A new paradigm for outer membrane protein biogenesis in the Bacteroidota – Nature

Bacterial strains and growth conditions

All strains and plasmids used in this work are listed in Supplementary Tables 2 and 3. F. johnsoniae was routinely cultured aerobically in Casitone yeast extract (CYE) medium⁴⁵ at 30 °C with shaking. For some physiological studies the cells were cultured in PY2 medium⁴⁶ as indicated below. For experiments testing growth on complex sugars cells were cultured in a 96-well plate in a CLARIOstarPlus plate reader using modified minimal A medium²⁷ and containing 0.25% (w/v) of either carob galactomannan (Megazyme, 11078-30-1) or tamarind xyloglucan (Megazyme, 37294-28-3) as the sole carbon source. E. coli strains were routinely grown aerobically in LB medium at 37 °C with shaking, or on LB agar plates. Where required, 100 µg ml⁻¹ erythromycin was used in the growth medium for F. Johnsoniae. 100 µg ml⁻¹ ampicillin or 50 µg ml⁻¹ kanamycin were used in the growth medium for E. coli. aTC (CAY10009542-50 mg, Cambridge Bioscience) was used as a final concentration of 0.2 µg ml⁻¹ (liquid culture) and 2 µg ml⁻¹ (agar plates).

Genetic constructs

Plasmids were constructed by Gibson cloning⁴⁷ using the primers and target DNA in Supplementary Table 4. Suicide and expression plasmids were introduced into the appropriate F. johnsoniae background strain by triparental mating as previously described⁴⁶. Chromosomal modifications were introduced using the suicide vector pYT313 harbouring the counter-selectable sacB gene as previously described⁴⁸. All plasmid constructs and chromosomal modifications were confirmed by sequencing.

Construction of a tightly regulated gene expression system for F. johnsoniae

The aTC-inducible systems for the depletion of essential Bam_FJ components (Extended Data Fig. 7a) were based on the native F. johnsoniae ompA and fjoh_0824 promoters and contain the 100 bp upstream of ompA or fjoh_0824. Guided by the observations of Lim et al.⁴⁹, a tetO2 site (TetR binding site) was inserted upstream of the conserved −33 motif in these promoters and another tetO2 site downstream of the conserved −7 motif generating the synthetic promoters P_ompAinduc and P_{fjoh_0824induc} (Extended Data Fig. 7b). The constructs also contain tetR under the control of an additional copy of the constitutive F. johnsoniae ompA promoter. The final inducible systems containing the gene to be induced were integrated into the chromosome at an assumed phenotypically neutral site^26,36 by replacing fjoh_4538 to fjoh_4540.

The designed inducible systems were validated using strains in which a NanoLuc reporter gene⁵⁰ was placed under the control of the chromosomally integrated aTC-inducible systems (Extended Data Fig. 7c). Overnight cultures of these strains were diluted 1:100 into fresh CYE medium in the absence or presence of 0.2 µg ml⁻¹ aTC and cultured for 6 h to mid-exponential phase (OD₆₀₀ ~ 0.6). Cells were collected and resuspended in PY2 medium to OD₆₀₀ = 0.6. A volume of 50 µl of cell resuspension was mixed with 50 µl of reaction solution (48 µl PY2 medium supplemented with 2 µl of furimazine (Promega)) in a 96-well plate and the luminescence signal measured in a CLARIOstar^Plus plate reader.

Strains to enable the depletion of the essential BAM_Fj subunits were constructed by introducing a copy of the target gene under the control of the designed inducible system into the chromosome at the phenotypically neutral site. The native copy of the target gene was then deleted in the presence of aTC to allow expression of the introduced copy of the gene.

