Coadsorption of Doxorubicin and Selected Dyes on Carbon

May 1, 2016 - Among some number of studied dyes revealing pKa in the range 5–7, only a few of them led to more or less interesting results. ... Next...
1 downloads 6 Views 3MB Size
Download PDF
Article pubs.acs.org/Langmuir

Coadsorption of Doxorubicin and Selected Dyes on Carbon Nanotubes. Theoretical Investigation of Potential Application as a pH-Controlled Drug Delivery System Tomasz Panczyk,*,† Pawel Wolski,† and Leszek Lajtar‡ †

Institute of Catalysis and Surface Chemistry, Polish Academy of Sciences UI. Niezapominajek 8, 30239 Cracow, Poland Department of Chemistry, Maria Curie-Sklodowska University Plac M. Curie-Sklodowskiej 3, 20031 Lublin, Poland

‡

S Supporting Information *

ABSTRACT: This work shows results of a theoretical survey, based on molecular dynamics simulation, of potential applicability of doxorubicin coadsorption with various dyes molecules on/in carbon nanotubes as a drug delivery system. The central idea is to take advantage of the dyes charge distribution change upon switching the pH of the environment from neutral (physiological 7.4) to acidic one (∼5.5 which is typical for tumor tissues). This work discusses results obtained for four dye molecules revealing more or less interesting behavior. These were bromothymol blue, methyl red, neutral red, and p-phenylenediamine. All of them reveal pKa in the range 5−7 and thus will undergo protonation in that pH range. We considered coadsorption on external walls of carbon nanotubes and sequential filling of the nanotubes inner hollow space by drug and dyes. The latter approach, with the application of neutral red and p-phenylenediamine as blockers of doxorubicin, led to the most promising results. Closer analysis of these systems allowed us to state that neutral red can be particularly useful as a long-term blocker of doxorubicin encapsulated in the inner cavity of (30,0) carbon nanotube at neutral pH. At acidic pH we observed a spontaneous release of neutral red from the nanotube and unblocking of doxorubicin. We also confirmed, by analysis of free energy profiles, that unblocked doxorubicin can spontaneously leave the nanotube interior at the considered conditions. Thus, that system can realize pH controlled doxorubicin release in acidic environment of tumor tissues.

1. INTRODUCTION Drug delivery systems are currently widely studied, as they represent an important component of novel therapies free of side effects and oriented toward a precise attack of disease sources.1 Most of the studies are focused on the development of new materials or optimization of the already known. It would be difficult to mention all the systems considered drug carriers, their application areas, or advantages and disadvantages. However, the most advanced results, in some cases including clinical trials, concern liposomes, emulsions, micelles, dendrimers, magnetic nanoparticles, or various other nanoparticulate systems.2 Among the nanoparticulate systems, carbon nanotubes (CNT) have attracted attention because of their unique physicochemical properties.3 They include, but are not limited to, large surface area available for functionalization by various moieties, large inner volume, which can accumulate large amounts of small molecule drugs, and needle like shape, which facilitates penetration of cell membrane.4 The drawbacks are insolubility in aqueous media and possible toxicity. However, both these problems can be removed by chemical treatments leading to shortening, purification, and development of hydrophilic functional groups located at the nanotube tips and sidewalls.5 Currently, it is very difficult to find a combination of CNT, drug, biological receptor, and triggering factor for drug release that has not already been studied in the literature in some way. © XXXX American Chemical Society

Normally, drug molecules are attached to CNT sidewalls by creation of chemical bonds (often indirectly through some spacer molecule) or by a simple physical adsorption. The same goes for biological receptors, which are desirable in order to enhance the targeting performance of the whole system. Another, though probably less popular strategy is encapsulation of drugs in the inner hollow cavity of CNTs. Such architecture of drug carrier offers perfect isolation of drug from the environment prior to its accumulation at the target site. The drug release at the target site is sometimes additionally triggered by using various physical factors like infrared radiation, and ultrasonic and magnetic fields.6,7 Biochemical triggering factors are also considered. Among them, a particularly useful one is pH change from neutral to acidic occurring in tumor tissue. Hydrolysis of hydrazone bonds is often utilized in practical realizations of such pH-triggered drug release mechanisms from various carriers.8 Carbon nanotubes were studied as carriers of a large number of anticancer drugs belonging to the class of topoisomerase inhibitors, anthracyclines, platinum-based drugs, antifolates, purine/pyrimidine antagonists, antimicrotubules, and others.4 It would be beyond the scope of this paper to discuss each mechanism of the anticancer action or mention all literature Received: January 26, 2016 Revised: March 25, 2016

A

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir Table 1. Neutral and Acidic Forms of the Selected Dye Molecules Analyzed in This Study

and probably involves multiple interactions between various functional groups localized on the CNT surface. Incorporation of DOX to the inner cavity of CNTs has been studied theoretically17 and the results suggest stronger binding of DOX inside the CNT than on its sidewall. Encapsulation of DOX inside carbon nanotubes has not been strictly addressed in experimental studies. This is quite surprising since there is evidence of successful loading of relatively big molecules to the inner hollow space of CNTs.19 On the other hand, there are numerous studies showing encapsulation of platinum-based drugs in the inner space of CNTs. The release rate in such systems is often controlled by formation of removable caps locking the nanotube tips.4,20 Methotrexate, an important member of folic acid antagonists has been attached to CNT sidewalls either covalently or by physical adsorption. One of the interesting examples was the attachment by disulfide bonds which are sensitive to pH and presence of esterase.21 Physically adsorbed methotrexate also revealed enhanced release rate at reduced pH22 though at neutral pH the release was observed either. Purine/pirimidine antagonists and antimicrotubules were conjugated with CNTs and in some cases pH sensitive drug carriers were obtained. Examples include: release of gemcitabine (a low molecular weight nucleoside and an antimetabolite that inhibits cellular DNA synthesis) at acidic pH23 and paclitaxel.24,25 The pH change from physiological value 7.4 to ca. 5.5 in tumor tissue or endosomes8 can also initiate changes in charge distribution of some molecules, usually dyes, which are

reports concerning the conjugates of carbon nanotubes with all molecules classified and approved as anticancer drugs. However, let us briefly mention about the most representative results obtained for the mentioned classes of anticancer agents. Topoisomerase inhibitors (e.g., camptothecin) were usually attached to CNT walls by physical adsorption.9,10 Tripisciano et al.11 studied encapsulation of irinotecan in open-ended multiwalled carbon nanotubes. They observed enhanced release rate at acidic pH though the system was not perfectly tight at neutral pH. Important member of anthracycline class of chemoterapeutic agents is doxorubicin, DOX. It was, in the majority of cases, attached to functionalized CNT sidewalls by physical adsorption, and the release was triggered by acidic pH.12−16 So, experimental data confirm attachment of DOX to the sidewalls of carbon nanotubes previously functionalized by some biocompatible and targeting factors like, for example, chitosan and folic acid. Those systems revealed low DOX release rate at physiological pH 7.4 and enhanced release at slightly acidic pH conditions, i.e., ∼ 5.5. However, the mechanism of pH triggered release of DOX seems to be unclear. Usually, this is attributed to the protonation of the NH2 group of DOX at low pH values, which enhances the hydrophilicity and solubility of DOX. 17 However, the protonation pKa of DOX amino group is 8.2,18 thus the protonated forms of DOX should be dominant already at neutral pH. So, the true mechanism of pH triggered release of DOX from functionalized CNTs is definitely more complex B

