Integral membrane proteins are firmly embedded into cellular membranes by virtue of their hydrophobic transmembrane domains. These are typically folded into either an α-helix bundle or a β-barrel (1). Both conformations allow the full saturation of the protein backbone’s hydrogen-bonding potential within the hydrophobic environment of the membrane, thereby ensuring their highest stability and adhering to the second law of thermodynamics (1,2). In fact, the thermodynamic stability of different bacterial β-barrel proteins ranges between -3.4 to -32.5 kcal/mol, indicative of their high stability (3). Helix bundle membrane proteins are present in virtually all cellular membranes, while β-barrel membrane proteins are prominent in the outer membrane (OM) of Gram-negative bacteria as well as their endosymbiotic descendants present in eukaryotic cells: mitochondria and chloroplasts (4). In the latter organelles, outer membrane proteins (OMPs) typically function as substrate-specific pores for the uptake of small molecules through passive diffusion. In humans, dysfunction of mitochondrial OMPs is associated with different pathologies, including diabetes and Parkinson’s disease. In bacteria, however, OMPs are functionally more diverse as they are not only involved in small molecule transport but are also important for maintaining structural integrity of the cell wall, lipopolysaccharide modification as well as attachment site for bacteriophages and toxins (4). Moreover, in pathogenic bacteria OMPs often serve as virulence factors that mediate infection. A mature OMP is made up of β strands that are arranged into antiparallel β sheets, which are folded into a cylindrical barrel through establishing hydrogen bonds between the first and last β strands. This interaction is called the seam. In the final barrel structure, the N and C-termini face the periplasm with apolar and polar residues located alternately throughout the transmembrane part of the β strand (2,4). The apolar residues are found at the exterior of the barrel (and thus face the lipid bilayer), while polar residues reside on the inside. The barrel is anchored into the bilayer through aromatic residues present at the water-membrane interface that interact with membrane lipids, thereby contributing to OMP stability (3,4).
Strikingly, within bacterial OMPs, charged residues are prominent in extracellular parts, whereas these residues are less favored intracellular parts and this asymmetric distribution is known as positive-outside rule (3). Within bacterial OMPs, the level of sequence conservation increases towards the C-terminal strands with a conserved sequence, the β-signal, present at the C-terminus of many OMPs. This sequence is recognized by chaperones and the OM insertion machinery (4). Although all OMPs contain a characteristic cylindrical β-barrel, they show a marked variety in size and shape, while, moreover, some are monomeric and others are known to oligomerize. Interestingly, all bacterial OMPs are even stranded, whereas mitochondrial OMPs are also odd stranded (4). As shown in figure 1, OmpA is one of the smallest bacterial OMPs with 8 strands (in magenta), while the porin PhoE and the maltose transporter LamB are made up of 16 (in magenta) and 18 strands (in red), respectively. PapC, that is involved in the assembly of pilli subunits on the bacterial cell surface, and LptD, which is required for the assembly of lipopolysaccharide into the OM, represent the largest bacterial OMPs as yet known with 24 (in red) and 26 strands (in yellow). The human mitochondrial OMP hVDAC1 represents a voltage-dependent anion channel of 19 strands (in yellow). The biogenesis of OMPs is a complex and multistep process, involving either protein export or import machineries, different molecular chaperones and the OM insertion machinery (5). The latter is a crucial player in OMP biogenesis and facilitates the final steps of their assembly into the OM. In Gram-negative bacteria, the BAM (barrel assembly machinery) complex is responsible for the insertion and folding of β-barrel proteins into the OM, while in mitochondria and chloroplasts these tasks are performed by the evolutionarily related SAM (sorting and assembly machinery) and Toc75 (translocase of the OM of chloroplasts), respectively (5). The bacterial BAM complex represents the best characterized system for the insertion and folding of OMPs. In the Gram-negative bacterium Escherichia coli, the BAM apparatus represents a heteropentameric complex that is made up of the β-barrel protein BamA and four lipoproteins, BamB-E (2). Despite the wealth of genetic and biochemical data on the bacterial BAM system, it is still not well understood how this complex processes a great diversity of OMPs. It is therefore important to note that recently detailed structures of the substrate-engaged E. coli BAM complex were presented, providing profound molecular insight into the way how newly synthesized β-barrel proteins are folded and inserted into the OM (6). Here, I will discuss these structures as well as their current functional understanding.
