1 INTRODUCTION

The intermolecular interactions play paramount roles in biological activities, and to learn the mechanisms created by Mother Nature needs to understand all the details of the interactions, especially for the binding constant and the number of binding sites, which could reversibly influence labeling, detecting/sensing and therapy efficiencies. A number of methods have been developed to study intermolecular interactions, such as gel electrophoresis (GE) [1-4], electrochemistry [5], spectroscopy [1, 3, 4, 6-11], quartz crystal microbalance (QCM) [12-14], surface plasmon resonance (SPR) [15, 16], and also computer simulations [17, 18] are popularly used in biomedical field. In general, the raw data obtained by the above methods normally needs to be further processed in combination with some classical equations like Stern-Volmer equation, Scatchard equation and Hill equation to get the binding constants, sites and even cooperativities in order to comprehensively understand intermolecular interactions. Let’s take a simple molecular interaction system as an example: R + L ⇌ RL, the binding constant can be described as K_a= [RL]/([R][L]), from which three concentrations must be figured out to calculate the final K_a. However, one may only know the concentrations of both [R] and [L] at the beginning of reaction. Alternatively, they could be indirectly obtained by subtracting the final concentration of [RL] from the initial concentration of [L] or [R] at the equilibrium state. Most of the above methods adapt to measuring the concentration of [RL], such as spectroscopy-, QCM- and SPR-based methods. However, the precision of results could be interfered somehow by the subtle deviations from the calibration curves made before binding. Hence, it will be more precise and reliable to directly measure the concentration of free [L] or [R] at the equilibrium.

Inspired by the separation principle of centrifugation and size exclusion chromatography (SEC) [19], we can learn that both these two separation methods are based on the significant intermolecular difference between either weights or sizes. It might not be a coincidence, most of the important interactions in biomedical researches can make best use of this principle, especially for pharmaceutic researches since almost all of designed drugs are much smaller than what they really interact with in cells. With the deliberate calculation equation invented by an American chemist George Scatchard, we would like to demonstrate that the combination of Scatchard plot and ultracentrifugation/SEC can be a facile and more precise method to solve the binding constant and the number of binding sites for small-large/light-weighted molecular interactions.

Eosin B (4’,5’-dibromo-2’,7’-dinitrofuorescein, EB) is a form of eosin dyes which belong to xanthenes class and have been used as histological stains [20], for protein estimations [21, 22], and sonosensitizers [23] for decades, and its analogue eosin Y (EY) has been used for staining proteins on acid-urea polyacrylamide gels [24]. Bovine serum albumin (BSA), also known as "fraction V" which refers to the fifth separation fraction by using the original Edwin Cohn purification methodology, has been considered a model protein because it shares a high sequence homology of 76.52% with human serum albumin (HSA) [3]. Serum albumin is the most abundant drug carrier protein in blood plasma playing an important role in the transportation and distribution for endogenous and exogenous substances.

In this work, BSA-EB interaction system has been studied by a comprehensive comparison of different methods since EB and BSA can be considered as the model small/light and large/weight molecules, respectively. Firstly, a combination of Scatchard equation and either ultracentrifugation or SEC for the physisorption of BSA-EB was investigated. Secondly, other methods like combination of photoluminescence (PL) quenching and either Stern-Volmer equation or Hill equation have been introduced to make comparisons. In addition, the conformational changes of BSA upon binding EB has been studied by using three-dimensional fluorescence spectroscopy (3DFS), meanwhile, the binding mode of BSA-EB with atomic resolution was simulated by using molecular docking methods. We hope this study could contribute to the investigation of small-larger molecular interaction system.

2 EXPERIMENTAL

3 RESULTS AND DISCUSSION

The absorption spectra comparison of BSA at 280 nm before and after ultracentrifugation have proved that the current ultracentrifugation condition is enough for a perfect separation of bound EB and free EB (Fig.1a), and this has largely simplified both measurements and calculations: on one hand, normally an ultracentrifuge can contain at least six vials and this will be enough for the plot; on the other hand, one can just stop the centrifuge and take about no more than half supernatant from the vials for the concentration measurement simply with a lab pipette. With an absorption maximum dependent concentration calibration curve (Fig.1b), the free EB’s concentration can be calculated from the maximum absorption of supernatants (Fig.1c) and thus the corresponding Scatchard plot curve can be obtained (Fig.1d). With the slope and intercept from the best linear fit of the Scatchard plot, the K_a and number of binding sites were determined to 0.624 µM^-1 (6.24 × 10⁵ M^-1) and 3.11, respectively.