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

2.2. Force Field and Computational Details. The topologies of the studied CNTs were build using self-designed scripts. We generated single-walled zigzag carbon nanotubes with chiralities (10,0) and (30,0). Their lengths were the same, i.e., 80 Å. Internal degrees of freedom of the CNTs were described by the many-body reactive empirical bond order potential ai-REBO.37 This is the most advanced potential for description of carbonaceous materials. CNT was thus treated as a flexible body that could even undergo structural rearrangements in the case of strong interactions with the environment. However, in these particular cases of relatively light molecules interacting with the CNTs, the deformations were small and transient. Therefore, the flexible nature of the nanotubes does not manifest itself strongly. Interaction of the CNT with other components of the system was described using the standard Lennard-Jones potential for carbon atoms. The LJ parameters for carbon atoms were set up according to the AMBER force field.38,39 No partial charges were set up for carbons creating the CNT, and the terminal carbons were left unsaturated by hydrogen or other species. All interaction parameters and topology associated with the dyes molecules and DOX were generated using the automatic atom and bond-type perception scheme implemented in AmberTools 12.38 Atomic partial charges were determined using the R.E.D. Tools.40 The full set of parameters is available in the Supporting Information file (Figures S1−S9 and Tables S1−S63). All calculations were performed using the Large-scale Atomic/ Molecular Massively Parallel Simulator (lammps) code41 in NPT ensemble using a 1.5 fs time step. The pressure and temperature were controlled using the Nose−Hoover barostat. The TIP3P water model was used, and the SHAKE algorithm was applied to make water molecules rigid. The total number of atoms varied depending on the nanotube and dye combination. The number of carbon atoms for the case of (10,0) nanotube was 800, while for the (30,0) nanotube it was 2400. The number of water molecules was about 25 000, and we also added about 60 Na+ and Cl− ions in order to mimic the physiological ionic strength 0.145 mol L−1. The cutoff for intermolecular interactions was 12 Å. The simulation time varied depending on the problem being analyzed. Before any production run, the systems were heated up to 400 K and cooled to 310 K. Additionally, after an equilibration period taking ca. 0.6−1 ns, the production runs were done taking at least 3.6 ns for each problem being studied. In the case of umbrella sampling simulations, the total simulation time reached 7.2 ns for each system studied in such a way. The equilibrium was justified by monitoring the total energy decay after cooling the system to the target temperature 310 K. We also monitored root mean squared displacements of dyes and DOX molecules during the equilibration stage. At the end of the equilibration stages, all molecules statistically passed through the simulation box at least once. In most cases, the molecules passed much larger distancesmore than 3 times the size of the simulation box. The initial configurations in the case of the narrow nanotube (10,0) were build by placing a given number of dye and DOX molecules at certain distances from the nanotube axis. The distances between dye molecules, DOX and the nanotube were properly adjusted in order to avoid any overlaps. (Figure S10A in the Supporting Information). Next, about 25 000 of water molecules were inserted to the simulation box at grid points separated by 5 Å. The initial size of the simulation box was 140 × 140 × 170 Å with 3D periodic boundary conditions. During the equilibration stage running in the NPT ensemble, the box compressed to ca. 85 × 85 × 107 Å, and the final water density reached 0.033 Å−3. A similar procedure of the simulation box construction was applied in the case of (30,0) nanotube. However, in this case, DOX molecules were placed inside the nanotube, close to its center and at distances preventing overlaps. Next, half of the prescribed number of dye molecules was placed on the remaining space of the nanotube on its right-hand side (r.h.s.) and the second half in its left-hand side (l.h.s.) (Figure S10B). The number of DOX molecules was 12 because this was the largest number possible to pack in the CNT cavity without overlaps and the remaining space for dye molecules was still enough. The number of dye molecules varied depending on the size of an individual molecule. Generally, we tried to

commonly called as pH indicators. We recently found that such a change of pH can significantly affect the adsorption of congo red dye on carbon nanotubes.26 Thus, the question is whether such a change in charge distribution may lead to phenomena affecting binding strength of some drug molecules to some carriers, particularly to carbon nanotubes. This seems to be still unexplored area in the literature since, to the best our knowledge, dyes molecules have not been applied to control drug release from carbon nanotubes at acidic conditions. Our survey of about dozen of various combinations of dyes and nanotubes structures led however to very interesting results. We focused on dyes molecules which must be nontoxic and reveal pKa value about 6, i.e. should transform from the unprotonated to the protonated form in the mentioned pH range. As a model drug we selected doxorubicin, DOX, a commonly used anticancer drug belonging to the class of anthracycline chemotherapeutic agents. This paper is organized as follows: first we define all the systems being analyzed in this paper by using molecular dynamics simulations. Next, we analyze results obtained from standard unbiased simulation method and we select a few architectures which reveal promising properties. Next, these selected architectures are subjected to a more detailed analysis based on determination of free energy changes. From the determined free energy profiles we draw conclusions concerning the macroscopic observations, particularly longterm behavior of these systems. Finally, we point out which of the studied systems is the most promising one and deserves experimental realization.

