Relationship investigation of molecular structure–binding affinity of antibiotics to bovine serum albumin using flow injection chemiluminescence analysis and molecular docking

Abstract

Luminescently probed by luminol, the interaction behaviors of bovine serum albumin (BSA) with different antibiotics (macrolides, tetracyclines and sulfonamides) at picomolar levels were investigated using flow injection chemiluminescence (FI-CL) analysis. It was found that BSA and antibiotics formed complexes with the binding ratio of 1 : 1, with the binding constants K within the range of 10³ to 10⁵ L mol⁻¹ generally following the order of macrolides < tetracyclines < sulfonamides. Results showed that the ester groups of –OOCC₃H₇ or –OOC(CH₂)₂COOC₂H₅ in macrolides led to increased K by factors of 3.1–42.8 times, –OH or –Cl in tetracyclines increased K by 1.1–1.3 times, and the additional –C(=NH)NH₂ in sulfonamides increased K about 1.3 times. The thermodynamic parameters demonstrated that the BSA–antibiotic binding process should be spontaneous mainly through hydrophobic interaction for macrolides, hydrogen bonding and van der Waals force for tetracyclines, and electrostatic interaction for sulfonamides. The further antibiotics to BSA molecular docking study revealed that the pocket at subdommain IIA of BSA should be the principle binding site for antibiotics, showing that the interaction parameters including K and ΔG agreed well with the results from the proposed FI-CL analysis. The interesting phenomenon that the log K had good linear correlation to molecular volume MV, molar refractivity MR, polarizability POL and partition coefficient log P of antibiotics was also observed.

---

1. Introduction

Protein–drug binding, an essential intermolecular interaction in the cellular or organ environment, which is helpful in understanding a drug's in vivo processes^1–4 and provides valuable information for drug designation,^5–8 has drawn increasing concern in scientific fields of medicine, biology and chemistry.^9–13 One of the prominent model proteins, bovine serum albumin BSA, which contains two major drug binding sites in the pockets at subdomain IIA and subdomain IIIA, has been applied extensively in the study of protein–drug interaction.^14–18 Several reports have shown individually the interaction mechanisms of antibiotics including midecamycin MID, chlortetracycline CHL and sulfamethazine SMZ binding to BSA,^19–21 and no comprehensive investigation of interaction behaviors between BSA and different antibiotics (macrolides, tetracyclines and sulfonamides) has been described so far.

Macrolides, tetracyclines and sulfonamides are three typical classes of natural or semisynthetic antibiotics, and are characterized by the existence of 14–16-membered lactone ring,²² octahydrotetracene-2-carboxamide skeleton²³ and sulfanilamide group,²⁴ respectively. As large antibiotic families which possess remarkably broad-spectrum activities, they are widely utilized in clinical therapy of human and animal infections.^25–27 It's known that the structural differences of drugs affect their absorption, transportation, metabolism and pharmacologic activities in vivo,^28–31 making structure–affinity of drug binding to protein become an important insight into protein–drug interaction.^32–35 Up to date, there has been nearly no report on the structural influence of the three kinds of antibiotics on their binding behaviors to BSA.

Herein, using luminol as the luminescent probe, the interaction behaviors of BSA with 14 different antibiotics (macrolides, tetracyclines and sulfonamides, Scheme 1) were investigated at picomolar levels by flow injection chemiluminescence (FI-CL) analysis, which were based on the quenching effect of antibiotics on the CL intensity from luminol–BSA system. The binding constants K and the numbers of binding sites n of BSA with antibiotics were given, which generally ranked in order of macrolides < tetracyclines < sulfonamides. The effect of structural differences of antibiotics (including ester groups at R₁–R₄ in macrolides, –Cl or –OH at R₁–R₂ in tetracyclines and –C(=NH)NH₂ at R₁ in sulfonamides) on the K values of BSA–antibiotics interactions was examined. The relationships of log K values vs. basic physicochemical descriptors (molecular volume MV, molar refractivity MR, polarizability POL and partition coefficient log P) of antibiotics were also studied.

Scheme 1 The chemical structures of antibiotics and CL intensity quenching proportionality^a.

Protein–ligand (P–L) molecular docking, a useful tool to model and predict the interaction behavior of protein with drug offering visual output,^36–39 was employed subsequently to identify the interaction sites and binding modes of antibiotics to BSA. The antibiotics to BSA docking analysis revealed that the principle binding site of antibiotics should be the pocket at subdomain IIA of BSA, with key residues involved in the interactions given. The correlation of interaction parameters from FI-CL analysis and P–L docking was also discussed.

