Article pubs.acs.org/biochemistry
Concatemers of Outer Membrane Protein A Take Detours in the Folding Landscape Kell K. Andersen,† Brian Vad, Sahar Omer, and Daniel E. Otzen* iNANO and Department of Molecular Biology and Genetics, Aarhus University, Gustav Wieds Vej 14, DK-8000 Aarhus C, Denmark ABSTRACT: Outer membrane protein A (OmpA) is the most abundant protein in the outer membrane of Escherichia coli. The N-terminal domain forms an eight-stranded membraneembedded β-barrel that is widely used as a model protein for in vitro folding into the membrane and into surfactant micelles. Under conditions that include a low surfactant concentration, OmpA can form stable higher-order structures by intermolecular association. Other β-barrel membrane proteins also associate to form noncovalently linked trimers in vivo. This inspired us to test how topological constraints imposed by intramolecular links between individual OmpA molecules affect this process. Here we report on the properties of concatemers consisting of two and three copies of the transmembrane part of OmpA. Both concatemers could be folded to a native state in surfactant micelles according to spectroscopy and electrophoretic band shifts. This native state had the same thermodynamic stability against chemical denaturation as the original OmpA. Above 1.5 M GdmCl, concatemerization increased both refolding and unfolding rates, which we attribute to entropic effects. However, below 1.5 M GdmCl, folding kinetics were 2−3 orders of magnitude slower and more complex, involving a greater degree of parallel folding steps and species that could be classified as off-pathway. Only OmpA2 could quantitatively be folded into vesicles (though to an extent lower than that of OmpA), while OmpA3 formed three species with different levels of folding. Thus, close spatial and sequential proximity of OmpA domains on the same polypeptide chain have a strong tendency to trap the protein in different misfolded states.
T
exit through the lateral opening of BamA, while loops and extracellular domains exit through a top substrate exit pore.11,12 Once inside the membrane, the β-strands will most likely have to assemble by themselves, though this is presumably aided by the controlled vectorial release from the BAM complex, starting from the N-terminal part of the OMP. Despite the existence of such an intricate folding machinery provided by periplasmic chaperones and the BAM complex, many OMPs fold spontaneously in vitro.13 These in vitro studies are aided by the fact that most OMPs, unlike their more hydrophobic α-helical membrane protein counterparts, are soluble in chemical denaturants such as urea and guanidinium chloride (GdmCl), so that folding can be accomplished by simply diluting out the denaturant-unfolded OMP in the presence of a suitable membranelike environment. There are significant kinetic barriers to folding into phospholipid vesicles, so that careful screening is often required to identify truly reversible folding−unfolding vesicle-insertion conditions.14,15 Folding of OmpA into membranes is a highly cooperative process that involves concomitant formation of both secondary
he membrane-embedded parts of all known bacterial outer membrane proteins (OMPs) form β-barrel structures, in which polar amide and carbonyl groups of neighboring β-strands are connected by hydrogen bonds. Formation of these β-barrels is a highly cooperative process; individual β-strands are not stable in the membrane because they cannot form hydrogen bonds within one strand.1−3 Accordingly, in vivo insertion of OMP into the outer membrane is a highly regulated process.4 After synthesis in the cytosol, OMPs are transported into the periplasm through SecYEG and escorted in the unfolded state via chaperones such as SurA and Skp5−7 to the outer membrane, where insertion occurs via the Omp85 or BAM (β-barrel-assembling machinery) complex.8−10 This complex consists of five subunits, of which BamA’s Cterminal domain forms a 16-strand membrane-embedded βbarrel structure while its N-terminal domains form a circular structure in the periplasm with the lipoproteins BamB−E;8 the lipoproteins are also linked to the membrane via lipidated Cys residues. Rotation of the lipoproteins is proposed to lead to a scissorlike movement of the first six β-strands of BamA, allowing it to switch from a periplasmic (bottom) open to a laterally open state and thus promote insertion of nascent OMP chains into the membrane.8 Consistent with this, a “budding” model has been proposed, in which membrane-bound β-strands © XXXX American Chemical Society
Received: November 12, 2016 Revised: December 1, 2016 Published: December 2, 2016 A
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry and tertiary structure,16 is envisaged to occur as a gradual insertion into the hydrocarbon layer, and usually requires minutes to hours to reach completion.17 However, folding into surfactant micelles is considerably easier and usually completed more rapidly.18 Refolding of OMPs into micelles has an additional advantage in that it can be performed over a broad denaturant concentration range, allowing extension to unfolding conditions. This allows us to dissect OMP folding with the analytical tools and formal framework developed over decades for the folding of globular proteins.19−21 Such an approach has led to the complete kinetic description of the folding and unfolding of the 176-residue transmembrane part of the archetypal OMP OmpA from Escherichia coli,18 a protein used as a model system for OMP folding for decades. Folding in surfactant micelles turned out to be remarkably straightforward and involved only three states, namely, GdmCl-denatured state D, a transient intermediate I that accumulates only at very low GdmCl concentrations, and native state N. Thus, under these conditions (high concentrations of surfactant, no phospholipids, and thorough denaturation in GdmCl), there is no indication that folding is complicated by the accumulation of kinetically trapped species. However, OmpA may form kinetically trapped states under other conditions. For example, cleaving OmpA into complementary pairs, unfolding the fragments in denaturant, and then allowing them to associate and refold by diluting out denaturant lead to a dramatic decrease in folding yields and very slow folding rates,22 demonstrating that partial unfolding or destabilization of the membrane protein can promote nonproductive non-native interactions. Furthermore, refolding at surfactant concentrations around the critical micelle concentration (i.e., low concentrations of surfactant micelles) promotes the formation of higher-order oligomeric species of OmpA that can exist in folded and (partially) unfolded states.23 Many of these states persist for hours, even at elevated temperatures. Similar phenomena are seen when OmpA refolds at relatively low concentrations of phospholipid vesicles.23 We have ascribed these oligomeric species to the formation of nativelike contacts between complementary β-strands from different polypeptide chains, promoted by the high local concentration of OmpA in the micellar or lipid environment. This suggests an intriguing degree of intermolecular cross-talk by membrane proteins when they fold in the spatially and energetically restrictive environment of a membrane or micelle and highlights the importance of biological control mechanisms for minimizing these potentially deleterious contacts during folding. We reasoned that such contacts are likely to be facilitated by the existence of multiple free termini, through either fragments or multiple copies of single transmembrane domains, which gives the proteins greater freedom in sampling and selecting different intermolecular topologies. Accordingly, reducing the number of free termini and linking several copies of an OMP on the same polypeptide chain might be expected to restrict the degree to which the polypeptide chain could explore conformational space as well as increasing the local concentration of individual folding domains. This should also impact the folding process by providing permanent proximity to potential folding partners, though it is not a priori obvious what the extent of the impact would be. The higher-order species formed by monomeric OmpA at low surfactant concentrations gradually disappear at higher denaturant concentrations and are unfeasible to study by spectroscopy because of their low population. In contrast, concatemers persist irrespective of denaturant concentration,
allowing us to extend the formalism of protein folding analysis to these constructs and thus gain mechanistic insight into how these alterations in connectivity affect membrane protein folding in vitro. There is a rich body of work on the folding of naturally occurring soluble protein concatemers of immunoglobulin-like domains24 and Ankyrin repeat25 that reveals important insights into the evolutionary constraints on such constructs, but no such studies have been performed on membrane proteins. We believe that such studies may provide important insight into the existence of intrinsic folding traps in the membrane protein conformational landscape that are biologically suppressed by appropriate use of chaperones and folding machinery. To address this scenario, we here describe a study in which we have prepared concatemeric constructs of the β-barrel domain of OmpA (TM-OmpA) through fusions at the DNA level. We have linked either two or three copies of the gene encoding TM-OmpA with linker sequences consisting of 12residue Gly-Ser stretch, leading to OmpA2 or OmpA3, respectively, and have analyzed the folding behavior in terms of both the thermodynamic stability and the mechanism of folding. Our data suggest that the folding behavior of TMOmpA is significantly skewed by these concatemeric structures, leading at low denaturant concentrations (like the complementary fragment pairs) to strongly retarded folding as well as the formation of off-pathway intermediates to an extent that increases with the number of copies of OmpA. Only at higher denaturant concentrations does this give rise to faster folding and unfolding, which we attribute to entropic effects. Thus, topological restrictions in combination with high local concentrations compromise efficient and direct formation of β-barrel outer membrane proteins.
