The function of species-specific cysteine-containing β-lactamases from cystic-fibrosis-associated pathogens depends on DsbA-mediated oxidative protein folding.

β-lactam MIC values for E. coli MC1000 expressing diverse disulfide-bond-containing β-lactamases (Ambler classes A, B and D) are substantially reduced in the absence of DsbA (MIC fold changes: >2; fold change of 2 is indicated by the black dotted lines). No changes in MIC values are observed for the aminoglycoside antibiotic gentamicin (white bars) confirming that absence of DsbA does not compromise the general ability of this strain to resist antibiotic stress. Minor changes in MIC values (≤ 2-fold) are observed for strains harboring the empty vector control (pDM1) or those expressing the class A β-lactamases L2-1 and LUT-1, which contain two or more cysteines (Table S1), but no disulfide bonds (top row). Graphs show MIC fold changes for β-lactamase-expressing E. coli MC1000 and its dsbA mutant from three biological experiments each conducted as a single technical repeat; the MIC values used to generate this figure are presented in File S2A (rows 2-7 and 9-20).

Absence of DsbA results in degradation or misfolding of species-specific cysteine-containing β-lactamases.

(A) The protein levels of most tested disulfide-bond-containing Ambler class A, B, and D β-lactamases are drastically reduced when these enzymes are expressed in E. coli MC1000 dsbA; the amount of the control enzyme L2-1, containing three cysteines but no disulfide bonds, is unaffected. An exception to this is the class A enzyme BPS-1m for which no decrease in abundance is observed in the dsbA mutant (compare lanes 11 and 12). Protein levels of StrepII-tagged β-lactamases were assessed using a Strep-Tactin-AP conjugate. A representative blot from three biological experiments, each conducted as a single technical repeat, is shown; molecular weight markers (M) are on the left, DnaK was used as a loading control and solid black lines indicate where the membrane was cut. Full immunoblots and SDS PAGE analysis of the immunoblot samples for total protein content are shown in File S3. (B) The hydrolysis of the chromogenic β-lactam nitrocefin by cysteine-containing β-lactamases is impaired when these enzymes are expressed in E. coli MC1000 dsbA. The hydrolytic activities of strains harboring the empty vector or expressing the control enzyme L2-1 show no dependence on DsbA. The “Enzyme stability” column informs on the abundance of each enzyme when it is lacking its disulfide bond(s); this was informed from the immunoblotting experiments in panel (A). The “Nitrocefin hydrolysis” column shows the amount of nitrocefin hydrolyzed per mg of bacterial cell pellet in 15 minutes. n=3, table shows means ±SD, significance is indicated by * = p < 0.05, ns = non-significant.

Absence of the principal DsbA analogue (DsbA1) allows treatment of multidrug-resistant Pseudomonas aeruginosa clinical isolates with existing β-lactam antibiotics.

(A) Deletion of dsbA1 in the AIM-1-expressing P. aeruginosa G4R7 clinical isolate sensitizes this strain to ceftazidime and results in reduction of the piperacillin/tazobactam MIC value by 192 μg/mL. (B) Deletion of dsbA1 in the AIM-1-expressing P. aeruginosa G6R7 clinical isolate sensitizes this strain to piperacillin/tazobactam and ceftazidime. (C) 100% of the G. mellonella larvae infected with P. aeruginosa G6R7 (blue curve) or P. aeruginosa G6R7 dsbA1 (light blue curve) die 18 hours post infection, while only 52.5% of larvae infected with P. aeruginosa G6R7 and treated with piperacillin (red curve) survive 28 hours post infection. Treatment of larvae infected with P. aeruginosa G6R7 dsbA1 with piperacillin (pink curve) results in 77.5% survival, 28 hours post infection. The graph shows Kaplan-Meier survival curves of infected G. mellonella larvae after different treatment applications; horizontal lines represent the percentage of larvae surviving after application of each treatment at the indicated time point (a total of 40 larvae were used for each curve). Statistical analysis of this data was performed using a Mantel-Cox test. The most relevant comparison is noted on the figure. Full statistical analysis is as follows: n=40; p=0.3173 (non-significance) (P. aeruginosa vs P. aeruginosa dsbA1), p<0.0001 (significance) (P. aeruginosa vs P. aeruginosa treated with piperacillin), p<0.0001 (significance) (P. aeruginosa dsbA1 vs P. aeruginosa treated with piperacillin), p=0.0147 (significance) (P. aeruginosa treated with piperacillin vs P. aeruginosa dsbA1 treated with piperacillin). (D) Deletion of dsbA1 in the GES-19/GES-26-expressing P. aeruginosa CDC #769 clinical isolate sensitizes this strain to piperacillin/tazobactam and aztreonam. (E) Deletion of dsbA1 in the GES-19/GES-20-expressing P. aeruginosa CDC #773 clinical isolate sensitizes this strain to piperacillin/tazobactam, aztreonam, and ceftazidime. (F) 100% of G. mellonella larvae infected with P. aeruginosa CDC #773 (blue curve), P. aeruginosa CDC #773 dsbA1 (light blue curve) or larvae infected with P. aeruginosa CDC #773 and treated with ceftazidime (red curve) die 21 hours post infection. Treatment of larvae infected with P. aeruginosa CDC #773 dsbA1 with ceftazidime (pink curve) results in 96.7% survival, 24 hours post infection. The graph shows Kaplan-Meier survival curves of infected G. mellonella larvae after different treatment applications; horizontal lines represent the percentage of larvae surviving after application of each treatment at the indicated time point (a total of 30 larvae were used for each curve). Statistical analysis of this data was performed using a Mantel-Cox test. The most relevant comparison is noted on the figure. Full statistical analysis is as follows: n=30; p<0.0001 (significance) (P. aeruginosa vs P. aeruginosa dsbA1), p>0.9999 (non-significance) (P. aeruginosa vs P. aeruginosa treated with ceftazidime), p<0.0001 (significance) (P. aeruginosa dsbA1 vs P. aeruginosa treated with ceftazidime), p<0.0001 (significance) (P. aeruginosa treated with ceftazidime vs P. aeruginosa dsbA1 treated with ceftazidime). For panels (A), (B), (D), and (E) the graphs show MIC values (μg/mL) from three biological experiments, each conducted as a single technical repeat; red dotted lines indicate the EUCAST clinical breakpoint for each antibiotic.

