Nanostructural Analysis of Water Distribution in Hydrated

May 6, 2015 - School of Pharmacy, University of East Anglia, Norwich, Norfolk NR4 ... Innovation Centre, Norwich Research Park, Norwich, Norfolk NR4 7...
1 downloads 0 Views 3MB Size
Article pubs.acs.org/molecularpharmaceutics

Nanostructural Analysis of Water Distribution in Hydrated Multicomponent Gels Using Thermal Analysis and NMR Relaxometry Doroty Codoni,†,§ Peter Belton,‡ and Sheng Qi*,† †

School of Pharmacy, University of East Anglia, Norwich, Norfolk NR4 7TJ, United Kingdom School of Chemistry, University of East Anglia, Norwich, Norfolk NR4 7TJ, United Kingdom § Procarta Biosystems Ltd., Innovation Centre, Norwich Research Park, Norwich, Norfolk NR4 7GJ, United Kingdom ‡

ABSTRACT: Highly complex, multicomponent gels and water-containing soft materials have varied applications in biomedical, pharmaceutical, and food sciences, but the characterization of these nanostructured materials is extremely challenging. The aim of this study was to use stearoyl macrogol-32 glycerides (Gelucire 50/13) gels containing seven different species of glycerides, PEG, and PEG-esters, as model, complex, multicomponent gels, to investigate the effect of water content on the micro- and nanoarchitecture of the gel interior. Thermal analysis and NMR relaxometry were used to probe the thermal and diffusional behavior of water molecules within the gel network. For the highly concentrated gels (low water content), the water activity was significantly lowered due to entrapment in the dense gel network. For the gels with intermediate water content, multiple populations of water molecules with different thermal responses and diffusion behavior were detected, indicating the presence of water in different microenvironments. This correlated with the network architecture of the freeze-dried gels observed using SEM. For the gels with high water content, increased quantities of water with similar diffusion characteristics as free water could be detected, indicating the presence of large water pockets in these gels. The results of this study provide new insights into structure of Gelucire gels, which have not been reported before because of the complexity of the material. They also demonstrate that the combination of thermal analysis and NMR relaxometry offers insights into the structure of soft materials not available by the use of each technique alone. However, we also note that in some instances the results of these measurements are overinterpreted and we suggest limitations of the methods that must be considered when using them. KEYWORDS: gels, thermal analysis, NMR relaxometry, microenvironment, water dynamics



INTRODUCTION Highly complex, multicomponent gels and water-containing soft materials have varied applications in biomedical, pharmaceutical, and food sciences. An understanding of the microstructure and networks within these gels is extremely important for further development of applications for these materials.1,2 As an example, in many pharmaceutical products mixtures of lipids and polymeric materials with surface-active properties are often used for improving the hydration and dissolution behavior of drugs.3,4 A clear understanding of the behavior of the material in biorelevant media can facilitate prediction of in vitro and in vivo behavior of these formulations. Depending on the miscibility of the components in the formulation with water, on hydration, water penetration into the material can lead to the formation of either homogeneous hydrogel-like networks or nanostructures with micro- and nanoscale phase-separated water pockets.5,6 © XXXX American Chemical Society

However, understanding the behavior of water in soft materials, when the systems contain multiple water species that increase the complexity of the material is extremely challenging. This study uses a combination of thermal, microscopic, and NMR relaxometry analysis to probe the distribution and interaction of water within a model hydrated multicomponent surface-active matrix formulation that has potential for use in the delivery of protein and peptide drugs. The model systems are hydrated stearoyl macrogol-32 glycerides, also known by the trade name of Gelucire 50/13. Gelucire 50/13 is a complex nonionic, water dispersible pharmaceutical excipient. It is composed of mainly of mono- and diacyl-polyoxyethylene glycols (72% w/w), Received: December 19, 2014 Revised: May 1, 2015 Accepted: May 6, 2015

A

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

point can be made very clearly with reference to water at the triple point. At this point, ice, water, and water vapor exist in equilibrium, each representing orders of magnitude differences in rotational and translational molecular correlation times. Below the triple point water vapor readily condenses directly as ice without an intervening liquid phase. The formation or not of ice, therefore, is no indication of the existence of a phase with reduced molecular mobility. NMR is complementary to thermal methods and helps provide further understanding of the gel nanostructure by probing the molecular dynamics of water in the hydrated matrix. Often, proton NMR relaxation time data of macromolecule/water systems are cited as evidence of “bound” or motionally inhibited water molecules.12,13 This interpretation needs to be treated with great caution. The relaxation behavior of macromolecular systems where there are exchangeable protons on the macromolecule are dominated by proton exchange processes and there is no evidencefor example, in the case of proteins14that any significant inhibition of the motion of the water at the interface with the macromolecule occurs. However, with careful interpretation NMR relaxometry can yield useful information on water behavior in gels.14−16 The Gelucire gels characterized in this study were prepared by two different methods, which was expected to provide different degrees of homogeneity. These two groups of gels were used to challenge the sensitivity of the characterization methods used in this study. The results obtained in this study can further be used to provide insights into the performance of the gels as potential carrier systems for delivering therapeutic agents such as proteins and peptides.

