Effect of Tetraphenylborate on Physicochemical Properties of Bovine Serum Albumin

The binding interactions of bovine serum albumin (BSA) with tetraphenylborate ions ([B(Ph)4]−) have been investigated by a set of experimental methods (isothermal titration calorimetry, steady-state fluorescence spectroscopy, differential scanning calorimetry and circular dichroism spectroscopy) and molecular dynamics-based computational approaches. Two sets of structurally distinctive binding sites in BSA were found under the experimental conditions (10 mM cacodylate buffer, pH 7, 298.15 K). The obtained results, supported by the competitive interactions experiments of SDS with [B(Ph)4]− for BSA, enabled us to find the potential binding sites in BSA. The first site is located in the subdomain I A of the protein and binds two [B(Ph)4]− ions (logK(ITC)1 = 7.09 ± 0.10; ΔG(ITC)1 = −9.67 ± 0.14 kcal mol−1; ΔH(ITC)1 = −3.14 ± 0.12 kcal mol−1; TΔS(ITC)1 = −6.53 kcal mol−1), whereas the second site is localized in the subdomain III A and binds five ions (logK(ITC)2 = 5.39 ± 0.06; ΔG(ITC)2 = −7.35 ± 0.09 kcal mol−1; ΔH(ITC)2 = 4.00 ± 0.14 kcal mol−1; TΔS(ITC)2 = 11.3 kcal mol−1). The formation of the {[B(Ph)4]−}–BSA complex results in an increase in the thermal stability of the alfa-helical content, correlating with the saturation of the particular BSA binding sites, thus hindering its thermal unfolding.


Introduction
In aqueous media, proteins undergo many physicochemical changes such as conformational alterations, denaturation, folding/unfolding processes, and ligand exchange [1][2][3][4][5][6]. These phenomena can be invoked by variations in the temperature and/or the pH of a solution. Consequently, the environmental conditions may affect the binding properties of a biomolecule, such as the affinity of low-molecular weight compounds to a protein, the number of potential binding sites, and the stoichiometry of the resulting protein-ligand complexes [7]. As a result, the biological and pharmacological activity of the protein is often highly influenced by the presence of the different ligands in the system. Generally, the nature and the concentration of some potential ligands present in the system, which are capable to compete with other compounds (for example drugs) to a biomolecule, may affect the binding properties and functions of a protein. In the field of physicochemical studies, this phenomenon is applied to the studying of the potential binding sites of BSA by competitive displacement assays [8,9]. There are many techniques used for studying the albumin-ligand interactions. However, due to the presence of two tryptophan residues in BSA, namely Trp-134 (subdomain IA) and Trp-213 (subdomain IIB) [10], fluorescence spectroscopy is the most exploited. An excellent review on the application of direct fluorescence titrations for studying serum albumin-ligand interactions provided by Macii Figure 1. The secondary structure of BSA (grey cartoon) with marked Trp residues (magenta VDW representation). The first and second potential binding sites of SDS are shown with red and green rings, respectively. In this work, we have exploited the previously gathered information on the SDS-BSA interactions for studying the binding interactions of tetraphenylborate ions ([B(Ph) 4 ] − ) with BSA. We have focused our attention on the [B(Ph) 4 ] − ions, as they represent the lowmolecular weight compounds with four, bulky hydrophobic phenyl moieties and a negative charge capable of binding through hydrophobic interactions and/or a combination of hydrophobic and electrostatic forces [21] (Figure 2). A set of some complementary methods, namely Isothermal Titration Calorimetry (ITC), steady-state fluorescence spectroscopy, Differential Scanning Calorimetry (DSC) and Circular Dichroism (CD) was employed for assessing the physicochemical nature of the interactions. The effect of the [B(Ph) 4 ] − ions binding on the protein structure was also discussed. Finally, new competition ITC experiments using SDS as a competitive ligand have been performed to find BSA binding sites of [B(Ph) 4

