The assembly of β-barrel outer membrane proteins

Abstract

The outer membranes of Gram-negative bacteria, mitochondria, and chloroplasts contain β-barrel integral membrane proteins. The five-protein β-barrel assembly machine (Bam) accelerates the folding and membrane integration of these proteins. The central component of the machine, BamA, contains a β-barrel domain that can adopt a lateral-open state with its N- and C-terminal β-strands unpaired. Recently, strategies have been developed to capture β-barrel folding intermediates on the Bam complex. Biochemical and structural studies provide support for a model in which substrates assemble at the lateral opening of BamA. In this model, the N-terminal β-strand of BamA captures the C-terminal β-strand of substrates by hydrogen bonding to allow their directional folding and subsequent release into the membrane.

Introduction

A characteristic of mitochondria, chloroplasts, and the Gram-negative group of bacteria is that they contain a double layer of membranes. The outer membrane contains proteins of β-barrel structure. Conserved multi-subunit machines assemble these β-barrel proteins into the outer membrane. In Escherichia coli, this machine is the β-barrel assembly machine (Bam) complex [1]. Homologous machines are present in the outer membranes of mitochondria (Sam, sorting and assembly machinery) [2,3] and in the outer membranes of chloroplasts (OEP80, outer envelope protein) [4]. How the Bam complex performs rapid folding and membrane integration of β-barrel membrane proteins, repeatedly and without an input of energy (e.g., from ATP), has been a longstanding question since the discovery of the first components [1,5,6].

To learn how the Bam complex catalyzes folding, several related aspects of the process must be understood: how substrates are held by the Bam components during folding, how the machine promotes folding, and how folded substrates are released to allow catalyst turnover. Observing folding intermediates in complex with the machine is required to answer these questions. In this review, we highlight major milestones in our understanding of β-barrel assembly, emphasizing work that has been done in the past few years.

Structure and function of the Bam complex

The multi-subunit machines that assemble β-barrel membrane proteins in Gram-negative bacteria, mitochondria, and chloroplasts each contain a β-barrel component. It has long been believed that this protein plays the central role in folding β-barrel substrates. In Gram-negative bacteria, the central component is BamA, an essential protein that belongs to the Omp85 superfamily of outer membrane proteins [6]. The Bam complex from E. coli also includes four lipoproteins: BamB, BamC, BamD, and BamE [1,7]. BamA is the only transmembrane component and catalyzes the folding process [8-12]. BamA contains five N-terminal soluble periplasmic polypeptide transport-associated (POTRA) domains and a C-terminal β-barrel transmembrane domain. The POTRA domains act as a scaffold that mediates interaction with the lipoproteins [13-17]. Although all four lipoproteins are required for maximal folding efficiency [18], BamD is the only essential lipoprotein [19]. Several reports have shown that BamA alone catalyzes the chemistry required for cell survival [11,20,21], providing support for the longstanding belief that the roles of the lipoproteins must be regulatory [20-25].

The budding model for β-barrel assembly

Structural studies of the Bam complex from bacteria provided the first clear ideas about the mechanism of substrate folding [8-10,14-17]. These structures showed that BamA can exist either in a closed state or an open state based on the pairing of the N- and C-terminal β-strands (Figure 1a-e). From the structures it was proposed that this open seam pairs by hydrogen-bonding with substrate β-strands to nucleate β-sheet formation [8,26]. A model was proposed in which β-strands of the substrate assemble at the seam and folded portions of the substrate bud through this opening to enter the membrane (Figure 1f). This “budding model” addresses a key aspect of the folding mechanism, which is the pathway by which substrates enter the membrane. As we discuss below, several recent discoveries support the proposal that the β-barrel of BamA opens and β-strands of nascent substrates pass through this opening into the membrane. Recently, structures of the Sam complex were reported that showed the central component, Sam50, in a lateral-open state [27]. These structures suggest that a similar mechanism for β-barrel folding might exist in Gram-negative bacteria and mitochondria.

Figure 1. Structures of the Bam complex and the budding model for folding.

