A periplasmic phospholipase that maintains outer membrane lipid asymmetry in Pseudomonas aeruginosa

Significance

Gram-negative bacteria are surrounded by an outer membrane consisting of lipopolysaccharide in the outer leaflet and glycerophospholipids in the inner leaflet. This unique architecture prevents many clinically relevant antibiotics from entering the cell. Here, we show that the MlaA-like protein PA3239 (renamed MlaZ) and the putative lipase PA3238 (renamed MlaY) in the highly antibiotic-resistant pathogen Pseudomonas aeruginosa remove and degrade glycerophospholipids that have mislocalized to the outer leaflet of the outer membrane. The MlaYZ system plays an analogous role to the outer membrane phospholipase PldA in the model Gram-negative Escherichia coli. Convergent evolution of systems that degrade outer leaflet glycerophospholipids highlights the importance of this process in Gram-negative physiology.

Abstract

The outer membrane of Gram-negative bacteria is unique in both structure and function. The surface-exposed outer leaflet is composed of lipopolysaccharide, while the inner leaflet is composed of glycerophospholipids. This lipid asymmetry creates mechanical strength, lowers membrane permeability, and is necessary for virulence in many pathogens. Glycerophospholipids that mislocalize to the outer leaflet are removed by the Mla pathway, which consists of the outer membrane channel MlaA, the periplasmic lipid carrier MlaC, and the inner membrane transporter MlaBDEF. The opportunistic pathogen Pseudomonas aeruginosa has two proteins of the MlaA family: PA2800 and PA3239. Here, we show that PA2800 is part of a canonical Mla pathway, while PA3239 functions with the putative lipase PA3238. While loss of either pathway individually has little to no effect on outer membrane integrity, loss of both pathways weakens the outer membrane permeability barrier and increases production of the secondary metabolite pyocyanin. We propose that mislocalized glycerophospholipids are removed from the outer leaflet by PA3239 (renamed MlaZ), transferred to PA3238 (renamed MlaY), and degraded. This pathway streamlines recycling of glycerophospholipid degradation products by removing glycerophospholipids from the outer leaflet prior to degradation.

The opportunistic bacterial pathogen Pseudomonas aeruginosa causes acute nosocomial infections as well as chronic infections in those with cystic fibrosis (1, 2). These infections can be challenging to treat as P. aeruginosa are resistant to many of the antibiotics in our current arsenal (3, 4). As with all Gram-negative bacteria, antibiotics must first cross the P. aeruginosa outer membrane to reach their intracellular target. The outer membrane is an asymmetric lipid bilayer with glycerophospholipids in the inner leaflet and the glycolipid lipopolysaccharide (LPS) in the outer leaflet (5–7). Electrostatic interactions between LPS and neighboring LPS molecules or outer membrane proteins create mechanical strength and prevent lipophilic molecules from entering the cell (8–13). Nutrients and small, hydrophilic molecules cross the outer membrane through porins, some of which are substrate specific, while others discriminate by substrate size (8). The overall structure of the outer membrane resembles that of a sieve, allowing small and hydrophilic molecules to diffuse freely into the cell while rejecting those that are large or hydrophobic (8, 9). P. aeruginosa lack some of the size-specific porins found in other Gram-negative bacteria, such as OmpF and OmpC in the model Gram-negative Escherichia coli (14). As a result, fewer molecules can penetrate the P. aeruginosa outer membrane, a factor that contributes to the enhanced antibiotic resistance of this pathogen (3, 4).

When the LPS interaction network is compromised, glycerophospholipids move into the outer leaflet (5, 11, 15). This creates patches of glycerophospholipid bilayer that disrupt the mechanical strength and permeability properties of the outer membrane (10, 11, 16). In E. coli, mislocalized glycerophospholipids are removed from the outer leaflet by the phospholipase PldA or the maintenance of lipid asymmetry (Mla) pathway (17–19) (Fig. 1A). PldA is an integral outer membrane β-barrel protein that breaks down glycerophospholipids and, to a lesser extent, lysoglycerophospholipids by removing an acyl chain from the sn-1 or sn-2 position (20–23). The lysoglycerophospholipid, fatty acid, and/or glycerophosphodiester products of this reaction are released into the outer leaflet, where they may then be transported to the cytoplasm and recycled (24–34). Both lysoglycerophospholipids and free fatty acids have detergent-like properties, and accumulation of these compounds in the outer leaflet may induce further membrane damage (18, 34, 35). The Mla pathway, which consists of the outer membrane protein MlaA, the periplasmic protein MlaC, and the inner membrane transporter MlaBDEF, transports mislocalized glycerophospholipids to the cytoplasmic/inner membrane (17). Outer leaflet glycerophospholipids move passively across the outer membrane through a channel in MlaA (36, 37). They are then shuttled across the periplasm by MlaC and inserted into the inner membrane by the MlaBDEF complex in an ATP-dependent manner (17, 38–40).

Fig. 1.

Maintenance of lipid asymmetry systems in E. coli and P. aeruginosa. (A) Outer membrane lipid asymmetry in E. coli is maintained by the Mla pathway (red) and PldA (yellow). The Mla pathway is highly conserved across proteobacteria and consists of the outer membrane protein MlaA, the periplasmic protein MlaC, and the inner membrane transporter MlaBDEF. Mislocalized glycerophospholipids in the outer leaflet are translocated across the outer membrane through a channel in MlaA, carried across the periplasm by MlaC, and inserted into the inner membrane by MlaBDEF in an ATP-dependent manner. PldA is an outer membrane phospholipase that hydrolyzes an acyl chain from outer leaflet glycerophospholipids and lysoglycerophospholipids. The degradation products of this reaction are released into the outer leaflet. (B) Outer membrane lipid asymmetry in P. aeruginosa is maintained by the MlaABCDEF (red) and MlaYZ (blue) pathways. The MlaABCDEF pathway is similar to that of E. coli and transports outer leaflet glycerophospholipids to the inner membrane. In the MlaYZ pathway, mislocalized outer leaflet glycerophospholipids are translocated across the outer membrane by MlaZ, delivered to the lipase MlaY, and degraded. While the products of this reaction are unknown, it is likely that they are released into the periplasm. ATP, adenosine triphosphate; ADP, adenosine diphosphate; IM, inner membrane; P_i, inorganic phosphate; LPS, lipopolysaccharide; OM, outer membrane; PL, glycerophospholipid.

