Non-lamellar lipid assembly at interfaces: controlling layer structure by responsive nanogel particles

Abstract

Biological membranes do not only occur as planar bilayer structures, but depending on the lipid composition, can also curve into intriguing three-dimensional structures. In order to fully understand the biological implications as well as to reveal the full potential for applications, e.g. for drug delivery and other biomedical devices, of such structures, well-defined model systems are required. Here, we discuss the formation of lipid non-lamellar liquid crystalline (LC) surface layers spin-coated from the constituting lipids followed by hydration of the lipid layer. We demonstrate that hybrid lipid polymer films can be formed with different properties compared with the neat lipid LC layers. The nanostructure and morphologies of the lipid films formed reflect those in the bulk. Most notably, mixed lipid layers, which are composed of glycerol monooleate and diglycerol monooleate with poly(N-isopropylacrylamide) nanogels, can form films of reverse cubic phases that are capable of responding to temperature stimulus. Owing to the presence of the nanogel particles, changing the temperature not only regulates the hydration of the cubic phase lipid films, but also the lateral organization of the lipid domains within the lipid self-assembled film. This opens up the possibility for new nanostructured materials based on lipid–polymer responsive layers.

1. Background

Biological membranes do not only form planar bilayers but can also curve into structures such as those that occur in the endoplasmic reticulum, inner mitochondrial membrane and bacterial membranes [1–3]. Highly organized nano-architectures with three-dimensional periodicity (so-called cubic membranes) have been identified in many biological systems [1,4]. However, knowledge about the formation and function of these highly organized three-dimensional periodic membrane structures is lacking, both due to the shortcomings of experimental techniques as well as the transitional nature of some of these structures. Deng et al. [5] have shown that the mitochondria membrane of the amoeba Chaos, when exposed to starvation conditions, undergoes a reversible morphological transition from random tubular into cubic morphology that is linked to changes in the lipid composition. The formation of local and temporal curved bilayer structures as intermediates in the process of fusion and fission of lipid bilayers is another topological membrane transformation that is not fully understood [4,6]. In particular, it is not known what triggers such transitions. However, it is known that these local changes in membrane curvature and membrane topology, due to factors like local variations in lipid composition, have significant effects on the activity of membrane-associated proteins [7], especially curvature-sensitive proteins [8].

All naturally occurring cubic liquid crystalline (LC) phases have unit cell dimensions that are one order of magnitude larger than in simple lipid systems. Tyler et al. [7] have shown that it is possible to swell a monoolein-based bicontinuous Im3m cubic phase to five times its original size by (i) increasing the bilayer stiffness with cholesterol and (ii) inducing electrostatic repulsion between the opposite curved bilayers with anionic lipids. The ease with which such LC phases can be altered and tuned makes them promising candidates for new controlled materials. While the formation of supported lipid bilayers or lamellar structures is much studied, the non-lamellar lipid-based surface layers are not. In fact, very few studies have been dedicated to the formation of non-lamellar self-assembled structures at interfaces, despite the biological relevance and numerous potential applications of such structures. This study focuses on non-lamellar lipid-based structures on surfaces.

One approach to obtain surface layers with nanostructure beyond flat bilayers that mimic the curvature of membranes observed in nature is by using a nanostructured template as a substrate. Recently, we showed that fluid supported bilayers can be formed on vertical nanowire forests from vesicles in solution [9]. The bilayers follow the contours of the nanowires to form continuous and locally highly curved model membranes to which proteins and vesicles can be attached. The nanowire supported bilayer approach allows studies of the specific effect of curvature when a protein is confined to a curved lipid surface. However, the templated bilayer provides only a single bilayer and some biological processes involve confinement within the aqueous cavity or lipid domains of an LC phase. This includes low molecular weight molecules with pharmaceutical activity which can be incorporated in lipid-based cubic LC phases and dispersions [10]. Certain proteins can be entrapped within non-lamellar, e.g. bicontinuous cubic, lipid phases aqueous domains [11–13] and if they are enzymes, they retain their activity as determined by electrochemical methods [11,12]. Thus, the possibility to fabricate layers of non-lamellar phase-forming lipids provides means to form bioactive LC surfaces. The proteins have dimensions of about the same size as the water channels of the bicontinuous cubic phase, but size depends on the lipid system and can be regulated by different additives [14]. For example, by introducing hydration-enhancing sugar esters into the lipid matrix, the expansion of the cubic phase water channels can be regulated in such a way that the confinement of an entrapped enzyme can be relaxed, as has been demonstrated by the increase in the enzymatic activity of entrapped horseradish peroxidase [12].

Another approach to obtain lipid non-lamellar LC surface layers is to deposit the lipid/lipid mixture from a solution using spin-coating, followed by hydration in the aqueous solution [15,16]. This approach ensures that the LC structure expected from the phase diagram is maintained, and hence the aqueous cavities of the LC phase are intact [15,17]. The structure and dynamics of films formed by mixtures of soy phosphatidylcholine and glycerol dioleate spin-coated onto silicon substrates and hydrated in D₂O were studied by specular and off-specular neutron reflectometry and grazing incidence neutron spin echo spectroscopy [17]. Both reverse hexagonal (H_II) and micellar cubic phase (Fd3m) gave rigid films, but the H_II film was more flexible, appearing as a modified undulation spectrum of cylinders due to the interaction with the substrate.

