Since polymeric micelles are promising and have potential in drug delivery systems, people have become more interested in studying the compatibility of polymeric carriers and drugs, which might help them to simplify the preparation method and increase the micellar stability. In this article, we report that cationic amphiphilic drugs can be easily encapsulated into PEGylated phospholipid (PEG–PE) micelles by self-assembly method and that they show high encapsulation efficiency, controllable drug release and better micellar stability than empty micelles. The representative drugs are doxorubicin and vinorelbine. However, gemcitabine and topotecan are not suitable for PEG–PE micelles due to lack of positive charge or hydrophobicity. Using a series of experiments and molecular modelling, we figured out the assembly mechanism, structure and stability of drug-loaded micelles, and the location of drugs in micelles. Integrating the above information, we explain the effect of the predominant force between drugs and polymers on the assembly mechanism and drug release behaviour. Furthermore, we discuss the importance of pKa and to evaluate the compatibility of drugs with PEG–PE in self-assembly preparation method. In summary, this work provides a scientific understanding for the reasonable designing of PEG–PE micelle-based drug encapsulation and might enlighten the future study on drug–polymer compatibility for other polymeric micelles.
Polymeric micelles composed of amphiphilic block copolymers are of particular interest in drug delivery systems and several micellar formulations have been in different stages of clinical trials [1–3]. These nano-sized drug carriers have many advantages, including easy preparation, biodegradability, controllable drug release, long systemic circulation time and enhanced accumulation in tumours via the enhanced permeability and retention effect . In order to better develop polymeric micelle delivery system, on the one hand, people are engaged in developing new polymers to be carriers; on the other hand, people have become more interested in studying the compatibility of polymers and drugs [5,6]. If we can identify the scope of suitable drugs for a specific polymeric micelle, we will not only simplify the method of preparing drug-loaded micelles, but also easily find the best application of newly developed polymeric micelles for drug delivery.
Poly(ethylene glycol)–phosphatidylethanolamine (PEG–PE) is a kind of biodegradable amphiphilic diblock copolymer, which can self-assemble to spherical micelles with core–shell structure in aqueous solution . This polymer is negatively charged in water due to the electric dissociation of the phosphate group that exists between hydrophilic PEG and hydrophobic PE (figure 1b). PEG–PE micelles are usually used to encapsulate hydrophobic drugs by dissolving them in micellar phospholipid core . In order to successfully load these poorly soluble drugs, polymers and drugs are needed to be dissolved together in an organic solvent, and then the organic solvent is removed by evaporation or dialysis. This kind of preparation procedure has some disadvantages. One is that the remains of an organic solvent might be hazardous because of incomplete evaporation or dialysis. Another one is that the size distribution of drug-loaded micelles might be non-uniform due to the asynchronous precipitation of drugs and polymers from an organic solvent .
Recently, we reported that PEG–PE micelles can also successfully encapsulate some water-soluble drugs by a very simple one-step self-assembly preparation method in aqueous solutions. These drugs included doxorubicin hydrochloride and vinorelbine tartrate . We found that both doxorubicin-loaded PEG–PE micelles and vinorelbine-loaded PEG–PE micelles showed high drug encapsulation efficiency (more than 99%), sustainable controlled drug release and superior anti-cancer therapy efficacy to free drugs in vitro and in vivo [11,12]. This one-step self-assembly preparation method is simple, fast and efficient, but it cannot be applied to every drug. For example, gemcitabine hydrochloride and paclitaxel cannot be successfully encapsulated into PEG–PE micelles using this method .