■
MATERIALS AND METHODS Materials. Urea, EDTA, and proteinase K were from Sigma. Glycine, SDS, and guanidinium chloride (GdmCl) were from AppliChem. 1-Palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phospho(1′-racglycerol) (POPG), 1,2-dilauroyl-sn-glycero-3-phosphocholine (DLPC), and 1,2-dilauroyl-sn-glycero-3-phospho(1′-rac-glycerol) (DLPG) were from Avanti Polar Lipids. n-Octyl β-Dmaltopyranoside (OM) was from Anatrace. Preparation of OmpA2 and OmpA3. TM-OmpA consists of residues 1−176 of the original full-length OmpA from E. coli (NCBI reference sequence YP_489229.1, GI 388477041) with an eight-residue extension (RSH6) to facilitate purification on a Ni-NTA column. OmpA2 consists of a starter residue M, 176 residues from the transmembrane part of OmpA (residues 22−197 in the original sequence) followed by the 12-residue linker GGSGGGSGGGSG, the same 176 residues, and the eight-residue extension RSH6 (373 residues in all). OmpA3 consists of a starter residue M, the 176 residues, linker GGSGGGSGGGSG, 176 residues, linker GGSGGGSGGGSG, 176 residues, and the extension RSH6 (561 residues in all). The two constructs were cloned into expression vector pET22b using NdeI and XhoI restriction sites and standard protocols. Proteins were expressed as inclusion bodies in E. coli and purified in 8 M urea using anion exchange chromatography as described previously.18 Refolding of OmpA Constructs into Surfactant. Refolding buffer consisted of 10 mM Gly, 2 mM EDTA (pH 10), and different concentrations of octyl maltoside (OM). OmpA, OmpA2, and OmpA3 were refolded by 16-fold dilution B
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry from a stock of ∼3 mg/mL protein in 8 M urea into refolding buffer and different concentrations of OM, incubated at 37 °C for 24 h, and analyzed by sodium dodecyl sulfate− polyacrylamide gel electrophoresis (SDS−PAGE). Samples were not boiled prior to being loaded on the gel unless stated. Proteolysis Experiments. TM-OmpA, OmpA2, and OmpA3 were diluted into refolding buffer [250 mM OM, 10 mM glycine (pH 10), and 2 mM EDTA] to final concentrations of 1 mg/mL and incubated at room temperature for 4 days. Proteinase K stock solutions (1 mg/mL) were made by dissolving the protein in 50 mM Tris-HCl (pH 8) and 10 mM CaCl2. Proteinase K was added to the folded protein solution to a final concentration of 50 μg/mL, and the samples were incubated at 37 °C for 1 h. Proteolysis was stopped by the addition of phenylmethanesulfonyl fluoride to a final concentration of 5 mM. Samples were analyzed by SDS− PAGE. Chemical Denaturation of OmpA Concatemers. Unfolding Experiments. Proteins were initially folded by incubation at 0.25 mg/mL in 250 mM octyl maltoside and glycine buffer [10 mM glycine (pH 10) and 2 mM EDTA] at 37 °C for 2 days. They were then diluted 5-fold into glycine buffer, 250 mM OM, and appropriate amounts of GdmCl. The samples were equilibrated for 4 days at 25 °C and analyzed by monitoring Trp fluorescence at 25 °C using a PerkinElmer LS55 fluorimeter. The excitation wavelength was 295 nm, and emission was measured from 320 to 400 nm at a scanning speed of 100 nm/min. Excitation and emission slit widths were 5 and 6 nm, respectively. Refolding Experiments. Protein (3 mg/mL) in 8 M urea was diluted 60-fold into glycine buffer, 250 mM OM, and different concentrations of GdmCl. The samples were equilibrated for 4 days at room temperature and analyzed the same way as in the unfolding experiments. Data were analyzed by plotting the ratio of the intensities at 340 and 361 nm (corresponding to peak intensities of the folded and unfolded states, respectively) versus GdmCl concentration and fitting the data to a two-state unfolding model18 using the following equation:
of SDS to the protein. Accordingly, the sample loading buffer was diluted from 6× to 2× in a 10% SDS solution, giving a SDS concentration of 350 mM. This sample loading buffer was then mixed in a 1:1 ratio with the protein mixture, making the final SDS concentration 175 mM. The samples were then run on SDS−PAGE gels, and the bands were quantitated using ImageJ. Kinetics of OmpA Folding and Unfolding in OM Micelles and GdmCl Monitored by Trp Fluorescence. Measurements were taken with a Cary-Varian Eclipse fluorimeter in 10 mm quartz cuvettes with magnetic stirring. Excitation and emission wavelengths were 280 and 333 nm, respectively. The excitation slit width was 2.5 nm, and the emission slit width was 5 nm. The detector voltage was set to medium (600 V). In unfolding experiments, 0.6 mg/mL OmpA3 and OmpA2 were initially folded into glycine buffer and 250 mM OM and incubated at 37 °C for 2 days. Contents of cuvettes containing glycine buffer with OM and appropriate GdmCl concentrations were equilibrated at 25 °C using the instrument thermostat. Folded OmpA2 and OmpA3 were then quickly diluted 12-fold to final concentrations of 0.05 mg/mL, and unfolding was followed by fluorescence. In refolding experiments, glycine buffer and 250 mM OM with and without GdmCl were equilibrated in cuvettes. Concentrated and unfolded OmpA2 and OmpA3 were then added to the same final concentrations that were used for unfolding experiments, and refolding was followed by fluorescence. In the range of 0− 1.8 M GdmCl, refolding data were fitted to a triple-exponential decay: signal = Amp1 × exp( −k1t ) + Amp2 × exp(−k 2t ) + Amp3 × exp(−k 3t ) + c
where Amp represents the amplitude, k is the associated rate constant for a given decay phase, and c is an offset constant. Refolding rates at 1.9−2.2 M GdmCl and all unfolding data were fitted to a double-exponential decay equation in which the term containing Amp3 was omitted from eq 1. Analysis of Kinetic Data. The off-pathway folding scheme (Scheme 2) was analyzed according to the following equation:26
signal ratio = ⎡⎣αN + βN[GdmCl] + (αD + βD[GdmCl]) 50%
)⎤
50%
)⎤
× 10mD‐N([GdmCl] − [GdmCl]
+ 10mD‐N([GdmCl] − [GdmCl]
⎡ ⎦ ⎣1 ⎦
(2)
⎛ 10 log k fwater + mf [GdmCl] log kobs = log⎜⎜ water ⎝ 1 + 10 log KC + mC[GdmCl] (1)
water
+
where αN and αD are the ratios of the native and denatured states, respectively, βN and βD describe their linear dependence on [GdmCl], [GdmCl]50% is the midpoint of denaturation where the native and denatured states are equally populated, and mD‑N is the cooperativity of unfolding that describes the linear dependency of the log of the equilibrium constant of unfolding on denaturant concentration. Kinetic Analysis of OmpA Folding in OM Micelles Analyzed by SDS−PAGE. Unfolded OmpA2 and OmpA3 were added into refolding buffer containing 250 mM OM, 10 mM glycine (pH 10), and 2 mM EDTA to final concentrations of 1 mg/mL preceded by immediate mixing. All folding kinetics of OmpA2 and OmpA3 were conducted at 25 °C. At different time points, aliquots of the reaction mixture were immediately mixed with reducing SDS sample buffer to arrest folding and subsequently analyzed by SDS−PAGE without prior boiling (unless otherwise stated). It was found that the high concentration of surfactant (OM) interfered with the binding
10 log k 2
1 + 10−log
+ m2[GdkCl]
K Cwater − mC[GdmCl]
water
+ 10 log k u
⎞ ⎟ ⎠
+ m u[GdmCl]⎟
(3)
where kobs is the observed rate constant for folding or unfolding and the parameters mf, m2, mu, and mC describe how the microscopic parameters log kf, log k2, log ku, and log KC, respectively, depend linearly on GdmCl concentration. In practice, m2 is set to 0 because the chevron plot is essentially horizontal at low GdmCl concentrations. Folding of OmpA2 and OmpA3 in Phospholipid Vesicles. DLPC and DLPG were dissolved in chloroform and dried under a stream of nitrogen air to thin lipid films in glass vials. They were then dissolved in 10 mM glycine buffer (containing 2 mM EDTA) (pH 10) and mixed to give a final composition of 92.5% DLPC and 7.5% DLPG at a final concentration of 10 mM. After sonication for 10 min, the solution was centrifuged and incubated overnight at 4 °C. For folding experiments, TM-OmpA, OmpA2, and OmpA3 were diluted out into a lipid solution from a 5 mg/mL stock solution C
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 1. OmpA2 and OmpA3 can be refolded at micellar concentrations of surfactant. (A) OmpA2 and (B) OmpA3 were diluted 16-fold from 8 M urea into refolding buffer with 0−30 mM octyl maltoside (OM), incubated at 37 °C for 24 h, and analyzed by SDS−PAGE without samples being boiled before being loaded (left panels). In the right panels, the intensities of bands corresponding to the unfolded and folded states were quantified by densitometric analysis and plotted versus [OM]. Dotted lines indicate the critical micelle concentration of OM (18 mM).
Figure 2. OmpA2 and OmpA3 refold to concatemers of OmpA. (A) Like TM-OmpA, folded OmpA2 and OmpA2 both undergo a band shift upon being boiled in SDS sample loading buffer, leading to the denatured state. (B) Far-UV CD spectra of all three proteins after refolding contain the minimum around 215 nm characteristic of β-sheet proteins. In contrast, the denatured states in urea all show a random coil spectrum. (C) Proteinase K digestion of folded TM-OmpA, OmpA2, and OmpA3 leads to monomeric TM-OmpA in all cases. Unfolded states are completely degraded.
in 8 M urea to a final lipid concentration of 9.5 mM and protein concentrations of 0.23, 0.11, and 0.06 mg/mL, corresponding to lipid:protein ratios (on an OmpA monomer concentration scale) of 800:1, 1600:1, and 3200:1, respectively. The solutions were incubated at 25, 40, and 55 °C. The lipid solutions were equilibrated at these temperatures for 10−15 min before the protein was added. At different time points, 35 μL samples were removed, mixed with 7 μL of SDS loading buffer (7 μL) to quench folding, and run on SDS−PAGE gels without boiling. For experiments involving urea denaturation of lipid-folded OmpA concatemers, OmpA and OmpA3 were each diluted to a final concentration of 0.23 mg/mL in the presence of 200 nm vesicles of 92.5% DLPC and 7.5% DLPG (5 mg/mL), giving a molar protein:lipid ratio of ∼1:800 in 10 mM Gly buffer (pH
10) and 2 mM EDTA, and allowed to fold for 2 h at 40 °C. Samples were then mixed in a 1:2 ratio with different urea concentrations to give final concentrations of 0−6 M urea in steps of 0.6 M and incubated for 2 h at 20 °C before being analyzed by SDS−PAGE without boiling as described above.
