Inactivation of the Mismatch Repair System in Pseudomonas aeruginosa Attenuates Virulence but Favors Persistence of Oropharyngeal Colonization in Cystic Fibrosis Mice

Abstract

The inactivation of the mismatch repair (MMR) system of Pseudomonas aeruginosa modestly reduced in vitro fitness, attenuated virulence in murine models of acute systemic and respiratory infections, and decreased the initial oropharyngeal colonization potential. In contrast, the inactivation of the MMR system favored long-term persistence of oropharyngeal colonization in cystic fibrosis mice. These results may help in understanding the reasons for the low and high prevalences, respectively, of hypermutable P. aeruginosa strains in acute and chronic infections.

The establishment of Pseudomonas aeruginosa chronic infections is mediated by a complex adaptive process that includes physiological changes produced by the activation of specific regulatory pathways, including the induction of the biofilm mode of growth or the differential expression of virulence genes (24), and the selection of an important number of adaptive mutations required for long-term persistence (20).

A common feature of P. aeruginosa chronic lung infections, including those occurring in patients suffering from cystic fibrosis (CF), bronchiectasis, or chronic obstructive pulmonary disease, is a very high prevalence (30 to 60%) of hypermutable (or mutator) strains deficient in the DNA mismatch repair (MMR) system, in contrast to that observed in acute processes (<1%) (1, 6, 9, 11, 16, 17). The presence of hypermutable strains has been found to be linked to the high antibiotic-resistance rates of P. aeruginosa clinical isolates recovered from patients with chronic lung infections (1, 9, 11, 16), and in vitro and in vivo experiments have shown that hypermutation dramatically speeds up resistance development during exposure to antimicrobial agents (1, 12). Nevertheless, except for antimicrobial resistance development, a link between hypermutation and the genetic adaptation required for the long-term persistence of chronic infections has not yet been proved.

Laboratory and theoretical approaches have shown that, under particular circumstances, such as exposure to new environments or stressful conditions, mutator cells are selected in bacterial populations by hitchhiking with the adaptive mutations that they produce more frequently than the regular cells, therefore playing a role in evolution (13, 21, 23). Various in vivo models have also shown that hypermutation may favor the adaptation and persistence of bacterial pathogens. Giraud et al. (4), using a murine model of Escherichia coli intestinal colonization, found that hypermutation was initially beneficial because it allowed a faster adaptation to the mouse gut environment, although this advantage disappeared once adaptation was reached and the transmissibility of the hypermutable strains was then considerably reduced due to the accumulation of deleterious mutations for secondary environments. A similar result was obtained by Nilsson et al. (15) when studying the adaptation of Salmonella enterica serovar Typhimurium to the reticuloendothelial system of mice. Finally, it has recently been shown that the inactivation of the MMR system in E. coli favors the persistence of urinary tract infections in a mouse model (10).

In addition, various researchers have also determined whether the inactivation of the MMR system has a direct effect on pathogenicity, with variable results. Merino et al. demonstrated that the deletion of the mutSL locus in Listeria monocytogenes reduces the virulence of this intracellular pathogen in a murine model of infection (14), whereas, on the contrary, mutS inactivation does not seem to have any significant direct effect on virulence in animal models of infection by E. coli or Salmonella (19, 25).

In this work, we evaluated the direct effect of mutS inactivation on the virulence and colonization potential of P. aeruginosa and explored whether the high mutation rates produced by the inactivation of the MMR system could favor the adaptation process required for long-term persistence in the CF setting.

The inactivation of the MMR system in P. aeruginosa reduces fitness and attenuates virulence.

In vitro competition experiments between PAOΔmutS and its parent wild-type strain PAO1 revealed that the inactivation of the MMR system determines a modest but significant (P, <0.001) reduction in fitness when growing in a rich medium (Luria-Bertani [LB] broth), as shown by the median PAOΔmutS/PAO1 ratio (competition index [CI]) of 0.6 after 20 generations of growth (Fig. 1A). Furthermore, in vitro competition experiments performed between both strains harboring plasmid pUCPMS (which contains a copy of the PAO1 wild-type mutS gene) (18) demonstrated that the fitness cost was indeed associated with mutS inactivation, since the complementation of PAOΔmutS with the wild-type mutS gene almost completely abolished the reduction in fitness (median CI, 0.9) (Fig. 1A).

