The Big BAM Theory: An Open and Closed Case?

Abstract

The β-barrel assembly machinery (BAM) is responsible for the biogenesis of outer membrane proteins (OMPs) into the outer membranes of Gram-negative bacteria. These OMPs have a membrane-embedded domain consisting of a β-barrel fold which can vary from 8–36 β-strands, with each serving a diverse role in the cell such as nutrient uptake and virulence. BAM was first identified nearly two decades ago, but only recently has the molecular structure of the full complex been reported. Together with many years of functional characterization, we have a significantly clearer depiction of BAM’s structure, the intra-complex interactions, conformational changes that BAM may undergo during OMP biogenesis, and the role chaperones may play. But still, despite advances over the past two decades, the mechanism for BAM-mediated OMP biogenesis remains elusive. Over the years, several theories have been proposed that have varying degrees of support from the literature, but none has of yet been conclusive enough to be widely accepted as the sole mechanism. We will present a brief history of BAM, the recent work on the structures of BAM, and a critical analysis of the current theories for how it may function.

Brief introduction to BAM

The outer membrane of a Gram-negative bacterium is a natural barrier that protects from the often-harsh extracellular milieu (1–3). It is composed of an asymmetric bilayer with phospholipids in the inner leaflet and lipopolysaccharide (LPS) in the outer leaflet, and is host to a class of integral membrane proteins that are often referred as outer membrane proteins (OMPs) (4–6). These OMPs share a common membrane-embedded β-barrel domain which anchors them into the outer membrane (7, 8). Examples include OmpA, OmpF, OmpT, FepA, and PhoE which serve various roles for the cell including adhesion, signaling, and nutrient import (7–11). The biogenesis of these OMPs is mediated by a multi-component complex called the β-barrel assembly machinery (BAM) (4, 12, 13). As depicted in Figure 1, OMPs are first synthesized in the cytoplasm and then transported into the periplasm through the Sec translocation machinery. With the help of chaperones such as SurA, the OMPs are further escorted to the outer membrane where BAM mediates folding and insertion (4, 6, 14). For simplicity, for this review, we will focus on monomeric OMPs, where one polypeptide forms a single barrel domain. A recent review also discusses the biogenesis of OMPs with added focus on trimeric autotransporter adhesins (TAAs), where three polypeptides are required to form a signal barrel domain (15). BAM has been shown to play a critical role in folding TAAs, however, the exact mechanism here has also remained elusive (16–19).

Figure 1. Biogenesis of β-barrel outer membrane proteins.

β-barrel outer membrane proteins (OMPs) are synthesized in the cytoplasm and routed into the periplasm by the Sec pathway. Chaperones then further escort the OMPs to the β-barrel assembly machinery (BAM) for biogenesis into the outer membrane.

In E. coli, BAM consists of five components called BamA – E, yet the composition of BAM can vary from bacteria to bacteria (i.e. Neisseria lack a BamB ortholog) (20–27). BamA is an OMP itself while BamB-E are lipoproteins anchored into the periplasmic leaflet of the outer membrane. BamA (88 kDa) is a member of the Omp85 superfamily and plays a central role within BAM as both the scaffold which assembles the complex and as the workhorse which orchestrates OMP biogenesis (20, 28–31). BamA and BamD (26 kDa) are the core components of the complex that are essential for cell survival (20, 32). And while BamB (40 kDa), BamC (34 kDa), and BamE (11 kDa) are thought to play accessory roles in BAM, cells that lack these components are comprised and would not likely survive the harsh host conditions during an infection (33–36). Despite early reports that multiple BAMs may constitute the functional biological unit, it has been generally accepted that a single BAM is the active biological unit consisting of a 1:1:1:1:1 ratio of each component having a size of 200 kDa (12, 37–40). Recently though, it has been shown that multiple BAMs co-localize within the outer membrane in precincts which may be mediated by inter-complex BamB-BamB interactions, however, more work is needed to verify the role of BamB here (41).

