Near-Infrared-Triggered Reversible Transformations of Gold Nanorod-Laden Lipid Assemblies: Implications for Cellular Delivery

Abstract

Robust drug and gene delivery systems require innovative methods to control payload release and tune delivery efficiency. The most promising delivery materials are lipid-based and their efficiency often hinges on structural transformations activated by endogenous pH changes. Exogenously driving phase transitions in lipid assemblies is a tantalizing idea that could lead to better control of cargo release dynamics. Multiple reports have demonstrated phase transitions induced in lipid systems, achieved via plasmonic heating of entrained gold nanorods. However, undesirable nonlocalized heating is common due to the size mismatch between the nanorods and the lipid architecture in these systems. Lipid assemblies often exhibit lattice dimensions of just a few nanometers, rendering gold particles challenging to integrate due to their incommensurate sizes, especially in lipid nanoparticle or colloidal forms. We investigate these processes using a judiciously chosen ternary lipid system with entrained small gold nanorods that undergoes transitions between bicontinuous cubic and inverse hexagonal phases on exposure to near-infrared light. Utilizing small-angle X-ray scattering alongside electron reconstruction, we show that gold nanorods integrate into the lipid assembly core lattice by colocalizing in the water nanochannels. We also found that plasmonically activated transformations occur in a couple of minutes and are reversible.

Graphical Abstract

1. INTRODUCTION

Lipid-based nanoparticles (LNPs) are among the most promising drug and gene delivery materials for the treatment and prevention of a variety of diseases.^1–3 In fact, we are all witnessing the tremendous impact of LNP gene delivery with the development of the COVID-19 vaccine.^4–6 Expanding the application space of lipid-based delivery systems hinges on our ability to engineer materials that effectively penetrate cells, a process most often regulated via endocytosis. Once encased in the endosome, lipid materials must disrupt the endosomal membrane,⁷ and transformation processes such as membrane fusion, pore formation, or phase changes from bilayer lamellar to tubular reversed hexagonal are known to facilitate the process.⁸ Bicontinuous cubic structures self-assembled from lipids have shown promise as improved nanotherapeutic vectors for insoluble and immunogenic drugs as well as nucleic acids.^9–25 The delivery efficiency of bicontinuous cubic phase lipid materials has been correlated with the fact that these lipid nanostructures are more prone to fuse with target membranes,^26–29 thus increasing the rate of endosomal escape.¹⁸

The elastic energy cost of membrane transformations is well described by the Helfrich theory,³⁰ a successful continuum-elastic description of membrane energetics that depends quadratically on curvature. Briefly, the total curvature-elastic energy of a membrane is the integral of the energy density over the entire membrane area, $E/A=\frac{\kappa }{2}{\left(J–J_0\right)}^2+\bar{\kappa }K$, where the total or extrinsic curvature $J$ is a sum of the membrane’s local principal curvatures, $J=C_1+C_2$, the Gaussian curvature is $K=C_1{C}_2,J_0$ is the spontaneous membrane curvature, $\kappa$ is the (normal) bending modulus, and $\kappa \circledR$ is the Gaussian curvature modulus. Bicontinuous cubic phases reduce the elastic cost of forming a membrane fusion pore because of their intrinsically positive $\kappa \circledR$ and negative $K$.^27–29,31 Another way to modulate elastic energy density is to render $J_0<0$, as in inverse hexagonal lipid nanostructures, which have been deemed “fusogenic”.³² LNPs comprising inversed hexagonal phases (hexosomes) are promising nanovectors, but unlike cubosomes, they cause significant cytotoxicity when used to deliver siRNA to mammalian cells. This cytotoxicity was presumed to stem from destabilization of the plasma membrane before the onset of endocytosis,³³ an effect similar to the mechanism of action of antimicrobial peptides.³⁴ We sought to combine the low cytotoxicity and high loading efficiency of cubosome LNPs with the high fusogenicity of hexosome LNPs by designing a material system with remotely activated phase transformations between the cubic and the hexagonal nanostructures. By doing so, we aim to lay the groundwork for future colloidal materials that could be remotely phase-activated after endocytosis, allowing heightened control over cellular delivery.