How are outer membrane proteins made?
Much of what we know about OMP biogenesis has been acquired through genetic and biochemical experiments in bacteria and E. coli in particular. It has been estimated that the E. coli proteome contains about 100 potential OMPs, although their precise number is uncertain because the primary sequence of these proteins is difficult to recognize by in silico analysis (7). The basic steps in the biogenesis of E. coli OMPs are presented in figure 2 (obtained from 8) and reveals that they are initially produced by cytosolic ribosomes as precursors with an N-terminal signal sequence. With the help of the SecB chaperone that recognizes the signal sequence and prevents folding, newly synthesized OMPs are delivered at the Sec translocon (2). Following binding at the translocon, the OMP precursor is transported across the cytoplasmic membrane towards the periplasmic space through the Sec translocon in an unfolded state. Translocation of the precursor is powered by the SecA ATPase at the expense of ATP hydrolysis. At the periplasmic side of the cytoplasmic membrane, the signal sequence is proteolytically removed and the mature polypeptide is released into the periplasm.
Here, it is captured by periplasmic chaperones, such as Skp and SurA, that keep the mature OMP in an unfolded state and guide it to the OM, while the DegP protease degrades misfolded OMPs in the periplasm primarily under stress conditions (2). Skp and SurA are functionally redundant as neither of these chaperones alone is essential for cell viability, while the absence of both is lethal (9). At the OM, the BAM complex catalyzes folding and insertion of β-barrel proteins into the OM that were shipped across the periplasm by Skp and SurA. Interestingly, the steps that occur in the periplasm and OM don’t require ATP, PMF or a redox potential, which implies that they are spontaneous and thermodynamically favorable (4).
Biological and structural features of the BAM complex
Biological features of BamABCDE
Despite the well-established initial steps in the biosynthesis of bacterial OMPs, the mechanism of how these proteins are assembled in the OM remained enigmatic for a long time. The first proteinaceous component implicated in the assembly of OMPs was identified in 2003. Specifically, it was shown in Neisseria meningitidis that Omp85, a highly conserved OMP, is essential for cell viability and its absence unfolded OMP precursors accumulate in the periplasm (10). In the same study evidence was, moreover, presented that Omp85 is part of a high molecular weight complex. Likewise, it was established that in Pseudomonas aeruginosa Opr86, a member of the Omp85 family, is essential for viability and plays a key role in the assembly of OMPs (11). Consistent with these results, it was also reported that BamA, the Omp85 homolog of E. coli, is an essential protein and mediates OMP assembly (12). BamA is an OM-embedded β-barrel protein of about 90 kDa with five N-terminal POTRA (polypeptide transport-associated) domains that are localized to the periplasm and a C-terminal transmembrane barrel. BamA forms a heteropentameric complex of ~ 203 kDa with four lipoproteins termed BamB-E that are anchored to the outer membrane through their lipid-modified N-terminal cysteine. Of these, BamB has a molecular weight of about 42, BamC 37, BamD 28 and BamE 12 kDa, respectively. BamA and BamD are the only subunits of the BAM complex that are essential for growth in E. coli, while deletion of the other subunits does not affect cell viability and only results in mild OMP assembly defects (12). However, all five Bam subunits are required for efficient OMP assembly in an in vitro reconstituted assay (13). One of the reasons why BamD is a crucial component of the Bam complex is that it probably captures incoming substrates because it binds unfolded OMPs through their conserved β signal and thereby determines the substrate specificity of the Bam complex (14). In addition to substrate recognition, BamD may also regulate the activity of BamA (2). While the other lipoprotein subunits are not essential for activity of the Bam complex, it is conceivable that they are required for the folding and insertion of a subset of large and difficult OMP substrates such as LptD (2). Thus, BamA and BamD are the essential core components that are ubiquitously present in Gram-negative bacteria, while BamC-E are less conserved and probably fulfill an auxiliary role in the folding and insertion of OMPs.
