, ICT and Future Planning (ABC-2015M3A6A2065700), and the Bio & Medical Technology Development Program of the NRF funded by the Korean government, MSIT (NRF-2017M3A9E4078017).

The authors have no financial conflicts of interest to declare.