Purification of BAM_Fj and SusCDE complexes

To purify complexes containing Twin-Strep tagged BamA, the relevant strain was cultured for 22 h in CYE medium using 1 l culture volume in 2.5 l flasks. A total culture volume of 12 l was used for sample preparations for structure determination, and 4 l of culture was used for analytical purifications of BAM_Fj variants. Cells were collected by centrifugation at 12,000g for 30 min and stored at −20 °C until further use. All purification steps were carried out at 4 °C. Cell pellets were resuspended in buffer W (100 mM Tris-HCl pH 8.0, 150 mM NaCl, 1 mM EDTA) containing 30 μg ml⁻¹ DNase I, 400 μg ml⁻¹ lysozyme and 1 mM phenylmethylsulfonyl fluoride (PMSF) at a ratio of 5 ml of buffer to 1 g of cell pellet. Cells were incubated on ice for 30 min with constant stirring before being lysed by two passages through a TS series 1.1 kW cell disruptor (Constant Systems) at 30,000 PSI. Unbroken cells were removed by centrifugation at 20,000g for 20 min. The supernatant was recovered and total membranes were collected by centrifugation at 230,000g for 75 min. Membranes were resuspended in buffer W to a protein concentration of 6.5 mg ml⁻¹ and solubilized by incubation with 1% (w/v) lauryl maltose neopentyl glycol (LMNG, Anatrace) for 2 h. Insoluble material was removed by centrifugation at 230,000g for 75 min. Endogenous biotin-containing proteins were masked by addition of 1 ml BioLock solution (IBA Lifesciences) per 100 ml of supernatant and incubation for 20 min with constant stirring. The solution was then circulated through a Strep-TactinXT 4Flow High Capacity column (IBA Lifesciences) overnight. The column was washed with 10 column volumes of buffer W containing 0.01% LMNG (buffer WD) and bound proteins were eluted with 6 column volumes Strep-TactinXT BXT buffer (IBA Lifesciences) containing 0.01% LMNG. The eluate was concentrated to 500 μl using a 100-kDa molecular weight cut-off (MWCO) Amicon ultra-15 centrifugal filter unit (Merck) and then injected onto a Superose 6 Increase 10/300 GL column (Cytiva) previously equilibrated in buffer WD. Peak fractions were collected and concentrated using a 100-kDa MWCO Vivaspin 500 column (Sartorius).

Purification of SusCDE complexes with a N-terminal Twin-Strep tag on SusC was carried out by the same protocol.

Peptide mass fingerprinting

Samples were excised from Coomassie-stained gels. For whole sample proteomic analysis, SDS–PAGE was carried out only until the sample had fully entered the gel and the protein smear at the top of the gel was excised. Samples were subject to in-gel trypsin digestion and electrospray mass spectrometry at the Advanced Proteomics Facility (University of Oxford, UK).

Immunoblotting

Immunoblotting was carried out as previously described¹⁹. Antibodies against BAM_Fj subunits, Sus proteins and SkpA were raised in rabbits against His-tagged recombinant proteins produced using the plasmids listed in Supplementary Table 3. Antiserum against OmpA³⁴ was provided by S. Shibata and antiserum against SprF³⁶ by M. McBride. The following commercial antisera were used: anti-Strep-tag (34850 Qiagen), anti-GroEL (G6532 Merck), anti-ALFA-Tag (N1582 Synaptic Systems GmbH), anti-His-tag (H1029-100UL Merck Life Science), anti-HA-tag (26183 Thermo Fisher Scientific), anti-mouse IgG peroxidase conjugate (A4416 Merck) and anti-rabbit IgG peroxidase conjugate (31462 Pierce). Antibodies were used at the following dilutions: anti-His-tag and anti-HA-tag, 1:1,000; anti-SprF, 1:2,500; anti-BamH, anti-BamM, anti-BamP, anti-SusC, anti-SusD, anti-SusE, anti-SkpA, anti-Strep-tag and anti-ALFA-tag, 1:3,000; anti-OmpA and anti-GroEL, 1:50,000.

Original uncropped gels and immunoblots are shown in Supplementary Fig. 2.

