In vitro release characteristics and cellular uptake of poly(D,L-lactic-co-glycolic acid) nanoparticles for topical delivery of antisense oligodeoxynucleotides

Abstract

The efficacy of antisense oligodeoxynucleotides (AsODNs) is compromised by their poor stability in biological fluids and the inefficient cellular uptake due to their size and negative charge. Since chemical modifications of these molecules have resulted in a number of non-antisense activities, incorporation into particulate delivery systems has offered a promising alternative. The aim of this study was to evaluate various poly(D,L-lactic-co-glycolic acid) (PLGA) nanoparticles for AsODN entrapment and delivery. PLGA nanoparticles were prepared using the double emulsion solvent evaporation method. The influence of formulation parameters such as PLGA concentration and volume ratio of internal aqueous phase volume (Va1) to organic phase volume (Vo) to external aqueous phase volume (Va2) on particle size, polydispersity index (PDI) and zeta potential (ZP) was investigated using a full factorial study. The particle size increased with increasing PLGA concentrations and volume ratios, with an interaction detectable between the two factors. AsODN entrapment efficiencies ranged between 49.97% and 54.95% with no significant difference between various formulations. By fitting the in vitro release profiles to a dual first order release model it was shown that the AsODN release occurred via two processes: a diffusion controlled process in the early phase (25 to 32% within one day) and a PLGA degradation process in the latter (39 to 70% after 14 days). Cellular uptake studies using primary corneal epithelial cells suggested active transport of nanoparticles via endocytosis. PLGA nanoparticles therefore show potential to successfully entrap AsODNs, transport them into cells and release them over time due to polymer erosion.

Keywords::

• PLGA nanoparticles
• antisense oligodeoxynucleotides
• in vitro release
• cellular uptake
• double emulsion solvent evaporation method

Abbreviations
PLGA,	=	poly(D,L-lactic-co-glycolic acid);
AsODN(s),	=	antisense oligodeoxynucleotide(s);
Cx,	=	connexin;
PDI,	=	polydispersity index;
ZP,	=	zeta potential;
EE,	=	entrapment efficiency

Introduction

Connexin 31.1 (Cx31.1) protein is predominantly expressed in the corneal epithelium and the skin epidermis (Chang et al., 2009; Goliger and Paul, 1994; Hennemann et al., 1992). It leads to formation of non-functional gap junction channels between cells and is therefore associated with cell apoptosis. It has previously been shown that knockdown of Cx31.1 in the cornea reduced apoptosis in superficial cell layers by 72% and therefore led to thickening of the corneal epithelium (Chang et al., 2009), which may be useful in the treatment of keratoconus. Knockdown of Cx31.1 in the epidermis of the skin, on the other hand, could result in skin thickening and rejuvenation, which may be of potential interest for cosmetic purposes.

Due to the increasing number of diseases associated with inappropriate or inadequate protein production, there has been a remarkable growth of nucleic acid-based therapeutics. Antisense oligodeoxynucleotides (AsODNs) typically consist of 15 to 30 base pairs and hybridize with the complementary mRNA in a sequence-specific manner through Watson-Crick base pairing to block the translation into the target protein (Zon, 1988). This can occur via two main mechanisms: activation of RNase H (Crooke, 1998; Leonetti et al., 1993) or translational arrest (Stein and Cheng, 1993). The main problems associated with these molecules include their poor stability and inefficient cellular uptake due to the size and charge of the molecules. A number of chemically modified AsODNs have been trialled over recent years (Chen et al., 2005) and even though modifications have provided an improvement in stability and cell penetration, they have also resulted in a variety of non-antisense activities. Recent approaches therefore focus on the development of appropriate delivery systems in order to improve the stability and cellular uptake of the AsODNs (Akhtar et al., 2000; Dokka and Rojanasakul, 2000; Fattal and Bochot, 2008; Remaut et al., 2010; Toub et al., 2006).