While loss of the Mla pathway in E. coli results in a slight increase in outer membrane permeability, loss of PldA has no effect (17). As such, it is believed that the Mla pathway plays a more important role in maintaining outer membrane lipid asymmetry than PldA. However, loss of both systems has a greater impact on outer membrane permeability than loss of the Mla pathway alone, indicating that these systems have distinct but overlapping roles in maintaining outer membrane lipid asymmetry.

How outer membrane lipid asymmetry is maintained in P. aeruginosa is less clear. While no ortholog of PldA has been identified, components of the Mla pathway in P. aeruginosa are encoded by PA2800 (mlaA or vacJ) and the PA4456-PA4452 operon (mlaFEDCB or ttg2ABCDE) (Fig. 1B). Although some studies have observed a slight increase in outer membrane permeability in mla mutants, one study observed no effect (41–45). As such, a role for the Mla system in maintaining integrity of the P. aeruginosa outer membrane has been questioned (44). Recent biochemical data indicate that P. aeruginosa MlaC binds glycerophospholipids, providing strong evidence that the Mla pathway has similar roles in P. aeruginosa and E. coli (41). The modest increase, or lack thereof, in outer membrane permeability in P. aeruginosa mla mutants despite a role for the Mla pathway in maintaining lipid asymmetry indicates that P. aeruginosa must encode additional, unidentified pathways involved in this process.

In this study, we show that PA3238 and PA3239 (renamed MlaY and MlaZ, respectively) are involved in maintaining outer membrane lipid asymmetry in P. aeruginosa (Fig. 1B). The outer membrane lipoprotein MlaY is a putative lipase of the α/β hydrolase family, while MlaZ is a homolog of E. coli MlaA. We propose that mislocalized glycerophospholipids are removed from the outer leaflet by MlaZ, transferred to MlaY, and degraded. Though functionally analogous to PldA, the MlaYZ system would bypass the need to remove potentially toxic glycerophospholipid degradation products from the outer leaflet, which may enhance their recycling.

Results

P. aeruginosa Encode Multiple Orthologs of MlaA, MlaE, and MlaF.

We began our search for additional pathways that maintain outer membrane lipid asymmetry in P. aeruginosa by looking for orthologs of PldA. Basic Local Alignment Search Tool (BLAST) analysis failed to identify protein sequences in P. aeruginosa that are significantly similar to E. coli PldA. We then searched the AlphaFold database (46, 47) for P. aeruginosa proteins that are structurally similar to E. coli PldA. As a proof of principle, we searched the AlphaFold database for P. aeruginosa proteins that are similar to E. coli PagP, which, like PldA, is an integral outer membrane β-barrel protein without a similar protein sequence in P. aeruginosa. However, a previous study determined that the P. aeruginosa PagP ortholog is encoded by PA1343 (48). As expected, the P. aeruginosa protein in the AlphaFold database that is most similar in structure to E. coli PagP [PDB: 1THQ (49)] is PA1343 (SI Appendix, Table S1). Likewise, the E. coli protein in the AlphaFold database that is most similar to the predicted structure of PA1343 is PagP (SI Appendix, Table S1). Our analysis revealed that there are 63 P. aeruginosa proteins that are structurally similar to E. coli PldA [PDB: 1QD6 (18), SI Appendix, Table S2]. If these proteins are indeed orthologs of PldA, we would expect that their predicted structure is more similar to that of E. coli PldA than any other E. coli protein. As such, we searched the AlphaFold database for E. coli proteins that are similar in structure to each of the 63 P. aeruginosa proteins identified in our initial search. This reciprocal analysis revealed that each of the 63 P. aeruginosa proteins are more similar to an E. coli protein other than PldA, indicating that they are not PldA orthologs (SI Appendix, Table S2). Overall, these results indicate that P. aeruginosa may not contain an ortholog of PldA.

While we did not identify an ortholog of PldA in P. aeruginosa, we did identify multiple proteins with sequence similarity to E. coli MlaA, MlaE, and MlaF (SI Appendix, Table S3). In addition to the proteins that have already been identified as components of the Mla system, BLAST analysis revealed that PA3239 (renamed MlaZ) and PA3211 encode homologs of MlaA and MlaE, respectively. The large number of MlaF homologs identified in our BLAST search likely reflects the high degree of similarity between ABC transporter nucleotide-binding domains. One homolog of E. coli MlaF is encoded in an operon with PA3211, along with an MlaD domain–containing protein that was not identified in our BLAST search. These proteins may be part of additional pathways that maintain outer membrane lipid asymmetry in P. aeruginosa.

Loss of Both MlaA and MlaZ Disrupts Outer Membrane Integrity.

To investigate the role of MlaZ in outer membrane function, we constructed clean deletions of mlaA, mlaZ, or both in P. aeruginosa strain PAO1. We noticed that the ΔmlaA ΔmlaZ double mutant is blue, reflecting an increase in production of the secondary metabolite pyocyanin (50) (Fig. 2A). The increase in pyocyanin production in the ΔmlaA ΔmlaZ double mutant can be complemented by expressing a wild-type copy of either mlaA or mlaZ from an ectopic locus, confirming that this phenotype is caused by loss of both proteins. Pyocyanin levels in the ΔmlaA and ΔmlaZ single mutants are comparable to wild type (Fig. 2A). Although the reason why pyocyanin production is increased in the ΔmlaA ΔmlaZ mutant is unclear, this phenotype is proven to be a reliable indicator for loss of MlaA and MlaZ activity. A possible explanation for this phenotype is presented in the discussion.

Fig. 2.

