Synthetic high-density lipoprotein nanoparticles delivering rapamycin for the treatment of age-related macular degeneration

Abstract

Synthetic high-density lipoprotein (sHDL) and rapamycin (Rap) have both been shown to be potential treatments for age-related macular degeneration (AMD). The low aqueous solubility of Rap, however, limits its therapeutic utility. Here we used an Apolipoprotein A-I mimetic peptide and phospholipid-based sHDL for the intravitreal delivery of Rap. By incorporation of Rap in sHDL nanoparticles (sHDL-Rap), we achieve 125-fold increase in drug aqueous concentration. When applied in vitro to retinal pigment epithelium cells, sHDL-Rap exhibited the abilities to efflux cholesterol, neutralize endotoxin, and suppress NF-κB activation. As an mTOR inhibitor, Rap induced autophagy and inhibited NF-κB-mediated pro-inflammatory signaling. Additionally, a greater reduction in lipofuscin accumulation and increased anti-inflammatory effects were achieved by sHDL-Rap relative to free drug or sHDL alone. In vivo studies demonstrated that sHDL reached the target retina pigment epithelium (RPE) layer following intravitreal administration in rats. These results suggest that sHDL-Rap holds potential as a treatment for AMD.

Graphical Abstract

Rapamycin loaded synthetic HDL (sHDL-Rap) was made by co-lyophilization and thermocycling, which resulted in uniform, 10 nm sized particles with the highest drug encapsulation efficiency (40%). These sHDL-Rap particles displayed little toxicity in ARPE-19 RPE cells and were able to modulate cholesterol homeostasis, autophagy and inflammation in RPE cells. sHDL nanoparticle was also able to target RPE layer in vivo following single intravitreal administration using fluorescently labelled DiO-sHDL nanoparticles in rats.

Introduction

Age-related macular degeneration (AMD) is the most common cause of visual impairment in developed countries (1). AMD is characterized by the loss of central vision, as the result of cell dysfunction and the death of cone photoreceptors in the macula. Photoreceptor degeneration is thought to be secondary to the dysfunction and death of underlying retinal pigment epithelium (RPE) cells (2). RPE cells serve three main functions in the visual cycle: 1) transport nutrients from the choroid to photoreceptors, 2) phagocytose outer segments shed from photoreceptors, and 3) recycle the visual pigment necessary for visual transduction. As an essential part of the visual cycle and photoreceptor health, dysfunction or loss of any of these processes performed by RPE cells may underlie the development of AMD (3).

Currently the only approved treatments for AMD are for the neovascular, or wet-AMD, with no treatments available for nonexudative, or dry-AMD. Given this lack of treatment options, there is a clear need to research potential therapies for dry-AMD. While it remains challenging to develop effective dry-AMD therapies, recent advances in the understanding of mechanisms underlying disease progression are providing novel avenues for treatment development. Increasing evidence has revealed that impaired autophagy, increased protein aggregation and inflammasome activation are all indicative of an AMD pathological phenotype.

For example, RPE cells phagocytose up to 10% of the photoreceptor outer segment (POS) length on a daily basis and process the vast amounts of lipids contained within these structures (3, 4). Therefore, dysfunctional RPE cells may have decreased autophagy and impaired lysosomal POS clearance. This could increase lipofuscin accumulation, leading to enhanced oxidative stress, protein aggregation, and drusen formation (5). Dysfunction in this pathway may also lead to exocytosis of damaged proteins which can activate an inflammatory response (6). Thus, therapeutic approaches regulating autophagy and suppressing chronic inflammation might offer some relief to patients by slowing or preventing AMD progression.

One such therapeutic option is rapamycin (Rap), a classic immunosuppressant approved by the FDA to prevent organ rejection following transplantation (7). Rap is a specific inhibitor of the mTOR (mechanistic target of rapamycin) protein which controls cell metabolism and cell growth through the regulation of protein synthesis and autophagy (8). Inhibition of mTOR in ARPE-19 cells leads to the activation of autophagy which increases autophagic flux and reduces lipofuscin-like aggregates (9). Rap can also inhibit NF-κB signaling, reducing inflammatory cytokine production (10, 11). Taken together, Rap has exciting therapeutic potential for the treatment and prevention of retinopathy in AMD.

Due to its broad therapeutic mechanism of action, Rap had been tested for the treatment of a variety of ocular diseases including AMD, diabetic macular edema, choroidal neovascularization and dry eye in multiple Phase 1/2 clinical trials (NCT00814944, NCT00656643, NCT00712491 and NCT00711490). However, it is difficult to safely administer therapeutic doses of rapamycin to the eye due to the extremely low aqueous solubility of the drug (~2μg/mL). Thus, in previous clinical studies Rap was administered in non-aqueous solutions containing 96% PEG 400 and 4% ethanol (12, 13). In these trials, Rap was administered by either a subconjunctival or an intravitreal injection at 0.3–1.3mg per dose. The high levels of solvents resulted in irritation of the eye as well as drug precipitation upon injection, leading to patient compliance issues and limiting the levels of Rap reaching the retina (14). Therefore, techniques to improve the solubility of Rap in an aqueous solution will greatly enhance its clinical utility. Nanoparticle-based systems for the effective delivery of therapeutically relevant doses of drugs to the retina are a promising technique for achieving this goal.