glycerides, a palmitostearic acid mixture (C16 and C18) (20% w/ w), and free PEG 1500 (8% w/w).7 Previously, Gelucire 50/13 has been reported to form bioadhesive gels after hydration, but no clear structure of the gels has been defined.8 To fully understand the behavior of water in such complex systems, thermodynamic, kinetic and molecular dynamic factors need to be investigated. Differential scanning calorimetry (DSC) explores thermodynamic properties but data interpretation may be affected by kinetic factors and nuclear magnetic resonance (NMR) can be used to give useful molecular dynamic data. A very common phenomenon in the thermal behavior of such systems is the observation of nonfreezing water.9 This is a fraction of the water that remains unfrozen when the bulk of the water in the system freezes. It is often assigned to some form of “bound water”, which is assumed to thermodynamically distinct from bulk water and also to have different molecular dynamic properties. These notions need to be addressed with extreme caution, particularly for many multicomponent pharmaceutical gels. Net transfer of matter from one phase to another occurs when the activity of one phase is slightly less than the activity of the other phase. When the activities of both are equal both phases coexist in equilibrium and although matter exchange may occur between phases there is no net transfer. Thus, ice is formed from liquid water when the thermodynamic activity of the ice is less than that of the liquid water. In a system where solute is present, cooling a water-rich system will cause ice formation and concentration of the solute in the remaining liquid. This process will continue until the concentration of the solute is such that its solubility in the liquid phase is reached and it precipitates out. The system, which exists as a eutectic, consists of solid phases of ice and crystalline solute. However, in a complex multicomponent system such as hydrated Gelucire, equilibrium may not be reached. The formation of ice is a dehydration process for the Gelucire components. As ice is formed, the components become more concentrated and, if equilibrium were achieved, would precipitate out in a crystalline form. However, this is inhibited by the heterogeneous nature of the Gelucire and complete crystallization does not occur. Thus, further cooling results in a concentration of the Gelucire components and a decrease in the water activity. When the water activity in the liquid phase is the same as that in the ice, no more ice formation takes place.10 In general terms, then, it is straightforward to explain nonfreezing water without recourse to notions of bound water. However, other factors may complicate the situation. Where small pores or other high surface area to volume systems are involved, the surface free energy becomes a significant part of the total free energy of the phase in the pore. In this case, the thermodynamics of the pore water may be changed and the ice−water transition temperature affected.11 Another problem that arises in interpreting the phase changes that occur, are the kinetic effects of nucleation and inhibition of crystal growth. In DSC the scanning rates are well known to affect apparent phase transition temperatures. In addition, ice nucleation may be inhibited by the absorption of foreign molecules on to the surface of the growing crystals and by the lack of nuclei available in pores. These considerations are not important in melting point studies but could affect the amount of ice formed. The temperature of a phase transition from fluid to solid is not coupled to the molecular dynamics of the fluid phase. This



MATERIALS AND METHODS Materials. Stearoyl macrogol-32 glycerides (Gelucire 50/ 13) was kindly donated by Gattefossé S.A.S. (St. Priest, France). A range of Gelucire 50/13 gels with from 10% to 90% (w/w) water were prepared using two procedures: heating and hydration methods (see Table 1). In this work, ultrapure, type I Table 1. Freezeable water contents (WF,%) calculated for the gels prepared by heating and hydration methods (n=3)

gel samples (% water) 30 40 50 60 70 80 90

heating method (HEG)

hydration method (HYG)

2 °C/min

2 °C/min

Tp (°C) −11.9 −6.5 −5.2 −1.3 −1.1 −0.7

± ± ± ± ± ±

0.6 1.2 2.9 0.4 0.1 0.5

WF (%)

23 38 40 45

± ± ± ±

Tp (°C)

8.1 15.7 3.8 23.3

−16.23 −12.77 −8.63 −4.9 −1.67 −0.7 −1.09

± ± ± ± ± ± ±

5.06 3.85 0.69 0.03 0.21 0.30 1.19

WF (%)

23 45 61 81

± ± ± ±

2.0 1.6 5.0 1.8

water (Milli-Q grade, 18 MΩ·cm at 25 °C) produced by Barnstead Nanopure system (Thermo Scientific, U.K.) was used for all sample preparation. For the heating method, Gelucire 50/13 was melted using a hot plate at 62 ± 2 °C. The appropriate amount of water, heated to the same temperature, was added to the melted lipid. A hand-held disperser (UltraTurrax T10 basic, IKA, Germany) was used for agitating the mixture at 8000 rpm for 2−3 min. The gels prepared by the heating method are given abbreviation HEG in this paper. For the hydration method, room-temperature water was added to B

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 1. SEM images of the freeze-dried HEG gels with 10−90% water content (w/w).

