Skip to main content
eLife logoLink to eLife
. 2020 Jul 30;9:e56518. doi: 10.7554/eLife.56518

Structure and dynamics of a nanodisc by integrating NMR, SAXS and SANS experiments with molecular dynamics simulations

Tone Bengtsen 1, Viktor L Holm 2, Lisbeth Ravnkilde Kjølbye 3,4, Søren R Midtgaard 2, Nicolai Tidemand Johansen 2, Giulio Tesei 1, Sandro Bottaro 1, Birgit Schiøtt 3, Lise Arleth 2,, Kresten Lindorff-Larsen 1,
Editors: Lucie Delemotte5, José D Faraldo-Gómez6
PMCID: PMC7426092  PMID: 32729831

Abstract

Nanodiscs are membrane mimetics that consist of a protein belt surrounding a lipid bilayer, and are broadly used for characterization of membrane proteins. Here, we investigate the structure, dynamics and biophysical properties of two small nanodiscs, MSP1D1ΔH5 and ΔH4H5. We combine our SAXS and SANS experiments with molecular dynamics simulations and previously obtained NMR and EPR data to derive and validate a conformational ensemble that represents the structure and dynamics of the nanodisc. We find that it displays conformational heterogeneity with various elliptical shapes, and with substantial differences in lipid ordering in the centre and rim of the discs. Together, our results reconcile previous apparently conflicting observations about the shape of nanodiscs, and pave the way for future integrative studies of larger complex systems such as membrane proteins embedded in nanodiscs.

Research organism: E. coli

Introduction

Nanodiscs are widely used membrane models that facilitate biophysical studies of membrane proteins (Bayburt et al., 2002). They are derived from, and very similar to, the human ApoA1 protein from high density lipoproteins (HDL particles) and consist of two amphipatic membrane scaffold proteins (MSPs) that stack and encircle a small patch of lipids in a membrane bilayer to form a discoidal assembly. The popularity of nanodiscs arises from their ability to mimic a membrane while at the same time ensuring a small system of homogeneous composition, the size of which can be controlled and can give diameters in a range from about 7 to 13 nm (Denisov et al., 2004; Hagn et al., 2013).

Despite the importance of nanodiscs in structural biology research and the medical importance of HDL particles, we still lack detailed structural models of these protein-lipid particles. The nanodisc has so far failed to crystallize, so a range of different biophysical methods have been used to provide information about specific characteristics. For example, mass spectrometry experiments have provided insight into lipid-water interactions and heterogeneous lipid compositions (Marty et al., 2014; Marty et al., 2015), solid state NMR has been used to quantify lipid phase transition states and lipid order (Mörs et al., 2013; Martinez et al., 2017) and small angle X-ray scattering (SAXS) and -neutron scattering (SANS) have provided insight into the size and low resolution shape of nanodiscs in solution (Denisov et al., 2004; Skar-Gislinge et al., 2010; Midtgaard et al., 2014). These experiments have been complemented by molecular dynamics (MD) simulations that provided both pioneering insights into the structure (Shih et al., 2005; Shih et al., 2007) as well as a better understanding of the assembly process, lipid-protein interactions and how much a nanodisc mimicks the membrane bilayer (Siuda and Tieleman, 2015; Debnath and Schäfer, 2015; Vestergaard et al., 2015).

A high resolution structure of the MSP protein belt encircling the nanodisc was recently obtained from the small, helix-5-deleted nanodisc, MSP1D1ΔH5 (henceforth ΔH5), reconstituted with DMPC lipids (ΔH5-DMPC) (Bibow et al., 2017) by combining nuclear magnetic resonance (NMR) spectroscopy, electron paramagnetic resonance (EPR) spectroscopy and transmission electron microscopy (TEM) (Bibow et al., 2017). While these experiments were performed on lipid-loaded nanodiscs, the study focused on the protein components, and on determining a time- and ensemble averaged structure of these, but left open the question of the role of the lipids (Martinez et al., 2017) as well as any structural dynamics of the overall nanodisc. Intriguingly, the resulting structure of the belt proteins corresponded to that of an almost circularly-shaped disc, while our previous SAXS/SANS investigations are clearly consistent with discs with an on-average elliptical cross-section (Skar-Gislinge et al., 2010; Midtgaard et al., 2015).

Here, we build upon this work to study the structure and dynamics of the nanodisc and the lipid properties in the disc. We performed SAXS and SANS experiments on the ΔH5-DMPC variant, and integrated these with MD simulations and the NMR data (Bibow et al., 2017) through an integrative Bayesian/maximum entropy (BME) approach (Hummer and Köfinger, 2015; Różycki et al., 2011; Bottaro et al., 2020; Bottaro et al., 2018; Orioli et al., 2020). We thereby obtain a model of the conformational ensemble of the ΔH5-DMPC nanodisc that is consistent with the structural information obtained from each method, as well as our molecular simulations, and which successfully explains differences in previous structural interpretations. In addition, we study the lipid ordering in our ensemble, and use the results to aid in the interpretation of Differential Scanning Calorimetry (DSC) measurements of the melting transition of DMPC in differently sized nanodiscs. Our study exemplifies how these integrative methods can be used to protein-lipid systems, possibly paving the way for future studies of membrane proteins embedded in nanodiscs.

Results and discussion

Structural investigations of ΔH5-DMPC and ΔH4H5-DMPC nanodiscs by SAXS and SANS

We determined optimal reconstitution ratios between the DMPC lipids and the ΔH5 and ΔH4H5 protein belts to form lipid-saturated nanodiscs based on a size-exclusion chromatography (SEC) analysis (Figure 1—figure supplement 1 and Materials and methods). In line with previous studies (Hagn et al., 2013), we found that reconstitution ratios of 1:33 for ΔH4H5:DMPC and 1:50 for ΔH5:DMPC were optimal in order to form single and relatively narrow symmetric peaks. Building upon earlier work for other discs (Denisov et al., 2004; Skar-Gislinge et al., 2010) we performed combined SEC-SAXS and SEC-SANS experiments to determine the size and shape of DMPC loaded ΔH5 and ΔH4H5 nanodiscs (Figure 1). These experiments were performed at 10°C, and based on results from previous NMR experiments on nanodiscs (Martinez et al., 2017) as well as a melting temperature TM ≈ 24°C for DMPC, where we expect the lipids to be in the gel-phase. Our SAXS and SANS data all exhibit a flat Guinier region at low q and indicate no signs of aggregation (Figure 1A,B). In both the ΔH5-DMPC and ΔH4H5-DMPC systems, the SAXS data exhibit an oscillation at medium to high q ([0.05:0.2] Å-1) arising from the combination of a negative excess scattering length density of the hydrophobic alkyl-chain bilayer core and positive excess scattering length densities of the hydrophilic lipid PC-headgroups and the amphipathic protein belt. The SANS data decreases monotonically as a function of q in accordance with the homogeneous contrast situation present here. These two different contrast situations, core-shell-contrast for SAXS and bulk-contrast for SANS, are also clearly reflected in the obtained p(r)-functions (Figure 1C,D), which also confirm that the ΔH5-DMPC nanodiscs are slightly larger than the ΔH4H5-DMPC nanodiscs.

Figure 1. SEC-SAXS and SEC-SANS analysis of nanodiscs.

(A) SEC-SAXS (dark purple) and SEC-SANS (light purple) for ΔH5-DMPC nanodiscs at 10°C. The continuous curve show the model fit corresponding to the geometric nanodisc model shown in E. (B) SEC-SAXS (dark orange) and SEC-SANS (light orange) data for the ΔH4H5-DMPC nanodiscs at 10°C. (C,D) Corresponding pair-distance distribution functions. (E, F) Fitted geometrical models for the respective nanodiscs (drawn to scale relative to one another).

Figure 1.

Figure 1—figure supplement 1. SEC analysis of the reconstitution of ΔH4H5 and ΔH5 nanodiscs with DMPC.

Figure 1—figure supplement 1.

(A) ΔH4H5 with DMPC at variable molar ratios of DMPC to ΔH4H5 with the molar stoichiometry indicated in the plot. (B) ΔH5 with DMPC at variable molar ratios of DMPC to ΔH4H5. In both plots, a reconstitution of MSP1D1:DMPC is inserted as reference (black line). The SEC analysis is performed using a GE Healthcare Life Science Superdex 200 10/300 GL column.
Figure 1—figure supplement 2. Model-based interpretation of the SAXS/SANS data on DMPC based nanodiscs obtained under different conditions.

Figure 1—figure supplement 2.

Top left) SAXS data from ΔH5ΔHis (I.e. ΔH5 with removed his-tags) obtained at 30°C (red) and 10°C (blue). Experimental data (points) and model fits (full lines). Top right) His-tagged ΔH5-DMPC nanodiscs measured at 10°C with SEC-SAXS (dark violet) and SEC-SANS (light violet). Data were fitted with the analytical model for nanodiscs with elliptical cross-section (see description in main article). Bottom) Table with the parameter values of the shown best model fits for the different samples. In all cases, that is, with/without His-tag and below and above the DMPC melting temperature, we found an axis ratio of the formed discs different from unity (between 1.2 and 1.4). Hence neither the variation of temperature nor the removal of the His-tag affects the overall conclusion that the elliptically-shaped nanodiscs describe the obtained small-angle scattering data.
Figure 1—figure supplement 3. Varying the axis ratio in the model.

Figure 1—figure supplement 3.

We repeated the parameterization of the coarse-grained model by scanning a range of fixed values of the axis ratio and refitted the remaining parameters to optimize the fit. (A) Comparison between experimental SAXS data and those calculated from the model with different values of the axis ratio (AR). (B) Quantification of the agreement between experiment and model. (C and D) show zoom ins on regions highlighted in A.
Figure 1—figure supplement 4. Introducing polydispersity in the model.

Figure 1—figure supplement 4.

We implemented a model for the nanodiscs that included a normally distributed dispersity around the average number of embedded lipids, where the width of the Gaussian was defined by its relative standard deviation in the number of embedded lipids, σlip, and truncated the Gaussian at ±3σlip. An upper hard limit for the number of lipids in the distribution was furthermore defined by the value that yielded circular and hence fully loaded discs. A lower hard limit was defined by the value that yielded discs with axis ratios exceeding 2. (A) Comparison between experimental SAXS data and those calculated from the model with different values of σlip, with σlip=10-4 representing a monodisperse system. (B) Quantification of the agreement between experiment and model. (C and D) show zoom ins on regions highlighted in A.

Our data are in qualitative agreement with the SAXS and SANS data obtained for MSP1D1 nanodiscs (Denisov et al., 2004; Skar-Gislinge et al., 2010) and similar systems (Midtgaard et al., 2014; Midtgaard et al., 2015), and indicate an ‘on average’ discoidal structure. Therefore, w first analyzed the scattering data by global fitting of a previously developed molecular-constrained geometrical model for the nanodiscs (Skar-Gislinge et al., 2010; Skar-Gislinge and Arleth, 2011; Pedersen et al., 2013). The model (see Materials and methods) describes the lipid interior of the nanodisc as a stack of flat, elliptically-shaped discs that consists of the hydrophobic bilayer that is sandwiched in between the two hydrophilic headgroup layers. The inner lipid part of the nanodisc is encircled by a hollow cylinder with an elliptical cross-section, which models the two protein MSP-belts stacked upon one another (Figure 1E,F). Using this model, we obtained excellent simultaneous fits to SAXS and SANS data for both the ΔH4H5-DMPC and ΔH5-DMPC nanodiscs (Figure 1A,B).

We find the area per headgroup, Ahead, for DMPC for both systems (ca. 55 Å2 ; Table 1 left), somewhat higher than the Ahead of gel-phase DMPC (47.2 ± 0.5 Å2 at 10°C) (Tristram-Nagle et al., 2002), but in agreement with the very broad melting transition observed in our DSC data (see below). We find 65 ± 13 and 100 ± 14 DMPC molecules in the nanodiscs for ΔH4H5 and ΔH5, respectively, in agreement with the reconstitution ratios reported above.

Table 1. Parameters of the SAXS and SANS model fit.

Left: Parameters for the simultaneous model fits to SEC-SAXS and SEC-SANS of His-tagged nanodiscs (denoted -His) for both ΔH4H5-DMPC and ΔH5-DMPC. Both measurements were obtained at 10°C. Right: Standard solution SAXS measurements of the ΔH5-DMPC nanodisc without His-tags (denoted -ΔHis) obtained at two different temperatures, in the gel phase at 10°C and in the liquid phase at 30°C. * marks parameters kept constant.

SEC-SAXS+SEC-SANS SAXS
ΔH4H5-His ΔH5-His ΔH5-ΔHis ΔH5-ΔHis
T 10°C 10°C 10°C 30°C
χ2reduced 1.95 5.12 3.76 2.40
Fitting Parameters
Axis Ratio 1.3 ± 0.4 1.2 ± 0.2 1.4 ± 0.1 1.3 ± 0.1
AHead 55 ± 5 Å2 54 ± 2 Å2 52 ± 2 Å2 60 ± 3
HBelt 24* Å2 24* Å2 24* Å2 24* Å2
NLipid 65 ± 13 100 ± 14 102 ± 7 104 ± 9
CVbelt 1* 1* 1* 0.97 ± 0.02
CVlipid 1.00 ± 0.02 1.01 ± 0.01 1.003 ± 0.007 1.044 ± 0.007
Scalex–ray 1.13 ± 0.28 1.1 ± 0.2 1.2 ± 0.1 1.2 ± 0.2
Scaleneutron 1.7 ± 0.5 0.8 ± 0.2 - -
Results From Fits
Hlipid 40 Å 41 Å 41 Å 38 Å
Htails 28 Å 28 Å 29 Å 26 Å
Rmajor 27 Å 32 Å 34 Å 36 Å
Rminor 21 Å 27 Å 25 Å 28 Å
Wbelt 10 Å 9 Å 9 Å 9 Å

Temperature dependence probed by SAXS and SANS

We continued to investigate the impact of temperature and His-tags on both the SAXS measurements and the resulting geometrical model of ΔH5-DMPC. We acquired standard solution SAXS data for a new preparation of the ΔH5-DMPC nanodiscs, this time without His-tags and measured at both 10°C and 30°C. At these two temperatures the DMPC is expected to be dominantly in the gel and liquid phase, respectively, as they are below and above the melting transition temperature (Martinez et al., 2017) (see also DSC analysis below). We used a standard solution SAXS setup for these measurements, as this at present provides a better control of both the sample temperature and sample concentration than in the SEC-SAXS based measurement. The effect of the DMPC melting transition is clearly reflected in the SAXS data (Figure 1—figure supplement 2) where both the position of the first minimum and the shape of the oscillation changes as the DMPC transitions from the gel to the molten state. We observe that the intensity of the forward scattering decreases significantly with increasing temperature, a result of the small but significant temperature-dependent change of the partial specific molecular volume of the DMPC.

