Structural characterization of the POTRA domains from A. baumannii reveals new conformations in BamA

Summary

Recent studies have demonstrated BamA, the central component of the β-barrel assembly machinery (BAM), as an important therapeutic target to combat infections caused by A. baumannii and other Gram-negative pathogens. Homology modeling indicates BamA in A. baumannii consists of five polypeptide transport associated (POTRA) domains and a β-barrel membrane domain. We characterized the POTRA domains of BamA from A. baumannii in solution using SEC-SAXS analysis and determined crystal structures in two conformational states that are drastically different than those previously observed in BamA from other bacteria, indicating that the POTRA domains are even more conformationally dynamic than has been observed previously. Molecular dynamics simulations of the POTRA domains from A. baumannii and E. coli allowed us to identify key structural features that contribute to the observed novel states. Together, these studies expands on our current understanding of the conformational plasticity within BamA across differing bacterial species.

eTOC Blurb

In A. baumannii, BamA has an N-terminal periplasmic domain consisting of five tandem POTRA domains. Overly Cottom et al. determined crystal structures of AbBamA POTRA 1–4, revealing new conformations that are rotated by ~180°, indicating even more conformational plasticity than has been observed previously and offering a possible explanation for surface exposure of accessory proteins.

Graphical Abstract

Introduction

Acinetobacter baumannii is a Gram-negative pathogen which causes a variety of infections in humans including pneumonia, bloodstream infections, wound infections, urinary tract infections, and meningitis¹. A. baumannii is a significant cause of hospital-derived infections and it has a unique ability to persist on surfaces, which makes the emergence of drug-resistant strains particularly problematic^2–7. In 2019, the World Health Organization designated carbapenem-resistant A. baumannii as an urgent threat⁸, necessitating the rapid development of new treatments to combat A. baumannii infections.

A. baumannii, like all other Gram-negative bacteria, is characterized by the presence of an inner membrane (IM) and an outer membrane (OM) separated by the periplasm and sandwiching the peptidoglycan layer. The OM uniquely contains a host of β-barrel outer membrane proteins (OMPs) which perform many cellular functions including nutrient import, cell signaling, and adhesion^9–14. Nascent OMPs are trafficked across the IM by the Sec translocon, shuttled across the periplasm with the help of periplasmic chaperones, and are folded and inserted into the OM by the β-barrel assembly machinery (BAM) complex^11,15–17. In E. coli, the BAM complex consists of BamA, which is itself an OMP, and BamB, C, D, and E, which are soluble periplasmic domains that are attached to the inner leaflet of the OM by lipid anchors^18–22. Both BamA and BamD are required for viability in E. coli²³, and are highly conserved among other Gram-negative bacteria^18,24–26. There is, however, variability in the number and type of BAM complex lipoprotein subunits present across different bacterial species^18,24–28. Due to its essential role in the cell, the BAM complex has recently become an emerging target for the development of novel antimicrobial therapeutics targeting Gram-negative bacteria^13,29–33.

The A. baumannii genome codes for homologs of all five canonical BAM complex subunits (BamA-E) present in E. coli, so the BAM complex might represent one possible target for novel antimicrobial therapeutics capable of treating Acinetobacter infections. Recent studies have shown that recombinant A. baumannii BamA can effectively be used to immunize mice against infection with A. baumannii^13,34. However, no structural characterization of A. baumannii BamA, or any other BAM complex component, has been performed to date. Elucidation of the structure of the BAM complex and its individual components from Acinetobacter would provide the structural basis for targeting this complex for new therapeutics against pathogenic Acinetobacter.

BamA is the central component of the BAM complex and performs the insertase activity mediating OMP biogenesis^35–41. BamA is composed of an N-terminal periplasmic domain consisting of five polypeptide transport associated (POTRA) domains with a C-terminal β-barrel domain that is inserted in the outer membrane. The barrel domain consists of 16 β-strands and has a lateral seam joining strands 1 and 16 forming the final barrel shape of the protein³⁹. BamA functions by inducing local disruptions with the OM to allow OMPs to insert into the membrane. Additionally, the barrel of BamA has been shown to serve as a ‘template’ for which the new OMP nucleates and forms^42–44.The periplasmic POTRA domains of BamA interact directly with the accessory lipoproteins to form an extensive assembly beneath the BamA barrel^45–49.

The number of POTRA domains present across bacterial species is variable, as Fusobacterium nucleatum BamA is predicted to contain four, N. gonorrhoeae, E. coli and A. baumannii contain five, and Myxococcus xanthus contains seven^24,27,28. Additionally, the minimal number of POTRA domains that are essential for viability also varies among Gram-negative species⁵⁰. For example, POTRA 3–5 are essential in E. coli^23,41, while only POTRA 5 is essential in N. meningitidis⁵¹. In E. coli BAM, BamB interacts with POTRA 2 and POTRA along the hinge between them^45–47,52 and BamD interacts primarily with POTRA 5 and minimally with POTRA 2^45–47,53.

The POTRA domains provide the scaffold for the association of the Bam lipoproteins to assemble into the higher order complex^45–47. POTRA domains typically have a conserved secondary structure, with a β₁α₁α₂β₂β₃ topology, despite sequence divergence^41,54,55. Each POTRA repeat within E. coli BamA was found to be structurally conserved, except for a β-bulge present in POTRA 3^41,55. Conformational flexibility in the POTRA domains is important for mediating changes in the full complex that assist in initiation and folding of the new OMPs. It has been postulated that the POTRA domains may serve as a chaperone to assist the unfolded OMPs with pre-forming β-strands through β-templating using the exposed edges of strands present in the POTRA domains^54,56.

