The β-barrel assembly machinery in motion

Abstract

In Gram-negative bacteria, the biogenesis of β-barrel outer membrane proteins is mediated by the β-barrel assembly machinery (BAM) complex. Over the past decade, structural and functional studies have collectively contributed to advancing our understanding of the BAM complex structure and function; however the exact mechanism remains elusive. In this Progress article, we discuss recent structural studies that have revealed that the accessory proteins may regulate essential unprecedented conformational changes within the core component BamA during function. We also review the mechanistic insights that have been gained from structural data, mutagenesis studies, molecular dynamics simulations, and explore two emerging models for BAM-mediated OMP biogenesis in bacteria.

ToC

In this Progress article, Buchanan and colleagues discuss recent structural studies that have advanced our understanding of the structure of the fully assembled β-barrel assembly machinery (BAM) complex and the interactions between the individual components. They also review the mechanistic insights that have been gained and explore two emerging models for BAM-mediated OMP biogenesis in bacteria.

Transmembrane proteins consist of either an α-helical fold or a β-barrel fold. α-helical membrane proteins contain one or more transmembrane-spanning domains (TMDs) that consist of hydrophobic α-helices, whereas monomeric β-barrel proteins contain between 8–26 TMDs consisting of amphipathic β-strands. The β-strands are arranged in a linear antiparallel β-sheet, with the first strand and last strand interacting at a junction to form the classic barrel-like shape from which their name is derived (Fig. 1a, b). Whereas α-helical transmembrane proteins can be found in all cellular membranes, β-barrel transmembrane proteins are only found within the outer membranes of Gram-negative bacteria and the endosymbionts apicoplasts, chloroplasts, and mitochondria; and more generally referred to as β-barrel outer membrane proteins (OMPs)^{1, 2}. OMPs are essential for many cellular processes, including nutrient import, outer membrane biogenesis and motility³. Additionally, for pathogenic bacteria, OMPs can function as virulence factors that mediate pathogenesis and infection^{4, 5}. Despite their essential roles, the mechanism for the biogenesis of OMPs remains poorly understood.

Figure 1. A journey toward the biogenesis of a β-barrel outer membrane protein.

a. Schematic of the architecture of a generic β-barrel outer membrane proteins (OMP), which consists of a linear arrangement of an even number of antiparallel β-strands (ranging from 8–26) where the first strand and last strand (also referred to as the β-signal) interact at the junction site to form a barrel shape within the membrane. Residues oriented inside the barrel domain (indicated by cyan arrows) are primarily hydrophilic, interacting with solvent, other domains, and/or other proteins, while the residues oriented outside the barrel domain (indicated by red arrows) are almost exclusively hydrophobic, mediating interactions with the hydrophobic core of the membrane. b. An example of an OMP, the structure of Ail (PBD ID 3QRA) from Y. pestis containing eight strands. The first strand is shown in green and the last strand is shown in yellow. c. Gram-negative bacteria contain both an inner membrane and an outer membrane, where the inner membrane partitions the cytoplasm and periplasm and peptidoglycan. The inner membrane contains exclusively α-helical membrane proteins, whereas the outer membrane contains almost exclusively β-barrel outer membrane proteins; only a few examples of the outer membrane containing α-helical membrane proteins are known. d. The biogenesis of an OMP begins within the cytoplasm where it is translated with an N-terminal leader sequence (green). It is then translocated across the inner membrane into the periplasm by the Sec machinery. The periplasmic chaperones SurA (primary pathway) and Skp (rescue pathway) then stabilize and further escort the nascent OMP to the BAM complex for final insertion into the outer membrane. If OMPs become misfolded and cannot be rescued by Skp, they will be degraded by DegP to prevent toxicity to the cell. e. In E. coli, the BAM complex consists of five components: the essential core component BamA, which is an OMP itself, and four accessory lipoproteins termed BamB, BamC, BamD and BamE, each containing an N-terminal post-translational lipid modification that anchors them to the inner leaflet of the outer membrane. Based on microscopy studies, the majority of the soluble domain of BamC is found at the surface of the bacteria, as depicted. However, recent structural studies show close association with BamD, and further studies are required to determine whether the proposed presence of BamC at the surface serves as part of its role within the BAM complex. f. The interaction network within the BAM complex, based on experimental studies, including results from recent structural studies. The BAM complex has been purified as two separate modules containing BamAB and BamCDE.