To analyze the data, we again applied the molecular constrained geometrical model for the nanodiscs (Table 1, Right). Here, the effect of the DMPC melting transition can clearly be seen on the obtained DMPC area per headgroup which increases significantly as a result of the melting. Qualitatively similar observations of the melting transition of DMPC and DPPC based nanodiscs were previously reported in the MSP1D1 and MSP1E3D1 nanodiscs using DSC, SAXS and fluorescence (Denisov et al., 2005; Graziano et al., 2018). Regarding the shape of the ΔH5 nanodiscs without the His-tag (Figure 1—figure supplement 2), we find parameters similar to those derived from SEC-SAXS/SANS experiments including an elliptical shape with ratios of the two axes between 1.2 and 1.4. This observation is in apparent contrast to the recently described integrative NMR/EPR structural model of the ΔH5-DMPC nanodisc which was found to be more circular (Bibow et al., 2017). We therefore examined the fit to the model varying the axis ratios from 1.0 to 1.6 and indeed find that a number of features are best explained with a slightly asymmetric model (Figure 1—figure supplement 3). Both in the SEC-SAXS/SANS experiments, but perhaps particularly in the standard solution SAXS setup, it is possible that polydispersity in the number of lipids embedded in the nanodiscs is present (Skar-Gislinge et al., 2018), and contributes to the shapes obtained from our models (Caponetti et al., 1993). We therefore analyzed our data using a model where we include polydispersity through a normally-distributed number of lipids, parameterized via the relative standard deviation (σlip). Our results show that while a modest level of polydispersity (ca. 1%) cannot be ruled out, greater levels lead to worsening of the fit to the data (Figure 1—figure supplement 4).

Molecular dynamics simulations

The results described above suggest an apparent discrepancy of the solution structure of the ΔH5-DMPC nanodisc when viewed either by NMR/EPR or SAXS/SANS. In particular, the NMR/EPR structure revealed a circular shape whereas the SAXS/SANS experiments suggested an elliptical shape. The two kinds of experiments, however, differ substantially in the aspects of the structure that they are sensitive to. Further, both sets of models were derived in a way to represent the distribution of conformations in the experiments by a single ‘average’ structure.

In order to understand the structural discrepancies between the two solution methods better, and to include effects of conformational averaging, we performed atomistic MD simulations of the His-tag truncated ΔH5-DMPC nanodisc. In these simulations, we mimicked the experimental conditions of the standard solution SAXS measurements obtained at 30°C and used 100 DMPC lipids in the bilayer as found above. We performed two simulations (total simulation time of 1196 ns) using the CHARMM36m force field (Huang et al., 2017). We visualized the conformational ensemble of the ΔH5-DMPC nanodisc by clustering the simulations, and found that the three most populated clusters represent 95% of the simulations. Notably, these structures all have elliptical shapes, but differ in the directions of the major axis (Figure 2A).

Figure 2. Comparing MD simulations with experiments.

(A) Visualization of the conformational ensemble from the MD simulation by clustering (blue). Only the protein parts of the nanodisc are visualized while the lipids are left out to emphasize the shape. The top three clusters contain 95% of all frames. The previous NMR/EPR-structure is shown for comparison (red). (B) Comparison of experimental standard solution SAXS data (red) and SAXS calculated from the simulation (blue). Green dotted line is the back-calculated SAXS from the integrative NMR/EPR-structure (labelled PDB). Residuals for the calculated SAXS curves are shown below. Only the high q-range is shown as the discrepancy between simulation and experiments are mainly located here (for the entire q-range see Figure 2—figure supplement 1). (C) Comparison of average distances from simulations (blue) to upper-bound distance measurements (red) between methyl NOEs. The labels show the residues which the atoms of the NOEs belong to.

Figure 2.

Figure 2—figure supplement 1. Comparing simulations with SAXS data.

Figure 2—figure supplement 1.

This figure is an expanded version of that in the main text, which shows only part of the q-range (marked in white).
Figure 2—figure supplement 2. HN-NOE.

Figure 2—figure supplement 2.

Comparison of average distances from simulations (blue) to upper-bound distance measurements (red) between HN-NOEs.
Figure 2—figure supplement 3. Example of a His-tagged nanodisc used for SANS calculations.

Figure 2—figure supplement 3.

Figure 2—figure supplement 4. Comparing MD simulations with SANS data.

Figure 2—figure supplement 4.

Figure 2—figure supplement 5. Comparing MD simulations with PRE data.

Figure 2—figure supplement 5.

Each panel corresponds a different probe position as indicated by the labels. We show the experimental values (black), those calculated from the structure determined using these and other data (grey) and our MD simulations both before (blue) and after (red) reweighting. For many probe positions and residues, the values calculated from the PDB structure and our simulations are very similar, so that the colored lines appear hidden beneath the grey line.
Figure 2—figure supplement 6. Comparing MD simulations with EPR data.

Figure 2—figure supplement 6.

Each panel corresponds a different probe position as indicated by the labels. We show the distance distributions estimated from the experimental measurements (black), and compare to those calculated from the structure determined using these and other data (grey) and our MD simulations both before (blue) and after (red) reweighting.

We then examined the extent to which the simulations agree with the ensemble-averaged experimental data, focusing on the SAXS experiments and NOE-derived distance information from NMR. We calculated the SAXS intensities from the simulation frames using both FOXS (Schneidman-Duhovny et al., 2013; Schneidman-Duhovny et al., 2016, Figure 2B) and CRYSOL (Svergun et al., 1995, Figure 2—figure supplement 1) and compared to the corresponding standard solution SAXS experiments obtained at 30°C. Similarly, we used r-3-weighted averaging to calculate the effective distances in the simulations and compared them to the previously reported methyl (Figure 2C) and amide NOEs (Figure 2—figure supplement 2, Bibow et al., 2017). The discrepancy observed between the simulation and the experiments were quantified by calculating χ2 (Table 2).

Table 2. Comparing experiments and simulations.

We quantify agreement between SAXS and NMR NOE experiments by calculating the χ2. The previously determined NMR structure (Bibow et al., 2017) (PDB ID 2N5E) is labelled PDB, the unbiased MD simulation by MD, and simulations reweighted by experiments are labelled by MD and the experiments used in reweighting. Srel is a measure of the amount of reweighting used to fit the data (Bottaro et al., 2018) (see Methods for more details).

Data for integration Srel χ2
SAXS NOE
PDB 2.9 9.5
MD 0 10.0 8.2
MD + SAXS -1.7 1.5 7.9
MD + NOE -1.9 8.9 4.2
MD + SAXS + NOE -1.7 1.9 6.0

The comparison between experiments and simulations reveal an overall good agreement between the two. Interestingly, the simulations agree well with the SAXS data in the q-region where scattering is dominated by the lipid bilayer and where our geometric fitting of the models for SAXS generally are very sensitive. The MD simulation trajectory captures accurately the depth of the SAXS minimum around q = 0.07 Å–1; however, the shoulder observed in the experiments in the range 0.15 Å–1–0.20 Å–1 is not captured accurately.

Direct comparison of the previously determined integrative NMR/EPR structure (Bibow et al., 2017) to the SAXS data is made difficult by the missing lipids in the structure. We thus built a model of the lipidated structure by first adding DMPC lipids to the NMR/EPR solved structure (PDB ID 2N5E), and then equilibrating only the lipids by MD, keeping the protein conformation fixed. When we use this structure to calculate the SAXS data, the back-calculated data overshoots the depth of the SAXS minimum but captures well the shoulder observed in the experimental data (Figure 2B). Thus, neither the MD trajectory nor the NMR/EPR structure fit perfectly with the measured SAXS data.

When comparing the simulations to the NMR-derived distances between methyl groups (Figure 2C), we generally find good agreement, but observe a few distances that exceed the experimental upper bounds. A similar trend is observed in the comparison to amide NOEs (Figure 2—figure supplement 2) which shows overall good agreement but with a few NOEs violating at similar positions as for the methyl NOEs. As the amide NOEs are mostly sensitive to the local helical structure, the good agreement with this data mostly reflects that the secondary structures are maintained during the simulations.

We also compared the simulations to the SANS data for ΔH5-DMPC. The scattering contrast is very different in SAXS and SANS, and the scattering from the lipid bilayer has a relatively higher amplitude in the latter. This gives an independent check that the simulation provides a good description of the structure of the lipid bilayer. As the SEC-SANS data were measured on a His-tagged ΔH5-DMPC nanodisc, we therefore simulated this situation by creating an ensemble of His-tag structures and randomly sampled and attached these to the outer MSP-belts in the simulation frames under the assumption that the His-tags are disordered on the nanodiscs (Figure 2—figure supplement 3). As for the SAXS and NOE data we also here find a generally good agreement (Figure 2—figure supplement 4).

As a final consistency check, we compared our simulations to NMR paramagnetic relaxation enhancement (PRE) data (Figure 2—figure supplement 5) and EPR data (Figure 2—figure supplement 6), that both use spin-labels to probe longer range distances. As reference, we used the calculation of the PRE and EPR data from the structure that was derived using these and the remaining NMR data (Bibow et al., 2017) and find comparable agreement.

Integrating experiments and simulations

While the MD simulations are overall in good agreement with the SAXS and NMR NOE data, there remain discrepancies that could contain information about the conformational ensemble of ΔH5-DMPC in solution. We therefore used a previously described Bayesian/Maximum Entropy (BME) approach (Hummer and Köfinger, 2015; Różycki et al., 2011; Bottaro et al., 2018; Cesari et al., 2016; Bottaro et al., 2020) to integrate the MD simulations with the SAXS and NMR data. Briefly, the BME method refines the simulation against measured experimental averages by balancing (1) minimizing the discrepancy between the simulation and the observed experimental averages and (2) ensuring as little perturbation of the original simulation as possible thereby limiting chances of overfitting. The outcome is a conformational ensemble that is more likely to represent ΔH5-DMPC in solution. In practice, this is achieved by changing the weight of each configuration in the ensemble obtained from the MD simulations, and we therefore call this a ‘reweighted ensemble’ (Bottaro and Lindorff-Larsen, 2018; Bottaro et al., 2020). The amount of reweighting can be quantified by an entropy change (Srel) that reports on how much the weights had to be changed to fit the data (Bottaro et al., 2018) (see Materials and methods). Alternatively, the value ϕeff=exp(Srel) reports on the effective ensemble size, that is what fraction of the original frames that were used to derive the final ensemble (Orioli et al., 2020). We note that we reweight each individual conformation in the ensemble, and thus that the clustering is only used for presenting the results. In this way we avoid uncertainties that come from difficulties in clustering heterogeneous ensembles.

We used both the SAXS and NOE data individually, as well as combined, to understand the effects of each source of data on the reweighted conformational ensemble (Table 2). We note that when a specific type of data is used to generate the ensemble, the resulting χ2 simply reports on how well the simulation has been fitted to the data; because of the maximum-entropy regularization to avoid overfitting, we do not fit the data as accurately as possible. The two types of experimental data complement each other in structural information content. Specifically, the SAXS data report on the overall size and shape, and is sensitive to both protein and lipids through atom-atom pair distributions in a range starting from ≈10 Å, whereas the NOEs contain local, specific atom-atom distances from the protein belts of the ΔH5 but not any direct information about the lipids.

We find that refining against a single of the two data types only improves the MD trajectory with respect to the structural properties it is sensitive to, highlighting the orthogonal information in the two sources of information. In addition, we performed reweighting with the methyl NOEs and the amide NOEs separately (Figure 3—figure supplement 4). The already low discrepancy of the amide NOEs barely improves while the discrepancies of both methyl NOEs and SAXS are unaffected by integration with amide NOEs alone, implying that the structural information content contained in the amide NOEs (mostly secondary structure) is already correctly captured by the force field and starting structure. Because the NOE and SAXS experiments provide independent information we refined the ensemble against both sets of data (Figure 3). We find that we can fit both sources of data at reasonable accuracy without dramatic changes of the weights away from the Boltzmann distribution of the force field (ϕeff=18%).

Figure 3. Integrating simulations and experiments.

(A) SAXS data calculated from the simulation before and after reweighting the ensemble using experimental data. Only the high q-range is shown as the discrepancy between simulation and experiments are mainly located here (for the entire q-range see Figure 3—figure supplement 1). Agreement with the NOEs before and after integration are likewise shown in Figure 3—figure supplement 2. (B). Histogram of the acylindricity of the simulations (C) both before integration (dark blue) and after integration (light blue). (C) Visualization of the conformational ensemble showing structures sampled every 100 ns in cartoon representation (blue), the original NMR/EPR structure is shown in rope representation for comparison (red). The table below shows the acylindricity of the entire conformational ensemble before and after integration and compared to the original NMR/EPR (NMR) structure and the SAXS/SANS model fit. (D) Weights and acylindricity of the three main clusters of the MD simulation (blue) before and after integration.

