Complexity associated with caprylate binding to bovine serum albumin: Dimerization, allostery, and variance between the change in free energy and enthalpy of binding

Abstract

Isothermal titration calorimetry (ITC), differential scanning calorimetry (DSC), and pressure perturbation calorimetry (PPC) were used to study different aspects of the diverse interaction between the fatty acid caprylate and bovine serum albumin (BSA). The ITC thermogram was consistent with exothermic binding to a single site on BSA, which was electrostatic but had little or no hydrophobic contribution. ITC revealed that small changes to solution conditions and temperature were associated with apparent enthalpy–entropy compensation, causing large changes in enthalpy (ΔH) during binding, but with little corresponding changes in free energy (ΔG). ITC also detected a slower endothermic interaction at a low mole ratio. Dynamic light scattering suggested that this was due to dimerization or similar self‐association. DSC demonstrated that further interactions took place at higher mole ratios. This was consistent with weak binding of caprylate to multiple binding sites which had a considerable impact on the structural conformation of BSA. PPC showed that the conformational change of BSA was accompanied with a reduction in surface hydrophobicity of the protein. PPC also demonstrated that in solution caprylate's hydrocarbon tail is hidden from water as no clathrate‐like water is evident, which is consistent with the lack of hydrophobic contribution during binding. Cumulatively, the three calorimetric techniques offer a comprehensive view of caprylate and BSA interactions, highlighting the role of electrostatic interaction in binding accompanied by probably dimerization and considerable structural change associated with weaker binding to BSA.

1. INTRODUCTION

Binding between macromolecules and smaller molecules or other macromolecules is an essential feature in living systems. While binding is often described in terms of simple interactions, such as hydrogen bonding, electrostatic interactions, and water displacement, this view overlooks the complexity inherent in many binding interactions. The association of a protein and ligand in an aqueous solution comprises multiple processes (Fox et al. 2015). These include changes associated with the ligand; (1) change in ligand conformation, and related dynamics, and (2) change in ligand hydration. Additionally, changes associated with the protein as the ligand–protein complex forms; (3) ligand–protein bonding through electrostatic and hydrogen bonding, (4) change in hydration at the binding site, (5) changes in protein conformation (allostery) and related dynamics, and (6) alterations in hydration associated with change in protein conformation. There are also possible protein–protein interactions resulting from the protein's conformational change. Calorimetric techniques provide insights into the binding event itself, structural changes in the protein, and changes in the hydration layers of both the protein and ligand.

At the close of the twentieth century, ultrasensitive calorimetry instruments became available. Isothermal titration calorimetry (ITC) is a commonly used technique for measuring binding through detecting the heat utilized or generated by the binding event by measuring the power required to maintain a constant temperature while titrating in one binding partner into a cell containing the other. ITC can also detect other exothermic or endothermic events, such as structural rearrangements of the macromolecule and reorganization of the hydration layers around the macromolecule and ligand. Early adoption of isothermal titration calorimetry (ITC) demonstrated its ability to determine the change in enthalpy (ΔH) from the area under the peaks. Additionally, the stoichiometry (n), association constant (K _a), and change in free energy (ΔG) could be derived from the thermogram curve shape, with entropy (ΔS) subsequently calculated (Cooper 1998; Freyer and Lewis ²⁰⁰⁸; Ladbury ²⁰⁰⁴). The applications for ITC have grown to include a vast range of binding interactions and include studying reaction kinetics involving heat change (Falconer et al. 2021).

The advent of ultrasensitive differential scanning calorimetry (DSC) enabled the study of protein thermal unfolding, allowing measurement of the temperature of maximum unfolding temperature (T _M), enthalpy (ΔH), and heat capacity changes (ΔC _p) associated with unfolding (Plotnikov et al. 1997). Protein conformational changes can be studied when they produce measurable effects on T _M and ΔH of unfolding. DSC can also be used to calculate changes in free energy (ΔG) and the activation energy of unfolding (Falconer 2018). A DSC study of the thermal stabilization of defatted human serum albumin (HSA) by n‐alkyl fatty acid anions demonstrated that hydrocarbon chain length of fatty acids significantly affects the T _M and ΔH values during unfolding, suggesting significant conformational change of HSA initiated by fatty acid binding (Shrake et al. 2005).

