Membrane proteins are difficult to work with due to their insolubility in aqueous solution and quite often their poor stability in detergent micelles. Here, we present the peptidisc for their facile capture into water-soluble particles. Unlike the nanodisc, which requires scaffold proteins of different lengths and precise amounts of matching lipids, reconstitution of detergent solubilized proteins in peptidisc only requires a short amphipathic bi-helical peptide (NSPr) and no extra lipids. Multiple copies of the peptide wrap around to shield the membrane-exposed part of the target protein. We demonstrate the effectiveness of this ‘one size fits all’ method using five different membrane protein assemblies (MalFGK2, FhuA, SecYEG, OmpF, BRC) during ‘on-column’, ‘in-gel’, and ‘on-bead’ reconstitution embedded within the membrane protein purification protocol. The peptidisc method is rapid and cost-effective, and it may emerge as a universal tool for high-throughput stabilization of membrane proteins to advance modern biological studies.
10.7554/eLife.34085.002eLife digest
Surrounding every living cell is a biological membrane that is largely impermeable to water-soluble molecules. This hydrophobic (or “water-hating”) barrier preserves the contents of the cell and also regulates how the cell interacts with its environment. This latter function is critical and relies on a class of proteins that are embedded within the membrane and are also hydrophobic.
The hydrophobic nature of membrane proteins is however inconvenient for biochemical studies which usually take place in water-based solutions. Therefore, membrane proteins are under-represented in biological research compared to the water-soluble ones, even though roughly one quarter of a cell’s proteins are membrane proteins. Researchers have developed a few tricks to keep membrane proteins soluble after they have been extracted from the membrane. An old but popular technique makes use of detergents, which are chemicals with opposing hydrophobic and hydrophilic properties (hydrophilic literally means “water-loving”). However, even mild detergents can damage membrane proteins and will sometimes lead to experimental artifacts. More recent tricks to stabilize membrane proteins without detergents have been described but remain laborious, costly or difficult to perform.
To overcome these limitations, Carlson et al. developed a simple method to stabilize membrane proteins without detergent. Called the “peptidisc”, the method uses multiple copies of a unique peptide – a short sequence of the building blocks of protein – that had been redesigned to have optimal hydrophobic and hydrophilic properties. The idea was that the peptides would wrap around the hydrophobic parts of the membrane protein, and shield them from the watery solution. Indeed, when Carlson et al. mixed this peptide with five different membrane proteins from bacteria, all were perfectly soluble and functional without detergent. The ideal ratio of peptide needed to form a peptidisc around each membrane protein was reached automatically, without having to test many different conditions. This indicates that the peptidisc acts like a “one size fits all” scaffold.
The peptidisc is a new tool that will allow more researchers, including those who are not expert biochemists, to study membrane proteins. This will yield a better understanding of the structure of a cell’s membrane and how it interacts with the environment. Since the approach is both simple and easy to apply, more membrane proteins can now also be included in high-throughput searches for potential new drugs for various medical conditions.
peptidemembrane proteinnanoparticlesstructural biologyself-assemblyResearch organismE. colihttp://dx.doi.org/10.13039/501100000024Canadian Institutes of Health Research74525MOPDuongFranck Van Hoahttp://dx.doi.org/10.13039/501100000038Natural Sciences and Engineering Research Council of CanadaDiscovery Grant 2796BeattyJ Thomashttp://dx.doi.org/10.13039/501100000233Genome British ColumbiaSOF153BeattyJ ThomasThe funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.Author impact statementA simple, yet elegant method for robust self-assembly of diverse membrane proteins into soluble peptide nanoparticles for their structural and functional analysis in detergent-free solutions.Introduction
Membrane proteins play essential roles, such as membrane transport, signal transduction, cell homeostasis, and energy metabolism. Despite their importance, obtaining these proteins in a stable non-aggregated state remains problematic. Membrane proteins are generally purified in detergent micelles, but these small amphipathic molecules are quite often detrimental to protein structure and activity, in addition to interfering with downstream analytical methods. This drawback has led researchers to develop detergent-free alternatives such as amphipols (Popot, 2010), SMALPs (Lee et al., 2016), saposin-lipoparticles (Frauenfeld et al., 2016), and the popular nanodisc system (Bayburt et al., 2006; Denisov et al., 2004). In the latter case, two amphipathic membrane scaffold proteins (MSPs) derived from apoA1 wrap around a small patch of lipid bilayer containing the target membrane protein (Bayburt et al., 2006; Denisov et al., 2004; Denisov and Sligar, 2016). However, in spite of an apparent simplicity, the formation of a nanodisc depends on several factors such as lipid to protein ratio, scaffold length, nature of lipids, rate of detergent removal and overall amenability of the target for re-assembly into lipid bilayer (Bayburt et al., 2006; Denisov et al., 2004; Hagn et al., 2013). The method is therefore not trivial and small deviations from optimal conditions often leads to low-efficiency reconstitution, or else liposome formation or protein aggregation (Bayburt et al., 2006).
Peptides have been considered as an alternative to scaffold proteins. Peptergents (Corin et al., 2011), lipopeptides (Tao et al., 2013), nanostructured [beta]-sheet peptides (Privé, 2009), and bi-helical derivatives of the ApoA1-mimetic 18A peptide, termed ‘beltides’ (Larsen et al., 2016), have been reported in the recent years. Yet, these systems have not been widely adopted for structural or functional characterization of membrane proteins for various reasons. Peptergents and lipopeptides can solubilize membrane proteins directly from a lipid bilayer, but same as detergent these peptides form mixed micelles that readily aggregate and precipitate below a certain critical micellar concentration (Corin et al., 2011; Tao et al., 2013). Beltides were shown to trap bacteriorhodopsin in solution, but the method required prior incubation with specific amounts of lipids and the particles formed were reportedly unstable at physiological temperatures (Larsen et al., 2016). Cost and complexity of the peptide can also be problematic. Nanostructured [beta]-sheet peptides contain extended alkyl chains covalently linked to glycine residues, while lipopeptides need to be covalently linked to a lipid molecule (Tao et al., 2013; Privé, 2009). These hydrophobic peptides are also difficult to work with given their low solubility. Peptergents require titration of base to become soluble in aqueous solution (Corin et al., 2011), and [beta]-sheet peptides can form extended filament clusters without detergent (Privé, 2009). Thus, a peptide-based reconstitution method that is cost-effective, rapid, unhindered by issues of solubility and generally applicable to membrane proteins remains to be developped.
We present the peptidisc. The peptidisc is made by multiple copies of an amphipathic bi-helical peptide (hereafter termed NSPr) wrapping around its target membrane protein. In contrast to the nanodisc, no additional lipids are necessary during reconstitution, except those that have co-purified with the protein (i.e. annular lipids). The peptide design we employ is a reverse version of the original nanodisc scaffold peptide (NSP), which consists of two repeats of the ApoA1-derived 18A peptide joined by a flexible linker proline (Kariyazono et al., 2016; Chung et al., 1985), in addition to two leucine residues which are substituted by phenylalanines to increase lipid affinity (Kariyazono et al., 2016; Mishra et al., 2008) (Supplementary file 1). The original NSP can stabilize lipid particles (Kariyazono et al., 2016), but issues of water-solubility and adaptability to membrane proteins were not demonstrated. We show here the NSPr design is very effective for stabilizing both α-helical and β-barrel membrane proteins of different size, topology, and complexity.
Results
The original NSP is able to capture lipids from a lipid bilayer and forms discoidal particles of varying size (Kariyazono et al., 2016). However, due to its hydrophobicity, the NSP peptide is difficult to solubilize; the solution is cloudy after resuspension in water (Figure 1B). To increase peptide solubility, we reversed the amino acid sequence of NSP, so that the two amphipathic helices have a slightly altered orientation, leading to a lower hydrophobic moment and higher overall electropotential (Figure 1AC, Supplementary file 1). We also forewent acetylation and amidation of the peptide termini so that modifications such biotinylation remain possible. These small modifications allowed the final design (termed NSPr) to be fully dissolved in water at concentrations up to 25 mg/ml (Figure 1B).
10.7554/eLife.34085.003
Solubility test of NSP and NSP<sub>r</sub>.
(A) Peptide models computed by the 3D-hydrophobic moment peptide calculator. The direction of hydrophobic moment is indicated by a red line. Peptides are oriented with their N to C-terminus from bottom (red) to top (blue). (B) Turbidity measurement of peptide suspension. The absorbance of light at 550 nm for NSP (blue squares, 15 mg/mL) and NSPr (red circles, 25 mg/mL), re-suspended in distilled water (dH2O) were compared to a dH2O control (green triangles). (C) Calculated electropotential and hydrophobic moment of peptide variants. Calculations were performed using the 3D-hydrophobic moment peptide calculator as described in Materials and methods.
We tested the ability of NSPr to capture the ABC transporter MalFGK2 using an ‘on-column’ reconstitution method (Figure 2). The NSPr peptide was mixed with MalFGK2 in dodecyl maltoside and the mixture applied immediately onto a size exclusion column equilibrated in a detergent-free buffer (Figure 2A). The collected particles (hereafter termed peptidisc or MalFGK2-NSPr) were soluble and monodisperse, as shown by clear-native CN-PAGE and blue-native BN-PAGE (Figure 2B). We determined the approximate peptide content using SDS-PAGE (Figure 3A). Also, since annular lipids are tightly bound to membrane proteins (Bechara et al., 2015), we determined the final lipid content in the peptidisc using thin layer chromatography and photocolorimetric methods (Figure 4A and B). This analysis indicated a stoichiometry of 10 ± 2 peptides and 41 ± 10 lipids per MalFGK2 (Table 1). This stoichiometry allowed us to calculate the mass of the MalFGK2 peptidisc to 251 ± 12 kDa (Table 1). Interestingly, the lipids identified in the TLC analysis were predominantly negative phospholipids, cardiolipin and phosphatidylglycerol (Figure 4B). These lipids play a role in regulation of MalFGK2 by stabilizing interactions with the regulatory protein EIIA (Bao and Duong, 2013). To corroborate the peptide and lipid stoichiometry, we determined the molecular weight of the intact complex by native mass spectrometry (247 ± 24 kDa; Table 1, Figure 5A and B), and size exclusion chromatography coupled multi-angle light scattering (SEC-MALS) (250 ± 17 kDa; Table 2, Figure 6A). The SEC-MALS analysis showed that the peptidisc remains perfectly stable during storage (e.g. 3 days at 4°C). We then examined the particles by single particle negative-stain electron microscopy (Figure 2C). The 2D-class averages revealed a structure very similar to MalFGK2 in nanodiscs, (Fabre et al., 2017) with distinctly visible elements such as the MalK2 dimer, the periplasmic P2 loop and a larger discoidal density corresponding to the NSPr peptides wrapping around the MalFG membrane domain. The measured diameter of the peptidisc was 11.7 ± 1.4 nm, which is consistent with a stoichiometry of 12 ± 2 peptides per MalFGK2 complex when arranged in the double-belt model (Table 2). Finally, the ATPase activity of MalFGK2 in peptidisc was similar to that reported in proteoliposomes and in nanodiscs (Bao and Duong, 2012), in sharp contrast to the high and unregulated ATPase activity observed in detergent micelles (Figure 2D). Importantly, the structural integrity of the peptidisc remained stable at elevated temperatures, with ~80% of MalFGK2 peptidisc intact after incubation for 3 hr at 30°C (Figure 6B).
10.7554/eLife.34085.004
The ‘on-column’ reconstitution of MalFGK<sub>2</sub>.
(A) Typical size-exclusion chromatography of MalFGK2 in peptidisc (MalFGK2-NSPr) using the ‘on-column’ method. (B) CN-PAGE and BN-PAGE analysis of MalFGK2 in detergent micelle (DDM), nanodisc (MSP1D1), and peptidisc (NSPr). (C) Top panel: Field of view of particles stained with uranyl formate. Bottom panel: Selected class averages representing three characteristic views of MalFGK2 in peptidisc. The nucleotide-binding domains (MalK2), the transmembrane domain (MalFG), and periplasmic P2-loop are indicated with yellow, red and blue arrows, respectively. (D) Maltose-dependent ATPase activity of MalFGK2 (0.5 µM) reconstituted in detergent (DDM), proteoliposomes (PL), peptidiscs (NSPr), and nanodiscs (MSP1D1) obtained at 30°C in the presence or absence of MalE (2.5 µM). Error bars represent standard deviations from three separate experiments.
10.7554/eLife.34085.005
Quantification of NSP<sub>r</sub> in peptidiscs.
(A) Left panel; 15% SDS-PAGE analysis of MalFGK2 in peptidisc or DDM. NSPr runs at the bottom of the gel and can be visualized with Coomassie blue staining. Dye fluorescence was measured on a LICOR Odyssey scanner and quantified by Image J. Right panel; Standard curve derived from NSPr titration measurement (black dots), and average intensity of NSPr fluorescence from MalFGK2 peptidisc (red dot). (B) Left Panel: Western Blot of FhuA-peptidisc reconstituted into NSPrbio, and visualized by incubation with Streptavidin-Alexa 680. Fluorescence of the Alexa 680 dye was measured on a LICOR Odyssey scanner (700 nm, excitation 680 nm) and quantified in Image J. Right Panel: Standard curve as in A. (C) 15% SDS-PAGE analysis of BRC in peptidisc. The MLH subunits of BRC partially resist denaturation by SDS, resulting in a higher molecular weight band located above the single subunits. Each gel was repeated in triplicate with independent standard curves to calculate the values reported in Table 1.