Figure 3.

Figure 3—figure supplement 1. Comparison of the experimental SAXS data from simulation before and after integration.

Figure 3—figure supplement 1.

This figure is an expanded version of that in the main text and shows agreement with the simulation after reweighting. SAXS data were calculated using FOXS.
Figure 3—figure supplement 2. NOEs from simulation before and after reweighting.

Figure 3—figure supplement 2.

Figure 3—figure supplement 3. Determination of θ.

Figure 3—figure supplement 3.

θ is a hyperparameter that tunes the balance between fitting the data accurately (low χr2) and not deviating too much from the prior (low Srel) thereby avoiding overfitting. It is here determined by plotting Srel vs χr2 and selecting a value of θ near the natural kink and at a step where a similar decrease in χr2 gives rise to a much lower Srel, indicating that we cannot fit the to the experiments further without a risk of overfitting. The value of θ that produce the given (Srel,χr2) is annotated above the given point together with a measure of the effective number of frames used from the original simulation this gives rise to. Red dot marks the chosen θ.
Figure 3—figure supplement 4. Combining experiments and simulations.

Figure 3—figure supplement 4.

Similar analysis to the main text, but with the methyl- and HN-NOEs integrated individually. As can be seen, the methyl-NOE distances have a larger impact, likely due to the longer distances measured in methyl-NOE whereas the HN-NOEs mainly report on distances between atom pairs of 4 residues or less apart in the sequences and, thus, likely mainly on the helical secondary structure.

Finally, we used the PRE and EPR data to validate the refined ensemble. In general we find comparable and overall good agreement between the original NMR/EPR structure and our MD refined ensembles, suggesting that our ensembles are in good agreement with data that was not used directly as input in the refinement (Figure 2—figure supplement 5 and Figure 2—figure supplement 6). We further find that reweighting the MD simulations against the SAXS and NOE data generally improves the agreement to the EPR data. We thus proceed with our analysis of the structural features of ΔH5-DMPC using an ensemble of conformations that is based on integrating the MD simulations with both the SAXS and NOE experiments.

Analysis of the measured SAXS and SANS revealed an elliptical shape of the ΔH5-DMPC upon fitting of a single structure to the data. In contrast, the structure obtained by fitting the NMR/EPR data to a single structure gave rise to a more circular configuration. Combining the results of the two studies, we hypothesized that the nanodisc possesses underlying elliptical fluctuations with the major axis changing within the nanodisc. In such a system NMR and EPR measurements, which build on ensemble averaged information of specific atom-atom interactions, will give rise to an on-average circular structure. SAXS and SANS, on the contrary, which build on distributions of global distances rather than specific atom-atom distances, will not distinguish between the orientation of the major axis within the nanodisc and thus give rise to observations of an elliptical shape. By complementing the experiments with MD simulations we obtained a ensemble with structural features that support this hypothesis.

We thus quantified the degree of ellipticity in terms of an acylindricity parameter, C, defined as the difference between the x and y components of the gyration tensor (see Materials and methods for details). C is thus a measure of how far from a perfect circular cylinder the shape is, and C = 0 corresponds to a circular shape. We calculated both the average and distribution of the acylindricity from the simulated ensemble both before and after reweighting against the experimental data (Figure 3B and C). In addition, we calculated the acylindricity of both the integrative NMR structure and from the structural model obtained from the SAXS and SANS measurements.

We find that the acylindricity decreased from C=17 Å in the original MD simulation trajectory to C=15 Å after integration of the NMR and SAXS data, showing that the experiments indeed affect the structural features. This value is in the middle of that obtained from the analytical geometric model fitted to the SAXS data (C=22 Å) and that of the integrative NMR/EPR structure (C=8 Å) (Bibow et al., 2017). Thus, the acylindricity of the final, heterogenous ensemble lies between that of the two conformations that were fitted as single structures to fit either the NMR or SAXS data.

To understand better the elliptical shape of the ΔH5-DMPC nanodisc and the role played by reweighting against experiments, we calculated the average acylindricity for each cluster of conformations of ΔH5-DMPC both before and after integration with experimental data (Figure 3D). We note that because our reweighting procedure acts on the individual conformations and not at the coarser level of clusters, the average acylindricity changes slightly for each cluster upon reweighting. Clusters 1 and 2, which together constitute about 80% of the conformational ensemble (both before and after reweighting), are both clusters with high acylindricity, but with almost orthogonal directions of the major axis in the elliptical structure. The major change after integration is the exchange in populations of the two clusters resulting in cluster 2 to be weighted highest, underlining the influence and importance of the integration. Thus, our MD simulations and the integration with the experiments support the hypothesis of underlying elliptical fluctuations with the major axis changing direction inside the nanodisc, and we note that the detailed molecular description of this was only possible by combining the MD simulations with both the SAXS and NMR data.

Analyses of the lipid properties in nanodiscs

Nanodiscs are often used as models for extended lipid bilayers, but the presence and interactions with the protein belt — and the observed shape fluctuations — could impact the properties of the lipid molecules in the nanodisc compared to a standard bilayer. Building on earlier experimental (Mörs et al., 2013; Martinez et al., 2017) and simulation work (Siuda and Tieleman, 2015; Debnath and Schäfer, 2015) work, we therefore used our experimentally-derived ensemble of nanodisc structures to investigate the properties of lipids in the small ΔH5-DMPC nanodisc, and compared them to those in a DMPC bilayer. Specifically, we calculated the thickness of the DMPC bilayer (Figure 4A) and the order parameters, SCH, of the DMPC lipids (Figure 4B,C).

Figure 4. Lipid properties from simulations.

Figure 4.

2D plots of the (A) thickness and (B) order parameters averaged over the ensemble and all C-H bonds in the two aliphatic tails of the DMPC lipids in the ΔH5 nanodisc. The core and rim zones are indicated in panel B. Arrows indicate the average value in simulations of a DMPC bilayer. (C) The order parameters as a function of carbon number in the lipid tails in the ΔH5-DMPC disc. The rim zone is defined as all lipids within 10 Å of the MSPs, while the core zone is all the lipids not within 10 Å of the MSPs.

As done previously (Siuda and Tieleman, 2015; Debnath and Schäfer, 2015), we subdivide the lipid area in the nanodisc into zones dependent on the distance from the MSP protein belts (above or below 10 Å). The results of both the thickness and order parameter analyses show the same trend: a clear difference between the lipids close to the protein belt and those more central in the nanodisc. The results illustrates that the DMPC lipid bilayer in the ΔH5 nanodiscs is not homogeneous but rather thinner and un-ordered near the protein belt and thicker and more ordered in the core of the nanodisc, which in turn is more similar to a pure bilayer (Figure 4). These results are in line with previous simulation studies on the larger DMPC nanodiscs, MSP1, MSP1E1 and MSP1E2 (Siuda and Tieleman, 2015; Debnath and Schäfer, 2015), albeit performed without experimental reweighting, as well as with solid state NMR data on the both ΔH5-DMPC and the larger MSP1-DMPC (Mörs et al., 2013; Martinez et al., 2017).

We proceeded by performing DSC experiments on nanodiscs of different sizes to examine the impact of the differentiated lipid order in the core and rim of the nanodisc. Specifically, we examined the lipid melting transition of DMPC inside ΔH4H5, ΔH5 and the larger MSP1D1 nanodiscs, and used pure DMPC vesicles as reference. In line with earlier DSC experiments (Shaw et al., 2004), our results show that the melting transition peak broadens significantly in all three nanodisc systems compared to that of pure DMPC vesicles (Figure 5). The broader melting transition is in line with the observed differentiated lipid ordering in nanodiscs from the reweighted simulations, as the observed differences in how ordered the lipids are necessarily will cause differences in the melting temperature and thus give rise to the broader peaks. Furthermore, the broadened peaks are in line with results observed in previous solid state NMR experiments which found a substantially broadened and diminished lipid gel-liquid phase transition in the ΔH5-DMPC nanodisc in the temperature range 10-28°C (Martinez et al., 2017). Our results show that the transition enthalpy per mole of DMPC, i.e. the area under the curves, increases with the nanodisc size, in line with previous observations for, respectively, DMPC and DPPC in MSP1D1 and in the larger MSP1E3D1 systems (Denisov et al., 2005), where it was proposed to be due to the absence of a cooperative melting transition of the lipids at the nanodisc rim (Denisov et al., 2005).

Figure 5. DSC analysis of lipid melting in nanodiscs.

Figure 5.

The three DMPC-filled nanodiscs studied, listed by increasing size, are ΔH4H5-DMPC (orange), ΔH5-DMPC (purple) and MSP1D1-DMPC (green). DSC data from plain DMPC-vesicles (black) are shown for comparison. Arrows indicate the temperature with maximal heat capacity. DSC data from the three nanodisc samples are normalized by DMPC concentration, while the data from the DMPC liposome is on an arbitrary scale.

Interestingly, we observe that the maximum of the melting transition, TM, depends on nanodisc belt and can fall both below and above the TM of plain DMPC vesicles (24°C). In the smallest ΔH4H5 nanodisc, the DMPC has a TM ≈ 22.5°C. In ΔH5 the DMPC has TM at 24.5°C which is close to that of the DMPC vesicles, while the larger MSP1D1 nanodisc has a TM ≈ 28°C. This TM value is similar to the value of 28.5°C measured for DMPC melting in MSP1D1 by Denisov et al., 2005, who in addition measured a TM value of 27.5°C in the even larger MSP1E3D1 discs.

Together our results are in line with previous NMR experiments (Martinez et al., 2017), and suggest that the state of the ordering of the lipids in the nanodiscs is inhomogeneous compared to the DMPC vesicle, and that the behaviour of the lipids is modulated by their interaction with the membrane scaffold proteins. Our results point towards a non-trivial effect of the DMPC-MSP interactions. They can both destabilize DMPC in the gel-phase in the smaller nanodiscs (ΔH4H5-DMPC) where the low area-to-rim ratio leads to the lower TM compared to the DMPC vesicles, but also stabilize the DMPC gel-phase in larger nanodiscs with larger area-to-rim ratios such as MSP1D1-DMPC and MSP1E3D1-DMPC. Thus, when using nanodiscs as membrane mimics it is relevant to keep in mind that the given lipid gel/liquid state might be affected. We also note that even if lipids in larger discs are less perturbed than those in the smallest discs, introduction of membrane proteins into the discs might in itself perturb the lipids in ways similar to the MSPs.

Conclusions

Lipid nanodiscs are versatile membrane mimetics with a wide potential for studies of the structure, function and dynamics of membrane proteins. Despite their widespread use and numerous studies, we still do not have a full and detailed understanding of the structural and dynamic features of nanodiscs. This in turn limits our ability to interpret e.g. solution scattering experiments when membrane proteins are embedded into such nanodiscs. In order to further our understanding of the conformations and structural fluctuations of both the protein and lipid components in nanodiscs, we have performed a series of biophysical experiments on DMPC-loaded ΔH5 and ΔH4H5 nanodiscs.

Using SEC-SAXS and SEC-SANS measurements, we investigated the solution structure of the ΔH4H5-DMPC and ΔH5-DMPC discs. Model-based analysis of this data showed an ‘average’ elliptical shape of both nanodiscs. In contrast, a previously determined integrative NMR/EPR (Bibow et al., 2017) method gave rise to a more circular average structure of the ΔH5 nanodisc.

We reconcile these two apparently opposing views and provide a richer and more detailed view of the nanodisc proteins and lipids and their dynamics by performing MD simulations. In particular, we used a Bayesian/Maximum Entropy approach to integrate the MD simulations with the SAXS and NMR data to uncover the existence of underlying fluctuations between elliptical shapes with orthogonal major axes in consistency with both sources of data. We note that the NMR/EPR-derived structure, and our MD simulations initiated from this structure, provide good agreement with the SAXS data even without reweighting. Because our SAXS data are rather precise, however, we were able to detect subtle deviations that enabled us to refine our model. An interesting avenue for further analysis might be to use our structural ensembles to interpret electron microscopy data of nanodiscs. Negative stain transmission electron microscopy of ΔH5 appears to show discs of different shapes (Bibow et al., 2017), whereas class-averaged cryo-electron microscopy of a membrane protein embedded in a different nanodisc appears more symmetric (Frauenfeld et al., 2011). Direct comparisons between solution structures and electron microscopy data should also take into account any possible changes in shape that might happen during the freezing process. We have previously used contrast-variation to prepare specifically deuterated nanodiscs that become invisible to neutrons in D2O (Maric et al., 2014). In the future it would be interesting to use a similar strategy to study the belt proteins and lipids independently by matching out each component separately.

In addition to studying the overall shape fluctuations, we also analyzed the lipid structure and dynamics in the nanodiscs, and find an inhomogeneous distribution. Specifically, we find substantially perturbed lipid properties near the belt proteins, whereas the lipids more central in the disc behaved more similar to those in a pure DMPC bilayer. We used DSC to investigate the lipid melting transition in the small nanodiscs in comparison to the lipid vesicles and found that the melting takes place over a much broader temperature range in the small nanodiscs. The observed correlation between the size of the belt proteins and the lipid melting enthalpy give support to the proposition (Denisov et al., 2005), that the arrangement of the lipids near the nanodisc rim must be substantially perturbed. In particular, our results suggest that the belt proteins induces additional disorder to the lipid tails near the rim.

