Author response: Competition between lysogenic and sensitive bacteria is determined by the fitness costs of the different emerging phage-resistance strategies

Abstract

Article Figures and data Abstract Editor's evaluation Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Many bacterial genomes carry prophages whose induction can eliminate competitors. In response, bacteria may become resistant by modifying surface receptors, by lysogenization, or by other poorly known processes. All these mechanisms affect bacterial fitness and population dynamics. To understand the evolution of phage resistance, we co-cultivated a phage-sensitive strain (BJ1) and a polylysogenic Klebsiella pneumoniae strain (ST14) under different phage pressures. The population yield remained stable after 30 days. Surprisingly, the initially sensitive strain remained in all populations and its frequency was highest when phage pressure was strongest. Resistance to phages in these populations emerged initially through mutations preventing capsule biosynthesis. Protection through lysogeny was rarely observed because the lysogens have increased death rates due to prophage induction. Unexpectedly, the adaptation process changed at longer time scales: the frequency of capsulated cells in BJ1 populations increased again because the production of the capsule was fine-tuned, reducing the ability of phage to absorb. Contrary to the lysogens, these capsulated-resistant clones are pan-resistant to a large panel of phages. Intriguingly, some clones exhibited transient non-genetic resistance to phages, suggesting an important role of phenotypic resistance in coevolving populations. Our results show that interactions between lysogens and sensitive strains are shaped by antagonistic co-evolution between phages and bacteria. These processes may involve key physiological traits, such as the capsule, and depend on the time frame of the evolutionary process. At short time scales, simple and costly inactivating mutations are adaptive, but in the long term, changes drawing more favorable trade-offs between resistance to phages and cell fitness become prevalent. Editor's evaluation The overarching question of the manuscript is important and the findings inform the patterns and mechanisms of phage-mediated bacterial competition, with implications for microbial evolution and antimicrobial resistance. The strength of the evidence in the manuscript is compelling, with a huge amount of data and very interesting observations. The conclusions are well supported by the data. This manuscript provides a new co-evolutionary perspective on competition between lysogenic and phage-susceptible bacteria, that will inform new studies and sharpen our understanding of phage-mediated bacterial co-evolution. https://doi.org/10.7554/eLife.83479.sa0 Decision letter Reviews on Sciety eLife's review process Introduction Parasites shape the life history and fitness of their hosts. They also impact community structure via predation and competition, and thereby affect numerous ecological and evolutionary processes (Koskella and Brockhurst, 2014; Pedersen and Fenton, 2007; Lefèvre et al., 2009). Bacteriophages (phages) are very abundant predators of bacteria (Brüssow and Hendrix, 2002; Suttle, 2007). Temperate phages either follow a lytic cycle in which they replicate within bacterial cells and release infectious virions, or a lysogenic cycle in which they integrate the bacterial genome and replicate with it. Nearly half of the sequenced bacterial genomes are lysogens (Touchon et al., 2016). The dual lifestyle of temperate phages is costly, but can also provide the host with multiple advantages. During lysogeny, prophages may increase biofilm formation (Gödeke et al., 2011), phosphate acquisition (Sullivan et al., 2005), or express virulence factors (Busby et al., 2013; Fasano et al., 1991; Varani et al., 2013). Inactivated prophages leave genes in the genome that are co-opted by the host and result in functional innovation, e.g., as bacteriocins used in bacterial warfare (Winstanley et al., 2009; Bobay et al., 2014; Nakayama et al., 2000). Prophages also protect bacteria from closely related phages, a process called superinfection resistance (Bondy-Denomy et al., 2016). Furthermore, when the lytic cycle is initiated in a small subpopulation, it may facilitate colonization by directly mediating competition within communities (Li et al., 2017; Joo et al., 2006; Harrison and Brockhurst, 2017; Wendling et al., 2021), because the released virions will infect and lyse closely related but not identical strains. This can promote the acquisition of adaptive traits from bacterial competitors (Wendling et al., 2021; Haaber et al., 2016). Hence, it is suggested that prophage induction affects bacterial population dynamics, community structure, and evolution (Bondy-Denomy and Davidson, 2014; Bossi et al., 2003; De Paepe et al., 2016; Gama et al., 2013; Nanda et al., 2015). How