Antibiotic treatment can perturb bacterial metabolism and increase production of reactive metabolic byproducts (RMB), such as reactive oxygen species (ROS). These reactive metabolites are well known for their ability to damage macromolecules, such as DNA and proteins, and they can be mutagenic (Markkanen, 2017; Seixas et al., 2022; Wong et al., 2022). Although even sub-lethal doses of antibiotics are sufficient to increase mutagenesis that leads to increased minimum inhibitory concentrations, scavenging these reactive metabolites could prevent this effect (Kohanski et al., 2007; Kohanski et al., 2010).

Pribis and colleagues found that a small subpopulation of ciprofloxacin-treated Escherichia coli generate ROS which leads to mutagenic DNA repair and subsequent resistance development in this subpopulation (Shee et al., 2011; Pribis et al., 2019). To combat this problem, they treated E. coli with the antioxidant edaravone to reduce ROS levels, which ultimately reduced the number of resistance mutants that arose and preserved ciprofloxacin’s killing efficacy (Pribis et al., 2019). It must be acknowledged, however, that hindering RMB generation could reduce antibiotic efficacy. In contrast to the results of Pribis and colleagues, another group found that treating E. coli with RMB scavengers and ciprofloxacin, kanamycin, or mecillinam reduced the antibiotics’ killing effects (Wong et al., 2022). More research into the mechanisms of RMB involvement in antibiotic killing is necessary to determine whether prescribing antibiotics along with compounds to limit these stressors could be a valuable resistance-resistant strategy.

In contrast to lessening the stressors, bacterial evolvability can be reduced by inhibiting the evolution-driving pathways that respond to these stressors ( Figure 1A ) (Galhardo et al., 2007; Ragheb et al., 2019; Merrikh and Kohli, 2020). Given that many antibiotics can either directly (e.g., fluoroquinolones) or indirectly (e.g., antibiotics that stimulate RMB production) damage DNA integrity, one major pathway that allows bacteria to respond to antibiotic treatment and guard against deleterious DNA breaks is the SOS response (McKenzie et al., 2000; Maslowska et al., 2019; Podlesek and Bertok, 2020). This DNA repair process can involve the activation of mutagenic DNA polymerases that lack proofreading activity (Baharoglu and Mazel, 2014), potentially leading to resistance-conferring mutations (Podlesek and Bertok, 2020).

It has been hypothesized that SOS response inhibitors could limit adaptive mutagenesis and curb resistance development (McKenzie et al., 2000; Maslowska et al., 2019; Podlesek and Bertok, 2020). Cirz and colleagues found that SOS-deficient E. coli were unable to evolve resistance against ciprofloxacin or rifampicin (Cirz et al., 2005). Similarly, nanobodies or phages that prevent the cleavage of LexA, a repressor that undergoes proteolysis in response to DNA damage to induce the SOS response, block DNA repair and antibiotic resistance development (Lu and Collins, 2009; Maso et al., 2022). Other members of the SOS pathway have also been implicated as potential targets for inhibition along with antibiotic treatment in order to limit resistance mutations (Yakimov et al., 2021). Beyond targeting SOS response proteins, Ragheb and colleagues found that inhibiting Mfd, a DNA translocase found to increase mutagenesis and antibiotic-resistance development, could decrease the likelihood of resistance-conferring mutations (Ragheb et al., 2019). Although the regulation of enzymes responsible for mutagenesis varies across organisms, combining antibiotic therapy with the inhibition of conserved evolutionary drivers, like error-prone DNA polymerases, has immense potential to aid in the fight against antibiotic resistance development.

Antibiotic treatment can select for pre-existing resistance mutations, but combining one antibiotic with another, or with an adjuvant that has a different target, can limit such selection ( Figure 1B ) (Wright, 2016). It is much less likely for a bacterium to simultaneously resist two treatments with different targets than it is to develop resistance toward monotherapy (Maron et al., 2022). Some examples of antibiotic combinations that have been shown to inhibit resistance development include ciprofloxacin plus amikacin against Pseudomonas aeruginosa, streptomycin plus para-aminosalicylic acid against Mycobacterium tuberculosis, and daptomycin plus rifampicin against methicillin-resistant Staphylococcus aureus (Bamberger et al., 1986; Fasching et al., 1987; Mouton, 1999; Garrigós et al., 2010; Mitchison and Davies, 2012; Worthington and Melander, 2013). Unfortunately, while many studies on antibiotic combinations focus on killing efficacy and synergy between combinations, few directly examine how these combinations affect resistance development (Deresinski, 2009; Samonis et al., 2012; Roemhild et al., 2022; Wang et al., 2022). Though counterintuitive, it has been shown that the combination of two antibiotics that suppress each other’s activity can actually hinder the development of resistance against either drug (Chait et al., 2007; Singh and Yeh, 2017; Liu et al., 2020). For example, the combination of daptomycin and rifampicin is less effective at killing S. aureus than treatment with either drug alone; however, this suppressive antibiotic combination prevents the emergence of rifampicin resistance because cells that do develop rifampicin resistance can resume growth and become more sensitive to daptomycin than the wild-type cells (Stein et al., 2016; Liu et al., 2020). These unexpected fitness dynamics indicate that the success of suppressive antibiotic combinations will hinge on the prevention of resistance evolution (Singh and Yeh, 2017; Liu et al., 2020).