The unique properties of OMPs offer some clues about the machinery that is responsible for placing them into the outer membrane. OMPs have a much different fold than α-helical membrane proteins; they consist of a barrel-shaped β-sheet containing an intricate hydrogen bonding network that creates an extremely stable tertiary structure which is often resistant to denaturants such as sodium dodecyl sulfate^{6, 7}. Each amphipathic strand contains hydrophobic residues that point towards the outside of the barrel and mediate interactions with the membrane bilayer, whereas the residues that point towards the inside or the lumen of the barrel are typically hydrophilic residues that interact with solvent, protein substrates and/or accessory proteins (Fig. 1a). These properties of OMPs make their biogenesis a unique and complex folding problem that requires complex machinery to accomplish.

Our current understanding of OMP biogenesis has come primarily from studies in Gram-negative bacteria such as Escherichia coli, Neisseria meningitidis, and Neisseria gonorrhoeae. Unlike α-helical membrane proteins, OMPs must be transported across the inner membrane, periplasm and peptidoglycan before they are inserted into the outer membrane ^8–11 (Fig. 1c). This journey begins in the cytoplasm, where nascent OMPs are fully translated with an N-terminal leader sequence that directs them to the Sec export machinery, which transports the nascent OMPs across the inner membrane into the periplasm^{8–10, 12} (Fig. 1d). Chaperones in the periplasm then stabilize the nascent OMPs and escort them across the periplasm to the outer membrane¹³. The chaperone SurA has been implicated to be the key chaperone during targeting of OMPs ; however, Skp has also been shown to have a role during chaperone-mediated biogenesis for some OMPs, even rescuing ‘off-pathway’ OMPs that are usually chaperoned by SurA¹³. However, recent studies indicate a more complicated pathway where OMPs may experience complex chaperone networks that are driven by energy landscapes within the periplasm^14–16. Misfolded OMPs that cannot be rescued are degraded by the periplasmic protease DegP to prevent their accumulation and thus cellular toxicity^{13, 17–19}. Once at the outer membrane, the nascent OMPs are delivered to the β-barrel assembly machinery (BAM) complex, which mediates their biogenesis into the outer membrane^{8–10, 20}. The pathway to the BAM complex and hence OMP biogenesis is complicated by a number of factors, including the peptidoglycan, which is part of the cell wall that forms a rigid mesh within the periplasm, separating the inner membrane and the outer membrane (Fig. 1c). The chaperone-stabilized nascent OMPs must pass through the peptidoglycan to reach the BAM complex within the outer membrane. Another barrier for OMP biogenesis is the absence of a known energy source, such as nucleotide hydrolysis in the periplasm or at the outer membrane²¹. Without a novel mechanism, the energy required to insert OMPs into the inner membrane or outer membrane would be too great to overcome in vivo. However, once nascent OMPs arrive at the outer membrane, it has been postulated that the BAM complex acts as a catalyst on the membrane itself to facilitate insertion of OMPs^22–24. No analogous catalyst has yet been identified within the inner membrane, which possibly explains why no OMPs have been found at that membrane.