Compared to other nanocarriers, such as liposomes, micelles, and polymeric nanoparticles, synthetic high-density lipoprotein (sHDL) is uniquely positioned for the treatment of AMD (15–17). sHDL nanoparticles have been shown to enhance the removal of excess cholesterol from macrophages in atheroma, reduce vascular inflammation, and improve endothelial function in cardiovascular disease (18–20). In the context of AMD, recent studies found that intravitreal administration of the ApoA-1 mimetic peptide, 4F (Peptide sequence: Ac--DWFKAFYDKVAEKFKEAF-NH₂), reduced lipid deposition in murine Bruch’s membrane, while topical administration of sHDL reduced the size of in mice (21, 22). Reduction of lipid deposition is critical for AMD because drusen, or accumulated lipid molecules, are a hallmark of dry-AMD. Similarly, decreasing the extent of a choroidal neovascularization injury, thus decreasing the potential for vision loss, is vital for slowing the progression of AMD. Another benefit in using sHDL as the delivery vehicle is that it is readily able to incorporate water-insoluble drugs and deliver them to target tissues (15). Combining the therapeutic utility of sHDL for reducing lipid deposition via enhanced lipid export with the anti-inflammatory and pro-autophagic properties of Rap may prove to be a novel treatment method for dry-AMD.

To test this hypothesis, we present efficacy and safety data on Rap-loaded sHDL nanoparticles tested in vitro and in vivo. We designed an sHDL-Rap compound optimized to have enhanced lipid clearance, autophagic activation, and anti-inflammatory effects. The optimized lipid composition of sHDL was made by co-lyophilization and thermocycling, which resulted in uniform, 10nm sized particles. We found the optimal drug encapsulation efficiency to be 40%. These sHDL-Rap particles displayed very low toxicity in ARPE-19 cells (>90% ARPE-19 viability) and were able to modulate cholesterol homeostasis, autophagy and inflammation in RPE cells (Schematic Design). We also showed that sHDL is targeted to RPE in vivo following single intravitreal (IVT) administration using fluorescently labeled DiO-sHDL nanoparticles in rats.

Schematic design:

Rapamycin-loaded synthetic HDL (sHDL-Rap) delivered to the retina, inducing autophagy, removing excess cholesterol, and resolving inflammation through synergistic actions.

Methods and Methods

Materials

22A peptide (PVLDLFRELLNELLEALKQKLK) was custom synthesized by GenScript (Piscataway, NJ). Rapamycin was purchased from Cayman (Ann Arbor, MI). Phospholipids, 1,2-dimyristoyl-sn-glycero-3-phosphocholine (DMPC), egg sphingomyelin (SM), 1,2-dipalmitoyl-sn-glycero-3-phosphocholine (DPPC), 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), and 1,2-dioleoyl-sn-glycero-3-phosphocholine (DOPC) were purchased from Avanti Polar Lipids (Alabaster, AL) and NOF America Corporation (White Plains, NY). Fluorescent label, 3,3′-dioctadecyloxacarbocyanine perchlorate (DiO), was purchased from Invitrogen (Carlsbad, CA). All other reagents were analytical grade and were purchased from commercial suppliers.

Preparation and characterizations of rapamycin-loaded sHDL (sHDL-Rap)

sHDL nanoparticles were prepared by co-lyophilization followed by thermocycling (16). Briefly, 22A peptide and various phospholipids (DMPC, DPPC, POPC or DOPC) were dissolved in about 1mL acetic acid at a 22A:lipid ratio of 1:2 by weight (Equivalent molar ratios are presented in Table 1). Rapamycin was dissolved in methanol as a stock solution at 10mg/mL. Components (lipid, 22A, rapamycin) were then mixed together and subjected to lyophilization overnight to remove the organic solvent. sHDL-DiO was generated by adding 10μg DiO per 1mg peptide directly into the acetic acid mixture of peptide and DMPC. The lyophilized powder was hydrated with PBS, vortexed and subjected to heating/cooling cycles to form a clear, homogeneous solution of sHDL nanoparticles.

Table 1.

