Impact of Spontaneous Prophage Induction on the Fitness of Bacterial Populations and Host-Microbe Interactions

SOS-INDUCED SPI

Because an induced SOS response is responsible for the induction of many lambdoid lysogens, it was discussed by several researchers that spontaneous SOS induction in single cells might trigger the induction of prophages (10). During growth, ongoing (multifork) replication has been shown to cause sporadic DNA damage resulting in the derepression of the SOS genes (16). Recent single-cell studies revealed a small fraction of SOS-induced cells in clonal populations grown under standard conditions, and it is reasonable to infer that a prolonged induction will also lead to the activation of resident prophages (13, 17, 18).

Lwoff first described free phage appearing in the supernatant of noninduced cultures of lysogenic bacteria (8), and it was later shown that recombination-deficient recA mutant strains of Escherichia coli showed no discernible spontaneous induction of the prophage (19–21). Different parameters were shown to influence the spontaneous induction of phage λ expression in E. coli, such as overexpression of CI and deletion of recA, thus decreasing the growth rate, or exchanging glucose with glycerol, thus affecting cyclic AMP (cAMP) levels, which have an influence on the SOS response as well (22). In their recent study, Little and Michalowski found that the intrinsic switching rate of E. coli lambda lysogens is almost undetectably low (<10⁻⁸/generation) in a recA mutant background (21). Remarkably, the intrinsic stability very much depends on the cultivation conditions as well as on the occurrence of mutations which render the lysogen unstable. These findings emphasize that SPI has coevolved with its specific trigger (e.g., the SOS response) to optimize the switching frequency over a wide range of naturally occurring inducing conditions.

Recently, it was shown that single cells of Corynebacterium glutamicum undergo spontaneous SOS induction under standard cultivation conditions (13). This spontaneous induction of the SOS response is sufficient to induce key promoters within prophage CGP3, among them the genes encoding a putative integrase and lysin (13). Mutant strains lacking recA showed only a basal level of spontaneous CGP3 induction, which might be due to stochastic effects (see below) or mutations of the prophage. That stochasticity might play a rather minor role is also emphasized by a study by Livny and Friedman, who could show that SPI of the easily inducible H-19B prophage in almost all cases coincided with the induction of a second prophage in the same cells of Shiga toxin-producing E. coli (STEC) lysogens (23).

Another layer of control is added by RecA-independent and yet SOS-dependent SPI. In the case of coliphage 186, the viral Tum protein functions as an antirepressor which binds to the phage repressor, resulting in the induction of lytic growth (24, 25). Expression of tum itself is under the control of host LexA, and phage induction is thereby again linked to the SOS response system of the bacteria (Fig. 1). Similar mechanisms were also described for the N15 linear plasmid-prophage of E. coli and the Fels-2 Salmonella enterica phage (26, 27). Another interesting example is the induction of the CTX prophage of Vibrio cholerae, which encodes cholera toxin. Here, expression of the genes required for virion production is directly regulated by host LexA, which binds just upstream of the RstR phage-encoded repressor (28, 29). Repressor inactivation can also be achieved by the transcriptional regulator of the cps gene cluster RcsA or by the small RNA DsrA, which relieves repression of rcsA (30).

