Directed-complement killing of Pseudomonas aeruginosa protects against lethal pneumonia

Abstract

Background: Multidrug-resistant Pseudomonas aeruginosa raises major clinical concerns due to its capacity to cause a wide-array of infections in individuals with compromised immune defences and to withstand standard-of-care therapeutic treatments. Antibody-based approaches have proven to be efficient in the treatment of diverse infections. Here we propose an innovative approach harnessing the complement at the surface of bacteria for further killing.

Methods: We developed two Complement-activating Multimeric immunotherapeutic compleXes (CoMiX) targeting the bacterium through a single-chain variable fragment directed against the exopolysaccharide Psl, and carrying one of two different effector functions, Factor H Related protein 1 (FHR1) or a Fc dimer. Each CoMiX was assessed in vitro for their antibacterial activity, and further evaluated in a mouse model of acute pneumonia.

Findings: Both CoMiX-FHR1 and CoMiX-Fc effectively deposit C1q (for CoMiX-Fc), C3b, and C5b9 at the surface of multidrug-resistant clinical isolates, promoting their direct killing and/or opsonisation and subsequent phagocytosis for CoMiX-Fc (p < 0.001). Both CoMiX synergise with amikacin and protect epithelial cells against P. aeruginosa-induced cytotoxicity. Importantly, CoMiX administered intranasal to acutely infected mice significantly improve their survival (p < 0.001) by reducing local bacterial burden through the higher induction of C3b (opsonisation) and C5a (neutrophils recruitment and activation) and by decreasing lung inflammation.

Interpretation: Our proof-of-concept demonstrates the efficient, direct and indirect killing of P. aeruginosa by the complement, highlighting the therapeutic potential of CoMiX to combat multidrug-resistant bacteria.

Funding: Luxembourg National Research Fund, Ministry of Higher Education and Research of Luxembourg, COST action CA21145 EURESTOP, Institut National de la Santé et de la Recherche Médicale, and Tours University.

Keywords: Complement system; FHR1; Immunotherapy; Multidrug resistance; Pseudomonas aeruginosa.

Fig. 1

Schematic representation and characterisation of CoMiX-Fc (a) and CoMiX-FHR1 (b) proteins. The dimerisation scaffold derived from the β-chain of C4BP is used to combine the targeting function of the scFv recognising P. aeruginosa, to either an Fc-region (CoMiX-Fc, a) or the CCP3-5 domains of Factor H-related protein 1 (CoMiX-FHR1, b). Both proteins were purified with protein G affinity columns for CoMiX-Fc and a Nickel column binding to the His-tag at the C-terminal end of CoMiX-FHR1. (c) SYPRO Ruby Staining: 3 μg of CoMiX-Fc and CoMiX-FHR1 proteins were loaded onto 4–15% SDS-PAGE gels and stained with SYPRO Ruby. (d) Western Blotting: Proteins, in non-reduced (NR) and reduced (R) conditions, from the SDS-PAGE gel were transferred to a low-fluorescence PVDF membrane and probed with specific fluorochrome-conjugated antibodies: anti-Fc AF647 for CoMiX-Fc and anti-His AF647 for CoMiX-FHR1. Protein bands from the two techniques were detected using an Amersham Typhoon scanner. The predicted molecular weight of CoMiX-Fc is approximately 122 kDa in normal conditions and 61 kDa when reduced, while CoMiX-FHR1 has a smaller size of around 107 kDa in normal conditions, and 53 kDa when reduced.

Fig. 2

Anti-Psl CoMiX bind specifically to P. aeruginosa, but not to other bacterial strains, with CoMiX-FHR1 competing with FH. The binding of anti-P. aeruginosa CoMiX, irrelevant anti-A. fumigatus CoMiX, and irrelevant anti-M. catarrhalis CoMiX to reference strain PAO1 (a), clinical isolate IT 2 (b), and Gram-bacterial strain M. catarrhalis(c) was analysed by whole-cell ELISA. Immobilised bacteria (1 × 10⁷ CFU/mL) or bacterial lysate (20 μg/mL) were incubated with 10 μg/mL of CoMiX and irrelevant controls. Bound CoMiX was detected using specific antibodies: anti-His for CoMiX-FHR1 and anti-Fc for CoMiX-Fc. CoMiX targeting the second bacterial strain and the fungi were used as negative control. Data are presented as the mean values ± SEM. Results correspond to two-three pooled independent experiments (2–3 replicates per experiment). Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test. ∗∗∗∗p < 0.0001. To investigate the competition between FH and CoMiX-FHR1, recombinant FH (20 μg/mL) and increasing concentrations of CoMiX-FHR1 (and CoMiX-Fc as control) were added to PAO1 bacteria. The binding of FH to bacterial cells was detected with the FH-specific monoclonal antibody OX-24 (d). Data are presented as the mean values ± SEM. Results correspond to two-three pooled independent experiments (3 replicates per experiment). Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test: it indicates significant differences for the binding of FH in absence versus presence of CoMiX-FHR1. ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001.