■
RESULTS
Concatemers of OmpA Can Be Refolded in Surfactant Micelles and Consist of Independent Domains with the Same Stability as TM-OmpA. The purpose of this study was to investigate how covalent linking of several copies of the transmembrane domain of OmpA (TM-OmpA) would affect the folding of OmpA. The individual TM-OmpA domains were D
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 3. Denaturation of OmpA variants in GdmCl followed by the change in Trp fluorescence. TM-OmpA, OmpA2, and OmpA3 show the same stability toward unfolding in GdmCl with a denaturation midpoint around 2.2 M GdmCl, based on fits of eq 1 to the data. Furthermore, denaturation curves are fully superimposable regardless of whether the starting point is the denatured or refolded state. Data are summarized in Table 1.
separated by a 12-residue linker GGSGGGSGGGSG, where Gly was chosen to impart flexibility and Ser to impart a hydrophilic character. We first investigated whether concatemers consisting of two (OmpA2) or three (OmpA3) copies of TM-OmpA could be folded at all. We employed the conventional SDS−PAGE band shift assay, which exploits the fact that folded OmpA is resistant to unfolding in SDS (unless boiled) and therefore migrates faster than unfolded OmpA.27,28 Both OmpA2 and OmpA3 were obtained from inclusion bodies produced upon overexpression in E. coli and were solubilized and purified in 8 M urea. Folding was then attempted by diluting out the urea in an appropriate refolding medium. As refolding medium we used the nonionic surfactant octyl maltoside (OM), which has a critical micelle concentration of around 18 mM under our buffer conditions.18 Folding of singledomain TM-OmpA in surfactants requires the presence of surfactant micelles,29 and this is also the case for OmpA2 and OmpA3 (Figure 1). In both cases, we see a shift in the size of the dominant band around 20 mM OM. For OmpA2, the band shifts from ∼40 to ∼36 kDa, while for OmpA3, there is a shift from ∼62 kDa to a more fuzzy band centered around 55 kDa. The molecular weights of OmpA2 and OmpA3 are 40.3 and 60.3 kDa, respectively, which agree nicely with the estimated molecular weights of the denatured species from SDS−PAGE.
Once folded, both OmpA2 and OmpA3 can then be unfolded again by being heated in SDS−PAGE loading buffer (Figure 2A). Further proof of the folded nature of the two concatemers is provided by far-UV circular dichroism spectra (Figure 2B): in 8 M urea, they show the hallmarks of a random coil just like TM-OmpA, but when folded, they give rise to a minimum at 215 nm, just like folded TM-OmpA, which is characteristic of β-sheet structure (Figure 2B). When exposed to proteinase K, a broad-specificity protease that attacks all flexible or unfolded parts on a protein, folded TM-OmpA remains intact, while unfolded TM-OmpA is completely degraded; in the case of OmpA2 and OmpA3, the unfolded form is completely degraded while the folded form is cleaved to monomer size (Figure 2C). This is consistent with separate folding of each domain of OmpA. To further corroborate that each domain was folded separately from the other domain(s), we analyzed the stability of the OmpA constructs against unfolding in chemical denaturants. Micelle-embedded OmpA does not unfold reversibly in urea within the normal experimental time scale of hours to days, showing considerable hysteresis between unfolding and refolding.18 However, in guanidinium chloride (GdmCl), it is possible to obtain essentially superimposable unfolding curves regardless of whether the GdmCl titration E
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
concentrations, where the refolding intermediate transiently accumulates as the equilibrium constant KI = [I]/[D] becomes ≫1.18 Here we describe similar studies with OmpA2, where the data indicate a significantly more complex process. The raw Trp refolding (Figure 4A) and unfolding time profiles (Figure 4B) are best fitted to multiple-exponential decay phases. Three phases were needed for refolding in the range of 0−1.8 M GdmCl, while refolding at higher (1.9− 2.2M) GdmCl concentrations and all unfolding data were fitted to double-exponential decays. Unfolding rates increase greatly with denaturant concentration as shown by the time courses in Figure 4B; this leads to a simple linear relationship between the log of the two unfolding rate constants and the denaturant concentration (Figure 4C). However, the situation is more complex with the refolding rates. The slow rate constant was difficult to determine precisely because of its very long half-life (10−24 h), which made it difficult to reach completion within normal experimental time windows; furthermore, its amplitude constitutes ∼70% of the total amplitude at low (0−0.4 M) GdmCl concentrations but dwindles to zero in an essentially linear fashion between 0.4 and 2 M GdmCl (Figure 4D). [In contrast, the amplitudes for the two unfolding phases remain reasonably constant above ∼2.5 M GdmCl up to around 4.5 M GdmCl (see Figure 4E).] Accordingly, we have not included this phase in a detailed analysis. The picture is clearer for the two faster phases. Starting from 0 M GdmCl, we found the fast refolding rate constant k1 actually increases to a modest but measurable degree (∼0.7 log unit, corresponding to an increase in the rate constant of a factor of 5) as the GdmCl concentration increases to 1−1.2 M, after which it starts to decrease, reaching a minimum around 2−2.2 M GdmCl. This concentration range corresponds to the midpoint of the transition between N and D in the equilibrium denaturation diagram (Figure 3B), where the dynamics of exchange between the two states is slowest. A similar appearance is observed for the chevron plot of the relaxation phase associated with intermediate rate constant k2. The first question to ask is whether these different relaxation phases represent different steps of the same overall folding process (such as a serial D → I1 → I2 → N step, where the first step is fast, the next is intermediate in rate, and the third is slow) or simply represent three parallel routes to the native state. In other words, does N form as a single event or as three separate events? To be able to distinguish between serial and parallel folding steps, we need to be able to monitor the accumulation of N over time. This information is not provided directly by the Trp fluorescence time profiles but can be obtained via SDS−PAGE. Although this technique does not have the same high time resolution as Trp fluorescence, it provides a more direct measure of the formation of native (SDS-resistant) OmpA as well as other (partially) SDS-resistant OmpA species, because of their distinct migration behavior. A restriction of SDS−PAGE is that it is incompatible with GdmCl because of coprecipitation of GdmCl with SDS. Therefore, we have limited our analysis to the folding of OmpA2 at 0 M GdmCl, where we observe three different relaxation phases in the Trp time profiles (Figure 4C). Remarkably, we observe three different bands on the SDS−PAGE gel (Figure 5A). In addition to the D and N states, there is also a faint band of intermediate size between D and N that is observed from the earliest time point and gradually disappears over the next several hours. The faintness of the band makes it difficult to quantify and analyze according to a kinetic scheme, but its
curve starts from the folded or unfolded state, provided the samples are incubated for at least 4 days.18 For all three constructs, folding and/or unfolding could be monitored by the considerable red shift and the decrease in overall Trp fluorescence emission intensity accompanying unfolding (Figure 3A−C). As summarized in Table 1, all three constructs Table 1. Midpoints and Cooperativities of Unfolding of Different OmpA Constructs in GdmCla construct
initial stateb
TM-OmpA
N D N D N D
OmpA2 OmpA3
[GdmCl]50% (M)c 2.10 2.22 2.17 2.10 2.04 2.11
± ± ± ± ± ±
0.06 0.04 0.03 0.04 0.03 0.03
mD‑N (M−1)d 4.74 3.10 3.91 5.02 2.80 3.48
± ± ± ± ± ±
2.40 0.72 0.89 2.13 0.41 0.64
a
All folding performed in 250 mM OM, 10 mM glycine, and 2 mM EDTA (pH 10) at 25 °C. Data measured after incubation for 4 days. b State of OmpA before GdmCl was added. OmpA constructs were folded to N as described in Materials and Methods. The D state was OmpA purified in 8 M urea. cMidpoint of denaturation obtained by fitting eq 1 to the data.18 dCooperativity of denaturation obtained by fitting eq 1 to the data, describing how the log of the equilibrium constant of denaturation depends on GdmCl concentration.18
undergo an unfolding transition with a midpoint around 2.2 M GdmCl. The midpoints ([GdmCl]50%) from both refolding and unfolding curves and the parameter describing the cooperativity of unfolding (mD‑N) are identical within error for each construct, and there is no systematic difference among the three constructs. Thus, all our data indicate that concatemers of OmpA show the same structure and stability as monomeric TM-OmpA and there is no significant interaction among the different transmembrane domains. Chevron Plots of Folding of OmpA2 Indicate Two Parallel Triangular Folding−Unfolding Schemes. Having established that OmpA under equilibrium conditions folds to independent domains, we now turn to the kinetics of the folding process. We expect this to be affected to a much greater extent by the concatemerization of several OmpA domains. We monitor folding through the change in the protein’s endogenous Trp fluorescence over time. The same approach was used in our study of the folding and unfolding of singledomain TM-OmpA in OM micelles in the presence of 0−5 M GdmCl and 250 mM OM.18 The high OM concentration was needed to ensure that micelles were present throughout the range of denaturant concentrations. The TM-OmpA folding process yielded remarkably simple kinetics and could be modeled as a process in which TM-OmpA folds via one major folding intermediate I according to Scheme 1. This intermediate is only transiently populated during the folding reaction and is not observed by SDS−PAGE. We deduced the presence of this intermediate from TM-OmpA’s chevron plot (a plot of the logarithm of the folding and unfolding rate constants vs [GdmCl]). The plot is overall Vshaped but shows a slight rollover at the lowest GdmCl Scheme 1
F
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 4. Kinetics of folding and unfolding of OmpA2. (A) Change in Trp fluorescence as a function of time when OmpA2 is refolded by transfer from the urea-denatured state into 250 mM OM and different concentrations of GdmCl. The refolding curve at 0 M GdmCl has been displaced up by 30 units to improve visualization. The units of the right y-axis refer to the 2 M GdmCl refolding curve. (B) Time courses for unfolding of refolded OmpA2 into different concentrations of GdmCl. Note the logarithmic time axis and the much greater variation in time profiles for unfolding than for refolding curves. For the sake of clarity, the unfolding curve at 5 M GdmCl has been displaced down by 10 units. (C) Plots of obtained rate constants of folding and unfolding vs GdmCl concentration. Joined lines for the fast and intermediate phase show the best fit of a triangular folding pathway (Scheme 2, eq 3). Rate constants obtained by fitting a double- or triple-exponential decay (eq 2) to data in panels A and B. Data are summarized in Table 2. (D) Amplitudes of refolding at different GdmCl concentrations obtained from fits to data in panel A. (E) Amplitudes of unfolding as a function of GdmCl concentration. Only two unfolding relaxation phases are observed.