EXTRINSIC FACTORS

In addition to the intrinsic factors which affect genomic DNA or RecA and induce the SOS response, extrinsic factors, such as reactive oxygen species (ROS) generated within macrophages (31–33) and UV radiation, both of which induce DNA damage, or the effects of antibiotics such as mitomycin C (MmC) (34) and fluoroquinolones (Fq) (35), should be taken into account as well. Fqs and MmC are both able to induce pneumococcal prophages in multiple strains of Streptococcus pneumoniae (36). In this bacterium, induction acts via RecA and yet does so in an SOS-independent manner and represents a response to the topoisomerase IV-Fq complex formed or to transcriptional regulation of phage or bacterial genes by DNA supercoiling, which is a result of Fq treatment (36). Further factors which influence the lysogenic maintenance or induction of the prophage should be taken into account as well. The effects of pH, temperature, organic carbon, and the presence of chromium (VI) and the toxic compound potassium cyanide on prophage induction were studied in the ammonia-oxidizing Nitrospira multiformis bacterium. Whereas increasing levels of Cr (VI) and KCN led to an increase in prophage induction, a shift toward acidic pH had the most dramatic effect (37). The neutralophile Helicobacter pylori B45, which inhabits the human gastric mucosa, was shown to induce its phiHP33 bacteriophage after UV irradiation and at a low pH (38, 39). The opposite relationship was observed for prophage ϕLC3 of Lactococcus lactis, where a decrease in pH reduced the amount of spontaneously induced prophage (40). In their multifactorial experimental setup, the authors revealed that an increase in the levels of simultaneously acting stressors increased SPI. It is tempting to speculate that RecA and the SOS response are responsible for the prophage induction dynamics in this system, because the repressor of L. lactis prophage ϕLC3 belongs to the CI-like repressor family. Indeed, earlier studies showed that changes in intracellular pH influence the stability and binding properties of LexA or CI repressor and thus influence the transcription of target genes (41, 42).

IMPACT OF STOCHASTICITY IN GENE EXPRESSION

Gene expression can be surprisingly dynamic and heterogeneous. Cell-to-cell variation, even in clonal populations of cells grown under the same conditions, has been observed in a variety of different organisms, from mammalian stem cells to bacteria (9, 43). Hence, a plausible mechanism for SPI is that it could result from spontaneous fluctuations in the levels of repressor such that, below a threshold level, the lytic genes would be expressed, leading to switching to the lytic state. This model does not apply in the best-studied case, phage lambda, since SPI requires RecA; in addition, phage mutants with noncleavable repressors do not show SPI, despite the likelihood that expression levels of these mutant proteins would fluctuate in the same way as in the wild type. In contrast, the unstable mutant described by Little and Michalowski likely switches by this mechanism (21). Other examples of switching also appear to result from stochastic fluctuations in gene expression. Spontaneous excision of the mobile element ICEclc (integrative and conjugative element encoding clc genes) was, for example, shown from the genome of Pseudomonas knackmussii B13 (12, 44). Integrative and conjugative elements reside in the host genome and are able to excise and be transferred via conjugation. The excision of ICEclc depends on variations in the levels of the RpoS stationary-phase sigma factor among individual cells. RpoS levels reach a threshold at which ICEclc-encoded excision factors are expressed, and this state is locked in these single cells, thus maintaining ICEclc excision in the form of a bistable state (12). In their recent study on several mycobacteriophages, which are not SOS inducible, Broussard et al. described a novel class of simple switches relying on the site-specific recombination catalyzed by integrases as a key to decision (45). They also observed spontaneous switching from lysogeny to the lytic state and speculated that this might be due to a drop in the repressor level below the threshold or to sporadic expression of the integrase gene (45).

Noise in gene expression is ubiquitous. It can provide a selective advantage by increasing phenotypic heterogeneity within a clonal population of one species or even within microbial communities (e.g., biofilms). It is therefore conceivable that several mobile genetic elements, such as ICEclc, pathogenicity islands (PIs), and prophages, exploit noise as a function to modulate the frequency of their spontaneous activation and transfer (Fig. 1).

IMPACT OF LYSOGENIC PROPHAGES ON THE FITNESS OF BACTERIAL POPULATIONS

In their recent review, Bondy-Denomy and Davidson summarized the general positive impact that prophages can have on a population's fitness (46). For instance, this positive impact was impressively demonstrated by the work of Wang et al., who deleted nine cryptic prophages in E. coli K-12 (47). These cryptic prophages are unable to propagate into infectious virus particles, and it was assumed that functions which assist their host to propagate in adverse environments were retained. The study revealed not only a decreased growth rate in the prophage-cured strain but also that it was more sensitive to antibiotics, was impaired in adaptation to osmotic stress, and showed a decrease in biofilm formation (see the next section for more details on biofilm formation). Interestingly, seven of the nine deleted prophages were shown to excise spontaneously (47). As we go on to show, this spontaneous activity of prophage elements can have a high impact on the general fitness of bacterial populations under diverse conditions.