The effect of phospholipid composition and rapamycin loading on sHDL size and drug encapsulation efficiency

Lipid (mg)	Peptide (mg)	Mole ratio (peptide: lipid)	Rapamycin (μg)	EE%	Particle size (nm)	Appearance
DMPC (10)	22A (5)	1: 7.15	500	43.4 ± 2.3	9.1. ± 0.2	clear
			100	43.4 ± 1.1	9.8 ± 0.2	clear
			25	38.4 ± 2.7	9.8 ± 0.2	clear
SM (10)	22A (5)	1: 7.38	100	23.1 ± 1.8	11.0 ± 0.3	clear
DPPC (10)	22A (5)	1: 7.15	100	23.5 ± 1.8	10.1 ± 0.2	clear
POPC (10)	22A (5)	1: 6.90	100	43.5 ± 2.6	18.8 ± 0.6	cloudy
DOPC (10)	22A (5)	1: 7.15	100	31.8 ± 1.8	20.6 ± 1.2	cloudy

Size distribution was determined by dynamic light scattering (DLS) on a Malvern Zeta sizer (Westborough, MA). Particle purity was determined by gel permeation chromatography (GPC) with UV detection at 220 nm using a Tosoh TSK gel column (Tosoh Bioscience, King of Prussia, PA) separated by a Waters HPLC, as previously described (16).

Encapsulation efficiency (EE) and drug-loading (DL) capacities were determined after removing un-encapsulated rapamycin using Zeba Spin Desalting columns (5 mL7K MWCO, Thermo Fisher). The sHDL-Rap nanoparticles collected before and after desalting were analyzed by UPLC detection for drug content. The UPLC measurement was carried out with an Acquity UPLC BEH C18 column (1.7μm, 2.1×100mm, Waters, Milford, MA, USA) at 40°C and the mobile phase was methanol:water (with 0.1% formic acid) at a ratio of 80:20 at 0.3mL/min with an injection volume of 10μL. The concentration of rapamycin was detected by UV absorbance at a wavelength of 278 nm. EE and DL were calculated according to the following formulas: EE (%) = rapamycin encapsulated in sHDL/total amount of rapamycin×100 %, DL (%) = Theoretical loading × EE%.

For stability evaluations, sHDL-Rap (rapamycin 100μg/mL) and Rap solution (10μg/mL) were incubated in PBS (pH 7.4) at 37°C for 72 hours. The particle size of sHDL-Rap at different time points was measured by DLS. Undegraded Rap contents in rapamycin solution or sHDL-Rap were analyzed by UPLC as described above. In a parallel experiment, at different time points during incubation, sHDL-Rap was passed through Zebra Spin Desalting Columns (MWCO 7kD) to remove unencapsulated rapamycin. The encapsulated intact Rap was then quantified using UPLC as described above to calculate drug retention percentage.

Culture of RPE cells

ARPE-19 cell line was purchased from ATCC (CRL-2302, Manassas, VA) and routinely maintained in DMEM/F12 medium supplemented with 10% FBS (Gibco, Grand Island, NY). Highly-polarized RPE monolayers were established by culturing in DMEM/F12 medium for 1 month prior to experiments. For transepithelial electrical resistance (TEER) measurements, cells were grown on Transwell^® cell culture inserts (6.5mm diameter, 0.4mm pore size, polyester, Corning, Tewksbury, MA, USA). TEER was measured using the Millicell^® ERS-2 voltohmmeter (MilliporeSigma).

Cytotoxicity assay

A cytotoxicity assay was conducted using slight deviations from the existing MTT assays (23). Briefly, ARPE-19 cells were seeded on a 96-well plate at a density of 1×10⁴/per well. After 1 month of culture, cells were incubated with sHDL, Rap or sHDL-Rap at different concentrations for 24 hours. MTT solution (Sigma, 5mg/mL in PBS) was added into each well and incubated for 4 hours. The media was removed, and cells were dissolved in dimethyl sulfoxide. Absorbance was measured at 550 nm using a microplate reader (SynergyTM NEO HTS Multi-Mode Microplate Reader, Bio-Tek). PBS-treated cells were set as the baseline for 100% viability.

Cholesterol efflux assay

Phenol red-free media was used to avoid any interference with physiological cholesterol pathways. RPE cells were seeded in 24-well transwell insert with 0.4 μm pore size (Corning, NY) and cultured for one month (transepithelial electrical resistance (TEER) value reached around 50Ω/cm²). They were then labeled with [1,2-³H]-cholesterol at a concentration of 1μCi/mL (Perkin Elmer) in DMEM containing 0.5% fatty acid-free bovine serum albumin for 24 hours. Cells were then washed twice with PBS and equilibrated in fresh DMEM containing 0.5% fatty acid-free bovine serum albumin supplemented with 5μg/ml of the ACAT inhibitor Sandoz 58–035 (Sigma) for 24 hours. Rapamycin (1μg/mL), sHDL (50μg/mL) or sHDL-Rap (1μg/mL of Rap and 50μg/mL of sHDL) was then added into the medium for 16 hours. To measure the cholesterol efflux, the media were collected and centrifuged to remove cell debris while the cells were lysed with 0.1% SDS/0.1M NaOH solution for 2 hours. Radioactivity in the media and cell lysate were counted on a PerkinElmer scintillation counter. Cholesterol efflux in the media was expressed as a percentage of total cell [³H]-cholesterol content.

