Here, we present a protocol for the production and purification of proteins that are labeled with stable isotopes, and subsequent characterization of protein-protein interactions using Nuclear Magnetic Resonance (NMR) spectroscopy and MicroScale Thermophoresis (MST) experiments.
Filamentous proteins such as vimentin provide organization within cells by providing a structural scaffold with sites that bind proteins containing plakin repeats. Here, a protocol for detecting and measuring such interactions is described using the globular plakin repeat domain of envoplakin and the helical coil of vimentin. This provides a basis for determining whether a protein binds vimentin (or similar filamentous proteins) and for measurement of the affinity of the interaction. The globular protein of interest is labeled with 15N and titrated with vimentin protein in solution. A two-dimensional NMR spectrum is acquired to detect interactions by observing changes in peak shape or chemical shifts, and to elucidate effects of solution conditions including salt levels, which influence vimentin quaternary structure. If the protein of interest binds the filamentous ligand, the binding interaction is quantified by MST using the purified proteins. The approach is a straightforward way for determining whether a protein of interest binds a filament, and for assessing how alterations, such as mutations or solution conditions, affect the interaction.
Interactions between proteins allow the formation of molecular machines that create order within cells. The individual interactions are often weak but usually contribute to multivalent complexes that can be cooperative and dynamically regulated. Sensitive assays that provide atomic resolution and quantitative information about such complex interactions are needed to deduce mechanisms and design interventions such as drug-like molecules. NMR spectroscopy is an efficient method for obtaining such information about protein interactions, and also is used for fast screening for ligands including those that bind weakly1. The NMR methods used can be categorized into those that are protein observe or ligand observe. This manuscript uses the former approach in which a spectrum of a stable-isotope labeled protein that is comparatively small (usually under 20 kDa) is acquired and the unlabeled ligand is titrated. This allow the labeled residues involved in the interaction to be mapped in favorable cases. Once the complex forms, there are changes in the chemical environments of interacting residues that manifest themselves as changes in the chemical shift and shape of their NMR signals. The extent of such changes correlates with the degree of involvement of these groups in the interaction. Chemical shift perturbations (CSPs) can be measured by comparing a series of NMR spectra of the protein collected in the absence and presence of varying amounts of the ligand. For larger ligands or complex interactions, the change in peak shape or intensity can be measured to deduce interactions.
The most common 2D experiment used for detecting ligand interactions is the 15N-heteronuclear single quantum correlation (HSQC) experiment2. This requires that one protein be uniformly labeled with 15N, which is typically achieved by expressing them as affinity-tagged versions in E. coli bacterial cultures grown in 15N-enriched media. Binding is apparent when the HSQC spectra collected during the titration are superimposed, revealing peak changes for a subset of residues involved in the complex formation. The interaction can occur in the fast exchange regime where the free and ligand-saturated state signals collapse into one population averaged peak. Alternatively, in the case of slow exchange between the states, both signals are observed with integrals that represent their relative amounts. While NMR lineshape analysis can be used to estimate the binding affinities in some cases, methods such as MST have also proven convenient and provide cross-validation of genuine interactions.
The example provided is of two proteins found within desmosomes. They mediate junctions between cell surfaces and the cytoskeleton and mediate multivalent interactions between cell adhesion machines and intermediate filaments to maintain the integrity of skin and heart tissues and withstanding of shear forces. Diseases can result when desmosomal proteins such as desmoplakin or vimentin are compromised by mutations or autoantibodies, leading to destabilization of cell-cell junctions, and hence their interactions are of critical importance3. The structural basis of ligand binding by desmosomal proteins can be characterized by NMR spectroscopy, while the interactions can be quantified by MST. Methods herein were used to characterize the interactions between plakin repeat domains (PRDs) which often are present as tandem sets that offer basic grooves, and vimentin, an intermediate filament that interacts through an acidic surface offered by its helical bundle4. These complexes are formed at the cell membrane where they anchor for intermediate filaments of the cell cytoskeleton to desmosomes that connect to adjacent cells, thus forming a network of adhesive bonds that radiates throughout a tissue.
