Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 217 (2019) 155–163

Contents lists available at ScienceDirect

Spectrochimica Acta Part A: Molecular and BiomolecularSpectroscopy

j ourna l homepage: www.e lsev ie r .com/ locate /saa

Polymyxins interaction to the human serum albumin: A thermodynamicand computational study

A. Poursoleiman a, M.H. Karimi-Jafari a, Z. Zolmajd-Haghighi a, M. Bagheri a, T. Haertlé b,c, G. Rezaei Behbehani d,A. Ghasemi a, Y.Y. Stroylova e, V.I. Muronetz e, A.A. Saboury a,⁎a Institute of Biochemistry and Biophysics, University of Tehran, Tehran, Iranb Poznan University of Life Sciences, Department of Animal Nutrition, Poznan, Polandc Biopolymers, Interactions, Assemblies, UR 1268, Institute National de la Recherche Agronomique, Nantes, Franced Chemistry Department, Imam Khomeini International University, Qazvin, Irane Belozersky Institute of Physico-Chemical Biology, Lomonosov Moscow State University, Moscow, Russia

⁎ Corresponding author at: Institute of BiochemistryTehran, P.O. Box 13145-1384, Tehran, Iran.

E-mail address: Saboury@ut.ac.ir (A.A. Saboury).

https://doi.org/10.1016/j.saa.2019.03.0771386-1425/© 2019 Elsevier B.V. All rights reserved.

a b s t r a c t

a r t i c l e i n f o

Article history:Received 18 October 2018Received in revised form 18 March 2019Accepted 22 March 2019Available online 25 March 2019

Polymyxin B and E (colistin), are a groupof cationic charged cyclic antibiotic lipopeptides that are frequently usedin the clinics to treat infections caused by the multidrug-resistant gram-negative bacteria. Since the interactionswith the blood plasma drug-transport proteins may play a critical role in determining their pharmacological andpharmacokinetic profiles, we studied the binding properties of polymyxins to the human serum albumin (HSA)under simulated physiological conditions by the combination of biophysical approaches, such as isothermal titra-tion calorimetry (ITC),fluorescence anisotropy, circular dichroism (CD) buttressed by computational studies. TheHSA binding to the polymyxins was relatively strong (Ka ≈ 1.0 × 107 M−1). Molecular docking indicated thatpolymyxins bind to the cleft of HSA between domains I and III via the electrostatic interactions. This evidencewas further confirmed by the entropy-driven interaction for the polymyxins bound HSA. Far UV-CD experimentsshowed that the secondary structure of HSA doesn't alter and its stable structure is preserved. Collectively, theseinvestigations revealed that the polymyxins bind preferentially to the partially unfolded intermediate forms ofthe protein structure; however, HSA molecule does not undergo any significant conformational changes uponbinding. This is promising as it may limit the unfavorable side effects of the medicine. On the whole, the resultsprovide quantitative and qualitative insight of the binding interaction between HSA and polymyxins, which isimportant in understanding their effect as therapeutic agents.

© 2019 Elsevier B.V. All rights reserved.

Keywords:ColistinPolymyxin BIsothermal titration calorimetry (ITC)Fluorescence anisotropySimulated-annealing

1. Introduction

The increasing emergence of resistant bacteria to available antibi-otics and the lack of new antimicrobial agents poses a major challengeto medical researches [1]. Multidrug-resistant (MDR) gram-negativebacteria like Pseudomonas aeruginosa, Acinetobacter baumannii, andKlebsiella pneumoniae are on a hit-list of the most dangerous pathogensand have been identified as one of the 3 greatest threats to humanhealth, according to the World Health Organization (WHO) and Infec-tious Diseases Society of America (IDSA) [2,3].

Considering the fact that introduction of new antibiotics activeagainst gram-negative bacteria is unlikely in the near future [4], it is im-portant to optimize the activity of already existing drugs, such as poly-myxins, which are the last-ditch antibiotics for defense against

and Biophysics, University of

untreatable serious infections [5]. Efficacy of these antibiotics can be im-proved by understanding their pharmacodynamics and pharmacokinet-ics, chemistry, side effects and location of the binding interaction. Forthis purpose, the study of interactions of these peptides with plasmaproteins at the molecular level is seriously needed [6].

Polymyxins are a group of antibiotics that are composed of a posi-tively charged cyclic heptapeptide ring (due to the presence of 5-L-diaminobutyric acid (Dab), and γ-amino groups) and of a tripeptidechain acylated on its N end with a long fatty acid (Fig. 1). Polymyxin B(PMB) and E (PME) are used currently to treat infections caused bymultidrug-resistant (MDR) gram-negative bacteria. Polymyxin B andcolistin differ in amino acid N° 6, which is DPhe in PMB and DLeu inPME [7,8].

Ligand binding to plasma proteins is a reversible reaction and thereis usually an equilibrium between the concentrations of unbound andbound drugs. Among many components in the human blood plasma(e.g., HSA, α-1-acid glycoprotein (AGP), globulins and lipoproteins),which interact with drugs, the serum albumin is the most importantcandidate protein because it can bind drugs in relatively high quantities

Fig. 1. Chemical structures of PMB and PME. These peptides contain 10 amino acids displayed by the three-letter code for amino acids, and DAB is 1,3-diamino butyric acid. R in PMB is D-Phe and in PME is D-Leu. DAB4 has an additional amide linkage between its 3-amino group and the C-terminal carboxylate of Thr10 to form a cycle. The N terminus of DAB1 is end-cappedwith a long alkyl chain.

156 A. Poursoleiman et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 217 (2019) 155–163

[9]. TheHSA has amolecularweight of 66.5 kDawith a usual concentra-tion of around 42 mg/mL. It is involved in binding, transportation,deposition, distribution and metabolism of a range of endogenousand exogenous (e.g. drugs) ligands, including fatty acids, aminoacids, nutrients, steroids, hormones, enzymes, metal ion and numer-ous pharmaceuticals [10]. The crystal structure of the HSA revealed asingle non-glycosylated chain protein with 585 amino acids residuecomposed of three structurally homologous α-helical domains (I, II,and III), each consisting of two subdomains (named A and B). Also,the HSA consists of an equilateral triangle with sides of about 80 Åand a depth of about 30 Å with 17 disulfide bridges, one free thiol(Cys34) and a single tryptophan residue (Trp214), which are locatedin domain IIA [11]. Crystallographic studies revealed several drugbinding sites in the HSA including: sites 1, 2 and 3 (located insubdomains IIA, IIIA and IB, respectively), site 4 (betweensubdomains IIB and IIA), site 5 (between subdomains IB and IIA)and site 6 (on the interface between subdomains IIIA and IIIB).Also, there is a cleft between domains I and III. Besides, nine fattyacid sites (FAs) are localized on the HSA [12,13].

