A Fluorescence-Based Assay to Determine PDZ–Ligand Binding Thermodynamics

Abstract

Postsynaptic density-95, disks-large, and zonula occludens-1 (PDZ) domain interactions with cognate linear binding motifs (i.e., PDZ-binding motifs or PBMs) are important for many biological processes and can be pathological when disrupted. There are hundreds of PDZ–PBM interactions reported but few have been quantitatively determined. Moreover, PDZ–PBM interactions have been identified as potential therapeutic targets. To thoroughly understand PDZ–PBM binding energetics and their specificity, we have developed a sensitive and quantitative equilibrium binding assay. Here, we describe a protocol for determining PDZ–PBM binding energetics using fluorescence anisotropy-based methodology.

1. Introduction

Postsynaptic density-95, Disks-large, and Zonula occludens-1 (PDZ) proteins are ubiquitously found in many types of mammalian cells and regulate the spatial and temporal function of a diverse set of signaling pathways. A distinguishing feature of these proteins is the small (~90 amino acids, ~10 kDa), structurally conserved protein–protein interaction module known as a PDZ domain that selectively interacts with linear C-terminal and internal peptide motifs (i.e., PDZ-binding-motifs or PBMs) [1]. PDZ–PBM interactions and their specificity are critical for many biological processes including the maintenance of cell polarity, neuronal development, and signal transduction. Thus, it is not surprising that genetic mutations in PDZ proteins or perturbation of PDZ–PBM interactions can contribute to pathologies such as neuronal disorders and complications form brain injury, cancer, cystic fibrosis, and viral infections (reviewed in [1]).

Although PDZ–PBM interactions have been extensively characterized, there remains inadequate understanding of the general molecular mechanisms that determine PDZ–PBM specificity, particularly for internal PBMs. This is the result of the low sequence identity among PDZ domain homologs, promiscuous binding profiles, and context-dependent interaction mechanisms. Physiological PDZ–PBM interactions have relatively weak binding affinities, with a dissociation constant (K_d) ranging from μM to low mM [2–4]. To thoroughly characterize PDZ–PBM interactions it is necessary to determine the binding energetics (i.e., ΔG_b, Gibbs free energy of binding) of PDZ–PBM interactions. The binding energetics coupled with high-resolution structural information and mutagenesis can provide deep insights into the binding mechanism and specificity of PDZ–PBM interactions [5–14]. Importantly, this information can be used to design potential PDZ–PBM protein–ligand inhibitors. Indeed, over the past ~10 years PDZ–PBM interactions have been identified as potential therapeutic targets (reviewed in [1]). Here, we describe a general protocol for determining the binding energetics of PDZ–PBM interactions using a robust and simple fluorescence anisotropy-based assay sensitive to interactions with dissociation constants in the 1 to ~500 μM range [15–17].

2. Materials

2.1. Equipment

1. A spectrofluorometer equipped with excitation and emission polarizers and a magnetic stirrer is used to collect fluorescence anisotropy data [15]. Here, we use a Fluorolog-3 (Jobin Yvon, Horiba, NJ) controlled by the FluorEssence V3.8 software program (Jobin Yvon, Horiba, NJ). The spectrofluorometer is set to an excitation wavelength at 340 nm and an emission wavelength at 550 nm, specific for the dansyl [5-(dimethyl amino)naphthalene-1-sulfonyl] chloride fluorophore (see Note 1), with constant stirring at 25 °C. The instrument light slit widths are adjusted in the range of 3–9 nm to optimize the signal-to-noise ratio and maximum output intensity—aiming for ~one million counts per second on the detector (see Note 2).

2. A quartz cuvette containing 4 polished windows, compatible with a magnetic stirring platform is used. We use a 2 mL,10 mm length path cuvette equipped with a stopper and stir bar (Hellma, NY; catalog #119F-10-40).

2.2. Constructs, Medium, and Reagents for PDZ Domain Purification

1. PDZ domains cloned into bacterial expression plasmids are used. Here we use the CASK PDZ domain cloned into pET28a (Novagen) and the Scribble PDZ1 cloned into a modified pET21a (Novagen) [18].

