Interplay of protein primary sequence, lipid membrane, and chaperone in β‐barrel assembly

Abstract

The outer membrane of a Gram‐negative bacterium is a crucial barrier between the external environment and its internal physiology. This barrier is bridged selectively by β‐barrel outer membrane proteins (OMPs). The in vivo folding and biogenesis of OMPs necessitates the assistance of the outer membrane chaperone BamA. Nevertheless, OMPs retain the ability of independent self‐assembly in vitro. Hence, it is unclear whether substrate–chaperone dynamics is influenced by the intrinsic ability of OMPs to fold, the magnitude of BamA–OMP interdependence, and the contribution of BamA to the kinetics of OMP assembly. We addressed this by monitoring the assembly kinetics of multiple 8‐stranded β‐barrel OMP substrates with(out) BamA. We also examined whether BamA is species‐specific, or nonspecifically accelerates folding kinetics of substrates from independent species. Our findings reveal BamA as a substrate‐independent promiscuous molecular chaperone, which assists the unfolded OMP to overcome the kinetic barrier imposed by the bilayer membrane. We additionally show that while BamA kinetically accelerates OMP folding, the OMP primary sequence remains a vital deciding element in its assembly rate. Our study provides unexpected insights on OMP assembly and the functional relevance of BamA in vivo.

Abbreviations

Ail

attachment invasion locus protein

BAM

barrel assembly machinery

DMPC

1,2‐dimyristoyl‐sn‐glycero‐3‐phosphocholine

DMPE

1,2‐dimyristoyl‐sn‐glycero‐3‐phosphoethanolamine

LUV

large unilamellar vesicles

nOMP

nascent OMP

OMP

outer membrane protein

1. INTRODUCTION

Gram‐negative bacteria and eukaryotic mitochondria and chloroplast are embedded with transmembrane proteins in their outer membrane that are collectively referred to as outer membrane proteins (OMPs). These proteins share a common β‐barrel architecture, but carry out versatile functions as adhesins, invasins, efflux pumps, selective transporters, and porins. ¹ , ² Defects in OMP assembly lead to protein misfolding and aggregation, which link to stress and cellular toxicity in prokaryotes. ² The insertion and assembly of an OMP into the bacterial outer membrane is a highly regulated process, ³ , ⁴ , ⁵ enabled by assistance from periplasmic holdases (e.g., SurA and Skp) and the outer membrane chaperone. ⁴ , ⁵ , ⁶ , ⁷ , ⁸ , ⁹ This phylogenetically conserved transmembrane molecular chaperone is a multiprotein complex referred to as the barrel assembly machinery (BAM). ¹⁰ The central and functionally autonomous component of the BAM complex is the transmembrane 16‐stranded β‐barrel protein BamA. ⁴ , ⁶ , ¹⁰ , ¹¹ , ¹² , ¹³ Assembly defects in the BAM complex cause dysfunctioning in OMP biogenesis, leading to cell death. ¹⁰ , ¹⁴ , ¹⁵

The general process of OMP biogenesis and its translocation to the target membrane is fairly well understood in bacteria. ⁷ , ¹¹ , ¹⁶ , ¹⁷ , ¹⁸ Yet, several unresolved ambiguities and open questions exist in the underlying molecular mechanism of OMP assembly and its regulation in the membrane. ⁷ , ¹⁶ For example, bacterial membranes are enriched with phosphoethanolamine (PE) lipids, which are known to impose kinetic retardation in OMP folding in experiments done in vitro. ¹⁸ , ¹⁹ , ²⁰ , ²¹ The BAM complex helps the unfolded nascent OMP (nOMP) overcome the bilayer‐induced kinetic barrier during its insertion into the outer membrane. ¹⁹ Three currently accepted hypotheses for BAM‐assisted folding of nOMPs involves deformation in the membrane architecture due to the asymmetric BamA barrel structure, augmented assembly of the nOMP on the BamA β‐barrel scaffold, and formation of a lateral gate by BamA that forms the entry point for the client nOMP during its assembly. ⁵ , ⁷ , ¹³ , ²² , ²³ , ²⁴ BamA recognizes the highly conserved C‐terminal β‐signal of the nOMP substrate and facilitates assembly of the nOMP by lowering the kinetic barrier imposed by the PE‐rich membrane. ²⁵

The kinetic barrier imposed on OMP folding by PE lipids of the bacterial outer membrane endorses the existence and necessity of the evolutionary conserved BAM complex. ¹² , ¹⁴ , ¹⁹ Extensive studies on variants and mutants of the BAM complex have delineated the importance of this chaperone for OMP insertion and assembly, and for bacterial survival. ⁸ , ⁹ , ¹⁰ , ¹⁴ , ¹⁶ Despite this need for a chaperone in vivo, several in vitro studies have independently demonstrated that the folding of a nOMP is intrinsically encoded in its primary sequence. ¹⁷ , ²⁰ , ²⁶ , ²⁷ , ²⁸ Advancements in studies on membrane protein folding have highlighted that the folding reaction of nOMPs is thermodynamically controlled, and depends on their primary sequence. ³ , ¹² , ¹⁶ , ²⁰ , ²⁵ , ²⁹ , ³⁰ , ³¹ , ³² Indeed, innumerable studies have demonstrated that nOMPs can spontaneously fold into their native conformation in both artificial detergents and lipid bilayer membranes. ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ , ³³ , ³⁴ , ³⁵ Of the various factors that influence the rate of nOMP insertion and assembly kinetics in the membrane, membrane thickness, fluidity, composition, lipid headgroup, temperature, and pH are known to play major roles. Not surprisingly, in a given lipid membrane and microenvironment, different nOMPs would be expected to display varied folding tendencies in vitro. ²⁰

The ability of a nOMP to independently exhibit correct and spontaneous assembly allowed us to ask whether BamA functions as a translocase and holdase to direct the nOMP to the outer membrane and prevent its aggregation, or as a molecular chaperone that specifically accelerates nOMP assembly by several magnitudes. On similar lines, we also found it of interest to address promiscuity in substrate recognition by BamA of the BAM complex. Earlier studies have characterized the mechanism by which BamA folds β‐barrels of different sizes (8–26 strands), ¹⁹ , ²⁰ , ²³ , ²⁴ , ³⁶ , ³⁷ , ³⁸ , ³⁹ , ⁴⁰ and have found interesting differences in the nOMP folding rate and assembly pathway. Hence, the length of the substrate polypeptide influences how BamA interacts with the nOMP during its assembly and the pathway it adopts. However, when presented with near‐identical substrates (of similar chain lengths, final structure, and similar BamA recognition motifs), from different bacterial origins, whether BamA demarcates these nOMPs is not known. Further, what attributes of the intrinsic folding efficiency of the nOMP influence the observed nOMP–BamA interplay is also unclear.

