Summary Paragraph
Bin/Amphiphysin/RVS (BAR) domain proteins belong to a ubiquitous superfamily of coiled-coil proteins that influence membrane curvature in eukaryotes and are associated with vesicle biogenesis, vesicle-mediated protein trafficking, and intracellular signaling1–6. BAR domain proteins have not been identified in bacteria, despite certain organisms displaying an array of membrane curvature phenotypes7–16. Here we identify a prokaryotic BAR domain protein, BdpA, from Shewanella oneidensis MR-1, an iron reducing bacterium known to produce redox active membrane vesicles and micrometer-scale membrane extensions. BdpA is required for uniform size distribution of outer membrane vesicles and is responsible for scaffolding outer membrane extensions (OMEs) into membrane structures with consistent diameter and curvature. While a strain lacking BdpA produces OMEs, cryogenic transmission electron microscopy reveals more lobed, disordered OMEs rather than the membrane tubes produced by the wild type strain. Overexpression of BdpA promotes OME formation even during planktonic conditions where S. oneidensis OMEs are less common. Heterologous expression also results in OME production in Marinobacter atlanticus CP1 and Escherichia coli. Based on the ability of BdpA to alter membrane curvature in vivo, we propose that BdpA and its homologs comprise a newly identified class of prokaryotic BAR (P-BAR) domains that will aid in identification of putative P-BAR proteins in other bacterial species.
Introduction
Eukaryotic Bin/Amphiphysin/Rvs (BAR) domain-containing proteins generate membrane curvature through electrostatic interactions between positively charged amino acids and negatively charged lipids, scaffolding the membrane along the intrinsically curved surface of the antiparallel coiled-coil protein dimers17–20. Some BAR domain-containing proteins, such as the N-BAR protein BIN1, contain amphipathic helical wedges that insert into the outer membrane leaflet and can assist in membrane binding21. Other BAR domains can be accompanied by a membrane targeting domain, such as PX for phosphoinositide binding22,23, in order to direct membrane curvature formation at specific sites, as is the case with sorting nexin BAR proteins4. The extent of accumulation of BAR domain proteins at a specific site can influence the degree of the resultant membrane curvature24, and tubulation events arise as a consequence of BAR domain multimerization in conjunction with lipid binding25. Interactions between BAR domain proteins and membranes resolve membrane tension, promote membrane stability, and aid in localizing cellular processes, such as actin binding, signaling through small GTPases, membrane vesicle scission, and vesicular transport of proteins1,26,27. Despite our knowledge of numerous eukaryotic BAR proteins spanning a variety of modes of curvature formation, membrane localizations, and subtypes (N-BAR, F-BAR, and I-BAR), characterization of a functional prokaryotic BAR domain protein has yet to be reported.
Bacterial cell membrane curvature can be observed during the formation of outer membrane vesicles (OMV) and outer membrane extensions (OME). OMV formation is ubiquitous and has many documented functions9. OMEs are less commonly observed, remain attached to the cell, and various morphologies can be seen extending from single cells including Myxococcus xanthus14,15, flavobacterium strain Hel3_A1_488, Vibrio vulnificus10, Francisella novicida28, Shewanella oneidensis7,29–31, and as cell-cell connections in Bacillus subtilis32–34 and Eschericia coli35. Several bacterial proteins have demonstrated membrane tubule formation capabilities in vitro16,36–40, but despite the growing number of reports, proteins involved in shaping bacterial membranes into OMV/Es have yet to be identified. Recently, researchers have begun to suspect that OMV and OME formation has some pathway overlap8, and it is proposed that proteins are necessary to stabilize these structures13.
Shewanella oneidensis is a model organism for extracellular electron transfer (EET), a mode of respiration whereby electrons traverse the inner membrane, periplasm, and outer membrane via multiheme cytochromes to reach exogenous insoluble terminal electron acceptors, such as metals and electrodes41,42. It is also known to produce redox-active OMVs43 and OMEs coated with mulitheme cytochromes, particularly upon surface attachment7,30,43. However, little is known about their formation mechanism, control of shape or curvature, and electrochemical properties that influence EET function.
