Ueo

KAa

b

c

d

S

a

ARRAA

KNBTPM

1

lwastftat

0h

Colloids and Surfaces B: Biointerfaces 116 (2014) 720–726

Contents lists available at ScienceDirect

Colloids and Surfaces B: Biointerfaces

journa l homepage: www.e lsev ier .com/ locate /co lsur fb

nraveling the binding mechanism of polyoxyethylene sorbitansters with bovine serum albumin: A novel theoretical model basedn molecular dynamic simulations

arelia H. Delgado-Magneroa,1, Pedro A. Valientea,1, Miriam Ruiz-Penab,urora Pérez-Gramatgesc, Tirso Ponsd,∗

Laboratorio de Biología Computacional, Centro de Estudios de Proteína (CEP), Facultad de Biología, Universidad de La Habana, Habana 10400, CubaDepartamento de Radioquímica, Instituto Superior de Tecnologías y Ciencias Aplicadas (InSTEC), P.O. Box 6163, La Habana 10400, CubaDepartamento de Química, Pontifícia Universidade Católica do Rio de Janeiro, Rio de Janeiro 22453-900, BrazilStructural Biology and Biocomputing Programme, Spanish National Cancer Research Centre (CNIO), C/Melchor Fernández Almagro 3, Madrid E-28029,pain

r t i c l e i n f o

rticle history:eceived 21 March 2013eceived in revised form 6 November 2013ccepted 10 November 2013vailable online 19 November 2013

eywords:on-ionic surfactantsSAweenolysorbates

a b s t r a c t

To gain a better understanding of the interactions governing the binding mechanism of proteins withnon-ionic surfactants, the association processes of Tween 20 and Tween 80 with the bovine serum albu-min (BSA) protein were investigated using molecular dynamics (MD) simulations. Protein:surfactantmolar ratios were chosen according to the critical micelle concentration (CMC) of each surfactant inthe presence of BSA. It was found that both the hydrophilic and the hydrophobic groups of the BSAequally contribute to the surface area of interaction with the non-ionic surfactants. A novel theoreticalmodel for the interactions between BSA and these surfactants at the atomic level is proposed, where bothsurfactants bind to non-specific domains of the BSA three-dimensional structure mainly through theirpolyoxyethylene groups, by hydrogen bonds and van der Waals interactions. This is well supported by thestrong electrostatic and van der Waals interaction energies obtained in the calculations involving surfac-

olecular dynamics simulations tant polyoxyethylene groups and different protein regions. The results obtained from the MD simulationssuggest that the formation of surfactant clusters over the BSA structure, due to further cooperative self-assembly of Tween molecules, could increase the protein conformational stability. These results extendthe current knowledge on molecular interactions between globular proteins and non-ionic surfactants,and contribute to the fine-tuning design of protein formulations using polysorbates as excipients forminimizing the undesirable effects of protein adsorption and aggregation.

© 2013 Elsevier B.V. All rights reserved.

. Introduction

Non-ionic surfactants like the fatty acid esters of polyoxyethy-ene sorbitans (a.k.a. Tween 20 and Tween 80, polysorbates) are

idely used in the pharmaceutical industry to prevent proteindsorption [1] and aggregation under various processing conditionsuch as refolding [2], mixing [3], reconstitution [4,5], freeze-hawing [6] and freeze-drying [7]. Tween 20 and Tween 80 areorming part of several protein pharmaceutical products as inac-

ive protective excipients, in liquid or solid dosages [8]. Thesemphiphilic molecules have common polyoxyethylene subunits inheir polar heads and differ only in the fatty acid nature of their

∗ Corresponding author. Tel.: +34 91 732 80 59; fax: +34 91 224 69 76.E-mail addresses: tpons@cnio.es, tirsopons@gmail.com (T. Pons).

1 Contribute equally to this work.

927-7765/$ – see front matter © 2013 Elsevier B.V. All rights reserved.ttp://dx.doi.org/10.1016/j.colsurfb.2013.11.018

hydrocarbon tails; i.e. monolaurate for Tween 20 and monooleatefor Tween 80 [9,10].

The ability of Tween 20 and Tween 80 to stabilize proteinsis attributed primarily to their ability to outcompete proteinmolecules for hydrophobic surfaces such as air–water interfaces,thereby preventing proteins from unfolding at these hydropho-bic interfaces [11–13]. These non-ionic surfactants can also blockprotein molecules from adsorbing to other hydrophobic surfacespresent during processing [12–14]. In addition, it has been reportedthat Tween 20 and Tween 80 may prevent the aggregation bythe direct interaction with hydrophobic patch regions in pro-teins because the protein:surfactant complexes become morehydrophilic compared to the pure proteins [11–13,15–17].

Several biophysical methods such as isothermal titration calori-metric [18], surface tensiometry and fluorescence [15,19], andcircular dicroism spectroscopy [2,20] have been used to study thestabilizing effect of Tween 20 and Tween 80 against the protein

Surfa

ap[tiMtpdpntaoBtt

ptfipaubicwtaBplda2ob

2

2

f(ottfteaowbgitf

3(i

K.H. Delgado-Magnero et al. / Colloids and

ggregation [13]. However, the protein stabilization mechanismsroposed for the action of these surfactants are still quite unclear21]. An additional difficulty arises from the fact that, due toheir production process, commercial polysorbates commonly usedn the pharmaceutical industry are actually complex mixtures.