Multiple lipid phase transition stimuli are potential targets for remote activation, including temperature, pressure, pH, irreversible chemical processes such as lipid degradation and cleavage, ultrasound, and others.^35–39 Of these, temperature is usually favored due to its ubiquity in existing studies, as well as the existence of many common lipid systems with biologically relevant transition temperatures.⁴⁰ To achieve remote phase activation via locally induced heating, energy must be transferred into the lipid particle. This has been achieved with many different forms of electromagnetic radiation including ultraviolet light (UV),⁴¹ visible light,⁴² and near-infrared (NIR) light,^43–45 as well as magnetic fields.⁴⁶ Of specific interest for colloidal applications in biological systems are particles photothermally activated by NIR light due to the high NIR transparency of most biological tissue and the lower risk of cell damage from ionization.⁴⁷ Of note is the work of Boyd et al. showing phase transitions between lamellar and nonlamellar lipid systems by remote NIR stimulation of entrained micro- and nanoparticles,^48,49 including gold nanorods.^44,45,50

Gold nanoparticles are promising biocompatible materials being developed as photothermal agents for a number of therapeutic and diagnostic applications, and they notably function through localized surface plasmon resonance (LSPR), which enables absorption wavelength tuning by changing particle shape (e.g., gold nanorods).^51–53 The length of the nanorods used in previous lipid photothermal phase transformation studies is approximately one-half to one-fourth the target diameter of an LNP (ca. 80–150 nm).^44,45 While these nanorods worked well for initial studies, the large particle sizes could cause problems such as nonuniform heating and inefficient and irreproducible particle encapsulation for colloidal applications. In this manuscript, we build upon the aforementioned efforts by tackling two critical barriers to the development of photothermal LNPs: (1) limiting the size of entrained nanoparticles such that they may be incorporated into nanoscale gene/drug delivery vectors and (2) ensuring precise integration within the core lattice of the lipid nanostructure to eradicate off-target effects.

With these goals in mind, we synthesize gold nanorods (AuNRs) with diameters small enough to allow their incorporation into the tight water domains of lipid lattices in a biocompatible ternary lipid system, allowing structural transformations through plasmonic stimulation. Small-angle X-ray scattering (SAXS) is used to monitor in situ reversible transformations from bicontinuous cubic structures of two different spacegroup symmetries $(Pn\stackrel{-}{3}m,Im\stackrel{-}{3}m)$ to the 2D inverse hexagonal phase upon NIR irradiation. Real-space electron reconstructions of the $Pn\stackrel{-}{3}m$ phase obtained from the experimental SAXS data reveal that the AuNRs reside in the water channels of the structure. Our study shows for the first time that small AuNRs incorporate at the core lattice of complex lipid assemblies, permitting precise and reversible control of nanostructural transformations via plasmonic stimulation.

2. RESULTS AND DISCUSSION

Our study focuses on the formulation of tunable lipid mesophases that undergo thermally induced transformations from bicontinuous cubic to inverse hexagonal phase via local heating for future application in LNP-triggerable drug/gene delivery. We synthesize small AuNRs and integrate them in a biocompatible ternary lipid system, which we term GOPEG, comprising lipids already used in pharmaceuticals such as monoglyceride glycerol monooleate (GMO) and small amounts of both anionic fatty acid oleic acid and PEGylated phospholipid DOPE-PEG (1,2-dioleoyl-sn-glycero-3-phosphoethanolamine-N-[methoxy(polyethylene glycol)-2000]). The rationale is to attain transformations triggered by NIR that are quickly activated, are reversible, and have a transition temperature only slightly above human biological temperatures. Each lipid in GOPEG is added to serve a specific function. GMO is used as a base lipid due to its documented ability to form bicontinuous cubic phases.^54–57 Oleic acid is added for dual purposes of (1) lowering the transition temperatures of the system by increasing extrinsic curvature⁵⁸ and (2) adding charge, increasing LNP colloidal stability as well as aiding the incorporation of cationic cetrimonium bromide (CTAB)-coated AuNRs. Finally, DOPE-PEG is added to increase systemic circulation time and decrease immunogenicity of future LNP formulations.^59,60 The lipids used are shown in Figure 1A, along with a simple depiction of the planned cubic lipid–AuNR complex phases.