2. Materials and methods

2.1. Materials

All chemicals used in this work were of at least analytical reagent grade. BSA (Sigma-Aldrich, St. Louis, MO, USA) was used as received without any further purification. Macrolides (erythromycin ERY, kitasamycin KIT, MID, josamycin JOS, erythromycin ethylsuccinate ERE, josamycin propionate JOP and midecamycin acetate MIA), tetracyclines (tetracycline TET, oxytetracycline OXY and CHL) and sulfonamides (sulfanilamide SNM, sulfaguanidine SGD, sulfadizine SDZ and SMZ) were obtained from National Institute for the Control of Pharmaceutical and Biological Products, China. Luminol (Fluka, Biochemika, Switzerland) was purchased from Xi'an Medicine Purchasing and Supply Station, China. Deionized water was passed through a Milli-Q water purification system (Millipore, Bedford, MA, USA, 18.2 MΩ cm) prior to usage.

Macrolides standard solutions (2.0 × 10⁻⁴ mol L⁻¹) were prepared in ethanol–water solution (1 : 4, v/v). Tetracyclines standard solutions (1.0 × 10⁻⁴ mol L⁻¹) were prepared in water. Sulfonamides standard solutions (1.0 × 10⁻⁴ mol L⁻¹) were prepared in 5.0 × 10⁻² mol L⁻¹ NaOH solution except SGD standard solution which was dissolved in 0.2 mol L⁻¹ H₂SO₄ solution. BSA standard solution (1.0 × 10⁻⁶ mol L⁻¹) was prepared in water. Luminol standard solution (2.5 × 10⁻² mol L⁻¹) was prepared in 0.1 mol L⁻¹ NaOH solution and kept in dark.

2.2. Experimental details

The utilized FI-CL apparatus (Xi'an Remax Analysis Instrument Co. Ltd., Xi'an, China, Fig. S1†) consisted of a sampling system (IFFM-E), a CL detector (IFFS-A) and a recorder (a computer with IFFM-E client system). At a constant rate of 2.0 mL min⁻¹, 100 μL luminol was injected into the flow line by six-way valve and then merged with the homogenous stream of BSA and antibiotic. Thereafter, the whole mixture was delivered to the CL cell in an alkaline medium and the produced CL signal was detected without wavelength discrimination by the PMT at a high voltage of −750 V. The recorded CL signal of luminol–BSA system with and without antibiotic were defined as I and I₀, respectively, and antibiotic concentration was quantified by the CL intensity decrement (ΔI = I₀ − I) accordingly. The temperature of the solution was controlled by a water bath (HW.SY11-K6B/C, T ± 0.1 °C).

Fluorescence study of BSA with antibiotics was carried out using the F-4500 fluorophotometer (Hitachi, Kyoto, Japan, ESI†).

2.3. P–L docking analysis

The crystal structure of BSA was taken from the RCSB Protein Data Bank with the entry code of 4F5S (http://www.rcsb.org/pdb/explore/explore.do?structureId=4F5S), and the initial structures of antibiotics were generated by Chem-Bio-Office 2010 with energy optimized using Merk Molecular Force Field 94. AutoDock 4.2 suit of packages (http://autodock.scripps.edu/), a popular and proven robust approach with good docking accuracy and realibility,^40–42 was utilized to study the interaction modes of BSA with antibiotics. Gasteiger partial charges, nonpolar hydrogen atoms, and rotatable bonds were dealt with the aid of AutoDock Tools 1.5.6 (http://mgltools.scripps.edu). The docking grid box was set as 60 Å × 60 Å × 60 Å with the grid center of (−3.560 Å, 27.687 Å, 100.906 Å) and a grid box point spacing of 0.375 Å. The population size and the maximum number of evaluations were 150 and 2 500 000, respectively. Lamarckian Genetic Algorithm was chosen as the searching and scoring procedure for van der Waals force, electrostatic interaction, entropy loss upon binding process, hydrogen bonding and solvation; and the docking results were clustered with a root-mean-square deviation tolerance of 2.0 Å. The output results were graphically analyzed using Pymol 1.6.0.0 and Ligplus 1.3.2.

3. Results and discussion

3.1. The optimum experimental conditions and stability of present FI-CL system

The study of optimum experimental conditions showed that a stable and strong CL intensity can be obtained at 5.0 × 10⁻⁵ mol L⁻¹ luminol, 5.0 × 10⁻⁹ mol L⁻¹ BSA and 2.5 × 10⁻² mol L⁻¹ NaOH. Considering a compromise of higher signal/noise ratio, better precision and less reagent consumption, 2.0 mL min⁻¹ flow rate, 5.0 cm mixing tubing length and −750 V high voltage of PMT were chosen as optimal.

The stability of this FI-CL system was representatively evaluated by repetitive determination of ERY (1.0 × 10⁻¹², 1.0 × 10⁻¹⁰ and 1.0 × 10⁻⁹ mol L⁻¹) in five consecutive days with the flow system being operated continuously over 8 h per day. The derived relative standard deviations (RSDs) were less than 1.5% (n = 7), indicating the acceptable stability of this FI-CL system.