10.7554/eLife.34085.006
Quantification of phospholipids trapped in peptidiscs.
(A) Calculated number of phospholipids per peptidisc. Phospholipid content was determined by Malachite green assay after acid digestion of lipid extracts. Error bars represent standard deviation derived from three separate measurements. B) TLC analysis of lipid extracts obtained from 10 µg MalFGK2 peptidisc, 10 µg FhuA peptidiscs and 20 µg BRC peptidiscs, as well as pure lipid standards Cardiolipin (CL), 1,2-dioleoyl-sn-glycero-3-phosphoglycerol (PG), and 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine (PE).
10.7554/eLife.34085.007
Native mass spectrometry of intact peptidiscs.
Panels A, C, E are mass spectra acquired in positive ion mode for aqueous ammonium acetate solutions (100 mM, pH 7, 22°C) of MalFGK2-NSPr, BRC-NSPr and FhuA-NSPr, respectively. Panels B, D, and F are deconvoluted mass spectra of the peptidiscs shown in A, C, and E, respectively.
10.7554/eLife.34085.008
Stability of MalFGK<sub>2</sub> peptidisc.
(A) Multi-angle light scattering analysis of MalFGK2 reconstituted in peptidisc. MalFGK2-NSPr (100 µg) was left for 3 days at 4°C before analysis by SEC-MALS. Protein sample was injected and protein concentration tracked through differential refractive interferometry (dRI, black trace). Molecular weight was calculated for the fractions corresponding to the peak of MalFGK2-NSP (red trace). (B) Structural stability of MalFGK2-NSPr. The MalFGK2 peptidisc was incubated at 30°C in Buffer A for the indicated time, then analyzed by BN-PAGE.
10.7554/eLife.34085.009
Calculated and observed molecular weight and scaffold stoichiometry of peptidiscs
Peptidisc
Molecular weight measured by ESI-MS (kDa)
Measured NSPr stoichiometry (NSPr/disc)
Measured lipid stoichiometry (Lipid/Disc)
Calculated molecular weight* (kDa)
MalFGK2-NSPr
247 ± 24
10 (±2): 1
41 (±10): 1
251 ± 12
BRC-NSPr
138 ± 18
9 (±1): 1
4 (±1): 1
138 ± 5
FhuA-NSPr
137 ± 18
10 (±2): 1
8 (±3): 1
131 ± 9
*The formula for the calculated molecular weight is as follows: MWpeptidisc = MW(protein)+n(MWNSPr)+m(MWLipid); where n is the measured NSPr stoichiometry, m is the measured lipid stoichiometry, MWLipid = 0.8 kDa, MWNSPr = 4.5 kDa, and MWprotein = 173 kDa, 80 kDa, and 94 kDa for MalFGK2, FhuA, and BRC, respectively. For NSPr and Lipid stoichiometry, the standard deviation is derived from three separate measurements.
10.7554/eLife.34085.010
Molecular weight, diameter, and scaffold stoichiometry of MalFGK<sub>2</sub> reconstituted in peptidisc.
Measured molecular weight (kDa)*
Measured diameter †
Calculated stoichiometry (scaffold/disc)‡
MalFGK2 Peptidisc
250 ± 17
11.7 ± 1.4
12 (±2):1
*Molecular weight calculated from SEC-MALS data (Fig. S1). The standard error is derived from three independent SEC-MALS experiments. †Diameter of MalFGK2-peptidiscs determined by negative stain electron microscopy Figure 1B, assuming a perfectly circular shape. ‡Stoichiometry (n), based on the measured diameter of the particles, was calculated with the following formula: π(ddisc-2dα-helix) = (n/2)LNSPr; where dα-helix represents the diameter of an alpha-helix (0.5 nm), ddisc represents the measured disc diameter, and LNSPr represents length of the NSPr peptide.
We also incorporated β-barrel membrane proteins in peptidisc, using the FhuA receptor as a model protein (Figure 7A). Analysis of FhuA-peptidisc by native mass spectrometry indicated a molecular weight of 138 ± 17 kDa (Figure 5E and F). Quantitation of the individual peptide and lipid components, following the approach applied to MalFGK2 above, indicated an average of 8 ± 3 lipids and 10 ± 2 NSPr per FhuA (Table 1, Figure 3B and Figure 4A). These measurements were used to estimate a molecular mass of 131 ± 9 kDa (Table 1). Binding analysis on CN-PAGE further showed that FhuA in peptidisc is functional for both TonB and colicin M (Figure 7B). The binding of TonB and colicin M is modulated by the ligand ferricrocin (Figure 7B), as previously reported in vivo and in vitro (Mills et al., 2014; Wayne et al., 1976). The peptidisc is therefore suitable for the functional reconstitution of both α-helical and β-barrel membrane proteins.
10.7554/eLife.34085.011
Binding activity of FhuA in nanodiscs and peptidiscs.
(A) Typical SEC profile of FhuA reconstituted in peptidisc (FhuA-NSPr) using an ‘on-column’ reconstitution protocol as described in Materials and methods. (B) The FhuA transporter reconstituted in nanodiscs (FhuA-MSPL156) or peptidiscs (FhuA-NSPr) was incubated with the C-terminal TonB23-329 fragment (2 µg) or with colicin M (5 µg), with or without ferricrocin as indicated. Samples were analyzed by CN-PAGE and Coomassie-blue staining of the gel.
The ‘in-gel’ method (Figure 8) was developed to determine optimal reconstitution conditions in a time- and cost-effective manner. Small amounts of peptides (0–2.5 µg) were mixed with the target protein (~1.25 µg) in detergent solution, and the resulting mixture immediately loaded on native gel. In that case, removal of the non-ionic detergent occurs during electrophoresis when the protein-peptide mixture enters the detergent-free part of the gel. At the correct NSP ratio, the target membrane protein does not aggregate at the top of the gel but instead migrates in a soluble form to its expected molecular weight position. This simple method allowed us to estimate the effective NSP concentrations required to trap four different integral membrane complexes into a peptidisc: MalFGK2 (Figure 8A), FhuA (Figure 8B), the trimeric OmpF porin (Figure 8C), and the membrane translocon SecYEG (Figure 8D). These membrane proteins were generally reconstituted at similar peptide concentrations, with a half-maximal molar ratio (RR50) of 20 (Figure 8E and F). This is significantly higher than the measured stoichiometry of ~10 peptides per protein complex (Table 1), suggesting that excess peptide is needed to achieve efficient assembly. This analysis also showed that the SecYEG complex can be trapped as a dimer and higher order oligomeric form in peptidisc (Figure 8D), probably due to the self-association of this complex in detergent solution, as shown before (Bessonneau et al., 2002). This later observation further differentiates the peptidisc from the nanodisc. In the nanodisc, the selective reconstitution of the SecYEG monomer and dimer requires MSP proteins of different lengths (Figure 8—figure supplement 1) (Dalal et al., 2012).
10.7554/eLife.34085.012
Express ‘in-gel’ method for determining optimal reconstitution ratio.
(A) NSP and MalFGK2 were mixed at the indicated molar ratio for 2 min in Buffer A containing a low amount of detergent (~0.008% DDM) before loading onto CN-PAGE. Peptidisc reconstitution occurs during migration in the detergent-free gel environment. The same experiment was performed with. (B) FhuA,. (C) OmpF3,. (D) SecYEG. (E) Reconstitution efficiency of FhuA as a function of the NSP concentration. The protein band FhuA-NSP in B) was quantified with Image J and the data plotted as log (mol NSP/mol FhuA). The data were fitted with a Boltzmann sigmoidal function to generate a curve describing the reconstitution efficiency and the half-maximal reconstitution ratio (RR50). (F) The RR50 was determined for other target proteins as in (E). Error bars represent the standard deviation from three separate reconstitution experiments.
10.7554/eLife.34085.013
Capture of SecYEG monomer and dimer in peptidisc and nanodisc.
(A) Left panel; the SecYEG complex (2.5 µg) was incubated with the indicated scaffolds proteins (1.25 µg each) in Buffer A + 0.02% DDM. The sample was diluted threefold in detergent-free Buffer A and immediately analyzed by CN-PAGE. Right panel; the experiment on left panel was performed without addition of SecYEG. (B) Illustration representing possible reconstitution products of the SecYEGn complex into MSP1D1 and NSPr. The subunits SecY, SecE, and SecG are represented as blue, pink, and magenta, respectively.
To save on peptide consumption, as well as to minimize exposure of the target to detergent, we developed the ‘on-beads’ reconstitution method (Figure 9), wherein affinity-purification of the protein and incorporation into peptidisc are carried out simultaneously. As illustrated in Figure 9A, the peptide was added in excess while the membrane protein still bound to the beads, followed by detergent dilution and eventually elution in a detergent-free buffer (Figure 9A, step 5). The method was tested using the his-tagged MalFGK2 complex (Figure 9B). Analysis of the eluted complex by BN-PAGE and CN-PAGE showed that MalFGK2 is readily incorporated into peptidiscs (Figure 9C), with purity and yield as good as with conventional detergent-based chromatography (Figure 9C). As an added advantage, the excess peptide collected before the elution step was re-used for the next round of ‘on-beads’ reconstitution, thereby reducing the cost of large-scale reconstitution.
10.7554/eLife.34085.014
Direct ‘on-beads’ reconstitution during membrane protein purification.
(A) Principle of the ‘on-beads’ reconstitution. Step 1: the tagged protein is extracted from the membrane with excess of detergent buffer (>CMC) and incubated with the affinity resin. Step 2: The beads are washed twice with the detergent buffer near its critical micelle concentration (~CMC). Step 3: The beads are incubated with buffer containing excess NSP and limited amount of detergent (<CMC). Step 4: The beads are washed in detergent-free buffer to remove unbound NSP and residual detergent. Step 5: The protein captured in peptidiscs is eluted from the column in detergent-free solution. (B) SDS-PAGE and. (C) Native-PAGE analysis of the his-tagged MalGFK2 complex purified following conventional detergent method and ‘on-beads’ peptidisc detergent-free method.
Finally, we reconstituted the photosynthetic bacterial reaction center (BRC) from Rhodobacter sphaeroides, given its potential for biotechnological application. The BRC is employed in bio-hybrid solar cells due to its ability to absorb light in the near-infrared with high quantum efficiency (Blankenship et al., 2011; Ravi and Tan, 2015; Yaghoubi et al., 2017). However, sustained heat and light exposure lead to irreversible loss of pigments and protein denaturation (Hughes et al., 2006; Scheidelaar S et al., 2014). Thermal stability and solubility at high protein concentration are therefore important parameters for successful application. The molecular weight of the BRC peptidisc measured by native mass spectrometry is 138 ± 18 kDa (Table 1, Figure 5C and D). It has been shown that purification of BRC in LDAO delipidates the complex (Scheidelaar S et al., 2014), and accordingly the BRC peptidisc contained only 4 ± 1 phospholipids (Table 1, Figure 4A). Analysis by SDS-PAGE indicated a stoichiometry of 9 ± 1 NSP per BRC (Table 1, Figure 3C). The calculated molecular weight was 138 ± 5 kDa, in excellent agreement with native mass spectrometry data (Table 1). We next measured the stability of the BRC pigments in peptidiscs or in LDAO detergent (Figure 10—figure supplement 1A and B). The spectral properties of the BRC complex were similar in both environments (Figure 10A). However, the BRC in peptidisc resisted denaturation 65°C for 1 hr, while it was fully denatured in less than 4 min in LDAO (Figure 10B). This difference corresponds to ~100 fold increase of the half-life of the BRC complex at elevated temperatures (Figure 10C).
10.7554/eLife.34085.015
Thermostability of the BRC complex in peptidiscs.
(A) Absorbance scans of the BRC (1 µM) in detergent solution (0.03% LDAO, red trace) and in peptidisc (black trace). Scans were normalized to the value measured at 803 nm (the absorbance peak of the accessory bacteriochlorophylls). (B) Decrease in absorbance of the BRC at 803 nm after incubation at 65°C for the indicated time. (C) Calculated half-life of the BRC in peptidisc and LDAO at 65°C. The data in B) were fit with an exponential decay function to determine the corresponding half-life. Error bars represent the standard deviation from three separate experiments.
10.7554/eLife.34085.016
Effect of peptidisc on BRC stability.
(A) Absorbance scans of the BRC complex (1 µM) in peptidisc after incubation at 65°C for up to 1 hr (blue traces). Incubation at 90°C leads to full release of the bacteriochlorophyll pigment (magenta trace). (B) Absorbance scans of the BRC (1 µM) after incubation at 65°C in 0.03% LDAO for up to 4 min. (C) Fluorescence of the BRC in peptidisc (green trace), 0.1% LDAO (red trace), 0.02% DDM (blue trace), and 0.1% SDS (black trace). The BRC (1 µM) was incubated for 5 min at the indicated temperature before fluorescence was measured (700 nm; excitation at 680 nm). (D) The experiment repeated as in C), with BRC reconstituted into MSP1D1 (1:2 BRC:MSP1D1 molar ratio, brown trace), SMA (0.1%, grey trace), Proteoliposomes (1:1600:400 BRC:DOPC:DOPG), and peptidiscs (green trace). Fluorescence were normalized to 100% after denaturation for 5 min at 90 degrees. (E) Data were fitted using a Boltzmann sigmoidal function to calculate the melting temperature (Tm). (F) Analysis of reconstituted BRC fractions on BN-PAGE (left panel) and SDS-PAGE (right panel).