2. Luria–Bertani (LB) medium: 10 g/L tryptone, 5 g/L yeast extract, and 10 g/L sodium chloride.

3. 100 mg/mL ampicillin (pET21a) or 50 mg/mL kanamycin (pET28a) stock solution.

4. E. coli bacterial strain BL21(DE3) (Novagen).

5. 1 M isopropyl 1-thio-β-d-galactopyranoside (IPTG) stock solution.

2.3. Reagents and Solutions

All buffers should be of the highest purity available.

1. Binding buffer: 20 mM sodium phosphate pH 6.8, 50 mM sodium chloride, and 0.5 mM ethylenediaminetetraacetic acid (EDTA).

2. Peptides containing C-terminal PBMs are typically commercially synthesized. The peptides used here were chemically synthesized by GenScript Inc. (Piscataway, NJ) and used at >95% purity. Peptides corresponding to the C-terminus of partner proteins were 8 amino acids long, N-terminally dansylated and contained a free carboxylate at the C-terminus. An internal peptide from SH3-containing guanine nucleotide exchange factor (SGEF) contained 14 amino acids and was N-terminally dansylated. In addition, the C-terminal carboxylate group was amidated (see Note 3).

3. The amino acid sequence of the C-termini of the following human proteins was used in the development of this protocol: Neurexin-1 (residues 1470–1477: NKDKEYYV), Caspr4 (residues 1301–1308: ENQKEYFF), and Syndecan-1 (residues 303–310: TKQEEFYA). The internal SGEF peptide was derived from residues 42–55: KPNGLLITDFPVED [18].

4. Resuspend ~2 mg aliquot of lyophilized peptide in 1 mL of binding buffer and adjust the pH to 6.8 to obtain a highly concentrated (1–2 mM) master stock solution.

5. Prepare a working stock solution of 0.130 mM stock dansylated-PBM by diluting the master stock with binding buffer (see Notes 4 and 5).

6. Store all PBM peptide stock solutions at −20 °C in the dark (see Note 6).

3. Methods

3.1. Purified CASK and Scribble PDZ Domains

A protocol for the expression of CASK and Scribble PDZ domains is described below. Purification of these recombinant PDZ domains is beyond the scope of this chapter but can be achieved using similar methods and buffers as previously published [7, 8, 11]. In brief, CASK PDZ was purified using cation exchange and size-exclusion chromatography (Superdex 75, GE Healthcare Life Sciences) while purification of the first PDZ domain of Scribble was carried out using nickel-chelate affinity chromatography. The N-terminal 6 × His affinity tag of Scribble PDZ1 was removed by proteolysis with recombinant tobacco etch virus (rTEV) protease for 36 h at 4 °C. Undigested protein, cleaved 6 × His tag, and His-tagged rTEV were separated from the Scribble PDZ1 domain by nickel-chelate chromatography. The digested PDZ1 protein was further purified using Superdex 75 size-exclusion chromatography. All proteins were concentrated to ~0.5–2.0 mM in binding buffer (see Note 5). The prepared samples were generally used immediately (see Note 7).

1. Supplement LB medium with the appropriate antibiotic for the desired PDZ domain (final concentration of 100 μg/mL ampicillin for the mpET21a CASK PDZ and 50 μg/mL kanamycin for pET28a Scribble PDZ1).

2. Grow bacterial cells transformed with either CASK or Scribble PDZ constructs in LB medium supplemented with antibiotic at 37 °C under vigorous agitation to an optical density of a 0.6–0.8 measured at 600 nm wavelength.

3. Cool cultures to 18 °C.

4. Induce protein expression by adding IPTG to 1 mM final concentration.

5. Incubate for an additional 16–18 hrs at 18 °C.

6. Harvest bacteria by centrifugation.

7. Proceed with the purification of proteins and concentrate to ~0.5 to 2.0 mM in binding buffer (see Note 5). The prepared samples are generally used immediately (see Note 7).

