Antibiotic resistance is a growing global public-health problem1. One group of bacteria, called Gram-negative bacteria, is particularly difficult to treat, because the cells are shielded by a double-membrane envelope, which constitutes a formidable barrier to antibiotics2. When antibiotics do breach the membranes, these bacteria often use efflux pumps to remove the drugs3,4. Three papers (two in Nature5,6 and one in the Proceedings of the National Academy of Sciences7) now describe antibiotics that overcome these obstacles by targeting, directly or indirectly, a protein integral to the outer membrane.
The outer membrane of Gram-negative bacteria contains lipopolysaccharide (LPS) molecules in its outer leaflet, with outer-membrane proteins (OMPs)8 spanning the entire outer membrane. OMPs are folded into the membrane by a protein complex called the β-barrel assembly machine (BAM), the central component of which, BamA, is an OMP itself (Fig. 1). Because BamA is exposed to the extracellular space, it could be an Achilles heel in the bacterial shield — inhibitors that access BamA would not need to penetrate the cell. Indeed, a proof-of-concept study9 has shown that this approach inhibits OMP folding and compromises membrane integrity, albeit by an unknown mechanism.
The three current studies took different approaches to develop antibiotics against Gram-negative bacteria. In the first, Imai et al.5 turned to Gram-negative bacteria that live symbiotically in the gut of nematode worms and can secrete antibiotics to fend off competing bacteria — including other Gram-negative species. A screen of the secretions from 22 of these symbionts revealed a Gram-negative-targeting antibiotic, which the authors named darobactin.
Darobactin displayed antibiotic activity against multiple Gram-negative bacteria, both in vitro and in infected mice, including against several drug-resistant human pathogens such as polymyxin-resistant Pseudomonas aeruginosa and β-lactam-resistant Klebsiella pneumoniae and Escherichia coli. Darobactin was not toxic to human cells at the concentrations at which it was an effective antibiotic.
Next, Imai et al. asked what bacterial molecule darobactin targets. The group identified three strains of E. coli that were resistant to darobactin and showed that each harboured mutations in the bamA gene. The mutations all changed amino-acid residues in the same region of BamA’s protein structure, suggesting a putative binding site for darobactin that would be accessible from the extracellular space.
The authors provided evidence that darobactin and BamA bind to each other directly, using a technique called isothermal titration calorimetry, which measures the heat changes associated with physical interactions between molecules. The results of nuclear magnetic resonance (NMR) spectroscopy experiments were also consistent with direct binding, and suggested that the antibiotic stabilizes the protein in a potentially inactive conformation.
The researchers next showed that darobactin inhibits the ability of an isolated BAM complex to perform its OMP-folding function in vitro, consistent with direct BamA targeting. However, only one of the resistant BamA mutants showed reduced inhibition by darobactin in this assay. A test of whether darobactin–BamA binding is impaired in the bamA mutants could be used in the future to confirm BamA as the molecular target.
In the second of the current studies, Luther et al.6 focused on analogues of an existing antibiotic, murepavadin10, which targets a surface-exposed protein called LptD that is involved in assembling LPSs in the outer membrane8. Murepavadin displays potent but narrow antibiotic activity against P. aeruginosa10. The authors therefore screened for murepavadin analogues that had antibiotic activity against other Gram-negative species.
Luther and colleagues chemically linked the compounds identified through this screen to a portion of another antibiotic, polymyxin B, that binds to LPS directly11. Intact polymyxins efficiently disrupt bacterial membranes and kill cells, but are rather toxic to humans12. The researchers hoped that linking just the LPS-binding portion of polymyxin B could increase the membrane targeting of their murepavadin analogues. Indeed, their strategy produced several chimaeras that had potent activity, both in vitro and in mice infected with K. pneumoniae, P. aeruginosa, E. coli and other Gram-negative bacteria, including drug-resistant strains. Notably, the chimaeras showed low toxicity in mice.
It might be expected that the chimaeras would target LptD, but when Luther and colleagues tested for interacting partners, they found evidence of BamA targeting. The authors analysed strains of K. pneumoniae that showed resistance to the chimaeras. They found that resistant strains carried mutations in several genes, including bamA and genes responsible for LPS modification. Reintroduction of the wild-type bamA gene into the resistant strains led to increased sensitivity to the chimaera, indicating that BamA has a role in the antibiotic’s mechanism of action.
Direct chimaera–BamA binding was confirmed with in vitro assays in which the authors fluorescently labelled the chimaeras and monitored changes in fluorescence that indicate binding to a large protein such as BamA. As with darobactin, NMR experiments suggested that chimaera binding stabilizes BamA in a potentially inactive conformation, consistent with direct BamA targeting. However, when the bacteria were treated directly with the chimaeras, both the outer and inner membranes were rapidly permeabilized; this suggests that the compounds might act directly on the membrane. The results raise the possibility that the chimaeras act in a similar way to polymyxins, with binding to BamA strengthening their membrane targeting.
In the third study, Hart et al.7 identified a compound, MRL-494, that had similar antibiotic potency against both wild-type E. coli and a mutant defective in outer-membrane integrity and efflux mechanisms, suggesting that this antibiotic might not need to penetrate the cell to exert its activity. In vitro, MRL-494 exhibited moderate potency against Gram-negative pathogens, including K. pneumoniae and P. aeruginosa. The efficacy of MRL-494 in animal models remains to be tested.
The authors showed that treatment of E. coli with the compound resulted in decreased abundance of OMPs in the outer membrane, indicating BamA as a possible target. In support of this possibility, Hart et al. identified a bamA mutation that confers resistance to MRL-494 in E. coli. They showed that, whereas MRL-494 inhibited normal folding of a model OMP in E. coli cells expressing wild-type bamA, it had less effect on the resistant cells. The researchers found that MRL-494 stabilizes BamA against heat-induced protein aggregation in cells, suggesting an interaction between the two. However, MRL-494 stabilizes the resistant bamA mutant to a similar extent. Furthermore, MRL-494 displays similar potency against Gram-positive bacteria, which lack BamA. Therefore, in Gram-negative bacteria, MRL-494 might inhibit BamA directly or might target the outer membrane and affect BamA function indirectly.
Together, these studies describe new antibiotics that are active against difficult-to-treat Gram-negative bacteria. Given the compounds’ size and chemistry, they are likely to act at the cell surface, bypassing the need to breach the permeability barrier. Imai et al. provided compelling evidence that BamA is the target of darobactin, including a putative binding site, to be confirmed by demonstrating reduced binding to resistant mutants. The chimaeric compounds both seem to bind BamA and LPS. But, as is also the case for MRL-494, further experiments will be required to determine whether their activity is caused by direct effects on BamA.
Future research to identify specific BamA binding sites for any of the compounds, and to examine the mechanism by which antibiotic binding impairs BamA activity, would provide a platform for further antibiotic development. Such research might also shed light on how BAM mediates the insertion and folding of OMPs, which is poorly understood.
Darobactin and MRL-494 are initial lead compounds, and medicinal-chemistry efforts could yield more-potent and effective analogues. Preclinical studies aimed at determining their toxicity in animal models will also be important. Luther and colleagues’ chimaeras are at a more advanced stage of development, because, as the authors show, they have potent in vivo activity as well as favourable toxicity, pharmacokinetics and pharmacodynamics in animal models. The future looks promising for this newly discovered class of antibiotic.
Nature 576, 389-390 (2019)