parasite pressure may alter co-evolving bacterial populations has been seldom addressed, and most of these studies focused on virulent phages (Brockhurst et al., 2007; Fazzino et al., 2020; Lourenço et al., 2020; Weinbauer, 2004; Weinbauer and Rassoulzadegan, 2004). A few other studies have tested the impact of coevolution between lysogens and non-lysogens and the advantages the former provides in vivo by mediating bacterial interactions (Joo et al., 2006; De Paepe et al., 2016; Burns et al., 2015; Davies et al., 2016; Frazão et al., 2022). However, the relevance of poly-lysogeny for population dynamics during hundreds of generations remains unknown. Further, little is known about the interactions between the different resistance mechanisms, how they affect the cost of resistance, and how they may provide an opportunity for the emergence of novel mechanisms. To test this, we co-evolved two natural isolates of K. pneumoniae, an ubiquitous species, of which at least 75% of the species’ genomes are polylysogenic (de Sousa et al., 2020): (i) the hypervirulent BJ1 strain without inducible or cryptic prophages, that was isolated from a liver abscess (ST380) and (ii) a polylysogenic multidrug-resistant K. pneumoniae strain (ST14) isolated from a urinary tract nosocomial infection. ST14 produces multiple infectious virions for which the BJ1 is known to be sensitive (Supplementary file 1; de Sousa et al., 2020). Based on previous studies, cell defense mechanisms, such as restriction-modification systems, are not expected to impact population dynamics (de Sousa et al., 2020). We hypothesized that resistance would rather emerge by lysogenization under strong phage pressure and by inactivation of the extracellular capsule, the main surface receptor of phage at intermediate and low phage pressure (de Sousa et al., 2020). To test if, and how, prophage induction affects the competition outcome between the two strains, we followed their population dynamics through time. We then tested for the emergence of phage resistance in the susceptible strain. This revealed the diversity and interactions of the emerging mechanisms of phage resistance. It also provided unique insight into how these different mechanisms coexist within a population and evolve through time in response to infection pressure. Results Temperate phages provide fitness advantage during competition We first aimed at understanding if the prophages of strain ST14 provide a fitness advantage during competition with phage-susceptible strain BJ1. To limit confounding factors such as competition for resources, we grew the cells in a rich environment. To modulate the amount of phage produced, and the ability of the latter to infect, we defined three conditions: (i) LB, (ii) LB supplemented with 0.2% citrate to inhibit phage infection due to calcium chelation (de Sousa et al., 2020; Shafia and Thompson, 1964), and (iii) LB with mytomicin C (MMC, 0.1 µg/mL) to increase the phage titers in the environment. MMC was added at a concentration that did not significantly affect the growth of BJ1 (Figure 1—figure supplement 1A, B), and despite the consumption of citrate by Klebsiella, after 24 hr there is still a large amount of citrate remaining, that is sufficient to inhibit infection (Figure 1—figure supplement 1). We also quantified the amount of PFU/mL produced by strain ST14 in the different growth conditions. As expected, phage production in ST14 was significantly higher in MMC compared to the two other treatments. Interestingly, ST14 grown in citrate resulted in a marginal increase in phage production relative to the control (LB) (Figure 1—figure supplement 1C, D). To test whether phages could contribute to the competitive fitness of their host, we co-inoculated both strains (BJ1 and ST14) at an initial ratio of 1:1 for 24 hr. First, we tested if mixing the two strains affected the total growth or the population yield. We observed no increased cell death due to the competition in the different growth conditions (Figure 1—figure supplement 2A). Then, we calculated the competitive index of the strains and observed that there was a large fitness advantage for strain ST14 in all three conditions (Figure 1A). This was most likely because ST14 has both a higher growth rate and population yield than BJ1, except in the presence of MMC (Figure 1—figure supplement 1A, B). Most importantly, we also observed differences in the competitive index depending on the amount of phage released, or its ability to infect the BJ1 (Kruskal-Wallis, dF = 2, p=0.006). This effect could be due to the different population yields of each strain in each environment (Figure 1—figure supplement 2B). To take into account only the fitness effects due to phages, we quantified the strain-interaction effects using Ci(j), which measures the effect of mixing two strains i and j on the viable population size of strain i, relative to pure culture controls. This measure accounts for