This study is prompted by an interest in making such layers responsive to external triggers, like temperature, to an extent that goes beyond what can be achieved by a pure lipid–aqueous LC phase. Here, we chose poly(N-isopropylacrylamide) (PNIPAM) nanogels, because it is a polymer with well-known and well-defined thermoresponsive behaviour, i.e. a reversible lower critical solution temperature (LCST) transition where it becomes dehydrated and thus decreases in volume [18,19]. This transition also occurs in a physiologically relevant temperature range around 30°C. We were also able to synthesize the nanogels with a diameter of about 44 ± 1 nm (figure 1), which is about 10 times larger than a typical enzyme. This size is the smallest possible for such particles, in order to minimize the impact on the lipid nanostructure, i.e. the cubic phase structure, without losing the capability to undergo a volume phase transition. Our idea was to be able to design a system where the polymer particles are located in the boundaries between different LC domains, where it could affect the hydration and potentially other structural properties without protruding into the aqueous channels of cubic lipid LC phase. This also makes it possible to include biofunctional molecules, like proteins or peptides in the aqueous channel. The lipid mixture was chosen because the dimensions of the water channels in the glycerol monooleate (GMO) and diglycerol monooleate (DGMO) cubic phase can be increased by inclusion of DGMO [20]. The ease with which such LC phases can be altered and tuned makes them promising candidates for new controlled materials. By combining the well-known controlled release properties of the nanostructured lipid cubic phase with the responsiveness of the PNIPAM nanogel, we aim to develop new hybrid structures for biomedical applications. Owing to the temperature sensitivity of PNIPAM, the nanogels act as thermoresponsive controllers for the extension of the lipid films.

Figure 1.

Hydrodynamic radius of the nanogel as a function of temperature. Three measurements were performed for each temperature at a scattering angle, θ/2, of 90°. From the obtained diffusion coefficient, a hydrodynamic radius, R_h, was calculated as described in the text.

We recently showed that such spherical nanogels can be embedded within films of lipid non-lamellar LC films to more than 10% of the lipid component [21]. In this study, we have used specular and off-specular neutron reflectometry in combination with SAXS, confocal microscopy and rheology to reveal the nanostructural and morphological changes that take place in the lipid–nanogel hybrid layer. Knowledge of the temperature response of nanostructured lipid-based surface films will provide new leads for controlled delivery in food, cosmetics, consumer products as well as pharmaceuticals and bioanalytical systems.

2. Material and methods

2.1. Preparation of nanogels

Precipitation polymerization in the presence of the surfactant sodium dodecyl sulfate (SDS) was used to synthesize the PNIPAM nanogels. Briefly, SDS (0.193 g) was dissolved in water (96 g) in a three-necked round-bottomed flask. Re-crystallized N-isopropylacrylamide (1.48 g) was added to the reaction mixture together with the cross-linker N,N′-methylenbis(acrylamide) (0.065 g). The reaction mixture was bubbled with argon for 30 min, after which time, the temperature was increased to 70°C. To initiate the polymerization, potassium persulfate (0.055 g) dissolved in water (2 g) was added to the reaction mixture. The reaction was left to proceed for 6 h (under a constant argon atmosphere and at a constant stirring rate of 250 rpm). The solution was left to cool overnight under constant stirring. The cooled solution was then filtered and dialysed, with the dialysis water replaced once a day for two weeks until the conductivity was found to be below 1 μS cm⁻¹.

2.2. Dynamic light scattering

Dynamic light scattering (DLS) measurements were conducted to determine the hydrodynamic radius of the particles versus temperature using a three-dimensional LS Spectrometer from LS Instruments (Switzerland) operating at a wavelength, λ = 660 nm. Three measurements were performed for each temperature at a scattering angle, θ/2, of 90° and the results were evaluated using a first-order cumulant analysis of the initial part of the correlation function. The obtained decay constant is Γ(q), where q is the magnitude of the scattering vector, q = 2(πn/λ)sin(θ/2), where n is the refractive index of solution. From the Γ(q), the diffusion coefficient can be obtained from Γ = Dq² and the obtained value of D was used to determine the hydrodynamic radius, R_h, from the Stokes–Einstein equation, D = kT/(6πηR_h), where η is the viscosity of the solution.

2.3. Preparation of lipid formulations

To prepare the bulk lipid phases, stock solutions were prepared from mixtures of commercial samples of glycerol monooleate (GMO-50) and diglycerolmonooleate (DGMO), i.e. GMO-50/DGMO, at 40/60 wt ratio by mixing 85 wt% of the appropriate amounts of lipids with 15 wt% ethanol. The mixture was sealed and mixed on a roller mixer for 24 h. Nanogel particles were added to the stock solution at concentrations of 1, 5, 7.1, 10 or 20 wt%. Samples were mixed for a further 24 h on a roller mixer. The vials with the samples were then centrifuged alternating in up-side-up and up-side-down until they appeared homogeneous by visual inspection under polarized light. In order to form the bulk phases in excess water, small amounts of the formulation were added to MilliQ water (less than 95 wt%) followed by roller mixing for 7 days at room temperature prior to measurements.

2.4. Small angle X-ray scattering

Lipid mixtures with and without PNIPAM nanogel in ethanol were investigated on the Ganesha 300XL (SAXSLAB ApS, Skovlunde, Denmark). The instrument has a Genix 3D X-ray source, with a λ = 1.54 Å⁻¹ and a two-dimensional 300 K Pilatus detector from Dectris. Samples were exposed for 45 min at seven different temperatures (20, 25, 27.5, 30, 32.5, 35 and 40°C) controlled by an external recirculating water bath. They were sealed between two thin mica sheets in a metallic block to prevent ethanol evaporation. The lack of vaporization was confirmed by weighing the sample holders before and after analysis. I(q) was obtained by radially average the two-dimensional scattering pattern using the SAXSGui software.