SPI can also promote the spread of phages and increased survival of lysogens when they are grown in mixed populations. Salmonella enterica serovar Typhimurium harbors four to five full-size prophages, with most being able to undergo lytic development (48–50). When clonal populations of strains carrying these prophages were cultivated, spontaneous prophage induction had no apparent consequences for the population due to its immunity to superinfection (51). However, when prophage-carrying strains and those cured of the prophages of the same origin (52) or of different origins (51) were cocultivated, a “selection regime” that forced maintenance and spread of viral DNA was set, with a portion of bacteria being killed due to lytic development of prophages and survivors undergoing lysogenic conversion. One may think of this process as lysogenic bacteria using spontaneous lytic development of their prophages as a weapon, giving them a competitive advantage against nonlysogenized cells. At the same time, the prophage uses this war to its own advantage, with the ultimate goal of spreading its DNA by lysogenic conversion of the nonlysogens (52).

The presence of inducible prophage elements can also be a selective disadvantage, as shown for Staphylococcus aureus in the nasopharynx. It is displaced by its relative Streptococcus pneumoniae, which generates H₂O₂ and is thus responsible for the formation of hyperoxides via the Fenton reaction. S. pneumoniae itself, however, is resistant to them (53). RecA-inducible prophages of S. aureus are induced, leading to cell death (54) and, ultimately, displacement of the strain. Thus, despite its positive effect, the presence of lysogenic bacteriophages may add a potential Achilles heel with respect to the fitness of the bacteria which is exploited by bacterial competitors and eventually might even be exploited by the human host.

PHAGES AFFECT MICROBIAL BIOFILM FORMATION

In nature, the vast majority of bacteria is thought to exist in surface-associated communities, commonly referred to as biofilms (55). In biofilms, the cells are encased in a matrix mainly consisting of various extracellular polymeric substances (EPS). The most prominent hallmark of cells growing in such communities is their increased tolerance of a wide range of environmental perturbations, including antibacterial and antibiotic treatments. Biofilm formation varies widely between different species but is often described as a developmental process (56). Phage-induced lysis within biofilms generally leads to an accumulation of extracellular DNA (eDNA) within the community. In concert with the high cell densities commonly occurring in biofilms, this provides a huge pool for horizontal gene transfer (reviewed in references 57 and 58).

An increasing number of studies on several bacterial species have provided evidence that (pro)phages also directly affect the progression of biofilm formation through all stages (Fig. 2). The presence of the phages can lead to biofilm dispersal resulting from cell lysis while at the same time providing the enzymes needed to degrade the extracellular matrix. On the other hand, lysis of a cellular subpopulation may provide biofilm-promoting factors such as matrix components. And finally, phages drive the diversification of the biofilm community, a major factor responsible for the overall fitness of bacterial communities (59).

• Open in new tab
• Download powerpoint

FIG 2

Impact of SPI on host physiology. Spontaneous induction of lysogenic prophages can have manifold effects, such as enhancing biofilm formation, playing a vital role in bacterial virulence or leading to horizontal gene transfer of its own DNA (including virulence factors) and of host genes.

All three aspects of bacterium-phage interaction in biofilm formation could be identified in P. aeruginosa PAO1. When grown in flow cells (hydrodynamic growth conditions), this strain undergoes an intricate biofilm developmental cycle which is strongly affected by the Pf4 filamentous bacteriophage (60). During the initial steps of biofilm formation, Pf4 activity is strongly controlled by production of the PhdA “prevent-host-death” factor encoded by the host (61). However, activation of the Pf4 prophage in later stages of biofilm formation (62, 63), likely due to elevated ROS levels, results in formation of a hyperinfective form of Pf4 and extensive cell lysis within the three-dimensional structures, leading to dispersal of the biofilm (62, 63). In contrast, complete loss of the prophage or overproduction of the PhdA prevent-host-death factor leads to decreased cell lysis and, notably, less-stable biofilms, a result which was attributed to the lack of eDNA as an important structural component (60, 61). Thus, it was proposed that controlled Pf4-mediated cell lysis is required for the release of biofilm-promoting factors, such as eDNA, in P. aeruginosa biofilm formation (64, 65). Activity of Pf4 phages coincided with formation of small-colony variants of P. aeruginosa PAO1, which were characterized by increased biofilm formation capacities (66). In addition, loss of the phage significantly decreased the virulence of P. aeruginosa PAO1 in mice, which has been attributed to phage-induced strain variations (60). This example nicely illustrates how the components of phage-host interactions are intimately linked and have evolved to benefit microbial group behaviors.