(a) Structure of the BamABCDE complex in which the lateral gate of BamA is closed (PDB: 5D0O). (b) Top-down view of complex in (a). (c) Structure of the BamABCDE complex in which the lateral gate of BamA is open (PDB: 5LJO). (d) Top-down view of complex in (c). (e) Overlay of the β-barrel domain of BamA from a structure with a closed lateral gate (teal; PDB: 5D0O) and a structure with an open lateral gate (orange; PDB: 5EKQ). In the lateral-open structure, the initial β-strands are flipped outward approximately 60 degrees. (f) The budding model for β-barrel assembly into the outer membrane.

Capturing β-barrel folding intermediates

The early structural studies of BamA and the Bam complex identified a pathway for how substrates enter the membrane once the β-strands are assembled into a β-sheet but did not explain how the β-sheet itself forms. More recently, several groups have developed strategies to capture β-barrel folding intermediates by crosslinking them to their assembly machines [12,28-30].

One crosslinking study on Sam50, the BamA homolog in mitochondria, used a series of radiolabeled C-terminal fragments of substrates that were designed to mimic sequential β-hairpin intermediates [28]. These substrates all formed strong crosslinks from the C-terminal β-strand in the substrate to the N-terminal β-strand of the Sam50 β-barrel; however, residues in the N-terminal β-strand of a substrate fragment formed crosslinks to both the N- and C-terminal edges of the Sam50 β-barrel. Because crosslinking was stronger to the C-terminal β-strand of the substrate, it was suggested that the N-terminal β-strand of Sam50 formed a more stable interaction with the C-terminal β-strand of the substrate during folding. Interactions between N-terminal β-strands of the growing substrate with the C-terminal edge of the Sam50 β-barrel were proposed to be transient, forming as new β-strands inserted at the C-terminal edge of the lateral gate and then breaking to allow the additional β-strands to insert. In this model, the N- and C-terminal edge of the lateral gate templates β-sheet formation.

Crosslinking studies of the Bam complex were also reported using substrates accumulated in vivo. One study looked at the folding of an autotransporter containing a C-terminal β-barrel domain. This family of outer membrane proteins also contain an N-terminal passenger (extracellular) domain, which is translocated through the β-barrel during its folding by the Bam complex [31]. This secretion step is required to complete folding of the β-barrel domain [32]. Therefore, to stall the β-barrel on the Bam complex during folding, a full-length autotransporter substrate was fused to a soluble maltose-binding protein at the N-terminus, which prevented translocation of the passenger domain [29]. Crosslinking studies suggested that this substrate interacted asymmetrically with the BamA β-barrel, with the N-terminus of the BamA β-barrel forming strong interactions to the C-terminus of the autotransporter substrate, and the C-terminus of the BamA β-barrel forming two different interaction patterns with the N-terminus of the substrate. These two interaction patterns were proposed to reflect a folding state before and a folding state after membrane integration. The authors proposed a model in which the β-sheet of the substrate forms in the periplasm [33], and then BamA mediates the membrane integration of the largely folded β-barrel through a swinging motion into the membrane.

Another study of folding intermediates accumulated on the Bam complex examined the assembly of LptD, which forms a β-barrel around a lipoprotein plug, LptE [12]. This two-protein complex forms the translocon that inserts lipopolysaccharide into the outer membrane (see review by Ruiz and colleagues in this issue), and LptD assembles orders of magnitude more slowly than other β-barrel substrates [34]. Variants of LptD that fold even more slowly had previously allowed extensive crosslinking experiments to probe interactions between the β-barrel of BamA and the folding substrate [35]. This more recent study showed that the C-terminus of LptD is held at the N-terminal edge of the BamA β-barrel [12]. However, the N-terminal region of LptD was found to form a β-sheet along the concave interior wall of the BamA β-barrel. Changes to residues in strands of the LptD β-sheet that contact the BamA interior, or changes to residues on the interior surface of BamA itself, can increase rate of folding of a substrate that would otherwise accumulate. It was proposed that the interior wall of the BamA β-barrel forms a confined, cage-like environment that serves as an active site for β-sheet formation by reducing the loss of entropy upon folding. Once sufficient nascent structure forms, the substrate could exit the interior of BamA and enter the membrane through the lateral gate.