Pressure perturbation calorimetry (PPC) is a rarely used calorimetric technique that has been available since the year 2000 (Cooper et al. 2001). Historically, PPC has been utilized to calculate the apparent molar expansivity and thermal expansion coefficient of macromolecules in solution (Kamerzell et al. 2008; Lin et al. ²⁰⁰²; Pandharipande and Makhatadze ²⁰¹⁵; Ravindra and Winter ²⁰⁰⁴). Its most valuable application has likely been the study of solute–water interaction. The theoretical basis for this approach was established by Loren Hepler (Hepler 1969), which was used to disprove the structure maker–breaker explanation for Hofmeister series salts and the effects of organic chaotropes and osmolytes on protein stability (Batchelor et al. 2004). It has been demonstrated that PPC can detect both clathrate‐like structures around hydrophobic moieties on small organic molecules and the disruption of tetrahedral water by most inorganic and organic solutes (Bye et al. 2017; Bye and Falconer ²⁰¹⁵; Toronjo Urquiza et al. ²⁰²⁴).

The interaction between bovine serum albumin (BSA) and the fatty acid caprylate has been studied using a variety of techniques. BSA and closely related HSA were among the earliest protein structures studied (Brown 1975; Carter and Ho ¹⁹⁹⁴). X‐ray crystallography and nuclear magnetic resonance (NMR) have previously been applied to examine the fatty acid binding sites (Bhattacharya et al. 2000; Curry et al. ¹⁹⁹⁸; Hamilton ²⁰⁰²; Krenzel et al. ²⁰¹³; Sarver et al. ²⁰⁰⁵; Simard et al. ²⁰⁰⁵; Sugio et al. ¹⁹⁹⁹). While NMR has effectively been applied to study caprylate binding to serum albumin (Cisticola et al. ^1987a; Cisticola et al. ^1987b; Hamilton ¹⁹⁸⁹), X‐ray crystallography approaches have been less productive due to the difficulty in crystallizing serum albumin with caprylate (Bhattacharya et al. 2000). DSC has also been employed to study interactions between fatty acids of varying sizes with HSA (Carter and Ho 1994). Caprylate is thought to bind to fatty acid binding site 2 in HSA (Bhattacharya et al. 2000). This site is located between subdomains IA and IIA and is well enclosed within the protein. The carboxyl group forms hydrogen bonds with amino acid sidechains Y150, R257, and S287. Site 2 also contains an arginine (R257) that interacts with the fatty acids' carboxyl group near the hydrophobic channel (Hamilton 2002). Both BSA and HSA possess a fatty acid binding site 2, with high level of sequence homology (Bujacz 2012). Allosteric changes occur in serum albumins during fatty acid binding (Fang et al. 2006; Oleszko et al. ²⁰¹⁸), and both HSA and BSA are known to self‐associate, forming dimers (Levi and Gonzalez Flecha ²⁰⁰²; Squire et al. ¹⁹⁶⁸). The hydration layer around BSA has also been subject to study using terahertz and sub‐terahertz frequency spectroscopy, revealing detectable changes to the water 15 and 25 Å beyond the protein surface, respectively (Bye et al. 2014; Sushko et al. ²⁰¹⁵). These combined studies make serum albumins among the most well‐understood proteins.

In this paper, we used a well characterized binding interaction between caprylate (octanoate) and BSA to interrogate how ITC thermograms are interpreted, particularly focusing on enthalpy–entropy compensation, which will be further discussed. We also employed three calorimetry techniques ITC, PPC, and DSC to demonstrate how each calorimetric technique provides unique insight on binding, dimerization and allostery in addition to changes in the hydration of both caprylate and BSA.

2. MATERIALS AND METHODS

2.1. Materials

Fatty acid free BSA, sodium caprylate, sodium chloride (NaCl), sodium iodide (NaI), ultrapure water (for HPLC, <0.0003% residue on evaporation) were sourced from Merck (Darmstadt, Germany).

2.2. Isothermal titration calorimetry

ITC experiments were performed using a MicroCal PEAQ‐ITC calorimeter (Malvern Instruments Limited, Worcestershire, UK) with 5 mM sodium caprylate solution as the titrant and solutions containing BSA in the cell sample.

