This is an open access article published under an ACS AuthorChoice License, which permits copying and redistribution of the article or any adaptations for non-commercial purposes.
Article pubs.acs.org/JPCB
Release of Solubilizate from Micelle upon Core Freezing Jing Dai,† Zahra Alaei,† Beatrice Plazzotta,‡ Jan Skov Pedersen,‡ and István Furó*,† †
Division of Applied Physical Chemistry, Department of Chemistry, KTH Royal Institute of Technology, SE-10044 Stockholm, Sweden ‡ Interdisciplinary Nanoscience Center (iNANO) and Department of Chemistry, Aarhus University, Gustav Wieds Vej 14, 8000 Aarhus C, Denmark S Supporting Information *
ABSTRACT: By combining NMR (yielding 1H chemical shift, spin relaxation, and self-diffusion data) and small-angle X-ray scattering experiments, we investigate the complex temperature dependence of the molecular and aggregate states in aqueous solutions of the surfactant [CH3(CH2)17(OCH2CH2)20OH], abbreviated as C18E20, and hexamethyldisiloxane, HMDSO. The latter molecule serves as a model for hydrophobic solubilizates. Previously, the pure micellar solution was demonstrated to exhibit core freezing at approximately 7−8 °C. At room temperature, we find that HMDSO solubilizes at a volume fraction of approximately 10% in the core of the C18E20 micelles, which consists of molten and thereby highly mobile alkyl chains. Upon lowering the temperature, core freezing is found, just like in pure micelles, but at a temperature shifted significantly to 3 °C. The frozen cores contain immobile alkyl chains and exhibit a higher density but are essentially devoid (volume fraction below 1%) of the solubilizate. The latter molecules are released, first gradually and then rather steeply, from the core in the temperature range that is roughly delimited by the two core freezing temperatures, one for pure micelles and one for micelles with solubilizates. The release behavior of systems with different initial HMDSO loading follows the same master curve. This feature is rationalized in terms of loading capacity being strongly temperature dependent: upon lowering the temperature, release commences once the loading capacity descends below the actual solubilizate content. The sharp release curves and the actual release mechanism with its molecular features shown in rich detail have some bearing on a diverse class of possible applications.
1. INTRODUCTION Aggregates of amphiphilic molecules, ranging from simple surfactants to block copolymers and biomimetic species, provide environments where molecules not directly soluble in the main solvent environment can solubilize.1 Hence, for example, hydrophobic molecules can be made part of aqueous solutions in a thermodynamically stable manner. The applications of this phenomenon are wide-ranging, where the delivery of hydrophobic drugs is just one of many examples.2−8 The current study concerns a narrow, yet important subfield. Namely, in many applications, the solubilized molecules are supposed to be released from within the aggregates to provide some particular function. For example, the hydrophobic drugs mentioned earlier are supposed to reach their particular targets within tissues and cells. Often, one wishes to regulate the release. This can be performed in a fashion that sets the release rate in a predefined manner to a particular value or to a particular timeline. Yet, in many cases, the desirable way of action is to release upon a particular trigger or stimulus. It is easy to see advantages with this latter strategyif, for example, release of a drug is triggered by either temperature or light conditions having been changed at a particular location, the drug load can locally be made high without harming the rest of © 2017 American Chemical Society
the organism. Other advantages in other applications are abundant. Hence, triggered release of solubilizates (often and particularly, drugs) is a rich and burgeoning field both within current physical and colloid chemistry and molecular pharmacology. Often, the molecular units exploited for release are rather large block copolymers that, in turn, lead to rather large (∼100 nm) aggregates, frequently termed nanoparticles. Typical release stimuli are pH, dilution, and temperature.2−8 In the case of a temperature trigger, one commonly exploited molecular mechanism is lower/upper critical solution temperature for one of the molecular blocks present.9,10 Recently, we investigated11 a conceptually simple, yet seldom12−16 observed phase transition behavior in micelles made up by CH3(CH2)(n−1)(OCH2CH2)mOH (abbreviated to CnEm, note that this definition was incorrect in the abstract of our previous paper on a related topic11) surfactants consisting of a short (n = 18) alkyl chain connected to a longer m = 20− 100 ethylene oxide chain (indeed, one discussion point is Received: September 7, 2017 Revised: October 18, 2017 Published: October 20, 2017 10353
