1. Microbiology and Infectious Disease
Download icon

The Acinetobacter baumannii Mla system and glycerophospholipid transport to the outer membrane

  1. Cassandra Kamischke
  2. Junping Fan
  3. Julien Bergeron
  4. Hemantha D Kulasekara
  5. Zachary D Dalebroux
  6. Anika Burrell
  7. Justin M Kollman
  8. Samuel I Miller  Is a corresponding author
  1. University of Washington, United States
  2. The University of Sheffield, United Kingdom
Research Article
  • Cited 0
  • Views 891
  • Annotations
Cite this article as: eLife 2019;8:e40171 doi: 10.7554/eLife.40171

Abstract

The outer membrane (OM) of Gram-negative bacteria serves as a selective permeability barrier that allows entry of essential nutrients while excluding toxic compounds, including antibiotics. The OM is asymmetric and contains an outer leaflet of lipopolysaccharides (LPS) or lipooligosaccharides (LOS) and an inner leaflet of glycerophospholipids (GPL). We screened Acinetobacter baumannii transposon mutants and identified a number of mutants with OM defects, including an ABC transporter system homologous to the Mla system in E. coli. We further show that this opportunistic, antibiotic-resistant pathogen uses this multicomponent protein complex and ATP hydrolysis at the inner membrane to promote GPL export to the OM. The broad conservation of the Mla system in Gram-negative bacteria suggests the system may play a conserved role in OM biogenesis. The importance of the Mla system to Acinetobacter baumannii OM integrity and antibiotic sensitivity suggests that its components may serve as new antimicrobial therapeutic targets.

https://doi.org/10.7554/eLife.40171.001

eLife digest

Gram-negative bacteria are a large group of single-celled organisms that share a typical external envelope. This casing is formed of an inner and an outer membrane, which have different structures and properties.

The outer membrane lets nutrients penetrate inside the cell, but blocks out other compounds, such as antibiotics. It is made of a complex assembly of molecules, including glycerolphospholipids (GPL) that are produced inside the cells. Very little is known about how this external shield is created and maintained. For example, it was still unclear how GPLs were exported through the inner membrane to the outer one.

To investigate these questions, Kamischke et al. exposed a species of Gram-negative bacteria to a molecule that is normally blocked by the outer membrane. If the outer membrane is not working properly, the compound can cross it and the cell turns blue.

Kamischke et al. then introduced genetic changes at random locations in the genomes of the bacteria. If colonies became blue, this meant that the mutations had happened in a gene essential for the outer membrane. Sequencing these blue bacteria revealed 58 genes involved in keeping the outer membrane working properly. Amongst them, four genes coded for a transport machine, the Mla system, which allowed GPLs to cross the inner membrane and reach the outer membrane. The experiments also showed that a working Mla system was required for bacteria to survive antibiotics.

Certain dangerous Gram-negative bacteria are now resistant to many drugs, having evolved unique envelopes that keep antibiotics at bay. By learning more about the outer membrane, we may be able to create new treatments to bypass or to disable this shield, for example by targeting the Mla system.

https://doi.org/10.7554/eLife.40171.002

Introduction

Gram-negative bacteria are enveloped by two lipid bilayers, separated by an aqueous periplasmic space containing a peptidoglycan cell wall. The inner membrane (IM) is a symmetric bilayer of glycerophospholipids (GPL), of which zwitterionic phosphatidylethanalomine (PE), acidic phosphatidylglycerol (PG), and cardiolipin (CL) are among the most widely distributed in bacteria (Zhang and Rock, 2008). In contrast, the outer membrane (OM) is largely asymmetric and composed of an inner leaflet of GPL and an outer leaflet of lipopolysaccharide (LPS) or lipooligosaccharide (LOS) (Pelletier et al., 2013). The OM forms the first line of defense against antimicrobials by functioning as a molecular permeability barrier. The asymmetric nature of its lipid bilayer and the structure of LPS/LOS molecules facilitates barrier function, as the core region of LPS impedes the entry of hydrophobic molecules into the cell while the acyl chains within the bilayer also help to prevent the entry of many hydrophilic compounds (Bishop, 2014). Although progress has been made in understanding many aspects of OM assembly – including the discovery of an LPS transport system and the machinery for proper folding and insertion of outer membrane proteins (Okuda and Tokuda, 2011; Okuda et al., 2016) – little is known about the molecular mechanisms for transport of the GPLs necessary for OM formation and barrier function.

Acinetobacter baumannii is an important cause of antibiotic-resistant opportunistic infections and has significant innate resistance to disinfectants and antibiotics. Similar to other Gram-negative opportunistic pathogens such as Pseudomonas aeruginosa and Klebsiella spp., individuals with breached skin or damaged respiratory tract mucosa are most vulnerable (Chmelnitsky et al., 2013; Abbo et al., 2005). We performed a genetic screen to identify genes important for the OM barrier of A. baumannii. This led to the identification of an ABC (ATP-binding cassette) transporter complex that promotes GPL export to the OM. Transporter disruption attenuates bacterial OM barrier function, resulting in increased susceptibility of A. baumannii to a wide variety of antibiotics.

The homologous system for E. coli has previously been termed Mla for its suggested role in the maintenance of outer membrane lipid asymmetry via the removal of GPL from the outer leaflet of the OM to the IM. While this is a reasonable hypothesis, there is not direct biochemical evidence that the Mla system functions to return GPL from the OM to the IM. In this work, we present evidence that the A. baumannii Mla system functions to promote GPL movement from the IM to the OM. This conclusion is based on the observation that newly synthesized GPLs accumulate at the IM of mla mutants, akin to how LPS molecules accumulate at the inner membrane in bacteria with mutations in the lpt genes encoding the LPS ABC transport system (Okuda et al., 2016). Given the broad conservation of Mla in prokaryotic diderm organisms, the anterograde trafficking function of Mla might be exploited by a variety of species.

Results

A screen for activity of a periplasmic phosphatase identifies genes required for A. baumannii OM barrier function

We identified strains with mutations in genes required for maintenance of the Acinetobacter baumannii OM barrier by screening transposon mutants for the development of a blue colony phenotype on agar plates containing the chromogenic substrate BCIP-Toluidine (XP). Although A. baumannii carries an endogenous periplasmic phosphatase enzyme, colonies remain white on agar plates containing XP. We reasoned that lesions in genes necessary for the OM barrier function should result in a blue colony phenotype, as the XP substrate becomes accessible to the periplasmic enzyme (Strauch and Beckwith, 1988; Lopes et al., 1972). Screening roughly 80,000 transposon-containing colonies for the blue colony phenotype yielded 364 blue colonies whose insertions were mapped to 58 unique genes (Supplementary file 1). We confirmed the results of the screen by assaying for OM-barrier defects using ethidium bromide (EtBr) and N-Phenyl-1-naphthylamine (NPN) uptake assays (Helander and Mattila-Sandholm, 2000; Murata et al., 2007). We also tested for resistance to antimicrobials, including trimethoprim, rifampicin, imipenem, carbenicillin, amikacin, gentamicin, tetracycline, polymyxin B, and erythromycin. Greater than 85% of the strains identified in the screen demonstrated decreased OM barrier function compared to wild type. Out of the 58 strains with transposon insertions, 23 demonstrated OM permeability defects by NPN and EtBr uptake assays, and 49 out of 58 resulted in increased sensitivity to two or more antibiotics compared to the parent strain, indicating that the screen identified lesions causing OM barrier defects leading to increased permeability to small charged and hydrophobic molecules, including commonly used antibiotics.

The Mla system is necessary for A. baumannii OM integrity

Four mutants with a blue colony phenotype contained unique transposon insertions in the genetic loci A1S_3103 and A1S_3102, predicted to encode core components (mlaF and mlaE) of a multicomponent ABC transport system. These genes are within a five-gene operon that encodes for a conserved proteobacterial ABC transport system homologous to the E. coli mla system previously implicated in OM integrity (Malinverni and Silhavy, 2009). The A. baumannii operon includes: mlaF and mlaE, respectively predicted to encode the nucleotide-binding and transmembrane domains of an ABC transporter; mlaD, encoding a protein containing an IM-spanning domain and a predicted periplasmic soluble domain; mlaC, encoding a soluble periplasmic protein; and mlaB, an additional gene predicted to encode a cytoplasmic sulfate transporter and anti-sigma factor antiagonist (STAS)-domain protein (Figure 1A). An additional putative OM lipoprotein is encoded on mlaA, or vacJ, which is clustered with the rest of the mla operon in some Gram-negative bacteria, although it is at a different chromosomal location in A. baumannii. MlaA has been functionally associated with the rest of the Mla components in E. coli, as mutations in mlaA yield comparable phenotypes to mutations in other components of the system (Strauch and Beckwith, 1988).

Figure 1 with 2 supplements see all
Disruption of the Mla system results in an altered outer membrane barrier.

(A) Genomic organization of the A. baumannii mlaFEDCB operon and its predicted products. Triangles indicate the position of four independent transposon insertions, isolated in a screen for genes involved in outer membrane integrity. (B) Ethidium bromide uptake assay of outer membrane permeability of ∆mla mutants and complemented strains. A.U., arbitrary units. Lines shown depect the average of three technical replicates. (C) Ethidium bromide uptake assay of outer membrane permeability following plasmid-based expression of MlaF, compared to its dominant negative version, MlaFK55L. Lines shown depict the average of three technical replicates. (D) Minimum inhibitory concentration (MIC) of select antibiotics in A. baumannii. *Indicates wild type A. baumannii containing pMMB plasmid constructs, and cultures grown with the addition of kanamycin (25 μg/mL) to maintain plasmids and 50 μM IPTG for induction.

https://doi.org/10.7554/eLife.40171.003

Bioinformatic analysis predicts that the mlaC and mlaF genes respectively encode the soluble periplasmic component and cytoplasmic ATPase component of the ABC transport system, and we chose to focus on mutants of these genes for further experiments to elucidate the function of the mlaFEDCB operon. Chromosomal deletions were created by allelic exchange, and these mutations resulted in OM permeability defects as measured by EtBr uptake assays. We complemented the OM defect for the ∆mlaC and ∆mlaF deletion mutants by repairing the original deletion event in the chromosome and confirmed complementation of the observed permeability defect (Figure 1B). Deletions in mlaF and mlaC also rendered A. baumannii increasingly sensitive to a variety of antibiotics as determined by MIC measurements (Figure 1D). Increased sensitivity to antibiotics whose uptake is not mediated by OM porins is consistent with a direct effect on the membrane component of the OM permeability barrier (Nikaido, 2003; Vaara, 1992). In addition to OM defects, the mla mutants display phenotypes that may correlate with OM stress, including increased production of extracellular carbohydrates as evidenced by crystal violet staining of pellicles following growth in broth culture (Figure 1—figure supplement 1A). These data indicate a role for Mla in the maintenance of the outer membrane barrier of A. baumannii.

ATPase activity of MlaF is required for maintenance of the OM barrier of A. baumannii

To exclude the possibility that the membrane defect was the result of the disruptive effect of a partially formed Mla protein complex, we engineered an enzymatically inactive ATPase component and expressed the defective enzyme from a plasmid. We reasoned that by expressing this allele in the wild type bacteria we could create a dominant-negative effect on Mla function. The cytoplasmic ATPase component of the Mla system, MlaF, contains the consensus sequence GxxxxGKT at residues 49–56, characteristic of a Walker A motif. Downstream residues 173–178 contain the sequence LIMYDE, typical of a Walker B motif. The Walker motifs form highly conserved structures critical for nucleotide binding and hydrolysis (Walker et al., 1982). The lysine residue of the Walker A motif is particularly essential for the hydrolysis of ATP. Mutations in this lysine residue are inhibited for nucleotide binding, and the mutated protein is rendered inactive (Hanson and Whiteheart, 2005). Additionally, ATPase mutants in the key lysine residue have been shown to have a dominant-negative effect on ATP hydrolysis when co-expressed with their wild-type ATPase counterparts, as typical ABC transporters have a structural requirement for two functional nucleotide-binding proteins which dimerize upon substrate transport (Davidson and Sharma, 1997; Wilkens, 2015).

Therefore, we created a version of the MlaF coding sequence with a leucine substitution of the Walker A lysine residue (MlaFK55L), and then cloned the mutated mlaF into the low-copy pMMBkan vector under control of the mlaF native promoter. We observed that expression of MlaFK55L in wild type A. baumannii had a dominant-negative effect on membrane permeability as measured by EtBr uptake (Figure 1C), and expression of MlaFK55L also resulted in increased exopolysaccharide production as demonstrated by increased staining by crystal violet (Figure 1—figure supplement 1B). Correspondingly, expression of MlaFK55L rendered A. baumannii more sensitive to a variety of antibiotics, resulting in reduced MICs when compared to A. baumannii expressing the empty pMMBkan vector (Figure 1D). Therefore, expression of a defective ATPase results in a dominant-negative mutant with a comparable phenotype to deletion of components of the mla operon. These results demonstrate a requirement for ATP hydrolysis by MlaF for the maintenance of OM barrier function in A. baumannii, and indicate that the phenotypes of deletion mutants were likely a result of a lack of transport function, rather than formation of a toxic incomplete membrane protein complex.

Structure of the A. baumannii MlaBDEF complex

The genetic arrangement and conservation of the components of this ATPase-containing transport complex indicated it was likely that the individual components formed a higher order protein structure. To define whether the Mla components form a stable protein complex, we expressed the entire operon (mlaFEDCB) from A. baumannii ATCC 17978 in E. coli with a carboxy-terminal hexahistidine tag on the MlaB component. Affinity purification of MlaB revealed three additional bands, with sizes corresponding to MlaF, MlaD, and MlaE (Figure 2—figure supplement 1) and confirmed by MALDI-TOF mass spectrometry analysis, indicating that these four proteins form a stable complex. We did not detect MlaC, suggesting it might interact only transiently with the other components, consistent with results recently reported by Thong et al. (2016).

We next used cryo-electron microscopy to characterize the architecture of the A. baumannii MlaBDEF complex (abMlaBDEF). This complex is uniformly dispersed in vitreous ice (Figure 2—figure supplement 2A), and 2D classification demonstrated the presence of a range of views suitable for structure determination (Figure 2—figure supplement 2B). Following 2D- and 3D-classification, we obtained a final dataset of ~14,000 particles with which we obtained a structure to a resolution of 8.7 Å (Figure 2—figure supplement 2D). The structure possesses significant visible features in agreement with the nominal resolution (Figure 2—figure supplement 2C). Based on the bioinformatically-predicted localization of individual proteins and work recently performed on the similar E. coli Mla complex (ecMlaBDEF) (Thong et al., 2016), we propose that MlaD localizes to the periplasmic side of the IM, MlaE forms the central transmembrane region, and MlaF and MlaB form the bottom layer on the cytoplasmic face of the IM with two visible hetero-dimers (Figure 2—figure supplement 2E). We note that the structure of ecMlaBDEF, at lower resolution, was reported recently (Ekiert et al., 2017). The overall features of both structures, solved independently, are identical, suggesting that they correspond to the correct structure for the complex. However, the limited resolution of the ecMlaBDEF complex structure did not allow modeling of its individual subunits, in contrast to the abMlaBDEF structure reported here.

We note that a clear six-fold symmetry is present for the region of the map attributed to MlaD (Figure 2B), despite the fact that we only imposed a 2-fold symmetry averaging. This agrees with the proposed hexameric state of its E. coli homologue (ecMlaD) (Thong et al., 2016). We next modeled abMlaD, using an evolution restraints-derived structural model of ecMlaD (Ovchinnikov et al., 2017) as a template, and used our previously-published EM-guided symmetry modeling procedure (Bergeron et al., 2013) to model its hexameric state. The obtained abMlaD hexameric model is at a low-energy minimum (Figure 2—figure supplement 3B) and fits the EM map density well (Figure 2B and Figure 2—figure supplement 4B). A crystal structure of the periplasmic domain of ecMlaD published recently (Ekiert et al., 2017) formed a crystallographic hexamer, suggesting that this corresponds to the native hexomeric arrangement for this domain. Our abMlaD hexameric model is very similar to the crystallographic ecMlaD structure (Figure 2—figure supplement 3C), supporting the proposed domain arrangement in the MlaBDEF complex. We note, however, that one region of density in the EM map is not accounted for by our MlaD hexamer model (Figure 2B). The localization of this extra density suggests that it corresponds to a ~ 45 amino-acid insert present between strands 4 and 5 of the abMlaD β-sheet (Figure 2—figure supplement 4A). The role of this insert, uniquely found in the A. baumannii orthologue, is not known.