3.2. The relative CL intensity–time profiles of different systems

The relative CL intensity–time profiles of different systems are tested with results given in Fig. 1 (I_max refers to the maximum CL intensity, and T_max refers to the time for the I_max). It is clear that the I_max for luminol–BSA system (curve 1) and luminol–O₂ system (curve 16) are 285 and 140, with the T_max of 4.1 and 4.4 s, respectively; while in the presence of antibiotics, the I_max for luminol–BSA system is obviously quenched with the T_max remaining 4.1 s (curves 2–15), and the I_max for luminol–O₂ system shows no significant changes (curve 17, taking luminol–O₂ CL system with ERY as the example). It can also be seen that the I_max for luminol–BSA system declines differently by 1.0 × 10⁻¹⁰ mol L⁻¹ antibiotics, with the quenching proportionalities (Scheme 1) following macrolides (ERY < KIT < MID < JOS < ERE < JOP < MIA) < tetracyclines (TET < OXY < CHL) < sulfonamides (SNM < SGD < SDZ < SMZ).

Fig. 1 The relative CL intensity–time profiles of different systems, luminol: 5.0 × 10⁻⁵ mol L⁻¹; BSA: 5.0 × 10⁻⁹ mol L⁻¹; antibiotics: 1.0 × 10⁻¹⁰ mol L⁻¹. Curve 1: luminol–BSA CL system; curves 2–15: luminol–BSA CL system with ERY, KIT, MID, JOS, ERE, JOP, MIA, TET, OXY, CHL, SNM, SGD, SDZ and SMZ, respectively; curve 16: luminol–O₂ CL system; curve 17: luminol–O₂ CL system with ERY.

3.3. The possible mechanism of luminol–BSA–antibiotic CL system

It is clear in Fig. 1 that the I_max for luminol–BSA system is 2.0-fold that of luminol–O₂ system and the T_max shortens 0.3 s, demonstrating that BSA could accelerate the electron transfer rate of excited 3-aminophthalate. According to previous report,⁴³ it is hypothesized that luminol could enter into the site of Trp134 at subdomain IB in BSA, which induces the hyperchromic effect on luminol leading to the enhanced CL intensity. It is also clear that there are no obvious differences of the I_max for luminol–O₂ system with and without antibiotics, indicating that the interactions of luminol with antibiotics can be neglected. Hence, the CL intensity quenchment for luminol–BSA system by antibiotics might attribute to the interactions of BSA with antibiotics, and the possible CL mechanism of luminol–BSA–antibiotic system can be elucidated as follows: the hyperchromic effect of BSA on luminol causes the complexation enhancement of CL (CEC), producing the enhanced CL intensity from luminol–O₂ system; while in the presence of antibiotics, the BSA–antibiotics interactions lead to the complexation enhancement of quenching (CEQ), resulting in the significantly quenched CL intensity from luminol–BSA system.

3.4. The stoichiometric ratios of BSA with antibiotics

Using Continuous Variations method, the molar ratios of BSA with antibiotics are obtained and results reveal that BSA and antibiotics can form 1 : 1 complexes. By taking BSA–ERY complex as a representative example in Fig. 2, it is clearly shown that the CL intensity (I_i) has an inflexion point with the value of 189.5 at 0.5 of C_BSA/(C_ERY + C_BSA), which indicates the stoichiometric ratio of BSA–ERY complex is 1 : 1. The theoretical CL intensity (I_t) at the inflexion of 190.5 is estimated by extrapolating the linear parts of the curve to their intersection. Using the equation of α = (I_t − I_i)/I_t, the degree of dissociation (α) for BSA–ERY complex is obtained as 0.52%.

Fig. 2 The binding ratio of BSA with ERY by Continuous Variation method I_i: the CL intensity at the inflexion; I_t: the theoretical CL intensity at the inflexion. The total concentration of BSA and ERY is set as 1.0 × 10⁻⁹ mol L⁻¹.

3.5. The linear relationship of ΔI vs. C_Antibiotic

It is found that the decrement of the CL intensity from luminol–BSA system shows linear response to the logarithm of antibiotic concentration within 1.0 to 5000 pmol L⁻¹, yielding a general calibration equation of ΔI = A log C_Antibiotic + B (correlation coefficient r > 0.99, Table 1). Herein, the slope A is known as the determination sensitivity factor, while the intercept B, representing the characteristic CL intensity decrement (ΔI = B) from luminol–BSA system with antibiotic at the low terminal concentration of 1.0 pmol L⁻¹, is first defined as the quenching efficiency factor. It can be seen that both the CL parameters A and B obey the same increasing order of quenching proportionalities, suggesting that the quenching proportionalities are positively related to the CL parameters.