pKa and of drugs are two important factors to evaluate their electrostatic and hydrophobic interaction with polymers, respectively, and the interaction pattern and intensity further affect drug encapsulation efficiency and micellar stability [13,14]. In order to identify the scope of suitable drugs in our case, we picked four drugs to investigate their compatibility with PEG–PE micelles in terms of self-assembly ability, drug release behaviour, structure and in vivo stability. Three drugs are doxorubicin, vinorelbine and topotecan that are all cationic but have different values. The fourth drug is gemcitabine, which is not charged in a neutral solution (figure 1a). In our study, we firstly investigated the drug–polymer assembly process by isothermal titration calorimetry (ITC) to clarify whether the dominant drug–polymer interaction was electrostatic force (enthalpy-driven) or hydrophobic force (entropy-driven). Secondly, we analysed the structure of drug-loaded PEG–PE micelles using transmission electron microscopy (TEM), dynamic light scattering (DLS), static light scattering (SLS) and small-angle X-ray scattering (SAXS). A series of basic physico-chemical parameters were measured, including molecular weight, aggregation number, size, shape and core size. Furthermore, with the help of molecular dynamics (MD) simulation to verify the SAXS data, we confirmed the location of cationic amphiphilic drugs in PEG–PE micelles and discussed the consistency among drug location in micelles, drug–polymer interaction and drug physico-chemical properties. Integrating these results, we also explained the interesting phenomena that only doxorubicin-loaded micelles showed pH-dependent drug release behaviour, and on the other hand, only vinorelbine released from micelles exhibited great sensitivity to the change in the drug–polymer molar ratio. Since the drug encapsulation might bring different forces to stabilize the micelles, drug-loaded micelles might be more stable than empty micelles. In the last part, we compared the stability of empty micelles and doxorubicin-loaded micelles by detecting their disassembly through the fluorescence resonance energy transfer (FRET) technique. In summary, we provided a scientific understanding for the reasonable designing of PEG–PE micelle-based drug encapsulation, and analysed the importance of pKa and in evaluating compatibility of drugs for PEG–PE micelles in self-assembly preparation method.
2. Results and discussion
(a) Self-assembly of water-soluble drugs and PEG–PE micelles
When we tried to encapsulate doxorubicin, vinorelbine, topotecan or gemcitabine into PEG–PE micelles using one-step self-assembly method, we found they had very different drug encapsulation efficiency and micellar stability. Gemcitabine could not be successfully loaded into PEG–PE micelles, reflected by its encapsulation efficiency which was only 1.2%. By contrast, doxorubicin, vinorelbine and topotecan could easily self-assemble to PEG–PE micelles and obtained close to 100% encapsulation efficiency under an appropriate drug–polymer molar ratio . In addition, doxorubicin-loaded micelles and vinorelbine-loaded micelles presented sustainable controlled drug release and both of them retained drugs in micelles in agarose gel electrophoresis, but topotecan-loaded micelles had an initial burst drug release and disassembled in electric field (figure 2a,b).
PEG–PE is constituted of a hydrophilic PEG block and a hydrophobic PE block, and there is a phosphate group between these two blocks (figure 1b). Because of the electric dissociation of the phosphate group, PEG–PE micelles are negatively charged at pH=7. In a previous study, we brought up the idea that negative charge of PEG–PE played an important role in triggering the assembly of cationic drugs and PEG–PE, and the hydrophobic PE block provided a stable reservoir for drug loading . According to the pKa value, gemcitabine is not charged at pH 7 (figure 1a), which means that it does not satisfy the prerequisite for self-assembly due to lack of electrostatic attraction. But the other three drugs are positively charged in water (figure 1a), so they can easily self-assemble to PEG–PE micelles. In terms of hydrophobic interaction, the values of vinorelbine, doxorubicin and topotecan are 4.58, 0.92 and −0.36, respectively, implying only doxorubicin and vinorelbine can adhere to the PE block whose value is as high as 12.11. The hydrophobicity difference explained why the use of doxorubicin and vinorelbine rather than topotecan can lead to controllable release and stable micellar structure.
In order to measure the proportion of electrostatic force and hydrophobic force in PEG–PE micelles encapsulating drugs, we performed ITC to study the self-assembly process of PEG–PE micelles with doxorubicin, vinorelbine and topotecan in water and in 0.15 M sodium chloride at 313 K. The ITC data of PEG–PE micelles titrating into the three drugs in water are presented in figure 3 and the thermodynamic parameters are summarized in table 1. The control of PEG–PE micelles titrating into pure water had no thermal change (data not shown). For all three cationic drugs, the titrations were exothermic (ΔH<0) and spontaneous (ΔG<0), indicating that drugs can self-assemble to PEG–PE micelles in water. All three titration curves looked like a reverse S; this is because PEG–PE has only one electrostatic binding site for positively charged drugs. Doxorubicin, vinorelbine and topotecan have one, two and one positive charge at pH=7 (figure 1a), respectively; therefore, their measured maximum binding number was close to 1, 2 and 1, respectively (table 1). This result not only proved that the electrostatic attraction was the prerequisite for drug–polymer self-assembly, but also provided information about how to choose the best drug–polymer molar ratio when preparing drug-loaded PEG–PE micelles.