Figure 2 with 4 supplements see all
Structure of the abMlaBDEF complex.

(A) Cryo-EM map of abMlaBDEF (grey), with structural models for MlaD, MlaB and MlaF (in magenta, cyan and green respectively) docked at their putative location. The density for the TM helices is clearly resolved. (B–D) Cartoon representation of the MlaD hexamer (B), the MlaB-MlaF hetero-tetramer (C), and the MlaE dimer (D) region of the abMlaBDEF atomic model. (E) Comparison of the MlaF domain arrangement in the EM map to that of the Maltose transporter ATPase MalK. The two chains of MlaD (in light and dark green) superimpose well to those of MalK (in cyan and dark blue) in the pre- translocation conformation (top, PDB ID: 4 KHZ), while a clear rotation is observed compared to the ATP-bound outward-facing conformation (bottom, PDB ID: 4KI0).

https://doi.org/10.7554/eLife.40171.006

We next modeled the structures of MlaB and MlaF and fitted their respective coordinates in the corresponding region of the EM map (Figure 2C and Figure 2—figure supplement 3A). For both proteins, most helices are well resolved, which allowed us to place the models unambiguously. We then compared the conformation of the ATPase MlaF to that of the maltose transporter ATPase MalK, which has been trapped in several conformations of the transporter; that is the inward-facing state, the pre-translocation state, and the outward-facing state (Khare et al., 2009; Oldham et al., 2013). Interestingly, the arrangement of MlaF clearly resembles the pre-translocation state of MalK (Figure 2D). This suggests that we have trapped a similar conformation of the abMlaBDEF complex. It is possible that MlaD and/or MlaF, for which there are no equivalent in other ABC transporters, stabilizes this conformation. Alternatively, it is possible that the presence of detergents, which were present to solubilize the complex, mimics the natural ligand in the transporter’s active site. Finally, the transmembrane (TM) region of the map is well resolved, and density for the transmembrane (TM) helices can be clearly identified. We therefore modeled abMlaE, using an evolution restraints-derived structural model of ecMlaE (Ovchinnikov et al., 2017) as a template, and fitted the obtained coordinates in the corresponding region of the map, with the orientation corresponding to the predicted topology. The resulting MlaE dimer model (Figure 2D) fits well to the EM map density (Figure 2—figure supplement 4C), and clearly corresponds to a closed transporter, with no solvent channel between the subunits. Interestingly, we also noted clear density for three TM helices that likely correspond to the MlaD N-terminal helices (Figure 3A). However, they lacked continuity, and we observed that only two form a direct interaction with MlaE. It is possible that this is due to heterogeneity in the orientation of MlaD relative to the rest of the complex. To verify this, we performed further 2D classification of the particles used for reconstruction (Figure 3B), which revealed a range of positions for the MlaD region relative to the rest of the complex. We therefore performed further 3D classification leading to a smaller dataset of ~8000 particles. This produced a structure of lower resolution (~11.5 Å) but with the six MlaD N-terminal TM helices clearly visible (Figure 3B). While the periplasmic domain possesses 6-fold symmetry, the TM domains of MlaD do not appear symmetrical, with two forming close contacts with the density attributed to MlaE while the other four do not appear to contact any other proteins. This observation likely explains the asymmetry of contacts between the dimeric MlaE and the hexameric MlaD. A higher-resolution structure will be required to determine if additional contacts are formed between the outward-facing loops of MlaE and the periplasmic domain of MlaD.

Figure 3 with 1 supplement see all
Localization of the 6 TM helices from MlaD.

(A) lateral section of the abMlaBDEF EM map, with the MlaE model in yellow. Density attributed to the MlaD N-terminal helices are indicated with a red star. (B) 2D classes generated from the set of particles used to generate the abMlaBDEF structure, corresponding to side views. A range of orientations for the periplasmic domain is observed. (C) Structure of abMlaBDEF, generated using a subset of the most homogeneous ~8000 particles. Some features of the map shown in Figure 3C are not present, but the overall structure is similar. Six well- defined helices in the central TM region are visible. (D) Sections along the vertical axis, corresponding to the three red lines shown in C, is shown on the left. The six- fold axis of MlaD is visible in the periplasmic region, but this breaks down in the TM region, where the six helices are asymmetric. An angular representation of the six helices at each three cross-sections is represented on the right.

https://doi.org/10.7554/eLife.40171.011

Components of the mla system interact directly with GPL

The crystal structure of MlaC has been solved from Ralstonia solanacearum. The structure contains a single phosphatidylethanolamine molecule oriented such that the hydrophobic acyl chains are located inside the protein while the hydrophilic head group is exposed (Huang et al., 2016). More recently, the crystal structure for MlaC has been solved from E. coli and shown to bind lipid tails (Ekiert et al., 2017). As noted in previous work performed on the E. coli Mla system, this is strong evidence that the substrates of the Mla system are GPL (Malinverni and Silhavy, 2009). In order to confirm that the periplasmic components of the Mla pathway in A.baumannii interact with GPL, we purified the soluble domains of both MlaC and MlaD by expressing histidine-tagged proteins followed by Ni-affinity FPLC purification. After overnight dialysis of the proteins, we performed Bligh-dyer chloroform extraction on the purified proteins to isolate any bound GPL and analyzed the results by LC-MS/MS. GPL analysis revealed both phosphatidylglycerol and phosphatidylethanolamine of varying acyl chain lengths. This suggests the possibility that the periplasmic substrate binding components of the system may bind diacylated GPL molecules with limited polar head group specificity (Figure 4—figure supplement 1).

Mla mutants have decreased abundance of outer membrane GPL

Given the OM defect of mla mutants, as well as the system’s apparent direct association with GPL, we chose to further characterize the overall membrane GPL composition of the mla mutants. Previous work on the Mla system in E.coli has demonstrated an increase in hepta-acylated lipid A in mla mutants, indicating activation of PagP that acylates GPL and lipid A in the outer leaflet of the OM in enterobacteria (Malinverni and Silhavy, 2009; Dalebroux et al., 2014). From this data it has been suggested that the system may serve to maintain lipid asymmetry within the OM, although it is well known that GPL displacement to the OM outer leaflet is a general reflection of chemical damage to the OM (Jia et al., 2004; Bishop et al., 2000; Dekker, 2000). However, biochemical analysis of the membrane GPL composition for mla mutants has not been published for any organism to our knowledge, so we sought to apply our lab’s methods of GPL quantification to test the hypothesis of retrograde transport function. To determine whether A. baumannii mla mutations cause changes in the membrane GPL concentration, GPL were extracted from inner and outer membrane fractions separated by density centrifugation. As can be seen on Figure 5—figure supplement 3, density centrifugation results in nice separation of the outer and inner membranes of Acinetobacter baumannii, with the OM contain the vast majority of OmpA and the inner membrane containing all the NAPPH oxidase. Thin-layer chromatography (TLC) and electrospray-ionization time-of-flight mass spectrometry (ESI-MS) were used to qualitatively assess GPL composition from these well separated membrane fractions. TLC and ESI-MS indicated ΔmlaC A. baumannii had a dramatically decreased abundance of all major phospholipid species in the OM compared to wild type. (Figure 4A and Figure 4—figure supplement 2).

Figure 4 with 2 supplements see all
Outer membrane glycerophospholipid levels are reduced in ∆mlaC mutant.

(A) Identification of inner and outer membrane phospholipids of wild type A. baumannii and ∆mlaC using 2D thin-layer chromatography. PE, phosphatidylethanolamine; PG, phosphatidylglycerol; CL, cardiolipin. (B) LC-MS/MS quantification of isolated inner and outer membrane glycerophospholipids. Error bars indicate ±s.e.m. (n = 3).

https://doi.org/10.7554/eLife.40171.013

To better analyze the differences in membrane GPL, we quantified GPL by normal phase liquid-chromatography collision-induced-dissociation mass spectrometry (LC-MS/MS). We quantified the ratio of individual GPL within each membrane by normalizing to an internal standard of known quantity. We then normalized the quantified GPL to the protein content of isolated IM and OM. Quantitative LC-MS/MS confirmed the overall reduction in outer membrane GPLs observed by ESI-MS and TLC, with the reduced levels observable across multiple GPL species for ΔmlaC mutants relative to wild type (Figure 4B). Therefore, mutations in the components of the Mla system result in a decrease in OM GPL, whereas the retrograde transport hypothesis would predict an increase in OM GPL. Therefore, these results instead suggest a possible role for Mla in outward GPL trafficking.

Mla mutants demonstrate an accumulation of newly synthesized GPL in the IM

The overall decrease in outer membrane glycerophospholipids of A. baumannii mla mutants suggests that either the Mla system is functioning to deliver GPLs from the inner membrane to the outer membrane, or alternatively, mutations in the Mla system may disrupt the outer membrane in a manner that leads to the activation of outer membrane phospholipases, which then degrade GPL. Work performed on the Mla system in E.coli has demonstrated that disruption of genes in the Mla pathway results in activation of both the OM acyl-transferase PagP, which cleaves a palmitate moiety from GPL and transfers it to LPS and PG, creating a hepta-acylated LPS molecule and palmitoyl-PG and the OM phospholipase PldA (Malinverni and Silhavy, 2009; Bishop et al., 2000). A. baumannii has no known PagP enzyme but similar activity of the multiple predicted OM phospholipases could account for the reduction in OM GPL as observed by TLC and quantitative mass spectrometry. Therefore, we designed a mass spectrometry-based assay to study intermembrane GPL transport using 13C stable isotope labeling (Figure 5—figure supplement 1A), to better analyze the directionality of GPL transport by the Mla system between the bacterial membranes. When grown in culture with sodium acetate as the sole carbon source, many bacteria directly synthesize acetyl-CoA using the conserved enzyme acetyl-CoA synthase (Kumari et al., 2000). Acetyl CoA, the precursor metabolite for fatty acid biosynthesis, is first converted to malonyl-CoA and enters the FasII (fatty acid biosynthesis) pathway that supplies endogenously synthesized fatty acids to macromolecules such as lipopolysaccharides, phospholipids, lipoproteins, and lipid-containing metabolites. By growing cultures in unlabeled acetate then ‘pulsing’ with 2-13C acetate and analyzing separated membrane fractions from set time points, we can observe the flow of newly synthesized GPLs between the IM and OM of A. baumannii (Figure 5—figure supplement 1B) (Dalebroux et al., 2014).

Upon introducing the 2-13C acetate as the sole carbon source, 13C-labeled GPL were immediately synthesized in the bacterial cytoplasm. We reasoned that continued growth in 13C acetate should result in a mixed pool of unlabeled and labeled IM GPL molecules. As the GPL are then fluxed from the IM to the OM, the likelihood that an individual GPL molecule is transported is directly proportional to the ratio of labeled to unlabeled GPL in the IM pool. As the bacteria continue to grow in 13C acetate, the ratio of labeled to unlabeled GPL in the IM will gradually increase as new GPL are synthesized and inserted in the IM. As such, the likelihood of transporting labeled GPL to the OM will also increase. A comparison of the ratios of labeled to unlabeled GPL in the IM and OM will thus reflect the efficiency of transport between the membranes, and analysis of transport in wild type A. baumannii will establish reference for transport efficiency with which to compare our mutants. Additionally, OM phospholipases, some of which may be activated upon membrane damage (Istivan and Coloe, 2006), will not distinguish between labeled and unlabeled GPL and therefore will not affect the ratio of labeled to unlabeled GPL obtained from this assay.

Membrane separation and analysis of wild type A. baumannii revealed near-identical rates-of-change between the two membranes in ratios of 13C-labeled to unlabeled GPLs, indicating that newly synthesized GPLs are transported and inserted into the OM at a rate equivalent to their rate of synthesis and assembly within the IM. Furthermore, the ratios of labeled to unlabeled GPLs were nearly equal in the IM compared to the OM at the time points evaluated, indicating that GPL transport likely occurs rapidly, consistent with earlier pulse-chase experiments performed in E. coli that estimate the half-life of translocation of various GPLs at between 0.5 and 2.8 min (Donohue-Rolfe and Schaechter, 1980). By contrast, mutants in the Mla system accumulate newly synthesized GPLs in their IM at a greater rate than occurs in the OM as evidenced by the increasing disparity in the ratio of labeled to unlabeled GPLs between the IM and OM over time (Figure 5A). The discrepancy in ratios of labeled to unlabeled GPLs between the IM and OM of ∆mlaF is apparent for PG and PE of varying acyl chain lengths corresponding to the most naturally abundant species C16:0/C16:0, C18:1/C18:1, or C16:0/C18:1 (Supplementary file 2). Further, the effects of MlaFK55L expression on GPL trafficking were similar to what was observed in the ∆mlaF strain (Figure 5B). Therefore, ATP hydrolysis by MlaF appears to be a requirement for extraction of these GPLs from the IM of A. baumannii for subsequent transport to the OM.

Figure 5 with 3 supplements see all
Newly synthesized glycerophospholipids accumulate at the inner membrane of Mla mutants.

(A) LC-MS/MS quantification of 13C labelled/unlabeled glycerophospholipids in isolated membrane fractions over time after growth in 2–13C acetate in ∆mlaF and complemented strain. Facet labels on the right indicate the specific glycerophospholipid species analyzed and the acyl chain length. PG, phosphatidylglycerol; PE, phosphatidylethanolamine. Shown is representative data from repeated experiments. (B) LC-MS/MS quantification of 13C labelled/un- labeled glycerophospholipids in isolated membrane fractions following plasmid-based expression of MlaF compared to its dominant negative version, MlaFK55L. Facet labels on the right indicate the specific glycerophospholipid species analyzed and the acyl chain length. PG, phosphatidylglycerol; PE, phosphatidylethanolamine. Shown is representative data from repeated experiments. (C) Relative proportion of newly synthesized GPL on IM and OM after one hour growth in 2–13C acetate. Error bars represent ± s.d. (n = 2). Statistical analyses performed using a Student’s t test. p-Value: *, p < 0.05; **, p < 0.01.

https://doi.org/10.7554/eLife.40171.016

To better characterize the role of the periplasmic substrate binding component MlaC, we performed similar stable isotope pulse experiments to observe the flow of newly synthesized GPLs in the ∆mlaC strains. Stable isotope experiments on ∆mlaC mutants reveal IM accumulation of newly synthesized GPLs similar to the result in ∆mlaF mutants (Figure 5—figure supplement 2A), indicating that in the absence of the periplasmic component GPLs are not efficiently removed from the IM by the remainder of the Mla system. We also sought to characterize the potential role of the putative OM-lipoprotein MlaA, which has been implicated as a component of the Mla system in E. coli. A chromosomal deletion strain of mlaA was created by allelic exchange, and complemented by expression of MlaA from a pMMB67EH-Kan plasmid. The results of the stable isotope pulse experiments in the ∆mlaA strain revealed results consistent with those obtained from ∆mlaC and ∆mlaF, in which the ratio of labeled to unlabeled GPL is consistently higher in the inner membrane than the outer membrane after one hour of exposure to 13C-acetate (Figure 5—figure supplement 2B and C). These results are consistent with a model in which the IM-localized ABC transporter complex MlaBDEF first transfers GPLs to the periplasmic binding protein MlaC, which then transports GPL to the OM, whereupon MlaA facilitates GPL insertion into the OM.