Non-viral delivery systems based on biocompatible polymers are preferred due to their safety, stability, relative ease of large-scale production and lack of intrinsic immunogenicity (Li and Huang, 2000). One of the most commonly used polymers is the biodegradable and biocompatible poly(D,L-lactic-co-glycolic acid) (PLGA), which has been approved by the Food and Drug Administration (FDA) for human use in therapeutic devices. PLGA nanoparticles have been extensively investigated for sustained and targeted or localized delivery of various agents such as AsODNs (Aukunuru et al., 2003; Nafee et al., 2007), DNA (Ravi Kumar et al., 2004) as well as protein and peptide drugs (Elamanchili, 2007). They have shown great efficiency as delivery vehicles by increasing the amount of drug crossing the various biological barriers as well as protecting the molecules from enzymatic degradation.

This study evaluated the use of PLGA nanoparticles prepared by the double emulsion solvent evaporation method for the delivery of Cx31.1 AsODNs. The influence of formulation parameters such as PLGA concentration and volume ratio of internal aqueous phase volume (Va1) to organic phase volume (Vo) to external aqueous phase volume (Va2) on particle size, polydispersity index (PDI) and zeta potential (ZP) was investigated using a full factorial study. Moreover, the effect of these physical parameters on the entrapment efficiency as well as the release kinetics of the AsODNs from the nanoparticles was evaluated. Finally, the cellular uptake mechanism of the AsODN-loaded PLGA nanoparticles was determined.

Materials and Methods

Materials

Poly(D,L-lactic-co-glycolic acid) (50:50, MW 13,600, inherent viscosity 0.19 dl/g) was a generous gift from Transo-Pharm (Siek, Germany). Polyvinyl alcohol (PVA, MW 130,000, hydrolysis degree 95–97%) from Ajax Finechem (Auckland, New Zealand) was used as an emulsifying agent. Analytical grade dichloromethane (DCM) was purchased from Scharlau (Barcelona, Spain). Cy3-labelled Cx31.1 AsODNs (5′ Cy3-AGA GGC GCA CGT GAG ACA C 3′, MW 6379.4 g/mol) and Cy3-labelled Cx43 AsODNs (5′ Cy3-GTA ATT GCG GCA GGG GGA ATT GTT TCT GTC 3′, MW 9815.4 g/mol) were purchased from Sigma (St. Louis, MO, US). Primary corneal epithelial cells were derived from human donor tissue (New Zealand National Eye Bank, Auckland, New Zealand). Opti-MEM I, reduced serum medium modification of Eagle’s Minimal Essential Medium (MEM), fetal calf serum (FCS), penicillin-streptomycin-L-glutamine, trypsin (0.25%)-EDTA (1 mM) and phosphate saline buffer (PBS) were purchased from Gibco Laboratories (Grand Island, NY, US). Sterile T75 flasks, 8-well slide chambers, centrifuge tubes, pipettes and tips were purchased from BD Falcon (Auckland, New Zealand). Paraformaldehyde was obtained from Ajax Finechem (Auckland, New Zealand). 4′,6-diamidino-2-phenylindole (DAPI) and wheat germ agglutinin (WGA) Alexa 488 were purchased from Sigma (St. Louis, MO, US). Mounting medium (CitiFluor anti-fading) was purchased from Agar Scientific (Stansted, UK). Water used throughout the experiments was ion-exchanged, distilled and passed through a Milli-Q water purification system (Millipore, Rockland County, NY, US).

Experimental design

In order to evaluate the effect of PLGA concentration and volume ratio of internal aqueous phase volume (Va1) to organic phase volume (Vo) to external aqueous phase volume (Va2) (independent variables) on particle size, PDI and ZP of blank nanoparticles (dependent variables), and to see whether there was an interaction between these two parameters, factorial studies were designed. The first study used a complete factorial design and included the independent variables PLGA concentration at three levels (2.5, 3 and 3.5 % w/v) and volume ratio at two levels (Va1:Vo:Va2 = 1:10:20 and 1:20:50). The second study also used a complete factorial design and evaluated the effect of storage time at four levels (0, 3, 8 and 15 days) on PLGA concentration at two levels (3 and 3.5 % w/v) and volume ratio at two levels (Va1:Vo:Va2 = 1:10:20 and 1:20:50). Finally, a partial factorial study assessed entrapment of AsODNs into the nanoparticles. This was done by incorporating Cx31.1 AsODNs using PLGA at 3% w/v and phase ratios of 1:10:20 and 1:20:50 or PLGA at 2.5% w/v and a phase ratio of 1:20:50, while Cx43 AsODNs were incorporated into nanoparticles using PLGA at 2.5% w/v and phase ratios of 1:20:50. All experiments were performed in triplicate.