Western blot

After different treatments, ARPE-19 cells were washed twice with cold PBS and lysed in RIPA buffer containing an EDTA-free protease inhibitor cocktail (Sigma) and kept on ice for 30 min. Protein concentration in the supernatant was measured with the Bradford assay reagent (BioRad, Cressier, Switzerland). Protein lysates (50μg/lane) were separated by SDS-PAGE then transferred to PVDF membranes. Membranes were blocked in 5% non-fat blocking milk (BioRad) for 1 hour at room temperature and incubated overnight with the following primary antibodies: rabbit anti-LC3 (CST 4108S, 1:1000 dilution), anti-P65 (CST 8242S, 1:1000 dilution) and anti-GAPDH (CST 5174S, 1:1000 dilution). After washing, membranes were incubated with appropriate HRP conjugated secondary antibodies and their signal was developed using an enhanced chemiluminescence (ECL, Amersham Biosciences) substrate. Images were acquired on a Protein Simple FluorChem M imaging system (San Jose, CA). Intensity of bands was quantified using ImageJ and normalized to GAPDH levels.

Lipofuscin images by confocal microscope

For confocal microscopy imaging, ARPE-19 cells were cultured on gelatin-coated coverslips in a 6-well plate and treated with POS (4μg protein per cm² cell growth area, Invision Bioresources, Seattle, WA) daily for 7 days. Cells were also treated with Rap (1μg/mL), sHDL (50μg/mL) or sHDL-Rap (1μg/mL of Rap) on days 6 and 7. Cells were then washed with PBS to remove non-internalized POS, fixed with 4% paraformaldehyde, and nuclei were stained with 5μg/mL DAPI. Finally, cells were imaged by a Nikon A-1 confocal microscope.

Anti-inflammation effects

Confluent ARPE-19 cells were treated with PBS, Rap, sHDL or sHDL-Rap for 18 hours, followed by lipopolysaccharide (LPS, 10μg/ml; Sigma–Aldrich) and 4-hydroxynonenal (HNE, 15μM; Sigma–Aldrich) for 24 hours. Media samples were collected, and the concentrations of soluble pro-inflammatory cytokines IL-6, TGF-β, and IL-1β were measured in triplicate using commercial ELISA kits (Bio-legend) according to the manufacturer’s protocols.

Biodistribution in vivo

The biodistribution of sHDL following intravitreal (IVT) injection was examined using 3,3′-dioctadecyloxacarbocyanine perchlorate (DiO)-loaded sHDL nanoparticles. Male, Sprague-Dawley rats (age ~2–3 months) were purchased from Charles River (Wilmington, MA, USA). 3–4 animals per group were used. 2μL of DiO-sHDL was injected into the left eye of each animal and free DiO dissolved in DMSO (DMSO-DiO) was injected into the right eye. Rats were euthanized 24 hours and 72 hours post-injection, and eyes were enucleated and immediately flash frozen in OCT (optimal cutting temperature) media. 10μm thick sections were prepared and coverslipped using ProLong^™ Gold Antifade Mountant with DAPI (Thermo Fisher Scientific, Waltham, MA, USA Cat# P36930) and DiO fluorescence in the outer retina was evaluated using a Leica DM6000 wide field fluorescent microscope with a 40x objective. Additional frozen sections were prepared and subjected to H&E staining as previously described (24).

Statistical analysis

All data are presented as mean ± standard deviation. Statistical comparisons were performed by one-way ANOVA for multiple groups. p < 0.05, 0.01, and 0.001 were considered statistically significant and marked with *, **, ***, respectively.

Results

Optimizing sHDL nanoparticles for rapamycin delivery

In this study, we made sHDL consisting of the ApoA-I mimetic peptide, 22A, and phosphatidylcholines that were previously utilized by our lab to deliver a variety of drugs (16, 25, 26). This ApoA-I mimetic peptide retains many biological functions of endogenous ApoA-I protein, easily forms homogeneous sHDL nanoparticles, and demonstrated favorable safety and pharmacokinetic profiles in human clinical trials (27, 28). The selection of phospholipid is critical as fatty acid chain lengths and degrees of saturation impact the fluidity of the sHDL membrane, which, in turn, may affect the efficiency of drug molecule incorporation (17). We developed a panel of sHDL formulations containing the 22A peptide with various lipids to optimize the loading of Rap (Fig. 1A). Phospholipids with low transition temperature (T_m) (for example, DOPC with T_m = −17°C and POPC with T_m = −2°C, respectively) made sHDL suspensions murky, whereas phospholipids with high T_m (for example, DPPC and DMPC with T_m of 41°C and 24°C, respectively) formed clear sHDL solutions. sHDL nanoparticles with clear solutions had an average diameter of 10nm, as determined by dynamic light scattering (Table 1). Transmission electron micrographs of sHDL particles containing DMPC phospholipids also showed uniform spherical shapes (Fig. 1B) with a purity >95% as evidenced by GPC (Fig. 1C). To determine encapsulation efficiency of Rap, unencapsulated rapamycin was removed via a desalting column, and sHDL-Rap particles were assayed. UPLC was used to quantify encapsulation efficiency. The highest rapamycin encapsulation efficiency was achieved using DMPC-based sHDL (43%), followed by SM (23%) and DPPC (23%) based sHDL. DMPC-sHDL encapsulation efficiency did not change significantly when generating sHDL-Rap particles using 25, 100 or 500μg/ml or rapamycin, showing ample capacity to incorporate rapamycin. When compared to the low aqueous solubility of rapamycin (2μg/mL), incorporation of rapamycin into sHDL showed an effective solubility of 250μg/mL resulting in 125-fold improvement in solubility without using organic solvents. Based on sHDL appearance, particle size homogeneity and encapsulation efficiency, we chose DMPC as the lipid component of our sHDL-Rap formulation for further investigation.

