Engineering

Overview

"Failure is central to engineering." This statement has always inspired our team.

To realize the vision of NanoDisguiser, the wet lab group members were divided into different teams to construct different parts of NanoDisguiser sequentially, and finally complete the assembly of the nanodisc. We hope to engineer the assembly process of the nanodisc to make it more repeatable. For example, we conducted a large number of repeat experiments in the purification of MSP to explore the best experimental conditions (see Notebook page). After the assembly was completed, we conducted a rich characterization and measurement of NanoDisguiser and carried out functional verification. In short, our engineering design revolves around the following four parts:

1. Express MSP, using models to help us refine the protein structure design step by step.

2. Use sGFP technology to construct stable membrane protein dimers.

3. Assemble NanoDisguiser and conduct multifaceted characterization and measurement.

4. Verify the antiviral function of NanoDisguiser using virus-like particles (VLPs).

Our entire engineering process is logically carried out under the DBTL (Design, Build, Test, Learn) principles recommended by iGEM, continuously modifying and iterating to improve our project.

Production of MSP

Research

The inspiration for nanodiscs comes from high-density lipoprotein (HDL) in the human body that transports cholesterol. By engineering the membrane scaffold protein (MSP) through the modification of ApoAI, it has become a common tool in structural biology. Whether in the laboratory or on the production line, MSP faces engineering issues in terms of length design, purification process, and cyclization methods. We aim to conduct comprehensive engineering iterations on MSP, starting with the common MSP1E3D1. Through literature research, we learned that adding SpyTag and SpyCatcher at both ends of MSP can achieve self-cyclization of MSP, enhancing its stability and reducing the difficulty of purification. We refer to these monomer MSPs as Mono-MSP. During the purification of Mono-MSP, we encountered some difficulties, which forced us to consider how to more effectively produce large nanodiscs, leading us to create Multi-Polymerized MSP, which expresses three MSPs that can link to each other, greatly reducing the difficulty of induction expression and protein extraction.

Cycle1 Design and presentation validation of Mono-MSP

1. Design

MSP has a very high degree of editability. To achieve project goals, we need to produce different types of nanodiscs, characterize their properties and functions. Starting with cyclization methods and length design, we have designed experiments for four types of Mono-MSP: MSP1E3D1 – for making the most basic and most fluid nanodiscs; spMSP1D1 and spNW15 – for making more stable and controllable self-cyclized nanodiscs; spNW50 – for making self-cyclized nanodiscs with a larger membrane area.

2. Build

We obtained plasmids and gene sequences from Addgene and synthesized plasmids in GENEWIZ (Figure 1). BL21(DE3) containing plasmids were induced with 0.2 mM IPTG and cultured at 16°C for 16 hours.

Next, we need to further extract and purify Mono-MSP. Although MSP has amphiphilic properties similar to membrane proteins, due to its special conformation and support from related literature, we simplified the purification process of amphiphilic proteins, assuming that MSP could be purified in a manner similar to water-soluble proteins, while controlling some conditions to stabilize MSP. We used affinity chromatography to purify the bacterial liquid, and then further purified it by size exclusion chromatography (SEC).

3. Test

We verified the results of induction and purification of Mono-MSP through SDS-PAGE. We found that the initial induction effects of spNW50 and MSP1E3D1 were not good. For spNW15, spMSP1D1, and spNW50, we observed impurities with a molecular weight twice that of these three, and the amount of this impurity for spNW50 was particularly high, leading us to speculate that MSP might have undergone dimerization, and spNW50 is more prone to dimerization.

4. Learn

In the SDS-PAGE results after the purification of MSP1E3D1, we did not observe dimerization bands, which convinced us that the occurrence of dimerization was due to the interaction between the tag and catcher. For the phenomenon that spNW50 is more prone to dimerization, we believe it is because the longer the MSP, the farther the distance between the tag and catcher, and the lower the probability of self-cyclization.

We found that the purification process of MSP1E3D1 was very difficult, and we could only obtain a protein with a very low concentration. We analyzed that this was because MSP could not be simply approximated as a soluble protein, and the purification of MSP protein might need to partially refer to the purification methods of membrane proteins. In contrast, the purification of MSP with tags and catchers was more successful, with a higher yield, which is consistent with the phenomena reported in the literature^[1].