Preparation of PLGA nanoparticles

Blank nanoparticles and nanoparticles loaded with Cy3-labelled AsODNs were prepared using the double emulsion solvent evaporation technique with polyvinyl alcohol (PVA) as emulsifier. Briefly, an aqueous solution (with or without 20 μM of AsODNs) was emulsified in the organic phase (DCM containing 25 to 35 mg/ml of PLGA) using a 2 mm microtip probe sonicator (Hielscher, Teltow, Germany) at an amplitude of 50 W for 1 min. The formed primary water-in-oil (w/o) emulsion was further emulsified in an aqueous solution of PVA (3% w/v) to form a multiple water-in-oil-in-water (w/o/w) emulsion. The organic phase was then evaporated under constant stirring at room temperature and the resultant nanoparticles were recovered by ultracentrifugation at 30,000 rpm and 4 °C. After two washing steps with water, particles were lyophilized for 24 h (VirTis, SP Scientific, Gardiner, NY, US).

Characterization of PLGA nanoparticles

Mean particle size, PDI and ZP were determined using a Malvern Zetasizer Nano (Malvern Instruments, Worcestershire, UK) with samples diluted in MilliQ water. All measurements were performed in triplicate and results were expressed as mean values with a standard deviation (SD). Particle morphology was also investigated using scanning electron microscopy (SEM). Briefly, freeze-dried samples were mounted onto aluminum stubs using double-sided adhesive tape and then sputter coated with gold under argon atmosphere using a Polaron SC 7640 sputter coater. Images were taken on a Philips XL30S FEG Scanning Electron Microscope (Eindhoven, Netherlands) at an acceleration voltage of 5 kV.

Entrapment efficiency of AsODNs

A 4 mg mass of Cy3-labelled AsODN-loaded nanoparticles was added to 0.25 ml of DMSO to allow for polymer dissolution. Subsequently, 0.5 ml water was added and the mixture was allowed to stand for 15 min at room temperature before shaking for 30 min to facilitate extraction of Cy3-tagged AsODNs into the aqueous phase. Samples were centrifuged at 13,000 rpm and 4 °C for 5 min and AsODN concentrations in the supernatant were quantified using a fluorometer at an excitation wavelength of 543 nm and an emission wavelength of 570 nm using a microplate reader (Spectra Max M2 Microplate Readers, Molecular Devices, US). Entrapment efficiencies (EE) were calculated according to the following equation (Equation 1):

1

All measurements were carried out in triplicate and results were expressed as mean values with a standard deviation (SD).

In vitro release studies

A 5 mg mass of AsODN-loaded nanoparticles was suspended in 0.8 ml of PBS and shaken at 150 rpm and 33 °C over the period of the release experiment. 0.6 ml of supernatant was withdrawn at pre-determined time points (0, 4, 10 h and 1, 2, 4, 8, 14 days) and replaced with fresh medium. Samples were centrifuged at 30,000 rpm and the supernatant was assayed for Cy3-tagged AsODNs using a fluorometer. Cumulative amounts of AsODNs released were calculated according the following equation (Equation 2), where D_t is the cumulative amount of AsODNs released at time t and D_∞ is the total amount of drug loaded into the nanoparticles.

2

Cumulative amounts released were plotted against time and release profiles were fitted to a dual first order release model appropriate to the bimodal release observed.

In addition, nanoparticles were added to 0.25 ml of DMSO at the end of the release experiment and further treated as described above for the entrapment efficiency. This was done to allow for complete polymer dissolution and to determine any unreleased AsODN still trapped inside the polymer matrix.