Drawing inspiration from the potential success stemming from administering a combination of antibiotics, single compounds featuring two distinct antibacterial mechanisms that act synergistically have been developed. Several compounds have been developed in which two antibiotics (most commonly one fluoroquinolone and one DNA or protein synthesis inhibitor) are connected by either a cleavable or non-cleavable linker, and compounds based on these designs are currently undergoing clinical trials (Bremner et al., 2007; Pokrovskaya and Baasov, 2010; Ma and Lynch, 2016; Jubeh et al., 2020; Theuretzbacher, 2020). Additionally, Pentobra, which consists of the aminoglycoside tobramycin functionalized with an antimicrobial peptide, has been shown to better permeate bacterial membranes than tobramycin alone. This allows Pentobra to kill bacteria that normally resist aminoglycosides through poor uptake (Schmidt et al., 2015). Like studies of traditional antibiotic combinations, studies of these dual-mechanism compounds usually focus on killing efficacy rather than directly investigating their ability to prevent resistance development. Additionally, until recently, such two-pronged compounds often worked well against Gram-positive bacteria but failed against Gram-negative species (Gupta and Datta, 2019; Theuretzbacher, 2020). Excitingly, Irresistin-16 features one moiety that targets folate metabolism and another that targets membrane integrity; together, these moieties act to decrease survival of both Gram-negative and Gram-positive pathogens all while leaving no detectable resistant mutants (Martin et al., 2020).

Some bacteria have enzymes that confer resistance by deactivating certain antibiotics (Blair et al., 2015; Wright, 2016; Markley and Wencewicz, 2018). While the antibiotic alone can kill members of the population lacking these enzymes, combining the antibiotic with inhibitors of these enzymes allows it to kill resistant bacteria as well (Wright, 2016; Liu et al., 2018; Markley and Wencewicz, 2018). The most successful example of this strategy is the combination of β-lactam antibiotics with β-lactamase inhibitors (Zhanel et al., 2013; Coleman et al., 2014; Brown and Wright, 2016; Wright, 2016). However, more research is needed before inhibitors of other antibiotic-inactivating enzymes can be used in the clinic. Park and colleagues found that anhydrotetracycline (aTc) binds to the active site of tetracycline destructases, thereby sparing some tetracycline from these enzymes. Treating E. coli expressing the tetracycline destructase Tet(56) with aTc significantly increased susceptibility to tetracycline (Park et al., 2017). Further research into the structures of these enzymes and the mechanisms by which they destroy or deactivate antibiotics may lead to the development of additional inhibitors that can preserve the efficacy of our antibiotic arsenal.

In some cases, instead of deactivating or destroying an antibiotic, bacteria simply use efflux pumps to reduce the antibiotic’s intracellular concentration (Amaral et al., 2014; Li et al., 2015). Some efflux pump-encoding genes can be acquired through horizontal gene transfer, but several bacterial species naturally express a repertoire of efflux pumps that provide resistance to multiple antibiotics. Therefore, inhibiting efflux pumps to prevent the bacteria from escaping antibiotic attack is a promising strategy. Despite many compounds that are able to inhibit efflux pumps, this strategy has not been applied clinically because many of these inhibitors are toxic to humans (Ferrer-Espada et al., 2019). To combat this toxicity, Ferrer-Espada and colleagues found that antimicrobial peptides that increase the membrane permeability of P. aeruginosa increased the effectiveness of efflux pump inhibitors, which in turn increased the bacteria’s susceptibility to all antibiotics tested (Ferrer-Espada et al., 2019). This increased effectiveness should mean that a lower, and hopefully more clinically safe, dose of efflux pump inhibitor could be given to great effect. Additionally, it was recently reported that increasing the expression of multidrug efflux pump genes acrAB in E. coli reduces the expression of mutS, which is involved in DNA mismatch repair, and promotes spontaneous mutagenesis (El Meouche and Dunlop, 2018). These findings suggest that strategies to reduce acrAB expression can potentially limit mutagenesis and resistance development.