The BAM complex consists of multiple components which can vary by species^{23, 25–28}. In E. coli, the BAM complex consists of five components: the essential core component BamA, which is an OMP itself, and four accessory lipoproteins termed BamB, BamC, BamD and BamE, each containing an N-terminal post-translational lipid modification that anchors them to the inner leaflet of the outer membrane^{9, 10, 29, 30} (Fig. 1e). BamB and BamD have been shown to interact directly with the periplasmic domain of BamA while BamC and BamE interact with BamD, leading to the isolation of primarily BamAB and BamCDE modules in vivo^{9, 10, 29, 31–35} (Fig. 1f). Both BamA and BamD are essential for cell viability; however, all components are required for efficient OMP biogenesis in vivo^{31–33, 36, 37}. In vitro studies have further demonstrated the vital role of BamB for optimal folding of the two OMPs OmpT and EspP, possibly mediating the delivery of substrate from SurA to the BAM complex, a process that is still not well understood^{36, 37}.

In this Progress article, we discuss recent structural studies that have advanced our understanding of the structure of the fully assembled BAM complex and the interactions between the individual components. We also review the mechanistic insights that have been gained from structural data, mutagenesis studies, molecular dynamics simulations, and explore two emerging models for BAM-mediated OMP biogenesis in bacteria.

The structure of the BAM complex

The structures of the Bam proteins

In the past decade, the structures of all of the periplasmic domains of the Bam accessory proteins have been reported^{29, 35} (Fig. 2a). BamB adopts an eight-bladed β-propeller fold containing WD40-like motifs, suggesting that it may play a role as a scaffold within the BAM complex^38–41. The structure of BamC contains an unstructured N-terminal domain followed by two helix-grip domains, each of which was solved individually^42–44. Interestingly, in vivo studies have shown the helix-grip domains of BamC to be surface-exposed in E. coli⁴⁵. Here, fluorescently labeled antibodies, which were raised against fully folded BamC, were used to specifically label BamC located outside of E. coli. While labeling was still observed when the C-terminal helix-grip domain was removed, no labeling was observed with a construct lacking both helix-grip domains. Exactly how surface exposure of BamC occurs or what role this may have within the BAM complex remains to be determined. BamD consists of five tetratricopeptide repeat (TPR) domains, and it has been suggested that BamD has a role in activating BamA and/or interacts with substrates^{43, 46, 47}. BamE contains an ααβββ fold and has been shown to enhance the interaction of BamD with BamA^{33, 43, 48, 49}. Although the structures of these Bam accessory components have been solved, they offered few clues about how the BAM complex interacts with nascent OMPs or how it mediates their biogenesis.

Figure 2. The structures of the β-barrel assembly machinery complex.

a. The individual structures of all the Bam components. BamA (green) contains an N-terminal periplasmic domain consisting of five polypeptide transport-associated (POTRA) repeat domains (P1 – P5) and a C-terminal 16-stranded β-barrel domain. BamB is an eight bladed β-propeller structure containing multiple WD40-like repeats and may serve as a scaffolding protein. BamC, which has been shown to be presented at the surface, contains multiple domains, including an unstructured N-terminal domain followed by two helix-grip domains. BamD contains five tetratricopeptide repeat (TPR) domains and forms the primary interaction with BamA. BamE contains an ααβββ globular fold and interacts with both BamA and BamD, helping to stabilize the intact complex. b,c. The structure of the BAM complex was recently reported with (PDB ID 5D0O and 5AYW) (b) and without BamB (PDB ID 5D0Q and 5EKQ) (c) by X-ray crystallography, revealing an unprecedented 45° shift of the first half of the barrel domain of BamA. The direction of the conformational changes observed within the POTRA domains are indicated. d. More recently, the structure of the BAM complex with BamB was reported using cryoEM (PDB ID 5LJO), revealing a similar conformation as observed in the X-ray crystal structure lacking BamB (part c). e. These structures suggest that the role of the Bam accessory proteins may be to regulate the conformation of BamA during OMP biogenesis. In all structures reported to date, two major conformations have now been observed for BamA. In the first, the barrel domain of BamA shows an ‘inward open’ conformation where the exit pore is occluded, with POTRA5 being located away from the barrel domain to enable access to the lumen from the periplasm. In the second, the barrel domain of BamA shows an ‘outward open’ conformation where the exit pore is now fully open, as POTRA5 shifts to occlude access to the barrel lumen from the periplasm. These conformational states were initially thought to be modulated by BamB, however, both have now been observed in the presence and absence of BamB. Exactly what role the conformational changes serve or how they are regulated remains unknown. The structure of BamA alone (PDB ID 4C4V and 4N75) has also been reported and was found in an ‘inward open’ state; however, BamA alone would presumably be rare in vivo.