In vitro cellular uptake

Primary corneal epithelial cells were routinely cultured in Opti-MEM I medium supplemented with 5% fetal calf serum (FCS) and 1% penicillin-streptomycin-L-glutamine. Cells were normally grown in 75 cm² tissue culture flasks to confluence and were then seeded onto 8-well chamber slides. After 48 h of settling and adhesion, nanoparticle suspensions (5 mg/ml) were added and cells were incubated for 2 h and 8 h at 4 °C to check for passive uptake and 34 °C to allow for active uptake. Cellular uptake and intracellular distribution were detected by fluorescence microscopy (Leica DMRA, Leica Microsystems, Heidelberg, Germany) with Cy3-AsODNs loaded nanoparticles appearing in red, while cell nuclei were stained with DAPI (blue) and cell membranes were labeled with WGA Alexa 488 (green). Images were recorded at 20x magnification using a digital camera (Nikon Digital Sight DS-U1) and NIS-Elements BR imaging software. In addition, confocal images were taken using a Confocal Laser Scanning Microscope (Olympus FV1000, Olympus, Heidelberg, Germany) equipped with U-MWBV (wideband blue), U-MNIBA3 (narrowband blue) and U-MWIG3 (wideband green) filters and a 60x/1.35 oil immersion lens. A step motor was used to image 40 nm optical slices of the cells and images were processed using FluoView 2.0 software.

Results and Discussion

Physical properties and stability of blank nanoparticles

Table 1 lists the responses for the first factorial study (interaction plots are not displayed). It was shown that polymer concentration and volume ratio had a significant effect (P < 0.001) on the particle size. An increase in the PLGA concentration generally resulted in increased particle size. This was in agreement with previous studies which revealed the formation of larger droplets due to an increase of the emulsion droplet viscosity, which in turn resulted in poorer dispersibility of the PLGA solution into the aqueous phase (Mainardes and Evangelista, 2005; Rizkalla et al., 2006; Song et al., 2008). Increasing the volume ratio (Va1/Vo/Va2) from 1:10:20 to 1:20:50 generally increased the particle size due to increased encapsulated material relative to the internal aqueous phase, which resulted in increased emulsion droplet viscosity and therefore led to poorer dispersibility of the continuous phase. However, there was a significant interaction between the PLGA concentration and the volume ratio (P < 0.001), with the highest PLGA concentration (3.5% w/v) and an increased volume ratio from 1:10:20 to 1:20:50 not resulting in an increased particle size, as the higher polymer concentration of the organic phase led to improved stability of the w/o droplet.

Table 1.  Effect of PLGA concentration and volume ratio (Va1/Vo/Va2) on the physical properties of blank nanoparticles (data represents mean values ± SD, n = 3).

Formulation	Particle Size (nm)	PDI	ZP (mV)
PLGA2.5%-1:10:20	138.03 ± 2.16	0.15 ± 0.01	−21.47 ± 3.38
PLGA2.5%-1:20:50	146.67 ± 1.55	0.10 ± 0.01	−22.20 ± 0.93
PLGA3%-1:10:20	140.60 ± 1.06	0.11 ± 0.01	−33.01 ± 1.73
PLGA3%-1:20:50	159.83 ± 2.25	0.10 ± 0.02	−29.56 ± 1.86
PLGA3.5%-1:10:20	165.93 ± 2.38	0.14 ± 0.01	−28.67 ± 0.39
PLGA3.5%-1:20:50	149.70 ± 1.54	0.11 ± 0.02	−20.07 ± 0.40

The PLGA concentration and the volume ratio both had a significant effect on the PDI (P = 0.032 and P = 0.003 respectively), however, there was no interaction between these two factors (P = 0.301). Both factors also had a significant effect on the ZP (P < 0.001 and P = 0.003 respectively), with an interaction occurring between the two (P = 0.010). This may be related to the relationship between PLGA concentration and volume ratio as explained above for the particle size.