Results and Discussion
S. oneidensis OMVs are redox-active and enriched with BdpA
OMVs were purified from cells grown in batch cultures to characterize the redox features and unique proteome of S. oneidensis OMVs, as well as to identify putative membrane shaping proteins. Cryogenic transmission electron microscopy (cryo-TEM) tomography reconstruction slices of the purified samples showed uniform OMVs with the characteristic single membrane phenotype and an approximate diameter of 200 nm (Fig. 1a). Previous measurements suggest OMVs can reduce extracellular electron acceptors43 and that vesicles from G. sulfurreducens can mediate electron transfer44. Electrochemical activity of multiheme cytochrome complex MtrCAB and their ability to mediate micrometer-scale electron transport has been characterized in whole cells45, but no electrochemical characterization of OME/Vs has been reported that link activity to multiheme cytochromes. Here, electrochemical measurements of isolated OMVs were performed to determine if purified OMVs maintain the redox features when detached from cells. Cyclic voltammetry (CV) of isolated membrane vesicles adhered to a gold electrode via self-assembled monolayers show redox activity demonstrating electron transfer to and from the electrode interface (Fig. 1b). The first derivative (Fig. 1b inset) revealed a prominent peak with a midpoint potential of 66 mV and a smaller peak at −25 mV versus a standard hydrogen reference electrode (SHE). This midpoint potential is consistent with the characteristics of multiheme cytochromes such as MtrC/OmcA from previous microbial electrochemical studies45,46, suggesting that the extracellular redox molecules of the cellular outer membrane extends to OMVs.
The proteome of the OMVs was compared to the proteome of purified outer membranes extracted from whole cells. Using a label-free quantification method47, significant differences in the ratio of individual proteins in the vesicle to the outer membrane could be computed (log fold change) (Fig 1c). The proteome of the purified OMVs showed ~300 proteins were significantly enriched in the vesicles as compared to the outer membrane, and ~300 proteins were significantly excluded from the vesicles (Fig. 1c). MtrCAB cytochromes were neither significantly enriched nor excluded from the vesicles, consistent with the interpretation that vesicles could extend the respiratory surface area. Active protein sorting into eukaryotic vesicles is a coordinated process involving a protein sorting signal, localized membrane protein recruitment, initiation of membrane curvature induction, and coating nascent vesicles with membrane scaffolds48. Several proteins significantly enriched in the vesicles might contribute to OMV formation, such as murein transglycosylase, the peptidoglycan degradation enzyme holin, cell division coordinator CpoB, and a highly enriched putative BAR domain-containing protein encoded by the gene at open reading frame SO_1507, hereafter named BAR domain- like protein A (BdpA) (Fig. 1d).
Vesicle enrichment of BdpA led us to the hypothesis that BdpA could be involved in membrane shaping of OMVs based on the role of such proteins in eukaryotes. The C-terminal BAR domain of BdpA is predicted to span an alpha-helical region from AA 276-451 (E-value = 2.96e-03); however, since the identification of the protein is based on homology to the eukaryotic BAR domain consensus sequence (cd07307), it is possible that the BAR domain region extends beyond these bounds (Fig. 1d). Coiled coil prediction49 suggests BdpA exists in an oligomeric state of antiparallel alpha-helical dimers, as is the case for all known BAR domain proteins18,50–52. BdpA has an N-terminal signal peptide with predicted cleavage sites between amino acids 22-23, suggesting non-cytoplasmic localization (Fig. 1d). A galactose-binding domain-like region positioned immediately downstream of the signal peptide supports lipid targeting activity seen in other BAR domain proteins, such as the eukaryotic sorting nexins3 which have phox (PX) domains that bind phosphoinositides53. The S. oneidensis rough-type lipopolysaccharide (LPS) contains 2-acetamido-2-deoxy-D-galactose54, which suggests possible localization of the protein to the outer leaflet of the outer membrane.
BdpA controls size distribution of vesicles
To determine whether BdpA influences vesicle morphology, OMVs were harvested from wild type (WT) cells and cells in which the gene for BdpA had been deleted (ΔbdpA), and their diameters were measured by dynamic light scattering (DLS). WT OMVs (n=11) had a median diameter of 190 nm with little variability in the population (±21 nm), while the diameters of ΔbdpA OMVs (n=9) were distributed over a wider range with a median value of 280 nm ± 131 nm (Fig 2a). The data suggest BdpA controls vesicle diameter in membrane structures ex vivo, potentially acting by stabilizing OMVs. OMV frequency and size distribution was also measured in live cultures using a perfusion flow imaging platform and the membrane stain FM 4-64, as described previously29. S. oneidensis strains were monitored for OME/V production over the course of 5 hours (>5 fields of view per replicate, n=3). Spherical membrane stained extracellular structures were classified as OMVs, while larger aspect ratio (i.e. length greater than the width) structures were classified as OMEs. The duration of time-lapse imaging allowed tracking the progression of an OME/V over time. It was possible to quantify the proportion of cells producing ‘large’ vesicles, defined as those where the membrane was clearly delineated from the interior of the vesicles, typically >300 nm. ΔbdpA cells produced significantly more large vesicles compared to WT cells (Fig. 2b) even though both the overall frequency of vesiculation and extensions were the same (Fig. 2c). The size of S. oneidensis vesicles was more discrete than vesicles produced by other bacteria55,56 that do not contain a BdpA homolog, making it likely that BdpA is responsible for precise regulation of vesicle size. Previous studies showed that OMEs transition between large vesicles and OMEs over time29. BdpA appears to be involved in this transition due to the increased frequency of large vesicles from ΔbdpA cells.