Deleting both mlaA and mlaZ disrupts outer membrane integrity. (A) Oxidized pyocyanin levels in the clarified supernatant from stationary phase cultures of wild-type PAO1, the ΔmlaA and ΔmlaZ single mutants, the ΔmlaA ΔmlaZ double mutant, and the ΔmlaA ΔmlaZ double mutant expressing either mlaA or mlaZ at the Tn7 site. Data correspond to the means and SDs of three biological replicates. Asterisks indicate a statistically significant difference from wild type [****P < 0.0001 (one-way ANOVA with Dunnett’s multiple comparisons test)]. +, wild type; Δ, null allele. (B) Plating efficiency of wild-type PAO1 and the indicated mutants on LB agar and LB agar containing 0.5mM EDTA and 0.5% SDS. Data are representative of at least three independent replicates. Dilution is indicated under the first image.

We assessed outer membrane permeability in each mutant by measuring sensitivity to the anionic detergent sodium dodecyl sulfate (SDS) in the presence of ethylenediaminetetraacetic acid (EDTA), a divalent cation chelator that disrupts the LPS interaction network (51). While the ΔmlaA and ΔmlaZ and single mutants can grow on medium containing 0.5% SDS and 0.5mM EDTA, the ΔmlaA ΔmlaZ double mutant cannot (Fig. 2B). The SDS/EDTA sensitivity of the ΔmlaA ΔmlaZ double mutant can be rescued by expressing a wild-type copy of either mlaA or mlaZ (Fig. 2B). These data show that the outer membrane permeability barrier is more greatly disrupted in the ΔmlaA ΔmlaZ double mutant than either single mutant.

As previous studies have shown that P. aeruginosa lacking components of the Mla pathway are sensitive to tetracycline-like antibiotics (42, 44), we measured the sensitivity of the ΔmlaA ΔmlaZ double mutant to doxycycline. We found that wild-type P. aeruginosa, the ΔmlaA single mutant, and the ΔmlaZ single mutant can grow on rich medium containing 3, 6, or 12 µg mL⁻¹ doxycycline (SI Appendix, Fig. S1). The ΔmlaA ΔmlaZ double mutant, on the other hand, grows poorly on rich medium containing 3 µg mL⁻¹ doxycycline and not at all at higher concentrations (SI Appendix, Fig. S1). As loss of both mlaA and mlaZ has a synergistic, not additive, effect on pyocyanin production, SDS/EDTA sensitivity, and doxycycline sensitivity, these proteins likely have overlapping functions.

E. coli PldA Restores Outer Membrane Integrity in P. aeruginosa Lacking Both mlaA and mlaZ.

Antibiotic sensitivity of E. coli mla mutants can be suppressed by overexpressing the phospholipase PldA, strongly suggesting that glycerophospholipids accumulate in the outer leaflet of the outer membrane in the absence of the Mla system (17). To determine whether a similar defect occurs in the outer membrane of P. aeruginosa lacking both mlaA and mlaZ, we measured pyocyanin production, SDS/EDTA sensitivity, and doxycycline sensitivity in the ΔmlaA ΔmlaZ double mutant expressing E. coli PldA. As observed previously, loss of both mlaA and mlaZ increases pyocyanin production, SDS/EDA sensitivity, and doxycycline sensitivity compared to wild type (Fig. 3 A and B and SI Appendix, Fig. S2). We found that all three conditions can be suppressed by E. coli PldA. These data indicate that glycerophospholipids accumulate in the outer leaflet of the outer membrane in P. aeruginosa lacking both mlaA and mlaZ, providing further evidence that the P. aeruginosa Mla proteins are involved in maintaining outer membrane lipid asymmetry.

Fig. 3.

E. coli PldA restores outer membrane integrity in the ΔmlaA ΔmlaZ double mutant. (A) Oxidized pyocyanin levels in the clarified supernatant from stationary phase cultures of wild-type PAO1, the ΔmlaA and ΔmlaZ single mutants, the ΔmlaA ΔmlaZ double mutant, and the ΔmlaA ΔmlaZ double mutant expressing E. coli pldA at the Tn7 site. Data correspond to the means and SDs of three biological replicates. Asterisks indicate a statistically significant difference from wild type [****P < 0.0001 (one-way ANOVA with Dunnett’s multiple comparisons test)]. +, wild type; Δ, null allele. (B) Plating efficiency of wild-type PAO1 and the indicated mutants on LB agar and LB agar containing 0.5% SDS and 0.5mM EDTA. Data are representative of three independent replicates. Dilution plated is indicated under the first image.

MlaA and MlaZ Are Part of Different Maintenance of Lipid Asymmetry Pathways.

In addition to an MlaA-like protein, the Mla glycerophospholipid transport pathway consists of the lipid carrier MlaC and the inner membrane transporter MlaBDEF (17). To determine which of MlaA or MlaZ functions with MlaBCDEF, we compared pyocyanin levels, SDS/EDTA sensitivity, and doxycycline sensitivity in the ΔmlaA, ΔmlaZ, and ΔmlaBCDEF mutants to that of the ΔmlaBCDEF mutant lacking either mlaA or mlaZ. We found that pyocyanin levels are comparable in P. aeruginosa lacking both mlaA and mlaBCDEF to those lacking either mlaA or mlaBCDEF alone (Fig. 4A). However, much like the ΔmlaA ΔmlaZ double mutant, pyocyanin levels are increased in the ΔmlaZ ΔmlaBCDEF mutant. Likewise, SDS/EDTA and doxycycline sensitivity are increased in the ΔmlaZ ΔmlaBCDEF mutant but not in the ΔmlaA ΔmlaBCDEF mutant (Fig. 4B and SI Appendix, Fig. S3). As the phenotypes of the ΔmlaA ΔmlaBCDEF mutant are similar to that of the ΔmlaA and ΔmlaBCDEF mutants, we conclude that MlaA and MlaBCDEF are in the same Mla pathway. However, as the ΔmlaZ ΔmlaBCDEF mutant has increased levels of pyocyanin and is more sensitive to doxycycline and SDS/EDTA than either the ΔmlaZ or ΔmlaBCDEF mutant, we conclude that MlaZ and MlaBCDEF are in different Mla pathways.