Together, our results provide an integrated view of both the protein and lipid components of nanodiscs. Approaches such as the one described here takes advantage of the increasing possibilities for accurate NMR and scattering data in solution, improved computational models for lipid bilayers as well as new approaches to integrate experiments and simulations. In this way, our study exemplifies how integrating multiple biophysical experiments and simulations may lead to new insight into a complex system and paves the way for future studies of membrane proteins inside nanodiscs.

Materials and methods

Expression of membrane scaffold protein (MSP) variants

We used previously reported constructs for ΔH4H5, ΔH5 and MSP1D1 (Hagn et al., 2013; Ritchie et al., 2009). We expressed and purified the proteins as previously described (Ritchie et al., 2009), with minor modifications to the purification protocol: The cells were opened in lysis buffer containing 50 mM Tris/HCl pH 8.0, 300 mM NaCl, 1% Triton X-100 and 6 M GuHCl by vigorous shaking for 15 min. Insoluble material was subsequently removed by centrifugation at 18,000 rpm for 1 hr using an SS-34 rotor. The supernatant was loaded on Ni-NTA resin pre-equilibrated in lysis buffer and washed extensively with the same buffer. Extensive washes using lysis buffer without GuHCl and subsequently wash buffer containing 50 mM Tris/HCl pH 8.0, 300 mM NaCl, 20 mM Imidazole and 50 mM Cholate was performed in order to remove GuHCl and Triton X-100. Protein was eluded in buffer containing 50 mM Tris/HCl pH 8.0, 300 mM NaCl, 500 mM Imidazole, concentrated, flash frozen and stored at −80°C until further use. Cleavage of the TEV-site was performed by addition of 1:20 TEV protease, and dialysing at room temperature for 6–12 hr against 20 mM TrisHCl pH 8, 100 mM NaCl, 0.5 mM EDTA, 1 mM DTT. TEV protease and any un-cleaved MSP was removed by passing the solution over Ni-NTA resin again.

Reconstitution of ΔH5-DMPC and ΔH4H5-DMPC nanodiscs

Before assembly, the DMPC lipids (Avanti Polar Lipids) were suspended in a buffer containing 100 mM NaCl, 20 mM Tris/HCl pH 7.5, and 100 mM sodium cholate detergent to a final lipid concentration of 50 mM. We determined optimal reconstitution ratios between the DMPC lipids and the ΔH5 and ΔH4H5 by first mixing the lipid and MSP stock solutions at a series of different molar concentration ratios in the range from 1:9 to 1:80 depending on the MSP type (Figure 1—figure supplement 1). In all samples, cholate was removed after mixing by addition of an excess amount of Amberlite detergent absorbing beads to start the assembly of the nanodiscs. The samples were left in a thermomixer for 4 hr at 28°C and the Amberlite was removed by centrifugation at 5000 rpm. Purification was performed using size exclusion chromatography (SEC) on an Äkta purifier (FPLC) system with a Superdex200 10/300 column from (GE Healthcare Life Science; S200). We found that reconstitution ratios of 1:33 for ΔH4H5:DMPC and 1:50 for ΔH5:DMPC resulted in a single and relatively narrow symmetric peak, in good agreement with the previously reported ratios of 1:20 for ΔH4H5:DMPC and 1:50 for ΔH5:DMPC (Hagn et al., 2013). More narrow and well-defined SEC-peaks were obtained if the reconstitution took place at or above the melting temperature, TM, of DMPC at 24°C (Ritchie et al., 2009).

Differential scanning calorimetry (DSC)

The measurements were performed on a VP-DSC (MicroCal) using a constant pressure of 1.7 bar (25 psi) and a scan rate of 1°C/min between 6°C and 40°C. All samples had been purified in PBS buffer prior to the measurement. We used the Origin instrument software for background subtraction and baseline correction using a ‘Cubic Connect’ baseline correction. Finally, the data were normalized by the lipid concentration of the individual samples.

SEC-SANS

SEC-SANS was performed at the D22 small-angle scattering diffractometer at the ILL, Grenoble, France using a recently developed SEC-SANS setup (Jordan et al., 2016; Johansen et al., 2018). Briefly, the setup was as follows: the in situ SEC was done using a modular HPLC system (Serlabo) equipped with a Superdex 200 GL gel filtration column (GE) with a void volume of approximately 7.5 ml and a flow rate of 0.25 ml/min. The SmartLine 2600 diode-array spectrophotometer (Knauer) was connected via optic fibers either to an optic cell of 3 mm path length placed at the outlet of the chromatography column, enabling the simultaneous recording of chromatograms at four different wavelengths, including 280 nm which we used for the concentration determination. All components of the HPLC setup including buffers and the column were placed in a closed cabinet connected to an air-cooling system set to 10°C to control the temperature of the sample. Before measurements, we equilibrated the column in a D2O-based buffer, and the buffer in the sample was exchanged to a D2O-based buffer using an illustra NAP-25 gravity flow column (GE). The D2O buffer contained 20 mM Tris/DCl pH 7.5 and 100 mM NaCl.

The experiments were carried out with a nominal neutron wavelength, λ, of 6.0 Å and a wavelength distribution, Δλ/λ=10% FWHM, a rectangular collimation of 40 mm × 55 mm and a rectangular sample aperture of 7 mm × 10 mm. The distance of the sample-detector used for the characterization of the nanodiscs was 5.6 m (with collimation of 5.6 m), covering a momentum transfer range, q, of 0.0087 Å−1 to 0.17 Å−1, with q=4πsin(θ)/λ, where θ is half the angle between the incoming and the scattered neutrons. Measured intensities were binned into 30 s frames. Sample transmission was approximated by the buffer measured at the sample-detector distance of 11.2 m. The measured intensity was brought to absolute scale in units of scattering cross section per unit volume (cm−1) using direct beam flux measured for each collimation prior to the experiment. Data reduction was performed using the GRASP software (https://www.ill.eu/fr/users-en/scientific-groups/large-scale-structures/grasp/). The SANS data appropriate for buffer subtraction was identified based on when the 280 nm absorption during the SEC curve showed no trace of protein.

SEC-SAXS

SEC-SAXS was performed at the BioSAXS instrument at BM29 at the ESRF, Grenoble, France (Pernot et al., 2013). Briefly, the setup at BM29 included an HPLC controlled separately from the SAXS measurement, coupled to a UV-Vis array spectrophotometer collecting absorption from 190 nm to 800 nm. Data were collected with a wavelength of 0.9919 Å using a sample-detector distance of 2.87 m which provided scattering momentum transfers ranging from 0.003 Å−1 to 0.49 Å−1. The capillary was cooled to 10°C, however, the HPLC including the SEC-column was placed at ambient temperature. Size exclusion chromatography was performed using the same column as for SEC-SANS and equivalent H2O-based buffer. A flow rate of 0.5 ml/min was used. Data reduction was carried out using the in-house software, and subsequent conversion to absolute units was done with water as calibration standard (Orthaber et al., 2000). The 1 s frames recorded were subsequently averaged in 10 s bins.

Standard solution SAXS

Standard solution SAXS data were obtained at the P12 beamline at the PETRA III storage ring in Hamburg, Germany (Blanchet et al., 2015) using a wavelength of 1.24 Å, a sample-detector distance of 3 m, providing a momentum transfers covering from 0.0026 Å−1 to 0.498 Å−1 and a variable temperature in the exposure unit. 20 exposures of 0.045 s were averaged, background subtracted and normalized to absolute scale units (cm−1) using Bovine Serum Albumin, BSA as calibration standard by the available software at the beamline. The measurements were performed at both 10°C and 30°C.

SAXS and SANS data analysis

The output of the SAXS and SANS experiments were small-angle scattering data in terms of absolute intensities I(q). I(q) was transformed into the pair distance distribution function, p(r), by indirect Fourier transformations using BayesApp (Hansen, 2014). Further SAXS/SANS modelling was carried out using our previously developed WillItFit software (Pedersen et al., 2013) (https://sourceforge.net/projects/willitfit/). The applied structural models (see further description below) are an adaptation of similar models previously developed to analyse SAXS and SANS data from MSP1D1 nanodiscs (Skar-Gislinge et al., 2010; Skar-Gislinge and Arleth, 2011). Briefly, the model describes the nanodiscs as coarse-grained elliptical shapes and is based on analytical form factors (Pedersen, 1997; Skar-Gislinge et al., 2010; Skar-Gislinge and Arleth, 2011). The ellipticity, in terms of the axis ratio of the embedded bilayer patch is allowed to vary in the fit and can also take the size of unity corresponding to a circular disc. The model is fitted on absolute scale and utilizes information on the composition of the protein belt and lipids, and the molecular volumes, v, of the DMPC lipids and the different belts with/without His-tag. These are taken to be νDMPC=1085Å-3, νΔH4H5=20349Å-3, νΔH4H5=24298Å-3, νHis=3142Å-3. The X-ray and neutron scattering lengths of the different components are calculated from their chemical composition.

Apart from the parameters listed in Table 1, the model also fits a small constant background added to the model, and includes a term accounting for interface roughness, fixed to 2 Å in the present analysis, and where relevant, a Gaussian random coil description of the linked TEV-His-tag with RG = 12.7 Å consistent with the assumption that the 23 amino acids of the tag are in a fully disordered state (Kohn et al., 2004). As our measurements are on a calibrated absolute intensity scale, we can compare the observed intensities with those expected from the composition of the sample. Both the SAXS and SANS data had to be re-scaled by a constant close to unity to fit the data (Table 1), but in the case of the ΔH4H5-DMPC SANS data, the scaling constant (1.7 ± 0.5) was larger than expected, most likely the result of a less accurate protein concentration determination for this system.

MD simulations

We initiated our MD simulations from the first model in PDB ID 2N5E (Bibow et al., 2017). A total of 50 pre-equilibrated DMPC lipids (Domański et al., 2010; Dickson et al., 2012) were inserted into each monolayer inside the protein belt. The number of lipids was chosen from the measured optimal reconstitution ratio, and in accordance with the reconstitution ratio used in the experiments for the NMR structure (Bibow et al., 2017) as well as obtained from our fit of the geometric model to the SAXS and SANS data. The MD simulations were performed using GROMACS 5.0.7 (Pronk et al., 2013; Abraham et al., 2015) and the CHARMM36m force field (Huang et al., 2017). The system was solvated in a cubic box and neutralized by addition of Na+ counter ions followed by a minimization of the solvent. Equilibration was performed in six steps following the protocol from CHARMM-GUI (Lee et al., 2016) with slow decrease in the positional restraint forces on both lipids and protein. The volume of the box was then equilibrated in the NPT ensemble at 303.15 K and 1 bar giving a final box with side lengths 13.2 nm. The production run was performed in the NVT ensemble at 303.15 K (above the phase transition of the DMPC lipids) using the stochastic velocity rescaling thermostat (Bussi et al., 2007), 2 fs time steps and the LINCS algorithm to constrain bonds. We performed two production runs (lengths 600 ns and 595 ns) starting from the same equilibrated structure. We concatenated these two MD simulations into a single trajectory, which then represents our sample of the dynamics of the system. We clustered the conformations from the simulations (one structure extracted for every nanosecond) with the RMSD based Quality Threshold method (Heyer et al., 1999; Melvin et al., 2016) using Cα atoms only and with a cluster diameter cutoff of 0.58 nm; this resulted in six clusters. We also performed a 50ns-long simulation of a pure DMPC bilayer. The simulation parameters were the same as for the nanodisc system apart from using the NPT ensemble and anisotropic pressure control.

Calculating SAXS and SANS from simulations

We performed SAXS calculations using both CRYSOL (Svergun et al., 1995) and FOXS (Schneidman-Duhovny et al., 2013; Schneidman-Duhovny et al., 2016) on structures extracted every 1 ns from the simulations and for the q-range from 0.0 Å−1 to 0.25 Å−1. Most of the overall structural information is contained within this q-range, and the calculations of SAXS intensities from the structures are also less accurate in the wide-angle regime. We used standard solution SAXS data recorded at 30°C on the ΔH5-DMPC (without His-TEV-tags) to compare to our simulations, as this setup is most similar to that used to derive the NMR/EPR structure. The SAXS profile of the NMR/EPR structure was calculated by adding DMPC lipids to the first model of the PDB entry and subsequent equilibration of the lipids by MD (fixing the protein), and then using FOXS to back-calculate the SAXS.

Both CRYSOL and FOXS are implicit solvent methods that use fitting parameters to take into account the buffer subtraction and the solvation layer around the solute. The programs automatically optimize these parameters by fitting to experimental data for each input frame, but applying this approach to many frames in a molecular dynamics trajectory could lead to over-fitting. Instead, we calculated the average of each fitted parameter over the trajectory and re-calculate the SAXS with the parameters fixed to this average. FOXS has two parameters, c1 (scaling of atomic radius for adjustment of excluded volume) and c2 (solvation layer adjustment) which, after the fitting, are set to small intervals around the averages [1.01:1.02] and [-0.148:-0.140], respectively. Narrow intervals are used as the program only takes an interval for the parameters. CRYSOL’s fitting parameters dro (Optimal hydration shell contrast), Ra (Optimal atomic group radius) and ExVol (relative background) are set to [0.0090:0.0098], [1.72:1.76] and [162300:162320], respectively. Both CRYSOL and FOXS calculations were performed with hydrogens explicitly included in order to limit artifacts from the excluded volume parameter settings, that is, buffer subtraction, that is suspected to arise from the lipid tails (Chen and Hub, 2015). For CRYSOL the additional settings Maximum order of harmonics was set to 50, the Order of Fibonacci grid to 18 while the Electron density of the solvent was set to 0.334 e3 .