General architecture of BamABCDE
Several high-resolution structures of different Bam proteins have been solved, including a detailed structure of the complete BamABCDE complex at 2.9 Å (REF). This structure is shown in figure 3 in both surface (top panel) and ribbon representation (bottom panel) with subunits colored as indicated. Overall, the Bam complex resembles a top hat with the β-barrel of BamA forming the crown, while the POTRA domains and associated subunits make up the brim in the periplasm. The inner surface of this periplasmic ring is formed by BamD and the POTRA domains. The barrel domain of BamA is made up of 16 β-strands that are folded into a cylindrical barrel with a water-filled lumen. When viewed from the top, the BamA barrel is closed by several extracellular loops, thereby preventing the transport of small molecules through the BamA cylinder. The POTRA domains of BamA lack significant sequence similarity but display a similar overall architecture, comprising two α-helices that cover three β-sheets (β α α β β fold). Functionally, the POTRA domains serve as scaffold for the attachment of BamB-E (bottom panel of figure 3). Specifically, BamB is associated with POTRA 2 and 3. The N-terminal part of BamC is bound to BamD as well as POTRA 1 of BamA, whereas its C-terminal part is associated with BamD and POTRA 2. BamD is attached to POTRA 1, 2 and 5 as well as a periplasmic loop of the BamA barrel. The functional significance of the interactions with POTRA 1 and 2 is not clear because the interaction with POTRA 5 is probably the most important one as its deletion strongly affects assembly of the entire complex as evidenced by the formation of different subcomplexes. BamE is located at the back of the complex and is bound to the C-terminal domain of BamD and POTRA 4 and 5 of BamA. In addition to serving as a scaffold for the attachment of the four lipoprotein subunits, the POTRA domains of BamA also guide OMP substrates into the barrel.
Open and closed BamA barrel structures
One of the final steps in the folding of OMPs comprises closure of the cylindrical barrel at the seam through hydrogen bond formation between the N and C-terminal β strands. This allows complete membrane integration of the mature OMP. The seam of BamA is formed by strands β1 and β16 and, interestingly, BamA exhibits a relatively low thermodynamic stability as evidenced by its melting temperature of 37oC (2). This suggests that BamA is highly dynamic under physiological conditions and may therefore be structurally unstable. Consequently, structures of BamA in different conformations were reported with poor hydrogen bond formation at the seam. Figure 4, for example, shows a close up of the BamA barrel domain in ribbon representation with a closed seam (left panel) or open seam with unpaired N- and C-terminal β-strands (right panel) (15,16). Based on this it was proposed that conformational changes with BamA induce opening of the lateral gate at the seam, allowing membrane insertion of new OMP barrels (2). The functional importance of the seam was probed by expressing BamA variants in E coli cells containing disulfide crosslinks that prevent its lateral opening. This prevented the growth of these cells, showing that lateral opening is required for BamA activity (17). Moreover, it was also shown by reconstituting the assembly of OmpT in vitro that lateral gating of BamA is required for full BamA activity, although proteoliposomes containing BamA with a crosslinked seam did not completely prevent the assembly of OmpT (18). Furthermore, in the same study a Cryo-EM structure of the complete BamABCDE complex of E. coli was presented with a partially open barrel domain of BamA (18).