Discussion

We performed a screen to identify A. baumannii proteins that are essential for its OM barrier that led to the identification of an ABC transport system whose ATPase activity maintains OM barrier function. IM and periplasmic components of this system can be purified, bind GPLs, and assemble into a defined protein complex with significant symmetry, indicating that this system could function to transport GPLs from the IM to the OM. Consistent with the possibility that Mla functions as an anterograde transporter, the OM of mutants show an overall reduction of GPL along with an excess accumulation of newly synthesized GPL on the IM. Therefore, these results lead us to propose that the function of the A. baumannii Mla system is the trafficking of GPL from the IM, across the periplasm, for delivery to the outer membrane (Figure 6). According to this model, ATP hydrolysis by MlaF provides the initial energy to extract GPL from the IM, while the substrate binding components MlaD and MlaC take up lipids for delivery to the OM. It has been observed by van Meer and colleagues that complete extraction of GPLs from the membrane bilayer requires a Gibbs free energy of ~100 kJ/mol (Abreu et al., 2004; van Meer et al., 2006), whereas ATP contains just 30 kJ/mol of energy. To account for the energy difference, a hydrophobic acceptor molecule is proposed to allow the lipids to fully dissociate from the rest of the ABC transporter and facilitate complete removal from the bilayer. The GPL-binding component, MlaD, contains an IM spanning domain and is shown here, and in orthologous systems, to be in complex with the MlaE and MlaF proteins within the IM (Ekiert et al., 2017; Roston et al., 2012). The close association of MlaD with the outer leaflet of the IM may allow it to extract lipids from the IM by hydrophobic interaction with the acyl chains after ATP hydrolysis by MlaF. Subsequent trafficking across the periplasm then involves the periplasmic GPL binding protein MlaC, which likely accepts GPL from MlaD and then carries them to the OM. We note however the observed effect of mlaC deletion on GPL accumulation in the IM, while statistically significant for most of the analyzed diacyl-glycerophospholipids, appears to be less than that of deletion of the ATPase component (Figure 5C), suggesting that while MlaC may participate in transfer to the OM, there may be redundant mechanisms by which the IM complex can transport or remove IM GPL in the absence of MlaC. While the precise mechanism of GPL insertion into the OM is not yet known, work performed on the E.coli Mla system has shown that MlaC interacts with both the IM MlaFEDB complex, as well as with the putative OM lipoprotein MlaA, and our results support a role for MlaA in the function of the overall Mla system and delivery of GPL to the OM.

The multicomponent Mla system transports glycerophospholipids from the inner membrane to the outer membrane of A. baumannii.

A schematic of glycerophospholipid transport to the Gram-negative bacterial outer membrane by the Mla system.

https://doi.org/10.7554/eLife.40171.020

In this work, we designed a method to monitor lipid transport between Gram-negative bacterial membranes using stable 13C isotope labeling. We considered the possibility of loss of GPL from the outer membrane due to outer membrane vesicle formation or, more likely, by the possible increased activation of outer membrane phospholipases. Both would have the effect of removing GPL from the outer membrane and would result in lower GPL levels in the outer membrane of mla mutants. Neither of these mechanisms would operate specifically on either labeled or unlabeled lipids. We understood the decrease in outer membrane GPL to be insufficient evidence of an anterograde transport function for mla, and for this reason we developed the stable isotope assay to control for these possibilities. The stable isotope assay gives insight into whether these results are due simply to mislocalization or degradation of outer membrane GPL, or if they can in fact be attributed to deficient anterograde GPL transport. This is because the peak intensity of each newly synthesized, C13-labeled GPL is normalized to the corresponding unlabeled version of that GPL species with each sample injected into the LC-MS/MS. The result is a ratio of labeled to unlabeled GPL for every membrane sample. With C13-acetate as the sole carbon source, we observe a gradual increase in the ratio of labeled GPL relative to unlabeled GPL over time. In wild type bacteria, these ratios for the inner and outer membranes track closely over time, which indicates that under these conditions GPL transport from the inner to the outer membrane occurs quite rapidly. In mla mutants the ratio is both higher and typically increases at a greater rate in the inner membrane. Phospholipases and budding outer membrane vesicles will not distinguish between labeled and unlabeled GPL species, and so will not impact the ratios obtained with this assay.

Our results using this assay are consistent with the Mla system functioning as an anterograde GPL transporter, however they do not exclude the possibility of a dual role for Mla components in the maintenance of OM lipid asymmetry. Previous work performed on the orthologous Mla system in E.coli has been interpreted to suggest that the function of the system is to remove GPL from the outer leaflet of the OM for retrograde transport back into the cytoplasm based on the observation that E.coli mla mutants likely contain GPLs on the outer leaflet of the OM. (Malinverni and Silhavy, 2009; Benning, 2008). This proposed function was inferred from the observation that gene deletions resulted in an increased activation of the OM-phospholipase enzymes PagP and OMPLA, suggesting an increased amount of GPL in the outer leaflet of the OM (Malinverni and Silhavy, 2009). The interpretation of retrograde transport function was also based on the existence of an orthologous system in plant chloroplasts that transports lipids from the endoplasmic reticulum (ER) into the organelle. Many plants require this retrograde transport function because certain lipids in the chloroplast thylakoid membrane derive from GPL originating in the ER (Hurlock et al., 2014). However, since Gram-negative bacteria synthesize GPL within the IM, retrograde transport of GPL would only be necessary for the recycling of GPL mislocalized to the OM outer leaflet. Although this is a reasonable inference based on data available at the time, we would point out that the directionality of transport by the E. coli Mla system had not been thoroughly probed experimentally using membrane analysis or with a functional assay of the type performed here. It is conceivable that the import function of the orthologous chloroplast TGD system is a result of adaptation to the intracellular environment, the system in this case having evolved to aid in the transfer of GPL from the nearby ER to the chloroplast. Furthermore, while it is possible that the Mla system in E. coli serves a different primary function than in A. baumannii, we demonstrate here that both complexes possess a similar architecture, pointing to a conserved function. The outer membrane defect phenotypes observed in E. coli mla mutants might also be explained by a disruption of OM structure stemming from decreased concentrations of OM GPL, leading to activation of the PagP enzyme. It is well established that for E. coli, GPL displacement to the OM outer leaflet and subsequent activation of these enzymes reflects OM instability and can be achieved by chemical disruption of the bilayer (Jia et al., 2004; Bishop et al., 2000; Dekker, 2000). It may be the case that the OM of E. coli mla mutants contain GPL in the outer leaflet, but the possibility remains that OM GPL can flip into the outer leaflet under conditions of OM damage resulting from an imbalance of LPS-to-GPL ratios, along with perhaps the corresponding disruption of OM proteins. However, final determination of the directionality of GPL transport by the Mla system in E.coli and other organisms will require intermembrane transport studies similar to what has been done here for A. baumannii, along with studies similar to those performed for the Lpt LPS transport system for which molecular transfer of LPS from molecule to molecule of the Lpt system is functionally defined.

Following the introduction of the retrograde transport model for Mla function into the existing literature, a number of studies have examined the phenotypic effects of Mla disruption in various organisms. In a recent study in PNAS, Powers and Trent first obtained A. baumannii deficient in lipooligosaccharide (LOS) by selection in the presence of polymyxin B (Powers and Trent, 2018). They then performed an evolution experiment, passaging the strains in cultures containing polymyxin B over 120 generations, at which point they observed significantly improved growth in the populations. These evolved populations were also observed to have increased resistance to antibiotics including vancomycin, bacitracin, and daptomycin, and to appear more morphologically consistent relative to the unevolved strains when observed microscopically. Whole genome sequencing of the evolved strains revealed mutations in mla genes in seven of the 10 evolved populations. They also observed frequent disruptions in pldA, as well as in other envelope genes. To further study these effects, they then introduced clean deletions of mlaE and pldA to ATCC 19606, and selected for LOS-deficient bacteria by plating on polymyxin B. These double mutants demonstrated improved growth and resistance to antibiotics but continued to display altered cellular morphology. The authors present their data as evidence in support of Mla as a retrograde transport system, and assume that a lack of removal of GPL from the outer leaflet is promoting the OM barrier. We would point out that lacking in their data is examination of the membrane glycerophospholipid (GPL) profile in their LOS-deficient mutants. It is assumed by the authors that mla and pldA mutations have the effect of stabilizing a symmetric outer membrane produced in the absence of LOS by allowing GPL to fill in the outer leaflet, resulting in improved growth and antibiotic resistance. Given that the data suggests that Mla and PldA are selected against when LOS is absent, examination of the outer membrane GPL content might have supported the authors’ conclusions if it revealed an increase in GPL in mla and pldA mutants. Absent such data, it is not obvious to us that the authors have sufficiently ruled out alternative explanations for their observed phenomena. For example, we would question the mechanisms regulating the homeostasis of both the inner and outer membranes and the entire periplasmic space in the absence of LOS. The authors acknowledge earlier work that observed an increase in expression of mla genes upon initial loss of LOS in 19606 (Henry et al., 2012; Boll et al., 2016). Genes in the mla pathway were shown to have an up to 7.5-fold increase in gene expression upon loss of LOS. Powers and Trent assert that the function of Mla is deleterious in the absence of LOS, but perhaps what is deleterious is the profound upregulation of mla expression in the absence of LOS, combined with an active PldA. If Mla is an anterograde transporter, we can imagine this might create a situation in which GPL are rapidly removed from the inner membrane and then degraded in the outer membrane in excess of what the cell can support and limiting both of these processes together simply allows the cell to achieve a new homeostasis. Understanding of the myriad processes regulating bacterial outer membrane assembly and integrity remains limited even when LOS is present, and so interpreting results such as these as providing direct evidence of function may exceed the limits of the data.

The gene for MlaA, the proposed OM component, is at a different chromosomal location from the remainder of the mla operon. Recent structural studies on MlaA have revealed that MlaA forms a ring-shaped structure localized the inner leaflet of the OM, and have shown it to form stable complexes with the outer membrane proteins OmpF and OmpC (Abellón-Ruiz et al., 2017). The proposed structure of MlaA in the OM supports the argument that MlaA is involved in removal of GPL from the outer leaflet, and it is suggested that GPL from the outer leaflet travel through a pore in MlaA where they are received by MlaC, yet our data reveals that A. baumanniimlaA mutants are defective in delivery of GPL from the IM to the OM. These data can be reconciled by a model in which MlaA functions both to remove mislocalized GPL from the outer leaflet of the OM, and additionally serves to facilitate delivery of GPL to the OM by MlaC, perhaps by enabling MlaC localization to the surface of the inner leaflet. By this model, mutations in MlaA will be phenotypically similar to mutations in other components of the Mla system, and we would expect to observe a decreased rate of anterograde GPL transport. We would here point out that while previous work has implicated the Mla system in the maintenance of OM lipid asymmetry through observation of increased activity of PagP, the role of the MlaFEDB complex and MlaC in retrograde GPL transport has previously only been inferred from homology to the chloroplast TGD system. It is established that cellular mechanisms exist in Gram-negative bacteria to resist stressful conditions that lead to OM disruption. For example, OM phospholipase enzymes, such as PldA, are activated under conditions of membrane stress to digest GPL in the outer leaflet of the OM, as high levels of GPL in the outer leaflet destabilize the OM barrier function. The model of retrograde GPL transport by the Mla system proposes that growing cells expend cellular energy in the form of ATP in order to transport undigested GPL from the OM, across the periplasm, and back into the IM, at which point some of those same molecules will be transported back to the OM by an unknown mechanism. However, the available data points most clearly to a model of anterograde GPL transport by MlaFEDB and MlaC, facilitated in some way by MlaA.

The first three genes of the mla operon – comprising an ATPase, permease, and substrate-binding components of the ABC transporter complex – are conserved in Mycobacteria spp, Actinobacteria, and chloroplasts, while the entire five-gene operon appears to be conserved in Gram-negative bacteria (Casali and Riley, 2007). Given the conservation of the system across Gram-negative species, our results may shed light on a generalized mechanism contributing to OM biogenesis. Additionally, we have here demonstrated that the function of this ABC transport system is crucial for maintaining the integrity of the A. baumannii OM. The fact that mla mutations are tolerated, and that levels of OM GPL are reduced but not abolished, suggests the intriguing possibility of additional undiscovered mechanisms of GPL delivery to the OM. Also of interest is the potential role of the increased exopolysaccharide observed upon disruption of the Mla system. It is possible this exopolysaccharide plays a partially compensatory role in A. baumannii resulting from decreased OM GPL, given that recent work has shown that A. baumannii exopolysaccharides can contribute to antibiotic resistance, likely through improved barrier function (Geisinger and Isberg, 2015).

The progression towards a more complete understanding of intermembrane GPL transport and OM barrier function should ultimately have relevance in the development of novel drug targets to undermine emerging antibiotic resistance in Gram-negative pathogens. The emergence of antibiotic resistant Gram-negative bacteria for which few or no antibiotics are available therapeutically is an important medical concern. This issue is typified by current isolates of A. baumannii that can only be treated with relatively toxic colistin antibiotics. This has led many individuals and agencies to propose the development of single agent antimicrobials which could be used for organisms such as A. baumannii and P. aeruginosa that have significant antibiotic resistance. Therefore, work furthering the understanding of the OM barrier could lead to the development of drugs which target the barrier and allow the therapeutic use of many current antibiotics.

Materials and methods

Key resources table
Reagent type
(species)
or resource
DesignationSource
or reference
IdentifiersAdditional
information
Gene (Acinetobacter baumannii)mlaCNAGenbank Accession: AKA30172.1
Gene (A. baumannii)mlaFNAGenbank Accession: AKA30169.1
Gene (A. baumannii)mlaENAGenbank Accession: AKA30170.1
Gene (A. baumannii)mlaDNAGenbank Accession: AKA30171.1
Gene (A. baumannii)mlaBNAGenbank Accession: AKA30173.1
Gene (A. baumannii)mlaANAGenbank Accession: AKA32955.1
Gene (A. baumannii)phoUNAGenbank Accession: AKA33305.1
Strain, strain background (A. baumannii)Acinetobacter baumannii ATCC 17978Pelletier et al., 2013Baumann et al., 1968Source: ATCCGenBank ACCESSION: CP000521
Genetic reagent (A. baumannii)ATCC 17978 ∆phoUThis paperChromosomal deletion in ATCC 17978 by allelic exchange using pEX2tetRA vector
Genetic reagent (A. baumannii)∆mlaFThis paperChromosomal deletion in ATCC 17978 by allelic exchange using pEX2tetRA vector
Genetic reagent (A. baumannii)mlaCThis paperChromosomal deletion in ATCC 17978 by allelic exchange using pEX2tetRA vector
Genetic reagent (A. baumannii)mlaAThis paperChromosomal deletion in ATCC 17978 by allelic exchange using pEX2tetRA vector
Antibodyanti-OmpA (rabbit polyclonal)This paperMade to purified OmpA by GenScript Biotech Corp (1:1000)
Recombinant DNA reagentpMarKT (plasmid)This paperProgenitors: C9 Himar (PCR), TetRA from Tn10 (PCR), pACYC184, pBT20.
Recombinant DNA reagentpEX2tetRA (plasmid)This paperProgenitors: tetRA (PCR), pEXG2
Recombinant DNA reagentpMMBkan (plasmid)This paperKanamycin resistance marker inserted at DraI site of pMMB67EH
Recombinant DNA reagentpMMBkan:MlaF (plasmid)This paperpMMBKan expressing mlaF
Recombinant DNA reagentpMMBkan:MlaFK55L (plasmid)This paperpMMBKan expressing Walker box mutant of mlaF
Recombinant DNA reagentpET28a:MlaFEDCB-His (plasmid)This paperpET28a expressing MlaFEDCB with C-terminal HISX6 tag on MlaB.
Recombinant DNA reagentpET15b-mlaC-SD-His (plasmid)This paperpET15b expression vector containing MlaC soluble domain with C-terminal HisX6 tag.
Recombinant DNA reagentpET15b-mlaD-
SD-His (plasmid)
This paperpET15b expression vector containing MlaD soluble domain with C-terminal HisX6 tag.
Chemical compound, drug2–13C acetateCambridge Isotope Laboratories, Inc.
Chemical compound, drugBCIP-Toluidine (XP)Gold BiotechnologyB-500–10
Chemical compound, drugN-Phenyl-1-naphthylamine (NPN)Sigma Aldrich104043–500G
Chemical
compound, drug
NADHSigma Aldrich606-68-8
Chemical compound, drugCCCP (Carbonyl cyanide 3-chlorophenylhydrazone)Sigma-AldrichC2759-1G
Software, algorithmMotionCorr2Thong et al., 2016Dr. Agard Lab, University of CA San Francisco
Software, algorithmCTFFIND4Rohou and Grigorieff, 2015Dr. Grigorieff Lab, University of MA Medical Center
Software, algorithmAppionLander et al., 2009Dr. Carragher Lab, The Scripps Research Institute
Software, algorithmRelion 2Scheres, 2012Dr. Scheres Lab, MRC Lab of Molecular Biology
Software, algorithmEMAN2Tang et al., 2007Dr. Ludtke Lab, Blue Mountain College
Software, algorithmChimeraPettersen et al., 2004UCSF Resource for Biocomputing, Visualization, and Informatics
Software, algorithmModellerWebb and Sali, 2016Dr. Sali Lab, University of CA San Francisco
Software, algorithmRosettaDiMaio et al., 2011Dr. Baker Lab, University of WA

Bacterial strains

Transposon mutagenesis and subsequent chromosomal deletions of mla genes were performed in Acinetobacter baumannii ATCC 17978.