FIG. 1.

(A) Results of in vitro competition experiments between PAOΔmutS and wild-type PAO1 strains and between both strains complemented with the plasmid harboring the wild-type mutS gene (pUCPMS). (B) Results of in vivo competition experiments between PAOΔmutS and wild-type PAO1 strains in the mouse models of acute systemic and acute respiratory infections.

The inactivation of the MMR system in PAOΔmutS also determined an important biological cost and reduction in virulence in two in vivo models of acute infection. In the murine model of acute systemic infection [ICR (CD-1) mice; intraperitoneal inoculation], the mortality of mice infected with PAOΔmutS (19%) was significantly lower (P, <0.001) than that of mice infected with PAO1 (94%). Furthermore, PAOΔmutS was strongly outcompeted by the wild-type PAO1 strain, as shown by the median CI of 0.03 in bacteria recovered from the spleen 24 h after the intraperitoneal inoculation (Fig. 1B). In the murine model of acute respiratory infection (C57BL/6 mice; transtracheal inoculation), the mortality of mice infected with PAOΔmutS (37%) was also significantly lower (P, 0.04) than that of mice infected with PAO1 (67%). PAOΔmutS was also clearly outcompeted by wild-type PAO1 (median CI, 0.38; P, <0.001) in bacteria recovered from the lungs 48 h after the intratracheal inoculation, although this effect was milder than that observed in the systemic model (Fig. 1B).

Hypermutation favors persistent P. aeruginosa oropharyngeal colonization in CF mice.

The procedure used for the establishment of the CF mouse oropharyngeal colonization model was a modification of that described by Coleman et al. (2) using homozygous Cftr^tm1Unc-Tg(FABPCFTR) 1 Jaw/J mice (CF mice). A total of 36 CF and 36 wild-type (C57BL/6J) mice were exposed to approximately 5 × 10⁷ CFU/ml of exponentially growing cells of PAO1 and PAOΔmutS, mixed in a 1:1 ratio, in their drinking water for 10 days. Oropharyngeal swabs and bronchoalveolar lavages were obtained and plated in P. aeruginosa selective media at weeks 0, 1, 4, 8, 12, 16, 20, and 24.

As expected, oropharyngeal colonization by P. aeruginosa was found to be significantly favored in CF mice compared to wild-type controls (Fig. 2A). Nevertheless, despite their higher susceptibility, CF mice readily tended to eliminate this microorganism from the oropharynx, since the percentage of colonized mice decreased over time (for instance, only 20% of the mice were colonized after 16 weeks). Furthermore, a significant progression to chronic colonization of the lower respiratory airways was not achieved with this model: the percentages of CF mice showing positive bronchoalveolar lavage cultures was low and decreased over time (from 39% at week 0 to 7% at ≥16 weeks) and was always associated with a low bacterial load (<10³ CFU/ml).

FIG. 2.

Percentages of CF and wild-type mice showing P. aeruginosa-positive cultures from oropharyngeal swabs over time (A). Results of in vivo competition experiments between PAOΔmutS and wild-type PAO1 strains in the CF mouse model of oropharyngeal colonization: CIs over time (B) and percentages of CF mice showing P. aeruginosa-positive cultures in which PAOΔmutS outcompeted PAO1 (CI > 1) over time (C).

Competition experiments also revealed an initial disadvantage of PAOΔmutS for the oropharyngeal colonization of CF mice, as shown by the CI of 0.40 at week 1 (P, 0.03) (Fig. 2B). In contrast to what was documented for the acute infection models, a wide range of variation in the CIs (from below 0.1 to above 10) was noted for the different mice (Fig. 2B), and after 4 weeks, it was a frequent event to find just one of either strain in the cultures. These results, together with those described above (Fig. 2A), show a strong clearing pressure of P. aeruginosa from the oropharynx, even when colonizing CF mice. The overall disadvantage of PAOΔmutS continued up to week 8, when, in 75% of the P. aeruginosa-positive mice, PAO1 was the dominant strain (Fig. 2C). Nevertheless, from then on, the tendency was inverted, and in all positive CF mice after ≥16 weeks (20%), PAOΔmutS was the dominant strain (P = 0.01) (Fig. 2C). Namely, PAO1 was no longer detected in two of the six CF mice showing positive cultures after 16 weeks (only PAOΔmutS was detected) and for the other four, the CIs ranged from 2.2 to 48. Overall, these results show a clear positive selection of the hypermutable strain during long-term oropharyngeal colonization of CF mice.