LC phase formulations in excess water, with and without PNIPAM nanogels, were also characterized by small angle X-ray scattering (SAXS). Hydrated samples in excess water were confined between Kapton windows in a steel sample holder and analysed using the I911–4 SAXS beamline at the MAXIV Laboratory (Lund University, Sweden). Measurements were performed at a range of temperatures (25, 27.5, 30, 32.5, 35, 37.5 and 40°C) with an exposure time of 60 s. The sample-to-detector distance (1919 mm) and detector positions were calibrated with silver behenate. A wavelength of 0.91 Å⁻¹ and a Pilatus 1 M two-dimensional detector were used. Fit2D software (http://www.esrf.eu/computing/scientific/FIT2D/) was employed to analyse the two-dimensional SAXS profiles.

2.5. Spin-coating of lipid films

Prior to spin-coating, all formulations were diluted in ethanol to a lipid concentration of 20 wt%. The dilute formulations were dropped onto the silica surfaces (1 ml for 25 cm² NR samples) and these spread to cover almost the whole surface. Immediately after spreading, spin-coating was performed on an SCS 6800 (Speciality Coating Systems, USA) using the following programme: 10 s at 700 r.p.m. followed by 60 s at 2000 r.p.m. The spin-coating resulted in a uniformly spread, visually glossy and smooth lipid layer on the surface and most of the ethanol is expected to evaporate. Measurements by spectroscopic ellipsometer revealed that the film was about 1.9 µm thick (electronic supplementary material, figure S3). After spin-coating, the substrate was immediately immersed in excess water, then sealed into a flow-cell type sample holder. In order to remove any excess contamination and the remaining ethanol, an excess of water (at least 25 times sample cell volume) was flowed through the sample cell, unless otherwise specified.

2.6. Rheology on formulations

The rheological measurements were carried out on an Anton Paar Physica MCR 301 instrument equipped with a cone–plate geometry (50 mm, 1°). The temperature was controlled by a Peltier system. Steady and small amplitude oscillatory shear rheological experiments were performed on the non-hydrated samples with 0, 1, 5, 7.1, 10 and 20 wt% of nanogels. Frequency sweep experiments were performed in the linear viscoelastic regime.

2.7. Confocal laser scanning microscopy

A Leica CLSM TCS SP2 confocal microscope, operated in the inverted mode, using a 63 × 1.3 numerical aperture water immersion objective was used to image the lipid-based surface structures in excess water. In order to visualize the layers, lipid formulations were labelled with 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) for the lipid phase (0.01 mol%) and rhodamine 110 for the water phase (100 nM). The red fluorescence of rhodamine B-labelled lipid and the green fluorescence of rhodamine 110 were excited by a 561 nm DPSS and a 488 nm Ar ion laser, respectively. The samples were contained in a single-well sample holder with a round cover glass (diameter of 15 mm) and hydrated with 2 ml of water. Images were taken with a resolution of 512 × 512 pixels using a 400 Hz bidirectional scan with each scanning line averaged four times. Leica software was used to create three-dimensional reconstructions of the z-stacks.

2.8. Neutron reflectivity

Specular and off-specular neutron reflectivity (NR) experiments were performed on the OffSpec beamline at the ISIS Facility, Rutherford Appleton Laboratory (Chilton, Oxfordshire, UK). The basic principle of neutron reflectometry involves directing a collimated neutron beam onto a flat interface at a low incidence angle, θ, and measuring the ratio of reflected to incident intensity, R, as a function of momentum transfer Qz = 4π sin θ/λ, which is dependent on the incident angle, θ, and the neutron wavelength, λ. Data were collected at two incident angles (0.5° and 1.6°) at a wavelength range of 2.0–14.0 Å. Specular NR yields information about the vertical (out-of-plane) layer structure of the sample: in most cases, a layer model is fitted to the data and values for each layer's thickness, scattering length density (or composition) and roughness are obtained [22,23]. Best results are achieved if there is large scattering contrast between the individual layers of interest. This can be achieved or enhanced by using selective deuteration. Another possibility is the use of a ferromagnetic under-layer. For this purpose, our films were formed by spin-coating of lipid onto silicon blocks coated with ferromagnetic Co₉₀Zr₁₀ layers (see electronic supplementary material). The purpose of the magnetic layers was to provide added contrast in the NR measurements. When the reflectivity of the magnetically saturated sample is measured using a polarized neutron beam, the two different (in the absence of spin flip) neutron spin states give rise to different optical potentials in the magnetic layer and thus two different contrasts can be acquired without changing the isotopic composition of the sample (as described in Holt et al. [24]). In this way, magnetic layers were used in tandem with polarized NR to attain additional contrasts without the need for deuteration of the lipids in order to enable the precise determination of the structure of the interfacial lipid films.

The linear detector also recorded the off-specular reflectivity in the vertical direction, which yields information about the in-plane structure at the interface. The detector consists of 768 wavelength shifting fibres with 0.5 mm pitch, resulting in an observable Q_x range of −6.5 × 10⁻⁴ to 6.5 ×10⁻⁴ Å⁻¹. The observable Q-range corresponds to real space distances of approximately 1–40 µm (real and reciprocal space are inversely proportional and related via d ∼ 2π/Q). For each condition, NR data were collected in two aqueous bulk solvents (H₂O and D₂O) under two polarizations, resulting in a total of four datasets with different contrasts for the same structure. The bulk solvents were exchanged using an HPLC pump at a rate of 0.5 ml min⁻¹. In order to analyse the NR data, the four NR profiles of the bare layers recorded for different contrasts were modelled with Motofit Program run in Igor Pro (WaveMetrics) [25]. The scattering length density of each material was calculated as the total scattering length of the constituent atoms divided by the molecular volume. The molecular volume of H₂O and D₂O was calculated as 30 ų from the known density. For the NR profiles of lipid films, each Bragg peak was fitted with a Lorenzian function using PeakFit software (GraphPad Software, Inc.) in order to obtain the peak position in Q.