Fig. 4.

MlaA, but not MlaZ, functions with MlaBCDEF. (A) Oxidized pyocyanin levels in the clarified supernatant from stationary phase cultures of wild-type PAO1 and the ΔmlaA, ΔmlaZ, ΔmlaBCDEF, ΔmlaA ΔmlaZ, ΔmlaA ΔmlaBCDEF, and ΔmlaZ ΔmlaBCDEF mutants. Data correspond to the means and SDs of three biological replicates. Asterisks indicate a statistically significant difference from wild type [****P < 0.0001 (one-way ANOVA with Dunnett’s multiple comparisons test)]. +, wild type; Δ, null allele. (B) Plating efficiency of wild-type PAO1 and the indicated mutants on LB agar and LB agar containing 0.5% SDS and 0.5mM EDTA. Data are representative of three independent replicates. Dilution is indicated under the first image.

MlaZ Functions with the Putative Lipase MlaY.

The data collected thus far indicate that P. aeruginosa encode at least two Mla pathways. The first pathway consists of MlaA, the lipid carrier MlaC, and the inner membrane transporter MlaBDEF, while the second pathway consists of MlaZ and an additional, unidentified protein(s). One protein that may be part of the second Mla pathway is the MlaE homolog PA3211 (SI Appendix, Table S3). To test this hypothesis, we measured the SDS/EDTA sensitivity of wild-type P. aeruginosa, the ΔmlaBCDEF mutant, and the ΔmlaBCDEF mutant lacking either mlaZ or PA3211. Growth of the ΔmlaBCDEF and ΔmlaBCDEF ΔPA3211 mutants in the presence of 0.5% SDS and 0.5mM EDTA is similar to wild type, while growth of the ΔmlaBCDEF ΔmlaZ mutant is decreased (SI Appendix, Fig. S4A). Similar results were obtained when assessing doxycycline sensitivity (SI Appendix, Fig. S4B). Together, these results indicate that outer membrane integrity in the ΔmlaBCDEF mutant is not affected by loss of PA3211. As such, PA3211 is not part of the second Mla pathway.

To identify additional proteins of the second Mla pathway, we looked for mutations that increase doxycycline sensitivity in P. aeruginosa lacking MlaABCDEF by transposon-directed insertion site sequencing. High-density transposon mutant libraries were constructed in wild-type PAO1 and the ΔmlaA ΔmlaBCDEF mutant and grown in the presence or absence of a sublethal dose of doxycycline in duplicate. The location and abundance of each transposon were determined by next-generation sequencing. Four genes had fewer transposon insertions in the ΔmlaA ΔmlaBCDEF mutant library treated with doxycycline than the untreated control in both replicates: mlaZ, PA3238 (renamed mlaY), PA3596, and PA5153 (Fig. 5A). As expected, mutants with transposons in mlaZ are less abundant, indicating that our experiment can successfully identify proteins of the second Mla pathway. Importantly, the number of transposons junctions mapped to mlaZ, mlaY, PA3596, and PA5153 was unaffected by doxycycline treatment in the wild-type background (Fig. 5A).

Fig. 5.

MlaY maintains outer membrane lipid asymmetry in the absence of MlaABCDEF. (A) Number of transposon junction sequencing reads mapped to each gene in the wild-type PAO1 and ΔmlaA ΔmlaBCDEF mutant libraries treated with 1 µg mL⁻¹ doxycycline or an equivalent volume of dimethyl sulfoxide (DMSO). Genes with less than one read are not shown. Data are representative of two independent replicates. (B) Oxidized pyocyanin levels in the clarified supernatant from stationary phase cultures of wildtype PAO1; the ΔmlaA ΔmlaBCDEF, ΔmlaZ, ΔmlaYZ, ΔmlaA ΔmlaBCDEF ΔmlaZ, and ΔmlaA ΔmlaBCDEF ΔmlaYZ mutants; and the ΔmlaA ΔmlaBCDEF ΔmlaZ and ΔmlaA ΔmlaBCDEF ΔmlaYZ mutants expressing mlaZ from the Tn7 site. (C) Plating efficiency of wildtype PAO1 and the indicated mutants on LB agar and LB agar containing 0.5% sodium dodecyl sulfate (SDS) and 0.5mM ethylenediaminetetraacetic acid (EDTA). (D) Oxidized pyocyanin levels in the clarified supernatant from stationary phase cultures of wildtype PAO1; the mlaY_H415A, ΔmlaA ΔmlaBCDEF, and ΔmlaA ΔmlaBCDEF mlaY_H415A mutants; and the ΔmlaA ΔmlaBCDEF mlaY_H415A mutant expressing mlaY from the Tn7 site. (E) Plating efficiency of wildtype PAO1 and the indicated mutants on LB agar and LB agar containing 0.5% SDS and 0.5mM EDTA. Data in (B) and (D) correspond to the means and standard deviations of three biological replicates. Asterisks indicate a statistically significant difference from wildtype (****, P <0.0001 [one-way ANOVA with Dunnett’s multiple comparisons test]). Data in (C) and (E) are representative of three independent experiments. Dilution plated is indicated under the first image in each panel. +, wildtype; Δ, null allele.

Here, we focus on MlaY, which is a putative lipase of the α/β hydrolase family (52). Primary sequence analysis indicates that MlaY is an outer membrane lipoprotein and a previous study observed MlaY in the periplasm (53). As mlaY is encoded directly upstream of mlaZ, deleting mlaY may have a polar effect on MlaZ expression. To circumvent this issue, we deleted the mlaYZ operon and expressed mlaZ from a constitutive promoter at the Tn7 site. To determine whether MlaY is a component of the second Mla pathway, we measured pyocyanin levels, SDS/EDTA sensitivity, and doxycycline sensitivity in the ΔmlaA ΔmlaBCDEF ΔmlaYZ mutant expressing mlaZ at the Tn7 site. While levels of pyocyanin are comparable in the wild-type and ΔmlaA ΔmlaBCDEF, ΔmlaZ, and ΔmlaYZ mutants, they are increased approximately four-fold in the ΔmlaA ΔmlaBCDEF mutant lacking either mlaZ or mlaYZ (Fig. 5B). Expressing mlaZ at the Tn7 site in the ΔmlaA ΔmlaBCDEF ΔmlaZ mutant restores pyocyanin levels to that of the wild type. However, pyocyanin levels remain elevated in the ΔmlaA ΔmlaBCDEF ΔmlaYZ mutant expressing mlaZ at the Tn7 site, indicating that both Mla pathways are disrupted. Likewise, expressing mlaZ at the Tn7 site restores SDS/EDTA and doxycycline resistance in the ΔmlaA ΔmlaBCDEF ΔmlaZ mutant but not the ΔmlaA ΔmlaBCDEF ΔmlaYZ mutant (Fig. 5C and SI Appendix, Fig. S5A).

