Bacterial strains used in this study are listed in Table 1 . S. maltophilia K279a (Abda et al., 2015) was used as a model strain to investigate the role of the mla operon in this species. P. aeruginosa PAO1 was used in competition and polymicrobial biofilm formation assays. Unless otherwise stated, all strains were routinely grown at 37°C in Miller’s LB medium (1% tryptone, 0.5% yeast extract, 1% NaCl) at 200 rpm on a rotatory shaker, or on LB supplemented with 1.5% (w/v) agar (LBA), except E. coli SM10 (λpir)/pUX-BF13 (Bao et al., 1991) that was routinely cultured at 30°C. For biofilm formation assays, Brain Heart Infusion (BHI) broth from Oxoid (cat. No. CM1135) or modified BM2-glucose minimal medium (Overhage et al., 2007) (62 mM potassium phosphate buffer, pH 7.0, 2 mM MgSO₄, 10 μM FeSO₄, 0.4% glucose, supplemented with 0.5% casamino acids) was used. In general, bacterial growth was measured in a Novaspec II spectrophotometer at 600 nm (OD_600nm). When necessary, antibiotics were added to the agar plates at the indicated concentration.

Markerless S. maltophilia K279a mutants were constructed using the pGPI-SceI/pDAI-SceI-SacB system, which was initially described for species of the genus Burkholderia (Flannagan et al., 2008; Aubert et al., 2014). Briefly, pΔmlaF-B-US’DS’ and pΔmlaA-US’DS’ deletion plasmids were derived from the mobilizable suicide vector pGPI-SceI-XCm containing upstream and downstream flanking regions of the target genes in the K279a genome and maintained in E. coli SY327 (see Tables 1 , 2 for plasmid construction and primer details, respectively). The deletion plasmids were then transferred to the recipient S. maltophilia K279a by triparental mating using E. coli DH5α/pRK2013 as a helper strain and E. coli SY327 carrying the deletion plasmid as a donor strain. The K279a co-integrates were selected at 37°C on LBA plates containing 5 µg/mL norfloxacin in order to counterselect against E. coli donor and helper strains and 60 µg/mL chloramphenicol to select for K279a transconjugants. To confirm the integration of the suicide plasmids, single colonies were streaked and sprayed with 0.45 M pyrocatechol. Next, the plasmid pDAI-SceI-SacB was introduced into S. maltophilia K279a co-integrates by triparental mating using E. coli DH5α/pDAI-SceI-SacB and E. coli DH5α/pRK2013 as donor and helper strains, respectively. Transconjugants were selected at 30°C on LBA plates containing 5 µg/mL norfloxacin and 50 µg/mL tetracycline. The loss of the integrated pGPI-SceI-XCm plasmid was confirmed by the absence of yellow color upon pyrocatechol exposure and susceptibility to 60 µg/mL chloramphenicol. The presence of the mutant allele was confirmed by PCR screening and DNA sequencing. Markerless mutants were obtained by curing the plasmid pDAI-SceI-SacB using sucrose counterselection.

In trans complementation of the deleted genes was achieved by cloning the entire region containing the wild-type mlaF-B operon and the mlaA gene into the broad-host range plasmids pBBR1MCS1 and pBBR1-BAD-Cm, respectively. To generate the expression vector pBBR1-BAD-Cm, in which mlaA transcription is driven by the arabinose promoter, the pBAD promoter and AraC-encoding cassette from pBAD18-Cm were cloned into the vector pBBR1MCS1 on compatible SpeI/XhoI sites using standard techniques and primers pBAD18-Up and pBAD18-Lw ( Table 2 ).