3.2. Experimental Sample Preparation

The experimental design calls for using serial dilutions of the PDZ domain protein to cover the concentration range of 1–400 μM in discrete steps. Thus, a 0.5 mM stock provides reliable quantification ranging from 1 to 100 μM K_d, while a 2.0 mM stock provides reliable quantification ranging from 20 to 200 μM K_d. The PDZ domain concentration range and number of titration points can be adjusted in subsequent experiments to optimize the titration and obtain a more reliable K_d determination.

3.2.1. Cuvette and PBM Peptide Preparation

1. Rinse the cuvette with distilled and deionized H₂O (ddH₂O) and ethanol using a vacuum cuvette washer (see Note 8).

2. Air dry the cuvette after rinsing and/or cleaning.

3. Add stir bar and 1290 μL of binding buffer.

4. Add 10 μL of dansylated-PBM peptide stock solution to the cuvette to obtain a 1.0 μM PBM peptide concentration (see Note 2).

5. Gently mix the sample using a pipette being careful to avoid introducing bubbles (see Note 9).

6. Cover the cuvette to prevent introducing dust particles and to minimize the dansyl-fluorophore light exposure (see Note 10).

3.2.2. Preparation of PDZ Domain Dilution Stock Solutions

All the PDZ domain diluted solutions described below should be prepared in advance and on ice.

1. Prepare 26 μL of a 100-fold diluted PDZ domain stock per experiment (concentration 5–20 μM).

2. Prepare 10 μL of a tenfold diluted PDZ domain stock per experiment (concentration 50–200 μM).

3. Prepare 656 μL PDZ domain stock per experiment (concentration 0.5–2 mM).

3.3. Binding Assay Parameters and Data Collection

Below are the experimental parameters defined in the FluorEssence V3.8 software. Anisotropy data is collected after each addition of PDZ protein.

1. Experiment type: anisotropy.

2. Temperature: 25 °C (or other desired temperature).

3. Number of data points is ~29 (see Note 11).

4. After adding the appropriate volume of PDZ domain, gently mix the sample using a micropipette to avoid out-gassing followed by the measurement of fluorescence anisotropy (three measurements are taken and averaged). A typical titration experiment uses the following PDZ domain concentrations and volumes per titration step.

5. Data point 1: the initial background measurement without any PDZ domain. This serves as the baseline control.

6. Data point 2 and 3: add 3 μL of 100-fold diluted stock PDZ domain (concentration 5–20 μM) at each step.

7. Data point 4 and 5: add 5 μL of 100-fold diluted stock PDZ domain (concentration 5–20 μM) at each step.

8. Data point 6: add 10 μL of 100-fold diluted stock PDZ domain (concentration 5–20 μM).

9. Data point 7 and 8: add 5 μL of tenfold diluted stock PDZ domain (concentration 50–200 μM) at each step.

10. Data point 9 and 10: add 3 μL of PDZ domain stock (concentration 0.5–2 mM) at each step.

11. Data point 11 and 12: add 5 μL of PDZ domain stock (concentration 0.5–2 mM) at each step.

12. Data point 13 and 14: add 10 μL of PDZ domain stock (concentration 0.5–2 mM) at each step.

13. Data point 15 to 21: add 20 μL of PDZ domain stock (concentration 0.5–2 mM) at each step.

14. Data point 22 to 29: add 60 μL of PDZ domain stock (concentration 0.5–2 mM) at each step.

15. Each PDZ–PBM binding assay is collected in triplicate (using either biological or technical replicates). Figure 1 shows an example of the output from a typical titration dataset.

Fig. 1.

Sample data output of a PDZ–ligand PBM binding assay. The “PDZ ^addedV_stock” column shows the volume of stock solution for each data point of the titration. The “Anisotropy” column shows the average anisotropy value measured for each titration data point. The “Trials” column shows the number of measurements used for calculating the average anisotropy. The “StdErrAniso” column shows the standard deviation of the anisotropy

3.4. Data Processing

Data processing requires the calculation of PDZ concentration for each titration step, [PDZ]_n, where n is the individual data point collected.