We also compared the thermal stability of the BRC complex when reconstituted without additional lipids in SMA polymer and nanodiscs (Figure 10—figure supplement 1C and D). Without added lipid, the diameter of the BRC is too small for the MSP1D1 belt, resulting in some protein aggregation and heterogenous nanodisc preparation (Figure 10—figure supplement 1F, left panel, lane 4). Reconstitution into the SMA polymer without lipids was monodisperse (Figure 10—figure supplement 1F, left panel, lane 5) and the thermostability was higher than in LDAO (Figure 10—figure supplement 1E). This observation is contrary to previous reports which suggest that the lipid environment in the SMA particle is what increases BRC thermostability. (Scheidelaar S et al., 2014) In our hands, reconstitution of the BRC into peptidiscs, proteoliposomes, low-lipid nanodiscs and SMA particles all result in comparable thermostability (Figure 10—figure supplement 1E). From these results, it appears more important to remove detergents than include lipids to increase thermal tolerance of the BRC.
Discussion
Detergents remain the most effective way to extract and purify membrane proteins, yet these surfactants have many undesired effects on protein stability and downstream biochemical analysis. To circumvent these difficulties, and to handle these proteins like their water-soluble counterparts, membrane proteins are more often reconstituted with amphipathic scaffolds. However, current methods of reconstitution are difficult because each scaffold system has specific properties and limitations, and each requires substantial optimization. The aim of our work was to develop a ‘one-size fits all’ method to streamline the capture of membrane proteins in detergent-free solution.
We show that the peptidisc is a simple and efficient way to replace detergent. The system works with membrane proteins of different size, fold and complexity, and does not require addition of exogenous lipids. The peptidisc captures membrane proteins regardless of their initial lipid content, and therefore the use of exogenous lipids to match the diameter of the scaffold such as in the nanodisc system is avoided. Since the binding of the peptide is essentially guided by the size and shape of the protein template, the peptide stoichiometry is also self-determined. As a direct consequence, the preparation of peptidisc is possible through rapid detergent removal techniques such as ‘in-gel’, ‘on-column’ and ‘on-bead’ methods. Each of these methods has considerable advantages compared to overnight dialysis and ‘biobeads’ techniques traditionally used for scaffold reconstitution. The ‘in-gel’ method is fast, high throughput and requires only 1–2 µg of the precious target protein in order to identify the optimal peptide ratio (Figure 8). The ‘on-column’ method allows direct preparation of large quantities of peptidiscs with simultaneous removal of salts, detergent and excess peptides. However, gel filtration requires concentrated protein samples and unbound peptide is wasted. The ‘on-bead’ method is therefore ideal for the peptidisc reconstitution because protein purification, detergent removal, trapping in peptidisc and recovering of unbound peptide are carried out simultaneously in the same tube. Additionally, because long exposure or storage of the protein in detergent is avoided, the ‘on-bead’ method can be especially advantageous in the case of unstable membrane proteins.
Beside these methodological considerations, we show that the peptidisc maintain proteins in a functional state. For instance, both FhuA and MalFGK2 retain their ability to interact with their soluble binding partners. In contrast to the detergent, the ATPase activity of MalFGK2 in peptidisc regains dependence to substrate and maltose-binding protein MalE, indicating a return to the transporter’s native membrane conformation (Bao and Duong, 2012). Negative-stain electron microscopy of MalFGK2 shows good resolution for the periplasmic P2 loop and cytosolic ABC domains, whereas the transmembrane domains MalFG are surrounded by an extra density corresponding to the peptide belt and naturally present lipids (Figure 2C). Lipid quantitation indeed indicates that ~ 40 lipids molecules, mostly acidic, are captured with MalFGK2 (Table 1, Figure 4A). In the case of the BRC complex however, there must be direct protein-peptide contact because almost no lipids are detected in the final assembly (Table 1, Figure 4A and B). Despite the absence of lipids, the thermal stability of the BRC in peptidisc is still much higher than in detergent, with a melting temperature similar to that observed proteoliposomes (Figure 10—figure supplement 1E). Clearly, the peptidisc is more than a simple surrogate of the detergent molecules; the peptide assembly forms an environment that stabilize and support the transmembrane domains of a membrane protein from aqueous solution.
The peptidisc offers distinct advantages when compared to other amphipathic scaffolds. Unlike beltides, peptergents, lipopeptides, and nanostructured [beta]-sheet peptides, the NSPr does not require modifications at its N- or C-termini (Corin et al., 2011; Tao et al., 2013; Privé, 2009; Larsen et al., 2016). Both ends of NSPr remain available for modifications, such as biotinylation, which we show does not affect peptidisc assembly (Figure 3B). Importantly, peptergents and lipopeptides have critical micelle concentrations and dynamically exchange in solution. Because of this instability, an excess of these peptides is always needed to keep membrane proteins in solution (Corin et al., 2011; McGregor et al., 2003). This is not the case for the peptidisc because NSPr binds strongly to its target, allowing free peptides to be removed without compromising peptidisc stability. The length of the NSPr also does not need to be adjusted to the diameter of the target protein. This property may be especially advantageous when capturing macromolecular membrane protein complexes or proteins that exist in oligomeric state, as is the case with the SecYEG complex (Figure 8—figure supplement 1B). Lastly, the NSPr is synthetically made and can be obtained in large quantity, high purity and free of immunogens usually found in recombinant cell expression systems (Schwarz et al., 2014). The structural homogeneity of NSPr is also high compared to other synthetic scaffolds, such as amphipols and styrene-maleic acids (SMA). This is because peptide synthesis is sequential, whereas addition of carboxylate and other repeating units in amphipols and industrially prepared SMA polymers is randomly distributed along the polymer chain during assembly. Due to the ‘one-pot’ synthesis, these synthetic polymer preparations are often polydisperse mixtures of different lengths (Popot, 2010; Smith et al., 2017). This compositional heterogeneity can be problematic for functional and structural studies (Deller et al., 2016). Finally, due to both positive and negative charged amino acids, the solubility of the NSPr remains high at various pH or with divalent cations. Both these conditions can destabilize assemblies formed by synthetic polymers because they largely depend on charged carboxylate groups for solubility (Popot, 2010; Gulati et al., 2014).
How exactly the peptidisc wraps around the membrane protein template remains an important question. In ApoA1 lipid nanodiscs, the two scaffold proteins arrange themselves in an anti-parallel ‘double belt’ configuration (Bibow et al., 2017). If NSPr were also arranged in a double belt, then the peptide to protein ratio would expectedly vary by factor of 2, as an odd number of scaffold would leave part of the protein exposed to the environment. However, the native mass spectrometry profiles for FhuA and BRC indicate peptidisc populations which can differ in mass by one peptide only (Figure 5D and F), suggesting an arrangement that is flexible. Possibly, the NSPr could be arranged in an orthogonal ‘picket fence’ orientation as proposed for lipopeptides, nanostructured [beta]-sheet peptides, and single helix ApoA1 mimetic peptides.(Tao et al., 2013; Privé, 2009; Islam et al., 2018). However, the length of NSPr (37 amino acids) is too long to be orthogonal while maintaining contact with hydrophobic parts of the protein or alkyl chains of annular lipids. Thus, the NSPr perhaps simply lies in a tilted orientation. A tilted orientation would facilitate optimal binding and stoichiometry as the peptide shifts in angle of association, adapting to best fit the target membrane protein template.
In conclusion, the peptidisc offers several advantages. The method is cheap, fast and seamlessly integrated in existing protein purification protocols, such as size exclusion and affinity chromatography. The peptide is relatively simple to synthesize and it can be recycled via the ‘on-bead’ method to decrease consumption. The peptidisc is not hindered by issue of buffer instability or heterogeneity as observed with other synthetic scaffold. Since the peptide self-associates to its template and without added lipids, this could be advantageous for structural studies which are affected by compositional heterogeneity. There are of course applications where other scaffold systems are better suited, such as direct protein solubilization with SMA polymers, or control over lipid environment offered by nanodiscs. Nevertheless, the current advantages of the peptidisc surely diminish the challenges associated with biochemical, structural and pharmacological characterization of purified membrane proteins. The peptidisc may emerge as a very practical way to analyse or exploit membrane proteins in a detergent-free environment.
Materials and methodsBiological reagents
Tryptone, yeast extract, NaCl, imidazole, Tris-base, acrylamide 40%, bis-acrylamide 2% and TEMED were obtained from Bioshop, Canada. Isopropyl β-D-1-thiogalactopyranoside (IPTG), ampicillin, and arabinose were purchased from GoldBio. Detergents n-dodecyl-β-d-maltoside (DDM) and octyl-β-D-glucoside (β-OG) were from Anatrace. Detergent N,N-dimethyldodecylamine N-oxide (LDAO) was from Sigma. Total E.coli lipids were purchased from Avanti Polar Lipids. Resource 15Q, Fast Flow S, Superdex 200 HR 10/300 GL and 5/150 GL were obtained from GE Healthcare. Ni2+-NTA chelating Sepharose was obtained from Qiagen. All other chemicals were obtained from Fischer Scientific Canada.
Peptides
Peptide NSPr (Nter-FAEKFKEAVKDYFAKFWDPAAEKLKEAVKDYFAKLWD-Cter) and NSP (Nter- DWLKAFYDKVAEKLKEAAPDWFKAFYDKVAEKFKEAF-Cter) were obtained from A+ peptide Co. Ltd. and Genscript (each with purity >80%). Peptide NSPrbio was obtained from KareBay. The effective purity of the peptides employed is actually higher as mass spectrometry analysis finds that main contaminants (>10%) consists of peptides missing the final aspartate residue. To aid accessibility to the academic community, bulk NSPr peptides and core protocols are available at www.peptidisc.com
Preparation of the NSP peptides
For solubility test experiments, lyophilized NSP and NSPr (purity of 82% and 85%, respectively) were resuspended in dH20 at room temperature to final concentrations of 15 mg/mL and 25 mg/mL, respectively. Peptide concentration was determined by absorbance at 280 nm. Residual TFA from peptide synthesis results in a low pH solution (pH 2–3). For all other experiments, peptides were solubilized in dH2O at 6 mg/mL. Solubilized peptides were stored at 4°C for up to 5 weeks. Immediately before use, the pH of the peptide solution was modified by addition of 20 mM Tris-HCl, pH 8 to form the so-called Assembly Buffer. Immediately before use in peptidisc reconstitutions, peptide concentration in Assembly Buffer was verified by Bradford assay (Prehna et al., 2012).