2. METHODS 2.1. Definitions of the Analyzed Systems. The pH change from neutral to acidic one occurring in tumor tissue due to hypoxia is a natural and promising factor tha can be utilized for triggering drug release. That range of pH change is not very wide (from 7.4 to ca. 5.5) but enough to initiate changes in charge distribution of some organic dyes molecules. Among some number of studied dyes revealing pKa in the range 5−7, only a few of them led to more or less interesting results. They are listed in Table 1. BTB is a sulfonphthaleine dye with numerous applications in biological and medical fields. It is not only sensitive to pH, but is also used in sensor applications for detecting pesticides, CO2 and ammonia.28 Due to strong pH-dependent color change near the neutral pH region, the neutral red (NR) is used as an intracellular pH indicator. NR is also very useful biological probe and has been widely used for various purposes in many biological systems.33 MR belongs to the class of azo dyes.34 PPD is commonly used as hair dye.35 We consider coadsorption of those dyes with doxorubicin (DOX) on carbon nanotube in physiological conditions. DOX is a member of anthracycline class of chemotherapeutic agents used for the treatment of many common human cancers.36 However, the main adverse effect of anthracyclines is cardiotoxicity, which considerably limits their usefulness. Carbon nanotubes are widely studied as carriers of small molecule drugs.4 They provide large inner volume and external surface area, which can be used either for encapsulation of small molecule drugs or for attachment of various functional moieties on the sidewalls. The studies were carried out according to two strategies. The first one is the application of narrow nanotube with the inner diameter 7.8 Å (chirality (10,0)). In this case, the CNT cannot collect dye and DOX molecules in its inner cavity. The coadsorption occurs on the CNT sidewall only. The second approach is the application of a wider nanotube with chirality (30,0) and inner diameter of 23.4 Å. That nanotube can collect both species in its inner space. Thus, we consider the process of blocking/unblocking DOX molecules in the CNT interior by dye molecules incorporated to the CNT in a second impregnation cycle. C

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir pack as much dye molecules as possible in the remaining space of the nanotube. Initially, no water molecules were inserted to the inner space of the CNT. They, however, spontaneously filled the free interatomic space during the equilibration stage. The filling of CNT by water molecules occurred according to an unbiased molecular diffusion mechanism.

3. RESULTS AND DISCUSSION 3.1. Coadsorption of BTB, MR, and NR with Doxorubicin on (10,0) CNT. Calculations concerning the coadsorption of DOX with BTB, MR, and NR on the narrow nanotube (10,0) led to interesting observations, though, in the context of controlled release of DOX, the results are barely significant. Let us briefly discuss each system in the process of switching the pH from the neutral to acidic one. Such a change of pH leads to the transitions of deprotonated forms of dyes into their protonated counterparts, that is, the molecular charges and topologies change according to the schemes in Table 1. Technically, both cases of pH were treated independently, that is, starting configurations were subjected to heating, cooling, and equilibration steps separately. Thus, we compare equilibrium structures obtained for protonated and deprotonated forms of each dye in independent simulation runs. Figure 1A shows simulation snapshots obtained for the system containing BTB and DOX. It also shows the definitions of two parameters used for analysis of changes in molecular distributions as a function of pH. These two parameters rz and rc are necessary because the nanotube surface is not uniquely defined. The rz defines a radial distance from the nanotube axis. Using this parameter, we determine average numbers of molecules located at the distance rz + drz, but the averaging is performed only for molecules for which z-coordinates are in the range covered by the nanotube length. Other molecules, i.e., the molecules for which projection on the nanotube sidewall cannot be defined, are analyzed using the rc parameter. This parameter is defined as the distance from the center of mass of the terminal nanotube rings. Using this parameter, we count molecules that are located at the hemispheres of the radius rc and thickness drc. Thus, when the numbers of molecules obtained from the analysis of these two distances are summed up, we get the total number of molecules in the system. Figure 1B shows the molecular distributions of BTB and DOX in the considered systems. NBTB(rz) shows how the number distribution of BTB molecules changes with the distance from the CNT sidewall, whereas NBTB(rc) describes the number distribution of BTB molecules as a function of the distance from the CNT tips (both tips are included). Analogously, we defined the functions for the number distributions of DOX molecules. We can notice that at neutral pH, BTB actually does not adsorb on the nanotube. The dye molecules prefer to form smaller clusters distributed quite uniformly in the bulk. The NBTB(rz) does not reveal any significant increase of BTB density in the vicinity of the nanotube. Similar behavior can be observed for DOX, that is, no preferential adsorption on the surface of the nanotube occurs, and only some small increase of DOX concentration is observed at large distance from the CNT, i.e., ∼ 15 Å. Both components weakly localize on the CNT tips. Of the total 80 DOX molecules in the system, only 12.7 (on average) were detected on the tips. In the case of DOX, only 0.35 molecule localized on the tips, while the rest, i.e., 19.65 were detected on the sidewall. At acidic pH, a significant increase of the number

Figure 1. (A) Snapshot of the system state for coadsorption of BTB with DOX on (10,0) CNT at temperature 310 K and ionic strength of solution 0.145 mol L−1. Blue: deprotonated BTB; red: protonated BTB; green: DOX. The bottom part shows definitions of two distances used for analysis of changes in molecular distributions as a function of pH. rz is the radial distance from the CNT axis, while rc is the radial distances (hemispherial) from the centers of masses of the terminal rings of the CNT. (B) Number distributions of DOX and BTB as functions of the distance from the CNT sidewall (left panel) and CNT tips (right panel) for the case of coadsorption of these components on (10,0) CNT at temperature 310 K and ionic strength of solution 0.145 mol L−1. Blue (solid) lines show the cases of neutral pH while red (dashed) lines are for the acidic pH.

of BTB molecules at close distances from the CNT occurs. The number of DOX molecules increases at larger distance from the CNT than at neutral pH, and they actually disappear in the tips region. However, the interaction energies with the CNT are still rather small, as shown in Table 2, and the increase of BTB density can be attributed to the formation of a big cluster, which partially attached to the CNT sidewall. That attachment is the effect of high BTB concentration in solution rather than an enhanced affinity to the CNT. The increase of DOX density is due to weakening its interaction with BTB at acidic pH. A D