Determination of the equilibrium (drug binding) constants is one ofthe major aspects of pharmacokinetic and pharmacodynamics studies[14] because only the free drug fraction is able to exert the biological ac-tion and/or to show pharmacological and toxicological effects [9,15]. Asdisruption of biological activity and the possible side effectwould be theresult of irreversible structural changes in the protein molecule, furtherresearch and thorough knowledge around plasma protein and drugs in-teractions, is extremely needed to find the optimal dose of drug admin-istration [16].

As the pharmacology of polymyxins is complex and there is an ev-ident scarcity of pharmacological and pharmacokinetic knowledgeabout the behavior and fate of polymyxins [5,17,18]. Therefore,deeper understanding of the plasma protein binding of polymyxinsis badly needed. Several studies of the binding of cationic peptidesby plasma proteins have been reported for example binding of poly-myxins by BSA (Bovine Serum Albumin) and AGP and binding ofshort cationic antimicrobial peptides by HSA [19,20]. Investigationsof polymyxins interactions with plasma proteins in rats show thatover 50% of polymyxins is bound by plasma proteins and the circulat-ing drug is a minor fraction of the total concentration of the drug[19,21]. No study to date has examined the conformational changesand mechanism of the interactions of polymyxins with the HSA. Forthis reason, we have studied these interactions using isothermal ti-tration calorimetry, circular dichroism spectroscopy, fluorescenceanisotropy and molecular modeling.

2. Materials and methods

2.1. Materials

Albumin from human serum lyophilized powder, essentially fattyacid free and globulin-free (N99% purity), dialysis membranes (cut-off,12,000 Da), polymyxin E, and polymyxin B were purchased from theSigma-Aldrich (USA). Dimethyl sulfoxide (DMSO) and Nile Red werefrom Merck (Germany). Fluorescein 5-isothiocyanate (FITC) was ob-tained from Thermo-Fisher Scientific (USA).

2.2. Preparation of solutions

The HSA stock solution (20 μM) was freshly prepared in the phos-phate buffer of pH7.4. The pHwas adjustedwith a suitably standardizedusing Orion-868 pH meter (Thermo Scientific, USA) at room tempera-ture. The concentration of HSA was measured by NanoDrop 2000/2000c spectrophotometer (Thermo Fisher Scientific, USA) using the ex-tinction coefficients of (36,500 M−1 cm−1) at 280 nm [22]. The poly-myxins' working solution was prepared by dissolving appropriateamounts of the peptides in the same buffer.

2.3. Isothermal titration calorimetry (ITC)

Experiments on binding interaction of the polymyxins with HSAwere performed using a VP-ITC Microcalorimeter (MicroCal, LLC,USA). The sample cell was filled with 1.8 mL of protein solution(0.21mM) and the syringewas filled with 300 μL of polymyxin solution(2 mM). The samples were degassed under vacuum prior to using forITC runs. The reference cell was filled with 0.1 M phosphate buffer(pH7.4). The systemwas allowed to equilibrate and a stable baseline re-corded before initiating an automated titration. The titration of the pro-tein with polymyxins involved 30 consecutive injections of 10 μLpolymyxin solution with 3-min intervals. The first injection was set to5 μL. The cell content was stirred constantly at 307 rpm. The ITC titra-tionswere performed at 25 °C. The enthalpies of colistin+HSA interac-tions, ΔH, and the enthalpies of dilution for colistin in the buffersolution, ΔHdilut, are listed in Supporting information Table S1.

2.4. Circular dichroism (CD)

The circular dichroic spectra were recorded on Aviv 215 Spectropo-larimeter (USA) in the far-UV region (190–260 nm) using a 1-mmquartz cells at 25 °C. The molar ratio of polymyxin analogs to HSA was

157A. Poursoleiman et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 217 (2019) 155–163

changed from 17 to 50. The signal of buffer solution as reference wassubtracted from the sample spectra. The results were expressed as themolar ellipticity in [θ]λ (deg cm2 dmol−1), which is based on a meanamino acid residue weight of about 114 (MRW). The molar ellipticitywas determined as [θ]λ=( θobs × MRW / 10 dc), where θobs is the ob-served ellipticity (in degrees) at a given wavelength, c is the proteinconcentration (in units of g/mL) and d is the length of the light path(cm) [23]. Percentage of secondary structures was performed usingthe CDNN CD spectra deconvolution software (version 2.1) [24].

2.5. Fluorescence anisotropy measurements

Lysine residues on the surface of HSA molecules were labeled ac-cording to the previously mentioned protocol [25]. The labeling wasachieved as follows: protein was dissolved in phosphate buffer(pH 9.0). Subsequently, the FITC was dissolved in a few drops of di-methyl sulfoxide and added to the protein solution. Then, tube waswrapped in foil and incubated at room temperature in the dark undergentle stirring during 8 h. Then, the solution was first extensively dia-lyzed against 0.1M phosphate buffer (pH 7.0) to remove free unreacteddyes using dialysis membrane (cut-off, 12,000 Da) and immediatelyused. Dialysis was considered complete when free FITC fluorescence inthe outer solution was no longer detectable [25]. Fluorescence spectraand steady-state emission anisotropies of all HSA samples were deter-mined with a Cary Eclipse fluorimeter (Agilent, Palo Alto, CA). Anisot-ropy was measured with a manual polarizer and the G factorcorrection was done manually, with a reference fluorophore.

Stoichiometry of labeling for HSA-FITC was determined by measur-ing the absorbance of the solutions at 280 and 490 nm and using thesame extinction coefficients as above for proteins and FITC. The F/Pratio of labeled HSA samples was computed according to standardmethods [25–27]. The anisotropy measurement of the labeled proteinwas then done in the presence of ligands (PMB and PME). (All thedata are the average of three repeated measurements of anisotropy ±SD).

2.6. Anisotropy of Nile Red

To investigate the binding interaction of the peptides and HSA, thepeptides were labeled with Nile Red and the anisotropy of the com-pounds was measured using a Cary Eclipse fluorimeter (Agilent, PaloAlto, CA) equipped with a pair of manual polarizers. The anisotropy ofNile Red solution in ethanol was measured. The solution was thenadded to the peptides and the anisotropy was measured again. After-wards, the labeled peptides were added to the protein and the anisot-ropy was measured again after 3 min of incubation. All the data arethe average of five repeated measurements of anisotropy ± SD.

