The relentless increase in antibiotic resistance among all major groups of bacterial pathogens shows no sign of abating. The situation is exacerbated by a marked decline in the number of new antibiotics entering the marketplace. It is essential that new ways to treat severe bacterial infections are investigated before the antibiotic well runs dry. This review covers many promising approaches, some novel and some based on old ideas that were not considered viable when clinicians were able to exploit a wide palette of cheap and effective antibacterial chemotherapeutics. These approaches include the use of photosensitive dyes, bacteriophage and phage-encoded proteins, and agents that compromise virulence and antibiotic-resistance machineries. I also make a case for continuing in some form with tried and trusted platforms for drug discovery that served society well in the past.
We continue to witness the erosion of antibiotic efficacy due to the emergence and spread of resistance genes in major groups of bacterial pathogens, as bacteria respond to the enormous selective pressure imposed by the worldwide use and overuse of antibacterial chemotherapy and the often unjustified application of antibacterial drugs in agriculture and animal husbandry. To limit the impact of antibiotic resistance, it is essential that the development pipeline is replenished with novel active structures that possess the properties that would allow them to be transformed into successful medicines. The reasons for the failure to generate new agents for the treatment of bacterial disease are many and varied and have received much recent attention [1–4], with little impact on the rate or extent of drug discovery.
The steady decline in the rate of discovery of novel molecules, drug scaffolds and antibacterial pharmacophores increased the difficulty and the cost of identifying novel antibiotics by traditional methods. This coincided with the introduction within the pharmaceutical industry of target-based drug discovery and an over-reliance on emerging molecular technologies such as whole-genome sequencing . It has been often assumed that genomic information will provide an abundance of new targets and enhance the capacity to identify new agents within existing chemical diversity [6,7]. Gene deletion will identify essentiality, but essential gene products may not represent viable targets for chemical intervention; many are components of metabolic pathways that lead to compounds required for growth in vitro but may not be necessary during infection, when nutrients, intermediates and biosynthetic products may be obtained from the host. In this context, platensimycin, a secondary metabolite from Streptomyces platensis that selectively inhibits bacterial type II fatty acid synthesis (FASHII) in Gram-positive bacteria , has well-established efficacy in experimental murine infections, but it has been claimed that FASHII is not essential if streptococcal cultures are supplemented with fatty acids or human serum [9–11]; this controversy is currently unresolved. More than three-quarters of approved, clinically useful antibacterial agents are natural products or drugs based on natural product scaffolds  derived without exception from moulds and actinomycetes, notably Streptomyces spp. . The recent explosion in genome sequencing has indicated that much chemical diversity and bioactivity remains to be discovered and that cryptic secondary metabolites may be expressed at very low levels under laboratory conditions [14,15]. The identification of biosynthetic clusters in genomes and metagenomes of uncultured populations and rare taxa suggests that traditional screening methods could be adapted along with new technologies to evaluate less abundant novel compounds from Streptomyces and other genera.
It is noteworthy that useful antibiotics target, not soluble gene products located in the bacterial cytoplasm, but components of machineries involved in cell wall biosynthesis , protein synthesis  and DNA replication , as well as associated macromolecular structures such as the cytoplasmic membrane . Here, I explore alternative therapeutic paradigms that recognise this fundamental tenet.
More of the same?
Traditional, historically highly successful natural product screening was largely discarded by the pharmaceutical industry to be replaced by high-throughput robot-guided biochemical assays employing soluble protein targets and large compound libraries, in part because conventional, ‘mature’ screening programmes invariably yielded many more known antibacterial compounds than novel structures. A mathematical model based on the cumulative number of secondary metabolites obtained from Streptomyces spp. has been used to determine the number of antimicrobial compounds from this genus to be of the order of 100 000, only a small fraction of which has been unearthed so far . The relative decline in discovery appeared to be due to a decline in screening efforts rather than an exhaustion of compounds; the authors suggest that if screening is maintained or increased, the rate of discovery of new compounds will not decline for a period of several decades.