Three 350 μM fatty acid free BSA stock solutions were prepared with ultrapure water, 1.75 mM NaCl solution, and 1.75 mM NaI solution as the solvents, self‐buffering pH 4.4 ± 0.1. Small molecules in the fatty acid free BSA were reduced with two solvent changes using a 5–20 mL Pierce™ Protein Concentrator PES with a 3 k molecular weight cut‐off (Thermo Scientific, UK). Centrifugation was performed using an Eppendorf 5920 R Centrifuge (Hamburg, Germany) set at 4°C and 10,000 rpm. The fatty acid free BSA solutions were returned to a volume of 20 mL using ultrapure water, 1.75 mM NaCl, and 1.75 mM NaI solutions.

Fatty acid free BSA dissolved in different solvents (water, NaCl solution, and NaI solution) were utilized for the first set of experiments as the cell sample. The ITC's reference cell contained degassed ultrapure water. The titrations for the first set of experiments were completed at a temperature of 25°C, high feedback, reference power of 41.9 μW, and stirrer speed of 500 rpm. There were 19 injections with the first injection adding 0.4 μL of titrant for 0.8 s into the cell after a delay of 60 s, and remaining injections added 2 μL of titrant for 4 s. An equilibration time of 210 s for each injection was also set to attain a good baseline and separation between injections. For the second set of experiments, fatty acid free BSA solutions dissolved in ultrapure water was titrated with sodium caprylate at different temperatures. All parameters were unchanged from the first set of experiments with exception of both temperature and contents of the cell. Titration temperatures were set to 15 and 35°C. All experiments were performed in triplicate and caprylate‐free control titrations conducted.

The MicroCal PEAQ‐ITC Analysis Software version 1.41 (Malvern Instruments Limited, Worcestershire, UK) was used to fit the injection data from the experiments. The area under the peak for each injection was determined against the molar ratio of the titrant and protein in the cell (mol titrant/mol protein), providing the enthalpy change associated with the introduction of titrant into the sample. The software used an iterative Levenberg‐Marquadt algorithm to fit the enthalpy–molar ratio curve and calculate the number of sites (n), dissociation constant (K _D), enthalpy change associated to binding (ΔH), change in Gibbs free energy (ΔG), and entropy change (−TΔS) by fitting a one site model.

2.3. Pressure perturbation calorimetry

PPC measurements were obtained using a capillary Nano‐DSC (TA Instruments, New Castle, DE). Samples were degassed under vacuum for 15 min at 20°C to prevent bubble formation during the scan. Heat effects were measured in 0.3 mL samples during alternating pressure pulses from 1 to 4 bar, then from 4 to 1 bar, each pressure change occurred at 1°C intervals. A heating rate of 0.1°C min⁻¹ was used from 10 to 95°C. NanoAnalyze software (TA Instruments, New Castle, DE) was used to calculate the difference in heat supplied to the sample and the reference cell (ΔQ _exp). The difference in the change in heat capacity with respect to changes in pressure at constant temperature between the sample and ultrapure water (Δ(∂C _P/∂P)_T) was calculated using the slope of the ΔQ _exp versus temperature data. The strategy for operating pressure perturbation calorimetry using a rate of 0.1°C min⁻¹ has been validated to match those generated under isobaric conditions (Dragan et al. 2009). The BSA concentration was 1 mg mL⁻¹ (14.4 μM) in pure water, pH 4.4 ± 0.1. The sodium octanoate concentration was varied.

2.4. Differential scanning calorimetry

DSC measurements were obtained using a capillary Nano‐DSC (TA Instruments, New Castle, DE). The BSA concentration was 1 mg mL⁻¹ (14.4 μM) in pure water, pH 4.4 ± 0.1. The sodium octanoate concentration was varied. Samples were degassed under vacuum for 15 min at 20°C to prevent bubble formation during the scan. The scan rate was 1.5°C min⁻¹ from 10 to 105°C. NanoAnalyze software (TA Instruments, New Castle, DE) was used analyze the data.

2.5. Dynamic light scattering

The dynamic light scattering (DLS) experiments were conducted using a Zetasizer Ultra Red Label with ZS Xplorer software version 3.2.1.11 (Malvern Panalytical, UK). Four clear sided polystyrene cuvettes (SARSTEDT AG, Nümbrecht, Germany) were used to hold the samples. Measurements were taken using 173° back scatter. The BSA concentration was 1 mg mL⁻¹ (14.4 μM) in pure water, pH 4.4 ± 0.1. The sodium octanoate concentration was varied. Samples were degassed under vacuum for 15 min at 20°C to minimize microbubbles from the sample.