2.7. Computational section

2.7.1. CHARMM Recipes for PMB and PMEThe initial 3D models of the peptides were built from scratch since

there is not any experimentally determined geometry for PMB andPME. All residues were built in natural L form with the exception ofPhe6 or Leu6 which were considered as D-amino acids in PMB andPME, respectively. Available topologies and parameters in CHARMM27forcefield [28,29] (NAMD2.10 program) [30]were extended to supportthe unnatural structural components available in considered cyclic pep-tides. By analogy with lysine, a new CHARMM residue was defined forthe unusual amino acid DAB. As seen in Supporting informationFig. S1, two middle charge groups of lysine that contain CG and CDwere deleted and atoms of the last charged groupwere renamed to cre-ate a new topology for DAB. A new CHARMM topology patch was alsodefined to tackle the unusual cyclic structure of peptides which is nota normal N- to C-terminal attachment. Instead, the cyclic topology isformed via an amide bond between the amine group in side-chain of

DAB4 and the C-terminal of Thr10 (see Fig. 1 and Supporting informa-tion Fig. S2). This attachment necessitated some new parameters andall of themwere extracted by analogy from other amino acids includingalanine, glycine and threonine (see Supporting information Fig. S2). An-other CHARMM topology patch was also defined for capping of the N-terminal (DAB1) by a fatty acid chain (6-methyloctanoyic acid). All ofthe new topologies and parameter definitions are provided inSupporting materials. The CHARMM-compatible recipes for PMB andPME peptides were used in subsequent MD and SAMD simulations.

2.7.2. Sampling of ligand conformations in waterFull consideration of the flexibility of ligand and protein is a major

challenge in calculation of energetic and structure of interactions. Thisissue becomes more critical in the case of peptide structures withmany torsional degrees of freedom and results in the so called combina-torial explosion of search space in docking problem [31–33]. To avoidthis problem, a conformational search of PMB and PME in water wasconsidered to obtain stable conformations of these peptides and the di-verse set of generated structures were used separately in docking simu-lations. The SAMD simulations of peptides were performed in a cubicbox of TIP3P water models [34] with box size of 52 × 52 × 52 Å. Eachsystem was neutralized by addition of 5 chloride ions and minimizedfor 10,000 steps of conjugated gradient method.

All MD simulations were performed with NAMD 2.10 package usingthe CHARMM27 force field for proteins with the above mentioned ex-tensions made to support PMB and PME peptides. An equilibrationphase was then performed for 200 ps of NPT dynamics at 300 K and1 atm. A time step of 2 fs was taken with SHAKE algorithm [35] tokeep frozen all hydrogen-heavy atom bonds. Berendsen pressure bathcoupling [36] and Langevin dynamics [37,38] were used to maintainpressure and temperature around their reference values, respectively.A cutoff of 12 Å was used for non-bonded interactions with a switchingscheme started at 10 Å. The particle mesh Ewald method [39] was usedin calculation of long-range electrostatic interactions over adopted peri-odic boundary conditions. After equilibration,five rounds of SAMDwereconducted separately that each of them includes 100 cycles of coolingthe system from 800 to 300 K with a rate of 5 K/ps and minimizationof the final structure for 1000 steps of conjugated gradient. Each cyclewas started by a new set of randomized velocities to enhance the diver-sity of conformational search and the final geometrywas checked to en-sure correct chirality of all amino acids. The whole procedure results in500 conformations for each of the PMB and PME systems and all of themwere used in subsequent docking simulations.

2.7.3. Docking of PMB and PME conformational ensembleInitial binding modes of PMB and PME to the HSA were searched by

docking of all 500 stable conformations of ligands to the protein. A sin-gle chain complete structure of HSA was extracted from reported X-raystructure of protein (PDB ID 1AO6 [40]). Missing hydrogen atoms wereadded to the structure using the REDUCE program [41] which also de-cides on preferred protonation states of amino acids. The protein wasthen solvated in a box of water and minimized for 20,000 steps of con-jugated gradient method while harmonic restraints were imposed onposition of backbone atoms. The minimized structure was used indocking simulations. Since there was no predefined binding site forPMB and PME peptides, the whole structure of protein was consideredto be searched by defining three docking boxes with enough overlap(see Supporting information Fig. S3). Using 500 conformations of pep-tide, this results in 1500 docking experiments for each of the PMB orPME systems. The Autodock Vina docking software [42] were usedwith an exhaustiveness value of 200. Input files of docking were pre-pared automatically with Python scripts provided by MGLTools [43].Predicted complex geometries with best binding affinity on eachdocking experiment were collected together. This results in a complexensemble of 1500 peptide-protein structures. The RMSDof alpha carbonatoms of peptide structures within a cutoff of 10 Å were used to cluster

Table 1Binding parameters for polymyxins-HSA interactions computed using Eq. (1).

HSA-PME HSA-PMB

ρ 0.76 0.86δA° 0.66 ± 0.02 0.61 ± 0.02δB° 0.50 ± 0.01 0.91 ± 0.01ΔH°/kJ mol−1 0.38 ± 0.04 −0.55 ± 0.04TΔS°/kJ mol−1 41.21 ± 0.33 41.00 ± 0.33ΔG°/kJ mol−1 −40.58 ± 0.43 −40 ± 0.43Ka/mol−1 L 1.31 × 107 ± 1200 1.00 × 107 ± 1100g (number of binding sites) 3 1.25

158 A. Poursoleiman et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 217 (2019) 155–163

binding poses and to identify the most important binding sites of HSAfor PMB and PME. Ten structures with highest affinity over the complexensemble were selected for subsequent refinement with SAMDsimulations.

2.7.4. Refinement of complexes with SAMDGeometry of 10 best scored complexes from docking ensemble were

used as initial structures in SAMD simulations that aim to depict a morerefined picture of the atomistic details of the interaction of HSA withPMB and PMEpeptides. Such simulationswere reported to be successfulin the case of protein-ligand complexes [44]. The whole procedure in-cludes multiple cycles of cooling the system from high temperaturesin order to reach more stable minima over the interaction energy land-scape. Details of SAMD simulations are the same as those used for con-formational search of peptides with following differences. The solventeffects were included by Generalized Born Implicit Solvent (GBIS)model [45–48]. Non-bonded interactions were gradually switched tozero over a range of 14.0–16.0 Å. In GBIS model, the maximum distancebetween atom pairs was set to 14 Å in calculation of Born radius. Usingan RMSD collective variable [49] on alpha carbon atoms, the structure ofHSA was restrained to not deviate drastically from the minimized ge-ometry while the peptide geometry was free to change. The SAMD pro-cedure includes cooling from randomized velocities at 500 K to 50 Kwith a rate of 5 K/ps. For each complex, 100 separate SAMD cycleswere performed and final geometries were minimized. All structureswere analyzed and visualized by VMD 1.9 package [50].

3. Results and discussion

3.1. Experimental section

3.1.1. Isothermal titration calorimetry (ITC)Wehave previously shown that the heats of the biomolecules and li-

gands interactions in the aqueous solvent systems can be described byEq. (1) [51–54].