Protein expression and purification
Unless otherwise stated, all proteins were expressed in E.coli BL21(DE3) (New England Biolabs) for 3 hr at 37°C after induction at an OD of 0.4–0.7 in LB medium supplemented with required antibiotic. Cells were harvested by low-speed centrifugation (10,000 x g, 6 min) and resuspended in Buffer A (50 mM Tris-HCl: pH 8; 100 mM NaCl; 10% glycerol). Resuspended cells were treated with 1 mM phenylmethylsulfonyl fluoride (PMSF) and lysed using a microfluidizer (Microfluidics) at 10,000 psi. Unbroken cell debris and other aggregates were removed by an additional low-speed centrifugation. Cytosolic and crude membrane fractions containing the overexpressed protein of interest were subsequently isolated by ultracentrifugation (100,000 x g, 45 min) and crude membrane fraction resuspended in Buffer A (50 mM Tris-HCl: pH8, 100 mM NaCl, 10% glycerol). MalE and His-tagged MalFGK2 were purified as previously described (Bao and Duong, 2012), expressed from plasmids pBAD33-MalE and pBAD22-FGKhis, respectively. Crude membrane containing His-tagged MalFGK2 were solubilized at 4°C overnight in Buffer A + 1% DDM and clarified by ultracentrifugation. Solubilized MalFGK2 was isolated by Ni2+-chelating chromatography in Buffer A + 0.02% DDM, washed in five column volumes (CV) of Buffer B (50 mM Tris-HCl: pH 8; 200 mM NaCl; 15 mM imidazole; 10% glycerol)+0.02% DDM, and then eluted in Buffer C (50 mM Tris-HCl: pH 8; 100 mM NaCl; 400 mM imidazole; 10% glycerol)+0.02% DDM. Protein MalE was isolated on Resource 15Q column, concentrated using a 30 kDa polysulfone filter (Pall Corporation), and then further purified on Superdex 200 HR 10/300 GL column equilibrated in Buffer EQ (50mM Tris-HCl: pH 8, 50 mM NaCl, 10% glycerol). His-tagged-MSPL156 and His-tagged TonB23-329 were purified by Ni2+-chelating chromatography as previously described (Mills et al., 2014). His-tagged Colicin M was expressed and purified according to established protocols from plasmid pMLD189 in the E. coli strain BW25113 (Mills et al., 2014). His-tagged FhuA, encoded by plasmid pHX405, was expressed in E. coli strain AW740 (ΔompF, ΔompC) in M9 minimal media and was purified in LDAO as previously described (Mills et al., 2014) OmpF was expressed from E.coli JW2203 (ΔOmpC) as previously described (Jun et al., 2014). Prepared crude membrane was resuspended in Buffer A and the inner membrane solubilized by addition of 1% Triton X-100. The outer membrane fraction (OM) was isolated by ultracentrifugation, resuspended in Buffer A + 1% LDAO at a concentration of 3 mg/mL, and incubated overnight at 4°C with gentle rocking. Insoluble material was removed by an additional ultracentrifugation step, and the clarified lysate was applied onto a Resource 15Q column pre-equilibriated in Buffer EQ + 0.1% LDAO. OmpF was eluted by a linear 20 mL gradient of 50–700 mM NaCl, and further purified by Superdex 200 HR 10/300 in Buffer A + 0.1% LDAO. Expression and purification of His-tagged SecYEG was performed from the plasmid pBad22-His-EYG as previously described (Dalal et al., 2012). Crude membranes were solubilized for 1 hr at 4°C in Buffer A + 1% DDM. Solubilized material was clarified by ultra-centrifugation and passed over a 5 mL Ni2+-NTA column. After extensive washing in Buffer A + 0.02% DDM, SecYEG was eluted in Buffer A + 0.02% DDM over a 20 mL gradient of 0–600 mM imidazole. The most concentrated fractions were pooled and diluted fivefold in Buffer O (50 mM Tris-HCl: pH 8, 10% glycerol +0.02% DDM) before being applied to a 5 mL Fast Flow S cation exchange column pre-equilibrated in Buffer EQ + 0.02% DDM. Bound protein was eluted over a 20 mL gradient from 50 to 600 mM NaCl in Buffer EQ + 0.02% DDM. Plasmids pET28 encoding his-tagged MSPD1 and MSP1D1E3 proteins were transformed into BL21 cells and protein expression and purification was performed as previously described (Dalal et al., 2012). All proteins, with the exception of BRC, were flash frozen in liquid nitrogen immediately after purification and stored at −80°C for later use. BRC was purified as previously described (Bradford, 1976). In brief, His-tagged BRC was expressed in Rhodobacter Sphaeroides RcX (ΔpuhA, ΔpufQBALMX, ΔrshI, ΔppsR) using plasmid pIND4-RC1. A preculture of 10 mL in RLB media (LB medium; 810 µM MgCl2; 510 µM CaCl2)+25 µg/mL kanamycin was transferred into 100 ml of RLB-kan and grown overnight at 30°C before transfer into 1 L of freshly prepared RLB-kan. After growth for 8 hr at 30°C, BRC production was induced with 1 mM IPTG for an additional 16 hr. During growth and purification, light exposure was kept to a minimum. Cells were harvested by low-speed centrifugation, resuspended in Buffer A and lysed by French press (10,000 psi). Unbroken cells and cell debris were removed by low speed centrifugation, and the supernatant treated with 1% LDAO overnight at 4°C. After removal of insoluble material by ultracentrifugation, the supernatant was supplemented with 10 mM imidazole and the BRC purified by Ni2+-chelating affinity chromatography. BRC bound to affinity resin was washed overnight at 4°C in 20 column volumes of Buffer B + 0.03% LDAO, before elution in Buffer C + 0.03% LDAO. The complex was further purified on a Superdex 200 HR 10/300 GL in Buffer A + 0.03% LDAO, and stored in the dark at 4°C before use in thermostability assays.
‘On-column’ peptidisc reconstitution
MalFGK2 (300 µg) in Buffer E + 0.02% DDM was mixed with NSP (480 µg) in Assembly Buffer in a total volume of 100 µL. The mixture was immediately injected onto a 100 µL loop connected to a Superdex 200 HR 5/200 GL column running at 0.4 ml/min in Buffer AC (50 mM Tris-HCl, pH 8; 100 mM NaCl). Fractions were collected, pooled, concentrated using a 100 kDa polysulfone filter (Pall Corporation, USA), and stored at 4°C. For on-column reconstitution of FhuA, 500 µL of the protein (1 mg) was mixed with NSP (1.8 mg) in Buffer A + 0.05% LDAO, and injected onto a 500 µL loop connected to a Superdex 200 HR 10/300 GL column running at 0.5 mL/min in Buffer AC.
‘In-gel’ peptidisc reconstitution
The target membrane protein (~1.25 µg) was mixed with increasing concentrations of NSPr(0–2.5 µg) and allowed to incubate for 1–2 min at room temperature. The mixture was then supplemented with Buffer A to bring the final detergent concentration below its CMC (0.008% and 0.01% for DDM and LDAO, respectively) while keeping the final volume to 15 µL. A solution of glycerol was added to 10% final to facilitate loading on 4–12% CN-PAGE. The electrophoresis was set constant at 25mA for 1 hr at room temperature. Bands were visualized by Coomassie Blue G250 staining.
‘On-bead’ peptidisc reconstitution
Crude membranes (10 mL at 7.5 mg/ml total protein content) containing overexpressed MalFGK2 were solubilized in Buffer A + 1% DDM for 1 hr at 4°C before removal of insoluble aggregate by ultracentrifugation (100,000 x g, 1 hr, 4°C). The solubilized membrane proteins were incubated with 200 µl of Ni-NTA resin (Qiagen) pre-equilibriated in Buffer A + 0.02% DDM for 1 hr at 4°C. The Ni-NTA beads were collected by low-speed centrifugation (3000 x g, 3 min), washed twice with 10 CV of Buffer B supplemented with 0.02% DDM. Post-washing, 10 CV of Assembly Buffer (1 mg/mL NSPr in 20 mM Tris-HCl pH 8) was added to the beads and allowed to incubate for 5 min on ice. The Assembly Buffer was removed and the beads loaded into a gravity column with 10 CV of Buffer B (50 mM Tris-HCl, pH 8; 200 mM NaCl, 10% glycerol, 15 mM imidazole). The assembled peptidiscs were subsequently treated with 500 µL of Buffer C (50 mM Tris-HCl, pH 8; 100 mM NaCl;10% glycerol; 400 mM imidazole) to elute the peptidisc from the affinity resin. The same procedure was done in parallel, except the NSPr was omitted from the Assembly Buffer and 0.02% DDM was included in Buffer A, B, and C.
Reconstitution of the BRC in Peptidiscs, low lipid nanodiscs, and styrene maleic acid nanoparticles
The purified BRC complex (1 mg/mL) was mixed at a 1:1.8 (µg/µg) ratio with NSPr followed by 10-fold dilution in Buffer A to decrease the LDAO concentration to 0.003%. For formation of low-lipid nanodiscs, the purified BRC complex was instead mixed at a 1:2 (mol/mol) ratio with MSP1D1 before dilution. Alternatively, an equivalent amount of BRC was diluted in Buffer A supplemented with 0.03% LDAO, 0.02% DDM, 0.1% SMA or 0.1% SDS as described. After incubation for 10 min on ice, aggregated proteins were removed by centrifugation (13,000 x g, 10 min at 4°C). Peptidisc formation was confirmed by analysis on CN-PAGE.
Reconstitution of MalFGK<sub>2</sub> and BRC in proteoliposomes
Proteoliposomes were prepared at a molar protein:lipid ratio of 1:2000. Total E. coli lipids were dissolved in chloroform, dried under nitrogen and resuspended in Buffer A + 0.8% β-OG. Purified MalFGK2 was added to the solubilized lipids, and the detergent was removed by overnight incubation at 4°C with Amberlite XAD-2 adsorbent beads (Supelco). The proteoliposomes were isolated by ultracentrifugation (100,000 × g, 60 min at 4°C) and resuspended in 20 mM Tris–HCl, pH 8 before use in ATPase assays. The same procedure was employed for the BRC, but a lipid mixture of DOPC:DOPG (80:20 mol/mol) was utilized in place of total E.coli lipids.
Native gel electrophoresis
Equal volumes of 4% and 12% acrylamide solutions were prepared in advance (Supplementary file 2). Linear gradient gels were formed by gradual mixing of the two solutions (35 mL each) at a flow rate of 2 ml/min using a 100 mL gradient mixer (Sigma). The cross-linking agents, TEMED and ammonium persulfate, were added immediately before gradient mixing. Once poured, plastic wells (Biorad) were inserted and gels allowed to cure for 90 min before storage at 4°C. For clear-native PAGE, anode and cathode buffers consisted of Buffer N (37 mM Tris-HCl; 35 mM Glycine; pH 8.8). For blue-native PAGE, anode buffer consisted of Buffer N + 180 µM Coomassie Blue G-250, and cathode buffer contained Buffer N only.
Dynamic and static light scattering analysis
Aliquots of MalFGK2-NSPr were analyzed by static light scattering. Static light scattering analysis were performed using a WTC-050S5 column (Wyatt Technologies) connected to a miniDAWN light scattering detector and interferometry refractometer (Wyatt Technologies). Data were recorded in real time and the molecular masses were calculated using the Debye fit method using the ASTRA software (Wyatt Technology).
Sample preparation and EM image acquisition
MalFGK2-peptidisc sample (0.035 mg/mL) reconstituted by on-column method was applied onto negatively glow-discharged carbon-coated grids (400 mesh, copper grid) for 1 min, and excess liquid was removed by blotting with filter paper. Freshly prepared 1.5% uranyl formate (pH 5) was added (5 µl) for 1 min and then blotted. Around 200 digital micrographs were collected using a FEI Tecnai G2 F20 microscope operated at 200 kV and equipped with a Gatan Ultrascan 4k × 4 k Digital CCD Camera. The images were recorded at defocus between 0.7 and 1.4 µm at a magnification of 67,000X at the camera and a pixel size of 2.24 Å.
EM data processing and image analysis
Contrast transfer function parameters were determined using CTFFIND3 (Mindell and Grigorieff, 2003). We selected 31188 protein particles using e2boxer from the EMAN2 software suite (Tang et al., 2007) and extracted with a box size of 96 × 96 pixels. Particles were classified using a likelihood 2D classification with 16 seeds with the RELION-1.3 software suite (Scheres, 2012). A 2D variance of particles contained in side view (3175 particles) was computed with SPARX to estimate the peptidisc diameter variation (Hohn et al., 2007). The measurements were done using e2display from the EMAN2 software suite (Corin et al., 2011) on the side views shown in Figure 2C.
FhuA binding assay
FhuA-MSPL156 nanodiscs were prepared as previously described (Lee et al., 2016). FhuA-NSPr was prepared by on-column peptidisc reconstitution. About 2 µg of FhuA reconstituted into either MSPL156 or NSPr was incubated with TonB23-329 (2 µg) or ColM (5 µg) in the presence or absence of ferricrocin for 5 min at room temperature. The protein complexes were separated by CN-PAGE and visualized by Coomassie blue staining. Neither monomeric TonB nor ColM migrate on CN-PAGE due to their isoelectric points > pH 8.8.
Mass spectrometry
BRC peptidisc, MalFGK2 peptidisc, and FhuA peptidisc were prepared by ‘on-column’ reconstitution in 100 mM ammonium acetate, pH 7.0 at protein to NSPr (g/g) ratios of 1:1.8, 1:1.6, and 1:1.8, respectively. Mass spectrometry measurements were performed in positive ion mode on a Synapt G2S quadrupole-ion mobility separation-time-of-flight (Q-IMS-TOF) mass spectrometer (Waters, Manchester, UK) with a nanoflow electrospray ionization ESI (nanoESI) source. Borosilicate capillaries (1.0 mm o.d., 0.68 mm i.d.) were pulled in-house using a P-1000 micropipette puller (Sutter Instruments, Novato, CA). A voltage of ~1.0 kV was applied to a platinum wire was inserted into the nanoESI tip. A source temperature of 60°C and a Cone voltage of 30 V were used. Argon was used in the Trap and Transfer ion guides, at pressures of 2.77 × 10−2 mbar and 2.84 × 10−2 mbar, respectively, and the Trap and Transfer voltages were 5 V and 2 V, respectively. All data were processed using MassLynx software (v4.1). Spectral deconvolution was performed with the UniDec (Marty et al., 2015) deconvolution algorithm using the following parameters: m/z range – 7000 to 9500 (MalFGK2 peptidisc), 5500 to 9000 (BRC peptidisc), 5000 to 10000 (FhuA peptidisc); Subtract minimum - 50.0; Gaussian Smoothing - 10.0; Bin every 1.0; Linear m/z (constant delta m/z); Charge Range - 20 to 40 (MalFGK2 peptidisc), 10 to 30 (BRC peptidisc), 10 to 30 (FhuA peptidisc); Mass range - 200,000 to 300,000 (MalFGK2 peptidisc), 100,000 to 180,000 (BRC peptidisc), 100,000 to 170,000 (FhuA peptidisc); Sample Mass Every 1.0 Da; Peak FWHM (Th) 4.0; Peak Shape Function - Gaussian; Charge Smooth Window - 1.0; Mass Difference - 4474.0; Mass Smooth Window - 1.0; Maximum number of iterations - 1000. Spectral files were loaded as text files containing intensity and m/z values.
