Identification of Potential Antimicrobial Targets of Pseudomonas aeruginosa Biofilms through a Novel Screening Approach

Since the 1950s, antimicrobial drugs have been mainly developed in screens using planktonic bacteria (1). However, more than 80% of human chronic infections are associated with biofilms (2). Biofilms are communities of microorganisms embedded in a self-produced matrix of extracellular polymeric substances adhering to each other and/or to a surface (2). Sessile bacteria in biofilms are protected from immune system defenses and can tolerate up to 1,000 times higher antibiotic concentrations than planktonic cells, requiring doses that cannot be administered in humans (3). New therapeutic options are urgently needed to combat biofilm-related infections, especially those caused by the multidrug-resistant pathogen Pseudomonas aeruginosa (4, 5). We hypothesized that studying biofilm-relevant genes would advance the mechanistic understanding of P. aeruginosa biofilm tolerance to antibiotics and highlight new targets for antibiofilm drug development. This approach differed from other global screenings by rationally reducing the number of analyzed genes, allowing a more extensive characterization of their roles in biofilms.

Biofilm-relevant genes were selected from a transcriptomic study by Whiteley et al., in which 73 genes were upregulated or downregulated by more than 2-fold in P. aeruginosa biofilms compared with planktonic cells (6). Of the 73 genes, 42 were functionally characterized using a P. aeruginosa MPAO1 transposon mutant library (7) and our in vitro assay system (8, 9) (see Table S1 in the supplemental material for additional information on the selected transposon mutants). The antibiotic tolerance of P. aeruginosa biofilms was tested using gentamicin, an aminoglycoside commonly used to treat P. aeruginosa infections (10) (Fig. 1 and Table 1), and the last-resort antibiotic colistin (Fig. S1; see also Fig. S2 to S4 for detailed results). To account for the potential influence of the transposon Tn5 background (9), MPAO1 mutants missing the fiuA and arnB genes, which encode a receptor for heterologous siderophores (11) and colistin resistance (12), respectively, were selected as control reference strains. As previously shown (9, 13), the inactivation of fiuA or arnB in P. aeruginosa MPAO1 did not impact biofilm formation and tolerance toward gentamicin and led to phenotypes representative of most analyzed mutants (Fig. 1).

The screening identified several genes that promoted biofilm formation and tolerance to gentamicin once inactivated. Mutations in the genes encoding the rod shape-determining protein MreC (14), the response regulator PprB (15), and the cytochrome c oxidase subunit CoxC (16) increased the P. aeruginosa biofilm biomass by approximately 4-fold compared with that of the reference mutant (Fig. 1A). These three mutants also showed high biofilm recovery after treatment with 100 μg/mL gentamicin (Fig. 1B). Further characterizations revealed similar minimal bactericidal concentrations of biofilms (MBC-B) of gentamicin for fiuA and pprB mutants (Table 1) but a significantly higher recovery of pprB mutants after exposure to a sub-MBC-B of gentamicin (Fig. S2 and S3). The latter observation concurred with previous studies which showed that pprB overexpression increased membrane permeability and aminoglycoside susceptibility (15). The coxC mutant exhibited a lower MBC-B of gentamicin than the fiuA mutant (Table 1), a similar MBC of planktonic cells (MBC-P), and higher biofilm recovery after exposure to a sub-MBC-B of gentamicin. Inactivating the genes encoding the aa₃-type cytochrome c oxidase (i.e., coxB and coxA) did not influence biofilm formation or tolerance to gentamicin (Fig. 1). Extensive research is needed to decipher the precise roles of mreC, pprB, and coxC in antibiotic resistance and biofilm formation. However, our results suggest that the decreased pprB and coxC expression levels in P. aeruginosa biofilms (6) represent an active mechanism of tolerance against gentamicin.

Our screening results revealed that the inactivation of the genes encoding the hypothetical protein PA3785, the bacteriophage protein PA0720, and the flagellin FliC significantly altered biofilm tolerance to gentamicin independently to biofilm biomass and growth rate (Fig. 1, Fig. S4, and Table 1).

The conserved hypothetical protein encoded by the PA3785 gene appeared to be important for biofilm tolerance to gentamicin (Fig. 1B). Despite the unchanged MBC-B value, PA3785 mutant biofilms exhibited the highest recovery among all tested mutants after exposure to a sub-MBC-B of gentamicin (Fig. 1B and Fig. S2). The PA3785 gene was downregulated in P. aeruginosa biofilms compared with planktonic cells and upregulated 5-fold higher in tobramycin-treated biofilms than in untreated biofilms (6). Its exact function remains unknown, and its role in P. aeruginosa remains to be elucidated through further research.