The fluorescently labelled S. maltophilia wild-type strain K279a::sfGFP was constructed previously with a mini-Tn7 delivery system that has been optimized for work with bacteria of the family Xanthomonadaceae (Mamat et al., 2023). The same procedure using transfer of the mini-Tn7 delivery plasmid pUC18T-mini-Tn7T-Gm-Pc-sfGFP to S. maltophilia K279a ΔmlaF-B by four-parental mating was applied herein to label the mla mutant strain with sfGFP. For the construction of P. aeruginosa PAO1::tdTomato, the plasmid pUC18T-mini-Tn7T-Gm-Pc-tdTomato was transferred to the P. aeruginosa PAO1 wild-type strain by conjugation, involving E. coli DH5α/pRK2013 (Figurski and Helinski, 1979) and E. coli SM10 λpir/pUX-BF13 as helper strains, E. coli DH5α/pUC18T-mini-Tn7T-Gm-Pc-tdTomato (Mamat et al., 2023) as the donor, and P. aeruginosa PAO1 as the recipient strain. The strains were grown overnight in LB media with kanamycin (30 µg/mL) for E. coli DH5α/pRK2013, 100 µg/mL of ampicillin for E. coli SM10 λpir/pUX-BF13, and ampicillin (100 µg/mL) and gentamycin (15 µg/mL) for E. coli DH5α/pUC18T-mini-Tn7T-Gm-Pc-tdTomato. The cells from 1 mL of each E. coli culture and 330 µL of P. aeruginosa PAO1 were harvested by centrifugation. For preparation of the mating mixture, the cell pellets of all strains were pooled in 1 mL of LB medium, sedimented by centrifugation, resuspended in 100 µL of super optimal broth (SOB) (0.5% yeast extract, 2% tryptone, 10 mM NaCl, 2.5 mM KCl, 20 mM MgSO₄) and then spotted onto an SOB agar plate essentially as described elsewhere (Aubert et al., 2014). The P. aeruginosa PAO1 transconjugants were selected at 37°C on Vogel-Bonner minimal medium (VBMM) (3 g/L trisodium citrate, 2 g/L citric acid, 10 g/L K₂HPO₄, 3.5 g/L NaNH₄PO₄·4H₂O, 1 mM MgSO₄, 100 µM CaCl₂, pH 7.0) agar plates containing 30 µg/mL of gentamycin (Choi and Schweizer, 2006). Chromosomal insertion of the mini-Tn7 transposon downstream of the glmS gene for glutamine–fructose-6-phosphate aminotransferase was verified by PCR, using the primer pairs PTn7R (Choi and Schweizer, 2006) and PAO1-glmS-Ctrl1, and PTn7L (Choi and Schweizer, 2006) and PAO1-PA5548-Ctr to amplify the flanking mini-Tn7 regions ( Table 2 ), respectively.

To compare the growth of mutant strains with that of their parent strains, fresh overnight cultures were adjusted to an OD_600nm of 0.01 with appropriate media, followed by transfer of 200 μL of each diluted culture to each well of a conventional 96-well microtiter plate (triplicate samples). The plates were incubated for 24 hours in a Multiskan FC microplate photometer (Thermo Fisher Scientific) with constant circular shaking at 30°C or 37°C for bacterial growth, and the OD_550nm was measured every 15 min.

For growth competition assays on agar plates, overnight cultures of S. maltophilia were adjusted to OD_600nm of 0.01 and spread onto LBA using a sterile cotton swab to prepare a bacterial lawn. Five microliters of logarithmic cultures of P. aeruginosa PAO1 were spotted onto the competitor lawns using a micropipette. Plates were incubated for 24 hours at 37°C. For growth competition assays in liquid cultures, overnight cultures in 0.5 × BHI medium were diluted to an OD_600nm of 0.05 (∼5 × 10⁸ CFU/mL) and were inoculated simultaneously at an equal ratio (1:1 v/v). Samples were collected at 24 hours, serially diluted in PBS, and then seeded onto LBA with 60 µg/mL chloramphenicol for P. aeruginosa PAO1 and LBA with 50 µg/mL streptomycin for S. maltophilia K279a. The agar plates were incubated at 37°C for 24 hours and the number of viable cells was determined.

For plating efficiency tests, strains were grown overnight in LB medium at 37°C, serially diluted 10-fold, and then replica plated on LBA or MacConkey agar (MacConkey broth from Oxoid with 1.5% (w/v) agar). Plates were then incubated overnight at 37°C.

The minimum inhibitory concentration (MIC) of a broad spectrum of antibiotics was determined using the broth microdilution method in accordance with CLSI guidelines (CLSI, 2015) and EUCAST recommendations for colistin (The European Committee on Antimicrobial Susceptibility Testing and Clinical and Laboratory Standards Institute, 2016). Briefly, MICs were determined in sterile 96-well plates by two-fold serial dilutions of the corresponding antibiotic in cation adjusted Mueller-Hinton broth (CAMHB). To prepare CAMHB, Mueller-Hinton Broth (MHB) from Oxoid (cat. No. CM0405) was supplemented with calcium and magnesium to final concentrations of 25 and 12.5 mg/L, respectively. First, two-fold serial dilutions of antibiotics were prepared in 100 μl of CAMHB in microtiter plate wells. Overnight cultures in CAMHB were diluted in the same media to contain an initial inoculum of 5 × 10⁵ cells/mL. Each well containing the corresponding two-fold serial antibiotic dilution was filled with 100 µL of the cell suspension (final volume of 200 µL/well). MIC plates were read after 20 hours of incubation at 37°C without shaking. The MIC was determined to be the lowest concentration at which no bacterial growth occurred, as assessed by visual inspection and confirmed by adding 30 µL of 0.01% resazurin to each well to determine cell viability. Susceptibility to the membrane-damaging agents EDTA (ethylenediaminetetraacetic acid) and SDS (sodium dodecyl sulfate) was also determined by the broth microdilution method using MHB.