Aside from OMP biogenesis, BAM also plays a role in the bacterial contact-dependent growth inhibition (CDI) system, a CdiA/CdiB family of two-partner secretion proteins, which suppresses the growth of neighboring target cells under nutrient-limited conditions (42, 43). Several studies have indicated that BamA is the receptor for CdiA while the accessory proteins BamB-E do not appear to be required. BAM’s role in OMP biogenesis does not appear to be necessary for its role in CDI, as the presence of only BamA seems sufficient and studies have localized the CdiA^EC93 binding site to extracellular loops L6 and L7 within the β-barrel domain of BamA (43).

BAM as a promising therapeutic target

Multi-drug resistant (MDR) bacteria have quickly become a major concern in the United States and worldwide (44–46). According to a report by the CDC published in 2013, each year about 2 million people in the United States get an antibiotic-resistant infection with 23,000 eventually dying from the infection (44, 47). Antibiotic resistant bacteria are able to survive by undergoing genetic changes in which they become resistant to antibiotics used to treat them or they develop new mechanisms for evading the host’s immune defenses (44, 48, 49). Several studies have identified BAM and its components as promising targets for the development of new antibiotics and antimicrobial therapies against such threats (50–52). This is due to BAM’s essential role in the insertion of OMPs in the outer membrane and its localization at the cell surface. Essential cellular processes such as nutrient acquisition and host cell adhesion make OMPs critical for infection and pathogenesis, and targeting the function of BAM would down-regulate the production of these virulence factors, which would then limit a pathogen’s ability to combat the host’s immune responses.

In 2005, a study found that BamB plays a role in the invasive ability of E. coli strain LF82 (53). Mutants with the BamB gene deleted showed a 95% decrease in their invasive ability compared to the wild type strain. Additional studies performed in Klebsiella pneumoniae, Yersinia enterocolitica and Salmonella enterica revealed that deletion of the BamB gene resulted in changes in OMP levels in the outer membrane (34, 54), a decrease in virulence, and an increase in sensitivity to some antibiotics (34, 35, 54). Other studies focused on BamA found similar results. A study done with Acinetobacter baumannii found that mice immunized against BamA had a 60–80% survival rate when subsequently infected with a lethal dose of A. baumannii (55). A subsequent study found that the addition of MAB1, an antibody antagonistic to BamA, to a culture of E. coli disrupted outer membrane integrity, inhibited protein folding, and inhibited bacterial growth (52).

Overall, these studies point to BAM and its components as a promising target for the development of new antibiotics against pathogenic Gram-negative bacteria, particularly for those that have developed multi-drug resistance. BAM represents a novel target found on the surface of the bacteria, thereby circumventing the need to design compounds which must permeate the outer and inner membranes to reach their targets.

The BAM structures

Over the past decade, all the structures of the individual domains of the Bam proteins have been reported from various bacteria (Figure 2), and more recently, structures of fully assembled BAM have been reported (56–77) (Tables 1 and 2). The first structures reported were of the polypeptide transport-associated (POTRA) domains of BamA, which revealed an elongated repeating structure which was thought to serve as the scaffold for the accessory proteins (Figure 2A) (60, 67, 75, 78). The POTRA domains have been observed in various conformations over the years and from SAXS and other experiments, it was concluded that the POTRAs are dynamic with varying regions of high/low flexibility; what role the flexibility of the POTRA domains may serve within BAM remains to be determined. The full length structure of BamA, along with several truncated versions, were later published revealing that BamA contained a membrane-embedded 16-stranded β-barrel domain (62, 79, 80). The barrel domain was found to have unique properties that were important for its role within BAM. The first was that the aromatic belt was thinned along the seam that closes the barrel domain, which was later shown to lead to a significant thinning of the local membrane accompanied by substantial destabilization of the lipid bilayer. And the second was that the seam between strands β1 and β16 undergoes a lateral separation that is required for BAM function (79, 81). Together, these observations offered clues that BamA was conformationally dynamic, particularly along the seam, and may be acting as a catalyst to destabilize the local membrane for OMP biogenesis, ideas later supported by other reports (82).

Figure 2. Structures of the Bam proteins.