3. RESULTS AND DISCUSSION

3.1. Isothermal titration calorimetry

The ITC thermogram of sodium caprylate titrated into fatty acid free BSA (Figure 1) indicates that binding is predominantly an exothermic interaction, with a competing slower endothermic interaction observed in the first four injections. The change in enthalpy, plotted against caprylate to BSA molar ratio, is shown in Figures 2a and 3a. Excluding the first four injections (excluding six injections at 35°C), the data can be explained using a one‐site independent binding model. This model assumes that n caprylate ligands bind per BSA macromolecule with identical thermodynamic values. The stoichiometry and thermodynamic values for caprylate binding to BSA are presented in Tables 1 and 2.

FIGURE 1.

(a) Isothermal titration calorimetry thermogram of sodium caprylate titrated into fatty acid free bovine serum albumin (BSA) in pure water. (b) Focus on the first five injections highlighting the downward exothermic binding interaction and the slower upward endothermic interaction seen in the first three injections.

FIGURE 2.

Isothermal titration calorimetry of caprylate titration into fatty acid free bovine serum albumin in no salt (x), 1.75 mM NaCl (+) and 1.75 mM NaI (○) at 25°C. (a) The area under each peak from the thermogram was integrated (symbols) and one‐site independent binding model fitted using the data after c. 0.8M ratio (line). (b) The endothermic interaction (the difference between the experimental data and the fitted binding model).

FIGURE 3.

Isothermal titration calorimetry of caprylate titration into fatty acid free bovine serum albumin in pure water at 15°C (▼), 25°C (+), and 35°C (▲). (a) The area under each peak from the thermogram was integrated (symbols) and one‐site independent binding model fitted using the data after c. 0.8M ratio (line). (b) The endothermic interaction (the difference between the experimental data and the fitted binding model).

TABLE 1.

Thermodynamic values derived from ITC analysis of caprylate binding to fatty acid free BSA (15 μmol) in the presence of pure water, sodium chloride (1.75 mmol), and sodium iodide (1.75 mmol).

	N	ΔH (kJ mol−1)	−TΔS (kJ mol−1)	ΔG (kJ mol−1)	K D ×10−6 (mol)
25°C	1.22 ± 0.01	−56.0 ± 0.9	30.0	−25.8 to −26.1	28.6 ± 1.4
25°C NaCl	1.32 ± 0.01	−46.1 ± 0.4	20.2	−25.7 to −25.8	30.4 ± 0.9
25°C NaI	1.55 ± 0.01	−30.7 ± 1.0	5.0	−25.4 to −26.0	31.8 ± 3.5

TABLE 2.

Thermodynamic values derived from ITC analysis of caprylate binding to fatty acid free BSA in pure water at different temperatures.

	N	ΔH (kJ mol−1)	−TΔS (kJ mol−1)	ΔG (kJ mol−1)	K D ×10−6 (mol)
15°C	1.34 ± 0.01	−33.9 ± 0.6	7.3	−26.4 to −26.8	15.3 ± 1.1
25°C	1.22 ± 0.01	−56.0 ± 0.9	30.0	−25.8 to −26.1	28.6 ± 1.4
35°C	1.16 ± 0.07	−112.2 ± 23.6	85.6	−26.1 to −26.7	34.0 ± 3.7

A solution containing 1.75 mM sodium halide (chloride or iodide) and 0.35 mM BSA corresponds to 5 halide ions per BSA molecule. At pH 4.4, the BSA has a net charge of approximately +1 (based on BSA's amino acid sequence, UniProt P02769) so 1.75 mM sodium halide will provide enough halide ions for ion pairing with positively charged sidechains and will negate the BSA's net charge.

The presence of 1.75 mM sodium chloride had a significant effect on the ΔH, reducing the exothermic ΔH, but had no discernible effect on ΔG (Figure 2a). This change in ΔH is evidence of ion pairing between charged arginine (R257) sidechain on the BSA and the chloride anion. Despite the anion's presence, the interaction between caprylate and BSA's binding site, which involves an electrostatic component, remained unaffected. The change in the ΔH due to the presence of the anion associated with the arginine (R257) may be due to two causes. Neutralization of the lysine charge by the anion could induce rearrangement of the BSA's structure, and the neutralization of the arginine charge by the anion would reduce the electrical field associated with the arginine (R257) and the associated modification by the electrical field of the water structure due to electrostriction. The role of the arginine (R257) sidechain on the BSA in fatty acid binding was originally determined using ¹³C NMR and validated by crystal structures (Hamilton 2002).