Table 2 lists physical properties for the second experimental design over a period of 15 days. The increase in PDI observed with time suggests that aggregation and disaggregation of nanoparticles occurred during storage. While the particle size increased steadily over the first eight days, there was a slight decrease after 15 days except for PLGA3.5%-1:10:20 nanoparticles which may have undergone extensive disaggregation between day 8 and 15. Interaction plots (not shown) revealed significant interactions between the polymer concentration and the volume ratio on all three dependent variables (size P < 0.001, PDI P = 0.001 and ZP P = 0.012 respectively). However, a significant interaction between the three independent variables (time, concentration and volume ratio) was only observed for the particle size (P < 0.001).

Table 2.  Effect of storage time on the physical properties of blank nanoparticles with different PLGA concentrations and volume ratios (Va1/Vo/Va2) (data represents mean values ± SD, n = 3).

Formulation	PLGA3%-1:10:20	PLGA3%-1:20:50	PLGA3.5%-1:10:20	PLGA3.5%-1:20:50
Particle size (nm)
day 0	140.62 ± 1.06	159.83 ± 2.25	165.93 ± 2.38	149.71 ± 1.26
day 3	151.93 ± 0.13	171.37 ± 1.71	167.21 ± 1.64	148.62 ± 0.95
day 8	156.67 ± 2.26	168.83 ± 1.78	195.41 ± 2.42	159.32 ± 1.65
day 15	148.67 ± 2.25	148.42 ± 2.12	109.93 ± 0.59	168.07 ± 4.62
PDI
day 0	0.11 ± 0.01	0.09 ± 0.02	0.14 ± 0.01	0.11 ± 0.01
day 3	0.21 ± 0.01	0.16 ± 0.01	0.19 ± 0.01	0.14 ± 0.03
day 8	0.21 ± 0.01	0.13 ± 0.01	0.31 ± 0.03	0.21 ± 0.00
day 15	0.23 ± 0.02	0.18 ± 0.00	0.30 ± 0.03	0.30 ± 0.00
ZP (mV)
day 0	−33.07 ± 1.73	−29.56 ± 1.86	−28.67 ± 0.39	−20.01 ± 0.33
day 3	−24.51 ± 0.62	−26.37 ± 0.98	−23.37 ± 1.30	−22.63 ± 0.69
day 8	−26.63 ± 1.24	−25.03 ± 2.46	−21.71 ± 1.20	−21.33 ± 0.84
day 15	−25.93 ± 1.95	−20.11 ± 0.43	−27.77 ± 0.90	−16.07 ± 0.59

SEM images showed nanoparticles of spherical shape with a narrow size distribution (100–200 nm) and a smooth, non-porous surface, with no obvious difference between different formulations, but also revealed agglomeration due to the freeze drying process (Figure 1).

Figure 1.  SEM micrograph of freeze-dried nanoparticles prepared with 3% w/v PLGA and a volume ration of 1:10:20 (8000× magnification, scale bar = 200 nm)

Physicochemical properties and entrapment efficiencies of AsODN-loaded nanoparticles

Table 3 summarizes physicochemical properties obtained for AsODN-loaded nanoparticles. Analogous to plain nanoparticles, size, PDI and ZP were found to be dependent on both the PLGA concentration and the volume ratio of Va1/Vo/Va2, with the particle diameter being significantly different (P < 0.05) between different percentages of PGLA, volume ratios of Va1/Vo/Va2 and molecular weight of the entrapped AsODN (Cx31.1 = 6379 g/mol, Cx43 = 9815 g/mol). Again, increasing the PLGA concentration resulted in an increased viscosity of the organic phase, reducing the shear stress and promoting the formation of larger droplets. There was no significant difference in size, PDI or ZP of the AsODN-loaded nanoparticles compared to the unloaded ones. This would suggest that most of the AsODN is entrapped inside the nanoparticles rather than adsorbed onto the surface, as adsorption of the negatively charged AsODN would result in reduction of ZP. Loading the nanoparticles with Cx43 AsODN, which is 1.5 times the molecular weight of and 11 base pairs longer than Cx31.1 AsODN, resulted in an observed increase in size, however, this was not significant (P > 0.05).

