No effect of host–parasite co-evolution on host range expansion


Correspondence: Pauline D. Scanlan, Department of Zoology, University of Oxford, South Parks Road, Oxford OX1 3PS, UK.

Tel.: +44 (1865) 271100; fax: +44 (1865) 310447;



Antagonistic co-evolution between hosts and parasites (reciprocal selection for resistance and infectivity) is hypothesized to play an important role in host range expansion by selecting for novel infectivity alleles, but tests are lacking. Here, we determine whether experimental co-evolution between a bacterium (Pseudomonas fluorescens SBW25) and a phage (SBW25Φ2) affects interstrain host range: the ability to infect different strains of P. fluorescens other than SBW25. We identified and tested a genetically and phenotypically diverse suite of co-evolved phage variants of SBW25Φ2 against both sympatric and allopatric co-evolving hosts (P. fluorescens SBW25) and a large set of other P. fluorescens strains. Although all co-evolved phage had a greater host range than the ancestral phage and could differentially infect co-evolved variants of P. fluorescens SBW25, none could infect any of the alternative P. fluorescens strains. Thus, parasite generalism at one genetic scale does not appear to affect generalism at other scales, suggesting fundamental genetic constraints on parasite adaptation for this virus.


Antagonistic host–parasite co-evolution, the reciprocal selection for host resistance and parasite counter-resistance, can select for parasites with the ability to infect a wider range of host genotypes other than those with which they have co-evolved. It is argued that parasites, notably viruses, with broad host ranges may be genetically predisposed to infect hosts that they have not previously been encountered (Woolhouse, 2002). Indeed, single mutations can significantly alter host range specificities (Woolhouse et al., 2005; Duffy et al., 2006). An alternative possibility is that co-evolved parasites may have a reduced ability to infect hosts that they have not been exposed to (despite showing an increased host range with hosts they co-evolve with), as a result of costs associated with generalism and mutation accumulation or as a result of increasing specialization (Elena et al., 2009). Here, we attempt to experimentally determine the ability of co-evolved parasites with varying host ranges (against co-evolved hosts) to infect hosts genetically similar to the host they had co-evolved with, but which the phages had not previously encountered. In other words, does increased infectivity range resulting from co-evolution with a single host strain affect the ability to infect other host strains?

To this end, we used the lytic bacteriophage SBW25Φ2 and the host bacterium Pseudomonas fluorescens SBW25 (Buckling & Rainey, 2002a), which have been shown to undergo extensive antagonistic co-evolution in vitro, characterized by the continual appearance of resistant host genotypes and infective phage genotypes. Critically, both individual- and population-level analyses have demonstrated that co-evolution in vitro (but see Gomez & Buckling, 2011) results in the evolution of phages that can infect an increasingly wide range of the evolved bacteria genotypes through time (Buckling & Rainey, 2002b; Scanlan et al., 2011). We then determined the ability of phages with varying host ranges with respect to Pseudomonas fluorescens SBW25 genotypes to infect other P. fluorescens strains.

Materials and methods

Choosing novel co-evolved phage

We isolated co-evolved phage from six independent co-evolving populations that were initiated with isogenic Pseudomonas fluorescens SBW25 and SBW25Φ2 (Hall et al., 2011b). We chose 55 unique and diverse co-evolved SBW25Φ2 phage genotypes from three time-points of the co-evolution experiment. This selection was based on a large-scale phenotypic assay (that tested phage infectivity against a wide range of sympatric and allopatric co-evolved hosts) and genetic analysis of the tail-fibre gene that is involved in host recognition and infectivity, see Scanlan et al. (2011). All co-evolved phage had a greater host range than the ancestral phage and had acquired a number of infectivity mutations at the tail-fibre gene, see Fig. 1a,b.

Figure 1.