The presence of 1.75 mM sodium iodide in the 0.35 mM BSA solution had a significant effect on the ΔH, reducing the exothermic ΔH more than sodium chloride, yet had no discernible effect on ΔG (Figure 2a). Iodide, a low‐density anion, is well known to associate with hydrophobic pockets and moieties on macromolecules (Gibb and Gibb 2011; Rogers et al. ²⁰²²; Sokkalingam et al. ²⁰¹⁶; Sullivan et al. ²⁰¹⁸). If the iodide anion were associating with a hydrophobic pocket, such as fatty acid binding site 2 on BSA, a reduction in the endothermic component of the binding thermogram would be expected, however, this was not observed. This suggests that fatty acid binding site 2 may not be accessible to iodide, despite the minimal native structure of the fatty acid free BSA. Iodide anions have been shown to be more effective at charge screening than smaller chloride anions (Zhang and Cremer 2009). This is related to the iodide larger size and the weaker attraction of water to the anion enabling more effective interaction and screening of an opposite charged species, in this case the arginine (R257) sidechain.

Hydrophobic interaction does not appear to have played a significant role in caprylate binding to BSA. The ΔH became more strongly exothermic as the temperature was increased (Figure 3a), while the ΔG was unaffected. Typically, hydrophobic interaction becomes more endothermic as temperature rises and the ΔG is temperature‐dependent, favoring binding at higher temperatures. This observation aligns with crystallographic and NMR studies, where the caprylate initially attracted to the basic amino acid at the mouth of the fatty acid binding site followed by the hydrocarbon chain slipping into the hydrophobic groove (Hamilton 2002).

The change in enthalpy (ΔH) plotted against change in entropy (TΔS) for caprylate binding to BSA at different temperatures and in salt free or 1.75 mM sodium chloride of 1.75 mM sodium iodide solutions (Figure 4). This suggests there is an apparent linear relationship between enthalpy and entropy. This is an example of the enthalpy–entropy compensation discussed later in this paper.

FIGURE 4.

The change in enthalpy (ΔH) plotted against temperature times change in entropy (TΔS) for octanoate binding to fatty acid free bovine serum albumin at 15°C (▼), 25°C (x), 35°C (▲) in pure water and at 25°C with NaCl (+) and 25°C with NaI (○) showing enthalpy–entropy compensation. Each data point is the calculated ΔH and TΔS for an individual ITC run, each condition was tested in triplicate.

The thermogram (Figure 1b) shows a clear exothermic peak followed by a smaller endothermic peak in the first three injections. This indicates a strong exothermic binding event followed by a slower endothermic reaction. Binding events with exothermic and endothermic components are not unique and have been observed with a mixed population of tannins and polyproline, where the hydrophobic interaction was endothermic and the hydrogen bonding was exothermic (McRae et al. ²⁰¹⁰). Another example is the interaction between phytate and lysozyme, where phytate binding was exothermic, but the phytate also crosslinked the protein, displacing water in an endothermic reaction (Darby et al. 2017).