Armed with this insight, we now turn to the analysis of the observed rate constants, where we can analyze the fast and intermediate rate constants as separate and nominally independent pathways of folding. The complex behavior shown by the fast and intermediate phases contrasts with the simple rollover observed for TM-OmpA but is consistent with a triangular folding scheme involving a “detour” species C (Scheme 2). Such a model was previously proposed for protein S6 when it refolds in the presence of the strongly stabilizing salt sodium sulfate.26,30 In this scheme, denatured state D is in equilibrium with a collapsed species C. D can fold directly to N (rate constant kf) but can also collapse to C (equilibrium constant KC = [C]/ [D]), which folds to N sufficiently slowly (rate constant k2) that a significant fraction of C, depending on the denaturant concentration, will have time to convert back to D and fold directly to N in this way, provided kf > k2. It is necessary to include k2 in Scheme 2; otherwise, log kobs will decline linearly in the left part of the chevron plot, because the fraction of D available for folding will continue to decrease with a decrease in
existence illustrates that additional species form during the folding of the OmpA2 concatemer. Importantly, the increase in the intensity of the N state follows a time course that is consistent with the three-exponential decays observed in our Trp fluorescence experiments (Figure 4B). The expected rate constants for refolding of OmpA2 at 0 M GdmCl are obtained from the chevron plot in Figure 4C. When we fit a tripleexponential decay with locked values for these rate constants (corresponding to half-lives of 15.2, 103, and 1383 min for the fast, intermediate, and slow phases, respectively) and allow only the amplitudes to vary, we obtain a satisfactory fit, given amplitudes of 0.33 ± 0.05, 0.06 ± 0.03, and 0.53 ± 0.04 for the fast, intermediate, and slow phases, respectively (Figure 5B). (The denatured state decays in a mirror fashion as the intensity of the N band is described as the fraction of the combined D and N band intensities; the intensity of the intermediate band is too faint to be included in the analysis.) Thus, SDS−PAGE data support the formation of the native state in three parallel pathways, and rate constants (at 0 M GdmCl) are consistent with the Trp-based kinetic data. G
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 5. Refolding of OmpA2 monitored by SDS−PAGE. (A) Time course for folding from the urea-denatured state into 250 mM OM in the absence of GdmCl. Note the formation of a band intermediate between the denatured and folded states. (B) Change in the relative intensity of the band corresponding to the folded state (relative to the sum of the band intensities of the folded and unfolded states) as a function of time. Data are fitted to a triple-exponential decay using rate constants obtained from Trp fluorescence time profiles in Figure 4 at 0 M GdmCl.
In stage 1 (0−0.6 M GdmCl), the dominant ground state from which folding occurs is C, which means that the fraction of protein that can fold along the fast track from D to N (kf) is insignificant. Rather, the process of folding is rate-limited by the slow transition from C to N (k2). The value m2, which describes how log k2 varies with GdmCl concentration, is ≈0 (Table 2). The fact that k2 does not vary with GdmCl concentration can be interpreted to mean that C and transition state TS between C and N have the same degree of compactness, so GdmCl does not favor one state over the other. m values for a transition between two species are generally understood to represent the difference in the degree of burial of surface area between the two states.31,32 For m values associated with kinetic processes, the two species are the ground state (from which the process starts) and the rate-determining transition state separating the ground state from the end state. Thus, m2 describes the transition from C to TS. In stage 2 (0.6−1.2 M GdmCl), D gradually approaches C in stability, so folding rates increase because it becomes faster to fold from C than from D. The rate of the D → N transition decreases with GdmCl concentration [according to the value of mf = −2.58 M−1 (Table 2)], but this is compensated by the
Scheme 2
GdmCl concentration. The relative populations of C and D can be calculated from the relationships fC = [C]/([D] + [C]) = K C/(1 + K C)
(4)
and
fD = 1 − fC
(5)
Calculated values, shown in Figure 6A, illustrate that C dominates overwhelmingly (>0.95) from 0 to 0.6 M GdmCl while D becomes the dominant species from ∼1.5 M GdmCl. The different stages of the refolding limb can be explained as follows. H
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 6. Analysis of the folding of OmpA2. (A) Relative populations of C and D in the ground state before refolding of OmpA2 as a function of GdmCl concentration. Values calculated from eqs 4 and 5 using data in Table 2. (B) Comparison of the measured and calculated amplitudes for the fast refolding phase of OmpA2. Calculated values obtained with eq 6 using data from panel A in combination with estimated amplitudes AmpC→N and AmpD→N as described in the text. (C) Similar analysis for the intermediate refolding phase of OmpA2.
Table 2. Summary of Kinetic Parameters Describing Folding and Unfolding of OmpA Concatemers in GdmCla OmpAb kwater f −1
−1
(k in min ) log mf (M ) c log Kwater I −1 c mI (M ) e log Kwater C mC (M−1)e log kwater (k in min−1)d 2 water log ku (k in min−1) mu (M−1)b ΔGD‑N (kcal/mol)
OmpA2
fast (k1)
fast (k1)
0.99 ± 0.08 −0.61 ± 0.34c 1.78 ± 0.27c −2.57 ± 0.27c N/Ad N/Ad N/Ad −6.37 ± 0.12c 1.50 ± 0.03c −12.43 ± 0.42g
2.99 ± 0.89 −2.58 ± 0.52c N/Ad N/Ad 4.86 ± 0.79 −3.77 ± 0.60 −1.34 ± 0.33 −4.39 ± 0.14d 1.12 ± 0.04d −10.0 ± 1.2h
c
d
OmpA3
intermediate (k2)
fast (k1)
intermediate (k2)
1.34 ± 0.61 −1.98 ± 0.34d N/Ad N/Ad 5.57 ± 0.90 −4.29 ± 0.85 −2.17 ± 0.08 −5.28 ± 0.13d 1.18 ± 0.03d −9.00 ± 0.85h
7.2 ± 2.9 −5.2 ± 1.7d N/Ad N/Ad 12.83 ± 3.15 −8.57 ± 2.03 −1.63 ± 0.08 −4.06 ± 0.12d 1.06 ± 0.03d −15.3 ± 3.9h
0.06 ± 0.82d −1.3 ± 0.48d N/Ad N/Ad NDf NDf −2.59 ± 0.05 −5.63 ± 0.37d 1.41 ± 0.11d −7.7 ± 1.2h
d
d
All parameters derived from chevron data in Figures 4C and 7A. bData from ref 18. cObtained from ref 18 by fitting chevron data to the equation for sequential folding (Scheme 1), where KI = [I]/[D] and kf and ku describe rate constants for folding from I to N and unfolding from N to I, respectively. dObtained from fitting kinetic data (Figure 4C for OmpA2 and Figure 7A for OmpA3) to eq 3 (Scheme 2) where KC = [C]/[D], k2 is the rate constant for folding of C to N, and kf and ku describe rate constants for folding from D to N and unfolding from N to D, respectively. eNot applicable because the parameter in question is not part of the scheme used to analyze the kinetic data. fCould not be determined with sufficient + log Kwater − log kwater ). hΔGD‑N = −RT ln(10) × (log kwater confidence because of the limited quality of the data. gΔGD‑N = −RT ln(10) × (log kwater f I u f − log kwater ). u a
I
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 7. Kinetics of folding and unfolding of OmpA3. (A) Chevron plots of refolding and unfolding of OmpA3. A triangular folding pathway (Scheme 2) is fitted to the data for the fast and intermediate phases using eq 3. (B) Amplitudes of the different refolding phases. (C) Amplitudes of different unfolding phases. (D) Comparison of the measured and calculated amplitudes for the fast refolding phase of OmpA3 (cf. Figure 6B).
possible to perform this analysis on the slow phase). The m values for the fast and intermediate process are very similar. For both processes, the refolding and unfolding rate constants combine to give the same stabilities for the N state. This suggests that the slow and fast processes monitor the same folding process though with different activation barriers, i.e., two parallel folding pathways, which both involve an offpathway folding intermediate. The refolding intermediate observed for single-domain OmpA is probably still present but does not play a kinetic role because the rate-limiting step for folding is entirely displaced to another folding process, namely folding from collapsed state C. Further support for the existence of an off-pathway state according to Scheme 2 can be obtained by inspecting the amplitudes. According to Scheme 2, for the fast and
even more pronounced increase in the rate of growth of the D population [according to the value of mC = −3.77 M−1 (Table 2)]. In stage 3 (1.2−2 M GdmCl), the population of C dwindles to insignificance and the main folding occurs from D to N. The rate of this process decreases with increasing denaturant concentrations. In stage 4 (2−5 M GdmCl), D is more stable than N so the dominant process is unfolding from N to D, which is accelerated by the denaturant. C does not accumulate under these conditions. The corresponding kinetic parameters for the fast and intermediate phases are summarized in Table 2 (because of the lack of a slow phase during unfolding as well as the limited refolding data in the region of 0−2 M GdmCl, it was not J
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 8. SDS−PAGE analysis of the folding of OmpA3. (A) Time course of refolding from the urea-denatured state into 250 mM OM in the absence of GdmCl. Note here the rise of an intermediate band I between D and N that slowly decays. Intensities of (B) bands I and N and (C) band D. A triple-exponential decay is fitted to data in panels B and C, using rate constants from Figure 7A at 0 M GdmCl but allowing amplitudes to vary.