(a) Co-evolution selects for phages with an increased host range relative to the ancestral WT phage (T0). Light grey bars show the mean proportion of allopatric and sympatric co-evolved hosts infected by co-evolved phage (±1 SD) isolated from various transfer time-points of the co-evolution experiment (see Hall et al., 2011b). (b) Co-evolved phages have a high number of amino acid substitutions at the tail fibre gene relative to the ancestral WT phage (T0). Mutations at the tail fibre are required for host range expansion and host infectivity in SBW25Φ2, see (Hall et al., 2011a; Scanlan et al., 2011).

Novel bacterial hosts

One hundred and fifty Pseudomonas fluorescens host strains from a well-studied strain collection were obtained, and an overview of these bacteria is provided in Table S1. Although many of the National Collection of Plant Pathogenic bacteria ( and Collection Française de Bactéries Phytopathogènes ( strains have different qualifier names, all fall within the P. fluorescens ‘complex’ (with a small number of exceptions), proposed in 2000 by Yamamoto et al. According to this classification, which is based on gyrB and rpoD nucleotide sequence analysis, the Pseudomonas fluorescens complex is composed of two different lineages (‘chlororaphis’ and ‘fluorescens’ lineage). Pf-5 and Pf0-1 fall within the ‘chlororaphis’ lineage and SBW25 in the ‘fluorescens’ lineage (Yamamoto et al., 2000), which also includes over half of the strains in this collection. Also included in this collection of strains are 14 strains that were isolated along with SBW25 in a field sampling experiment (Haubold & Rainey, 1996; Rainey & Bailey, 1996). One hundred and fifty strains is a high number of host strains compared with other studies (Ceyssens et al., 2006; Cornelissen et al., 2011; Sepulveda-Robles et al., 2012), with perhaps a few exceptions (Michel et al., 2010); however, it is possible that if additional strains were used, this may have altered the experimental result.

Testing co-evolved phage against novel hosts

To test for phage infectivity against novel Pseudomonas fluorescens host strains, 30 μL of each co-evolved phage genotype and the ancestral phage (~1 × 107 PFU mL−1) was spot-plated in triplicate onto growing lawns of each bacterial host (that had been reconditioned from stock by growing for 24 h in liquid KB at 28 °C, n = 150) using KB soft agar overlay plates. Plates were placed in a 28 °C incubator and checked for phage plaques (zones of lysis that indicates parasite infectivity) after 8, 12, 24, 48 and 72 h of incubation. P. fluorescens SBW25 was used as a control host for all phage genotypes. We also conducted additional experiments to assess whether host receptor binding is the key limiting step in phage infectivity and reproduction, see Appendix S1).

Results and discussion

Here, we determined whether intrastrain host range expansion of a bacteriophage resulting from real-time co-evolution also leads to changes in the ability to infect other related strains of bacteria. Even though all co-evolved phage had novel infectivity alleles (with some genotypes having up to 28 nonsynonymous amino acid mutations) and an increased host range relative to the ancestral phage (Fig. 1), not a single phage genotype was able to infect any of the novel 150 Pseudomonas fluorescens strains tested in this study. This result indicates that although co-evolution with bacterial hosts can select for generalist parasite phenotypes with novel infectivity alleles (Hall et al., 2011a), these generalist variants are confined in their infectivity range to sympatric and allopatric co-evolved variants of the ancestral host bacterium (Pfluorescens SBW25).