BdpA constrains membrane extension morphology
The median diameter of the OMVs is also the apparent maximum diameter observed in outer membrane extensions29 suggesting BdpA influences membrane morphologies of both structures. As with the vesicles, WT and ΔbdpA cells made the same number of extensions in perfusion flow conditions (Fig. 2c). The resolution of fluorescence microscopy was insufficient to identify morphological differences between OMEs. To minimize sample processing of unfixed OMEs for cryo-TEM sample preparation, cells were deposited onto a glass coverslip instead of a perfusion flow chamber. BdpA was also expressed from a 2,4-diacetylphloroglucinol (DAPG)-inducible promoter57 (PPhlF-BdpA) in the ΔbdpA strain containing the plasmid p452-bdpA. After 3 hours post deposition on cover glass, OMEs can be seen extending from WT, ΔbdpA, and ΔbdpA p452-bdpA cells (Fig. 3, Supplemental Fig. 1, 5 fields of view, n=3). Similar to perfusion flow experiments (Fig. 2c), no statistically significant difference in the overall frequency of OME production was observed between the cells in static cultures.
Cryo-TEM was used to assess any morphological differences between the OMEs in each of the strains at the ultrastructural level. S. oneidensis OMEs from unfixed WT, ΔbdpA, and ΔbdpA p452-bdpA strains were visualized at 90 minutes (Supplemental Fig. 2) and 3 hours (Fig. 3) post deposition onto EM grids. At 90 minutes, WT OME phenotypes appeared narrow, tubule-like, and seldom interspersed with lobed regions (Supplemental figure 2a). In ΔbdpA OMEs, lobed regions are prevalent with irregular curvature (Supplemental figure 2b). Several narrow ΔbdpA p452-bdpA OMEs evenly interspersed with slight constriction points or “junction densities” were observed extending from a single cell (Supplemental Fig. 2c), suggesting that BdpA expression rescues the phenotype by constricting and ordering OMEs into narrow tubules. By 3 hours post inoculation, images of WT cells consistently show narrow, tubule-like OMEs (Fig. 3b, n=31). The ΔbdpA OMEs generally appear as lobed, disordered vesicle chains with irregular curvature, and vesicles can be observed branching laterally from lobes on the extensions (Fig. 3b, n=13). Nascent WT OMEs from previous studies also exhibited lateral branching of vesicles and lobes, but they exhibited uniform curvature and diameter between lobes and were observed immediately following OME formation29. Tubules were not observed in any ΔbdpA OMEs at 3 hours. OMEs from ΔbdpA p452-bdpA cells appear as a narrow tubules of a uniform curvature or as ordered vesicle chains (Fig. 3b, n=3).
Expression of BdpA results in OMEs during planktonic growth
S. oneidensis OMEs are more commonly observed during surface attachment rather than planktonic cultures7,29. BAR domain proteins can directly promote tubule formation from liposomes in vitro24, so inducing expression of an additional copy of the bdpA gene prior to attachment could result in OME formation even during planktonic growth. Growth curves were similar in cultures with the pBBR1-mcs2 empty vector in either of the WT (MR-1 pBBR1-mcs2) or ΔbdpA (ΔbdpA pBBR1-mcs2) background strains, but induction of bdpA in ΔbdpA p452-bdpA cells at higher concentrations of 1.25 and 12.5 μM 2,4-diacetylphloroglucinol affected the growth rate (Supplemental figure 3). Planktonic cultures inoculated from overnight cultures were induced with 12.5 μM DAPG for 1 hour, labeled with FM 4-64, and imaged by confocal microscopy. Neither WT (Fig. 4) nor MR-1 pBBR1-mcs2 exposed to 12.5 μM DAPG (not shown) produced OMEs immediately following deposition onto cover glass. However, 12.5 μM DAPG-induced S. oneidensis MR-1 p452-bdpA cells displayed OMEs immediately, ranging between 1-7 extensions per cell (Figure 4, Supplemental video 1). OME formation combined with growth rate data suggests bdpA expression in planktonic cultures redirects membrane production necessary for cell division into OMEs. The ultrastructure of OMEs resulting from expression of bpdA from MR-1 p452-bdpA cells was examined by cryo-TEM, but in this case samples from planktonic cultures were vitrified on EM grids after induction rather than incubation during induction on the EM grids. OMEs appear as tubule-like segments interspersed with pearled regions proximal to the main cell body (Fig. 4b). OMEs from the MR-1 p452-bdpA strain are observed as thin, tubule-like outer membrane vesicle chains, suggesting BdpA involvement in the constriction of the larger outer membrane vesicle chains into longer, tubule-like extensions with more evenly interspersed junction densities. The BdpA OME phenotype more closely resembles membrane tubules formed by the F-BAR protein Pacsin1 from eukaryotic cells, showing a mixture of tubule regions interspersed with pearled segments58,59.