ΔH ° ¼ ΔHmaxx0B−δ °

A x0ALA þ x

0BLB

� �− δ °

B−δ °A

� �x

0ALA þ x

0BLB

� �x

0B ð1Þ

ΔH° are the enthalpies of polymyxins - HSA interactions, and ΔHmax

is the heat value at the saturation limit with respect to HSA.The parameters δA° and δB° reflect the net effect of ligands on the HSA

conformational changes in the low and high concentrations, respec-tively. The negative values for δA° or δB° indicate that the ligands formdestabilized PME-HSA or PMB-HSA complexes. The p b 1 indicates neg-ative cooperatively of the HSA molecule in binding of the peptides. The

Fig. 2. Comparison between the experimental heats, ΔH°, (dots) for HSA-polymyxinsinteraction and calculated data (lines) via Eq. (1).

binding of onemolecule at one site of HSA reduces the affinity for otherligands. The xB' can be expressed as follows:

x0B ¼ pxB

xA þ pxB¼ ν

gð2Þ

where xB is the molar ratio of either PMB or PME that is obtained fromthe concentration of peptide ligands after each injection ([L]) dividedby their maximum concentration upon saturation of all HSA ([L]max)as follows:

xB ¼ L½ �L½ �max

;xA ¼ 1−xB ð3Þ

and xA is the fraction of unbound HSA. In general, there will be “g” sitesfor binding of polymyxins per HSA molecule and ν is defined as the av-erage moles of bound polymyxins per mole of total HSA. The LA and LBare the relative contributions of the free and bound ligands in the en-thalpies of dilution with the exclusion of HSA. They can be calculatedfrom the enthalpies of dilution of the ligands in buffer,ΔHdilut as follows:

LA ¼ ΔHdilut þ xB∂ΔHdilut

∂xB

� �; LB ¼ ΔHdilut−xA

∂ΔHdilut

∂xB

� �ð4Þ

The enthalpies of HSA and the colistin interactions (ΔH° values,Supporting information Table S1) were fitted to Eq. (1) over thewhole concentrations. Enthalpies of PMB-HSA interactions are notshown. In the fitting procedure, the only adjustable parameter (p) waschanged until the best agreement between the experimental and calcu-lated data was attained (Fig. 2). δA° and δB° parameters are calculatedfrom the coefficients of the second and third terms of Eq. (1). Bindingparameters for the interactions between the peptides and HSA recov-ered from Eq. (1) are reported in Table 1. The Ka is the apparent associ-ation constant as a function of polymyxins concentrations. TheKa valueswas computed according to specified equation [48]. The Gibbs energiescan be obtained as follows:

ΔG ° ¼ −RTlnKa ð5Þ

Table 2Fraction of secondary structure elements as calculated from the respective CD spectra ofpolymyxin B and polymyxin E.

PME PMB

57.2%

57.7%

57.7%

58.1%

58.0%

58.0%

%Helix

23.8%

22.2%

23.0%

22.0%

22.2%

22.2%

%Beta-Sheet

19.0%

20.1%

19.3%

19.9%

19.8%

19.8%

%Random.Coil

Fig. 3. CD spectra of PMB–HSA and PME-HSA system; at the molar ratios of 0,17 and 50 polymyxins/HSA.

159A. Poursoleiman et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 217 (2019) 155–163

where T is absolute temperature (298 K) and R is the gas constant(8.314 J mol−1 K−1). Moreover the TΔS° values were calculated fromthe Gibbs energies. The agreement between the calculated and experi-mental results was striking and gave considerable support to the useof Eq. (1) (Fig. 2).

Very close δA° and δB° values indicate that there is no significantchanges in theHSA structure. p b 1 indicates that colistin and polymyxinB bind preferentially to the partially unfolded forms of the HSAstructure.

The values of the thermodynamic parameters shed light on the in-termolecular forces involved in the interaction between the polymyxinsand HSA. The negative sign of Gibbs energies value suggested spontane-ous nature of the binding reaction. The magnitude and sign of ΔS° andΔH° are used in the predicting kind of forces that may take place in var-ious ligand–protein binding processes, as described below. In overall,

Fig. 4. Profile of dihedral energy changes in the course of reaching to thefirstfive localminimaduin water (b). Number of conformers in each cluster was shown in each box with their represenhighlighted in green. (For interpretation of the references to color in this figure legend, the rea

the interaction between the HSA and polymyxins is driven by entropy.The changes in the molar enthalpy of the interactions are negligibleand the ΔS° is positive for both colistin and polymyxin B binding tothe HSA that points to the significant role of the electrostatic interac-tions as themajor stabilizing forces in the lipopeptide–protein complexformation [55]. The negative values for ΔH° and ΔS° values are consid-ered for the van der Waals force and hydrogen bonding formation inlow dielectric media, thus, the thermodynamic data obtained forpolymyxins–HSA system clearly excluded the participation ofhydrophobic interactions [56]. Taken together, the binding constantsshowed the strong interactions between the peptides and HSA (Ka

≈ 1.0 × 107 M−1 for PMB, and Ka ≈ 1.31 × 107 M−1 for PME; seeTable 1); however, minor changes of HSA structure observed fromstructural analysis by the CD and fluorescence spectroscopies suggeststhe possible role of the electrostatic interactions as the main driving

ring the SAMD sampling (a). Agglomerative clustering of generated conformations of PMBtative structure depicted below them. More similar clusters are grouped in five main setsder is referred to the web version of this article.)

Fig. 5. Potential energy distribution of PMB and PME conformations in water. Representative structures were depicted for selected bins in the lowest, highest and middle energy values.

160 A. Poursoleiman et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 217 (2019) 155–163

force for the peptide-protein association. The above results were furthersupported by our molecular modeling results, as described in the rele-vant section.

3.1.2. Circular dichroism (CD)Circular dichroism (CD) spectroscopy can be helpful to illustrate the

geometrical structure of a protein [57]. The far-UV CD spectroscopy canbe used to find changes in the secondary structure of the protein.Changes in the secondary-structure contents leads to protein structuralalterations, and causes a deprivation of their biological activities. The re-sults of the quantitative analysis using the CDNNprogram [24] are givenin Table 2. The far UV-CD spectrum for the intact protein shows twominima at 208 and 222, which are consistent with available data in

Fig. 6. The results of docking simulations clustered by their pose of interaction. The first five clusthe references to color in this figure legend, the reader is referred to the web version of this ar

the literature [14]. In the presence of different concentrations of drugsat the molar ratios of 0,17 and 50 drugs/HSA (the recommended dailydose of this type of product [58]), the content of α-helix, β-sheet,turns and random coil structures remains without notable change(Fig. 3).

