Chimeric bacteriocin S5-PmnH engineered by domain swapping efficiently controls Pseudomonas aeruginosa infection in murine keratitis and lung models

Identification of Pseudomonas putative pore-forming bacteriocin sequences

Putative pore-forming bacteriocins from genus Pseudomonas have been retrieved from NCBI by BLAST search using as query pore-forming domain of pyocin S5 (pfam01024). After the analysis of BLAST results, we selected six most divergent putative pore forming bacteriocins from different Pseudomonas species: Pflu095 (P. fluorescens WP_016979095), Pflu373 (WP_014717373 from strain A506), Pflu794 (WP_081041794 from strain ATCC 17400), Pflu618 (WP_034155618 from strain H16), Pput259 (WP_098964259 from P. putida strain FDAARGOS_376) and PmnH from P. synxantha strain BG33R. PmnH is the only one of selected proteins that has been published previously. It has unusual architecture as it harbors two cytotoxic domains, colicin-M like domain and pore-forming domain. So far, only activity of its pore-forming domain have been demonstrated²².

Clustal W amino acid sequence alignment of pore-forming domains of six selected proteins showed 32–49% of identity with pyocin S5 (Fig. 1A). The amino acid sequences of pfam 01024 domains of bacteriocins were subjected to phylogenetic tree analysis along with pore-forming domains of some described pore-forming colicins, klebicins and pyocin S5 (Fig. 1B). Two major phylogenetic groups could be distinguished: pore forming domains of Pflu794, Pflu373 and Pflu095 are most related to pyocin S5 and belong to the group of E1-type proteins, and Pflu618 and Put259 are most related to PmnH and belong to A-type protein group (Fig. 1B).

Construction and plant expression of chimeric pyocins

Pore-forming domains of six selected Pseudomonas putative porins were used for the construction of the chimeric proteins. All chimeric proteins contain identical N-terminal end of first 309 a.a. of pyocin S5 including translocation, FptA binding and CPA binding domains¹⁸. The S5 fragment was fused to the cytotoxic domain of the putative pore-forming bacteriocins.

All six chimeric proteins were successfully expressed in Nicotiana benthamiana transient expression system and purified by two-step chromatography (Fig. 2, Supplementary Text S1).

In vitro activity of chimeric pyocins

A panel of 25 P. aeruginosa strains (from culture collections and clinical isolates, Supplementary Table S1) were subjected to genomic DNA extraction and PCR analysis using pyocin S5, pyocin S5 immunity protein and FptA receptor-coding gene specific primers. All 25 strains tested positive for the presence of fptA. Six strains (PA14, PAO1, HP6, HP7, ATCC 1960 and NCTC 13921) tested positive for amplification of both pyocin S5 and pyocin S5 immunity protein coding genes (Supplementary Fig. S1).

Next, all 25 P. aeruginosa strains were subjected to agar disc-diffusion assay by spotting different amounts of purified pyocin S5 (0.3 μg, 3 μg and 30 μg) on bacteria lawns. Six P. aeruginosa strains were resistant to pyocin S5. Four of these six strains were pyocin S5 and immunity protein-encoding strains—PA14, PAO1, ATCC 1960 and NCTC 13921. Two cystic fibrosis isolates, 12-35708 and 12-29165 were also completely resistant to pyocin S5, despite the absence of S5, or S5 immunity protein coding genes. Surprisingly, pyocin S5 still had week inhibition effect on the lawns of two remaining S5 producer strains, clinical isolates HP6 and HP7. Turbid inhibition zones were detected when 3 μg and 30 μg of S5 were spotted on the lawn of these strains (Table 1, Fig. 3).

As the next step we analyzed the activity of S5 chimeras in agar disc diffusion assay. Two chimeric proteins, S5-Pflu095 and S5-PmnH demonstrated broadened activity spectrum in comparison to pyocin S5. Both chimeras formed inhibition zones on the lawns of all six S5-producing strains. The inhibition zones on HP6 and HP7 lawns were significantly larger and clearer than those formed by S5. These chimeric proteins also demonstrated activity similar to S5, with only small variations, on all other remaining P. aeruginosa strains (Table 1, Fig. 3). Between the two proteins, S5-PmnH demonstrated slightly superior activity and was selected for further experiments.

Chimeric bacteriocin S5-Pflu373 also demonstrated good activity profile, similar to S5-Pflu095, with exception that it was not active on one of S5 producers, ATCC 19660. S5-Pflu628 and S5-Pput259 chimeras performed less well and had weaker and less broad activity in comparison to pyocin S5, while S5-Pflu794 demonstrated only week activity on 6 tested strains (Table 1, Fig. 3). None of the chimeric proteins inhibited growth of two cystic fibrosis isolates, 12-35708 and 12-29165.