Further literature review revealed that during the purification process of MSP with tags/catcher, dimerization and oligomerization phenomena do occur (Figure 2), and the literature characterized this^[1]. However, we do not want MSP to oligomerize too much, leading to protein loss and the inability to form the expected size of nanodiscs.

Cycle 2 Exploration of the dimerization of spNW15, spMSP1D1 and spNW50

1. Design

The literature^[1] mentioned that to reduce the formation of oligomers, they chose to induce protein expression at 16°C, and our induction temperature is consistent with theirs, but still cannot avoid the combination between MSP molecules. Therefore, we further pondered how to make MSP have a preference for the latter between molecular combination and molecular self-cyclization?

In response to this question, we first designed simple control experiments and western blot to further verify the existence of intermolecular interactions; secondly, we optimized the overall induction and purification conditions.

2. Build and Test

We purified the bacterial liquid without IPTG induction under the same conditions and did not observe any protein bands through SDS-PAGE, which proved that only when the protein was successfully expressed, this large molecular weight protein would appear, and the molecular weight of this protein was an integer multiple of the monomer protein, which was enough to prove the occurrence of dimerization. We verified the bacterial liquid induced by IPTG through western blot, but unfortunately, we did not observe the bands of dimerization. We placed the protein at different temperatures and verified through SDS-PAGE after 24 hours, but did not find obvious differences between the bands of different temperature groups.

3. Learn

We successfully demonstrated the existence of protein dimers through a control experiment, but the western blot was not successful. However, it should be noted that errors may occur during the western blot process, such as insufficient transfer time. We verify that storage temperature is not responsible for the dimerization of MSP. It has been hypothesized that proteins themselves are relatively prone to dimerization, and that the dimerization of proteins intensifies with longer storage time.

The purification of spNW50 was very difficult, which inspired us that to build large nanodiscs, we cannot simply increase the length of MSP. We then turned to design a new Polymer Linker Systems, creating a new efficient method for building large nanodiscs.

Cycle 3 Testing of Polymer Linker Systems

1. Design

Next, we aim to connect MSP monomers through protein linker systems, which enables flexible control over MSP length, and facilitates the construction of large nanodiscs.

Given the efficient protein ligation properties of the SpyTag-SpyCatcher system and successful examples of adding linkers to the N- and C-termini of monomeric MSP to enhance self-cyclization^[2], we aim to test the feasibility of using these linkers to create multi-polymerized MSP. Additionally, to achieve the connection of multiple MSP segments, we have also identified linker systems with similar properties- SnoopTag/SnoopCatcher and Sdytag/SdyCatcher, and initially investigated their performance through literature review.

To rapidly characterize linker interactions, we turn to the BiFC system- by constructing fusion proteins of the test proteins and fluorescent protein fragments, fluorescence could be observed when the test proteins interact. Meanwhile, to avoid potential interference of MSP cyclization characterization from the characterization of dimeric membrane proteins, we have selected the split-mCherry system for fluorescent labeling, distinguishing it from the GFP fluorescence associated with membrane proteins.

2. Build

To initially assess the feasibility of achieving cyclization using the tag/catcher systems, we constructed MSP fused with SpyCatcher/SpyTag or SpyCatcher/SpyTag variants, SnoopCatcher/SnoopTag, and SdyCatcher/SdyTag at both ends as the basic model. mCherry fragments were inserted into these constructs, and successful cyclization of monomeric MSP was used as the criterion for validation (validation of mCherry takes SpyTag/SpyCatcher system as an example).

3. Test

Due to time constraints, we utilized AlphaFold 2 to model and validate these constructs, obtaining protein models for the different fusion proteins. Observations revealed that SpyCatcher-SpyTag and SpyCatcher-SpyTag variants, SnoopCatcher-SnoopTag, and SdyCatcher-SdyTag, could all interact normally, with the conserved regions of MSP exhibiting regular and stable α-helix structures. However, when each components of SpyCatcher-mCherry[1-10]-MSP-mCherry[11]-SpyTag were directly concatenated, there was a high confidence in the model that the SpyCatcher-SpyTag system interfered with the split-mCherry system, resulting in cross-linking as shown in the diagram, which made it impossible to ensure normal interactions.

4. Learn

The modeling results indicated that direct connection between components might interfere with the interaction of linkers, thus failing to ensure the correct conformation of the protein. Therefore, we introduced GSlinker—a flexible linker peptide composed of Gly and Ser residues which connects different components to reduce interference.