Fig. 3

Anti-Psl CoMiX enhance C1q deposition (for CoMiX-Fc), C3b opsonisation and the formation of membrane attack complex (C5b9/MAC) on P. aeruginosa. Immobilised bacterial cells of the P. aeruginosa reference strain PAO1 and the clinical isolate IT 2 were incubated with 10 μg/mL of CoMiX, before the addition of either 2% (C3b) or 4% (C1q, C5b9/MAC) of normal human serum (NHS) or heat-inactivated human serum (ΔNHS) in GVB++ at 37 °C for 30 min. Complement deposition at the bacterial surface was measured by ELISA, using (a) an anti-human C1q mAb, (b) an anti-human C3/C3b/iC3b mAb, and (c) an anti-human C5b9 mAb, followed by an HRP-conjugated anti-mouse IgG mAb. Data are presented as the mean values ± SEM. Results correspond to three pooled independent experiments. Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test. ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. GFP-expressing PAO1 bacteria were incubated with 10% C5-deficient serum (C3b) or normal human serum (C5b9/MAC) at 37 °C for 1 h, in the presence or absence of 10 μg/mL of CoMiX. Bacteria were then incubated with either a goat anti-human C3b or goat anti-human C5b9 antibody, followed by a secondary AF647-conjugated anti-goat mAb. Once stained, bacteria were mounted onto an agarose gel pad, visualised on a confocal Leica Sp8 microscope (C3b), or on a wide field Axio observer microscope (C5b), and analysed by ImageJ to detect C3 cleavage product (C3b, iC3b, and C3c) deposition (d) or C5b9 deposition (e). Two to four fields have been acquired for each conditions. Representative images of the deposition are presented here. Green = GFP-expressing PAO1, red = C3b or C5b9 deposition by anti-goat AF647. Scale bar = 5 μm; 10 μm.

Fig. 4

CoMiX enhance complement-mediated killing of P. aeruginosa and display synergy with amikacin. (a) Bacterial reference strain PAO1, and clinical isolate IT 2 were incubated without serum (w/o NHS), with 10% normal human serum (NHS) or with 10% decomplemented serum (ΔNHS) in the presence or absence of 30 μg/mL CoMiX or irrelevant control for 2 h at 37 °C. Bacteria were plated and CFU were enumerated to assess bacterial viability. Data are presented as the mean values ± SEM. Results correspond to two pooled independent experiments (2–3 replicates per experiment). Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test. ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. (b) A mutated PAO1 bacterial strain with a luminescence-based reporter system (PAO1-Lux) was used in the same protocol to monitor in real-time bacterial growth. The luminescence of bacteria (RLU) was measured after 2 h of incubation on a POLARStar Omega microplate reader, as it is known that luminescence of bacteria correlates well with its concentration. Data are presented as the mean values ± SEM. Results correspond to two pooled independent experiments (2–3 replicates per experiment). Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test. ∗∗∗∗p < 0.0001. (c) The complement potency of mouse serum (MS) and decomplemented MS (ΔMS) was assessed using the same protocol as in a. Bacteria were plated and CFU were enumerated to assess bacterial viability. Data are presented as the mean values ± SEM. Results correspond to two pooled independent experiments (2–3 replicates per experiment). Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test. ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. (d) Following the same protocol as in a, the membrane permeabilisation of both PAO1 and clinical isolate IT 2 was quantified after 2 h through ethidium homodimer fluorescence, measured on a GloMax® Discover (Ex = 520 nm, Em = 580–640 nm). Data are presented as the mean values ± SEM. Results correspond to two pooled independent experiments (3 replicates per experiment). Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test. ∗∗p < 0.01; ∗∗∗∗p < 0.0001. (e) To assess the potential of CoMiX in combination with the antibiotic amikacin, both PAO1 and clinical isolate IT 2 were incubated with 10% NHS in the presence or absence of 30 μg/mL of CoMiX or irrelevant control, as well as various sub-MIC concentrations of amikacin (2 μg/mL for PAO1, 10 μg/mL for clinical isolate IT 2) for 2 h at 37 °C. Bacteria were plated, and CFU were enumerated to assess bacterial viability. A synergy was admitted when the synergy coefficient (log(C) − log(SA) − log(SB) + log(SAB) < 0 (where C is CFU without treatment, SA is CFU with amikacin only, SB is CFU with CoMiX only, and SAB is CFU with both)), was negative (S). Data are presented as synergy coefficient calculated from the mean of three independent experiments (3 replicates per experiment).