Darobactin inhibition experiments

E. coli or F. johnsoniae strains were cultured, respectively, in LB and CYE medium (supplemented with erythromycin if carrying p^TSBamP or p^TSBamP4 plasmids). Five-millilitre starter cultures were grown aerobically overnight at 30 °C, then diluted into 5 ml fresh medium to OD₆₀₀ = 0.02 and then grown to OD₆₀₀ between 0.6 to 0.8. The cultures were then diluted with fresh medium to OD₆₀₀ = 0.006. Fifty microlitre aliquots were transferred into a 96-well plate and mixed with 50 µl of the required concentration of darobactin solution in growth medium. The minimum inhibitory concentration (MIC) for darobactin was assessed after overnight incubation at 30 °C in a CLARIOstarPlus plate reader.

BamP pull-downs

Strains with pCP11-derived plasmids expressing N-terminal Twin-Strep-tagged BamP or BamP4 under the control of a remA promoter were grown aerobically overnight at 30 °C in erythromycin-supplemented CYE medium. The culture was diluted into 100 ml fresh medium to OD₆₀₀ = 0.02 and grown to an OD₆₀₀ = 0.8–1.0. Cells were then collected by centrifugation at 8,000g for 10 min and resuspended in 3 ml of buffer W containing 30 µg ml⁻¹ DNase I, 400 µg ml⁻¹ lysosome and 1 mM PMSF. The cells were incubated for 30 min at 4 °C, and then lysed by sonication for 3 min on ice using a Sonics Vibra Cell Ultrasonic Processor VCX 130 with a 6 mm probe at 40% amplitude, with a 10 s on to 10 s off cycle. Unbroken cells were removed by centrifugation at 20,000g for 20 min. The supernatant fraction was then centrifuged at 200,000g for 1 h to pellet total membranes. The membrane pellets were resuspended to a protein concentration of 6.5 mg ml⁻¹ with buffer W and solubilized by incubation with 1% (w/v) LMNG for 2 h. Insoluble material was removed by centrifugation at 230,000g for 1 h, and the recovered supernatant supplemented with 1% BioLock solution before mixing with 50 µl Strep-TactinXT 4Flow lbeads (IBA Lifesciences) that had been equilibrated in buffer WD. Samples were rotated slowly at 4 °C for 2 h and then transferred into Mini Bio-Spin Chromatography columns (Bio-Rad, 7326207), and centrifugation at 100g for 1 min. The beads were washed 3 times with 250 µl buffer WD and bound proteins then eluted with 150 µl of 1× Strep-TactinXT BXT buffer containing 0.01% LMNG. The elute was concentrated to 30 µl using a 10 kDa MWCO Vivispin500 centrifugal concentrator (VS0102, Sartorius).

BAM_Fj subunit depletion experiments

The desired depletion strain was grown overnight in CYE medium supplemented with 0.2 µg ml⁻¹ aTC. Cells from 1 ml of the overnight culture were collected, washed once in 1 ml CYE, and resuspended in 1 ml of CYE medium. Cells from this sample were then used to inoculate 15 ml of CYE medium, either with or without 0.2 µg ml⁻¹ aTC, to OD₆₀₀ = 0.02. The cells were then cultured aerobically at 30 °C and cell samples collected into SDS sample buffer every 2 h for subsequent analysis by immunoblotting. Samples for imaging or membrane preparation were collected and analysed as detailed below.

To purify BamA complexes after depleting the essential BamG or BamH subunits, a 200 ml overnight culture of the appropriate strain grown in the presence of 0.2 µg ml⁻¹ aTC was collected and resuspended in the same volume of fresh CYE medium without aTC. This sample was used to inoculation 8 l of CYE without aTC to OD₆₀₀ = 0.1 which was then cultured aerobically at 30 °C for 6 h. Cells were collected and BamA complexes processed for purification as described above.

Microscopic analysis of cells during BAM subunit depletions

Live cells were imaged directly in growth medium by spotting samples taken from depletion cultures onto a 1% agarose pad prepared in PY2 medium. Phase contrast images were acquired on an inverted fluorescence microscope (Ti-E, Nikon) equipped with a perfect focus system, a 100× NA 1.4 oil immersion objective, a motorized stage, and a sCMOS camera (Orca Flash 4, Hamamatsu).