The crystal structure of BamA offered the first hint to how the BAM complex may function in OMP biogenesis^{23, 24, 50–54}. BamA consists of an N-terminal periplasmic domain that contains five polypeptide transport associated (POTRA) domains and a C-terminal 16-stranded β-barrel domain^{20, 23, 35} (Fig. 2a). Shortly after the BAM complex was first discovered, it was shown that both BamB and BamD interact with BamA via its POTRA domains^{31, 32, 34}. Other studies suggested that the POTRA domains may also interact with the nascent OMP substrate prior to insertion into the outer membrane^{53, 54}. More recently, several reports revealed that the atomic structure of the β-barrel domain of BamA is novel and has unique properties that were not previously seen in other OMPs with known structure^{20, 23, 24, 35}. Whereas BamA contains sixteen β-strands as predicted, strands β1 and β16, which are located at the junction site that closes the barrel, were found to be substantially shorter, resulting in a drastically thinned hydrophobic belt (phenylalanine, tryptophan, and tyrosine residues found along the outside of the barrel domain that assist in maintaining buoyancy within the membrane)^{20, 23, 24, 51, 52}. Furthermore, in two of the reported crystal structures, strand β16 was found partially unzipped from β1, sitting in a bent conformation pointing toward the lumen of the barrel domain, leaving only a few hydrogen bonds at the junction^{23, 51, 52}. These studies offered the first evidence that the barrel domain of BamA contains a lateral gate into the membrane, which would create a direct portal for the insertion of nascent OMPs directly into the outer membrane during biogenesis^{23, 24}. Moreover, an exit pore was identified that has been proposed to mediate the formation of extracellular loops of nascent OMPs^{24, 52}.

The structure of the fully assembled complex

The first complex crystal structure of Bam components was that of BamD bound to a BamC construct that lacks the C-terminal helix-grip domain. The structure revealed that the unstructured N-terminal domain of BamC almost exclusively mediates the interaction with BamD⁵⁵. The next complex reported was that of a fusion between POTRA3, POTRA4 and POTRA5 of BamA with BamB^{56, 57}. This crystal structure further confirmed previous studies that indicated that BamB binds along POTRA3 in proximity to the linker with POTRA2. This interaction may have a role in regulating the conformational changes within the POTRA domains that were previously observed along this hinge point between POTRA2 and POTRA3^{50, 53, 54}. Later, the crystal structure of a fusion between POTRA4 and POTRA5 with BamD was reported, showing that TPR3 and TPR4 of BamD mediate the primary interactions with only POTRA5 of BamA, agreeing well with previous studies^{31, 58}. The exact role of BamD binding to BamA remains unclear; however, studies have suggested that BamD may activate BamA⁵⁹.

