International Journal of Pharmaceutics 477 (2014) 12–20

Contents lists available at ScienceDirect

International Journal of Pharmaceutics journal homepage: www.elsevier.com/locate/ijpharm

Pharmaceutical nanotechnology

Nanoparticles based on naturally-occurring biopolymers as versatile delivery platforms for delicate bioactive molecules: An application for ocular gene silencing Jenny E. Parraga a , Giovanni K. Zorzi a,1, Yolanda Diebold c , Begoña Seijo a,b , Alejandro Sanchez a,b, * a Department of Pharmacy and Pharmaceutical Technology, Faculty of Pharmacy, University of Santiago de Compostela (USC), Campus Sur, Santiago de Compostela 15782, Spain b Molecular Image Group, Health Research Institute, University Clinical Hospital of Santiago de Compostela (IDIS), A Choupana, Santiago de Compostela 15706, Spain c Institute for Applied Ophthalmobiology (IOBA), University of Valladolid, Campus Miguel Delibes, Paseo de Belén 17, Valladolid 47011, Spain

A R T I C L E I N F O

A B S T R A C T

Article history: Received 18 June 2014 Received in revised form 23 September 2014 Accepted 26 September 2014 Available online 30 September 2014

Nanoparticles based on naturally-occurring biopolymers, most of them endogenous macromolecules, were designed as a versatile generation of delivery platforms for delicate bioactive molecules. The design of these nanosystems was specifically based on our recent finding about the ability of endogenous polyamine spermine (SPM) to interact with anionic biopolymers (ABs) generating ionically cross-linked nanosystems. The initial first generation of these delivery platforms, based on glycosaminoglycans and other polysaccharides, showed a very high association capacity for some delicate bioactive proteins such as growth factors, but a limited capacity to associate negatively charged molecules, such as pDNA and siRNA. However, the versatility of these nanosystems in terms of composition allowed us to customise the association of active ingredients and their physicochemical characteristics. Concretely, we prepared and incorporated gelatine cationized with spermine (CGsp) to their composition. The resulting modified formulations were characterised by a nanometric size (150–340 nm) and offer the possibility to modulate their zeta potential (from
35 to 28 mV), providing an efficient association of nucleic acids. The biological evaluation of these optimised nanosystems revealed that they are able to be internalised in vivo into corneal and conjunctival tissues and also to provide a significant siRNA gene silencing effect. ã 2014 Elsevier B.V. All rights reserved.

Keywords: Spermine Polyanionic biopolymers Nanoparticles Cationized gelatine siRNA Gene therapy

1. Introduction From a historical perspective, a premise of biomaterial-related sciences, when designing new components for medical and pharmaceutical use, was bioinertness as a major requirement; the material should not provoke any reaction from the body (Peppas and Langer, 1994). However, this former position was

Abbreviations: SPM, spermine; ABs, anionic biopolymers; HA, hyaluronic acid; CA, colominic acid; ChS, chondroitin sulfate; DS, dextran sulfate; HS, heparan sulfate; CGsp, modified cationic gelatine with spermine. * Corresponding author at: Department of Pharmacy and Pharmaceutical Technology, Faculty of Pharmacy, University of Santiago de Compostela, Campus Sur s/n, 15782, Spain. Tel.: +34 981 563 100x15105; fax: +34 981 547 148. E-mail address: [email protected] (A. Sanchez). 1 Current address: Graduate Studies Program in Pharmaceutical Sciences (PPGCF)–Federal University of Rio Grande do Sul (UFRGS), Ipiranga 2752, Porto Alegre 90610-000, Brazil. http://dx.doi.org/10.1016/j.ijpharm.2014.09.049 0378-5173/ ã 2014 Elsevier B.V. All rights reserved.

reconsidered after it became clear that the use of biomacromolecules could improve targeting towards specific tissues/organs and provide an added biological value (Byrne et al., 2008; Farokhzad and Langer, 2009; Phillips et al., 2010; Pirollo and Chang, 2008). As a consequence, biomaterials with specific abilities to interact with biological structures have recently emerged as new potential components of different devices intended for biomedical use (Kamaly et al., 2012). Within this frame, the interest in some naturally-occurring biomaterials as delivery system components has gained increasing attention. However, when designing delivery systems at the nanoscale level, one of the main limitations of these biomaterials is related to the preparation technique, which usually requires organic solvents or harsh conditions that can modify the natural and desired properties of these biomolecules. With the aforementioned potentials and limitations in mind, we have recently patented new delivery nanosystems based on