By aligning the predicted structure of MlaY to that of a known lipase of the α/β hydrolase family, a previous study proposed that the active site of MlaY is formed by residues S187, D388, and H415 (52). As point mutations are unlikely to be polar, another strategy we employed to disrupt MlaY activity without affecting expression of MlaZ was to mutate H415 to an alanine. While pyocyanin levels, doxycycline sensitivity, and SDS/EDTA sensitivity in the ΔmlaA ΔmlaBCDEF and mlaY_H415A mutants are similar to wild type, all are affected in the ΔmlaA ΔmlaBCDEF mutant expressing MlaY^H415A (Fig. 5 D and E and SI Appendix, Fig. S5B). The mlaY_H415A allele can be complemented by expressing a wild-type copy of mlaY at the Tn7 site, revealing that the phenotypes associated with MlaY^H415A are due to loss of MlaY activity and not due to polar effects on MlaZ. Overall, these results demonstrate that both MlaY and MlaZ maintain outer membrane lipid asymmetry in the absence of the MlaABCDEF pathway. As the ΔmlaZ and ΔmlaYZ mutants have similar phenotypes, we conclude that MlaY and MlaZ are part of the same Mla pathway.

To determine whether MlaY and MlaZ are sufficient to maintain outer membrane lipid asymmetry, we expressed both proteins in E. coli lacking mlaC and pldA. The ΔmlaC ΔpldA double mutant is sensitive to SDS/EDTA, and this sensitivity can be rescued by expressing a wild-type copy of PldA from a plasmid (SI Appendix, Fig. S6). Likewise, expressing both MlaY and MlaZ from a plasmid can rescue the SDS/EDTA sensitivity of the E. coli ΔmlaC ΔpldA mutant. However, expressing MlaZ and the nonfunctional MlaY^H415A does not rescue the SDS/EDTA sensitivity of the ΔmlaC ΔpldA mutant. These results provide further evidence to suggest that MlaY and MlaZ maintain outer membrane lipid asymmetry and demonstrate that these proteins are both necessary and sufficient.

Evolutionary Analysis of MlaA, MlaY, and MlaZ.

To assess the evolutionary relationship between MlaA and MlaZ, we constructed a phylogenetic tree of 33 P. aeruginosa MlaA homologs from diverse Gram-negative bacteria (Fig. 6A). The results show that MlaA is in a monophyletic group with homologs from Burkholderia multivorans, Neisseria meningitidis, and Bordetella pertussis, while MlaZ is in a different but closely related monophyletic group containing homologs from Geobacter sulfurreducens, Desulfovibrio vulgaris, and Fusobacterium nucleatum. We then searched for homologs of MlaY in the proteomes of organisms included in the MlaA phylogenetic tree. Of these, only four have an MlaY homolog: P. aeruginosa, G. sulfurreducens, D. vulgaris, and F. nucleatum (Fig. 6A). Strikingly, these are the same organisms with MlaA homologs that cluster with MlaZ. In each organism, the MlaY homolog is encoded directly upstream of the MlaZ-clustering MlaA homolog. Overall, these results suggest that MlaZ has diverged from MlaA and coevolved with MlaY.

Fig. 6.

Evolutionary analysis of MlaA, MlaY, and MlaZ. (A) Tree of P. aeruginosa (PSEAE) MlaA homologs from Acinetobacter calcoaceticus (ACICP), Aliivibrio fischeri (ALIF1), Bordetella pertussis (BORPE), Burkholderia multivorans (BURM1), Campylobacter jejuni (CAMJE), Caulobacter vibrioides (CAUJE), Cereibacter sphaeroides (CERS4), Citrobacter koseri (CITK8), Desulfovibrio vulgaris (DESVH), E. coli (ECOLI), Enterobacter lignolyticus (ENTLS), Erwinia tasmaniensis (ERWT9), Francisella tularensis (FRATT), Fusobacterium nucleatum (FUSNN), Geobacter sulfurreducens (GEOSL), Haemophilus influenzae (HAEIN), Klebsiella pneumoniae (KLEPH), Neisseria meningitidis (NEIMB), Pasteurella multocida (PASMU), Photorhabdus laumondii (PHOLL), Photobacterium profundum (PHOPR), Rhodopirellula baltica (RHORT), Rickettsia prowazekii (RICPR), Salmonella typhimurium (SALTY), Shewanella oneidensis (SHEON), Shigella flexneri (SHIFL), Stenotrophomonas maltophilia (STRMK), Vibrio cholerae (VIBCH), Xanthomonas campestris (XANCP), and Yersinia pestis (YERPE). The tip of each branch is labeled with the UniProt accession number and organism. The location of P. aeruginosa MlaA and MlaZ is indicated by a red and dark blue marker, respectively. Organisms containing a MlaY homolog are labeled in light blue. (B) Alignment of the predicted structures of MlaY from P. aeruginosa and a protein of the MlaA family (UniProt accession number A0A3A4W9L3) from an unclassified Desulfobacteraceae. The AB_Hydrolase and MlaA domains of A0A3A4W9L3 are indicated in tan and red, respectively. The rmsd is 0.671 over 1,903 atoms. (C) Amino acid sequence identity between A0A3A4W9L3 from an unclassified Desulfobacteraceae (9DELT) and MlaY, MlaZ, and MlaA from P. aeruginosa calculated using Clustal MUCLE. The AB_Hydrolase and MlaA domains of A0A3A4W9L3 were identified by InterPro (54). Approximate regions compared are indicated in gray.