Four crystal structures of the assembled BAM complex were recently reported, providing the molecular details for how the full complex is assembled, the interactions between the individual Bam components, and associated conformational changes^60–62 (Fig. 2b, c). Two of the reported structures lack BamB, whereas two of the structures contain BamB, yet are the same structures with identical space groups and cell parameters^{61, 62}. More recently, a fifth structure of the BAM complex with BamB was reported using cryoEM⁶³. BamC was found interacting with BamD as observed previously, yet the helix-grip domains were found in various states and conformations along the periplasmic side of the complex with one or both disordered, indicating the lack of a stable interaction with other Bam components. While this seemingly conflicts with the previous in vivo studies indicating the helix-grip domains are surface exposed⁴⁵, the X-ray and cryoEM structures of the BAM complex were determined in non-native environments and in the absence of substrates, which may be required to observe the native topology of BamC. Proper controls were demonstrated in the in vivo studies, with a construct lacking both helix-grip domains showing no surface labeling, providing the most convincing evidence that BamC truly is surface exposed. And since no stable interaction of BamC with the surface of BamA has been described, it is not surprising that BamC was observed to be dynamic in the reported structures. Aside from BamD, where flexing of the molecule was observed within the complex, the individual structures of BamB, BamC and BamE remained mostly unchanged with only slight localized conformational changes. In all structures of the BAM complex, the binding interactions of BamC with BamD, BamD with BamA, and BamB with BamA closely mirror those observed in the previously described complex crystal structures^55–58. In addition, BamE was found to interact with TPR4 and TPR5 of BamD and unexpectedly also with POTRA5 of BamA, which may explain previous studies reporting that BamE enhances the interaction of BamD with BamA³³. Further, TPR1 and TPR2 of BamD were found to interact with POTRA1 and POTRA2 of BamA, which may stabilize a ring-like quaternary structure along the periplasmic portion of the BAM complex^60–62. Although this ring-like structure has been suggested to be important for OMP biogenesis, no mechanism has been postulated describing exactly how it contributes to the function of the BAM complex. While the recent structures show BamD interacting with POTRA1, POTRA2 and POTRA5, only the interaction with POTRA5 was consistently observed and supported by in vivo studies³¹, suggesting the other interactions may be artefacts.

Conformational cycling of BamA

The most important observations taken from the recent structures of the BAM complex are the drastic conformational changes observed in BamA^60–63 (Fig. 2b, c, d). In total four conformational states are observed for the POTRA domains. Two states were observed in the crystal structures that lack BamB, in which POTRA1-4 undergo ~30 Å shift from one another, whereas POTRA5 remains mostly unchanged, being located away from the barrel domain, thus enabling access to the lumen of the barrel domain^{60, 61} (Fig. 2c). The third conformational state was observed in the cryoEM structure which closely resembles one of the crystal structures that lack BamB; however only slight shifts were detected, primarily in POTRA1 and POTRA2⁶³ (Fig. 2d). The fourth conformational state was observed in the crystal structure with BamB, in which POTRA1-5 undergo ~40 Å shift with POTRA5 now being located directly under the barrel domain fully occluding access the barrel lumen^{61, 62} (Fig. 2b). The POTRA domains of BamA were previously shown to be dynamic, and it is expected that the BAM complex itself is also dynamic. Therefore, it remains to be determined whether the observed conformations of BamA are truly due to binding of the Bam accessory proteins or just an artifact of crystallization and/or in vitro sample preparation^{23, 50, 53, 54}.

Based on the X-ray crystal structures, two conformations were observed for the β-barrel domain of BamA, and the different states first seemed to be dependent on whether BamB was present or not^60–62. In the presence of BamB, BamA contains a classic barrel domain with a fully formed junction site (inward open conformation)^{61, 62} (Fig. 2b). By contrast, in the absence of BamB, an unprecedented conformational change was observed, where the first half of the barrel domain undergoes a ~45° rotation, starting with strand β1 and diminishing up to strand β9, with extracellular loops 1–3 fully opening along the previously identified exit pore (outward open conformation)^{60, 61} (Fig. 2c). However, in the recent cryoEM structure which contains BamB⁶³, the barrel domain of BamA was also found in an outward open conformation (Fig. 2d). Exactly why different conformations are observed between the X-ray crystal structure and the cryoEM structure of the fully assembled BAM complexes in the absence of BamB is not known; however, one possible explanation may be that different detergents were used during structure determination.