Although PEG–PE micelle self-assembly with the three cationic drugs was favoured by both enthalpy and entropy (ΔH<0,TΔS>0), we found electrostatic force dominating in doxorubicin-loaded micelles (the biggest |ΔH| among the three drugs), while hydrophobic force played a bigger role in vinorelbine-loaded micelles (the biggest |TΔS| among the three drugs). This conclusion was further confirmed by the titration results in 0.15 M sodium chloride. Sodium ion and chloride ion can interfere with the drug–polymer assembly by blocking electrostatic attraction; therefore, the dominant force in micelles was more obvious. As shown in table 1, the ΔG of doxorubicin and vinorelbine increased 1.1 and 1.5 kcal mol−1 in sodium chloride solution, respectively, but the major force in micelles was not affected by sodium chloride, reflected by the fact that the ΔH of doxorubicin and the TΔS of vinorelbine had no obvious change. Because topotecan lacked enough hydrophobic interaction with PEG–PE due to its negative , its TΔS changed from 3.6 to −0.9 kcal mol−1, and its binding constant Kb decreased by 103. Weak hydrophobic force explained the poor stability of topotecan in PEG–PE micelles, which was the reason topotecan-loaded micelles showed initial burst drug release in the accelerated release experiment and easy disassembly in electrophoresis (figure 2a,b).
Dominant force in drug-loaded micelles not only determined the self-assembly mechanism, but also induced different drug release behaviour. If we increase the drug–polymer molar ratio, the averaged hydrophobic force for each drug will be attenuated and drug retention ability in micelles will decrease. Finally, it will result in accelerated drug release from micelles. Vinorelbine has stronger hydrophobic interaction with PEG–PE than doxorubicin because of its larger value. For this reason, vinorelbine release from PEG–PE micelles was more sensitive to the change of drug–polymer molar ratio (figure 2c,d). On the other hand, although both doxorubicin and vinorelbine are cationic drugs and have electrostatic attraction with PEG–PE, only doxorubicin-loaded PEG–PE micelles showed pH-dependent drug release behaviour (figure 2e,f). The reason might be the difference in the nitrogen atom between the two drugs. The primary amine in doxorubicin (pKa=8.94) is the only binding site for PEG–PE, and its increased dissociation at pH=5 rather than pH=7.4 increased the electrostatic repulsion of the drug and attenuated its electrostatic attraction to the phosphate group of PEG–PE. This change destabilized the micellar structure and accelerated the drug release. However, vinorelbine has two cationic amines and two anionic amines (figure 1a). The electrostatic attraction between cationic amines and the phosphate group of PEG–PE was interfered with by the anionic amines, and therefore electrostatic attraction played a minor role in stabilizing vinorelbine-loaded micelles. For this reason, vinorelbine release was not dependent on the solution pH.
In summary, we concluded that the scope of drugs suitable for loading to PEG–PE micelles by self-assembly method was positively charged, water soluble and having enough hydrophobicity, in other words, they need to be cationic and amphiphilic.
(b) Structure of drug-loaded PEG–PE micelles
In the second part, we made a comprehensive investigation of the structure of doxorubicin-loaded micelles and vinorelbine-loaded PEG–PE micelles. We changed the drug–polymer molar ratio and measured different structure parameters using TEM, SAXS, DLS and SLS.
TEM images showed that both empty micelles and drug-loaded micelles were spherical nanoparticles whose diameter was about 20 nm, but the size distribution of drug-loaded micelles was more uniform than empty micelles (figure 4a). SAXS results verified the size and the shape of PEG–PE micelles, and further supplied information about micellar internal structure. As shown in figure 4b,e, the collimation-corrected relative scattering function I(q), I(q)=S(q)P(q), showed the typical features of sphere scattering of empty micelles and drug-loaded PEG–PE micelles. Pair-distance distribution function p(r) is the inverse Fourier transformation of P(q) calculated by the GIFT technique (figure 4c,f). From p(r), we calculated the entire micellar diameter and the hydrophobic core diameter Dc (table 3). The diameter of the empty PEG–PE micelle was about 18 nm and the diameter of its hydrophobic core was about 4.3 nm (Dc), which equals the length of two PE molecules. This result confirmed the core–shell structure of PEG–PE micelles and agreed well with a previous report . Loading doxorubicin did not change the core size and the entire size of PEG–PE micelles, but loading vinorelbine increased the core size (Dc) owing to its larger molecular volume (figure 1a).