Efficacy of S5-PmnH in P. aeruginosa keratitis models of disease

Two types of P. aeruginosa strains can be isolated from keratitis cases, cytotoxic P. aeruginosa strains are mainly causing keratitis related to contact-lens wear, while invasive strains are mostly causing disease in post-surgical complications²³. We aimed to investigate if both type of strains could be targeted by S5-PmnH in disease models. Cytotoxic and invasive strains can be distinguished by genotyping the effector protein coding genes of their type III secretion systems (TTSS). Invasive strains were found to possess both exoS and exoT genes whereas cytotoxic strains appeared to have lost exoS but presented exoT and exoU genes^24,25,26. We selected for our experiments cytotoxic strain ATCC 19660 (exoT, exoU) and invasive strain PAO1 (ExoY, exoT, exoS)²⁷. Both strains are pyocin S5 producers and immune to pyocin S5, but both are sensitive to S5-PmnH.

S5-PmnH treatment can reduce P. aeruginosa bacterial numbers in porcine corneas ex vivo

We first investigated the possibility to use S5-PmnH for eradication of P. aeruginosa colonizing the cornea in an ex vivo model, the dissected porcine corneas. Porcine corneas were colonized with invasive strain PAO1 or cytotoxic strain ATCC 19660. S5-PmnH MIC determined by agar dilution method against PAO1 was 4 µg/ml (0.07 nmol/ml) and against ATCC 19660 was 32 µg/ml (0.57 nmol/ml).

In order to obtain P. aeruginosa colonization, porcine corneas were incubated with 3 × 10⁴ CFU of P. aeruginosa ATCC 19660 or 0.4 × 10⁴ CFU of P. aeruginosa PAO1 for 16 to 20 h. Then 5 µg of S5-PmnH were applied to the cornea and incubated for additional 16–20 h. At the end of the experiment PAO1 burden in untreated corneas reached an average of 7.6 log₁₀ CFU/cornea, while S5-PmnH-treated corneas the burden was only an average of 10 CFU per cornea, demonstrating 6.6 log₁₀ reduction. In ATCC 19660 colonized corneas S5-PmnH treatment reduced CFU number by 5.3 log₁₀ (Fig. 4). Thus, in ex vivo porcine corneas, S5-PmnH efficiently reduced P. aeruginosa colonization by both strains.

S5-PmnH efficiently kills bacteria and prevents acute disease in murine keratitis models

Infection by cytotoxic strain ATCC 19660

For induction of keratitis the mice were anaesthetized, the cornea of left eye was scratched with a sterile needle and P. aeruginosa ATCC 19660 (4 × 10⁶ CFU) was applied to the corneal surface. The treatment by S5-PmnH, tobramycin or PBS (mock-treatment) was started 30 min post infection or 6 h post infection. When the treatment was started 30 min after the infection, no viable P. aeruginosa were isolated from infected eyes in both S5-PmnH and tobramycin-treated groups on 1, 3 or 5 dpi. By contrast, the bacterial burden in infected and untreated eyes reached 6–7 log₁₀ CFU/cornea (Fig. 5a, left panel).

The visual inspection by microscope of infected untreated eyes revealed acute disease signs: slight to dense opacity of cornea at 1 dpi, and dense opacity and sometimes cornea perforation at 3 dpi. At 5 dpi all the mice had cornea perforations. No signs of disease were observed in majority of samples treated by S5-PmnH or tobramycin (Fig. 5a, right panel, Supplementary Fig. S2). The histopathology examination of infected and mock-treated eyes revealed strong corneal inflammation at 1 dpi, and disrupted structure of the eye at 3 dpi and 5 dpi. Only weak oedema in the corneal stroma was observed in all infected eyes treated with PyoS5-PmnH and with tobramycin (Fig. 5c). Thus, both S5-PmnH and tobramycin completely eradicated P. aeruginosa ATCC 19660 and prevented the disease when the treatment was started almost immediately, 30 min, after infection.

We repeated the experiment with delayed treatment time, allowing the infection to establish for 6 h. Similar to the previous experiment, the average CFU burden in infected mock-treated eyes reached 6.1 to 6.6 log₁₀ CFU/cornea. In the tobramycin-treated group of mice, no viable bacteria were isolated on 1, 3 or 5 dpi. In the S5-PmnH-treated group viable bacteria were isolated from one mouse at 1 dpi (3.6 log₁₀ CFU/cornea) and from one mouse at 3 dpi (5 log₁₀ CFU/cornea). No viable bacteria were isolated from neither of three mice at 5 dpi (Fig. 5b, left panel). The clinical score evaluation of infected and mock-treated eyes revealed very similar picture to the previous experiment: the opacity of cornea started at 1 dpi and 3 out of 4 mice had cornea perforations at 5 dpi. Only mild disease signs were observed in two S5-PmnH-treated and one tobramycin-treated mice at 1 and 3 dpi and all corneas were completely clear at 5 dpi (Fig. 5b, right panel, Supplementary Fig. S3). The histopathology examination of mock-treated eyes revealed the thinning of corneal epithelium, thickening of stroma, acute inflammation at 1 dpi and acute suppurated inflammation of cornea at 3 dpi and 5 dpi. PyoS5-PmnH-treated infected eyes at 1 dpi presented signs of acute inflammation of cornea and no marked aberrations at 3 and 5 dpi, just weak oedema of corneal stroma was observed. Tobramycin-treated infected eyes at 3 and 5 dpi showed slight thickening of epithelium (Fig. 5c, right panel). Thus, here again S5-PmnH treatment was efficient in eradicating P. aeruginosa from infected corneas and in preventing the establishment and progress of disease.