Figure 1.

(A) Project goal is to attain cubic lipid–AuNR composite materials from the base lipids shown (left). (B) A depiction of the phase behavior of GOPEG incorporating 10 nM AuNR under externally applied heating. Light blue refers to the $Im\stackrel{-}{3}m$ bicontinuous cubic phase ($Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}$), dark blue refers to the $Pn\stackrel{-}{3}m$ bicontinuous cubic phase $\left(Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}\right)$, and orange refers to the inverse hexagonal phase $\left(H_{\mathrm{I}\mathrm{I}}\right)$.

2.1. Analyzing the Structural Phase Space of GOPEG under External Heat.

Figure 1B shows the results of an external heating study of GOPEG incorporating AuNRs in excess water. Oleic acid molar content was varied from 8 to 9% (mol), while DOPE-PEG content ranged from 0 to 1% (mol). The amount of GMO was calculated as the remainder. AuNR concentration was held constant at 10 nM. SAXS patterns were acquired at eight temperature steps ranging from 20 to 52 °C to probe the temperature-dependent phase behavior of each lipid mixture. We found that the increased oleic acid content correlated with an overall decrease in phase transition temperature between bicontinuous cubic and inverse hexagonal phases. This result aligns with previous findings showing increased stability of the inverse hexagonal phase at low temperatures with the addition of oleic acid to a GMO-dominant system.⁵⁸ However, a more stark change in phase behavior occurred with the addition of DOPE-PEG to the lipid system. With no DOPE-PEG added, the binary GMO–oleic acid system exhibited a $Q_{\text{II}}^{\mathrm{D}}(Pn\stackrel{-}{3}m$ spacegroup) phase close to room temperature, with a phase change to the $H_{\text{II}}$ phase at elevated temperatures. After the addition of as little as 0.5% DOPE-PEG, the $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}(Im\stackrel{-}{3}m$ spacegroup) phase was instead found at room temperature. Even though a different type of bicontinuous cubic phase formed at room temperature with the addition of PEGylated lipid, a phase change to inverse hexagonal was still present at elevated temperatures. This result establishes the feasibility of an LNP formulation with the desired phase behavior since PEGylated lipids can help stabilize colloidal lipid nanoparticles, as well as increase their circulation time in vivo. Lipid samples without AuNRs exhibited similar phase behavior upon external heating (Figure S1), with only minor changes to cubic-hexagonal phase transitions observed relative to samples with included AuNRs. A comparison of integrated SAXS patterns is shown in Figures S2 and S3 for compositions without and with DOPE-PEG, respectively.

2.2. Probing AuNR Incorporation into the Water Domains of the Lipid Lattice.

We sought to determine the precise location of the AuNRs in the GOPEG lattice. Structure factor intensities were extracted from the azimuthally integrated 2D-SAXS intensity profiles using Le Bail refinement performed in the JANA crystallographic software.⁶¹ The intensities were used as input for the charge-flipping algorithm of SUPERFLIP⁶² to yield electron density maps that were visualized as 90% isosurfaces using the VESTA software package.⁶³ The reconstructions obtained from the $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$ structure in the presence and absence of AuNRs, with a sample composition of 91% GMO and 9% oleic acid, are shown in Figure 2A. The 4-fold connector is apparently enlarged in the presence of the AuNRs, and the water channels appear discontinuous on the map. We posit that the AuNRs localize in the 4-fold connectors of the bicontinuous double diamond structure, which ostensibly amplifies the electron density of the nodes relative to the water channels around them. To confirm this hypothesis, we calculated the diameter of the water-filled rods at the node from the electron density of the (100) plane of the two $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$ phases (shown in Figure 2B). It is evident that the rods near the node are larger in diameter for the structure with the metal dopants. Note that the lattice parameters in the presence $(a=87.0\mathrm{\AA })$ and absence $(a=87.9\text{Å})$ of AuNRs are similar. The radius of the water channels $\left(r_w\right)$ estimated from the 2D electron density maps are 19.6 and 8.9 Å for the phases with and without Au, respectively. Using the minimal surface formulae by Anderson et al.⁶⁴ as simplified by Kulkarni et al.,⁵⁴ the radius of the water channels for $a=87.0\mathrm{\AA }$ of the $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$ phase is estimated to be 9.6–10.0 Å, which matches closely with $r_{\mathrm{w}}=8.9\mathrm{\AA }$, as found from the electron density reconstruction of the undoped structure. This implies that AuNRs localize at the 4-fold connectors of the $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$ structure, leading to apparent swelling of the node.