Figure 1.

Schematic representation of the sHDL nanoparticle preparation and characteristics (A) Schematic representation of protocol for generating rapamycin-encapsulated sHDL nanoparticles. These particles are comprised of DMPC, 22A peptide and rapamycin. (B) Representative TEM image of sHDL-Rap nanoparticles. The scale bar is set to 20 nm. (C) Size distribution measured by DLS. (D) The purity analysis analyzed by GPC. (E) Particle sizes of sHDL-Rap at different time points under 37°C incubation. (F) Percentage of the intact rapamycin in sHDL-Rap and rapamycin PBS solution. (G) Percentage of encapsulated and intact rapamycin in sHDL-Rap. Data are reported as mean ± SD, n = 3.

The stability of optimized sHDL-Rap was evaluated in terms of particle size and chemical stability of rapamycin. As shown in Fig 1E, the particle size of sHDL-Rap remained stable in prolonged incubation under 37°C without signs of aggregation. As rapamycin is prone to degradation under physiological pH (29), the chemical stability of rapamycin in sHDL-Rap was evaluated in comparison with free rapamycin solution. Results showed that less than 20% rapamycin remained undegraded in PBS solution within 24 hours. However, the chemical stability of rapamycin was greatly increased in sHDL-Rap formulation, with ~80% rapamycin, either encapsulated in sHDL particles or released as free form, staying intact at 24 hour time point (Fig 1F). The percentage of the intact rapamycin encapsulated in sHDL-Rap was determined after removing released drug molecules using desalting columns. Results showed that around 90% drug was released or degraded within 24 hours of incubation, and a complete drug release/degradation was observed in the 72 hour time point (Fig 1G).

The effect of sHDL-Rap on cell viability and cholesterol efflux

We next tested for any cytotoxicity of our sHDL and sHDL-Rap formulations in RPE-like cells (ARPE-19). sHDL has been widely tested in various cell types and is considered nontoxic over a range of concentrations (16, 30). In ARPE-19 cells, drug-free DMPC-based sHDLs show little cytotoxicity at all concentrations tested (10–100μg/mL of 22A peptide) (Fig. 2A). Treatment with rapamycin displayed clear cytotoxicity in ARPE-19 cells at the concentration of 10μg/mL (73.01% cell viability ). Concentrations of rapamycin lower than 5μg/mL exhibited no significant toxic effects on ARPE-19 cells. (Fig. 2B). Similar results were obtained by treating cells with sHDL-Rap. We chose an effect-concentration of 1μg/mL for rapamycin in the following in vitro studies based on previously published research (31–33). Recent studies have suggested that the pathophysiology of AMD is related to the dysregulation of cholesterol metabolism (34–36). In order to evaluate the ability of sHDL-Rap particles to enhance cholesterol efflux, ARPE-19 cells were assayed for their ability to retain labeled cholesterol (Fig. 2C). Rap alone had no effect on cholesterol efflux as expected. More importantly, incorporation of Rap into sHDL did not attenuate sHDL’s efflux capacity. sHDL and sHDL-Rap exhibited similar cholesterol efflux in RPE cells (~40%) (Fig. 2D). This efflux capacity was similar to values shown for sHDL treated atheroma macrophages (16).

Figure 2.

In vitro cell viability and cholesterol efflux assays (A) MTT assay on ARPE-19 cells treated with sHDL for 24 hours. ARPE-19 cells displayed no significant decrease in cell viability when treated with a wide range of sHDL concentrations. (B) MTT assay on ARPE-19 cells treated with Rap or sHDL-Rap for 24 hours. (C) Schematic representation of the cholesterol efflux assay. (D) sHDL and sHDL-Rap induced significant efflux of [3H]-cholesterol efflux from ARPE-19. Radioactivity counts were measured in cell lysate and media. Calculated % efflux = (media counts)/(media + cell counts)*100. Data is reported as mean ± SEM, n = 3.