Important insight into the role of the proposed lateral gate of BamA was provided recently by different crosslinking studies to map the molecular path of OMP substrates that are assembled by the Bam complex. Specifically, LptD was used as a model protein to study its initial interactions with BamA through site-specific in vivo photo crosslinking (19). To this end, BamA variants were engineered containing para-benzoylphenylalanine (pBPA), a photo-reactive phenylalanine analog, at various positions at the lateral gate of BamA were β1 and β16 meet as well as in the sixth extracellular loop (EL6). All these positions are believed to be important in OMP assembly. Following UV irradiation of E. coli cells expressing LptD and the respective BamA variants, the latter were purified and BamA-substrate crosslinks were investigated by immunoblotting. This showed that LptD contacts both β1 and β16 at the lateral gate, while EL6 of BamA was also crosslinked to LptD, thereby confirming that these regions of BamA directly interact with substrate OMPs. EL6 is located within the barrel domain, which suggests that substrates gain access to the lumen of the barrel. This was investigated by incorporating pBPA inside the barrel domain, resulting in several BamA-substrate crosslinks, showing that the interior of barrel directly interacts with substrates. Collectively, these results indicate that the interior surface of the BamA barrel domain and the lateral gate comprise a substrate-binding site. By incorporating pBPA into the C-terminus of LptD crosslinks to BamD were found. This is consistent with the finding that the C-terminus of OMP substrates (the β signal) interacts with BamD. Moreover, extensive contacts between LptD and the inside of the BamA β-barrel were also found through introducing pBPA substitutions in the N-terminal part of LptD, thereby verifying that a large part of BamA’s interior surface serves as substrate-binding site probably to chaperone β-sheet folding of OMP substrates. In sum, the C-terminus of LptD is bound by the N-terminus (β1) of BamA, while the N-terminal part of LptD is assembled inside the barrel domain. Following assembly of the substrate, its N-terminus will pair with its C-terminus to close the β-barrel, thereby allowing insertion into the OM. Another study employed disulfide crosslinking to probe the interactions of a β-barrel substrate protein bound by BamA at a late stage of assembly in E. coli cells (20). Therefore, the autotransporter EspP was used as a model protein. Autotransporters are a specialized class of proteins that typically serve as virulence factors and are made up of three domains, namely: A N-terminal signal sequence, an effector domain that is fully secreted and a C-terminal β-barrel that is embedded into the OM. To arrest OM assembly after association of the β-barrel, the passenger domain was extended with maltose-binding protein (MBP) that rapidly folds in the periplasm. This system, in which the assembly of a β-barrel is arrested, was used next to map the interactions with BamA by incorporating pairs of cysteine in the C-terminal β strands near the lateral gate (β15 and 16) of BamA as well as N-terminal positions in EspP-MBP. Following chemically-induced disulfide bond formation, crosslinks between β15 and the model protein were found, thereby showing that the C-terminal β-strands of BamA are in close contact to the N-terminal part of EspP-MBP. Moreover, crosslinks between the N-terminal β-strand of BamA and the C-terminal β-strand of the model protein were also observed. Collectively, these results show that BamA and EspP-MBP interact via two distinct interfaces, namely: the β-signal of the model protein and the first β-strand of BamA and the C-terminal β-strands of BamA and the N-terminal part of EspP-MBP. A disulfide crosslinking approach was, moreover, also used recently to study the molecular environment of a mitochondrial OMP (VDAC1) with Sam50, the Omp85 homolog in mitochondria, during OM assembly (21). This showed that VDAC1 also interacts with the lateral gate, resulting in hybrid-barrel. Specifically, the β signal of VDAC1 associates with β1 of Sam50, while the N-terminal part of the substrate interacts with β16 of Sam50. Similar to LptD, VDAC1 interacts with loop 6 inside the Bam50 pore, while its N-terminal part associates with the interior of the β-barrel as it probably moves through the Sam50 channel. In conclusion, the in vivo crosslinking studies discussed above, employing endogenous OMP substrates in different biological systems, point towards the formation of a hybrid-barrel during assembly with one interface between the N-terminal part of the OMP substrate and the N-terminal β-strand of the lateral gate and another interface between the β signal of the substrate and the C-terminal β-strand of the lateral gate of the bacterial and mitochondrial Omp85 homologs.