Here, we specifically examine the interplay of the primary sequence of the substrate nOMP with BamA, by presenting Escherichia coli BamA with structurally similar (eight‐stranded) OMPs from various bacterial sources. We additionally study the importance of the lipid‐imposed physicochemical membrane barrier by comparing the folding of these eight‐stranded OMPs in phosphocholine and PE‐doped membranes. We find that thicker bilayer membranes and the presence of PE retard the folding kinetics of nOMPs. However, BamA accelerates nOMP folding kinetics in both phosphocholine and PE‐doped membranes, which is in good agreement with previous findings. Additionally, we find that BamA acts as a species‐ and substrate‐independent promiscuous chaperone, wherein it accelerates the assembly of all nOMPs from different bacteria. Our studies also reveal that the enhancement in nOMP folding rate by BamA, and the folding efficiency achieved therein, depends largely on the intrinsic folding ability of the nOMP. This influences the measured kinetics of BamA‐assisted folding of nOMPs in both phosphocholine and PE membranes. Our observations also imply a common pathway for BAM‐assisted acceleration of nOMP folding in vivo for eight stranded barrels, across species.

2. RESULTS

2.1. Electrophoretic mobility shift to assay BamA‐assisted folding of diverse nOMPs in PE membranes

The assembly of nOMPs into the bacterial outer membrane is influenced by several factors. Of these, the two major factors causing kinetic retardation are the chemical nature of the lipid headgroups and thickness of the bilayer. Despite these physico‐chemical factors, the outer membrane chaperone BamA sustains cell survival through a controlled and highly regulated process of nOMP biogenesis and assembly. ¹⁰ , ¹⁴ , ¹⁵ Loss in controlled nOMP targeting coupled with overexpression of nOMPs can cause bacterial cell death. Despite the intrinsic limitations to in vivo folding, spontaneous folding of nOMPs does occur in vitro. Whereas several studies have addressed the kinetic contribution and mechanism of BamA in folding nOMPs, ¹⁶ , ¹⁸ , ²⁴ , ⁴¹ the precise role(s) of BamA need additional characterization. ⁷ In the absence of the BAM complex, native membrane extracts are known to be unsuitable in supporting the folding of nOMPs. ¹⁹ As multiple factors such as substrate concentrations, protein–lipid interaction dynamics, and molecular overcrowding interfere with accurately monitoring the ephemeral process of nOMP folding, we used in vitro reconstitution to monitor the direct implications of BamA on nOMP assembly in synthetic lipid vesicles.

The in vitro assembly process of membrane proteins is routinely monitored through various spectroscopic and calorimetric methods. An interesting intrinsic property of several bacteria transmembrane β‐barrels is the unique differential electrophoretic migration of the folded and unfolded protein bands on cold SDS‐PAGE gels. The electrophoretic mobility shift assay ⁴² has therefore been used to both establish the folded state of OMPs as well as monitor the kinetics of the folding process. We used this simplistic approach of utilizing the differential mobilities of OMPs in their folded and unfolded states, to investigate the role of BamA and the lipid headgroup on the assembly kinetics of nOMPs. We used seven different eight‐stranded OMPs, namely, E. coli outer membrane protein X (OmpX), transmembrane domain of OmpA (tOmpA), PhoPQ‐activated gene P (PagP), Salmonella typhimurium PagP (PagP‐St), PagC‐St, PagN‐St, and Yersinia pestis attachment invasion locus (Ail‐Yp) protein (Table 1; also see Table S1 and Figure S1), to address the mechanistic pathway of OMP assembly. Studying the role of E. coli BamA on the assembly of these seven structurally similar nOMPs of identical barrel sizes allows us to address whether BamA displays species‐ and substrate‐specificity, that is, promiscuity in its function.

TABLE 1.

OMPs used in this study

OMP	UniProt ID	MW (kDa)	No. β‐strands	Source	pI	Attributed functions a
BamA b	P0A940	90.7	16	Escherichia coli	4.87	Chaperone for insertion and assembly of nOMPs
tOmpA c	P0A910	18.8	8	E. coli	5.74	Porin, resistance to environmental stress; implicated in outer membrane stability and shape, peptidoglycan binding
OmpX	P0A917	16.5	8	E. coli	5.30	Binds human antigen presenting cells; linked with fimbriae production
PagP‐Ec	P37001	19.1	8	E. coli	5.50	Lipid A palmitoyltransferase; helps evade host immune defense
PagP‐St	Q8ZR06	19.0	8	Salmonella typhimurium	5.89	Lipid A palmitoyltransferase; helps evade host immune defense; confers resistance to antimicrobial peptides
PagC	P23988	17.9	8	S. typhimurium	6.28	Virulence protein; involved in pathogenesis; essential for survival within macrophages
PagN	Q8ZRJ9	23.5	8	S. typhimurium	5.60	Adhesin/invasin protein; involved in pathogenesis; facilitates adhesion and invasion of mammalian epithelia; hemagglutinin
Ail	Q8D0Z7	17.6	8	Yersinia pestis	7.84	Virulence and adhesin/invasion protein; involved in pathogenesis; adhesion and invasion of host cell; Yersinia outer protein delivery

Abbreviation: nOMP, nascent outer membrane protein.

^aObtained from www.uniprot.org.

^bContains an N‐terminal His₆ tag and a thrombin cleavage site that were retained in the mature protein.

^cTransmembrane domain only (residues 22–192).

Notably, all nOMPs possess a β‐signal (Aro‐Xaa‐Aro) at the C‐terminus, which forms a vital part of the BamA recognition motif. We reasoned that any variations in nOMP folding rates must not arise from differences in substrate recognition efficiency by BamA. ²⁵ Choosing similar nOMPs allowed us to address whether, despite similarities in sequence (Table S1), β‐signal recognition motif (Figure S1), and structure, E. coli BamA shows substrate nOMP‐specific acceleration. Further, this allowed us elucidate whether BamA functions as a promiscuous or species‐specific chaperone, under identical lipidic conditions. We specifically included PagP from E. coli and S. typhimurium, as they share >80% sequence identity (Table S1), but possess different thermodynamic properties. ³² The seven structurally similar nOMPs from the three different species, as well as BamA, exhibit differential electrophoretic mobility upon folding (unfolded nOMPs show retarded gel mobility compared to their folded counterparts; Figures 1, S2, and S3). Using this electrophoretic mobility assay, we addressed the importance of the protein's intrinsic ability to fold, wherein, despite adopting a common β‐barrel fold, each nOMP has evolved to different structural characteristics and functional entities in different bacterial hosts.