Here we report the expression, purification, and structural characterization of the POTRA domains of BamA from A. baumannii (AbBamA) and do a structural comparison to POTRA domains from other bacterial species. We characterized the POTRA domains 1–4 of AbBamA in solution using SEC-SAXS analysis, molecular dynamics (MD) simulations, and determined crystal structures in two conformational states. These conformational states are drastically different from those previously observed in BamA from other bacteria, indicating that the POTRA domains are even more conformationally dynamic than has been observed previously. Our study further expands on our current understanding of the plasticity within BamA and BAM during OMP biogenesis.

Results

Structure determination of the POTRA 1–4 domains from A. baumannii

The N-terminal POTRA 1–4 domains of BamA (residues 25–355) from A. baumannii (strain 19606) (AbBamA) were cloned into the pHisParallel2 vector and expressed in BL21(DE3) cells. (Fig. 1A). AbBamA POTRA 1–4 was purified by nickel affinity chromatography, followed by removal of the His-tag by cleavage with TEV protease, and then size-exclusion chromatography (SEC) as a final purification step. Purified AbBamA POTRA 1–4 migrated at an expected size of ~41 kDa on an SDS-PAGE gel (Fig. 1B, C). For SAD phasing experiments, expression was performed in the methionine-auxotroph B834(DE3) cells which were grown in SelenoMethionine Medium Complete (Molecular Dimensions). Selenomethionine-substituted BamA POTRA 1–4 was purified as described above. Both native and selenomethionine-substituted AbBamA POTRA 1–4 were concentrated to 10 mg/mL for crystallization.

Figure 1. Sequence comparison of POTRAs 1–4 and purification of this domain from A. baumannii.

A. Alignment of the AbBamA POTRA 1–4 amino acid sequence with other structurally characterized orthologs of BamA. The secondary structure of BamA POTRA 1–4 from A. baumannii is shown based on our studies here. The boundaries for each POTRA domain are indicated in blue, purple, green, and brown. B. The SEC profile from native and selenomethionine substituted (panel C) AbBamA POTRA 1–4 showing a pure sample running at ~40 kDa (expected mass is 41 kDa including the His tag).

High-throughput broad matrix crystallization screening was performed using hanging drop method (1:1 protein:well solution ratio) and all lead conditions looped directly from the trays and screened for diffraction quality. Initial crystallization conditions consisted of 100 mM citric acid, pH 4.0, and 800 mM ammonium sulfate for both native and selenomethionine-substituted (Se-met) AbBamA POTRA 1–4 (Fig. 2A). AbBamA POTRA 1–4 crystals grew as thin fragile plates. Despite extensive optimization methods including seeding, all AbBamA POTRA 1–4 crystals were clusters of thin plates which were difficult to separate and loop individually. We identified ethylene glycol in our additive screen as an additive which allowed us to obtain some larger, single thin crystal plates (Figure S1A). The best native crystals were grown in 100 mM citric acid, pH 3.8, 630 mM ammonium sulfate, and 13% ethylene glycol. Additionally, lowering the protein concentration of AbBamA POTRA 1–4 in the drop from 10 mg/ml to 5 mg/ml for native sample or 2.5 mg/ml for the Se-met sample also helped produce crystals more suitable for looping. To confirm that the crystals we obtained were indeed BamA POTRA 1–4, we washed them briefly in a mix of well solution and sample buffer, redissolved in LDS loading buffer by gentle pipetting, and visualized by SDS-PAGE (not shown). The molecular weight observed for the crystallized protein was ~41 kDa, which matches the molecular weight we observed for purified BamA POTRA 1–4 after SEC, indicating that the protein crystals obtained were AbBamA POTRA 1–4.

Figure 2. Structure determination of AbBamA POTRA 1–4.

A. AbBamA POTRA 1–4 crystals showing small thin finely stacked plates. B. The structure of AbBamA POTRA 1–4 highlighting the selenomethionine sites used for phasing. C. Representative electron density (blue isosurface; 1 σ) for the AbBamA POTRA 1–4 structure. D. The asymmetric unit of the structure in space group P3₂21 showing two molecules with pseudo two-fold non-crystallographic symmetry. Orthogonal views are shown.

Data collection was performed at the GM/CA beamline at the Advanced Photon Source and the data processed using HKL2000⁵⁷ to space group P1 with unit cell parameters a = 59.93 Å, b = 65.92 Å, c = 87.02 Å, with α = 67.76°, β = 88.28° and γ = 65.82°. Analysis of the processed dataset using Xtriage⁵⁸ predicted 3 copies of BamA POTRA 1–4 per ASU with a Matthews coefficient of 2.6 and a solvent content of 52%. Attempts to use existing structures of POTRA domains as search models, as well as computational models, for molecular replacement for initial phasing were unsuccessful. This was likely due to the elongated and flexible nature of the periplasmic domain itself. We then collected multiple selenium-SAD datasets from a single crystal. The data were again processed in space group P1, but with cell parameters a = 56.52 Å, b = 61.51 Å, c = 85.90 Å, with α = 97.59°, β = 107.35°, and γ = 112.45°; again with 3 copies per ASU predicted with a Matthews coefficient of 2.3 and a solvent content of 46% (Table 1). Attempts at experimental phasing with both Se-SAD and Se-MAD phasing were also unsuccessful. We attributed this again to partially being in space group P1, but mostly due to the observation that the crystal did not appear to be uniformly isotropic, as evidenced by the varying cell parameters from one end of the crystal to the other. Therefore merging multiple datasets was not assisting to amplify the anomalous signal as would typically be the case.

