The following experimental workflow was implemented to address the main questions raised in our study (Fig. 1).
Experiments were conducted as follows: culture preparation (1); biofilm formation (2); phage infection with single or cocktail preparations (3); incubation (4); biofilm and planktonic populations sampling (5); culture plating on TSA agar and isolation of discrete colonies (6); phage typing determination (7); to select isolates with unique patterns (8) for further phenotypic (9) and genome sequencing analyses (10).
The P. aeruginosa PAO1 reference strain and four other clinical representatives were infected with distinct lytic phages in a single or different cocktail combination. Randomly picked colonies from the surviving cultures were then tested in terms of susceptibility to inoculated phages as well as to the others from the Pseudomonas phages panel (Table 1). We were interested in exploring the broadest clonal variability developed in phage infected Pseudomonas population. Therefore, the first phase of the study was focused on examining the phenotypic heterogeneity of PAO1 reference mutants (phage typing) within planktonic and biofilm populations. Since the consequences of introducing lytic phages into the bacterial population are difficult to predict, a representative pool of bacterial clones that have survived infection was sampled. A total of 780 P. aeruginosa PAO1 clones were typed with phages (planktonic (320), biofilm populations (400) and 60 control clones). No resistance to phages was observed among the control clones taken from untreated biofilm or plankton. Therefore, three biofilm and three planktonic representatives and the wild-type PAO1 were selected for further genetic and fitness analyses (Table S1). Finally, a pool of 95 isolates has been filtered, representing seventeen different phage susceptibility patterns (Tables S1, 2). This selection was based on the maximum variety of phage-type profiles, without accounting for the origin of the isolate (biofilm/plankton), as the infected planktonic bacteria turned out to be less diverse and all phage types were also present in the biofilm population.
Since we did not aim to analyse the differences of planktonic versus sessile cells response to phage infection but rather look for maximum population heterogeneity, we decided to focus the investigation on the biofilm population for the other clinical strains during the second stage of this research. Accordingly, 880 (30 clones from every condition plus 10 control clones for each strain) isolated colonies from A5803, AA43, CHA, and PA biofilm populations were first subjected to phage typing. No phage resistance was observed among clones taken from phage-untreated samples compared to the wild-type strain. Ultimately, 35 phage-treated colonies, three controls, and the wild-type from each strain were selected for further investigation, resulting in a pool of 156 clones in total (39 × 4 strains) representing over twenty different phage susceptibility patterns (Table S1 and S3).
Do phages always select for cross-resistance to other phages recognising the same bacterial receptor?
The application of monovalent phage against reference PAO1 population generally led to the selection of cross-resistance against phages that recognise the same receptor as the applied one (Table S2). This was observed for 12/17 and 23/24 PAO1 clones isolated after LPS- and T4P-dependent phages treatment, respectively. Similar relation (15/20) was only observed for other clinical cultures infected with phiKZ phage (T4P-dependent) (Table S3). The resistance to both groups of phages was less frequent in monovalent infections (14.5% in PAO1 and 32.5% for other clinical strains) compared to polyvalent infections (61.1%; 33/54) and 51.6% (31/60) for PAO1 and clinical strains, respectively. The use of a cocktail of two phages recognising LPS selected for PAO1 clones resistant only to LPS-dependent phages. In contrast, LPS-dependent phages application was mostly accompanied by the emergence of resistance to phages recognising alternative receptors in clinical strains (28/60 cases).
The introduction of a particular phage into the population did not guarantee the isolation of clones resistant to this phage. This event was recorded in the case of single phages, as well as for polyvalent combinations (23 PAO1 mutants). However, the cross-resistance to other phages recognising the same or both receptors did also occur. Interestingly, LUZ7 and KTN6 phages could still infect surviving clinical populations with a frequency of 23/60 and 44/80, respectively. Indicating that the resistance to LPS-dependent phages in clinical strains was more difficult to develop compared to those impaired by giant viruses, with 11/60 and 1/20 still sensitive to phiKZ and PA5oct phages, respectively. Almost all PAO1 (80/95) and clinical (127/140) clones treated with phages developed resistance to phage PA5oct, whereas the resistance to the entire phage panel emerged regardless of the single or cocktails application.
To conclude, the selection of cross-resistance to other phages recognising the same bacterial receptor was mostly valid in the PAO1 model, whereas the other clinical strains primarily developed the cross-resistance to T4P-dependent phages.
Do phages from different taxonomy groups recognising the same receptor cause the emergence of the same type of resistant mutants? Are the defence response and genome changes correlated with the receptor specificity of infecting phage?