One consequence of the conformational change from an inward open to an outward open conformation of the barrel domain is that the junction gets completely disrupted with no hydrogen bonds being formed between strands β1 and β16^{60, 61}. Despite the dynamic conformational changes observed for BamA, the Bam accessory proteins accommodate these changes in BamA primarily as rigid bodies. With the existing structural data, it is tempting to speculate that the roles of BamB and BamCDE may be to regulate the cycling of BamA between ‘inward open’ and ‘outward open’ conformations (Fig. 2e). Crosslinking studies have provided further evidence for the observed structural changes^{24, 61}, which may explain previous observations from genetic and functional studies that indicated that conformational cycling of BamA during OMP biogenesis is regulated by Bam components ⁶⁴.

These conformational changes in BamA could possibly help drive the biogenesis of OMPs by directly threading nascent OMPs into the outer membrane and/or by further priming the membrane for OMPs insertion. However, additional experimental verification is certainly required to fully test this hypothesis.

Mechanistic insights into OMP biogenesis

Molecular dynamics simulations reveal clues towards mechanism

Molecular dynamics simulations have had a key role in unraveling the mechanisms of OMP biogenesis since the first BamA structures were solved^{23, 24}. Although these structures revealed a closed barrel, molecular dynamics simulations on the 1–2 μs time scale showed the junction between strands β1 and β16 of BamA to be dynamic^{23, 24}. The simulations also showed a distinct thinning of the membrane near the proposed lateral gate^{23, 24}. Notably, at the time the simulations used a simplified lipid membrane and/or were performed at an increased temperature, which leaves in question the biological relevance of the observed dynamics. However, biochemical experiments have confirmed the importance of lateral gating; disulfide cross-linking the gate at various points rendered BamA non-functional, leading to a loss of viability, whereas reducing the cross-link restored function^{24, 61}. Similarly, a hypothetical exit pore between the extracellular loops of BamA that was observed in simulations was confirmed to be relevant to BamA function in cross-linking studies^{24, 61}.

More recent molecular dynamics simulations of BamA have focused on the dynamics of the five POTRA domains of BamA⁶⁵. It was found that in addition to a high degree of flexibility between the domains, the POTRA domains can also bind to the membrane surface. Only certain membrane-bound conformations were observed to be compatible with BamB and/or BamD binding⁶⁵. Finally, simulations of the entire BAM complex, namely the 3.9-Å BamACDE structure (PDB 5D0Q) and the 2.9-Å BamABCDE structure (PDB 5D0O), have also been carried out⁶¹. These simulations showed that the distinct conformations of each complex are stable on the 100-ns time scale. Removal of some or all Bam accessory proteins from the complexes showed an increase in dynamics for the remaining components; for example, POTRA3 and POTRA4 move away from the membrane when BamB is removed⁶¹. Simulations of over 2.8 μs together with functional studies led to the conclusion that the POTRA domains of BamA (which are arranged in a ring-like architecture) and accessory proteins that are located below the BamA barrel shift between distinct conformations to modulate the conformation of the β-barrel, thereby alternating exposure to the periplasm or to the membrane⁶¹.

Mechanistic models for BAM-mediated biogenesis

Although several hypotheses have been proposed for how the BAM complex mediates OMP biogenesis, two popular mechanistic models are summarized here. Neither of the models has been conclusively demonstrated experimentally to apply to all OMPs and it may be that different types of OMPs interact with the BAM complex in a distinct way, depending on their complexity. The first model proposes that, based on their sequence, OMPs fold themselves and just need a locally disturbed membrane to organically fold into. In this model, the role of the BAM complex is to function as a catalyst by locally destabilizing the membrane bilayer and to traffic the nascent OMP into close proximity to the primed membrane for insertion into the outer membrane^{20, 22–24, 66} (Fig. 3a).