ALDI-TOF mass spectrometric studies of polysorbate formula-ions have revealed a complex mixture of oligomers that includeolyethylene glycol esters, sorbitan polyethoxylates, polysorbateiesters, and sorbitol polyethoxylate esters, where the major com-onent of the fatty acids esters of polyethoxy sorbitan defines theame of the product [10]. This diversity of chemical structures inhe commercial products, which composition varies greatly evenmong different batches of the same material, makes the physic-chemical studies and the interpretation of results more difficult.ecause of this, an efficient formulation of surfactant/protein mix-ures on a physicochemical basis is not so far developed, limitingheir practical use to empirical approaches.

Therefore, new insights from computational simulations torovide structural information about protein:non-ionic surfac-ant binding mechanism at the atomic level are relevant for thene-tuning design of protein formulations using polysorbates asrotective excipients. In this work, we studied for first time thessociation process of Tween 20 and Tween 80 with BSA by molec-lar dynamics (MD) simulations. BSA was chosen as protein modelecause its physicochemical properties are well characterized, and

t is a typical globular protein similar to those used in pharmaceuti-al formulations [22–25]. This globular protein is highly soluble inater and plays important physiological roles carrying out different

ypes of amphiphilic biological molecules (e.g., fatty acids, aminocids, drugs and pharmaceuticals) [26]. Few studies about drug-SA interactions at atomic level using MD simulations have beenublished. Recently, Niu and colleagues reported 25 ns MD simu-

ations to investigate the preferential binding mode of BSA withifferent flavonoids [27]. In the present work, we aimed to proposetheoretical model for the binding mechanism of BSA with Tween0 and Tween 80, taking into account binding to different domainsf BSA, which will be assessed from the calculations of hydrogenonding and van der Waals interaction energies.

. Methods

.1. Molecular models

The three-dimensional (3D) structure of BSA was takenrom the SWISS-MODEL repository (accession code: P027692)http://swissmodel.expasy.org/repository/) [28]. The 3D structuresf Tween 20 and Tween 80 were generated with the Spar-an Student v3.1.0 program (Wavefunction Inc., USA). Molecularopologies files for Tween molecules based on the general amberorce field (GAFF) [29] were generated using the following pro-ocol. First, we optimized the potential energy surface (PES) ofach surfactant by an ab initio quantum mechanical calculationt the MP2/6-31G** level with the Gaussian03 program [30]. Sec-nd, we calculated the atoms partial charges for each surfactantith the RESP methodology [31,32] implemented in the Antecham-

er program [33]. We also used Antechamber to generate a set ofeometrical parameters for bonds, angles, atom pairs, proper andmproper dihedrals angles based on the GAFF force field [29]. All

hese calculations were carried out considering a null total chargeor the Tween molecules.

2 During the manuscript preparation process, the 3D structure of BSA (PDB code:V03) was solved at 2.7 A resolution. The calculated root mean square deviationRMSD = 0.49 nm) between the backbone atoms of 3D-model and the X-ray structures a clear indication of the structural similitude of these 3D structures.

ces B: Biointerfaces 116 (2014) 720–726 721

3. Simulation procedure

All molecular dynamics simulations were carried out withthe GROMACS (version 4.6.1) software package [34] using theAMBER03 force field [35], and the TIP3P water model [36]. Tosetup the BSA:Tween 20 system, we solvated the system with48,226 water and 30 Tween 20 molecules. In the case of theBSA:Tween 80 system, one protein molecule was solvated with107,460 water and 64 Tween 80 molecules. A dodecahedron boxwas used in both systems. The BSA:Tween molar ratios corre-spond to the critical micelle concentration (CMC) reported for eachsurfactant in the presence of BSA [15]. To neutralize the chargeof each system, 14 Na+ ions were added. The protonation statesof the BSA ionizable residues at pH = 7 were assigned using thePROPKA option [37] implemented by the PDB2PQR web server(http://agave.wustl.edu/pdb2pqr/server.html).

For energy minimization, we used the steepest descent algo-rithms (preceded by a position restrained stage for protein atoms)and a conjugate gradient, until an energy gradient of less than10 kJ mol−1 nm−2 was reached. The MD simulations were per-formed according to the following criteria: 300 ps of MD runwith position restraints on the protein to allow for relaxationof the solvent molecules; followed by another 150 ps of MD runwith position restraints on the protein backbone’s atoms to allowa gradually liberation of the system. We also prepared threereplicates for both systems BSA:Tween 20 and BSA:Tween 80.Initial velocities of replicates were randomly generated from aMaxwell distribution at 310 K, in accordance with the atomicmasses. The first equilibration run was followed by a 50 ns MDrun without position restraints, using the leapfrog integrator[38].

The electrostatic interactions were calculated at every step withthe particle-mesh Ewald method [39,40]. The van der Waals inter-actions were described by a Lennard-Jones potential with a cut-offof 1.4 nm. The SETTLE [41] algorithm was used to constrain bondsand angles of water molecules and LINCS [42] was used for all otherbonds, allowing a time step of 2 fs. Both temperature and pressurewere controlled by a weak coupling to a thermostat at 310 K andbarostat at 1 atm, using the Berendsen algorithm [43].