Effects of Autophagy Induction or Inhibition on RPE Cell Autofluorescence

Recent evidence indicates that decreased autophagic flux is associated with RPE impairment and AMD pathology (37, 38). Autophagy decline in RPE cells can result in the accumulation of aggregation-prone proteins, cellular degeneration and ultimately cell death. This presents itself in the form of accumulated auto-fluorescent lipid-protein aggregates, called lipofuscin, in the lysosomes of RPE cells (6). Thus, regulation of autophagy signaling pathways is a potential therapeutic target for AMD treatment. The signaling of mechanistic target of rapamycin (mTOR) has proven to be the key regulator for autophagy activity. Recent findings demonstrated that rapamycin can inhibit mTOR complex 1 (mTORC1), and, in turn, activate autophagy and slow the aging and neurodegenerative processes in mice (39, 40). A commonly accepted criterion to estimate autophagic flux is to measure the ubiquitin-like process. The microtubule-associated protein 1 light chain (LC3-I) will notably become LC3-II after lipidation. LC3-II is then inserted into the inner and outer membranes of the autophagosomes and eventually degraded (41). We measured activation of autophagy after treating ARPE-19 cells with Rap, sHDL, or sHDL-Rap. We used the ratio of LC3-II relative to LC3-I as a quantitative index of autophagy—with an increased ratio representing autophagy activation. Our results show that ARPE-19 cells displayed strong activation of autophagy after rapamycin treatment (Fig. 3A), and displayed a similar response following treatment with sHDL-Rap, suggesting that encapsulation in an sHDL nanocarrier did not alter the effect of rapamycin. sHDL alone had little effect on autophagic flux (Fig. 3A).

Figure 3.

Treatment with sHDL-Rap induces autophagy and reduces lipofuscin formation. (A) Western blot results for the marker of autophagy-microtubule associated protein 1 light chain 3 (LC3). Upper band: LC3-I (cytoplasmic form); lower band: LC3-II (autophagosomal form). ARPE-19 cells were treated for 24 hours with PBS, Rap (1μg/mL), sHDL, or sHDL-Rap (1μg/mL). LC3-II/LC3-I ratios were calculated (mean ± SEM, n = 3, *** p < 0.001). (C) ARPE-19 cells were examined by confocal microscopy following 7 days of incubation with: PBS without POS, PBS with POS, Rap (1μg/mL) with POS, sHDL with POS, or sHDL-Rap (1μg/mL) with POS. Lipofuscin-like cellular autofluorescence (green) was detected using a confocal microscopy (excitation, 480nm; detection, 535nm). Nuclei were visualized by DAPI staining (blue) (Scale bar: 20μm). Fluorescence intensity levels were quantified by Image J (B).

Autophagy is critical for the breakdown of photoreceptor outer segments (POS) and a loss of this breakdown can lead to the accumulation of lipofuscin, an early indicator of AMD pathology. ARPE-19 cells were incubated daily with POS for 7 days and cellular autofluorescence of lipofuscin was detected by confocal microscopy (excitation, 480nm; detection, 535nm). POS induced significant formation of lipofuscin, characterized by autofluorescence (green dots, Fig. 3B). A significant reduction of autofluorescence was observed in Rap groups. This suggests that rapamycin suppresses mTOR, which stimulates autophagosome activity, leading to a more complete degradation of material, resulting in decreased lipofuscin accumulation (9). sHDL alone also showed a significant reduction of lipofuscin formation. Between this study and others, sHDL has shown its ability to remove excess cholesterol and photoreceptor outer segment-derived cholesterol (42). However, further experiments are required to study the physiological role and regulation of sHDL in lipofuscin removal. The sHDL-Rap group here had the most noticeable decrease in fluorescence. Our data fit well with the hypothesis that sHDL-Rap has the strongest ability to inhibit lipofuscin formation.

sHDL-Rap suppresses inflammatory cytokines in ARPE-19 cells

We next evaluated the anti-inflammation effects of sHDL-Rap in ARPE-19 cells. Phosphorylated NF-κB p65, a transcription factor, is generally upregulated during inflammation. After incubation with LPS and 4-hydroxynonenal (HNE), p65 protein levels increased significantly, but this effect was inhibited by treatment with Rap (Fig. 4A). sHDL alone was also capable of reducing p65 protein levels after an inflammatory event, consistent with a previous study in macrophages (43). Interestingly, sHDL-Rap further reduces p65 protein levels after treatment with LPS & HNE, suggesting a synergistic effect. Similarly, the levels of the inflammatory cytokines IL-6 and IL-1β were increased after LPS and HNE incubation, but were attenuated by rapamycin (Fig. 4B, 4C). Both rapamycin and sHDL exhibited drastic reductions in IL-6 and IL-1β compared to PBS controls. As expected, sHDL-Rap displayed superior anti-inflammatory effects compared to rapamycin or sHDL alone. Taken together, these results suggest that sHDL is an effective carrier of rapamycin, retaining the anti-inflammatory effects of both rapamycin and sHDL, allowing for better protective effects in vitro.