Absorbance spectroscopy
Absorption spectra were recorded using a Hitachi U-3010 spectrophotometer. A blank measurement was recorded in Buffer A (+0.03% LDAO for detergent purified BRC). Samples were incubated in a PCR thermocycler at the indicated temperature, and then measured at the desired time points in a quartz cuvette at room temperature. Spectra were collected between 600 nm and 1100 nm (scan time ∼20 s) at intervals of 1.5 min. For comparisons of spectra between conditions, spectra were normalized to a value of 1.0 at 804 nm.
Fluorescence measurements
The BRC complex into the indicated detergent or reconstituted into peptidiscs was incubated at varying temperatures in a PCR thermocycler for 5 min, then 3 µL of the mixture dotted onto nitrocellulose paper pre-wetted in Buffer A. The dot blot was imaged using a LICOR odyssey infrared fluorescence scanner (excitation 680 nm, emission 700 nm). Fluorescence intensity was quantified by Image J.
NSP<sub>r</sub> quantification
The MalFGK2 and BRC peptidiscs were prepared by on-column reconstitution on a Superdex 5/25 column equilibrated in Buffer A, followed by one additional gel filtration step to ensure full removal of free NSPr. MalFGK2 (1 µg), FhuA (2 µg), and BRC (2 µg) peptidiscs were analyzed by 15% SDS-PAGE. Gels were stained with Coomassie Blue G-250, and destained overnight before fluorescence measurement (excitation 680 nm, emission 700 nm) on a LICOR Odyssey scanner. The band corresponding to the NSPr peptide was quantified by densitometry using Image J and compared to a standard curve of NSPr(0–2 µg) loaded on the same gel. The determined NSPr amount was then subtracted from the total amount of protein loaded on the gel to determine the amount of reconstituted membrane protein in the peptidisc. Membrane protein content in peptidisc (g) = total protein in peptidisc (g) - measured NSPr content (g). We used these calculated mass measurements and the molecular weight (MW) for NSPr (4.5 kDa), MalFGK2 (173 kDa), FhuA (80 kDa) and BRC (94 kDa) to calculate NSPr stoichiometry as follows;NSPrStoichiometry=MWMembraneprotein(g/mol)MWNSPr(g/mol)×MeasuredNSPrcontent(g)Membraneproteincontentinpeptidisc(g)
Each experiment was repeated in triplicate on three different gels. We note that detergent-purified FhuA co-purified with a contaminant, thought to be short chain lipopolysaccharides, that migrated to the same position as NSPr, therefore FhuA was reconstituted using NSPr labelled with a biotin group (NSPrbio). To quantify NSPrbio, western blots were incubated with streptavidin conjugated to Alexafluor 680 in phosphate buffered saline (PBS), followed by several washes in PBS + 0.1% Tween. Western blots were imaged on a LICOR Odyssey scanner fluorescence (excitation 680 nm, emission 700 nm), and the bands corresponding to NSPrbio quantified in Image J.
Lipid extraction and quantification
The MaFGK2 and BRC peptidiscs were prepared on-bead, and the FhuA peptidisc was prepared on-column. MalFGK2 (40 µg), FhuA (40 µg), and BRC (80 µg) peptidiscs were diluted to a final volume of 200 µL of Buffer A, then mixed with 800 µL of a 2:1 solution of methanol:chloroform for 10 min at 25°C in glass screw cap vials. 200 µL of chloroform and 200 µL of distilled water were added sequentially, vortexed briefly, and the resulting two phase system separated by low-speed centrifugation (3000 r.p.m., 10 min). The organic phase was dried under nitrogen, and stored at −20°C. Total phosphate content was determined by a modified version of the malachite green assay (Lanzetta et al., 1979). Malachite green reagent was prepared as follows: ammonium molybdate (4.2 g) was dissolved in 100 mL of 4M HCl, then mixed with 300 mL malachite green (135 mg) dissolved in distilled water. The solution was mixed for 1 hr at 4°C, filtered, and stored at 4°C before use. Dried lipid extracts were subsequently incubated with 1 mL of 70% perchloric acid for 3 hr at 130°C, and then 20 µL of the resulting solution mixed with 500 µL of the malachite green reagent for 5 min at room temperature before absorbance measurement at 660 nm. Phosphate standards (KH2PO4) were diluted into perchloric acid and used to prepare a standard curve with phosphate concentrations ranging from 0.01 nmol to 1 nmol PO4. For thin layer chromatography (TLC) analysis, dried lipids were resuspended in 30 µL of chloroform, and 10 µL were dotted onto a TLC Silica gel 60 (Millipore). The TLC was developed in a solution of 35:25:3:28 chlorofrom:triethylamine:dH2O:ethanol. Plates were dried in an oven for 5 min at 150°C. Lipids were visualized by lightly wetting plates in a solution of 10% Cu2S04 in 8.5% phosphoric acid, followed by heating for 5 min at 150°C.
Other methods
The MalFGK2 ATPase activity was determined by monitoring the release of inorganic phosphate using the malachite green method (Lanzetta et al., 1979). Protein and peptide concentrations were determined by Bradford assay (Prehna et al., 2012). SMA polymer containing 2:1 styrene to maleic acid ratio was prepared following the procedure described by Dörr et al. (2014). In brief, 10% of SMA 2000 (Cray Valley), was refluxed for 3 hr at 80°C in 1M KOH, resulting in complete solubilization of the polymer. Polymer was then precipitated by dropwise addition of 6M HCl accompanied by stirring and pelleted by centrifugation (1500 x g for 5 min). The pellet was then washed 3 times with 50 mL of 25 mM HCl, followed by a third wash in ultrapure water and subsequent lyophilization. Lyophilized SMA was later re-suspended at 10% wt/vol in 25 mM Tris-HCl, and the pH of the solution adjusted to 8 with 1M NaOH. Peptide hydrophobic moment and electropotential was calculated using the 3D-HM calculator (Reißer et al., 2014). Sequences corresponding to NSP or NSPr were calculated with the C-terminus specified as (COO-) and N-terminus specified as (NH3+). UV absorbance of solubilized peptides was measured by Nanodrop.
Acknowledgements
This work was supported by operating grants from the Canadian Institutes of Health Research (74525MOP to FD), the Natural Sciences and Engineering Research Council of Canada (Discovery Grant 2796 to JTB), and Genome British Columbia (SOF153 to JTB).
Additional informationCompeting interests
No competing interests declared
Has opened a website to distribute peptides to the academic community, registered the Peptidisc term and filed a provisional patent via the University of British Columbia.
Author contributions
Conceptualization, Data curation, Formal analysis, Validation, Investigation, Visualization, Methodology, Writing—original draft, Writing—review and editing, Conceived the study, Contributed to discussion and writing, Performed light scattering experiments, Did all sample preparation and data analysis unless otherwise stated
Conceptualization, Methodology, Writing—review and editing, Prepared detergent purified SecEYG, Conceived the study, Contributed to discussion and writing
Investigation
Data curation, Formal analysis, Investigation, Writing—original draft, Prepared and analyzed electron microscopy data
Resources, Data curation, Investigation, Writing—review and editing, Performed BRC stability experiments
Formal analysis, Investigation, Visualization
Data curation, Formal analysis, Investigation, Visualization, Writing—original draft, Performed and analyzed intact native MS experiments
Investigation, Prepared samples for analysis by mass spectrometry
Investigation, Performed peptide titrations, Analyzed and presented the data
The length of the scaffold proteins was calculated by multiplying the number of amino acids (aa) by 1.5 Å, which is the rise given by an amino acid structured in an alpha-helix. For the MSPs scaffolds, the number of amino acids was from the TEV cleavage site (ENYLFQ//GXXX) to the C-terminus of the proteins.
10.7554/eLife.34085.018
Native gel buffer recipes.
10.7554/eLife.34085.019
Protein molecular weights.
10.7554/eLife.34085.020Data availability
All data generated or analysed during this study are included in the manuscript and supporting files.
ReferencesBayburtTHGrinkovaYVSligarSG2006Assembly of single bacteriorhodopsin trimers in bilayer nanodiscsArchives of Biochemistry and Biophysics45021522210.1016/j.abb.2006.03.01316620766MindellJAGrigorieffN2003Accurate determination of local defocus and specimen tilt in electron microscopyJournal of Structural Biology14233434710.1016/S1047-8477(03)00069-8BaoHDuongF2012Discovery of an auto-regulation mechanism for the maltose ABC transporter MalFGK2PLoS ONE7e3483610.1371/journal.pone.003483622529943BaoHDuongF2013Phosphatidylglycerol directs binding and inhibitory action of EIIAGlc protein on the maltose transporterJournal of Biological Chemistry288236662367410.1074/jbc.M113.48956723821551BecharaCNöllAMorgnerNDegiacomiMTTampéRRobinsonCV2015A subset of annular lipids is linked to the flippase activity of an ABC transporterNature Chemistry725526210.1038/nchem.2172BessonneauPBessonVCollinsonIDuongF2002The SecYEG preprotein translocation channel is a conformationally dynamic and dimeric structureThe EMBO Journal21995100310.1093/emboj/21.5.99511867527BibowSPolyhachYEichmannCChiCNKowalJAlbiezSMcLeodRAStahlbergHJeschkeGGüntertPRiekR2017Solution structure of discoidal high-density lipoprotein particles with a shortened apolipoprotein A-INature Structural & Molecular Biology2418719310.1038/nsmb.334528024148BlankenshipRETiedeDMBarberJBrudvigGWFlemingGGhirardiMGunnerMRJungeWKramerDMMelisAMooreTAMoserCCNoceraDGNozikAJOrtDRParsonWWPrinceRCSayreRT2011Comparing photosynthetic and photovoltaic efficiencies and recognizing the potential for improvementScience33280580910.1126/science.120016521566184BradfordMM1976A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye bindingAnalytical Biochemistry7224825410.1016/0003-2697(76)90527-3942051ChungBHAnatharamaiahGMBrouilletteCGNishidaTSegrestJP1985Studies of synthetic peptide analogs of the amphipathic Helix. correlation of structure with functionThe Journal of Biological Chemistry26010256102624019511CorinKBaaskePRavelDBSongJBrownEWangXWienkenCJJerabek-WillemsenMDuhrSLuoYBraunDZhangS2011Designer lipid-like peptides: a class of detergents for studying functional olfactory receptors using commercial cell-free systemsPLoS ONE6e2506710.1371/journal.pone.002506722132066DalalKChanCSSligarSGDuongF2012Two copies of the SecY channel and acidic lipids are necessary to activate the SecA translocation ATPasePNAS1094104410910.1073/pnas.111778310922378651DellerMCKongLRuppB2016Protein stability: a crystallographer's perspectiveActa Crystallographica Section F Structural Biology Communications72729510.1107/S2053230X15024619DenisovIGGrinkovaYVLazaridesAASligarSG2004Directed self-assembly of monodisperse phospholipid bilayer nanodiscs with controlled sizeJournal of the American Chemical Society1263477348710.1021/ja039357415025475DenisovIGSligarSG2016Nanodiscs for structural and functional studies of membrane proteinsNature Structural & Molecular Biology2348148610.1038/nsmb.319527273631DörrJMKoorengevelMCSchäferMProkofyevAVScheidelaarSvan der CruijsenEADaffornTRBaldusMKillianJA2014Detergent-free isolation, characterization, and functional reconstitution of a tetrameric K+ channel: the power of native nanodiscsPNAS111186071861210.1073/pnas.141620511225512535FabreLBaoHInnesJDuongFRouillerI2017Negative stain Single-particle EM of the maltose transporter in nanodiscs reveals asymmetric closure of MalK2 and Catalytic Roles of ATP, MalE, and MaltoseJournal of Biological Chemistry2925457546410.1074/jbc.M116.75789828188291FrauenfeldJLövingRArmacheJPSonnenAFGuettouFMobergPZhuLJegerschöldCFlayhanABriggsJAGaroffHLöwCChengYNordlundP2016A saposin-lipoprotein nanoparticle system for membrane proteinsNature Methods1334535110.1038/nmeth.380126950744GulatiSJamshadMKnowlesTJMorrisonKADowningRCantNCollinsRKoenderinkJBFordRCOverduinMKerrIDDaffornTRRothnieAJ2014Detergent-free purification of ABC (ATP-binding-cassette) transportersBiochemical Journal46126927810.1042/BJ2013147724758594HagnFEtzkornMRaschleTWagnerG2013Optimized phospholipid bilayer nanodiscs facilitate high-resolution structure determination of membrane proteinsJournal of the American Chemical Society1351919192510.1021/ja310901f23294159HohnMTangGGoodyearGBaldwinPRHuangZPenczekPAYangCGlaeserRMAdamsPDLudtkeSJ2007SPARX, a new environment for Cryo-EM image processingJournal of Structural Biology157475510.1016/j.jsb.2006.07.00316931051HughesAVReesPHeathcotePJonesMR2006Kinetic analysis of the thermal stability of the photosynthetic reaction center from Rhodobacter sphaeroidesBiophysical Journal904155416610.1529/biophysj.105.07002916533858IslamRMPourmousaMSviridovDGordonSMNeufeldEBFreemanLAPerrinBSPastorRWRemaleyAT2018Structural properties of apolipoprotein A-I mimetic peptides that promote ABCA1-dependent cholesterol effluxScientific Reports8295610.1038/s41598-018-20965-229440748JunDSaerRGMaddenJDBeattyJT2014Use of new strains of Rhodobacter sphaeroides and a modified simple culture medium to increase yield and facilitate purification of the reaction centrePhotosynthesis Research12019720510.1007/s11120-013-9866-623765434KariyazonoHNadaiRMiyajimaRTakechi-HarayaYBabaTShigenagaAOkuhiraKOtakaASaitoH2016Formation of stable nanodiscs by bihelical apolipoprotein A-I mimetic peptideJournal of Peptide Science2211612210.1002/psc.284726780967LanzettaPAAlvarezLJReinachPSCandiaOA1979An improved assay for nanomole amounts of inorganic phosphateAnalytical Biochemistry100959710.1016/0003-2697(79)90115-5161695LarsenANSørensenKKJohansenNTMartelAKirkensgaardJJJensenKJArlethLMidtgaardSR2016Dimeric peptides with three different linkers self-assemble with phospholipids to form