disappears and the amplitude of the fast phase declines. Combined refolding and unfolding data for the fast and intermediate phases can be analyzed according to Scheme 2, and the results are listed in Table 2. Note that the variation in fitted rate constants and the very steep transitions between 1 and 2 M GdmCl in the refolding limbs of the chevron plot for OmpA3, which are even more pronounced than for OmpA2, impart a large degree of error in extrapolating values for KC and kf to 0 M GdmCl and consequently make it difficult to estimate ΔGD‑N values based on these constants, as seen in Table 2. Consequently, the apparent difference in ΔGD‑N calculated on the basis of the fast (k1) and intermediate (k2) phase may not be significant. There is a reasonable correspondence between the measured amplitude and the amplitude calculated according to eq 2 (data for the fast phase shown in Figure 7B). A SDS−PAGE analysis of OmpA3 refolding at 0 M GdmCl, undertaken in a fashion analogous to that of OmpA2, uncovers some remarkable features of the folding process (Figure 8A). As the intensity of the band for denatured state D decays, the intensity of a band I of size intermediate between the sizes of D and N increases and decays only slowly over time. The intensity of the lowest band, corresponding to the N state, increases initially to a value lower than that of I but overtakes it after around 1000 min (Figure 8B), though the intensity of the I band only slowly declines over the next several hours. When the intensities are plotted versus time, there is a rapid burstphase increase in intensity of I from time zero to the first 30 s, followed by a rise to a plateau over the next 20 min. The N state does not show the burst-phase increase, though there is still a rapid increase in intensity over the first ∼20 min. We fitted intensity data for all three states to a triple-exponential decay analogous to Figure 5B using rate locked constants for folding in 0 M GdmCl obtained from the chevron plots in Figure 7A. As shown in panels B and C of Figure 8, there is good agreement for the time interval of 20−6000 min but considerable deviations for the interval of 0−20 min, where
intermediate phases, the amplitude consists of weighted contributions from two processes, namely, folding directly from D (AmpD→N) as well as folding from C (AmpC→N). This must be weighted by the relative fractions of D (f D) and C (f C) as well as the fraction of protein that actually folds to N (f N, which becomes significantly less than 1 only as we approach the midpoint of denaturation): Amptotal = AmpD → N × fD × fN + AmpC → N × fC × fN (6)
The values of the two amplitudes, AmpC→N and AmpD→N, are obtained as the amplitudes around 0 and 1.2 M GdmCl, where C and D dominate, respectively (Figure 6A). Thus, the variation in amplitude with GdmCl concentration can simply be calculated using the known variation in the concentrations of D, C, and N obtained from the chevron plots together with singlepoint values of amplitudes at 0 and 1.2 M GdmCl. Amplitudes calculated this way for the fast and intermediate folding processes fit quite well to the measured amplitudes (Figure 6B,C). In contrast, the unfolding amplitudes are relatively constant once we reach the concentration range (>2.4 M GdmCl) where the D state is the dominant species so the whole protein population unfolds. This emphasizes the simple one-step unfolding that is now occurring in parallel here. OmpA3 Folds through a Triangular Folding Pathway with Even More Stabilized Intermediate States. We reach very similar conclusions when we analyze the folding of OmpA3 by Trp fluorescence. As for OmpA2, unfolding is characterized by two exponential decays, the log of whose rate constants depends linearly on GdmCl concentration (Figure 7A). Refolding occurs in three phases at low (0−1.7 M) GdmCl concentrations, but the slower phase disappears around 1.8 M GdmCl just as for OmpA2, after which refolding and unfolding signals follow a double-exponential decay (Figure 7B). The two unfolding phases remain relatively constant between 2 and 4 M GdmCl, after which the intermediate phase K
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
vesicles (Figure 9B,C). OmpA2 gives rise to three bands at 25 and 40 °C that can be assigned to the unfolded state, a state with one domain presumed folded and a domain with two domains folded. Thus, significant parts of OmpA2 remain unfolded or incompletely folded after 2 h at 25−40 °C, and this situation does not change upon prolonged incubation [several days (data not shown)]. Only at 55 °C is the protein essentially completely folded after 2 h. For OmpA3, on the other hand, there are four bands visible at all three temperatures. The unfolded state dominates at all three temperatures, though there seems to be an optimum for folding around 40 °C. Changing the lipid:protein ratio from 800:1 to higher values (up to 2400:1) did not alter these outcomes for either OmpA2 or OmpA3 (data not shown). We sought to address the states of folding represented by the three lower bands of OmpA3 and to identify possible partially folded states. Our strategy was to construct their denaturation profiles in urea, using monomeric OmpA as a representative for the properly folded native state. OmpA3 and OmpA were refolded in lipids at 40 °C and then incubated with 0−6 M urea before being analyzed by SDS− PAGE. The intensities of the bands display a distinct pattern. OmpA shows the two expected bands corresponding to the N and D states (Figure 10A), which follow a sigmoidal curve very similar to the equilibrium denaturation profile determined by Trp fluorescence in micelles (Figure 3A). The curve can be fitted to a standard two-state denaturation curve with a midpoint at 3.43 ± 0.18 M urea, which is completely consistent with the ∼3.4 M midpoint range determined by us in a previous study in micelles.18 Of the three bands below the denatured state of OmpA3, the band labeled F3 in Figure 10B migrates the fastest and therefore is the most compact and likely the most folded. The F3 band intensity follows a denaturation curve similar to that OmpA, with a midpoint of 3.38 ± 0.08 M urea (F1 in Figure 10B). Interestingly, the second fastest band (F2) shows a midpoint of 2.77 ± 0.12 M urea. The highest of the three bands, F1, shows a transition to a higher population level with a midpoint of approximately 2.5 M urea but then decays at higher concentrations, though significant levels are still left around 5−6 M urea (Figure 10B). The simplest interpretation is that F3 represents a state in which all three OmpA units are folded independently, leading to the same level of stability as monomeric OmpA (consistent with the GdmCl equilibrium data in Figure 3 and the high mobility of the F3 band). If F2 and F1 simply contained two folded units and one folded unit, respectively, and one and two completely unfolded units, then they should follow the same denaturation profile as F3, which is clearly not the case. F2’s reduced midpoint indicates that it must contain one or more species that are less stable than the N state, and the decay in the F2 population coincides with the increase in the F1 population but precedes the decay in F3 and increase in the D population (Figure 10B). Thus, the simplest (but far from only) interpretation is that F2 contains a combination of several non-native folded domains and that unfolding of one of these domains converts F2 to F1, which retains (at least) one non-native folded domain that is either thermodynamically or kinetically more stable than a conventional native domain and therefore unfolds only at very high urea concentrations (see the diagram in Figure 10B to illustrate these possible transitions). We return to this issue in the Discussion. When the kinetics of folding of TM-OmpA and OmpA2 into DOPC/DOPG vesicles are monitored at 25 °C by SDS−PAGE over a 2 h period, it is characteristic that numerous higher-
SDS−PAGE data indicate that the relaxation of the fast phase occurs faster than it does according to Trp fluorescence data. We are unable to determine the reason for this disparity. It might be envisaged that the existence of partially folded regions in the intermediate state affects its behavior in the presence of SDS encountered in the gel loading buffer, so that SDS micelles accelerate its formation compared to that under the SDS-free conditions monitored by Trp fluorescence; this chaperone-like behavior could also help explain the burst phase in the formation of I. It is unclear whether such a phenomenon could explain the rapid formation of the band we attribute to the native state, though it might be suggested that SDS micelles could modify both I and N states by establishing them on individual micelles in such a way that different domains could interact productively. Data for the folding of OmpA3 in lipid vesicles (next section) allow the possibility of the existence of such “mixed structures”, though we hesitate to indiscriminately transfer such considerations to SDS micelles. Nevertheless, as for OmpA2, the SDS−PAGE data are consistent with the formation of a native state in three parallel steps, though an intermediate stage is much more prominent in the case of OmpA3. Folding of OmpA2 and OmpA3 into Lipids. Finally, we tested whether it was possible to fold OmpA2 and OmpA3 into phospholipid vesicles rather than surfactant micelles. Ureaunfolded OmpA2 and OmpA3 were diluted to low urea concentrations at 25, 40, and 55 °C in the presence of vesicles containing 92.5% DOPC and 7.5% DOPG, a combination known to favor OmpA folding.33 For TM-OmpA, folding is quantitative, and within 2 h, all OmpA has folded to the native state at all three temperatures (Figure 9A). However, OmpA2 and OmpA3 are more reluctant to fold into the phospholipid
Figure 9. Folding of OmpA variants into DLPG/DLPC phospholipid vesicles. Proteins were diluted from the urea-denatured state into solutions of vesicles at a lipid:protein molar ratio of 800:1 (OmpA monomer equivalents) and incubated for 2 h at the indicated temperatures before samples were subjected to SDS−PAGE. Prolonged incubation did not alter the band distribution. L
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 10. Stability of (A) OmpA and (B) OmpA3 in DLPG/DLPC vesicles measured by urea denaturation. Proteins were folded into vesicles at 40 °C for 2 h, after which they were incubated at different urea concentrations for 2 h and subjected to SDS−PAGE. The intensities of the indicated bands corresponding to single-polypeptide chain species were quantified by ImageJ and plotted in the graphs. An equilibrium denaturation model (eq 1) was fitted to all data except those for F1 and D in OmpA3. The models below the graph depict the transitions that are proposed to occur for the different bands. F3 contains three natively folded OmpA domains that unfold independently and show the same midpoint as monomeric OmpA (3.4 M urea). F2 is proposed to consist of several non-natively folded domains; unfolding of one leads to F1, whose single folded domain is kinetically trapped and thus unfolds at concentrations of urea higher than those at which natively folded OmpA unfolds.