Table 1.

Data collection and refinement statistics.

Data Collection	Data Set 1	Data Set 2
λ (Å)	0.9793	0.9793
Space group	P3221	P1
a, b, c (Å)	94.9 94.9 220.3	56.52 61.51 85.90
α, β, γ (º)	90 90 120	97.6 107.3 112.4
Resolution (Å)*	47.46 – 2.60 (2.69–2.60)	39.37 – 2.80 (2.90 – 2.80)
Completeness (%)*	100 (100)	87.0 (88.0)
Redundancy*	19.6 (19.8)	2.5 (2.4)
Rsym †*	0.114 (1.23)	0.146 (1.32)
I / σ (I)*	27.7 (1.7)	10.5 (1.0)
CC1/2	0.998 (0.79)	0.992 (0.284)
#Se found	10 sites/ASU
FOM	0.307
Bayes-CC	52.46
Skew	0.20
Refinement
Resolution (Å)	47.46 – 2.60 (2.69–2.60)	39.37 – 2.78 (2.85 – 2.78)
No. reflections	35362 (2,278)	23,193 (1,445)
R§/Rfree¶	0.22/0.24	0.24/0.29
r.m.s. deviations
Bonds (Å)	0.004	0.003
Angles (°)	0.78	0.57
No. Protein atoms	5167	4897
No. Waters	41	39
B-factors (Å2)
Wilson B	78.94	85.99
Protein	79.15	86.20
Waters	56.52	56.19
Ramachandran Analysis ¥
Favored (%)	97.24	95.25
Allowed (%)	2.45	4.45
Outliers (%)	0.31	0.31
PDB ID	9CX5	9CX4

^†$R_{\text{sym}}={\sum }_{hkl,j} \left(|{\mathrm{I}}_{\mathrm{h}\mathrm{k}\mathrm{l}}<{\mathrm{I}}_{\mathrm{h}\mathrm{k}\mathrm{l}}>\mid \right)/{\mathrm{\Sigma }}_{hkl,j}{\mathrm{I}}_{\text{hkl}}$, where $<{\mathrm{I}}_{\mathrm{h}\mathrm{k}\mathrm{l}}>$ is the average intensity for a set of j symmetry related reflections and ${\mathrm{I}}_{\text{hkl}}$ is the value of the intensity for a single reflection within a set of symmetry-related reflections.

^§$R\mathrm{ }\mathrm{f}\mathrm{a}\mathrm{c}\mathrm{t}\mathrm{o}\mathrm{r}={\mathrm{\Sigma }}_{hkl}\left(\left|\right|{\mathrm{F}}_{\mathrm{o}}\left|-\left|{\mathrm{F}}_{\mathrm{c}}\right|\right|\right)/{\mathrm{\Sigma }}_{hkl|}{\mathrm{F}}_{\mathrm{o}}\mid$ where ${\mathrm{F}}_{\mathrm{o}}$ is the observed structure factor amplitude and ${\mathrm{F}}_{\mathrm{c}}$ is the calculated structure factor amplitude.

^¶$R_{\text{free}}={\mathrm{\Sigma }}_{hkl,T}\left(\left|\right|F_{\mathrm{o}}\left|-\left|F_{\mathrm{c}}\right|\right|\right)/{\mathrm{\Sigma }}_{hkl,T}\left|F_{\mathrm{o}}\right|$, where a test set, T (5% of the data), is omitted from the refinement.

^¥Performed using Molprobity within PHENIX.

^*Indicates statistics for last resolution shell shown in parenthesis.

We revisited our initial crystallization screening in hopes that we could find another condition with more favorable packing, but had limited options given that only the current condition produced high-resolution diffraction. Since we could produce the sample at relatively high yields, to overcome this limitation, we performed reductive alkylation of the selenium-substituted sample and rescreened it using broad matrix high-throughput screening. This led to large single crystals in multiple conditions. These crystals were then screened for diffraction quality with one condition, 0.1 M BIS-TRIS propane pH 7.0, 1.2 M DL-Malic acid, pH 7.0, producing the best diffraction (Figure S1B, C). The data were processed in space group P3₂21 with cell parameters a = 94.92 Å, b = 94.92 Å, c = 220.30 Å, with α = 90°, β = 90° and γ = 120°; again with 3 copies per ASU predicted with a Matthews coefficient of 2.6 and a solvent content of 52%. Se-SAD phasing was performed using AutoSol in PHENIX⁵⁸ which produced an interpretable map (10 Se-sites per ASU with FOM of 0.307, Bayes-CC of 52.5, and Map Skew of 0.2) (Fig. 2B, C). Using COOT, models of the individual POTRA domains 1–4 could be placed into the density, with a final solution of two molecules being built into the ASU with R_work and R_free values of 0.22 and 0.25 at a resolution of 2.6 Å (Fig. 2D).

Once the Se-SAD structure in space group P3₂21 was solved, it was used as a starting search model for solving the structure of the selenium-substituted crystals in space group P1 using molecular replacement, which also contained only two molecules in the ASU with R_work and R_free values of 0.24 and 0.29 at a resolution of 2.78 Å. We then solved the native structure in space group P1 and found the same packing, however, at lower resolution (2.93 Å). Therefore, we only report the highest resolution structure in P1 in our study here.