Isothermal Titration Calorimetry (ITC)
Thermodynamic parameters of the BSA interaction with [B(Ph) 4 ] − ions determined by the ITC technique in the cacodylate buffer solution of a pH 7, at 298.15 K, were subsequently compared with the results obtained under the same experimental conditions for the BSA-SDS system [20]. In both investigated cases, namely {[B(Ph) 4 ] − }-BSA and SDS-BSA, nonlinear least-squares procedures were applied for fitting isotherms to a model that assumes two sets of binding sites. The stoichiometry of the ligand-albumin complexes (N = [ligand]/[BSA]), binding constants (K ITC ) and the enthalpy change (∆H ITC ) were obtained directly from ITC measurements. The free energy of binding (∆G ITC ) and entropy change (∆S ITC ) were calculated using the standard thermodynamic relationships: ∆G ITC = −RTlnK ITC = ∆H ITC -T∆S ITC . The obtained parameters are so-called conditiondependent parameters and can be compared with those obtained under the same pH, temperature and the type of a buffer solution [22]. Representative binding isotherms for the investigated systems are shown in Figure 3, whereas conditional parameters of the interactions are summarized in Table 1. assumes two sets of binding sites. The stoichiometry of the ligand-albumin complexes (N = [ligand]/[BSA]), binding constants (KITC) and the enthalpy change (ΔHITC) were obtained directly from ITC measurements. The free energy of binding (ΔGITC) and entropy change (ΔSITC) were calculated using the standard thermodynamic relationships: ΔGITC = −RTlnKITC = ΔHITC-TΔSITC. The obtained parameters are so-called condition-dependent parameters and can be compared with those obtained under the same pH, temperature and the type of a buffer solution [22]. Representative binding isotherms for the investigated systems are shown in Figure 3, whereas conditional parameters of the interactions are summarized in Table 1.   [17].  (1) Literature data [17].
ITC data revealed that BSA has two binding sites capable of binding tetraphenylborate ions, as has been previously observed for SDS monomers ( Figure 3, Table 1). The simplified chemical equations of the ligand (SDS or [B(Ph) 4 ] − )-BSA complex formation which take into account the stoichiometry of the resulting species are given below: The first binding site The second binding site 5 SDS + (SDS) 2 -BSA = (SDS) 2 -BSA-(SDS) 5 logK (ITC)2 = 5.29 The first site of albumin binds two [B(Ph) 4 ] − ions followed by the saturation of the second site by four tetraphenylborate ions. The binding constants (K (ITC) ), and consequently the free energy of binding (∆G ITC ), for both SDS-BSA and [B(Ph) 4 ] − -BSA complexes are comparable to one another (in the range of experimental error) indicating their similar thermodynamic stability (Table 1). In contrast to the SDS-BSA interactions in which the charge-charge type or/and van der Waals interactions are involved in the ligand-protein complex formation (|∆H (ITC) | > |T∆S (ITC) |), the binding of [B(Ph) 4 ] − ions in the first site of the albumin is both enthalpy-and entropy-driven. However, the hydrophobic interactions seem to prevail (|∆H (ITC) | < |T∆S (ITC) |) [23].
A different scenario is observed for binding [B(Ph) 4 ] − ions in the second site. The positive value of ∆H (ITC)2 shows that the thermodynamic stability of the resulting complexes strongly depends on the entropy change, while the charge-charge type or/and van der Waals interactions are involved in the SDS binding to the second site (Table 1) (Table 1).     solution. This is related to the fact that under such experimental conditions, the total number of SDS monomers bound to the protein is lower, and the SDS monomers are bound to their second binding site, which is far from sensitive to SDS binding of the Trp-134 and Trp-213 residues [20].