OmpA in which two or three OmpA molecules are forced into permanent physical proximity irrespective of the micellar concentrations. One disadvantage of this concatemeric approach is that we effectively lose the ability to monitor domain swapping within the polypeptide chain by SDS−PAGE because the main interactions will now stay within one polypeptide chain and will therefore not affect mobility to the same extent as the formation of intermolecular contacts by TMOmpA. However, this is outweighed by the advantage of being able to perform kinetic analyses of folding and unfolding over a full range of GdmCl concentrations, with the rich mechanistic insight that this provides, without concerning ourselves with staying close to the cmc as needed for TM-OmpA. Increasing GdmCl concentrations will also increase the cmc because of its solvation properties,18 just like the denaturant urea,34 and it would become unacceptably complicated to maintain the same low micelle concentrations between 0 and 5 M GdmCl. Furthermore, several outer membrane proteins form trimers in vivo (see below), and concatemers allow us to probe the consequences of expressing trimers on one polypeptide chain. Concatemerization Does Not Affect the Stability of OmpA. The concatemers behave like a collection of isolated OmpA molecules from a thermodynamic perspective, being denatured at the same GdmCl concentration as TM-OmpA and also showing the general membrane protein dependence on the presence of micelles for folding. Furthermore, folded
molecular weight bands emerge over time (Figure 11A,B). We have previously reported such bands for TM-OmpA and ascribe them to folded and (partially) unfolded oligomeric species.23 Interestingly, OmpA2 shows a propensity to form these bands that is significantly stronger than that of TM-OmpA. Focusing on the single-chain species, we can quantify band intensities to monitor the kinetics of formation of N (Figure 11C). It is clear that the two proteins reach a plateau at around the same rate. The important difference is that ∼93% of single-domain OmpA, but only 69% of OmpA2, reaches the folded state, emphasizing that OmpA2 is much more prone to kinetic traps in the folding process than single-domain TM-OmpA.
■
DISCUSSION This study of the folding of OmpA concatemers was inspired by the ability of TM-OmpA to form partially folded oligomers at surfactant concentrations close to the critical micelle concentration (cmc).23 We reasoned that the low micelle concentrations led to a high local concentration of OmpA molecules, which needed a micellar environment to fold to the native state; juxtaposition of OmpA molecules could then promote inter- and intramolecular contacts, leading to formation of domain-swapped structures in which β-strands from one molecule could form native-type interactions with complementary β-strands from another OmpA molecule. To validate this hypothesis, we therefore produced concatemers of M
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
Figure 11. Time course for the folding of TM-OmpA and OmpA2 into DLPC/DLPG vesicles at 55 °C. Folding of (A) TM-OmpA and (B) OmpA2 into vesicles was monitored by SDS−PAGE. Note the pronounced formation of higher-order species, particularly for OmpA2. (C) Plot of the intensity of the native monomer band of TM-OmpA and OmpA2 (relative to the total intensity of monomeric native and denatured bands at each time point) vs time. A double-exponential decay function is used to fit the data.
flexibility (achieved, e.g., by inserting Gly or removing Pro residues) can promote swapping.40 In the case of OmpA, the lower stability of domain-swapped constructs can be attributed to strain and topological barriers to making two or three circular OmpA domain swaps (as schematically illustrated in Figure 12). The possibility that domain-swapped OmpA oligomers could be stabilized in a concatemeric context by judicious changes to loops and linkers, such as the shortening of some of the four long surface-exposed loops, cannot be ruled out, though OmpA is known to tolerate substantial shortening of these loops.41 However, in this study, it is advantageous that the conventional nonswapped form is the most stable species because it allows us to focus on the kinetic traps that are encountered en route to this state. These traps arise because there is kinetic competition between the formation of the native state and other states, leading to folding kinetics significantly more complex than that for monomeric TM-OmpA. Kinetic Detours Dramatically Reduce the Rate of Folding of OmpA Concatemers. On the basis of our SDS− PAGE and Trp fluorescence kinetic studies, it is striking that the introduction of concatemers leads to refolding kinetics significantly slower than that of TM-OmpA. We have previously shown that TM-OmpA folds quite rapidly into OM micelles at low denaturant concentrations;18 the major phase (accounting for half the total signal) has a half-life of ∼7 s, while the intermediate and slow phases (both with ∼25% of
concatemers can be cleaved to yield monomers of the expected weight. This is to be entirely expected. Evolutionarily optimized multidomain proteins have arisen through duplication, shuffling, and evolutionary drift and make up 70−80% of the proteome in eukaryotes and 40−70% in prokaryotes.35 They often contain domains that are incapable of independent stability but are propped up by stabilizing interactions at the interdomain interfaces, typically through hydrophobic contacts.36 In contrast, the OmpA concatemer contains no such designed intermolecular (here interdomain) contacts and would be expected to show the same stability as isolated OmpA molecules unless domain-swapped OmpA constructs were more thermodynamically stable than the individual OmpA molecules. This is clearly not the case for our constructs. In principle, the preference for isolated domains versus domain swapping could be modified by protein engineering. When intertwined homodimers undergo three-dimensional domain swapping, substitution of specific intramolecular contacts with their intermolecular equivalents usually involves a hinge loop region that either folds back on itself in the monomer or adopts an extended conformation in the domain-swapped domain.37,38 The energetics of domain swapping is regulated by the hinge loop. Thus, loop shortening makes it difficult for the chain to fold back on itself, encouraging domain swapping and formation of either reciprocal dimers or higher-order aggregates depending on the details of the design,39 while increased N
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
forced close together as concatemers, and furthermore restricted by the need to be inside a micelle (possibly one shared between different OmpA domains) to fold, this will increasingly drive formation of highly collapsed (yet still nominally denatured) species involving contacts between different OmpA domains. As a consequence, it will be even more difficult to reconfigure the polypeptide chain to allow folding of individual domains of OmpA. Furthermore, these collapsed states can according to Scheme 2 collapse or fold further to a separate state C that contains levels of stabilizing structure higher than that of the D state. As seen in Table 2, the longer the concatemer (OmpA3 vs OmpA2) and the more collapsed the initial denatured state (intermediate phase vs fast phase), the greater the propensity to form the C state, i.e., the greater the log Kwater value. The C state could be the band C observed as an intermediate state in Figure 8A. Folding of Concatemers in Lipids Highlights Mixtures of Differently Folded States of OmpA. Backtracking from this C state to reach the native state through D (or in a direct slow track to N) requires a great deal of conformational flexibility, which is possible in surfactant micelles because of their high level of dynamics and ability to regroup around the protein.42,43 However, the greater degree of packing in lipid vesicles than in micelles clearly increases the activation barriers for this process, because there is only a low level of formation of full-length folded OmpA2 in vesicles at room temperature. We used lipid vesicles made of DLPC and DLPG, which are the shortest lipid chains that lead to vesicles and therefore have a rather thin layer of hydrocarbon chains packing against the protein molecules. Therefore, the packing restrictions will be even more pronounced in vesicles made of longer chains wherein OmpA is known to fold more slowly.16 The situation improves as the temperature is increased to 40 and 55 °C, indicating that it is a question of overcoming activation barriers to folding. Nevertheless, OmpA3 forms a multitude of differently folded states at both 40 and 55 °C where the most compact state does not dominate under any conditions. Both concatemers form states in these lipids with enough residual structure to migrate as species between the denatured and fully folded state, indicating that they are trapped in energetic minima from which they find it difficult to extricate themselves. A more extensive analysis of the stabilities of the different states of OmpA3 suggests that they consist of mixtures of natively and non-natively folded OmpA domains. The question of whether the phospholipid-stabilized species that are observed in urea denaturation experiments can be related to the compact state predicted from folding kinetics of OmpA3 in micelles in the presence of GdmCl arises. Making the reasonable assumption that the denaturation potencies of urea and GdmCl scale proportionally, we reason as follows. Using the stability parameters log KC and mC for compact state C of OmpA3 based on the fast folding phase (Table 2), we calculate that C remains more stable than D until 1.5 M GdmCl, where D becomes more stable (cf. calculations for OmpA2 in Figure 6A). Given that the native state of OmpA3 has a midpoint denaturation point of 2.04 M GdmCl (Table 1) corresponding to 3.38 M urea (Figure 10B), the midpoint of the C state in OmpA3 must be around 1.5/2.04 × 3.38 M = 2.48 ± 0.12 M urea. This value is roughly equal to that of the midpoint of the F2 band. While this may be a coincidence, it does highlight the possibility that at least one of the compact states observed for OmpA3’s folding in micelles could give rise to a more permanently trapped state when OmpA3 is refolded
Figure 12. Model highlighting the possibility of concatemers of OmpA either folding directly or forming domain swaps under folding conditions, which can either unfold back to the denatured state or fold directly (but very slowly) to the native concatemeric state.