The large majority of antibacterial agents in use today are derived from secondary metabolites produced by cultured soil microorganisms using the discovery platform introduced after the Second World War by Selman Waksman at Rutgers University . Soil is a rich source of microbial diversity, with up to 5000 species in 1 g of soil containing ∼109 bacteria , and has yielded a wide array of chemical structures with an impressive range of pharmacological activities . The intensive and successful search for antibacterial metabolites from soil microorganisms was driven by the perception that elaboration of antibiotics conferred a competitive advantage in a hostile and crowded environment, although Davies and colleagues [23,24] have recently emphasised that microbial communities are highly communicative and the principal role of microbial secondary metabolites may be to facilitate cross-talk to maintain rather than perturb microbial diversity in complex environments. Conventional culture techniques lead to the recovery of only ∼1% of the microbial soil population and this realisation precipitated the development of novel methods to isolate ‘uncultivable’ members of complex microbial communities as an untapped source of new antibiotics [25,26]. This approach has already led to the discovery from soil of a new antibiotic belonging to a previously unrecognised structural class ; teixobactin inhibits Gram-positive and mycobacterial cell wall synthesis by binding to precursors of peptidoglycan and teichoic acid and is at an early stage of preclinical development .
It is to be hoped that this promising work leads to a revival of screening of bacterial cultures for small-molecule active metabolites, especially when supplemented with new technologies to facilitate pure cultures of bacteria that often exist only in complex communities. Soil is one of many environments that play host to diverse and complex microbial communities and the search for novel antibiotics could be extended to include lakes, rivers, sediments, marine environments and other underexploited niches . Oceans cover over 70% of the world's surface with a median depth >3000 m, and their flora and fauna have proved to be rich sources of pharmacologically active agents that are being exploited as anticancer, antiviral and antipsychotic drugs [30,31]. However, no antibiotic has been developed from marine sources, and marine microorganisms remain relatively unexploited as potential sources of antibacterial agents [29,32], even though the estimated 1029 marine bacteria and archaea represent enormous potential for antibacterial drug discovery that is only now beginning to be addressed . The availability of large culture collections of marine microorganisms and compound libraries of marine-derived metabolites would lend support to searches for novel lead compounds.
Disruption of the bacterial cell
Antibiotics possess discrete modes of action that present target bacteria with an opportunity to bypass the susceptible metabolic step, prevent the drug from reaching its target or inactivate the agent by enzymatic modification or degradation. Treatment modalities that kill by physical disruption of bacterial integrity by simultaneously attacking multiple biomolecular sites within the cell are likely to both circumvent antibiotic-resistance mechanisms and prevent the development of resistance to the disrupting agent. Proliferating cells, including bacteria, may selectively accumulate photosensitive dyes and the cells are killed if the dye is activated by light of appropriate wavelength. Thus, photoactivation of tetrapyrrolic molecules, such as porphyrins, phthalocyanines and bacteriochlorins, with red light (λ650–800 nm) in the presence of oxygen leads to cell death due to singlet oxygen (1O2)-mediated peroxidative damage and, to a lesser extent, the generation of free radicals . 1O2 has a short lifetime of <0.04 µs and, as a consequence, a radius of action of <0.02 µm ; its lethal effect is therefore restricted to the site of photoactivation. Although photodynamic effects were initially realised with bacteria, photodynamic therapy (PDT) in the modern era has been exploited predominantly as a modality for the ablation of solid tumours  and to a lesser extent wet age-related macular degeneration, psoriasis, atherosclerosis and viral infections . The most coherent light sources for PDT are lasers and focusing light on the affected area provides a degree of selectivity that minimises damage to healthy tissue. Gram-positive bacteria are susceptible to photoinactivation with a wide variety of dyes and second-generation cationic photosensitisers have been developed that overcome the barrier imposed by the outer membrane to efficiently sensitise Gram-negative bacteria , antibiotic susceptible and multi-drug-resistant pathogens alike . 1O2 and free radicals will cause damage to a wide range of cellular components, including proteins, nucleic acids and lipids, and only dyes of the phenothiazinium class have been found to be efflux pump substrates . Clearly, PDT will have limited utility for the treatment of disseminated infections but has been considered for localised conditions, such as wound infections, burns, soft tissue infections and abscesses, and surface infections of the cornea and skin . It is particularly effective at disrupting biofilms, killing the component bacteria, and as such is undergoing clinical investigation for the treatment of oral conditions such as periodontitis . This interesting treatment modality may find niche applications for reduction in bioburden of complex polymicrobial dental infections.