DOI: 10.1021/acs.jpcb.7b08912 J. Phys. Chem. B 2017, 121, 10353−10363
Article
The Journal of Physical Chemistry B distinguishing between surfactants and block copolymers16). At room temperature, the micelles formed in aqueous solution were small (∼5 nm hydrodynamic radius) globular objects consisting of a small (approx. 2 nm radius, containing approx. 6% of the total micellar volume) hydrophobic alkyl core and a more extensive hydrophilic ethylene oxide-dominated shell. As is typical for small micelles, at room temperature, the core exhibited liquid-like molecular mobility, as witnessed by the narrow 1H NMR peaks for the alkyl moieties. Upon lowering the temperature, the core froze in the 0−10 °C temperature region. Freezing was easy to detect by the broadening of the alkyl 1H NMR peaks; at the same time, the NMR signal from the moieties in the shell remained unaffected, thereby indicating the shell retained its high molecular mobility. Small-angle X-ray scattering (SAXS) measurements demonstrated that core freezing led to tighter packing of the alkyl chains leading to a slight decrease of the core radius and a significant change of the X-ray contrast of the core. The difference in core packing of the C18 chains of the two surfactants ultimately led to a phase separation in mixtures of C18E20 and C18E100 surfactants. Namely, at room temperature, the surfactants were found to form mixed micelles, whereas upon core freezing, we observed a reversible separation of C18E20 and C18E100 into distinct micellar species.11 This observation supplemented by the notion of tight chain packing within the frozen cores led us to inquire what may happen to solubilizates17 in micelles exhibiting core freezing. In addition, it has been established that in larger polymeric micelles, semicrystalline and amorphous cores exhibit different (up to a factor of 2) loadings by solubilizates.18,19 In one particular study, it has also been demonstrated that temperature change around the glass transition of the poly(D,L-lactide)-based core had an influence on the amount of solubilizate.20 As we are going to show below, lowering the temperature below the core freezing point for micelles without added solubilizates leads to release of model hydrophobic solubilizates in small C18E20 micelles. The release profile is sharp (large release within a narrow temperature range). The focus of this study is the generic mechanism, and at this stage, we do not seek practical applications. Hence, the solubilizate investigated is not a drug but a spectroscopically convenient hydrophobic probe molecule.
approximately 7.5 cm height. Subsequently, this NMR tube was centrifuged again at 1250 rpm for 30 min. This latter procedure was necessary to remove small amounts of residual nonsolubilized HMDSO that sometimes appeared (either because of temperature increase during centrifugation or mechanical disturbance during the handling of the vial) in the transferred micellar liquid. Centrifugation collected the nonsolubilized residual HMDSO at the top of the long liquid column; this was not detectable by eye. The tube was placed in the probe so that the NMR signal was collected from the lower part of the long liquid column and not influenced by any HMDSO collected at the top. The procedure above, carried out at room temperature 22 ± 1 °C, provided reproducible (within a few percentage as established by repeated preparations, see also Figure 3 below) HMDSO content, as measured by 1H NMR (see below), in the final solutions at a given amount of added HMDSO. 2.2. NMR Experiments. The 1H NMR spectra (see Figure 1, the full assignment has been provided previously and is recapitulated in the Supporting Information (SI)) of the various samples were recorded on a Bruker 500 Advance III HD spectrometer, equipped with a 5 mm conventional tripleresonance probe. The temperature dependence of the various parameters was recorded by decreasing the temperature from 20 to 0 °C at 2 °C steps from 20 to 10 °C and at 1 °C steps from 10 to 0 °C. After having set a new temperature, a stabilization time of 5 min passed before performing any new experiments. The integral intensity of the HMDSO peak at 20 °C, at 0 ppm, relative to that of the ethylene oxide peak was used to obtain the concentration of HMDSO in the prepared solutions (recall that the surfactant concentration was known). As we are going to show below, the volume of the solubilized HMDSO was below 5 μL in the 1 mL volume of the micellar solution in the NMR tube. This added volume has a negligible (that is, relative to experimental error) effect on the ethylene oxide concentration. The longitudinal spin relaxation time for those groups was detected by an inversion recovery experiment. The linewidth Δ1/2 of the peaks arising from the oxyethylene (at approx. 