Zeta potential, ζ, hydrodynamic radius, Rh, and particle dispersion index (PDI) of micelles were measured by DLS at 298 K in water. Loading doxorubicin or vinorelbine did not change the Rh of micelles, indicating that the PEG shell which is related to the micellar diffusion and viscosity was not affected by drugs. However, the decreased PDI value after drug encapsulation suggested that the size distribution was improved, which was consistent with the TEM images. Empty PEG–PE micelles had negative ζ due to the dissociation of the phosphate group. Loading cationic drugs increased micellar ζ, and this increase was dependent on the drug loading amount (table 2). This change was also reflected in the agarose gel electrophoresis of doxorubicin-loaded micelles (figure 2b). Carboxyfluorescein-labelled empty PEG–PE micelles had the fastest movement towards the positive pole, while M-Dox1/1 (doxorubicin/PEG–PE=1/1) had the slowest movement.
Molecular weight Mw, aggregation number Nagg and gyration radius Rg of micelles were measured by SLS. The Nagg of empty PEG–PE micelles was about 90, which was consistent with the result obtained by SAXS reported before . Loading doxorubicin or vinorelbine did not change Nagg, implying that drugs replaced only the water in micelles and did not induce re-assembly of micelles. Although Rh of drug-loaded micelles had no obvious change, Rg decreased after drug loading, resulting in the change of Rg/Rh. Rg/Rh can be used to evaluate whether the particle has a compact or loose structure. The Rg/Rh of spherical micelles with swollen shell is about 1.3, while the Rg/Rh of solid sphere is about 0.77 . As shown in table 2, the Rg/Rh of empty PEG–PE micelles was about 1.235. After loading drugs, this parameter decreased, indicating drug-loaded micelles were more compact than empty micelles.
In conclusion, our results demonstrated that loading cationic amphiphilic doxorubicin or vinorelbine into PEG–PE micelles using self-assembly method did not change micellar size, shape and aggregation number, but improved their size distribution and made them become compact. Usually, dissolving hydrophobic drugs into the hydrophobic core of micelles swells the core size, and sometimes it induces re-assembly of micelles and a change in the aggregation number . Once the micelles re-assemble, it becomes hard to predict the property and stability of drug-loaded micelles. Besides, using evaporation or dialysis method to load hydrophobic drugs might induce non-uniform size distribution of drug-loaded micelles . However, our self-assembly method to encapsulate cationic amphiphilic drugs to PEG–PE micelles showed advantages over the above two aspects. Its advantages include uniform size distribution, increased stability and easy way of predicting the properties of drug-loaded micelles. This finding proved the importance and convenience of choosing compatible drugs for a specific polymeric micelle.
(c) Localization of cationic amphiphilic drugs in PEG–PE micelles
Since doxorubicin and vinorelbine are positively charged and amphiphilic, it is highly possible that these drugs are distributed at the core–shell interface of PEG–PE micelles where the micelles have negative charges apart being from amphiphilic. In order to verify this hypothesis, we combined SAXS experiments and MD simulation to study the location and status of doxorubicin in PEG–PE micelles.
In SAXS experiments, we de-convoluted p(r) functions by the Decon program and obtained the radial electron density fluctuations Δρ(r) (subtraction of water in arbitrary units; figure 4d,g). The r-value at Δρ(rC)=0 is the radius of the hydrophobic core (rC). And the r-value at the maximum of Δρ, , reveals the place drugs locate in micelles. These data are listed in table 3. After encapsulating doxorubicin or vinorelbine, the change of Δρ(r) mainly occurred at r=2–4 nm, indicating that these drugs are distributed at the interface between hydrophobic PE core and hydrophilic PEG shell. Doxorubicin encapsulation did not alter rC and , but both of these two parameters increased slightly after loading vinorelbine because of its larger molecular volume.