Uptake of 1-N-phenylnaphthylamine (NPN) was investigated as previously described (Helander and Mattila-Sandholm, 2000), with a few modifications (Loh et al., 1984; Fernández et al., 2012). Stock solutions of NPN at 50 mM and the uncoupler carbonyl cyanide m-chlorophenylhydrazone (CCCP) at 5 mM, both reagents purchased from Sigma-Aldrich, were prepared in advance in acetone. NPN was then diluted in 5 mM HEPES buffer (pH 7.2) to a concentration of 40 µM to prepare a working solution. Bacterial overnight cultures were diluted 100-fold in LB medium with appropriate antibiotic and grown to mid-exponential phase (OD_600nm of 0.6). Cells were harvested by centrifugation at room temperature and resuspended in 5 mM HEPES buffer (pH 7.2) supplemented with CCCP at 5 µM (HEPES-CCCP buffer). The cell suspension (OD_600nm = 0.5) was left at 23°C for an hour before adding any reagent. The cell suspension (100 µL), 40 µM NPN working solution (50 µL) and HEPES-CCCP buffer (50 µL) were mixed in flat-bottomed black 96-well plates immediately before measuring the fluorescence (excitation, 340 nm; emission, 415 nm) in a Victor 3V 1420 multilabel plate reader (PerkinElmer^®). To increase membrane permeability, 0.5 mM EDTA was added. Values were recorded up to 3 min to calculate NPN uptake kinetics. The results were expressed as relative fluorescence units (fluorescence value for the cell suspension with NPN, minus the corresponding value for the control), per triplicate samples, and two-way analysis of variance (ANOVA) with Bonferroni’s multiple-comparison test was applied to the data (GraphPad Prism v 9.0; GraphPad Inc, San Diego, USA).

DSF determination was performed using the reporter strain Xanthomonas campestris pv. campestris (Xcc) 8523 pL6engGUS as previously described (Huedo et al., 2015). The Xcc reporter strain was routinely grown in NYG medium (0.5% peptone, 0.3% yeast extract and 2% glycerol) supplemented with 10 μg/mL of tetracycline at 28°C. Strains to be tested were pin inoculated onto NYG agar plates containing 80 μg/mL of 5-bromo-4-chloro-3-indolyl-β-d-glucopyranoside (X-Glu) and the DSF-reporter strain and incubated for 24 hours at 28°C. The presence of a blue halo around the colony indicated DSF activity.

Overnight cultures were streaked onto BHI agar containing 3% sucrose and 0.08% (wt/v) Congo red dye. Plates were incubated at 30°C for 24 hours, and colonies were then examined for absorption of Congo red dye.

To evaluate single-species biofilm formation under static conditions, overnight cultures of the strains were grown aerobically (200 rpm) in 0.5 × BHI broth or modified BM2-glucose minimal medium at 37°C, followed by dilution of each culture with the corresponding fresh medium to an OD_600nm of 0.05. Only the overnight cultures of the complemented strain carrying the plasmid pBBR1MCS1-mlaF-B were supplemented with 60 µg/mL of chloramphenicol. Sterile, untreated 96-well flat-bottom microtiter plates were then inoculated with the fresh bacterial suspensions (200 μL per well) and incubated at 30°C for 24 hours in a humidified atmosphere. Quantification of the biofilm biomass was performed by crystal violet (CV) staining as described previously (Yero et al., 2020). The amount of biofilm was quantified by measuring the OD₅₅₀ of dissolved CV using a microplate reader (Multiskan FC microplate photometer, Thermo Fisher Scientific).