Steady-State Fluorescence Spectroscopy
with the ITC results, and confirm the fact that upon saturation of BSA with [B(Ph)4] − the changes in the Trp fluorescence intensity are not as pronounced as for the [B(Ph)4] − -free solution. This is related to the fact that under such experimental conditions, the total number of SDS monomers bound to the protein is lower, and the SDS monomers are bound to their second binding site, which is far from sensitive to SDS binding of the Trp-134 and Trp-213 residues [20]. To obtain a better insight into the mechanisms of the interactions between BSA (and its mixtures with [B(Ph)4] − ) and SDS, the obtained results were analyzed according to a well-known Stern-Volmer equation: 1 1 , where F0 and F denote the fluorescence intensities in the absence and presence of a quencher (SDS), [Q] is the quencher concentration, KSV is the Stern-Volmer quenching constant, kq is the bimolecular quenching rate constant, while τ0 is the lifetime of the fluorophore (BSA) in the absence of quencher [25,26]. The graphs of versus [Q] plotted according to the Stern-Volmer equation are shown in Figure 6. Consequently, Table 2 presents the newly determined values of Stern-Volmer quenching constants along with linear correlation coefficients (R 2 ) and bimolecular quenching rate constants (kq) recovered for the studied systems. The latter ones were calculated based on the value of the average fluorescence lifetime τ0 of free BSA (in Tris-HCl, pH 7.0), which equals approximately 6.3 ns [27]. It can be observed that the quenching (binding) constants decrease with the increase in the [B(Ph)4] − concentration of the system, which can be assigned to the competition of SDS with [B(Ph)4] − in BSA [28,29]. Furthermore, the estimated values of kq for all systems are of the order 10 12 M −1 ·s −1 , which is approximately a few hundred times higher than the maximum value possible for the diffusion-controlled quenching-rate constant (2.0·10 10 To obtain a better insight into the mechanisms of the interactions between BSA (and its mixtures with [B(Ph) 4 ] − ) and SDS, the obtained results were analyzed according to a wellknown Stern-Volmer equation: , where F 0 and F denote the fluorescence intensities in the absence and presence of a quencher (SDS), [Q] is the quencher concentration, K SV is the Stern-Volmer quenching constant, k q is the bimolecular quenching rate constant, while τ 0 is the lifetime of the fluorophore (BSA) in the absence of quencher [25,26]. The graphs of F 0 F versus [Q] plotted according to the Stern-Volmer equation are shown in Figure 6. Consequently, Table 2 presents the newly determined values of Stern-Volmer quenching constants along with linear correlation coefficients (R 2 ) and bimolecular quenching rate constants (k q ) recovered for the studied systems. The latter ones were calculated based on the value of the average fluorescence lifetime τ 0 of free BSA (in Tris-HCl, pH 7.0), which equals approximately 6.3 ns [27]. It can be observed that the quenching (binding) constants decrease with the increase in the [B(Ph) 4 ] − concentration of the system, which can be assigned to the competition of SDS with [B(Ph) 4 ] − in BSA [28,29]. Furthermore, the estimated values of k q for all systems are of the order 10 12 M −1 ·s −1 , which is approximately a few hundred times higher than the maximum value possible for the diffusion-controlled quenching-rate constant (2.0 × 10 10 M −1 ·s −1 ) [30]. Since the values of bimolecular quenching rate constants are considered to be definitive in differentiating between dynamic and static quenching mechanisms, the predominant role of ground-state complexation (static quenching) in the investigated systems was confirmum albumin was measured using an extinction coefficient eqed.

Differential Scanning Calorimetry (DSC)
The influence of [B(Ph) 4 ] − ions on a thermal unfolding transition of BSA was assessed based on DSC experiments. The representative, original DSC curves are shown in Figure 7. Two distinct peaks are observed in the heat-capacity curve for the sample of free BSA. The transition midpoints T m1 = 328.9 K and T m2 = 346.7 K were estimated by fitting the experimental C p vs. T curves to a non-two-state model using Marquardt nonlinear least-squares method (T m denotes the temperature value at which a maximum endothermic effect appears). Previously, it has been reported that the appearance of these peaks corresponds to the low-temperature (T m1 ) and the high-temperature (T m2 ) transition resulting from the melting of structurally independent parts of the protein [31][32][33][34]. The less thermally stable domain comprises the subdomains IIB, IIIA, and IIIB, whereas the subdomains IA, IB, and IIA represent the region (the energetic domain) more resistant to thermal unfolding [35].
The M −1 s −1 ) [30]. Since the values of bimolecular quenching rate constants are considered to be definitive in differentiating between dynamic and static quenching mechanisms, the predominant role of ground-state complexation (static quenching) in the investigated systems was confirmed.