SANS calculations were performed using CRYSON (Svergun et al., 1998) setting the maximum order of harmonics to 50, the order of the Fibonacci grid to 18 and the fraction of D2O in solution to 1.0 in accordance with the experimental measurements. The experimental SANS data were measured on a His-TEV tagged nanodisc. For comparison, we used the simulation frames and added His-TEV tags computationally by extracting conformations from our simulation (w/o His-tags) every 1 ns and attaching a random His-tag structure generated from Flexible Meccano (Ozenne et al., 2012) and Pulchra (Rotkiewicz and Skolnick, 2008) from a pool of 10000 structures to the tails of the nanodisc. If there we detected any clash of the attached His-TEV-tag structure with the protein belt or lipids of the nanodisc or with the second His-TEV-tag, the His-TEV-tag was discarded and a new random structure from the pool was attached. By sampling randomly from a pool of 10.000 His-tag structures together with having in total 1195 frames from the simulation of the nanodisc (1ns per frame) we assume that the His-TEV-tags represents a sufficiently realistic distribution to model the impact on the SANS data.

Comparing simulations to NOEs

We calculated distances corresponding to the experimentally observed NOEs on structures extracted every 1 ns from the simulations. To compare with the experimental distances, available as upper bounds, we averaged the distances, R, between the respective atoms (or the geometric center for pseudo atoms) as R-3-1/3 (Tropp, 1980). When calculating χ2 for validation we only include those distances where this average exceeded the experimentally-determined upper-bounds.

Calculating EPR and PRE data from simulations

We used a previously developed rotamer library for MTSL spin-label probes (Polyhach et al., 2011; Klose et al., 2012) to calculate both EPR and PRE data using the DEER-PREdict software (Tesei et al., 2020; https://github.com/KULL-Centre/DEERpredict). In the case of the EPR DEER data, we calculated the distance distribution of spin-label probes and compared to those estimated from experiments (Bibow et al., 2017). For the NMR data we used a Model Free approach to calculate the PREs (Iwahara et al., 2004) and estimated intensity ratios as previously described (Battiste and Wagner, 2000) using R2,dia=60s1, τc=34ns, τt=1ns and an INEPT delay of 10 ms.

Integrating experiments and simulations

We used a Bayesian/maximum entropy approach (Różycki et al., 2011; Hummer and Köfinger, 2015; Bottaro et al., 2018), as implemented in the BME software (Bottaro et al., 2020) (github.com/KULL-Centre/BME), to integrate the molecular simulations with the SAXS and NMR experiments. The name originates from the two equivalent approaches, Bayesian and Maximum Entropy ensemble refinement, which are equivalent when the errors are modelled as Gaussians (Hummer and Köfinger, 2015; Cesari et al., 2016; Bottaro et al., 2020). We here provide a brief overview of the approach and refer the reader to recent papers for more details (Hummer and Köfinger, 2015; Cesari et al., 2016; Bottaro et al., 2020; Orioli et al., 2020).

Given that our MD simulations provide a good, but non-perfect, agreement with experiments the goal is to find an improved description of the nanodisc that simultaneously satisfies two criteria: (i) the new ensemble should match the data better than the original MD ensemble and (ii) the new ensemble should be a minimal perturbation of that obtained in our simulations with the CHARMM36m force field in accordance with the maximum entropy principle. In a Bayesian formulation, the MD simulation is treated as a prior distribution and we seek a posterior that improves agreement with experiments. This may be achieved by changing the weight, wj, of each conformation in the MD-derived ensemble by minimizing the negative log-likelihood function (Hummer and Köfinger, 2015; Bottaro et al., 2020):

(w1wn)=m2χr2(w1wn)-θSrel(w1wn). (1)

Here, the reduced χr2 quantifies the agreement between the experimental data (FiEXP) and the corresponding ensemble values, (F(𝐱)), calculated from the weighted conformers (𝐱):

χr2(w1wn)=1mim(jnwjFi(𝐱j)-FiEXP)2σi2. (2)

The second term contains the relative entropy, Srel, which measures the deviation between the original ensemble (with initial weights wj0 that are uniform in the case of a standard MD simulation) and the reweighted ensemble Srel=-jnwjlog(wjwj0). The temperature-like parameter θ tunes the balance between fitting the data accurately (low χr2) and not deviating too much from the prior (low Srel). It is a hyperparameter that needs to be determined (Figure 3—figure supplement 3). In practice it turns out that minimizing can be done efficiently by finding Lagrange multipliers in an equivalent Maximum Entropy formalism and we refer the reader to previous papers for a full description and discussion of the approaches including how to determine θ (Hummer and Köfinger, 2015; Cesari et al., 2016; Bottaro et al., 2020). The weights from the BME analysis, the MD simulations as well as the various data that we analyzed are available online at https://github.com/KULL-Centre/papers/tree/master/2020/nanodisc-bengtsen-et-al (copy archived at https://github.com/elifesciences-publications/KULL-CENTREpapers/tree/master/2020/nanodisc-bengtsen-et-alKULL-Centre, 2020).

Acylindricity

In order to quantify how ‘elliptical’ the different nanodisc conformations are, we calculated the square root of the acylindricity, C, where the acylindricity is defined from the principal components of the gyration tensor as C:=λx2-λy2, where the z-axis is orthogonal to the membrane and has the smallest principal component. In our calculations we included only the protein backbone atoms (excluding also the flexible tails from residues 55–63). This choice also makes it possible to compare with a similar calculation from the geometric model fitted from the SAXS and SANS data where the acylindricity was calculated using the major and minor axes from the geometric fit.

Lipid properties

We calculated the bilayer thickness and lipid order parameters for both the nanodisc and a simulated DMPC lipid bilayer. The values obtained for the nanodisc were from the reweighted ensemble every 1 ns. We defined the bilayer thickness as the minimum distance along the bilayer normal between two phosphate headgroup pairs in the two leaflets. The headgroup pairs were identified and saved for each leaflet, top and bottom, along with the corresponding thickness and xy-coordinates. The pairs were further distributed unto a 6 × 6 grid in the xy-plane with each bin corresponding to 22 Å for both the top and bottom leaflet. An averaged grid was then obtained from the two grids of the leaflets. The order parameters SCH where calculated as (Piggot et al., 2017): SCH=123cos2θ-1, where θ is the angle between the C-H bond and the bilayer normal. The order parameters were calculated for each lipid and each carbon along the two lipid tails every 1 ns. The values were further averaged across the two lipid tails before distributed unto a 6 × 6 grid. An average across frames and lipids were then obtained for each bin. In order to study the profile of the lipid tails, an average across frames, lipids, and tails were likewise obtained. Parameters were calculated from the simulations of the DMPC bilayer in the same way.

Acknowledgements

We thank Stefan Bibow, Gunnar Jeschke, Yevhen Polyhach and Roland Riek for discussions and input, and for sharing experimental data, and Jesper Ferkinghoff-Borg for discussions and input to analysis of lipid structures. The research described here was supported by a grant from the Lundbeck Foundation to the BRAINSTRUC structural biology initiative (LA and KL-L), a Hallas-Møller Stipend from the Novo Nordisk Foundation (KL-L), the Velux Foundations (KL-L) and the Danish Council for Independent Research in Technology and Production (BS).

Funding Statement

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Contributor Information

Lise Arleth, Email: arleth@nbi.ku.dk.

Kresten Lindorff-Larsen, Email: lindorff@bio.ku.dk.

Lucie Delemotte, KTH Royal Institute of Technology, Sweden.

José D Faraldo-Gómez, National Heart, Lung and Blood Institute, National Institutes of Health, United States.

Funding Information

This paper was supported by the following grants:

  • Danish Council for Independent Research to Birgit Schiøtt.

  • Lundbeckfonden BRAINSTRUC to Lise Arleth, Kresten Lindorff-Larsen.

  • Novo Nordisk Foundation Hallas-Møller Stipend to Kresten Lindorff-Larsen.

  • Villum Fonden Block grant to Sandro Bottaro, Kresten Lindorff-Larsen.

Additional information

Competing interests

No competing interests declared.

Author contributions

Data curation, Software, Formal analysis, Investigation, Visualization, Methodology, Writing - original draft, Writing - review and editing.

Data curation, Formal analysis, Investigation, Writing - original draft.

Software, Formal analysis, Visualization, Methodology, Writing - review and editing.

Formal analysis, Investigation, Writing - review and editing.

Data curation, Formal analysis, Investigation.

Software, Formal analysis, Methodology.

Software, Formal analysis, Methodology.

Resources, Supervision, Methodology, Writing - review and editing.

Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Methodology, Project administration, Writing - review and editing.

Conceptualization, Resources, Formal analysis, Supervision, Funding acquisition, Investigation, Methodology, Project administration, Writing - review and editing.

Additional files

Transparent reporting form

Data availability

Scattering data, molecular simulations and results from reweighting are available at https://github.com/KULL-Centre/papers/tree/master/2020/nanodisc-bengtsen-et-al (copy archived at https://github.com/elifesciences-publications/KULL-CENTREpapers/tree/master/2020/nanodisc-bengtsen-et-al).

The following previously published dataset was used:

Bibow S, Polyhach Y, Eichmann C, Chi CN, Kowal J, Stahlberg H, Jeschke G, Guentert P, Riek R. 2017. The 3D solution structure of discoidal high-density lipoprotein particles. RCSB Protein Data Bank. 2N5E