This is further exemplified in E. coli, in which the CP4-57, DLP12, e14, and rac cryptic prophages affect the biofilm formation of their host (47, 67). Presumably, under conditions of nutrient limitation and the presence of uncharged tRNAs, the Hha regulator activates lytic prophage genes in CP4-57 and DLP12, resulting in cell death and dispersal (68). Notably, excision of CP4-57 and its subsequent loss from the chromosome predominantly occurred in biofilm cells. These cells exhibit enhanced flagellum-mediated motility and decreased nutrient metabolism, thereby broadening the diversity of the population (67).

The diverse roles phages might play during biofilm formation are reflected in several further studies. Propagation of biofilm formation through phage-mediated lysis and eDNA release has been proposed for two other bacterial species, Streptococcus pneumoniae and Shewanella oneidensis MR-1. Spontaneous SV1-mediated cell lysis was shown to occur in S. pneumoniae, and, accordingly, deletion of the SV1 phage lysin led to a decrease in cell lysis, eDNA release, and biofilm formation (69). In S. oneidensis, three prophages, MuSo1, MuSo2, and LambdaSo, jointly affect lysis and eDNA release, and consequently, a mutant devoid of all three phages is severely impaired in all stages of biofilm formation (70). Recent studies on this species have demonstrated that environmental iron levels are an important factor involved in inducing expression of LambdaSo in a RecA-dependent fashion during biofilm formation (71). Another highly interesting example for host-phage interactions is the lysogenic infection of P. aeruginosa PA14 with bacteriophage DMS3. Lysogeny results in decreased biofilm formation and bacterial swarming. Notably, this effect is dependent on the clustered regularly interspaced short palindromic repeat (CRISPR)-CAS system of the host (72).

These examples illustrate the complex interaction between (pro)phages and their hosts with respect to group behaviors such as biofilm formation. Given the fact that prophage genes are commonly among the highly regulated genes during biofilm formation and that many potential cell surface factors are encoded by prophages, it is likely that we have as yet seen only the tip of the iceberg.

SPONTANEOUS PROPHAGE ACTIVITY PROMOTES HOST VIRULENCE

A major reason we are interested in bacteria is their impact on humans. Early in the last century, d'Hérelle discovered that it is not only the relationship between humans and bacteria that is of importance but rather that bacteriophages need to be included as well (73). This seems plausible, especially when changing focus to the emergence of virulence in pathogenic bacterial strains. The formation of biofilms, which we have just covered, is in itself already an important factor for pathogen virulence. In many bacterial infections, the first step is attachment and adhesion to the right host cells (74).

Once they are established within the host, the release of extracellular toxins is a hallmark of several pathogens. For example, Shiga toxin-producing E. coli (STEC) bacteria release Shiga toxins in the gut. The life cycle of STEC bacteria and their Stx-encoding prophages exemplifies how the spontaneous induction of prophages in a small number of cells is important for pathogenic traits (Fig. 2). The colonization of the gastrointestinal tract of ruminants was assumed to rely on the spontaneous induction of a subset of STEC serovar O157:H7 cells (75). It was proposed that the presence of a small fraction of induced STEC cells would lead to a sufficient release of Shiga toxin and thus to diarrhea, enabling the spread of the remaining uninduced lysogenic cells (23). Indeed, a basal level of spontaneous induction could be shown for Stx-encoding prophage H-19B of STEC serovar O26:H19 (23). Remarkably, the level of SPI was significantly higher than the level of induction of λ in the same strain background, suggesting that this higher frequency of SPI coevolved to meet the requirements for toxin release of the pathogenic host strain.