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

strongly on the CNT sidewall as well (the energy is about 100 kJ mol−1 per molecule); however, there is a visible tail in the NMR(rz) function, which suggests that a few MR molecules diffused to the bulk. After lowering pH, the adsorption of MR becomes stronger (sharp peak in NMR(rz) function at small distances) and the sidewall of the CNT becomes quite tightly wrapped by the layer of MR molecules. Some number (5.6 molecules) of MR molecules belonging to the tips region has been shifted to the sidewall at acidic pH. DOX is, in turn, repelled from the adsorbed layer of MR and diffuses to the bulk mainly in the form of a few molecule clusters. It is confirmed by the behavior of NDOX(rz) function, which reveals a long tail at larger distances from the nanotube. DOX also appears at the tips region after lowering pH, but this is not a very strong effect. Generally, in this case, the change of pH from neutral to acidic significantly affects the interactions of DOX with MR; we observe more than 10-fold decrease of the interaction energy. Thus, DOX molecules entrapped in the potential well created by the adsorbed on the CNT MR molecules at neutral pH become released to the bulk at acidic pH. The drawback of this system is probably too weak adsorption of MR at neutral pH (or quite strong repulsion between MR molecules). It is likely that at lower concentrations of MR its adsorbed layer would be not more than monomolecular. It is enough to trap the DOX molecules at neutral pH but probably too little to repel them from the adsorbed layer at acidic pH. However, this system definitely deserves more detailed studies, as it reveals quite promising properties. Co-adsorption results of NR and DOX are presented in Figure 3A,B. In this case, we observe quite intense change of the adsorbed phase structure upon lowering pH. However, a closer inspection of the NNR(rz;rc) and NDOX(rz;rc) functions leads to the conclusion that the change concerns only the NR structure. DOX is affected little by the pH change; in both conditions, its adsorption on the CNT is relatively weak (the interaction energy of DOX with CNT is ∼ −20 kJ mol−1), and this conclusion is confirmed by long tails in NDOX functions. NR is strongly adsorbed at neutral pH, and it locates only on the sidewalls. At acidic pH we observe spreading of NR molecules over the whole simulation box. Thus, this system cannot act as pH controlled carrier of DOX. However, the intense desorption of NR from the CNT surface is an important observation. The interaction energy with CNT per single NR molecule decreases from −167 to −35 kJ mol−1 when going from neutral to acidic pH. Definitely, at neutral pH, NR adsorbs strongly on the CNT surface, leading to almost uniformly covered surface by the layer of this dye. The protonation of NR, which should occur at pH close to 6, leads to actual total removal of dye from the CNT surface. That effect is obviously due to the dominant role of electrostatic repulsion between protonated forms of NR molecules and it can be utilized in other circumstances as will be discussed soon. 3.2. Blocking of DOX in the Inner Cavity of (30,0) Nanotube by PPD and NR. The diameter of (30,0) nanotube is 23.4 Å and doxorubicin can locate in its inner cavity. Thus, this is the second strategy in utilizing coadsorption of dyes with DOX as pH sensitive drug carrier. In order to get notions about possible effects occurring after switching the pH, we assumed the following protocol. Twelve DOX molecules were placed inside the nanotube. Next, some number of dye molecules in their deprotonated forms were incorporated to the remaining volume of the CNT. As a result, DOX molecules became locked by a layer of dye

Table 2. Interaction Energies between Various Components of the Systemsa

BTB MR NR

neutral acidic neutral acidic neutral acidic

dye− dye

DOX− dye

dye− CNT

DOX− CNT

DOX− DOX

−4478 −472 +297 −78 +10 +247

−841 −584 −1200 −56 −45 +4

−14 −23 −108 −132 −167 −35

−12 −22 −55 −30 −19 −24

−737 −736 −634 −752 −749 −747

a

The systems contain (10,0) CNT, 80 dye molecules, and 20 DOX molecules. The energies are expressed in kJ mol−1 per single molecule.

general conclusion is that the coadsorption of BTB with DOX does not lead to immobilization of DOX on the nanotube, and lowering the pH cannot induce its release from that coadsorbed phase. Although some shift of DOX from the nanotube at acidic pH is observed, this effect is too weak to be utilized in pH-triggered drug release. Figure 2A,B shows analogous results for the case of methyl red, MR. In this case, we can observe enhanced concentration of DOX on the CNT sidewall at neutral pH. MR adsorbs quite

Figure 2. (A) Snapshot of the system state for coadsorption of MR with DOX on (10,0) CNT at temperature 310 K and ionic strength of solution 0.145 mol L−1. Blue: deprotonated MR; red: protonated MR; green: DOX. (B) Number distributions of DOX and MR as functions of the distance from the CNT sidewall (left panel) and CNT tips (right panel) for the case of coadsorption of these components on (10,0) CNT at temperature 310 K and ionic strength of solution 0.145 mol L−1. Blue (solid) lines show the cases of neutral pH while red (dashed) lines are for acidic pH. E

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

Figure 4. Coadsorption of PPD with DOX in (30,0) CNT at temperature 310 K and ionic strength of solution 0.145 mol L−1. Blue: deprotonated PPD; red: protonated PPD; green: DOX. The bottom part shows distributions of a given component within the nanotube interior. The z-coordinate defines the position within the nanotube (z = 0 corresponds to half of the CNT length). Vertical dashed lines illustrate positions of the nanotube ends. Animation showing the process of lowering the pH value is provided as a web-enhanced object (Supporting Information). The pH change occurs when the dye molecules change color from blue to red. Figure 3. (A) Snapshot of the system state for coadsorption of NR with DOX on (10,0) CNT at temperature 310 K and ionic strength of solution 0.145 mol L−1. Blue: deprotonated NR; red: protonated NR; green: DOX. (B) Number distributions of DOX and NR as functions of the distance from the CNT sidewall (left panel) and CNT tips (right panel) for the case of coadsorption of these components on (10,0) CNT at temperature 310 K and ionic strength of solution 0.145 mol L−1. Blue (solid) lines show the cases of neutral pH while red (dashed) lines are for acidic pH.

molecules within the volume of the CNT. PPD also form clusters because the interaction between neutral forms of PPD is strong as seen in Table 3. It leads to the formation of “corks” Table 3. Interaction Energies between Various Components of the Systemsa

PPD

adsorbed on each side of the CNT. The systems were subjected to standard heating, cooling, and equilibration steps without application any extra forces preventing the adsorbed molecules from escaping to the bulk. Two dyes led to interesting results, namely PPD and NR. After full equilibration and production runs, we stated that the systems are stable at the considered conditions, i.e., pH and temperature. In the next step we switched the deprotonated forms of dyes into their protonated forms, which mimic lowering the pH of solution below the pKa. The replacement of the deprotonated forms of dyes into their protonated counterparts is accompanied by a slight change of the molecular topology of the considered dyes (see Table 1). Incorporation of extra hydrogen to each dye molecule, existing in the already equilibrated structure at neutral pH, leads to small overlaps mainly with water molecules. Thus, it was necessary to run a few tens of integration steps in the NVT ensemble with self-control of a maximum displacement occurring in the systems. So, after those few tens of relaxation steps the simulations were continued in the standard isothermal−isobaric ensemble. Figure 4 shows results for the case of blocking a cluster of DOX molecules by PPD. At neutral pH DOX is effectively blocked by PPD, as we did not observe spreading of drug

NR

neutral acidic neutral acidic

dye− dye

DOX− dye

dye− CNT

DOX− CNT

DOX− DOX

−382 −170 +45.0 +386

−255 +198 −58.0 +534

−49.8 −18.8 −141 −134

−124 −221 −134 −143

−659 −680 −682 −610

a

The systems contain (30,0) CNT, 12 DOX molecules and 60 PPD or 24 NR molecules. The energies are expressed in kJ mol−1 per single molecule.