Mixed biofilms of S. maltophilia and P. aeruginosa strains were investigated with sfGFP- or tdTomato-labelled cells, respectively, in ibiTreat µ-Slide 8 well plates (Ibidi, Cat.No: 80826). To grow mixed biofilms of fluorescently tagged S. maltophilia and P. aeruginosa strains, preparation of the samples included overnight pre-cultures of the strains grown with agitation at 37°C in 0.5 × BHI medium and containing 60 µg/mL of chloramphenicol for the complemented mutant strain. The next day, the OD_600nm was adjusted to 0.05 for the initial inocula in 0.5 × BHI medium at 30°C, and 200 µL of each strain (in duplicate) were inoculated into 8-well chambers and incubated at 30°C in a humidified atmosphere. Every 12 hours, the medium was aspirated from each chamber and fresh medium was added, up to a total incubation time of 72 hours. Prior to observation under the microscope, all excess medium was aspirated to image the biofilm. Imaging was performed with a TCS SP5 inverse confocal laser-scanning microscope (Leica Microsystems, Mannheim, Germany) equipped with a Leica 63x/NA 1.40 HCX Plan Apochromat CS oil immersion objective. The sfGFP protein was excited with 488 nm laser light, and fluorescence emission was detected between 495 and 550 nm. The excitation wavelength of laser light was 561 nm for tdTomato and emission of its fluorescence was detected between 580 and 630 nm. The fluorescent signal from a minimum of three fields per chamber was collected (50 z-stacks) for processing. The images were analyzed with LAS AF software (version 2.73) and three-dimensional biofilm images were generated from confocal image stacks using the daime computer program (Daims et al., 2006). Images were split into individual color channels for biovolume quantification using the daime software. Image acquisition and processing was carried out at the core facility Fluorescence Cytometry of the Research Center Borstel (Borstel, Germany). ANOVA with Bonferroni’s multiple-comparison test was applied to the data from all biofilm experiments using GraphPad Prism v 9.0.

The visualization of S. maltophilia cells was performed on a Scanning Electron Microscope (SEM) LEO SEM 1525 (Zeiss, Germany) at 5 kV at the Hamburg Zoological Institute (Hamburg, Germany). Overnight cultures of each strain were prepared in 5 mL of LB medium with or without a sub-inhibitory concentration of 0.5 mM EDTA. The suspensions were centrifuged at 13,000 rpm for 2 min, and the cell sediments were resuspended in 500 µL of sterile PBS. Samples were fixed in 1% paraformaldehyde and gradually dehydrated in increasing concentrations of ethanol. The dehydration step was completed by drying the pellets at the critical point with a Balzers CPD 030 Critical Point Dryer instrument. Prior to visualization, the samples were coated with gold particles using an SCD 050 Sputter Coater (Bal-Tec, United States).

Visualization of internal structures of the different strains of S. maltophilia was carried out using the transmission electron microscope (TEM) Biotwin CM120 (Philips, Netherlands) operating at 120 kV in the facilities of the Biocenter Klein Flottbek of the University of Hamburg (Germany). Overnight cultures were prepared under the same conditions as for SEM. Cells were fixed overnight with 2.5% glutaraldehyde and post-fixed with 1% osmium tetraoxide for 2 hours. Prior to observation, the cells were dehydrated in ethanol gradients. Ultrathin sections were obtained using a UC7 ultramicrotome (Leica Microsystems) and deposited on a nickel-coated grid and counterstained with uranyl acetate and lead citrate. The images obtained with the Gatan MSC 794 camera were recovered with the built-in Digital Micrograph software.

To produce N-terminally His-tagged MlaC (Smlt4573) protein in E. coli BL21(DE3)pLysS, the coding region was cloned into the NcoI/XhoI sites of pET28b using standard techniques and primers MlaC_His_Up and MlaC_Lw ( Table 2 ). Since MlaC is naturally located in the periplasm, the signal sequence was removed from the coding sequence of the recombinant protein. The cells of an overnight culture of E. coli BL21(DE3)pLysS containing the pET28b-H6-MlaC plasmid were diluted 1:100 in 2 L of pre-warmed LB and grown to mid-exponential phase (OD_600nm = 0.6). Isopropyl-ß-d-thiogalactopyranoside (IPTG) was added to the culture at a final concentration of 1 mM for protein expression induction. After four hours, cells were harvested (OD_600nm = 3.5) and washed twice with sterile PBS. In order to verify the presence of the MlaC protein in the soluble fraction, the cell pellet was resuspended in lysis buffer (5 mM imidazole, 0.5 M NaCl, 0.1% Triton X-100, 5% glycerol, 20 mM Tris-HCl, pH 7.9), lysed by sonication in an ice bath, and separated by 12% SDS-PAGE. Protein purification was performed by the Protein Production Platform (PPP) of Nanbiosis facilities at the Institute of Biotechnology and Biomedicine of the Universitat Autònoma de Barcelona (UAB, Spain). Briefly, cells were disrupted by sonication in lysis buffer containing 4 µg/mL lysozyme, 8 µg/mL DNase and 2 mM MgCl₂. Protein was purified from the soluble fraction by immobilized metal affinity chromatography (IMAC) on a HisTrap HP 5 mL (GE, 17-5248-01) column with washing buffer (40 mM imidazole, 0.33 M NaCl, 13.3 mM Tris-HCl, pH 7.9) and elution buffer (125 mM imidazole, 62.5 mM NaCl, 2.5 mM Tris-HCl, pH 7.9). Eluted protein was immediately dialyzed against 50 mM Na₂HPO₄ (pH 7.0) to remove the imidazole.