4. Energetic and structural analysis

All the energetic and structural analyses were carried outwith the GROMACS program tools [44]. The root-mean-squaredeviations (RMSD) were calculated with g rms, taking the initialstructure as frame reference. The number of contacts between BSAand Tween 20 or Tween 80 was calculated with g mindist, usinga 0.4 nm cut-off. The van der Waals and electrostatic interactionenergies between BSA and the surfactant polar heads and tails werecalculated with g energy after rerunning the equilibrium ensembleusing a 2.0 nm cut-off. The number of hydrogen bond interactionsbetween BSA and the surfactant polar heads were calculated withg hbond. A Donor (D) and Acceptor (A) distance cut-off of 0.35 nmand a D–A–H angle cut-off of 45◦ were used to define a hydrogenbond [44]. The surface interaction area (SIA) of BSA upon pro-tein:surfactant binding were calculated as the differences in thesolvent accessible area between the bound and the unbound stateof BSA with g sas. For both surfactants, the SIA was calculated sim-ilarly. A solvent probe of 0.14 nm and 24 dots per sphere wereused as parameters for the surface interaction area calculations. Thetotal SIA was decomposed into hydrophobic and hydrophilic com-

ponents. All the energetic and structural parameters average andblock averaging error estimations were calculated with g analyzeover the last 6 ns of the MD simulations, to minimize convergenceartifacts.

722 K.H. Delgado-Magnero et al. / Colloids and Surfaces B: Biointerfaces 116 (2014) 720–726

F timed

5

BstmbCpeapcas

6

6s

ttts

TA

ig. 1. (A) The root-mean-square deviations (RMSD) of BSA backbone vs simulationeviations of the RMSD values during the 50 ns of simulations.

. Principal component analysis

For Principal Component Analysis (PCA), the coordinates of aSA recorded during each MD trajectory were first fitted to thetarting structure to eliminate the overall translation and rota-ion, followed by the construction of the mass-weighted covariance

atrix, C, of the atomic positional fluctuations [45] of the back-one atoms for each system. The covariance matrix is defined as:ij =〈 (Xi − 〈 Xi 〉)(Xj − 〈 Xj 〉) 〉 where 〈....〉 is the average over all sam-led data. The covariance matrices were diagonalized to obtain theigenvectors (PC) and the eigenvalues (�) that provide informationbout correlated motions throughout the protein [45]. We com-ared the conformational stability of BSA upon protein:surfactantomplexation by projecting the MD trajectory of the BSA:Tween 20nd BSA:Tween 80 systems over the eigenvectors of the BSA freeystem.

. Results and discussion

.1. Nature and extent of interactions between polyoxyethyleneorbitan esters and BSA

The hydrophobic interaction of non-ionic surfactants with pro-

eins has been considered as one of the key factors for preventinghe aggregation and the agitation-induced denaturation of pro-eins during biotechnological processes [16,18,19]. However, othertudies have showed that just hydrophobic interactions alone are

able 1verage and standard deviations of the total number of atomic contacts, interaction ener

Replicate P%a Atomic contacts (at 4.0 A)

Head Tail

BSA:Tween20 (1:30)1 80% 4219 ± 288 997 ± 662 80% 5182 ± 26 930 ± 213 83% 5974 ± 95 1183 ± 158

BSA:Tween80 (1:64)1 28% 3659 ± 108 938 ± 152 38% 3287 ± 142 547 ± 253 42% 4457 ± 156 486 ± 103

a Percentage of surfactant molecules that are binding to BSA during the last 6 ns of sim

for BSA free and BSA with surfactants between replicates. (B) Average and standard

not sufficient enough to stabilize the binding of amphiphilic com-pounds, such as fatty acids, to albumin proteins [46]. In general,depending on the specific protein and surfactant used, the stabi-lization effect is quite different, and cannot be predicted up to now.One reason for this is the inadequate understanding of the prin-ciples that govern the stabilization of proteins in the presence ofsurfactants.

Here, we studied the molecular interactions of the non-ionicsurfactants Tween 20 and Tween 80 with bovine serum albuminthrough molecular dynamics simulations. The molecular struc-tures of polyoxyethylene sorbitan esters of fatty acids were chosenas representative of the major components in the commercialpolysorbates. In order to study the nature and extent of interactionsat critical concentrations, where important changes could be occur-ring in the micellar solution properties, we used protein:surfactantmolar ratios corresponding to the CMC of Tween 20 (1:30) andTween 80 (1:64) in the presence of BSA [15].

Initially, the RMSD of the protein backbone (C C� N), with andwithout surfactants, were calculated as a function of the simulationtime and with respect to the initial structure, used as frame refer-ence (Fig. 1). We also calculated the number of contacts of Tween20 and Tween 80 with the BSA, as a function of time, for all simu-lations, and the total number of contacts reached the equilibrium

about 40 ns; therefore, we decided to do the analysis over the last6 ns of MD simulations.

In addition, the contribution of the surfactant head and tail to thenumber of contacts between each non-ionic surfactant and the BSA

gies and hydrogen bond interactions between Tweens and BSA.