Table 1 The linear equation with concentration linear range of antibiotic to BSA by FI-CL at 298 K

Antibiotics	ΔI = A log CAntibiotic + B	Linear range (pmol L^−1)	r^a
a r: correlation coefficient.
ERY	ΔI = 4.6 log CERY + 14.2	1.0–1000	0.9966
KIT	ΔI = 5.0 log CKIT + 17.6	1.0–1000	0.9953
MID	ΔI = 6.1 log CMID + 22.8	1.0–5000	0.9960
JOS	ΔI = 7.0 log CJOS + 30.3	1.0–1000	0.9951
ERE	ΔI = 7.2 log CERE + 35.6	1.0–1000	0.9972
JOP	ΔI = 7.4 log CJOP + 37.4	1.0–1200	0.9952
MIA	ΔI = 8.0 log CMIA + 41.5	1.0–1000	0.9985
TET	ΔI = 8.3 log CTET + 46.6	1.0–2000	0.9982
OXY	ΔI = 9.2 log COXY + 52.4	1.0–2000	0.9978
CHL	ΔI = 9.9 log CCHL + 60.8	1.0–1500	0.9963
SNM	ΔI = 11.7 log CSNM + 70.5	1.0–500	0.9949
SGD	ΔI = 12.7 log CSGD + 77.4	1.0–1000	0.9952
SDZ	ΔI = 13.4 log CSDZ + 85.6	1.0–1000	0.9992
SMZ	ΔI = 13.8 log CSMZ + 89.1	1.0–700	0.9956

---

3.6. The interaction parameters and binding modes of BSA with antibiotics

Based on the steady CL intensity from luminol with protein reaction, the FI-CL model of log[(I₀ − I)/I] = log K + n log C_Ligand (K: binding constant, n: the number of binding sites, C_Ligand: the ligand concentration, the details of the derivation are given in ESI†) for studying protein–ligand interaction was successfully constructed by our group.^44,45 Using the homemade FI-CL model, the binding constants K and the numbers of binding sites n of BSA with antibiotics at 298 K are listed in Table 2, with plots of log[(I₀ − I)/I] against log C_Antibiotic given in Fig. S2† (the K and n at 288/308 K are listed in Table S1†). It can be seen that the K values in the range of 10³ to 10⁵ L mol⁻¹ rank in order of macrolides (ERY < KIT < MID < JOS < ERE < JOP < MIA) < tetracyclines (TET < OXY < CHL) < sulfonamides (SNM < SGD < SDZ < SMZ); and the n values within 0.73–1.04 indicate that there is one independent class of binding sites on BSA for antibiotics, agreeing well with the stoichiometric ratio study. It is found that the log K values are linearly proportional to the n values (log K = 7.4n − 1.8, r = 0.9910), confirming the suitable application of this homemade FI-CL model in the present study.⁴⁶ Also there are monotonic increasing correlations of log K vs. A or B, revealing the quenching efficiency factor B as well as the determination sensitivity factor A can be used as a positive predictor for the binding affinity of antibiotic to BSA.

Table 2 Binding parameters of antibiotic to BSA by FI-CL/FQ/P–L docking^a

Antibiotics	K (L mol^−1) FI-CL/FQ/P–L docking	n FI-CL/FQ	Linear range^b (pmol L^−1)
a The binding parameters were obtained at 298 K.b The shown antibiotic linear ranges were used in FI-CL model.
ERY	2.30 × 10^3/2.35 × 10^3/1.93 × 10^3	0.73/0.74	1.0–900
KIT	7.03 × 10^3/6.74 × 10^3/6.84 × 10^3	0.76/0.75	1.0–1000
MID	1.49 × 10^4/1.40 × 10^4/1.26 × 10^4	0.81/0.80	1.0–3500
JOS	4.27 × 10^4/4.43 × 10^4/4.05 × 10^4	0.86/0.86	1.0–1000
ERE	8.25 × 10^4/7.97 × 10^4/8.18 × 10^4	0.91/0.89	1.0–750
JOP	9.84 × 10^4/9.15 × 10^4/9.47 × 10^4	0.92/0.90	1.0–1000
MIA	1.14 × 10^5/1.02 × 10^5/1.07 × 10^5	0.94/0.92	1.0–1000
TET	1.87 × 10^5/1.57 × 10^5/1.51 × 10^5	0.95/0.95	1.0–1500
OXY	2.10 × 10^5/2.29 × 10^5/2.46 × 10^5	0.96/0.96	1.0–1500
CHL	2.46 × 10^5/2.63 × 10^5/2.61 × 10^5	0.97/0.97	1.0–1000
SNM	4.17 × 10^5/4.04 × 10^5/4.32 × 10^5	0.98/1.01	1.0–500
SGD	4.88 × 10^5/4.34 × 10^5/4.95 × 10^5	1.01/1.02	1.0–900
SDZ	5.57 × 10^5/5.76 × 10^5/5.63 × 10^5	1.03/1.05	1.0–1000
SMZ	6.24 × 10^5/6.67 × 10^5/6.68 × 10^5	1.04/1.08	1.0–700

---

The thermodynamic parameters (including enthalpy change ΔH, entropy change ΔS and binding free energy ΔG) are obtained using the following equations:⁴⁷

ΔG = −RT ln K = ΔH − TΔS (2)

The ΔG values from −33.82 to −18.35 kJ mol⁻¹ (Table 3, 298 K; Table S2,† 288/308 K) show that the interaction of BSA with antibiotic should be a spontaneous process, with the binding tendencies generally obeying macrolides < tetracyclines < sulfonamides. According to Ross theory,⁴⁸ the predominant binding force for antibiotics to BSA is hydrophobic interaction for macrolides (ΔH > 0, ΔS > 0), hydrogen bond (H-bond) and van der Waals force for tetracyclines (ΔH < 0, ΔS < 0), and electrostatic interaction for sulfonamides (ΔH < 0, ΔS > 0). It is observed that the absolute values of ΔS increase with increasing absolute values of ΔH for all the three kinds of antibiotics binding to BSA, revealing there is enthalpy–entropy compensation relationship in the interaction processes.