In Shiga toxin-producing E. coli (STEC) bacteria, the spontaneous induction of prophages increases the fitness of the entire bacterial population within the host environment by priming the host's epithelial cells for infection (23, 75, 76). In the proposed model, a small fraction of cells induces the Stx-encoding prophages, leading to Shiga toxin release and, in consequence, priming the gut epithelial cells by the expression and relocation of nucleolin and further receptors to the epithelial cell surface. These receptors were previously shown to bind to the bacterial surface protein intimin, thus promoting STEC colonization (77–79). In these bacteria, expression of a type 3 secretion system (T3S), which is required for colonization, was shown to be influenced by the presence of Stx-encoding prophages. Reconstitution in an E. coli K-12 background provided evidence for a negative impact of lysogeny (in particular, that associated with cII) on expression of the T3S (reduction by about 2-fold). However, the effects described in this study were rather small and their actual impact on host-microbe interaction remains to be elucidated.

Evidence for the role of SPI in other human diseases is only beginning to emerge. One example is seen in infective endocarditis, which is an inflammation of the inner lining of the heart. It is often caused by Streptococcus mitis. In this organism, the surface expression of øSM1 prophage-encoded PblA and PblB is dependent on the presence of holin and lysin, which are expressed in the course of prophage induction (80). Upon permeabilization and lysis of the bacterial host cell, PblA and PblB are released into the culture medium, bind choline within the bacterial cell wall of unlysed cells, and mediate platelet binding and aggregation (81) (Fig. 2; deletion of pblA and pblB leads to a 40% reduction in binding). The bacterial lysin itself also plays a role in platelet binding. Through its interaction with choline, it becomes cell wall associated and can directly bind platelet fibrinogen (82). Enterococcus faecalis is a leading cause of hospital-acquired bacterial infections and may cause urinary tract infections, intra-abdominal infections, and infective endocarditis (83). E. faecalis strain V583 is polylysogenic for 7 prophage-like elements (termed V583-pp1 to V583-pp7) (Table 1). It was shown that three of these, pp1, pp4, and pp6, are important for adhesion to human platelets, which is possibly achieved by expression of the genes encoding PblA and PblB which are homologous to those of S. mitis (84). Thus, these examples nicely reveal that SPI triggered lysis of a small fraction of cells, possibly promoting the adhesion of the remaining noninduced part of the population.

Besides having a direct effect on bacterial virulence, the role that spontaneous prophage activity plays in horizontal gene transfer of virulence-associated factors is of great concern (85, 86). S. aureus pathogenicity islands encoding toxins (SaPI) are highly mobile and packaged into small infectious particles (87, 88). SaPIbov1 (89) and SaPIbov2 (90) are two SaPIs which contain the tst gene encoding the toxic shock syndrome toxin (TSST) and the bap (biofilm-associated protein) gene (91), which is needed for persistence in mammary tissue (90), respectively. Once excised, both PIs are transferred due to the action of resident prophages. It was shown for both PIs that they spontaneously excise from S. aureus genomes. Consequently, it was postulated that, due to the spontaneous induction of resident prophages in combination with the low-level activation of SaPIs, packaging into transducible phage particles is ensured and represents a serious case of horizontal gene transfer in habitats colonized by staphylococci (92).

PHAGE-PHAGE INTERACTION

Besides the effect that phages have on their bacterial host or on human physiology, phages can also influence one another. For STEC cells, it was shown that repressor proteins of different Stx-encoding prophages mutually influence their spontaneous induction, in turn influencing the stability of the lysogenic state and, ultimately, the virulence due to Stx production (93). Another example is found in Salmonella species. Infection of Salmonella enterica with P22 or SE1 leads to a kil-dependent induction of the SOS response (94). This in turn activates the LexA-controlled antirepressors which bind the Gifsy phage repressors (Fig. 1) (95). The phage repressor of Gifsy-2, however, is insensitive to the antirepressors of Gifsy-1 and Gifsy-3. Rather, it is inactivated by the antirepressor of the Fels-1 prophage. This Fels-1 antirepressor is neither regulated by LexA nor involved in Fels-1 induction at all (Fels-1 induction relies on a λ CI-like repressor cleavage mechanism). Thus, the ability of prophages to influence the stability of other prophages and the properties of antirepressors with respect to recognizing noncognate substrates allow multilayered induction regulation in polylysogenic strains.