Table 3.  Physicochemical properties of AsODN-loaded nanoparticles (data represents mean values ± SD, n = 3)

Formulation	Particle size (nm)	PDI	ZP (mV)	EE (%)
Cx31.1-PLGA3%-1:10:20	128.81 ± 1.21	0.26 ± 0.02	−29.00 ± 0.44	54.95 ± 9.01
Cx31.1-PLGA3%-1:20:50	165.97 ± 0.95	0.18 ± 0.01	−14.03 ± 0.21	54.53 ± 6.91
Cx31.1-PLGA2.5%-1:20:50	145.69 ± 0.39	0.11 ± 0.01	−22.53 ± 0.85	49.97 ± 0.87
Cx43-PLGA2.5%-1:20:50	183.07 ± 13.11	0.41 ± 0.04	−24.97 ± 0.66	54.03 ± 7.56

As the AsODNs were dissolved in the internal aqueous phase, the organic phase served as a barrier between the two aqueous phases, preventing the diffusion of the active component to the external aqueous phase (O’Donnell and McGinity, 1997) and therefore allowing relatively high recovery of the AsODNs (between 50 and 55%). Even though previous studies have shown that the entrapment of hydrophilic drugs can be improved by using higher concentrations of PLGA and surfactant (Labhasetwar et al., 1997), no significant difference (P > 0.05) was seen between different formulations in this study.

In vitro release studies

Cumulative percentages of AsODNs released versus time are shown in Figure 2. Formulations exhibited a two-phase release pattern: an initial burst release (25–32% within one day) of AsODNs adsorbed onto or near the surface of the nanoparticles followed by a diffusion-controlled delayed release (38–65% after 14 days) of AsODNs embedded in deeper polymer pores together with erosion of PLGA. Addition of DMSO to the remaining nanoparticles after 14 days revealed that up to 30% of the AsODN was still entrapped inside the polymer matrix and was not released over the investigated time frame.

Figure 2.  In vitro release profiles of various AsODN-loaded nanoparticles (data points represents mean values + SD, n = 3)

In order to further analyze the mechanism of drug release and compare the different formulations, in vitro release date was fitted to zero-, first- and second-order (Costa and Sousa Lobo, 2001), as well as Higuchi (Higuchi, 1961), Hixson-Crowell cube root (Hixson and Crowell, 2002) and Baker & Lonsdale models (Akbuga and Durmaz, 1994). The highest R² values (0.696 to 0.889) were obtained after the application of the Higuchi model, which describes the amount of drug released as a function of the square root of time. This is typical for systems where drug release is governed purely by diffusion. The second best R² values (0.445 to 0.753) were observed for the Hixson-Crowell cube root model, which is generally used to describe erosion-based mechanisms. A two exponential model was therefore proposed (Equation 3), with the first part (A) accounting for rapid release due to diffusion and the second part (B) relating to delayed release from the polymer matrix after erosion of PLGA (Equation 3).

3

Fitting the release data to the above equation, fractions of drug released (f), release rate constants (k), half-life (t_1/2) and lag times (t_lag) were calculated for the initial burst release (A) and the delayed slow release (B) (Table 4).

Table 4.  Kinetic parameters for the hypothesized release model (data represents mean values ± SD, n = 3), *(P < 0.05)).

Formulation	Cx31.1-PLGA3%-1:10:20	Cx31.1-PLGA3%-1:20:50	Cx31.1-PLGA2.5%-1:20:50	Cx43-PLGA2.5%-1:20:50
fA (%)	28.15 ± 1.84	23.01 ± 1.82	28.48 ± 1.25	11.19 ± 5.34*
kA (day^−1)	2.40 ± 1.42	4.10 ± 0.11	4.46 ± 2.80	27.31 ± 15.56*
t1/2 A (day)	0.35 ± 0.14	0.17 ± 0.00	0.26 ± 0.20	0.04 ± 0.03
fB (%)	34.28 ± 8.95	15.87 ± 0.73	41.23 ± 18.21	32.08 ± 9.46
kB (day^−1)	0.17 ± 0.08	0.23 ± 0.03	0.26 ± 0.19	0.68 ± 0.33
t1/2 B (day)	4.55 ± 1.03	3.05 ± 0.35	4.89 ± 1.44	1.20 ± 0.45
tlag (h)	< 1	< 1	< 1	< 1
R^2	0.999 ± 0.00	0.996 ± 0.00	0.996 ± 0.00	0.983 ± 0.01