Structural alignment reveals conformational plasticity in AbBamA POTRAs 1–4

While cell content analysis for both structures predicted three molecules per ASU, only two were observed for each space group (Fig. 3A, B). An RMSD of 3.6 Å was calculated when comparing chain A to chain B in the P3₂21 structure, where chains A and B refer each to a molecule within the ASU, indicating significant conformational flexibility (Fig. 3C). However, an RMSD of 0.7 Å was calculated when comparing chain A to chain B in the P1 structure, where minimal differences were observed between the monomers within the ASU (Fig. 3D). When comparing chain A from the P1 space group to chain A and chain B from the P3₂21 structure, RMSDs of 6.3 Å and 8.5 Å were calculated, respectively (Fig. 3E, F). This indicated an even larger conformational deviation among the structures in the two space groups. In total, we observed three conformational states: 1) space group P1, chains A and B, 2) space group P3₂21, chain A, and 3) space group P3₂21, chain B.

Figure 3. The structures of AbBamA POTRA 1–4 in condensed and extended conformational states.

A. The assembly of AbBamA POTRA 1–4 in the asymmetric unit in space group P3₂21 and P1 (panel B), both with pseudo two-fold non-crystallographic symmetry. C. A superposition of the two chains in space group P3₂21 (RMSD value of 3.6 Å) and P1 (panel D; RMSD value of.7 Å). The conformation of the structure in P1 is more extended. E. A superposition of chains A (RMSD of 6.3 Å) and B (panel F; RMSD of 8.5 Å) from each respective space group.

An alignment of both chains of POTRA 1–4 from space group P3₂21 with chain A from space group P1 shows significant conformational flexibility, with the structures in P1 being more extended (Fig. 4A, B). An alignment along POTRAs 3–4 of the three conformational states observed in our AbBamA POTRA 1–4 structures shows ~45° sweep spanning ~54 Å of POTRAs 1 and 2 along the hinge between POTRA 2 and 3 (Fig. 4B). Comparing these conformations to previously reported structures from E. coli revealed conformations not previously observed in the POTRA domains of BamA. Comparing the POTRA domains from AbBamA in space group P1, chain A, to E. coli ( PDB ID 3EFC) shows ~180° shift of POTRA 1 and 2, with others ranging from ~130–150° ( PDB ID 5EKQ, 2QDF, and 2QCZ) (Fig. 5A). A similar observation was observed in comparison to the POTRA domains of BamA from N. gonorrhoeae (PDB ID 4K3B) and H. Influenzae (PDB ID 6J09) (not shown).

Figure 4. SAXS analysis of AbBamA POTRA 1–4.

A. An overall superposition of chain A from space group P1 to both chains in space group P3₂21. Orthogonal views are shown. B. The same structural alignment as in panel A, yet superimposed along POTRAs 3 and 4, demonstrating a large ~45° swing of POTRA 1 and 2 moving ~54 Å. C. SEC-SAXS analysis of AbBamA POTRA 1–4 gave an R_g of 35 Å, an average mass of 41.5 kDa, and a D_max value of 126 Å (panel D). E. A Kratky plot is consistent with a globular fold containing flexible linkers on its termini.

Figure 5. AbBamA POTRA 1–4 reveals new conformations in BamA.

A. A structural alignment of POTRA 1–4 from AbBamA with EcBamA alone (PDB ID 2QCZ, 2QDF, and 3EFC) and EcBamA within the BAM complex (PDB ID 5EKQ). The left panel is the same view as in Fig 4. panels A and B, while the top-right view is rotated ~45° from the left view, with the right -bottom being orthogonal to the top-right. B. CRYSOL was used to compare calculated scattering curves of the structures from panel A to the experimental scattering curve in panel Fig. 4C. χ² values indicate the closest match is chain A from space group P1 of AbBamA POTRA 1–4 (pink; χ² value of 1.3), while the structures for EcBamA have χ² values ranging from 1.6 – 7.2; NgBamA has a χ² value of 1.6. SREFLEX was used to improve the fit of chain A from space group P1 of AbBamA POTRA 1–4, producing UC01 (red), which gave a χ² value of 1.1. C. A superposition of chain A from space group P1 of AbBamA POTRA 1–4 (pink) and UC01 (red) with the DAMMIN/IF SAXS envelope. D. Modeling the new conformations with the full length E. coli BamA structure (PDB ID 5D0Q). E. A surface depiction of the two conformations from panel D, where the POTRA domains are in the periplasm and at the membrane surface.

A structural comparison of the individual POTRA domains of BamA from A. baumannii, E. coli, and N. gonorrhoeae shows good pairwise structural agreement for all four POTRA domains with RMSDs ranging from 0.9 – 3.3 Å (Figure S2A and Table S1). The most notable structural difference was found in POTRA3, where the loop between α1 and α2 has an additional small helix in A. baumannii that is absent in both E. coli and N. gonorrhoeae (Figure S2B, C). Importantly, this loop sits at the junction where POTRA12 are shifted in the opposite direction in E. coli and N. gonorrhoeae, suggesting this loop may be contributing factor leading to the unique conformation observed in A. baumannii.

SEC-SAXS authenticates conformational plasticity in AbBamA POTRAs 1–4

To investigate the structure of POTRA 1–4 in solution, we performed SEC-SAXS at the BioCAT beamline at the Advanced Photon Source (Fig. 4C). The scattering curve and Guinier plot indicated an aggregation-free sample with an R_g of 35.0 Å and a D_max of 126 Å (Fig. 4C, D), with an average mass calculation of ~41 kDa (Table 2). These results are consistent with a monomer in solution and with our reported crystal structures which have a max dimension of 110 Å (range from 80–110 Å). A Kratky plot is also consistent with a globular folded protein containing an elongated disordered N-terminal linker region (Fig. 4E).

Table 2. Summary of SEC-SAXS data collection parameters and results.