For transmission electron microscopy, cells were collected at the required time points during depletion by centrifugation at 8,000g for 5 min. After carefully removing the supernatant, cell pellets were gently resuspended in 1 ml of fixative solution (2.5% glutaraldehyde, 4% formaldehyde in 0.1 M PIPES buffer, pH 7.4) and incubated at room temperature for 1 h. Following fixation cells were washed with TEM buffer (100 mM PIPES NaOH pH 7.2), treated with TEM buffer containing 50 mM glycine, washed again in TEM buffer, and then subjected to secondary fixation with TEM buffer containing 1% (w/v) osmium tetroxide and 1.5% (w/v) potassium ferrocyanide. Samples were then washed extensively with Milli-Q water, stained with aqueous 0.5% (w/v) uranyl acetate overnight, then washed again with Milli-Q water. The samples were dehydrated through an ethanol series and infiltrated with and embedded in TAAB low viscosity epoxy resin ahead of polymerization at 60 °C for 24 h. Sections of 90 nm were cut from the resin blocks using a Leica UC7 Ultramicrotome and collected onto 3 mm copper grids. The sections were then post-stained with lead citrate and imaged using a JEOL Flash 120 kV TEM equipped with a Gatan Rio camera.

Whole-membrane proteomics

Fifteen millilitres of cells at the 6 h time point of the standard depletion experiment were collected by centrifugation at 8,000g for 5 min at 4 °C. The cells were resuspended in 1 ml of buffer W and lysed on ice using a probe sonicator (Sonics Vibra Cell, probe 630-0422) at 40% power by 12 repeats of a 10 s on/10 s off pulse cycle. After lysis, the samples were centrifuged at 20,000g for 20 min at 4 °C to remove cell debris. The supernatant was then centrifuged at 135,000g for 45 min at 4 °C to pellet the membranes. The membranes were resuspended in buffer W and the protein contents of the samples normalized by A_280 nm. The samples were run together on SDS–PAGE gels and stained with Coomassie Blue (Extended Data Fig. 7i) to confirm that normalization had been correctly implemented. Statistical methods were not used to determine sample size. Randomization and blinding were not used.

Membrane fractions were resuspended in lysis buffer containing 1% SDS, 0.1 M ammonium bicarbonate pH 8.0. Samples were sonicated for 5× 15 s in a water bath with 15 s incubations on ice between each pulse cycle. The samples were clarified by centrifugation at 17,500g for 30 min and 50 µg of total protein lysate was taken for analysis. Samples were reduced for 30 min using 10 mM tris(2-carboxyethyl)phosphine (TCEP) followed by alkylation for 30 min in the dark using 2-chloroacetamide. SpeedBeads Magnetic Carboxylate Modified Particles (GE Healthcare) were mixed with the sample in a 10 volumes beads: 1 volume sample ratio and the samples shaken for 10 min at 1,000 rpm. The beads were then washed twice with 70% ethanol followed by 100% acetonitrile. This procedure was repeated 8 times. 100 mM ammonium bicarbonate was added to the washed beads and pre-digestion with endoprotease LysC (Wako; 1:100) was carried out at 37 °C for 2 h. This was followed by 16 h digestion with trypsin (Promega, 1:40) at 37 °C. The supernatant was collected and any remaining bound peptides were eluted from the beads using 2% dimethyl sulfoxide (DMSO). Digested peptides were loaded onto C18 stage tips, pre-activated with 100% acetonitrile and 0.1% formic acid and centrifuged at 4000 rpm. The tips were then washed with 0.1% formic acid and eluted in 50% acetonitrile/0.1% formic acid. Eluted peptides were dried in a speed-vac.