The conjugation of an antibiotic to a monoclonal antibody limits the number of bacteria that survive treatment compared to treatment with the antibiotic alone (Lehar et al., 2015; Kajihara et al., 2021; Cavaco et al., 2022). If fewer bacteria survive the treatment, then there are fewer bacteria left to develop and spread resistance (Bakkeren et al., 2019). This approach’s increased killing of the pathogen plus its ability to spare the microbiota from antibiotic exposure make it a promising resistance-resistant strategy. Often in an infection, pathogens are able to persist inside phagocytic or even non-phagocytic host cells while avoiding being digested (Helaine et al., 2010; Mariathasan and Tan, 2017; Peyrusson et al., 2020). In combined antibody-antibiotic therapy, an antibody carrying the antibiotic binds to a pathogen and opsonizes it for phagocytosis by immune cells. Once in the acidic internal environment of the phagocyte, the antibiotic will be cleaved off of the antibody and allowed to attack the pathogen, making it less able to persist inside the phagocyte (Dubowchik et al., 2002; Lehar et al., 2015; Cavaco et al., 2022). As mentioned before, a major benefit of this approach is that by targeting pathogens of interest with highly specific antibodies, the host microbiota should be spared from antibiotic exposure, reducing the likelihood of selecting for resistant mutants among commensal bacteria (Zurawski and McLendon, 2020). One pitfall of this approach is that designing and producing these conjugates remains very difficult because of the specific characteristics required for the antibody, antibiotic, and linker to each have maximum efficacy (Beck et al., 2017; Wei et al., 2018). Additionally, just as monoclonal antibodies designed to target pathogens for phagocytosis rely on at least partial immune system functionality (ter Meulen, 2007), antibody-drug conjugates may not be effective for immunocompromised patients.

The combination of phage therapy with antibiotics has shown promise in reducing resistance development (Torres-Barceló et al., 2022). Bacteriophages specifically infect certain species or strains of bacteria and ultimately kill them through cell lysis (Lin et al., 2017). The range of bacteria that a given phage can infect is limited by the receptors available on the bacterial cell surfaces (de Jonge et al., 2019). With phage mining/hunting efforts and the development of modern genetic engineering techniques, the potential repertoire of phages that could be used against pathogens has greatly expanded (Chen et al., 2019; Gauthier et al., 2022). As with the other combination therapies mentioned, phage therapy and antibiotic therapy present two distinct selective pressures to the pathogen. Among the defenses that bacteria employ to resist being killed by phages (Egido et al., 2022), the most relevant are mutations that either remove or change the structure of the cell-surface receptor a given phage must bind in order to infect the bacterium. The result of this selection is that the population shifts toward decreased expression of the normal receptor on their surfaces. Then, if this receptor is important for the bacteria’s defense against antibiotics, their antibiotic susceptibility will be increased. In this way, phage therapy that targets surface proteins important for antibiotic resistance can shift the population toward antibiotic susceptibility and vice versa. For example, using phages that dock onto efflux pump components can select for mutants with defects such as non-functional multidrug efflux pumps, preventing them from expelling their usual antibiotic substrates (Chan et al., 2016; Gurney et al., 2020). Additionally, combined antibiotic-phage therapy can select for bacteria that only survive treatment by incurring mutations that reduce their fitness in the absence of treatment (Zhang and Buckling, 2012). Antibiotic-phage therapy has even shown promise in the clinic—an example is the successful treatment of a patient with an aortic arch prosthesis infected by P. aeruginosa (Chan et al., 2018). The phages targeted the multidrug efflux pumps of P. aeruginosa, making the pathogen more susceptible to treatment with ciprofloxacin and ceftazidime (Chan et al., 2018; Hatfull et al., 2022). These examples highlight the fascinating potential for phages and antibiotics together to prevent antibiotic resistance and effectively cure infections that are difficult to treat with either approach alone.

However, the use of bacteriophages has some limitations. First, unlike broad-spectrum antibiotics, many phages can only infect a few species or even a few strains of bacteria, making it difficult to effectively utilize phage therapy before identifying the infection’s causative agent (Loc-Carrillo and Abedon, 2011; Göller et al., 2021). This problem can be mitigated by prescribing a cocktail of several phages each with specificities for different bacteria (McVay et al., 2007; Kutateladze and Adamia, 2010; Loc-Carrillo and Abedon, 2011). Additionally, lysogenic phages integrate their DNA into the host’s genome and may transfer virulence or antibiotic-resistance genes to susceptible pathogens and commensal bacteria. Therefore, it is recommended that researchers focus on lytic phages that kill bacteria without integrating their DNA in order to prevent horizontal transfer of these harmful genes (Skurnik et al., 2007; Colomer-Lluch et al., 2011; Loc-Carrillo and Abedon, 2011). Finally, there are concerns that phages could cause unwanted immune responses. One study showed that phage therapy can increase serum levels of inflammatory markers (Oechslin et al., 2017; Liu et al., 2021). Additionally, the patient’s immune system can make antibodies against phages (Sunagar et al., 2010; Jun et al., 2014; Liu et al., 2021), but one study showed that production of such antibodies likely begins after the phages have already exerted their killing effects (Eskenazi et al., 2022). Phage therapy research has gained traction in the past decade after being largely overlooked in the 20^th century (de Jonge et al., 2019). Progress on this front can potentially illuminate strategies to improve the efficacy and overcome current limitations of phage therapy.