Purified phage lysins present an alternative to traditional antibiotics and work by hydrolysing peptidoglycan. Phage lysins have been developed against Gram-positive pathogens such as Bacillus anthracis and Streptococcus pneumoniae, where the peptidoglycan layer is exposed on the cell surface. Addition of the lysin to a bacterial culture results in rapid death of the organism. Gram-negative bacteria are resistant to phage lysins because they contain an outer membrane that protects the peptidoglycan from degradation. We solved crystal structures of a Yersinia pestis outer-membrane protein and the bacteriocin that targets it, which informed engineering of a bacterial–phage hybrid lysin that can be transported across the outer membrane to kill specific Gram-negative bacteria. This work provides a template for engineering phage lysins against a wide variety of bacterial pathogens.
Bacteriophages are viruses that infect bacteria in order to proliferate. They were recognized as a potential treatment for bacterial infections over 100 years ago , but the development of small-molecule antibiotics was more aggressively pursued because they are often smaller, cheaper to manufacture and broadly efficacious. More recently, the emergence of pan-resistant bacterial strains  has renewed interest in the use of phages as an alternative to traditional antibiotics . Rather than using whole phages, purified phage lysins have been produced to target specific Gram-positive pathogens [4,5], where the peptidoglycan (cell wall) is exposed on the cell surface. This approach, targeting specific pathogens rather than the entire gut flora, may reduce the development of multidrug-resistant bacterial strains.
Until now, Gram-negative pathogens were resistant to treatment by phage lysins because the peptidoglycan layer is sandwiched between the inner and outer membranes. A lysin would need to be transported across the outer membrane in order to access and degrade the peptidoglycan. By targeting an outer membrane iron transporter and modifying a bacterial toxin that uses it for entry into the cell, we discovered a way to apply phage therapy to Gram-negative bacteria. Concurrent with our work, Braun and co-workers  performed very similar experiments with the same outcome.
TBDTs (TonB-dependent transporters) and bacteriocins
TBDTs normally transport ferric iron or cobalamin across the outer membrane . In Yersinia pestis, a TBDT called FyuA is required for virulence in the early stages of a bubonic plague infection [8,9]. The bacterium synthesizes the siderophore yersiniabactin , secretes it into the extracellular space where it binds Fe3+ and imports the iron–yersiniabactin complex into the cell to meet its iron requirements. Transport across the outer membrane requires FyuA as the transporter and TonB–ExbB–ExbD plus protonmotive force to energize the process. Y. pestis also makes a bacteriocin called pesticin  that requires FyuA for transport across the outer membrane. Pesticin is encoded on a plasmid  that also contains the genes for the Pim (pesticin immunity protein) and Pla (plasminogen activator). Pla is important for virulence so pesticin targets and kills Yersinia strains that express FyuA on the cell surface but have lost the bacteriocin-containing plasmid. Thus the TBDT–bacteriocin couple maintains maximal virulence in the strain.
Crystal structures of FyuA and pesticin inform engineering of a Yersinia-targeting phage lysin
We solved the crystal structures of FyuA and pesticin to better understand what these proteins look like and how they function . As mentioned above, FyuA is a TBDT that spans the outer membrane. We expressed FyuA in Escherichia coli outer membranes, then extracted, purified and crystallized it using various detergents. FyuA is a 71 kDa protein whose structure consists of two domains (mirroring all known TBDTs): a 22-stranded β-barrel spans the outer membrane and an N-terminal plug domain blocks the pore (Figure 1). From studies on FyuA and other TBDTs, we know that the substrate is bound by loops of the plug domain and extracellular loops of the β-barrel. The plug domain contains a TonB box near its N-terminus that is required for import of either iron–yersiniabactin or pesticin. In FyuA, the TonB box consists of residues 8–14, STLVVTA, and this region is disordered in the structure.
Ribbon diagram of the FyuA crystal structure
We solved the structure of pesticin (40 kDa) after expressing it in E. coli, purifying and crystallizing it using standard techniques. The pesticin structure consists of two domains joined by a short linker (Figure 2). The N-terminal domain is a mixed α-helix/β-sheet structure, whereas the CTD (C-terminal domain) is primarily α-helical. The N-terminal domain contains a TonB box motif at the N-terminus, DTMVV, which is required for import of pesticin by FyuA . This region is disordered in our structure. Although pesticin had no structural or sequence similarity to any known colicin structure, we discovered that it has significant structural similarity to phage T4 lysozyme, even conserving two of the three active-site residues . Both T4 lysozyme and pesticin degrade peptidoglycans [15,16].
Ribbon diagrams of pesticin and the hybrid lysin
Structure determination of pesticin revealed how similar its CTD is to T4 lysozyme, so we undertook engineering of a bacterial–phage hybrid protein to create a phage lysin that targets Gram-negative bacteria. The hybrid protein consists of the N-terminal domain of pesticin, which we determined to be essential for binding to FyuA, fused to T4 lysozyme in place of the C-terminal pesticin (muramidase/killing) domain. Determination of the hybrid structure showed that the domain architecture is preserved and the two proteins (pesticin and hybrid) are similar in size and shape (Figure 2).