FIGURE 1.

BamA‐assisted outer membrane protein (OMP) folding. Electrophoretic mobility shift assay reveals folding efficiency of nascent OMP (nOMPs) in 1,2‐dimyristoyl‐sn‐glycero‐3‐phosphocholine (DMPC) vesicles. The faster gel mobility in Tricine gels observed upon folding of two representative OMPs, namely Escherichia coli OmpX (left) and Salmonella typhimurium PagP (right) monitored with time (shown in log₁₀ scale) are presented as percent folded. Insets show representative gel images for both proteins with(out) BamA, and monitored with time (F: folded protein; UF: unfolded protein; lanes 1,2:2 hr; 3,4:4 hr; 5,6:6 hr). Note the accumulation of an additional species between the F and UF bands in PagP‐St (right, inset). The origin of this species (seen predominantly in PagP) is uncertain and has not been observed in previous studies that employed 10‐ and 12‐C vesicles. Nevertheless, to maintain consistency in our densitometry analysis, we have only considered the F and UF band intensities. The contribution of BamA‐Ec (black circles or −, without BamA; red triangles or +, with BamA) to the folding efficiency of the substrate OMP is also presented. Fits of the mean rates to an exponential function are shown as solid lines. See Figures S2–S5 for the complete graphs and gel images of all proteins in DMPC and PCPE vesicles, and for details of the fitting

We chose 14‐C lipids to address the effect of thicker bilayers on the folding efficiency of nOMPs, allowing us to compare our findings with folding kinetics reported in 10‐C–12‐C lipids. ¹⁸ , ¹⁹ , ²⁰ Here, large unilamellar vesicles (LUVs) formed using the 14‐C phosphatidylcholine 1,2‐dimyristoyl‐sn‐glycero‐3‐phosphocholine (DMPC), which form lamellar membranes that support the folding of nOMPs, were employed. PE, found in abundance in the bacterial outer membrane (~75% ⁴³ , ⁴⁴ , ⁴⁵ in all the three organisms used in this study) imposes a physicochemical retardation to the folding and insertion of nOMPs in vitro. ¹⁹ , ²¹ , ²⁶ Hence, we compared the folding kinetics in DMPC vesicles with that obtained from DMPC doped additionally with PE (14‐C, 1,2‐dimyristoyl‐sn‐glycero‐3‐PE [DMPE]), and coupled our analysis to the influence of prefolded BamA in these vesicles (Figure 1). Put together, we studied 28 different experimental conditions involving BamA, seven nOMPs, and two lipidic membranes.

The folding and insertion of BamA in PC and PCPE LUVs were achieved using well‐established protocols. ⁹ , ¹⁸ , ¹⁹ Here, rapid dilution of urea‐unfolded BamA into the folding buffer containing preformed LUVs resulted in BamA folding, with an efficiency of ~50–70% (assessed from densitometry of electrophoretic mobility shift assay; see Figures S2 and S3 also for details of BamA folding efficiency). Folded BamA is resistant to unfolding by residual urea in the folding reaction, as reported earlier. ⁹ , ¹⁹ , ²⁰ , ²⁶ Hence, we matched previously reported experimental conditions by supplementing 1.0 M urea in our experiments, and also retained the BamA‐substrate ratios used therein, ¹⁹ thereby allowing us to compare our measurements of BamA‐assisted k _obs in 14‐C lipids with rates obtained in 10‐C lipids. ¹⁹

2.2. Assembly of nOMPs is not retarded by the PE headgroup

PE lipids, which are found in the bacterial outer membrane, structurally induce negative curvature and higher packing density in the membrane due to their smaller headgroups. ²¹ This tighter packing reduces defects in the bilayer architecture, and along with the bilayer thickness, imposes a kinetic retardation in the folding process of nOMPs into the bacterial outer membrane. ¹⁸ , ¹⁹ , ²⁰ , ⁴⁶ Earlier in vitro studies have shown that practical limitations exist in preparing PE‐rich vesicles and the experimental temperature conditions limits us with the maximum molar doping limit of PE in PC LUVs. ⁹ , ¹⁸ , ¹⁹ , ⁴⁶ Higher PE content in vesicles also increases the phase transition temperature (50°C for 100% DMPE), and affects the folding efficiency of BamA in these vesicles. ¹⁹ , ²⁰ PC lipids, although absent in bacterial outer membranes, support LUV structures due to their lamellar nature; hence, they are widely used to study bacterial OMPs. ⁹ , ¹⁸ , ¹⁹ , ⁴⁶ Lamellar lipids with smaller headgroups, such as PE, are doped in PC‐containing vesicles (20% PE doping could be achieved in our experiments), to obtain a more native‐like membrane. To address the contribution of PE, we monitored the folding kinetics of all the nOMPs in PC vesicles (Figures S2 and S4) and in PE‐doped PC vesicles (PCPE) (Figures S3 and S5) using the electrophoretic mobility shift assay.

Comparison of the measured intrinsic folding rates (k _obs) of nOMPs shows no significant kinetic retardation in the PCPE LUVs as compared to the PC vesicles for nearly all the OMPs (Figure 2a). Surprisingly, the folding kinetics of PagN‐St is enhanced considerably in the presence of PE (Figure 2a). Further, the folding kinetics of PagP‐Ec, PagP‐St, and PagC‐St are not significantly affected by PE addition. Unlike previous observations wherein a PE‐imposed retardation in nOMP assembly was observed, ⁹ , ¹⁹ , ²⁰ we see a limited effect of PE doping on both the nOMP folding rates (Figures 2a and S6a) and the end‐point folding efficiency (see Figure S8; discussed later). The most prominent effect of PE is obtained only in Ail‐Yp (the evolutionary homolog of E. coli OmpX; both proteins share ~45% sequence identity [see Table S1]), which leads to protein aggregation. Thus, PE doping and the associated change in bilayer thickness significantly affects the kinetic partitioning and folding efficiency of Ail.

FIGURE 2.