Infection by invasive strain PAO1

Next, we examined the efficacy of S5-PmnH for treatment of cornea infection by invasive P. aeruginosa PAO1. The infection and treatment procedures were similar to the previous experiment and treatment was started 6 h post infection. The control group of infected mice suffered from severe disease and were euthanized at 3 dpi. At 1 dpi, bacterial burden in mock-treated group of mice reached 6.26 log₁₀ CFU/cornea. The S5-PmnH and tobramycin treatment reduced burden by average 1.04 log₁₀ CFU/cornea and 1.42 log₁₀ CFU/cornea, respectively. At 3 dpi no viable bacteria were isolated from all three tobramycin-treated mice, and from two S5-PmnH-treated mice. The cornea from the third S5-PmnH-treated mouse contained 2.38 log₁₀ CFU of viable bacteria. By contrast, burden in the control group of mice reached 6.73 log₁₀ CFU/cornea. At 5 dpi, no bacteria were isolated from two tobramycin-treated and one S5-PmnH-treated mice. One mouse from the tobramycin treated group contained 2.8 log₁₀ CFU/cornea. The two remaining mice from the S5-PmnH treated group contained 5.14 log₁₀ and 3.34 log₁₀ CFU/cornea (Fig. 6a, left panel).

Clinical examination revealed stronger disease symptoms compared to the cytotoxic strain ATCC 19660. Mock-treated eyes presented severe disease signs at 1 dpi (clinical score 2–3). Several S5-PmnH-treated and tobramycin-treated eyes presented mild clinical scores (1–2) starting from 1 dpi to the end of experiment (Fig. 6a, right panel, Supplementary Fig. S3).

Histological examination of uninfected eyes revealed no marked aberrations, although weak edema in corneal stroma was observed in most samples. Infected S5-PmnH-treated eyes presented signs of acute inflammation of cornea at 1 dpi and 3 dpi. Infected tobramycin-treated eyes presented signs of acute inflammation of cornea at 1 dpi, thinning of cornea epithelium and strong edema of the corneal stroma at 3 dpi and local inflammation of cornea, thinning and degeneration of cornea epithelium at day 5 (Fig. 6B).

In conclusion, S5-PmnH efficiently reduced bacterial burden and prevented acute disease regardless of whether a cytotoxic or invasive strain was used for infection. However, despite lower S5-PmnH MIC against P. aeruginosa PAO1 than against P. aeruginosa ATCC 19660, the chimeric pyocin more efficiently eradicated the cytotoxic strain of P. aeruginosa and prevented the establishment of disease; a similar effect was observed for tobramycin.

S5-PmnH efficiently eradicates lung colonization by P. aeruginosa in a murine model of disease

Mice were infected intranasally (IN) with P. aeruginosa ATCC 27853 strain. One hour later, 2.5, 25 and 250 µg/mouse of S5-PmnH were administered IN once to both nares of the mouse. 5 h later mice were euthanized and lung burden of P. aeruginosa ATCC 27853 was evaluated.

5 h post infection, P. aeruginosa ATCC 27853 burden in the mock-treated mice reached 1.24 × 10⁷ CFU/g of lung tissue, corresponding to an increase of 1.53 log₁₀ CFU/g from 1 h post infection. S5-PmnH administered IN at 2.5 µg/mouse reduced lung burden by 2.1 log₁₀ CFU/g, administered at 25 µg/mouse by 2.31 log₁₀ CFU/g and administered at 250 µg/mouse by 2.66 log₁₀ CFU/g. The bacterial burden was reduced to below the level of stasis (pre-treatment) in all S5-PmnH treatment groups: in 2.5 µg/mouse group by 0.58 log₁₀ CFU/g, in 25 µg/mouse group by 0.78 log₁₀ CFU/g and in 250 µg/mouse group by 1.13 log₁₀ CFU/g. Increased reduction in burden was observed with higher dose levels of S5-PmnH, however the differences were not statistically significant. Tobramycin administered IN once at 200 µg/mouse reduced bacterial burden by 2.75 log₁₀ CFU/g compared to vehicle, corresponding to 1.23 log₁₀ CFU/g below the level of stasis. Higher variability was observed in this group compared to S5-PmnH or vehicle treatments (Fig. 7).