Structure of substrate-engaged BAM complex
Despite the molecular insight provided by these crosslinking studies, a complete understanding of how the BAM complex accelerates the folding of OMP substrates without any obvious external energy source requires structures of substrates that are trapped on the BAM complex. With regards to this, it is important to note that recently a detailed structure of E. coli BamA caught in the act of folding an endogenous OMP substrate was presented (6). To obtain this structure, a slow folding BamA variant, lacking extracellular loops, was used as model substrate. This variant is expected to accumulate on the BAM machinery due to its slow folding, which is a prerequisite to trap it successfully during OM assembly. Indeed, in vivo site-specific photo crosslinking confirmed that this BamA variant accumulates on the BAM complex. To stabilize its interaction with BamA, a disulfide bond was introduced between the C-terminal region of wild-type BamA and the N-terminal region of the BamA substrate. Subsequently, the BamABCDE complex containing the BamA substrate was purified and its structure was solved by cryo-EM at 4.1 Å resolution. This structure is shown in figure 5 in surface representation (top left panel) and reveals that all subunits were resolved as well as the trapped BamA substrate. The overall structure of the substrate-engaged Bam machinery is similar to that of the substrate-free complex. The organization of the periplasmic Bam parts is comparable to that of the structures without substrate with BamA in a lateral open conformation. Although the structures of the individual subunits of the substrate-engaged complex are similar to previously reported structures, some significant conformational changes are observed in BamA when compared to its structure without substrate. For example, β-strand 1 and 2 of substrate-engaged BamA are rotated towards the outside, thereby separating them from the C-terminal β-strand, while similar conformational changes are also observed in β3, 4 and 5. As indicated by crosslinking studies, the barrel domains of BamA and the substrate associate during OM assembly, resulting in a hybrid barrel. Indeed, within the structure of the substrate-bound BAM complex the β domains of BamA and its client BamA form a hybrid barrel (center top panel), thereby confirming the aforementioned crosslinking studies. The N-terminal β-strand of substrate-engaged BamA (β1) and the C-terminal β-strand of the BamA substrate (β16) are joined together through six hydrogen bonds. This results in a hybrid barrel with a continuous lumen as shown in the top-down view of the complex presented in the top right panel. Both β-domains have, moreover, a lateral opening. In addition to the paired edge between β1 and β16, the hybrid barrel contains an unpaired edge made up of the C-terminal β-strand of substrate-bound BamA and the N-terminal β-strand of the substrate that are not hydrogen bonded. Both ends are folded into the lumen (top right panel) in such a way that the barrel domains interact along their exterior surface. The outside of the substrate’s barrel domain is mainly exposed to the OM, while its N-terminal β-strands interact with substrate-engaged BamA. Thus, the hybrid barrel comprises an unpaired edge at one end made up of substrate-bound BamA and the BamA substrate with the unpaired β-strand tucked away into the lumen of the hybrid barrel.
A close-up of the paired interface between the N-terminal β-strand of substrate-bound BamA (β1) and the C-terminal β-strand of the BamA substrate (β16) is provided in the bottom left panel of figure 5. Five residues on either β-strand (shown as raspberry or orange or sticks) interact through six hydrogen bonds, thereby creating a thermodynamically stable interaction. Moreover, the paring between the exposed N-terminal edge of substrate-engaged BamA (mature BamA) and the C-terminal edge of the substrate creates a template for the folding of new β-strands that occurs in the C to N-terminal direction. The stable interaction between the six hydrogen bonds at the paired interface prevents premature product release. This raises the question how the release of folded OMPs from the BAM complex is achieved? The structure of substrate-bound BamA provides a potential answer. Firstly, no hydrogen bonds need to be broken at the unpaired interface comprising the N-terminal β-strand of the substrate and the C-terminal β-strand of mature BamA, which are turned inside the lumen of the hybrid barrel (top right panel). Secondly, the hydrogen bonds at the paired edge of the hybrid barrel are not the same as the hydrogen bonds at the seam of a substrate’s folded barrel domain. Within the structure of the substrate-bound BAM complex, BamA wedges open at the extracellular side of its N-terminal β-strand, thereby facilitating interaction with residues in the C-terminal β-strand of the substrate. These residues, however, normally interact with residues of the opposing β-strand in the fully folded barrel domain of the substrate. The pairing between β16 of the substrate and β1 of mature BamA creates an overhang at the C-terminal β-strand of the substrate (right bottom panel) due to a steric clash with β-strands 14 and 15 of mature BamA. This overhang is formed by residues Gly807 to Trp810 (shown as red sticks). Upon folding of the substrate’s barrel domain, the N and C-termini begin to form hydrogen bonds, while sequentially disrupting the bonds at the paired interface. As the seam of the substrate’s barrel domain closes, its C-terminal β-strand peels away in a step-wise fashion from the β-barrel of mature BamA. Thus, hydrogen bond forming between the C-terminal and N-terminal β-strand of the substrate facilitate its release from the BAM complex.