3.6 ppm) and methylene (at approx. 1.2 ppm) groups of Brij S20 was obtained at half amplitude. Note that this linewidth has a composite character and is, especially when it is small, contributed to by several features such as transverse spin relaxation, J-coupling, and, for the methylene peak in particular, chemical shift variation over the contributing moieties. The spectral intensities of the two peaks within the split HMDSO peak manifold at low temperatures were estimated by fitting two Lorentzian peaks to the experimental spectrum in the HMDSO range and extracting the two corresponding integrals. The reproducibility of those intensities was better than 2% of the total intensity. The diffusion NMR measurements21 were carried out on a Bruker 500 Advance III spectrometer, equipped with a Bruker DIFF 30 probe and GREAT 60 gradient amplifier. Because temperature-dependent diffusion experiments are potentially sensitive to convection artifacts, the convection-canceling double-stimulated-echo pulse sequence22 was used for recording the diffusional decays of the various peaks in the spectrum. The decay of spectral integrals was acquired with increasing gradient strength g and the self-diffusion coefficient D was obtained by fitting the Stejskal−Tanner expression23,24
2. EXPERIMENTAL SECTION 2.1. Materials and Sample Preparation. C18E20 (with commercial name Brij S20), hexamethyldisiloxane (HMDSO, NMR grade), and heavy water (99.9 atom % D) were purchased from Sigma-Aldrich. All materials were used as received. Micelle stock solutions were prepared by dissolving C18E20 in heavy water at a weight concentration of 1% (corresponding to approx. 9.7 mM). A set (1.5 mL) volume of the stock solution was filled into high-performance liquid chromatography vials that could be closed by a rubber septum. A set volume (from 0.5 to 80 μL) of HMDSO was added to a vial, which was then closed and vortex-mixed for at least 5 min. The vial was then centrifuged for 30 min at 6500 rpm with its septum down. For large added HMDSO volumes, this resulted in a two-phase sample, with the less dense (ρ = 0.76 g/cm3) HMDSO-rich phase making a layer at the top. From the vial kept with septum down, 1 mL of solution from the lower aqueous part was removed through the septum by a syringe and added to a 5 mm NMR tube. The liquid filled the tube up to 10354
DOI: 10.1021/acs.jpcb.7b08912 J. Phys. Chem. B 2017, 121, 10353−10363
Article
The Journal of Physical Chemistry B
tration of C18E20 is much (by about 200 times) higher than the critical micelle concentration (46 μM25), most surfactants are collected within micelles and the recorded diffusion coefficient reflects the motion of the micelles. Hence, from Ds, the hydrodynamic radius Rh of the micelles was calculated via the Stokes−Einstein equation Ds =
kBT 6πηTR h
(2)
where kB is the Boltzmann constant, T is the absolute temperature, and ηT is the viscosity of the solution at the given temperature T. 2.3. SAXS Experiments and Data Evaluation. SAXS data were acquired at the flux- and background-optimized Bruker AXS NanoSTAR instruments at Aarhus University. The pure C18E20 samples were measured on an older instrument, which uses a rotating anode Cu source, homebuilt scatterless slits,26 and a homebuilt thermostated flow-through capillary.27 The oilloaded samples were measured on a newer instrument, which uses a Ga liquid metal jet source (Excillum),28 homebuilt scatterless slits,26 a homebuilt thermostated flow-through capillary, and an automated sample-handler based on Gilson components. Water was measured as background and also used for absolute scale calibration.27 The intensity is displayed as a function of the modulus of the scattering vector q = 4π sin(θ)/ λ, where λ = 1.54 and 1.34 Å is the X-ray wavelength for Cu and Ga sources, respectively, and 2θ is the angle between the incident and scattered X-rays. The SAXS data (see Figure 2) for loaded and unloaded micelles display a large variation with temperature, and there
Figure 1. 1H NMR spectrum of C18E20 dissolved at 1 wt % in D2O, solubilizing 0.25 mg/mL hexamethyldisiloxane within the micelles. (A) The full spectrum at 20 °C. The chemical shift scale is defined by having set the water peak to 4.75 ppm. See peak assignment in SI. The spectral range around the HMDSO peak, indicated by the dashed lines, is shown expanded in (B) at 20 °C, in (C) at 7 °C, in (D) at 4 °C, and in (E) at 1 °C. All shifts were referenced to the water chemical shift at 20 °C set to 4.75 ppm.