Recombinant MlaC protein produced in E. coli was delipidated using an HPLC system and a C18 column (Phenomenex Jupiter 5U C18 300A) in 0.1% TFA. Protein was eluted from the column with a gradient of acetonitrile in 0.1% TFA.

Protein identity was confirmed by peptide mass fingerprinting using matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) at the Proteomics Laboratory of the Consejo Superior de Investigaciones Científicas and UAB. Analysis of MlaC-bound PLs was done under denaturing conditions by MALDI-TOF MS in the negative ion mode with 9-aminoacridine as matrix. The identification of E. coli PLs present in the sample was done according to Oursel et al. (2007) and Gidden et al. (2009), using Lipidomics Gateway (http://www.lipidmaps.org) based on the m/z values of MS spectra.

The binding of phosphatidylethanolamines PE14/14 and PE16/16 to MlaC was first evaluated by nano Differential Scanning Fluorimetry (nanoDSF), which determines the melting temperature of the protein in the presence and absence of compound. The reaction buffer was optimized to 50 mM MOPS pH 7.0, 100 mM NaCl and 0.05% Tween, for a final reaction volume of 10 μl; final protein and compound concentrations were 5 μM and 100 μM, respectively, and DMSO concentration was 1%; the reaction temperature was 20-95°C, with a heating rate of 1°C min^-1. Binding and corresponding K_d values were then evaluated by MicroScale Thermophoresis (MST), which measures protein movement along a temperature gradient in the presence and absence of compound. The reaction buffer was as for nanoDSF; final dye and protein concentrations were 5 nM and 100 nM, respectively; MST excitation power was 20%. Both assays were performed by Proteros Biostructures GmbH (Germany).

S. maltophilia K279a homologues of P. aeruginosa PAO1 genes coding for Ttg2ABCDE (PA4452-PA4456) or VacJ (PA2800) and E. coli K-12 substr. MG1655 genes encoding MlaFEDCB (b3195- b3191) or MlaA (b2346) were identified as follows. Protein sequences were first analyzed using BLAST, PSI-BLAST and CDD within NCBI (http://ncbi.nlm.nih.gov/). Orthologous sequences in other species were searched using BLASTp and initially aligned with Clustal Omega (https://www.ebi.ac.uk/Tools/msa/clustalo/). Promoter prediction was done by means of PRODORIC database (Dudek and Jahn, 2022). Operon organization was inferred from the BioCyc data collection (Caspi et al., 2016) and MicrobesOnline database (Alm et al., 2005). Transcription units were identified by mapping reads from previously published RNA-seq experiment (GEO GSE206442) (Coves et al., 2023) on the K279a chromosome and visualized by the Artemis genome browser (Rutherford et al., 2000).

Protein structures of MlaC orthologues from P. aeruginosa (6HSY) (Yero et al., 2021), E. coli (5UWA) (Ekiert et al., 2017) and Ralstonia solanacearum (2QGU) were obtained from the RCSB PDB database (https://www.rcsb.org/). For structural comparisons, all selected structures are lipid-bound and hydrogen atoms were removed from the 6HSY structure. Multiple sequence alignment was constructed using hmmalign tool from the HMMER v3.3.2 package (Mistry et al., 2013) using Pfam PF05494.15 profile (https://www.ebi.ac.uk/interpro/) and presented with Jalview (Waterhouse et al., 2009). DeepMind’s AlphaFold (Jumper et al., 2021) version 2.2.4, locally installed and running in a Docker container, was used to model the Smlt4673 protein from residue A28 onward. AlphaFold was obtained from https://github.com/deepmind/alphafold. The script provided by the developers was used to download all required databases (software and databases were downloaded on October 18th, 2022). The AlphaFold script was run with default parameters using the “monomer_ptm” model, full databases and multiple sequence alignments constructed using all templates available at the download date. 3D structure supersposition was carried out using DaliLite.v5.1 (Holm, 2019) and cavities were analyzed with the web server CASTp (Tian et al., 2018). PyMOL v2.4.2 was used for protein structure visual analysis and figure preparation (The PyMOL Molecular Graphics System, Version 2.4.2, 2022).