Figure 3. Mechanistic models for the role of the BAM complex in OMP biogenesis.

a. The first model proposes that β-barrel outer membrane proteins (OMPs) fold intrinsically and just need a locally disturbed membrane to organically fold into. In this model, nascent OMPs are first synthesized in the cytoplasm with an N-terminal leader sequence (removed by proteases at the inner membrane) which directs them to the Sec translocon for transport across in the inner membrane into the periplasm (step 1). Next, the periplasmic chaperones SurA or Skp (not shown) binds the nascent OMPs and delivers them to the BAM complex (step 2), which serves as a catalyst to locally destabilize the membrane bilayer as indicated) and to traffic the nascent OMP into close proximity to the primed membrane for insertion into the outer membrane (step 3). The current hypothesis is that trafficking to the BAM complex may be regulated by the β-signal of the nascent OMP. b. The second model proposes a more systematic mechanism whereby the SurA- and/or Skp-stabilized nascent OMP is delivered to the BAM complex, which then systematically folds it into the membrane. This model shares the initial two steps (step 1 and step 2) with the first model (part a), where nascent OMPs are imported into the periplasm and stabilized by chaperones such as SurA and Skp (not shown). Next, the nascent OMPs are delivered to components of the BAM complex (possibly BamB, BamD, and/or the POTRA domains of BamA) and folding and insertion are initiated by the β-signal of the nascent OMP (step 3)). Upon opening of the lateral gate of BamA, a β-hairpin has been hypothesized to bind the exposed N-terminal strand of BamA by β-augmentation. The barrel of the nascent OMP integrates directly into the barrel of BamA, thus satisfying the requirement that hydrophilic residues point inside the barrel, while hydrophobic residues point outside and mediate interaction with the membrane. The nascent OMP continues to grow either strand by strand or by a β-hairpin at a time (steps 4–6)). To prevent the formation of a super-pore in the outer membrane, the nascent OMP eventually ‘buds’ away from the barrel domain of BamA. We hypothesize that maturation occurs when the first strand of the nascent OMP comes into proximity of the junction site, whereby the last strand of the nascent OMP unpairs from BamA and then pairs with its own first strand (step 7)). This results in the termination of the folding process, forming an independent barrel which then diffuses into the outer membrane, while the BAM complex is left primed for another round of OMP folding.

The first model can be rationalized by the fact that some OMPs can readily and efficiently fold into perturbed membranes; however, this is not true for larger, more complex OMPs, including those that are multimeric. To address this, the second model proposes a more systematic mechanism whereby nascent OMPS that are stabilized by SurA and/or Skp are delivered to BamA and/or other Bam proteins and then systematically threaded into the membrane, possibly strand by strand or by one β-hairpin at a time^{29, 67, 68} (Fig. 3b). To satisfy the amphipathic properties of the new strands, each is formed by β-augmentation, a process in which existing β-strands serve as templates for the formation of new strands, using the exposed edges of the BamA barrel following lateral opening or following the large conformational change observed in the most recent structures of the assembled BAM complex lacking BamB. Strands would be added and an enlarged barrel is formed, which would expand only to a certain point until it begins to bud away from the BamA barrel, forming its own barrel within the membrane. It has been hypothesized that the C-terminal strand (also referred to as the β-signal) of the nascent OMPs would interact with the first strand of the BamA barrel as an initiating step for biogenesis; however, this notion remains to be shown experimentally. While the lateral gate formed by BamA enables the formation of strands directly into the membrane, the exit pore would mediate the formation of extracellular loops, some of which can be as large as 100 residues or more⁴. Barrel growth of the nascent OMP would continue until the termination step, which is triggered when the first strand of the new OMP forms and comes into close proximity with the transient junction. We hypothesize that the C-terminal strand of the new OMP has an inherently higher affinity for its native first strand than it does for that of BamA. Therefore, strand exchange occurs, which completes the biogenesis of the new OMP where it can freely diffuse into the outer membrane. This concept can be rationalized by the fact that the junctions of most OMPs share many more hydrogen bonds than the five or six that the first strand of BamA can offer.