The endothermic component of binding is clearly illustrated if the ΔH values calculated using the one‐site independent binding model are subtracted from the experimental ΔH values (Figures 2b and 3b). The stoichiometry is approximately 0.5, indicating one caprylate per two BSA molecules. A plausible explanation is that caprylate binding to BSA promotes the association of BSA with other BSA molecules, including fatty acid free BSA molecules. BSA is known to dimerize (Levi and Gonzalez Flecha ²⁰⁰²; Squire et al. ¹⁹⁶⁸), a process that displaces water and is evidently endothermic. While it is unclear whether the ΔG of this association was influenced by the presence of the sodium halide salts, the ΔH values were demonstrably affected. Under the assumption that dimerization is a weak association where the ΔH is primarily due to water displacement, the effect of the salts and temperature likely altered the structure of the displaced water as dimerization took place. DLS measurement of fatty acid‐free BSA calculated the hydrodynamic diameters to be 5.48 ± 0.11 nm and BSA with 30 μM caprylate (2 caprylate molecules per BSA) the hydrodynamic diameter was 6.98 ± 0.22 nm (Figure 5). The addition of 2 caprylate molecules per BSA caused a 1.6‐fold increase in particle volume. This is consistent with weak dimerization where the association constant is close to 1. At 150 μM caprylate (10 caprylate molecules per BSA), the hydrodynamic diameter was 8.53 ± 0.30 nm, which is equivalent to 2.4‐fold increase in particle volume, and at 750 μM caprylate (50 caprylate molecules per BSA) the hydrodynamic diameter was 7.91 ± 0.16 nm, which is equivalent to 2.1‐fold increase in particle volume. The hydrodynamic diameter at 750 μM caprylate is consistent with published value of 7.8 nm for BSA that was not subject to fatty acid removal and dissolved in pure water (Zhang et al. 2015). The hydrodynamic diameters (Figure 5) are consistent with fatty acid free BSA being in a monomeric form, BSA in 1.5 mM caprylate (100 caprylates per BSA) being a dimer, and BSA at caprylate concentrations above 30 μM and below 1.5 mM undergoing dimerization and possibly some non‐specific self‐association.

FIGURE 5.

Hydrodynamic diameter of bovine serum albumin (BSA) with different ratios of octanoate to BSA, measured using dynamic light scattering. The DLS experiments were conducted using a Zetasizer Ultra Red Label (Malvern Panalytical, UK) using 173° back scatter.

3.2. Pressure perturbation calorimetry

PPC provides unique insights into the interactions between small molecules and macromolecules with water, detecting changes in hydrogen bonding in the water around the solute (Batchelor et al. 2004; Bye and Falconer ²⁰¹⁵; Hepler ¹⁹⁶⁹). Increased hydrogen bonding, indicative of structures such as clathrate‐like structures around hydrophobic moieties, can be seen as a decrease in ΔQ _exp with an increase in temperature, which can be presented as a negative Δ(∂C _P/∂P)_T. Conversely, an increase in ΔQ _exp with an increase in temperature, or a positive Δ(∂C _P/∂P)_T value, indicates a decrease in water structure. These values reflect the average for the interactions between water and the solute.

PPC analysis of caprylate in water suggests that clathrate‐type water cannot be detected by this method (Figure 6). The steady rise in the ΔQ value is consistent with a positive Δ(∂C _P/∂P)_T value, which in turn is indicative of caprylate disrupting the tetrahedral structure of water. A negative Δ(∂C _P/∂P)_T value would have been consistent with clathrate‐like water formation around the hydrocarbon chain (Bye and Falconer 2015). This implies that caprylate forms clusters in solution, reducing the exposure of hydrocarbon tails to water, even at concentrations well below the micelle concentration (Stanley et al. 2009). Previous work on small organic molecules in water have demonstrated that most molecules do not readily form clathrate‐type water but instead self‐associate to reduce hydrophobic moieties exposure to water (Bye and Falconer 2015). Alcohols are an exception to this rule, forming clathrate‐type water at low alcohol concentrations (Bye et al. 2017).

FIGURE 6.

Δ(∂C_P/∂P)_T measured for sodium octanoate‐water mixtures using pressure perturbation calorimetry with a pressure rise from 1 to 4 bar, for the temperature range of 15–35°C plotted against the mole fraction; demonstrating no negative Δ(∂C_P/∂P)_T even at very low sodium octanoate concentrations. Negative Δ(∂C_P/∂P)_T is indicative of clathrate‐like water formation around hydrophobic moieties.

PPC of fatty acid free BSA is shown in Figure 7. The fatty acid free BSA displays a slow decrease in ΔQ with an increase in temperature. This is consistent with some hydrophobic moieties in the fatty acid free BSA being exposed to water causing a negative Δ(∂C _P/∂P)_T. The ΔQ values for BSA in 10 mM caprylate (Figure 8) do not have an apparent slope, suggesting the hydrophobic moieties observed in fatty acid free BSA have been hidden as the BSA forms in the presence of caprylate.

FIGURE 7.

Pressure perturbation calorimetry of bovine serum albumin with 0 mM caprylate. The PPC used a series of pressure changes between 1 and 4 bar (lower line). The difference in power is plotted against temperature. The BSA concentration was 1 mg mL⁻¹ and the scan rate was 0.1°C min⁻¹.

FIGURE 8.