the total signal) have half-lives of ∼1.1 and ∼110 min, respectively. In contrast, OmpA2’s half-lives of refolding at 0 M GdmCl are around 11, 110, and 1400 min (Figure 4C) while those of OmpA3 (Figure 7C) are 28, 280, and 4400 min. Furthermore, for TM-OmpA, the fast folding phase dominates in terms of amplitude both during refolding in GdmCl and in urea throughout the whole refolding range, and in GdmCl, it becomes the sole refolding phase above 0.6 M GdmCl.18 The situation is quite the reverse for the concatemers (Figures 4D and 7B), where the slow phase is by far dominant in terms of amplitude at low GdmCl concentrations and the fast phase has the smallest amplitude. This can to only some extent be attributed to the difference in signal when folding from the C state versus the D state in Scheme 2. Thus, concatemer refolding is heavily skewed toward slow folding. The multiple refolding phases can be explained as parallel folding pathways from the denatured to the native state, just as seen for TM-OmpA for refolding in urea.18 For TM-OmpA, we attributed these parallel pathways to formation of different species of denatured state D with different levels of collapse at the onset of the folding process when the denatured state is diluted out from high to low denaturant concentrations in the presence of micelles. The slowest refolding arose because of “overcollapse” and the need to extricate the polypeptide chain to a more extended state prior to refolding. This could therefore be seen as a type of collapsed species akin to the C state suggested for the concatemers, though the low-amplitude, scattered kinetic data and merger of the different phases at intermediate denaturant concentrations precluded a more detailed analysis. In the case of OmpA concatemers, on the other hand, the kinetic data are sufficiently good to allow us to observe folding complexity at several levels. These include (1) parallel folding pathways, (2) multiple species between D and N that can be detected on gels and probably represent species with different numbers of folded domains, and (3) a species C that according to Scheme 2 can be classified as being offpathway, though with the possibility of slow folding to the native state at rate k2, which is around 3−4 orders of magnitude slower than the direct folding route with rate kf (cf. Table 2). This development is nicely consistent with the collapse scenario proposed for TM-OmpA. When several OmpA molecules are O
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
An Unexpected Consequence: Faster Folding and Unfolding in the Absence of the Trapped State Could Be Due to Entropic Effects. This does not mean, however, that concatemers have no positive effects. A comparison of the folding and unfolding rate constants of the three OmpA constructs (Figure 13A) illustrates some important points in
in lipids, given that the more well-packed and organized lipid environment generally leads to slower conformational rearrangements and folding transitions [cf. kinetics of folding of OmpA in micelles (ref 18 vs ref 44)]. It is also tempting to speculate that the compact state formed during the intermediate phase of refolding of OmpA3 (whose kinetic parameters cannot be determined because of the abruptness of the transition in Figure 7A) could be involved in the stabilization of F1. However, we emphasize that this is highly speculative. Nevertheless, these observations indicate that there is a possible connection between the kinetic intermediates observed in micellar refolding and states that can be trapped during folded in a more biologically representative lipid environment. Natural Concatemers Have Evolved To Prevent Unwanted Interdomain Contacts. Outer membrane proteins naturally form trimers in vivo, e.g., porins such as PhoE.45 In contrast to the OmpA3 construct used in this study, OMP trimers are formed in vivo by association of monomers rather than by concatemers. Interestingly, PhoE can fold to monomers in vitro and subsequently can trimerize via the help of phospholipids and lipopolysaccharides;46 all the same, they have been suggested not to fold to the monomeric state first in vivo but to form a metastable trimeric state that subsequently converts to the final folded trimer.47 Nevertheless, the complications associated with folding of OmpA concatemers suggest that it is much more advantageous to express the proteins as individual domains and then let them assemble without the topological restrictions caused by interdomain linkage. This also makes it much easier to direct the process with the help of periplasmic and molecular chaperones,5 which guide the proteins to the outer membrane and the BAM complex. Water-soluble protein concatemers containing domain repeat proteins are found in vivo, having arisen through internal gene duplications, and constitute up to 20% of the total protein population.48 However, their folding behavior is regulated by various evolutionary adjustments, which minimizes potentially deleterious aggregation and misfolding. One such adjustment is a forced reduction in the level of sequence similarity: adjacent domains in immunoglobulin-like proteins such as titin generally show levels of sequence similarity much lower than those of nonadjacent domains,24 thus favoring internal rather than external molecular contacts. Other strategies include gatekeeper residues49 where specific residues are selected to prevent aggregation rather than promote folding.50,51 A sophisticated system is provided by the 33-residue Ankyrin repeat:25 a low level of internal homology among tandem Ankyrin proteins prevents significant aggregation, and differences in thermodynamic and kinetic stability mean that a subset of the repeats is highly structured in the major folding transition state while others undergo folding only upon binding other components. Remarkably, the greater the number of repeats, the more pronounced the downward curvature in the unfolding arm of the chevron plot, which reflects sequential unravelling of the repeat array, repeat by repeat.52 Another characteristic feature of the unfolding arm includes mutationally induced upward curvature due to the shift from one path to another. 52 The absence of such curvature in OmpA concatemers emphasizes yet again the difference between evolved tandem repeats and those that are made by simple juxtaposition of identical sequences.
Figure 13. Kinetic analysis of OmpA constructs. (A) Comparison of the fast phases for folding and unfolding of the different OmpA constructs compared in this study. Data for OmpA are from ref 18 and are fitted to the D ⇔ I ⇔ N folding scheme, where I is an on-pathway intermediate. Data for OmpA2 and OmpA3 are fitted to eq 3 associated with Scheme 2. (B) Speculative scenario for the acceleration of folding and unfolding kinetics caused by OmpA concatemers in the fast phase in the absence of kinetic traps. Both the denatured state and the native state could be destabilized for different entropic reasons; these effects would be absent (or reduced) in the transition state, leading to a relative stabilization of this state and a consequent lowering of the barrier to both folding and unfolding.
this regard. All three constructs have the same minimum in kinetics around 2 M GdmCl, nicely consistent with their similar stability. However, over most of the denaturant range (≳1.5 M GdmCl), both folding and unfolding rate constants are considerably faster for the two concatemers than for singledomain OmpA. Although the two concatemers have very similar rate constants, OmpA3 unfolds slightly faster than OmpA2, suggesting that the acceleration effect is (weakly) cumulative. The acceleration means that over this concentration range, the concatemers experience an activation barrier to both folding and unfolding smaller than that of the singledomain OmpA (i.e., a more stable transition state), despite having similar levels of stability at the ground state. Thus, in the absence of conformational detours caused by the accumulation of the C species, the concatemers should actually fold significantly faster than OmpA at low GdmCl concentrations. We can only speculate about the basis for this phenomenon at present. However, it may be expected that intermediate to high concentrations of GdmCl will particularly weaken the interactions that otherwise stabilize kinetically trapped species such as the C state and overcollapsed species. This allows OmpA to profit from the potential topological advantages of concatemers without the drawbacks from trapped states. One such advantage could be entropic. In a concatemer, the loss of conformational entropy in the folding transition state during folding of individual domains may not be as large as that of folding of separate domains, because the unfolded state of each state is already more restricted in its flexibility because of the covalent connection with other domains. However, in the final folded state, this advantage may be canceled by loss of conformational freedom such as strain or other steric hindrances around the linkers, leading to the same stability overall as for individual domains. The net effect would be a P
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry
OmpA3, three copies of the transmembrane part of OmpA joined by linker sequences.
destabilization of the D and N states, leaving the TS state at the same energy level (or at least not destabilized to the same extent). This will lower the activation barrier, accelerating both folding and unfolding, but not leading to an overall increase in stability (Figure 13B). The reduction in m values can be reconciled with this scenario. m values for a reaction going from A to B (e.g., refolding from D to TS) are usually interpreted to be proportional to the change in compaction between these two states.20 The equilibrium m values for OmpA constructs (Table 1) have error bars that are too large to be considered here, but apart from the fast phase of OmpA3 (which also has large errors associated with it), mf values for OmpA2 and OmpA3 (Table 2) are clearly smaller than their equivalent value for OmpA (namely mf + mI, which represents the step from D to TS via I). The same applies for mu values, which are more robustly determined. Thus, the change in compaction for the two steps between D and N (D to TS and TS to N) appears to be reduced for concatemers compared to that for OmpA, most likely because of an overall increase in the extent of compaction of the D state that extends to some degree to the TS. Nevertheless, the high concentrations of denaturant required to reach this situation suggest that this will not be physiologically relevant. Periplasmic chaperones or lipopolysaccharides, which also can control OmpA refolding,53 may prevent unwanted intermolecular interactions such as domain swapping, but their binding will also reduce the entropic advantages incurred from concatemerization. In summary, our work illustrates the complications that can arise when concatemers force identical copies of individual protein domains into the proximity of each other. The situation is no doubt exacerbated for a membrane protein in which a high intrinsic hydrophobicity provides a strong driving force to form intra- and extradomain contacts under conditions favoring the native state and highlights the many potential complexities that face proteins as they navigate through a perilous energy landscape in search of a safe native haven.
■
■
REFERENCES
(1) White, S. H., and Wimley, W. C. (1999) Membrane protein folding and stability: physical principles. Annu. Rev. Biophys. Biomol. Struct. 28, 319−365. (2) Bishop, C. M., Walkenhorst, W. F., and Wimley, W. C. (2001) Folding of beta-sheets in membranes: specificity and promiscuity in peptide model systems. J. Mol. Biol. 309, 975−988. (3) Wimley, W. C. (2003) The versatile beta-barrel membrane protein. Curr. Opin. Struct. Biol. 13, 404−411. (4) Costello, S. M., Plummer, A. M., Fleming, P. J., and Fleming, K. G. (2016) Dynamic periplasmic chaperone reservoir facilitates biogenesis of outer membrane proteins. Proc. Natl. Acad. Sci. U. S. A. 113, E4794. (5) Mogensen, J. E., and Otzen, D. E. (2005) Interactions between periplasmic chaperones and bacterial outer membrane proteins. Mol. Microbiol. 57, 326−346. (6) Lyu, Z. X., and Zhao, X. S. (2015) Periplasmic quality control in biogenesis of outer membrane proteins. Biochem. Soc. Trans. 43, 133− 138. (7) Sklar, J. G., Wu, T., Kahne, D., and Silhavy, T. J. (2007) Defining the roles of the periplasmic chaperones SurA, Skp, and DegP in Escherichia coli. Genes Dev. 21, 2473−2484. (8) Gu, Y., Li, H., Dong, H., Zeng, Y., Zhang, Z., Paterson, N. G., Stansfeld, P. J., Wang, Z., Zhang, Y., Wang, W., and Dong, C. (2016) Structural basis of outer membrane protein insertion by the BAM complex. Nature 531, 64−69. (9) Wu, T., Malinverni, J., Ruiz, N., Kim, S., Silhavy, T. J., and Kahne, D. (2005) Identification of a multicomponent complex required for outer membrane biogenesis in Escherichia coli. Cell 121, 235−245. (10) Ricci, D. P., and Silhavy, T. J. (2012) The Bam machine: a molecular cooper. Biochim. Biophys. Acta, Biomembr. 1818, 1067−1084. (11) Noinaj, N., Kuszak, A. J., Balusek, C., Gumbart, J. C., and Buchanan, S. K. (2014) Lateral opening and exit pore formation are required for BamA function. Structure 22, 1055−1062. (12) Kim, K. H., Aulakh, S., and Paetzel, M. (2012) The bacterial outer membrane beta-barrel assembly machinery. Protein Sci. 21, 751− 768. (13) Andersen, K. K., and Otzen, D. E. (2013) Folding of outer membrane proteins. Arch. Biochem. Biophys. 531, 34−43. (14) Moon, C. P., Kwon, S., and Fleming, K. G. (2011) Overcoming hysteresis to attain reversible equilibrium folding for outer membrane phospholipase A in phospholipid bilayers. J. Mol. Biol. 413, 484−494. (15) Burgess, N. K., Dao, T. P., Stanley, A. M., and Fleming, K. G. (2008) Beta-barrel proteins that reside in the Escherichia coli outer membrane in vivo demonstrate varied folding behavior in vitro. J. Biol. Chem. 283, 26748−26758. (16) Kleinschmidt, J. H., and Tamm, L. K. (2002) Secondary and tertiary structure formation of the β-barrel membrane protein OmpA is synchronized and depends on membrane thickness. J. Mol. Biol. 324, 319−330. (17) Kleinschmidt, J. H. (2015) Folding of β-barrel membrane proteins in lipid bilayers - Unassisted and assisted folding and insertion. Biochim. Biophys. Acta, Biomembr. 1848, 1927−1943. (18) Andersen, K. K., Wang, H., and Otzen, D. E. (2012) A Kinetic Analysis of the Folding and Unfolding of OmpA in Urea and Guanidinium Chloride: Single and Parallel Pathways. Biochemistry 51, 8371−8383. (19) Fersht, A. R. (1999) Structure and mechanism in protein science. A guide to enzyme catalysis and protein folding, Freeman & Co., New York. (20) Tanford, C. (1970) Protein denaturation. Part C. Theoretical models for the mechanism of denaturation. Adv. Protein Chem. 24, 1− 95. (21) Baldwin, R. (1996) On-pathway versus off-pathway folding intermediates. Folding Des. 1, R1−R8.