peptide nanodiscs that stabilize membrane proteinsSoft Matter125937594910.1039/C6SM00495D27306692LeeSCKnowlesTJPostisVLJamshadMParslowRALinYPGoldmanASridharPOverduinMMuenchSPDaffornTR2016A method for detergent-free isolation of membrane proteins in their local lipid environmentNature Protocols111149116210.1038/nprot.2016.07027254461MartyMTBaldwinAJMarklundEGHochbergGKBeneschJLRobinsonCV2015Bayesian deconvolution of mass and ion mobility spectra: from binary interactions to polydisperse ensemblesAnalytical Chemistry874370437610.1021/acs.analchem.5b0014025799115McGregorCLChenLPomroyNCHwangPGoSChakrabarttyAPrivéGG2003Lipopeptide detergents designed for the structural study of membrane proteinsNature Biotechnology2117117610.1038/nbt77612524549MillsALeH-TCoultonJWDuongF2014FhuA interactions in a detergent-free nanodisc environmentBiochimica Et Biophysica Acta (BBA) - Biomembranes183836437110.1016/j.bbamem.2013.09.022MishraVKPalgunachariMNKrishnaRGlushkaJSegrestJPAnantharamaiahGM2008Effect of leucine to phenylalanine substitution on the nonpolar face of a class A amphipathic helical peptide on its interaction with lipid: high resolution solution NMR studies of 4F-dimyristoylphosphatidylcholine discoidal complexThe Journal of Biological Chemistry283343933440210.1074/jbc.M80638420018845546PopotJL2010Amphipols, Nanodiscs, and fluorinated surfactants: three nonconventional approaches to studying membrane proteins in aqueous solutionsAnnual Review of Biochemistry7973777510.1146/annurev.biochem.052208.11405720307193PrehnaGZhangGGongXDuszykMOkonMMcIntoshLPWeinerJHStrynadkaNC2012A protein export pathway involving Escherichia coli porinsStructure201154116610.1016/j.str.2012.04.01422658749PrivéGG2009Lipopeptide detergents for membrane protein studiesCurrent Opinion in Structural Biology1937938510.1016/j.sbi.2009.07.00819682888RaviSKTanSC2015Progress and perspectives in exploiting photosynthetic biomolecules for solar energy harnessingEnergy & Environmental Science82551257310.1039/C5EE01361EReißerSStrandbergESteinbrecherTUlrichAS20143D hydrophobic moment vectors as a tool to characterize the surface polarity of amphiphilic peptidesBiophysical Journal1062385239410.1016/j.bpj.2014.04.02024896117ScheidelaarSvan GrondelleRKillianJAJonesMRJonesMR2014Bacterial reaction centers purified with styrene maleic acid copolymer retain native membrane functional properties and display enhanced stabilityAngewandte Chemie International Edition53118031180710.1002/anie.20140641225212490ScheresSH2012RELION: implementation of a bayesian approach to cryo-EM structure determinationJournal of Structural Biology18051953010.1016/j.jsb.2012.09.00623000701SchwarzHSchmittnerMDuschlAHorejs-HoeckJ2014Residual endotoxin contaminations in recombinant proteins are sufficient to activate human CD1c+ dendritic cellsPLoS ONE9e11384010.1371/journal.pone.011384025478795SmithAAAAutzenHELaursenTWuVYenMHallAHansenSDChengYXuT2017Controlling styrene maleic acid lipid particles through RAFTBiomacromolecules183706371310.1021/acs.biomac.7b0113628934548TangGPengLBaldwinPRMannDSJiangWReesILudtkeSJ2007EMAN2: an extensible image processing suite for electron microscopyJournal of Structural Biology157384610.1016/j.jsb.2006.05.00916859925TaoHLeeSCMoellerARoyRSSiuFYZimmermannJStevensRCPotterCSCarragherBZhangQ2013Engineered nanostructured β-sheet peptides protect membrane proteinsNature Methods1075976110.1038/nmeth.253323817067WayneRFrickKNeilandsJB1976Siderophore protection against colicins M, B, V, and Ia in Escherichia coliJournal of Bacteriology126712131121YaghoubiHSchaeferMYaghoubiSJunDSchlafRBeattyJTTakshiA2017A ZnO nanowire bio-hybrid solar cellNanotechnology2805400605400810.1088/1361-6528/28/5/05400610.7554/eLife.34085.022Decision letterDötschVolkerReviewing EditorJ.W. Goethe-UniversityGermany
In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.
[Editors’ note: this article was originally rejected after discussions between the reviewers, but the authors were invited to resubmit after an appeal against the decision.]
Thank you for submitting your work entitled "The Peptidisc, a Simple Method for Stabilizing Membrane Proteins in Detergent-free Solution" for consideration by eLife. Your article has been evaluated by a Senior Editor and three reviewers, one of whom, Volker Dötsch (Reviewer #1), is a member of our Board of Reviewing Editors. The following individual involved in review of your submission has agreed to reveal their identity: Markus A Seeger (Reviewer #3).
Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife.
While all reviewers see the potential usefulness and impact of the peptidiscs it seems that the use of peptides to solubilize membrane proteins has been proposed and implemented before. As you will see in the individual reviews the lack of comparison to these existing methods has been criticized. In addition, a more detailed analysis of how the peptides achieve solubilization is lacking in comparison to published peptide based methods. And finally, a real proof that peptidiscs are superior to existing methods for structure determination as claimed would have been beneficial. Overall, the manuscript describes a very interesting alternative to existing methods to solubilize membrane proteins, it seems however less evident that the method will provide a major breakthrough.
Reviewer #1:
Carlson et al. describe a peptide based system that keeps membrane proteins soluble with tightly attached lipid molecules. In contrast to the well-established nanodisc system this system is based on a variable number of peptides that surround the membrane protein and stabilize it. No addition of lipids is necessary and the high flexibility of the system does not make the engineering of differently long peptides as scaffolds necessary. They show with five different membrane proteins that they can be stably solubilized in these discs and provide a very careful assessment of the peptide/protein/lipid composition of these discs. They also show that these discs can be formed in several different ways starting with detergent solubilized membrane proteins.
This is an interesting system that can provide an attractive alternative to the existing hydrophobic systems to solubilize membrane proteins. The study is well documented and carefully performed.
1) The proteins used are relatively large membrane protein complexes or b-barrel proteins that are stable even in detergent micelles. For those membrane proteins the peptidisc will be a good alternative. However, for membrane proteins that are not stable in detergent micelles, are expressed cell-free or resolubilized from inclusion bodies it is not clear if they can be stabilized with these discs. A GPCR or a transporter would have been good alternatives.
2) One of the advantages over nanodiscs that the authors advertise is that the peptidiscs can be used for structural investigations. It would have been good to show for example an NMR spectrum of a membrane protein in such a peptidisc and to show that this is advantages relative to larger nanodiscs.
Reviewer #2:
The manuscript by Carlson et al. includes a considerable amount of data on five different membrane proteins solubilized into bi-helical peptides. This is combined with three reconstitution methods to transfer proteins into peptidiscs on columns, beads or gels. As such, it potentially provides a useful methodological advance and benchmark within the rapidly developing field of preparation and analysis of membrane proteins in nanometer sized discs.
However, there are overstatements and lack of clarity of key points that need to be addressed.
The Abstract states that "the peptidisc just requires a short amphipathic bi-helical peptide (NSP) and no extra lipids". However, the proteins are initially prepared using detergents and are expressed in E. coli. Hence recombinant expression in for e.g. E. coli and detergent appear to be also required.
The authors go on to say that "This drawback has led researchers to develop detergent-free alternatives such as amphipols,[Popot, 2010] SMALPs,[Lee et al., 2016] saposin-lipoparticles[Frauenfeld et al., 2016] and the popular nanodisc system.[Bayburt, Grinkova and Sligar, 2004; Denisov et al., 2004]" Again, some of these methods also require detergent to be added. This is a distinction that needs to be clarified. The statement "We present here the peptidisc as a simple assembly method to support membrane protein in detergent-free solution" similarly needs to be corrected as it is not clear that the methods are simple or detergent-free. The concluding statement in the Introduction that "we show that the NSP peptide may well be the universal scaffold for stabilizing both α-helical and β-barrel membrane proteins of different size, topology, and complexity" and following the discussion that "These advantages combined suggest that the peptidisc should diminish the challenges associated with biochemical, structural and pharmacological characterization of membrane proteins, making the peptidisc an efficient and perhaps universal tool for stabilizing these proteins in membrane- and detergent-free solution" are overstated. For this to be a useful paper, the authors need to indicate the limitations of the method. In the Materials and methods used for on column peptidisc reconstitution detergents such as LDAO are used. Doesn't the peptide itself denature in such detergents, and would this not limit the effectiveness of this reagent?
The yield, purity and activity should be given quantitatively for each protein in a peptidisc vs. in detergent alone, and ideally also vs. in a liposome. This would allow the method to be objectively compared. Do the proposed methods not require detergent that could strip away natively bound lipids and destabilize membrane proteins? Is this not a limitation? Ideally assays should be given across a temperature range to ensure that folded protein is being measured for activity.
The identities of bound lipids should also be stated, rather than non-specific statements about lipid content like "Also, because lipids (i.e. annular lipids) can remain tightly bound to membrane proteins during purification [Bechara et al., 2015], we also determined the lipid content by thin layer chromatography and photocolorimetric methods" and "Following the same approach applied to MalFGK2 above, we quantified the individual peptide and lipid components of the FhuA peptidisc (Figure 3B and A), resulting in an average of 8 ± 3 phospholipids and 10 ± 2 NSP per FhuA peptidisc". The low number of lipid molecules present (4 and 8 in the cases of BRC and FhuA) indicates that only the most tightly bound lipids remain in the peptidisc, and that a disc shape cannot be assumed.
The presence of an apparently non-physiological multimer (Sec(EYG)n) in Figure 2D is glossed over. This needs to be explained as it indicates that use of NSP is leading to potentially artefactual multimeric states.
In Figure 2F, why do proteins with different molecular weight share the same RR50? Would one not expect to see a higher number of peptides interacting with larger assembly or with different membranes or cell types? The authors didn't explain their perspective on this, nor is it clear how this was optimized, despite being a significant cost and determinant of success of the method. A recommended molar concentration of peptide for reconstitution should be indicated and justified.
The description of the peptide being like a belt around the protein and of the disc shape of peptidiscs needs to be justified with experimental data and/or references. There is one set of negative stain EM data of MalFGK2 in peptidiscs showing the presence of a number of pairs and triplets of discs. Is this not significant and indicative of disc-disc interactions, perhaps mediated by NSP peptides? If so shouldn't lower peptide concentrations be used to minimize such stacks in biophysical assays? Also, the scale bar in the bottom panel of Figure 1C appears to indicate 50 (not 5) Å. If so this is inconsistent.
Reviewer #3:
The manuscript by Carlson et al. describes a protocol for membrane protein stabilization using an amphiphilic 37 amino acid peptide called nanodisc scaffold protein (NSP) to result in a peptidisc, which corresponds to the membrane protein surrounded by NSP and annular lipids. The method is applied and validated by reconstituting an ABC importer (maltose transporter), two outer membrane proteins, SecEYG and a bacterial reaction center (BRC). The ratio between protein and NSP peptide as well as the number of bound phospholipids was determined using a set of different biophysical methods. The ATPase activity of the maltose transporter embedded in a peptidisc was found to be strongly coupled to the presence of the maltose binding protein MalE, which is a hallmark of samples reconstituted in proteoliposomes or nanodiscs, whereas very poor coupling is observed in detergent solution. For BRC, the peptidisc was shown to stabilize the membrane protein as compared to the (rather harsh) detergent LDAO.
The peptidisc is proposed in the manuscript to be equivalent to other membrane reconstitution methods such as the widely used nanodisc system and the classical proteoliposomes with regard to conformational coupling (shown for the maltose transporter) as well as membrane protein stabilization. On the other hand, the NSP peptide is used as a surrogate for a detergent, i.e. it is very simple to use and just needs to be added to a solubilized membrane protein to replace the initial detergent as is often done with short chain detergents replacing for example DDM, which is widely used to solubilize membrane proteins.