Peptide analysis employed a Thermofisher Scientific Ultimate RSLC 3000 nano liquid chromatography system coupled in-line to a Q Exactive mass spectrometer equipped with an Easy-Spray source (Thermofisher Scientific). Peptides were separated using an Easy-Spray RSLC C18 column (75 µm internal diameter, 50 cm length, Thermofisher Scientific) using a 60 min linear 15% to 35% solvent B (0.1% formic acid in acetonitrile) gradient at a flow rate 200 nl min⁻¹. The raw data were acquired on the mass spectrometer in a data-dependent acquisition (DDA) mode. Full-scan mass spectra were acquired in the Orbitrap (Scan range 350–1,500 m/z, resolution 70,000, AGC target 3 × 10⁶, maximum injection time 50 ms). The 10 most intense peaks were selected for higher-energy collision dissociation (HCD) fragmentation at 30% of normalized collision energy. HCD spectra were acquired in the Orbitrap at resolution 17,500, AGC target 5 × 10⁴, maximum injection time 120 ms with fixed mass at 180 m/z.

Mass spectrometry data were analysed using MaxQuant 2.5.1.0 as previously described⁵¹ to obtain label-free quantification values that were then used for data processing in Perseus 2.1.3.0⁵². Label-free quantification values were log₂-transformed and categorically grouped by replicates. Rows were filtered based on two valid values in each group and then missing values were replaced using a normal distribution with a width of 0.3 and down shift of 1.8 (default values). Then, dataset was normalized by subtracting the medians of each sample. After visually verifying a normal distribution and a linear correlation, sample pairs were subjected to a two-tailed t-test using a false discovery rate (FDR) of 0.1 and a S₀ of 0.1 to define a threshold of statistical significance. Proteins were represented in a volcano plot, according to the log₂ of their enrichment and the −log₁₀ of the t-test P value.

An ANOVA test was carried out for indicated groups of proteins using the Benjamini–Hochberg method with a FDR of 0.05 for truncation. Then, a post hoc Tukey’s honest significant difference test for one-way ANOVA using a FDR of 0.05 was carried out. Proteins were then filtered by ANOVA significance and by category to represent in a heat map their honest significant difference scores, as indicated.

A batch normalization using empirical Bayes method was carried out with the ComBat script⁵³ for PerseusR package 0.3.4⁵⁴ to make the heat map for all depletions (Extended Data Fig. 8). Then, samples were subjected to the statistical test previously described.

The proteins obtained from the mass spectrometry experiments were categorized as follows. Proteins with signal peptides or lipoprotein signal peptides were first extracted using SignalP 6.0⁵⁵ to obtain datasets containing only OM plus periplasmic proteins, or lipoproteins, respectively. Proteins were then manually sorted to the categories OMP or SLP. This sorting was carried out using Uniprot entry data that included AlphaFold²³ models. Lipoproteins were only classified as SLPs if they were either SusD homologues or if they were found at a locus coding SusCD systems.

Determination of cell surface exposure of SusE

The strain for analysis was transformed with plasmid pXL184 which expresses His-tagged SusE. The cells were then grown overnight in CYE supplemented with erythromycin, and for BAM subunit depletion strains with 0.2 µg ml⁻¹ aTC. Cells were collected, resuspended in CYE medium, and then used to inoculate 10 ml of erythromycin-containing CYE medium to OD₆₀₀ = 0.02, supplementing with 0.2 µg ml⁻¹ aTC as required. The cells were cultured for 6 h before being collected by centrifugation and resuspended in phosphate buffered saline (PBS) containing 10 mM MgCl₂ to a total volume of 80 µl and OD₆₀₀ = 1. Samples were supplemented as appropriate with 200 μg ml⁻¹ proteinase K (Thermo Fisher) and 1% (v/v) Triton X-100 (Merck) and incubated for 20 min at room temperature. Reactions were stopped by the addition of 5 mM PMSF (ITW Reagents) followed by incubation at 100 °C for 5 min, addition of SDS–PAGE sample buffer, and further incubation at 100 °C for 5 min before analysis by immunoblotting.

Isolation of outer membrane vesicle fraction

The isolation of outer membrane vesicles (OMVs) was performed essentially as in ref. ³⁸. In brief, cells were separated from culture supernatant by centrifugation at 8,000g for 5 min and the pellets reserved as the whole-cell fraction. Culture supernatant from the equivalent of 2 ml of culture at OD₆₀₀ = 1 was filtered through a 0.2 µm filter (MilliporeSigma, SLGPR33RB) and concentrated using a 100 kDa molecular weight cut-off Amicon Ultra-4 centrifugal filter (MilliporeSigma, UFC810096) to produce the OMV fraction. Samples were adjusted to equal volume before analysis by immunoblotting.