A Mariner-based transposon vector for use in Acinetobacter baumannii:

To perform transposon mutagenesis a Mariner-based transposon vector was designed for use in Acinetobacter baumannii ATCC 17978. The new transposon vector, derived from pBT20, termed pMarKT, contains an outward facing pTac promotor as well as a selectable kanamycin resistance marker followed by an omega terminator within the Mariner arm sites (Kulasekara et al., 2005). The plasmid backbone contains the Mariner transposase gene C9 Himar, a tetRA resistance marker from Tn10, a p15A origin from pACYC184, and an oriT site for mobilization. The plasmid was constructed by PCR of select fragments followed by restriction digest and ligation of the cleaved ends. The new transposon vector was confirmed by restriction digest and partial sequencing.

Transposon mutagenesis

Initial mutagenesis revealed that many hits occurred in the high affinity phosphate uptake transcriptional repressor phoU (A1S_0256). Subsequent rounds of mutagenesis were conducted on an ATCC 17978 phoU chromosomal deletion strain, and plated on high phosphate media to reduce the background level of cleavage of the chromogenic substrate. Chromosomal deletions were performed by allelic exchange using a pEX2tetRA vector, which was created by insertion of the tetRA tetracycline resistance marker from Tn10 into the pEXG2 plasmid (Rietsch et al., 2005). Roughly 1000 bp regions upstream and downstream of the genes of interest were amplified for homologous recombination with the ATCC 17978 chromosome. Sucrose was used to counter-select against cells retaining the pEX2tetRA backbone, and deletions were confirmed by PCR. Complementation of deletions was accomplished by repairing the original deletion in the chromosome, again using the pEX system and allelic exchange.

Donor E. coli containing the pMarKT transposon vector were suspended in LB broth to an OD600 of 40 and mixed with an equal volume of the recipient A. baumannii suspended to OD600 of 20. 50 µL aliquots of this mixture were then plated in spots on a dried LB agar plate and incubated for 2 hr at 37°C (Kulasekara et al., 2005). Each 50 µL spot resulted in about 80,000 colonies of A. baumannii containing Mariner transposon insertions. The mutants were plated on LB agar containing 1X M63 salts, 50 µg/mL kanamycin, 30 µg/mL chloramphenicol, and 40 µg/mL XP substrate. Plates were incubated for at least 36 hr at 30°C to allow for the appearance of the blue color from cleavage of the XP substrate. Sequencing of the transposon insertions was adapted from the method described in Chun et al. (1997), including semi-arbitrary two-step PCR amplification of transposon regions followed by sequencing.

Ethidium bromide uptake assay

Bacteria were grown in 5 mL LB cultures to mid-log OD600 (0.3–0.6), then spun down and normalized in PBS to OD600 0.2. Prior to measurement, CCCP was added at 200 µM to inhibit the activity of efflux pumps. Ethidium bromide was added immediately prior to measurement to final concentration of 1.2 µM in 200 µL total reaction volume. Permeability was assessed using a PerkinElmer EnVision 2104 Multilabel Reader using a 531 nm excitation filter, 590 nm emission filter, and a 560 nm dichroic mirror. Readings were taken every 15 s for 30 min with samples assessed in triplicate in a Greiner bio-one 96-well flat bottom black plate.

MIC measurements

MICs were determined in 96-well microtiter plates using a standard two-fold broth dilution method of antibiotics in LB broth. The wells were inoculated with 104 bacteria per well, to a final well volume of 100 μL, and plates were incubated at 37°C with shaking unless stated otherwise. Experiments were performed thrice using two technical replicates per experiment. MICs were interpreted as the lowest antibiotic concentration for which the average OD600 across replicates was less than 50% of the average OD600 measurement without antibiotic.

Crystal violet assay for exopolysaccharide production

Strains were inoculated to OD600 0.05 and grown overnight at 37°C in 2 mL LB broth with shaking in glass tubes. The next day, liquid was carefully decanted and the tubes left to dry for 2 hr at 37°C. Pellicles were stained with the addition of 0.1% crystal violet, then gently washed three times in dH2O. Crystal violet was solubilized in a 80:20 solution of ethanol:acetone and read at 590 nm. P values were determined from a Student’s t-test over three biological replicates per sample.

MlaFEDB protein expression and purification

The mlaFEDCB operon from the genome of A. baumannii ATCC 17978 was subcloned into the pET-28a vector (Novagen, US) with a hexahistidine (−6HIS) tag fused at the C-terminus of the MlaB protein. The nucleotide sequence of the operon was confirmed using DNA sequencing. The plasmid was transformed into E. coli RosettaBlue strain. Cells were grown at 37°C in LB medium until the cell density reached an OD600 of 1.0. The temperature was then reduced to 16°C before induction with 1 mM isopropyl β-D-thiogalactoside (IPTG). After growth at 16°C for 18 hr, cells were harvested by using centrifugation at 4,200 g. Cells were resuspended in ice-cold buffer A (20 mM Tris-HCl (pH 8.0), 150 mM NaCl, 5% (v/v) glycerol) and subjected to three runs of homogenization at 10,000–15,000 psi using Avestin EmulsiFlex-C3 high pressure homogenizer (Avestin, Ottawa, Ontario, Canada). The homogenate was centrifuged at 17,000 g for 10 min at 4°C, and then the supernatant was ultra-centrifuged at 100,000 g for 60 min. The membrane fraction was resuspended in buffer A supplemented with 1% (w/v) dodecyl-β-d-maltopyranoside (DDM) and was slowly stirred for 1 hr at 4°C. After another ultra-centrifugation at 100,000 g for 30 min, the supernatant was collected and loaded on 2 ml of Ni2+-nitrilotriacetate affinity resin (Ni-NTA from Qiagen, Germany) pre-equilibrated with buffer A supplemented with 5 mM imidazole and 0.025% (w/v) DDM. After incubating for 1 hr, the resin was washed with 50 ml buffer A supplemented with 20 mM imidazole and 0.025% (w/v) DDM. The protein sample was eluted with 10 ml elution buffer containing buffer A, 300 mM imidazole, and 0.025% (w/v) DDM, and was concentrated to 0.5 ml. The concentrated protein sample was then loaded onto a Superdex-200 column (10/30, GE Healthcare, US) pre-equilibrated with 20 mM Hepes (pH 7.0) 150 mM Nacl 0.025% DDM. Peak fractions were collected and the pooled protein sample was concentrated to 1 mg/ml.

Cryo-EM sample preparation, data acquisition and image processing

Purified Mla complex at ~1 mg/ml was applied to glow-discharged holey grids, blotted for 6 s, and plunged in liquid ethane using a Vitrobot (FEI). Images were acquired on a FEI Tecnai G2 F20 200 kV Cryo-TEM equipped with a Gatan K-2 Summit Direct Electron Detector camera with a pixel size of 1.26 Å/pixel. 500 micrographs were collected using Leginon (Suloway et al., 2005) spanning a defocus range of −1 to −2 µm.

Movie frames were aligned with MotionCorr2 (Zheng et al., 2017) and the defocus parameters were estimated with CTFFIND4 (49). 333 high-quality micrographs were selected by manual inspection, from which ~ 55,000 particles were picked with DOG in Appion (Lander et al., 2009). Particle stacks were generated in Appion using a box size of 200 pixels. Several successive rounds of 2D and 3D classification were performed in Relion 2 (Scheres, 2012; Kimanius et al., 2016) using an initial model generated by Common Lines in EMAN2 (53) leading to a final stack of ~14,000 particles for 3D structure refinement in Relion.

Structure modeling and docking in the EM density

The structures of MlaB and MlaF were modeled using the threading server Phyre (Kelley et al., 2015) based on the structures of the anti-sigma factor antagonist tm1081 (PDB ID 3F43, 18% sequence identity to MlaB) and the ABC ATPase ABC2 (PDB ID 1OXT, 36% sequence identity to MlaF) respectively. Two copies of each structural model were positioned in their putative location within the EM map using Chimera (Pettersen et al., 2004) and their position was optimised using the Fit to EM map option. The abMlaD and abMlaE structures were modelled on ecMlaD and ecMlaE structural models deposited in the Gremelin database (Ovchinnikov et al., 2017), using Modeller. For abMlaD, the N-terminal TM helix and the insert region were modelled ab initio using the Rosetta suite (DiMaio et al., 2011) and positioned in their putative localisation in Chimera. The MlaD hexamer, as well as the MlaE dimer, were modelled with Rosetta using a EM-guided symmetry modelling approach described previously (Bergeron et al., 2013). The final model was refined with Rosetta.

Membrane isolation and separation

Cells were resuspended in 20 mL of 0.5 M sucrose, 10 mM Tris pH 7.8, 75 µg freshly prepared lysozyme (Roche 10837059001), and 20 mL of 0.5 mM EDTA, and kept on ice with gentle stirring for 20 min. Samples were homogenized (Avestin EmulsiFlex-C3) and spun down at 17,000 g for 10 min to removed un-lysed cells prior to ultracentrifugation. Membranes were spun down using a Ti45 Beckman rotor at 100,000 g for 1 hr and then added to the top of a sucrose gradient. IM and OM were separated by 18 hr ultracentrifugation using a SW-41 rotor in a Beckman Coulter Optima L90X ultracentrifuge. Spheroplast formation and sucrose gradient separation of IM and OM was adapted from a method by Osborn et al. (1972) by use of a defined 73%–53–20% sucrose gradient as described in Dalebroux et al. (2015). Our sucrose gradients contain three distinct concentrations of sucrose, and inner and outer membranes separate into distinct bands that are collected individually (Figure 5—figure supplement 3F). To limit any potential mixing of the membranes, the inner membrane is collected from the top of the tube while the outer membrane is collected by puncturing the bottom of the tube and allowing the bottom band to be collected. The purity of membrane separation by this method was confirmed by NADH assay and by Western blotting for the A. baumannii OM-localized OmpA protein, with 10 µg of total protein loaded into each lane as measured by Bradford protein assay (Figure 5—figure supplement 3).

GPL extraction and TLC

GPLs from isolated membranes were extracted using a 0.8:1:2 ratio of water: chloroform: methanol as per the method of Bligh and Dyer (Bligh and Dyer, 1959). Two-dimensional TLC was performed using silica gel 60 plates and immersion in Solvent System A (60:25:4 CHCl3:CH3OH:H2O), followed by Solvent System B (80:12:15:4 CHCl3:CH3OH:CH3COOH:H2O) in the orthogonal direction.

MlaC and MlaD protein purification and GPL extraction

Primers were designed to amplify the mlaC gene of ATCC 17978, excluding the signal sequence for export from the cytoplasm, and the periplasmic domain of mlaD of ATCC 17978, excluding the membrane-spanning domain. These fragments were cloned into pET29b and expressed with a carboxy-terminal hexahistidine (−6HIS) tag in BL21 E. coli with 2 hr induction. Cells were pelleted and resuspended in Tris-buffered saline containing 10% glycerol (TBSG) and protease inhibitor cocktail (Roche, Complete EDTA-free). Cells were lysed by homogenization (Avestin) and ultracentrifuged at 100,000 g for 1 hr to spin down membranes. The supernatants were then applied to a 5 mL-HiTrap(TM) Chelating HP Ni-affinity column pre-loaded with 0.1 M NiSO4 and equilibrated with TBSG. The proteins were eluted from the column using FPLC (Akta) by applying a stepwise gradient of 25 mM, 50 mM, and finally 300 mM imidazole for protein elution. Elution was monitored by UV-absorption at 280 nm. The MlaC- and MlaD-containing fractions were then further purified by injecting into a HiLoad 120 ml-6/600 Superdex(TM) 200 preparative grade size-exclusion column equilibrated in TBSG using a flow rate of 1 mL/min. The purity of the collected protein fractions was confirmed by SDS polyacrylamide gel electrophoresis. Proteins were diluted to 2 mg/mL and dialyzed overnight in 1 L TBSG at 4°C with stirring. GPLs were extracted from 1 mg each of purified proteins MlaC and MlaD by the method of Bligh and Dyer and analyzed by LC-MS/MS as previously described.

Lc-ms/MS

Retention of PG, CL, PE, and Lyso-CL was achieved at a flow rate of 0.3 mL/min using mobile phase A [CHCl3/CH3OH/NH4OH (800:195:5 v/v/v)] and mobile phase B [CHCl3/CH3OH/NH4OH (600:340:5 v/v/v)]. The chromatography method used is a three-step gradient as described in the SI Materials and methods of Dalebroux et al. (2014). The samples were run on an Agilent Zorbax Rx-SIL silica column (2.1 × 100 mm, 1.8-Micron) using an Agilent HPLC autosampler. Mass spectrometry was performed using an AB Sciex API4000 Qtrap with multiple reaction monitoring (MRM). The identities of the major GPLs present in the A. baumannii membrane were predicted by parent ion scans.

Stable isotope assay development

The Q1/Q3 transitions of glycerolphospholipids from cells grown in 2-13C acetate were determined using a Thermo Orbitrap LTQ. The integrated peak areas of both 13C-labeled and unlabeled GPLs from the AB Sciex API4000 Qtrap were used to calculate the ion-current ratios for each GPL species. The ratio of labeled GPL for each unique species can be calculated based on the following equation:

Rlab = Ri Rb (MacCoss et al., 2001)

Where Ri is the ion-current ratio of labeled GPL to unlabeled GPL within the sample and Rb is the ion-current ratio of samples before the administration of the tracer, 13C-acetate, and represents the natural background abundance of the stable isotope species within the bacterial membrane. Rlab approximates the molar ratio of labeled species to unlabeled species (nlab/nun) according to the equation (nlab/nun) = [Ri-Rb]/k, where k is the molar response factor of the instrument and is ideally equal to unity (MacCoss et al., 2001).

To demonstrate that OM phospholipases will not distinguish between labeled and unlabeled GPL and therefore will not affect the ratio of labeled to unlabeled GPL obtained from this assay, we compared ratios of labeled and unlabeled GPL from wild type A. baumannii and deletion mutants in pldA. Bacteria were grown carrying either the empty pMMB::kan vector, or expressing the Walker box mutant MlaFK55L. Accumulation of newly synthesized GPL was observed in those strains expressing MlaFK55L when compared to the vector control, across various species of GPL. Of strains expressing the vector control, on average 51.84 ± 1.07% and 52.07 ± 1.23% of newly synthesized PG C16:0/18:1 appeared on the inner membrane of wild type and ∆pldA, respectively, after one hour incubation with 13C acetate, while 66.33 ± 1.23% and 62.60 ± 1.70% of newly synthesized PG C16:0/18:1 accumulated at the inner membranes of wild type and ∆pldA expressing MlaFK55L. In vector controls strains, 48.53 ± 1.37% and 51.01 ± 0.55% of newly synthesized PG C16:0/16:0 appeared on the inner membrane of wild type and ∆pldA, respectively, after one hour incubation with 13C acetate, while 62.98 ± 1.01% and 60.41 ± 1.25% of newly synthesized PG C16:0/16:0 accumulated at the inner membranes of wild type and ∆pldA expressing MlaFK55L. In vector controls strains, 50.17 ± 1.31% and 50.49 ± 1.15% of newly synthesized PE C16:0/18:1 appeared on the inner membrane of wild type and ∆pldA, respectively, while 60.14 ± 0.93% and 62.06 ± 1.07% of newly synthesized PE C16:0/18:1 accumulated at the inner membranes of wild type and ∆pldA expressing MlaFK55L.

Stable isotope GPL analysis and culture conditions

Cultures of A. baumannii ATCC 17978 were grown in M63 media containing 5 mM sodium acetate and 4 mM MgCl2 to OD600 0.4, then washed and resuspended in media containing 5 mM 2-13C sodium acetate (Cat. No. CLM-381–0, Cambridge Isotope Laboratories, Inc.). Membrane fractions were isolated from both wild type and mla mutant A. baumannii at simultaneous time points, and GPL were extracted and assessed using previously established LC-MS/MS methods with additional MRM values to account for the increased m/z ratios of 13C-labeled GPL. MRMs were selected to account for PG and PE having acyl chains of either C16:0/16:0, C16:0/18:1, and C18:1/18:1 as these were determined by total ion scan MS to be the predominant species of PG and PE GPL. Pulse experiments were performed at least twice for each mutant.