2.9. Chemicals

N-isopropylacrylamide and the cross-linker N,N′-methylenbis(acrylamide) for the nanogel synthesis was obtained from Acros Organics and Sigma-Aldrich, respectively. SDS and KPS were obtained from Duchefa Biochemie and Sigma-Aldrich, respectively. DGMO was obtained from Danisco A/S (Brabrand, Denmark). DGMO is composed of 88% of diglycerol monoester and 4.9% free glycerol and polyglycerols, in which the main fatty acid components are: oleic (90.7%), linoleic (4.2%), saturated (2.9%), eicosenoic (1.2%) and linolenic (0.8%) acids. Capmul GMO-50 (GMO-50, lot no. 140721-6) was purchased from Abitec (Janesville, WI). GMO-50 is a blend of 64.52% monoacylglycerol, 31.70% diacylglycerol and 3.78% triacylglycerol. The fatty acid constituents are 86.1% oleic, 5.7% linoleic, 4.0% palmitic and 3.2% stearic acid. The fluorescent lipid probe, 1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-(lissamine rhodamine B sulfonyl) (ammonium salt) (C₆₈H₁₀₉N₄O₁₄PS₂; Rhod-PE; greater than 99% purity) was purchased from Avanti Polar Lipids (Alabaster, AL) while rhodamine 110 was purchased from Invitrogen (Heidelberg, Germany). Sodium chloride (NaCl) was purchased from Sigma-Aldrich (Schnelldorf, Germany). Ethanol (99.7%) was purchased from Solveco AB (Rosersberg, Sweden). Chloroform (HPLC quality, 99.8% stabilized with ethanol) was purchased from VWR International (Leuven, Belgium). Water was purified to a resistivity of 18.2 MV cm at 25°C using the MilliQ system (Millipore). D₂O was purchased from Sigma-Aldrich (UK).

3. Results

In this study, we will first discuss the thermal response of the nanogels and the cubic LC phase alone followed by the response of the bulk nanogel–lipid mixture, including the rheological properties of the mixture. We will then discuss the film formation followed by an analysis of the nanostructural and morphological changes of the lipid–nanogel hybrid layer as a function of temperature. We have previously shown that PNIPAM nanogels can be embedded within films containing the glycerol monooleate and diglycerol monooleate with cubic phase nanostructure LC phases to more than 10 wt% of the lipid component [21].

First, we consider the nanogel particles themselves. At only 44 ± 1 nm in diameter, the particles used in this study are smaller than commonly used microgels of PNIPAM. These uniquely small particles were chosen in order to perturb the nanostructure of the LC phase as little as possible, but at the same time provide the responsiveness typical of PNIPAM. When heated above 32°C, the interactions of PNIPAM with water become unfavourable and PNIPAM undergoes a reversible LCST transition where it dehydrates and thus decreases in volume. As the temperature stimulus reaches the volume phase transition of the PNIPAM (32°C), each nanogel particle becomes more hydrophobic with water expelled from its polymer matrix resulting in a decrease in particle size. Figure 1 shows the hydrodynamic radius of the nanogel in water as a function of increasing temperature. The particle size decreases as we approach the expected volume phase transition of 32°C. It is interesting to note that it was not possible to measure particle sizes at or above 32°C as the particles aggregated in aqueous solution, which is expected from the poor miscibility of the collapsed particles in water [18,19].

The thermal response of the LC phase itself was then investigated. We have previously studied the structure and phase behaviour of this system extensively [20,26]. Based on these data, we selected the conditions where we expect to have a single reverse bicontinuous cubic Pn3m phase. Figure 2 shows the effect of the nanogels on the LC phase structure over a range of temperatures, from 25 to 40°C. The lipid LC phases with and without nanogels both give rise to six distinct Bragg reflections with relative positions, √2 : √3 : √4 : √6 : √8 : √9, which can be indexed as the diffraction from a cubic bicontinuous phase of the Pn3m space group, where the lattice parameters are given in electronic supplementary material, figure S1. It is apparent that there are only minute changes in the structure if any when the particles are added, at all the temperatures investigated. Both with and without the particles, we noticed a shift to higher q-values with increasing temperature, indicating a decrease in the dimensions of the structure. At 25.0 and 27.5°C, the lattice parameters with and without nanogels show almost the same values of 14.1 ± 0.1 nm (electronic supplementary material, figure S1). Above 27.5°C, the lattice parameter linearly decreases from 14.1 ± 0.1 to 12.4 ± 0.1 nm for the lipid phase without the nanogels, and from 13.9 ± 0.1 to 12.2 ± 0.1 nm for the lipid phase with the nanogels.

Figure 2.

SAXS diffractograms for lipid formulations alone (black line) and containing 10 wt% of nanogels (lower curve in blue) shown at different temperatures. The peaks observed correspond to, from left to right, six distinct Bragg reflections with relative positions, √2 : √3 : √4 : √6 : √8 : √9. The vertical line marks the √2 peak position for the formulation without the particles at 25°C. The lattice parameters obtained are plotted in electronic supplementary material, figure S1.