BdpA-mediated membrane extensions in Marinobacter atlanticus CP1 and E. coli
To test the effect of expressing BdpA in an organism with no predicted BAR domain-containing proteins and no apparent OME production, BdpA was expressed in Marinobacter atlanticus CP160. Marinobacter and Shewanella are of the same phylogenetic order (Alteromonadales) and have been used for heterologous expression of other S. oneidensis proteins, such as MtrCAB61,62. Upon exposure to DAPG, M. atlanticus containing the p452-bdpA construct (CP1 p452-bdpA) form membrane extensions (Figure 4). OMEs ranged from small membrane blebs to OME tubules extending up to greater than 10 μm in length from the surface of the cell (Supplemental Fig. 4). As noted previously, variation in the tubule phenotypes are commonly seen in tubules from eukaryotic F-BAR proteins58,59, showing possible mechanistic overlap of mutable membrane curvature functionalities between these two separate BAR domain proteins.
In previous membrane curvature formation experiments with eukaryotic BAR domain proteins, localized BAR domain protein concentrations affected the resultant shape of the membranes, ranging from bulges to tubules and branched, reticular tubule networks at the highest protein densities63–65. We predicted that expression of BdpA in cells optimized for protein overexpression, such E. coli BL21(DE3), would show OMEs resembling structures previously observed from eukaryotic BAR protein experiments in vitro. While the uninduced E. coli BL21(DE3) p452-bdpA cells had uniform, continuous cell membranes similar to those of plasmid-free BL21(DE3) cells under the conditions tested, E. coli BL21(DE3) cells containing the p452-bdpA vector induced with DAPG had outer membrane extensions and vesicles (Figure 4). When visualized over time, OMEs progressed towards a network of reticular membrane structures extending from the cell (Fig. 4c). After 30 minutes, additional membrane blebs were observed that developed into elongated OMEs by 60 minutes. Growth of E. coli OMEs was coincident with shrinking of the cell body (from initial cell length = 4.457 μm to 3.479 μm at 60 minutes), supporting direct membrane sculpting activity of BdpA.
P-BAR: a new BAR domain subtype
The discovery of a novel, functional BAR domain protein in prokaryotes provokes questions into the evolutionary origin of BAR domains, such as whether the BdpA BAR domain in Shewanella arose as a result of convergent evolution, a horizontal gene transfer event, or has a last common ancestor across all domains of life. BdpA homologs were identified by PSI-BLAST in several other organisms, ranging from other species of Shewanella to Alishewanella, Rheinheimera, and Cellvibrio (Supplemental Fig. 5). The current BAR domain pfam Hidden Markov Model (HMM) prediction analysis identified BAR domain features in only 5 of the 52 prokaryotic homologs despite greater than 90% homology to S. oneidensis BdpA. Functional analysis will be necessary to determine if these homologs contain unpredicted BAR domains and merit inclusion in the generation of a new BAR domain pfam seed alignment. The resultant alignment was used to generate a maximum likelihood phylogenetic tree showing evolutionary relatedness of BdpA orthologs to the BAR domain prediction sequences (Supplemental Fig. 5). The 5 BdpA orthologs predicted to contain a BAR domain based on the current model were subsequently aligned with representative known BAR proteins from the various BAR domain subtypes (N-BAR, F-BAR, and I-BAR)66. BdpA and its prokaryotic orthologs cluster separately from the eukaryotic BAR proteins in their own distinct clade (Fig. 5), suggesting that while BdpA contains a functional BAR domain, it represents its own class of BAR domain, hereafter named P-BAR (Prokaryotic BAR). It seems likely that the P-BAR domain arose as a result of horizontal gene transfer from a eukaryote due to the prevalence of eukaryotic coiled-coil proteins with predicted homology to BdpA after 2 iterations of PSI-BLAST. However, the branch lengths and low bootstrap values supporting the placement of P-BAR relative to other BAR domain subtypes make it challenging to directly infer the evolutionary history of P-BAR domains. Discovery of other putative P-BAR proteins would help to build this analysis, and if future comparative proteomics analysis of OME/Vs demonstrates overlapping activity of BdpA with preferential cargo loading into OME/Vs, it could hint at the evolutionary origins of vesicle-based protein trafficking. Conservation of BAR domain proteins supports the notion that three-dimensional organization of proteins in lipid structures is as important to prokaryotes as it is eukaryotes, and suggests additional novel P-BAR proteins are waiting to be discovered.