1. The PDZ concentration of the first data point is 0: [PDZ]₁ = 0.

2. The equation for calculating the PDZ concentration at [PDZ]_n data point is.

   $${[\text{PDZ}]}_{\text{n}}=\left(\left({[\text{PDZ}]}_{n-1}{\times }^{\text{tot}}{V}_{n-1}\right)+\left({[\text{PDZ}]}_{\text{stock}}{\times }^{\text{added}}{V}_{\text{stock}}\right)\right){/}^{\text{tot}}{V}_{\text{n}}$$ (1)

   where ^totV_n is the total volume added over n titration steps, ^totV_n-1 is total volume of PDZ protein added at the n₋₁ titration point, and ^addedV_stock is the volume added of stock PDZ domain ([PDZ]_stock) (i.e., ^totV_n = ^totV_n-1 + ^addedV_stock summed over n titration steps). This calculation can be conveniently performed in spreadsheet software (e.g., Microsoft Excel).

3. Baseline correction: The anisotropy value of the first data point (A₁ at [PDZ]₁ = 0) is subtracted from the anisotropy value of each subsequent data point (A_n at [PDZ]_n) to obtain the corrected anisotropy at each PDZ concentration (^corrA_n = A_n − A₁). Figure 2 shows an example of processed data computed in a spreadsheet program.

Fig. 2.

Processed data of the CASK PDZ–SDC1 PBM binding assay. The concentration of CASK PDZ domain is indicated. The column labeled “Stock Conc.” shows the concentration of the PDZ stock used for the collection of each data point. The column labeled “Conc.” indicates the concentration of PDZ domain sample in the cuvette for each data point. The “Base line corr.” column is the anisotropy value after baseline correction (^corrA)

3.5. Data Analysis and Binding Curve Presentation

1. The binding curves are fit to a standard hyperbolic binding model:

   $${}^{\text{corr}}A=\left[\frac{B_{\text{max}}[\text{PDZ}]}{K_{\text{d}}+[\text{PDZ}]}\right]$$ (2)

   where ^corrA is the corrected anisotropy at each titration step, B_max is the maximum anisotropy at PDZ domain saturation, K_d is the dissociation constant, and [PDZ] is the total concentration of the PDZ domain in solution. SigmaPlot (Systat Software Inc., CA) was used to determine B_max and K_d by fitting the binding data to Eq. 2 using nonlinear regression analysis (see Note 12) [6, 15]. The K_d of each PDZ–PBM pair is measured in triplicate and reported as the mean and standard error of the mean.

2. Each data point is normalized to the fitted B_max for graphical presentation of multiple binding curves in a single plot. Figure 3 shows the presentation of binding curves for CASK PDZ– and Scribble PDZ1–ligand binding reactions.

3. The Gibbs free energy of binding (ΔG_b) is calculated by

   $$\Delta G_{\text{b}}=-\text{RT}*\text{ln}\left({\text{K}}_{\text{d}}\right)$$ (3)

   where R is the universal gas constant and T is the given experimental temperature. The error in free energy can be obtained by propagation of the error in K_d.

Fig. 3.

PDZ–PBM binding curves and energetics. (a) Representative binding curves for PDZ–PBM interactions. The Caspr4 binding curve is an example of negative control C-terminal peptide that does not bind (i.e., N.B., no binding) the CASK PDZ domain. The CASK PDZ domain binds C-terminal NRXN1 and SDC1 peptides. The Scribble PDZ1 domain binds an internal peptide derived from SGEF. (b) Dissociation constants and Gibbs free energy of binding for several PDZ–PBM interactions. The reported dissociation constants are the average and standard error derived from at least three independent experiments

4. Notes

1. The dansyl fluorophore can affect binding affinity by directly interacting with PDZ domains [7, 8]. For relative affinity measurements this may not be an issue. PDZ domain–fluorophore interactions can be minimized by using other fluorophores or longer peptides [15].