Therapy with lytic bacteriophage, like PDT, relies on physical disruption of the infectious agent, with a high degree of selectivity (usually species-specific) determined by the nature of the receptor-mediated host–phage interaction that initiates the lytic cycle: phage therapy has, therefore, the potential to eradicate an infection without disturbing the host microbiota, but needs to be combined with precise diagnosis and species identification for maximum effect. The relentless increase in global antibiotic resistance has precipitated a re-appraisal of phage therapy following its abandonment by the early 1940s in the West (but not in the former Soviet Union), and phage-based therapeutic strategies are currently being pursued commercially, predominantly against difficult-to-treat infections caused by Pseudomonas aeruginosa, Staphylococcus aureus and Clostridium difficile . There is no doubt that phages are effective in eradicating a broad range of non-systemic infections in experimental animals , but regulatory conundrums have in the past proved to be a major barrier to the commercial exploitation of phage therapy  and development of phage medicines will require a major change in the mindset of regulatory authorities [44,45]. As a consequence, recent clinical studies have been expedited in Europe under national Medical Ethical Committee guidelines . Non-systemic administration of lytic phage preparations is considered safe  as phages are inherently harmless to eukaryotic cells but, as lysogenic phages often carry genes that alter the pathogenic potential of their hosts , their use should be strongly discouraged. Some evidence of lowering of regulatory barriers is now evident: the U.S. Food and Drug Administration and the U.S. Department of Agriculture's Food Safety and Inspection Service have approved the use of phage cocktails against Listeria monocytogenes as direct food additives in ready-to-eat meat and poultry , a Salmonella-specific cocktail for direct application onto poultry, fish, shellfish and fresh and processed fruits and vegetables , and a cocktail for application to red meat parts and trim, which will subsequently be ground, to reduce Escherichia coli O157:H7 contamination .
Resistance to phage may develop in the bacterial host by mutation, recombination and horizontal transfer of resistance genes, by activation of existing restriction–modification or abortive infection systems, and by stationary or other phases of growth that confer temporary resistance . The capacity to generate phage-resistant bacterial mutants during therapy could be minimised by a deeper understanding of the non-linear kinetics of phage–bacteria interactions  and by incorporating features of molecular interactions between host and parasite into the selection and design of a therapeutic phage . For example, the pathogenesis of neuroinvasive E. coli strains carrying the K1 capsule depends on the expression of the surface-located polysialic acid K1 structure : most laboratory-generated mutants resistant to K1-specific phage are non-capsular and therefore unlikely to survive during neonatal systemic infections caused by the parental bacteria. Thus, most if not all phages targeting essential virulence determinants at the bacterial surface will not select fully virulent variants able to participate in the infectious process. Phages could also be advantageously applied to persistent biofilm infections, where antibiotics often fail.
The emergence of resistance may also be minimised by consideration of pharmacokinetic and pharmacodynamic parameters: kinetic models show that active therapies that rely on phage amplification in the host bacterium to generate sufficient progeny through relatively low doses of phage (the ‘living medicine’ approach) are likely to give rise to resistant mutants if only a single phage type is used . Phage combinations may suppress resistance if they are closely matched with respect to the rate at which each phage generates progeny. Resistance could be circumvented by passive phage therapy where doses are sufficiently large to prevent bacterial growth. In this mode, phage is used in a fashion analogous to an antibiotic, although it may not always be possible to deliver sufficient number of phage particles to the site of infection. This important study  shows that combination passive therapy will close the window through which resistant mutants are selected, provided that the concentration of each component phage is greater than the concentration needed to counter bacterial growth. As phages are not required to undergo replication in the target bacterial host in passive therapy, restriction–modification, abortive infection mechanisms and interference between phages in multiply infected cells should not influence outcome, with a consequent reduction in emergence of resistance. Whether or not it is practical for such recommendations to be included in the design of phage therapy regimes remains to be determined. If not, the use of stand-alone phage therapy may be restricted by the rapid emergence of resistant bacteria during the course of infection.
Compromising virulence and antibiotic resistance
Interfering with the expression of the determinants of bacterial pathogenesis or antibiotic resistance is at first sight an attractive approach to antibacterial therapy. Directed suppression of virulence would not kill bacterial targets, but allow host defences to overwhelm the compromised invader. Suppression of antibiotic resistance could facilitate the wider clinical use of drugs that have lost chemotherapeutic utility as a result of the global spread of resistance genes. It has frequently been suggested that such modifications of the bacterial phenotype would reduce the selective pressure for emergence of resistance and preserve the host microbiota. There is, however, little experimental evidence to support these potentially game-changing effects, although a study of co-operation between P. aeruginosa populations containing cells both susceptible and resistant to quorum-sensing (QS) inhibitors suggests that resistance to these modifying compounds may not evolve within co-operating mixed cultures  and flags QS inhibition as a viable and novel approach to antibacterial chemotherapy.