A. BamA structures that include the β-barrel domain (top) and those that include just portions of the POTRA domains (bottom). B. All full length BamB structures. C. All BamC structures, which are fragments rather than the full structure. D. All full length BamD structures. E. All full length BamE structures; 2YH9 was found as a domain swap dimer, only one monomer is shown. F. Binary complexes of fragments of Bam proteins. See Table 1 for more information.

Table 1.

Summary of Bam protein structures.

Protein name	Organism	Method	Resolution (Å)	PDB ID
Individual proteins and domains
BamA (barrel + P5)	E. coli	X-ray	3	4C4V
BamA (β-barrel)	E. coli	X-ray	2.60	4N75
BamA (β-barrel)	E. coli	X-ray	2.6	6FSU
BamA (full length)	N. gonorrhoeae	X-ray	3.2	4K3B
BamA (barrel + P45)	H. ducreyi	X-ray	2.91	4K3C
BamA (β-barrel)	S. typhimurium	X-ray	2.92	5OR1
BamA (P1234)	E. coli	X-ray	2.7	2QCZ
BamA (P1234)	E. coli	X-ray	2.2	2QDF
BamA (P1234)	E. coli	X-ray	3.3	3EFC
BamA (P45)	E. coli	X-ray	2.69	3OG5
BamA (P45)	E. coli	X-ray	1.5	3Q6B
BamA (P12)	E. coli	NMR		2V9H
BamB	E. coli	X-ray	1.65	3Q7M
BamB	E. coli	X-ray	1.77	3Q7N
BamB	E. coli	X-ray	2.09	3Q7O
BamB	E. coli	X-ray	2.6	2YH3
BamB	E. coli	X-ray	2.00	3Q54
BamB	E. coli	X-ray	1.8	3PRW
BamB	E. coli	X-ray	2.6	3P1L
BamB	P. aeruginosa	X-ray	1.85	4HDJ
BamB	M. catarrhalis	X-ray	2.33	4IMM
BamC N-terminal	E. coli	X-ray	1.55	2YH6
BamC C-terminal	E. coli	X-ray	1.5	3SNS
BamC C-terminal	E. coli	NMR		2LAE
BamC N-terminal	E. coli	NMR		2LAF
BamD	E. coli	X-ray	1.8	2YHC
BamD	E. coli	X-ray	2.60	3Q5M
BamD	N. gonorrhoeae	X-ray	2.50	5WAQ
BamD	R. marinus	X-ray	2.15	3QKY
BamE	E. coli	X-ray	1.8	2YH9
BamE	E. coli	NMR		2KM7
BamE	E. coli	NMR		2KXX
BamE	N. gonorrhoeae	X-ray	2.45	5WAM
Binary complexes
BamA(P3-5)B	E. coli	X-ray	2.15	4XGA
BamA(POTRA3-5)B*	E. coli	X-ray	3.1	4PK1
BamA(POTRA45)D*	R. marinus	X-ray	2	5EFR
BamCD	E. coli	X-ray	2.9	3TGO

^*fusion constructs

Table 2.

Summary of BAM structures.

PDB ID	Method; Resolution	Bam proteins	Detergent(s)	BamC region	BamA POTRA conformation	BamA β-barrel conformation
5D0Q	X-ray; 3.5 Å	ACDE	C8E4/β-NG	N-term, HG1, HG2	Closed	Outward-open
5D0O*	X-ray; 2.9 Å	ABCDE	OG/LDAO	N-term	Open	Inward-open
5AYW*	X-ray; 3.6 Å	ABCDE	C8E4/LDAO/OG	N-term	Open	Inward-open
5EKQ	X-ray; 3.4 Å	ACDE	C8E4	N-term, HG1	Closed	Outward-open
5LJO	cryo-EM; 4.9 Å	ABCDE	DDM	N-term, HG1†	Closed	Outward-open

^*these structures are the same, having the same space group, cell parameters, conformations, and an overall RMSD of 0.46 Å.