The POTRA domains 1–4 from AbBamA (without His tag) has a calculated molecular weight of 38 kDa.

Data Collection
Instrument/data processing		Advanced Photon Source, BioCAT beamline with Pilatus3 1M detector
Wavelength (Å)		1.0332
Beam size (μm)		~30 x 150
Camera length (m)		3.631
Q measurement range (Å−1)		0.0047 – 0.3525
Exposure time (s)		0.5
Sample configuration		SEC-MALS-SAXS
Sample temperature (°C)		22
Data processing
Protein Complex	Rg (Å)	Molecular Weight (kDa)				Dmax (Å)
		Vc	Vp	Bayes	MALS
AbBamA (POTRA 1–4)	35.0	38.1	46.1	40.2	39.0	126

To determine what conformation was observed in solution, CRYSOL was used to compare calculated scattering curves from the POTRA 1–4 structures from AbBamA and EcBamA to the experimental SEC-SAXS scattering curve (Fig. 5A, B). Values for χ² ranged from 1.3 – 1.7 for the AbBamA structures with chain A from space group P1 having the lowest value and therefore most closely matching the observed conformation. Other structures had elevated χ² values of 1.6 (PDB ID 3EFC, 4K3B), 2.5 (PDB ID 5EKQ), 6.1 (PDB ID 2QDF), and 7.2 (PDB ID 2QCZ), demonstrating the deviation from the conformational states observed in AbBamA. SRE-FLEX was used to optimize the fit of the P1A structure to the experimental SEC-SAXS scattering curve, producing a slightly modified structure (UC01) with a χ² value of 1.1. An alignment of the P1A structure, UC01 structure, and the DAMMIF/IN envelope shows good agreement along POTRAs 1–3, with the largest conformational deviation along POTRA 4, suggesting we likely have a mix of conformational states in solution as well (Fig. 5C).

MD simulations reveal differences in conformational plasticity between AbBamA and EcBamA POTRAs 1–4

We performed molecular dynamics (MD) simulations for two AbBamA models (chains A and B of the structure solved here) and two EcBamA models (chains A and B of PDB ID 2QCZ) to investigate the dynamic behavior of both AbBamA and EcBamA POTRA domains (Fig. 6A). During the simulations, the POTRAs exhibited significant flexibility. Both AbBamA and EcBamA displayed two types of states: an “open” state, when no hydrogen bonds were formed between POTRA1 and POTRA4 (Fig. 6B, C), and a “closed” state, when stable interactions formed between POTRA1 and POTRA4 (Fig. 6E, F). The distribution of angles formed by POTRA1–2, POTRA2–3, and POTRA3–4 for both AbBamA and EcBamA were calculated to compare their respective conformations (Fig. 6H, I). For both AbBamA and EcBamA, the POTRA1–2 angle distributions are located around 30–80°, although EcBamA has a narrower distribution. Significant distribution differences were observed for POTRA2–3 and POTRA3–4 angles. AbBamA shows two peaks around 50° and 100° for the POTRA2–3 angle distribution, while EcBamA had a single peak around 115°. Specifically, for AbBamA, the closed state mostly corresponded to the 100° peak, and the open state to the 50° peak. For POTRA3–4, AbBamA had one peak near 80°, while EcBamA had two peaks at 60° and 120°, corresponding to the closed and open states, respectively.

Figure 6. The conformation of AbBamA and EcBamA POTRA 1–4 in MD simulations.

AbBamA POTRAs 1,2 and 4 are shown in dark grey, and POTRA3 is in yellow. EcBamA POTRAs 1,2 and 4 are shown in blue and POTRA3 is in orange. A. The initial conformation for AbBamA and EcBamA POTRAs. B,C. Snapshots of “open” (B) and “closed” states (C) of AbBamA. E,F. Snapshots of “open” (B) and “closed” states (C) of EcBamA. D,G. Overlapping conformations of AbBamA (D) and EcBamA (G) chain A during the trajectories, with one snapshot every 20 ns. Structures are aligned based on POTRA3 β-sheet structures in panels A, D, and G. H,I. Distribution of angles formed by different POTRA domains of AbBamA (H) and EcBamA (I).

The structural differences between AbBamA and EcBamA POTRA domains include a longer α-helix in AbBamA at POTRA3 (Fig. 6A). This steric hindrance directly influenced the dynamics. During simulations, the POTRA2 domain tended to fold in different directions for AbBamA and EcBamA. For EcBamA, POTRA2 folded towards the longer α-helix side (Fig. 6G), while for AbBamA, it folded towards the shorter ɑ-helix side (Fig. 6D). In the AbBamA simulations, stable salt bridges formed by Lys204 on POTRA3 with Glu142 and Asp154 on POTRA2 were observed (with occupancies of 47.8% and 34.7%, respectively; Figure S3A), but similar stable H-bonding was not observed in EcBamA simulations.

The shorter helical structure of POTRA3 in EcBamA provides more flexibility, leading to a larger and more unstable RMSD value compared to AbBamA’s POTRA3 (Figure S3C, D). When POTRA2 folded towards the longer α-helix side, it pushed the α-helix upward to POTRA4. This movement causes POTRA4 to orient itself between the two helices of POTRA3 due to steric hindrance, which prevents it from reaching the closed state (Figure S3B).

In summary, the structural differences between the POTRA3 domains of EcBamA and AbBamA play a critical role in their dynamic behavior. The binding interactions and steric hindrances contribute to the preferential states observed in AbBamA and EcBamA. EcBamA, characterized by its shorter α-helix, demonstrates greater flexibility and spends a larger proportion of time in the closed state (30.2% for EcBamA is in the close state, compared to 17.5% for AbBamA). In contrast, AbBamA, with its more stable and longer POTRA3 α-helix, tends to maintain an extended, open conformation.