(A-D) Impairment of disulfide bond formation allows the treatment of Stenotrophomonas maltophilia clinical strains with β-lactam and colistin antibiotics. (A, B) Deletion of dsbA dsbL in the S. maltophilia AMM and S. maltophilia GUE clinical isolates results in drastic decrease of their ceftazidime MIC values. (C) Deletion of dsbA dsbL in the S. maltophilia AMM clinical strain results in drastic decrease of its colistin MIC value. (D) Use of a small-molecule inhibitor of DsbB against the S. maltophilia AMM clinical strain results in decrease of its ceftazidime and colistin MIC values. For panels (A-D) graphs show MIC values (μg/mL) from three biological experiments; for β-lactam MIC assays each experiment was conducted as a single technical repeat, whereas for colistin MIC assays each experiment was conducted in technical triplicate. In the absence of EUCAST clinical breakpoints for S. maltophilia, the black dotted lines indicate the EUCAST clinical breakpoint for each antibiotic for the related pathogen P. aeruginosa. (E) Protection of P. aeruginosa by S. maltophilia clinical strains is dependent on oxidative protein folding. The susceptible P. aeruginosa strain PA14 can survive exposure to ceftazidime up to a maximum concentration of 4 μg/mL when cultured in isolation (white bars). By contrast, if co-cultured in the presence of S. maltophilia AMM, which can hydrolyze ceftazidime through the action of its L1-1 β-lactamase enzyme, P. aeruginosa PA14 can survive and actively grow in concentrations of ceftazidime as high as 512 μg/mL (dark pink bars). This protection is abolished if P. aeruginosa PA14 is co-cultured with S. maltophilia AMM dsbA dsbL (light pink bars), where L1-1 is inactive (as shown in Fig. 4A and [23]). The graph shows P. aeruginosa PA14 colony forming unit counts (CFUs) for each condition; three biological replicates were conducted in technical triplicate, and mean CFU values are shown. The black dotted line indicates the P. aeruginosa PA14 inoculum. The mean CFU values used to generate this figure are presented in File S2B.

Inhibition of oxidative protein folding counters antibiotic resistance and inter-species interactions in CF-associated pathogens.

(Left) After Sec translocation to the periplasm and DsbA-assisted folding, cysteine-containing species-specific β-lactamase enzymes from recalcitrant pathogens, like P. aeruginosa or S. maltophilia, are active and can hydrolyze β-lactam antibiotics. However, in the absence of their disulfide bonds, DsbA-dependent β-lactamases either degrade or misfold, and thus can no longer confer resistance to β-lactam compounds. (Right) In multispecies bacterial communities, bacteria that degrade antibiotics, for example species producing β-lactamases, can protect antibiotic-susceptible strains. Targeting disulfide bond formation impairs interbacterial interactions that are reliant on the activity of DsbA-dependent β-lactamase enzymes, allowing eradication of both bacterial species.