BamA dependence of the folding kinetics of nascent outer membrane proteins (nOMPs) in vesicles. (a,b) Comparison of the observed rate (k _obs) of folding of nOMPs with(out) BamA. The BamA unassisted folding rates of nOMPs derived in 1,2‐dimyristoyl‐sn‐glycero‐3‐phosphocholine (DMPC) (PC, gray) and PCPE (white) vesicles are shown in (a) and BamA‐assisted folding rates DMPC (PC, black) and PCPE (checkered) vesicles are shown in (b). Note that in the absence of BamA, Ail exhibits aggregation in both PC and PCPE lipids, and the rates in (a) are measured from the fraction of protein that did not precipitate in the reaction. A direct comparison of the effect of BamA on tOmpA‐Ec and OmpX‐Ec folding in DMPC and PCPE is shown in (c). The influence of prefolded BamA on the observed rates of folding across all nOMPs in DMPC (d) and PCPE (e) are also compared. The observed rate of folding of PagP from the two different species Escherichia coli (Ec) and Salmonella typhimurium (St) sharing >85% sequence identity and in the two lipidic vesicles is compared directly in (f) to highlight that the chaperone function of BamA is not species‐specific. Error bars in all cases represent the SD obtained by fitting three independent datasets to an exponential function. Statistical analysis was carried out using Student's t test (p = .05 [*], p = .01 [**], p = .002 [***]). Also see Figure S6 for rates obtained from fitting the mean data to an exponential function (errors shown therein represent goodness of fit)

The 10–12‐C PC lipids used in the earlier studies have higher dynamicity and lower phase transition temperatures than the 14‐C lipids used herein. Previous studies have also shown substantial kinetic retardation of nOMP folding in membranes of longer hydrocarbon chains. ¹⁸ Hence, we propose that the kinetic retardation imposed by thicker membranes can parallel (or supersede) the PE headgroup‐imposed retardation in nOMP folding. Additionally, we find that the k _obs differ across the nOMPs within species and across species (e.g., compare tOmpA, OmpX, and PagP from E. coli, as well as PagP from E. coli and S. typhimurium in Figure 2a). These observations together suggest the likely existence of nOMP sequence dependence in the barrel assembly rate that supersedes any kinetic retardation that may be imposed by the PE membrane.

2.3. BamA enhances nOMP assembly kinetics in the presence of PE

Previous studies have shown that the PE‐imposed kinetic retardation in nOMP folding is overcome by the well‐structured functional BAM complex. ⁷ , ¹³ , ²⁴ , ⁴¹ Here, we find that the measured k _obs of all nOMPs are significantly enhanced in the presence of prefolded BamA, and this is lipid‐dependent (Figures 2b–f and 3). For example, the k _obs of PagP‐St and PagC‐St are enhanced when prefolded BamA is provided in DMPC (Figures 2d and 3). However, BamA does not enhance the folding efficiency of tOmpA‐Ec, OmpX‐Ec, PagP‐Ec, and PagN‐St in PC, irrespective of the measured k _obs for the intrinsic folding.

FIGURE 3.

Role of the lipid headgroup in BamA‐assisted nascent outer membrane protein (nOMP) folding kinetics. Contribution of BamA to the fold increase in the observed rate of folding of nOMPs in 1,2‐dimyristoyl‐sn‐glycero‐3‐phosphocholine (DMPC) (PC, filled histograms) and PCPE (open histograms) vesicles. Notably, BamA acts as a chaperone and enhances the folding rate of OMPs in both lipidic vesicles irrespective of the bilayer thickness. Also note that the magnitude of enhancement by BamA is different for each OMP, does not correlate directly with the lipid headgroup, and is independent of the species. The dotted line indicates no effect of BamA on the k _obs. Values are calculated from the k _obs presented in Table 2

In PCPE, the presence of prefolded BamA enhances the folding rate for several nOMPs examined herein (Figures 2b,c,e and 3; also see Figure S6). The highest increase in folding rate of approximately ninefold is seen with PagP‐St in PC and ~3.0 to 3.5‐fold in PCPE membranes for both E. coli (tOmpA, OmpX) and S. typhimurium (PagP, PagC, PagN) proteins, whereas the folding efficiency in the presence of BamA is not significantly enhanced for PagP‐Ec (Figure 3). Interestingly, and unlike PC membranes, BamA enhances the folding efficiency of both tOmpA and OmpX in PCPE (Figure 2c). Therefore, BamA assists both nOMPs overcome the intrinsic retardation due to the thicker membrane additionally doped with PE. Our findings are in excellent agreement with previous results from tOmpA–BamA in other lipidic systems. ⁹ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁶ Overall, the fold‐increase in k _obs is different for each nOMP, suggesting contributions of the substrate and the lipidic environment in the measured assembly kinetics.

Additionally, we find that BamA does not substantially increase the folding efficiency (end‐point % folded) of OmpX in both conditions (Figure S8). Furthermore, BamA increases the folding efficiency of tOmpA in DMPC (Figure S8a) but not in PCPE (Figure S8b). This observation can be explained when we compare % folded for both proteins with(out) BamA (Figure S7). Although the k _obs vary (Figure 2c), the % folded for OmpX is largely similar in all conditions. In tOmpA, the % folded is lower for the unassisted assembly in DMPC (Figure S7a versus S7B). Hence, we infer that BamA enhances both the folding rate and the folding efficacy of tOmpA in DMPC.

We also find that BamA assists nOMPs whose folding efficiency is lowered by the lipid membrane, and are additionally aggregation‐prone. Particularly with Ail‐Yp, which shows the poorest folding efficiency in both lipids, BamA facilitates the nOMP overcome the PE‐induced kinetic barrier. An enhanced folding rate (Figures 2 and 3) for all nOMP substrates and an increase in the nOMP folding efficiency supports the role of BamA as a molecular membrane chaperone. BamA additionally scavenges the nOMP from aggregation in the case of Ail‐Yp (wherein we observed protein aggregation in the unassisted reactions), and aids its insertion and folding into vesicles. BamA achieves this by increasing the folding flux and concomitantly reducing the flux through the aggregation pathway. In summary, the presence of BamA enhances the folding rate of select nOMPs in PC membranes (PagP‐St, PagC‐St) and in PE membranes (tOmpA‐Ec, OmpX‐Ec, PagC‐St, and PagN‐St) (Figures 2 andS6), and additionally influences the folding efficiency for those nOMPs that are aggregation prone (Ail‐Yp). Our findings support the proposed mechanisms for BAM‐assisted OMP biogenesis, ⁵ , ²⁴ , ⁴⁷ in which nOMPs circumvent the membrane‐associated kinetic retardation with the help of BamA.

