Membrane proteins govern the majority of input and output signals of cells and represent the largest class of pharmaceutical drug targets, making the analysis of their molecular interactions critical to mapping the interactome as well as drug discovery efforts. Due to their integration into a lipid bilayer, in vitro characterization of molecular interactions of membrane proteins presents a unique challenge compared to soluble proteins. By capturing 150 nmsized lipoparticles that contain a targeted membrane protein on the surface of a biosensor, ForteBio’s Octet platform enables the kinetic analysis (kon, kdiss and KD) of membrane protein-analyte interactions. Due to the large size of the lipoparticle, the observed data trace is often inverted, requiring a “flip” during data processing. Using a membrane protein-antibody model system, data processing and analysis of the membrane protein-analyte interaction on a lipoparticle is discussed and validated.
Lipoparticles, available from Integral Molecular (www.integralmolecular.com), offer expression of integral membrane proteins in a native lipid bilayer format for nanoscale in vitro protein analysis research (Figure 1). The proprietary retroviralbased technology incorporates a high concentration of conformationally intact membrane protein into a ~150 nm virus-like particle (VLP) composed of natural cell membranes. This cell-free solution for membrane protein analysis creates a unique biochemical tool to study GPCRs, ion channels, and other membrane proteins. Lipoparticles do not contain a genome and are thus non-infectious.
Octet platform and BLI
ForteBio’s label-free detection platform provides instruments, biosensors, reagents and assay kits for analyzing biomolecular interactions in 96- and 384-well microplate formats. A proprietary subtraction Bio-Layer Interferometry (BLI) technology enables real-time analysis of interactions on the surface of a fiber optic biosensor (Figure 2), providing information on affinity, kinetics and concentration. The efficient workflow of the platform provides easy, fast and highquality characterization of drug candidates in drug development applications where existing methods have limitations in throughput, performance and cost.
Biomolecular analysis of the interaction between a lipoparticle associated membrane protein and an antibody begins with immobilization of the lipoparticle upon the biosensor surface. This example exploits the high affinity between wheat germ agglutinin and lipoparticle carbohydrates for lipoparticle immobilization. The basic experiment contains 6 steps (Figure 3): Step 1, biosensor hydration; Step 2, immobilization of biotinylated wheat germ agglutinin (WGA) on a ForteBio Streptavidin Biosensor; Step 3, capture of the VLP utilizing the immobilized WGA; Step 4, wash and establish baseline; Step 5, measure membrane protein-antibody association kinetics; and Step 6, measure membrane protein-antibody dissociation kinetics. Alternative methods of lipoparticle capture or immobilization, such as amine coupling, are available.
Tips For Optimal Peformance
- Octet RED, Octet RED96 and Octet RED384 instruments are recommended.
- 180 μL to 200 μL/well of reagent is required in a 96-well microplate. 100 μL/well is required for a 384-well microplate and 40 μL/well is required for a 384-tilted well plate.
- Evaluate shake speeds between 500 and 1000 rpm for optimal performance with each ligand-analyte system.
- Due to their large size, lipoparticles can generate a negative response in biolayer interferometry and, consequently, all binding events subsequent to the loading step may also be expected to generate negative responses. This negative response can be managed in the Octet Data Analysis software (Version 6.3 or greater) by signal inversion using the “flip” feature, followed by standard processing protocols.
- Employ one of three referencing methods:
- Use null lipoparticles (without over-expressed target membrane protein) as a reference biosensor either as a single biosensor in a column of biosensors (reference biosensor) or as a parallel column of biosensors against the same analyte series (parallel reference subtraction).
- Use receptor-containing lipoparticles to probe a “reference well” of buffer (no analyte or negative control antibody).
- Combine the use of reference biosensor(s) and a reference well with double referencing.
- Choose controls that closely match the analytes of interest (e.g., if testing murine-derived antibodies, the control should be a murine-derived antibody with an epitope not present on the protein of interest or the lipoparticles).
- If inter-step correction is to be used during data processing, the same well should be for both baseline and dissociation steps.
- Streptavidin (SA) Biosensor Tray (ForteBio part no. 18-5019)
- Black 96-well (Greiner part no. 655209), 384-well (Greiner part no. 781209) or 384-tilted well (ForteBio part no. 18‑5080) polypropylene plate
- PBS supplemented with 1 mg/mL BSA (PBS-B; BSA Fraction V, Sigma part no. A3059, 0.2 μm filtered)
- Biotinylated Wheat Germ Agglutinin (WGA) (Vector Laboratories part no. B-1025)
- 1 to 7 units lipoparticles (ligand) per biosensor
- Antibody (analyte) of interest
- Negative control analyte
Optional Reagents For Regeneration
- 1% Empigen (Sigma part no. 45165)
- 500 mM NaCl
- 500 mM N-acetyl-D-glucosamine (NAG) (Alfa Aesar part no. A13407)
- 10 mM HCl
Assay Protocol With Single Reference Well
Sample Plate Preparation
Prepare a microtiter plate (96-well, 384-well or 384-tilted well) with the required ligands, analytes and wash buffers (refer to Figure 4A for representative plate setup for single reference well subtraction).