References

  1. 1
  2. 2
  3. 3
  4. 4
    Study of the moraxella group. I. genus moraxella and the neisseria catarrhalis group
    1. P Baumann
    2. M Doudoroff
    3. RY Stanier
    (1968)
    Journal of Bacteriology 95:58–73.
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
    Leakage of periplasmic enzymes by mutants of Escherichia coli and salmonella typhimurium: isolation of "periplasmic leaky" mutants
    1. J Lopes
    2. S Gottfried
    3. L Rothfield
    (1972)
    Journal of Bacteriology 109:520–525.
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
    Mechanism of assembly of the outer membrane of salmonella typhimurium. site of synthesis of lipopolysaccharide
    1. MJ Osborn
    2. JE Gander
    3. E Parisi
    (1972)
    The Journal of Biological Chemistry 247:3973–3986.
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49
  50. 50
  51. 51
  52. 52
  53. 53
  54. 54
  55. 55
  56. 56
    Agents that increase the permeability of the outer membrane
    1. M Vaara
    (1992)
    Microbiological Reviews 56:395–411.
  57. 57
  58. 58
  59. 59
  60. 60
  61. 61
  62. 62

Decision letter

  1. Wendy S Garrett
    Senior Editor; Harvard TH Chan School of Public Health, United States
  2. Tâm Mignot
    Reviewing Editor; Aix Marseille University-CNRS UMR7283, France

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The first decision letter after peer review is shown below.]

Thank you for submitting your work entitled "An ABC transporter delivers glycerophospholipids to the Acinetobacter baumannii outer membrane" for consideration by eLife. Your article has been reviewed by three peer reviewers, and the evaluation has been overseen by a Reviewing Editor and a Senior Editor. The reviewers have opted to remain anonymous.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.

While all reviewers agree that the manuscript addresses an important topic, they all think that the evidence provided is not sufficient to support the main claim of the manuscript, that the Gpt system delivers phospholipids to the outer membrane (OM). This is an especially sensitive issue because the E. coli ortholog Mla complex has been described to function in the opposite direction from the OM to the IM. While the reviewers do not exclude that your conclusions are correct, they all think that this controversy can only be resolved if a large body of additional experiments is provided. Briefly, clear membrane fractionation controls should be provided, but it should be explained how the PL decrease in the OM is compensated. Other approaches such as EM and live microscopy would also strengthen the conclusions.

The policy at eLife is to only allow revisions of manuscript that could be performed within a two-months time period. Given, the amount of perceived additional experiments we have are rejecting the manuscript. The individual reviews are appended below to help prepare the manuscript for submission elsewhere.

Reviewer #1:

This work by Kamischke et al. presents the interesting finding of a glycerolipid transport system, which appears to have a major role in establishing the composition of the OM.

The transporter is present in other gram negative bacteria and the finding would thus be an important insight into the mechanism of OM biogenesis.

The work nicely provides genetic, structural and biochemical assays to support the authors conclusions

Overall, I am enthusiastic about the work, but I have a number of concerns.

– First, the discovery of the Gpt system is not exactly novel, the Mla complex, apparently the direct homolog, has been previously identified in E. coli (which the author acknowledge). However contrary to previous interpretations, Gpt is proposed to transport glycerolipids from the IM to the OM while Mla (f I understand correctly) was proposed to maintain lipid asymmetry in the OM, transporting lipids from the OM to the IM. I am not an expert in this field but because this creates a controversy, I am wondering whether the data presented is sufficient to make this claim. The conclusion mostly comes from the assay presented in Figure 7 showing that newly synthesized phospholipids accumulate at the IM in gpt mutants. However, I do not understand why the OM composition appears mostly unaffected in the same mutants in that assay? Also given that the authors have this assay in hand it should not be too complicated to test the E. coli mlA mutant and see if they get similar results? That would help solving this controversy.

Bottom line, this part should be absolutely convincing for the paper to be accepted.

– Some microscopy of the gpt mutants (EM, phase contrast) would be helpful to describe the phenotypes in these mutants. Could permeability of the mutants be shown directly in individual cells?

– Statistics are missing in several instances especially in in Figure 1 and Figure 7 which only shows representative experiments.

Reviewer #2:

This work deals with the most poorly understood aspect of envelope biogenesis in Gram-negative bacteria: transport of phospholipids (PLs) between the inner membrane (IM) and outer membrane (OM). Kamischke et al. characterize the Gpt system of Acinetobacter baumannii, which is orthologous to the previously identified Mla system in E. coli. A portion of this work describes how some of what we know about Mla also applies to Gpt: mla/gpt mutants have OM permeability defects; IM components of Mla/Gpt system form a complex; part of the Mla/Gpt system is a predicted ABC transporter and activity of its ATPase is required for transport; and finally, some periplasmic components of Mla/Gpt bind PLs.

The remaining part of this manuscript describes two novel findings but raises the following serious concerns:

1) Cryo-EM structural information of the GptBDEF IM complex. This is a novel structure that is only validated by the fitting of Phyre model structures of the cytoplasmic components (GptBF) and previous data showing that MlaD (GptD) is hexameric. Because the structure is of low-resolution (~11 Å), its impact is minimal – it does not provide insight into the mode of function or guidance for future work on function.

2) The authors conclude that Gpt transports PLs from the IM to OM. This is very surprising because both the Mla system of E. coli and its chloroplast orthologous system have been reported to function in retrograde PL transport based on in vivo data. Previously, Malinverni and Silhavy (reference #8) concluded that Mla functions in retrograde transport because they obtained in vivo data demonstrating that PLs accumulate in the outer leaflet of the OM: in mla mutants, PagP was activated and defects were suppressed by PldA overexpression. It is impossible to explain those data in a model where Mla transports PLs from the IM to the OM, as it is suggested here for Gpt. It is also very difficult to imagine how a multi-protein transport system that is driven by an ABC transporter to transport PLs from one membrane to another could have evolved to run in opposite directions in different organisms.

Here, Kamischke et al. conclude that Gpt is involved in anterograde transport because of data presented in Figures 6 and 7, where the authors measure PL content in the IM and OM. These data rely on the separation of IM and OM using a density ultracentrifugation procedure. How well these two membranes were separated is crucial to validate this work. Therefore, the authors should show the data they mentioned in the Materials and methods that illustrate proper separation of the IM and OM. In fact, the authors should at the very least show stained protein gels (not just immunoblots) of all fractions so that readers can assess the fractionation procedure. Furthermore, I worry that this technique might have misled the authors because of the following issue. Separation of IM and OM only works when the density values of these membranes are different enough. There are two factors contributing to this difference: the different protein:lipid ratio between the two membranes and the LOS:phospholipid ratio in the OM. If the Gpt system was responsible for retrograde transport, as previously proposed for Mla, a gpt mutant would accumulate PLs in the OM. This would cause the density of the OM (or domains/areas in the OM) to be more IM-like, making the fractionation more difficult or even impossible depending on the extent of the defect. In fact, it is likely that if PLs accumulated in the OM, the OM would be a mixture of two types of bilayers: some areas would be like wild-type OM (asymmetric bilayer containing both LOS and PLs), while other areas would be IM-like (symmetric PL bilayer). That is, some OM domains (or areas) would fractionate as expected (heavy density), while others would fractionate with the IM or somewhere in between the OM and IM fraction (lighter density). If this happened, fractionations would appear normal except that one would expect excess PLs in the IM, compared to the OM. This is what the authors' data show, so I wonder if they are misinterpreting the data. Can the authors rule out this possibility? Furthermore, the authors use a fractionation procedure that is based on spheroplasting cells, which leaves domains/areas of the OM still connected to the IM via trans-envelope complexes (e. g. Tefsen J Biol Chem. 2005). Would the same results be obtained if cells were lysed by high pressure?

Another issue that their data raises is this: The OM should cover more surface area than the IM. Here, the authors say that the OM of a gpt mutant has a lot less PLs than the OM of the wild type and that the IM of the mutant has more PLs than that of the wild type. Their data suggest that the OM in the gpt mutant covers less surface area that in IM. How is this possible? What is filling the void in the OM caused by the decrease in PLs? The authors do not provide any information about the overall composition of the OM in gpt mutants that could explain how it can have less PLs while the IM has more. Again, different composition of the OM could affect how it (or domains of it) fractionates.

The authors should also address these additional major problems:

3) Why isn't Figure 6B showing accumulation of PG and PE in the IM? These data seem to contradict the model. Even Figure 6A does not appear to show much of an increase of PLs at the IM.

4) The authors state (without providing data) that cardiolipin is made at the OM. If that is true, based on the authors' model (Gpt transports the substrates needed to make cardiolipin at the OM), one would expect that the overall cardiolipin content in gpt mutants would be very low, especially in comparison to the other PL species. That is not what Figure 6A shows. Why?

5) Could the authors please show error bars for Figure 7?

6) Discussion section: "While it is possible that the Gpt system in E. coli serves a different primary function than in A. baumannii, the phenotypes observed in E. coli mla mutants might also be explained by a disruption of outer membrane structure stemming from decreased concentrations of outer membrane PLs, leading to activation of the PagP enzyme." This weak argument does not make any sense. PagP is activated by an increase in PLs at the outer leaflet of the OM.

Reviewer #3:

The article is written and presented in a seamless manner, where figures, hypotheses, and figures are clearly articulated. This is a testament to the overall quality of the work and presentation. However, this is not to say that it has no weaknesses, some of which need to be addressed prior to publication. The authors' conclusions are never heavy-handed or written in absolute terms; as such they accommodate alternative opinions and possibilities; especially important, as their final figure, serves to upturn the existing dogma in the field, regarding the function of a homologous operon best studied in E. coli. Some might argue that renaming these homologs is unnecessary. The authors should strongly consider their proposal to rename these genes for the following reasons: i) the buried mention (Materials and methods) that cardiolipin is not transported by this system, argues that it is not in fact a general "glycerophospholipid" transporter (see below additional comments on cardiolipin); ii) these authors have only demonstrated in one strain of A. baumannii that some degree of PE and PG transport to the OM likely occurs as a consequence of this operon, and have not overturned the likely function of this operon to maintain asymmetry of the outer membrane. It's possible that context, such as organism or environments, favors utilization of this machinery in one direction or another. Also of great importance is the issue as to what fills in for the loss of GPL in the outer membrane as it is highly unlikely that it is LOS at the inner leaflet of the outer membrane. The authors' consideration on these matters, and their ability to adequately address these concerns will likely impact publication.

Major concerns/comments:

1) As noted above the Mla (Gpt) system has been previously characterized in E. coli and hypothesized as a retrograde GPL transporter. Clearly the mla system is not required for growth, so if it is functioning as an anterograde transporter what is replacing GPLs in the inner leaflet of the OM in mla mutants. In the view of this reviewer, the burden of proof is now higher given the previous work on the E. coli mla system.

2) The audience must be provided with inner membrane and outer membrane separation data throughout the manuscript. This is critical as the reader must be convinced that changing the ratio of GPL/LOS/OM protein allows for proper separation of Acinetobacter membranes.

3) Do the authors have data to demonstrate that the K55L "dominant negative" mutation truly abrogates ATP hydrolysis, since it is a critical assumption in interpreting the data in Figures 1 and 2?

4) Could the authors please speculate why only A1S_3102 and A1S_3103 were identified using BCIP-Toluidine screen of Tn library? Where there not Tn insertions in the other members of the operon?

5) Why do the authors not discuss MlaA and why was a mutant not made in this gene? Also, it is not labeled in the model? Does it not function in the pathway in Acinetobacter?

6) Figure 6. Several comments on this figure:

a) In 6A – it appears there are more GPLs by TLC in the OM of the wild type which is odd

b) In 6B, – if gpt was an anterograde transporter one would expect to see accumulation of GPLs in the inner membrane and decreases of GPLs in the outer membrane. When you look at the bar graph, the levels of PE and PG don't increase in the inner membrane in the gptC mutant and they actually decrease a little. Why?

7) Have the authors performed electron microscopy to see if they observe changes in the inner membrane (membrane invaginations) because of loss of gpt mediated GPL transport? This sort of data would greatly strengthen their hypothesis.

8) Discussion paragraph two: The Mla system was previously reported as a retrograde transport system for E. coli and in mla mutants PagP is activated. Since PagP is only functional when PLs are present at the surface, the idea was that in mla mutants GPLs are mislocalized in the outer leaflet of the OM. In the discussion, the authors rationalize that the phenotypes observed in E. coli mla mutants might also occur because GPL transport to the OM is not as efficient and because of this the LPS-to-GPL rations are off, resulting in flipping of GPLs to the outer leaflet and this is why PagP is activated. Is this likely, as PagP needs GPLs for activity and in the authors' model the Gpt (Mla) system is responsible, at least in part, for transport of GPLs to the OM?

9) Discussion paragraph three: The authors note that since the Mla (Gpt) system is upregulated in LOS deficient Acinetobacter, it suggests that Mla serves as a anterograde transporter as increased transport of GPLs may serve to compensate for lack of LOS in the outer membrane. Why couldn't it be the reverse, that the Mla system is upregulated in order to prevent too many GPLs in the OM. This goes back to what fills in the gaps when LOS is missing. After searching on PubMed, this reviewer found a recent paper from Boll et al., reporting that key lipoproteins seemed to be present in the outer leaflet of LOS deficient Acinetobacter.

[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration. They were asked to provide a plan for revisions before the editors issued a final decision.]

Thank you for sending your article entitled "The Acinetobacter baumannii Mla system and glycerophospholipid transport to the outer membrane" for peer review at eLife. Your article is being evaluated by two peer reviewers, and the evaluation is being overseen by a Reviewing Editor and Wendy Garrett as the Senior Editor.

Given the list of essential revisions, including new experiments, the editors and reviewers invite you to respond within the next two weeks with an action plan and timetable for the completion of the additional work. We plan to share your responses with the reviewers and then issue a binding recommendation.

The current manuscript proposes that the Acinetobacter baumannii Mla system transports glycerophospholipid to the outer membrane. This statement challenges a large body of literature on both the E. coli and A. Baumannii systems (and to some extent V. Cholerae and H. influenza), all of which favor that Mla is instead involved in OM turnover and retrograde transport. After consultation with the reviewers, several key points would have to be clarified for publication in eLife:

1) The reference to existing literature must be improved in particular to reconcile and potentially explain the differences between this study and published studies. If needed, additional experiments must be provided to resolve/address these conflicts. Along these lines:

– Powers and Trent, 2018, published evidence for retrograde transport in the same organism, which is not commented by the authors.

– Contrarily to this study, multiple laboratories have reported that deletion of mlA genes is not associated with increased sensitivity to vancomycin and in fact linked to vancomycin resistance gene induction.

– Roier et al. (Nat Comm, 2016) have shown that mlA deletion is linked to OMV production in V. Cholerae and H. influenza, which is not taken into account in this study. Can the authors rule out that lower GPL levels in the OM are not due to the loss of OM materials?

2) The pulse chase assay is an important innovation of this study and the major experimental evidence to support the proposed directionality of Mla. Proving the consistency of this assay is therefore paramount to support the authors’ conclusions. The authors need to provide all the controls confirming that the IM and the OM were correctly separated in sucrose density gradients, which apparently is performed for the first time in A. Baumannii. For this, sucrose gradients need to be shown, with a complete set of IM and OM controls. A series of helpful recommendations is provided in the individual reviews. Variations in the assay starting point and in particular the high variance in total labeled lipids should also be explained.

Reviewer #2:

In "The Acinetobacter baumannii Mla system and glycerophospholipid transport to the outer membrane," Kamischke et al. report that the previously characterized Mla system functions in an anterograde manner. This is a well-written manuscript and the subject matter is important. The authors used a transposon library to identify genes required for OM integrity in A. baumannii by monitoring for release of an endogenous periplasmic phosphatase. Among the 58 genes identified in the screen, two genes encoded for components of the Mla system. The finding that mutations in mla genes results in OM permeability defects in Ab aligns with previous data reported in E. coli and other Gram-negative organisms (e.g. Vibrio cholerae, Haemophilus influenza). Previously, the Silhavy Laboratory demonstrated that Mla mutants show increased sensitivity to SDS-EDTA and deletion of any of the Mla components resulted in mislocalization of phospholipids to the outer leaflet of the outer membrane. It was found that the outer membrane phospholipase, PldA, was a suppressor of the mla mutant SDS-EDTA sensitivity (see below for additional details on previous work). This is a key point as it suggests that the role of the Mla system is to remove phospholipids from the outer leaflet.