1. Recombinant Protein Expression
2. Immobilized Metal Affinity Chromatography (IMAC) Purification of VimRod and E-PRD
3. NMR Methods
4. MicroScale Thermophoresis (MST)
The E-PRD domain (residues 1822-2014 cloned into pProEX-HTC) of the human envoplakin gene and the VimRod domain (residues 99-249 cloned into pET21a) of human vimentin4 were expressed with His6 tags and purified. Figure 6 and Figure 7 demonstrate the levels of purity of VimRod (18.8 kDa) and E-PRD (21.8 kDa) obtained from this method of protein purification. The removal of the His6 tag from the E-PRD construct is essential for the MST experiments as the VimRod protein is labeled using a His6 tag binding dye and any E-PRD retaining its His6 tag may compete for binding of the dye. The second IMAC column after cleavage of the tag with TEV protease removes the TEV protease, the cleaved tag and any uncleaved His6-E-PRD that remained. The final polishing step of the purification is size exclusion chromatography. Despite both proteins being of a similar size, the VimRod elutes from the column at a 51 mL while the E-PRD elution peak is centered at 72 mL where a protein monomer of this size would be expected. The apparent increase in size of VimRod is likely due to the its characteristics as a filamentous long rod shaped protein as analytically ultracentrifuge experiments demonstrated that VimRod was monomeric4. Lower yields of protein are obtained from the cultures grown in M9 than those from rich broth due to a lower amount of cells being produced in the minimal media. The initial growth of larger starter cultures for M9 preparations in TB allows improvement of cell yields while maintaining the extent of 15N labeling necessary for the NMR experiments.
The 15N-1H HSQCs were acquired for the wild type and R1914E mutant of E-PRD in presence or absence of VimRod (Figure 8A–8D). The spectrum of E-PRD in Figure 8A shows the expected number of well resolved peaks, indicative of a properly folded protein. In the presence of VimRod (Figure 8B) the spectrum shows extensive line broadening and peak disappearance, corresponding to binding between the E-PRD and VimRod. This binding is lost by mutation of R1914E as evidenced by comparison of Figure 8C and 8D. Little change is observed in the spectrum upon addition of VimRod to the R1914E mutant indicating a lack of binding between this mutant E-PRD and VimRod. The E-PRD peak intensities in the presence/absence of VimRod were compared and plotted as the relative peak intensities in Figure 8E, which indicates the range of peak broadening in the E-PRD complex. The R1914E mutant of E-PRD (not shown) retained about 97% of peaks at 20% or higher peak intensities in presence of VimRod compared to about 20% for the wild type (Figure 8E). This represents a loss of function point mutant, with additional mutants having intermediate effects also having been studied4.
To validate and quantitate the binding of VimRod and E-PRD MST analysis using His6-VimRod labeled with fluorescent RED-tris-NTA dye as the target mixed with decreasing concentrations of the ligand E-PRD from 1.28 mM to 39.1 nM were performed. Three binding titrations were carried out and the results are averaged and shown in Figure 9. The data were fit with a standard model of one-site ligand binding and gave a KD of 25.7 ± 2.1 μM. Evaluation of the binding between VimRod and E-PRD by surface plasmon resonance gave a similar KD value of 19.1 ± 1.3 μM4.
Figure 1: Screen Capture of the Setup of the NMR Experiment. The window shown is used to set up a standard experiment to collect a HSQC dataset. Experiment parameters are read in adjacent to Experiment. The ZGPR experiment shown is chosen as an initial experiment to load the standard and solvent dependent proton parameters. The Title window is used to input experimental details for record keeping purposes. To collect the HSQC spectrum the ZGPR experiment is replaced with SFHMQC3GPPH. Please click here to view a larger version of this figure.
Figure 2: Adjusting of NMR Experimental Parameters. The window shown is used for entering the basic parameters for the NMR pulse sequence in order to optimize the signal. Please click here to view a larger version of this figure.
Figure 3: NMR Data Processing. Parameters used for processing each of the two dimensions of the NMR spectrum are shown, with arrows indicating those that are typically adjusted. Please click here to view a larger version of this figure.
Figure 4: Parameters for NMR Peak Picking. The parameters used for picking NMR peaks in the processed NMR spectrum are shown with typical values. Adjust the ppm range, intensity and number of peaks to optimize the spectra. Please click here to view a larger version of this figure.
Figure 5: Representative Peaklist with Intensities. Each peak that is picked in the NMR spectrum is given a number, and its 1H and 15N chemical shifts and signal intensity are displayed. This peaklist can then be used to compare spectra obtained in the presence/absence of an interacting partner. Please click here to view a larger version of this figure.