Figure 4.

sHDL-Rap is effective at preventing pro-inflammatory signaling. (A) NF-κB subunit p65 protein expression in ARPE-19 cells was measured by western blot. Cells were treated with PBS, Rap, sHDL or sHDL-Rap for 18 hours, followed by LPS+HNE for 24 hours. “No treat” indicates cells which received no treatment of sHDLs or LPS+HNE. (B, C) Pro-inflammatory cytokines (IL-1β, IL-6) were measured by ELISA after treatments as previously described (mean ± SEM, n = 3, *p < 0.05, **p < 0.01, ***p < 0.001 compared to PBS, # p < 0.05 compared to sHDL-Rap).

Intraocular distribution of sHDL

The target tissues for sHDL-Rap are in the posterior segment of the eye, and nanoparticles will likely need to be administered by the intravitreal route. Intravitreal injection is routinely used in clinical practice for the administration of anti-VEGF, and in clinical trials for the administration of organic solvent solutions of Rap. The ability of sHDL-Rap to target RPE cells following intravitreal administration was examined by measuring the biodistribution using fluorescently labeled DiO-sHDL nanoparticles in Sprague-Dawley rats. At 24 hours post-injection, DiO labeled sHDLs diffused rapidly from the vitreous to the retina and accumulated within the RPE layer, indicating transretinal movement of the sHDL, whereas DiO delivered in a solution of DMSO did not efficiently target the outer retina or RPE (Fig. 5A). The intensity of the green fluorescence signal in the retina 72 hours post-injection was much lower compared to that of 24 hours post-injection, indicating the elimination of intra-retinal sHDL nanoparticles (Fig. 5A). In addition, there was no observed histological evidence of toxicity at 24 and 72 hours post-dose following IVT administration of sHDL particles (Fig. 5B). These results indicate that sHDL are capable of efficiently and safely delivering drug cargo to RPE cells following IVT administration. This accumulation of sHDL nanoparticles in the posterior segment of the eye is superior to the delivery of the hydrophobic drug by organic solution (DMSO) injection known to elicit local retinal toxicity (44).

Figure 5.

Successful delivery of sHDL nanoparticles to the retina and RPE. (A) Representative images of frozen sections show fluorescent signals in the outer retina and RPE layers of sHDL-DiO at 24 hours post injection. The fluorescent signal from DiO is mostly absent by 72 hours post injection. (B) H&E staining of frozen sections shows no gross morphological defects associated with sHDL-DiO injection. DMSO – Dimethylsulfoxide, sHDL – synthetic High Density Lipoprotein, ONL – Outer Nuclear Layer, IS/OS – Inner Segment/Outer Segment region, RPE – Retinal Pigment Epithelium. 3–4 animals were included per group.

Discussion

In this work, we have optimized the composition of sHDL for the incorporation of Rap. The highest encapsulation efficiency of Rap (~40%) was achieved by preparing sHDL particles using DMPC. sHDL-Rap nanoparticles, consisting of 22A peptide and DMPC, contain 0.27% (w/w) rapamycin. These sHDL-Rap nanoparticles showed homogeneous shape and were approximately 10nm in diameter. The small and stable particle size of sHDL-Rap is particular favorable to achieving efficient retina delivery, as previous studies have suggested that nanoparticles with small particle sizes diffuse faster within the vitreous and have higher efficiency in retina penetration (45, 46). Encapsulating rapamycin to sHDL particles significantly increased the effective aqueous solubility of Rap and greatly decreased the degradation of rapamycin in physiological pH, which improves the bioavailability of Rap after intravitreal injection. We have demonstrated that sHDL-Rap are effective at inducing cholesterol efflux from ARPE-19 cells and are more effective than Rap alone at inducing autophagy to clear POS, preventing the accumulation of lipofuscin. This, combined with the enhanced solubility of sHDL-Rap in aqueous solutions, creates a promising new delivery mechanism for drugs to the retina.

The advantage of our approach is the utilization of the dual functionality of sHDL. It serves as the nanocarrier to deliver a hydrophobic drug, in this case Rap, and as the cholesterol acceptor, removing excess cholesterol from target tissues (16, 47). One of the key features of dry-AMD is the accumulation of drusen on the basal side of RPE cells. These deposits are rich in lipids, making a therapeutic system which can aid in the clearance of these deposits very useful. In addition, retinal macrophages and monocytes display lower cholesterol clearance capacities, exacerbating the progression of AMD (48, 49). We first demonstrated that sHDL alone enhances the efflux of radiolabeled cholesterol from ARPE-19 cells (Fig. 2) and that loading sHDL nanoparticles with Rap did not inhibit this effect. Other studies also demonstrated that both sHDL and ApoA-1 mimetic peptides can reduce lipid deposition under the RPE (50, 51). Federica et al. demonstrated that in human primary RPE cells, ³H-cholesterol-labeled POS was effluxed to ApoA-I (52), although the mechanisms underlying this observation require further study. The dysfunction of proper cholesterol clearance from RPE is a driver of dry-AMD pathology, therefore restoring this function to RPE may be very beneficial for patients (53, 54).