2.4. BamA is a species‐independent promiscuous chaperone

In vivo, BamA subunit recognizes the evolutionarily conserved β‐signal recognition motif present at the C‐terminal of nOMPs (Figure S1). ²⁵ It has been proposed that BamA then actively inserts and folds the nOMP into the outer membrane by altering local membrane properties and overcoming the membrane‐imposed kinetic barrier to nOMP folding. ¹⁸ , ²⁶ , ³⁸ , ⁴¹ While BamA is indispensable for nOMP assembly in vivo, it is unclear whether it exhibits species specificity in substrate nOMP recognition, whether BamA selectively accelerates the folding of specific substrates, and whether BamA functions as a holdase or chaperone.

From our experiments, we find that the presence of prefolded BamA selectively accelerates nOMP folding rates when compared with the unassisted folding rates in both PC and PCPE vesicle conditions (Figures 2, 3, and S6). This elevated nOMP folding rate and folding efficiency is evident when the BamA‐assisted rates are compared to the unassisted folding rates in all the substrate OMPs irrespective of their origin, and in both lipidic conditions with variable bilayer thickness (Figure 3). The results support the function of BamA as a molecular chaperone in the folding of all substrate OMPs. However, we find that the contribution of BamA varies between different substrates and in both membrane environments, suggesting the existence of other molecular factors which dictate the nOMP assembly rate.

To address whether E. coli BamA shows species specificity, we compared the folding rates and amplitudes across the different nOMPs (Table 2). Our analysis reveals that E. coli BamA does not differentially recognize nOMPs from E. coli, namely OmpA, OmpX, and PagP, over nOMPs from other species namely PagN, PagC and PagP from S. typhimurium and Ail from Y. pestis. The species‐independent function of BamA is evident further when we compare the two nearly identical proteins from two different sources that is, PagP‐Ec and PagP‐St (Figure 2f). Both PagP proteins differ only by 15 of the 144 residues in their transmembrane region, but vary by twofold and ~15‐ to 20‐fold in their thermodynamic and enzymatic properties, respectively. ³² However, BamA from E. coli nonspecifically accelerates the intrinsic folding rate of PagP‐St in both PC and PCPE (Figure 2f and 3), which is in agreement with our conclusion that E. coli BamA is a nonspecific chaperone. An approximately ninefold enhancement in the folding rate of PagP‐St in PC and ~3.5‐fold in PCPE membranes (Figure 3) suggests that E. coli BamA helps PagP‐St overcome the physical retardation caused by bilayer thickness irrespective of its composition. BamA also selectively accelerates the folding of thermodynamically less stable OMPs such as Ail (approximately threefold in PCPE; Figure 3). Put together, we demonstrate that the BamA‐dependent acceleration of folding of nOMPs is independent of the source of these OMPs, and validates our conclusion that BamA acts as a promiscuous membrane chaperone.

TABLE 2.

Role of BamA in the assembly kinetics and amplitudes of nOMPs in lipid vesicles

OMP	Lipid vesicles a	−BamAb		+BamAb
		k obs (hr−1)	Amplitude	k obs (hr−1)	Amplitude
tOmpA	PC	0.85 ± 0.49	60.82 ± 3.50	0.70 ± 0.06	86.63 ± 1.81
	PCPE	0.32 ± 0.09	84.38 ± 6.47	1.15 ± 0.46	70.63 ± 3.27
OmpX	PC	0.77 ± 0.23	78.85 ± 2.43	1.16 ± 0.15	86.90 ± 2.01
	PCPE	0.60 ± 0.27	74.01 ± 4.96	2.23 ± 0.47	84.93 ± 1.61
PagP‐Ec	PC	0.20 ± 0.04	57.98 ± 1.16	0.18 ± 0.03	67.29 ± 5.77
	PCPE	0.24 ± 0.05	53.35 ± 4.89	0.36 ± 0.10	66.02 ± 1.44
PagP‐St	PC	0.01 ± 0.01	59.52 ± 29.72	0.10 ± 0.04	38.56 ± 3.84
	PCPE	0.03 ± 0.02	39.77 ± 24.87	0.09 ± 0.06	50.04 ± 7.16
PagC	PC	0.35 ± 0.13	58.22 ± 3.76	0.99 ± 0.35	71.43 ± 1.01
	PCPE	0.35 ± 0.10	68.93 ± 5.46	1.00 ± 0.19	67.96 ± 3.85
PagN	PC	0.15 ± 0.09	44.50 ± 9.31	0.21 ± 0.03	52.80 ± 10.48
	PCPE	0.35 ± 0.10	31.07 ± 4.48	0.80 ± 0.07	49.74 ± 5.68
Ail	PC	0.31 ± 0.25	20.94 ± 4.92	0.29 ± 0.06	36.75 ± 3.67
	PCPE	0.08 ± 0.02	28.95 ± 0.86	0.28 ± 0.17	44.23 ± 5.76

^aPC: 100% 14:0 PC; PCPE: 80% 14:0 PC + 20% 14:0 PE.

^bAll rates and amplitudes are presented as mean ± SD, derived from fits of 0–48 hr data of three independent experiments (0–108 hr was used for PagP‐St).

2.5. nOMP primary sequence outweighs BamA functioning and lipid compositional variation

Our experimental investigation, together with previous studies on nOMP assembly kinetics, provides the conclusion that prefolded BamA enhances the folding rate for any nOMP with a functional β‐signal, irrespective of the lipid condition. However, enhancement in this folding efficiency and folding rate differs across nOMPs. The kinetics of nOMPs that exhibit fast folding rates are not enhanced further by BamA (including PagP‐Ec in PC vesicles). Mismatch or alteration in the β‐signal, and more specifically in the Aro‐Xxx‐Aro motif present at the C‐terminus, greatly impacts and inhibits nOMP biogenesis by hampering the recognition process of OMPs by the BAM complex. ²⁵ The nOMPs studied here have nearly identical β‐signals (Figure S1), despite which the folding efficiency and BamA‐mediated increase in folding rate are different across the nOMPs (Table 2).