Many bacteria express virulence determinants in temporal fashion, with factors associated with colonisation appearing at an early stage of infection. The synthesis of proteins damaging to the host may become evident only after the pathogen has established a colonising foothold. Control of gene expression during the various stages of the infection cycle is mediated by small secreted molecules of the QS system and by global gene regulators [55,56]. QS is a cell-to-cell signalling process that promotes collective behaviour within a bacterial population, enabling gene expression when the cell density reaches a threshold level [55,57]. Thus, the responses elicited by QS signals contribute directly to pathogenesis by synchronising the elaboration of toxins, proteases and immune-evasive factors; QS may also determine the rate and extent of biofilm development, enabling the pathogen to attain an antibiotic-resistant phenotype [57,58]. There has been substantial effort to develop QS inhibitors. Several distinct families of QS molecules have been investigated in both Gram-positive and Gram-negative bacteria and many competitive inhibitors of these small-molecule extracellular autoinducers developed [57–62]. Some of these compounds have shown excellent in vivo efficacy in experimental small animals, but no preclinical development programmes or clinical trials have been reported to date.
Some excellent reviews have highlighted the potential of antivirulence strategies for antibacterial therapies [63,64], but there have so far been few practical examples, in particular demonstration of in vivo efficacy. However, René Dubos and Oswald Avery at the Rockefeller Institute for Medical Research showed in the pre-antibiotic era that selective removal of a major virulence determinant from a bacterial pathogen can alter the course of infection [65–68]. They used an enzyme preparation from cultures of a peat soil bacterium to selectively remove the polysaccharide capsule from type III pneumococci. Intraperitoneal administration to mice prior to bacterial challenge gave rise to type III-specific protection and intravenous administration to rabbits with type III dermal infections resulted in early termination of the normally fatal infection. Excellent efficacy was also shown in infected non-human primates . We have used similar principles to investigate the capacity of a ‘capsule-stripping’ enzyme to alter the course of lethal E. coli K1 neonatal systemic infection in rats [70,71] and inhalation anthrax in mice [72,73]; in both cases, capsule depolymerase administration prevented death in all infected animals. This therapeutic paradigm, illustrated in Figure 1 for Bacillus anthracis modification, has been reviewed in depth elsewhere . The ‘capsule-stripping’ approach is limited by the selective nature of the enzymes employed, but may be applicable in situations where the infecting bacterium expresses a protective capsule that does not vary in structure between isolates, as with B. anthracis. These enzymes, in common with other therapeutic antibacterial proteins, would be most advantageously employed to counter acute infections that require urgent resolution, as repeated systemic administration may lead to elaboration of neutralising antibodies.
Impact of intravenous administration of the recombinant capsule depolymerase rEnvD on inhalation anthrax in mice.
Reversible modification of the bacterial phenotype could also be exploited to counter antibiotic resistance; agents which perturb the antibiotic resistance machinery may have the capacity to restore the efficacy of antibiotics that have lost utility due to emergence and spread of genes conferring resistance to established antibiotics. This concept has been successfully exploited in the clinic through the longstanding use of β-lactam drugs in combination with β-lactamase inhibitors . Compounds have been recently described that restore susceptibility to β-lactam agents in methicillin-resistant S. aureus (MRSA) through binding to the cell division Z-ring protein FtsZ , opening the way to treat MRSA infections with a combination of β-lactam drug and modifier. We have established that (−)-epicatechin gallate, a polyphenol found in large quantities in green tea, abrogates β-lactam resistance in MRSA by disrupting the cell division machinery at the septal site of cell division, perturbing the physical association between penicillin-binding protein PBP2 and the mecA-encoded resistance protein PBP2a [77–81]. Unfortunately, this in vitro effect cannot be replicated in a model of staphylococcal infection in the zebrafish embryo , and it is to be hoped that other antibiotic-resistance modifiers do not follow this trend.
This short survey of recent activities aimed at providing alternatives for the treatment of bacterial infections in light of seemingly relentless increases in the incidence of acquired antibiotic resistance shows that there is no shortage of promising novel approaches and it is hoped that a few will undergo clinical evaluation in the near future. The list is incomplete; thus vaccination strategies, immune modulation , phage-encoded endolysins  and predatory bacteria  are not covered here, but have been the subject of comprehensive review elsewhere.
New antibiotics and therapeutic modalities are needed to preserve the ability to treat bacterial infections.
Novel therapeutic principles that delay the emergence of drug resistance should be pursued.
Traditional drug discovery platforms may continue to yield new drugs if supplemented with modern molecular insights.
The Author declares that there are no competing interests associated with this manuscript.