Fig. 5

CoMiX-Fc enhances significantly phagocytosis of P. aeruginosa by PBMCs-derived M1 macrophages and more importantly by Neutrophils-Like Cells, nevertheless, both CoMiX slightly improve NLCs-dependent antimicrobial activity against the bacteria. PMA-activated M1 macrophages, from four different healthy donors, were co-incubated with pHrodo-stained P. aeruginosa (PAO1) at a 12:1 bacteria-to-cell ratio with 10% NHS and in the presence or absence of 15 μg/mL CoMiX-Fc, CoMiX-FHR1 or CoMiX-irrelevant. Engulfment of bacteria was assessed after 30 min (left panels) and 1 h (right panels) by real-time Incucyte® microscope (a–c). The percentage of pHrodo positive cells (a) and the intensity of pHrodo rationalised over the surface of cells and calculated as integrated intensity (b) were recorded. Data are presented as the mean values ± SEM. Results correspond to two independent experiments with macrophages from 4 healthy donors (three technical replicates per donor). Statistical analysis was performed using paired one-way ANOVA (to smooth inter-donor variability), followed by Tukey's post-hoc test: ∗p < 0.05; ∗∗p < 0.01. (c) Representative incucyte images for the phagocytosis induced by serum and CoMiX-Fc over time. (d) Fluorescence microscopy image obtained on a wide field Axio Observer Z1, and treated on ImageJ of phagocytosed P. aeruginosa bacteria by PBMCs-derived M1 macrophages in presence of 10% serum and 15 μg/mL CoMiX-Fc. Red = wheat germ agglutinin Alexa-647, staining carbohydates residues of macrophages membranes; green = CellTrace™ CFSE stained P. aeruginosa bacteria. Scale bar = 25 μm. P. aeruginosa PAO1 strain and PAO1-GFP strain were co-cultured with NLCs at a 10:1 bacteria-to-cell ratio, treated with 2% NHS and CoMiX (15 μg/mL) and incubated for 20 min at 37 °C under agitation (200 rpm). After washes and treatment with gentamicin to eliminate non-phagocytosed bacteria from the cells samples were (1) fixed and read by flow cytometry. (e) Gating of NLCs with FSC and SSC to eliminate cell debris and free bacteria from the analysis (left graph). Representative histogram describing the total population of NLCs, and composed of GFP negative cells and GFP positive cells (right graph). To assess phagocytosis, the percentage of GFP positive cells (f) and the mean fluorescence (g) were acquired. Data are presented as the mean values ± SEM. Results correspond to three pooled experiments (2 replicates per experiment). Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test: ∗∗∗∗p < 0.0001. (h) After washes and treatment with gentamicin, cells were also (2) lysed by Triton X-100, diluted in PBS and plated onto petri dishes. CFUs, corresponding to phagocytosed bacteria were counted in duplicates. Data are presented as the mean values ± SEM. Results correspond to three pooled independent experiments (2 replicates per experiment). Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test: ∗∗∗∗p < 0.0001. To assess NLCs-dependent killing, P. aeruginosa PAO1 strain at a MOI of 2:1 was co-cultured with NLCs, treated with 10% NHS and CoMiX (15 μg/mL) for 1 h, plated on petri dishes to count the final bacterial CFUs in duplicates (i). Conditions which were not co-cultured with NLCs were used as control (100% cell survival). Results are expressed as surviving bacteria compared to bacterial growth under the same conditions in the absence of NLCs. Data are presented as the mean values ± SEM. Results correspond to three pooled independent experiments (2 replicates per experiment). Statistical analysis was performed using one-way ANOVA followed by Tukey's post-hoc test: ∗p < 0.05; ∗∗p < 0.01.