Methods
Bacterial strains, plasmids, and medium
The bacterial strains used in this study can be found in Supplemental Table 1. S. oneidensis strains were grown aerobically in Luria Bertani (LB) media at 30°C with 50 μg/mL kanamycin when maintaining the plasmid. To observe membrane extensions, cells were centrifuged and resuspended in a defined media comprised of 30 mM Pipes, 60 mM sodium DL-lactate as an electron donor, 28mM NH4Cl, 1.34 mM KCl, 4.35 mM NaH2PO4, 7.5 mM NaOH, 30 mM NaCl, 1mM MgCl2, 1 mM CaCl2, and 0.05 mM ferric nitrilotriacetic acid30. Marinobacter atlanticus CP1 strains were grown in BB media (50% LB media, 50% Marine broth) at 30°C with 100 μg/mL kanamycin to maintain the plasmids as described previously60.
Inducible BdpA expression plasmids were constructed for use in S. oneidensis MR-1, M. atlanticus CP1, and E. coli BL21(DE3) using the pBBR1-mcs2 backbone described previously60. The Marionette sensor components (phlF promoter, consitutively expressed PhlF repressor, and yellow fluorescence protein (YFP)) cassette from pAJM.45257 was cloned into the pBBR1-mcs2 backbone, and the YFP cassette was replaced with the gene encoding BdpA by Gibson assembly (primers in Supplemental Table 1). The resulting plasmid was given the name p452-bdpA. The Gibson assembly reactions were electroporated into E. coli Top10 DH5α cells (Invitrogen), and the sequences were confirmed through Sanger sequencing (Eurofins genomics). Plasmid constructs were chemically transformed into conjugation-competent E. coli WM3064 cells for conjugative transfer into the recipient bacterial strains of S. oneidensis MR-1 and M. atlanticus CP1. The same BdpA expression vector was transformed into E. coli BL21(DE3) cells (Invitrogen) by chemical transformation.
Generation of a scarless ΔbdpA knockout mutant of S. oneidensis was performed by combining 1 kilobase fragments flanking upstream and downstream from bdpA by Gibson assembly into the pSMV3 suicide vector. The resultant plasmid pSMV3_1507KO was transformed into E. coli UQ950 cells for propagation. Plasmid sequences were confirmed by Sanger sequencing before chemical transformation into E. coli UQ950 for conjugation into S. oneidensis. Conjugation of pSMV3_1507KO into S. oneidensis MR-1 was performed as described previously31.
Purification of Outer Membrane Vesicles
S. oneidensis MR-1 cells were grown in LB in 1L non-baffled flasks at 30° C at 200 RPM. When an OD600 of 3.0 was reached, cells were pelleted by centrifugation at 5000 x g for 20 min at 4°C, resulting supernatant was filtered through a 0.45 μm filter to remove remaining bacterial cells. Vesicles were obtained by centrifugation at 38,400 x g for 1 h at 4°C in an Avanti J-20XP centrifuge (Beckman Coulter, Inc). Pelleted vesicles were resuspended in 20 ml of 50 mM HEPES (pH 6.8) and filtered through 0.22 μm pore size filters. Vesicles were again pelleted as described above and finally resuspended in 50 mM HEPES, pH 6.8, except for vesicle preparations used for electrochemistry which were suspended in 100 mM MES, 100 mM KCl, pH 6.8. Extracellular DNA, flagella, and pili can all be co-purified. Protocol was adapted from67.
Dynamic Light Scattering
Distribution of vesicle diameters were measured with Wyatt Technology’s Möbiuζ dynamic light scattering instrument.