Hydrogen bondinteractions

Ecoulomb (kJ/mol) ELJ (kJ/mol)

BSA:head BSA:head BSA:tail

22 ± 4 −2613 ± 354 −1761 ± 80 −410 ± 2326 ± 1 −2729 ± 107 −2130 ± 13 −390 ± 626 ± 4 −2455 ± 254 −2608 ± 46 −525 ± 5717 ± 1 −2026 ± 249 −1485 ± 10 −423 ± 717 ± 1 −2395 ± 194 −1445 ± 44 −260 ± 1024 ± 2 −2853 ± 536 −1901 ± 63 −224 ± 62

ulations.

K.H. Delgado-Magnero et al. / Colloids and Surfaces B: Biointerfaces 116 (2014) 720–726 723

Table 2Hydrophilic and hydrophobic interaction areas between Tweens and BSA.

Surface Interaction (nm2)

Replicate Protein Surfactant

Hydrophobic Hydrophilic Hydrophobic Hydrophilic

BSA:Tween20 (1:30)1 24 ± 3 20 ± 2 31 ± 3 43 ± 52 28 ± 3 18 ± 2 25 ± 3 55 ± 53 38 ± 3 24 ± 3 23 ± 3 61 ± 5

wortbotalcsihid

BhachtoZabti

aBtmbbktna

Fit

BSA:Tween80 (1:64)1 23 ± 32 20 ± 33 24 ± 2

as calculated for all replicates (Table 1). There is a predominancef interactions though the polar head of non-ionic surfactants,eflected in the higher number of atomic contacts of the surfac-ant heads compared to surfactant tails. This result is consistent foroth surfactants, despite the differences in the chemical structuref their hydrocarbon tails. On the other hand, highly similar elec-rostatic interaction energies were obtained for both surfactants,s well as the number of hydrogen bonds of BSA with the ethoxy-ated (EO) groups of the surfactant heads. Since these surfactantsontain the same number of EO units around the sorbitan ring, aimilar interaction can be expected. Furthermore, the van der Waalsnteraction energies of BSA with the EO moieties are considerablyigher than those with the surfactant hydrocarbon tails, confirm-

ng the hypothesis that interactions occur through both surfactantomains.

Therefore, we can hypothesize that the interaction betweenSA and these non-ionic surfactants is not solely driven by theydrophobic interaction between the surfactant tails and the apolarmino acids of the protein, as expected, but there is also a significantontribution from the interactions through the polar heads. Thereave been a few experimental reports referring to the possibilityhat hydrogen bonds can be formed between ethylene oxide chainsf non-ionic surfactant and BSA molecules [47–49]. In particular,adymova and coworkers found that the BSA:surfactant complex-tion occurs owing to specific interactions (hydrogen bonding)etween polar groups of Tween 80 and protein molecules, withhe participation of tryptophan residues and amide groups that aren the immediate vicinity to these residues [49].

Therefore, in this novel theoretical model based on atomic inter-ctions we proposed that the non-ionic surfactants bind to theSA 3D structure both through their polyoxyethylene groups andheir hydrocarbon tails, with a significant contribution of the for-

er. This approach extends the accepted criteria that associationetween proteins and non-ionic surfactants is preferentially guidedy hydrophobic interactions [16,50]. Nevertheless, it should be

ept in mind that these computational simulations are limitedo the role of the major component in the commercial products,amely polyoxyethylene sorbitan monolaurate in polysorbate 20,nd polyoxyethylene sorbitan monooleic in polysorbate 80. Thus,

ig. 2. Tween 20 and Tween 80 binding to different regions on the BSA surface. Residues ofn yellow surface. BSA domains are represented in blue (IA), cyan (IB), green (IIA), green lext, the reader is referred to the web version of this article.)

15 ± 3 25 ± 2 47 ± 413 ± 2 50 ± 3 79 ± 318 ± 2 51 ± 2 80 ± 3

experimental results using commercial polysorbates in formula-tions containing globular proteins might include other effects dueto heterogeneity in the samples.

Other interesting results obtained from the MD calculations thatsupport the previous finding arise from the analysis of the surfaceinteraction areas between the different regions of the non-ionicsurfactants and the protein (Table 2). As can be seen, the similarcontribution of the hydrophilic and hydrophobic components ofthe surface interaction area of the protein is a clear indication ofthe direct association of the non-ionic surfactants with BSA throughdifferent regions on the protein surface. On the other hand, the anal-ysis of the surfactant surface interactions shows there is a highercontribution of the hydrophilic component in both systems, whichcan be explained based on the larger van der Waals radius of oxygen(O) compared to that of carbon (C) atoms, even if we also con-sider a similar C:O ratio in the polyoxyethylene moieties presentin the surfactant polar heads. This result is consistent with the pre-vious findings, discussed above, being the difference between theareas less marked for the BSA:Tween 80 system, likely due to thehigher hydrophobic character of the C-18 hydrocarbon tail of thissurfactant.