Pesticin and the hybrid lysin specifically target cells expressing FyuA
We tested the activities of pesticin and the hybrid lysin in E. coli by expressing FyuA in the outer membrane . When FyuA is present, both pesticin and the hybrid lysin are transported across the outer membrane, where they attack the peptidoglycan layer to kill the cell. E. coli cells that did not express FyuA were unaffected by both toxins. The Pim confers protection by residing in the periplasm where it inactivates pesticin that has been transported across the outer membrane . When we expressed FyuA in the outer membrane and Pim in the periplasm of E. coli, cells were no longer killed by pesticin, but they were still killed by the hybrid lysin. We visualized these events using EM (electron microscopy) on whole frozen E. coli cells and learned that the two toxins appear to kill differently: pesticin made discrete holes in the outer membrane which eventually resulted in cell lysis, whereas the hybrid lysin caused massive vesiculation of inner and outer membranes, resulting in total destruction of the cell. Although it appears to degrade peptidoglycan more aggressively than pesticin, the hybrid lysin may be imported less efficiently because the overall efficacy of pesticin was higher than that of the hybrid lysin. Nonetheless, since Pim does not inactivate the hybrid lysin, it can target bacterial strains expressing pesticin and Pim, extending its potential use to any organism expressing FyuA on the cell surface.
The efficacy of pesticin and the hybrid lysin against Yersinia and UPEC (uropathogenic E. coli) strains
We evaluated killing by pesticin and the hybrid lysin in two strains of Y. pestis and also in Yersinia pseudotuberculosis . Y. pestis KIM6+ expresses pesticin and Pim, and is killed by the hybrid lysin, but not by pesticin. Y. pestis KIM10+ does not make pesticin or Pim, so both toxins can kill this strain. Similarly, Y. pseudotuberculosis does not make pesticin or Pim and is killed by both toxins. FyuA is also expressed in certain pathogenic E. coli strains that cause urinary tract and kidney infections (UPEC) . We obtained 18 clinical isolates and evaluated our toxins in these strains too. In every case, if FyuA is expressed on the cell surface, pesticin and the hybrid lysin were able to kill the cells (none of these strains had the plasmid expressing pesticin and Pim). Therefore both pesticin and the hybrid lysin can kill clinically relevant bacterial cultures in broth and plate assays, and the next experiment to undertake is to determine efficacy in an animal model of disease.
Considerations for further development
In order to develop modified bacteriocins for clinical use, we need to improve both efficiency of transport across the outer membrane and enzymatic activity in the organism. Our EM observation that pesticin disrupts cell membranes at discrete locations, whereas the hybrid lysin causes massive destruction, and yet pesticin still kills more effectively , suggests that the hybrid lysin is transported across the outer membrane less efficiently than pesticin. Because both pesticin and the hybrid lysin appear to be too large to be transported by FyuA without some degree of unfolding , we made a thermostable hybrid lysin containing two disulfide bonds  that should prevent unfolding of the lysozyme domain. Although the thermostable hybrid lysin was still able to kill cells, it was 10-fold less active than hybrid lysin containing the wild-type T4 lysozyme domain. Additional data come from Braun and co-workers where single disulfides conferred a complete loss of activity . These results indicate that by increasing rigidity of the lysozyme domain, we have reduced the efficacy of the hybrid lysin. There is still much to be learned about how the killing domains of bacteriocins are imported and unanswered questions concern the following: (i) the precise pathway(s) remains unknown, (ii) whether energy is required at this stage is unclear, (iii) whether electrostatics or interactions between specific residues play a role, and (iv) whether size, shape and flexibility (or ability to unfold) are important. These are questions relevant not only to pesticin and the hybrid lysin, but also to the bacteriocin field as a whole. Once we better understand the mechanism(s) by which killing domains are imported, we will be better able to engineer effective phage therapy reagents.
In addition to improving the transport efficiency, it may be possible to increase and/or change the enzymatic activity of the killing domain. To increase in vivo activity of the pesticin–T4 lysozyme hybrid lysin, we could change the type of lysozyme used for killing. This might be beneficial because E. coli makes a periplasmic protein that inhibits vertebrate lysozymes called Ivy, and homologues are also found in Shigella, Yersinia, Klebsiella, Pseudomonas and other genera. Ivy preferentially not only inhibits vertebrate lysozymes, such as hen's-egg white lysozyme, but also inhibits the activity of phage T4 lysozyme by 40% . In contrast, some lysozymes are resistant to Ivy, with (bacteriophage) λ lysozyme, cauliflower lysozyme and mutanolysin from Streptomyces globisporus showing no reduction in activity. By substituting an Ivy-resistant lysozyme for the T4 lysozyme domain in our hybrid lysin, we might increase activity by 40% or more.
Another way to increase activity might be to change it. Rather than attacking peptidoglycan using a lysozyme-type activity, we could take advantage of the fact that bacteriocins are modular protein toxins with killing domains that include RNA and DNA nucleases, ion channels and domains that target peptidoglycans . By substituting a nuclease or ion channel killing domain for T4 lysozyme, we could completely change the mode of killing by the hybrid toxin while retaining specificity for FyuA. Since FyuA is required for virulence [8,9] and is primarily expressed in virulent strains [8,17,21–27], this strategy targets specific pathogens rather than causing widespread bacterial eradication. Conversely, we could retain T4 lysozyme and substitute a receptor-binding domain specific for a TBDT in a different organism for the FyuA-binding domain. This strategy should allow us to target other bacterial strains, extending the phage therapy approach beyond Yersiniae and E. coli.
How Bugs Kill Bugs: Progress and Challenges in Bacteriocin Research: A Biochemical Society Focused Meeting held at University of Nottingham, U.K., 16–18 July 2012. Organized and Edited by Colin Kleanthous (Oxford, U.K.), Chris Penfold (Nottingham, U.K.) and Dan Walker (Glasgow, U.K.).
P.L., N.N. and S.K.B. are supported by the Intramural Program of the National Institute of Diabetes and Digestive and Kidney Diseases of the National Institutes of Health.