The manuscript appears solid with regard to the biophysical analyses of the peptidisc complexes and their content of NSP peptides and annular lipids. Nevertheless, I am a bit skeptical whether a peptidisc represents a reconstitution in the classical sense, because the NSP peptides directly interact with the transmembrane helices (driven by hydrophobic protein-protein interactions mediated by amino acid side chains) instead of lipids via their aliphatic fatty acid chains. Nevertheless, the method (although not being a novel concept at as outlined below) seems to be rather easy and versatile in its application.
1) As a potential future user (and a current user of the nanodisc method), it is critical to know how expensive the synthesis (or any alternative preparation) of the NSP peptide is. I assume it is more expensive than detergents, but in contrast to the MSP protein used for nanodiscs, it might be less expensive. Along the same lines, the authors stated that the purity of their chemically synthetized NSP peptide was "more than 80%". Is this sufficient? What are the impurities. Is this rather low purity linked to the price of the synthesis?
2) How aggregation-prone are peptidiscs? The authors hypothesize that the NSP peptides build a regular belt around the membrane protein, akin to the MSP belt. But is this really feasible? It is clear that the peptides arrange such that the hydrophobic parts of the membrane protein are covered by its hydrophobic side, while the hydrophilic face of the peptides remains exposed to the solvent. But this process is likely to be stochastic in the sense that there remain hydrophobic gaps on the membrane protein surface, as well as on some of the not perfectly placed NSP peptides. These remaining hydrophobic surfaces would then serve as nucleation points for protein aggregation. Remaining detergent molecules originating from the solubilization and purification of the membrane protein may shield these remaining hydrophobic surfaces.
3) Related to the above comment: Do peptidiscs still contain detergents? It is well known that quantitative detergent exchange is not a trivial endeavor. Often the initial detergent used to solubilize the membrane protein remains present to some degree. The authors need to show, whether in the peptidiscs there are still some detergents remaining (next to annular lipids and the NSP peptide).
4) How big is the NSP belt compared to detergent belts of different sizes? This question is highly relevant for membrane protein crystallization. The peptidisc appears more compact than a nanodisc and membrane proteins in peptidiscs might be crystallized. Did the authors try to crystallize their five membrane proteins used in this study, for which high resolution crystal structures (determined in detergent) exist?
5) Does the peptidisc really mimic the natural environment of a membrane protein? The authors carefully worked out that the peptidisc contains annular lipids. However, annular lipids are also contained in detergent-purified samples. The authors should address this question by comparing the content of annular lipids side-by-side and over time using the same transporter purified in detergent as well as in peptidiscs and measuring by mass spectrometry the mass (or loss of mass over time) of the entire complex.
6) The manuscript lacks a proper Discussion. Most problematically, the authors do not discuss at all that the concept of peptides used as detergent surrogates is not novel at all. Rather, it was first described in the 80s and 90s as peptitergent (you can find this word even on Wikipedia). See also in Schafmeister CE, Miercke LJ, Stroud RM. Structure at 2.5 A of a designed peptide that maintains solubility of membrane proteins. Science. 1993 Oct 29;262(5134):734-8. In addition, there are lipopeptide detergents (LPDs), nano-structured β-sheets (see Tao H, Lee SC, Moeller A, Roy RS, Siu FY, Zimmermann J, Stevens RC, Potter CS, Carragher B, Zhang Q. Engineered nanostructured β-sheet peptides protect membrane proteins. Nat Methods. 2013 Aug;10(8):759-61). Further, there was the development of short peptides for membrane stabilization, which are much cheaper to produce than the 37 aa NSP peptide (see Kiley P, Zhao X, Vaughn M, Baldo MA, Bruce BD, Zhang S. Self-assembling peptide detergents stabilize isolated photosystem I on a dry surface for an extended time. PLoS Biol. 2005 Jul;3(7):e230. Epub 2005 Jun 21.)
Important is the question whether at all or to what extent the NSP peptide is superior to these previously described methods using amphiphilic peptides, in particular the short designer peptides as described in Kiley et al., 2005). Critically, this point needs to be addressed experimentally by a functional assay including the maltose transporter and/or the BRC.
The Discussion should also include the concept of membrane protein stabilization via peptide-protein interactions (peptidisc) versus lipid protein interactions (nanodisc/saposins/proteoliposomes). And finally, it needs to address the amphipols (which are conceptually most closely linked to peptitergents and peptidiscs) in more detail.
[Editors’ note: what now follows is the decision letter after the authors submitted for further consideration.]
Thank you for resubmitting your work entitled "The Peptidisc, a Simple Method for Stabilizing Membrane Proteins in Detergent-free Solution" for further consideration at eLife. Your revised article has been favorably evaluated by Richard Aldrich (Senior Editor), a Reviewing Editor, and two reviewers.
The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:
1) What is interesting in the revised version is the fact that the sequence of the original NSP peptide as it is present in ApoA1 was in fact reversed and that this resulted in an increase of solubility of the peptide. Why was this information completely lacking in the first version?
2) The exact sequence of the NSPr peptide should be provided directly in the Materials and methods section under section "Biological reagents and peptides" (I had difficulties finding it in Supplementary file 1).
3) The authors are now stating in the manuscript that the method is inexpensive. However, I would still be interested to know how much a mg or a gram costs and how much you will need for an experiment (by providing a reasonable range).
4) Concerning the impurities of the peptide, the authors gave an interesting answer in the response letter (i.e. that the great majority of the impurity is in fact the same peptide missing its final amino acid). This should be mentioned in the manuscript as well.
5) The authors refer to a homepage to order/obtain the peptide: www.peptidisc.com. However, the homepage does work yet. It seems that the authors will or have already founded a start-up company to commercialize the peptidisc. Potential financial competing interests have to be declared.
10.7554/eLife.34085.023Author response
[Editors’ note: the author responses to the first round of peer review follow.]
Reviewer #1:
[…] 1) The proteins used are relatively large membrane protein complexes or b-barrel proteins that are stable even in detergent micelles. For those membrane proteins the peptidisc will be a good alternative. However, for membrane proteins that are not stable in detergent micelles, are expressed cell-free or resolubilized from inclusion bodies it is not clear if they can be stabilized with these discs. A GPCR or a transporter would have been good alternatives.
We agree with the reviewer, the peptidisc like other mimitics will not work with proteins that are hard to purify or too unstable in detergent. However, in the case of the BRC complex, we observe 100 fold stability increase in peptidic versus LDAO detergent at elevated temperature. The immediate aim of our work is to present the peptidisc as a new tool, so that it can be tested by other researchers on their favorite target. For example, current biochemical work on GPCRs involves amphipols, and not so much nanodiscs perhaps due to the required addition of exogenous lipids which complicate the reconstitution. The peptidisc therefore has also potential to replace amphipols which are inherently polydisperse. Further studies characterizing the peptidisc’s usefulness for cell-free expression systems is certainly of interest, however beyond the scope of this introductory article. We feel that a cell-free expression should be investigated as a separate study with all the different possible membrane mimetics (nanodiscs/SMA/peptergents etc.) to be truly useful.
2) One of the advantages over nanodiscs that the authors advertise is that the peptidiscs can be used for structural investigations. It would have been good to show for example an NMR spectrum of a membrane protein in such a peptidisc and to show that this is advantages relative to larger nanodiscs.
We only mentioned that peptidisc, given their compositional homogeneity, should be advantageous for structural studies. Our laboratory is however not equipped for such structural analysis but we are actively seeking collaborators.
Reviewer #2:
The manuscript by Carlson et al. includes a considerable amount of data on five different membrane proteins solubilized into bi-helical peptides. This is combined with three reconstitution methods to transfer proteins into peptidiscs on columns, beads or gels. As such, it potentially provides a useful methodological advance and benchmark within the rapidly developing field of preparation and analysis of membrane proteins in nanometer sized discs. However, there are overstatements and lack of clarity of key points that need to be addressed.
The Abstract states that "the peptidisc just requires a short amphipathic bi-helical peptide (NSP) and no extra lipids". However, the proteins are initially prepared using detergents and are expressed in E. coli. Hence recombinant expression in for e.g. E. coli and detergent appear to be also required.
To prevent further confusion, we have modified this statement to “Reconstitution of a detergent solubilized membrane protein into a peptidisc only requires a short,”
The authors go on to say that "This drawback has led researchers to develop detergent-free alternatives such as amphipols,[Popot, 2010] SMALPs,[Lee et al., 2016] saposin-lipoparticles[Frauenfeld et al., 2016] and the popular nanodisc system.[Bayburt, Grinkova and Sligar, 2004; Denisov et al., 2004]" Again, some of these methods also require detergent to be added. This is a distinction that needs to be clarified.
We have introduced a new section in the Introduction to present and to explain the main differences between synthetic scaffolds and the other detergent free alternatives.
The statement "We present here the peptidisc as a simple assembly method to support membrane protein in detergent-free solution" similarly needs to be corrected as it is not clear that the methods are simple or detergent-free.
It is both. We have rephrased this statement.
The concluding statement in the Introduction that "we show that the NSP peptide may well be the universal scaffold for stabilizing both α-helical and β-barrel membrane proteins of different size, topology, and complexity" and following the discussion that "These advantages combined suggest that the peptidisc should diminish the challenges associated with biochemical, structural and pharmacological characterization of membrane proteins, making the peptidisc an efficient and perhaps universal tool for stabilizing these proteins in membrane- and detergent-free solution" are overstated. For this to be a useful paper, the authors need to indicate the limitations of the method.
The initial Discussion was very short because we initially submitted this paper as a method. We now have completely revisited the Discussion to highlight the pro and con of the various scaffolds and where other methods may be more appropriate. We have also increased the Introduction to present those other methods and why the peptidisc is evidently superior on certain specific aspects.
In the Materials and methods used for on column peptidisc reconstitution detergents such as LDAO are used. Doesn't the peptide itself denature in such detergents, and would this not limit the effectiveness of this reagent?
Reconstitution in peptidisc, like in nanodisc, occurs upon dilution of the detergent. This collapses the micelle, allowing the peptide to fold around the target proteins. Unfolding of the peptide is not an issue. Our data show that LDAO is not limiting the effectiveness of the method.
The yield, purity and activity should be given quantitatively for each protein in a peptidisc vs. in detergent alone, and ideally also vs. in a liposome. This would allow the method to be objectively compared. Do the proposed methods not require detergent that could strip away natively bound lipids and destabilize membrane proteins? Is this not a limitation?
We have performed a side-by-side comparison of the BRC complex reconstituted in liposomes, nanodiscs, peptidiscs, SMA polymer, and LDAO detergent and presented the results in the revised manuscript. All membrane mimetic systems show a similar increase in thermostability. Delipidation is a general problem with detergent solubilization. However, we show that the peptide increases the thermostability of the delipidated BRC complex. This is likely due to i) removal of denaturing detergents, and ii) a more stable hydrophobic environment as compared to a detergent micelle. Thus, even without extra lipids, the peptidisc is able to stabilize the BRC as much as in a proteoliposome (Figure 10—figure supplement 1). Detergent delipidation can also be a useful for increasing purity of membrane proteins. For example, during purification of FhuA, the inner membrane is first removed by solubilization in Triton X-100 before addition of LDAO to release FhuA from the outer membrane. This significantly increases the final purity.
Ideally assays should be given across a temperature range to ensure that folded protein is being measured for activity.
We have shown that MalFGK2 in peptidisc remains folded throughout the reconstitution using BN-PAGE and ATPase activity assays.
The identities of bound lipids should also be stated, rather than non-specific statements about lipid content like "Also, because lipids (i.e. annular lipids) can remain tightly bound to membrane proteins during purification [Bechara et al., 2015], we also determined the lipid content by thin layer chromatography and photocolorimetric methods" and "Following the same approach applied to MalFGK2 above, we quantified the individual peptide and lipid components of the FhuA peptidisc (Figure 3B and A), resulting in an average of 8 ± 3 phospholipids and 10 ± 2 NSP per FhuA peptidisc". The low number of lipid molecules present (4 and 8 in the cases of BRC and FhuA) indicates that only the most tightly bound lipids remain in the peptidisc, and that a disc shape cannot be assumed.
We have included data on the identities of the bound lipids to MalFGK2. We also show that FhuA and BRC have very low lipid content. However, we do not agree that this precludes a disc shape, as both the periplasmic and cytoplasmic faces of FhuA are still accessible to its soluble binding partners. Therefore, the bulk of the peptide must be located around the transmembrane domains of the transporter. Electron microscopy data included in the manuscript support further this. We have extended the Discussion to describe the possible mode of peptide association to target protein.
The presence of an apparently non-physiological multimer (Sec(EYG)n) in Figure 2D is glossed over. This needs to be explained as it indicates that use of NSP is leading to potentially artefactual multimeric states.
The oligomeric propensity of SecEYG have been reported by many researchers in the past. This is not an artifact and we have referenced this in the manuscript. Our data show that the peptidisc can capture these oligomers.
In Figure 2F, why do proteins with different molecular weight share the same RR50? Would one not expect to see a higher number of peptides interacting with larger assembly or with different membranes or cell types? The authors didn't explain their perspective on this, nor is it clear how this was optimized, despite being a significant cost and determinant of success of the method. A recommended molar concentration of peptide for reconstitution should be indicated and justified.