3.1. Preparing the hybrid layers

In order to achieve a homogeneous surface coating, we chose to use the technique of spin-coating a mixed ethanolic solution of nanogel and lipids. The formed layer was then incubated in excess water to remove any traces of ethanol and fully hydrate the lipid phase of the layer. Prior to layer deposition, SAXS was used to determine the structure in the bulk solution. The key findings are reported in electronic supplementary material, figure S2 and the diffractograms clearly show that the formulations had reverse micellar structure prior to hydration. It is interesting to note that this is in agreement with the phase diagram of GMO-50/DGMO/water, when water content is low, i.e. with water instead of ethanol [20,26]. Even in the non-hydrated samples, the nanogels did not affect the structure. However, we note a slight shift of the peak corresponding to the reversed micellar phase towards lower q (electronic supplementary material, figure S2), which is probably due to an expansion of the polar core of the micelles, at the highest concentration of particles. It is noteworthy that this is observed both at 25°C and 40°C and the peak does not shift with temperature.

We then studied how the presence of the particles affected the flow properties of particle dispersion in the lipid matrix before hydration. For this purpose, rheology measurements were carried out to reveal the interactions in the system. Here, it is useful to recall the different types of rheological properties of materials. For a Newtonian fluid, the stress is proportional to the strain rate. Bingham pseudoplastic behaviour refers to a material that behaves as a rigid body at low stresses but flows as a viscous fluid at high stress. In other words, when the stress exceeds a certain value (yield stress), the fluid behaves as a Newtonian liquid. A common example of such material is mayonnaise. While the flow behaviour of all rubber compounds is strongly non-Newtonian, viscosity decreases rapidly as the strain rate increases due to the elastic properties. The Maxwell model, i.e. a purely viscous damper and a purely elastic spring connected in series, describes the simplest viscoelastic behaviour of a material. A common example of such material is blood plasma. The rheology measurements show that formulations show a Newtonian behaviour between 0 and 1 wt% of nanogel, whereas at nanogel contents of above 1 wt%, the lipid reversed micellar solution with the embedded nanogels become stiffer compared with the lipid-only formulation. Such a stiffer film is desirable for maintaining the formulation on a substrate or a biological surface, such as the mucous membrane. Above 5 wt% of the nanogel in the formulation, shear thinning was observed. Such a shear thinning is desirable in applications for spreading of the formulation. From 7.1 to 20 wt% of nanogels, the shear thinning effect is strong as apparent from figure 3a. Figure 3b shows the corresponding shear rate/shear stress profiles, which reveal a Bingham pseudoplastics behaviour of the formulation. The frequency sweep behaviour of the sample without nanogels shows a rubbery behaviour at 20°C with a crossover at approximately 1 rad s⁻¹ (figure 3c). However, already by adding 1% of nanogels, a Maxwell behaviour can be easily recognized from the G′ and G″ slopes. This transition was not observed in the nonlinear rheology experiments. The Maxwellian behaviour is preserved also at 5% of nanogels. On the other hand, with 7.1 wt% of nanogels in the formulation, an additional transition occurs and the frequency sweep shows a behaviour between a rubbery and the intermediate regime before the glassy state with a crossover at approximately 3 rad s⁻¹. At higher concentration, the solutions appear to be in the intermediate regime before the glassy state. Three transitions can be appreciated between 0 and 1%, 5 and 7.1% and 7.1 and 10% of nanogels. At 7.1%, the layer shows the behaviour of a viscoelastic gel because G′ and G″ are almost equal. When the temperature is increased to 40°C except for 0% of nanogels, changes in the rheological properties do not indicate any change from a structural point of view (compare figure 3c,d). In fact, the changes observed in the absolute values of G′ and G″ are consistent with what is expected by applying an Arrhenius scaling factor. This suggests that the nanogels do not undergo a volume phase transition between 25 and 40% contrary to what is observed in water (figure 1). This is in agreement with previous observations that PNIPAM microgels collapse in the presence of certain concentration of organic solvent in the aqueous phase and therefore do not exhibit a volume phase transition [18,27,28]. Based on the rheology results, it is possible to optimize the conditions for spin-coating, i.e. at high concentration of particles when the viscoelastic behaviour appears, it is necessary to decrease the spin-coating rate.

Figure 3.

(a) Flow curves at 0, 1, 5, 7.1, 10 and 20 wt% of nanogels at 20°C. (b) Shear rate/shear stress profiles at 0, 1, 5, 7.1, 10 and 20 wt% of nanogels at 20°C. Frequency sweep in the linear viscoelastic regime at 20°C (from a) in (c) and 40°C (from b) in (d).

Spin-coating was used to coat solid surfaces with thin films of the mixed lipid-only formulation. The centrifugal force drives the bulk phase materials radially outward to form a visually smooth and homogeneous surface layer. We used spectroscopic ellipsometry to optimize the process of spin-coating, using a wavelength range of 450–750 nm range to measure the thickness and estimate the homogeneity of the lipid films (electronic supplementary material, figure S3). A good fit was found using an optical model that describes the film as a single layer, with the refractive index being that of the lipids. The results are consistent with the findings of our previous study investigating the effects of spin-coating parameters on the formation of films made only of non-lamellar lipids, where the films were found to retain their reverse cubic (Fd3m) structure when deposited on solid surfaces [15,17]. Similar observations were made for bicontinuous cubic phases based on phytantriol with space group Ia3d and Pn3m at two different degrees of hydration [29].