Pressure perturbation calorimetry of bovine serum albumin with 10 mM caprylate. The PPC used a series of pressure changes between 1 and 4 bar (lower line). The difference in power is plotted against temperature. The BSA concentration was 1 mg mL⁻¹ and the scan rate was 0.1°C min⁻¹.

3.3. Differential scanning calorimetry

Combining ITC and DSC experimentation is not a new approach (Jelesarov and Bosshard 1999). In this study, ITC and DSC provide markedly different insights into the interaction between caprylate and BSA.

The DSC thermogram for fatty acid free BSA (Figure 9) shows a low, broad peak with a T _M of 66.7°C. Both ΔH and ΔC _p were too small to be accurately measured, indicative of the BSA having little secondary and tertiary structure. At the ratio of 2 caprylates to 1 BSA (30 μM caprylate) the DSC thermogram is similar to that of fatty acid free BSA. There were low broad peaks with T _M 56.2°C and 70.9°C. The ΔH and ΔC _p were too small to be accurately measured. While the ITC data is consistent with caprylate binding and subsequent dimerization, DSC only detects a trivial ΔH in the DSC thermogram. At the ratio of 10 caprylates to 1 BSA (150 μM caprylate) there were two broad peaks with T _M 64.0°C and 72.2°C, a total ΔH 780 kJ mol⁻¹ and ΔC _p 2.2 kJ °C⁻¹ mol⁻¹, suggestive of significant secondary and/or tertiary structure forming. At the ratio of 50 caprylates to 1 BSA (750 μM caprylate) there was a single peak with a T _M 72.3°C and ΔH 850 kJ mol⁻¹. The peak with a T _M 64.0°C seen at 10:1 ratio has disappeared. The ΔC _p was 3.1 kJ °C⁻¹ mol⁻¹. At the ratio of 665 caprylates to 1 BSA (10 mM caprylate), there was a pronounced single peak with a T _M of 84.5°C and a ΔH 1240 kJ mol⁻¹, with a ΔC _p of 5.0 kJ °C⁻¹ mol⁻¹. This indicates that BSA is clearly structurally more stable at a higher caprylate concentration than at lower caprylate concentrations, which is indicative of the BSA having pronounced secondary and tertiary structure.

FIGURE 9.

Differential scanning calorimetry thermograms for fatty acid free bovine serum albumin (15 μM BSA) with 0 μM (blue line), 30 μM (green line), 150 μM (orange dot dash line), 750 μM (red line), and 10 mM (black line) sodium octanoate. The fatty acid free BSA was dissolved in pure water, 1 mg mL⁻¹, pH 4.4, and the scan rate was 1.5°C min⁻¹.

The failure of caprylate at low concentrations to stabilize the BSA is consistent with crystallography studies, where inability to crystallize was attributed to the hydrocarbon tail of caprylate being too short to establish stabilizing contact with the IA subdomain (Bhattacharya et al. 2000).

The DSC results suggest that ITC is only detecting the first binding event between caprylate and BSA, with weaker binding occurring at the other eight potential binding sites located in BSA. These weak binding events, characterized by a low ΔG, which resulted in small even peaks in the ITC thermogram, were not distinguishable from the reference. DSC is able to detect a clear change in both T _M and ΔH associated with the weak binding events. As the BSA is further stabilized with 150 μM, 750 μM, and 10 mM caprylate, a measurable ΔC _p value was detected, consistent with the exposure of the hydrophobic interior of BSA during thermal unfolding of the structure.

3.4. Fatty acid binding to serum albumin

PPC analysis of caprylate in solution reveals that the fatty acid does not remain isolated in water (under normal circumstances), where an energetically unfavorable clathrate‐like water structure would form around the hydrocarbon tail. Instead, caprylate evidently forms mesoscopic clusters, reducing hydrocarbon–water contact. The ITC of caprylate into fatty acid free BSA suggests the interaction has an electrostatic component, but little or no evidence of endothermic water displacement from the caprylate or the BSA, and follows the sequence shown in Figure 10 where caprylate singletons are transitory, not contributing to the thermogram.

FIGURE 10.

Proposed sequence of events during octanoate binding to bovine serum albumin. Note that octanoate binding to BSA is followed by conformational change and dimerization of the protein.