Mechanism of BAM-mediated outer membrane protein assembly
Several models have been proposed to describe the folding and insertion of OMPs. Of these, the budding and assisted model are currently the most favored ones (figure 6 adapted from 2). Both are based on the finding that the seam of BamA is structurally unstable and can open laterally. The budding model proposes that β-strands of unfolded OMPs pass through the barrel domain of BamA and are templated at the open BamA seam, thereby forming a hybrid BamA/substrate barrel. Once the barrel domain of the substrate OMP is folded, it closes and is released into the OM. In the assisted model, OMP folding occurs at the periplasmic side of the OM and once it is complete the newly folded β-barrel is inserted into the OM as a single unit by the BAM complex. It is thought that BamA lowers the energy barrier for membrane insertion through creating a local membrane defect (outer membrane thinning) near its seam (2).
The structure of the substrate-bound BAM complex is consistent with models in which substrate OMPs move from the side of the BamA β-barrel into the OM. Moreover, it verifies crosslinking studies that have revealed that the C-terminal β-strand of the substrate is bound by the N-terminal β-strand of mature BamA. Unexpectedly, the structure also shows that one edge of the hybrid barrel is unpaired, while the paired edge contains an overhang in the C-terminal β-strand of the substrate. Here, release of the substrate’s barrel domain is initiated through sequential replacement of hydrogen bonds as the seam of the substrate’s β-barrel closes. The novel structure of the substrate-bound E. coli BAM complex therefore supports the budding model for the assembly of bacterial OMPs. Based on the wide spread conservation of Omp85 homologs, it is conceivable that the mechanism of β-barrel closure and release has also been retained during evolution. Indeed, it was recently shown that folding and insertion of mitochondrial OMPs also occurs via the budding model (21).
Despite their structural diversity, the basic architecture of integral membrane proteins is relatively simple. Two general protein folds are known that span the membrane and firmly anchor membrane proteins into the lipid bilayer, namely: the helix bundle and β-barrel (1). Membrane proteins of the helical bundle class are common in all biological membranes, while β-barrel proteins are prominent in the outer membrane of Gram-negative bacteria, mitochondria and chloroplasts. Mitochondrial OMPs are mainly involved in small molecule transport, whereas bacterial OMPs are functionally more diverse, ranging from structural cell wall component to lipopolysaccharide modification and virulence factors that mediate infection in pathogenic bacteria. In contrast to the detailed mechanistic understanding of how α-helical membrane proteins are produced, the mechanism of how OMPs are assembled in the OM remained enigmatic for a long time. Important insight into this was, however, provided by Gram-negative bacteria (2). For example, it was shown in Neisseria meningitidis that Omp85, a highly conserved OMP, is essential for cell viability and its absence unfolded OMP precursors accumulate in the periplasm. Likewise, it was also reported that BamA, the Omp85 homolog of E. coli, is an essential protein and mediates OMP assembly (12). BamA forms a heteropentameric complex with four lipoproteins termed BamB-E. Currently, detailed structures of individual BAM subunits as well as the entire complex are available but despite this precise information on how OMPs are folded and inserted into the OM is scant. Nevertheless, several models for OMP assembly have been put forward of which the favored ones propose either an active or passive role for BamA. In the budding model, BamA is thought to catalyze the folding of new β-strands in OMP substrates at its open seam. This results in a hybrid barrel comprising the barrel domain of BamA and the substrate, which collapses after the folding of the substrate’s β-barrel is complete and is released into the OM. In the assisted model, however, folding of new OMPs occurs in the periplasm after which the complete barrel domain is inserted by BamA through lowering the energy barrier for membrane insertion (2). Recent data obtained by crosslinking studies points towards the formation of a hybrid barrel during OMP assembly, which is confirmed by a novel structure of substrate-bound BamA (6). Thus, strong biochemical and structural evidence supports the budding model for the folding and insertion of new OMPs. Moreover, BamA has emerged as a promising target for the development of novel antibiotics against Gram-negative pathogens, including multi-drug resistant bacteria. In fact, the first members of this novel class of antibiotics are currently clinically evaluated (22). It can be expected that the new structure of substrate-bound BamA will facilitate the design of additional small molecule drugs that interfere with the function of BamA.