Electrochemistry
CHA Industries Mark 40 e-beam and thermal evaporator was used to deposit a 5 nm Ti adhesion layer and then a 100 nm Au layer onto cleaned glass coverslips (43X50 NO. 1 Thermo Scientific Gold Seal Cover Glass, Portsmouth NH, USA). Self-assembled monolayers were formed by incubated the gold coverslip in a solution of 1mM 6-mercaptohexanoic acid in 200 proof ethanol for at least 2 hours. Electrode was then rinsed several time in ethanol followed by several rinses in milliQ water. The SAMs layer was then activated by incubation in 100 mM N-(3-Dimethylaminopropyl)-N′-ethylcarbodiimide hydrochloride and 25 mM N-hydroxysuccinimide, pH 4, for 30 minutes. A sample of outer membrane vesicles was deposited on the surface of the electrode and incubated at room temperature overnight in a humid environment. Cyclic voltammetry was performed in a 50 mL 3 electrode half-cell completed with a platinum counter electrode, and a 1 M KCl Ag/AgCl reference electrode electrical controlled by a Gamry 600 potentiostat (Gamry, Warminster, PA). The whole experiment was completed in an anaerobic chamber with 95% nitrogen, 5% hydrogen atmosphere.
Proteomics
Vesicle samples were prepared as described above. S. oneidensis outer membrane (OM) was purified via the Sarkosyl method described by Brown et al.68. A 50 mL overnight culture of cells was harvested by centrifugation at 10,000 × g for 10 min. The cell pellet suspended in 20 mL of 20 mM ice-cold sodium phosphate (pH 7.5) and passed four times through a French Press (12000 lb/in2). The lysate was centrifuged at 5,000 × g for 30 min to remove unbroken cells. The remaining supernatant was centrifuged at 45,000 × g for 1 h to pellet membranes. Crude membranes were suspended in 20 mL 0.5% Sarkosyl in 20 mM sodium phosphate and shaken horizontally at 200 rpm for 30 min at room temperature. The crude membrane sample was centrifuged at 45,000 × g for 1 h to pellet the OM. The pellet of OM was washed in ice-cold sodium phosphate and recentrifuged.
To prepare for mass spectrometry samples were treated sequentially with urea, TCEP, iodoactinamide, lysl endopeptidase, trypsin, and formic acid. Peptides were then desalted by HPLC with a Microm Bioresources C8 peptide macrotrap (3×8mm). The digested samples were subjected to LC-MS/MS analysis on a nanoflow LC system, EASY-nLC 1200, (Thermo Fisher Scientific) coupled to a QExactive HF Orbitrap mass spectrometer (Thermo Fisher Scientific, Bremen, Germany) equipped with a Nanospray Flex ion source. Samples were directly loaded onto a PicoFrit column (New Objective, Woburn, MA) packed in house with ReproSil-Pur C18AQ 1.9 um resin (120A° pore size, Dr. Maisch, Ammerbuch, Germany). The 20 cm x 50 μm ID column was heated to 60°C. The peptides were separated with a 120 min gradient at a flow rate of 220 nL/min. The gradient was as follows: 2–6% Solvent B (7.5 min), 6-25% B (82.5 min), and 25-40% B (30 min) and to 100% B (9min). Solvent A consisted of 97.8% H2O, 2% ACN, and 0.2% formic acid and solvent B consisted of 19.8% H2O, 80% ACN, and 0.2% formic acid. The QExactive HF Orbitrap was operated in data dependent mode with the Tune (version 2.7 SP1build 2659) instrument control software. Spray voltage was set to 2.5 kV, S-lens RF level at 50, and heated capillary at 275 °C. Full scan resolution was set to 60,000 at m/z 200. Full scan target was 3 × 106 with a maximum injection time of 15 ms. Mass range was set to 300−1650 m/z. For data dependent MS2 scans the loop count was 12, target value was set at 1 × 105, and intensity threshold was kept at 1 × 105. Isolation width was set at 1.2 m/z and a fixed first mass of 100 was used. Normalized collision energy was set at 28. Peptide match was set to off, and isotope exclusion was on. Data acquisition was controlled by Xcalibur (4.0.27.13) and all data was acquired in profile mode.
Bioinformatics
Putative BAR domain SO_1507 (BdpA) was identified in search of annotation terms of S. oneidensis MR-1. The conserved domain database (CDD-search)(NCBI) was accessed to identify the position-specific scoring matrix (PSSM) that matched and specific region of SO_1507 that represented the BAR domain. It was confirmed that a region 276-421 matched to BAR superfamily cl12013 and specifically to the family member BAR cd07307. LOGICOIL multi-state coiled-coil oligomeric state prediction was used to determine the presence of coiled-coils within BdpA. SignalP 6.1 was used to detect the presence of the signal peptide and cellular localization of BdpA.