at the nanotube ends by clusters of PPD molecules. As mentioned, that state was stable within the considered simulation time (up to 3.6 ns), and no release of dye or DOX was observed. After lowering the pH (switching to the protonated form of PPD), the dye molecules escaped quickly from the CNT interior. This is due to electrostatic repulsion between dye molecules at short distances. Each dye molecule becomes positively charged as a result of protonation of amine groups at acidic pH. This induces another factor driving the release of PPD, namely, a strong repulsion between DOX (carrying positive charge18) and PPD. Looking at Figure 4 and the distributions NPPD(z) and NDOX(z), we can see that CNT became rapidly unloaded from PPD (within ca. 1.2 ns), and DOX molecules spread out within the whole volume of the CNT. It is accompanied by the increase of interaction energy of DOX with the CNT. The F

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir remaining space, after PPD removal, was filled in water molecules as can be deduced from Nw(z) distributions. As can be seen, no release of DOX was observed within the 3.6 ns simulation time. The above qualitative analysis of PPD-DOX-CNT system leads to very promising results. The uncorking of the nanotube will occur quickly, though it also depends on the nanotube length. Important factors in this case are the long-term stability of the corked state at neutral pH and release dynamics of DOX after uncorking at acidic pH. These factors need more detailed quantitative analysis. Let us consider another combination of dye and DOX filled CNT. In Figure 3, we observed strong adsorption of NR on the CNT surface at neutral pH and its intense desorption at acidic conditions. This suggests application of NR as a pH-removable plug in the case of a wider (30,0) nanotube. Figure 5 shows the relevant results.

molecules during the 3.6 ns simulation time. Moreover, these released NR molecules seem to readily adsorb on the external wall of the CNT. Thus, the above qualitative observations suggest that the system NR−DOX−CNT is even more promising than its analogue with PPD. This is because the stability of the corked form of the nanotube is supposed to be higher. Also, the release of NR at acidic pH will be slower. Definitely, this system deserves more detailed studies. As mentioned in the Introduction section, there are no experimental evidence for DOX loading into the hollow space of carbon nanotubes. However, theoretical studies based on quantum chemical computations prove that the interaction energy of DOX with the inner walls of CNTs is much bigger than with the external walls.17 Our analysis confirms that observation as well (see DOX-CNT components in Tables 2 and 3). Thus, DOX should spontaneously fill the inner space of CNT by capillary suction mechanism. Obviously, that process depends on the DOX concentration in solution and the refractive index of solvent applied. So, we can only speculate that in cases where open-ended multiwalled nanotubes were used, DOX probably partially incorporated to those nanotubes. 3.3. More Detailed Discussion of the Most Promising Architectures. The above, rather qualitative, analysis of those various architectures leads to the conclusion that coadsorption of doxorubicin and dyes on the sidewalls of carbon nanotubes is the least interesting in terms of their application as pH controlled drug carriers. The most promising seems to be blocking of DOX inside the CNT by strongly interacting dyes at neutral pH and their release at acidic pH. Then DOX becomes unlocked and can be released to the bulk according to molecular diffusion mechanism. However, DOX seems to interact with CNT quite strongly either at neutral or acidic pH. Therefore, it is necessary to estimate the probability of spontaneous leaving the interior of the CNT by DOX molecules. To that purpose we applied biased simulations using the umbrella sampling methodology. The organization of the studies was the following: the system containing 12 DOX molecules encapsulated in the CNT interior was subjected to standard heating and equilibration stages without any bias potential. During those stages (lasting more than 3 ns) DOX did not escape from the CNT interior but spread out quite uniformly within the internal space of the CNT. Next, we selected one DOX molecule located at the vicinity of the CNT end and imposed to it a series of the umbrella potentials centered at various distances from the nanotube. As a result, we obtained a series of systems with that DOX molecule fixed at growing distances from the nanotube end. According to the umbrella sampling methodology, it is possible to extract the potential of mean force (pmf) along the studied trajectory by constructing histograms of that molecule positions and applying the weighted histogram analysis.42 Figure 6 shows the relevant results, i.e., pmf (which can be identified with the free energy) versus position of the DOX center of mass. The same procedure was applied to the systems where DOX was blocked by the coadsorbed dyes at neutral pH, i.e., the systems from Figures 4 and 5. But in this part of the studies, more important is the stability of a given dye molecule than the DOX molecule, which is blocked by the clusters of dye molecules. Therefore, the dragging was imposed on a dye molecule located in the vicinity of the nanotube end. The relevant results, i.e., the obtained potentials of mean force

Figure 5. Co-adsorption of NR with DOX in (30,0) CNT at temperature 310 K and ionic strength of solution 0.145 mol L−1. Blue: deprotonated NR; red: protonated NR; green: DOX. The bottom part shows distributions of a given component within the nanotube interior.. Animation showing the process of lowering the pH value is provided as a web enhanced object (Supporting Information). The pH change occurs when the dye molecules change color from blue to red.

As can be seen, NR locks DOX molecules at neutral pH in the inner cavity of the nanotube. We did not observe any escape of these molecules during the 3.6 ns simulation time and without any extra barriers preventing the release. Looking at intermolecular interaction energies, we state that NR is adsorbed much stronger than PPD at analogous conditions. Particularly, at neutral pH, the NR molecules are kept in the CNT interior mainly due to interaction with the CNT walls. The component coming from interaction with DOX is less important here, contrary to the case of PPD. At acidic pH, interaction of NR with DOX and with the other NR molecules becomes strongly repulsive. This effect drives the release of NR from the nanotube interior; however, the release rate is much slower than in the case of PPD. Therefore, the distribution functions shown in Figure 5 are weakly affected by the pH change, as we could detect the removal of only a few NR G

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

concept of transition state theory. The probability of the release is given by the Boltzmann factor, as mentioned already. The attempt frequency is, however, unknown. But we can roughly estimate that factor by assuming that the molecules residing in the CNT interior perform one-dimensional diffusion. The diffusion length l, i.e., the distance which a given molecule has traveled in time t, in the one-dimensional case, is given by l = 2 Dt where D is the diffusion coefficient. A single release attempt appears when a molecule approaches the CNT end. Thus, on average, a given molecule reaches the CNT end not later than L2/4D, where L is the CNT length. We can thus assume that the attempt frequency for the release is not smaller than 4D/L2. The diffusivity can be easily determined in MD simulations by monitoring the mean squared displacements of the molecules. For pure DOX encapsulated in the CNT interior we found that DDOX = 3.17 × 10−5 cm2s−1. It compares well to the diffusivity of DOX in PBS buffer determined experimentally, i.e., 2.96 × 10−5 cm2s−1.43 The diffusivities of NR, PPD and DOX blocked by these dyes were also determined and are collected in Table 4. As can be seen all of them are of the same order of magnitude but in the cases of larger loadings of the CNT internal space, the diffusivities are smaller than that for pure DOX. The least mobile are the DOX molecules blocked by NR. This is due to strong interaction of NR with the CNT walls; that interaction reduces the mobility of NR what, in turn, affects the mobility of DOX locked from both sides by the layers of NR. Trying to estimate the release rate of DOX, we can use its diffusivity, energetic barrier, and the nanotube length for the computation of the mean residence time t of DOX in the nanotube interior:

Figure 6. Potential of mean force for the dragging of DOX, PPD and NR molecules from the nanotube to the bulk obtained by applying the weighted histogram analysis to the umbrella sampling results. The applied force constant in the umbrella potential was 0.1 eV Å−2 and the successive umbrella windows were shifted by 2 Å. The dragging was imposed along the direction parallel to the nanotube axis. The snapshots show the starting configurations for the dragged molecules at the first umbrella window, i.e. when the molecule is still in the CNT interior. The nanotube end is located at z = 42 Å.

accompanied the release of dye molecules are provided in Figure 6. The transfer of DOX (when it is not blocked by dyes molecules) from the CNT interior to the bulk is accompanied by an energetic barrier of E = 22 kJ mol−1. Other molecules should experience similar barrier against spontaneous release because we started from the configuration that was reached spontaneously. Thus, other DOX molecules can approach to the CNT end, and their probability of release will be governed by the similar energetic barrier. Other energetic parameters determined for pure DOX encapsulated in the interior of (30,0) nanotube are similar as in the case of coadsorption with PPD at acidic conditions because that system also led to spreading of DOX within the whole volume of the CNT. The energy of interaction with CNT, UDOX‑CNT = −204 kJ mol−1, while UDOX‑DOX = −637 kJ mol−1. The energetic barrier for spontaneous DOX release is not big, but it is significant. Thus, the DOX molecules approaching the CNT end will be able to escape with the probability given by the Boltzmann factor, exp(−E/RT) = 1.82 × 10−4 at the considered temperature 310 K, as shown in Table 4. That probability is quite high in the molecular scale, and we can expect that spontaneous release of DOX from the CNT interior will be quite fast. Determination of the real release rate of DOX is not possible within the time scale available in the MD simulations. However, we can make some rough estimation of that value by using the

t=

⎛ E ⎞ L2 ⎟ exp⎜ ⎝ RT ⎠ 4D

Thus, the mean residence time of DOX in the nanotube considered in this study (L = 8 nm) would be 2.8 × 10−5 s. For a more realistic length of about 100 nm that time increases to 4.3 × 10−3 s. Therefore, we can conclude that DOX should leave the nanotube quickly for both cases, and its spontaneous release is feasible. Just to get a notion about the blocking performance of PPD and NR, we estimated their mean residence times using the same concept. As can be seen in Table 4, the blocking effect of PPD at neutral pH is actually an illusion because the mean residence time of PPD is not bigger than 100 ns. In the case of dyes, the estimated mean residence time should be rather viewed as its upper limit because the dyes molecules do not need to travel across the whole nanotube length in order to make a single release attempt. In fact, we observed a spontaneous escape of two PPD molecules at neutral pH during umbrella sampling calculations. So, we have to conclude that the present architecture of the system consisting of PPD

Table 4. Diffusivities, Energetic Barriers and Mean Residence Times Found for Pure DOX and Dye−DOX Encapsulated in the Interior of (30,0) Nanotube at Neutral pH mean residence time, s system DOX PPD+DOX NR+DOX

2

Ddye, cm s

−1

2.40× 10−5 1.67× 10−5

2

−1

DDOX, cm s

3.17 × 10−5 1.27 × 10−5 1.07× 10−5

−1

barrier height, kJ mol 22.2 (for DOX) 6.8 (for PPD) 82.8 (for NR) H

Boltzmann factor at 310 K

L = 8 nm

L = 100 nm

1.82 × 10−4 7.15 × 10−2 1.12 × 10−14

2.8 × 10−5 9.9 × 10−8 8.6 × 105

4.3 × 10−3 1.5 × 10−5 1.3 × 108

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir

based on the determination of free energy barriers against the release, and mean residence times in the nanotube interior, led to the conclusion that PPD cannot be the effective blocker of DOX at neutral pH. It is expected that PPD will leave the CNT faster than the DOX itself. On the other hand, every parameter determined for NR suggests that this particular molecule can act as an effective long time blocker of DOX at neutral pH. This system seems to be perfectly tight in terms of DOX leakage beyond the acidic tumor microenvironment. At the same time, the NR release at acidic pH is accompanied by negligible free energy barriers, so the nanotube will be quickly unloaded from NR at acidic pH and the release of DOX can occur. Thus, the system NR-DOX-CNT reveals very promising properties as a pH-controlled drug delivery system. It is definitely worth further experimental studies.

molecules as blockers of DOX is not functional. Perhaps further modifications of the nanotube tips would enhance the stability of such a system at neutral pH. On the other hand, the estimated mean residence time of NR turns out to be very long. It reaches macroscopic values for both nanotube lengths. Obviously, we should be careful while drawing conclusions from such a crude estimation method. However, even if we assume that the residence times are overestimated by 2 or 3 orders of magnitude, we still get very promising results. The times of the order of 105 s should be enough for stable blocking of drug molecules in the interior of nanotubes with lengths of a few hundreds of nanometers. Thus, the system where DOX is blocked by NR at neutral pH should work properly as pH triggered drug carrier. Because we observed the spontaneous escape of a few NR molecules at acidic pH (Figure 5), we conclude that the barrier for NR release from the (30,0) nanotube at acidic pH is small and of the order of kBT. Thus, DOX release at acidic pH will be accompanied by a barrier comparable to the barrier determined for the pure component (it can be higher by a factor of the order of kBT). It is well-known that the dynamics of a complex process is determined by the dynamics of its slowest stage. In this case of acidic pH, the barrier for DOX release is much higher than the barrier for NR release; therefore the dynamics of DOX release will be actually the same as in the case of pure DOX encapsulated in the CNT interior. Experimental realization of the drug carrier containing DOX molecules blocked by NR in the inner cavity of a carbon nanotube does not seem to be difficult. Thus, our theoretical results should stimulate further research in order to verify that unexplored strategy in construction of pH sensitive drug carriers.