The Alpha Fold 2 prediction results showed that after the use of GSlinker, mCherry[1-10] and mCherry[11] could correctly assemble into a β-barrel conformation. This demonstrated that the flexible connection between structural domains plays a crucial role in the function of our MSP, advancing our design towards a mature MSP tag/catcher universe.

Cycle 4 Design of Multi-polymerized MSP

1. Design

Inspired by protein logic gates^[3], to avoid self-cyclization of individual MSP proteins, we added different Catcher/Tag sequences at both ends of MSP respectively. Only when the three sequences are simultaneously connected can the multi-polymerized MSP be constructed, as shown in the figure.

2. Build

We added different series of Tag/Catcher sequences on both ends of MSP, and added split-mCherry sequences at the SpyTag/Catcher junctions to characterize protein interactions. Based on this, we constructed the initial multi-polymerized MSP sequence.

3. Test

We used AlphaFold 2 to model and validate the above constructs. The results showed that cross-reactivity between SnoopTag and mCherry[1-10], and between SdyCatcher and mCherry[11], might lead to self-cyclization of monomeric MSP, while Fragment 2 of multi-polymerized MSP, which did not contain the mCherry sequence, formed the correct linear non-cyclized conformation.

4. Learn

The results indicated that while the initial design of MSPs prevented self-cyclization to some extent, there were unexpected cross-reactions that made it difficult to assemble multi-polymerized MSP.

Based on this, we modified the linker design of the first and third proteins in the multi-polymerized MSP to ensure that MSPs could not self-assemble into cyclic structures. We also modeled and simulated the conformations of the two modified MSP proteins to verify that each part of the multi-polymerized MSP had the correct tertiary structure. (Refer to the Results page for modeling results.)

Cycle 5 Designing Plasmids for Protein Expression Induction

1. Design

After designing the protein sequences, we synthesized plasmids through GENEWIZ and transformed them into BL21(DE3) for protein expression. Our goal was to obtain high-concentration MSP and then mix and incubate the three components to construct multi-polymerized MSP.

2. Build

Activated punctured bacteria were cultured and induced with 0.8 mM IPTG at OD600 0.6-0.8, then cultured at 16°C with shaking for 16 hours. After centrifugation, lysis, and nickel column purification, size exclusion chromatography further purified the protein to obtain high-purity MSP.

3. Test

The induced expression of SCSdC-mCh[1-10] was relatively successful, with a large number of target protein bands visible on the SDS-PAGE results after nickel column purification. Purer protein was also collected after SEC purification. However, the results for SnCSdT and SnTST-mCh[11] were not ideal.

4. Learn

We speculated that during the protein purification process, SnCSdT and SnTST-mCh[11] were degraded by proteases. We decided to add protease inhibitors to the buffer and re-extract the proteins.

Cycle 6 Optimizing Expression Conditions and Constructing Multi-polymerized MSP

1. Design

We planned to add protease inhibitors to the buffers involved in the purification process of SnCSdT and SnTST-mCh[11] and re-purify the proteins after exploring the optimal induction conditions, and finally attempted to connect the three MSPs to construct multi-polymerized MSP.

2. Build

We added serine protease inhibitor AEBSF to the buffers during purification. After observing no significant improvement in the results, we speculated that other types of protease inhibitors might be causing MSP degradation and switched to using a protease inhibitor cocktail. We mixed three types of MSP in a 1:1:1 ratio, incubated them at room temperature in the dark, and validated using SDS-PAGE.

3. Test

We did not observe any obvious target bands (i.e., multi-polymerized MSP formed by the cyclization of the three MSPs) on the SDS-PAGE, indicating that the three MSPs were not successfully connected.

4. Learn

These results indicated that there were issues during the connection of multi-polymerized MSP. We further speculated that this might be due to insufficient protein concentration or inappropriate incubation conditions.

Cycle 7 Optimizing Connection Methods and Successfully Constructing Multi-polymerized MSP

1. Design

We hoped to increase the concentration of the three proteins by changing the mixing time, thereby enhancing the efficiency of protein connection to counteract the low sensitivity of SDS-PAGE.

2. Build

In a new round of protein purification, we attempted the in-vitro construction of multi-polymerized MSP using two methods:

Firstly, we mixed three proteins after resuspension, and incubated in the dark at 4°C for a period of time. Then, we used an ultrafiltration tube to concentrate the protein mixture after incubation to increase the protein concentration.