All steps in the following example were performed at a shake rate of 600 rpm, except for the association and dissociation steps, which were performed at a shake rate of 1000 rpm. Optimal shake speeds may vary.
Prepare and Load Biosensor Surface with WGA
- Move biosensors to PBS-B (Figure 4A, column 1) and equilibrate for 5 minutes.
- Perform a 3-minute wash step (Figure 4A, column 2).
- Load biosensors with WGA by incubating with 50 μg/mL biotinylated WGA in PBS-B (Figure 4A, column 3) for 3 minutes.
- Quickly rinse biosensors by running a 3-minute baseline in PBSB (Figure 4A, column 4).
Load Lipoparticles on Biosensors
- Load lipoparticles for 30 minutes. Typically, the lipoparticles are diluted to a concentration of 1–10 μg/mL in PBS-B with 1–7 units of membrane protein per well (Figure 4A, column 5). A shift of 1 to 2 nm during target loading is expected. The sign of the observed nm shift may be negative due to the large size of the lipoparticles. Loading levels can be modulated by varying either the particle concentration or the loading time.
- Place biosensors back in PBS-B (Figure 4A, column 6) and wash for approximately 45 minutes. The quality of any subsequent kinetic data is strongly dependent on baseline stability. Higher levels of lipoparticle loading may require longer stabilization times.
- Biosensors can be regenerated back to the WGA surface (either on- or off-line).
- Place biosensors in a solution of 1% Empigen/500 mM NaCl/500 mM NAG for 1 minute (Figure 4A, column 8).
- Place biosensors in 10 mM HCl for 1 minute (Figure 4A, column 9).
- Repeat Steps 1 and 2 three to five times.
- Rinse biosensors in PBS or assay buffer to remove residual detergent or acid (Figure 4A, column 10).
Kinetic data for the interaction of lipoparticle-displayed CXCR4 and the mouse anti-CXCR4 antibody was obtained in a 9-step assay (Figure 5). After biosensor equilibration and wash steps, loading of the biotinylated WGA produced a large positive shift of approximately 3.5 nm (Figure 5, Step 3). Excess WGA was washed away (Figure 5, Step 4) and the CXCR4 displaying lipoparticles were loaded (Figure 5, Step 5). Due to the large size of the lipoparticles, the amplitude of the signal generated by their binding to the biosensor was negative, inducing an observed amplitude shift from approximately 3.5 nm to approximately 1.75 nm. The lipoparticleloaded biosensors were washed extensively (Figure 5, Step 6) in preparation for establishing a baseline signal (Figure 5, Step 7). Lastly, the kinetics of association (Figure 5, Step 8) and dissociation (Figure 5, Step 9) for the interaction between the anti-CXCR4 antibody and the lipoparticle-displayed CXCR4 were measured. The nm shift of the association and dissociation steps were significantly smaller than the nm shift than the loading steps, and are therefore expanded in Figure 6A and 6B for improved visualization.
The raw data acquired for the interaction between the lipoparticledisplayed CXCR4 and the anti-CXCR4 antibody was processed and fit to a curve in order to extract values of kon, kdiss and KD. Processing began with reference correction to compensate for signal drift of the immobilized biosensor with the assay buffer. Using a reference biosensor, the signal generated by a biosensor with CXCR4- containing lipoparticles that probed a blank matrix (PBS-B) was subtracted from both the association and dissociation steps for the interaction between the lipoparticle-displayed CXCR4 and the anti-CXCR4 antibody. Subsequently, Y-axis alignment, inter-step correction and Savitzky-Golay filtering were applied.
|Step Number||Step Name||Time (seconds)||Shake Speed||Step Type|
Due to its large size, the lipoparticle placed the interaction surface approximately 150 nm away from the biosensor surface. In this example, the extended distance caused the amplitude of observed data to be inverted, but otherwise identical (Figure 6A). The “flip” function of the Octet Data Analysis software was used to invert the signal. In the results table of the analysis tab, all rows were selected and “flip” was selected from the right-click menu. All data inverted so that the nm shift was transformed from negative to positive. Inversion of the data to a positive signal allowed curve fitting of both the association and dissociation steps with a 1:1 binding model using the global fitting function (grouped by color, Rmax unlinked by biosensor). Kinetic fitting results are displayed as red lines in Figure 6B and as quantitative values in Figure 6C. Specific processing settings can vary due to both the experimental setup and analyte-ligand system probed.