Differential Scanning Calorimetry (DSC)
The influence of [B(Ph)4] − ions on a thermal unfolding transition of BSA was assessed based on DSC experiments. The representative, original DSC curves are shown in Figure  7. Two distinct peaks are observed in the heat-capacity curve for the sample of free BSA. The transition midpoints Tm1 = 328.9 K and Tm2 = 346.7 K were estimated by fitting the experimental Cp vs. T curves to a non-two-state model using Marquardt nonlinear leastsquares method (Tm denotes the temperature value at which a maximum endothermic effect appears). Previously, it has been reported that the appearance of these peaks corresponds to the low-temperature (Tm1) and the high-temperature (Tm2) transition resulting from the melting of structurally independent parts of the protein [31][32][33][34]. The less thermally stable domain comprises the subdomains IIB, IIIA, and IIIB, whereas the subdomains IA, IB, and IIA represent the region (the energetic domain) more resistant to thermal unfolding [35].
The formation of fairly stable Na[B(Ph)4]-BSA complexes makes the BSA structure more compact and thermostable. This is reflected in the shift of the heat-induced transitions towards a higher range of temperatures with the increase in the Na[B(Ph)4]:BSA molar ratios in the mixture, namely Tm1 = 341.8 K and Tm2 = 351.9 K for the

Circular Dichroism (CD) Spectra Analysis
The effect of Na[B(Ph)4] ions' binding and temperature on the secondary structure of BSA was assessed by temperature-dependent CD measurements in the range of temperatures from 298.15 K to 368.15 K (Figures 8-10

Circular Dichroism (CD) Spectra Analysis
The effect of Na[B(Ph) 4 ] ions' binding and temperature on the secondary structure of BSA was assessed by temperature-dependent CD measurements in the range of temperatures from 298.15 K to 368.15 K (Figures 8-10). The performed experiments showed that the saturation of BSA with [B(Ph) 4 ] − ions at 298.15 K does not contribute to noticeable changes in the secondary structure of the protein. On the other hand, the thermal stability of the α-helix structure increases on account of the formation of the Na[B(Ph) 4 ]-BSA complexes ( Table 3). These findings are in agreement with the DSC results.

Circular Dichroism (CD) Spectra Analysis
The effect of Na[B(Ph)4] ions' binding and temperature on the secondary structure of BSA was assessed by temperature-dependent CD measurements in the range of temperatures from 298.15 K to 368.15 K (Figures 8-10

Molecular Modelling
To gain a deeper understanding of the experimental results at the atomistic level, MD simulation was conducted for the [B(Ph) 4 ] − -BSA system. In particular, the MD approach was applied to predict potential binding sites of [B(Ph) 4 ] − on the BSA surface and to further characterize and compare them in terms of the binding affinities for these ions. The protein was surrounded by 15 randomly distributed [B(Ph) 4 ] − ions and the simulation was performed for 100 ns at 300 K and a pH of 7. After the careful examination of the frames for each [B(Ph) 4 ] − ligand, the last 40% of the simulation was used for the further LIE analysis (Table 4). Based on our results, two bindings sites were localized for three molecules ( Figure 11). Interestingly, one binding site is in the proximity to Trp-134, as was observed in the case of our previous SDS-BSA study [20]. Two ions bound to site I and one ion to site II after 40 and 60 ns, respectively, and remained there until the end of the simulation. Their RMSD as well as the RMSD of the protein are shown in Supplementary Figure S1 to demonstrate the convergence. LIE energy values for the ions bound in sites I and II are shown in Supplementary Figure S2, supporting the convergence achieved for the binding in these sites. In order to strictly define the binding event in sites I and II, a 5 Å cutoff criteria was applied for the distance between the residues of the sites and [B(Ph) 4 ] − ion atoms. According to LIE calculations, site I has slightly higher affinity towards [B(Ph) 4 ] − than site II due to the more favorable van der Waals component of the free binding energy (the electrostatic and van der Waals components calculated for each [B(Ph) 4 ] − ion are provided in Table 4). Site I comprises Lys116, Pro119, Glu140, Ile141, Arg144, His145, Leu178, Glu182, Arg185, and Val188, while site II is formed by Leu386, Asn390, Phe402, Leu406, Arg409, Thr410, Lys413, Thr448, Ser479, Arg484, and Phe487. Clearly, site I has a more charged nature and lower propensity for establishing pi-pi and cation-pi interactions than site II. Additionally, when looking at the right panel of Figure 11, it seems that theoretical calculations confirm that hydrophobic rings of phenyl groups contribute to the stabilization of interactions between BSA and [B(Ph) 4 ] − .   Table 4. On the right panel, a representative frame for one [B(Ph)4] − ion and its protein surrounding.