Complementation of dsbA restores the β-lactam MIC values for E. coli MC1000 dsbA expressing β-lactamase enzymes.

Re-insertion of dsbA at the attTn7 site of the chromosome restores representative β-lactam MIC values for E. coli MC1000 dsbA harboring (A) pDM1-blaBEL-1 (ceftazidime MIC), (B) pDM1-blaCARB-2 (cefuroxime MIC), (C) pDM1-blaAIM-1 (ceftazidime MIC), (D) pDM1-blaOXA-50 (ampicillin MIC), (E) pDM1-blaBPS-1m (ceftazidime MIC), and (F) pDM1-blaBPS-6 (ceftazidime MIC). Graphs show MIC values (µg/mL) and are representative of two biological experiments, each conducted as a single technical repeat.

Assessment of off-target effects for clinical strains of P. aeruginosa and S. maltophilia that are deficient in oxidative protein folding.

(A) P. aeruginosa CDC #769 and its mutant lacking dsbA1 have identical gentamicin MIC values, confirming that absence of DsbA does not compromise the general ability of the strain to resist antibiotic stress. (B) Re-insertion of the dsbA1 gene from P. aeruginosa PAO1 at the attTn7 site of the chromosome restores representative antibiotic MIC values for P. aeruginosa CDC #769 dsbA1 (left, piperacillin/tazobactam MIC; right, aztreonam MIC). (C) S. maltophilia AMM and its mutant lacking dsbA and dsbL have near-identical gentamicin MIC values, confirming that absence of DsbA and DsbL does not compromise the general ability of the strain to resist antibiotic stress. (D) Re-insertion of the dsbA1 gene from P. aeruginosa PAO1 at the attTn7 site of the chromosome restores representative antibiotic MIC values for S. maltophilia AMM dsbA dsbL (left, ceftazidime MIC; right, colistin MIC). (E) Changes in MIC values observed using the DSB system inhibitor (compound 36) are due solely to inhibition of the DSB system. The gentamicin MIC value of S. maltophilia AMM remains unchanged upon addition of the inhibitor (left), and the same is observed for the colistin MIC value of S. maltophilia AMM dsbA dsbL in the presence of the compound (right). This indicates that the chemical compound used in this study only affects the function of the DSB system proteins. For all panels, graphs show MIC values (µg/mL) and are representative of three biological experiments. β-Lactam MICs were conducted as a single technical repeat and colistin MICs were conducted in technical triplicate; red dotted lines indicate the EUCAST clinical breakpoint for each antibiotic, where applicable. In the absence of EUCAST clinical breakpoints for S. maltophilia, the black dotted lines indicate the EUCAST clinical breakpoint for each antibiotic for the related pathogen P. aeruginosa, where applicable.

Protection of P. aeruginosa by S. maltophilia clinical strains is dependent on oxidative protein folding.

(A) Comparison of the colony forming units (CFUs) of S. maltophilia AMM with the CFUs of S. maltophilia AMM dsbA dsbL after six hours of growth, prior to P. aeruginosa PA14 addition. The two S. maltophilia strains display equivalent growth. (B) Complementary analysis to Fig. 4E; here the CFUs of all P. aeruginosa and S. maltophilia strains were enumerated in isolation and in mixed culture conditions for a more limited set of antibiotic concentrations. Equivalent trends to Fig. 4E are observed. The susceptible P. aeruginosa strain PA14 can survive exposure to ceftazidime up to a maximum concentration of 4 μg/mL when cultured in isolation (white bars). By contrast, if co-cultured in the presence of S. maltophilia AMM (dark blue bars), which can hydrolyze ceftazidime through the action of its L1-1 β-lactamase enzyme, P. aeruginosa PA14 (dark pink bars) can survive and actively grow in higher concentrations of ceftazidime (see 128 μg/mL of ceftazidime). This protection is abolished if P. aeruginosa PA14 (light pink bars) is co-cultured with S. maltophilia AMM dsbA dsbL (light blue bars). In this case, L1-1 is inactive (as shown in Fig. 4AB and [1]), resulting in killing of S. maltophilia AMM and, in turn, eradication of P. aeruginosa PA14 (see 128 μg/mL of ceftazidime, absence of light pink bars). Three biological replicates were conducted in technical triplicate and mean CFU values are shown. The grey line indicates the P. aeruginosa PA14 inoculum. The mean CFU values used to generate this figure are presented in File S2C.

The activity of additional species-specific β-lactamases depends on disulfide bond formation.