Several previous studies also indicate that conformational changesoccur in HSA due to interaction with drug compounds, for example,binding of the Taxol, Lomefloxacin, Gefitinib, Chlorogenic acid, Ferulicacid, Bicalutamide, Camptothecin, Idarubicin, Teniposide, Etoposide,Palladium(II) complexes of Imidazole and etc. with HSA [6,56,59–62].These interactions result inmajor protein secondary structural changes,known as the side effects of such drugs. In our experiment, the far UV-CD shows that there is a little modification in the secondary structure

ters with higher interaction energies were shown in different colors. (For interpretation ofticle.)

161A. Poursoleiman et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 217 (2019) 155–163

upon complexation of polymyxins. The electrostatic interaction be-tween polymyxins and HSA, can be one of the drugs advantages, sinceitmay cause no change in the second and third structures of the protein.It may also prevent the consequent effects (toxicity and other side ef-fects) of the conformational changes as the protein keeps its stablestructure [16].

3.1.3. Anisotropy of Nile RedNile Red was added to the peptides as a marker to study binding in-

teractions of peptidewith proteins [63]. The anisotropy of unboundNileRed in ethanol was about 0.05 ± 0.002. Upon addition to peptides, theanisotropy increases to 0.10 ± 0.002. The observed increase in anisot-ropywas attributed to binding of Nile Red to the peptide and the conse-quent increase in rotational correlation time of the dye [26,63–66].Upon addition of Nile Red-peptide solution to the protein, however,the anisotropy changed only slightly and raised to 0.12 ± 0.002. Asthe peptides masses were only 2% of the molecular weight of the pro-tein, the rotational correlation time of the labeled peptides shouldhave more extensively increased if there were a binding interaction be-tween the molecules. However, the slight increase in the anisotropy of

Fig. 7. Refined bindingmodes of PMB (a) and PME (b) toHSA and hotspot residues involved in sand protein residues involved in hydrogen bonding are drawn in extended form while those w

the peptides was interpreted as a slight conformational change in theprotein [26,27].

3.1.4. FITC labeled HSA and its interaction with the peptidesThe labeled FITC to protein at ratio of 1.5 showed that the protein

was only slightly labeled with FITC. The anisotropy of the solution ofthe labeled protein was 0.26 ± 0.001 at 512 nm as compared with thevalues 0.23 ± 0.006 and 0.21 ± 0.001 in the presence of PMB andPME, respectively. Themoderate decrease in anisotropies of the samplescould be the sign of resonance energy transfer between the FITC mole-cules labeled to different proteins. This can be an indication of the pro-tein folding in the presence of the polypeptides that had brought thefluorescent dyes to a distance where the chance of resonance energytransfer had increased [25–27,67].

3.2. Computational section

3.2.1. Structural and energetic diversity of conformations of polymyxinsThe simulated annealingmolecular dynamics (SAMD) protocol gen-

erates a diverse set of unbound conformations for PMB and PME. InFig. 4A variation of dihedral energy as a measure of conformational

pecific interactionswith ligand. Both binding sites correspond to the cleft of protein. Ligandith hydrophobic and van der Waals interactions are depicted in a compact form.

162 A. Poursoleiman et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 217 (2019) 155–163

change of peptideswas depicted for thefirstfive rounds of annealing. Asseen, the system overcomes energy barriers and reaches to differentlocal minima in the course of sampling. The resulting conformationswere then clustered hierarchically to obtain patterns of structural simi-larities. An agglomerative algorithm with Ward linkage was used. Thedendrogram of clustering were shown in Fig. 4B and Supporting infor-mation Fig. S4 for PMB and PME, respectively. Cutting the dendrogramat an appropriate level yields twenty clusters with different populations(denoted by numbers in leaf boxes), which their representative struc-tures shown below them. Main structural similarities were highlightedin five separate green areas along the clustering hierarchy. Alongsideconformational variations in the ring, the relative orientation of thetail was important factors of structural variations. While a compactscorpion-like form with the “tail folded on the ring” was more popu-lated for the PMB, there seems to be a minor shift towards more ex-tended structures in the PME. The energetic distribution of conformerswas also plotted in Fig. 5. TherewasN120 kcal/mol energy difference be-tween themost stable conformation and that with highest potential en-ergy, what indicates the success of adopted protocol in sampling fromdifferent regions of the energy landscape of the peptides. The extendedring conformations are preferred energetically to those with twistedring geometry.

3.2.2. Refined binding modes of PMB and PME to HSAClustered peptide-HSA complexes predicted from docking simula-

tions are presented in Fig. 6. Bindingmode of the first most stable com-plexes was shown with different colors. As seen, cluster 1 is morepreferred not only for its binding affinity, but also for its considerablehigher population. This corresponds to the binding of PMB or PME tothe cleft of HSA between domains I and III. Ten best scored dockedposes were then used in SAMD simulations for further refinement andchecking their binding stability. The interaction energies between pep-tides and HSA were plotted in Supporting information Fig. S5. In mostcases, in the course of SAMD simulations a refined geometry withmore negative interaction energy was obtained. Interestingly, the bestscored geometry obtained from docking (the first panel in Supportinginformation Fig. S5) found the most favorable interaction energy (de-noted by black circles) after refinement. However, in some complexesthe peptide was detached from HSA after some steps indicated by thecurve flattened on zero interaction energies. More atomistic details ofcontacts and noncovalent interactions between peptides and HSAwere depicted in Fig. 7. Both PMB and PME form numerous H-bonds,as well as electrostatic interactions with the HSA residues including incommon Leu112, Asp183, Asp187 and Glu425. These interactions areoften provided by the positive nitrogen center on side chain of DAB res-idues. On the other hand, a veryweak hydrophobic interactionwas seenonly between the PME acyl chain and the Leu463 from the HSA asshown in Fig. 7. Despite slight changes in the polymyxins' structure(see Leu6/Phe6 in PMB/PME), the overall pattern of interactions arevery similar for both peptides and the type of interactions do not changearound this position (Lys190, Pro421 for PMB and Ile115, Ile142 forPME). As shown above these results are also consistentwith the analysisof the experimental studies indicating that the binding interaction ofpolymyxins with HSA is dominated mainly by the electrostatic interac-tion and hydrogen binding.

4. Conclusions

One way to investigate the side effects of a drug or new synthesizedcompound, is to study its interaction with the blood carrier proteinssuch as human serum albumin [59]. In the present study, the bindingof colistin and polymyxin B with HSA is investigated for the first time.All assay techniques such as ITC, Far-UV CD spectroscopy, and fluores-cence anisotropy, as well as the computational study were indicativeof minor changes in the HSA conformational structure in interactionswith colistin and polymyxin B. For both lipopeptides, the interaction

with the protein was entropy-driven suggesting that electrostatic inter-actions may play a dominant role. This evidence was further proved byour results from the modeling studies. The electrostatic interaction ofthe polymyxins-HSA is supposed tomainly be as the result of close con-tact between the side chains of Asp/Glu and Dab residues from the HSAand polymyxins, respectively. Besides, the hydrogen bonds betweenHSA residues as shown in Fig. 7 and the positive nitrogen center onside chain of DAB residues are seen as a lateral interaction for thepeptide-protein complex formation [61]. Additionally, the computa-tional studies revealed that polymyxins bind to the cleft of HSA betweendomains I and III. Furthermore, the overall pattern of interaction is verysimilar for both peptides despite a slight structural difference betweenLeu6 and Phe6 where almost had no impact upon the interactionsaround this position; with exceptions to their interacting partners onproteins that are Lys190 and Pro421 for polymyxin B and Ile115 andIle142 for colistin.