Figure 2.

(A) 90% isosurfaces of the $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$ phase obtained in the absence of AuNRs (left) and with incorporated AuNRs (right). The maps show the water nanochannels in the structure. (B) Electron density reconstructions of the (100) plane in the two structures show the larger spatial extent of the 4-fold connector node for the double diamond $Pn\stackrel{-}{3}m$ morphology formed with AuNR inclusions (right) than that of the same morphology without AuNRs (left).

2.3. Plasmonic Heating of the GOPEG–AuNR System.

After characterization of the lipid system under external heat, bulk lipid samples hydrated with a 10 nM aqueous solution of AuNRs (6.6 ± 0.1 nm × 20 ± 0.6 nm; Figure S4) were probed prior to, during, and after irradiation with an 808 nm NIR laser. The X-ray beam and NIR laser were aligned such that the sample volume being probed was within the area exposed to the laser. We found that under NIR irradiation, GOPEG–AuNR systems underwent fast (sub-100 s), repeatable, and often fully reversible transformations from bicontinuous cubic to inverse hexagonal phases. Control systems were tested to rule out X-ray degradation-induced phase transformation (Figure S5A), as well as significant heating effects from AuNR X-ray absorption (Figure S5B). Figure 3A,B shows integrated SAXS profiles representing time-resolved plasmonic heating experiments. The three data sets were taken to show, from bottom to top, the first scan prior to NIR exposure, diffraction after the laser has heated the sample enough to cause a complete phase transformation, and a scan after irradiation was ceased and the sample recovered for 160 s. The peaks in the diffraction data—as well as those not shown but represented in Figure 3C—were indexed to three nanostructures. The first phase, a $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}(Pn\stackrel{-}{3}m$ spacegroup) structure, was indexed to seven peaks at the reciprocal lattice vectors $q/(2\pi /\mathrm{a})=G_{hkl}/(2\pi /a)={\left(h^2+k^2+l^2\right)}^{1/2}=\surd 2,\surd 3,\surd 4,\surd 6,\surd 8,\surd 9$, and $\sqrt{10}$, corresponding to {110}, {111}, {200}, {211}, {220}, {221}, and {310} plane families, respectively. These peak indices satisfy the $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$ structure rules: (i) $0k\mathrm{l}(k+l=2n)$ and (ii) $h00(h=2n)$ (with $h,k,l$ permutable and $n$ is an integer).⁶⁵ The second phase, a $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{p}}(Im\stackrel{-}{3}m$ spacegroup) structure, was indexed to eight peaks at the reciprocal lattice vectors $q/(2\pi /a)=G_{hkl}/(2\pi /a)={\left(h^2+k^2+l^2\right)}^{1/2}=\surd 2,\surd 4,\surd 6,\sqrt{10},\sqrt{12},\sqrt{14},\sqrt{16}$, and $\sqrt{18}$, corresponding to {110}, {200}, {211}, {310}, {222}, {321}, {400}, and {411} plane families, respectively. These peak indices satisfy the $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}$ structure rules: (i) $hkl(h+k+l=2n)$, (ii) $0kl(k+l=2n)$, (iii) $hhl(l=2n)$, and (iv) $h00(h=2n)$. The final phase, the inverse hexagonal phase $H_{\mathrm{I}\mathrm{I}}$, was indexed to three peaks at the reciprocal lattice vectors $q/(4\pi /a\surd 3)=G_{hk}/(4\pi /a\surd 3)={\left(h^2+hk+k^2\right)}^{1/2}=\surd 1,\surd 3$, and $\surd 4$, corresponding to {10}, {11}, and {20} plane families, respectively.