References

  1. Abraham MJ, Murtola T, Schulz R, Páll S, Smith JC, Hess B, Lindahl E. Gromacs: High performance molecular simulations through multi-level parallelism from laptops to supercomputers. SoftwareX. 2015;1-2:19–25. doi: 10.1016/j.softx.2015.06.001. [DOI] [Google Scholar]
  2. Battiste JL, Wagner G. Utilization of site-directed spin labeling and high-resolution heteronuclear nuclear magnetic resonance for global fold determination of large proteins with limited nuclear overhauser effect data. Biochemistry. 2000;39:5355–5365. doi: 10.1021/bi000060h. [DOI] [PubMed] [Google Scholar]
  3. Bayburt TH, Grinkova YV, Sligar SG. Self-Assembly of discoidal phospholipid bilayer nanoparticles with membrane scaffold proteins. Nano Letters. 2002;2:853–856. doi: 10.1021/nl025623k. [DOI] [Google Scholar]
  4. Bibow S, Polyhach Y, Eichmann C, Chi CN, Kowal J, Albiez S, McLeod RA, Stahlberg H, Jeschke G, Güntert P, Riek R. Solution structure of discoidal high-density lipoprotein particles with a shortened apolipoprotein A-I. Nature Structural & Molecular Biology. 2017;24:187–193. doi: 10.1038/nsmb.3345. [DOI] [PubMed] [Google Scholar]
  5. Blanchet CE, Spilotros A, Schwemmer F, Graewert MA, Kikhney A, Jeffries CM, Franke D, Mark D, Zengerle R, Cipriani F, Fiedler S, Roessle M, Svergun DI. Versatile sample environments and automation for biological solution X-ray scattering experiments at the P12 beamline (PETRA III, DESY) Journal of Applied Crystallography. 2015;48:431–443. doi: 10.1107/S160057671500254X. [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Bottaro S, Bussi G, Kennedy SD, Turner DH, Lindorff-Larsen K. Conformational ensembles of RNA oligonucleotides from integrating NMR and molecular simulations. Science Advances. 2018;4:eaar8521. doi: 10.1126/sciadv.aar8521. [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Bottaro S, Bengtsen T, Lindorff-Larsen K. Integrating molecular simulation and experimental data: a Bayesian/maximum entropy reweighting approach. In: Gáspári Z, editor. Structural Bioinformatics. Clifton, NJ: Springer-Nature; 2020. pp. 219–240. [DOI] [PubMed] [Google Scholar]
  8. Bottaro S, Lindorff-Larsen K. Biophysical experiments and biomolecular simulations: a perfect match? Science. 2018;361:355–360. doi: 10.1126/science.aat4010. [DOI] [PubMed] [Google Scholar]
  9. Bussi G, Donadio D, Parrinello M. Canonical sampling through velocity rescaling. The Journal of Chemical Physics. 2007;126:014101. doi: 10.1063/1.2408420. [DOI] [PubMed] [Google Scholar]
  10. Caponetti E, Floriano MA, Di Dio E, Triolo R. On the shape of the radial distribution function of an assembly of monodisperse ellipsoidal scatterers. Journal of Applied Crystallography. 1993;26:612–615. doi: 10.1107/S0021889893001335. [DOI] [Google Scholar]
  11. Cesari A, Gil-Ley A, Bussi G. Combining simulations and solution experiments as a paradigm for RNA force field refinement. Journal of Chemical Theory and Computation. 2016;12:6192–6200. doi: 10.1021/acs.jctc.6b00944. [DOI] [PubMed] [Google Scholar]
  12. Chen PC, Hub JS. Structural properties of Protein-Detergent complexes from SAXS and MD simulations. The Journal of Physical Chemistry Letters. 2015;6:5116–5121. doi: 10.1021/acs.jpclett.5b02399. [DOI] [PubMed] [Google Scholar]
  13. Debnath A, Schäfer LV. Structure and dynamics of phospholipid nanodiscs from All-Atom and Coarse-Grained simulations. The Journal of Physical Chemistry B. 2015;119:6991–7002. doi: 10.1021/acs.jpcb.5b02101. [DOI] [PubMed] [Google Scholar]
  14. Denisov IG, Grinkova YV, Lazarides AA, Sligar SG. Directed self-assembly of monodisperse phospholipid bilayer nanodiscs with controlled size. Journal of the American Chemical Society. 2004;126:3477–3487. doi: 10.1021/ja0393574. [DOI] [PubMed] [Google Scholar]
  15. Denisov IG, McLean MA, Shaw AW, Grinkova YV, Sligar SG. Thermotropic phase transition in soluble nanoscale lipid bilayers. The Journal of Physical Chemistry B. 2005;109:15580–15588. doi: 10.1021/jp051385g. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Dickson CJ, Rosso L, Betz RM, Walker RC, Gould IR. GAFFlipid: a general amber force field for the accurate molecular dynamics simulation of phospholipid. Soft Matter. 2012;8:9617. doi: 10.1039/c2sm26007g. [DOI] [Google Scholar]
  17. Domański J, Stansfeld PJ, Sansom MS, Beckstein O. Lipidbook: a public repository for force-field parameters used in membrane simulations. The Journal of Membrane Biology. 2010;236:255–258. doi: 10.1007/s00232-010-9296-8. [DOI] [PubMed] [Google Scholar]
  18. Frauenfeld J, Gumbart J, Sluis EO, Funes S, Gartmann M, Beatrix B, Mielke T, Berninghausen O, Becker T, Schulten K, Beckmann R. Cryo-EM structure of the ribosome-SecYE complex in the membrane environment. Nature Structural & Molecular Biology. 2011;18:614–621. doi: 10.1038/nsmb.2026. [DOI] [PMC free article] [PubMed] [Google Scholar]
  19. Graziano V, Miller L, Yang L. Interpretation of solution scattering data from lipid nanodiscs. Journal of Applied Crystallography. 2018;51:157–166. doi: 10.1107/S1600576717018441. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hagn F, Etzkorn M, Raschle T, Wagner G. Optimized phospholipid bilayer nanodiscs facilitate high-resolution structure determination of membrane proteins. Journal of the American Chemical Society. 2013;135:1919–1925. doi: 10.1021/ja310901f. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Hansen S. Update for bayesapp: a web site for analysis of small-angle scattering data. Journal of Applied Crystallography. 2014;47:1469–1471. doi: 10.1107/S160057671401315. [DOI] [Google Scholar]
  22. Heyer LJ, Kruglyak S, Yooseph S. Exploring expression data: identification and analysis of coexpressed genes. Genome Research. 1999;9:1106–1115. doi: 10.1101/gr.9.11.1106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Huang J, Rauscher S, Nawrocki G, Ran T, Feig M, de Groot BL, Grubmüller H, MacKerell AD. CHARMM36m: an improved force field for folded and intrinsically disordered proteins. Nature Methods. 2017;14:71–73. doi: 10.1038/nmeth.4067. [DOI] [PMC free article] [PubMed] [Google Scholar]
  24. Hummer G, Köfinger J. Bayesian ensemble refinement by replica simulations and reweighting. The Journal of Chemical Physics. 2015;143:243150. doi: 10.1063/1.4937786. [DOI] [PubMed] [Google Scholar]
  25. Iwahara J, Schwieters CD, Clore GM. Ensemble approach for NMR structure refinement against (1)H paramagnetic relaxation enhancement data arising from a flexible paramagnetic group attached to a macromolecule. Journal of the American Chemical Society. 2004;126:5879–5896. doi: 10.1021/ja031580d. [DOI] [PubMed] [Google Scholar]
  26. Johansen NT, Pedersen MC, Porcar L, Martel A, Arleth L. Introducing SEC-SANS for studies of complex self-organized biological systems. Acta Crystallographica Section D Structural Biology. 2018;74:1178–1191. doi: 10.1107/S2059798318007180. [DOI] [PubMed] [Google Scholar]
  27. Jordan A, Jacques M, Merrick C, Devos J, Forsyth VT, Porcar L, Martel A. SEC-SANS: size exclusion chromatography combined in situ with small-angle neutron scattering. Journal of Applied Crystallography. 2016;49:2015–2020. doi: 10.1107/S1600576716016514. [DOI] [PMC free article] [PubMed] [Google Scholar]
  28. Klose D, Klare JP, Grohmann D, Kay CW, Werner F, Steinhoff HJ. Simulation vs. reality: a comparison of in silico distance predictions with DEER and FRET measurements. PLOS ONE. 2012;7:e39492. doi: 10.1371/journal.pone.0039492. [DOI] [PMC free article] [PubMed] [Google Scholar]
  29. Kohn JE, Millett IS, Jacob J, Zagrovic B, Dillon TM, Cingel N, Dothager RS, Seifert S, Thiyagarajan P, Sosnick TR, Hasan MZ, Pande VS, Ruczinski I, Doniach S, Plaxco KW. Random-coil behavior and the dimensions of chemically unfolded proteins. PNAS. 2004;101:12491–12496. doi: 10.1073/pnas.0403643101. [DOI] [PMC free article] [PubMed] [Google Scholar]
  30. KULL-Centre Nanodisc-Bengtsen-Et-Al/ e0c89feGitHub. 2020 https://github.com/KULL-Centre/papers/tree/master/2020/nanodisc-bengtsen-et-al
  31. Lee J, Cheng X, Swails JM, Yeom MS, Eastman PK, Lemkul JA, Wei S, Buckner J, Jeong JC, Qi Y, Jo S, Pande VS, Case DA, Brooks CL, MacKerell AD, Klauda JB, Im W. CHARMM-GUI input generator for NAMD, GROMACS, AMBER, OpenMM, and CHARMM/OpenMM simulations using the CHARMM36 additive force field. Journal of Chemical Theory and Computation. 2016;12:405–413. doi: 10.1021/acs.jctc.5b00935. [DOI] [PMC free article] [PubMed] [Google Scholar]
  32. Maric S, Skar-Gislinge N, Midtgaard S, Thygesen MB, Schiller J, Frielinghaus H, Moulin M, Haertlein M, Forsyth VT, Pomorski TG, Arleth L. Stealth carriers for low-resolution structure determination of membrane proteins in solution. Acta Crystallographica Section D Biological Crystallography. 2014;70:317–328. doi: 10.1107/S1399004713027466. [DOI] [PubMed] [Google Scholar]
  33. Martinez D, Decossas M, Kowal J, Frey L, Stahlberg H, Dufourc EJ, Riek R, Habenstein B, Bibow S, Loquet A. Lipid internal dynamics probed in nanodiscs. ChemPhysChem. 2017;18:2651–2657. doi: 10.1002/cphc.201700450. [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Marty MT, Zhang H, Cui W, Gross ML, Sligar SG. Interpretation and deconvolution of nanodisc native mass spectra. Journal of the American Society for Mass Spectrometry. 2014;25:269–277. doi: 10.1007/s13361-013-0782-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
  35. Marty MT, Baldwin AJ, Marklund EG, Hochberg GK, Benesch JL, Robinson CV. Bayesian deconvolution of mass and ion mobility spectra: from binary interactions to polydisperse ensembles. Analytical Chemistry. 2015;87:4370–4376. doi: 10.1021/acs.analchem.5b00140. [DOI] [PMC free article] [PubMed] [Google Scholar]
  36. Melvin RL, Gmeiner WH, Salsbury FR. All-Atom molecular dynamics reveals mechanism of zinc complexation with therapeutic F10. The Journal of Physical Chemistry B. 2016;120:10269–10279. doi: 10.1021/acs.jpcb.6b07753. [DOI] [PMC free article] [PubMed] [Google Scholar]
  37. Midtgaard SR, Pedersen MC, Kirkensgaard JJ, Sørensen KK, Mortensen K, Jensen KJ, Arleth L. Self-assembling peptides form nanodiscs that stabilize membrane proteins. Soft Matter. 2014;10:738–752. doi: 10.1039/C3SM51727F. [DOI] [PubMed] [Google Scholar]
  38. Midtgaard SR, Pedersen MC, Arleth L. Small-angle X-ray scattering of the cholesterol incorporation into human ApoA1-POPC discoidal particles. Biophysical Journal. 2015;109:308–318. doi: 10.1016/j.bpj.2015.06.032. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Mörs K, Roos C, Scholz F, Wachtveitl J, Dötsch V, Bernhard F, Glaubitz C. Modified lipid and protein dynamics in nanodiscs. Biochimica Et Biophysica Acta (BBA) - Biomembranes. 2013;1828:1222–1229. doi: 10.1016/j.bbamem.2012.12.011. [DOI] [PubMed] [Google Scholar]
  40. Orioli S, Larsen AH, Bottaro S, Lindorff-Larsen K. How to learn from inconsistencies: integrating molecular simulations with experimental data. Progress in Molecular Biology and Translational Science. 2020;170:123–176. doi: 10.1016/bs.pmbts.2019.12.006. [DOI] [PubMed] [Google Scholar]
  41. Orthaber D, Bergmann A, Glatter O. SAXS experiments on absolute scale with Kratky systems using water as a secondary standard. Journal of Applied Crystallography. 2000;33:218–225. doi: 10.1107/S0021889899015216. [DOI] [Google Scholar]
  42. Ozenne V, Bauer F, Salmon L, Huang JR, Jensen MR, Segard S, Bernadó P, Charavay C, Blackledge M. Flexible-meccano: a tool for the generation of explicit ensemble descriptions of intrinsically disordered proteins and their associated experimental observables. Bioinformatics. 2012;28:1463–1470. doi: 10.1093/bioinformatics/bts172. [DOI] [PubMed] [Google Scholar]
  43. Pedersen JS. Analysis of small-angle scattering data from colloids and polymer solutions: modeling and least-squares fitting. Advances in Colloid and Interface Science. 1997;70:171–210. doi: 10.1016/S0001-8686(97)00312-6. [DOI] [Google Scholar]
  44. Pedersen MC, Arleth L, Mortensen K. WillItFit : a framework for fitting of constrained models to small-angle scattering data. Journal of Applied Crystallography. 2013;46:1894–1898. doi: 10.1107/S0021889813026022. [DOI] [Google Scholar]
  45. Pernot P, Round A, Barrett R, De Maria Antolinos A, Gobbo A, Gordon E, Huet J, Kieffer J, Lentini M, Mattenet M, Morawe C, Mueller-Dieckmann C, Ohlsson S, Schmid W, Surr J, Theveneau P, Zerrad L, McSweeney S. Upgraded ESRF BM29 beamline for SAXS on macromolecules in solution. Journal of Synchrotron Radiation. 2013;20:660–664. doi: 10.1107/S0909049513010431. [DOI] [PMC free article] [PubMed] [Google Scholar]
  46. Piggot TJ, Allison JR, Sessions RB, Essex JW. On the calculation of acyl chain order parameters from lipid simulations. Journal of Chemical Theory and Computation. 2017;13:5683–5696. doi: 10.1021/acs.jctc.7b00643. [DOI] [PubMed] [Google Scholar]
  47. Polyhach Y, Bordignon E, Jeschke G. Rotamer libraries of spin labelled cysteines for protein studies. Physical Chemistry Chemical Physics. 2011;13:2356–2366. doi: 10.1039/C0CP01865A. [DOI] [PubMed] [Google Scholar]
  48. Pronk S, Páll S, Schulz R, Larsson P, Bjelkmar P, Apostolov R, Shirts MR, Smith JC, Kasson PM, van der Spoel D, Hess B, Lindahl E. GROMACS 4.5: a high-throughput and highly parallel open source molecular simulation toolkit. Bioinformatics. 2013;29:845–854. doi: 10.1093/bioinformatics/btt055. [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Ritchie TK, Grinkova YV, Bayburt TH, Denisov IG, Zolnerciks JK, Atkins WM, Sligar SG. Chapter 11 - Reconstitution of membrane proteins in phospholipid bilayer nanodiscs. Methods in Enzymology. 2009;464:211–231. doi: 10.1016/S0076-6879(09)64011-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Rotkiewicz P, Skolnick J. Fast procedure for reconstruction of full-atom protein models from reduced representations. Journal of Computational Chemistry. 2008;29:1460–1465. doi: 10.1002/jcc.20906. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Różycki B, Kim YC, Hummer G. SAXS ensemble refinement of ESCRT-III CHMP3 conformational transitions. Structure. 2011;19:109–116. doi: 10.1016/j.str.2010.10.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Schneidman-Duhovny D, Hammel M, Tainer JA, Sali A. Accurate SAXS profile computation and its assessment by contrast variation experiments. Biophysical Journal. 2013;105:962–974. doi: 10.1016/j.bpj.2013.07.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Schneidman-Duhovny D, Hammel M, Tainer JA, Sali A. FoXS, FoXSDock and MultiFoXS: single-state and multi-state structural modeling of proteins and their complexes based on SAXS profiles. Nucleic Acids Research. 2016;44:W424–W429. doi: 10.1093/nar/gkw389. [DOI] [PMC free article] [PubMed] [Google Scholar]
  54. Shaw AW, McLean MA, Sligar SG. Phospholipid phase transitions in homogeneous nanometer scale bilayer discs. FEBS Letters. 2004;556:260–264. doi: 10.1016/S0014-5793(03)01400-5. [DOI] [PubMed] [Google Scholar]
  55. Shih AY, Denisov IG, Phillips JC, Sligar SG, Schulten K. Molecular dynamics simulations of discoidal bilayers assembled from truncated human lipoproteins. Biophysical Journal. 2005;88:548–556. doi: 10.1529/biophysj.104.046896. [DOI] [PMC free article] [PubMed] [Google Scholar]
  56. Shih AY, Arkhipov A, Freddolino PL, Sligar SG, Schulten K. Assembly of lipids and proteins into lipoprotein particles. The Journal of Physical Chemistry B. 2007;111:11095–11104. doi: 10.1021/jp072320b. [DOI] [PubMed] [Google Scholar]
  57. Siuda I, Tieleman DP. Molecular models of nanodiscs. Journal of Chemical Theory and Computation. 2015;11:4923–4932. doi: 10.1021/acs.jctc.5b00668. [DOI] [PubMed] [Google Scholar]
  58. Skar-Gislinge N, Simonsen JB, Mortensen K, Feidenhans'l R, Sligar SG, Lindberg Møller B, Bjørnholm T, Arleth L. Elliptical structure of phospholipid bilayer nanodiscs encapsulated by scaffold proteins: casting the roles of the lipids and the protein. Journal of the American Chemical Society. 2010;132:13713–13722. doi: 10.1021/ja1030613. [DOI] [PMC free article] [PubMed] [Google Scholar]
  59. Skar-Gislinge N, Johansen NT, Høiberg-Nielsen R, Arleth L. Comprehensive study of the Self-Assembly of phospholipid nanodiscs: what determines their shape and stoichiometry? Langmuir. 2018;34:12569–12582. doi: 10.1021/acs.langmuir.8b01503. [DOI] [PubMed] [Google Scholar]
  60. Skar-Gislinge N, Arleth L. Small-angle scattering from phospholipid nanodiscs: derivation and refinement of a molecular constrained analytical model form factor. Phys. Chem. Chem. Phys. 2011;13:3161–3170. doi: 10.1039/C0CP01074J. [DOI] [PubMed] [Google Scholar]
  61. Svergun D, Barberato C, Koch MHJ. CRYSOL – a program to evaluate x-ray solution scattering of biological macromolecules from atomic coordinates. Journal of Applied Crystallography. 1995;28:768–773. doi: 10.1107/S0021889895007047. [DOI] [Google Scholar]
  62. Svergun DI, Richard S, Koch MHJ, Sayers Z, Kuprin S, Zaccai G. Protein hydration in solution: experimental observation by x-ray and neutron scattering. PNAS. 1998;95:2267–2272. doi: 10.1073/pnas.95.5.2267. [DOI] [PMC free article] [PubMed] [Google Scholar]
  63. Tesei G, Martins JM, Kunze MBA, Wang Y, Crehuet R, Lindorff-Larsen K. DEER-PREdict: software for efficient calculation of Spin-Labeling EPR and NMR data from conformational ensembles. bioRxiv. 2020 doi: 10.1101/2020.08.09.243030. [DOI] [PMC free article] [PubMed]
  64. Tristram-Nagle S, Liu Y, Legleiter J, Nagle JF. Structure of gel phase DMPC determined by X-ray diffraction. Biophysical Journal. 2002;83:3324–3335. doi: 10.1016/S0006-3495(02)75333-2. [DOI] [PMC free article] [PubMed] [Google Scholar]
  65. Tropp J. Dipolar relaxation and nuclear Overhauser effects in nonrigid molecules: the effect of fluctuating internuclear distances. The Journal of Chemical Physics. 1980;72:6035–6043. doi: 10.1063/1.439059. [DOI] [Google Scholar]
  66. Vestergaard M, Kraft JF, Vosegaard T, Thøgersen L, Schiøtt B. Bicelles and other membrane mimics: comparison of structure, properties, and dynamics from MD simulations. The Journal of Physical Chemistry B. 2015;119:15831–15843. doi: 10.1021/acs.jpcb.5b08463. [DOI] [PubMed] [Google Scholar]