AUTHOR INFORMATION
Corresponding Author
*E-mail:
[email protected]. ORCID
Daniel E. Otzen: 0000-0002-2918-8989 Present Address †
K.K.A.: Agro Business Park A/S, Niels Pedersens Allé 2, DK8830 Tjele, Denmark. Author Contributions
K.K.A. and B.V. contributed equally to this work. K.K.A. and D.E.O. designed the experiments. K.K.A., S.O., and B.V. performed the experiments. D.E.O. analyzed data and wrote the manuscript. Funding
K.K.A., B.V., and D.E.O. are supported by grants from the Danish Research Council|Technology and Production (Grant 12-126 186 to K.K.A. and Grant 6111-00241B to B.V. and D.E.O.). Notes
The authors declare no competing financial interest.
■
ABBREVIATIONS OmpA, outer membrane protein A; OmpA2, two copies of the transmembrane part of OmpA joined by a linker sequence; Q
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX
Article
Biochemistry (22) Debnath, D., Nielsen, K. L., and Otzen, D. E. (2010) In vitro association of fragments of a β-sheet membrane protein. Biophys. Chem. 148, 112−120. (23) Wang, H., Andersen, K., Vad, B. S., and Otzen, D. E. (2013) OmpA can form folded and unfolded oligomers. Biochim. Biophys. Acta, Proteins Proteomics 1834, 127−136. (24) Wright, C. F., Teichmann, S. A., Clarke, J., and Dobson, C. M. (2005) The importance of sequence diversity in the aggregation and evolution of proteins. Nature 438, 878−881. (25) Rowling, P. J., Sivertsson, E. M., Perez-Riba, A., Main, E. R., and Itzhaki, L. S. (2015) Dissecting and reprogramming the folding and assembly of tandem-repeat proteins. Biochem. Soc. Trans. 43, 881−888. (26) Otzen, D. E., and Oliveberg, M. (1999) Salt-induced detour through compact regions of the protein folding landscape. Proc. Natl. Acad. Sci. U. S. A. 96, 11746−11751. (27) Schweizer, M., Hindennach, I., Garten, W., and Henning, U. (1978) Major proteins of the Escherichia coli outer cell envelope membrane. Interaction of protein II with lipopolysaccharide. Eur. J. Biochem. 82, 211−217. (28) Dornmair, K., Kiefer, H., and Jähnig, F. (1990) Refolding of an integral membrane protein. OmpA of Escherichia coli. J. Biol. Chem. 265, 18907−18911. (29) Kleinschmidt, J. H., Wiener, M. C., and Tamm, L. K. (1999) Outer membrane protein A of E. coli folds into detergent micelles, but not in the presence of monomeric detergent. Protein Sci. 8, 2065− 2071. (30) Otzen, D. E. (2005) Conformational detours during folding of a collapsed state. Biochim. Biophys. Acta, Proteins Proteomics 1750, 146− 153. (31) Tanford, C. (1968) Protein Denaturation. Part A. Characterization of the denatured state. Adv. Protein Chem. 23, 121−217. (32) Tanford, C. (1968) Protein Denaturation. Part B. The transition from native to denatured state. Adv. Protein Chem. 23, 121−282. (33) Hong, H., and Tamm, L. K. (2004) Elastic coupling of integral membrane protein stability to lipid bilayer forces. Proc. Natl. Acad. Sci. U. S. A. 101, 4065−4070. (34) Broecker, J., and Keller, S. (2013) Impact of urea on detergent micelle properties. Langmuir 29, 8502−8510. (35) Han, J. H., Batey, S., Nickson, A. A., Teichmann, S. A., and Clarke, J. (2007) The folding and evolution of multidomain proteins. Nat. Rev. Mol. Cell Biol. 8, 319−330. (36) Bhaskara, R. M., and Srinivasan, N. (2011) Stability of domain structures in multi-domain proteins. Sci. Rep. 1, 40. (37) Bennett, M. J., Schlunegger, M. P., and Eisenberg, D. (1995) 3D domain swapping: a mechanism for oligomer assembly. Protein Sci. 4, 2455−2468. (38) Rousseau, F., Schymkowitz, J., and Itzhaki, L. S. (2012) Implications of 3D domain swapping for protein folding, misfolding and function. Advances in experimental medicine and biology 747, 137− 152. (39) Ogihara, N. L., Ghirlanda, G., Bryson, J. W., Gingery, M., DeGrado, W. F., and Eisenberg, D. (2001) Design of threedimensional domain-swapped dimers and fibrous oligomers. Proc. Natl. Acad. Sci. U. S. A. 98, 1404−1409. (40) Rousseau, F., Schymkowitz, J. W., Wilkinson, H. R., and Itzhaki, L. S. (2001) Three-dimensional domain swapping in p13suc1 occurs in the unfolded state and is controlled by conserved proline residues. Proc. Natl. Acad. Sci. U. S. A. 98, 5596−5601. (41) Koebnik, R. (1999) Structural and functional roles of the surface-exposed loops of the beta-barrel membrane protein OmpA from Escherichia coli. J. Bacteriol. 181, 3688−3694. (42) Garavito, R. M., and Ferguson-Miller, S. (2001) Detergents as tools in membrane biochemistry. J. Biol. Chem. 276, 32403−32406. (43) Døvling Kaspersen, J., Moestrup Jessen, C., Stougaard Vad, B., Skipper Sørensen, E., Kleiner Andersen, K., Glasius, M., Pinto Oliveira, C. L., Otzen, D. E., and Pedersen, J. S. (2014) Low-resolution structures of OmpA-DDM protein-detergent complexes. ChemBioChem 15, 2113−2124.
(44) Kleinschmidt, J. H., and Tamm, L. K. (1996) Folding intermediates of a beta-barrel membrane protein. Kinetic evidence for a multi-step membrane insertion mechanism. Biochemistry 35, 12993−13000. (45) Arunmanee, W., Pathania, M., Solovyova, A. S., Le Brun, A. P., Ridley, H., Basle, A., van den Berg, B., and Lakey, J. H. (2016) Gramnegative trimeric porins have specific LPS binding sites that are essential for porin biogenesis. Proc. Natl. Acad. Sci. U. S. A. 113, E5034−E5043. (46) de Cock, H., Pasveer, M., Tommassen, J., and Bouveret, E. (2001) Identification of phospholipids as new components that assist in the in vitro trimerization of a bacterial pore protein. Eur. J. Biochem. 268, 865−875. (47) Jansen, C., Heutink, M., Tommassen, J., and De Cock, H. (2000) The assembly pathway of outer membrane protein PhoE of Escherichia coli. Eur. J. Biochem. 267, 3792−3800. (48) Apic, G., Gough, J., and Teichmann, S. A. (2001) Domain combinations in archaeal, eubacterial and eukaryotic proteomes. J. Mol. Biol. 310, 311−325. (49) Otzen, D. E., Kristensen, P., and Oliveberg, M. (2000) Designed protein tetramer zipped together with an Alzheimer sequence: a structural clue to amyloid assembly. Proc. Natl. Acad. Sci. U. S. A. 97, 9907−9912. (50) Richardson, J. S., and Richardson, D. C. (2002) Natural β-sheet proteins use negative design to avoid edge-to-edge aggregation. Proc. Natl. Acad. Sci. U. S. A. 99, 2754−2759. (51) Parrini, C., Taddei, N., Ramazzotti, M., Degl’Innocenti, D., Ramponi, G., Dobson, C. M., and Chiti, F. (2005) Glycine residues appear to be evolutionarily conserved for their ability to inhibit aggregation. Structure 13, 1143−1151. (52) Werbeck, N. D., Rowling, P. J., Chellamuthu, V. R., and Itzhaki, L. S. (2008) Shifting transition states in the unfolding of a large ankyrin repeat protein. Proc. Natl. Acad. Sci. U. S. A. 105, 9982−9987. (53) Bulieris, P. V., Behrens, S., Holst, O., and Kleinschmidt, J. H. (2003) Folding and insertion of the outer membrane protein OmpA is assisted by the chaperone Skp and by lipopolysaccharide. J. Biol. Chem. 278, 9092−9099.
R
DOI: 10.1021/acs.biochem.6b01153 Biochemistry XXXX, XXX, XXX−XXX