Figure 3.

In situ plasmonic heating experiments. (A) SAXS I vs q data obtained for GOPEG exposed to three representative times of NIR laser heating. The GOPEG composition is 91% GMO, 9% oleic acid, and 0% DOPE-PEG. Scans shown were taken before NIR irradiation, after 20 s of exposure, and 170 s after halting exposure. Peaks are indexed and color-coded. (B) The GOPEG composition is 90.5% GMO, 9% oleic acid, and 0.5% DOPE-PEG. Scans shown were taken before NIR irradiation, after 70 s of exposure, and 160 s after halting exposure. Peaks are indexed and color-coded. (C) Calculated lattice dimensions plotted against time for the data shown in A and B, where the top and bottom plots in C correspond to A and B, respectively. Circles represent SAXS measurements and are color-coded by the phase(s) indexed in the said measurements. X symbols indicate scans shown in the data of part A. (D) Cyclic plasmonic heating experiment, with five repeated cycles of 100 s irradiation and 100 s cool-down periods.

The plasmonic heating experiment represented in Figure 3A,C (top) was performed with a GOPEG composition of 91% GMO, 9% oleic acid, and 0% DOPE-PEG, hydrated with AuNR solution. Immediately upon laser exposure, the cubic phase began to shrink, shown by a decrease in lattice parameter a. After ca. 20 s of NIR exposure, the sample began transforming from $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$ to $H_{\mathrm{I}\mathrm{I}}$, the transition reaching completion after ca. 30 s of exposure (middle integrated plot, Figure 3A). After a total of 100 s of exposure, the observed phase appeared to stabilize, and the laser was turned off. The sample immediately began to cool and swell, as shown by the increase in both $H_{\mathrm{I}\mathrm{I}}$ and $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$ lattice dimensions (a) in Figure 3C (top). After a cool-down period of ca. 170 s, the sample reverted to a phase state similar to what was demonstrated prior to laser exposure. This reversible phase behavior aligns directly with what was seen upon external heating of the same sample composition (Figure 1B), and the ability for the sample to return fully to its initial phase shows promise for applications in sustained drug delivery and in providing more fine-tuned control over release profile.⁶⁶

A second plasmonic heating experiment (Figure 3B,C (bottom)) was performed with a GOPEG composition of 90.5% GMO, 9% oleic acid, and 0.5% DOPE-PEG, hydrated with AuNR solution. After ca. 10 s of NIR exposure, the sample began showing both $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}$ and $H_{\mathrm{I}\mathrm{I}}$ phases, reaching nearly complete phase transformation after ca. 70 s of exposure (middle integrated plot, Figure 3B). This sample composition diverged in cool-down behavior from the sample shown in Figure 3A, in that it did not return completely to its original nanostructure. While the sample without DOPE-PEG fully returned to its original phase of $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$, the sample with DOPE-PEG cooled to a phase coexistence of $Q_{\mathrm{I}}^{\mathrm{P}}$ and $H_{\mathrm{I}\mathrm{I}}$. It is unclear whether this is an effect of the DOPE-PEG itself or if it is unique to the relationship between the $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}$ and $H_{\mathrm{I}\mathrm{I}}$ phases, but regardless of its origin, it appears to be a kinetically trapped state, most likely due to fast cooling after the laser is turned off. The broad peaks of the $Q_{\text{II}}^{\mathrm{P}}$ phase post NIR exposure are most likely indicative of small crystalline grain size. Metastable lipid phases in highly viscous lyotropic and thermotropic bicontinuous cubic structures are very common. We have also previously observed that both $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{D}}$ and $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}$ phases can form large single-crystal structures with extended incubation at ambient temperature.¹⁵ Additionally, researchers have previously documented supercooling effects in colloidal GMO cubosome–hexosome phase transitions, with the initial preheating phase sometimes taking hours to fully recover.⁶⁷ It is therefore not unusual that a full return to initial phase and grain size in the samples shown would take far longer than the presented experimental scope of minutes.