We first examined the genomic organization of the mla genes (Smlt4670-Smlt4675) in the S. maltophilia K279a reference genome and compared it to that previously found in other Pseudomonadota ( Supplementary Figure 1A ). BioCyc predicted that the mla operon is divided into two distinct transcription units: smlt4670 (mlaF) and smlt4671 (mlaE), and smlt4672 (mlaD) to smlt4675 (mlaA). However, several lines of evidence suggest that all six genes belong to the same transcription unit: (i) the mla genes are oriented in the same direction on the chromosome including a maximum intergenic region of 59 bp between them (mlaE to mlaD), (ii) the lack of canonical promoters between mlaF and mlaA as predicted by the PRODORIC database, (iii) the existence of a complete mla operon with similar organization in other species, such as Neisseria gonorrhoeae (Baarda et al., 2019), (iv) transcriptomic studies for S. maltophilia K279a in different growth phases revealed a common deregulation of the expression of these six genes (Coves et al., 2023), and (v) mapping of reads indicated the existence of a single transcription unit, although there appear to be internal promoters upstream to some CDS such as smlt4674 (mlaB) and smlt4675 (mlaA) ( Supplementary Figure 1B ). Taken together, these observations strongly suggest the organization of the genes of the Mla system in a single polycistronic operon in S. maltophilia.

The MlaC (Smlt4673) protein of S. maltophilia is the soluble periplasmic PL-binding component of the Mla system. First, the primary amino acid sequence of Smlt4673 was compared with that of orthologous proteins from P. aeruginosa, E. coli, and R. solanacearum, whose 3D structures have already been solved ( Supplementary Figure 2 ). Despite low sequence identity between the MlaC protein of S. maltophilia and other organisms ( Supplementary Figures 1A , 2 ), the protein structure is expected to be highly conserved as shown for other species (Yero et al., 2021). The AlphaFold prediction for the 3D structure of Smlt4673 (UniProt accession B2FP85) yielded five models with pLDDT scores between 91.9 and 89.52 ( Figure 1A ), showing great confidence in the central zone of the sequence, which aligns perfectly with the Hidden Markov Model profile that defines the Pfam family MlaC (PF05494), leaving the areas with less confidence at the extremes. A superposition of all models over the best-ranking one resulted in root-mean-square deviation (RMSD) values between 0.9Å and 1.3Å, indicating a high similarity between them ( Supplementary Table 1 ). The C-terminal region was predicted as unstructured (Tunyasuvunakool et al., 2021), likely due to the presence of glycine and proline residues (Hutchinson and Thornton, 1994).

Superposition of the best-scoring model for Smlt4673 with the 3D structures of orthologous proteins whose amino acid sequences are aligned in Supplementary Figure 2 confirmed that structure is conserved among members of this protein family ( Supplementary Figure 3 ). As shown in Supplementary Table 1 , the most similar structure is that of P. aeruginosa, with an RMSD of 2.3Å, 176 out of 190 superposed amino acids, and a sequence identity of 25%. Note that all structures used for this comparison were solved in complex with a lipid ligand in its hydrophobic pocket. Finally, an analysis of the ligand-binding cavity in the five AlphaFold models revealed volumes between 1777.2 and 2751.6 ų ( Figure 1B ), which were compared to the values for the already known ortholog structures in Supplementary Table 1 (see Supplementary Figure 4 for a surface representation of the cavity in orthologous structures). According to the models, the Smlt473 cavity would be smaller than that of the MlaC ortholog of P. aeruginosa, but larger than those in the other MlaC proteins used for comparison (E. coli and R. solanacearum). Amino acids and secondary structures that could contribute to the size and flexibility of the cavity and ligand binding are shown in Supplementary Figures 3 , 5 .

The original description of the first structure solved for a MlaC protein, from E. coli (PDB 5UWA) (Ekiert et al., 2017), identified its fold with a transport protein domain known as NTF2. In a later study, the presence of a second domain, similar to the small helical domain of AAA+ ATPases, was identified and investigated (Yero et al., 2021). It was suggested that this domain may be responsible for the increase in cavity volume of MlaC proteins relative to other proteins with NTF2-type domains. In particular, the capacity of different MlaC orthologs to simultaneously transport two PLs was evaluated. This capacity was attributed to a bend caused by a glycine residue in the C-terminal helix and an adjacent tryptophan (G195 and W196 in 6HSY). The sequence of MlaC from S. maltophilia does not have either of these amino acids at these positions (E205 and L206) and the generated model displays a straight helix with no bend at this position. In fact, MlaC from S. maltophilia has a sequence very similar to that of E. coli (Q196 and L197), whose structure also shows a straight helix. However, the high presence of glycine residues down the sequence (G209, G214, P215, G217, G219) is likely to provide high flexibility to the region (Hutchinson and Thornton, 1994), as predicted by AlphaFold. This could translate in a capacity to accommodate one or two substrates, as shown for the P. aeruginosa protein ( Supplementary Figures 2 , 3 ).