Figure 6: Purification of His6-tagged VimRod by IMAC and S. A. The chromatogram for the elution from the IMAC column shows one major peak of VimRod. B. The chromatogram for the elution from the S column shows one major peak. C. SDS-PAGE of fractions collected over the course of purification: MW standards with the MW indicated in kDa to the left of the gel (M), cell lysate (1), IMAC flow-through (2), wash (3), pooled elution (E1), pooled S elution (E2). Bands visible at higher molecular weights in lanes E1 and E2 are oligomers of pure VimRod as confirmed by western blot (data not shown). Please click here to view a larger version of this figure.
Figure 7: Purification of E-PRD by IMAC and S. A. SDS-PAGE of the IMAC purification showing the molecular weight standards with the MW indicated in kDa to the left of the gel (M) and the eluate from the first IMAC column (E1), the TEV cleavage products (+TEV), and the flow through from the second IMAC column (FT). B. The chromatograph from the S column shows one major peak. C. SDS-PAGE of the molecular weight standards (M) and the fractions from S peak. Please click here to view a larger version of this figure.
Figure 8: HSQC Spectra of Wild-type and R1914E Mutant of E-PRD in the Presence and Absence of VimRod. The HSQC spectra show wild-type E-PRD (100 µM) in 20 mM Tris-HCl, 150 mM NaCl, 1 mM DTT, pH 7 in the absence (A) or presence of 50 µM VimRod (B). Panels C and D are the HSQC spectra of the R1914E mutant (100µM) in the absence or presence of 50 µM VimRod, respectively. In panel E the relative 1H-15N peak intensities of the E-PRD with or without VimRod binding are shown as a function of the peak number, which is arbitrarily assigned and not based on sequence position. These values can be used to define a significance cutoff for peak intensity reduction upon addition of a ligand. If assignments are available, the significant values can often be seen to map to a binding area. Please click here to view a larger version of this figure.
Figure 9: Binding of E-PRD to VimRod. E-PRD was diluted in a series of two-fold dilutions from 1.28 mM to 39.1 nM and incubated with labeled VimRod before performing MST analysis. Data from three independent assays were combined. The data were fit to a KD model giving a KD of 25.7 μM with a KD confidence of ± 2.1 μM. Please click here to view a larger version of this figure.
Reagent | Quantity |
Sodium phosphate, dibasic (anhydrous) | 6.0 g |
Potassium phosphate, monobasic (anhydrous) | 3.0 g |
Sodium Chloride | 0.5 g |
H2O | Up to 950 mL |
Table 1. M9 media for Isotopic Labeling.
Reagent | Quantity |
15NH4Cl | 1.0 g |
Glucose (or 13C-glucose) | 2.0 g |
1 M MgSO4 | 2 mL |
50 mM CaCl2 | 4 mL |
20 mg/mL Thiamine | 1.0 mL |
3 mM FeCl3 | 400 µL |
Metal Mix (Table 3) | 500 µL |
H2O | Up to 50 mL |
Table 2. Nutrient Mix for Supplementation of M9 media.
Reagent | Quantity |
4 mM ZnSO4 | 323 mg |
1 mM MnSO4 | 75.5 mg |
4.7 mM H3BO3 | 145 mg |
0.7 mM CuSO4 | 55.9 mg |
H2O | Up to 500 mL |
Table 3. Metal Mix Supplement for Enriching the MT Nutrient Mix.
Sample | 1 mM E-PRD in buffer A1 (µL) | 1mM VimRod in buffer A (µL) | Buffer A (µL) | 200 µM DSS in D2O (µL) | Buffer B2 (µL) | Total Volume (µL) |
E-PRD alone | 50 | 0 | 50 | 50 | 350 | 500 |
E-PRD + VimRod | 50 | 50 | 0 | 50 | 350 | 500 |
1Buffer A: 20 mM Tris-HCl, 1 mM DTT, pH 7 | ||||||
2Buffer B: 23 mM Tris-HCl, 1.14 mM DTT, pH 7 |
Table 4. NMR Sample Preparation.
The 2D 15N-resolved NMR experiment is one of the most widely used methods to show how two molecules interact. It is the most information-rich method that allows both partners' signals to be continuously monitored throughout a titration experiment in solution state. Although typically qualitative in the case of large complexes, the method can also be used in favorable cases to measure binding affinities where NMR signals can be tracked in high resolution spectra. Where assignments can be conveniently made, such as in the case of many proteins under 20 kDa in size, the binding sites can also be mapped. Complementary assays such as MST provide quantitative information about interactions in solution, and require less protein in unlabeled states. Comparison of mutant binding data is useful for providing controls to ensure that interactions evidenced by NMR line broadening are genuine and not artifacts of, for example, aggregation or viscosity changes.