■

ASSOCIATED CONTENT

S Supporting Information *

The Supporting Information is available free of charge on the ACS Publications website at DOI: 10.1021/acs.langmuir.6b00296. Detailed description of the force field parameters and their values (PDF) Animation showing the temporal evolutions of the system from Figure 4 while changing pH (AVI) Animation showing the temporal evolutions of the system from Figure 5 while changing pH (AVI)

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]; phone: +48 815375620. Notes

4. SUMMARY AND CONCLUSIONS In this work, we performed a theoretical analysis, based on molecular dynamics simulations, of coadsorption of doxorubicin and about dozen of various dyes revealing pKa close to 6 on the surface or in the inner cavity of carbon nanotubes. That theoretical survey led to selection of only a few combinations of systems that reveal potentially promising properties as pHcontrolled drug delivery systems. We found that coadsorption of doxorubicin and dyes on the external walls of carbon nanotubes leads to interesting phenomena but with rather low significance in terms of their application as pH-sensitive drug carriers. Only the neutral red (NR) and methyl red (to much lower extent) (MR) revealed the ability of binding DOX molecules in the adsorbed layer on the nanotube surface at neutral pH. At the same time, they desorbed from the nanotube together with DOX or extracted DOX molecules from the adsorbed layer at acidic pH. However, the observed effects could not be classified as unequivocal evidence of success, and a closer analysis of those systems was postponed. Another studied architecture was encapsulation of DOX inside the inner cavity of CNT and its further blocking by layers of dyes molecules incorporated to the nanotube in a second impregnation cycle. That construction of the systems led to very interesting results for both neutral red and p-phenylenediamine (PPD). We found, in standard unbiased simulation runs, that at neutral pH the layers of dyes forcibly lock the DOX molecules against the release. But, at acidic pH, they spontaneously escape from the nanotube interior leaving the DOX unlocked and ready to the release. More detailed studies,

The authors declare no competing financial interest.

■ ■

ACKNOWLEDGMENTS This work was supported by Polish National Science Centre Grant DEC-2012/07/E/ST4/00763. REFERENCES

(1) Allen, T. M. Drug Delivery Systems: Entering the Mainstream. Science 2004, 303, 1818−1822. (2) Arruebo, M.; Fernández-Pacheco, R.; Ibarra, M. R.; Santamaría, J. Magnetic Nanoparticles for Drug Delivery. Nano Today 2007, 2, 22− 32. (3) Prato, M.; Kostarelos, K.; Bianco, A. Functionalized Carbon Nanotubes in Drug Design and Discovery. Acc. Chem. Res. 2008, 41, 60−68. (4) Wong, B. S.; Yoong, S. L.; Jagusiak, A.; Panczyk, T.; Ho, H. K.; Ang, W. H.; Pastorin, G. Carbon Nanotubes for Delivery of Small Molecule Drugs. Adv. Drug Delivery Rev. 2013, 65, 1964−2015. (5) Pastorin, G. Crucial Functionalizations of Carbon Nanotubes for Improved Drug Delivery: A Valuable Option? Pharm. Res. 2009, 26, 746−769. (6) Chaban, V. V.; Prezhdo, O. V. Water Boiling Inside Carbon Nanotubes: Toward Efficient Drug Release. ACS Nano 2011, 5, 5647− 5655. (7) Panczyk, T.; Warzocha, T. P. Monte Carlo Study of the Properties of a Carbon Nanotube Functionalized by Magnetic Nanoparticles. J. Phys. Chem. C 2009, 113, 19155−19160. (8) Ulbrich, K. Polymeric Anticancer Drugs with pH-Controlled Activation. Adv. Drug Delivery Rev. 2004, 56, 1023−1050. (9) Sahoo, N. G.; Bao, H.; Pan, Y.; Pal, M.; Kakran, M.; Cheng, H. K. F.; Li, L.; Tan, L. P. Functionalized Carbon Nanomaterials as Nanocarriers for Loading and Delivery of a Poorly Water-Soluble I

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX

Article

Langmuir Anticancer Drug: A Comparative Study. Chem. Commun. 2011, 47 (18), 5235. (10) Tian, Z.; Yin, M.; Ma, H.; Zhu, L.; Shen, H.; Jia, N. Supramolecular Assembly and Antitumor Activity of Multiwalled Carbon Nanotube−Camptothecin Complexes. J. Nanosci. Nanotechnol. 2011, 11 (2), 953−958. (11) Tripisciano, C.; Rümmeli, M. H.; Chen, X.; Borowiak-Palen, E. Multi-Wall Carbon Nanotubes - a Vehicle for Targeted Irinotecan Drug Delivery. Phys. Status Solidi B 2010, 247 (11−12), 2673−2677. (12) Huang, H.; Yuan, Q.; Shah, J. S.; Misra, R. D. K. A New Family of Folate-Decorated and Carbon Nanotube-Mediated Drug Delivery System: Synthesis and Drug Delivery Response. Adv. Drug Delivery Rev. 2011, 63 (14−15), 1332−1339. (13) Ji, Z.; Lin, G.; Lu, Q.; Meng, L.; Shen, X.; Dong, L.; Fu, C.; Zhang, X. Targeted Therapy of SMMC-7721 Liver Cancer in Vitro and in Vivo with Carbon Nanotubes Based Drug Delivery System. J. Colloid Interface Sci. 2012, 365 (1), 143−149. (14) Dinan, N. M.; Atyabi, F.; Rouini, M.-R.; Amini, M.; Golabchifar, A.-A.; Dinarvand, R. Doxorubicin Loaded Folate-Targeted Carbon Nanotubes: Preparation, Cellular Internalization, in Vitro Cytotoxicity and Disposition Kinetic Study in the Isolated Perfused Rat Liver. Mater. Sci. Eng., C 2014, 39, 47−55. (15) Anbarasan, B.; Babu, S. V.; Elango, K.; Shriya, B.; Ramaprabhu, S. pH Responsive Release of Doxorubicin to the Cancer Cells by Functionalized Multi-Walled Carbon Nanotubes. J. Nanosci. Nanotechnol. 2015, 15 (7), 4799−4805. (16) Ali Mohammadi, Z.; Aghamiri, S. F.; Zarrabi, A.; Talaie, M. R. A Comparative Study on Non-Covalent Functionalization of Carbon Nanotubes by Chitosan and Its Derivatives for Delivery of Doxorubicin. Chem. Phys. Lett. 2015, 642, 22−28. (17) Wang, Y.; Xu, Z. Interaction Mechanism of Doxorubicin and SWCNT: Protonation and Diameter Effects on Drug Loading and Releasing. RSC Adv. 2016, 6 (1), 314−322. (18) Fülöp, Z.; Gref, R.; Loftsson, T. A Permeation Method for Detection of Self-Aggregation of Doxorubicin in Aqueous Environment. Int. J. Pharm. 2013, 454, 559−561. (19) Ren, Y.; Pastorin, G. Incorporation of Hexamethylmelamine inside Capped Carbon Nanotubes. Adv. Mater. 2008, 20 (11), 2031− 2036. (20) Pastorin, G.; Li, J.; Yoong, S. L.; Goh, W. J.; Czarny, B. M. S.; Yang, Z.; Poddar, K.; Dykas, M.; Patra, A.; Thirumalai Venkatesan, T.; et al. In Vitro Controlled Release of Cisplatin from Gold-Carbon Nanobottles via Cleavable Linkages. Int. J. Nanomed. 2015, 7425. (21) Samorì, C.; Ali-Boucetta, H.; Sainz, R.; Guo, C.; Toma, F. M.; Fabbro, C.; da Ros, T.; Prato, M.; Kostarelos, K.; Bianco, A. Enhanced Anticancer Activity of Multi-Walled Carbon Nanotube−methotrexate Conjugates Using Cleavable Linkers. Chem. Commun. 2010, 46 (9), 1494−1496. (22) Modi, C.; Patel, S.; Desai, A.; Murthy, R. S. R. Functionalization and Evaluation of PEGylated Carbon Nanotubes as Novel Drug Delivery for methotrexate. J. Appl. Pharm. Sci. 2011, 1 (5), 103−108. (23) Singh, R.; Mehra, N. K.; Jain, V.; Jain, N. K. GemcitabineLoaded Smart Carbon Nanotubes for Effective Targeting to Cancer Cells. J. Drug Target. 2013, 21 (6), 581−592. (24) Lay, C. L.; Liu, H. Q.; Tan, H. R.; Liu, Y. Delivery of Paclitaxel by Physically Loading onto Poly(ethylene Glycol) (PEG)-Graftcarbon Nanotubes for Potent Cancer Therapeutics. Nanotechnology 2010, 21 (6), 065101. (25) Sobhani, Z.; Dinarvand, R.; Atyabi, F.; Ghahremani, M.; Adeli, M. Ghahremani. Increased Paclitaxel Cytotoxicity against Cancer Cell Lines Using a Novel Functionalized Carbon Nanotube. Int. J. Nanomed. 2011, 705−719. (26) Panczyk, T.; Wolski, P.; Jagusiak, A.; Drach, M. Molecular Dynamics Study of Congo Red Interaction with Carbon Nanotubes. RSC Adv. 2014, 4, 47304−47312. (27) Sabnis, R. W. Handbook of Acid-Base Indicators; CRC Press: Boca Raton, 2008. (28) De Meyer, T.; Hemelsoet, K.; Van Speybroeck, V.; De Clerck, K. Substituent Effects on Absorption Spectra of pH Indicators: An