^†observable only when density map is visualized at lower contour levels

BamB was determined to be have an eight-bladed β-propeller fold that shares structural homology with eukaryotic proteins containing WD40 repeat domains which often serve as scaffolds for larger protein complexes (Figure 2B) (63, 64). Therefore, BamB has been proposed to serve as a scaffold protein for optimally orienting the flexible periplasmic domain of BamA for interaction with other BAM components and/or substrates and for assisting in the handoff of substrates from chaperones to BAM (61, 63, 64, 83, 84). Structures of BamB bound to fragments of the POTRA domains of BamA show BamB interacts primarily along the hinge region between POTRA2 and POTRA3, which would indeed sterically restrict the conformations of the periplasmic domain of BamA (61, 77). Another scaffolding role for BamB was also recently reported for BAM (41, 85). Here, BamB was shown to be important for the formation of BamA-BamB inter-complex interactions, thereby leading to the formation of organized assembly precincts within the outer membrane. The molecular details of these interactions and why BAM may need to form these localized precincts remain unknown.

The structure of BamC has been solved in fragments with the N-terminal flexible domain only being observed while in complex with BamD, while the structures of the two C-terminal helix-grip domains have been determined individually (Figure 2C) (66, 68–70). While the N-terminal domain of BamC interacts directly with BamD, no stable interactions between BamC’s helix-grip domains with BamD (or any other Bam protein) have been reported. Therefore, the role of BamC alone or the role of BamC’s interaction with BamD remains unknown. Further, the topology of BamC remains a topic of controversy, as studies have shown portions of it to be surface exposed rather than being found solely within the periplasm; this will be addressed more later in this section as the full BAM structures are presented (41, 85, 86). Whether this observation is important for BamC’s role within BAM is not known.

BamD binds directly to POTRA5 of BamA and the lack of BamD or BamA is lethal (20, 68). The structure of BamD alone, which contains five tetratricopeptide repeat (TPR) domains, and structures in complex with POTRA5 have been reported (Figure 2D) (66, 71, 72, 87). Studies have reported that BamD recognizes unfolded OMP substrates specifically along a recognition sequence termed a ‘β-signal’, however, these studies have been focused on one or few substrates and will need to be studied more comprehensively to determine if such as a mechanism is true for all BAM substrates (88, 89). BamD has been suggested to activate BamA, yet exactly how remains unknown. In these studies, a suppressor mutant of BamD (R176L) was found that reversed the lethal phenotype of the E373K mutation in BamA. In this strain of E. coli, OMP biogenesis was fully restored and the phenotype was indistinguishable from wild type, however, there was no observable interaction between BamD_R176L and BamA_E373K; therefore, the nature of this interaction remains unknown (90). A similar finding was reported in studies of LptD/E biogenesis, but again, these studies were only suggestive and not conclusive of a BamA-activation mechanism by BamD (88).

BamE, which has an ααβββ fold, interacts directly with BamD and enhances the association of BamD with BamA, which is accomplished by bridging additional interactions as observed in the BAM structures discussed later (Figure 2E) (32, 56–58). Similar to BamC, though nonessential for viability, BamE may be required for certain substrates or other unidentified roles within BAM (91). For example, deletion of BamE abolishes the assembly of OMP/RcsF complexes, yet the role of BamE here remains unknown (92, 93).

The fully assembled structure of BAM has recently been reported by X-ray crystallography and cryo-EM (Figure 3) (56–59) (Table 2). BamB-E were all found interacting with BamA along the POTRA domains and base of the barrel domain, with a trimeric BamCDE complex interacting primarily with POTRA5 and portions of POTRA4, along with periplasmic turns of the barrel domain. BamB, however, was found interacting primarily with POTRA3 along the POTRA2/3 hinge. Two structures of BAM include BamB (PDB IDs 5D0O/5AYW and 5LJO) while two of the structures lack the BamB component (PDB IDs 5D0Q and 5EKQ) (12, 94). The conformational variations between the BAM structures is discussed in more detail in the following section.

Figure 3. Structures of assembled BAM.

A. The structures of BAM found in the outward-open state, including various different conformational of the POTRA domains and both in the presence and absence of BamB. B. The structures of BAM found in the inward-open state, in the presence of BamB. See Table 2 for more information.