In addition to P. aeruginosa, B. multivorans, and D. vulgaris encode two homologs of MlaA (Fig. 6A). The homologs from B. multivorans are closely related and in the same monophyletic group as P. aeruginosa MlaA. The homologs from D. vulgaris are part of different monophyletic groups. One of the homologs is encoded in an operon with homologs of MlaC, MlaD, MlaE, and MlaF, indicating that this protein is likely an ortholog of MlaA. The other MlaA homolog in D. vulgaris clusters with P. aeruginosa MlaZ and is encoded immediately downstream of a MlaY-like protein. These findings suggest that this second MlaA homolog from D. vulgaris is an ortholog of MlaZ.

Review of the MlaA protein family on the InterPro database revealed that some MlaA homologs contain an α/β hydrolase domain (54). The predicted structure of one such homolog shows that the MlaA domain sits on top of the domain containing the α/β hydrolase fold (Fig. 6B). The α/β hydrolase domain of this fusion protein is homologous to MlaY, as indicated by a shared amino acid sequence identity of 40.46% (Fig. 6C). Furthermore, the predicted structure of the α/β hydrolase domain is similar to that of MlaY (Fig. 6B). The finding that some proteins of the MlaA family are fused to homologs of MlaY provides strong support for their shared function.

Discussion

In this study, we show that P. aeruginosa contain two proteins of the MlaA family. These proteins are encoded by the genes PA2800 and PA3239, which we have named mlaA and mlaZ, respectively. While mutants lacking mlaA or mlaZ individually behave like wild type, loss of both genes increases production of pyocyanin, sensitivity to SDS/EDTA, and sensitivity to doxycycline. These phenotypes can be suppressed by the E. coli phospholipase PldA, which degrades glycerophospholipids in the outer leaflet of the outer membrane (18, 19). Overall, these results suggest that MlaA and MlaZ have overlapping roles in maintaining outer membrane lipid asymmetry. In the absence of one, glycerophospholipids are efficiently removed from the outer leaflet by the other. In the absence of both, however, glycerophospholipids accumulate at the cell surface and weaken outer membrane integrity.

Our data suggest that MlaA and MlaZ are part of different pathways (Fig. 1B). MlaA belongs to a canonical glycerophospholipid transport pathway that, along with MlaC and MlaBDEF, transports mislocalized glycerophospholipids to the inner membrane (17). MlaZ functions with the putative lipase PA3238, which we have renamed MlaY. We found that residue H415 in the putative active site of MlaY is required to maintain outer membrane lipid asymmetry, indicating that MlaY has enzymatic activity and likely degrades mislocalized glycerophospholipids. However, as MlaY is a periplasmic-facing outer membrane lipoprotein, how it accesses outer leaflet glycerophospholipids is unclear. Bioinformatic analysis revealed that homologs of MlaY are fused to a MlaA-like protein in some Gram-negative bacteria, indicating that MlaY and MlaZ may interact in P. aeruginosa. We propose that mislocalized outer leaflet glycerophospholipids are funneled to MlaY through the channel in MlaZ. Once bound by MlaY, the glycerophospholipids are degraded.

We were unable to identify a homolog of PldA in P. aeruginosa by sequence or structural comparison. This and the complementation studies mentioned above with E. coli pldA argue strongly that P. aeruginosa lacks this protein. Lysoglycerophospholipids, fatty acids, and glycerophosphodiesters generated by PldA are deposited into the outer leaflet prior to uptake and recycling, providing an opportunity for these products to be lost to the environment or disrupt outer membrane integrity (18, 22, 34, 35). We believe that the MlaYZ system enhances recycling of degradation products by importing mislocalized glycerophospholipids into the cell prior to degradation. Furthermore, this system would avoid accumulating potentially toxic lysoglycerophospholipids and fatty acids in the outer leaflet. While we did not observe a growth defect in the ΔmlaA ΔmlaZ mutant expressing E. coli PldA, it is possible that outer leaflet lysophospholipids and fatty acids are harmful to P. aeruginosa in their native environment. Notably, however, outer leaflet lysophospholipids can be created in P. aeruginosa by the lipid A palmitoyltransferase PagP, indicating that some amount of lysophospholipid is tolerated.

Aside from its role in maintaining outer membrane lipid asymmetry, PldA regulates LPS biosynthesis in E. coli (33, 55). The presence of glycerophospholipids in the outer leaflet of the outer membrane indicates that more LPS is needed at the cell surface. Fatty acids released upon degradation of mislocalized glycerophospholipids by PldA are imported into the cytoplasm and ligated to coenzyme A by the acyl-CoA synthetase FadD. Acyl-CoA stimulates LPS biosynthesis by slowing degradation of LpxC, the enzyme that catalyzes the first committed step of LPS biosynthesis (56, 57). How acyl-CoA stabilizes LpxC is not fully understood. It will be interesting to see whether the MlaYZ pathway plays a similar signaling role in P. aeruginosa.

We and others have observed an increase in pyocyanin production in P. aeruginosa mla mutants (44). This increase can be suppressed by expressing E. coli PldA, indicating that mislocalized outer leaflet glycerophospholipids are the cause. Pyocyanin synthesis is controlled by the quorum sensing signaling cascade (58–60), which involves the synthesis, secretion, and detection of small molecules known as autoinducers (61). Given that many of the autoinducers produced by P. aeruginosa are hydrophobic (62), it is likely that their uptake from the environment is impeded by the outer membrane. We hypothesize that autoinducers are able to pass through patches of glycerophospholipid bilayer in the outer membrane of the ΔmlaA ΔmlaZ double mutant and inappropriately activate the quorum sensing circuit. In support of this hypothesis, a previous study found that expression of the quorum sensing regulator RhlR is increased in P. aeruginosa lacking MlaA (42). With this in mind, it is intriguing that autoinducers produced by many Gram-negative bacteria are hydrophobic small molecules while those produced by Gram-positive bacteria are small peptides (62).