In the current work, the authors went on to characterize the Ab Mla inner membrane complex via cyro-EM. In the view of this reviewer, this is the most substantial contribution of the current work. Previously, Ekiert and co-workers characterized multiple MCE-domain containing complexes, including the Mla complex (Cell, 2017), and the work completed here is a nice addition to the field. Unfortunately, the other findings reported here are highly problematic. The authors reach the conclusion that Mla functions as an anterograde phospholipid transporter, which is in direct conflict with previous work from multiple groups suggesting that Mla functions in a retrograde manner including a recent paper in A. baumannii (Powers et al., 2018). The Mla mutants presented here have very different phenotypes than those reported for other Gram-negatives and the biochemical data to support the anterograde claim are not convincing for various reasons.

Major Points:

1) Kamischke and co-workers report that deletion of mlaC or mlaF results in increased sensitivity to vancomycin (~10 fold). First, this goes against previous literature, reported by multiple labs, that mla mutants do not show increased sensitivity to vancomycin (or other antibiotics such as novobiocin). Actually, according to others the level of vanc resistance actually increases upon deletion of mla genes. How do the authors explain these differences?

2) Subsection “The Mla system is necessary for A. baumannii OM integrity”: What is the rationale for authors to complement mutations in original locus? This does not address the possibility of off-target gene expression alterations. This is even more relevant since it has already been reported that previous mla insertational mutants show polar effects (Silhavy, 2009).

3) Figure 4 and 5 and supplemental: Authors do not adequately illustrate the effectiveness of their membrane separation. It is established that many diderm organisms cannot be physically fractionated into OM and IM. Furthermore, to the reviewer's knowledge, this is the first publication to include membrane separations for Acinetobacter baumannii. The authors need to include significantly more information to assure the readers that their membrane fractionation works robustly in A. baumannii. Currently, the authors only include an outer membrane marker for an unknown fraction of the gradients. This is insufficient in the context of this work. The authors should include an inner membrane marker and an outer membrane marker across the gradient. Also, protein distribution and turbidity measures over the entire gradient would be very helpful.

4) Figure 4B: Where do the authors hypothesize the GPLs go? They presumably see a decrease in PG and PE in the outer membrane, but no subsequent increase in the inner membrane at steady state. Could this not be explained by vesicular blebbing, which would be an expected consequence after deletion of a retrograde transport system? In fact, an argument could be made that these findings are entirely an artifact of vesicular blebbing in the absence of Mla. In fact, Roier et al. have shown that deletion of mla components in both V. cholerae and H. influenza results in increased outer membrane vesicles. However, the authors do not consider this option or discuss this previous work.

5) Also, in reference to the Roier paper. In the current work the authors make the claim that "biochemical analysis for the membrane GPL composition for Mla mutants has not been published for any organism" This is not correct and these data were reported by Roier et al. Also, Roier did not observe the drastic reduction in OM phospholipid abundance reported in the current manuscript.

6) Discussion paragraph three: Authors suggest the function of MlaA is to allow MlaC to deposit phospholipids into the periplasmic-leaflet of the OM. The structure of MlaA identified a loop, preventing periplasmic-leaflet phospholipids from entering the MlaA channel. MlaA* lacks this loop, resulting in periplasmic-leaflet phospholipids entry into the MlaA channel. The result is a gain of function increased sensitivity to membrane stress.

7) Figure 5 and Figure 5—figure supplement 2: The authors devised a clever pulse chase assay. Why do the authors see such different starting points for the labeled cultures? Based on the variability of labeling by lipid species (even for the wildtype cultures), there is a concern regarding the consistency of the method.

8) Supplementary file 2: Furthering the concern about the pulse chase assay is the data presented in Supplementary file 2. We analyzed 3 replicates of the data provided, and it appears that the total lipid labeled achieved by this pulse chase assay is highly variable. How do the authors feel confident in drawing the conclusions presented when consistent data cannot be achieved with wild-type? In some replicates, entire lipid species were unable to be detected. How are the figures presented using this data representative, when the raw data is not?

Reviewer #3:

The outer membrane (OM) of Gram-negative bacteria is an asymmetric lipid bilayer, composed of an inner leaflet of glycerophospholipids (GPL) and an outer leaflet of lipopolysaccharides (LPS). Whereas we now have a reasonable understanding of how LPS molecules are transported across the periplasm and inserted into the OM, we still do not know how GPL molecules are transported from the inner to the outer membrane. In this article, Kamischkke et al. propose an answer to this long-standing question.

First, the authors carried out a genetic screen to identify genes important for the integrity of the OM in Acinetobacter baumannii, an opportunistic pathogen. They screened a library of transposon mutants for colonies with lesions in the OM, reasoning that bacteria with an impaired OM will allow the chromogenic substrate XP to reach the periplasm where it can be degraded by a phosphatase, yielding a blue color. The results of the screen were validated using different assays. This led them to identify mutants with transposon insertions in genes encoding core components of the Mla system. The Mla system has been identified in E. coli where it was proposed to play a role in OM maintenance, transporting GPL from the OM back to the IM. The authors also confirmed that the phenotype exhibited by the mutants resulted from a lack of transport function and then used cryoEM to characterize the structure of the MlaBDEF complex. They also confirmed that components of the system interact with GPL.

The work described in this first part of the manuscript is convincing, the results are solid and very well presented. However, this first part mostly confirms work previously carried out in E. coli where similar screens were performed and where the Mla system was characterized, including at the structural level. In particular, the structure of the E. coli Mla complex was recently reported, although at a lower resolution. Overall, this part of the work is interesting and carefully done, but its novelty is not very high.

The second part of the article contains what I think could warrant publication in eLife. Indeed, the authors provide experimental evidence that the Mla system transports GPL from the IM to the OM, which would make of this system the first identified transporting GPL in this direction. This conclusion is based on a number of experiments. First, analysing IM and OM fractions separated by density centrifugation with TLC and LC-MS/MS, they showed that GPL levels were reduced in the OM of the ∆mlaC mutant. Second, they designed an MS-based strategy to quantitatively follow intermembrane transport of radiolabelled GPL in membrane fractions and showed that GPL accumulate in the IM of Mla mutants.

These results are extremely interesting and are consistent with the model that the Mla system transports GPL from the IM to the OM, which would represent a major finding. However, I have a number of important concerns that need to be addressed.

1) Given that the authors quantify GPL in membrane fractions, they need to provide all the controls confirming that the IM and the OM were correctly separated in sucrose density gradients. Only one control, showing the localization of OmpA in the OM fraction is provided. I think that more fractions of the sucrose gradients need to be shown, with additional IM and OM controls. My concern is that the accumulation of lipids in one membrane could have an impact on how the two membranes are separated and therefore on the results of the quantitative analysis. In addition, if Mla mutants have portions of the OM with GPL both in the inner and outer leaflet, where will these "GPL bilayers" be found in the sucrose gradients?

2) If GPL accumulate in the IM, what is the impact of this accumulation on the IM? A thorough microscopy analysis should also be carried out, which would strengthen the conclusions. Does the total surface area of the IM increase as a result of lipid accumulation?

3) The authors should also verify that the lower GPL levels in the OM do not result from the loss of OM material via increased vesiculation.

Providing these data is even more important that the direction proposed here for GPL transport is in contradiction with the work performed by others indicating that the Mla system is involved in the retrograde transport of GPL from the OM to the IM. In this regard, the authors need to take into account the recent publication by Powers and Trent in PNAS (2018) providing "strong biological evidence" for the retrograde transport of GPL by the Mla system in the same bacterium (A. baumannii). They need to explain how these data are compatible with theirs.

[Editors’ note: formal revisions were requested, following approval of the authors’ plan of action.]

Thank you for your careful response to the reviewers' comments. After further consultation with the reviewers we decided that your response could adequately address the comments provided that in addition to the proposed changes the revision further contains the following new experiments:

1) Given the existing literature, we would like further that the discrepancy on the vancomycin sensitivity be resolved by verification of this phenotype in a different A. baumannii strain.

2) Completion of sucrose gradients with the proper controls to verify inner and outer membrane separation for both wild type and mla mutants (e.g. NADH oxidase assays). A clear description of the gradient methodology and images showing separation of the membrane as provided in the rebuttal letter should also be added.

https://doi.org/10.7554/eLife.40171.029

Author response

[Editors’ note: the author responses to the first round of peer review follow.]

Reviewer #1:

[…] Overall, I am enthusiastic about the work, but I have a number of concerns.

– First, the discovery of the Gpt system is not exactly novel, the Mla complex, apparently the direct homolog, has been previously identified in E. coli (which the author acknowledge). However contrary to previous interpretations, Gpt is proposed to transport glycerolipids from the IM to the OM while Mla (f I understand correctly) was proposed to maintain lipid asymmetry in the OM, transporting lipids from the OM to the IM. I am not an expert in this field but because this creates a controversy, I am wondering whether the data presented is sufficient to make this claim. The conclusion mostly comes from the assay presented in Figure 7 showing that newly synthesized phospholipids accumulate at the IM in gpt mutants. However, I do not understand why the OM composition appears mostly unaffected in the same mutants in that assay?

It is important to note that the data presented in Figure 7 of the original paper (now Figure 5 in the revised manuscript) is not a representation of the absolute value of newly synthesized phospholipids. Instead, we are presenting on the y-axis the molar ratio of newly synthesized, C13-labeled glycerophospholipid (GPL) to unlabeled glycerophospholipid. The data is presented in this manner to limit the need for normalization to something such as membrane protein content, which might be argued to be unreliable given that these mutations have been shown to severely disrupt the outer membrane. However, we recognize that this manner of presentation may cause unnecessary confusion about the assay, and so we have added additional figures to show the relative amount of newly synthesized GPL in the inner and outer membranes of wild type and mutant A. baumannii in the hope that this will more clearly represent the data. As detailed below, the data clearly shows that in the mla mutants newly synthesized GPL accumulate on the inner membrane.

At any given time, as GPL are transported from the IM to the OM by any mechanism, mla or otherwise, the likelihood that a single GPL molecule which is transported from the IM to the OM will be labeled is proportional to the ratio of labeled to unlabeled GPL in the inner membrane. Although the OM ratio is similar in the mutants and wild type, what is significant is the discrepancy in the ratios between the inner and outer membranes. As the ratio of labeled to unlabeled increases in the IM over time, it should also increase at the same rate in the OM, provided anterograde transport between the membranes is not inhibited, as demonstrated by the results of these experiments in wild type. The significance of our result is that newly synthesized GPL accumulate in the IM at a faster rate than in the OM in our mutants, indicating that some mechanism of anterograde transport is inhibited in these strains. However, as we note in the discussion, transport to the OM is not completely abolished, as there are still labeled GPL that appear in the OM under these conditions. The fact that the ratio of labeled to unlabeled GPL in the OM of mutants increases can be explained by the existence of additional unexplained mechanisms of GPL transport, perhaps including a passive flow of GPL between membranes that may occur during cell division, combined with the fact that the likelihood of transporting a labeled GPL is much higher in the mutants strains due to the excess accumulation of labelled GPL on the inner membranes, giving them a much greater proportion of labeled GPL on the inner membrane after 60 minutes.

Also given that the authors have this assay in hand it should not be too complicated to test the E. coli mlA mutant and see if they get similar results? That would help solving this controversy.

Bottom line, this part should be absolutely convincing for the paper to be accepted.

We appreciate the reviewer’s suggestion and agree that it would be interesting to repeat our experiments in Enterobacteriaceae. However, the focus of our study is mechanisms of outer membrane permeability in Acinetobacter, and as such we have written this manuscript in order to present our findings about the Mla system in that organism. While we have preliminary data that the function of the system may be similar in Salmonella typhimurium, we think that repeating this work in another organism would be beyond the scope of this manuscript, and might serve as the basis of another manuscript entirely. We would also like to point out that we have made efforts to explain that while we believe the system functions primarily for anterograde rather than retrograde transport of GPL, we have not eliminated the possibility that the system does play some role in the maintenance of outer membrane asymmetry. In particular, the precise molecular function of the system’s outer membrane component, MlaA, is not entirely clear, and it is possible to imagine it might be in some way facilitating both the maintenance of lipid asymmetry and the transfer of GPL from MlaC to the outer membrane. It is not our intention in putting forth this data to stoke controversy, but to publish what we have observed about the function of this system in Acinetobacter. Pursuing the characterization of Mla in E. coli could conceivably put our model in direct conflict with the current paradigm and necessitate an extremely high burden of proof requiring additional experiments that could further delay publication of the data we already have.

– Some microscopy of the gpt mutants (EM, phase contrast) would be helpful to describe the phenotypes in these mutants. Could permeability of the mutants be shown directly in individual cells?

We have included in this revised manuscript phase contrast images of our mutants to help describe the phenotypes (Figure 1—figure supplement 1). It is our opinion that specific dye uptake or other visual methods would not add much over the quantitative assays of outer membrane permeability and antibiotic susceptibility we have performed.

Reviewer #2:

[…] 1) Cryo-EM structural information of the GptBDEF IM complex. This is a novel structure that is only validated by the fitting of Phyre model structures of the cytoplasmic components (GptBF) and previous data showing that MlaD (GptD) is hexameric. Because the structure is of low-resolution (~11 Å), its impact is minimal – it does not provide insight into the mode of function or guidance for future work on function.

The overall features of both structures, solved independently, are identical, suggesting that they correspond to the correct structure for the complex. However, the limited resolution of the ecMlaBDEF complex structure did not allow modeling of its individual subunits, in contrast to the abMlaBDEF structure reported here. We have been able to further improve the resolution to 8.7 Å in this revised manuscript. Based on the modeling we saw more detail, which was not shown from the ecMlaBDEF structure. Similar to MlaB and MlaF proteins, most helices are well resolved, which allowed us to place the models unambiguously, the arrangement of MlaF clearly resembles the pre-translocation state of MalK. This suggests that we have trapped a similar conformation of the abMlaBDEF complex. We also show that while the periplasmic domain possesses 6-fold symmetry, the TM domains of MlaD do not appear symmetrical. Two of the TM domains of MlaD form close contacts with the density attributed to MlaE while the other four do not appear to contact any other proteins. We believe that this data will in fact prove useful to guide future work regarding the molecular function of the system.

2) The authors conclude that Gpt transports PLs from the IM to OM. This is very surprising because both the Mla system of E. coli and its chloroplast orthologous system have been reported to function in retrograde PL transport based on in vivo data. Previously, Malinverni and Silhavy (reference #8) concluded that Mla functions in retrograde transport because they obtained in vivo data demonstrating that PLs accumulate in the outer leaflet of the OM: in mla mutants, PagP was activated and defects were suppressed by PldA overexpression. It is impossible to explain those data in a model where Mla transports PLs from the IM to the OM, as it is suggested here for Gpt. It is also very difficult to imagine how a multi-protein transport system that is driven by an ABC transporter to transport PLs from one membrane to another could have evolved to run in opposite directions in different organisms.

As the reviewer describes, Malinverni and Silhavy demonstrated that mutations in the Mla system lead to activation of PagP, and PldA overexpression suppresses both the PagP activation phenotype as well as increased sensitivity observed in these mutants to SDS-EDTA, suggesting the accumulation of GPL in the outer leaflet of the outer membrane under those conditions. However, we strongly disagree with the reviewer’s statement that it is impossible to reconcile these observations with an anterograde transport model. As we note in our manuscript, conditions that perturb the outer membrane, such as chemical disruption via EDTA exposure, have been shown to result in an increase in membrane disorder, i.e. a reduction in membrane asymmetry following the spontaneous flipping of glycerophospholipids to the outer leaflet and activation of PagP (see references 27-29). It is reasonable therefore to suppose that mutation in any protein that contributes significantly to membrane barrier function might result in GPL exposure on the outer leaflet, and it does not necessarily mean that such a protein is directly responsible for maintaining lipid asymmetry under wild type conditions. Pulse-chase experiments have been performed to demonstrate that the orthologous system in the plant chloroplast functions to transport GPL originating in the endoplasmic reticulum towards the interior of the chloroplast for eventual delivery to the thylakoid membrane within the chloroplast (see reference 36). The Gram-negative bacteria we are discussing here are not contained within a eukaryotic cell exposed to organelles such as the ER that might be a source for GPL. Therefore, we do not find it difficult to imagine that this orthologous system might have evolved to have reverse functionality in the chloroplast from what we observe here in A. baumannii.