The dynamic and repeatable nature of the $Q_{\text{II}}$ to $H_{\text{II}}$ transformations of GOPEG activated by NIR was studied by performing a cyclic plasmonic heating experiment, in which five on–off cycles (100 s on, 100 s off) were performed in succession. The results in this experiment are shown in Figure 3D and were performed on the same GOPEG sample as in Figure 3B. The NIR-activated phase change is remarkably repeatable, with both $Q_{\text{II}}^{\mathrm{P}}$ and $H_{\text{II}}$ phases being present at almost all times. It is worth noting that each time after laser exposure ceases and right before laser exposure begins again, the lattice dimensions a of each phase return to almost the exact same values. One possible reason for this is that the thermodynamic balance between the amount of heat applied by the laser and the convective and conductive cooling of the sample impose a maximum temperature the sample can reach. This would explain the repeatability and low degree of hysteresis, assuming each heating cycle is long enough to reach a maximum temperature, as it would account for the lack of accumulated heat between heat–cool cycles. This would also explain the plateauing of the lattice parameters at the end of each heating cycle.

2.4. Effect of Relative AuNR Concentration on Plasmonic Heating.

An experiment was conducted to qualitatively study the effect of AuNR concentration on the observed phase behavior of the GOPEG material described above. GOPEG samples of identical composition (90.5% GMO, 9% oleic acid, 0.5% DOPE-PEG) were hydrated with three aqueous solutions containing either 100-, 50-, or 10%-concentrated AuNR solution (ca. 10, 5, and 1 nM, respectively). NIR irradiation experiments were performed as described previously, with SAXS plots of maximum heating shown in Figure 4A.

Figure 4.

Comparison of plasmonic heating behavior of samples with differing AuNR concentrations. Each sample was made with the same GOPEG composition (90.5% GMO, 9% oleic acid, 0.5% DOPE-PEG) but were hydrated with 100% AuNR solution (ca. 10 nM, dark blue), 50% AuNR solution (ca. 5 nM, light blue), and 10% AuNR solution (ca. 1 nM, gray). The remainder of the latter two hydrations were conducted with deionized water. (A) Single SAXS data of each of the three GOPEG–AuNR hybrid systems at times of maximum perceived heating for each separate plasmonic heating experiment. Each SAXS scan presented was the last taken during the plasmonic heating regime of each sample. The $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}$ {110} and $H_{\text{II}}${11} peaks are labeled by arrows. The inset plot shows the fractional H_II peak area plotted against the approximate AuNR concentration. (B) Fraction of fitted $H_{\text{II}}${11} compared to the sum of area under both $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}$ {110} and $H_{\text{II}}${11} peaks. Symbols are open during NIR laser exposure and filled during nonexposure. (C) Calculated lattice dimensions of the H_II phase during plasmonic heating experiment. Absent (unplotted) data in each plot represent times at which the phase in question could not be identified in the SAXS data. Symbols are open during NIR laser exposure and filled during nonexposure.

Integrated diffraction peaks were selected and fitted with a Gaussian peak profile to give an idea of the relative phase amount during phase transitions. Figure 4B shows the $H_{\mathrm{I}\mathrm{I}}${11} peak area fraction—the fitted peak area of the $H_{\mathrm{I}\mathrm{I}}${11} peak divided by the sum of $H_{\mathrm{I}\mathrm{I}}${11} and $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}${110} peak areas—for each of the three AuNR concentrations. All peak areas are corrected for peak multiplicity.⁶⁸ Since the SAXS peak area is determined by the amount of scattering sample, this ratio was used to qualitatively compare the progress of the phase transitions between the $Q_{\text{II}}$ and $H_{\text{II}}$ phases in the three samples. From the ratios, it is clear that a larger mass of lipid has shifted from $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}$ to $H_{\mathrm{I}\mathrm{I}}$ in samples with higher AuNR concentration, most likely due to higher plasmonic heat flux. Figure 4C shows the calculated $H_{\text{II}}$ lattice dimensions. The lattice shrinking effect—which correlates with increased temperature—increases with increasing AuNR concentration. This is also indicative of the increased plasmonic heat flux; since each sample would have similar cooling rates, the samples with more AuNRs would thereby be able to reach a higher maximum temperature and with that a higher ratio of $H_{\text{II}}$ phase within the X-ray beam path.