Secondly, we used the original method but reduced the volume of the buffer used to elute the nickel column to achieve higher concentrations of each protein component.

3. Test

We verified the results using SDS-PAGE. Compared to the individual components of the trimer, the mixture of the three proteins showed two new bands at around 100 and 180 kDa. We speculated that these might be trimers or dimers formed by the connection of MSP monomers.

To further verify the construction of the trimer, we observed the protein mixture that had been incubated for a period of time using a fluorescence inverted microscope. As shown in the figure, we observed significant mCherry red fluorescence under green excitation light.

4. Learn

The display of red fluorescence indicated the possible successful construction of the trimer. However, we could not accurately determine whether SnCSdT was connected to the other two proteins, which may require further characterization.

Membrane Protein Polymerization

Cycle 1: Purification of sGFP1-10 and sGFP11 Proteins

1. Design

Our inspiration comes from the split-GFP self-assembly technology, which provides a concept for constructing membrane protein dimers through the in vitro association of sGFP1-10 and sGFP11. However, due to the difficulty of directly linking membrane proteins to sGFP, we further adopted the streptavidin-biotin system. We decided to use the Escherichia coli BL21 (DE3) strain to express fusion proteins of streptavidin with sGFP (i.e., sGFP1-10 tether and sGFP11 tether) and purchased membrane proteins with an Avi tag, which can be biotinylated using a kit.

2. Build

We reviewed the literature and, based on the experiences of previous iGEMers (mainly from the 2023 BNU-China data), established an experimental plan for expressing and purifying sGFP1-10/sGFP11 tethers. Firstly, a simple examination of the induction conditions for protein expression was conducted by setting the IPTG induction concentrations at 0.5mM and 1mM, respectively, and inducing for 16 hours at 16°C. Next, we purified the protein using nickel affinity chromatography, taking advantage of the His-tag on the fusion protein.

3. Test

1.Protein Crude Extraction:

The SDS-PAGE results show that the induction of sGFP1-10 tether expression was relatively successful and consistent with the experimental results from 2023 BNU-China. The protein expression level was relatively high at an IPTG concentration of 1 mM; therefore, in this round of experiments, we maintained the IPTG induction concentration at 1 mM.

2.Ni-NTA Purification:

After gradient imidazole elution, SDS-PAGE analysis showed that the molecular weight of sGFP1-10 tether was 39.2 kDa, and that of sGFP11 tether was 17.0 kDa. Observing the gel, there were no distinct bands at the corresponding molecular weights, indicating that the purification efficiency of sGFP1-10/11 tethers from the supernatant was poor, and it could not be determined whether the protein was present.

3.Molecular Sieve Selection Purification:

To determine whether the target protein was present in the supernatant and to obtain a higher purity of the target protein, we performed further purification using gel filtration chromatography (molecular sieve). The protein peak profile during purification is shown below. It can be observed that there are many background peaks, and the target protein peak cannot be identified. Subsequent SDS-PAGE experiments also confirmed this, indicating that the purification of the protein from the supernatant was unsuccessful.

4. Learn

By analyzing the above experimental results and referencing relevant literature and the experiences of 2023 BNU-China, we believe that the main reason for the unsuccessful protein purification is that mSA in sGFP1-10/11 tethers causes a large amount of protein to exist as inclusion bodies in the lysate precipitate. Additionally, we conducted further analysis and reflection on the experimental conditions, and if necessary, we can revisit the IPTG induction conditions to obtain soluble protein. Therefore, next, we first validated Hypothesis 1 by reviewing literature on the extraction of inclusion body proteins.

Cycle 2: Purification of sGFP1-10 and sGFP11 Inclusion Body Proteins

1. Design

When expressing exogenous proteins in bacteria, the protein products often aggregate into inactive solid particles known as inclusion bodies. While the primary structure of inclusion body products is completely correct, their stereoconformation is incorrect, leading to misfolding and lack of biological activity. Inclusion body proteins are usually dissolved after being washed, followed by subsequent purification operations. Since dissolution often causes protein denaturation, refolding is necessary after purification to proceed with membrane protein dimer construction.