Structure of a folding intermediate on the Bam complex reveals how turnover can occur

All experimental studies to characterize how substrates interacted with the Bam complex during folding showed a strong interaction between the C-terminal β-strand of the substrate and the N-terminal β-strand of the BamA β-barrel. It makes sense that a β-barrel assembly machine would hold onto a substrate tightly until folding is complete. To complete folding, the N- and C-terminal β-strands of the substrate must pair with one another so the β-barrel can close. Therefore, the C-terminal β-strand of the substrate must somehow dissociate from the N-terminal edge of BamA. Proposed folding models did not provide a mechanism for how release occurs. It was observed that the folded β-barrel is stable in the membrane, but a thermodynamic driving force cannot explain how the Bam complex catalyzes folding, which depends on kinetic barriers. This machine carries out rapid, repeated assembly of β-barrel substrates, and operates without an exogenous energy source [18,36]. How the Bam complex can hold substrates stably at its N-terminus during folding without releasing them prematurely, but then release them rapidly to complete folding, remained a question.

The structure of a nascent β-barrel as it folds on the Bam complex was reported recently and suggested a mechanism for release that was supported experimentally [30] (Figure 2). BamA itself is folded by the Bam complex and a series of BamA substrates was generated, each lacking one of the eight extracellular loops. Loop deletions were expected to slow folding of the substrate to allow accumulation and capture on the machine. Experiments with these mutants in cells established that folding proceeds from the C-terminus of the substrate towards the N-terminus. A structure of the Bam complex bound to a loop-deleted substrate showed a snapshot of a very late-stage folding intermediate. At this stage, the C-terminal β-strand of the substrate is paired with the N-terminal β-strand of the machine via a series of six hydrogen bonds. The nascent β-barrel protrudes into the membrane. The N-terminal edge of the substrate is not hydrogen-bonded to the C-terminal edge of the BamA β-barrel, but instead curves inward and points toward its own C-terminal β-strand with the two β-barrels contacting each other on their exterior surfaces. This closed interface ensures that the unpaired edges of each β-barrel face into an aqueous lumen where they are solvated by water, while also ensuring that membrane lipids do not fill the interior.

Figure 2. Structure of a late-stage folding intermediate on the Bam complex.

(a) Side view of the cryo-EM structure of the Bam complex bound to a late-stage BamA folding intermediate. (b) Top-down view of complex in (a). (c) Overlay of the β-barrel domain of BamA from the substrate-engaged complex (green; PDB: 6V05) and a substrate-free, lateral-open complex (orange; PDB: 5EKQ).

In order for the substrate to release from the machine, a total of six hydrogen bonds must be broken so that the N- and C-terminal strands of the substrate can pair. In a membrane where the dielectric is low, these bonds are worth ~25 kcal/mol, and breaking them simultaneously would have a huge kinetic barrier. Given that turnover occurs in the absence of ATP, there must be a mechanism for release that occurs through a pathway that does not require simultaneous rupture of all of these bonds.

The structure showed an overhang at the very C-terminus of the stalled substrate that did not pair with the N-terminal edge of the BamA β-barrel. It was proposed that this region of the substrate can initiate pairing with the free N-terminal β-strand of the substrate, allowing a stepwise exchange of hydrogen bonds to the machine for hydrogen bonds between the two terminal edges of the substrate. Multiple lines of evidence showed the importance of the C-terminal overhang in release. Dramatically, single amino acid changes at the C-terminus of the substrate were found to stall an otherwise completely intact substrate on the machine.