In Figure 4A, SAXS data selected at times of maximum perceived heating for each separate plasmonic heating experiment are superimposed to qualitatively contrast the $Q_{\mathrm{I}\mathrm{I}}^{\mathrm{P}}/H_{\text{II}}$ phase coexistence with changing AuNR concentration. The increase of AuNR concentration in the aqueous hydrating solution seems to correlate with a decrease in the extent of the $Im\stackrel{-}{3}m$ phase at full heating. Consequently, a reverse effect is observed with the $H_{\text{II}}$ {11} peak, chosen for this comparison over the {10} peak due to the proximity of the latter to the $Im\stackrel{-}{3}m$ {111} and {200} peaks, which increases the relative peak area. These interpretations led us to conclude a correlation between AuNR concentration and resultant plasmonic heating capability of the final GOPEG–AuNR hybrid system.

3. CONCLUSIONS

In this study, we define and characterize a ternary lipid system that utilizes entrained small gold nanorods to provide remote phase activation via plasmonic heating. This ternary system demonstrates nimble, reversible, and repeatable phase transitions from both bicontinuous primitive and double diamond cubic phases to an inverse hexagonal phase, with the primitive/hexagonal system arising upon the inclusion of DOPE-PEG. Increasing the mole fraction of oleic acid leads to a decrease in the onset temperature of the inverse hexagonal phase, as does increasing DOPE-PEG. Increased DOPE-PEG content also shows a marked increase in phase coexistence between the primitive cubic and inverse hexagonal phases. These observations establish a lipid system with tunable phase transition onset temperature. Small gold nanorods were loaded into the lipid matrix during hydration and were later shown by electron density reconstruction to be most probably situated in the nodes of the primitive bicontinuous cubic phase water channels. These smaller nanorods would provide even and highly localized heating unlike larger nanorods used in previous studies, and their incorporation into the water channels themselves allow for their inclusion in the phase as opposed to their being situated at defects and grain boundaries in these LNP mesophases. This is especially promising for the design of LNP formulations since limitation of the particle dimensions to the sub-100 nm scale also limits the ability to entrain larger nanoparticles. Studies on the effect of gold nanorod loading concentration on heating and phase behavior show that for the same lipid composition, heating effects are uniform to all tested nanorod concentrations, ca. 1–10 nM. For the three nanorod concentrations, at least partial cubic to hexagonal phase transitions were observed, with the degree of transition correlated to the concentration of added nanorods. This shows that relatively low nanorod concentrations are needed to achieve heating-induced nanostructural transformations. In an in vivo setting, the required concentration of nanorods may be even lower, as the required heating from the human body temperature to the transition temperature would be much smaller than in the described experiments. This shows promise for future LNP development and allows for flexibility when it comes to the concentration of entrained nanoparticles in the nanoscale formulations. This study provides a robust and biocompatible ternary lipid system with nanorods incorporated into the core nanostructure lattice, a system with significant promise for applications in triggered LNP drug and gene delivery.

4. EXPERIMENTAL DETAILS

4.1. Materials.

Glycerol monooleate (≥99%), oleic acid (≥99%), cetrimonium bromide (≥99%), sodium borohydride (≥99%), gold(III) chloride trihydrate (≥99.9%), silver nitrate (≥99.9999%), and L-ascorbic acid (reagent grade) were purchased from Sigma-Aldrich (MO, USA). DOPE-PEG2000 (≥99%) was purchased from Avanti Polar Lipids (AL, USA). Hydrochloric acid was purchased from Ricca Chemical Company (TX, USA). All chemicals were used without further purification, and lipids delivered in powder form were dissolved in fresh chloroform solvent (Macron Fine Chemicals, PA) to achieve the desired concentration. Quartz glass mark tubes were purchased from Hilgenberg GmbH (HE, DE). Lacey carbon transmission electron microscopy (TEM) grids with 50 μm holes on 200-mesh copper were purchased from Electron Microscopy Sciences (PA, USA). The 200 mW, 808 nm, and 12 mm lasers used for remote heating experiments were purchased from OdicForce (Surbiton, U.K.).