Moreover, we calculated the number of molecules that are inter-acting directly with the BSA (at a distance of 4 A), from the analysisof the data shown in Table 1. These values correspond to the last 6 nsof simulation. The results indicate that there are 24 ± 1 moleculesof Tween 20 and 23 ± 5 molecules of Tween 80 interacting withthe protein. This is a quite interesting finding since it suggests thatthere is a maximum number of Tween molecules able to interactdirectly at the protein surface, regardless of the surfactant type. Inturn, it is a supporting argument toward the hypothesis that pre-dominant interactions occur through the polar heads, which arestructurally the same in Tween 20 and Tween 80.

6.2. Tween 20 and Tween 80 binding to non-specific surfaceregions on the BSA 3D structure

Fig. 2 shows the binding of non-ionic surfactants to BSA struc-tures, as obtained in the three replicates of the MD simulation. Itis possible to see that binding occurs at different protein regions,

BSA that are interacting at a distance of at least 4 A from surfactants are representedight (IIB), red (IIIA), and pink (IIIB). (For interpretation of the references to color in

724 K.H. Delgado-Magnero et al. / Colloids and Surfaces B: Biointerfaces 116 (2014) 720–726

F ructury eferen

nfrptScsm

i2

iTfimt

ig. 3. Formation of surfactant clusters on protein surface. The protein secondary stellow (head). (A) BSA:Tween 20 and (B) BSA:Tween 80. (For interpretation of the r

ot showing significant differences between those correspondingor each surfactant. Nevertheless, it is worth noting that someesidues are preserved during the interaction, mostly charged andolar residues, which supports the idea of interactions occurringhrough the polar surfactant heads (Supplementary material Table1). Based on these results, we can state that there are not spe-ific regions of BSA that make clearly preferential contacts withurfactant molecules of Tween. Therefore, we can conclude thatonomeric surfactant interactions with BSA are non-specific.Supplementary material related to this article can be found,

n the online version, at http://dx.doi.org/10.1016/j.colsurfb.013.11.018.

The results obtained from MD simulations regarding bind-ng to non-specific surface regions on the BSA also show thatween molecules self-assemble in clusters over the protein sur-

ace (Fig. 3). The explanation for this behavior can be foundn the spontaneous and entropy-driven association of surfactant

olecules, some of which are initially interacting at the pro-ein surface via their polar heads, and then assemble as micelles

Fig. 4. Distribution probability plot of the con

e is shown in gray, and the surfactant molecules are represented in green (tail) andces to color in text, the reader is referred to the web version of this article.)

through cooperative hydrophobic interactions with other surfac-tant molecules in solution, thus increasing the solubility of thesurfactant:protein complex. A significant aggregation of Tween20 molecules over the protein surface is observed, which agreeswith the results from calorimetric experiments of Hoffman andcolleagues using the same system [21]. This is also in accordanceto the results of Garidel et al., with human serum albumin (HSA)[18], a human homologue protein that shares a 76% in sequenceidentity with BSA [26]. More extensive aggregation is observed inBSA:Tween 80 systems, as a direct consequence of the larger hydro-carbon tail of Tween 80 compared to Tween 20, responsible formicelle formation at lower concentrations. A previous experimen-tal work by Ruiz et al. reported that a higher number of surfactantmolecules of Tween 80 are involved in interactions with BSA com-pared to Tween 20, based on the surfactant:protein ratios obtained

at the CMC value of each surfactant in the presence of protein [15]. Itis worthy to note that these results can be quite interesting and rele-vant to pharmaceutical formulators since they offer a more detailedmolecular approach to interactions occurring between globular

formational subspace sample along PC1.

Surfa

pw

6c

afH8ftBttfitBit

2aoTaeits

7

ancInhheet

bssatmntists

R

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[[

[

[

K.H. Delgado-Magnero et al. / Colloids and

roteins and surfactants used as excipients in biomedical products,ith direct implications in the formulation procedures.

.3. Effect of Tween 20 and Tween 80 binding on theonformational stability of BSA

The effects of Tween 20 and Tween 80 in preventing thegitation-induced denaturation of proteins are of special interestor the development of novel pharmaceutical formulations [19].ere, we attempt to predict the effect of Tween 20 and Tween0 on the conformational stability of BSA by comparing the con-ormational space explored by the protein in the absence and inhe presence of these surfactants (Fig. 4). In the surfactant-freeSA system, we observed that the protein sampled four conforma-ions along the eigenvector 1 (PC1). Several works have indicatedhat most of the protein internal motions can be captured by therst principal modes [45,51,52]. It is remarkable that PC1 describeshe 40.68% of the total variance of BSA. For the BSA:Tween 20 andSA:Tween 80 systems, the conformational space explored by BSA

s reduced compared to that of BSA free (in the absence of surfac-ants).

It is important to note that in these simulations the role of Tween0 and Tween 80 for preventing BSA aggregation was not directlyssessed, due to the fact that the systems contain just one moleculef BSA each. Nevertheless, we expect that the presence of bothween 20 and Tween 80 in solution would prevent BSA aggregationt the stoichiometry ratio corresponding to the CMC values, as sev-ral experimental evidences have shown [2,16,50]. One examples the study of Bam et al., who reported that Tween 20 preventshe aggregation of the recombinant human growth hormone aturfactant:protein molar ratios higher than 4:1 [16].