Our findings also suggest that the presence of PE and BamA alter the folding rates of each nOMP. Owing to the absence of a clear correlation between both elements across the nOMPs, we compared the folding flux (measured by the reaction amplitudes) of each reaction in both membranes and with(out) BamA (Figure S7). Interestingly, we find that the results are largely similar between PC and PE‐doped membranes. Additionally, the presence of BamA results in only a marginal enhancement in the folding flux, and the total molecules attaining the folded state (compare Figure S7a,b). While PE membranes retard the assembly rates of the nOMPs, we deduce that the variation in the extent to which BamA accelerates folding for some substrates depends on how quickly the primary sequence is able to attain the folded native conformation. A similar conclusion is evident when the efficiency across all nOMPs is compared 24 hr postfolding (t ₂₄, Figure S8). We observe that in both lipidic conditions, the presence of BamA does not significantly increase the folding efficiency of each nOMP to achieve near‐complete folding.

Put together, our experimental results establish that the extent to which each nOMP folds, and this assembly rate, are determined primarily by the intrinsic energetics coded in its primary sequence. Factors such as target membrane environment and the presence of a chaperone regulate the rate of this nOMP folding. Our findings with the eight‐stranded β‐barrel substrates are in good agreement with the membrane thinning model for BamA‐assisted OMP folding, ⁸ , ¹⁸ , ³⁸ since the extent of acceleration by BamA that we observe (Figure 3) and the folding efficiency (Figure S8) are not similar across the various nOMPs used in our study. Here, nascent polypeptides with the intrinsic ability to adopt a folding nucleus rapidly, such as OmpX and tOmpA, achieve the final folded active confirmation in shorter folding timescales with higher folding efficiency. In PE membranes that can impose kinetic retardation, the membrane defect (thinning) induced by BamA facilitates the spontaneous folding of these OMPs into the membrane. Whether the differences we observe in the folding timescales for the different OMPs (despite their largely similar size) relates to the function these structures are intended to perform in vivo, or their thermodynamic stability, remains to be addressed. Additionally, for OMPs such as Ail that are aggregation prone, the primary function of BamA is to rescue the nascent polypeptide from nonspecific association in the periplasm.

3. DISCUSSION

The in vivo biogenesis of nOMPs is enabled by the outer membrane molecular chaperone called the BAM complex, along with assistance from periplasmic holdases SurA and Skp. ⁴ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ The core transmembrane protein of the BAM complex, namely, BamA is sufficient for assembly of substrate OMPs in vitro. One of the facets that has remained ambiguous is the extent to which BamA contributes to the folding kinetics of smaller nOMPs, the attributes of BamA–nOMP interplay dynamics, and whether this chaperone exhibits species specificity. Our results, along with findings from previous studies, suggest that for nOMPs which assemble into eight‐stranded β‐barrels, BamA primarily facilitates its insertion and folding into the membrane by functioning as a promiscuous molecular membrane chaperone, and does not appear to influence the nOMP folding pathway. This process is species independent, with BamA enhancing the intrinsic folding flux of most nOMPs. With E. coli nOMPs that intrinsically possess higher folding kinetics (such as OmpX and tOmpA), the folding efficiency is not substantially enhanced by BamA. Instead, BamA is likely to enhance the folding flux by assisting these nOMPs overcome the kinetic barrier imposed by PE‐rich membranes in vivo. ⁹ , ¹⁶ , ¹⁸ , ¹⁹ , ²⁰ For nOMPs that exhibit slower intrinsic folding rates and are prone to aggregation (such as Ail‐Yp), BamA significantly accelerates both the folding rate and efficiency. BamA achieves this by increasing the folding flux and concomitantly decreasing the flux through aggregation. Our observations explain how the nonselective nature of BamA allows this protein to assemble diverse β‐barrels ranging from eight‐stranded to 36‐stranded scaffolds ⁷ in the bacterial outer membrane, by augmenting the nOMP's intrinsic folding ability (summarized in Figure 4).

FIGURE 4.

Effect of BamA on nascent outer membrane protein (nOMP) assembly. The presence of prefolded BamA in 14‐C 1,2‐dimyristoyl‐sn‐glycero‐3‐phosphocholine (DMPC) (left) and PCPE (right) large unilamellar vesicles (LUVs) differentially affects the folding rate of each eight‐stranded nOMP, irrespective of its origin. Based on our findings, we propose that BamA is a promiscuous chaperone, which nonspecifically accelerates the intrinsic folding ability of an nOMP (shown by green upward arrow), particularly in phosphoethanolamine (PE)‐doped membranes. Additionally, BamA lowers the flux through the aggregation pathway in Ail (red downward arrow). Our results also indicate that membrane thinning at strands β1 and β16 of this membrane chaperone ⁹ , ¹² , ¹⁶ , ¹⁸ , ¹⁹ , ²⁶ supports the insertion and assembly of small eight‐stranded nOMPs in membrane bilayers through a generalized pathway

Of particular interest is to compare our findings for E. coli OmpX and tOmpA in 14‐C membranes, with previous measurements in identical experimental conditions but using 10‐C lipids. ¹⁹ For example, the unassisted folding of OmpX and tOmpA is both fast and of comparable rates in 10‐C PC doped with 20% 10‐C PE (k _fast for OmpX = 0.0015 s⁻¹; tOmpA = 0.0096 s⁻¹). The addition of BamA caused a 10‐fold increase in the folding rate only for OmpX (k _fast = 0.016 s⁻¹), while the fast folding kinetics of tOmpA remained unaffected (k _fast = 0.0083 s⁻¹). Increasing the chain length to 14‐C (PCPE) provides us with an interesting outcome. Here, the unassisted folding rates for both proteins (k _obs for OmpX = 0.60 hr⁻¹; tOmpA = 0.32 hr⁻¹) in PCPE membranes show a 3.7‐fold and 3.6‐fold increase for OmpX and tOmpA, respectively, in the presence of BamA (k _obs for OmpX = 2.23 hr⁻¹; tOmpA = 1.15 hr⁻¹). Based on these comparisons, we propose that the bilayer membrane thickness plays a significant role in determining BamA‐assisted assembly rates for similar nOMP polypeptides.