To verify the status of cationic amphiphilic drugs at the core–shell interface of PEG–PE micelles, we adopted MD simulation to analyse the nanostructure of doxorubicin-loaded micelles. Two PEG–PE micelle models were constructed in our experiments, one was an empty PEG–PE micelle that contains 90 PEG2000-DSPE molecules, and the other one was a doxorubicin-loaded micelle that contains 90 PEG2000-DSPE molecules and 90 doxorubicin molecules (M-Dox1/1). As shown in figure 5a, for the empty micelle, after 1.5 ns of equilibration, the long flexible PEG chain bent towards the micelle centre, and the system dimension rapidly reduced from 36 to 16.5 nm. Because the size reduced too quickly, we reconfigured the micelle system by replacing a 40 nm cubic water box with a 20 nm box and performed longer simulations. After another 10 ns of simulation, we found that the micelle size fluctuated at the 16.5 nm level until the end of the simulation (see electronic supplementary material, video S1).
In the model of the doxorubicin-loaded micelle, doxorubicin molecules were initially placed between the PE block and PEG block, so that the positively charged nitrogen in doxorubicin can interact with the negatively charged phosphate in PEG–PE by electrostatic attraction. Similarly, the size of the doxorubicin-loaded micelle reduced from 36 to 17.5 nm in a total of 2 ns of simulation (figure 5b). The micelle size rapidly decreased in the first 0.5 ns and decreased much slower thereafter. After reducing the solvation water box to 20 nm cubed, we found that the micelle size also reduced to 16.5 nm in 8 ns of simulation and fluctuated at this dimension during the rest of the simulation (see electronic supplementary material, video S2). The plot of micelle diameter change against simulation time is shown in figure 5c. In the last 2 ns of simulation, both the empty micelle and the doxorubicin-loaded micelle fluctuated at 16.5 nm, suggesting that these systems were fully relaxed and reached equilibration. And this result agreed well with our above experiments.
In addition, we calculated the mass density against the micelle radius (figure 5d), and these results were averaged over 21 selected frames. For each frame, the micelle centre was defined as the mass centre of phosphorus atoms for all the PEG–PE molecules and only heavy atoms were included in the calculation. After loading doxorubicin, mass density showed an obvious increase from 2 to 4 nm, which was the interface of the PE core and the PEG shell. The results of micelle size and drug location matched the previous SAXS experiments (figure 4d). Furthermore, we calculated the radial probability of several representative atoms in PEG–PE and doxorubicin. As shown in figure 5e,f, the nitrogen atom distribution of doxorubicin matched the phosphorus atom distribution of PEG–PE. The marked carbon atoms from PE block and oxygen atoms from PEG blocks showed that the packing of PE and PEG in micelles had no big difference between the empty micelle and the doxorubicin-loaded micelle. These results indicated that PEG–PE micelles had enough space at the core–shell interface to load doxorubicin without changing the micellar core and shell, which was also proved by DLS, SLS and SAXS experiments. Comparing the RC and RN location of doxorubicin, we concluded that doxorubicin inserted its hydrophobic anthracycline ring into the PE core of phospholipid (figure 5g).
In summary, the PEG–PE micelle is an 18 nm diameter sphere with a 4 nm diameter core. Doxorubicin molecules distribute at the r=2–4 nm core–shell interface of micelles and insert their anthracycline ring into the hydrophobic core.
(d) Increased stability of drug-loaded micelles
According to the DLS and SLS results, drug encapsulation made PEG–PE micelles more compact and improved the size distribution (table 2). Besides, MD simulations showed that adding doxorubicin accelerated the micelle to reach equilibration (figure 5c). The above information implied that doxorubicin-loaded micelles had better stability than empty micelles. In order to test this hypothesis, we prepared FRET empty PEG–PE micelles and FRET doxorubicin-loaded micelles, and then detected their disassembly in vitroand in vivo according to the FRET efficiency. FRET empty micelles were PEG–PE micelles, which were incorporated with 0.5% (molar) NBD–PE and 0.5% (molar) Rhodamine B (Rho)–PE (figure 6a, left). When FRET empty micelles were excited at 460 nm, donor NBD transferred its energy to the acceptor Rho and yielded the fluorescent emission peak of Rho at 590 nm. FRET drug-loaded micelles contained Alexa660- PEG–PE and doxorubicin (figure 6a, right). Alexa660 was attached to the outer surface of the PEG shell, while doxorubicin molecules were distributed at the core–shell interface. Owing to the close distance between doxorubicin and Alexa660 (the radius of the micelle is less than 10 nm), doxorubicin can transfer energy to Alexa660 and yield the Alexa660 emission peak at 690 nm when doxorubicin is excited at 543 nm.