A PSI-BLAST69 search against the NCBI nr database was performed using the BdpA BAR sequence as the initial search seed to determine how prevalent the BdpA BAR domain is in related species. Conserved BdpA orthologs were annotated as hypothetical proteins in all of the species identified. In the initial round, 24 proteins were found from other organisms identified as Shewanella with a high conservation among the proteins and another 28 proteins were found in more distant bacteria species that had similarity of 65% to 44 %. A second iteration identified a few proteins much more distantly related from bacterial species and then proteins from eukaryote phylum Arthropoda that were annotated as being centrosomal proteins. All of the found proteins from bacterial species were hypothetical proteins with no known function. Only five of the proteins from the search returned hits to the PSSM of the BAR cd07307. The identity among the proteins was very high and examination of the proteins suggests that a functional form similar to the BAR domain would result for all the found proteins. Overall this places the original protein SO_1507 as a protein that just barely meets criteria via PSSM models to be assigned a matching the BAR domain while the rest of the proteins found have enough differences to fail to match the BAR model while still being very similar to SO_1507. An attempt was made to build up a HMM (Hidden Markov Model) using HMMer to use for searching for other proteins that might match but as with the PSI-BLAST search only the proteins that formed the model returned as good matches. So there appear to be a tight clade of very similar proteins with very little differentiation in the sequence. This indicates that while sequence homology between BdpA and the existing BAR domain consensus sequence predicted the BAR domain region in BdpA using hmmer or NCBI tools, the sequence conservation is at the cusp of a positive hit by the HMM since other closely related (>90% homology) BdpA orthologs were not predicted to contain a BAR domain by this method. The most homologous eukaryotic protein to BdpA (27%) is a putative centrosomal protein in Vollenhovia emeryi (accession #: XP_011868153) that is predicted to contain an amino terminal C2 membrane binding domain and a carboxy-terminal SMC domain within a coiled-coil region. Despite CDD search failing to predict the presence of a BAR domain in this protein, it does not preclude the presence of one, pending an updated BAR pfam HMM.
Confocal microscopy
For in vivo imaging of intrinsic outer membrane extension production, S. oneidensis MR-1 strains were grown in LB media overnight, washed twice with SDM, and diluted to an OD600 of 0.05 in 1 mL of SDM with appropriate antibiotics. Prior to pipetting, ~1cm of the pipette tip was trimmed to minimize shear forces during transfer. 100 μL of each culture was labeled with 1 μL 1M FM 4-64 to visualize the cell membranes. After staining, 10 μL of the labeled cell suspension was gently pipetted onto 22 × 22 mm No.1 cover glass (VWR) and sealed onto glass slides with clear acrylic nail polish (for confocal imaging) or onto chambered cover glass (for widefield fluorescence). On average, intrinsic membrane extension formation could be observed starting after 45 minutes sealed onto cover glass. Diluted cells were induced with 12.5 μM DAPG for 1 hour at 30°C with 200 RPM shaking agitation for planktonic OME production. Cells were labeled with FM 4-64 and sealed onto glass slides as before. Induced OMEs were imaged immediately after mounting onto slides.
Confocal images were taken by a Zeiss LSM 800 confocal microscope with a Plan-Apochromat 63x/1.4 numerical aperture oil immersion M27 objective. FM 4-64 fluorescence was excited at 506 nm: 0.20% laser power. Emission spectra was detected from 592-700 nm using the LSM 800 GaAsP-Pmt2 detector. To capture the dynamics of the OMEs, images were collected over the designated length of time between 0.27 – 0.63 seconds per frame. Single frame time series images were collected of either a 50.71 μm by 50.71 μm (2x zoom) or a 20.28 μm by 20.28 μm (5x zoom) field of view. Images were recorded using the Zeiss Zen software (Carl Zeiss Microscopy, LLC, Thornwood, NY, USA).
Perfusion flow microscopy
For OME statistics comparing S. oneidensis strains MR-1 and ΔbdpA, cells were pre-grown aerobically from frozen (−80°C) stock in 10 mL of Luria-Bertani (LB) broth (supplemented with 50 μg/mL Kanamycin for strains with plasmid) in a 125-mL flask overnight at 30°C and 225 rpm. The next day, the stationary phase (OD600 3.0 – 3.3) preculture was used to inoculate 1:100 into 10 mL of fresh LB medium in a 125-mL flask. After ~6 hours at 30°C and 225 rpm, when the OD600 was 2.4 (late log phase), 5 mL of cells were collected by centrifugation at 4226 x g for 5 min and washed twice in defined medium. The perfusion chamber, microscope, and flow medium described previously7,29,30 were used for all perfusion flow OME statistics experiments. During each 5 hour imaging experiment, the perfusion chamber was first filled with this flow medium, then <1 mL of washed cells were slowly injected for a surface density of ~100-300 cells per 112 x 112 μm field of view on a Nikon Ti-E inverted microscope. Cells were allowed to attach for 5-15 minutes on the coverslip before perfusion flow was resumed at a volumetric flow rate of 6.25 ± 0.1 μL/s. Cells and OMEs were visualized with the red membrane stain FM 4-64FX in the flow medium (0.25 μg/mL of flow medium). A total of 1,831 wild type and 2,265 ΔbdpA cells were used for extension and vesicle quantification.