1. von Heijne G. (2007). The membrane protein universe: what's out there and why bother? J Intern Med. 261: 543-557.
2. Konovalova A, Kahne DE, Silhavy TJ. (2017). Outer Membrane Biogenesis. Annu Rev Microbiol. 71: 539-556.
3. Schiffrin B, Brockwell DJ, Radford SE. 2017. Outer membrane protein folding from an energy landscape perspective. BMC Biol. 15: 123.
4. Chaturvedi D, Mahalakshmi R. 2017. Transmembrane β-barrels: Evolution, folding and energetics. Biochim Biophys Acta Biomembr. 1859: 2467-2482.
5. Walther DM, Rapaport D, Tommassen J. 2009. Biogenesis of beta-barrel membrane proteins in bacteria and eukaryotes: evolutionary conservation and divergence. Cell Mol Life Sci. 66: 2789-804.
6. Tomasek D, Rawson S, Lee J. et al. 2020. Structure of a nascent membrane protein as it folds on the BAM complex. Nature. 583: 473-478.
7. Elofsson A, von Heijne G. 2007. Membrane protein structure: prediction versus reality. Annu Rev Biochem. 76: 125-140.
8. Wu R, Stephenson R, Gichaba A. et al. 2020. The big BAM theory: An open and closed case? Biochim Biophys Acta Biomembr. 1862: 183062.
9. Sklar JG, Wu T, Kahne D. et al. (2007). Defining the roles of the periplasmic chaperones SurA, Skp, and DegP in Escherichia coli. Genes Dev. 21: 2473-2484.
10. Voulhoux R, Bos MP, Geurtsen J. et al. 2003. Role of a highly conserved bacterial protein in outer membrane protein assembly. Science. 299: 262-265.
11. Tashiro Y, Nomura N, Nakao R. et al. 2008. Opr86 is essential for viability and is a potential candidate for a protective antigen against biofilm formation by Pseudomonas aeruginosa. J Bacteriol. 190: 3969-3678.
12. Wu T, Malinverni J, Ruiz N. et al. 2005. Identification of a multicomponent complex required for outer membrane biogenesis in Escherichia coli. Cell. 121: 235-245.
13. Hagan CL, Kim S, Kahne D. 2010. Reconstitution of outer membrane protein assembly from purified components. Science. 328: 890-892.
14. Hagan CL, Wzorek JS, Kahne D. 2015. Inhibition of the β-barrel assembly machine by a peptide that binds BamD. Proc Natl Acad Sci U S A. 112: 2011-2016.
15. Gu Y, Li H, Dong H. 2016. Structural basis of outer membrane protein insertion by the BAM complex. 531: 64-69.
16. Bakelar J, Buchanan SK, Noinaj N. 2016. The structure of the β-barrel assembly machinery complex. Science. 351: 180-186.
17. Noinaj N, Kuszak AJ, Balusek C. et al. 2014. Lateral opening and exit pore formation are required for BamA function. Structure. 22: 1055-1062.
18. Iadanza MG, Higgins AJ, Schiffrin B. et al. 2016. Lateral opening in the intact β-barrel assembly machinery captured by cryo-EM. Nat Commun. 7:12865.
19. Lee J, Tomasek D, Ma Santos T. et al. 2019. Formation of a β-barrel membrane protein is catalyzed by the interior surface of the assembly machine protein BamA. Elife. 8: e49787.
20. Doyle MT, Bernstein HD. 2019. Bacterial outer membrane proteins assemble via asymmetric interactions with the BamA β-barrel. Nat Commun. 10: 3358.
21. Höhr AIC, Lindau C, Wirth C. et al. 2018. Membrane protein insertion through a mitochondrial β-barrel gate. Science. 359: eaah6834.
22. Luther A, Urfer M, Zahn M. et al. 2019. Chimeric peptidomimetic antibiotics against Gram-negative bacteria. Nature. 576: 452-458.