In order to find out whether the selection of the hypermutable lineages during long-term persistence in the oropharynx of CF mice was driven by their coselection with mutations improving their oropharyngeal colonization capacity, competition experiments were performed between three different PAOΔmutS persisting lineages, recovered from CF mice 16 to 24 weeks after inoculation, and the PAO1 wild-type strain, using the same model. Competition experiments showed that the PAOΔmutS persisting lineages did not have an increased oropharyngeal colonization potential, but rather the contrary, since they were all clearly outcompeted by PAO1 during initial colonization, showing a median CI of 0.3 1 week after inoculation.

The results from this work are certainly a step forward in our understanding of the reasons for the low and high prevalences of mutators in acute and chronic infections, respectively. First, the observed attenuation of virulence produced by the inactivation of the MMR system should predict a reduced capacity of hypermutable strains to produce acute human infections, as well as a reduced possibility of their selection during the course of an acute infection process, even under the strong selective pressure exerted by antimicrobial treatments. Results from the CF mouse model of oropharyngeal colonization provide further clues in the same direction. Despite the mutS-deficient strain showing a reduced potential for the initial oropharyngeal colonization of CF mice compared to the potential of the wild type, it was invariably selected after long-term colonization in the absence of antibiotic pressure, clearly suggesting that hypermutation favors the selection of adaptive mutations required for long-term persistence. Nevertheless, long-term persistent mutS-deficient lineages did not show an increased capacity for the subsequent oropharyngeal colonization of new CF mice, but rather the contrary, since all adapted lineages were outcompeted by the wild-type strain during the initial oropharyngeal colonization of new CF mice. Similar results were obtained in a previous mouse model of intestinal colonization by mutS-deficient E. coli (4). These results, in concordance with those gathered from the study of collections of clinical P. aeruginosa strains isolated from CF patients (1, 9, 16), suggest that hypermutable lineages are selected as a consequence of adaptation to the environment of the respiratory tract of CF patients, but also that their capacity for efficient transmission to secondary hosts (other CF patients) is expected to be reduced.

Even though persistent oropharyngeal colonization by PAOΔmutS was documented in 20% of CF mice, we were not able to document clear signs of progressive lower respiratory tract colonization or subsequent pathological inflammatory response during the 6-month time frame studied. Mucoid variants, typical markers of P. aeruginosa chronic respiratory colonization of CF patients (5), were also not detected, although another phenotypic variant frequent in the CF setting, the small-colony variant (7), was readily selected over time in PAOΔmutS lineages (data not shown). Overall, these results suggest that during the 6-month time frame studied, only the first stages of the complex chronic respiratory colonization process were reached. One limitation of this model is that the evolution of the process cannot be followed much further, due to the natural life expectancy of mice. On the other hand, these observations make this model quite compatible with the natural course of the process in CF patients, in which the markers of chronic colonization typically appear several months after the initial colonization (3). The chronic respiratory process itself (but not its natural course) can certainly be more efficiently reproduced and studied by the direct inoculation into the respiratory tract of CF (or even non-CF) mice of natural (and therefore adapted) P. aeruginosa strains isolated from chronically infected CF patients (8). We intentionally used the nonadapted P. aeruginosa reference strain PAO1 (and its mutS-deficient derivative) recovered from a wound infection over 50 years ago (22) and exposure in the drinking water (instead of direct inoculation) in order to attempt to reproduce the natural adaptation process ultimately leading to chronic respiratory colonization in the CF setting.

In summary, we showed that the inactivation of the MMR system in P. aeruginosa reduced in vitro fitness, attenuated the virulence of acute infections, and diminished the initial oropharyngeal colonization potential but, on the other hand, favored the long-term persistence of oropharyngeal colonization in CF mice. These results may help in understanding the reasons for the low and high prevalences, respectively, of hypermutable P. aeruginosa strains in acute and chronic infections.