Decision letter

Editor: Lucie Delemotte1
Reviewed by: Lucie Delemotte2, Franz Hagn3

In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.

Acceptance summary:

Computer simulations are increasingly used in structural biology as a means to integrate different kinds of experimental biophysical data, so as to facilitate a coherent and physically realistic molecular-level interpretation. This paper reports a structural refinement strategy of a previously determined NMR structure of a lipid nanodisc using orthogonal scattering data and advanced MD simulations. The elegant integrative simulation/experimental Ansatz leads to the insightful conclusion that the nanodisc undergoes elliptical structural fluctuations. Such elliptic fluctuations are "averaged out" in the NMR signal, but not in the scattering experiments data, thus resolving the apparent discrepancy between the different experimental methods.

Decision letter after peer review:

Thank you for submitting your article "Structure and dynamics of a nanodisc by integrating NMR, SAXS and SANS experiments with molecular dynamics simulations" for consideration by eLife. Your article has been reviewed by four peer reviewers, including Lucie Delemotte as the Reviewing Editor and Reviewer #1, and the evaluation has been overseen by José Faraldo-Gómez as the Senior Editor. The following individual involved in review of your submission has agreed to reveal their identity: Franz Hagn (Reviewer #3).

The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.

We would like to draw your attention to changes in our revision policy that we have made in response to COVID-19 (https://elifesciences.org/articles/57162). Specifically, we are asking editors to accept without delay manuscripts, like yours, that they judge can stand as eLife papers without additional data, even if they feel that they would make the manuscript stronger. Thus the revisions requested below only address clarity and presentation.

Summary:

The present manuscript builds up on previous work by the same authors (e.g. Skar-Gislinge, 2010, Midtgaard, 2015) towards the goal of understanding the structure and dynamics of MSP1D1-ΔH5 and MSP1D1-ΔH4H5 nanodiscs and their membrane mimicking capabilities. In particular, small angle X-ray and neutron scattering experiments are performed and used together with previously determined NMR distance restraints (Bibow et al., 2017)) and molecular dynamics simulations to obtain additional information on the structural properties of a lipid nanodisc particle in solution.

In essence, this paper reports on a structural refinement strategy of a previously determined NMR structure of a lipid nanodisc using orthogonal scattering data and advanced MD simulations. According to a procedure established by the Lindorff-Larsen lab, the conformational clusters obtained in the simulations are re-weighted to obtain a better fit to the experimental data.

A key question addressed by the authors is whether the NDs are circular, as suggested by the previous NMR experiments, or slightly elliptical, as suggested by the scattering experiments carried out by the authors. The elegant integrative simulation/experimental Ansatz leads to the insightful conclusion that the ND undergoes elliptical structural fluctuations, with the major axes changing direction and being largely orthogonal to each other. Such elliptic fluctuations are "averaged out" in the NOEs from NMR, but not in the SAXS data, thus resolving the apparent discrepancy between the different experimental methods.

Essential revisions:

1) The reviewers agreed unanimously that the work was methodologically rigorous, technically well done, and the scientific conclusions fully supported by the data. They also agreed that this manuscript was clearly written, and that the presentation of the data was thorough and well organized. However, it was not entirely clear what aspects of this manuscript were novel. SAXS and SANS on MSP1D1 nanodiscs, lipid melting transition temperature dependence probed by SAXS on MSP1D1 and MSP1E3D1, analysis of lipid properties in MSP1 nanodiscs, broadening of gel-liquid phase transition in MSP1D1-ΔH5, size-dependent increase of the transition enthalpy in MSP1D1 and MSP1E3D1 had been reported beforehand. The MD methodology as well as SAXS/SANS model building and fitting has been previously published by the authors.

It thus appeared to the reviewers that the novel aspect of this paper was the combination of various structural data and simulations to obtain better structures of challenging systems. Thus, while this paper provided a valuable protocol for dealing with structural data of various sources, this was not entirely clear from the paper in the present form.

Additionally the relevance of an oval nanodisc shape for their membrane mimicking properties was not entirely clear, since the shape might change in presence of a membrane protein. It was also unclear whether the conclusions around ellipticity could be generalized for all nanodiscs.

The reviewers thus recommended that the paper be rewritten to reflect the novel aspects of the work and to highlight its strengths while clarifying which aspects were already tackled by previous work.

2) In addition to the circular NMR structure from Bibow et al., 2017, a circular nanodisc structure was solved in a cryo-electron microscopy study by Frauenfeld et al., 2011. How would the authors compare their results to this single particle cryo-electron microscopy structure? It would be interesting to use also these restraints into the MD simulation. Could the oval shape of an empty nanodisc also be seen in (cryo-)electron microscopy single particle data (before class averaging)?

3) The work by Bibow et al. also includes PRE distance restraints and EPR distance restraints (and for the latter also a size distribution) which are of medium range and thus on almost a similar level of information as the SAXS and SANS data. These restraints should also be used in the reweighing procedure and discussed.

4) The authors show the good fit between SAXS and SANS data towards an elliptical model. The authors should illustrate how a circular model would fit with the data as well as different elliptical models to access the request for an elliptical model.

5) The major problem of any bulk studies is the potential presence of a heterogenous sample likely to be the case for nanodiscs because they may contain a distinct number of lipids. While the authors state that a single SEC peak is present that of course is not sufficient evidence for a single entity. First, this potential issue must be stated. Second, it is suggested to calculate the SAXS and SANS curves for a heterogenous sample with several cyclic nanodiscs having a distinct radius. Third, some argumentation may be put forward that indicate that the heterogeneity is of dynamic nature (such as a single set of NMR peaks, the dynamics of the MD simulation).

6) What do the authors mean with "membrane mimicking capabilities"? Just physical properties of lipids in nanodiscs as compared to liposomes?

7) It is not clearly described how SAXS and SANS data have been scaled for data fitting. Is it just a hardware-dependent correction or does it have an underlying physical meaning? This requires a short explanation.

8) For comparing the NMR structure with the simulation results, have the authors considered the entire NMR structural bundle (10 structures) or just a single structure? This might have a marked impact on the chi-square value for the NMR structure.

9) Correlation of MD with amide NOES (subsection “Integrating experiments and simulations”): The authors might want to add here that the amide-amide NOEs report on the α-helical secondary structure of the individual MSPs and are not quite sensitive to slight changes in the shape of the overall MSP belt. Thus, a good correlation with these restraints is very much expected.

10) Subsection “Integrating experiments and simulations”, seventh paragraph: it seems like the square root(C) value for the NMR/EPR structure is missing.

11) Order parameters (Figure 4B): What C-H moieties in DMPC have been used to calculate the numbers plotted in the figure? Also, it seems counter-intuitive that lipids in the rim region are more mobile than lipids in the core.

12) It is very good that the authors make their MD data publicly available on GitHub. It would be nice if the refined ensemble (or some number of representative frames) could be made available to the scientific community also via a database, such as PDB-dev.

13) The clustering step of the MD simulations seems quite important for the rest of the paper, and the choice of methods and parameters is not rationalized. Were other methods/parameters tested and why did the authors settle on this specific choice?

14) Additionally, the reviewers had the following suggestion, but did not strictly require additional experiments to be performed: Have the authors tried to perform SANS contrast matching experiments to obtain scattering data of the protein belt and the lipid patch only, respectively? Since the NMR structure does not contain lipids, such experiments would be helpful to validate the "lipid-free" structure.

eLife. 2020 Jul 30;9:e56518. doi: 10.7554/eLife.56518.sa2

Author response


Essential revisions:

1) The reviewers agreed unanimously that the work was methodologically rigorous, technically well done, and the scientific conclusions fully supported by the data. They also agreed that this manuscript was clearly written, and that the presentation of the data was thorough and well organized. However, it was not entirely clear what aspects of this manuscript were novel. SAXS and SANS on MSP1D1 nanodiscs, lipid melting transition temperature dependence probed by SAXS on MSP1D1 and MSP1E3D1, analysis of lipid properties in MSP1 nanodiscs, broadening of gel-liquid phase transition in MSP1D1-ΔH5, size-dependent increase of the transition enthalpy in MSP1D1 and MSP1E3D1 had been reported beforehand. The MD methodology as well as SAXS/SANS model building and fitting has been previously published by the authors.

It thus appeared to the reviewers that the novel aspect of this paper was the combination of various structural data and simulations to obtain better structures of challenging systems. Thus, while this paper provided a valuable protocol for dealing with structural data of various sources, this was not entirely clear from the paper in the present form.

Additionally the relevance of an oval nanodisc shape for their membrane mimicking properties was not entirely clear, since the shape might change in presence of a membrane protein. It was also unclear whether the conclusions around ellipticity could be generalized for all nanodiscs.

The reviewers thus recommended that the paper be rewritten to reflect the novel aspects of the work and to highlight its strengths while clarifying which aspects were already tackled by previous work.

We have shortened and revised the manuscript in an attempt to clarify these matters, focusing both on what is new, and how the information might be useful in a wide context:

We and others have previously examined nanodiscs in solution using SAXS and SANS measurements, and used these to obtain information about the overall shape. Given the low resolution of small-angle scattering experiments, and the complexity of the system, this is clearly an underdetermined problem. Our earlier work thus made two simplifying assumptions that we go beyond in the current work. In particular, in our earlier work we (i) fitted the scattering data using a coarse-grained/continuum models of nanodiscs, and (ii) optimized the parameters in this to find a single “average” representation that explained the data.

In the current work we went beyond these assumptions, by (i) adding additional experimental information from (previously published) NMR data, (ii) using all-atom MD simulations to increase the resolution and making it possible to link the structure/shape to both SAXS/SANS and NMR and (iii) used our Bayesian/Maximum Entropy approach to integrate experiments and simulations. Together, this made it possible to provide a “dynamic ensemble” of this MSPΔH5 nanodisc in solution.

As the previous NMR experiments were performed on DMPC-loaded MSPΔH5 nanodiscs, we measured new SAXS/SANS data on the same system. We also measured new scattering data on MSPΔH4ΔH5 discs, as well as calorimetry data on MSPD1 to examine effects of changing the size.

It is correct that similar observations about the melting transitions have been made before, and we have attempted to make this clearer. We, however, use our experimentally-derived structural model to interpret these observations. In this way, we aim to bridge our computational/experimental view of the protein-lipid system to the thermodynamic properties of the discs. As our work is mostly focused on the MSPΔH5 discs, this is the system for which we can make the strongest conclusions. We, however, speculate that some of the observations will also generalize to other (larger) discs, appropriately considering the changed ratio of “rim” vs. “internal” lipids in discs of different sizes. We have integrated the sections discussing lipid properties of the MSPΔH5 discs with our DSC results to make the relationships clearer.

Finally, it is correct that the shape and properties may indeed be affected by embedding a membrane protein. Indeed, this was one of the motivations for this work – to provide a baseline and overall approach for interpreting experimental data on even more complex systems. While the current work goes some way towards this goal, it remains a complex problem.