I (g ) = exp( −γ 2g 2δ 2(Δ−δ /3)D) I(0)
Figure 2. SAXS data for C18E20 solutions with (dashed lines) and without (full lines) HDMSO at 1 °C (blue), 5 °C (black), 10 °C (green), and 15 °C (red). For clarity, the error bars of the individual experimental points are presented in Figures S7 and S8 in the SI.
are significant differences between the data with and without HMDSO at high temperature. The data were analyzed using a well-established29,30 model similar to the one exploited by us previously.11 The model has a spherically symmetric core−shell geometry with core radius R and outer radius R + Dshell, where the Dshell is the width of the shell. The interfaces are graded with a width of σcore for the core−shell interface and σouter for the outer interface. To reduce the number of fit parameters, σcore was fixed at 1.0 Å and Dshell = 2.5σouter. Additional fit
(1)
to the experimental decays. In eq 1, I(g) is the value of a specific spectral integral at a particular value of g, γ is the gyromagnetic ratio, δ is the duration of the gradient pulse, and Δ is the diffusion time, the latter values set to 1 and 200 ms, respectively. The surfactant self-diffusion coefficient Ds was derived from the oxyethylene peak that remained narrow over the whole explored temperature range. Because the concen10355
DOI: 10.1021/acs.jpcb.7b08912 J. Phys. Chem. B 2017, 121, 10353−10363
Article
The Journal of Physical Chemistry B
mg/mL of solubilized HMDSO. This value is about 500 times the aqueous solubility of HMDSO in D2O (approx. 0.0005 mg/ mL, as established by us after having equilibrated a water− HMDSO two-phase sample and measured the HMDSO 1H NMR intensity in the water phase). Clearly, with our current preparation procedure, a large excess of HMDSO was needed to reach saturation, which is probably a consequence of the low aqueous solubility of HMDSO. Yet, establishing the saturation level in this manner was more reliable than exploring a lower amount of HMDSO in combination with (far) longer preparation times. From the known (as obtained from solid densities) molar volumes of HMDSO (VHMDSO = 0.212 L/mol) and the C18 chains31 (VC18 = 0.314 L/mol) and the known molar concentrations cHMDSO and cs, the volume fraction ϕ of HMDSO in the micelle core can be obtained as
parameters were the relative scattering length density of the shell relative to the core, an overall scale factor, and a background. Note that the contribution of HMDSO to the core scattering could not be distinguished, probably because its amount is relatively small (see below) and/or because it is relatively homogeneously distributed within the core. The data for the pure micelles could not be fitted without including a structure factor to account for concentration effects. A hard-sphere structure factor with an effective hard-sphere volume fraction and hard-sphere interaction radius was used. The results for both of these had reasonable magnitudes (0.028 ± 0.003 and 57 ± 2 Å, respectively). Because of the different behavior at low q of the SAXS data of the loaded micelles, one model with sticky spherical micelles and another one with micelles with ellipsoidal cores were initially fitted to the data. However, the spherical model with the constraint Dshell = 2.5σouter could fit the data quite well without including more degrees of freedom in the model, so therefore, this model was used. When clustering was included in the sticky micelle model, the clustering was insignificant for temperatures from 7 °C and downwards for a reasonable separation of the micelles (2 times the outer radius of about 70 Å). For temperatures from 8 °C and upwards, the fits could be slightly improved by including a structure factor. But, to have the same model in the full temperature range, the simpler model without a structure factor and clustering was chosen. It should be noted that the change in results from including, respectively, structure factor and cluster formation relative to those obtained without such parameters is small (within twice the statistical error).
ϕ=
VHMDSO·c HMDSO VC18·cs + VHMDSO·c HMDSO
(3)
3. RESULTS AND DISCUSSION 3.1. HMDSO Loading and Saturation in C18E20 Micelles. The loading behavior and capacity of the micelles was established by displaying the HMDSO concentration in the final solution as a function of the amount of added HMDSO. As shown in Figure 3, the system gets saturated around 0.25
At saturation, approximately ϕ = 9.5% of the core volume was taken up by the solubilizate (see more below). Hence, the intermolecular environment for the alkyl chains changes only slightly, which is reflected by the small (