Filamentous Pf1-like bacteriophages (Pf phages) play major roles in biofilm physiology and antibiotic tolerance (17, 18) and correlate with increased antibiotic resistance in P. aeruginosa isolates from patients with cystic fibrosis (CF) (19). Encoding a single-stranded DNA binding protein, PA0720 is part of the Pf phage operon integrated in the P. aeruginosa genome (20). Our study suggested that PA0720 confers biofilm-mediated tolerance of P. aeruginosa MPAO1 to gentamicin. Inactivating PA0720 did not impact the planktonic resistance toward gentamicin but led to a 4-fold decrease in the MBC-B (Table 1). Gentamicin tolerance was only reduced by inactivating PA0720 but not PA0728, which is essential to produce Pf phages (21), or any other Pf phage genes (Fig. 1B). Therefore, these results highlighted the potential role of PA0720 in P. aeruginosa physiology, besides its role in Pf phage production. It is especially interesting that several transcriptomic and proteomic studies have found PA0720 to be one of the few genes systematically upregulated in P. aeruginosa biofilms (6, 22–24). In summary, PA0720 represents a promising target for drug development and has potential value as a clinically relevant marker for prediction of P. aeruginosa biofilm tolerance to gentamicin.

Our screening further revealed that the inactivation of fliC, which encodes flagellin type B, decreased the P. aeruginosa biofilm tolerance to gentamicin (Fig. 1B and Fig. S3). The MBC-B of gentamicin was 4-fold lower for the fliC mutant (400 μg/mL) than for the fiuA mutant (1,600 μg/mL), while the MBC-Ps were similar (Table 1). The biofilm tolerance to colistin was not affected by the inactivation of fliC (Fig. S1), which suggested a tolerance mechanism specific to gentamicin. In contrast to other motility gene mutants, fliC inactivation did not reduce the biofilm biomass of P. aeruginosa (Fig. 1A), in agreement with the findings of a previous study which showed that the nonmotile fliC mutant produced higher biofilm biomass, owing to an increased ability to adhere on abiotic surfaces compared to the wild type (25). fliC is downregulated in biofilms (6) and chronic CF infections (26), which has been attributed to an adaptive response to avoid phagocytic recognition and clearance (26). We hypothesize that fliC repression contributes to the biofilm-specific tolerance to antibiotics. However, further work is required to understand the precise role of FliC in biofilm physiology and assess its potential value for developing antibiofilm strategies.

The potential link between the antimicrobial resistance (AMR) phenotype and biofilm production is controversial (27, 28). Our results revealed that no clear correlation exists. Some mutants showed antibiotic tolerance relating to higher biofilm production, whereas others did not follow this trend (Fig. 1C). Concurring with a previous study (25), our screening showed that, compared with the reference mutant, the mutations in the motility genes (fliD, cupA1, cupA2, and pilA) led to significantly less biofilm biomass (Fig. 1A) but did not alter biofilm tolerance to gentamicin (Fig. 1B). Moreover, mutation of the gene encoding the sigma factor RpoS (29) increased the biofilm biomass by 250% (Fig. 1A) but did not increase tolerance to gentamicin (Fig. 1B) and displayed high sensitivity to colistin compared with the reference mutant (Fig. S1). These results suggested that biofilm biomass alone is not a good indicator for the AMR phenotype.

This study developed a novel screening method for biofilm research and identified candidate genes involved in biofilm antibiotic tolerance, thereby improving the understanding of biofilm-related infections and identifying relevant therapeutic targets. The screening results suggested that the level of biofilm biomass or planktonic cell resistance of a given strain is not a strong indicator of antibiotic failure. The inactivation of the Pf phage PA0720 and flagellin fliC significantly reduced the gentamicin tolerance of P. aeruginosa biofilms, without impacting the biofilm biomass or MBC-P. This study discovered that novel factors such as pprB, coxC, and PA3785 are involved in the gentamicin tolerance of P. aeruginosa biofilms. Thus, we have highlighted promising targets to develop antibiofilm treatments and relevant markers to predict gentamicin failure in the treatment of biofilm infections. The transposon mutant phenotypes remain to be confirmed with knockout strains. The present study highlights potential leads for future research in biofilm physiology and antibiofilm treatment.

This work was financially supported by the Joint Programming Initiative on Antimicrobial Resistance (JPIAMR), the Swiss National Science Foundation grant numbers 40AR40_173611 (to Q.R.) and 156320 and 197391 (to C.H.A.), German Federal Ministry of Education and Research (BMBF, grant #01KI1710), the Netherlands ZonMW grant 547001003, and the UK Medical Research Council (MRC MR/R005621/1).

We declare no competing interest.

J.D.P.V., H.C.v.d.M., and Q.R. designed the project experiments. J.D.P.V. and S.A. performed experiments. J.D.P.V., Q.R., and H.C.v.d.M. interpreted the data with contributions from A.R.V., C.H.A., F.S., and J.S.W. The manuscript was written by J.D.P.V. and Q.R. with scientific revision from A.R.V., C.H.A., F.S., J.S.W., and H.C.v.d.M.