An increase in bilayer thickness and the associated packing density of longer hydrocarbon chains retards the assembly rates of nOMPs. Previous measurements of nOMP assembly rates in thinner bilayers have reported faster and multiple folding rates, ⁹ , ¹⁶ , ¹⁸ , ¹⁹ suggesting that the hydrocarbon chain length, which contributes to bilayer fluidity and bilayer defects, alters the nOMP assembly rate. For example, comparison of the folding kinetics we obtain for tOmpA and OmpX in 14‐C membranes with previous studies in 10‐C and 12‐C membranes, ¹⁸ , ¹⁹ , ²⁰ , ²⁶ clearly shows that a thicker bilayer retards nOMP folding. Thinner hydrocarbon bilayers support faster folding and membrane insertion (completed in ≤15 min in 10‐C membranes ¹⁹ ) and yield two assembly rates (k _fast followed by k _slow), while the folding in thicker 14‐C bilayers is substantially slower (~12 hr), and follows a single rate (k _obs) (Figure 2, S6, 4). Similarly, the time required for tOmpA to attain 50% folding increases from 7.0 ± 0.3 min to 276.0 ± 6.0 min when the lipid chain length increases from 12‐C to 14‐C, ¹⁸ highlighting the important role of the kinetic barrier induced by the membrane architecture in retarding nOMP folding. The PE‐enriched bacterial outer membrane is also known to induce substantial kinetic retardation to nOMP folding ¹⁹ , ²⁰ , ²⁶ Our observations demonstrate that the retardation in the nOMP folding process by thicker bilayers supersedes the effect of PE. Our findings further highlight the dependence of nOMPs on BamA to fold into PE‐rich thicker membranes through a general mechanism irrespective of the substrate (Figure 4), wherein BamA accelerates the overall process by recognizing only the β‐signal motif. This observation is particularly relevant in vivo, as the thickness of the outer membrane (in addition to the headgroup characteristics) would mechanically retard nOMP assembly. The ability of BamA to introduce bilayer defects would therefore be vital in nOMP assembly in vivo, especially when the hydrocarbon chain lengths are sufficient to decelerate the β‐barrel folding process and promote off‐pathway aggregation.

The BamA structure, ¹¹ , ¹³ previous studies on BamA mediated OMP folding, ⁹ , ¹⁶ , ¹⁸ , ¹⁹ , ²⁰ , ²³ , ²⁴ , ³⁶ , ³⁷ , ³⁸ , ³⁹ and our findings from the assisted folding process together emphasize the functional importance of this chaperone in scavenging nOMPs, accelerating the nOMP folding process, and enhancing folding efficiency by allowing the nOMP overcome the kinetic retardation imposed by the high lateral packing in the membrane architecture. ⁴⁰ BamA‐Ec can recognize nOMPs from different species and help assemble these substrates with comparable (or better) efficiency in both PC and PE membranes. Our finding highlights the evolutionary conservation of this vital folding machinery across gram‐negative bacteria, and supports the general mechanism of BAM‐mediated biogenesis of nOMPs. We propose that the trimolecular interaction dynamics between the bilayer (composition and thickness), ⁴⁰ BamA (membrane thinning near strands β1 and β16 of the asymmetric BamA structure), ¹¹ , ¹³ and the nOMP sequence, are rate‐limiting steps in the biogenesis of OMPs in vivo. However, the dynamics of this interaction triggers and culminates in the spontaneous insertion and folding of eight‐stranded β‐barrel scaffolds, in biologically relevant timeframes (Figure 4). Our data identifies a key role of the nOMP primary protein sequence in influencing the β‐barrel folding and assembly rate in the membrane.

The primary sequence emerges as the foremost deciding element on the ability of nOMPs to fold into bilayer membranes. In other words, the extent to which nOMP folding is accelerated in the presence of BamA depends largely on the intrinsic ability of the protein primary sequence to spontaneously attain the final folded conformation in the bilayer membrane (Figure 4). The various molecular mechanisms of how BamA functions have been discussed. One such model is the budding (threading) mechanism involving opening of the lateral BamA gate ¹³ , ²² , ²³ , ⁴⁷ whereas another plausible swing mechanism has recently been elucidated in vivo using a 12‐stranded β‐barrel as substrate. ²⁴ Our results with small 8–stranded β‐barrel are, however, in good agreement with the membrane‐thinning (assisted) model for OMP biogenesis. ⁸ , ¹⁸ , ²⁰ , ³⁸ We propose that for eight‐stranded OMPs with the intrinsic ability for spontaneous self‐assembly, BamA functions as a freeway gatekeeper, facilitating entry of the nOMP into the outer membrane by lowering the kinetic barrier imposed by the lipid bilayer.

Although bacterial nOMPs are capable of attaining their native functional conformation in membranes in vitro, the time scales required to achieve this is not biologically relevant. ¹⁷ , ²⁰ , ³³ Various factors play antagonistic roles in the in vivo assembly of nOMPs into the outer membrane, including the kinetic retardation caused by the chemical nature of the lipid headgroup, and lipid bilayer thickness. ¹⁹ , ²⁶ Our study additionally establishes the interplay dynamics that exists between chemical nature of the amino acid residues of the nOMP sequence, the lipid‐mediated physiochemical barrier, and the assistance of BamA, all of which critically determine the nOMP assembly time and folding efficiency. Hence, transmembrane β‐barrel proteins with the obligatory requirement of chaperones for in vivo assembly are regulated by the same thermodynamic principles that govern the folding kinetics of nearly all soluble proteins. ²⁹ Our observations contribute to previous findings in this field, by highlighting the requirement of the evolutionarily conserved BAM complex in the outer membrane as a species‐ and substrate‐independent promiscuous chaperone which assists nearly all nOMPs overcome the kinetic barrier for spontaneous and correct assembly into the membrane. Further studies directed toward the assembly process of larger β‐barrels will reveal mechanistic insights on the magnitude of substrate‐specific contributions of membrane chaperones in both bacteria and mitochondrial outer membranes.

4. EXPERIMENTAL METHODS

4.1. Cloning and expression of BamA and OMPs from various sources

Full‐length bamA gene from E. coli (five POTRA domains and C‐terminal transmembrane domain) was cloned into pET15b without the signal sequence. The substrate OMPs used in the study (Table 1) were cloned into pET3a without the signal sequence. For OmpA, only the transmembrane domain (residues 22–192) was cloned. The plasmids were used to transform E. coli BL21 (DE3) cells, and the proteins were overexpressed as inclusion bodies. The inclusion bodies were isolated using reported protocols and then further purified on an anion exchange column under denaturing conditions of 8.0 M urea. ³² , ⁴⁸ Urea was removed by extensive dialysis, and proteins in their purified form were stored at −86°C as lyophilized powders. The lyophilized proteins were checked on 12% SDS‐PAGE before use. Note that the N‐terminal His₆ tag and thrombin cleavage site were retained in the full‐length BamA protein, while all substrate OMPs were tag‐free.