where Mt is the magnetization at time t, M0 is the magnetization at time zero, Pi is the fractional population of the component i with a T2 of T2i, and analyzed using a 1D inverse Laplace transform in WINDXP. The inversion recovery pulse sequence used for the 2D inverse Laplace transform experiments had the inversion recovery step, [180° − t1]−, a sequence of 64 steps (between 100 μs to 10 ms) followed by a CPMG sequence with a τ1 value of 300 μs as described above. The expression for the resulting experimental outcome is

the molten lipid. The mixture was cooled to room temperature and the system was left to hydrate at room temperature for a week. Manual stirring was performed during this period to facilitate mixing of the components. The gels prepared by hydration method are given abbreviation of HYG. Differential Scanning Calorimetry (DSC). Conventional DSC was performed under nitrogen flux with purge rate of 50 mL/min. A full instrument calibration was performed prior to the sample runs. Aluminum hermetic pans (PerkinElmer, U.S.A.) were used for all samples and an average sample weight of 2−5 mg was used throughout the study. The gel samples were equilibrated at −50 °C for 15 min, followed by heating up to 25 °C at 2 °C/min. The heat of fusion of ultrapure water (Milli-Q grade) was experimentally determined using the same temperature program described above. All experiments were performed in triplicate. Relaxometry NMR. The gels prepared by the heating method (which were proven to be more homogeneous) were analyzed using a benchtop NMR (Maran Ultra, Resonance, Oxford Instruments) equipped with a DRX console operating at a frequency 23 MHz for protons. A 90°−180° pulse gap (τ = 300 μs) and a relaxation delay of 10s was applied. Onedimensional Carr−Purcell−Melboom−Gill (CPMG) sequences were performed with a pulse spacing of 300 μs and between 2038 and 16 348 echoes sampled, depending on decay length. Decays were assumed to exponential of the form ⎛ t ⎞ M t = M 0 ∑ Pi exp⎜ ⎟ ⎝ T2i ⎠ i

M t1, t2 = M∞





t ⎞⎤

∬ P(T1, T2)⎢⎢⎣1 − 2exp⎝⎜− T1 ⎟⎠⎥⎥⎦exp

⎛t ⎞ − ⎜ 2 ⎟dt1dt 2 ⎝ T2 ⎠

1

(2)

where M∞ is the equilibrium magnetization in the inversion recovery part of the experiment and P(T1,T2) describes the map of T1 versus T2, which is the required correlation graph obtained by the 2D inverse Laplace transform. The transform was accomplished using the method of Song et al.17 T1−T2 correlation spectra obtained using a fast two-dimensional Laplace inversion17 in a home produced script in MatLab (MathWorks, Massachusetts, U.S.A.). One-dimensional diffusion measurements were made using the stimulated echo method,18 field gradients were stepped from 0 to 2.406 Tm−1 in 20 steps and for each step and the time between the two field gradient pulses was from 50 to 500 ms.

(1) C

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 2. SEM images of freeze-dried HYG gels with 10−90% water content (w/w).

Imaging Analysis of Freeze-Dried Gels. Gels prepared by heating and hydration methods were freeze-dried using a benchtop freeze-drier (VirTis adVantage 2.0, SP Scientific, Gardiner, NY, U.S.A.). In the thermal treatment step, the shelf temperature was set at −40 °C and maintained at that temperature for 120 min to ensure complete freezing of the gels. The primary drying phase was carried out in sequential temperature steps of 120 min each, from −35 to 25 °C with 5 °C increments between each step at a chamber pressure of 200 mTorr. The secondary drying was set at 25 °C for 60 min. After freeze-drying, the gels were stored in a desiccator at 0% RH prior to examination with SEM. A Phillips XL20 SEM (Phillips Electron Optics, Netherlands) was used to image all the freezedried gel samples. The samples were sputter coated with Au/Pd prior to examination. For every sample, at least 10 SEM images were taken across the freeze-dried gel to evaluate the homogeneity of the samples. The pore size was estimated using imaging software by considering the diameter of the pore on the x and on the y axis. The average between the two axes was calculated. The diameter of 10−25 pores for each image was calculated.

in Figure 1. For the gels with lower water contents (10−30%), no uniform pores can be observed, but a layered structure can be seen. In the gel with 40% water content, both a layered structure and a porous network are visible. This pore size increases with increasing water content. As seen in Figure 1, the gels with 50% and 60% water content have more uniform pores with sizes of 6 ± 0.24 μm and 7 ± 0.14 μm, respectively. The pore size increases to 18 ± 1.41 μm for the gel with 70% water content. These results provide the first indication of the existence of a certain degree of internal structure within the Gelucire gels with moderate to high water content (40−90% w/w). The SEM results of the HYG are largely similar to the HEG, as seen in Figure 2. The gels between 10 and 30% water contents show a layered structure, whereas the gels with 40% water and above are characterized by a porous structure. However, for these gels with moderate and high water contents, the pore sizes of the freeze-dried HYG samples are significantly smaller than those of HEG. The HYG with 80% and 90% water content show a more structured porous interior with some intact plates in comparison to the HEG with the same water content. This observation indicates the effect of the preparation method on the interior structures of the gels. Although it is possible that the pores may be formed through expansion of ice crystals during the freeze-drying process, the porous structure and the increase in pore size with water content are still likely to be a valid indication of the presence of water-rich (which are the areas formed the pores after freeze-drying) and Gelucirerich domains. Determination of Freezable Water Content in the Gels Using DSC. DSC was used as a fast screening method to