2. Both the peptide concentration and slit-width can be adjusted to optimize signal-to-noise ratio. However, the concentration of peptide should be kept ~tenfold lower than the K_d to obtain reliable fits of the data.

3. The peptide ligand can be either a C-terminal or internal ligand derived from the target full-length protein. For C-terminal ligands, generally 6–8 terminal residues are used but additional residues may be involved, and this should be determined empirically. Internal PBM sequences can also be used [16]. Again, the exact residues will vary for each interaction and should be determined empirically. Internal ligands also should have their N-terminus acetylated to mimic an internal protein sequence by avoiding an electrostatic dipole contribution (see [15] for additional details). In addition, the C-terminus should be amidated to neutralize the free carboxylate group.

4. Peptides are typically purified via high-performance liquid chromatography purification using acidic buffers (e.g., trifluoracetic acid) prior to lyophilization. Thus, care should be taken to adjust the pH of the buffer solution upon solubilizing the peptide. We use a benchtop pH meter equipped with microelectrode for small volume samples.

5. The concentration of peptide and protein in solution can be measured by UV absorbance at 280 nm wavelength using a spectrophotometer. The extinction coefficient can be calculated from the amino acid sequence (e.g., ExPASy—Prot-Param) [19, 20]. A fluorophore can contribute to the 280 nm wavelength absorbance. However, the excitation and emission wavelengths of the dansyl fluorophore are more than 10 nm away from 280 nm (λ_ex = 340, λ_em = 550); thus, the fluorophore should not significantly affect the peptide concentration determination at the 280 nm wavelength [15]. If the peptide lacks amino acids with chromophores, one can use the extinction coefficient of dansyl chloride (ε ~4350 M⁻¹ • cm⁻¹) [21] to estimate the peptide concentration. Alternatively, one can use a color-based protein assay (e.g., bicinchoninic acid or Bradford assay, Thermo Scientific).

6. Dansylated peptide solutions should be kept away from light by either covering them with aluminum foil or using tinted (light free) microfuge tubes. Stock dansylated peptides are generally stored in small aliquots to avoid repetitive freeze and thaw cycles.

7. Using the PDZ protein sample immediately after purification is highly recommended. Long-term storage of proteins at 4 °C or at −20 °C can significantly affect protein stability. If storage is required, the integrity of the PDZ sample should be tested periodically with a known reference peptide.

8. The Hellmanex^™ III (Hellma, NY) cleaning concentrate can be used periodically to remove biological material from the surface of cuvettes followed by thorough rinsing with ddH₂O and ethanol using a vacuum cuvette washer.

9. When preparing buffers for sample preparation, filtering and degassing are highly recommended. Particulate matter and air bubbles scatter light and disrupt the spectroscopic measurements. We filter buffers with a 0.45 μm membrane using a vacuum filter unit attached to a dry vacuum system (Welch) followed by continuous stirring under vacuum for 15–30 min.

10. To prevent fluorophore bleaching by light exposure, it is highly recommended to cover the cuvette (containing the fluorophore-peptide) in the spectrofluorometer with the lid closed. The sample is now ready for the data acquisition. If necessary, the sample can be equilibrated for several minutes (typically 1–5 min) or until the measured anisotropy value is stable over time.

11. The number of data points can vary depending on the sample and affinity of the PDZ–ligand interaction. Data collection is complete when the anisotropy value plateaus (i.e., three consecutive data points have similar anisotropy values). However, we typically collect two or three additional data points after approaching the anisotropy plateau to ensure binding saturation has been reached.

12. The following assumptions are made when fitting the binding data. First, the PDZ–PBM binding stoichiometry is 1:1, which is true for all known PDZ–PBM interactions. Second, the concentration of free PDZ domain is on the order of the total PDZ domain concentration. Third, there is no significant change in fluorescence intensity (<10%) upon PDZ binding. We have not observed a significant change in fluorescence intensity in our experiments. Otherwise, correction factors should be applied [15].