. Concluding remarks

In this work we propose for first time a novel theoretical modelt the atomic level for the interactions between BSA and theon-ionic surfactants Tween 20 and Tween 80, which are the majoromponents in polysorbates, based on results from MD simulations.n this model, the surfactants interact with the protein predomi-antly via the polyoxyethylene groups of their polar heads throughydrogen bonds and van der Waals interactions, in addition toydrophobic interactions through the hydrocarbon tails, as gen-rally accepted. This picture is well supported by the favorablelectrostatic and van der Waals interaction energies calculated forhe association between BSA and both Tweens.

In addition, it was shown that Tween 20 and Tween 80ind to non-specific surface regions on the BSA 3D structure,elf-assembling in clusters over the protein surface. It was thenuggested that the formation of these clusters of Tween moleculesround the BSA structure could increase the protein conforma-ional stability. Further studies including a higher number of BSA

olecules are needed to directly assess if the effects of theseon-ionic surfactants in preventing protein aggregation is concen-ration dependent, as well as simulations at longer times, to taken account surfactant relaxation. Finally, we plan to extend thesetudies to other relevant proteins for the pharmaceutical indus-ry, providing a detailed molecular approach of the role of theseurfactants in formulations.

eferences

[1] W.R. Gombotz, S.C. Pankey, L.S. Bouchard, D.H. Phan, A.P. MacKenzie, Stability,

Characterization, Formulation and Delivery System Development for Trans-forming Growth Factor-beta1, Plenum Press, New York, 1996.

[2] L. Bam, J.L. Cleland, T.W. Randolph, Molten globule intermediate of recombinanthuman growth hormone: stabilization with surfactants, Biotechnol. Prog. 12(1996) 801–809.

ces B: Biointerfaces 116 (2014) 720–726 725

[3] M. Katakam, L.N. Bell, A.K. Banga, Effect of surfactants on the physical stabilityof recombinant human growth hormone, J. Pharm. Sci. 84 (1995) 713–716.

[4] M.Z. Zhang, K. Pikal, T. Nguyen, T. Arakawa, S.J. Prestrelski, The effectof the reconstitution medium on aggregation of lyophilized recombinantinterleukinl-2 and ribonuclease A, Pharm. Res. 13 (1996) 643–646.

[5] M.Z. Zhang, J. Wen, T. Arakawa, S.J. Prestrelski, A new strategy for enhancingthe stability of lyophilized protein: the effect of the reconstitution medium onkeratinocyte growth factor, Pharm. Res. 12 (1995) 1447–1452.

[6] B.S. Chang, B.S. Kendrick, J.F. Carpenter, Surface-induced denaturation of pro-teins during freezing and its inhibition by surfactants, J. Pharm. Sci. 85 (1996)1325–1330.

[7] J.M. Sarciaux, S. Mansour, M.J. Hageman, S.L. Nail, Effects of buffer compositionand processing conditions on aggregation of bovine IGG during freeze-drying,J. Pharm. Sci. 88 (1999) 1345–1361.

[8] Physician’s Desk Reference, Medical Economics Company, PDR, Montvale, NJ,1999.

[9] F.O. Ayorinde, S.V. Gelain, J.H.J. Johnson, L.W. Wan, Analysis of some commer-cial polysorbate formulations using matrix-assisted laser desorption/ionizationtime-of-flight mass spectrometry, Rapid Commun. Mass Spectrom. 14 (2000)2116–2124.

10] B.A. Kerwin, Polysorbates 20 and 80 used in the formulation of protein bio-therapeutics: structure and degradation pathways, J. Pharm. Sci. 97 (2008)2924–2935.

11] T.W. Randolph, L.S. Jones, Surfactant-protein interactions, Pharm. Biotechnol.13 (2002) 159–175.

12] T.J. Kamerzell, R. Esfandiary, S.B. Joshi, C.R. Middaugh, D.B. Volkin,Protein–excipient interactions: mechanisms and biophysical characterizationapplied to protein formulation development, Adv. Drug Deliv. Rev. 63 (2011)1118–1159.

13] D. Otzen, Protein–surfactant interactions: a tale of many states, Biochim. Bio-phys. Acta 1814 (2011) 562–591.

14] S.H. Mollmann, U. Elofsson, J.T. Bukrinsky, S. Frokjaer, Displacement of adsorbedinsulin by Tween 80 monitored using total internal reflection fluorescence andellipsometry, Pharm. Res. 22 (2005) 1931–1941.

15] M. Ruiz-Pena, R. Oropesa-Nunez, T. Pons, S.R. Louro, A. Pérez-Gramatges,Physico-chemical studies of molecular interactions between non-ionic sur-factants and bovine serum albumin, Colloids Surf. B: Biointerfaces 75 (2010)282–289.

16] N.B. Bam, J.L. Cleleand, J. Yand, M.C. Manning, J.F. Carpenter, R.F. Kelley,T.W. Randolph, Tween protects recombinant human growth hormone againstagitation-induced damage via hydrophobic interactions, J. Pharm. Sci. 87 (1998)1554–1559.

17] S. Deechongkit, J. Wen, L.O. Narhi, Y. Jiang, S.S. Park, J. Kim, B.A. Kerwin, Physicaland biophysical effects of polysorbate 20 and 80 on darbepoetin alfa, J. Pharm.Sci. 98 (2009) 3200–3217.