Current model for β-barrel assembly

The recent work described here leads to a model for β-barrel assembly by the Bam complex (Figure 3). Initially, the substrate associates with Bam components at the outer membrane (step a). Interactions with BamD may trigger opening of the BamA β-barrel at its seam [24]. The C-terminal β-strand of the substrate is then captured by the N-terminal β-strand of BamA to form a continuous β-sheet (step b). The substrate enters the interior of the BamA β-barrel, where folding of the β-sheet, catalyzed by stabilizing interactions with the interior wall of the BamA β-barrel, results in the addition of β-strands from the C-terminus of the substrate towards the N-terminus (step c). Once a significant amount of folding occurs, the substrate passes into the membrane through the open seam of BamA (step d). Features at each end of the substrate promote release by a stepwise exchange of hydrogen bonds that result in the closure of the substrate β-barrel (step e).

Figure 3. Current model for β-barrel assembly.

(a) The substrate is recruited to the Bam complex. (b) Opening of the seam of the BamA β-barrel allows interactions with the unfolded substrate. (c) Folding proceeds in a C- to N-terminal direction as β-strands are added within the interior of BamA to form a β-barrel. (d) Full membrane integration occurs at a late stage of folding, and the remaining β-strands are added to complete the β-barrel. (e) Release of the substrate occurs by a hydrogen bond exchange mechanism. For simplicity, only BamA (containing N-terminal soluble domains and a C-terminal β-barrel transmembrane domain) and BamD are shown.

Conclusions and Future Directions

Important advances in our understanding of β-barrel assembly have been made over the past few years. One surprising discovery was that BamA captures only one end of the substrate by hydrogen bonding, while the other end remains unpaired. One remaining question is how folding is initiated to form the hybrid β-barrel. It has been proposed that BamD or the POTRA domains of BamA can recruit substrates [12,13,24,37,38], possibly by interactions with chaperones [39,40]; however, it is unclear how substrates are then passed to the open seam of BamA. Another interesting question is how β-barrels that form oligomers assemble in the membrane. It has been shown that multiple Bam complexes can co-localize in outer membrane “precincts” mediated by interactions between BamB within adjacent complexes, and these precincts were shown to be important in the assembly of β-barrel trimers [41]. It has also been assumed that substrates traverse the periplasm via chaperones that deliver them to the outer membrane [39]. Recently, however, it has been proposed that a super-complex consisting of the Sec translocon, the chaperone SurA, and the Bam complex exists to connect the inner and outer membrane for protein delivery [42]. Additional work will confirm the relevance of these protein bridges to β-barrel assembly in vivo. The Bam complex assembles a variety of substrates, some of which have a plug-and-barrel architecture [43-45]. A recent structure showed that the interior of the BamA β-barrel can accommodate a folded lipoprotein, RcsF (see review by Collet and colleagues in this issue) [46]. If the Bam complex is functioning properly, RcsF is believed to be transferred from BamA to the interior of some substrate β-barrels. The recent structural work described above provides a picture for how lipoprotein transfer could occur—when substrates are folding, RcsF can move from the interior of BamA to that of the substrate, since the interiors of the two β-barrels are connected. Future work will address whether other lipoproteins, such as LptE, which is involved in assembly of the β-barrel LptD, can also be held initially within the interior of BamA prior to transfer to LptD. Finally, future studies should be aimed at capturing different substrates at various stages of folding to construct a complete physical picture of the process.

Antibiotic resistant infections, especially ones caused by Gram-negative bacteria, are an emerging problem. Recently, an antibody targeting BamA has been shown to perturb outer membrane integrity when added to E. coli cells and can prevent proper folding of outer membrane proteins by binding extracellular loops of BamA [47]. Furthermore, small molecules that interact with the Bam complex have been identified that lead to cell death by preventing efficient function of this essential molecular machine, likely by stabilizing the closed state of BamA (see review by Walker and colleagues in this issue) [48-50]. Recent structural and functional insights into the folding of β-barrels may enable the design of new antibiotics, as proper assembly of these proteins is critical for survival of Gram-negative bacteria.

Highlights.

• The BamA β-barrel adopts an open state with its first and last β-strands unpaired.

• BamA β-strand 1 captures the C-terminal β-strand of substrates by hydrogen bonding.

• Folding proceeds from the C-terminus of the substrate towards the N-terminus.

• Folding occurs in the interior of the BamA β-barrel prior to membrane integration.

• Interactions between the N- and C-termini of the nascent β-barrel allow release.