3.2. The change in morphology as the temperature increases

In order to visualize the layers, we included a lipid analogue (rhodamine-labelled DOPE) to label the lipid phase and a hydrophilic dye (rhodamine 110) in the co-solvent used to formulate the lipid and nanogel. Surface layers formed from these fluorescently labelled lipid formulations were spin-coated onto glass substrates and imaged in excess water using confocal microscopy. Visualization in the z-direction (figure 4) shows that the lipid-only films were 5 ± 1 µm and smooth. However, the hybrid lipid films that contained the nanogel appeared to be thicker (up to 18 µm in regions) and had a more heterogeneous structure. The films appeared to have a network-like structure with lipid domains (labelled red) interwoven with micro-sized water-rich regions (figure 4b). A temperature increase caused very little change to the morphology of the film containing lipid LC phase only, as seen in figure 4a,b. However, for the films containing the nanogels, dramatic morphological changes were observed (compare figure 4b). There is an obvious loss of material from the surface and the remaining lipid-rich parts appear to be thicker. It appears as if the lipid-rich domains film are compressed and partly detached from the surface. This temperature-triggered release of material from the surface occurs only in the presence of the nanogel and only when the film is hydrated. Preliminary DLS shows that the released particles have a size of about 300 nm (data not shown).

Figure 4.

Three-dimensional reconstruction of a confocal fluorescence z-stack of images of (a) lipid-only and (b) lipid–nanogel layers (tilted surface area of 150 × 150 µm). The insets show the side view of the lipid layers. Scale bar, 20 µm. The transition in morphology of the layers shown occurs when the temperature is increased from 25 to 40°C.

3.3. The change in nanostructure as the temperature increases

The internal nanostructure of layers formed from lipids only or lipids with embedded nanogels on silica were further examined by NR. The lipid films were deposited onto silica-coated substrates with underlying magnetic layers to provide further neutron contrast (electronic supplementary material, figures S4 and S5). The use of magnetic reference layers allows us to record an additional set of data simultaneously in order to reduce the ambiguity of the fit. For clarity, mainly the spin-up NR data are shown (figure 5). Owing to the complex three-dimensional structure of these lipid-based films, it is at times difficult to separate quantitatively in-plane and out-of-plane structural information using simple fitting methods. This however does not preclude us from gaining an overall structural insight under different conditions, as will be discussed below.

Figure 5.

NR of (a) lipid-only and (b) lipid–nanogel layers at 25 and 40°C. The reflectivity from the bare magnetic surface is shown in grey. For clarity, only the data for the spin-up contrast in D₂O are shown.

First, we can conclude that the deposition of the lipid or lipid–nanogel mixtures significantly changes reflectivity profiles, which is illustrated in figure 5. The films made of the lipid-only mixture of GMO-50 : DGMO show evidence of strong structural order, with prominent Bragg peaks observed in all four contrasts (electronic supplementary material, figure S5a). Two notable peaks appear in the D₂O contrasts (0.072 and 0.080 Å⁻¹), figure 5a, and one peak in the H₂O contrasts (0.075 Å⁻¹) (electronic supplementary material, figure S5a). As the difference in contrast between the lipids and H₂O is small, it is expected that fewer structural features are resolvable in this contrast. The two peaks re-emerge when the solvent is changed back to D₂O and no changes were observed upon repeated rinsing showing that the structures were not affected by the flow of bulk solvent. Furthermore, we also analysed the off-specular reflectivity (figure 6). As discussed by Ott et al. [30], there are many different ways to plot the data. Here, we choose to plot it as Q_z versus Q_x and it is apparent that pronounced off-specular patterns with marked Bragg sheets are observed. These coincide with the Bragg peaks from the specular reflectivity. The diffraction pattern was extracted from the off-specular NR patterns in figure 6 by integrating the intensity over Q_x range from 0.00005 to 0.00015 Å⁻¹ in figure 6 for each Q_z value and the results are plotted in electronic supplementary material, figure S6. Over the limited Q-range available for each sample, two distinct Bragg peaks could be identified that reverse bicontinuous cubic Pn3m phase, although not unambiguously identify the cubic phase. The position of the two peaks in the diffractogram from the off-specular data can be indexed to √3 and √4 (electronic supplementary material, figure S6), corresponds to a cubic phase that is similar to the phase observed for the lipids in bulk (Pn3m), although slightly more swollen. We therefore conclude that the lipids deposited onto the surface adopt such a structure when equilibrated with excess water.

Figure 6.

Off-specular NR of (a,b) lipid-only and (c,d) lipid–nanogel layers at 25°C (a,c) and 40°C (b,d). The Q_z − Q_x plot is extracted from the detector image at an incidence angle of 1.6° and for the spin-up contrast. The specular ridge as well as the off-specular Bragg sheets are indicated. The white square in (a) indicates the part of the image used to produce the plot of intensity versus Q_z in electronic supplementary material, figure S6, by integrating over the Q_x range from 0.00005 to 0.00015 Å⁻¹. The corresponding areas were used in (b–d).

When the lipid solution used for spin-coating contained spherical nanogels (GMO-50 : DGMO at 40 : 60 weight ratio with 10 wt% nanogel), the hydrated surface layer shows only limited structural order (figure 5b). The Bragg peaks from the specular reflectivity were much lower (compared with the lipid-only film), suggesting that the nanogels cause some disordering or thinning of the lipid layer. In the case of the H₂O contrasts, where the difference between scattering length densities of the lipid and the solvent is small, the peaks are not visible. At 25°C, only one distinct peak is observed in the D₂O contrasts at 0.075 Å⁻¹ (figure 5b). However, the Bragg sheets observed in the off-specular pattern are very clear and indicate a significant structure in the layer. In contrast with the specular reflectivity, two distinct reflections appear in the off-specular pattern (electronic supplementary material, figure S6) corresponding to the √2 and √4 Bragg reflections. Comparing the cubic phase with the nanogel in the surface film with the neat lipid film, we note that the former is more hydrated than with only lipid with lattice parameter of 16.5 nm compared with 15.3 nm for the neat lipid aqueous LC phase.