Table 3 Thermodynamic parameters of antibiotic to BSA by FI-CL/FQ/P–L docking^a

Antibiotics	ΔH (kJ mol^−1) FI-CL/FQ	ΔS (J mol L^−1) FI-CL/FQ	ΔG (kJ mol^−1) FI-CL/FQ/P–L docking
a The listed ΔG values were the results at 298 K.
ERY	18.06/18.04	125.96/125.90	−19.17/−19.24/−18.75
KIT	19.10/18.59	138.44/136.44	−21.94/−21.36/−21.88
MID	21.48/21.53	151.66/151.49	−23.81/−23.66/−23.39
JOS	22.17/23.63	161.72/166.87	−26.42/−26.51/−26.28
ERE	23.16/24.05	170.61/173.36	−28.05/−27.96/−28.03
JOP	23.60/24.54	173.11/175.93	−28.49/−28.31/−28.39
MIA	28.04/25.89	189.40/189.74	−28.85/−28.57/−28.69
TET	−37.62/36.91	−24.61/−23.34	−30.08/−29.66/−29.55
OXY	−40.08/−39.87	−31.95/−31.19	−30.36/−30.58/−30.76
CHL	−40.67/−40.31	−33.34/−31.74	−30.96/−31.28/−30.90
SNM	−5.91/−6.72	87.22/84.81	−32.06/−31.99/−32.15
SGD	−7.12/−7.63	84.97/84.80	−32.22/−32.17/−32.49
SDZ	−8.28/−8.46	82.28/81.85	−32.58/−32.87/−32.81
SMZ	−8.30/−9.43	81.19/79.47	−33.06/−33.23/−33.24

---

By using the intrinsic fluorescence of BSA (λ_Ex/λ_Em = 280 nm/340 nm), the classic fluorescence quenching (FQ) method^38,49–51 is also applied to the interaction study of BSA (0.5 μmol L⁻¹) with antibiotics within ranges of 0.05–5.0 μmol L⁻¹ (Table S1†), showing that the binding parameters K and n (Table 2, 298 K; Table S1,† 288/308 K), and the thermodynamic parameters ΔH, ΔS and ΔG (Table 3, 298 K; Table S2,† 288/308 K) are very close to the results from FI-CL analysis, demonstrating that the K values by FQ method and the proposed procedure have no significant differences. It is clear that the proposed FI-CL analysis probed by luminol is more attractive and convenient with the conspicuous merit of sensitivity, which is higher than that of classic FQ method at least three orders of magnitudes.

3.7. P–L docking analysis for antibiotics binding to BSA

The results of antibiotics to BSA docking analysis are listed in Tables 2–4. It is found that antibiotics should locate in the pocket at subdomain IIA of BSA, and the binding modes of ERY, TET and SNM to BSA as examples are shown in Fig. 3 (other antibiotics to BSA are given in Fig. S3†). For macrolides, the binding cavity is observed to be mainly formed by hydrophobic residues Trp213, Leu237, Leu259, Ile263, Ile289 and Ala290, which suggests that hydrophobic interaction might be the dominant driving force, with the existence of H-bonds within 2.14–3.16 Å. For sulfonamides, docking results clearly reveal that they keep “V” style in the binding pocket, showing strong electrostatic interaction with residues His241, Arg256 and Ser286 of BSA, with H-bonds of average length of 3.07 Å. For tetracyclines, results reveal that they mainly have H-bonds towards residues Tyr149, Arg194, Arg198, Arg217, His241 and Arg256 of BSA; and the numbers of H-bonds for TET, OXY and CHL to BSA are 2, 3 and 4 with average length of 3.32, 2.99 and 2.74 Å, respectively, which indicates the effect of H-bonding on tetracyclines binding to BSA might obey the order of TET < OXY < CHL.