k_A was in the order of PLGA2.5%-1:20:50-Cx43 > PLGA2.5%-1:20:50-Cx31.1 > PLGA3%-1:20:50-Cx31.1 > PLGA3%-1:10:20-Cx31.1, with PLGA2.5%-1:20:50-Cx43 being significantly different (P < 0.05) to the other three formulations. This may be explained by the longer sequence of the Cx43 AsODNs, which results in less tight binding of the molecules onto the polymer surface and therefore more area for interaction with the surrounding medium leading to faster release of the adsorbed AsODNs. k_B showed no significant difference (P > 0.05) between formulations, suggesting that the erosion process occurs in a similar time-dependent manner. Calculated release half-life (t_1/2) indicated that the molecular weight of the AsODN played a major role in the rapid initial release of adsorbed AsODN molecules, with t_1/2 for Cx31.1 being 4 times longer (4.89 ± 1.44 days) than for Cx43 (1.20 ± 0.45 days). This again suggests that most of the Cx43 AsODNs were rather loosely bound to the surface of the nanoparticle than incorporated into the core when compared to Cx31.1 AsODNs.

Relating the release profile fitted parameters to the particle size it was seen that both the initial burst release and the total amount of Cx31.1 AsODNs released over time decreased with increasing particle size. This can be explained by a decrease in the surface area to volume ratio, resulting in less surface being available for AsODN adsorption as well as less surface area available for the medium to penetrate into the particles and erode the polymer matrix (Panyam et al., 2003). In addition, increasing the particle diameter also increased the length of the diffusion pathway, which led to drug molecules having to traverse a longer distance within the polymer matrix to reach the surface (Budhian et al., 2008). Looking at the effect of polymer concentration it was shown that the release rate of AsODNs from the 3% w/v PLGA nanoparticles was slower. This could be due to the fact that the higher PLGA concentration led to reduced porosity of the nanoparticles and therefore limited permeability of water, leading to an overall slower erosion process (Blanco and Alonso, 1998). If water was able to penetrate into the nanoparticle, the PLGA matrix would have become more and more hydrophilic, allowing even more water to penetrate, thereby further enhancing polymer degradation and therefore drug release (Cheng et al., 2008). Changing the volume ratio from 1:10:20 to 1:20:50 for the Cx31.1 AsODN-loaded 3% w/v PLGA nanoparticles, a decrease in the amount of total AsODNs released within 14 days was observed (from 60 to 40%). This could be explained by a decreased volume of the organic phase, resulting in higher PLGA concentrations. This would lead to reduced pore formation and therefore reduced polymer erosion as well as limited diffusion of AsODNs out of the polymer matrix. In addition, higher PLGA and PVA concentrations would stabilize the emulsion as well as the formed nanoparticles leading to slower erosion and therefore delay in drug release.

In vitro cellular uptake

Cells incubated at 4 °C showed nanoparticles adsorbed onto the cell surface especially along the pseudopodia (Figure 3, left), with no significant uptake of AsODN-loaded nanoparticles into the cells. However, nanoparticles with cells incubated at 34 °C were internalized into the cells rather than adsorbed onto the cell surface and dispersed throughout the cytoplasm and perinuclear region (Figure 3, right). This was confirmed by confocal microscopy (Figure 4), which revealed a large number of nanoparticles adsorbed onto the cell surface (yellow color resulting from an overlay of red (Cy3-AsODN-loaded nanoparticles) and green (WGA Alexa 488 labeled cell membranes), but also showed a significant amount of nanoparticles (red) internalized into the cell and dispersed throughout the perinuclear region.