Altogether, our results suggested that colistin or polymyxin B bindspreferentially to the partially unfolded intermediate forms of HSA, andthe protein does not undergo substantial conformational changesupon binding. The aforementioned interaction that occurs betweenpolymyxins and HSA is promising as this drug can still be used as thelast antibiotic line [16].

Acknowledgements

This work was financially supported by the Research Council of Uni-versity of Tehran and Iran National Science Foundation (INSF).

Appendix A. Supplementary data

Supplementary data to this article can be found online at https://doi.org/10.1016/j.saa.2019.03.077.

References

[1] S.A. Cochrane, C.T. Lohans, J.R. Brandelli, G. Mulvey, G.D. Armstrong, J.C. Vederas,Synthesis and structure-activity relationship studies of N-terminal analogues ofthe antimicrobial peptide tridecaptin A(1), J. Med. Chem. 57 (3) (2014)1127–1131, https://doi.org/10.1021/jm401779d.

[2] S. Patolia, G. Abate, N. Patel, S. Patolia, S. Frey, Risk factors and outcomes formultidrug-resistant Gram-negative bacilli bacteremia, Ther. Adv. Infect. Dis. 5 (1)(2017) 11–18, https://doi.org/10.1177/2049936117727497.

[3] D. van Duin, D.L. Paterson, Multidrug-resistant bacteria in the community: trendsand lessons learned, Infect. Dis. Clin. N. Am. 30 (2) (2016) 377–390, https://doi.org/10.1016/j.idc.2016.02.004.

[4] L. Tan, Y. Tao, T. Wang, et al., Discovery of novel pyridone-conjugatedmonosulfactams as potent and broad-spectrum antibiotics for multidrug-resistantgram-negative infections, J. Med. Chem. 60 (7) (2017) 2669–2684, https://doi.org/10.1021/acs.jmedchem.6b01261.

[5] L.M. Lim, N. Ly, D. Anderson, et al., Resurgence of colistin: a review of resistance, tox-icity, pharmacodynamics, and dosing, Pharmacotherapy. 30 (12) (2010)1279–1291, https://doi.org/10.1592/phco.30.12.1279.

[6] Z. Wang, J.X. Ho, J.R. Ruble, et al., Structural studies of several clinically importantoncology drugs in complex with human serum albumin, Biochim. Biophys. Acta1830 (12) (2013) 5356–5374, https://doi.org/10.1016/j.bbagen.2013.06.032.

[7] S.-E. Cheah, M.D. Johnson, Y. Zhu, et al., Polymyxin resistance in Acinetobacterbaumannii: genetic mutations and transcriptomic changes in response to clinicallyrelevant dosage regimens, Sci. Rep. 6 (2016), 26233. https://doi.org/10.1038/srep26233.

[8] T. Velkov, Z.Z. Deris, J.X. Huang, et al., Surface changes and polymyxin interactionswith a resistant strain of Klebsiella pneumoniae, Innate Immun. 20 (4) (2014)350–363, https://doi.org/10.1177/1753425913493337.

[9] G. Lambrinidis, T. Vallianatou, A. Tsantili-Kakoulidou, In vitro, in silico and inte-grated strategies for the estimation of plasma protein binding. A review, Adv.Drug Deliv. Rev. 86 (2015) 27–45https://doi.org/10.1016/j.addr.2015.03.011.

[10] A. Lacroix, T.G.W. Edwardson, M.A. Hancock, M.D. Dore, H.F. Sleiman, Developmentof DNA nanostructures for high-affinity binding to human serum albumin, J. Am.Chem. Soc. 139 (21) (2017) 7355–7362, https://doi.org/10.1021/jacs.7b02917.

[11] H. Greige-Gerges, R. Kaissi, J. Magdalou, A. Jraij, Reviewing the binding of a series ofparabens to human serum albumin, Biopharm. Drug Dispos. 34 (3) (2013) 186–194,https://doi.org/10.1002/bdd.1836.

[12] N. Jafari, R. Ahmed, M. Gloyd, J. Bloomfield, P. Britz-McKibbin, G. Melacini, Allostericsensing of fatty acid binding by NMR: application to human serum albumin, J. Med.Chem. 59 (16) (2016) 7457–7465, https://doi.org/10.1021/acs.jmedchem.6b00410.

[13] A. Bijelic, S. Theiner, B.K. Keppler, A. Rompel, X-ray structure analysis of indazoliumtrans-[tetrachlorobis(1H-indazole)ruthenate(III)] (KP1019) bound to human serum

163A. Poursoleiman et al. / Spectrochimica Acta Part A: Molecular and Biomolecular Spectroscopy 217 (2019) 155–163

albumin reveals two ruthenium binding sites and provides insights into the drugbinding mechanism, J. Med. Chem. 59 (12) (2016) 5894–5903, https://doi.org/10.1021/acs.jmedchem.6b00600.

[14] Y. Sun, Z. Ji, X. Liang, et al., Studies on the binding of rhaponticin with human serumalbumin by molecular spectroscopy, modeling and equilibrium dialysis,Spectrochim. Acta A Mol. Biomol. Spectrosc. 87 (Supplement C) (2012) 171–178,https://doi.org/10.1016/j.saa.2011.11.033.

[15] D.A. Smith, L. Di, E.H. Kerns, The effect of plasma protein binding on in vivo efficacy:misconceptions in drug discovery, Nat. Rev. Drug Discov. 9 (12) (2010)929–939https://doi.org/10.1038/nrd3287.

[16] M.A. Rub, J.M. Khan, A.M. Asiri, R.H. Khan, K-ud-Din, Study on the interaction be-tween amphiphilic drug and bovine serum albumin: a thermodynamic and spectro-scopic description, J. Lumin. 155 (2014) 39–46, https://doi.org/10.1016/j.jlumin.2014.06.009.

[17] A. Gallardo-Godoy, C. Muldoon, B. Becker, et al., Activity and predicted nephrotoxi-city of synthetic antibiotics based on polymyxin B, J. Med. Chem. 59 (3) (2016)1068–1077, https://doi.org/10.1021/acs.jmedchem.5b01593.