Discussion

Recent studies reporting the discovery of antimicrobial compounds targeting BamA have established that BAM is an exciting new target for the development of new antibiotics and vaccines against Gram-negative bacterial pathogens. Most functional and structural studies on BAM have been done in E. coli, with no structures reported for A. baumannii, which is at the top of the CDC’s list of urgent threats to public health due to multidrug resistance. To better understand how BAM functions in A. baumannii, here we report the structural characterization of the POTRA domains of BamA, the central and essential component of BAM (AbBamA). Our findings reveal new conformations of the POTRA domains that are significantly different than has been previously observed, providing new insights into the flexibility and dynamics of BamA. These new insights may be important for the general function of BamA in all Gram-negative bacteria or may represent a unique feature of BamA in A. baumannii.

Previous structural studies of BamA, either alone or within fully assembled BAM, have demonstrated flexibility within the POTRA domains of BamA. The first structures from E. coli of the POTRA domains only and the structures of fully assembled BAM highlight this conformational heterogeneity, something that was also observed in N. gonorrhoeae, H. ducreyi, and H. influenzae. Given that the accessory proteins BamB, C, D, and E assemble along the POTRA domains, the conformations of the POTRA domains are thought to be important in (i) mediating substrate interaction with BAM during biogenesis, and (ii) transducing conformational changes in the periplasmic region to the β-barrel of BamA that sits in the membrane. These conformational changes in the β-barrel domain of BamA are essential for the function of BAM and for turning on and off OMP biogenesis.

X-ray crystal and cryo-EM structures of BAM and BamA have demonstrated a range of conformations in the POTRA domains of BamA. These conformations are more restricted when bound with the accessory proteins; however, significant changes are present even then as demonstrated by the inward-open and outward-open states that BAM has been observed to adopt. These findings have been fully supported by computational studies which have been performed both in detergent and in membranes. The structures of POTRAs 1–4 of AbBamA reported here reveal that the POTRA domains of BamA have even more conformational freedom than has been previously observed experimentally or computationally. A comparison of POTRAs 1–2 in relation to POTRAs 3–4 shows a rotational movement ranging from ~130–180° compared to EcBamA, sweeping a distance of ~117–136 Å (Figure 5D). The exact role of these new conformations of BamA remains to be determined, but could explain multiple reports of Bam accessory proteins (BamC, BamD, and BamE) being surface exposed^59,60. In previously reported conformations, the POTRAs are fully within the periplasm. In our studies here, the POTRAs are projected up towards the membrane where they could assist in presenting the bound accessory proteins to the surface, although exactly how they would traverse the membrane is unknown (Fig. 5E). MD simulations revealed that conformational differences arise in part due to a longer α-helix in POTRA3 and salt bridges formed between POTRA2 and 3, which stabilize the conformations observed in the AbBamA POTRA1–4 crystal structures (Figure S3A, B). Another contributing factor for these conformations may be mechanistic differences between E. coli and A. baumannii. Given the varied conformations identified here (Fig. 6H,I), it is likely that BamA POTRA domains from other species will also display distinct conformational distributions.

STAR★Methods

RESOURCE AVAILABILITY

Lead Contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, Nicholas Noinaj (nnoinaj@purdue.edu).

Materials Availability

Primary data and other materials are available upon reasonable request to the lead contact.

Data and Code Availability

• Data generated in this study is available upon reasonable request from the lead contact. Crystal structures have been deposited to the PDB. These are publicly available as of the date of publication. Accession numbers are listed in the Key Resources table.

• This paper does not report any original code.

• Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Key Resources Table.

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Bacterial and virus strains
E. coli strain BL21(DE3)	New England Biolabs	C2527H
E. coli strain B834(DE3)	EMD Millipore	69041
Chemicals, peptides, and recombinant proteins
2xYT Media	Research Products International	X15600–5000.0
Selenomethionine Medium Complete	Molecular Dimensions	MD12–500
Ampicillin sodium salt	Fisher Bioreagents	BP1760–25
Isopropyl β-D-1-thiogalactopyranoside (IPTG)	Gold Biotechnology	P-470–25
β-mercaptoethanol (BME)	Fisher Chemical	O3446I-100
Phenylmethylsulfonyl fluoride (PMSF)	Gold Biotechnology	P-470–25
DNAseI	Gold Biotechnology	D-300–1
Imidazole	Fisher Chemical	O3196–500
Dithiothreitol (DTT)	GoldBio	DTT100
Ethylenediaminetetraacetic acid (EDTA)	Fisher Chemical	S311–500
Hydrochloric acid (HCl)	Fisher Chemical	A144–212
Formaldehyde	Fisher Chemical	F79500
Dimethylamine borane complex	Millipore Sigma	180238–5G
Deposited data
AbBamAPOTRA1–4 (P1 space group)	This study	PDB ID: 9CX4
AbBamAPOTRA1–4 (P3221 space group)	This study	PDB ID: 9CX5
Oligonucleotides
BamA-POTRA1–4 Forward Primer: AGGGCGCCATGGGAGCAGATGATTTCGTGGTCCG	This paper	N/A
BamA-POTRA1–4 Reverse Primer: GTGGTGCTCGAGTTAACGGCGAACGGTAACCTG	This paper	N/A
Recombinant DNA
pHis-Parallel2	Sheffield and Derewenda, 1999	N/A
pET20b	Novagen	69739
Software and algorithms
HKL2000	Otwinowski and Minor, 1997	https://hklxray.com/hkl-2000
Phenix	Adams et al., 2010	https://phenixonline.org/
Coot	Emsley et al., 2010	https://www2.mrclmb.cam.ac.uk/personal/pemsley/coot/
PyMOL	Schrodinger	https://pymol.org/2/
Photoshop	Adobe	https://www.adobe.com/products/photoshop.html
Illustrator	Adobe	https://www.adobe.com/products/illustrator.html
BioXTAS RAW	Hopkins et al., 2017	https://bioxtas-raw.readthedocs.io/en/latest/
ATSAS	Manalastas-Cantos et al., 2021	https://www.emblhamburg.de/biosaxs/software.html