Here, Kamischke et al. conclude that Gpt is involved in anterograde transport because of data presented in Figures 6 and 7, where the authors measure PL content in the IM and OM. These data rely on the separation of IM and OM using a density ultracentrifugation procedure. How well these two membranes were separated is crucial to validate this work. Therefore, the authors should show the data they mentioned in the Materials and methods that illustrate proper separation of the IM and OM. In fact, the authors should at the very least show stained protein gels (not just immunoblots) of all fractions so that readers can assess the fractionation procedure.

In addition to the immunoblots, we have here included a figure of stained protein gels of inner and outer membrane fractions in both mutant and wild type bacteria so that readers can better assess the fractionation procedure.

Furthermore, I worry that this technique might have misled the authors because of the following issue. Separation of IM and OM only works when the density values of these membranes are different enough. There are two factors contributing to this difference: the different protein:lipid ratio between the two membranes and the LOS:phospholipid ratio in the OM. If the Gpt system was responsible for retrograde transport, as previously proposed for Mla, a gpt mutant would accumulate PLs in the OM. This would cause the density of the OM (or domains/areas in the OM) to be more IM-like, making the fractionation more difficult or even impossible depending on the extent of the defect. In fact, it is likely that if PLs accumulated in the OM, the OM would be a mixture of two types of bilayers: some areas would be like wild-type OM (asymmetric bilayer containing both LOS and PLs), while other areas would be IM-like (symmetric PL bilayer). That is, some OM domains (or areas) would fractionate as expected (heavy density), while others would fractionate with the IM or somewhere in between the OM and IM fraction (lighter density). If this happened, fractionations would appear normal except that one would expect excess PLs in the IM, compared to the OM. This is what the authors' data show, so I wonder if they are misinterpreting the data. Can the authors rule out this possibility?

The reviewer is correct to point out that areas of symmetric GPL bilayer in the OM might conceivably fractionate with the IM upon density centrifugation, leading to the appearance of an excess of GPL in the IM relative to the OM. This is one of the reasons we developed the stable isotope assay after first observing the overall reduction in OM GPL by TLC and LC-MS/MS shown in Figure 4. Additionally, it occurred to us that activation of OM phospholipases such as PldA could result in this phenotype and lead to mischaracterization of the system. The stable isotope assay gives insight into whether these results are due simply to mislocalization or degradation of OM GPL, or if they can in fact be attributed to deficient anterograde GPL transport. This is because the peak intensity of each newly synthesized, C13-labeled GPL is normalized to the corresponding unlabeled version of that GPL species with each sample injected into the LC-MS/MS. The result is a molar ratio of labeled to unlabeled GPL for every membrane sample. With C13-acetate as the sole carbon source, we observe a gradual increase in the ratio of labeled GPL relative to unlabeled GPL over time. In wild type bacteria, these ratios for the inner and outer membranes track closely over time, which indicates that under these conditions GPL transport from the inner to the outer membrane occurs quite rapidly, as has been previously shown in E. coli reference. In mla mutants the ratio is both higher and typically increases at a greater rate in the inner membrane. Phospholipases will not distinguish between labeled and unlabeled GPL species, and so will not impact the ratios obtained with this assay. Areas of symmetric OM bilayers fractionating with the IM also cannot explain the data for our mutants because that would not result in a specific increase in labeled GPL relative to unlabelled GPL, as such theoretical micro domains would themselves be a mixture of labeled and unlabelled GPL. The results observed in mla mutants are best explained as demonstrating a decreased rate of anterograde transport relative to wild type bacteria, in which GPL are first synthesized and inserted in the IM but subsequent transport to the OM is inhibited.

Furthermore, the authors use a fractionation procedure that is based on spheroplasting cells, which leaves domains/areas of the OM still connected to the IM via trans-envelope complexes (e. g. Tefsen J Biol Chem. 2005). Would the same results be obtained if cells were lysed by high pressure?

As described in our Materials and methods, following the spheroplast procedure, samples were homogenized under high pressure with an Avestin EmulsiFlex-C3 and spun down at 17,000 g for 10 min to removed un-lysed cells prior to ultracentrifugation.

Another issue that their data raises is this: The OM should cover more surface area than the IM. Here, the authors say that the OM of a gpt mutant has a lot less PLs than the OM of the wild type and that the IM of the mutant has more PLs than that of the wild type. Their data suggest that the OM in the gpt mutant covers less surface area that in IM. How is this possible? What is filling the void in the OM caused by the decrease in PLs? The authors do not provide any information about the overall composition of the OM in gpt mutants that could explain how it can have less PLs while the IM has more. Again, different composition of the OM could affect how it (or domains of it) fractionates.

Respectfully, we do not state that the IM of the mla mutants have more GPL than wild type, and our TLC and LC-MS/MS quantification data in fact show that IM GPL levels are comparable or slightly decreased in mla mutants. While our TLC and initial quantification of GPL by LCMS/ MS in Figure 4 suggest a decrease in OM GPL, we acknowledge in our manuscript that these results could be due to indirect effects, which is why we developed the stable isotope assay, as described in our response above. That being said, these mutants were revealed to us in a screen for outer membrane integrity, and have been shown to have an outer membrane permeability defect, suggesting that damage to the outer membrane is not fully compensated. Previous work implicating orthologous systems in GPL transport, and our own data showing GPL binding to components of the system led us to focus on the GPL components of the membranes rather than attempting to profile the overall membrane composition. We have not attempted to determine whether the LOS or lipoprotein profiles are significantly altered in these mutants so as to compensate for the loss of GPL, although this might be an interesting avenue for future study.

Regarding this question, our data showing a significant increase in exopolysaccharide in mla mutants may suggest regulatory mechanisms responding to membrane damage to mitigate the effects.

The authors should also address these additional major problems:

3) Why isn't Figure 6B showing accumulation of PG and PE in the IM? These data seem to contradict the model. Even Figure 6A does not appear to show much of an increase of PLs at the IM.

It is likely that the inner membrane content of GPL is regulated by many factors including synthesis, degradation by phospholipases, outer membrane transport, and cell division. We can speculate that feedback mechanisms have resulted in maintenance of appropriate content of the inner membrane, which is critical to cellular integrity.

4) The authors state (without providing data) that cardiolipin is made at the OM. If that is true, based on the authors' model (Gpt transports the substrates needed to make cardiolipin at the OM), one would expect that the overall cardiolipin content in gpt mutants would be very low, especially in comparison to the other PL species. That is not what Figure 6A shows. Why?

Bioinformatic analysis of cardiolipin synthase genes in ATCC 17978 reveals two genes, one of which encodes a signal sequence for export to the periplasm. We do observe a reduction in OM CL by TLC as expected, however our stable isotope assay implicated the Mla system in anterograde transport of PE and PG, but not CL, indicating to us that there may be alternative mechanisms of increasing OM CL content, and we suggest that a periplasmic OM CL may explain this result, although we have not yet thoroughly investigated this as our focus is on the Mla system. We do not understand why the reviewer expects that the overall cardiolipin content in mutants would be low.

5) Could the authors please show error bars for Figure 7?

6) Discussion section: "While it is possible that the Gpt system in E. coli serves a different primary function than in A. baumannii, the phenotypes observed in E. coli mla mutants might also be explained by a disruption of outer membrane structure stemming from decreased concentrations of outer membrane PLs, leading to activation of the PagP enzyme." This weak argument does not make any sense. PagP is activated by an increase in PLs at the outer leaflet of the OM.

We respectfully disagree. PagP is activated by OM damage such as the use of EDTA, which may or may not cause migration or membrane disorder. Despite the “dogma” that PagP is activated on migration of phospholipids to the outer membrane (which comes from a detergent based crystal structure) it is activated in the absence of any membrane damage in Salmonella typhimurium. The idea that PagP is activated on movement of membrane asymmetry is reasonable but also not necessarily the only mechanism. It is also plausible that a membrane defect has resulted in membrane disorder that allows PagP access to substrate in the inner leaflet.

Reviewer #3:

The article is written and presented in a seamless manner, where figures, hypotheses, and figures are clearly articulated. This is a testament to the overall quality of the work and presentation. However, this is not to say that it has no weaknesses, some of which need to be addressed prior to publication. The authors' conclusions are never heavy-handed or written in absolute terms; as such they accommodate alternative opinions and possibilities; especially important, as their final figure, serves to upturn the existing dogma in the field, regarding the function of a homologous operon best studied in E. coli. Some might argue that renaming these homologs is unnecessary. The authors should strongly consider their proposal to rename these genes for the following reasons: i) the buried mention (Materials and methods) that cardiolipin is not transported by this system, argues that it is not in fact a general "glycerophospholipid" transporter (see below additional comments on cardiolipin); ii) these authors have only demonstrated in one strain of A. baumannii that some degree of PE and PG transport to the OM likely occurs as a consequence of this operon, and have not overturned the likely function of this operon to maintain asymmetry of the outer membrane. It's possible that context, such as organism or environments, favors utilization of this machinery in one direction or another. Also of great importance is the issue as to what fills in for the loss of GPL in the outer membrane as it is highly unlikely that it is LOS at the inner leaflet of the outer membrane. The authors' consideration on these matters, and their ability to adequately address these concerns will likely impact publication.

We thank the reviewer and agree with their points regarding renaming the system and acknowledge that we have not ruled out a role for this system in maintaining OM lipid asymmetry. In this version of the manuscript we will refer to the system as Mla.

Major concerns/comments:

1) As noted above the Mla (Gpt) system has been previously characterized in E. coli and hypothesized as a retrograde GPL transporter. Clearly the mla system is not required for growth, so if it is functioning as an anterograde transporter what is replacing GPLs in the inner leaflet of the OM in mla mutants. In the view of this reviewer, the burden of proof is now higher given the previous work on the E. coli mla system.

Please see our response above to the similar concerns of reviewer #2 regarding the composition of the OM in mla mutants.

2) The audience must be provided with inner membrane and outer membrane separation data throughout the manuscript. This is critical as the reader must be convinced that changing the ratio of GPL/LOS/OM protein allows for proper separation of Acinetobacter membranes.

These controls have been included in the new version of the manuscript.

3) Do the authors have data to demonstrate that the K55L "dominant negative" mutation truly abrogates ATP hydrolysis, since it is a critical assumption in interpreting the data in Figures 1 and 2?

We have not attempted to directly demonstrate that the K55L mutation abrogates ATP hydrolysis. The Walker box lysine residue is highly conserved in ATPases and has been previously demonstrated in other ATPases to abrogate ATP hydrolysis (See references 15-18).

4) Could the authors please speculate why only A1S_3102 and A1S_3103 were identified using BCIP-Toluidine screen of Tn library? Where there not Tn insertions in the other members of the operon?

It is likely our screen was not performed to completion and that subsequent rounds of transposon mutagenesis would reveal additional Tn insertions in other components of the operon.

5) Why do the authors not discuss MlaA and why was a mutant not made in this gene? Also, it is not labeled in the model? Does it not function in the pathway in Acinetobacter?

Following our initial submission we have mutated and analyzed MlaA and observed results consistent with mutations of other components in the system. That data is now included in the manuscript.

6) Figure 6. Several comments on this figure:

a) In 6A it appears there are more GPLs by TLC in the OM of the wild type which is odd.

We do not fully understand what the reviewer is referring to here with this observation.

b) In 6B, if gpt was a anterograde transporter one would expect to see accumulation of GPLs in the inner membrane and decreases of GPLs in the outer membrane. When you look at the bar graph, the levels of PE and PG don't increase in the inner membrane in the gptC mutant and they actually decrease a little. Why?

Gram-negative bacteria likely have multiple redundant mechanisms to ensure that the overall quality of IM GPL does not exceed a viable threshold relative to the OM. We reiterate our response to a similar concern raised by reviewer #2. It is likely that the inner membrane content of GPL is regulated by many factors including synthesis, degradation by phospholipases, outer membrane transport, and cell division. We can speculate that feedback mechanisms have resulted in maintenance of appropriate content of the inner membrane, which is critical to cellular integrity.

7) Have the authors performed electron microscopy to see if they observe changes in the inner membrane (membrane invaginations) because of loss of gpt mediated GPL transport? This sort of data would greatly strengthen their hypothesis.

This is an interesting idea but the lack of dramatic increase in GPL in the inner membrane has discouraged us from pursuing this line of research.

8) Discussion paragraph two: The Mla system was previously reported as a retrograde transport system for E. coli and in mla mutants PagP is activated. Since PagP is only functional when PLs are present at the surface, the idea was that in mla mutants GPLs are mislocalized in the outer leaflet of the OM. In the discussion, the authors rationalize that the phenotypes observed in E. coli mla mutants might also occur because GPL transport to the OM is not as efficient and because of this the LPS-to-GPL rations are off, resulting in flipping of GPLs to the outer leaflet and this is why PagP is activated. Is this likely, as PagP needs GPLs for activity and in the authors' model the Gpt (Mla) system is responsible, at least in part, for transport of GPLs to the OM?

We are not certain that PagP is only active on GPL in the outer leaflet. However there may be migration of GPL to the outer leaflet in the setting of the outer membrane defect of Mla mutants. It is possible even that MlaA could function to maintain lipid asymmetry by binding GPL in the outer membrane and presenting them to phospholipases and it could function as a receiver from MlaC to transport to the outer membrane. What seems unlikely is that there is retrograde transport of the GPL back to the inner membrane. PagP and phospholipases may function and the system may also transport from the inner membrane to the outer membrane.

9) Discussion paragraph three: The authors note that since the Mla (Gpt) system is upregulated in LOS deficient Acinetobacter, it suggests that Mla serves as a anterograde transporter as increased transport of GPLs may serve to compensate for lack of LOS in the outer membrane. Why couldn't it be the reverse, that the Mla system is upregulated in order to prevent too many GPLs in the OM. This goes back to what fills in the gaps when LOS is missing. After searching on PubMed, this reviewer found a recent paper from Boll et al., reporting that key lipoproteins seemed to be present in the outer leaflet of LOS deficient Acinetobacter.

The reviewer is correct to point out that the fact of Mla upregulation in LOS deficient Acinetobacter could be argued to support either a retrograde or an anterograde transport model. As such this point may not be useful for readers and we have removed it from the manuscript.

[Editors’ note: what follows is the authors’ plan to address the revisions.]

The current manuscript proposes that the Acinetobacter baumannii Mla system transports glycerophospholipid to the outer membrane. This statement challenges a large body of literature on both the E. coli and A. Baumannii systems (and to some extent V. Cholerae and H. influenza), all of which favor that Mla is instead involved in OM turnover and retrograde transport. After consultation with the reviewers, several key points would HAVE to be clarified for publication in eLife:

1) The reference to existing literature must be improved in particular to reconcile and potentially explain the differences between this study and published studies. If needed, additional experiments must be provided to resolve/address these conflicts. Along these lines:

– Powers and Trent, 2018, published evidence for retrograde transport in the same organism, which is not commented by the authors.

– Contrarily to this study, multiple laboratories have reported that deletion of mlA genes is not associated with increased sensitivity to vancomycin and in fact linked to vancomycin resistance gene induction.

– Roier et al. (Nat Comm, 2016) have shown that mlA deletion is linked to OMV production in V. Cholerae and H. influenza, which is not taken into account in this study. Can the authors rule out that lower GPL levels in the OM are not due to the loss of OM materials?

2) The pulse chase assay is an important innovation of this study and the major experimental evidence to support the proposed directionality of Mla. Proving the consistency of this assay is therefore paramount to support the authors’ conclusions. The authors need to provide all the controls confirming that the IM and the OM were correctly separated in sucrose density gradients, which apparently is performed for the first time in A. Baumannii. For this, sucrose gradients need to be shown, with a complete set of IM and OM controls. A series of helpful recommendations is provided in the individual reviews. Variations in the assay starting point and in particular the high variance in total labeled lipids should also be explained.