Analyzing the superposition of the known structures of the MlaC family, one recognizes the persistent presence of bulky residues inside the small helical AAA-like domain. In all but the 6HSY structure, the aromatic side chain of the residue superposed to F190 in Smlt4673 AlphaFold model (i.e., F181 in 2QGU and W181 in 5UWA) is occupying the same space. In P. aeruginosa, thanks to the bend produced by G195, this space is occupied by W196, which displaces F178 (the residue superposed to Smlt4673 F190) towards the interior of the domain, increasing the number of atoms housed there. Most likely, this causes the separation between the helices that form this domain in 6HSY, increasing the volume of the cavity in the NTF2 domain and enabling the binding of a second PL. Another position occupied in all cases by an aromatic residue except in 2QGU is Smlt4673 F104, which most likely would be interacting with the previously mentioned aromatic residue ( Supplementary Figures 2 , 4A, B ). On the other hand, all the structures studied show the presence of two bulky residues facing each other, marking the beginning of this domain and interacting with the PL ( Supplementary Figure 5C ).

Denaturing mass spectrometry was used to determine the lipids that were specifically bound to recombinant MlaC from S. maltophilia produced in the cytoplasm of E. coli ( Figure 2A ). Although this is clearly not the natural environment of the protein (i.e., the periplasmic space of S. maltophilia), several types of PLs were identified by the assay. A delipidated variant of the protein was used as a control. PLs bound by S. maltophilia MlaC belonged primarily to the phosphatidylglycerol (PG) and phosphatidylethanolamine (PE) types ( Figure 2B ), as previously described for other MlaC proteins. Lipid species with m/z of 719.5, 733.5 and 747.5 were found to be the most abundant under the assayed conditions (MALDI-TOF MS in the negative mode).

In vitro evaluation of the binding of delipidated MlaC to free PLs using microscale thermophoresis (MST) and nanoDSF provided the protein’s binding affinity for the specific PE variants C14:0/C14:0 and C16:0/C16:0 ( Figures 2C–E ). For C16:0/C16:0, no binding to MlaC was observed in either assay. For C14:0/C14:0, both assays showed potential binding to MlaC, albeit with a low-affinity K_d of 795 µM (MST).

We investigated the effects of the Mla system on OM permeability by analyzing two K279a mutant strains, one lacking the components of the mla operon except for mlaA (ΔmlaF-B) and the other lacking the mlaA gene (ΔmlaA). First, we showed that none of the mutants exhibited a growth defect under different standard laboratory conditions ( Supplementary Figure 6 ). In order to confirm an OM-barrier defect due to an incomplete Mla system, the sensitivities of the wild-type, mutants and their complemented strains to OM stressors, such as the chelating agent EDTA, SDS, and bile salts, were measured. The mutants were fourfold more susceptible to EDTA (MIC = 0.78 mM) than the parental wild-type strain (MIC = 3.125 mM). However, no difference was observed between the mutant and wild-type cells in their susceptibility to SDS, with an MIC value of 0.04% for all strains ( Table 3 ). On the other hand, plating efficiency assays in the presence of bile salts (MacConkey) showed that growth of cells lacking the Mla system was partially impaired in comparison with wild-type and complemented strains ( Figure 3A ).

To investigate whether the deletion of components of the Mla system affects antibiotic resistance, the susceptibility of all strains to a broad spectrum of antibiotics was analyzed. As shown in Table 3 , both the mlaF-B and mlaA mutants showed increased susceptibility to colistin and minocycline among all antibiotics tested. The complemented strains showed restored MIC values. However, the sensitivity to the rest of the antibiotics remained unchanged. After these results and the evidence that both mutants showed the same susceptibility phenotypes, we decided to proceed with the mutant strain ΔmlaF-B.