18] P. Garidel, C. Hoffmann, A. Blume, A thermodynamic analysis of the bindinginteraction between polysorbate 20 and 80 with human serum albumins andimmunoglobulins: a contribution to understand colloidal protein stabilization,Biophys. Chem. 143 (2009) 70–78.

19] D.K. Chou, R. Krishnamurthy, T.W. Randolph, J.F. Carpenter, M.C. Manning,Effects of Tween20 and Tween80 on the stability of Albutropin during agitation,J. Pharm. Sci. 94 (2005) 1368–1381.

20] L.S. Jones, T.W. Randolph, U. Kohnert, A. Papadimitriou, G. Winter, M.L. Hag-mann, M.C. Manning, J.F. Carpenter, The effects of Tween 20 and sucrose on thestability of anti-l-selectin during lyophilization and reconstitution, J. Pharm.Sci. 90 (2001) 1466–1477.

21] C. Hoffmann, A. Blume, I. Miller, P. Garidel, Insighty into protein–polysorbateinteractions analysed by means of isothermal titration and differential scanningcalorimetry, Eur. Biophys. J. 38 (2007) 557–568.

22] C. Giancola, C. De Sena, D. Fessas, G. Graziano, G. Barone, DSC studies on bovineserum albumin denaturation. Effects of ionic strength and SDS concentration,Int. J. Biol. Macromol. 20 (1997) 193–204.

23] M. Yamasaki, H. Yano, K. Aoki, Differential scanning calorimetric studies onbovine serum albumin: I. Effects of pH and ionic strength, Int. J. Biol. Macromol.12 (1990) 263–268.

24] M. Yamasaki, H. Yano, K. Aoki, Differential scanning calorimetric studies onbovine serum albumin: III. Effect of sodium dodecyl sulphate, Int. J. Biol. Macro-mol. 14 (1992) 305–312.

25] M. Yamasaki, T. Yamashita, H. Yano, K. Tatsumi, K. Aoki, Differential scanningcalorimetric studies on bovine serum albumin. IV. Effect of anionic surfactantswith various lengths of hydrocarbon chain, Int. J. Biol. Macromol. 19 (1996)241–246.

26] T.J. Peter, Serum albumin, Adv. Protein Chem. 37 (1985) 161–245.27] X. Niu, X. Gao, H. Wang, X. Wang, S. Wang, Insight into the dynamic interac-

tion between different flavonoids and bovine serum albumin using moleculardynamics simulations and free energy calculations, J. Mol. Model. 19 (2013)1039–1047.

28] J. Kopp, T. Schwede, The SWISS-MODEL repository: new features and function-alities, Nucleic Acids Res. 34 (Database issue) (2006) D315–D318.

29] D.A. Case, T.A. Darden, I.T.E. Cheatham, C.L. Simmerling, J. Wang, R.E. Duke,R. Luo, R.C. Walker, W. Zhang, K.M. Merz, B. Roberts, S. Hayik, A. Roitberg, G.

Seabra, J. Swails, A.W. Goetz, I. Kolossváry, K.F. Wong, F. Paesani, J. Vanicek, R.M.Wolf, J. Liu, X. Wu, S.R. Brozell, T. Steinbrecher, H. Gohlke, Q. Cai, X. Ye, J. Wang,M.-J. Hsieh, G. Cui, D.R. Roe, D.H. Mathews, M.G. Seetin, R. Salomon-Ferrer, C.Sagui, V. Babin, T. Luchko, S. Gusarov, A. Kovalenko, P.A. Kollman, AMBER 12(2012).

7 d Surfa

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

[

26 K.H. Delgado-Magnero et al. / Colloids an

30] G.W.T.M.J. Frisch, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, J.A.Montgomery Jr., T. Vreven, K.N. Kudin, J.C. Burant, J.M. Millam, S.S. Iyengar, J.Tomasi, V. Barone, B. Mennucci, M. Cossi, G. Scalmani, N. Rega, G.A. Petersson,H. Nakatsuji, M. Hada, M. Ehara, K. Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T.Nakajima, Y. Honda, O. Kitao, H. Nakai, M. Klene, X. Li, J.E. Knox, H.P. Hratchian,J.B. Cross, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev, A.J.Austin, R. Cammi, C. Pomelli, J.W. Ochterski, P.Y. Ayala, K. Morokuma, G.A.Voth, P. Salvador, J.J. Dannenberg, V.G. Zakrzewski, S. Dapprich, A.D. Daniels,M.C. Strain, O. Farkas, D.K. Malick, A.D. Rabuck, K. Raghavachari, J.B. Fores-man, J.V. Ortiz, Q. Cui, A.G. Baboul, S. Clifford, J. Cioslowski, B.B. Stefanov, G.Liu, A. Liashenko, P. Piskorz, I. Komaromi, R.L. Martin, D.J. Fox, T. Keith, M.A.Al-Laham, C.Y. Peng, A. Nanayakkara, M. Challacombe, P.M.W. Gill, B. Johnson,W. Chen, M.W. Wong, C. Gonzalez, J.A. Pople, Gaussian 03, Revision C.02 (orD.01), Gaussian, Inc., Pittsburgh, PA, 2003.