Outlook

In Gram-negative bacteria, it is generally accepted that the BAM complex is responsible for the biogenesis of most, if not all, OMPs. Given the amphipathic nature of the β-strand building blocks, the diversity in the number of strands, and the fact that there is no energy source such as nucleotide hydrolysis at the outer membrane, the mechanism for the biogenesis of OMPs is arguably more complex than that for α-helical membrane proteins. Collectively, studies over the past decade have greatly advanced our understanding of the journey OMPs must take to reach their final destination within the outer membrane. Furthermore, the recently solved structures of the BAM complex, which reveal unprecedented conformational changes in the β-barrel domain of BamA, cycling between ‘inward open’ and ‘outward open’, it is clear that the gap between what is known about the biogenesis of α-helical versus β-barrel membrane proteins is rapidly closing.

It was shown that the BAM complex can also partner with the translocation and assembly module (TAM) complex to mediate the biogenesis of some autotransporters and the usher protein FimD^69–72. More recently, in Borrelia burgdorferi, the TamB component of TAM was shown to interact directly with BamA and to be an essential component of the BAM complex⁷³. Exactly how the BAM and TAM systems communicate with one another, how they decide their substrates, and what mechanistic role each may have in OMP biogenesis remains to be determined.

With the new structures of the BAM complex hot off the press^60–63, we now have the molecular blueprints that are necessary to design experiments to precisely resolve the molecular and mechanistic details responsible for the biogenesis of OMPs. Moving forward, there are many unanswered questions that require further investigation. For example, what role does BamC have and does its proposed presentation at the cell surface contribute to OMP biogenesis? If so, how? Where does the energy come from that drives the BAM complex? Could the destabilized membrane and folding energy be all that are required? Do nascent OMPs really interact directly with the POTRA domains of BamA and if so, how? This idea has been around for many years but without any convincing evidence. Furthermore, if nascent OMPs do in fact interact directly with BamA or even another Bam protein, is the N-terminus or the C-terminus recognized first? How important are the sequences of the first and last strands of nascent OMPs? It can be rationalized that they must be complementary and this may even dictate the number of strands in an OMP. With the recent structures of the BAM complex, it is tempting to propose that the role of the accessory proteins is to regulate the conformational cycling of BamA; however, additional experiments are certainly required to support this hypothesis. Still, if true, this would align with the fact that BamA is the only evolutionarily conserved component of the BAM complex present also in mitochondria and in plastids such as apicoplasts and chloroplasts. Finally, the mechanism for how the BAM complex mediates OMP biogenesis has been a topic of much debate. Therefore, which proposed mechanism is correct, if any? Are there other, more plausible mechanisms that have been overlooked? Could the BAM complex mediate unique folding pathways tailored to the substrate OMP itself? With the structure of the fully assembled BAM complex now known, it is anticipated that many of these missing pieces of the mechanistic puzzle will soon be revealed.

Biographies

Nicholas Noinaj is an Assistant Professor in the Department of Biological Sciences at Purdue University. He received dual B.A. degrees in chemistry and mathematics at Berea College (Berea, Kentucky) and his Ph.D. in biochemistry at the University of Kentucky (Lexington, Kentucky). He did his postdoc within the Intramural Research Program at NIH/NIDDK. He joined the structural biology group at Purdue University in 2014 and his research focuses on structure determination of membrane protein folding machineries in pathogenic bacteria and chloroplasts.

James C. Gumbart is an Assistant Professor of Physics at the Georgia Institute of Technology (Atlanta, Georgia, USA). He received his B.S. from Western Illinois University (Macomb, Illinois, USA) and his Ph.D. from the University of Illinois (Urbana, Illinois, USA). His lab’s research is focused on the structure, mechanics, and function of the bacterial cell envelope.

Susan Buchanan is Chief of the Section on Structural Biology of Membrane Proteins, in the National Institute for Diabetes & Digestive & Kidney Diseases, at the National Institutes of Health. She joined the NIDDK as a Tenure Track Investigator in 2001 and is currently a Senior Investigator and Chief of the Laboratory of Molecular Biology, NIDDK. Research in Dr. Buchanan’s laboratory focuses on structure determination of integral membrane proteins using X-ray crystallography and functional analysis of these proteins using biophysical, biochemical, and cell biological techniques.