RESULTS AND DISCUSSION Interior Microstructure of Freeze-Dried Gels. Visualization of the gels prepared by the heating and hydration methods allows assessment of the effect of preparation method on the gels interior microstructure. The samples were freezedried and examined by SEM. The freezing temperature (−40 °C) was selected, as below this temperature no phase transitions were observed as determined by DSC (see below). The SEM images of the freeze-dried HEG are shown D

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 3. DSC results of the HEG (top) and HYG (bottom) gels determined at the scanning rate of 2 °C/min.

where WF is the freezeable water concentration (% w/w), ΔHF(exp) is the measured enthalpy value of the endothermic peak at 0 °C, ΔHF is the experimentally measured heat of fusion of pure water (346.5 ± 3.9 J/g at the heating rate of 2 °C/min). For the HEG gels, no endothermic transition was found in the gels with 10%, 20%, and 30% water content (data not shown) when the gels were reheated from subzero (−50 °C) freezing. The absence of an endothermic peak indicates that none of the contained water was able to crystallize, possibly because of entrapment of water molecules in the dense gel network. The gels with 40% water show an asymmetric

quantitatively estimate the amounts of freezeable water in the gels. DSC is a classic method reported in literature for such estimations, in particular for surfactant/water systems.9 The quantities of freezeable water in the HEG and HYG are expected to change with total water content in the gels. Assuming that the endothermic melting peak at 0 °C ± 5 °C is the melting of frozen free water, it is possible to quantitatively estimate the content of freezeable water (w/w%) in the gels using the following equation19 WF =

ΔHF(exp) × 100 ΔHF

(eq 3) E

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics endothermic peak with an onset temperature of −23.3 °C and a peak temperature of −13 °C (Figure 3a). The peak temperature of this endothermic transition further shifted to higher temperatures with increasing the water content in the gels (−7 °C and −5 °C for the HEG gel with 50% and 60% water, respectively), but little change can be observed in the onset temperature. An almost shoulder-like peak at a temperature below the main transition can be seen in all HEG gels with 40−90% water content. The presence of the shoulder and main transition indicates the coexistence of at least two different populations of water molecules in the frozen system. Although it must be a possibility that freezing causes some structural change in the gels, if we assume that the observed behavior represents the system in its unfrozen state, then the problem is to assign the peaks to water in different environments. As is pointed out, melting implies that water formed by the ice at the melting temperature has slightly lower thermodynamic activity than the ice at the same temperature. The multiple endothermic transitions observed by DSC may be an indication of differences in the space confinement of the water in the gels with 40% water and 50% or 60% water, which may be a reflection of different interior microarchitecture, such as layered planes, pores, and channels, as was observed using SEM for the gels with different water contents. Alternatively this may arise from the water being preferentially absorbed by the water-soluble components, such as free PEG, and forming a concentrated solution with a lowered water activity and thus freezing point. This hypothesis is supported by similar reported observations in aqueous PEG solutions with a wide range of concentrations (2−50%).20,21 The subzero endothermic transitions of frozen PEG 1500 (same molecular weight as the free PEG in Gelucire) solutions have been reported to occur between −10 °C to −17 °C in various literature reports.20,22,23 The HEG gels with 70−90% water contents show a clear main endothermic peak between −2 °C and 0 °C (Figure 3a). This is consistent with the presence of water with an activity close to that of bulk water, indicating the presence of pockets of freezeable water in these gels. For the HYG, the DSC results were largely similar to those obtained for the HEG (Figure 3b). However, in contrast to the HEG gel with 30% water (no melting endotherm), the HYG with 30% water showed a broad endothermic asymmetric peak with an onset at −22 °C and peak a temperature of −13 °C. This clear melting peak in the HYG gel with 30% water content may indicate that the more effective mixing induced by the heating preparation method leads to a higher homogeneity of the HEG than the HYG gels. The content of free freezeable water (WF) (with main peak temperature at 0 ± 5 °C) in the HEG and HYG gels is shown in Figure 4. Overall, free freezeable water was only present in the gels with water contents at and above 55% and increases in quantity with increasing the water content of the gels. However, the actual amount of free water content in the gels (corresponding to the enthalpy values of the peaks) is affected by the method of preparation. It can be seen that the free water contents in the HYG gels are higher than those of the HEG gels. As discussed earlier, with the heating method, there is a more homogeneous mixing of the aqueous and oleic phases at the melting temperature of Gelucire, which promotes a higher level of water−Gelucire interaction and possibly more intimate incorporation of water molecules in the nanostructure of the gels in comparison to the HEG gels.

Figure 4. Freezable water as a percentage of the total water content (WF) calculated using the DSC results with scanning rate of 2 °C/min.