In the in vitro experiments, we detected FRET spectra of the two kinds of FRET micelles in purified water, 1 mg ml−1 albumin and 20% serum. As shown in figure 6b, in the hydrophobic environments (albumin or serum), the FRET ratio of empty PEG–PE micelles decreased but the FRET ratio of doxorubicin-loaded Alexa660-PEG–PE micelles had no obvious drop. Similar results were obtained in the in vivo experiments. 100 μl of FRET micelles (10 mM PEG–PE) were injected into mice intravenously, and then the micellar fluorescence in the blood vessels of the mesentery was observed by confocal microscopy. At 15 min after injection, the NBD/Rho FRET of empty micelles was obvious. However, this FRET disappeared at 1 h after injection, suggesting empty micelles had disassembled (figure 6c). But the doxorubicin/Alexa660 FRET was observed at both 15 min and 1 h after injection, and the FRET peak showed no obvious decrease. Besides PEG–PE micelles, some other polymeric micelles were also reported unstable and easily disassembled in serum or in blood . This is because hydrophobic force is usually weak in empty micelles and sometimes electrostatic repulsion between the same ionic groups further destabilizes the micellar structure. However, loading cationic amphiphilic drugs into PEG–PE micelles brings the additional hydrophobic force between polymers and drugs to stabilize the structure; on the other hand, the electrostatic attraction between amine and phosphate groups replaces the electrostatic repulsion among phosphate groups. Taken together, that is why doxorubicin-loaded micelles became more stable than empty micelles and retained integrity in serum and in blood.
Compatibility of drugs and polymeric carriers is an important issue in drug delivery systems. In this study, we confirmed cationic amphiphilic drugs were suitable for PEG–PE micelles using self-assembly method, and pKa and were two important parameters to evaluate the drug–polymer compatibility. Electrostatic attraction triggered the drug–polymer assembly and hydrophobic force helped to stabilize the drug retention in micelles. Lack of positive charge (pKa<7) or hydrophobicity resulted in failed drug encapsulation or quick drug release, such as gemicitabine and topotecan. By contrast, doxorubicin and vinorelbine, which are classified as being cationic amphiphilic drugs, can bind the phosphate group of PEG–PE and distribute at the amphiphilic core–shell interface of PEG–PE micelles. These drug-loaded micelles exhibited uniform size distribution, controllable drug release and more stable structure than empty micelles. Besides, drug loading did not change micellar size, shape and aggregation, which will help us to easily predict the properties of drug-loaded PEG–PE micelles. In summary, our study provided a scientific understanding about the one-step self-assembly method to prepare drug-loaded PEG–PE micelles.
4. Material and methods
1,2-Distearoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000](ammonium salt) (DSPE-mPEG2000) and DSPE-mPEG2000-carboxyfluorescein were purchased from Avanti Polar Lipids (Alabaster, AL, USA). The drugs used were doxorubicin hydrochloride (Hisun Pharmaceutical Co. Ltd, Zhejiang, China), vinorelbine tartrate (Hansen Pharmaceutical Co. Ltd, Jiangsu, China), gemcitabine hydrochloride (Hansen Pharmaceutical Co. Ltd, Jiangsu, China) and topotecan hydrochloride (Lanbeizhihua Co. Ltd, Chengdu, China).
(b) Preparation of drug-loaded PEG–PE micelles
Drug-loaded PEG–PE micelles were prepared by one-step self-assembly method described before . Briefly, PEG–PE micelle solution and drug solution were mixed to obtain the desired drug–polymer molar ratio, and the mixture was incubated at 333 K for 30 min to allow for better drug encapsulation. Encapsulation efficiency detection and drug release experiments were performed using high-performance liquid chromatography as described before .