[18] K.S. Akers, M.P. Rowan, K.L. Niece, et al., Colistin pharmacokinetics in burn patientsduring continuous venovenous hemofiltration, Antimicrob. Agents Chemother. 59(1) (2015) 46–52, https://doi.org/10.1128/aac.03783-14.

[19] M.A.K. Azad, J.X. Huang, M.A. Cooper, et al., Structure–activity relationships for thebinding of polymyxins with human α-1-acid glycoprotein, Biochem. Pharmacol.84 (3) (2012) 278–291, https://doi.org/10.1016/j.bcp.2012.05.004.

[20] A. Sivertsen, J. Isaksson, H.K. Leiros, J. Svenson, J.S. Svendsen, B.O. Brandsdal, Syn-thetic cationic antimicrobial peptides bind with their hydrophobic parts to drugsite II of human serum albumin, BMC Struct. Biol. 14 (2014) 4, https://doi.org/10.1186/1472-6807-14-4.

[21] D. Landman, C. Georgescu, D.A. Martin, J. Quale, Polymyxins revisited, Clin.Microbiol. Rev. 21 (3) (2008) 449–465, https://doi.org/10.1128/cmr.00006-08.

[22] A. Sengupta, R.K. Koninti, K. Gavvala, N. Ballav, P. Hazra, An anticancer drug to probenon-specific protein-DNA interactions, Phys. Chem. Chem. Phys. 16 (9) (2014)3914–3917, https://doi.org/10.1039/C3CP54422B.

[23] S.M. Kelly, N.C. Price, The use of circular dichroism in the investigation of proteinstructure and function, Curr. Protein Pept. Sci. 1 (4) (2000) 349–384.

[24] J.T. Yang, C.S. Wu, H.M. Martinez, Calculation of protein conformation from circulardichroism, Methods Enzymol. 130 (1986) 208–269.

[25] Z. Zolmajd-Haghighi, Q.S. Hanley, When one plus one does not equal two: fluores-cence anisotropy in aggregates and multiply labeled proteins, Biophys. J. 106 (7)(2014) 1457–1466, https://doi.org/10.1016/j.bpj.2014.02.020.

[26] M. Brinkley, A brief survey of methods for preparing protein conjugates with dyes,haptens and crosslinking reagents, Bioconjug. Chem. 3 (1) (1992) 2–13, https://doi.org/10.1021/bc00013a001.

[27] G. Hungerford, J. Benesch, J.F. Mano, R.L. Reis, Effect of the labelling ratio on thephotophysics of fluorescein isothiocyanate (FITC) conjugated to bovine serum albu-min, Photochem. Photobiol. Sci. 6 (2) (2007) 152–158, https://doi.org/10.1039/b612870j.

[28] A.D. Mackerell Jr., M. Feig, C.L. Brooks 3rd, Extending the treatment of backbone en-ergetics in protein force fields: limitations of gas-phase quantum mechanics in re-producing protein conformational distributions in molecular dynamicssimulations, J. Comput. Chem. 25 (11) (2004) 1400–1415, https://doi.org/10.1002/jcc.20065.

[29] A.D. MacKerell, D. Bashford, M. Bellott, et al., All-atom empirical potential for molec-ular modeling and dynamics studies of proteins, J. Phys. Chem. B 102 (18) (1998)3586–3616, https://doi.org/10.1021/jp973084f.

[30] J.C. Phillips, R. Braun, W. Wang, et al., Scalable molecular dynamics with NAMD, J.Comput. Chem. 26 (16) (2005) 1781–1802, https://doi.org/10.1002/jcc.20289.

[31] D.A. Antunes, D. Devaurs, L.E. Kavraki, Understanding the challenges of protein flex-ibility in drug design, Expert Opin. Drug Discovery 10 (12) (2015) 1301–1313,https://doi.org/10.1517/17460441.2015.1094458.

[32] S.Y. Huang, X. Zou, Advances and challenges in protein-ligand docking, Int. J. Mol.Sci. 11 (8) (2010) 3016–3034, https://doi.org/10.3390/ijms11083016.

[33] F. Spyrakis, A. BidonChanal, X. Barril, F.J. Luque, Protein flexibility and ligand recog-nition: challenges for molecular modeling, Curr. Top. Med. Chem. 11 (2) (2011)192–210.

[34] W.L. Jorgensen, J. Chandrasekhar, J.D.Madura, R.W. Impey,M.L. Klein, Comparison ofsimple potential functions for simulating liquid water, J. Chem. Phys. 79 (2) (1983)926–935, https://doi.org/10.1063/1.445869.

[35] J.-P. Ryckaert, G. Ciccotti, H.J.C. Berendsen, Numerical integration of the cartesianequations of motion of a system with constraints: molecular dynamics of n-alkanes, J. Comput. Phys. 23 (3) (1977) 327–341, https://doi.org/10.1016/0021-9991(77)90098-5.

[36] H.J.C. Berendsen, J.P.M. Postma, W.F. van Gunsteren, A. DiNola, J.R. Haak, Moleculardynamics with coupling to an external bath, J. Chem. Phys. 81 (8) (1984)3684–3690, https://doi.org/10.1063/1.448118.

[37] G.J. Martyna, Douglas J. T, Michael L. Klein, Constant pressure molecular dynamicsalgorithms, J. Chem. Phys. 101 (5) (1994) 4177–4189.

[38] S.E. Feller, Y. Zhang, R.W. Pastor, B.R. Brooks, Constant pressure molecular dynamicssimulation: the Langevin pistonmethod, J. Chem. Phys. 103 (11) (1995) 4613–4621,https://doi.org/10.1063/1.470648.

[39] U. Essmann, Lalith Pe, Max L. Berkowitz, Tom Darden, Hsing Lee, Lee G.Pedersen, A smooth particle mesh Ewald method, J. Chem. Phys. 103 (19)(1995) 8577–8593.

[40] S. Sugio, A. Kashima, S. Mochizuki, M. Noda, K. Kobayashi, Crystal structure ofhuman serum albumin at 2.5 A resolution, Protein Eng. 12 (6) (1999) 439–446.

[41] J.M. Word, S.C. Lovell, J.S. Richardson, D.C. Richardson, Asparagine and glutamine:using hydrogen atom contacts in the choice of side-chain amide orientation, J.Mol. Biol. 285 (4) (1999) 1735–1747, https://doi.org/10.1006/jmbi.1998.2401.

[42] O. Trott, A.J. Olson, AutoDock Vina: improving the speed and accuracy of dockingwith a new scoring function, efficient optimization and multithreading, J. Comput.Chem. 31 (2) (2010) 455–461.

[43] M.F. Sanner, Python: a programming language for software integration and devel-opment, J. Mol. Graph. Model. 17 (1) (1999) 57–61.