DSC studies identified three types of endothermic events during the heating of frozen Gelucire gels. The number and type of transitions for each gel is highly water content dependent. The gels with high water content (60−90%) show large ice melting transitions close to 0 °C, which is associated with free freezeable water. Taking into account the SEM results, it is reasonable to suggest the presence of relatively large pockets of water in these gels (gels with 60− 90% water contents). The second type of transition between −15 °C to −10 °C, with low enthalpy values is similar to the eutectic melting transition observed in PEG solutions reported in literature, indicating the dissolution of free PEG in the Gelucire in water, possibly in micron to nanoscale phase separated domains. The lowest temperature phase transitions between −17 °C and −25 °C are similar to the subzero phase transitions observed in surfactant−water systems. These are often associated with the thermal behavior of so-called “interfacial water”, which is the water intimately bound at the interface between bulk free water and surfactant assemblies.9,19,24 These water molecules may have much deeper penetration and more intimate interactions with the components in Gelucire. However, this is highly speculative, as no study has been able to firmly prove this theory. Nevertheless, the DSC results provide valuable indications of the microstructure of the gels and the effect of preparation methods on this structure. These indicative hypotheses can be further confirmed by probing the mobility of water molecules in the gels using NMR relaxometry, as local microenvironments and Gelucire−water interactions can affect the motions of molecules. Molecular Understanding of Water Distribution in the Gels Studied by Relaxometry NMR. One-Dimensional (1D) Transverse Relaxation. Further identification of the different types of water molecules present in the Gelucire gels was carried out using relaxometry NMR. Typical 1D Laplace transforms of the transverse magnetization of selected gels are shown in Figure 5. The observed peaks are related to the detectable components. It can be seen that for the gels with 10−90% water content, the number of detectable components varies between 2 and 4. However, the identification of 2, 3, or 4 components does not seem to follow a straightforward pattern (Figure 6a). It should be recognized that the software offers the most parsimonious solution consistent with data that may or may not represent physical reality. In order to simplify F

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 5. Plots of 1D Laplace transforms of relaxation times for 10%, 30%, 70%, and 90% water content gels.

Figure 6. (a) Transverse relaxation times calculated from 1D Laplace transforms. Up to four (T1−4) components were resolved; (b) weighted transverse relaxation times calculated as described in the text; (c) populations of the fast and slow components in the samples. Triangles are the fractional proton population of the water content of the samples.

interpretation, therefore, relaxation times have been combined into two groups: Ts and Tf, where “s” is the slowest relaxation time and “f” is the population weighted average of all the faster relaxation times. P s,f are the corresponding fractional populations. Figure 6b shows that Ts has a strong dependence on water content but its relative population does not reflect the actual

water content and, with the exception of the data of gels with 10% and 90% water content, is consistently higher than the expected value if it were due entirely to water. The data of the gel with 10% water is anomalous in that the relative populations suggest that in this case the “f” and “s” data should be transposed. In general, then it must be concluded that the slow G

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics

Figure 7. 2D T1−T2 relaxation maps of the gels with 10% to 90% water contents. The dashed rectangles identify the reliable values set by the pulse sequence. The diagonal lines indicates the condition where T1 = T2.

Figure 7. It is noted that the inverse Laplace transform is numerically ill-conditioned and that interpretation must be conservative.25 However, from the 2D data some significant general features can be discerned for the gels at all water

relaxing signal cannot be due to water alone but must contain some magnetization from aliphatic protons. Two-Dimensional (2D) T1−T2 Correlation. 2D Laplace transforms for the inversion recovery experiment are shown in H

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics contents. The presence of peaks in the nonphysical region where T2 is greater than T1 (the boundary between the physical and nonphysical region is marked by the diagonal line in the diagrams in Figure 7) is clear evidence of exchange proton processes between various environments.26 At the lowest water contents (10% to 40%), the maxima are well separated and there is some evidence of fast relaxing peaks on the edge of the reliable detection zone. These could be the result of fast exchange, which tends to increase apparent relaxation rates, or components in which there is little mobility and are therefore fast relaxing. In the 10% gel, the peaks are well separated and discrete. Mechanisms of exchange in this system may be both by translational diffusion of water and other components and by spin diffusion due to slow motion in the aliphatic chains of the glycerides and glycols in Gelucire. Increasing the water content to 30 and 40% results in a smearing out of the peaks and may indicate the formation of number of different microenvironments with similar properties in which the water was entrapped. The occurrence of stronger peaks at the edge of the reliable detection zone may indicate an increase in exchange rates. These observations are consistent with the SEM images in which there are transitions in the 30−40% water content region from a clear layered structure to a more porous structure. At 40% water content, it is also the point at which ice formation can be observed in the system. These water contents thus represent a transitional region in which there are a wide variety of microenvironments. At 50% water, the peaks begin to separate and fast relaxing peaks diminish in intensity. Above 50% water, the maps show a rich pattern of discrete peaks that vary in number and position. This is due to both changes in intrinsic relaxation times and to changes in exchange rates. In the 60% water content gel, the distance and relative intensities of peaks 1 and 4 suggests that the exchange rate is close to the fast exchange limit27 and this may explain the appearance and disappearance of a peak in the position of peak 4 in the nonphysical zone in the 70%, 80%, and 90% water content gels as being close to the filter cut off for noise. In samples with water content above 60%, peak 1 persists close to the T1 = T2 line and it is tempting to assign this peak to bulk water; however, comparison of the transverse relaxation times of these peaks with those of T2f in the 1D experiments shows them to be very close in value and following the same trend with water content (Figure 8a). Because the intensities of these peaks is higher than that calculated for water alone they must be peaks associated with mixtures of water and other components from Gelucire. The increases observed in relaxation times are mirrored in the increasing pore size in the materials observed by SEM and an increase in the melting temperature of the ice formed on cooling detected by DSC. Diffusion Measurements. In order to determine the diffusion coefficient of water in the gels, one-dimensional diffusion experiments were conducted. The probability of spin displacing a distance R in time t is related to the attenuation of the signal (E(R, Δ)) by P(R , Δ) =