Open and closed conformations of BAM

A comparison of the recently reported structures of BAM reveals a number of intriguing observations, which may offer clues to how BAM functions. These include (1) the conformation of the POTRA domains of BamA and (2) the conformation of the barrel domain of BamA. Initially, the X-ray crystal structures of BAM appeared to suggest that BamB may regulate the transition of these changes, however, the cryo-EM structure refuted this idea, leaving us the task of trying to decipher exactly how the observed conformational changes in BAM help drive OMP biogenesis (Figure 3). One caution here is that all structures of BAM to date have been reported in the presence of different detergents rather than in a lipid bilayer or in the presence of substrates (Table 2). Therefore, the field must be cognizant that one or more of the conformations may be artifacts; more work is needed to verify the true conformational states of BAM in a lipid bilayer and what role these conformational changes may play in the biogenesis of OMPs. To address this, NMR has recently been used to investigate BAM in proteolipsomes (95–100). Results from solid-state studies suggest that the barrel domain of BamA and POTRA5 movements are attributed to complex formation in membranes and that the lateral gate and POTRA5 are locally dynamic. While these results are consistent with other observations, more work is needed to decipher exactly how the movements and dynamics along the lateral gate are promoting OMP biogenesis.

In the BAM structures, the POTRA domain of BamA was found in two states which we will refer to as POTRA_CLOSED and POTRA_OPEN (Figure 4 and Table 2). In the POTRA_CLOSED state, the POTRA domain is found along the base of the barrel domain fully occluding barrel access from the periplasm. In the POTRA_OPEN state, the POTRA domain is found swung away from the barrel domain allowing full access to the lumen of the barrel from the periplasm. Crosslinking the POTRA domains in the closed state was shown to disrupt the function of BAM suggesting that cycling of the POTRA domains between open and closed states was necessary for function (57). However, the role of the conformational cycling of the POTRA domains has not been determined, although it has been suggested to possibly help drive OMP insertion into the outer membrane. Further, the spiral ring-shape of the POTRA domains in the BAM structures has been loosely suggested to somehow correlate with the fact that BAM’s substrates are barrel-shaped, however, this seems unlikely given that substrates range significantly in size and strand numbers and that there is no evidence to support such a mechanism (57). In addition, recent molecular dynamics (MD) simulations indicate that the POTRA domain of BamA interacts with the inner leaflet of the outer membrane, which may further restrict the conformational freedom required for POTRA cycling from open to closed and vice versa (101).

Figure 4. POTRA domain conformations in BamA observed within the BAM structures.

A. Top-down view of the POTRA_OPEN state of BAM. BamA is shown in green, BamB in gray, BamC in cyan, BamD in magenta and BamE in blue. B. Top-down view of the POTRA_CLOSED state of BAM. Here, BamA is shown in dark gray. In panels A and B, the red dashed oval indicates the location and orientation of the barrel for reference. C. A top-down view of a structural comparison of the POTRA changes in BamA from the ‘open’ to ‘closed’ states. D. A side view depicting the motion and direction of the POTRA domains; only POTRA5 is shown for clarity. The red dashed arrows indicate the motion and direction of the conformational changes.

The other important conformational change observed was in the barrel domain of BamA, which was found in ‘inward-open’ and ‘outward-open’ states (Figure 5); this terminology is preferred over ‘lateral-open’ which can easily be confused with the unpaired C-terminal strand (unzipped or tucked) state. In the inward-open state, the base of the barrel domain is fully open to the periplasm, similar to what has been observed in nearly all other OMPs, except with a destabilized C-terminal strand which is found tucked inside the barrel domain (102, 103). In the outward-open state, the first half of the barrel domain undergoes a conformational twist where the periplasmic side constricts to close the base of the barrel by ~10 Å reducing the diameter of the barrel here by 25%, while the top of the barrel undergoes a significant opening by ~12 Å (Figures 5A and 5B). The outward-open conformation of the barrel domain has been observed in the presence and absence of BamB, countering the idea that this accessory protein is solely responsible for regulating the conformational state of the barrel domain of BamA (12). Therefore, it is still unclear what role the observed conformational states of BAM may serve in OMP biogenesis, or which are even important in vivo. While still preliminary, it seems clear that the conformation of the POTRA and barrel domains are coordinated such that when the POTRA domain is closed, the barrel is in the outward-open state, and when the POTRA domain is open, the barrel is in the inward-open state. Given the steric strain on the barrel caused from the closed state of the POTRA domain, it would be unfavorable for BAM to be found with a closed POTRA domain with the barrel in the inward-open state. If BAM has a preferred ‘active’ state or if cycling between the states truly is required remains to be determined, along with what role these conformational changes may be serving in OMP biogenesis.