Figure 1.Full-screen View

The Scatchard assay based on ultracentrifugation. a) The absorption comparison between BSA before (1) and after (2) ultracentrifugation (140, 000 rpm for 2h); b) The plotted calibration curve of EB’s absorption against concentration (C_f(EB)), and the inset is the absorption spectra of EB with varied concentration as numbered from 5 to 1: 30, 24, 18, 12, 6 μM, respectively; All of the data points were averaged from three latest replicates with a standard error less than 0.2%. c) The absorption spectra of supernatants after the ultracentrifugation of BSA-EB mixture (the molar ratios from numbered 5 to 1 as: ½, ¼, 1/6, 1/8 and 1/10, respectively). d) The plotted Scatchard curve (black dots) and the corresponding best linear fit (red line) of BSA-EB binding. The data were averaged from at least three replicates.

By using SEC, EB molecules can be also well repeatedly, precisely eluted during 20 to 40 min. One can see there are shoulder peaks at 24 min (Fig. 2a) of the elution spectra, which could be assigned to some micelle-like nanostructures formed by EB molecules due to the π- π interaction since it can be removed in the diluted solutions (lower than its critical micelle concentration). A calibration curve was then made by plotting the integrated eluted peaks against the elution time (Fig. 2b), which can facilitate the calculation of the free EB’s concentration. With a control sample of free BSA, it can be clearly seen that BSA and BSA-EB cannot separated with the current SEC columns (Fig.2c). Similar to the ultracentrifugation method, the free EB’s concentrations can be directly calculated from the SEC results by running the mixtures with the varied the molar ratios of EB/BSA (as shown in Table 1) on HPLC. With calculations, another Scatchard equation can be plotted (Fig.2d) and from which the binding constant K_a and the number of binding sites n were obtained as 0.57 µM^-1 (5.7 × 10⁵ M^-1) and 2.93, respectively.

Figure 2.Full-screen View

The Scatchard assay based on SEC. a) The plotted curves of absorption at 516 nm against elution time (t_e) with the varied EB’s concentrations as: 30, 24, 18, 12, 6 $\mu$ M, respectively (from 1 to 5); b) The plotted calibration curve and its best linear fit of EB’s integrated absorption peaks against its concentration (C_f(EB)). The intercept and slope from the linear fit are 0.00408 and 0.00154, respectively. All of the data points were averaged from three replicates with a standard error less than 0.2%. c) The plotted SEC elution with the molar ratio of EB/BSA varied as 10/1, 8/1, 6/1, 4/1, 2/1, and 0/1, respectively (from 1 to 6); The peaks shown in the figure reflect the free EB’s concentration (C_f(EB)), and the conjugate’s concentration was too low compared to the free EB’s to be seen in the figure, which is actually eluted out at 10 min; d) The plotted Scatchard curve (black dots) and the corresponding best linear fit (red line) of EB binding to BSA based on c). The intercept and slope from the linear fit are 1.67 and -0.57, respectively. The data were averaged from at least three replicates.

In contrast to the Scatchard method, all of Stern-Volmer equation (Eq.(3)), modified Stern-Volmer equation (Eq.(4)) and Hill equation (Eq.(5)) have been employed to analyze the binding constant as well, moreover, Hill equation can figure out the cooperativity in the binding procedure.

where F₀, F and F_sat are the PL intensities in the absence, presence and saturated state of quencher, respectively. The Stern-Volmer quenching constant is given by K_SV = k_qτ₀, where k_q is the quenching constant, τ₀ is the lifetime of the fluorophore in the absence of quencher, f_a, K_a, n and [Q] are the fraction of accessible fluorophores, the effective quenching constant or the binding constant for the accessible fluorophores, number of binding sites, and the concentration of the quencher, respectively (Table 2).

However, both Stern-Volmer and Hill equations based methods require PL variations, for an example the PL quenching upon binding. Owning to spectra overlap between the BSA’s PL and EB’s absorption (Fig.3a), the FRET can made best use to realize these two methods. To this end, the PL spectra were recorded for the mixtures with varied EB/BSA molar ratios (Fig. 3b), and the EB’s PL range didn’t show up because of the strong interruption from the second harmonic peak (@ 560 nm which is very close to EB’s PL @ 548 nm) of current excitation wavelength (280 nm), which has no real reference value since the measured samples were not purified. In practice, the Stern-Volmer plot showed a downwards bent curve (Fig. 3c inset), indicating a complicated binding procedure which most of time involves both static and dynamic quenching or even conformational changes upon binding [25]. Hence, a modified Stern-Volmer equation was plotted again with better linear results (Fig.3c), and K_a and f_a was determined to 5.33 × 10⁶ M^-1 and 0.94, respectively. Similarly, by applying Hill equation to the corresponding data, a double logarithmic plot and its best linear fit (Fig.3d) can give out 5.72 × 10⁴ M^-1 and 0.68 as an additional K_a and an important parameter for cooperativity n, respectively.