31] P. Cieplak, W.D. Cornell, C. Bayly, P.A. Kollman, Application of the multimoleculeand multiconformational RESP methodology to biopolymers: charge derivationfor DNA, RNA, and proteins, J. Comput. Chem. 16 (1995) 1357–1377.

32] C. Bayly, P. Cieplak, W. Cornell, P.A. Kollman, Advances and continuing chal-lenges in achieving realistic, J. Phys. Chem. 97 (1993) 10269–10280.

33] J. Wang, J. Caldwell, P.A. Kollman, D. Case, Development and testing of a generalamber force field, J. Comput. Chem. 25 (2004) 1157–1174.

34] B. Hess, C. Kutzner, D. van der Spoel, E. Lindahl, GROMACS 4: algorithms forhighly efficient, load-balanced, and scalable molecular simulation, J. Chem.Theory Comput. 4 (2008) 435–447.

35] V. Hornak, R. Abel, A. Okur, B. Strockbine, A. Roitberg, C. Simmerling, Com-parison of multiple Amber force fields and development of improved proteinbackbone parameters, Proteins: Struct. Funct. Bioinform. 65 (2006) 712–725.

36] L.J. William, C. Jayaraman, D.M. Jeffry, W.I. Roger, L.K. Michael, Comparison ofsimple potential functions for simulating liquid water, vol. 79, AIP, 1983, pp.926–935.

37] H. Li, J.H. Jensen, Very fast empirical prediction and rationalization of protein

pKa values, Proteins: Struct. Funct. Bioinform. 61 (2005) 704–721.

38] R.W. Hockney, The potential calculation and some applications, Methods Com-put. Phys. 9 (1970) 136–211.

39] T. Darden, D. York, L. Pedersen, Particle mesh Ewald: an N-log(N) method forEwald sums in large systems, J. Chem. Phys. 98 (1993) 10089–10092.

[

ces B: Biointerfaces 116 (2014) 720–726

40] U. Essmann, L. Perera, M.L. Berkowitz, T. Darden, H. Lee, L.G. Peder-sen, A smooth particle mesh Ewald potential, J. Chem. Phys. 103 (1995)8577–8592.

41] S. Miyamoto, P.A. Kollman, Settle an analytical version of the shake and rattlealgorithm for rigid water models, J. Comput. Chem. 13 (1992) 952–962.

42] B. Hess, H. Bekker, H.J.C. Berendsen, J.G.E.M. Fraaije, LINCS: a linear constraintsolver for molecular simulations, J. Comput. Chem. 18 (1997) 1463–1472.

43] H.J.C. Berendsen, J.P.M. Postma, W.F. van Gunsteren, A. Dinola, J.R. Haak, Molec-ular dynamics with coupling to an external bath, J. Chem. Phys. 81 (1984)3684–3690.

44] D. van der Spoel, E. Lindahl, B. Hess, A.R. van Buuren, E. Apol, P.J. Meulenhoff,D.P. Tieleman, A.L.T.M. Sijbers, K.A. Feenstra, R. van Drunen, H.J.C. Berendsen,Gromacs User Manual, version 4.5, 2010 www.gromacs.org

45] A. Amadei, A.B.M. Linssen, H.J.C. Berendsen, Essential dynamics of proteins,Proteins: Struct. Funct. Bioinform. 17 (1993) 412–425.

46] J.R. Simard, P.A. Zunszain, C.-E. Ha, J.S. Yang, N.V. Bhagavan, I. Petitpas, S. Curry,J.A. Hamilton, Locating high-affinity fatty acid-binding sites on albumin byX-ray crystallography and NMR spectroscopy, Proc. Natl. Acad. Sci. U.S.A. 102(2005) 17958–17963.

47] T. Arakawa, Y. Kita, Protection of bovine serum albumin from aggregation byTween 80, J. Pharm. Sci. 82 (2000) 646–651.

48] N.L. Almeda, C.L.P. Oliveria, I.L. Torriani, W. Loh, Calorimetric and structuralinvestigation of the interaction of lysozyme and bovine serum albumin withpoly(ethylene oxide) and its copolymers, Colloids Surf. B 38 (2004) 67–76.

49] N.M. Zadymova, G.P. Yampol’skaya, L. Yu, Filatova, interaction of bovine serumalbumin with nonionic surfactant Tween 80 in aqueous solutions: complexa-tion and association, Colloid J. 68 (2006) 162–172.

50] N.B. Bam, T.W. Randolph, J.L. Cleland, Stability of protein formulations: inves-tigation of surfactant effects by a novel EPR spectroscopic technique, Pharm.Res. 12 (1995) 2–11.

51] A. Amadei, A.B. Linssen, B.L. de Groot, D.M. van Aalten, H.J. Berendsen, An effi-

cient method for sampling the essential subspace of proteins, J. Biomol. Struct.Dyn. 13 (1996) 615–625.

52] J.D. Evanseck, L.S.D. Caves, M. Karplus, Locally accessible conformations ofproteins: multiple molecular dynamics simulations of crambin, Protein Sci. 7(1998) 649–666.