Figure 5. Barrel domain conformations in BamA observed within the BAM structures.

A. Inward-open (gray) and outward-open (green) conformations were observed within the BAM structures, where the first eight strands of the barrel domain undergo a twisting conformation as depicted. The inset shows a zoomed view of the change, with the red arrow indicating the motion of the barrel from inward-open to outward-open. B. The barrel of BamA is shown looking up the periplasmic face. Here, the base of the barrel domain undergoes ~10 Å reduction in diameter to the outward-open state. C. A surface representation of the inward-open state in gray. D. A surface representation of the outward-open state in green.

With the structures of BAM available, the topology of BamC becomes even more confusing, as three of the structures showed at least the N-terminal domain and the first helix-grip domain interacting with BamD, with one of the structures (PDB ID 5D0Q) also including the second helix-grip domain (57). While seemingly contradictory to the studies showing surface-exposure (41, 85, 86), it is likely that the detergent-based solubilization and extraction method disrupts the native topology of BamC which cannot be reconstituted in detergents alone. Additional support for surface-exposure of BamC, and possibly other accessory proteins, was recently reported from work in Neisseria, where BamE was found to be surface-exposed, along with BamD when BamE was knocked out (87). It remains to be determined if these observations are important for the function of BAM.

Mechanisms for the function of BAM

Two major mechanisms or ‘theories’ for BAM-mediated OMP biogenesis have been proposed over the past two decades, each with varying degrees of literature support. We have reviewed these relatively recently and therefore will only briefly summarize here (12). In the first, which we refer to as the BamA-assisted mechanism, BAM serves as a trafficking complex to recruit chaperone-stabilized OMP substrates to the outer membrane where they undergo insertion, catalyzed in part by the locally destabilized membrane created by the barrel domain of BamA (Figure 6A). This mechanism is supported by a vast number of in vitro studies showing that some OMPs can be refolded efficiently into a lipid bilayer without the need of a separate insertase and by the fact that refolding efficiencies are increased when a bilayer is artificially thinned or perturbed (104–106). It has also been shown that BamA alone is able to catalyze OMP folding in an isolated system, yet exactly how this is facilitated wasn’t determined (82). The idea that the presence of a static BAM/BamA alone is necessary and sufficient for mediating insertion doesn’t align with studies definitively demonstrating that the barrel of BamA must undergo conformational changes for BAM to be active, particularly at the lateral seam. While recently reported in vitro experiments using a BamA lateral seam-crosslinked mutant still retains significant activity, this set of experiments did not report the efficiency of crosslink formation which could have been used to estimate residual activity from the non-crosslinked species (59). Even so, the activity of BAM increases significantly when the crosslinked species is chemically reduced, further indicating that a static BAM/BamA is likely insufficient to handle the task of folding cellular quantities of OMPs in vivo. Another study with tOmpA came to a similar conclusion indicating that conformational cycling of BamA may not be required for folding smaller OMPs (107). A variation of the BamA-assisted mechanism is that the substrate OMPs are partially or fully pre-formed within the periplasm, possibly mediated and stabilized by chaperones and/or other Bam proteins, prior to insertion into the membrane (Figure 2B) (6, 108–110).

Figure 6. Two leading theories for the role of BAM in the biogenesis of OMPs.

A. In the BamA-assisted model, BAM serves as a trafficking complex to localize OMPs at the outer membrane. The OMPs are then inserted directly into the membrane assisted by the local destabilization of the membrane by BamA. B. A variation of the BamA-assisted model is depicted where the nascent OMPs are partially or fully pre-folded prior to insertion by BAM. C. In the BamA-budding model, BAM serves to recruit the OMPs to the outer membrane and then the exposed edges of BamA’s barrel serve as a template for catalyzing the formation of a hybrid BamA:OMP barrel via β-augmentation. Each strand added nucleates formation of the next strand until strand exchange occurs within the new OMP, thereby completing the biogenesis process. To prevent the formation of a super-pore within the membrane, the new OMP is thought to ‘bud’ or bleb away from the BamA barrel. D. A recently proposed variation of the budding model, the ‘swing’ model proposes that the nascent OMPs are folded along the interface of the membrane and must undergo a BamA-mediated swinging motion to put them into the membrane.

