Here, we addressed whether a combination of a phage and an antibiotic could prove effective in limiting resistance evolution in experimental P. fluorescens
populations and hence drive the populations extinct. Neither phage nor antibiotic alone was a very effective antimicrobial agent. The phage (SBW25Φ2) could not control bacterial growth: all wild-type bacterial populations under the phage treatment survived () presumably because of rapid evolution of bacterial resistance to the phage (Buckling and Rainey 2002
), although there were fitness costs associated with the resistance when measured in the absence of the phages (). The antibiotic (kanamycin) treatment had a better control on bacterial survival: only 50% of wild-type populations survived (), but the surviving populations showed little fitness cost (). A strong synergism was found between the phage and the antibiotic, with only one of 24 populations under the combined treatment evolving resistance (), presumably because of the requirement for multiple resistance mutations in the same genome to survive both phage and antibiotic attack; the surviving population also had very low fitness ().
Immigration of resistance alleles from source populations, in addition to de novo
mutation, is likely to play a crucial role in resistance evolution. In our experiment with the wild-type strain, bacterial immigration from the control (antibiotic free and phage free) environment increased the chance of resistance evolution under the antibiotic treatment, consistent with previous work (Perron et al. 2007
), but not under the combined antibiotic–phage treatment (). This is not surprising as immigrant populations from the control environment were likely to contain some mutants with antibiotic resistance (the spontaneous mutation rate to kanamycin resistance was approximately 10−7
per cell per generation, and the carrying capacity of each microcosm was >108
cells), but there was little chance of mutations conferring resistance to both phages and kanamycin being present in the same individual. By contrast, immigration of bacteria/phage (from the phage-treatment environment) greatly increased the chance of resistance evolution under the combined antibiotic–phage treatment (); the bacteria in the source environment had experienced selection from the phages and thus should have a fairly high probability to contain individuals with double resistance. Note that we found no evidence that phages evolving in the source microcosms in the absence of antibiotic selection showed greater infectivity than phages evolving in the sink microcosms in the presence of the antibiotic (Data S1; Figure S2
). The results imply that a combined antibiotic–phage therapy might be quite robust against immigration of bacteria from a reservoir where antibiotics and the specific therapeutic phages were absent. However, rational use is as important for phage therapy as for antibiotic therapy: environmental contamination with phages used for therapeutic purposes may diminish the efficacy of combination therapy of antibiotics and phages.
Phages, unlike antibiotics, may evolve novel counter-defense strategies to overcome bacterial resistance, at a rate that researchers developing antibiotics can never hope to replicate (Alisky et al. 1998
; Thiel 2004
; Brockhurst et al. 2007
; Pirnay et al. 2011
). Phages coevolving with host bacteria may or may not be effective at reducing bacterial population sizes, depending on the relative rates of evolution of bacteria and phages (Forde et al. 2008
; Poullain et al. 2008
). Nevertheless, such coevolving phages are likely to drive the bacteria to continuously evolve novel defense strategies and hence to suffer increasing fitness costs (Buckling et al. 2006
; Forde et al. 2008
). This is also confirmed by our work with the wild-type bacteria: fitness of populations surviving the phage treatment or the combined antibiotic–phage treatment decreased over time (). Therefore, where bacteria acquire resistance to, and thus survive, phage (or combined antibiotic–phage) treatment, the frequency of resistant mutants may decline when antimicrobial agents are absent.
High mutation rates in bacteria have been shown to result in very rapid evolution of resistance (Chopra et al. 2003
; Denamur et al. 2005
; Henrichfreise et al. 2007
; Perron et al. 2010
), and this was also the case in our experiment: even the combined antibiotic–phage treatment could not prevent resistance evolution in the mutator SBW25mutS
(). This constructed mutator has a very high mutation rate to phage resistance (Figure S1
); it may have evolved very quickly to outpace the phage and thus rendered the phages unable to persist (Data S1
), consistent with previous work of this system (Pal et al. 2007
). However, fitness of the mutator bacteria significantly decreased during the selection experiment, particularly in the presence of the antibiotic (). It is probable that deleterious mutations accumulated very quickly in the mutator, with the effect of these deleterious mutations enhanced in the presence of antibiotic resistance mutations (i.e., synergistic epistasis). Similarly, an increase in the effect of deleterious mutations has also been observed in the presence of phage resistance mutations (Buckling et al. 2006
). Our results support the view that mutators may not able to persist on long-term timescales in nature (de Visser 2002
Humans often adopt either a chemical or a biological treatment protocol, but not both in combination, to combat ‘harmful’ organisms such as insect pests or weeds (including those introduced to new habitats and becoming invasive species). In many cases, the biocontrol agents (e.g., parasitoids attacking herbivorous insects) can be negatively affected by the chemical agents (e.g., chemical pesticides); hence, there is antagonism rather than synergism between the biocontrol and the chemical control approaches. For bacterial control, however, nature provides us a group of special organisms, bacteriophages, which may work together with antibiotics to limit bacterial resistance evolution. While resistance to antibiotic cocktails may occur less frequently, phage–antibiotic combinations clearly have the following advantages: (i) cross-resistance to phages and antibiotics should be lower than that to multiple antibiotics and (ii) there is a potentially endless supply of therapeutic phages, whereas functional classes of antibiotics are finite.