OXA β-lactamases from Acinetobacter spp. are membrane bound and secreted into outer membrane vesicles

Abstract

β-lactamases from Gram-negative bacteria are generally regarded as soluble, periplasmic enzymes. NDMs have been exceptionally characterized as lipoproteins anchored to the outer membrane. A bioinformatics study on all sequenced β-lactamases was performed that revealed a predominance of putative lipidated enzymes in the Class D OXAs. Namely, 60% of the OXA Class D enzymes contain a lipobox sequence in their signal peptide, that is expected to trigger lipidation and membrane anchoring. This contrasts with β-lactamases from other classes, which are predicted to be mostly soluble proteins. Almost all (>99%) putative lipidated OXAs are present in Acinetobacter spp. Importantly, we further demonstrate that OXA-23 and OXA-24/40 are lipidated, membrane-bound proteins in Acinetobacter baumannii. In contrast, OXA-48 (commonly produced by Enterobacterales) lacks a lipobox and is a soluble protein. Outer membrane vesicles (OMVs) from A. baumannii cells expressing OXA-23 and OXA-24/40 contain these enzymes in their active form. Moreover, OXA-loaded OMVs were able to protect A. baumannii, Escherichia coli, and Pseudomonas aeruginosa cells susceptible to piperacillin and imipenem. These results permit us to conclude that membrane binding is a bacterial host-specific phenomenon in OXA enzymes. These findings reveal that membrane-bound β-lactamases are more common than expected and support the hypothesis that OMVs loaded with lipidated β-lactamases are vehicles for antimicrobial resistance and its dissemination. This advantage could be crucial in polymicrobial infections, in which Acinetobacter spp. are usually involved, and underscore the relevance of identifying the cellular localization of lactamases to better understand their physiology and target them.IMPORTANCEβ-lactamases represent the main mechanism of antimicrobial resistance in Gram-negative pathogens. Their catalytic function (cleaving β-lactam antibiotics) occurs in the bacterial periplasm, where they are commonly reported as soluble proteins. A bioinformatic analysis reveals a significant number of putative lipidated β-lactamases, expected to be attached to the outer bacterial membrane. Notably, 60% of Class D OXA β-lactamases (all from Acinetobacter spp.) are predicted as membrane-anchored proteins. We demonstrate that two clinically relevant carbapenemases, OXA-23 and OXA-24/40, are membrane-bound proteins in A. baumannii. This cellular localization favors the secretion of these enzymes into outer membrane vesicles that transport them outside the boundaries of the cell. β-lactamase-loaded vesicles can protect populations of antibiotic-susceptible bacteria, enabling them to thrive in the presence of β-lactam antibiotics. The ubiquity of this phenomenon suggests that it may have influenced the dissemination of resistance mediated by Acinetobacter spp., particularly in polymicrobial infections, being a potent evolutionary advantage.

Keywords: Acinetobacter spp.; OXA β-lactamases; dissemination of antimicrobial resistance; lipidated β-lactamases; outer membrane vesicles.

Fig 1

Class D contains the largest family of putative lipidated β-lactamases. (A) Pie chart indicating the number of β-lactamase sequences predicted to be translocated by the Sec and Tat systems. The bar at the right shows the distribution of β-lactamases translocated by the Tat system in each class, detailing if they are putative substrates of SpI (putative soluble proteins) or SpII (putative lipoproteins). (B) Putative substrates of SpI or SpII translocated by the Sec system, separated according to the β-lactamase class. The absolute numbers of each category are indicated inside each bar. The three subclasses of Class B enzymes (B1–B3) are indicated at the right. Class D enzymes show the largest number of predicted lipoproteins.

Fig 2

Residue-wise contact frequency (%) with cardiolipin lipids during the simulations. (A) OXA-23 and OXA-24/40 contact frequency (measured as percentage of simulation time) with cardiolipin, color-coded from white (0% frequency) to red (85.5% frequency). The lipidated cysteines (Cys18 and Cys20) are indicated in the N-terminus of both β-lactamases. Active site residues are shown as spheres. (B) A close-up of the residues (shown as sticks) on the opposite side of the active site (shown as spheres). These positively charged residues also contribute to membrane binding, positioning the active site toward the periplasm and preventing its occlusion.

Fig 3

OXA-23 and OXA-24/40 are membrane-anchored proteins, while OXA-48 is soluble periplasmic. (A) N-terminal sequences of the OXAs object of the experimental analysis. The lipoboxes are underlined and the cysteine target of lipidation is bolded. (B) Cell fractionation of A. baumannii ATCC 17978 expressing OXA-23 and OXA-24/40, and E. coli ATCC 25922 expressing OXA-48. (C) Cell fractionation of A. baumannii expressing the Cys18Ala-OXA-23 (CA_OXA-23) and Cys20Ala-OXA-24/40 (CA_OXA-24/40) variants. (D) Solubilization assays of OXA-23 and OXA-24/40 from A. baumannii total membranes. OXA-ST indicates the bands corresponding to anti-ST antibodies, which match the molecular weight of the enzymes fused to the tag. OmpA indicated the bands revealed by anti-OmpA antibodies which recognize the outer membrane protein OmpA. RNApol indicated the bands revealed using RNA polymerase antibodies.

Fig 4

Membrane-anchored OXAs are incorporated in higher proportions into OMVs than periplasmic soluble OXAs. (A) Anti-ST immunoblotting of whole cells (WCs) and outer membrane vesicles (OMVs) from A. baumannii ATCC 17978 expressing OXA-23 or OXA-24/40 and its soluble variants (CA_OXA-23 or CA_OXA-24/40). (B) Comparison between the percentages (%) of the levels of the soluble variants CA_OXA-23 and CA_OXA-24/40 into OMVs. The plotted values, normalized to the corresponding wild-type OXA (lipidated OXA) levels, were obtained as described in Materials and Methods. Data correspond to three independent experiments (black-filled symbols) and are shown as the mean value. Error bars represent standard deviations (SDs). P values according to the Student’s t test: **P ≤ 0.01, ***P ≤ 0.001. (C) Anti-ST immunoblotting of OMVs from A. baumannii carrying OXA-23 or OXA-24/40 treated with and without Proteinase K and 1% vol/vol Triton X-100.

Fig 5

OMVs loaded with lipidated OXAs provide enhanced protection to β-lactam-susceptible bacteria than OMVs containing soluble variants. MIC values (µg/mL) against (A) imipenem and (B) piperacillin of β-lactam susceptible A. baumannii, E. coli, and P. aeruginosa cells after treatment with OMVs purified from A. baumannii carrying empty vector (EV) or expressing lipidated enzymes: OXA-23 or OXA-24/40 or soluble enzymes: CA_OXA-23 or CA_OXA-24/40. Data correspond to mean values from two independent experiments. Error bars represent standard deviations (SD). The MICs of susceptible bacteria against imipenem or piperacillin (without incubation with OMVs) are shown in Table S3.