Introduction

As we struggle to find solutions to tackle the emergence of antibiotic resistance, phage therapy has experienced renewed interest. This treatment approach involves using bacterial viruses called bacteriophage or phage to infect and kill bacterial pathogens, and therefore represents a possible replacement or complementary treatment to antibiotics. Phage treatments are typically prepared by isolating phages that can infect a target pathogen in vitro in mono-culture. This approach favors phage infection: first, while bacteria are in exponential growth phase, their replication machinery can be co-opted by the phage; second, planktonic growth means that at sufficiently high concentrations, phage are more likely to encounter bacteria than if they were stuck together in aggregates, facilitating infection. Perhaps because of these reasons, phage treatments in vivo do not always match the efficacy of the in vitro tests. This highlights a need to understand how bacterial pathogens can escape from phages in their natural environment.
Here we focus on two key features of the natural environments of a patient that may be key to the success of phage therapy. First, a pathogen causing an infection will rarely colonize an empty patch, but will instead live alongside different microbial species, including a patient's resident microbial flora [4,5]. Second, bacteria - pathogens as well as commensals - tend to live in dense, surface-attached cell groups called biofilms. Biofilm-associated bacteria have a higher survival rate compared to their planktonic counterparts [6], particularly when exposed to antibiotics and importantly, also to phage \cite{Eriksen2018}.
Both of these factors - in addition to others that we are not considering here such as the host immune system - can be expected to greatly affect phage therapy. Phages tend to be quite host-specific, killing only a narrow range of bacterial strains. Nevertheless, the presence of resistant strains may alter treatment outcomes by affecting pathogen survival. Indeed, Harcombe & Bull \cite{Harcombe2005} have shown that competition with a co-inhabiting species can reduce the ability of the targeted pathogen to survive phage attack. Their study considered liquid cultures, however. The role of resistant strains in affecting treatment outcomes for a sensitive pathogen has been addressed in the context of antibiotic treatment in biofilms, with the opposite result: the presence of resistant cells can protect sensitive ones from antibiotics \cite{Sorg2016,Frost2018}. This is because a resistant strain that is in close proximity to the pathogen can break down antibiotics and "detoxify" its local environment. Whether a similar protective effect is expected with phage treatment in biofilms is unknown.
We also know that phage population dynamics change radically between liquid bacterial cultures and bacteria growing in or on solid surfaces \cite{Abedon2008}. Much of phage research and clinical work relies on amplifying phages in bacteria growing in semi-solid agar, resulting in the formation of plaques. Decades of experimental and theoretical work has therefore mapped out differences in the ecology and population dynamics of phage in liquid cultures and solid agar. However, these assays were developed to assess bacterial susceptibility to phages rather than to capture features of the natural environment.
In this paper, we explore the fate of a target pathogen, Pseudomonas aeruginosa PAO1, when infected with a phage in the presence of a second non-target strain, Pseudomonas aeruginosa PA14, that is resistant to the phage. We compare the outcome for the pathogen in a well-mixed liquid environment and a structured biofilm (colony) growing on a solid agar surface. In liquid, competition between the two strains can reduce the population size of the pathogen, giving a competitive advantage to the phage and eliminating the pathogen without the emergence of resistance. In contrast, in a solid environment, the presence of the phage-resistant strain - even if competing with the pathogen - allowed the pathogen to grow despite the presence of the phage. This occurred through two mechanisms: first, competition between the strains slowed down the growth of the pathogen, which reduces the replication rate of the phage, and second, the high density of cells - sensitive or resistant - as well as biofilm matrix create a barrier for phage diffusion. Indeed, a key parameter of the ability of phage to infect is their ability to diffuse through biofilm matrix.
We need to understand how bacteria can escape from phages.

Results

Inter-strain competition increases phage efficacy in liquid. We first sought to understand how the presence of the resistant strain P. aeruginosa PA14 (henceforth PA14) would affect the survival of the sensitive target strain P. aeruginosa PAO1 (henceforth PAO1) in the presence of phage in well-mixed liquid cultures. These liquid experiments involved growing bacteria in 96-well plates containing TSB and inoculated with mixtures of bacteria and phage, depending on the condition over a period of 48 hours. In control treatments involving PAO1 growing alone, we observed that phage treatment resulted in a drop in PAO1 population size after 6 hours, after which the population size recovered somewhat (Fig. 1A). Assays testing for phage resistance revealed that after 24 hours of culture 62 out of 63 tested colonies (98.41%) were resistant to the phage, while  after 48 hours, 24 out of 24 (100%) were resistant. As a control, resistant PA14 cells growing alone did not appear to be affected by the phage (Fig. 1B, two-sample t-test at 48 hours, df=2, P=0.4). Next, we co-cultured the two strains in the absence of phage, and found that PAO1 grew worse than when it was alone, presumably due to competition with PA14 (Fig. 1C). Finally, adding the phage to this co-culture eliminated all PAO1 within 6 hours of growth (Fig. 1D). Compared to growing alone then, PAO1 resistance could not emerge when growing with a competitor. This led us to hypothesize that the presence of PA14 prevented PAO1 from increasing its population size, thereby decreasing its potential to evolve resistance to the phage and survive the attack.
To test for the effect of population size on resistance evolution, we conducted two experiments. First, we grew PAO1 in the presence of phage with different starting population sizes. In agreement with our hypothesis, we found that resistance to the phage could evolve when the initial population size was large enough (greater than 104 CFU/ml, Fig. 1E).  Second, we kept the initial population size of PAO1 constant and varied the starting population size of its competitor PA14 in the presence of phage. Again, as predicted, we observed that phage resistance could emerge when there were fewer competitors, but as the number of competitors in the starting population grew beyond 106 CFU/ml, PAO1 cells were all killed by the phage at the end of 21 hours of co-culture (Fig. 1F). In all cases, PAO1 survival depended on becoming resistant to the phage. In sum, we find that in liquid culture, competition with resident strains can prevent a target pathogen from surviving phage attack, which is consistent with previous research \cite{Harcombe2005}.