the absolute performance of each competitor in mixed groups (see Methods). Negative Ci(j) values indicate that strain i have a lower population yield during growth in the presence of j than in pure culture, and positive values indicate the opposite. For the phage producer, strain ST14, the competition had no positive or negative effect on total population yield, most likely because the increased release of viruses resulting in ST14 death is outweighed by an exacerbated death of phage-sensitive BJ1 (Figure 1B). In the presence of citrate, a condition in which phage cannot infect, the growth of strain BJ1 was not significantly inhibited. However, in the absence of citrate, when phages can infect, Ci(j) was significantly lower than zero, indicating a negative effect on the growth of strain BJ1. This effect was dependent on the amount of phage released into the environment, as an increased production of phages by ST14 due to MMC leads to an even lower Ci(j) for BJ1 (Figure 1B, Kruskal-Wallis, dF = 2, p=0.007). Figure 1 with 2 supplements see all Download asset Open asset Fitness of strains during competition. (A) The competitive index is calculated as the final frequency of each strain divided by the initial frequency in the mixed cocultures. * P<0.05, Wilcoxon rank sum test adjusted by Benjamini-Hochberg correction. (B) The effect of mixing two strains during growth in coculture is given as Ci(j), expressed in log10, with i representing either strain BJ1 or strain ST14. Positive values represent increased cell numbers during coculture than those expected from the pure cultures. p-values correspond to a one-sample t-test for a difference of 0. *p<0.05, ***p<0.001. Each dot shape represents an independent biological replicate, N=5. Error bars indicate the standard deviation. We also tested whether ST14 could sense the presence of competition, for instance by quorum sensing mechanisms, and induce prophages and the production of viral particles (León-Félix and Villicaña, 2021). Our results showed that the growth of ST14 with spent supernatant of BJ1 did not result in increased viral release (Figure 1—figure supplement 1C, D, E). Taken together, our results showed that prophages can increase the fitness of their host in co-culture by disfavouring the non-lysogens. Resistance to temperate phages emerges rapidly during coevolution To assess whether ST14 prophages could provide a long-term fitness advantage to their hosts, and allow them to outcompete non-lysogens, we set up an experiment in which we allowed three independent mixed populations composed of phage-producing ST14 and phage-susceptible BJ1 strains to coevolve during 30 days, in the three previously defined environments (LB, LB supplemented with 0.2% citrate, and LB supplemented with MMC). To follow the evolution of each strain, we plated the populations every day on selective media and counted CFU. As expected, no significant changes in the group yield were observed (Figure 2—figure supplement 1). This is mostly explained because the dominant strain, the phage producer, does not change its population yield (Figure 2A). In contrast, the frequency of BJ1 decreased rapidly during the first four days, suggesting a large initial fitness disadvantage of this strain. This is observed in all three conditions, but it is accelerated in conditions in which phage release is exacerbated (with MMC) and bacterial infection is not restricted (without citrate). Decrease in BJ1 populations was also correlated with large increases in phage production during the first ten days (Figure 2—figure supplement 2). Interestingly, shortly after the beginning of the experiment, both phage production and evolved BJ1 populations seem to stabilize, except for one BJ1 population evolving in MMC (which increases significantly in frequency). This suggested the emergence of phage resistance. However, after day 15, BJ1 populations evolving in MMC remain stable whereas the ones in the other conditions showed a second decrease in CFUs. This second decrease continued until day 22, beyond which BJ1 populations once again stabilized at ca 103 CFU/mL. Taken together, in 30 days of coevolution, ST14 did not completely displace BJ1 from the populations, even in conditions where the production of the phage by strain ST14 is exacerbated. Indeed, across all conditions, and throughout all the experimental evolution, infectious virions were actively produced and released (Figure 2—figure supplement 2), representing a constant, active selective pressure. This suggests that prophage-mediated competition can be counter-balanced by the evolution of resistance mechanisms in the competitor strain. Figure 2 with 2 supplements see all Download asset Open asset Population yield and proportion of capsulated clones of the two strains during the coevolution experiment. (A) Total CFU per mL of each strain was estimated every day on selective media. Each line represents an independent coevolving