[44] Z. Huang, C.F. Wong, Docking flexible peptide to flexible protein by molecular dy-namics using two implicit-solvent models: an evaluation in protein kinase andphosphatase systems, J. Phys. Chem. B 113 (43) (2009) 14343–14354, https://doi.org/10.1021/jp907375b.

[45] D.E. Tanner, K.Y. Chan, J.C. Phillips, K. Schulten, Parallel generalized born implicit sol-vent calculations with NAMD, J. Chem. Theory Comput. 7 (11) (2011) 3635–3642,https://doi.org/10.1021/ct200563j.

[46] B. Roux, T. Simonson, Implicit solvent models, Biophys. Chem. 78 (1) (1999) 1–20,https://doi.org/10.1016/S0301-4622(98)00226-9.

[47] J. Kleinjung, F. Fraternali, Design and application of implicit solvent models in bio-molecular simulations, Curr. Opin. Struct. Biol. 25 (100) (2014) 126–134, https://doi.org/10.1016/j.sbi.2014.04.003.

[48] M. Feig, C.L. Brooks 3rd, Recent advances in the development and application of im-plicit solvent models in biomolecule simulations, Curr. Opin. Struct. Biol. 14 (2)(2004) 217–224, https://doi.org/10.1016/j.sbi.2004.03.009.

[49] R. Shen, W. Han, G. Fiorin, S.M. Islam, K. Schulten, B. Roux, Structural refinement ofproteins by restrainedmolecular dynamics simulations with non-interacting molec-ular fragments, PLoS Comput. Biol. 11 (10) (2015), e1004368. https://doi.org/10.1371/journal.pcbi.1004368.

[50] W. Humphrey, D. Andrew, Klaus Schulten, VMD: visual molecular dynamics, J. Mol.Graph. 14 (1) (1996) 33–38.

[51] G. Rezaei Behbehani, A.A. Saboury, A. Fallah Baghery, A. Abedini, Application of anextended solvation theory to study on the binding of magnesium ion with myelinbasic protein, J. Therm. Anal. Calorim. 93 (2) (2008) 479–483, https://doi.org/10.1007/s10973-007-8674-7.

[52] G.R. Behbehani, A.A. Saboury, E. Taleshi, A direct calorimetric determination of dena-turation enthalpy for lysozyme in sodium dodecyl sulfate, Colloids Surf. B:Biointerfaces 61 (2) (2008) 224–228, https://doi.org/10.1016/j.colsurfb.2007.08.007.

[53] G. Rezaei Behbehani, W.E. Waghorne, A high-performance theory for thermody-namic study of solvation in mixed solvents, Thermochim. Acta 478 (2008) 1–2):1-5https://doi.org/10.1016/j.tca.2008.08.014.

[54] G. Rezaei Behbehani, A. Divsalar, A.A. Saboury, A. Hekmat, A thermodynamic studyon the binding of PEG-stearic acid copolymer with lysozyme, J. Solut. Chem. 38(2) (2008) 219, https://doi.org/10.1007/s10953-008-9360-5.

[55] S. Tabassum, W.M. Al-Asbahy, M. Afzal, F. Arjmand, Synthesis, characterization andinteraction studies of copper based drug with human serum albumin (HSA): spec-troscopic and molecular docking investigations, J. Photochem. Photobiol. B Biol.114 (2012) 132–139, https://doi.org/10.1016/j.jphotobiol.2012.05.021.

[56] M.Z. Kabir, W.-V. Tee, S.B. Mohamad, Z. Alias, S. Tayyab, Interaction of an anticancerdrug, gefitinibwith human serum albumin: insights from fluorescence spectroscopyand computational modeling analysis, RSC Adv. 6 (94) (2016) 91756–91767,https://doi.org/10.1039/C6RA12019A.

[57] M.H.C. Koolhaas, G. van der Zwan, R.N. Frese, R. van Grondelle, Red shift of the zerocrossing in the CD spectra of the LH2 antenna complex of Rhodopseudomonasacidophila: a structure-based study, J. Phys. Chem. B 101 (37) (1997) 7262–7270,https://doi.org/10.1021/jp963368k.

[58] R.L. Nation, J. Li, Colistin in the 21st century, Curr. Opin. Infect. Dis. 22 (6) (2009)535–543, https://doi.org/10.1097/QCO.0b013e328332e672.

[59] M. Eslami Moghadam, A. Divsalar, A. Abolhosseini Shahrnoy, A.A. Saboury, Synthe-sis, cytotoxicity assessment, and interaction and docking of novel palladium(II)complexes of imidazole derivatives with human serum albumin, J. Biomol. Struct.Dyn. 34 (8) (2016) 1751–1762, https://doi.org/10.1080/07391102.2015.1090345.

[60] J. Kang, Y. Liu, M.-X. Xie, S. Li, M. Jiang, Y.-D. Wang, Interactions of human serum al-bumin with chlorogenic acid and ferulic acid, Biochim. Biophys. Acta 1674 (2)(2004) 205–214, https://doi.org/10.1016/j.bbagen.2004.06.021.

[61] J. Chamani, H. Vahedian-Movahed, M.R. Saberi, Lomefloxacin promotes the interac-tion between human serum albumin and transferrin: a mechanistic insight into theemergence of antibiotic's side effects, J. Pharm. Biomed. Anal. 55 (1) (2011)114–124, https://doi.org/10.1016/j.jpba.2010.12.029.

[62] M. Purcell, J.F. Neault, H.A. Tajmir-Riahi, Interaction of taxol with human serum al-bumin, Biochim. Biophys. Acta Protein Struct. Mol. Enzymol. 1478 (1) (2000)61–68, https://doi.org/10.1016/S0167-4838(99)00251-4.

[63] G. Mocz, Information content of fluorescence polarization and anisotropy, J.Fluoresc. 16 (4) (2006) 511–524, https://doi.org/10.1007/s10895-006-0095-7.

[64] N.G. James, D.M. Jameson, Steady-state fluorescence polarization/anisotropy for thestudy of protein interactions, Methods Mol. Biol. 1076 (2014) 29–42, https://doi.org/10.1007/978-1-62703-649-8_2.

[65] J.A. Ross, D.M. Jameson, Time-resolved methods in biophysics. 8. Frequency domainfluorometry: applications to intrinsic protein fluorescence, Photochem. Photobiol.Sci. 7 (11) (2008) 1301–1312, https://doi.org/10.1039/b804450n.

[66] D.M. Jameson, J.A. Ross, Fluorescence polarization/anisotropy in diagnostics and im-aging, Chem. Rev. 110 (5) (2010) 2685–2708, https://doi.org/10.1021/cr900267p.

[67] Time-dependent anisotropy decays, in: J.R. Lakowicz (Ed.), Principles of Fluores-cence Spectroscopy, Springer US, Boston, MA 2006, pp. 383–412, https://doi.org/10.1007/978-0-387-46312-4_11.