Isolation of a spontaneous suppressor of BamH depletion

The BamH depletion strain XLFJ_1140 was grown overnight in CYE medium supplied with aTC. One millilitre of cells was collected by centrifugation at 8,000g for 3 min, washed once with CYE and then diluted to a starting OD₆₀₀ = 0.2 in 10 ml fresh CYE medium without aTC. After culturing for 6 h, cells were diluted 1:200 into fresh CYE medium without aTC and cultured for a further 2 days before plating on CYE agar to obtain single colonies. Individual clones were cultured in parallel with and without aTC in CYE and the expression of BamH analysed by whole-cell immunoblotting. Clones that grew without aTC but still expressed BamH only following aTC induction (showing that they were not constitutively de-repressed for BamH synthesis) were subjected to genome sequencing (Plasmidsaurus). This identified the potential suppressor mutation bamA^Q801K, which was introduced into a BAM wild-type background, followed by successive deletions of bamH and bamH2 to produce the bamH^sup strain XLFJ_1198.

Cryo-EM sample preparation and imaging

Four microlitres of either fraction A (for the BAM_Fj complex, 1.3 mg ml⁻¹) or fraction B (for the BamAP complex, 1.3 mg ml⁻¹) of the BAM_Fj preparation (Fig. 1a), or of the BamP-deleted BAM complex (ΔBamP complex, 1.2 mg ml⁻¹) was adsorbed onto glow-discharged holey carbon-coated grids (Quantifoil 300 mesh, Au R1.2/1.3) for 10 s. Grids were blotted for 2 s at 10 °C, 100% humidity and frozen in liquid ethane using a Vitrobot Mark IV (Thermo Fisher Scientific).

Movies were collected in counted mode, in Electron Event Representation (EER) format, on a CFEG-equipped Titan Krios G4 (Thermo Fisher Scientific) operating at 300 kV with a Selectris X imaging filter (Thermo Fisher Scientific) and slit width of 10 eV, at ×165,000 magnification on a Falcon 4i direct detection camera (Thermo Fisher Scientific), corresponding to a calibrated pixel size of 0.732 Å. Movies were collected at a total dose ranging between 52.0–60.3 e⁻ Å⁻² (Extended Data Table 1), fractionated to ~1.0 e⁻ Å⁻² per fraction for motion correction.

Cryo-EM data processing

Patched motion correction, contrast transfer function (CTF) parameter estimation, particle picking, extraction and initial 2D classification were performed in SIMPLE 3.01⁵⁶. All downstream processing was carried out in cryoSPARC 4.5.3⁵⁷ or RELION 4.03⁵⁸, using the csparc2star.py script within UCSF pyem 0.5⁵⁹ to convert between formats. Global resolution was estimated from gold-standard Fourier shell correlations (FSCs) using the 0.143 criterion and local resolution estimation was calculated within cryoSPARC.