Figure 3.  Cellular uptake of Cx31.1-PLGA3% nanoparticles incubated at 4 °C (left) and 34 °C (right) respectively for 8 h; merged images of Cy3-AsODN-loaded nanoparticles (red), cell nuclei stained with DAPI (blue) and cell membranes labeled with WGA Alexa 488 (green); (scale bar = 10 µm)

Figure 4.  Confocal 3D cross-section of Cx31.1-PLGA3% nanoparticles incubated at 34 °C for 8 h; merged images of Cy3-AsODN-loaded nanoparticles (red), cell nuclei stained with DAPI (blue) and cell membranes labeled with WGA Alexa 488 (green), with the yellow color signifying an overlay of red and green; (scale bar = 10 µm)

The increase in uptake upon increase in temperature suggests that nanoparticles were mainly taken up by an active, energy-dependent process such as endocytosis rather than passive diffusion. Endocytosis can be subdivided into phagocytosis, fluid phase pinocytosis and receptor-mediated endocytosis (Foster et al., 2001; Suh et al., 1998). As there are no specific ligands for the nanoparticles, receptor-mediated uptake is unlikely. Panyam et al. (2002) demonstrated that inhibition of receptor-mediated endocytosis resulted only in 40% inhibition of nanoparticle uptake and suggested that particles could partially be taken up by non-specific receptor-mediated endocytosis, while the main uptake may occur via fluid phase pinocytosis. This was also supported by studies performed by Qaddoumi et al. (2003), who found that the uptake of PLGA nanoparticles into conjunctival epithelial cells was mostly independent of clathrin- and caveolin-1-mediated pathways but rather occurred through adsorptive endocytosis. As the lower size cut-off described for phagocytosis is 500 nm (Rupper and Cardelli, 2001), the most likely mechanism of nanoparticle uptake in this study is via fluid phase pinocytosis. However, further experiments using a variety of specific uptake inhibitors will need to be done in order to confirm the uptake mechanism.

In terms of nanoparticle release from the endosomes once inside the cell, the following can be proposed: Primary endocytic vesicles have a physiological pH (Mukherjee et al., 1997), at which PLGA nanoparticles are negatively charged. However, charge reversal occurs in the endo-lysosomes where the environment is slightly acidic (pH 4 to 5) (Panyam et al., 2002; Prabha and Labhasetwar, 2004). This positive charge destabilizes the endosomal membrane leading to rapid escape of the nanoparticles from the endo-lysosomes, which in turn results in less degradation of the AsODNs by lysosomal enzymes. In addition, AsODNs need to escape the endosomes in order to be able to bind to the mRNA and exhibit their action. Further experiments are required to confirm the endosomal escape and prove functional AsODN activity inside the cells.

Conclusion

PLGA nanoparticles for topical delivery of Cx31.1 AsODNs were prepared using the double emulsion solvent evaporation method. The nanoparticle size increased with increasing PLGA concentrations and volume ratios of the external water phase. Moreover, increasing the PLGA concentration resulted in a slightly higher entrapment efficiency of the AsODNs. In vitro release profiles exhibited a two-phase pattern with the initial burst release highly dependent upon the molecular weight of the adsorbed AsODNs. In addition, nanoparticle release decreased with increasing particle size and higher PLGA concentrations due to an increase in the diffusion pathway as well as lower porosity of the polymer matrix. It can therefore be concluded that release kinetics can be tailored by adjusting the PLGA concentration and volume ratio of the external phase. Finally, cellular uptake occurred via active, time-dependent endocytosis, most likely fluid phase pinocytosis. PLGA nanoparticles therefore show potential to successfully deliver Cx31.1 AsODNs to the skin and the eye and will be further evaluated using in vivo studies.

Acknowledgements

The authors would like to thank Dr Ally Chang, Jane McGhee and AP Trevor Sherwin from the Department of Ophthalmology for their guidance and advice with the cellular uptake experiments.

Declaration of Interest

The authors report no declarations of interest.