2) In addition to the circular NMR structure from Bibow et al., 2017, a circular nanodisc structure was solved in a cryo-electron microscopy study by Frauenfeld et al., 2011. How would the authors compare their results to this single particle cryo-electron microscopy structure? It would be interesting to use also these restraints into the MD simulation. Could the oval shape of an empty nanodisc also be seen in (cryo-)electron microscopy single particle data (before class averaging)?

High-resolution, single-particle cryo-electron microscopy data would indeed be very useful as an independent source of information about the shape.

We cannot compare our simulations directly to the refined structure from Frauenfeld as it (i) is a different disc where both the nanodisc belt and the lipids differ from those of our study (ii) contains a large transmembrane protein region and (iii) is the result of substantial averaging over individual particles. Also, depending on how well-defined the location of the membrane protein is in the disc, class-averaging could lead to averaging over multiple quite differently shaped discs, as particle classification is likely driven by the more asymmetric membrane protein.

Bibow et al. presented negative-stain transmission electron microscopy images of their MSPΔH5 discs. While these are of low resolution, they do appear to show different shapes including some that appear substantially elliptical. However, because they are negative-stain images, we have decided not to compare to them directly. Instead, it would be interesting in the future to compare simulations to individual particles in higher-resolution electron microscopy data.

We note, however, that caution should be exerted to control for possible changes during the cooling process, since the dynamical fluctuations and changes in lipid properties might occur on faster time scales, making it unclear to which extent images under cryogenic conditions would reflect properties in solution.

These issues are now discussed briefly in the revised manuscript.

3) The work by Bibow et al. also includes PRE distance restraints and EPR distance restraints (and for the latter also a size distribution) which are of medium range and thus on almost a similar level of information as the SAXS and SANS data. These restraints should also be used in the reweighing procedure and discussed.

We have now compared our simulation ensembles to the PRE and EPR/DEER spin-label data resulting in two new figures (Figure 2—figure supplements 5 and 6). We opted to use the data for validation rather than refinement. Thus, we used a spin-label rotamer library to calculate both the PREs and the distance-distribution from both the unweighted MD simulations, and the ensemble that we reweighted against the scattering and NMR data. We also compared the results to those calculated from the published structure, which used this data as part of the refinement procedure. Overall, we find that (i) the MD simulations are in good agreement with the PRE and EPR data, (ii) the agreement improves after refining the ensemble against the NOE and SAXS data and (iii) the agreement is comparable to that of the structures refined using the PRE and EPR (and other) data.

4) The authors show the good fit between SAXS and SANS data towards an elliptical model. The authors should illustrate how a circular model would fit with the data as well as different elliptical models to access the request for an elliptical model.

We explore this question in two ways.

First, as described in the manuscript, we compare both our (heterogenous) MD simulations and the (more “round”) NMR structure to the SAXS data and show that neither are in perfect agreement with the scattering data. When refining the MD ensemble, which displays a distribution of elliptical shapes, against the SAXS data we increase the population of less-extremely elliptical shapes, so that the final distribution is covers the range between the two “static” structures from NMR and the coarse-grained/continuum model.

Second, we have repeated the fitting of the SAXS data of the ΔH5-ΔHis sample at 30°C using the (static) coarse-grained/continuum model scanning a range of ellipticities (axis ratios 1.0 – 1.6 of the embedded bilayer patch). In this analysis we take into account that the model parameters are correlated, so that the changed axis ratios in part can be compensated by variation in some of the remaining model parameters. The results suggest strongly that a better fit is obtained when the axis-ratios are distinctly different from one. As seen from the new Figure 1—figure supplement 3 the best fits are obtained for an axis-ratio close to 1.3 as reported in Table 1 of the manuscript. For lower values, the model has too pronounced oscillations around q=0.1 Å-1 – 0.2 Å-1, while for larger values, the calculated values become too “smeared out” and compared to the experimentally observed features.

5) The major problem of any bulk studies is the potential presence of a heterogenous sample likely to be the case for nanodiscs because they may contain a distinct number of lipids. While the authors state that a single SEC peak is present that of course is not sufficient evidence for a single entity. First, this potential issue must be stated. Second, it is suggested to calculate the SAXS and SANS curves for a heterogenous sample with several cyclic nanodiscs having a distinct radius. Third, some argumentation may be put forward that indicate that the heterogeneity is of dynamic nature (such as a single set of NMR peaks, the dynamics of the MD simulation).

It is correct that there may be some remaining heterogeneity even after SEC-SAXS. The study by Bibow et al. found just a single set of peaks in their NMR spectra, suggesting (i) that the two belt proteins are in an identical (average) environment, and (ii) that under those conditions the sample was relatively homogenous.

It is known that it is generally not possible to separate dispersity in size or shape from SAS data alone, as e.g. previously discussed for ellipsoidal particles (Caponetti et al., 1993). That study shows that although the scattering from ellipsoidal particles can be represented by a polydisperse distribution of spheres, the required distribution is very peculiar and unlikely.

As we have previously discussed for the slightly larger MSP1D1 discs (Skar-Gislinge et al., 2010) the nanodiscs are atypical colloid particles in the respect that their circumference is defined by the mostly fixed length of the rather unelastic and chemically well-defined MSP belts. This has the consequence that less than maximal lipid loading will be manifested in symmetry broken discs, where an elliptical shape of the lipid bilayer patch is most simple, rather than in discs of a smaller radius. It also has the consequence that a dispersity in the number of lipids embedded in the nanodiscs for a given experimental sample will, to first order, be manifested in a sample of elliptical discs of varying axis ratios. Hence, assuming circular discs of different radii is not compatible with the knowledge we have about the nanodisc system.

To investigate the consequences of a variation in the number of embedded lipids, we implemented a model for the DMPC loaded ΔH5-ΔHis discs that included a normally distributed dispersity of the number of embedded lipids (mean value of Nlip=104), where the width of the Gaussian was defined by its relative standard deviation in the number of embedded lipids, σlip, and truncated the Gaussian at +/-3 σlip. An upper hard limit for Nlip in the distribution was furthermore defined by the Nlip that yielded circular and hence fully loaded discs. We chose a lower hard limit to be given by the Nlip that yielded discs of axis ratios ≥ 2. We evaluated this model using the parameters optimised for the SAXS data of the ΔH5-ΔHis sample at 30°C (Table 1, fourth column) using values of σlip of 0.01, 0.03 and 0.05 to represent increasing levels of dispersity, and used σlip = 0.0001 to represent the monodisperse case (Figure 1—figure supplement 4). The results show that as the relative dispersity increases beyond 0.01 we obtain a less good description of our data, in particular with less pronounced oscillations around q=0.1 Å-1 – 0.2 Å-1, thus resembling the effect of an increased axis ratio of the discs.

The analysis revealed a strong dependence between Nlip and the axis ratio of the lipid bilayer patch for the discs, so while the mean of the Gaussian of Nlip=104 yielded an axis ratio of 1.3, then already at Nlip=106, fully circular (axis ratio 1.0) and hence fully loaded discs were obtained. Conversely, an axis ratio of 2.0 was obtained already for Nlip=94. As for the investigations of the axis ratio (see above), the model parameters are correlated, so that an increase of the dispersity, e.g. to σlip=0.05, in part can be compensated by variation in some of the remaining model parameters. However, the re-optimized fits that could be obtained for the σlip=0.05 sample were still significantly less good due to the more smeared out features than the one obtained when assuming rather monodisperse discs (σlip≤0.01). This leads us to conclude that monodisperse model in terms of the number of embedded lipids does indeed provide a good description of the discs.

6) What do the authors mean with "membrane mimicking capabilities"? Just physical properties of lipids in nanodiscs as compared to liposomes?

Yes, this was poorly described and has been rephrased in the revised version. We indeed simply meant to which extent the order of the lipids resembles those in a more extended bilayer. We have updated the text to reflect this.

7) It is not clearly described how SAXS and SANS data have been scaled for data fitting. Is it just a hardware-dependent correction or does it have an underlying physical meaning? This requires a short explanation.

Briefly, our SAXS and SANS measurements are done on an absolute scale in units of cm-1. The data are calibrated during the experiment (SAXS data using water or BSA as secondary standards and SANS from an absolute count of the number of neutrons), so that we know the expected amount of forward scattering, i.e. I(q=0), to expect for a given sample of a given concentration. We can thus use agreement between expectation and observation as a quality control measure. In contrast, un-calibrated experiments would instead treat I(q=0) as a free parameter. In practice, however, uncertainties in e.g. actual sample concentration means that these numbers do not match exactly, and we generally accept variations in 10%-20% for a SAXS experiment, and slightly higher numbers for a SANS experiment like that described here. We have updated the manuscript to make this clearer.

8) For comparing the NMR structure with the simulation results, have the authors considered the entire NMR structural bundle (10 structures) or just a single structure? This might have a marked impact on the chi-square value for the NMR structure.

We have performed these calculations on a single structure, because the NMR ensemble (PDB ID 2N5E) is not a “dynamic ensemble”, but instead an “uncertainty ensemble” so that the collection of 10 structures are not meant to represent the heterogeneity of the nanodics. We note, however, that all 10 structures are very similar, with typical Cα RMSDs <2Å over nearly all residues (excluding just a few poorly-defined or mobile residues at the tails).

9) Correlation of MD with amide NOES (subsection “Integrating experiments and simulations”): The authors might want to add here that the amide-amide NOEs report on the α-helical secondary structure of the individual MSPs and are not quite sensitive to slight changes in the shape of the overall MSP belt. Thus, a good correlation with these restraints is very much expected.

We agree, and mention this in the revised manuscript.

10) Subsection “Integrating experiments and simulations”, seventh paragraph: it seems like the square root(C) value for the NMR/EPR structure is missing.

This value has been added in the revised manuscript.

11) Order parameters (Figure 4B): What C-H moieties in DMPC have been used to calculate the numbers plotted in the figure? Also, it seems counter-intuitive that lipids in the rim region are more mobile than lipids in the core.

The order parameters were calculated and averaged over all C-H groups in the two alkyl chains in DMPC; this is now described more clearly in the revised manuscript.

We agree that it is not obvious that the lipids in the rim region would be more conformationally heterogeneous than those in the centre of the disc. This observation is, however, in agreement with previous simulation results (Siuda and Tieleman, 2015; Debnath and Schäfer, 2015) and we speculate that the irregular belt-protein surface makes it difficult for the lipids to be highly ordered. This is now discussed briefly in the revised manuscript.

12) It is very good that the authors make their MD data publicly available on GitHub. It would be nice if the refined ensemble (or some number of representative frames) could be made available to the scientific community also via a database, such as PDB-dev.

We have opted not to deposit the ensemble in PDB-dev, in part because the technical requirements for reformatting the data are non-trivial, and in part because it was unclear to us that PDB-dev is the right place to deposit re-weighted molecular dynamics simulations. As the reviewers note, our simulations and analyses are already available online.

13) The clustering step of the MD simulations seems quite important for the rest of the paper, and the choice of methods and parameters is not rationalized. Were other methods/parameters tested and why did the authors settle on this specific choice?

We realize that we may have not adequately described the goal and impact of clustering. In particular, the clustering is only used for visualization and to gain an overview of the ensemble, but is not part of the reweighting procedure (see e.g. the last paragraph of the subsection “Integrating experiments and simulations” in the original manuscript). Indeed, this differentiates our approach from some of the earlier work on reweighting simulations against SAXS data. Instead, in our method we reweight each frame in the original MD simulation, and then accumulate the weights over the clusters. This makes our analyses much more robust against noise that would come from clustering. For these reasons we also did not try different clustering methods. We did examine different cutoffs, and found that the chosen value, giving rise to six clusters and with the top three clusters each having >10% population, represented the data sufficiently well. We have updated the text to make it clearer that the clustering is used only for analyses and not reweighting.

14) Additionally, the reviewers had the following suggestion, but did not strictly require additional experiments to be performed: Have the authors tried to perform SANS contrast matching experiments to obtain scattering data of the protein belt and the lipid patch only, respectively? Since the NMR structure does not contain lipids, such experiments would be helpful to validate the "lipid-free" structure.

We have indeed previously attempted such experiments based on contrast variation neutron scattering, but so far without the required signal-to-noise to provide useful information. We have previously published the production and analysis of “stealth”, neutron-invisible nanodiscs for use when examining membrane proteins embedded within such discs (Maric et al., 2014). Our pilot experiments on MDP1D1 with either lipids or belt-proteins matched out, however, suggest that we need stronger neutron sources to obtain sufficient signal to provide information beyond that intrinsically available in the multi-contrast SAXS experiments.

Associated Data

    This section collects any data citations, data availability statements, or supplementary materials included in this article.

    Data Citations

    1. Bibow S, Polyhach Y, Eichmann C, Chi CN, Kowal J, Stahlberg H, Jeschke G, Guentert P, Riek R. 2017. The 3D solution structure of discoidal high-density lipoprotein particles. RCSB Protein Data Bank. 2N5E [DOI] [PubMed]

    Supplementary Materials

    Transparent reporting form

    Data Availability Statement

    Scattering data, molecular simulations and results from reweighting are available at https://github.com/KULL-Centre/papers/tree/master/2020/nanodisc-bengtsen-et-al (copy archived at https://github.com/elifesciences-publications/KULL-CENTREpapers/tree/master/2020/nanodisc-bengtsen-et-al).

    The following previously published dataset was used:

    Bibow S, Polyhach Y, Eichmann C, Chi CN, Kowal J, Stahlberg H, Jeschke G, Guentert P, Riek R. 2017. The 3D solution structure of discoidal high-density lipoprotein particles. RCSB Protein Data Bank. 2N5E


    Articles from eLife are provided here courtesy of eLife Sciences Publications, Ltd

    RESOURCES