PPC analysis of BSA in the unfolded, fatty acid free state shows some exposed hydrophobic surfaces, as evidenced by a negative Δ(∂C _P/∂P)_T value, which disappears when folded in the presence of 10 mM caprylate. DSC analysis confirms this observation, as indicated by a positive ΔC _p value upon unfolding of the BSA in the presence of 150 μM or more caprylate.

The perspectives offered by ITC and DSC data on caprylate binding to BSA differ significantly. ITC detects the relatively strong binding between caprylate and BSA fatty acid binding site 2 but fails to detect any interaction between caprylate and other binding sites. In contrast, DSC detected BSA structural changes induced by caprylate binding. From the perspective of BSA structural change, the binding of caprylate to fatty acid binding site 2 had little impact on BSA structure. However, binding at the remaining fatty acid binding sites, though potentially weaker, resulted in significant structural changes in BSA. Conformational changes in serum albumin during fatty acid binding are well‐documented (Fang et al. 2006; Oleszko et al. ²⁰¹⁸). For instance, FTIR studies of palmitate (C₁₆H₃₂O₂) interaction with HSA detected significant secondary structural changes upon palmitate binding (Oleszko et al. 2018).

ITC in the presence of sodium chloride and sodium iodide suggests that binding of the fatty acid to the binding site had an electrostatic component, which is consistent with the carbonyl group on caprylate interacting with the arginine (R257) sidechain which was observed by NMR (Hamilton 2002). ITC suggests fatty acid in solution has minimal hydrocarbon chain–water contact, and that fatty acid binding to the site in BSA does not include significant displacement of water, consistent with the fatty acid binding site being closed off to water. This suggests the hydrophobic pocket in BSA does not have an associated “hydration layer” and is effectively closed to surrounding water (Hamilton 2002). NMR consistent with the ITC, did not detect the weak binding of caprylate to other binding sites.

The combined use of PPC, ITC and DSC offer a more comprehensive understanding of caprylate binding to BSA, highlighting that using ITC or DSC alone can provide a limited and potentially misleading perspective on this interaction.

3.5. Enthalpy–entropy compensation

Enthalpy–entropy compensation has been a subject of considerable debate regarding its reality, significance, and potential molecular origins (Chodera and Mobley 2013; Dragan et al. ²⁰¹⁷; Dunitz ¹⁹⁹⁵; Fox et al. ²⁰¹⁸; Lumry and Rajendra ¹⁹⁷⁰; Sharp ²⁰⁰¹).

There was apparent enthalpy–entropy compensation during caprylate binding to BSA, both in the presence of low salt concentrations (1.75 mM of sodium chloride or sodium iodide) and at different temperatures (Figure 4). Neither the change in salts nor temperature produced a measurable change in ΔG, but both caused measurable changes in the measured ΔH. Given that he TΔS value is calculated from the equation ΔG = ΔH – TΔS, it inherently remains equal to the ΔH.

The authors' interpretation of the apparent enthalpy–entropy compensation observed during caprylate binding to BSA is as follows: (1) The ΔH measured by ITC consists of multiple components these can be exothermic and endothermic; and the experimentally derived ΔH being the sum of the components. (2) The experimentally derived ΔH is not directly related to the ΔG of binding. (3) The component of the ΔH that is unrelated to the ΔG of binding, may be due to change in the structure of water around the protein, changes in water structure at the ligand binding site, or alterations in internal bonding within the protein. (4) The modulation of the ΔH by factors such as 1.75 mM sodium chloride, 1.75 mM sodium iodide, or changes in temperature are not directly related to the ΔG of binding, but causes the calculated TΔS to adjust accordingly, thus creating an apparent but artificial enthalpy–entropy compensation.

AUTHOR CONTRIBUTIONS

Marc Joseph A. Capili: Investigation; writing – review and editing; methodology; formal analysis; data curation. Sophie K. Oerlemans: Writing – review and editing; visualization; data curation; writing – original draft. Leah Wright: Supervision; writing – review and editing; visualization; validation; methodology. Robert J. Falconer: Supervision; conceptualization; investigation; funding acquisition; writing – original draft; writing – review and editing; methodology; formal analysis.

Capili MJA, Oerlemans SK, Wright L, Falconer RJ. Complexity associated with caprylate binding to bovine serum albumin: Dimerization, allostery, and variance between the change in free energy and enthalpy of binding. Protein Science. 2024;33(12):e5224. 10.1002/pro.5224