EXPERIMENTAL MODEL AND STUDY PARTICIPANT DETAILS

The genes used for cloning originated from Acinetobacter baumannii and cloning performed as described in the Methods Details section. The Escherichia coli strains used for plasmid propagation and expression in this study are listed in the key resources table. The cells were grown as described in the Method Details section.

METHOD DETAILS

Cloning

Acinetobacter baumannii BamA (strain ATCC 19606, EnsemblBacteria reference #HMPREF0010_00353) was amplified by polymerase chain reaction (PCR) from a plasmid (pET20b) containing a cassette coding for the entire codon-optimized A. baumannii BAM complex (BioBasic). A portion of the BamA gene coding for POTRA 1–4 (amino acids 25–355) was amplified using forward primer 5’- AGGGCGCCATGGGAGCAGATGATTTCGTGGTCCG −3’ and reverse primer 5’- GTGGTGCTCGAGTTAACGGCGAACGGTAACCTG −3’. The BamA POTRA 1–4 insert was then gel extracted (IBI Scientific), restriction enzyme-digested using NcoI and XhoI, and ligated into the pHISParallel2 vector (Novagen). The resulting construct was then verified by DNA sequencing and transformed into BL21(DE3) cells for native protein expression, or the methionine auxotrophic B834(DE3) cells for expression of selenomethionine-substituted BamA POTRA 1–4.

Protein expression and purification

For expression of native BamA POTRA 1–4, a fresh colony was used to inoculate 5 mL of 2xYT containing 100 μg/ml ampicillin, and the culture was grown overnight at 37°C. The next day, the starter culture was washed with 5 mL of fresh 2xYT-Amp, and then sub-cultured 1:400 into 1 L of fresh 2xYT-Amp. Large-scale cultures were grown with shaking at 37°C to an OD₆₀₀ of 0.5–0.6. The cultures were then induced with 200 μM IPTG and incubated at 20°C overnight. Induced cells were harvested by centrifugation using a JA-14 rotor (Beckman) at 6,000 RPM for 20 minutes.

The cells were then resuspended in lysis buffer (1x PBS, 10 μg/mL DNase, 200 μM PMSF and 20 mM imidazole) and passed three times through an Avestin Emulsiflex-C3 homogenizer. The lysate was clarified by centrifugation in a JA-20 rotor (Beckman) at 18,000 RPM for 15 minutes, and the supernatant was collected. Immobilized metal ion affinity chromatography (IMAC) was then performed by applying the soluble supernatants to a 5 mL column packed with HisPur resin (Thermo Scientific) using a PURE FPLC system (GE Healthcare). A linear gradient using Buffer A (1x PBS, 363 mM NaCl and 20 mM imidazole) and Buffer B (1x PBS, 363 mM NaCl and 1 M imidazole) was used to elute BamA POTRA 1–4 from the HisPur column (0 – 50% Buffer B). Fractions containing BamA POTRA 1–4 were pooled, and the His tag was removed by incubation with TEV protease. Briefly, approximately 1 mg of TEV protease was added to the BamA POTRA 1–4 HisPur fractions (~1:100 molar ratio), and the mixture was dialyzed overnight at 4°C K against 1 L of 1x PBS, 0.5 mM EDTA, and 1 mM DTT. Three dialysis exchanges against 1 L of 1x PBS were performed, and then the BamA POTRA 1–4 TEV cleavage mixture was run over a 5 mL HisPUR column again as an additional reverse-IMAC purification step. The flow through from the reverse-IMAC step containing cleaved BamA POTRA 1–4 was collected and concentrated to 500 μL using a 10 kDa MWCO Amicon centrifugal concentrator (Millipore). Size-exclusion chromatography (SEC) was then performed as a final purification step, using a Superdex 200 Increase 10/300 GL column (GE Healthcare) in 1x PBS. Purified BamA POTRA 1–4 was separated on a 4–15% SDS-PAGE gel and visualized using PageBlue Protein Staining Solution (Thermo Scientific).

Expression of selenomethionine-substituted BamA POTRA 1–4 was performed identically to the native protein, only using the Selenomethionine Medium Complete media/nutrient mix system (Molecular Dimensions) containing 40 mg/L L-selenomethionine. Purification of selenomethionine-substituted BamA POTRA 1–4 was identical to the purification of the native protein, except that 1 mM 2-mercaptoethanol was added throughout the purification.

Reductive alkylation of selenomethionine-substituted BamA POTRA 1–4

Selenomethionine-substituted AbBamA POTRA 1–4 protein samples were reductively methylated prior to crystallization to increase crystal quality, packing, and size. In brief, 20 μL of 1.0 M dimethylamine borane complex solution and 40 μL of 1.0 M formaldehyde were added to 1 mL of protein at 10 mg/mL, and the mixture was incubated at 4°C for 2 hours. Again, 20 μL of 1.0 M dimethylamine borane complex solution and 40 μL of 1.0 M formaldehyde were added to the mixture prior to another 2-hour incubation at 4°C. Then, 10 μL of 1.0 M dimethylamine borane complex solution was added and the mixture was incubated overnight at 4°C. The reaction was stopped by adding 125 μL of 1.0 M glycine and 125 μL of 50 mM DTT followed by a 2-hour 4°C incubation. As a final step, SEC was performed in 1x PBS prior to crystallization.