To assess the genetic basis of the resistance selected by phages, we performed single nucleotide polymorphisms (SNPs) and mapping analyses of 102 reference PAO1 clones and 156 clones derived from clinical strains (Figs. 2, 3, Table S2–S4). The wild-type P. aeruginosa strains were also re-sequenced with Illumina and PacBio technologies to ascertain their complete genomic background. Missense, nonsense, and frameshift mutation variants were taken into account in the analyses. Mutations that also occurred in control isolates were excluded from further consideration. The remaining mutations were divided into six groups: LPS-related genes, mucoidity-associated genes (EPS production, biofilm formation), T4P-related genes, global regulatory genes, and others (hypothetical or undefined function genes). The comparative analysis showed the presence of point mutations in 64 out of 95 examined PAO1 mutants. The frequency of mutations in PAO1 clones isolated after treatment with single or multiple phages was similar (73% vs 61%, respectively). In most of those isolates, only one gene mutation event was recorded (43%). However, in 23 cases SNPs occurred in two or three genes belonging to different metabolic gene groups. Five PAO1 isolates showed the presence of mutations in two genes from one gene group. The 33 cases of SNPs related to LPS synthesis were found in 29 mutants selected with single LPS-dependent phage preparations or in polyvalent combinations. Among these, the most frequent mutation (21/33 cases) was observed within the wzy gene, encoding the B-band O-antigen polymerase [30]. These frequent mutations in the LPS-biosynthesis cluster confirmed the phage resistance results emerging after LUZ7, KT28, and KTN6 phages propagation. In some cases, the LPS gene modification was accompanied by changes in EPS-related genes, leading to a mucoid phenotype. The T4P-dependent phage treatment also led to the selection of specific mutations in genes responsible for T4P expression, but also alterations in flagella-related genes (flgH, fliN, fliP, flhA). The mutations in global regulatory genes (most frequent yqjG and vfr) and “others” gene groups did not show any correlation to the type of phages used.
The colour dots refer to particular gene groups where the point mutations (accumulated results) were recorded within the genomes of examined mutant clones. The lower line contains information on the maximum and minimum size of large deletions (grey bands) and the presence of intact prophages (light blue bands). * means mutation in promoter region of the gene.
Panel (A) represents the PAO1 clones, and panel (B) represents the clinical strains populations. Populations were selected by specific phages targeting LPS (red dots) or T4P (blue dots) as a single treatment or in cocktails. The colour bars refer to particular gene groups where the point mutations were recorded within the genomes of examined mutant clones. N means the number of analysed clones for each strain.
Apart from point mutations, 23% of phage-resistant PAO1 isolates contained large genomic deletions (23,983 bp–544,729 bp) appeared regardless of the phage-type and cocktail composition used as selective pressure agents. All deletions were located in the same region and despite different starting/ending points, they hold a core element of 19,038 bp. This core element carries the galU gene (responsible for LPS biosynthesis), as well as the hmgA gene, which causes the accumulation of brown pigment in bacterial cells when absent. Besides, the cumulated deletion range contained a total of 706,374 bp, including many key genes involved in the bacterial metabolism.
Mutations detected in other clinical phage-resistant clones were classified according to the same criteria as in PAO1 (Figs. 2, 3, Table S3, S4). The genome-driven response to phage infection of A5803 was primarily located in global (71%, cpdA) and other genes (34%, PA2911); of AA43 in other genes (31%, PA2911); of CHA in T4P (34%) and global genes (34%, morA); and of PAK in T4P (25%) and other genes (23%, PA2911). Most of the mutations selected by LPS-dependent phage exposition were found in the global regulatory genes (9–11–25–54%) or “other” genes (17–23–31%), rather than in the LPS biosynthesis locus (0–3–6–17%) depending on the impacted strain (Table S3). That confirmed the phage-typing results where LUZ7 and KTN6 phages remained lytic towards surviving clones. In contrast, the application of phiKZ selected for the cross-resistance to T4P-dependent phages as well as for the genetic modifications in pili-related genes. Mutations in global regulatory and “others” genes show no correlation to the receptor specificity of phages used. Interestingly, a portion of phenotypically phage-resistant clones in each clinical P. aeruginosa population (5-9/35 clones) did not reveal any distinguishable genetic modifications. Consistent with PAO1, large genomic deletions were observed in A5803, AA43, and PAK strains ranging between 92,207 bp and 383,693 bp in size, encompassing the galU region. The MEME analysis of the regions flanking the deletions did not reveal specific motifs that would indicate recombination events. Interestingly, the unique large deletion found in CHA strain (15,126 bp) turned out to be the induced ssDNA filamentous Pf1-like phage.