In the second mechanism, which we refer to as the BamA-budding mechanism, chaperone-stabilized OMP substrates are delivered to BAM for systematic insertion, possibly initiated by the β-signal of the new OMP (Figure 6C) (12, 40, 111, 112). Rather than concerted insertion of the full barrel domain, the exposed edge of the first strand of the BamA barrel serves as a template for strand formation of the new OMP via strand templating, thereby forming a BamA:OMP hybrid barrel intermediate. Each added strand then nucleates formation of the next strand until folding of the new OMP is terminated when its β-signal undergoes dissociation from the first strand of the BamA barrel and binds with higher affinity and specificity to its own first strand during a process termed strand exchange. To prevent the formation of a super-pore in the outer membrane, the new OMP would undergo a ‘budding’ or blebbing away from the core barrel domain of BamA, such that the amphipathic strands of the new OMP form in the correct orientation with hydrophobic residues towards the outside of the barrel, mediating interactions with the membrane, and polar residues towards the inside. Support for this theory comes from several sources, including the observation that the barrel of BamA undergoes lateral opening and large conformational switches to outward-open and inward-open states. Preventing the lateral opening at the seam using crosslinking halts BAM’s function, suggesting that the role of BamA is more than just serving to locally destabilize the membrane (59, 81). Further, the β-barrel domain of all OMPs is organized as a linear arrangement of anti-parallel strands, which suggests a systematic folding process, rather than a stochastic one (6–8, 12). And recently, crosslinking studies with Sam50, a conserved ortholog of BamA in mitochondria, were consistent with the idea that a precursor protein enters the barrel domain of Sam50 and then gets inserted into the membrane through interaction with the lateral gate, in a systematic step-wise manner (113). Despite being proposed nearly a decade ago, even before the first structure of BamA was solved, this study represents the first direct observation in support of the budding mechanism. Recently, another study performed a similar study in bacteria demonstrated they could crosslink the barrel domain of EspP to the barrel of BamA, similar to what was shown for Sam50 (114). In these studies, a new ‘swing’ mechanism was proposed which aligns with the idea of barrel integration, however, differs in the proposed interactions with BamA and how the OMPs are inserted into the membrane; via a swinging motion mediated by BamA (Figure 6D). Still, more work is needed to verify if one or more of these mechanisms is truly used by bacteria, and to be conclusive, it should be demonstrated for a library of OMPs rather than for only one or a few to be accepted as a general mechanism for BAM-mediated OMP biogenesis.

Variations of these mechanisms have also been proposed. For example, studies have proposed that the N-terminus of nascent OMPs may interact with BAM immediately upon exiting the Sec translocon during translocation across the inner membrane, forming a super-complex spanning the periplasm (115). While attractive since it would be efficient and likely require a reduced role for chaperones, it opposes the concepts that the β-signal, which is typically at the C-terminus, of the new OMP initiates interaction with BAM (in both mechanisms) and that the β-signal may be the first strand to integrate into the barrel of BamA (in the BamA-budding mechanism). While still not well understood, studies suggest that the β-signals confer some species specificity and may directly modulate the conformational state of BamA itself (40, 105, 111, 112, 116–118). Therefore, this idea of a super-complex would also increase the risk of inserting an incompletely translated nascent OMP into the outer membrane, which could lead to compromised membrane integrity rendering the cell susceptible to environmental competition and host defenses. One advantage of having the β-signal initiate biogenesis is that it can also serve as a checkpoint to confirm that a fully translated nascent OMP will be inserted into the membrane, thereby ensuring membrane integrity is maintained.