Table 4 The H-bonding sites of antibiotics to BSA

H-bond in antibiotic	H-bond in BSA	Length Å	H-bond in antibiotic	H-bond in BSA	Length Å
ERY11O	Arg256NH2	2.21	CHL2O	His241NE2	2.87
ERY14O	Arg256NH2	2.31	CHL4O	Agr198NH2	2.29
ERY7O	Arg217NE	2.23	CHL7O	Agr194NH1	2.64
KIT8O	Arg217NE	2.14	SNM4′O	Tyr149OH	3.27
MID7O	Arg217NH2	2.99	SNM4′O	Arg256NE	3.05
JOS7O	Arg256NE	2.44	SNM4O	Arg256NH2	2.88
ERE7O	Arg217NH2	3.05	SGD4′O	His241NE2	3.06
JOP8O	Arg217NE	2.91	SGD4O	Arg256NH2	2.78
MIA14-2O	Arg194NH1	3.16	SGD1NH	Arg256O	3.41
TET2O	Tyr149OH	3.35	SDZ4′O	Tyr149OH	2.89
TET4O	Arg198NE2	3.28	SDZ4′O	Arg256NE	3.32
OXY1O	Arg217NH2	2.28	SDZ5H	Ser286O	2.94
OXY2O	Arg217NE	2.86	SMZ4′O	Arg217NE	2.96
OXY10O	Arg256NE	3.33	SMZ1N	His241NE2	3.06
CHL1O	Tyr149OH	3.15	SMZ5NH	Ile289O	3.18

---

Fig. 3 Results of antibiotics to BSA docking analysis, panels A–C present the three dimensional views of BSA binding with ERY, TET and SNM by Pymol, respectively. The residues of BSA are shown using line model and antibiotics are shown using stick-ball model. The H-bond is represented using black dashed line. Panels a–c present the two dimensional views of the corresponding hydrophobic interactions and H-bonds for BSA binding with ERY, TET and SNM by Ligplus. The residues and antibiotic atoms involved in hydrophobic interactions are highlighted using spokewise arcs and spokes, respectively. The H-bond is depicted using olive green dashed line.

It is clear that the binding constants K of 10³ to 10⁵ L mol⁻¹ (Table 2) and binding free energies ΔG of −33.58 to −18.75 kJ mol⁻¹ (Table 3) from P–L docking using Lamarckian Genetic Algorithm as the scoring algorithm are extremely close to the results from the proposed FI-CL analysis and follow the same increasing order.

3.8. The structural effect of antibiotics on the binding constants to BSA

It is discovered that the groups at R₁–R₄ in macrolides show hydrophobic contacts with BSA at different degrees, with the numbers of involved residues being 0, 4, 5, 6, 7, 11 and 13 for ERY, KIT, MID, JOS, ERE, JOP and MIA, respectively (Table S3†), which indicates that the ester groups substituting for –OH might differently affect the binding behaviors of macrolides to BSA. An additional ester group –OOCC₃H₇ or –OOC(CH₂)₂COOC₂H₅ in macrolides enhances the K about 3.1 to 42.8 times. Obviously, ERY with no ester group as the substitution for –OH has the lowest K of 2.30 × 10³ L mol⁻¹, and the K values of KIT, MID, JOS, ERE, JOP and MIA to BSA increase by the factors of 3.1, 6.5, 18.6, 42.8, 45.7 and 49.6 to that of ERY, respectively (Fig. 4). Apparently, OXY with R₂ as –OH, which generates an additional H-bond with Arg256 of BSA (Fig. S3G and S3g†), shows the K 1.1 times to that of TET; while CHL with R₁ as –Cl, being capable of improving the binding geometry with BSA⁵² and resulting in two additional H-bonds (Fig. S3H and S3h†), gives the K 1.3 times to that of TET. It is also discovered that the differences of R₁ group in sulfonamides obviously influence their steric conformations in the binding sites at BSA. The –C(=NH)NH₂ moiety substituting for –H at R₁ in SGD decreases the angle of “V” style from 109.5° to 101.1°, leading to the K about 1.1 times to that of SNM, and the further cyclization of –C(=NH)NH₂ to pyrimidine ring in SDZ with the angle of “V” style of 91.9° shows the K 1.3 times to that of SNM. However, the two electron-donating –CH₃ moieties at pyrimidine ring in SMZ declines the angle of “V” style to 84.7°, lead to the K 1.5 times to the value of SNM.

Fig. 4 The enhancement of ester groups substituting for –OH on the K of macrolides binding to BSA.

3.9. The correlations of log K vs. physicochemical descriptors of antibiotics

It is interestingly found that the log K values have good correlations to antibiotics' physicochemical descriptors including MV, MR, POL (data from ACD/I-LaB, http://ilab.acdlabs.com/iLab2/index.php) and log P (data from PubChem Public Chemical Database, http://pubchem.ncbi.nlm.nih.gov/) in Table 5, revealing that molecular size, steric property and hydrophobicity of antibiotics are essential factors for their binding affinities to BSA. It is worth noting that the slope values of log K vs. MR and POL follow the order of macrolides > tetracyclines > sulfonamides, suggesting the conformational stabilities of BSA–antibiotics complexes obey the opposite order;⁵³ and the results of log K vs. log P demonstrate hydrophobic effect on macrolides binding to BSA is more significant than tetracyclines and sulfonamides binding to BSA.