The uptake rate of 1-N-phenylnapthylamine (NPN) was also measured. The NPN probe fluoresces strongly in PL environments but only weakly in aqueous environments, and a damaged OM should be more permeable to hydrophobic substances such as NPN. The increase in fluorescence levels was significantly higher in the ΔmlaF-B strain (P<0.05) compared to the parental wild-type strain in the presence of 10 µM NPN ( Figure 3B ), indicating a compromised integrity of the OM, partially restored by in trans complementation of ΔmlaF-B with pBBR1MCS1-mlaF-B. To demonstrate that the mla-defective strain was more sensitive to external stressors, EDTA was added at a subinhibitory concentration of 0.5 mM, which resulted in more pronounced differences when comparing the wild-type strain and the mla mutant ( Figure 3C ). EDTA at this concentration did not affect the growth of the strains used in the assay ( Supplementary Figure 6 ).

To further investigate the effects that the mutation in the Mla system could have on S. maltophilia cells, we examined morphological changes in cell structure using electron microscopy. In a first step, samples from cell cultures of the wild-type strain, the mutant ΔmlaF-B and the complemented strain were examined by scanning electron microscopy (SEM). Under normal culture conditions, i.e., without addition of the chelating agent EDTA, the cells exhibited normal morphology with the characteristic rod-like shape of an S. maltophilia bacterium ( Figure 4A ). Despite a defective Mla system, no obvious morphological or structural differences were observed in ΔmlaF-B cells in the absence of EDTA ( Figure 4A ), suggesting that either the loss of components of the Mla system is not critical for maintaining the structural integrity of the membrane or that other systems are involved in compensating for this deficiency. Previous studies had also shown that cell structure was not affected in MlaA mutants of N. gonorrhoeae (Baarda et al., 2019) and Haemophilus influenzae (Fernández-Calvet et al., 2018).

In contrast, after treatment with EDTA at the subinhibitory concentration of 0.5 mM, mutant cells showed important morphological disorders. The ΔmlaF-B cells adopted a more spherical shape with smaller size in the presence of EDTA, indicating perturbations in the structural integrity of the cell envelope ( Figure 4B ). The detrimental effects on the integrity of the bacterial cell envelope were further confirmed by the detection of bacterial ghost cells with irregular shapes in SEM and TEM ( Figures 4B–D ), with the TEM images showing in particular empty cell envelopes devoid of cytoplasmic and nuclear contents ( Figures 4C, D ). Although the observed characteristics of the mutant strain upon contact with 0.5 mM EDTA indicated a decrease in cell viability, no effect of this concentration on the growth rate of the mutant strain could be detected ( Supplementary Figure 6 ).

We studied biofilm formation in microtiter plate under static conditions using K279a wild-type, mutant, and complemented strains grown in 0.5 × BHI or a modified BM2-glucose minimal medium ( Figure 5 ). The amount of biofilm formed by the ΔmlaF-B mutant was lower than that of the wild-type strain in the presence of EDTA (0.5 mM), but only when the cells were grown in minimal medium ( Figure 5 ). EDTA at this concentration did not affect the planktonic growth of the strains under the same conditions used for biofilm formation ( Supplementary Figure 6 ). To confirm the unique role of EDTA in biofilm reduction in the mutant strain, we also showed that production of both exopolysaccharide and the quorum sensing signal DSF (diffusible signal factor) were not affected in this strain ( Supplementary Figure 7 ). In S. maltophilia, exopolysaccharide biosynthetic genes are necessary for biofilm production (Huang et al., 2006) and DSF positively regulates biofilm formation (Alcaraz et al., 2019).

In addition, for formation of mixed biofilms, sfGFP-labelled strains of S. maltophilia were incubated in 8-well µ-Slide chambers with wild-type, tdTomato-labelled P. aeruginosa strain PAO1 ( Figure 6 ). We chose the BHI medium because it has proven to be the most robust medium for studying in vitro biofilm formation of a wide range of bacteria. We had previously shown that S. maltophilia and P. aeruginosa do not compete with each other under standard laboratory culture conditions and that S. maltophilia strains are not sensitive to compounds exogenously produced by P. aeruginosa ( Supplementary Figure 8 ). As observed in the S. maltophilia monospecies biofilms with EDTA, the mutant ΔmlaF-B also formed less biofilm in the presence of P. aeruginosa, although in this case biofilm formation was even more impaired ( Figure 6 ). In the competition assays, we observed that when the mixed culture conditions were static (without shaking), the planktonic population of S. maltophilia was reduced by one log order, but this was equally the case for all strains examined ( Supplementary Figure 8D ). Interestingly, biomass quantification of the 72-hours biofilms showed that P. aeruginosa cells were also affected in the mixed biofilm in the presence of the mutant ΔmlaF-B ( Figure 6B ), which seems to indicate a bidirectional effect.