Summary and Future Outlook

The simplest transmembrane domain of any membrane protein found in nature is a single α-helix which is composed entirely of hydrophobic residues. For decades, in vitro studies have been performed to show how efficiently a single transmembrane α-helix can be spontaneously inserted into an artificial lipid bilayer and how thinning the bilayer or perturbing it can further increase insertion efficiencies (119–121). Yet, in nature, rarely do spontaneous events occur in the cell, likely because these events are very challenging to regulate. And for something as critical as the biogenesis of a transmembrane protein, spontaneous insertion doesn’t occur in the cell, not even for a single transmembrane helix (121). N-terminal single transmembrane helix proteins are inserted during co-translation by complex machineries such as YidC and Sec, while C-terminal single transmembrane helix proteins (tail anchored) are inserted by a complex cascade of machineries via the GET pathway (122–125). Arguably more complex than a single transmembrane helix, β-barrel outer membrane proteins are composed of 8–36 amphipathic strands which must be inserted into the outer membrane in a systematic linear arrangement such that the hydrophobic residues are oriented towards the membrane (7, 8, 12). While the amphipathic nature of a nascent OMP may afford it a higher propensity to accommodate aqueous solution than a single transmembrane helix, there are added complexities including how to stabilize the large hydrophobic surface area prior to insertion and the amount of space and volume needed. This is complicated further when considering the biogenesis of trimeric autotransporter adhesins where three monomers make up the final barrel domain (15). Several theories for BAM-mediated OMP biogenesis are presented, however, we do not favor a spontaneous insertion mechanism for reasons noted previously, as evidence favors a more regulated and systematic mechanism. While it is possible that different OMPs may undergo different mechanisms during biogenesis, this isn’t observed for helical membrane proteins with varying numbers of transmembrane helices, although it is true that there are a host of accessory proteins which can assist Sec during biogenesis; a topic of debate and active research still within the helical membrane protein field. Admittedly, the idea that each strand of an OMP is individually inserted sounds messy and complex with many working parts that could potentially mess things up, especially in the absence of a known energy source. However, a similar process is observed for the biogenesis of α-helical membrane proteins with multiple transmembrane domains, which can range from a single helix to more than twenty. And yet, while also complex, the task is accomplished nearly flawlessly within the cell for nearly a third of the proteome.

Significant progress has been made over the past few years which has enabled a better understanding of the overall architecture of BAM, of the dynamics within BAM, and of how substrates may be recognized and inserted into the membrane by BAM. Still, the case of ‘BAM-mediated OMP biogenesis’ is not an open and closed case just yet, as there remain many aspects of OMP biogenesis that are yet to be fully deciphered. For example, the role chaperones such as SurA, Skp and FkpA play is still unclear; whether OMPs may be partially or mostly folded within the periplasm or not is yet to be conclusively determined; the process by which chaperones hand off OMPs to BAM remains mostly unexplored; the role of surface-exposed accessory proteins in OMP biogenesis remains a mystery; and what the conformational state of BAM is in a more native environment (i.e. nanodisc or lipid bilayer). To add clarity to these aspects of OMP biogenesis, future studies should aim (1) to decode the determinants for OMP substrate recognition by BAM and (2) to work out the mechanism for how BAM is able fold and insert new OMPs into the membrane and what role the observed conformational changes may serve. Since the structures of BAM have now been reported, we enter the era where the next breakthroughs in the field will require innovative experiments which aim to study the complex mechanism of BAM both at the biochemical level, but also at the near-atomic level using cryo-EM. And while it is easy to get lost in the science and wonderment of BAM and its complex role in OMP biogenesis, more studies targeting BAM for novel antibiotic discovery needs to be pursued, particularly against multi-drug resistant pathogens such as Neisseria, Pseudomonas, and Acinetobacter, where strains have been classified into the CDC’s highest threat levels.

Nonstandard Abbreviations

BAM

β-barrel assembly machinery

OMP

outer membrane protein

CDI

contact-dependent inhibition

LPS

lipopolysaccharide

POTRA

polypeptide transport-associated

TPR

tetratricopeptide repeat

MDR

multi-drug resistance

cryo-EM

cryo-electron microscopy

RMSD

root-mean-square deviation

MD

molecular dynamics

GET

guided entry of tail-anchored proteins