Figure 3.Full-screen View

The spectroscopy assay of BSA-EB interactions. a) Both absorption and emission spectra of free EB and BSA; b) The PL quenching of BSA by EB with different mixing molar ratios of EB/BSA (from 5 to 0 are: 25, 12.5, 6.3, 3.2, 1.6, 0, respectively) showed a Förster resonance energy transfer (FRET) phenomenon; Based on the PL quenching of BSA from b), the Stern-Volmer plot (inset is the modified) in c) and the Hill double logarithm plot (black dots) and its corresponding best linear fit in d) were drawn.

With comparison of several references’ report, we think that both Scatchard-based ultracentrifugation and SEC results should approach the truth. However, in contrast to the Scatchard-based methods, the binding constant K_a deduced from FRET-based methods are not well stable, because the difference of K_a values deduced from modified Stern-Volmer and Hill equations was around two orders of magnitude, though that f_a is less 1 makes much sense (the accessible fraction of EB to BSA cannot be more than 1). We attribute this huge deviation to the unstable PL of BSA due to the confirmation changes of BSA upon binding with EB. To further study the conformational information of BSA, we have employed 3DFS to see whether BSA conformation changes.

3DFS is a facile and powerful method to provide more detailed information about the conformational and micro-environmental changes of protein that combined with molecules [1, 7, 10]. The principle of 3DFS is based on that the intrinsic fluorophores (like Trp residues) in proteins, whose PL can be strongly influenced by the polarity of the local environment. For example, the PL of a protein containing a single Trp in its hydrophobic core will be red-shifted upon exposure this Trp to an aqueous environment [26]. From Figure 4, two typical PL peaks can be clearly observed in the 3DFS of BSA (peak 1: λ_ex / λ_em= 276 nm / 346 nm, peak 2: λ_ex / λ_em = 244 nm / 350 nm). Refers to the previous report [27], peak 1 mainly reveals the spectral characteristics of Trp and Tyr residues, is the primary PL peak, while peak 2 may mainly exhibit the PL characteristic of polypeptide backbone structures therefore its PL intensity can be correlated with the secondary structure of proteins. The corresponding spectral parameters were summarized in Table 3. Combining Figure 4 with Table 3, it is clear to see that both PL intensities of peak 1 and peak 2 of BSA decrease with the gradual addition of EB, whereas the Rayleigh scattering peak is almost unchanged. Analyzing from the intensity changes of both peak 1 and peak 2, the quenching results indicated that the interaction of EB with BSA induced a slight unfolding of the protein polypeptides, resulting in a conformational change of the protein, this could increase the exposure region of some hydrophobic regions that have previously been buried. Moreover, we realized that the maximum of PL of BSA occurred a blue shift due to water Ramon peak, which might be the one of the main reasons that cause the huge deviation of results obtained from two PL quenching-based methods.

Figure 4.Full-screen View

3-D excitation and emission matrix of BSA upon binding to EB. a) is for BSA alone; b), c) and d) are for the samples with different EB/BSA molar ratios as 2, 4 and 8, respectively. Peak a and b are two Rayleigh scattering peaks with its emission wavelength equals to and is two times (second-ordered harmonics) as its excitation wavelength, respectively. The number 1 and 2 indicate two peaks with the wavelengths of excitation/emission are 276 nm / 346 nm and 244 nm / 350 nm, respectively.

With the aim to get the detailed binding site information at the atomic level, the molecular docking method [19] was employed by using Auto dock software package. As shown in Fig. 5, EB can be perfectly docked into the binding pocket of the BSA. The benzoic acid ring of EB fits into bottom of the binding pocket of BSA, surrounded by the residues Trp²¹³, Arg²¹⁷, Arg¹⁹⁴ and Arg¹⁹⁸. Detailed analysis shows that carboxy group forms a hydrogen bond (2.9 Å) with residue Arg¹⁹⁸. Furthermore, carbonyl oxygen forms two hydrogen bonds (2.3 and 3.6 Å) with residue Arg¹⁹⁸, while the nitro group forms two hydrogen bonds (3.1 and 3.3 Å) with residue Asp⁴⁵⁰ and Arg¹⁹⁴, respectively, which could be the main binding affinity between EB and BSA.

Figure 5.Full-screen View

Molecular docking of EB binding to BSA. a) The EB is docked into the binding pocket of the BSA. b) The detailed analysis of main binding sites between BSA’s amino acid residues and EB structure. The unit of all distances is Å (0.1 nm).

4 Conclusion

The BSA-EB interaction has been quantified by successfully introducing a combinational method to quantify the interaction between two molecules by taking the advantages of significant differences of either sizes or weights. In comparison with PL quenching-based methods, the current introduced methods are simpler to operate and lead to results that are more precise. In addition, the conformational changes of BSA upon binding EB and the atomic resolved theoretical binding mode were studied by 3DFS and molecular docking method, respectively. The introduced facile method could be widely applied for bio-drug interaction researches or any two substances with large difference in sizes or weights. This study has also provided a good demonstration to the well combination of experimental measurements and the theoretical simulations.