The cryo-EM processing workflow for the BAM_Fj complex is outlined in Extended Data Fig. 1. In brief, particles were subjected to one round of reference-free 2D classification (k = 200) using a 240 Å soft circular mask within cryoSPARC resulting in the selection of 2,153,927 clean particles. A subset of these particles (180,179) was subjected to multi-class ab initio reconstruction using a maximum resolution cut-off of 7 Å, generating 4 volumes. These volumes were lowpass-filtered to 20 Å and used as references in a heterogeneous refinement against the full 2D-cleaned particle set. Particles (903,299) from the most populated and structured class were selected and non-uniform refined against their corresponding volume lowpass-filtered to 15 Å, generating a 3.0 Å map. Bayesian polishing in RELION followed by duplicate particle removal generated a 2.5 Å map after non-uniform refinement, which could be further improved to 2.3 Å after local and global CTF refinement (fitting beam tilt and trefoil only). These particles were then subjected to heterogeneous refinement against four compositionally distinct volumes previously generated by RELION 3D classification (k = 8, 3.75° sampling) of a particle subset of pre-polished particles. Particles (274,708) belonging to the class with strong BamD and POTRA densities were selected and non-uniform refined against their corresponding volume, generating a 2.4 Å map. Additional alignment-free 3D classification in RELION was performed (k = 6) using a soft mask covering BamD and the BamA POTRA domains yielding a class with stronger density. Particles (55,795) from this class were selected and non-uniform refined against a previous volume lowpass-filtered to 15 Å, generating a consensus 2.7 Å volume. Local refinements were performed against the consensus volume (lowpass-filtered to 7 Å) using soft masks covering the BamD/POTRA domains or extracellular density, yielding 3.2 Å and 2.7 Å volumes, respectively. ChimeraX⁶⁰ was used to generate a composite map from the consensus and individual focused maps.

The cryo-EM processing workflow for the BamAP complex is outlined in Extended Data Fig. 6. Two datasets were collected for this sample. In the first dataset particles were subjected to two rounds of reference-free 2D classification (k = 200) using a 200 Å soft circular mask resulting in the selection of 979,474 clean particles. These particles were then subjected to multi-class ab initio reconstruction (k = 4) using a maximum resolution cut-off of 8 Å, generating 4 volumes. Particles (514,326) belonging to the 2 most prominent volumes were combined and non-uniform refined against one of their corresponding volumes, lowpass-filtered to 15 Å, generating a 3.7 Å volume. The second particle dataset underwent four rounds of 2D classification (k = 200, 200 Å soft circular mask) followed by multi-class ab initio reconstruction using a maximum resolution cut-off of 7 Å, generating 6 volumes. Particles (438,412) from the most populated class were selected and refined against their corresponding volume lowpass-filtered to 15 Å, generating a 3.7 Å volume. Particles from both datasets were independently polished within RELION, combined, and non-uniform refined, fitting per-particle CTF parameters, yielding a 3.5 Å map. Alignment-free 3D classification was subsequently performed within cryoSPARC (k = 6), using a soft mask covering the full protein density of the complex. Particles (96,076) from the class demonstrating strong density for the N-terminal domain of BamP were selected and non-uniform refined against their corresponding volume, lowpass-filtered to 15 Å, generating a 3.7 Å map.

The cryo-EM processing workflow for the ΔBamP complex is outlined in Extended Data Fig. 5. In brief, particles were subjected to two rounds of reference-free 2D classification (k = 200) using a 180 Å soft circular mask within cryoSPARC resulting in the selection of 1,177,554 clean particles. These particles were then subjected to multi-class ab initio reconstruction using a maximum resolution cut-off of 6 Å, generating 6 volumes. Particles from volume classes containing BamA barrels were independently non-uniform refined against their corresponding volume, lowpass-filtered to 15 Å. These particles were subsequently combined and refined against a volume (lowpass-filtered to 15 Å) from the most populated class, generating a 3.6 Å consensus volume. Bayesian polishing in RELION followed by non-uniform refinement and fitting of per-particle CTF parameters plus beam tilt and trefoil generated a 3.5 Å map. Map quality was further improved by non-uniform refinement of a cleaner particle set (534,368 particles) generated by an additional round of 2D classification (k = 100, 180 Å soft circular mask), despite no increase in nominal resolution. A second β-barrel could be resolved in map density at low contour level (0.08). Attempts to improve map quality for this partner β-barrel, through extensive 3D classification and local refinement schemes, did not improve map quality for this region.

Model building, structure refinement and figure preparation

Iterative model building and real-space refinement using secondary structure, rotamer, and Ramachandran restraints was performed in Coot v0.9⁶¹ and Phenix 1.21⁶², respectively. Validation was performed in Molprobity 4.5.2⁶³ within Phenix. Cryo-EM data collection, image processing and structure refinement statistics are listed in Extended Data Table 1. Figures were prepared using UCSF ChimeraX v.1.9⁶⁰.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.