population. (B) Emergence of non-capsulated mutants in each strain. The proportion of capsulated clones in the population is depicted. The insert shows the area under the curve (AUC) during the first nine days of evolution, as calculated by the function trapz from the R package pracma. *p<0.05,**p<0.01,***p<0.001 for ANOVA with Tukey post hoc corrections. Phage pressure drives receptor inactivation as a mechanism of resistance It has been largely documented that the extracellular capsule is a main phage receptor in Klebsiella (de Sousa et al., 2020; Hesse et al., 2020; Tan et al., 2020; Venturini et al., 2020). Throughout the daily plating of coevolving populations, we observed the rapid emergence of non-capsulated clones in all independent populations across the three treatments (Figure 2B). We tested if non-capsulated clones could be under stronger selection for receptor inactivation when phage pressure is higher (higher density of phages). We observed that the emergence of non-capsulated clones in the BJ1 background is exacerbated in the environment in which phage pressure is greater (insert, Figure 2B), and is diminished when phages cannot infect. Hence, phage pressure accelerated capsule inactivation. Interestingly, this is also the case for ST14, the phage producer, which we had previously shown to be mildly susceptible to its own phages (de Sousa et al., 2020). Overall, within the first ten days, at least 50% of the population was composed of non-capsulated mutants. Towards the end of the experiment, increasing frequencies of capsulated clones were observed across many populations, suggestive of the emergence of other resistance mechanisms. We sequenced the gene wcaJ to identify the genetic causes of receptor inactivation because it encodes the first glycosyltransferase of the capsule biosynthesis pathway and is known to be largely responsible for capsule inactivation (Chiarelli et al., 2020; Haudiquet et al., 2021). This revealed that all but two non-capsulated clones (BJ1 (N=36); ST14 (N=18)) had mutations in wcaJ, most of which resulted in a loss-of-function (Supplementary file 2). In summary, phage pressure led to rapid resistance emergence by surface modifications. The emergence of new lysogens is rare and potentially unstable Some resistant clones of strain BJ1 are capsulated, which led us to hypothesize that they evolved other resistance mechanisms. To test this, we analyzed at different time points the resistance mechanisms of the capsulated clones in the population. We expected to find BJ1 lysogens, since super-infection exclusion due to the lysogenization of capsulated bacteria could prevent further infection by the same phages. Further, our previous work had already shown that, when infected with phage lysate at high titers, at least two of the four intact phages from strain ST14 could lysogenize BJ1 (Supplementary file 1; de Sousa et al., 2020). To quantify the proportion of lysogenized BJ1 cells, relative to other resistance mechanisms, we isolated over 1200 capsulated clones at different time points (Figure 3—figure supplement 1). We identified the clones that were resistant to purified phage lysates of strain ST14 and that produced phages when exposed to MMC in our culture conditions. More precisely, we analyzed the differences in the area under the growth curve of each clone, both when they were grown in LB (control), when phage lysate was added (to distinguish between resistant or susceptible), and when MMC was added (to induce prophages and identify newly lysogenic clones). Together with the resistant non-capsulated clones (Figure 2B), this provides a detailed overview of the different mechanisms of resistance, their proportion, and their temporal dynamics throughout the experiment (Figure 3). Figure 3 with 4 supplements see all Download asset Open asset Evolution of resistance mechanisms in strain BJ1. (A) Ratio of clones from each coevolving population that are susceptible (green), non-capsulated (light pink), capsulated lysogens (beige), or capsulated but resistant by other undefined mechanisms (dark pink). N.B. Dashed line indicates when the x-axis, no longer follows a linear scale. Dark arrows indicate the time points from which the lysogens tested in panel C were retrieved. (B) Growth of newly lysogenized clones reveals significant death during the exponential phase (in the absence of induction), as measured by the optical density. Black line corresponds to the control, BJ1 ancestor. (C) PFU/mL produced without induction by five selected new lysogens derived from BJ1 and isolated at day one for A4 and A6 (Population #2, MMC), at day four for B3 (Population #3, MMC), and day nine for H8 and H9 (Population #3, MMC). Dashed line indicates the limit of detection of the essay. Each black dot represents an independent biological replicate (independent strain lysate) and large red dots represent the mean. Error bars correspond to the standard deviation. Two-sided