4.2. Preparation of DMPC and DMPE‐doped DMPC LUVs

DMPC (14:0 PC) and DMPE (14:0 PE) lipids were obtained from Avanti Polar Lipids as solvent suspensions. Appropriate volumes of DMPC and DMPC +20% DMPE (PCPE) were aliquoted in vials and dried under a gentle stream of dry nitrogen gas. The dried film was flash‐frozen in liquid nitrogen, lyophilized, and stored at −86°C until use. The dried lipid film was hydrated by resuspending in 20 mM borate buffer (pH 10.0) to achieve a final lipid concentration of 10 mM, by alternating heating and vortexing cycles at 42°C for 30–60 min until a uniform suspension was obtained. LUVs of 100 nm size were prepared by extruding the resuspended lipid ~20–30 times at 42°C through a 0.1 μm polycarbonate membrane using an Avanti mini‐extruder.

4.3. Folding of BamA into LUVs

Unfolded BamA protein dissolved in 8 M urea (100 μM) was diluted 25‐fold into the folding buffer to a final concentration of 4 μM BamA in 3.2 mM LUV (DMPC or PCPE), 2 mM EDTA, 1 M urea and 20 mM borate buffer pH 10, and incubated overnight with constant rotation at 35°C. Folding efficiency of BamA was assessed using gel mobility shift on cold 10% Tricine‐PAGE (described below).

4.4. Folding of nOMPs into LUVs

Folding kinetics of the unfolded nOMP was measured in the absence or presence of prefolded BamA in DMPC and PCPE LUVs at 35°C. The purified nOMP powders (Table 1) was first dissolved in 8.0 M urea and then quantified to obtain a 100 μM solution. Folding was initiated by rapid 25‐fold dilution of this unfolded nOMP, into the folding reaction containing LUVs. The final reaction contained 4 μM OMP, 3.2 mM LUV (DMPC or PCPE), 2 mM EDTA, 1 M urea, and 20 mM borate buffer pH 10. The folding sample was incubated at 35°C with constant rotational mixing. To study the effects of BamA on OMP folding, 100 μM nOMP in 8.0 M urea was rapidly diluted 50‐fold into the folding mixture containing prefolded BamA, so as to obtain a final concentration of 2 μM prefolded BamA and 2 μM nOMP in 1.6 mM LUV (DMPC or PCPE), 2 mM EDTA, 1 M urea, and 20 mM borate buffer (pH 10). Defined aliquots were drawn at specific time intervals during the folding process, and the reaction was quenched with 5X SDS gel loading dye. The quenched (unboiled) samples were stored at 4°C. For the unfolded control, aliquots quenched with the SDS gel loading dye were boiled for 5 min at 100°C before subjecting it to electrophoresis. Two additional aliquots of folded BamA were collected to represent BamA folded and unfolded controls. The unboiled samples along with the controls were loaded onto 10% Tricine‐PAGE (Tris‐Tricine–SDS‐PAGE), and resolved at constant voltage (200 V) for 1.5 hr at 4°C. Protein bands were visualized by staining with Coomassie brilliant blue R‐250.

4.5. Electrophoretic mobility shift assay and quantification of folded bands

The folding efficiency of OMPs in DMPC and PCPE LUVs was assessed by electrophoretic mobility shift assay on cold 10% Tricine‐PAGE. The folded OMP exhibits faster gel mobility as compared to the unfolded protein, in Tricine‐PAGE. Folding efficiency as a function of time was calculated by densitometry analysis using Multi Gauge v2.3, as follows:

$$\mathsf{\text{Fraction folded}}=\left[\mathsf{\text{Folded band intensity}}\right]/\left[\mathsf{\text{folded band intensity}}+\mathsf{\text{unfolded band intensity}}\right]$$

With this calculation, artifacts from sample loading or staining efficiency were minimized. PagP‐Ec shows accumulation of an additional species during the folding (see Figures S2 and S3). To retain consistency across the analyses, only the folded and unfolded bands were used in the calculation. The fraction folded was converted to % folded and plotted as a function of time to monitor the folding process in the presence or absence of prefolded BamA. The folding efficiency of the OMP was also assessed as:

$$\mathsf{\text{Folding efficiency}}=\left[\mathsf{\text{Fraction folded}}\mathsf{at}{\mathsf{t}}_{\mathsf{24}}\right]\times \mathsf{100}$$

Here, the term folding efficiency denotes the fraction of OMP folded at 24 hr (t ₂₄) after initiation of the folding process.

4.6. Fitting of OMP folding and calculation of folding kinetics

The % folded was calculated from at least three independent folding reactions for unassisted and BamA‐assisted folding of each OMP in DMPC and PCPE LUVs. Here, the time course of the folding reaction (calculated as % folded) was plotted, and each plot was fitted to the exponential rise to maximum function (provided below) to determine the observed rate of folding for each independent reaction.

$$\mathsf{f}={\mathsf{y}}_{\mathsf{0}}+\mathsf{A}*\left(\mathsf{1}–{\mathsf{e}}^{\left(–\mathsf{k}*\mathsf{t}\right)}\right)$$

Here, f represents the % folded at a given time t, y ₀ + A denotes the total amplitude associated with the rate constant, and k (k _obs) represents the observed rate of folding of the substrate OMP. Data obtained for the first 48 hr was used for all nOMPs (108 hr for PagP‐St). The rates derived from three independent reactions in each lipidic condition (DMPC and PCPE) was used to calculate the mean (and SD), and also used for statistical analysis (Student's t test; data in Figure 2). In addition, the % folded was also plotted for each time point as mean (and SD) of three independent experiments (Figures S4 and S5), and fitted to the exponential rise function to obtain the mean rate of folding (Figure S6; errors in the rate of folding represent the goodness of fit).

CONFLICT OF INTEREST

Both authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

Radhakrishnan Mahalakshmi: Designed the research. Pankaj B. Tiwari: Performed the research. Both authors analyzed the data. Radhakrishnan Mahalakshmi: Wrote the paper with input from Pankaj B. Tiwari. Both authors have approved the final version of the manuscript.

Supporting information

Appendix S1: Supporting information

Tiwari PB, Mahalakshmi R. Interplay of protein primary sequence, lipid membrane, and chaperone in β‐barrel assembly. Protein Science. 2021;30:624–637. 10.1002/pro.4022