To obtain high-purity inclusion bodies, we washed the bodies along with some disrupted cell membranes and membrane proteins using low concentrations of denaturing agents and detergents. In the experiment, we used urea as a denaturant for dissolution. The refolding of proteins is very complex, and the refolding yield of inclusion body proteins is usually low, not exceeding 20%. Different proteins have distinct structures and functions, leading to variability in their refolding processes and differing levels of difficulty. Specific refolding methods for the target proteins should be designed and screened based on fundamental principles and preliminary experiments. Due to time constraints, we referenced literature and adopted refolding conditions similar to those in reference.

2. Build

Without changing the cultivation conditions of the bacteria and the protein expression induction conditions, we performed protein purification operations, collecting the precipitate after centrifugation of the lysate.

The precipitate obtained from bacterial lysis and centrifugation is washed sequentially with Inclusion Body Buffer I and II. The precipitate is then dissolved in Inclusion Body Solubilization Buffer, followed by incubation and centrifugation. The supernatant is collected to obtain the inclusion body extract, which is subjected to nickel column affinity chromatography. After protein purification, SDS-PAGE analysis is performed to verify whether the inclusion body protein has been successfully purified. For the successfully purified inclusion body protein, renaturation is carried out according to the specific experimental protocol provided.

3. Test

The SDS-PAGE analysis results of the purified inclusion body proteins are shown in the figure below, with distinct bands at the corresponding molecular weights, indicating successful extraction of the inclusion body proteins. Furthermore, the protein concentration and purity in some of the elution fractions meet the requirements; therefore, we will skip the molecular sieve purification and proceed directly to protein refolding.

After protein refolding and concentration, we also performed SDS-PAGE electrophoresis for verification, but no corresponding bands were observed.

To ensure the accuracy of the experimental results, we concurrently conducted an ELISA experiment to test the activity and concentration of the refolded protein. However, the absorbance values of the protein refolding samples were all relatively low, indicating that the concentration of active sGFP1-10/11 tether in the protein refolding samples was low, which was consistent with the results of the SDS-PAGE.

4. Learn

In this experiment, although we initially succeeded in extracting the target protein, we did not detect satisfactory concentration and activity after the subsequent refolding treatment. Reflecting on our experimental process and results, we accurately grasped the denaturing extraction of protein inclusion bodies and successfully extracted denatured proteins, but we underestimated the difficulty of protein refolding and did not thoroughly explore the refolding conditions for sGFP1-10/11 tethers.

Through subsequent discussions with Professor Yang Dong and Professor Wang Zhanxin, and considering the existing conditions and the challenges of protein refolding, we decided to re-explore the conditions for purifying soluble proteins. Through this cycle, we realized that our communication with the professors during the experiment was lacking, and our background research was insufficient, which hindered the progress of the experiment.

Cycle3 Exploration of Soluble Protein Purification and Dimer Construction

1. Design

We rethought how to obtain active target proteins. Reducing growth rates can help express proteins and consequently decrease the likelihood of inclusion body formation. Therefore, we considered directly altering the culture conditions to lower growth or expression rates, at the cost of potentially decreasing the total protein yield.

We first adjusted the IPTG working concentrations to create a gradient of 0 mM, 0.25 mM, 0.5 mM, 0.75 mM, and 1 mM, inducing at 16°C for 16 hours before proceeding with subsequent protein purification and activity assays. Additionally, we analyzed the possibility that the low protein concentration obtained from the supernatant purification led to the absence of target protein bands in the SDS-PAGE analysis. Therefore, we will attempt to purify soluble proteins from the supernatant again using low-temperature induction with 1 mM IPTG and concentrate the purified protein in hopes of obtaining active protein.

2. Build

1.Set IPTG gradient concentrations for induced expression and perform SDS-PAGE analysis on the crude extract.

2.Purify the supernatant obtained from low-temperature induction with 1 mM IPTG, validate it with SDS-PAGE and ELISA experiments, and upon obtaining active protein, attempt to connect sGFP1-10/11 tethers.

3. Test

1.Results of IPTG Gradient Induction

The results showed that the optimal protein expression for sGFP1-10 tether was achieved at a concentration of 0.75mM, while for sGFP11 tether, the optimal protein expression was achieved at a concentration of 0.5mM.

2.Purification and Concentration of Soluble Proteins

We observed a corresponding band for the target protein, but the band was faint, indicating that the purification of the supernatant proteins was successful, although the concentration of the purified protein was low. We incubated sGFP1-10/11 tethers at 4°C in the dark and observed whether they emitted fluorescence using a fluorescence microscope.