∫ E(R , Δ)e−iq·Rdq

Figure 8. (a) Variation in T2 as observed in the 1D (T2f) and 2D (P1) inverse Laplace transform experiments; (b) variation of the average value of the fast diffusion coefficients for q values where single exponential attenuation curves were observed; (c) variation of the slow diffusion coefficient with concentration and wave vector.

(4)

However, the gels contain seven components, all of which may diffuse. Because only one attenuation can be measured, the problem of inverting the equation to find a set of (E(R, Δ)) that correspond to the set of Pi(R, Δ) of the components is extremely challenging. One approach (which was used in this study) is to expand the attenuation, normalized to the

attenuation with no field gradient applied, as a multiple exponential time series at different values of q.28 Thus I

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics E(q , Δ) = E(0, Δ)

have been made with silica colloids.30 Alternatively dissolution of free polyethylene glycol or lipid micelles may result in an obstruction effect,31 which slows diffusion but has obstructions on length scale too short to be approached by the wave vectors accessible in our apparatus. Dilution would result in less obstruction and faster diffusion. The variation diffusion coefficient determined from the second component of the attenuation curve is shown in Figure 8c. The values at lower q have a large estimated error but converge to a consistent value at the highest q. The exception to this is the 90% water system, which shows consistently higher D values than the rest. The second component of diffusion could arise from restricted diffusion of water or from diffusion of one or more components of Gelucire. If it is from one of the Gelucire components then it is difficult to see why the diffusion coefficient would be dependent on the wave vector although it might be dependent on the amount of water present. A wave vector dependence of an apparent diffusion coefficient is what is expected of a restricted diffusion system. At the highest wave vector value, the values of D converge, indicating that gels of all water contents, with the exception of the 90% sample, have at least one restriction length in common. The exceptional behavior of the 90% sample might be explained by the observation in the SEM experiments that for the HEG samples there was a complete breakdown of structure in some regions of the materials at this water content. This would create large voids in which water could diffuse unrestricted and elsewhere larger distances between the barriers for water diffusion. Combining the results obtained from SEM, DSC, and NMR relaxometry, the change in the degree of order of the microstructure of the gels with changing water content in the gels can be proposed by Figure 9.

2 2

∑ Pi(q)e−(4π q D Δ) i

0, i

(5)

In general, this equation will reflect the different diffusion coefficients of the multiple components and possible restrictions to diffusion of those components. Varying the wave vector q varies the length scale over which structure can be determined by restriction of diffusion. Small values of q will tend to smooth out local variations in structure on the length scale of 1/q. The smallest value of q used here which gave reproducible results in the gels with 40−90% water was 1.04 × 104 m−1 corresponding to length scale of 100 μm. Diffusion was measured for a series of values of Δ from 54 to 240 ms. Under these conditions single exponential signal attenuation was observed. This implies that any structure that affects the apparent diffusion coefficients was smaller than the length scale determined by the q values (100 μm). This result correlates well with the microarchitecture observed by SEM in the freezedried gels. As seen in Figure 8b, the low q data are close to linear in the water content expressed as mole fraction of water protons (Pw). The value of D (2.0 (±0.1) ×10−9 m1 s−1) in the gel with 90% water is close to that of pure water with a diffusion coefficient of 2.3 × 10−9 m2 s−1.29 Increasing q resulted in the observation of multiple exponential decays. As the number of data points in the attenuation curves is small, attempts to analyze the data using eq 5 in terms of more than two exponents or 1dimensional inverse Laplace transforms would be not be justified. The fitting process consisted of first of all inspecting a logarithmic plot of attenuation versus Δ at different q values curves, which deviated from linearity and were fitted to a two exponential process using a Lever Marquand algorithm in the software package (Tablecurve). For the gels with 80 and 90% water content, single exponential behavior was observed with q values up to 1.57 × 104 m−1. For the remainder, at q values of 2.2 × 104 m−1, greater double-exponential behavior was observed. This approach resulted in good fits with r2 values greater than 0.99. The variation of D with water content and its value being close to the value for pure water in the 90% gel suggest that the fast-diffusing entity is water. No other substance present is likely to show such rapid diffusion. The variation of D with water content could be due to exchange with slower moving molecules. However, the scarcity of exchangeable protons in the rest of the system makes this unlikely. One model for explaining the gel structure could be based on a situation in which the water partitions between a slow-moving fraction associated with the nonaqueous phase and bulk water. Depending on the assumptions made this can lead to models which are linear or nonlinear in water content; however, attempts to fit the data to simple models in which partition coefficients are a constant of water content or Gelucire content failed. A variable partition coefficient model could be made to fit the data, but without supporting evidence this amounts to a parametrization of the data and cannot be considered as being of physical significance. A possible explanation may lie outside a simple exchange model: as water content above 40% increases, the pore size increases which was demonstrated by the SEM data. This implies freer diffusion and probably less tortuosity or fewer obstructions to diffusion. Increasing water content implies a reciprocally decreasing Gelucire concentration; this in turn reduces the obstructions to diffusion and results in the apparent increases in diffusion coefficient. Similar observations