Validation of “Flip” Data Software Feature
A model system was developed to investigate the validity of inverting the kinetic data obtained from a lipoparticle experiment (Figure 7). In this model system, the interaction between a mouse anti-CXCR4 antibody and a donkey anti-mouse IgG was studied both on the surface of captured lipoparticles (resulting in inverted data) and on the surface of a Streptavidin Biosensor (standard data). The values of kon, kdiss, and KD, derived from curve fitting data for the two assays, were compared to determine if inverting the data from the lipoparticle assay delivered valid kinetic measurements.
The lipoparticle-based model system was evaluated in a 8-step assay (Figure 8A). After biosensor equilibration (Figure 8A, Step 1), biotinylated WGA (50 μg/mL) was loaded onto the Streptavidin Biosensor (Figure 8A, Step 2). After a brief washing step (Figure 8A, Step 3), WGA was used to capture the lipoparticle displaying CXCR4 out of solution (10 μg/mL) (Figure 8A, Step 4), producing a large negative shift due to the binding of the 150 nm particle. After a wash step (Figure 8A, Step 5), the mouse anti-CXCR4 antibody (5 μg/mL) was loaded onto the surface of the lipoparticle (Figure 8A, Step 6). A baseline was established (Figure 8A, Step 7) and the interaction of interest, the binding of a donkey antimouse antibody with the mouse anti-CXCR4 antibody, was then observed (Figure 8A, Steps 8 and 9). The binding of the donkey anti-mouse antibody to the anti-CXCR4 antibody was measured in a half-log titration series from 3 to 100 nM, with the 3 nM and 100 nM samples run in duplicate. A 0 nM donkey anti-mouse IgG sample probed with CXCR4-containing lipoparticles was used as a reference control. The reference corrected data was inverted with the “flip” feature of the Octet Data Analysis software and fit to a 1:1 kinetic binding model (Figure 8B) to obtain values for kon, kdiss and KD (Figure 8C) as described in the Data Processing section.
The second model system assay, arranged without a lipoparticle to produce binding data with a positive nm shift, was performed in a 7-step assay (Figure 8D). After Streptavidin Biosensor equilibration (Figure 8D, Step 1), biotinylated goat anti-mouse antibody (0.63 μg/ mL) was loaded (Figure 8D, Step 2). Excess goat anti-mouse antibody was removed by washing (Figure 8D, Step 3) and mouse anti-CXCR4 antibody (5 μg/mL) was loaded (Figure 8D, Step 4). Excess mouse anti-CXCR4 was removed by washing (Figure 8D, Step 5) and the interaction of interest was then studied by binding donkey anti-mouse antibody to the mouse anti-CXCR4 antibody in a half-log titration (Figure 8D, Step 6). The donkey anti-mouse antibody was bound in a half-log titration series from 3 to 100 nM. Dissociation of the donkey anti-mouse antibody from the mouse anti-CXCR4 antibody was observed for one hour in PBS-B. A zero nM donkey anti-mouse antibody sample probed with a biosensor loaded with goat anti-mouse and mouse anti-CXCR4 antibodies was used as a reference control. The reference-corrected data was fit to a 1:1 kinetic binding model (Figure 8E) to obtain values for kon, kdiss and KD (Figure 8F) as described in the Data Processing section.
After reference subtraction, data for both systems was successfully fitted using a 1:1 binding model to obtain values for kon, kdiss and KD (Figures 8C and 8F). The kon values obtained with (1.83 E+4 M-1s-1) and without (1.26 E+4 M-1s-1) the lipoparticle agreed within 0.57 E+4 M-1s-1. The kdiss values with (1.42 E-4s-1) and without (7.13 E-5s-1) the lipoparticle agreed within 0.7 E-4s-1. The KD values, calculated as the ratio of kdiss/kon, differed by less than 2.1 nM. Raw binding data for both experiments produced concentration-dependent binding curves for analyte (donkey anti-mouse antibody) from 3 to 100 nM (compare Figures 8B and 8E). While the total nm shift observed for the model system without the lipoparticle was consistently greater than corresponding data in the presence of the lipoparticle, derivation of kinetic values is independent of the magnitude of the nm shift and the difference therefore does not bear relevance to the experimental objective of obtaining kinetic information. Comparison of the data for model system experiments performed in the presence and absence of the lipoparticle therefore demonstrates excellent agreement between the two results, validating kinetic characterization of membrane protein-ligand interactions on a lipoparticle using the ForteBio Octet platform.