Summarising the analyses one might say that phages from different taxonomy groups recognising the same receptor generally cause the emergence of a similar type of resistance within a particular strain. However, the defence response and genome changes correlated with the receptor specificity of infecting phage differ in a strain-dependent manner.
Do different strains of P. aeruginosa react similarly to a specific phage infection?
The next step aimed to assess the impact of gaining phage resistance in terms of population growth efficiency as an indicator for bacterial fitness. Three of the examined wild-type strains (PAO1, A5803, and CHA) have a naturally rapid growth rate, while the other two (AA43 and PAK) display moderate growth rates. For this reason, the final results are expressed as the cumulated OD600 (Fig. 4, Table S2, S3). Overall, the majority of PAO1 mutants (61/95; 64%, p < 0.001) showed a significant decrease in growth rate compared to the untreated population. The phage-infected P. aeruginosa AA43 behaved similarly to the PAO1 population with 51% of clones with significantly lowered multiplication efficiency. In contrast, the phages selected only 26%, 20%, and 29% of clones exhibiting reduced growth from P. aeruginosa A5803, CHA, and PAK populations, respectively. In most cases, these changes were associated with the occurrence of large genomic deletion, the persistence of phage DNA in the genome, or the combination of both. The correlation analysis of the phage-typing profile versus the reduction of growth rate (Table S2, S3) revealed statistically significant differences (p > 0.001) for the clones resistant to 6–7 phages but only in the PAO1 group. Moreover, only the selection done by phage cocktails gave a statistically significant reduction of bacterial growth (p > 0.001), while no differences were observed regarding groups treated with single LPS- or T4P- dependent phages. In contrast to the PAO1 reference strain, the statistical analyses conducted in the A5803, AA43, CHA, and PAK strains did not show any differences in terms of phage-typing profile nor phage-type selection pressure versus the population fitness reduction (growth rate).
Dots represent the growth of particular clones: the wild-type and control clones (green dots); mutants selected by LPS-dependent phage (red dots); mutants selected by T4P-dependent phage (blue dots); mutants selected by LPS/T4P-dependent PA5oct phage (orange dots); mutants selected by phage cocktail (black dots). * statistically different cumulative OD value compared to phage-untreated pool (p < 0.001).
Does prophage content in the genome affect the long-term persistence of lytic phages within the bacterial population?
This final question focused on the content of prophages versus the lytic phage maintenance in the infected population (carrier state). It should be emphasised that the carrier state was considered only when a full contig of phage DNA was present in the population phenotypically resistant to the infection by this particular phage. We were interested to assess differences in lytic phage DNA maintenance in P. aeruginosa strains versus the number and variability of temperate phages incorporated in the bacterial genome. The whole-genome analyses revealed that the only sequences of phage origin common to all strains encode pyocins with the similarity to phage phiCTX in A5803 and CHA strains while to phage YMC11/02/R656 in the remaining strains (Table S5). All strains, except AA43, contained a complete or incomplete genome of the filamentous phage Pf1, and only A5803 and CHA genomes contained intact dsDNA prophages. Two A5803 intact prophages showed very high sequence similarity to siphoviruses (A5803#1 → phage JBD25; A5803#4 → phage phi297), whereas the third and fourth were homologous to myoviruses (A5803#3 → Escherichia phage vB_EcoM_ep3; A5803#5 → phiCTX phage). The CHA#1 prophage was homologous to JBD18 siphovirus, and the CHA#2 prophage to phiCTX myovirus. The side-by-side comparisons of dsDNA prophage sequences revealed a very low similarity (73% at most for A5803#5 and CHA#2 prophages). In strain AA43, no phage genetic elements were found by PHASTER or manual curation (Table S5).
The genomic analysis of de novo assembled mutants showed that 69/70 isolates of A5803 and CHA strains did not maintain the DNA of lytic phages used for population treatment (no carrier state) (Table 3). On the other hand, this phenomenon was common in clones selected from the intact prophage-free P. aeruginosa strains. In the PAK mutants, the carrier state appeared in 8/35 clones, AA43 mutants contained phage DNA in 11/35 cases, and PAO1 mutants in 75/125 clones. Based on the frequency of individual lytic phage DNA detection in mutants, especially in those originating from PAO1, jumbo phages (PA5oct, phiKZ, and KTN4) appeared to be much more efficient in the long-term survival/maintenance (71/75 cases) in the bacterial population, compared to smaller phages.
Source: Ecology - nature.com