As the embedded nanogels are composed of a temperature-responsive polymer, we also investigated the effect of external temperature stimulus on surface layers using NR. For the lipid alone, as the temperature is increased from 25 to 40°C, there is a clear temperature-dependent response in terms of the decrease in hydration of the film as well as an increase in chain disorder with increasing temperature [4]. Both effects lead to a shift of the Bragg peaks to large Q-values, i.e. a decrease in the unit cell dimensions (and hydration). Upon increasing the temperature to 40°C, the specular peaks shift to higher Q-values (to 0.083 and 0.092 Å⁻¹ in the D₂O contrasts and to 0.087 Å⁻¹ in the H₂O contrasts). This trend is even more pronounced in the off-specular reflectivity. While further structural characterization of the lipid layers is needed, the diffraction data extracted from the off-specular reflectivity indicate that the lattice parameters of the lipid structure become smaller with temperature, for the lipid-only layer, they are 15.3 ± 0.1 and 13.4 ± 0.1 nm for 25 and 40°C, respectively. The corresponding values for the mixed lipid–nanogel layers are 16.5 ± 0.1 and 11.7 ± 0.1 nm for 25 and 40°C, respectively. That is, the effect of dehydration is significantly larger with the nanogels present. There is one more point to extract from the NR data and that is the reversibility of the layer. It was only possible to study this in detail in the case of the lipid film without nanogels as it is only here we have sufficiently sharp Bragg peaks. This is illustrated in electronic supplementary material, figure S5a, where the grey markers are the data recorded after heating to 40°C followed by cooling to 25°C. This shows that while one of the Bragg peaks seems to appear at the same place, and is even more pronounced, the high Q Bragg peak seems to have disappeared. It is tempting to speculate that this is a consequence of annealing of the structure, leading to changes in the LC domains, either in their size or their orientation.

4. Discussion

The approach used here was to prepare particles that were as small as possible for these types of PNIPAM nanogels to minimize the impact on the nanostructure in terms of blocking the connected aqueous network of the cubic phase nanostructure. This ensures that biomolecules, like e.g. enzymes, can be entrapped in the aqueous cavities of the cubic LC phase. However, the nanogels must be large enough to be responsive to a change in temperature. The responsiveness of the nanogels themselves is demonstrated by DLS (figure 1). Based on the set of data presented, we can now discuss the location of the nanogels in the lipid phases. First, we note that nanogels themselves are surprisingly soluble in the LC phase up to at least 10 wt% of the lipid and that these mixtures still form a Pn3m cubic phase in excess of water. We also note that the changes in the lattice parameter with temperature are not significantly affected by the presence of the particles. At least that is the case for the bulk sample prepared in excess of water based on the SAXS data reported in electronic supplementary material, figure S1. We also note that a similar behaviour was observed when the lipid–nanogel mixture was dissolved in ethanol (electronic supplementary material, figure S2). In the latter case, a reverse micellar phase is formed. Here, one could think of the particles as located in the polar region of the reverse micellar phase as we see a shift in the peak towards lower q, when the particle concentration is increased to 10%. However, the question remains as to how so much of the nanogel particles can be included in the cubic LC without visible phase separation and with only modest changes in the unit cell dimensions as the particles are an order of magnitude larger than the dimensions of the aqueous channels in the cubic phase. Thus, the nanogels are clearly too large to enter the aqueous cavities of the cubic phase. We suggest that they are located in the grain boundaries of the cubic phase mono-crystalline domains, strengthening the attractive interaction between the domains (figure 7). In this way, they could be solubilized to a relatively large degree without disturbing the LC structure. This could explain the large impact they have on the rheological properties of the mixture, in excess water, where the formulation with nanogels is more elastic and with neat lipid LC, the sample is viscoelastic [21]. Even before hydration, when the formulation is in the reverse micelle phase, already concentrations as low as 1 wt% of nanogels change the rheological properties (figure 3). Such a large effect could only be achieved if the nanogels are located in the boundary between different LC domains and not homogeneously distributed.

Figure 7.

Illustration of the proposed structural and morphological features of the mixed lipid–nanogel layer. Note that the figure is not to scale, but highlighting the main features. The layer comprises discrete regions of lipid crystalline grains (yellow), bordered by nanogels (blue spheres) as shown in the close-up. A further close-up depicts the Pn3m cubic structure of the lipids that exists within the lipid grains (green and yellow surfaces, indicates the presence of the two systems of water channels) and the surrounding nanogels (blue spheres). The fact that the nanogels are about 10 times the size of the water channels is highlighted.

We then turn our attention to films formed on the solid substrate. From the confocal microscopy, it is clear that the surface film is rougher in the presence of the nanogels than without (figure 4), while at the same time, the specular and off-specular reflectivity is giving clear indications that the cubic Pn3m structure is maintained (figures 5 and 6). This is consistent with the picture that the nanogels are accumulated in the grain boundaries. The rougher structure is also a consequence of the different rheological properties of the solution used for spin-coating, where we observed a more gel-like structure (figure 3). The interesting morphological transition for the lipid layer containing nanogels occurs when the temperature is increased to 40°C (figure 4). Here, we observe an apparent loss of material from the surface where the remainder of the layer appears thicker and more compact. The question here is if there is any cubic phase structure left on the surface layer. If we look at the specular data in figure 5b, there is a hint of a Bragg peak, but it is not obvious. However, if we look at the off-specular data (figure 6d), there is a clear indication of a cubic phase structure. Even though the limited q interval available from the off-specular reflectivity does not allow the unambiguous indexing of a cubic phase, the clear presence of Bragg sheets (extending also in Q_x direction) indicates the presence of a periodic in-plane (or 3D) structure. This becomes obvious when we integrate parallel to the Q_x axis to extract a diffractogram, where we plot intensity versus Q_z (electronic supplementary material, figure S6). In fact, here, the Bragg peaks appear sharper than in the specular reflectivity. It is also noteworthy that some of the Bragg peaks do not appear in the specular pattern as in the diffractogram extracted from the off-specular image. So this confirms the presence of a heterogeneous structure, both in terms of lateral and 3D organization as in the domain size and in their interactions.