We next investigated the regulatory function of Rap in autophagy signaling pathways. Impaired autophagy activity in senescent cells is related to numerous diseases, including neurodegenerative disorders, heart diseases and cancer (55, 56). Therefore, enhancing autophagy is expected to be beneficial to AMD therapy. Our studies indicate that treatment of ARPE-19 cells with Rap alone or sHDL-Rap can effectively induce autophagy (Fig. 3). The stimulation of autophagy by either Rap or sHDL-encapsulated Rap leads to a decrease in lipofuscin accumulation. These results demonstrate that the pharmacological efficacy of Rap is not lost after encapsulation in sHDL. Interestingly, sHDL alone also showed a relatively strong ability to decrease lipofuscin accumulation without activating autophagy. This could be explained by the ability of sHDL to efflux the outer segment (OS)-derived cholesterol. Thus, sHDL alone is capable of increasing lipid efflux from ARPE-19 cells, while the addition of Rap facilitates intracellular autophagy activity, followed by lipid clearance, thus decreasing overall cellular lipid accumulation. While additional mechanistic studies are required to elucidate individual contributions of both the sHDL carrier and Rap to overall sHDL-Rap efficacy, these results further highlight the advantage of using therapeutically active lipoprotein carriers relative to other traditionally used polymer or lipid-based nanocarriers. Lastly, sHDL-Rap was chosen for AMD treatment due to the anti-inflammatory potential of both sHDL and Rap. An inflammatory response may further accelerate AMD pathology and can occur when autophagic activity drops. In conjunction with this response, there is an increase in lipofuscin accumulation, leading to elevated protein aggregation and ROS production (57). Similarly, the induction of chronic inflammation can accelerate deficiency of autophagy and AMD progression (58). In addition to its role in autophagy, Rap has also been shown to suppress locally activated mTOR levels and related inflammatory molecules (IL-6, MCP-1) through NF-κB suppression. The anti-inflammatory functions of sHDL have been widely studied. These functions include neutralization of endotoxins and reduction of TLR4 recruitment to the lipid raft (59), inhibiting the activation of NF-κB. Furthermore, HDL can activate transcription factor 3 (ATF3) expression to suppress TLR-induced pro-inflammatory cytokine release (60). Our data show that sHDL-Rap exhibited superior anti-inflammatory activity in RPE cells compared to either Rap or sHDL alone, suggesting a synergistic effect.

Various other nanocarriers for ophthalmic delivery of Rap gained considerable attention due to the poor aqueous solubility of Rap and the toxicity of solvent formulations (61–63). MPEG-PCL micelles loaded with Rap increased drug stability and enhanced bioavailability following systemic administration (61). When this drug was delivered intravitreally in a model of experimental autoimmune uveitis, intraocular inflammation was abolished via the downregulation of the Th1 and Th17 response. Another liposomal-based nanoparticle formulation containing Rap (0.4mg/mL and 1mg/mL) was developed, and shown to improve dry eye pathology in a keratoconjunctivitis sicca dog model (62). These nanoparticle-rapamycin formulations show potential for the treatment of dry eye and ocular inflammation; however, they do not involve using a therapeutically active carrier like sHDL.

Beyond the inherent advantages of being a nanocarrier, sHDL has great potential as a therapeutic delivery option due to its other biological activities. sHDL particles possess unique physicochemical properties, including naturally synthesized components, amphipathic apolipoproteins, lipid-loading, hydrophobic agent-incorporating characteristics, and specific receptor-mediated uptake. Multiple sHDL products, CER-001, CSL-112, ETC-216 and ETC-642, have been proven safe and well-tolerated in both healthy and diseased human patients (27, 64, 65). Moreover, their manufacture under current good manufacturing practices (cGMP) is established and these processes could be easily adapted to incorporate rapamycin (15, 66). In addition, sHDL has been shown to facilitate cholesterol efflux and anti-inflammation efficacy in our previous studies (16, 17, 43), indicating the advantages of using it as a treatment for AMD.

In summary, here we applied a therapeutically active carrier, sHDL, for the delivery of a hydrophobic drug, rapamycin, to the back of the eye. Our future studies will involve the examination of sHDL-Rap’s ability to reduce the pathology of AMD in animal models. The combination of sHDL’s potential therapeutic utility in lipid dysregulation and its ability to effectively reach the posterior segment of the eye following IVT administration might be translatable in the delivery of other hydrophobic ocular therapeutics.