Figure 9. Illustration of the water distribution in nanostructured Gelucire 50/13 gels. J

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics



(5) Xia, L. W.; Xie, R.; Ju, X.-J.; Wang, W.; Chen, Q.; Chu, L.-Y. Nano-structured smart hydrogels with rapid response and high elasticity. Nat. Commun. 2013, 4, 2226. (6) Pan, L.; Yu, G.; Zhai, D.; Lee, H.; Zhao, W.; Liu, N.; Wang, H.; Tee, B.; Shi, Y.; Cui, Y.; Bao, Z. Hierarchical nanostructured conducting polymer hydrogel with high electrochemical activity. Proc. Natl. Acad. Sci. U.S.A. 2012, 109, 9287−9292. (7) Brubach, J. B.; Ollivon, M.; Mahler, J. B.; Bourgaux, C.; Lesieur, P.; Roy, P. Structural and Thermal Characterization of Mono- and Diacyl Polyoxyethylene Glycol by Infrared Spectroscopy and X-ray Diffraction Coupled to Differential Calorimetry. J. Phys. Chem. B 2004, 108, 17721−17729. (8) Qi, S.; Marchaud, D.; Craig, D. Q. M. An investigation into the mechanism of dissolution rate enhancement of poorly water-soluble drugs from spray chilled gelucire 50/13 microspheres. J. Pharm. Sci. 2010, 99, 262−274. (9) Cohen, Y.; Avram, L.; Frish, L. Diffusion NMR Spectroscopy in Supramolecular and Combinatorial Chemistry: An Old Parameter New Insights. Angew. Chem., Int. Ed. 2005, 44, 520−554. (10) Belton, P. S. NMR and the Mobility of Water in Polysaccharides. Int. J. Biol. Macromol. 1997, 21, 81−88. (11) Jaaehnert, S.; Vaca Chavez, F.; Schaumann, G. E.; Schreiber, G. E.; Schoenhoff, M.; Findenegg, G. H. Melting and freezing of water in cylindrical silica nanopores. Phys. Chem. Chem. Phys. 2008, 10, 6039− 6051. (12) Xu, F.; Leclerc, S.; Canet, D. NMR Relaxometry Study of the Interaction of Water with a Nafion Membrane under Acid, Sodium, and Potassium Forms. Evidence of Two Types of Bound Water. J. Phys. Chem. B 2013, 117, 6534−6540. (13) Hills, B. P.; Tang, H.-R.; Manoj, P.; Destruel, C. NMR Diffusometry of Oil-in-Water Emulsions. Magn. Reson. Imaging 2001, 19, 449−451. (14) Belton, P. S. Spectroscopic Approaches to the Understanding of Water in Foods. Food Rev. Int. 2011, 27, 170−191. (15) Niisson, P. G.; Lindman, B. Water Self-Diff usion in Nonionic Surfactant Solutions. Hydration and Obstruction Effects. J. Phys. Chem. 1983, 87, 4756−4761. (16) Gugring, P.; Lindman, B. Droplet and Bicontinuous Structures in Microemulsions from Multicomponent Self-Diffusion Measurements. Langmuir 1985, 1, 464−468. (17) Song, Y. Q.; Venkataramanan, L.; Hürlimann, M. D.; Flaum, M.; Frulla, P.; Straley, C. T1−T2 Correlation Spectra Obtained Using a Fast Two-Dimensional Laplace Inversion. J. Magn. Reson. 2002, 154, 261−268. (18) Ezrahi, S.; Aserin, A.; Fanun, M.; Garti, N. Sub-zero temperature behavior of water in microemulsions. In Thermal Behavior of Dispersed Systems; Garti, N, Ed.; Marcel Dekker, Inc.: New York, 2001; pp 59− 120. (19) Schulz, P. DSC analysis of the state of water in surfactant-based microstructures. J. Therm. Anal. Calorim. 1998, 51, 135−149. (20) Yamauchi, T.; Hasegawa, A. Determination of PEG concentration in its aqueous solution using differential scanning calorimetry. J. Appl. Polym. Sci. 1993, 49, 1653−1658. (21) Bhatnagar, B. S.; Martin, S. M.; Teagarden, D. L.; Shalaev, E. Y.; Suryanarayanan, R. Investigation of PEG crystallization in frozen PEGsucrose-water solutions: II. Characterization of the equilibrium behavior during freeze-thawing. J. Pharm. Sci. 2010, 99, 4510−24. (22) Bhatnagar, B. S.; Martin, S. M.; Teagarden, D. L.; Shalaev, E. Y.; Suryanarayanan, R. Investigation of PEG crystallization in frozen PEGsucrose-water solutions. I. Characterization of the nonequilibrium behavior during freeze-thawing. J. Pharm. Sci. 2010, 99, 2609−19. (23) Hager, S. L.; Macrury, T. B. Investigation of phase behavior and water binding in poly(alkylene oxide) solutions. J. Appl. Polym. Sci. 1980, 25, 1559−1571. (24) Garti, N.; Aserin, A.; Ezrahi, S.; Tiunova, I.; Berkovic, G. Water Behavior in Nonionic Surfactant Systems I: Subzero Temperature Behavior of Water in Nonionic Microemulsions Studied by DSC. J. Colloid Interface Sci. 1996, 178, 60−68.