4. Learn

We have obtained the target protein from the purified supernatant and observed fluorescence after mixed incubation. Next, we will further purify the protein under an IPTG concentration gradient to determine the optimal IPTG induction concentration for producing soluble proteins. Additionally, we will continue with protein purification to increase protein concentration and explore the optimal conditions for dimer formation by adjusting the temperature, time, and mixing ratios during incubation.

Currently, we are purifying the proteins separately before combining them; however, we will also attempt to mix and lyse both engineered bacteria for purification, investigating whether sGFP1-10 tether and sGFP11 tether can directly associate during the lysis process to obtain dimer proteins directly.

Fabrication and characterization of monomer MSP nanodiscs

Cycle1 The manufacturing process of nanodiscs

1. Design

We tried to use three kinds of purified MSP proteins with SpyTag and SpyCatcher to self-cyclide to construct nanodiscs with different particle sizes to meet the needs of our experimental design. According to the experimental conditions mentioned in the references, protein and lipid were co-incubated in corresponding proportions. In order to explore the optimal conditions, we constantly adjusted and improved protein-lipid ratio on the basis of literature, and tried to add detergent DDM^[4].

2. Build

In the preparation of the spNW15 nanodiscs, we made a gradient attempt on the protein-lipid ratio between 1：100 and 1：110, with a control group with or without the addition of detergent. For the production of the spMSP1D1 nanodiscs, we constantly modify the protein-lipid ratio from 1:60 to 1:200, and strive to achieve the optimal conditions. And we constructed spMSP1D1 nanodiscs using the protein-to-lipid ratio of 1:60 as reported in the literature. For the construction of the spNW50 nanodiscs, we attempted to construct nanodiscs using the protein-to-lipid ratio of 1:600 as described in the literature.

3. Test

For a series of trial samples, we conducted Native-PAGE test and inally confirmed the best protein-lipid ratio according to the test results.

According to the results of Native-PAGE, we found that spNW15 nanodisc strips are similar to those in the literature, and are banded and discrete, while spMSP1D1 and spNW50 are narrow strips that are inconsistent with the literature. At the same time, the same strip has different shapes and positions in different Native-PAGEs^[5].

4. Learn

According to the results, we believe that in the process of Native-PAGE characterization, the mobility of target bands is related to the isoelectric point of proteins, and the lipid charge may affect the characterization result after the addition of lipids. To solve the problem, we decided to visualize other representations.

Cycle2: Characterization of nanodiscs

1. Design

DLS can measure the particle size of our nanodiscs with high throughput and check whether it conforms to the literature records. Through negative stain electron microscopy, we can intuitively see whether we have constructed the disk structure of the nanodisc, and the particle size can be quantified according to the scale. The samples obtained in the last cycle were examined by DLS and negative stain electron microscopy.

According to the experimental design of modeling and the gradient experiment, we analyzed the main effect factors to make the subsequent construction and characterization of the nanodiscs more directional and save the number of tests.

2. Build

The gradient manipulation was carried out under the experimental conditions of lipid type, protein-lipid ratio, culture temperature and culture time, and the gradient results were characterized by negative stain electron microscopy and DLS(See more--Model).

3. Test

We successfully observed and quantified the samples using negative stain electron microscopy.

We not only actually saw the disk-like structure of the three kinds of nanodiscs through negative stain electron microscopy (Figure 21a, 21b, 21c), but even saw the side of the nanodiscs (Figure 21d).For the spMSP1D1 nanodiscs incorporating the Low-Density Lipoprotein Receptor(LDL-R) ，and we can saw the image of successfully containing membrane proteins in the nanodiscs (Figure 22a).This was consistent with the images of nanodiscs containing membrane proteins in the literature (Figure 22b)^[6].

The negative stain electron microscopic quantization results of spMSP1D1 (Figure 23a) and spNW15 (Figure 23b) were also consistent with the literature records, while the negative stain electron microscopic quantization results of spNW50 were the same as those of DLS (Figure 23c). Finally, the box diagram (Figure 23d) was drawn according to the particle sizes of the three nanodiscs. It can be seen that we produced three nanodiscs with different particle sizes.

A small number of negative stain electron microscope results showed that the nanodiscs clustered (Figure 24a) or formed flower-like structures (Figure 24b). However, the results of DLS (Figure 25) differ greatly from the theoretical values and the quantified values of negative stain electron microscopy.