We appreciate the opportunity to respond to the thoughtful comments of the reviewers and to share our plan to revise our manuscript to improve suitability for publication in eLife. We would like to thank the reviewers and editors for their time and careful discussions surrounding this work. Given the concerns of the editors and reviewers, we would like to offer the following clarifications of key points. The reviewers seemed to have three main points of concern for which we believe we can rapidly respond and facilitate the publication of our manuscript. The three concerns were 1) that our paper did not take into account existing literature on published phenotypes of Acinetobacter baumannii mutants; 2) that we did not provide adequate controls around membrane separation; and 3) about the initial labeling variability and potential alternative explanations of the radiolabeled lipid transport assay. In the case of the literature as discussed below in greater detail, we will take into account in the writing of the manuscript this additional paper that was published after we submitted the manuscript, though we do not believe our results are in conflict with it. As to membrane separation, we ran controls for outer and inner membrane separation that should have been included in the original manuscript, and they are very convincing in that they show that we are able to separate the membranes, and this should allay concerns in this regard. As to potential alternative explanations within the radiolabeled lipid assay (i.e., outer membrane blebbing or lipid hydrolysis), we discuss below how our choice to present data as a ratio of labeled to unlabeled lipid—rather than total lipid—controls for any loss of lipid through alternative methods. In short, while outer membrane blebbing or lipid breakdown may influence the total lipid content, it would not be expected to specifically target either labeled or unlabeled lipid and therefore would not affect the ratios presented. Further, we also discuss in detail below why the initial labeling may have variability but how it is similarly controlled by the measurement of ratios of labeled lipids. We hope this will allay the reviewers’ and editors’ concerns and allow us to publish this work, which is likely to be of interest to the field and stimulate discourse about this topic.

We intend to rewrite portions of the manuscript that refer to existing literature to further explain why we believe our hypothesis regarding the anterograde transport function of the Mla system is not disproven by existing literature. We are aware that our model of the Mla system conflicts with the dominant idea regarding the system’s function, and as such, our attempts to address previous work must be careful and deliberate. We intend to make every attempt in rewriting the manuscript to explain the limitations of our results, that they are in Acinetobacter baumannii, and that they do not rule out the possibility that under certain circumstance the system could additionally function to maintain outer membrane lipid asymmetry. However, we believe our conclusions regarding an anterograde transport function for the Mla system to be valid and supported by our data. We also intend to revise the manuscript to reflect the following:

1) The study by Powers and Trent was published in PNAS after our submission of this manuscript to eLife, and as such it has not been addressed in earlier versions. We intend to revise the manuscript to comment on this data, and to address those authors’ conclusions that their data reflect evidence of a retrograde transport function for Mla in Acinetobacter baumannii. Powers and Trent first obtained A. baumannii deficient in lipooligosaccharide (LOS) by selection in the presence of polymyxin B. They then performed an evolution experiment, passaging the strains in cultures containing polymyxin B over 120 generations, at which point they observed significantly improved growth in the populations. These evolved populations were also observed to have increased resistance to antibiotics including vancomycin, bacitracin, and daptomycin, and to appear more morphologically consistent relative to the unevolved strains when observed microscopically. Whole genome sequencing of the evolved strains revealed mutations in mla genes in seven of the 10 evolved populations. They also observed frequent disruptions in pldA, encoding an outer membrane phospholipase, as well as in other envelope genes. To further study these effects, they then introduced clean deletions of mlaE and pldA to ATCC 19606, and selected for LOS-deficient bacteria by plating on polymyxin B. These double mutants demonstrated improved growth and resistance to antibiotics, but continued to display altered cellular morphology. Mutations in mlaE in an LOS-deficient background were also shown by RNA-seq to be differentially regulated for 120 genes.

The authors present their data as evidence in support of Mla as a retrograde transport system. We would point out that lacking in their data is examination of the membrane glycerophospholipid (GPL) profile in their LOS-deficient mutants. It is assumed by the authors that mla and pldA mutations have the effect of stabilizing a symmetric outer membrane produced in the absence of LOS by allowing GPL to fill in the outer leaflet, resulting in improved growth and antibiotic resistance. Given that the data suggests that Mla and PldA are selected against when LOS is absent, examination of the outer membrane GPL content might have supported the authors’ conclusions if it revealed an increase in GPL in mla and pldA mutants. Absent such data, it is not obvious to us that the authors have sufficiently ruled out alternative explanations for their observed phenomena. For example, we would question the mechanisms regulating the homeostasis of both the inner and outer membranes and the entire periplasmic space in the absence of LOS. The authors acknowledge earlier work that observed an increase in expression of mla genes upon initial loss of LOS in 19606 (PMID: 22024825, PMID: 27681618). Genes in the mla pathway were shown to have an up to 7.5-fold increase in gene expression upon loss of LOS. Powers and Trent assert that the function of Mla is deleterious in the absence of LOS, but perhaps what is deleterious is the profound upregulation of mla expression in the absence of LOS, combined with an active PldA. If Mla is an anterograde transporter, we can imagine this might create a situation in which GPL are rapidly removed from the inner membrane and then degraded in the outer membrane in excess of what the cell can support, and limiting both of these processes together simply allows the cell to achieve a new homeostasis. The fact is, understanding of the myriad processes regulating bacterial outer membrane assembly and integrity remains limited even when LOS is present, and so interpreting results such as these is extremely difficult. Lastly, most data support a model in which the outer membrane is not a fluid membrane but an organelle assembled in chunks of proteins and lipids that are further assembled as a mosaic; and therefore it might be somewhat unpredictable whether adding GPL to pieces of the outer membrane without LOS would be detrimental or not, this idea is consistent with the large number of suppressor mutations the authors observed in the LOS mutant. Therefore, the authors’ interpretation of their data may not be as absolute as stated in their discussion, rather it in our view represents one possibility unsupported with biochemical data.

2) We are not aware of studies in A. baumannii that clearly refute our result of increased sensitivity to vancomycin in mla mutants. The reviewers did not indicate a specific reference in their comments. Powers and Trent observed an increase in vancomycin resistance in ATCC 19606 when both mla and pldA are mutated together, but the effect of either single mutation on vancomycin resistance is not shown. Recently, work performed in Burkholderia has shown that Bcc mla mutants are more sensitive to Gram-positive antibiotics such as macrolides and rifampin, as well as fluoroquinolones, tetracyclines, and chloramphenicol (PMID: 29986943), while those effects were not observed in E. coli or Pseudomonas aeruginosa mla mutants. This indicates that the effect of mla on antibiotic susceptibility may be species-dependent, and may differ in A. baumannii from other organisms that have been studied for reasons that are not yet known. In some organisms under the conditions grown in the laboratory, the amount of GPL transported and the expression of Mla may vary and reflect outer membrane integrity, perhaps leading to variability within species. However, we are confident in our results which have been repeated many times in multiple different independently isolated mutants.

3) The reviewers were interested in the idea that GPL could be lost from the surface by a mechanism that involved the formation of vesicles. While this could be true, it is not really relevant to our assay measuring GPL labeled ratios over time, which was developed in part because of this possibility. We considered the possibility of loss of GPL from the outer membrane due to outer membrane vesicle formation or more likely, by the possible increased activation of outer membrane phospholipases. Both would have the effect of removing GPL from the outer membrane and would result in lower GPL levels in the outer membrane of mla mutants. Neither of these mechanisms would operate specifically on either labeled or unlabeled lipids. We understood the decrease in outer membrane GPL to be insufficient evidence of an anterograde transport function for mla, and for this reason we developed the stable isotope assay to control for these possibilities. The stable isotope assay gives insight into whether these results are due simply to mislocalization or degradation of outer membrane GPL, or if they can in fact be attributed to deficient anterograde GPL transport. This is because the peak intensity of each newly synthesized, C13-labeled GPL is normalized to the corresponding unlabeled version of that GPL species with each sample injected into the LC-MS/MS. The result is a molar ratio of labeled to unlabeled GPL for every membrane sample. With C13-acetate as the sole carbon source, we observe a gradual increase in the ratio of labeled GPL relative to unlabeled GPL over time. In wild type bacteria, these ratios for the inner and outer membranes track closely over time, which indicates that under these conditions GPL transport from the inner to the outer membrane occurs quite rapidly, as has been previously shown in E. coli reference. In mla mutants the ratio is both higher and typically increases at a greater rate in the inner membrane. Phospholipases and budding outer membrane vesicles will not distinguish between labeled and unlabeled GPL species, and so will not impact the ratios obtained with this assay. We will include in our manuscript a discussion of OMV production in mla mutants and clarify that our stable isotope assay accounts for that possibility.

We appreciate the need to assure readers that the separation of the inner and outer membranes performed in this study is valid. We intend to provide enzymatic controls to further verify proper separation of the two membranes. We already have data that support the integrity of our samples using an NADH assay, as shown in Author Response Image 1. Our sucrose gradient contains three distinct concentrations of sucrose, and inner and outer membranes separate into distinct bands that are collected individually. To limit any potential mixing of the membranes, the inner membrane is collected from the top of the tube while the outer membrane is collected by puncturing the bottom of the tube and allowing the bottom band to be collected. We do not collect a gradient series as the reviewers may have seen performed by other laboratories. The image below may prove helpful for readers to understand this process and visualize the clear separation of A. baumannii membranes in this procedure. We can state that we visually observed no difference in membrane separation of mla mutants relative to wild type, but we did not record images of those membrane separations side by side. If those images would be helpful we can repeat the fractionation procedure to demonstrate.

Author response image 1

The reviewers were concerned about some variability in the initial incorporation of label in the radiolabeled assay. We are aware that over the first hour of exposure, the initial rate at which C13-acetate was taken up and used to synthesize GPL appears to vary within a single strain over different experiments, as can be seen on examination of Supplementary file 2. We attribute this variability to minor differences in the optical density and growth time of the cultures following overnight growth with acetate as the sole carbon source, as well as to potentially slightly varying exposure to the stable isotope, although the exact reasons are not entirely clear. Although the starting values of isotope labeling varied, we consistently observed an increase in labeling in the inner membranes of mla mutants compared to the outer membrane that we did not observe in wild type, across multiple experiments. Comparing the relative percentage of isotope labeling on the inner membrane relative to the outer membrane in fact revealed consistent results across replicates, as we have presented in Figure 5C. The initial labeling differences do not necessarily reflect specific biological differences but represent a starting point for the assay. We present some raw data and not data normalized as many might submit because the initial labeling is not relevant and could be altered by the amounts of unlabeled acetate versus labeled in the culture.

[Editors’ notes: the authors’ response after being formally invited to submit a revised submission follows.]

After further consultation with the reviewers we decided that your response could adequately address the comments provided that in addition to the proposed changes the revision further contains the following new experiments:

1) Given the existing literature, we would like further that the discrepancy on the vancomycin sensitivity be resolved by verification of this phenotype in a different A. baumannii strain.

2) Completion of sucrose gradients with the proper controls to verify inner and outer membrane separation for both wild type and mla mutants (e.g. NADH oxidase assays). A clear description of the gradient methodology and images showing separation of the membrane as provided in the rebuttal letter should also be added.

We have rewritten portions of the manuscript that refer to existing literature to further explain why we believe our hypothesis regarding the anterograde transport function of the Mla system is not disproven by existing literature.

1) The study by Powers and Trent was published in PNAS after our submission of this manuscript to eLife, and as such it has not been addressed in earlier versions. We have revised the manuscript to comment on this data, and to address those authors’ conclusions that their data reflect evidence of a retrograde transport function for Mla in Acinetobacter baumannii.

2) Recently, work performed in Burkholderia has shown that Bcc mla mutants are more sensitive to Grampositive antibiotics such as macrolides and rifampin, as well as fluoroquinolones, tetracyclines, and chloramphenicol (PMID: 29986943), while those effects were not observed in E. coli or Pseudomonas aeruginosa mla mutants. This indicates that the effect of mla on antibiotic susceptibility may be species-dependent and may differ in A. baumannii from other organisms that have been studied for reasons that are not yet known. In some organisms under the conditions grown in the laboratory, the amount of GPL transported and the expression of Mla may vary and reflect outer membrane integrity. Given the reviewers’ concerns about the vancomycin sensitivity data, we repeated our experiments to test the sensitivity to vancomycin in another strain of A. baumannii, ATCC 5075. Comparison of wild type and mlaF deletion mutants in this strain did not show an increase in vancomycin sensitivity for mlaF mutants as we observed in ATCC 17978. In light of this, and to avoid confusion, we have removed the vancomycin MIC data from our revised manuscript.

3) We have included in our manuscript a discussion of OMV production in mla mutants and clarify that our stable isotope assay accounts for that possibility.

4) We appreciate the need to assure readers that the separation of the inner and outer membranes performed in this study is valid. We now provide enzymatic controls to further verify proper separation of the two membranes. Our data support the integrity of our samples using an NADH assay, as shown in Author Response Image 2. The image now provided in Figure 5—figure supplement 3 should prove helpful for readers to understand this process and visualize the clear separation of A. baumannii membranes in this procedure.

Author response image 2
https://doi.org/10.7554/eLife.40171.030

Article and author information

Author details

  1. Cassandra Kamischke

    Department of Microbiology, University of Washington, Seattle, United States
    Contribution
    Conceptualization, Data curation, Formal analysis, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  2. Junping Fan

    Department of Microbiology, University of Washington, Seattle, United States
    Contribution
    Data curation, Investigation, Visualization, Writing—review and editing
    Competing interests
    No competing interests declared
  3. Julien Bergeron

    1. Department of Biochemistry, University of Washington, Seattle, United States
    2. Department of Molecular Biology and Biotechnology, The University of Sheffield, Sheffield, United Kingdom
    Contribution
    Data curation, Formal analysis, Investigation, Visualization, Writing—original draft, Writing—review and editing
    Competing interests
    No competing interests declared
  4. Hemantha D Kulasekara

    Department of Microbiology, University of Washington, Seattle, United States
    Contribution
    Formal analysis, Supervision, Investigation, Methodology, Writing—review and editing
    Competing interests
    No competing interests declared
  5. Zachary D Dalebroux

    Department of Microbiology, University of Washington, Seattle, United States
    Contribution
    Data curation, Supervision, Investigation, Methodology
    Competing interests
    No competing interests declared
  6. Anika Burrell

    Department of Biochemistry, University of Washington, Seattle, United States
    Contribution
    Data curation, Formal analysis, Investigation
    Competing interests
    No competing interests declared
  7. Justin M Kollman

    Department of Biochemistry, University of Washington, Seattle, United States
    Contribution
    Supervision, Funding acquisition
    Competing interests
    No competing interests declared
  8. Samuel I Miller

    1. Department of Microbiology, University of Washington, Seattle, United States
    2. Department of Genome Sciences, University of Washington, Seattle, United States
    3. Department of Medicine, University of Washington, Seattle, United States
    Contribution
    Conceptualization, Supervision, Funding acquisition, Project administration, Writing—review and editing
    For correspondence
    millersi@uw.edu
    Competing interests
    No competing interests declared
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1638-2181

Funding

National Institute of Allergy and Infectious Diseases (U19AI107775)

  • Samuel I Miller

National Institutes of Health (R01GM118396)

  • Justin M Kollman

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Dale Whittington and Dr. Scott Edgar at the Mass Spectrometry Center, Department of Medicinal Chemistry, University of Washington for technical help with MS analysis; and Mauna Edrozo for technical help. This work was supported by a grant from NIAID, U19AI107775.

Senior Editor

  1. Wendy S Garrett, Harvard TH Chan School of Public Health, United States

Reviewing Editor

  1. Tâm Mignot, Aix Marseille University-CNRS UMR7283, France

Publication history

  1. Received: July 17, 2018
  2. Accepted: January 11, 2019
  3. Accepted Manuscript published: January 14, 2019 (version 1)
  4. Version of Record published: February 6, 2019 (version 2)

Copyright

© 2019, Kamischke et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 891
    Page views
  • 193
    Downloads
  • 0
    Citations

Article citation count generated by polling the highest count across the following sources: Crossref, PubMed Central, Scopus.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Evolutionary Biology
    2. Microbiology and Infectious Disease
    Mohammed Kaplan et al.
    Short Report Updated
    1. Microbiology and Infectious Disease
    Philip V'kovski et al.
    Research Article Updated