(A) β-Lactam MIC values for E. coli MC1000 expressing disulfide-bond-containing β-lactamases from P. otitidis (left, POM-1; Table S1) and Serratia spp. (right, SMB-1; Table S1) are reduced in the absence of DsbA (MIC fold changes: >2; fold change of 2 is indicated by the black dotted lines). No changes in MIC values are observed for the aminoglycoside antibiotic gentamicin (white bars) confirming that absence of DsbA does not compromise the general ability of this strain to resist antibiotic stress. Graphs show MIC fold changes for β-lactamase-expressing E. coli MC1000 and its dsbA mutant. MIC assays were performed in three biological experiments each conducted as a single technical repeat; the MIC values used to generate this figure are presented in File S2A (rows 22-25). (B) Protein levels of disulfide-bond-containing β-lactamases are either unaffected (POM-1) or drastically reduced (SMB-1) when these enzymes are expressed in E. coli MC1000 dsbA. Protein levels of StrepII-tagged β-lactamases were assessed using a Strep-Tactin-AP conjugate. A representative blot from three biological experiments, each conducted as a single technical repeat, is shown; molecular weight markers (M) are on the left, DnaK was used as a loading control and solid black lines indicate where the membrane was cut. Full immunoblots and SDS PAGE analysis of the immunoblot samples for total protein content are shown in File S3. (C) The hydrolytic activities of both tested β-lactamases are significantly reduced in the absence of DsbA. The hydrolytic activities of strains harboring the empty vector or expressing the control enzyme L2-1 show no dependence on DsbA; the same data for the control strains are also shown in Fig. 2B. n=3 (each conducted in technical duplicate), table shows means ± SD, significance is indicated by * = p < 0.05, ns = non-significant.

Complementation of dsbA restores the β-lactam MIC values for E. coli MC1000 dsbA expressing β-lactamases.

Re-insertion of dsbA at the attTn7 site of the chromosome restores representative β-lactam MIC values for E. coli MC1000 dsbA harboring (A) pDM1-blaPOM-1 (imipenem MIC), and (B) pDM1-blaSMB-1 (ceftazidime MIC). Graphs show MIC values (µg/mL) and are representative of two biological experiments, each conducted as a single technical repeat.

Overview of the β-lactamase enzymes investigated in this study.

All tested enzymes belong to distinct phylogenetic clusters (see File S1), with the exception of BPS-1m and BPS-6. The “Cysteine positions” column states the positions of cysteine residues after amino acid 30 and hence, does not include amino acids that are part of the periplasmic signal sequence which is cleaved after protein translocation. All β-lactamase enzymes except L2-1 and LUT-1 (shaded in grey), which are used as negative controls throughout this study, have one or more disulfide bonds. Both L2-1 and LUT-1 contain two or more cysteine residues, but lack disulfide bonds as they are transported to the periplasm in a folded state by the Twin-arginine translocation (Tat) system; for L2-1 Tat-dependent translocation has been experimentally confirmed [2], whereas for LUT-1 this is strongly corroborated by signal peptide prediction software (SignalP 5.0 [3] likelihood scores: Sec/SPI = 0.0572, Tat/SPI = 0.9312, Sec/SPII (lipoprotein) = 0.0087, other = 0.0029). The “Mob.” (mobilizable) column refers to the possibility for the β-lactamase gene to be mobilized from the chromosome; “yes” indicates that the gene of interest is located on a mobile element, while “no” refers to immobile chromosomally-encoded enzymes. The “Spectrum” column refers to the hydrolytic spectrum of each tested enzyme; tested enzymes are narrow-spectrum β-lactamases (NS), extended-spectrum β-lactamases (ESBL) or carbapenemases. The “Inh.” (inhibition) column refers to classical inhibitor susceptibility i.e., susceptibility to inhibition by clavulanic acid, tazobactam or sulbactam. Finally, the “Organism” column refers to the bacterial species that most commonly express the tested β-lactamase enzymes.

Bacterial strains used in this study.

All listed isolates are clinical strains. “FNRCAR” refers to the French National Reference Centre for Antibiotic Resistance in Le Kremlin-Bicêtre, France, and “CDC AR Isolate bank” refers to the Centers for Disease Control and Prevention Antibiotic Resistance Isolate Bank in Atlanta, GA, USA.

Plasmids used in this study.

Oligonucleotide primers used in this study.

The “Brief description” column provides basic information on the primer design (restriction enzyme used for cloning, encoded protein or gene replaced by antibiotic resistance cassette, forward or reverse orientation of the primer (F or R); SQ stands for sequencing primers).

Sources of genomic DNA used for amplification of β-lactamase genes used in this study.

CRBIP stands for Centre de Ressources Biologiques de l’Institut Pasteur, France and FNRCAR refers to the French National Reference Centre for Antibiotic Resistance in Le Kremlin-Bicêtre, France.