Isothermal Titration Calorimetry (ITC)
All ITC experiments were performed at 298.15 K using an AutoITC isothermal titration calorimeter (MicroCal Inc. GE Healthcare, Northampton, MA, USA). The details of the measuring device and experimental setup have been described previously [20]. The reagents were dissolved directly in 10 Table 4. On the right panel, a representative frame for one [B(Ph) 4 ] − ion and its protein surrounding.

Isothermal Titration Calorimetry (ITC)
All ITC experiments were performed at 298.15 K using an AutoITC isothermal titration calorimeter (MicroCal Inc. GE Healthcare, Northampton, MA, USA). The details of the measuring device and experimental setup have been described previously [20].

Steady-State Fluorescence Spectroscopy and UV Spectrophotometry
The stock solutions of BSA, SDS and Na[B(Ph) 4 ] were prepared in 10 mM Caco buffer of pH 7 (all subsequent dilutions were made with this buffer). Theconcentration of bovine serum albumin was measured using an extinction coefficient equal to ε BSA 280 = 41, 180 M −1 cm −1 , calculated based on the content of tryptophan (ε W 280 = 5690 M −1 cm −1 ), tyrosine (ε Y 280 = 1280 M −1 cm −1 ), and cysteine (ε C 280 = 120 M −1 cm −1 ) [36]. A maximum ab-sorbance value of approximately 0.08 at 280 nm (corresponding to a protein concentration of 2 µM) was used to avoid the inner-filter effect. Steady-state fluorescence experiments were carried out with a Cary Eclipse Varian (Agilent, Santa Clara, CA, USA) spectrofluorometer, equipped with a temperature controller and a 1.  4 ], 0.015 mM BSA). All the samples were degassed before measurements. These measurements were recorded two times. The details of the measuring device, experimental setup and data processing have been described previously [20].  4 ] − (referred to as BPH for simplicity) was parameterized as by Kurt and Temel [38,39] and later built in xleap of AMBER16 package [40].

Molecular Dynamics Simulations
The MD approach was applied to localize potential binding sites and to characterize them in terms of the binding affinities. There are several reasons why such an approach is superior to the classical molecular docking approach, and therefore was used instead of molecular docking in our study: (1) Docking small ligands as SDS in our previous work [20] and [B(Ph) 4 ] − to BSA in this work did not allow for finding more than a single binding site.
(2) MD allows for the complete flexibility of the protein residues, which could be crucial for the binding of a small ligand, while rigid receptor molecular docking does not allow for it. Taking into account a structural ensemble of receptors in molecular docking is a possible alternative which is computationally more expensive than the MD approach. (3) In contrast to molecular docking, solvent is treated explicitly in the MD simulations. (4) Molecular docking scores are far less reliable than MD-derived binding affinities due to the fact that molecular docking scores single, static poses, while multiple frames corresponding to the statistical ensemble of the structures are considered in the MD-based calculations. (5) In most of the state-of-the-art modelling studies, the poses obtained by a molecular docking approach are further analyzed with the MD approach. Therefore, the MD approach, when applied for predicting binding sites, is already self-consistent in comparison to the two-step procedure that includes molecular docking followed by MD. Hence, the MD approach is preferable. (6) Furthermore, we previously demonstrated both convergence and high predictive power of the MD approach when applied to the protein-ion systems [41]. All all-atom molecular dynamics (MD) simulations were performed with the use of AMBER16 software package [40]. The initial structures of the complexes were created by random placement of 15 [B(Ph) 4 ] − ions around the BSA protein. A truncated, octahedron, TIP3P periodic box of an 8 Å water layer from the box's border to the solute was used to solvate complexes. The ff14SBonlysc force field for the protein [42] and gaff force field [43] with RESP charges [44] obtained in the antechamber module of AMBER16 [40] for [B(Ph) 4 ] − were used. The net negative charge of the system was neutralized with Na + counterions. Energy minimization was carried out in two steps: beginning with 500 steepest descent cycles and 10 3 conjugate gradient cycles with 100 kcal/mol/Å 2 harmonic force restraints on solute atoms, continued with 3 × 10 3 steepest descent cycles and 3 × 10 3 conjugate gradient cycles without any restraints. Following minimization steps, the system was heated up from 0 to 300 K for 10 ps with harmonic force restraints of 100 kcal/mol/Å 2 on solute atoms. Then, the system was equilibrated at 300 K and 10 5 Pa in isothermal isobaric ensemble for 100 ps. Afterwards, an independent prediction MD run was carried out in the same isothermal isobaric ensemble for 100 ns. The particle mesh Ewald method for treating electrostatics and the SHAKE algorithm for all the covalent bonds containing hydrogen atoms were implemented in the MD simulations. Such parameters within the MD simulation were shown to be sufficient for a system of ions with a protein, which was substantially bigger than BSA, allowing for proper prediction of the ion binding sites [41].