(c) Isothermal titration calorimetry
The heat flow of PEG–PE binding drugs was measured using ITC at 313 K (MicroCal, LLC, Northampton, MA, USA). In each titration, the 200 μl reaction cell was loaded with different drugs and sequences of 38 successive 1 μl aliquot injections were performed using a 40 μl auto-syringe filled with 10 mM PEG–PE aqueous solution using a stir speed of 400 r.p.m. The drug concentration was 1 mM. Control experiments were performed using PEG–PE titration into pure water to correct the heat effects of dilution and mixing. The calorimetric data were analysed and converted into enthalpy change using MicroCal Origin v. 7.0 software (OriginLab Corp., Northampton, MA, USA). The experimental data were fitted into a one-site binding model, and the binding constant (Kb), binding number (N), the enthalpy (ΔH), the entropy (ΔS) and the free energy of binding (ΔG) were calculated with the software.
(d) Transmission electron microscopy, dynamic light scattering and static light scattering
The size and morphology of micelles were detected by TEM as described before . Micelles were diluted to 1 mg ml−1 with pure water, and then stained with 1% uranyl acetate on a copper grid. The stained micelles were observed by a JEOL 100 CX electron microscope (JEOL USA, Inc., Peabody, MA, USA).
Rh (hydrodynamic radius) and ζ (zeta potential) of micelles were determined by DLS using a Malvern Zetasizer Nano ZS (λ=633 nm, Malvern Instruments Ltd, Malvern, UK) at 298 K. Rh was expressed as intensity-weighted radius. The micelles were diluted to 20 μM PEG–PE with pure water.
Rg (gyration radius), Mw (molecular weight) and Nagg (aggregation number) of micelles were detected by SLS. An 18-angle laser light scattering DAWN HELEOS (λ=658 nm, Wyatt Technology Corp., Santa Barbara, CA, USA) detector was connected to a refractive index detector Optilab rEX (λ=658 nm, Wyatt Technology Corp., Santa Barbara, CA, USA). In the batch model experiment, a series of dilutions of the sample in the range of 10–100 μM were injected into the detecting system from low to high concentration and data were collected by the ASTRA software. Data were normalized with 40 kD dextran and adjusted by alignment. Mw and Rg were calculated from the ASTRA software by global fit and assessed by Zimm plot based on Rayleigh–Gans–Debye equation. Nagg was evaluated based on Mw of the micelles. All the experiments were performed at 298 K.
(e) Small-angle X-ray scattering
SAXS measurements of micelles were performed at beamline 1W2A of BSRF (Beijing, China). The incident X-ray wavelength (λ) was chosen to be 0.155 nm by a triangle bending Si(111) monochromator. A two-dimensional CCD detector was used to record the two-dimensional scattering intensity distribution. A total of 10 mM PEG–PE micelles were investigated at 298 K. The SAXS pattern of pure water was collected as the scattering background. All these two-dimensional data were integrated into the one-dimensional I(q) profiles as a function of the magnitude of the scattering vector q (, where 2θ is the total scattering angle). The sample-to-detector distance was fixed to 1.6 m to cover a q-range of 0.15–3.00 nm−1. After normalizing the scattering intensities I(q) to the incident X-ray intensities, the contributions of scattering background from sample cell and solvent were removed. GIFT and Decon programs (Institut für Chemie Graz, Austria) were used to extract the information about micellar shape, size and electron density distribution function from I(q).
The total scattering intensity I(q) of one particle system (identical and randomly oriented) can generally be formulated as where n is the number density of the particles in solution, S(q) is the static structure factor, which comes from the inter-particle interactions, and P(q) is the oriented averaged form factor, which comes from the intra-particle interactions and contains all the information about the particle (shape, size and internal electron density fluctuation).
The pair-distance distribution function (PDDF), p(r), is the real-space coordinate associated with P(q) and connected with the Fourier transformation: Generalized indirect Fourier transformation (GIFT) [20–22] technique was used to calculate p(r) of micelles. In our calculation, we used the charged monodisperse sphere model of screened Coulomb (Yukawa) potential with the hypernetted chain closure relation. The details of this technique can be found elsewhere [21–23]. In p(r)−r plots, the maximum and minimum (or the bump) at the low-r part of p(r) come from the convolution of negative and positive electron density fluctuations of the hydrophobic core and hydrophilic shell in micelles, respectively. is the r value at which p(r) goes to zero, and a semi-quantitative measure of the hydrophobic core diameter Dc is calculated from the distance corresponding to the inflection point located at the high-r side of the local minimum.