Crystallization

For crystallization, BamA POTRA 1–4 was concentrated to 10 mg/ml in 1x PBS for the native protein, or 1x PBS with 1 mM 2-mercaptoethanol for the selenomethionine-substituted protein. Crystallization was performed at 20°C using the hanging drop vapor diffusion method in 96-well trays. Crystal trays were set using a Mosquito LCP (TTP Labtech) crystallization robot. Each hanging drop consisted of 0.2 μL protein solution and 0.2 μL well solution, and drops were equilibrated against 50 μL of well solution.

Commercially available broad-matrix screens were used for high-throughput crystallization screening (hanging drop) using an LCP-Mosquito crystallization robot (SPT Labtech). For native and selenomethionine-substituted samples, condition #49 (0.1 M citric acid, pH 4.0 and 0.8 M ammonium sulfate) from the AmSO₄ Suite (Qiagen) produced BamA POTRA 1–4 crystals within a month. Optimization of this condition was then performed by screening chemical conditions around the initial crystallization condition. A Dragonfly (STP Labtech) was used to generate 96-well optimization screens scanning different pHs (3.6 – 4.0) and ammonium sulfate concentrations (0 – 1.6 M). Additionally, an additive screen was performed using the Additive Screen HT (Hampton Research) in an attempt to identify any chemical additives which would improve crystal quality. For the alkylated selenomethionine-substituted sample, more conditions produced crystals during the initial screening than non-alkylated samples, with final crystals grown in 0.1 M bis-tris propane and 1.2 M DL-malic acid, pH 7.0 (SaltRx HT, #C9; Hampton Research).

Data collection and structure determination

AbBamA POTRA 1–4 crystals were flash-frozen in liquid nitrogen prior to data collection. X-ray diffraction data were collected at beamline 23-ID-B (GM/CA) of the Advanced Photon Source, Argonne National Laboratory, for the native crystals and at beamline 23-ID-D (GM/CA) for the selenomethionine-substituted crystals. Data processing and analysis were performed using HKL-2000⁵⁷ and Xtriage⁵⁸. Attempts to phase using molecular replacement were unsuccessful. Therefore, initial phases were calculated using data from the selenomethionine-substituted crystalsin space group P3₂21 using Se-SAD in AutoSol/PHENIX⁵⁸. Two molecules were observed per asymmetric unit containing a total of 10 selenium sites. The model from Se-SAD phasing was then used as a search model to solve the structure in space group P1 by molecular replacement within PHASER/PHENIX⁵⁸. Model building was performed using Coot⁶¹ and refined using Phenix⁵⁸. Data collection and refinement parameters and results are summarized in Table 1. Structural analysis and figure preparation were performed using PyMOL (Schrödinger). Final figures were assembled using Adobe Photoshop and Illustrator.

Size-exclusion chromatography small angle X-ray scattering (SEC-SAXS)

POTRAs 1–4 from AbBamA (5 mg/mL) were analyzed with SEC-SAXS (Superdex 200 Increase) to analyze its oligomeric state and overall shape. The SEC-SAXS data were collected at the BioCAT beamline at the Advanced Photon Source, Argonne National Laboratory. The data were analyzed and final plots made using BioXTAS RAW v2.2.1⁶² and ATSAS⁶³. First, the data were reduced, blank subtracted, and data range for scattering curves selected. Upon averaging of the data, the q-range and molecular weight information was obtained by Guinier analysis. Pair-distance distribution curves were calculated using GNOM. Theoretical scattering curves for the X-ray structures was calculated and compared with experimental scattering curves using CRYSOL and model refinement performed using the SREFLEX option.

Molecular dynamics simulations

We constructed simulation systems using two chains from the structure resolved here and two chains from EcBamA POTRA domains 1–4 (PDB ID 2QCZ). Each chain was individually placed in a TIP3P water box with sodium (Na⁺) and chloride (Cl⁻) ions added to achieve a salt concentration of 150 mM. The final system volume was approximately 159×159×159 ų and comprised around 400,000 atoms. Each simulation was conducted for 500 ns.

The simulations were performed using NAMD3⁶⁴ with the CHARMM36m⁶⁵ force field. Hydrogen mass repartitioning⁶⁶ was applied to enable a consistent time step of 4 fs. We maintained the simulations at a constant temperature of 310 K and a pressure of 1 atm using Langevin dynamics and a Langevin piston for temperature and pressure control, respectively.

Quantification And Statistical Analysis

Data collection and refinement statistics for X-ray crystallography are shown in Table 1. The summary of SEC-SAXS parameters is in Table 2.

Highlights.

1. The POTRA domains of BamA are known to be flexible.

2. Crystal structures of A. baumannii BamA POTRA domains reveal novel conformations.

3. MD simulations and SAXS data are consistent with the crystal structures.

4. These new conformations may explain how Bam components become surface exposed.

Abbreviations:

(POTRA)

polypeptide transport associated

(BAM)

β-barrel assembly machinery

(SEC)

size-exclusion chromatography

(SAXS)

small angle X-ray scattering

(RMSD)

room mean square deviation

(SAD)

single-wavelength anomalous dispersion

(MAD)

multiple-wavelength anomalous dispersion

(PDB)

Protein Data Bank