We have discussed this point in more detail in the Discussion, and propose a tilted conformation of the peptide to account for this interesting observation. We have also included a recommended molar concentration of peptide that is justified by the in-gel reconstitutions experiments.
The description of the peptide being like a belt around the protein and of the disc shape of peptidiscs needs to be justified with experimental data and/or references.
We now have an extended discussion on possible orientations of the peptide in the Discussion. The disc shape is clearly visible in our negative stain EM experiments.
There is one set of negative stain EM data of MalFGK2 in peptidiscs showing the presence of a number of pairs and triplets of discs. Is this not significant and indicative of disc-disc interactions, perhaps mediated by NSP peptides? If so shouldn't lower peptide concentrations be used to minimize such stacks in biophysical assays?
The MalFGK2 peptidisc preparation are monodisperse, please refer to the negative stain analysis Figure 1C. We are unclear where the reviewer is seeing disc stacking?
Also, the scale bar in the bottom panel of Figure 1C appears to indicate 50 (not 5) Å. If so this is inconsistent.
Figure 2C shows class average of a single nanodisc. 10 Å = 1 nanometre. The disc containing MalFGK2 is approximately 11-12nm across. The scale bar is therefore correct and consistent with stated measurements in the text.
Reviewer #3:
[…] The manuscript appears solid with regard to the biophysical analyses of the peptidisc complexes and their content of NSP peptides and annular lipids. Nevertheless, I am a bit skeptical whether a peptidisc represents a reconstitution in the classical sense, because the NSP peptides directly interact with the transmembrane helices (driven by hydrophobic protein-protein interactions mediated by amino acid side chains) instead of lipids via their aliphatic fatty acid chains. Nevertheless, the method (although not being a novel concept at as outlined below) seems to be rather easy and versatile in its application.
1) Interaction – we have included a full discussion on the possible mode of interaction of the peptide with the target membrane protein. Importantly, we show that the peptidisc represents a “true” reconstitution because it is entirely stable upon removal from excess peptide, in much the same manner as a nanodisc or amphipols. This is in contrast to other peptide scaffolds (lipopeptides and peptergents), which must be maintained in buffer above their CMC to maintain protein solubility.
2) Novelty – we have carefully addressed the issue of novelty in the section Introduction and Discussion by including comparisons (pro and con) with other peptides scaffolds, protein scaffolds and synthetic scaffolds. Although the concept of trapping membrane proteins with a scaffold is not novel, there isn’t to our knowledge other reports showing functional thermostable reconstitution of protein in peptidiscs.
1) As a potential future user (and a current user of the nanodisc method), it is critical to know how expensive the synthesis (or any alternative preparation) of the NSP peptide is. I assume it is more expensive than detergents, but in contrast to the MSP protein used for nanodiscs, it might be less expensive. Along the same lines, the authors stated that the purity of their chemically synthetized NSP peptide was "more than 80%". Is this sufficient? What are the impurities. Is this rather low purity linked to the price of the synthesis?
Even low-purity peptides are still “cleaner” compared to other MSPs membrane protein scaffolds which are contaminated with lipids and other proteins due to recombinant cell protein expression (a major issue for the biotech industry when generating antibodies against membrane proteins stabilized in nanodiscs). In peptidisc, the bulk of “contaminants” consists of the NSPr missing its final amino acid (aspartate) during synthesis. In comparison to MSP protein, referenced from Sigma Aldrich, the peptide is less expensive. We have included comments on these points in the Discussion.
2) How aggregation-prone are peptidiscs? The authors hypothesize that the NSP peptides build a regular belt around the membrane protein, akin to the MSP belt. But is this really feasible? It is clear that the peptides arrange such that the hydrophobic parts of the membrane protein are covered by its hydrophobic side, while the hydrophilic face of the peptides remains exposed to the solvent. But this process is likely to be stochastic in the sense that there remain hydrophobic gaps on the membrane protein surface, as well as on some of the not perfectly placed NSP peptides. These remaining hydrophobic surfaces would then serve as nucleation points for protein aggregation. Remaining detergent molecules originating from the solubilization and purification of the membrane protein may shield these remaining hydrophobic surfaces.
This reviewer raises very important and fundamental questions, and the exact same questions apply to the nanodisc and SMA polymer systems as well. From our experiments, detergent appears to be completely eliminated from the peptidisc, as characteristic effects of a detergent environment on protein activity and stability are not seen once the protein is transferred into the peptidisc. The experimental data show that aggregation is not occurring and the system are very stable in water solution. We have also revised the Discussion to clarify how the peptide may possibly orientate itself around the membrane protein in a peptidisc to account for the issue of exposed alkyl chains.
3) Related to the above comment: Do peptidiscs still contain detergents? It is well known that quantitative detergent exchange is not a trivial endeavor. Often the initial detergent used to solubilize the membrane protein remains present to some degree. The authors need to show, whether in the peptidiscs there are still some detergents remaining (next to annular lipids and the NSP peptide).
From the high degree of correlation between our calculated and observed masses for the peptidiscs, it is likely that most, if not all of the detergent is removed from the peptidisc. Furthermore, in all cases of exchange from detergent to peptidisc, significant changes in enzyme activity or thermostability are observed, suggesting that even if a small amount of detergent remains, the denaturing effect of a detergent environment does not. We have discussed this point in the expanded text.
4) How big is the NSP belt compared to detergent belts of different sizes? This question is highly relevant for membrane protein crystallization. The peptidisc appears more compact than a nanodisc and membrane proteins in peptidiscs might be crystallized. Did the authors try to crystallize their five membrane proteins used in this study, for which high resolution crystal structures (determined in detergent) exist?
We are also very excited with the perspective of getting crystals! Crystal trials are ongoing, and based on EM comparisons of MalFGK2, we also think the peptidisc is more compact than the nanodisc. However, the crystal optimization and use of the peptidisc for crystallization studies are outside the scope of this current paper.
5) Does the peptidisc really mimic the natural environment of a membrane protein? The authors carefully worked out that the peptidisc contains annular lipids. However, annular lipids are also contained in detergent-purified samples. The authors should address this question by comparing the content of annular lipids side-by-side and over time using the same transporter purified in detergent as well as in peptidiscs and measuring by mass spectrometry the mass (or loss of mass over time) of the entire complex.
This question applies to all other membrane scaffold systems as well. The peptidisc is clearly more gentle environment than the detergent micelle. Comparison of BRC reconstituted in SMA, nanodiscs, peptidiscs, and proteoliposomes show that all these membrane mimetics lead to a comparable increase in the protein’s thermostability. However, the peptidisc also preserves protein ability to bind surface ligands as shown in the manuscript with FhuA/ColM/Ferricrocin and MalFGK2-MalE. In detergents, these proteins show binding/activity that is inconsistent with their performance in a lipid system.
6) The manuscript lacks a proper Discussion. Most problematically, the authors do not discuss at all that the concept of peptides used as detergent surrogates is not novel at all. Rather, it was first described in the 80s and 90s as peptitergent (you can find this word even on Wikipedia). See also in Schafmeister CE, Miercke LJ, Stroud RM. Structure at 2.5 A of a designed peptide that maintains solubility of membrane proteins. Science. 1993 Oct 29;262(5134):734-8. In addition, there are lipopeptide detergents (LPDs), nano-structured β-sheets (see Tao H, Lee SC, Moeller A, Roy RS, Siu FY, Zimmermann J, Stevens RC, Potter CS, Carragher B, Zhang Q. Engineered nanostructured β-sheet peptides protect membrane proteins. Nat Methods. 2013 Aug;10(8):759-61). Further, there was the development of short peptides for membrane stabilization, which are much cheaper to produce than the 37 aa NSP peptide (see Kiley P, Zhao X, Vaughn M, Baldo MA, Bruce BD, Zhang S. Self-assembling peptide detergents stabilize isolated photosystem I on a dry surface for an extended time. PLoS Biol. 2005 Jul;3(7):e230. Epub 2005 Jun 21.)
Important is the question whether at all or to what extent the NSP peptide is superior to these previously described methods using amphiphilic peptides, in particular the short designer peptides as described in Kiley et al., 2005). Critically, this point needs to be addressed experimentally by a functional assay including the maltose transporter and/or the BRC.
We absolutely agree with the reviewer. The manuscript was initially submitted as a Short Communication, which has text limitations. We have now rewritten sections Introduction and Discussion so that many of the well thought out comments above can be addressed. In the revised manuscript, we explain why the peptidisc is superior than previous peptide-based methods, as well as addressing synthetic polymers protein scaffold based systems. We comment on possible orientations of the peptide in the peptidisc, as well as the possibility of detergents remaining in the particles.
The Discussion should also include the concept of membrane protein stabilization via peptide-protein interactions (peptidisc) versus lipid protein interactions (nanodisc/saposins/proteoliposomes). And finally, it needs to address the amphipols (which are conceptually most closely linked to peptitergents and peptidiscs) in more detail.
We have modified the Discussion to encompass these points in more detail.
[Editors’ note: the author responses to the re-review follow.]
The manuscript has been improved but there are some remaining issues that need to be addressed before acceptance, as outlined below:
1) What is interesting in the revised version is the fact that the sequence of the original NSP peptide as it is present in ApoA1 was in fact reversed and that this resulted in an increase of solubility of the peptide. Why was this information completely lacking in the first version?
The sequence utilized through this study has been consistently reported in the supplementary table and labeled generically as NSP (i.e. nano-scaffold peptide) because we did not find many differences in reconstitution performance compared to the NSP sequence described by Kariyazono et al. However, issues of lower solubility became more evident to us after submission of the first manuscript version. We therefore felt it important to present this information in the revision, include additional experimental data and also to determine the hydrophobicity moment to explain for this difference (Figure 8—figure supplement 1). To prevent confusion with Kariyazono et al., we had to rename this sequence to NSPr in the revised version, and it was an obvious mistake in hindsight not to label it as such in the first submission.
2) The exact sequence of the NSPr peptide should be provided directly in the Materials and methods section under section "Biological reagents and peptides" (I had difficulties finding it in Supplementary file 1).
The sequence of NSPr has been added to the manuscript Materials and methods, section entitled “Peptides”, in addition to Supplementary file 1.
3) The authors are now stating in the manuscript that the method is inexpensive. However, I would still be interested to know how much a mg or a gram costs and how much you will need for an experiment (by providing a reasonable range).
As an example, the peptide content of MsbA-peptidisc represents ¼ of the total molecular weight. Therefore the reconstitution of MsbA in peptidisc using the on-bead method in quantities suitable for crystallization screening (~20mg/mL, 200µL) utilizes at least of ~1mg peptide. The cost of the peptide can be as low as ~5-15$/mg, pricing dependent on supplier, quantity and purity, thus the peptide cost for crystallization screen can be ball parked in the ~5-15$ range. An equivalent amount of MSP (https://www.sigmaaldrich.com/catalog/product/sigma/m6574?lang=en®ion=CA) would cost close to $80. There is, of course, additional peptide lost during purification steps (IMAC, SEC, concentration steps). The final cost in our lab to prepare this amount of MsbA in peptidisc for crystal screening is $20, assuming we lose about 60% of the MsbA during the purification steps. In comparison, each Hampton sitting drop crystal plate will cost close to 5$. In the case of EM experiments, the cost is even lower since less protein is required (5-10 mg/mL). In either case, the on-bead method is important to limit the total cost because the peptidisc solution can be re-utilized multiple times, as only the requisite amount of peptide binds the protein. In comparison, the on-column reconstitution is more wasteful because the initial 10 fold excess of peptide cannot be recovered as it fractionates with small size impurities. The on-gel reconstitutions utilize approximately 2-8µg of NSPr, so the cost of optimization is extremely low. Another very important consideration is labor time generally associated with the preparation of other membrane scaffolds. This is not the case with the peptidisc since the product can be manufactured.
4) Concerning the impurities of the peptide, the authors gave an interesting answer in the response letter (i.e. that the great majority of the impurity is in fact the same peptide missing its final amino acid). This should be mentioned in the manuscript as well.
This information and the degree of peptide purity are now presented together in the same place in the Materials and methods, section entitled “Peptides”.
5) The authors refer to a homepage to order/obtain the peptide: www.peptidisc.com. However, the homepage does work yet. It seems that the authors will or have already founded a start-up company to commercialize the peptidisc. Potential financial competing interests have to be declared.
We recently started a website with the goal to deliver NSPr in a kit along model membrane proteins so that other researchers can easily reproduce the core reconstitution experiments we present in this manuscript before proceeding on their own target. We mentioned in the manuscript “To aid accessibility to the academic community, bulk NSPr peptides and core protocols are available at www.peptidisc.com”, and we can declare a competing financial interest where necessary. Our goal is to make the method accessible to all and the anticipated cost will reflect a small revenue to cover its distribution and efforts to ensure its proper application including technical advices. We had not released the website yet due to the paper remaining under review and will do soon after publication. We have also named the other various source of peptide manufacturers employed in this study. The peptidisc presents an additional interesting opportunity in comparison to other scaffolds, as its use is not protected by intellectual property so it can be easily utilized by academics and industry alike.