t-test ‘a’, p<0.001 compared to ancestor BJ1 (negative control, C-) and ‘b’, p<0.05 compared to ST14 (phage producer, positive control, C+). (D) Simulated temporal dynamics of the proportion of lysogens in the populations, as calculated by eVIVALDI. Each circle corresponds to the central tendency of replicate simulations, with the different colors indicating a given probability of spontaneous prophage induction (shown in the legend, values approximated to the nearest major integer). The error bars correspond to the standard deviation across the replicate simulations. In the represented simulations, the probability of acquisition of a phage resistance mutation (capsule loss) is 0.001, and the fitness cost of this mutation is 10% of the bacterial growth rate, as calculated in Buffet et al., 2021. We first observed that the proportion of susceptible clones quickly decreased, especially when phage pressure was high (under MMC, Figure 3). The majority of tested clones were resistant by day 2, 6, and 8 in populations evolving in MMC, LB, and citrate, respectively. As expected, lysogens emerged in all populations, but remained at low frequency and their numbers quickly dwindled after their emergence (Figure 3). We verified that the 94 identified lysogens, out of the 1209 screened clones, were bona fide lysogens. This could be confirmed by their production of phages infecting naïve BJ1 cells, both when induced by MMC (92 out of 94), and in the absence of induction (87 clones out of 94) (Figure 3—figure supplement 2). Interestingly, we observed that when new lysogens are grown in LB, in the absence of induction, there is a detectable amount of cell death, and growth delay at the end of the exponential phase in at least in 29 out of 94 tested clones (Figure 3B and Figure 3—figure supplement 4). This could correspond to a high frequency of spontaneous induction in the newly lysogenized bacteria. Indeed, we observed a large amount of phage release, as evidenced by large inhibition halos on an overlay of ancestral BJ1. We then selected five different lysogens that consistently showed large inhibition halos (Figure 3—figure supplement 2) and exacerbated death (Figure 3B and Figure 3—figure supplement 3) descending from three independently evolving BJ1 populations. We quantified the amount of phage released, in the absence of induction, on a lawn of ancestral BJ1. New lysogens produced between 100 and 1000 more PFU/mL than the ancestral phage producer (ST14) (Figure 3C). This suggests that protection by lysogeny results in significant fitness costs because prophage induction is frequent (Figure 3—figure supplements 2 and 3A). To study the impact of prophage acquisition on the long-term stability of lysogens in a population, we used eVIVALDI, an individual-based model for microbial interactions and evolution (de Sousa and Rocha, 2019). We used these simulations to explore different rates of induction of prophages, in the presence or absence of abiotic agents. We designed a scenario where a population of initially sensitive bacterial cells is exposed to an inoculum of temperate phages, and we followed the populations for a period of 150 iterations (e.g. approximately 150 generations). Simulated bacteria could either be infected by phage (thus either dying upon a lytic infection or becoming lysogens if the phage integrates the bacterial genome) or become resistant to phage by mutation (i.e. capsule inactivation, which decreases their growth rate). We then measured, over time, both the total number of cells and the proportions of lysogens. We observed two main patterns. When prophages have low spontaneous induction rates (1 to 5%), they generate stable, non-costly lysogens. Consequently, phages spread slowly in the bacterial population, which gives time for the phage-resistant mutants to emerge and increase to high frequencies. However, because these mutations are costly, lysogens slowly displace these mutants. This results in a sigmoidal-like temporal frequency of lysogens, where at the end of the simulations most of the resistant population is composed of lysogens (Figure 3D and lower left part of panel Figure 3—figure supplement 4). These dynamics are in contrast with the bell-shaped dynamics observed for high or intermediate rates of spontaneous prophage induction (i.e. ≥11%), where lysogens quickly invade the population but are absent at the end. Such high rates correspond to unstable lysogens that quickly die due to spontaneous induction of their prophages. These conditions facilitate the propagation of phage throughout the population (due to fast phage amplification), and result in the rapid emergence of new bacterial lysogens (t=10 in Figure 3D and Figure 3—figure supplement 4A). Becoming a lysogen can be advantageous if these cells are protected from new phage infections. However, if there are high rates of prophage induction, lysogeny may become less adaptive than other mechanisms