Experimental and Computational Study of Sulfonphthaleine Dyes. Dyes Pigm. 2014, 102, 241−250. (29) Das, D. K.; Goswami, P.; Barman, C.; Das, B. Methyl Red: A Fluorescent Sensor for Hg2+ over Na+, K+, Ca2+, Mg2+, Zn2+, and Cd2+. Environ. Eng. Res. 2012, 17, 75−78. (30) Shaikh, M.; Choudhury, S. D.; Mohanty, J.; Bhasikuttan, A. C.; Pal, H. Contrasting Guest Binding Interaction of Cucurbit [7−8] Urils with Neutral Red Dye: Controlled Exchange of Multiple Guests. Phys. Chem. Chem. Phys. 2010, 12, 7050−7055. (31) Dong, S.; Chi, L.; Yang, Z.; He, P.; Wang, Q.; Fang, Y. Simultaneous Determination of Dihydroxybenzene and Phenylenediamine Positional Isomers Using Capillary Zone Electrophoresis Coupled with Amperometric Detection. J. Sep. Sci. 2009, 32, 3232− 3238. (32) McMurry, J. Organic Chemistry, 8th ed.; Brooks/Cole, Cengage Learning: Belmont, CA, 2012. (33) Harrington, T. D.; Tran, V. N.; Mohamed, A.; Renslow, R.; Biria, S.; Orfe, L.; Call, D. R.; Beyenal, H. The Mechanism of Neutral Red-Mediated Microbial Electrosynthesis in Escherichia Coli: Menaquinone Reduction. Bioresour. Technol. 2015, 192, 689−695. (34) Spólnik, P.; Król, M.; Stopa, B.; Konieczny, L.; Piekarska, B.; Rybarska, J.; Zemanek, G.; Jagusiak, A.; Piwowar, P.; Szoniec, G.; et al. Influence of the Electric Field on Supramolecular Structure and Properties of Amyloid-Specific Reagent Congo Red. Eur. Biophys. J. 2011, 40, 1187−1196. (35) Zhang, Y.; Kim, C.; Zheng, T. Hair Dye Use and Risk of Human Cancer. Front. Biosci., Elite Ed. 2012, 4, 516−528. (36) Skowronek, M.; Stopa, B.; Konieczny, L.; Rybarska, J.; Piekarska, B.; Szneler, E.; Bakalarski, G.; Roterman, I. Self-Assembly of Congo RedA Theoretical and Experimental Approach to Identify Its Supramolecular Organization in Water and Salt Solutions. Biopolymers 1998, 46, 267−281. (37) Stuart, S. J.; Tutein, A. B.; Harrison, J. A. A Reactive Potential for Hydrocarbons with Intermolecular Interactions. J. Chem. Phys. 2000, 112, 6472. (38) Wang, J.; Wang, W.; Kollman, P. A.; Case, D. A. Automatic Atom Type and Bond Type Perception in Molecular Mechanical Calculations. J. Mol. Graphics Modell. 2006, 25, 247−260. (39) Wang, J.; Wolf, R. M.; Caldwell, J. W.; Kollman, P. A.; Case, D. A. Development and Testing of a General Amber Force Field. J. Comput. Chem. 2004, 25, 1157−1174. (40) Dupradeau, F.-Y.; Pigache, A.; Zaffran, T.; Savineau, C.; Lelong, R.; Grivel, N.; Lelong, D.; Rosanski, W.; Cieplak, P. The R.E.D. Tools: Advances in RESP and ESP Charge Derivation and Force Field Library Building. Phys. Chem. Chem. Phys. 2010, 12, 7821. (41) Plimpton, S. Fast Parallel Algorithms for Short-Range Molecular Dynamics. J. Comput. Phys. 1995, 117, 1−19. (42) Kumar, S.; Rosenberg, J. M.; Bouzida, D.; Swendsen, R. H.; Kollman, P. A. THE Weighted Histogram Analysis Method for FreeEnergy Calculations on Biomolecules. I. The Method. J. Comput. Chem. 1992, 13, 1011−1021. (43) Mansoor, I.; Lai, J.; Ranamukhaarachchi, S.; Schmitt, V.; Lambert, D.; Dutz, J.; Häfeli, U. O.; Stoeber, B. A Microneedle-Based Method for the Characterization of Diffusion in Skin Tissue Using Doxorubicin as a Model Drug. Biomed. Microdevices 2015, 17, 9967.

J

DOI: 10.1021/acs.langmuir.6b00296 Langmuir XXXX, XXX, XXX−XXX