4.2. Gold Nanorod Synthesis and Purification.

Gold nanorods were synthesized using a seedless protocol, as published by Ali et al. The specific recipe used is documented in Section 2.2 of the publication.⁶⁹ After synthesis, the nanorods were centrifuged at 15 000 rpm or 21 130 rcf (Eppendorf 5424R) for 15 min, after which the supernatant was removed and the nanorods were redispersed in water. This washing cycle was conducted a total of three times, before a fourth centrifugation cycle at 6000 rpm or 3381 rcf was used to better isolate nanorods from nanospheres, following the published procedure by Scaletti et al.⁷⁰ The final nanorod solution was once again centrifuged at 15 000 rpm or 21 130 rcf and concentrated to a final volume of approximately 50 μL. The resultant nanorods were deposited onto lacey carbon grids and imaged by TEM (JEOL 2100 Cryo TEM). A total of 344 nanorods were manually characterized from the resultant micrographs in MATLAB. Of these particles, 81% were counted as rods, defined as a particle with a length at least 25% greater than its width. The remaining 19% of particles were counted as spheres. The rods measured were 6.6 ± 0.1 nm in diameter and 20 ± 0.6 nm in length. The rod and sphere size distribution histograms can be seen in Figure S6. The total concentration of nanorods in solution was estimated to be 10 nM by inductively coupled plasma mass spectrometry (ICP-MS) analysis of an analogous nanorod solution. The final concentration was determined using metallic gold density, and the average nanoparticle volume that was seen in the micrographs, including both rods and spheres, assuming hemispherically capped cylinders.

4.3. Bulk Lipid–Nanorod Sample Preparation.

Lipid chloroform solutions were pipetted into quartz capillaries and dried in a fume hood at ambient temperature for 2 days, before being transferred to a vacuum chamber for an additional 1 day of vacuum drying. After drying, 10 μL of Millipore water or AuNR solution was pipetted into the capillaries for hydration to achieve a final lipid concentration of 500 mM. Samples were then flame-sealed and epoxysealed to prevent evaporation and were incubated at 45 °C for 3 days to induce hydration and self-assembly. After incubation, samples were centrifuged (Thermo Fisher Sorvall ST16R) end-over-end at 3000 rpm or 1693 rcf three times in 2 min increments for a total of 12 min of centrifugation to ensure even hydration of the lipid cake. Samples were analyzed after resting for at least 1 d at ambient temperature.

4.4. Small-Angle X-ray Scattering.

Hydrated ternary lipid systems with and without added gold nanorods were analyzed by Synchrotron SAXS at beamline 12-ID-B of the Advanced Photon Source at Argonne National Laboratory. This beamline has an average photon energy of 14 keV, along with a beam size of 300 μ m × 20 μ m (W × H). Data were acquired on a Pilatus 300K 20 Hz hybrid pixel detector (Dectris, Switzerland) and were radially integrated on-site using locally authored MATLAB software. For peak area comparisons, peaks were analyzed using the Peak Analyzer tool included in OriginPro 2019. Gaussian fits were conducted on the peaks, and backgrounds were hand-determined for each SAXS integration.

4.5. Electron Density Reconstruction.

Le Bail refinement was performed using the JANA2006 crystallographic software to extract structure factor intensities. These intensities were then used as input to the charge-flipping algorithm of the SUPERFLIP software to compute the final electron density profiles. The maps were visualized and exported using the VESTA software. SUPERFLIP input files for profiles shown in this manuscript, as well as agreement parameters and other information, are included in a dedicated Supporting Information section.