Fig. 6

Prophylactic treatment with anti-Psl CoMiX protects mice from lethal P. aeruginosa lung infection. (a) Eight-week-old female C57BL/6J mice were infected intranasal (i.n.) with a lethal dose (3 × 10⁶ CFU) of luciferase-expressing PAO1 strain. Mice were treated with anti-Psl CoMiX-Fc, anti-Psl CoMiX-FHR1 or an irrelevant CoMiX control 3 h before and after intranasal infection with luciferase-positive P. aeruginosa PAO1. Survival (b) and body weight (c) were assessed daily for six days following infection. Data are presented as the mean values ± SEM. Results correspond to two pooled independent experiments (n = 8–16 mice per group). Statistical analysis was performed using the log-rank test (Mantel–Cox). ∗∗∗∗p < 0.0001. (d, e) Lung infection was assessed by visualising P. aeruginosa-associated luminescence emission in live animals at 24 h and 48 h p.i. One representative mouse image for the time-point 24 h and for each group is displayed. Data are presented as a percentage of mice with negative versus positive signal for bacteria. The data were analysed using a Chi-square test for direct comparisons between anti-Psl CoMiX and the irrelevant control (∗p < 0.05; ∗∗p < 0.01).

Fig. 7

CoMiX protect mice from acute lung infection of P. aeruginosa. (a) Eight-week-old female C57BL/6J mice were infected intranasal (i.n.) with a lethal dose (3 × 10⁶ CFU) of luciferase-expressing PAO1 strain. Mice were treated 1 h later with 100 μg of CoMiX-Fc, CoMiX-FHR1 or an irrelevant CoMiX via i.n. administration. Survival (b) and body weight (c) were monitored for 7 days after infection. Data are presented as the mean values ± SEM. Results correspond to two pooled independent experiments (n = 16 mice per group). Statistical analysis was performed using the log-rank test (Mantel–Cox). ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. (d, e) Lung infection was assessed by visualising P. aeruginosa-associated luminescence emission in live animals at 24 h and 48 h p.i. One representative mouse image for the time-point 24 h and for each group is displayed. Data are presented as a percentage of mice with negative versus positive signal for bacteria. The data were analysed using a Chi-square test for direct comparisons between anti-Psl CoMiX and the irrelevant control (∗∗p < 0.01; ∗∗∗p < 0.001).

Fig. 8

Therapeutic administration of CoMiX in vivo results in enhanced bacterial clearance through local and systemic complement activation. Mice were infected/treated as described in Fig. 7a, sacrificed at 4 h, 8 h and 16 h p.i., and lungs, BAL and blood were collected for analysis. (a) Lungs were first perfused for a visual assessment of inflammation. Bacterial load in BAL (b) and lungs (c) were determined after serial dilution and plating on Petri dishes. Total protein was measured by BCA in the BAL (d). To assess the activation of the complement cascade, the concentration of activated fragments of the mouse complement protein C3 was determined by ELISA, systemically in the serum (e) and locally in the BAL (f). Concentration of local C3 activated fragments was correlated with the BAL bacterial load (g). The concentration of mouse complement protein C5a was determined by ELISA locally in the BAL (h) and the lungs (i). All data are shown as individual values and quoted as the mean values ± SEM and the results correspond to one experiment per time-point (n = 7–8 mice per group). Unless otherwise stated, all statistical analyses were performed using Kruskal–Wallis test followed by a Dunn's post-test for comparisons between the groups, ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001. Correlation between experimental variable were statistically analysed, and interpreted in regards to their Pearson correlation coefficient (ρ) and p-value.

Fig. 9

CoMiX in vivo protection is associated with transient neutrophil upregulation and improved control of lung inflammation. Mice were infected/treated as described in Fig. 7a, sacrificed at 4 h, 8 h and 16 h p.i., and lungs and BAL were collected for analysis. As a reflection of lung inflammation, immune cell populations were assessed by flow cytometry. The absolute number of (a) leukocytes (CD45⁺ cells), (b) alveolar macrophages (CD45⁺ SiglecF⁺ CD11b⁻), (c) eosinophils (CD45⁺ SiglecF⁺ CD11c⁺), and (d) neutrophils (CD45⁺ SiglecF⁻ Ly6G⁺ CD11b⁺) were quantified in the BAL. (e) CD11b expression on neutrophils was measured to assess their activation. (f) Concentration of local anaphylatoxin C5a was correlated with neutrophil number in the BAL at 8 h p.i. The production of the cytokines and chemokines (g) IL-6, (h) TNF-α, and (i) CXCL-1 were quantified in the BAL by a MSD assay. All data are shown as individual values and quoted as the mean values ± SEM and the results correspond to one experiment per time-point (n = 7–8 mice per group). Unless otherwise stated, all statistical analyses were performed using Kruskal–Wallis test followed by a Dunn's post-test for comparisons between the groups, ∗p < 0.05; ∗∗p < 0.01; ∗∗∗p < 0.001; ∗∗∗∗p < 0.0001. IL = interleukin; TNF = tumour necrosis factor. Correlation between experimental variable were statistically analysed, and interpreted in regards to their Pearson correlation coefficient (ρ) and p-value.