There are a number of possible explanations for the inability of our co-evolved viral parasites to infect novel hosts. Firstly, the resistant strains may lack the phage receptor present on the susceptible strains, SBW25. The exact host receptor site of phage SBW25Φ2 is currently under investigation, but based on a transposon mutagenesis study and homologies to T7 phage, the receptor site of SBW25Φ2 is most likely the O-antigen of lipopolysaccharide (LPS). LPS is a critical component of the outer surface of all Gram-negative bacterial cell walls and is a common receptor molecule that bacteriophage utilizes in host recognition and infectivity (Lindberg, 1973; Temple et al., 1986). LPS has three main components, namely the highly strain variable O-antigen, a core polysaccharide and the more conserved lipid A structure (Lam et al., 2011). If, as is the case for the related Escherichia coli phage T7 (Qimron et al., 2006), SBW25Φ2 variants use the host variable O-antigen for host binding and infection, then the inability of our co-evolved phage to infect novel hosts is perhaps unsurprising. By contrast, if the phage targeted a conserved ligand that is shared by multiple strains or species, then the adaptive potential for host range expansion in co-evolved parasites with novel infectivity alleles would likely be much greater. Note that additional experiments, where we attempted to directly electroporate phage DNA into bacteria, bypassing the need for receptor compatibility, also failed to cause infection. Although this finding is inconclusive, it might suggest that receptor binding is not the only constraint on successful infection.

Secondly, bacteria may be resistant to bacteriophage because they often possess multiple redundant resistance mechanisms encompassing both pre- and post-infection resistance strategies; examples include phage superinfection exclusion, CRISPR and restriction modification (Horvath & Barrangou, 2010; Labrie et al., 2010). As such, a parasite will require multiple susceptible pathogen effectors in the right combination to infect and replicate within the host. Starting co-evolution with a susceptible interaction allows sequential evolution of host resistance and parasite infectivity; however, starting with a resistant interaction or even a successful infection, several layers of mechanism could block infection – for example, CRISPR and DNA restriction modification (Labrie et al., 2010). Thus, the SBW25 antigen may be present in other strains but backed up by other mechanisms to deal with infection once bacteriophage has gained entry into the cell.

Finally, our perception of parasite host range may be confounded by the use of different taxonomic and phylogenetic classifications to group hosts (in this case, bacteria). For example, Salmonella, Shigella and Escherichia coli are examples of bacteriophage hosts that have been used in experimental evolution to address a number of questions including epistasis in host–parasite interactions and fitness trade-offs in host range expansion (Crill et al., 2000; Pepin & Wichman, 2007; Elena et al., 2009). Although the nomenclature of these bacteria would suggest they are members of distinct genera, these bacteria are very closely related (Fukushima et al., 2002) and Shigella spp. are in fact serotypes of E. coli (Ochman et al., 1983; Zhao et al., 2007). Conversely, consistent with earlier analysis (Yamamoto et al., 2000), recent whole-genome sequence analysis of P. fluorescens has defined this group of bacteria as a species complex with a much greater genetic diversity compared to other groups of Pseudomonas species (Silby et al., 2009). As such, bacterial and different host labels may have little evolutionary and ecological significance and the delineation of bacteria or other hosts into different groupings based on limited genetic information may greatly alter the perceived host range of a parasite.

Our results suggest that bacteria–phage antagonistic co-evolution, while commonplace in natural populations (Buckling & Brockhurst, 2012), is likely to contribute little to host range shifts unless there is a pre-existing host–parasite genetic compatibility and/or an absence of secondary defence mechanisms to counter parasite replication within the host. This view appears inconsistent with a recent experimental co-evolution study of a cyanobacteria and phage, where co-evolution resulted in phages able to infect an initially resistant strain that was genetically distinct to the cyanobacteria the phage co-evolved with (Marston et al., 2012). However, this resistant strain was experimentally derived from a strain that was originally sensitive to the phage; hence, co-evolution did not in fact result in an increase in the number of strains that could be infected. In the absence of some initial bacteria–phage compatibility, spontaneous mutation alone is unlikely to result in phage infection, but instead may require horizontal gene transfer (Haggard-Ljungquist et al., 1992; Jeltsch & Pingoud, 1996; Baumler et al., 1998; Lawrence et al., 2002).


We are grateful to the European Research Council for funding. The authors would also like to thank Dr Olivier Tenaillon and an anonymous reviewer for comments and suggestions that greatly improved the quality of the manuscript.