Protein Expression
Streamlining the expression process reduces the amount of labor intensive protein production. Part of this optimization process involves identification of an appropriate strain of E. coli for the recombinant expression of protein. Strain preference depends on elements including the nature of the vector in use and, more specifically, the ultimate stability of the recombinant protein being expressed7. The risk of degradation of the heterologous protein by endogenous E. coli proteases can be reduced by use of protease deficient E. coli such as the BL21 strain. For genes containing rare codons, a strain such as BL21-CodonPlus (DE3)RIPL may be preferred. This strain combines the protease deficient nature of the BL21 strain with additional endogenous copies of rare codon tRNAs for arginine, isoleucine, proline, and leucine. Alternatively, rare codons that can compromise overexpression may be avoided by ordering a codon-optimized construct from a commercial source. Many strains of E. coli are available for recombinant gene expression, each optimized for circumvention of a particular problem during expression7. In the case of this study, the standard protease deficient strain BL21(DE3) produced adequate quantities of soluble protein for subsequent purification and analysis.
Protein Purification
The purification protocol for a given protein is often unique in the sense that each protein remains stable and soluble under different conditions such as temperature, salt concentration, or pH. The overall effectiveness of purification through affinity chromatography is also sensitive to the concentration of eluting species such as imidazole at various steps during the purification process. In this work, critical buffer conditions for IMAC were pH for the E-PRD, and imidazole concentration for the VimRod. A pH of 7.5 was required to avoid precipitation of the E-PRD following initial elution from the IMAC column. For the IMAC purification of VimRod, increasing the concentration of imidazole from 30 to 50 mM during the column wash step was found to have a substantial improvement in the purity of the final elution fractions. For the elution step, increasing the concentration of imidazole from 250 to 350 mM also was found to improve the yield of the final elution. Initial attempts to elute protein using 250 mM imidazole led to incomplete elution of VimRod as revealed by a final 1 M imidazole strip of the column (data not shown). Increasing the imidazole concentration to 350 mM for the elution was sufficient to recover all of the protein bound to the column. S can serve a dual purpose because it acts as a polishing step for protein purification while simultaneously performing buffer exchange. Buffer exchange is a critical step for subsequent binding analysis since it removes the imidazole used to elute His6-tagged protein. It also serves as an opportunity to change conditions such as salt concentration or pH, which may impact the efficacy of certain downstream techniques or assays. Protein thermal shift (PTS) can be used to identify optimal buffers for downstream assays, especially for those requiring stable protein for prolonged periods of time at room temperature8,9.
Binding Analysis
Protein that is freshly prepared is critical for accurate binding assays, although frozen protein can also be used as long as the results are compared. Filamentous proteins such as vimentin multimerize in a salt and pH dependent fashion, and hence the solution condition need to be optimized and the oligomeric state estimated by a method such as SEC10,11, dynamic light scattering12 or analytical ultracentrifugation13,14,15. NMR spectroscopy is well suited for measuring ligand interactions of small proteins at atomic resolution. However, when a protein interacts with larger molecule, slower tumbling ensues, and this results in loss of signals, which can confirm binding although it does not necessarily allow mapping of binding sites, which would also require assignment of at least backbone resonances. In this scenario, NMR experiments do not allow identification of the interaction site. Hence site directed mutagenesis is applied to identify the critical residues needed for the binding. Such mutants therefore do not exhibit signal loss. In this protocol, a mutant form with a substitution at position 1914 retains the peak intensities in presence of VimRod and therefore confirms disruption of the interaction of E-PRD and VimRod. Assignment of the backbone and sidechain resonances would add value to this approach, particularly as the structure for the free E-PRD has been solved by X-ray crystallography4. Future applications of NMR include characterization of complex interactions between larger molecules and will benefit from ultra high field magnets and the use of other observable groups such as 13C-labeled and trifluoro methyl groups as reporters.
MST has a number of advantages for studying binding interactions16. The binding partners are free in solution and not immobilized. Analysis of the quality of the samples is built into the software with quality control reporting of aggregation, adsorption to the capillaries or insufficient fluorescent labeling of the target molecule. Small amounts of the target are typically used, the concentration of the labeled target is usually between 20-50 nM in a 10-20 μL volume/reaction. This protocol uses very small reaction volumes (10 μL) to maximize the concentration of ligand that can be achieved in the titrations allowing weak binding interactions to be characterized. This necessitates accurate pipetting and care being taken to avoid introducing bubbles while still thoroughly mixing. Adequate mixing is critical for accurate, consistent fluorescence measurements along the set of serial dilutions. The amount of Tween-20 in the MST experiments was reduced from a standard 0.05% to 0.015% to lower the tendency to create bubbles and improve mixing.