Table 5 Relationships of log K vs. MV, MR, POL and log P of antibiotics

Descriptor	Relationship	r
a Macrolides.b tetracyclines.c sulfonamides.
log K vs. MV	log K = 0.013MV − 4.8^a	0.9791
	log K = 0.020MV − 0.3^b	0.9905
	log K = 0.0014MV + 5.5^c	0.9705
log K vs. MR	log K = 0.052MR − 6.5^a	0.9781
	log K = 0.045MR + 0.1^b	0.9891
	log K = 0.0038MR + 5.4^c	0.9849
log K vs. POL	log K = 0.013POL − 6.5^a	0.9788
	log K = 0.012POL + 0.2^b	0.9969
	log K = 0.0099POL + 5.5^c	0.9894
log K vs. log P	log K = 0.82 log P + 2.1^a	0.9809
	log K = 0.27 log P + 5.7^b	0.9853
	log K = 0.097 log P + 5.7^c	0.9759

---

4. Conclusion

This work characterizes the binding behaviors of antibiotics to BSA at picomolar-level using FI-CL analysis for the first time. It is important to note that BSA and antibiotics form 1 : 1 complexes, with the K values ranking in order of macrolides (ERY < KIT < MID < JOS < ERE < JOP < MIA) < tetracyclines (TET < OXY < CHL) < sulfonamides (SNM < SGD < SDZ < SMZ). It is also found that the K values of antibiotics to BSA are significantly influenced by ester groups in macrolides, –OH or –Cl in tetracyclines, and –C(=NH)NH₂ in sulfonamides, and show high correlations to antibiotics' molecular size, steric property and hydrophobicity. The antibiotics to BSA docking study clearly reveals that antibiotics should enter into the pocket at subdomain IIA of BSA, yielding interaction parameters and binding modes in accordant with the results from FI-CL analysis. This comprehensive investigation with an insight into molecular structure–binding affinity relationship is helpful in understanding the binding property and mechanism of BSA–antibiotic interaction, and provides valuable information for new antibiotic designation.

Declaration of interest

The authors declare there is no conflict of interest.

Acknowledgements

The authors gratefully acknowledge the financial support from the National Nature Science Foundation of China (no. 21275118), the Open Funds from the Key Laboratory of Synthetic and Natural Functional Molecule Chemistry of Ministry of Education, China.