Some have proposed that the MlaABCDEF system can transport glycerophospholipids from the inner membrane to the outer membrane (63–66). The results presented in this study add to the mounting evidence that argues against this hypothesis. In Acinetobacter baumannii, loss of the Mla system improves fitness of a mutant lacking lipooligosaccharide (67). As both leaflets of the outer membrane in this mutant are composed of glycerophospholipids, constitutive removal of glycerophospholipids from the outer leaflet by the Mla system disrupts membrane integrity. Biochemical studies reconstituting the Mla system in vitro have shown that retrograde glycerophospholipid transport is strongly favored under physiologically relevant conditions (38, 39). Here, we show that the function of the canonical Mla system in P. aeruginosa overlaps with a putative lipase. Lipase activity is fundamentally inconsistent with anterograde glycerophospholipid transport.

The outer membrane is an essential cellular structure that protects Gram-negative bacteria from their environment. The asymmetric distribution of lipids within this membrane is central to its function, with LPS in the outer leaflet and glycerophospholipids in the inner leaflet (5–7). We have identified a system that maintains this lipid asymmetry in the highly antibiotic-resistant pathogen P. aeruginosa. We propose that this system removes and degrades mislocalized glycerophospholipids from the outer leaflet, playing an analogous role to PldA in E. coli. This discovery highlights the diverse mechanisms used by Gram-negative bacteria to monitor and maintain outer membrane lipid homeostasis.

Methods

Bacterial Strains and Growth Conditions.

All bacterial strains and plasmids used in this study are listed in SI Appendix, Table S4. Unless otherwise stated, bacteria were grown aerobically in LB broth (Lennox; Fisher Scientific) with constant agitation in a roller drum or on LB agar (1.5% w/v agar; Becton, Dickinson, and Company) at 37 °C. 125 µg mL⁻¹ Ampicillin; 10 µg mL⁻¹ chloramphenicol; 0.12% (v/v) DMSO; 1, 3, 6, or 12 µg mL⁻¹ doxycycline (dissolved in DMSO); 0.5 mM or 1 mM EDTA; 15 or 30 µg mL⁻¹ gentamicin; 25 µg mL⁻¹ kanamycin; 10 mM MgCl₂; 0.5% (w/v) SDS; or 5% (w/v) sucrose were added to media as necessary. Chemicals were purchased from Sigma-Aldrich unless otherwise stated.

Strain and Plasmid Construction.

All primers (Integrated DNA Technologies) used in this study are listed in SI Appendix, Table S5. P. aeruginosa mutants were constructed by allelic exchange as previously described (68). To generate markerless, in-frame chromosomal deletions or substitutions, DNA approximately 1,000 base pairs upstream and downstream of the target site was amplified by PCR. The fragments were joined and inserted into the suicide vector pEXG2 by Gibson assembly (New England Biolabs). pEXG2 was linearized prior to Gibson Assembly using the restriction enzymes SacI and XbaI (New England Biolabs). Plasmids were electroporated into P. aeruginosa as previously described (69), and recombinants were selected for on LB agar containing 15 or 30 µg mL⁻¹ gentamicin. Single, gentamicin-resistant colonies were inoculated into LB broth and grown to late logarithmic phase at 37 °C. The resulting culture was serially diluted and plated onto salt-free LB agar (10% w/v tryptone (Becton, Dickinson, and Company), 5% w/v yeast extract (Becton, Dickinson, and Company), 1.5% (w/v agar) containing 5% (w/v) sucrose, and cells were grown overnight at 30 °C. Sucrose-resistant, gentamicin-sensitive colonies were screened for the appropriate allele by PCR and DNA sequencing (Genewiz).

Wild-type alleles of mlaA, mlaZ, and pldA were inserted at the Tn7 site in the P. aeruginosa chromosome as previously described (70). Each gene and native ribosome binding site were amplified using the primers listed in SI Appendix, Table S5. The resulting PCR products and the mini Tn7 vector pUC18T-mini-Tn7T-Gm were digested with the restriction enzymes HindIII and SacI (New England Biolabs) and ligated using T4 DNA ligase (New England Biolabs) following the manufacturers protocols. Insertion of each construct at the Tn7 site was confirmed by PCR and DNA sequencing (Genewiz). Notably, the Tn7::mlaZ construct contains a GC to AT transition 11 base pairs upstream of the mlaZ coding region. Expression of each gene is driven from the strong, constitutive promoter J23119 (http://parts.igem.org/Part:BBa_J23119; 5′- TTGACAGCTAGCTCAGTCCTAGGTATAATACTAGT-3′).

pldA, mlaYZ, and mlaY_H415AZ were amplified from the chromosome of wild-type MC4100, wild-type PAO1, or ML128, respectively, using the primers listed in SI Appendix, Table S5. The DNA fragments and pZS21 vector were digested with XhoI and HindIII (New England Biolabs) and ligated using T4 DNA ligase. Each gene is expressed from the J23119 promoter, and the proteins are translated using the native ribosome binding site.

Bioinformatic Analyses.

P. aeruginosa homologs of E. coli MlaA, MlaB, MlaC, MlaD, MlaE, MlaF, and PldA were identified using the UniProt Basic Local Alignment Search Tool (BLAST) program blastp with an E-threshold of 0.0001 and the BLOSUM62 matrix (71). E. coli protein sequences were obtained from ecocyc (72).

The crystal structures of E. coli PldA (PDB: 1QD6) and PagP (PDB: 1THQ) were obtained from the RCSB Protein Data Bank and compared against the P. aeruginosa AlphaFold database (46, 47) using the DALI server (73). The predicted structures of P. aeruginosa proteins bearing similarity to E. coli PldA or PagP were obtained from AlphaFold and compared against the E. coli AlphaFold database using DALI. Source data can be found in Dataset S1.