We can now suggest a model of the formed nanogel–lipid hybrid structure and this is illustrated in figure 7. The main feature is the hierarchical structure where the lipid domains are surrounded by nanogels, which are likely to be fully hydrated but with no access water. The lipid domains themselves are LC with cubic bicontinuous phase, which the diffraction data show, are of the Pn3m type. This hierarchical structure accounts also for the response in morphology with temperature. A likely mechanism for the events observed on heating, based on the hypothesis that the nanogels are trapped in the grain boundaries, can now be suggested. First, we can conclude that very little effect is observed on the nanostructure of the lipid when the lipid–nanogel mixture is not exposed to water. So the likely response of the particles is that when the temperature is increased, they contract and aggregate. This is shown by DLS data in water. This contraction and aggregation tends to also compress the lipid grains, but also release of water from the particles. Both these effects lead to release of material from the surface and into the bulk solution. In fact, preliminary DLS data show particles of around 300 nm in diameter are released into the bulk solution (data not shown). The lipid domain size is not easily accessible from the recorded data, but the released particles from the layer upon heating are the order of 300 nm. Most likely, the size corresponds to that of a single domain. We have previously observed organized assembly of PNIPAM microgel particles at the lipid–aqueous interface of a lipid bilayer when increasing the temperature above volume phase transition temperature [31]. While in that paper, we were studying two-dimensional assembly at an interface, here we are considering the assembly in a three-dimensional system.

In this study, we also probed the feasibility of the composite layers to serve as suitable systems for controlled release. We envisage that the nanogel-guided release of small particles (domains of LC lipids) would be a first step with subsequent release of entrapped bioactive molecules from the LC lipid phase itself, which could be mediated by other triggers such as enzyme-controlled degradation, for example. While polymer microgels have been used for responsive delivery of model drugs [32], their potential cytotoxic effects renders them an unsuitable carrier. However, by encapsulating these nanogels within a protective lipid matrix, it would be possible to avoid this effect. Another important aspect is that in the present hybrid system, we only see the effect of the thermo-response from the nanogels when the layer is full hydrated, i.e. in a large excess of water.

5. Conclusion

Hybrid films composed of non-lamellar lipid liquid crystals and nanoparticles are emerging as fascinating materials both for fundamental studies and advanced applications. Here, we demonstrated the tunability of a system comprising of LC lipids with nanogels, spherical submicron-sized polymer particles. Via a simple spin-coating procedure, it is possible to form surface layers of nanogels embedded in a cubic phase lipid matrix. The embedded nanogels act as thermoresponsive controllers of layer hydration and adhesion. Upon heating above the phase transition of the polymer, the nanogels undergo a volume phase transition, which triggers a release of material from the surface. Using SAXS, we observed that the lipid domains retained their internal structure and it was possible to use NR to observe topological restructuring in the lipid films in response to heating. We also demonstrated that in the presence of 10 wt% of nanogels, the lipid matrix of these surface films retains a cubic structure in excess water, meaning that it has interconnected networks of nano-sized channels, which are an advantage for potential applications in controlled release. This approach opens up the possibility to tune the properties of the structure for lipid–polymer-based responsive layers.

Authors' contributions

A.P.D. coordinated the study, participated in the design of the study, carried out parts of the laboratory work, and helped draft the manuscript; M.V. and M.W. carried out parts of the laboratory work including the sample preparation and the SAXS measurements and M.V. interpreted the SAXS data; C.H. carried out the optimization of the spin-coating as well as ellipsometry and NR measurements; C.M. carried out the microscopy and data analysis; G.K.P. participated in the preparation of the magnetic reference layers and NR measurements; S.G. prepared the magnetic reference layers; S.N. synthesized and characterized the nanogel particles; L.G. carried out the rheology measurements and data analysis; J.B. participated in the design of the study and helped draft the manuscript; N.-J.S. and M.W.A.S. participated in the neutron studies and data analysis; G.S.-T. designed and produced figure 7; T.N. conceived of the study, took part in NR experiments/data analysis and drafted the manuscript. All authors gave final approval for publication.

Competing interests

We declare we have no competing interests

Funding

This work was supported by an infrastructure grant for the Super Adam reflectometer (CRG instrument at ILL, Grenoble, France) from the Swedish Research Council (2015-05988), which also supported the postdoc position for G.K.P. and A.P.D. A.P.D. wishes to acknowledge the support of a Young Investigators Training Program (YITP) scholarship financed by ACRI (Italian Banking Foundation Association). M.V. was supported by People Programme (Marie Curie Actions) of the European Union's Seventh Framework Programme FP7/2007-2013/ under REA grant agreement no. 606713 (Bibafoods). M.W. and T.N. were supported by the Knut and Alice Wallenberg Foundation (Framework grant ‘Anisotropic Forces in Colloid Chemistry’). S.N. was supported by Lund University and Swedish Research Council. NanoLund are acknowledged for funding the acquisition of the spin coater. The Knut and Alice Wallenber Foundation funded the acquisition of the spectroscopic ellipsometer.