Cryo transmission electron microscopy
Shewanella strains were streaked onto LB plates with or without kanamycin and allowed to incubate 3 days on a benchtop. The night before freezing, individual colonies were inoculated into 3 ml LB +/− kanamycin and incubated at 30 °C overnight with 200 rpm shaking. The following morning optical densities of the cultures were measured at 600nm and adjusted to a final OD600 of 1. Cells were pelleted at 8,000 rpm for three minutes for buffer exchange/washes. For the ΔbdpA_p452-bdpA transformed cells, 12.5 μM DAPG was added. A freshly glow discharged 200 mesh copper grid with R2/1 Quantifoil carbon film was placed into a concavity slide. Approximately 150 μl of a 1:10 dilution of the cell suspensions, with or without the inducer, was added to cover the grid. A glass coverslip was then lowered onto the concavity to exclude air bubbles. The edges of the coverslip were then sealed with nail polish to prevent media evaporation. The slide assembly was then incubated in a 30 °C incubator for 1.5 to 3 hours. Immediately prior to plunge freezing, the top coverslip was removed by scoring the nail polish with a razor blade. TEM grids with cells were gently retrieved with forceps and loaded into a Leica grid plunge for automated blotting and plunging into LN2-cooled liquid ethane. Vitrified grids were transferred to a LN2 storage dewar. Imaging of frozen samples was performed on either a Titan (ThermoFisher Scientific) microscope equipped with a Gatan Ultrascan camera and operating at 300 kV or a Talos (ThermoFisher Scientific) equipped with a Ceta camera and operating at 200 kV. Images were acquired at 10,000 to 20,000 X magnification and were adjusted by bandpass filtering.
Author Contributions
DP and LZ conceived the study independently then combined projects when complementary data on BdpA was discovered. LZ purified OMVs, prepared samples for LC MS-MS, and performed DLS measurements. LZ and SX made electrochemical measurements and analysis. DP conducted BdpA domain prediction and validation analysis, generated the p452-bdpA plasmid, ΔbdpA and p452-bdpA strains. DP and GC conducted fluorescence and confocal imaging experiments, and DP, LZ, and GC analyzed the data. LB adapted the Marionette sensor (PphlF-YFP) into pBBR1-mcs2. LZ and LAM performed cryo-TEM of OMVs. GC conducted perfusion flow imaging experiments. GC and LZ analyzed perfusion flow system data. CH, DP, and LD performed cryo-TEM experiments of OMEs and image processing / analysis. DP and AM generated phylogenetic data, and DP, AM, and BE analyzed the data. DP, LZ, CM, GC, AM, LAM, BE, GJJ, LD, MEN, and SG provided data interpretation. DP, LZ, MEN, and SG wrote the manuscript, with input from all coauthors.
Acknowledgements
We thank Dr. Jeffery Gralnick for helpful discussions and advice; Dr. Adam Meyer and Dr. Chris Voigt for the DAPG-inducible Marionette promoter; Dr. Annie Moradian and Dr. Mike Sweredoski and the California Institute of Technology Proteome Exploration Lab for useful discussions on the preparation and analysis of proteomics data. Some of the cryo-TEM work was done in the Beckman Institute Resource Center for Transmission Electron Microscopy at Caltech. This work was supported by the United States Department of Defense Synthetic Biology for Military Environments (SBME) Applied Research for the Advancement of Science and Technology Priorities (ARAP) program. Work in ME-N's lab was supported by the U.S. Office of Naval Research Multidisciplinary University Research Initiative Grant No. N00014-18-1-2632. LAZ was partially supported by the National Science Foundation grant DEB-1542527. SX was supported by the Division of Chemical Sciences, Geosciences, and Biosciences, Office of Basic Energy Sciences of the U.S. Department of Energy through grant DE-FG02-13ER16415. Work in GJJ’s lab was supported by the National Institute of Health (GM122588 to GJJ).