To build the MlaA phylogenetic tree, we searched for homologs of P. aeruginosa MlaA (PA2800) in a curated database containing the proteomes of Mycobacterium tuberculosis, Corynebacteria glutamicum, and 56 Gram-negative bacteria from diverse phyla using the UniProt BLAST program blastp with an E-value threshold of 0.0001 and the BLOSUM62 matrix (71). The organisms included in the curated database are listed in Dataset S2. Protein sequences of the MlaA homologs were aligned using MAFFT v7.490 in Geneious Prime with the BLOSUM62 scoring matrix, a gap open penalty of 1.53, and an offset value of 0.123 (74, 75). The maximum-likelihood tree was generated in IQ-TREE v1.6.12 using the best-fit model LG+F+I+G4 (76, 77). Branch supports were determined using 1,000 ultrafast bootstrap replicates and the SH-like approximate likelihood ratio test (78, 79). The maximum-likelihood tree is available as a Newick file at https://doi.org/10.5281/zenodo.7650644 (80). The amino acid sequence of P. aeruginosa MlaA was obtained from Pseudomonas Genome DB (81).

We searched for homologs of P. aeruginosa MlaY in the proteomes of organisms from the curated database that contain a MlaA homolog (Dataset S2) using the UniProt BLAST program blastp with an E-value threshold of 0.0001 and the BLOSUM62 matrix (71). The protein sequence for P. aeruginosa MlaY was obtained from Pseudomonas Genome DB (81).

The predicted structures of a MlaA-like protein from an unclassified Desulfobacteraceae (UniProt ID: A0A3A4W9L3) and P. aeruginosa MlaY (UniProt ID: Q9HZ02) were obtained from the AlphaFold database (46, 47) and aligned using the superposition/alignment plugin in PyMOL Molecular Graphics System, Version 2.3.5, Schrödiner, LLC. The AB_Hydrolase (residues L107-F454) and MlaA (residues R529-I741) domains of A0A3A4W9L3 were identified by InterPro (54). Protein sequences were aligned, and amino acid percent identities were determined, using the Clustal MUSCLE online tool. The AB_Hydrolase domain of A0A3A4W9L3 was aligned to the complete protein sequence of MlaY while the MlaA-like domain of A0A3A4W9L3 was aligned to the complete protein sequence of MlaA or MlaZ.

Pyocyanin Assay.

Oxidized pyocyanin levels were measured as previously described (82). A single colony was inoculated into 5 mL of LB broth and grown for 20 h at 37 °C with agitation in a roller drum. One milliliter of culture was pelleted by centrifugation at 3,900 rpm for 10 min, the supernatant was collected, and the cell pellet was resuspended in 1 mL phosphate-buffered saline (PBS; pH 7.4). The A₆₉₅ of the clarified supernatant and the A₆₀₀ of the bacterial culture resuspended in PBS were measured using the BioTek Synergy H1 plate reader. Pyocyanin production was calculated by dividing the A₆₉₅ of the supernatant by the A₆₀₀ of the bacterial culture. Source data for pyocyanin assays can be found in Dataset S3.

Efficiency of Plating.

P. aeruginosa strains were grown overnight in 5 mL LB broth containing 10 mM MgCl₂ at 37 °C with agitation in a roller drum. The cultures were diluted to an OD₆₀₀ of approximately 0.02 in 5 mL LB broth, grown to mid-logarithmic phase at 37 °C in a roller drum, and standardized by OD₆₀₀. E. coli strains were grown overnight in 5 mL LB broth at 37 °C with agitation in a roller drum, and the cultures were standardized by OD₆₀₀. Serial dilutions of the standardized cultures were replica plated onto the indicated agar media using a 48-pin replicator. Bacteria were grown overnight at 37 °C.

Transposon-Directed Insertion Sequencing.

To build the transposon mutant libraries, overnight cultures of the donor E. coli strain S17-1(λpir) carrying the transposon mutagenesis plasmid pBT20 were diluted 1:100 into 100 mL LB supplemented with 125 µg mL⁻¹ ampicillin and grown to an OD₆₀₀ of 0.5 to 0.6 at 37 °C with shaking at 200 rpm. The donor strain was pelleted by centrifugation, washed with 20 mL plain LB broth, and resuspended in LB to an OD₆₀₀ of 100. Overnight cultures of the recipient P. aeruginosa strains PAO1 and RLG926 were diluted 1:2 in LB broth and incubated statically at 42 °C for 3 h. Recipient strains were pelleted by centrifugation and resuspended to an OD₆₀₀ of 50 in LB. Equal volumes of the concentrated donor and recipient cultures were combined, spotted onto a prewarmed LB agar plate, and incubated at 37 °C for 2.5 h. Mating mixtures were resuspended in LB broth and plated in 100 µL aliquots on LB agar containing 30 µg mL⁻¹ gentamicin and 10 µg mL⁻¹ chloramphenicol. Approximately 100,000 and 150,000 individual exconjugants were pooled for the PAO1 and RLG926 transposon mutant libraries, respectively.

The PAO1 and RLG926 transposon mutant libraries were diluted to an OD₆₀₀ of approximately 0.001 in 25 mL LB broth and grown for 30 min at 37 °C with shaking at 200 rpm. Twenty-five milliliters of prewarmed LB broth containing 2 µg mL⁻¹ doxycycline (for a final concentration of 1 µg mL⁻¹) or an equivalent volume of DMSO was added to each culture, and cells were grown to an OD₆₀₀ of 0.5 at 37 °C with shaking at 200 rpm. Cells were collected, and genomic DNA was extracted using the DNeasy Blood and Tissue kit (Qiagen).

DNA libraries were prepared as previously described (83) and sequenced using the Illumina NovaSeq 6000. Primers used to amplify the transposon insertion sites are listed in SI Appendix, Table S5. Sequencing reads were mapped to the P. aeruginosa PAO1 genome (GenBank accession number: NC_002516) and quantified using Geneious Prime. Read counts can be found in Dataset S4.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (XLSX)

Data, Materials, and Software Availability

Newick Tree data have been deposited in Zenodo (https://doi.org/10.5281/zenodo.7650644) (80). All other data are included in the manuscript and/or supporting information.