SAXS can also supply inner structure information of nanoparticles via the convolution square of the electron density contrast, Δρ(r), which is the electron density difference between nanoparticles and their surroundings: where Δρ(r) is the electron density fluctuation at the position r and r is the distance between two scattering centres within the particle. For spherical scattering objects, the scattering length density profile is a function of the radial position. The radial electron density profile is related to the PDDF p(r): Program Decon [24,25], a numerical convolution square root technique, was used to calculate the radial electron density distribution, Δρ(r).
(f) Molecular dynamics simulations
All the molecular modelling was performed by VMD , with the assistance of energy minimization and short MD simulations with NAMD . The DSPE molecule was modelled from a 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphoethanolamine (POPE) molecule taken from a fully equilibrated POPE bilayer. A 44-mer of polyethylene glycol (PEG) chain was added to the DSPE molecule to construct a full PEG2000-DSPE molecule. The initial doxorubicin conformation was taken from the PDB entry 1I1E . Ninety PEG2000-DSPE molecules were used to construct an empty micelle. During the construction, the position of the phosphorus atom in each molecule was first distributed randomly and semi-uniformly on the surface of a sphere with 5.5 nm diameter. Then, the PEG2000-DSPE molecule was placed based on the position of the phosphorus atom and with the PEG chain pointing straight opposite to the micelle centre. Therefore, the DSPE fatty acid chains pointed to the centre with random axial orientation. To construct a micelle with doxorubicin, we first made 90 PEG2000-DSPE/doxorubicin pairs, with the nitrogen atom of the doxorubicin located 3.5 Å away from one of the free oxygen atoms of the DSPE phosphate and the relative orientation of the doxorubicin was selected randomly. The micelle with doxorubicin was constructed in a similar way as the empty micelle. The two micelles were then energy minimized sequentially with heavy atoms fixed, glycerol group fixed, only phosphorus atoms fixed and all atoms free, multiple 2000 steps for each case. This energy minimization step is crucial since many clashes exist between two PEG2000-DSPE molecules or between the PEG2000-DSPE and doxorubicin molecules. The structures were checked carefully after and between each minimization step to ensure the removal of the clashes and any possible un-physical configuration will be modified manually. At the end of the energy minimization steps, no clashes can be detected. The energy minimized structures were solvated with TIP3 waters in very large water boxes of 400×400×400 Å 3 to cover the full PEG chains, and then neutralized with Ca2+ and Cl− ions with an ion concentration of 150 mM, resulting in systems with more than six million atoms. The resulting systems were subjected to MD simulations for equilibration. For the empty micelle, again multiple steps of energy minimization were performed, followed by 1.5 ns equilibration with the phosphorus atoms constrained at the original positions. Because the random coil of the PEG chain and the micelle dimension were reduced, the equilibrated system was then reduced to 200×200×200 Å3 by removing the unnecessary waters, the reduced system was re-neutralized, and then energy minimized and equilibrated for 2 ns without any constraints. Finally, 8 ns of simulations were carried out as production run. The system with doxorubicin was simulated following similar procedures, except that an additional 0.5 ns equilibration was performed first on the big system with both phosphorus atoms and the doxorubicin nitrogen atoms constrained. All the MD simulations were carried out with NAMD  under CHARMM all-atom force field. We used the CHARMM force field for lipids  and ethers , and the missing parameters on the DSPE and PEG chain linker region and that for doxorubicin were adapted from these with similar structures in CHARMM general force field . A 12 Å cutoff was used for van der Waals interactions and particle mesh Ewald summation was used to calculate the electrostatic interactions in all simulations. During the equilibration phases, temperature was controlled at 310 K using Langevin dynamics with a damping coefficient of 5 ps−1, and pressure was controlled at 1 atm by the Langevin piston method. Similar controls were applied to the production run except that the damping coefficient was reduced to 0.5 ps−1. The last 2 ns of production run of each system was used for statistical analyses.
This work was supported by grants from National Natural Sciences Foundation of China (nos. 90606019, 30901869 and 31070827), the National Science and Technology Special Project of Major New Drugs Creation (2009ZX09501–025) and the China-Finland Inter-Governmental S&T Cooperation Project (2008DFA01510).
One contribution of 17 to a Theme Issue ‘Molecular nanostructure and nanotechnology’.
- © 2013 The Author(s) Published by the Royal Society. All rights reserved.