of protection, e.g., receptor loss (Figure 3—figure supplement 4B). As a result, when spontaneous induction rates are high and other, potentially fitter, resistant clones emerge, few or no lysogens are expected to survive as they will be outcompeted (top-right areas for the heatmaps in Figure 3—figure supplement 4A). This is consistent with our experimental results, where BJ1 clones quickly become lysogens with high induction rates which leads to their removal from the population by the end of the experimental evolution. In our simulations, the absence of capsule-inactivating mutations (resistance probability = 0, rightmost column of the heatmaps in Figure 3—figure supplement 4A), implies that populations either become extinct (if induction rates are too high) or are completely composed of lysogens. In contrast, our in vitro experiments revealed some resistant clones that were still capsulated and non-lysogens, indicating alternative mechanisms of resistance to phage. These novel clones were more frequent in populations under high phage induction pressure (MMC), when the cost of lysogeny is high, and less frequent undergrowth in LB with or without citrate (Kruskal-Wallis, dF = 2, P=0.03) (Figure 3). Taken together, our results show that most clones became resistant to capsule inactivation, a few by lysogenization, and others by novel mechanisms. These experimental results fit our simulations in suggesting the existence of competition between multiple phage resistance mechanisms. The cost of the resistance mechanism varies with phage pressure and across time scales To test the competitive fitness of the evolved clones having different phage-resistance mechanisms, we compared the area under growth curves of all BJ1 evolved clones isolated in the presence or in the absence of phage (Figure 4). In the first ten days of the evolution experiment, non-capsulated clones had a higher AUC than all other resistant clones in the presence of the phage and in the controls. This can be explained both by their intrinsic fitness advantage in nutrient-rich environments, compared to wild-type clones (Buffet et al., 2021) and by their efficient resistance to phage (de Sousa et al., 2020; Hesse et al., 2020; Tan et al., 2020). Hence, their fitness advantage likely drove their rapid expansion in the population (Figure 3). Figure 4 Download asset Open asset Growth of BJ1-resistant clones. All isolated clones that were capsulated and tested for resistance (Figure 3A and Figure 3—figure supplement 1) were grown in the absence of phage (control) or with phage. One or two non-capsulated clones per day were randomly selected and included in the analyses. The area under the growth curve of each clone was estimated, and compared to that of the BJ1 ancestor (dashed line). K-W, Kruskal-Wallis test. Post hoc tests for significant differences across groups were calculated and p-values adjusted for multiple testing with Bonferroni correction. a, p<0.05 difference from non-capsulated; b, p<0.05 difference from resistant by other mechanisms; c, p<0.05 difference from lysogens; d, p<0.05 difference from sensitive strains. AUC analyses also revealed that lysogens are less fit than the resistant capsulated clones, in the absence of phage. However, this is not so, in the presence of phage at short evolutionary time scales. At longer time scales (after ten days of evolution), resistant capsulated clones are fitter than lysogens, in the presence of phage, supporting our previous observations that lysogeny incurs a high fitness cost. This suggests that resistant mechanisms that emerge more frequently or that are accessible in evolutionary terms, such as capsule inactivation and lysogeny could be initially selected for, but become less advantageous at longer timescales where other less costly resistance mechanisms seem to provide higher fitness. Overall, our results show a hierarchy in the competitive advantages of phage resistance mechanisms, that varies across time scales and phage pressure. Several changes in the receptor production play a role in resistance to phages To identify the mechanisms of resistance to phages that involved neither receptor inactivation nor lysogeny, we characterized twelve random clones out of the 328 clones with such profiles. We measured their capsule production and resistance to purified phage lysate, either on a layer of melted agar or during growth in liquid culture. We then tested the ability of each clone to adsorb phage lysate to understand if resistance occurs prior to entering the cell. As controls, we used the ancestral strain (BJ1), as well as an ∆rcsB mutant, with reduced capsule expression, and a non-capsulated ∆wcaJ mutant (Figure 5). Additionally, we performed whole genome sequencing on all twelve resistant clones and looked for mutational targets, using the ancestral sequence as a reference (Table 1). Figure 5 Download asset Open asset Characteristics of phage-resistant clones. (A