The RED-Tris-NTA dye provides a quick, easy and convenient way to fluorescently label any protein that has a His tag. The labeling is effectively complete in only 30 minutes and is very tight so that no dye removal procedure is necessary. No modifications are made to amino acid residues in the protein that might alter the ligand binding properties. A caveat is that only the protein to be labeled should have a His6 tag. This required the cleavage of the tag from the ligand protein, E-PRD, and the removal of the tag and uncleaved E-PRD with a second IMAC column step. If possible, the ligand protein should be prepared without the use of a His tag. Alternatively, proteins may be covalently labeled with a fluorophore through amine coupling to lysine residues or thiol coupling to cysteine residues. However, care must be taken when using such systems since the covalent attachment of a fluorophore may affect electrostatic or polar binding interactions relying on lysine or cysteine residues. The quantification of binding affinity between the VimRod and E-PRD by MST was unusually sensitive to salt concentration. This problem was mitigated by initially dialyzing both the target and ligand into the same batch of assay buffer. Nonetheless, saturation of the MST binding curve could not be achieved when performing the MST assay in the presence of 150 mM NaCl due to the complex behavior of VimRod. Reliable, complete data was obtained once the concentration of NaCl was lowered to 10 mM allowing accurate calculation of the KD. Hence, careful optimization of solution conditions and comparison with complementary assays are recommended to achieve robust results. Furthermore, MST may be used to quantify the salt dependence for a given interaction, quantify stoichiometric properties of protein interactions, monitor protein folding, and probe into enzyme kinetics17.
The authors have nothing to disclose.
This project has been supported by NSERC RGPIN-2018-04994, Campus Alberta Innovation Program (RCP-12-002C) and Alberta Prion Research Institute / Alberta Innovates Bio Solutions (201600018), awarded to M.O and Genome Canada and Canada Foundation for Innovation grants awarded to The Metabolomics Innovation Centre (TMIC) and NANUC.
Monolith NT.115, includes control and analysis software | NanoTemper Technologies | MO-G008 | Instrument for microscal thermophoresis |
His-Tag Labeling Kit RED-Tris-NTA | NanoTemper Technologies | MO-L008 | RED-tris-NTA dye for MST |
Standard capillaries | NanoTemper Technologies | MO-K022 | capillaries for MST |
HisTrap HP, 5 mL | GE Healthcare | 17524801 | IMAC column |
HisPur Ni-NTA resin, 100 mL | Thermo Fisher Scientific | 88222 | IMAC resin |
HiLoad 16/600 Superdex 75 pg | GE Healthcare | 28989333 | SEC column |
TEV protease | Sigma Aldrich | T4455 | cleavage of his tag from E-PRD |
BL21(DE3) Competent E. coli | New England Biolabs | C2527H | cells for protein expression |
cOmplete Protease Inhibitor Cocktail Tablets EDTA-Free | Sigma Aldrich | 11873580001 | protease inhibitors for protein purification by IMAC |
TCEP, Tris(2-carboxyethyl)phosphine hydrochloride | Sigma Aldrich | C4706 | reducing agent for protein purification |
Reagents (HEPES, NaCl, etc) | Sigma Aldrich | various | Preparation of media and buffers |
Ammonium chloride (15N, 99%) | Cambridge Isotope Laboratories | NLM-467 | isotope labelling for NMR |
D2O, Deuterium Oxide (D, 99.8%) | Cambridge Isotope Laboratories | DLM-2259 | NMR sample preparation |
DSS, Sodium 2,2-dimethyl-2-silapentane-5-sulfonate-D6 (D,98%) | Cambridge Isotope Laboratories | DLM-8206 | reference for NMR |
Precision 5 mm NMR Tubes, 7” long | SJM/Deuterotubes | BOROECO-5-7 | NMR tubes |
NMR spectrometer (14.1 Tesla) | Bruker | acquisition of NMR data | |
TCI 5mm z-PFG cryogenic probe | Bruker | acquisition of NMR data | |
Name | Company | Catalog Number | Comments |
Software | |||
Bruker TopSpin 4.0.1 | Bruker | processing of NMR data | |
MO.Control | NanoTemper Technologies | included with Monlith NT.115 | |
MO.Affinity Analysis | NanoTemper Technologies | included with Monlith NT.115 |