CONCLUSIONS This study investigated the complex micro- and nanoarchitecture of Gelucire gels interior, which are produced from by the interaction of water with a complex mixture of seven different species of glycerides, PEG, and PEG-esters. The analytical approach taken in this study probes the thermal and diffusional behavior of water molecules within the microenvironments of the gels. The DSC and relaxometry NMR results indicated that the water molecules are in different microenvironments that have different degrees of confinement. The structure evolves with the water content in the gels. This indicates the presence of a complex interior structure that correlates well with the observed internal network of the gel observed after freeze-drying by SEM. The results of this study provide new insights into structure of Gelucire gel, which have not been reported before because of the complexity of the material. It also demonstrates that the combination of thermal methods and relaxometry NMR can be a powerful tool for investigating complex soft materials, including biological tissues and food products as well as pharmaceuticals. In our investigation, we have taken a view based on thermodynamic and molecular dynamic considerations. We have argued that changes in temperature of ice formation cannot be ascribed to notions of restriction of mobility of water and that NMR results need careful and conservative interpretation based on an understanding of the underlying physics and mathematical transformations employed. Structural information such as that obtained by this study can be used during formulation development to optimize gel formulations. For example, ensuring the selection of gels with an optimal network structure and that drug incorporation occurs in the appropriate subphase of the gel, maximizing the in vivo drug release and absorption performance as well as the physical stability of the formulations.



AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Notes

The authors declare no competing financial interest.



ACKNOWLEDGMENTS



REFERENCES

Doroty Codoni would like to thank the University of East Anglia for the financial support for the period of her Ph.D. The authors would like to sincerely thank Drs. Brian Hills and Ben Piggott for their invaluable effort on NMR training and discussions on preliminary data.

(1) Hoffman, A. S. Hydrogels for biomedical applications. Adv. Drug Delivery Rev. 2002, 54, 3−12. (2) Li, Y.; Rodrigues, J.; Tomas, H. Injectable and biodegradable hydrogels: gelation, biodegradation and biomedical applications. Chem. Soc. Rev. 2012, 41, 2193−2221. (3) Sek, L.; Boyd, B. J.; Charman, W. N.; Porter, C. J. Examination of the impact of a range of Pluronic surfactants on the in-vitro solubilisation behaviour and oral bioavailability of lipidic formulations of atovaquone. J. Pharm. Pharmacol. 2006, 58, 809−20. (4) Gao, P.; Morozowich, W. Development of supersaturatable selfemulsifying drug delivery system formulations for improving the oral absorption of poorly soluble drugs. Expert Opin. Drug Delivery 2006, 3, 97−110. K

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX

Article

Molecular Pharmaceutics (25) Bernin, D.; Topgaard, D. NMR diffusion and relaxation correlation methods: New insights inheterogeneous materials. Curr. Opin. Colloid Interface Sci. 2013, 18, 166−172. (26) Marigheto, N.; Venturi, L.; Hibberd, D.; Wright, K. M.; Ferrante, G.; Hills, B. P. Methods for peak assignment in lowresolution multidimensional NMR cross-correlation relaxometry. J. Magn. Reson. 2007, 187, 327−342. (27) Hills, B. P.; Benamira, S.; Marigheto, N.; Wright, K. T1-T2 correlation analysis of complex foods. Appl. Magn. Reson. 2004, 26, 543−560. (28) Hills, B. P.; Tang, H. R.; Manoj, P.; Destruel, C. NMR Diffusometry in Oil Water Emulsions. Magn. Reson. Imaging 2001, 19, 449−451. (29) Holz, M.; Heil, S. R.; Sacco, A. Temperature-dependent selfdiffusion coefficients of water and six selected molecular liquids for calibration in accurate 1H NMR PFG measurements. Phys. Chem. Chem. Phys. 2000, 2, 4740−4742. (30) Rogers, B. J.; Wirth, M. J. Obstructed Diffusion in Silica Colloidal Crystals. J. Phys. Chem. A 2013, 117 (29), 6244−6249. (31) Waters, D. J.; Frank, C. W. Hindered diffusion of oligosaccharides in high strength poly(ethylene glycol)/poly(acrylic acid) interpenetrating network hydrogels: Hydrodynamic vs. obstruction models. Polymer 2009, 50, 6331−6339.

L

DOI: 10.1021/mp5008508 Mol. Pharmaceutics XXXX, XXX, XXX−XXX