References

1. D. Stepensky, Eur. J. Pharm. Sci., 2011, 42, 91–98.
2. S. Schmidt, D. Gonzalez and H. Derendorf, J. Pharm. Sci., 2010, 99, 1107–1122.
3. J. A. Roberts, F. Pea and J. Lipman, Clin. Pharmacokinet., 2013, 52, 1–8.
4. S. Mizutani, E. Pauwels, V. Stoven, S. Goto and Y. Yamanishi, Bioinformatics, 2012, 28, 522–528.
5. F. Yang, Z. Y. Ma, Y. Zhang, G. Q. Li, M. Li, J. K. Qin, O. Lockridge and H. Liang, Eur. J. Pharm. Biopharm., 2013, 84, 549–554.
6. G. L. Warren, T. D. Do, B. P. Kelley, A. Nicholls and S. D. Warren, Drug Discovery Today, 2012, 17, 1270–1281.
7. S. M. Huang, H. Zhao, J. I. Lee, K. Reynolds, L. Zhang, R. Temple and L. J. Lesko, Clin. Pharmacol. Ther. Ser., 2010, 87, 497–503.
8. A. Genoni, M. Pennati, G. Morra, N. Zaffaroni and G. Colombo, RSC Adv., 2012, 2, 4268–4282.
9. C. M. Cheatum, Nat. Chem., 2013, 5, 152–153.
10. K. Vuignier, J. L. Veuthey, P. A. Carrupt and J. Schappler, Drug Discovery Today, 2013, 18, 21–22.
11. G. M. West, C. L. Tucker, T. Xu, S. K. Park, X. M. Han, J. R. Yates III and M. C. Fitzgerald, Proc. Natl. Acad. Sci. U. S. A., 2010, 107, 9078–9082.
12. A. Genoni, M. Pennati, G. Morra, N. Zaffaroni and G. Colombo, RSC Adv., 2012, 2, 4268–4282.
13. O. K. Abou-Zied, RSC Adv., 2013, 3, 8747–8755.
14. X. Y. Cao, D. B. Dong, J. L. Liu, C. Y. Jia, W. Liu and W. Yang, Chemosphere, 2013 DOI:10.1016/j.chemosphere.2012.12.069.
15. P. Bourassa, S. Dubeau, G. M. Maharvi, A. H. Fauq, T. J. Thomas and H. A. Tajmir-Riahi, Eur. J. Med. Chem., 2011, 46, 4344–4353.
16. B. Kaboudin, K. Moradi, M. R. Faghihi and F. Mohammadi, J. Lumin., 2013, 139, 104–112.
17. S. Patra, K. Santhosh, A. Pabbathi and A. Samanta, RSC Adv., 2012, 2, 6079–6086.
18. F. F. Tian, J. H. Li, F. L. Jiang, X. L. Han, C. Xiang, Y. S. Ge, L. L. Li and Y. Liu, RSC Adv., 2012, 2, 501–513.
19. H. R. Lv, X. J. Tan, Y. Zhang and Z. H. Song, Spectrosc. – Int. J., 2011, 26, 349–355.
20. Z. Z. Li, G. L. Jiao, G. D. Sun, L. Song and F. Sheng, J. Biochem. Mol. Toxicol., 2012, 26, 331–336.
21. A. D. Bani-Yaseen, J. Lumin., 2011, 131, 1042–1047.
22. H. A. Kirst, Van Nostrand's Encyclopedia of Chemistry, Wiley, New York, 5th edn, 2005.
23. M. O. Griffin, E. Fricovsky, G. Ceballos and F. Villarreal, Am. J. Physiol.: Cell Physiol., 2010, 299, 539–548.
24. J. W. Spoo and J. E. Riviere, in Veterinary Pharmacology and Therapeutics, ed. H. R. Adams, Blackwell Publishing, Iowa, 8th edn, 2001, p. 796.
25. P. Zarogoulidis, N. Papanas, I. Kioumis, E. Chatzaki, E. Maltezos and K. Zarogoulidis, Eur. J. Clin. Pharmacol., 2012, 68, 479–503.
26. C. Richards, L. Pantanowitz and B. J. Dezube, Pharmacol. Res., 2011, 63, 151–156.
27. S. Jones, Nat. Biotechnol., 2012, 30, 333.
28. J. B. Locke, J. Finn, M. Hilgers, G. Morales, S. Rahawi, G. C. Kedar, J. J. Picazo, W. Im, K. J. Shaw and J. L. Stein, Antimicrob. Agents Chemother., 2010, 54, 5337–5343.
29. J. B. Xiao and G. Y. Kai, Crit. Rev. Food Sci. Nutr., 2012, 52, 85–101.
30. S. Pal and C. Saha, J. Biomol. Struct. Dyn., 2013 DOI:10.1080/07391102.2013.811700.
31. C. Coderch, A. Morreale and F. Gago, Anti-Cancer Agents Med. Chem., 2012, 12, 219–225.
32. J. B. Xiao, G. Y. Kai, X. L. Ni, F. Yang and X. Q. Chen, Mol. BioSyst., 2011, 7, 1883–1890.
33. N. Kanoh, T. Suzuki, M. Kawatani, Y. Katou, H. Osada and Y. Iwabuchi, Bioconjugate Chem., 2013, 24, 44–52.
34. A. Lespine, S. Martin, J. Dupuy, A. Roulet, T. Pineau, S. Orlowski and M. Alvinerie, Eur. J. Pharm. Sci., 2007, 30, 84–94.
35. I. Miyazaki, H. Okumura, S. Simizu, Y. Takahashi, N. Kanoh, Y. Muraoka, Y. Nonomura and H. Osada, ChemBioChem, 2009, 10, 845–852.
36. S. Q. Harper, Proc. Natl. Acad. Sci. U. S. A., 2013, 110, 387–388.
37. P. J. Ballester and J. B. O. Mitchell, Bioinformatics, 2010, 26, 1169–1175.
38. A. Manna and S. Chakravorti, Mol. BioSyst., 2013, 9, 246–257.
39. Y. Li, W. M. Gao, F. Li, J. H. Wang, J. X. Zhang, Y. F. Yang, S. W. Zhang and L. Yang, Mol. BioSyst., 2013, 9, 2266–2281.
40. A. P. Norgan, P. K. Coffman, J. P. Kocher, D. J. Katzmann and C. P. Sosa, J. Cheminf., 2011, 3,  DOI:10.1186/1758-2946-3-12.
41. N. Shahabadi and S. M. Fili, Spectrochim. Acta, Part A, 2014, 118, 422–429.
42. D. Xiao, L. L. Zhang, Q. Wang, X. Lin, J. Y. Sun and H. Li, J. Lumin., 2014, 46, 218–225.
43. X. J. Tan, Z. H. Song, D. H. Chen and Z. M. Wang, Spectrochim. Acta, Part A, 2011, 79, 232–255.
44. Z. M. Wang, Z. H. Song and D. H. Chen, Talanta, 2010, 83, 312–319.
45. Z. M. Wang and Z. H. Song, Analyst, 2010, 135, 2546–2553.
46. L. M. Berezhkovskiy, J. Pharm. Sci., 2007, 96, 249–257.
47. J. H. Van't Hoff, Studies of Chemical Dynamics, Kessinger Publishing, Montana, 1984.
48. P. D. Ross and S. Subramanian, Biochemistry, 1981, 20, 3096–3102.
49. J. R. Lakowicz, Principles of Fluorescence Spectroscopy, Springer, US, New York, 3rd edn, 2006.
50. J. R. Blinks, D. G. Allen, F. G. Prendergast and G. C. Harrer, Life Sci., 1978, 22, 1237–1244.
51. D. Sarkar, RSC Adv., 2013, 3, 24389–24399.
52. M. Kolář, P. Hobza and A. K. Bronowska, Chem. Commun., 2013, 49, 981–983.
53. C. Hansch and T. E. Klein, Acc. Chem. Res., 1986, 19, 392–400.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/c3ra45885g

---