Binding Free-Energy Calculations
Energetic postprocessing of the trajectories and per-residue energy decomposition was performed for the system with the use of Linear Interaction Energy (LIE) with dielectric constant of 80. Frames from the MD simulation were carefully examined with VMD [45]. The last 40% of the simulation was chosen for the analysis, since in the corresponding frames no more events of ion dissociations or associations were observed. LIE takes into account only the electrostatic component (scaled by the dielectric constant) and the van der Waals component of the binding free energy. Although the binding free energies obtained by this method are not expected to represent absolute binding free energies, these calculations are useful to compare the binding strength in different binding sites.

Conclusions
A few complementary experimental methods supported by in silico analysis have successfully been applied to describe the physicochemical nature of tetraphenylborate ions' interactions with BSA. The obtained results were subsequently discussed in relation to a previously studied system, namely the interaction of sodium dodecyl sulphate (SDS) with BSA. The described approach enables us to find the BSA potential binding sites of [B(Ph) 4 ] − .
In conclusion, two main binding sites of [B(Ph) 4 ] − were unveiled. They are located in the subdomain IA and IIIA, respectively, of the investigated protein. Both tetraphenylborate ions and SDS molecules reveal a similar affinity to the first specific binding site in BSA which is located close to the Trp-134 residue. In turn, the second binding site on the BSA surface for investigated ligands are different. The binding of [B(Ph) 4 ] − ions in the subdomain IIIA is an entropy-driven process, whereas SDS in the subdomain IIB is enthalpy-driven.
It has been proven that [B(Ph) 4 ] − ions participate in the stabilization of the tertiary structure of BSA at higher temperatures on account of the formation of fairly stable [B(Ph) 4 ] − -BSA complexes of a different stoichiometry ([B(Ph) 4 ] − :BSA = 2:1 and 5:1). This phenomenon underlying its biological activity could be responsible for the protection of the protein from thermal unfolding.
The results clearly show that the presence of competitive ligands may have some implications on the concentration of the compounds tested (as has been found for [B(Ph) 4 ] − ions or SDS) in the cell-culture media containing albumin. On the one hand, BSA, an important component of the cell culture media, binding to the compounds of a hydrophobic nature, as [B(Ph) 4 ] − or SDS, can lead to a decrease in the concentration of free, active species. In consequence, this can affect the cytotoxic action of the tested compound. On the other hand, the saturation of the potential binding sites of BSA by a nontoxic, low-molecularweight ligand may prevent the interactions of albumin with other biologically active compounds, thus affecting their dose-dependent action.
The findings presented in this contribution are worth taking into consideration during the analysis of the cytotoxicity assays of the compounds being characterized.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.