J.E. Parraga et al. / International Journal of Pharmaceutics 477 (2014) 12–20

the capacity of the endogenous polyamine spermine to physically cross-link anionic macromolecules (Sanchez et al., 2010). Of course there are numerous nanoparticles based on ionic crosslinking processes. However, two main differences of the nanosystems proposed here can be mentioned. Firstly, such an invention refers to the “back to nature” nanoparticles, based on natural products. Secondly, as the main features of the developed nanosystems, we should mention that preliminary results suggested their potential versatility in terms of modifying their components and including new ones, which opens up the possibility of a tailored-design of stable nanoparticles with a tunable bio-inspired composition. The objective of the present work has been to verify and to exploit the aforementioned idea and features to develop nanoparticles based on biopolymers as versatile delivery platforms for delicate bioactive molecules in general, and for siRNA in particular, keeping in mind the recent necessity to develop new nanosystems as effective carriers for such molecules (Blow, 2007; de Fougerolles et al., 2007; Whitehead et al., 2009). For this purpose, we decided to select some endogenous anionic biopolymers (ABs) with high biological value as the initial components in the design of our first nanoparticles prototypes. However, we have also considered that, in general, nanocarriers for nucleic acids delivery share a common characteristic: they are based on positively charged polymers or lipids. These key elements play two main roles. First, they provide physical interaction between the positively charged moiety and the negatively charged nucleic acid, protecting them from nuclease degradation. Second, they provide a net positive charge, which in turn enables binding of the nucleic acid complex to anionic cell surface macromolecules (Kang et al., 2012; Li and Szoka, 2007). However, instead of incorporating conventional cationic polymers/lipids in the nanoparticles composition to associate the nucleic acid derivatives, we further continued with the aforementioned philosophy concerning the use of safe materials, and tried to improve the capacity of the described nanosystems to associate genetic material by the introduction of gelatine that was previously aminated by the reaction between the carboxylic acid present in its structure and the endogenous polyamine spermine (Zorzi et al., 2011). To understand the difficulty in developing such an approach, it is necessary to remember that the incorporation of proteins or modified proteins into nanoparticulate carriers often requires the use of organic solvents and/or chemical reagents (Maham et al., 2009). Our challenge has been to incorporate the cationized gelatine into the nanoparticles composition while avoiding all these previously mentioned harmful conditions. Finally, the physicochemical characterisation of the developed formulations was performed and they were biologically evaluated in terms of cellular toxicity and efficiency as siRNA carriers. 2. Materials and methods 2.1. Materials The anionic biopolymers (ABs) used for the development of the nanosystems were hyaluronic acid (HA, 136 kDa – Bioibérica, Spain), colominic acid (CA, 30 kDa – Sigma, Spain), chondroitin4-sulfate from bovine trachea (ChS, 40 kDa – Calbiochem, CA, USA), dextran sulfate (DS – 40 kDa – TdB Consultancy, Sweden), heparan sulfate (HS) (Sigma, Spain) and gelatine (type A, 137 kDa – Nitta Gelatin, Canada). The polyamine spermine hydrochloride was purchased from Fluka (Spain). The model drugs associated to the nanosystems were albumin (bovine serum albumin), kinetin and bFGF (Sigma, Spain), plasmid that codifies the fluorescent green protein (pEGFP) (Elim Biopharmaceutics, CA, USA), siRNA against GAPDH containing the sequence 50 -UGG

13

UUU ACA UGU UCC AAU ATT-30 (sense), 30 -UAU UGG AAC AUG UAA ACC ATG-50 (antisense), siRNA against GAPDH labelled with Cy31, non-specific siRNA against siEGFP containing the sequence 50 -GCA AGC UGA CCC UGA AGU UCTT-30 (sense), and 50 -GAA CUU CAG GGU CAG CUU GCTT-30 (antisense), a negative control siRNA containing the sequence 50 -GCA AGC UGA CCC UGA AGU UCTT-30 (sense) and 30 -GAA CUU CAG GGU CAG CUU GCTT-50 (antisense) (Ambion, CA, USA). N-(3-Dimethylaminopropyl)-N0 -ethylcarbodiimide hydrochloride (EDC), glycerol and all other chemicals not specified above were of the highest purity available, purchased from Sigma (Spain).

2.2. Preparation and characterisation of the ABs nanoparticles Polymer solutions of HA (1–1.5 mg/mL), ChS (2–2.5 mg/mL), HS (1.5–2 mg/mL), CA (1–1.2 mg/mL) or DS (1–1.5 mg/mL) were prepared in MilliQ water. The nanoparticles were formed spontaneously by adding an aqueous solution of the cationic spermine (0.1–1.5 mg/mL) over the polyanionic biopolymers solution, under magnetic stirring (volume ratio of 2:1; polymer: spermine). In the case of nanoparticles associating the selected bioactive molecules, these molecules were incorporated depending on their charge either in the anionic polymeric solution (albumin 5 mg/mL, bFGF 100 ng/mL, pDNA 5 mg/mL or siRNA 3 mg/mL) or in the cationic SPM solution (kinetin 1 mg/mL), in order to avoid ionic interactions prior to the formation of nanoparticles. Albumin was incorporated in ChS–SPM and DS–SPM nanoparticles at the theoretical loading of 13% w/w. Kinetin was incorporated in ChS–SPM and HA–SPM nanoparticles at the theoretical loading of 5% w/w. On the other hand, ChS–SPM and DS–SPM nanoparticles were selected to associate the growth factor bFGF at 0.006 and 0.009% w/w, respectively. pDNA and siRNA were incorporated at 5 and 3% w/w, respectively. The yield of production was determined previously for the selected formulations of nanoparticles: ChS–SPM (40:8) 40%, DS–SPM (20:5) 64% and DS–SPM (30:5) 70%, respectively. For kinetin formulations the mass ratio of HA–SPM was 8:1 and ChS:SPM 8:1 (yield of 48 and 52%, respectively). The mean particle size was determined by photon correlation spectroscopy (PCS). Each analysis was carried out at 25  C with a detection angle of 173 . The zeta potential was obtained by Laser Doppler Anemometry (LDA). For this purpose, the samples were diluted with a millimolar solution of KCl. Both analyses were performed with a Zetasizer Nano (Malvern, UK). The association efficacy was calculated indirectly by the difference between the total amount of bioactive molecules incorporated and the amount that was not associated, which remained in the preparation medium. For this purpose, free albumin and bFGF were recovered in supernatant samples collected by centrifugation of the nanoparticles (60 min, 10,000 rcf) (Beckman CR412, Beckman Coulter, Spain). The association efficacy of the albumin was determined using a micro BCA kit (Pierce co, USA) and bFGF was quantified using an ELISA kit (Anogen, Canada). The association efficacy of kinetin was determined by spectrophotometry (l = 265 nm) after separation of the free kinetin by ultrafiltration (Amicon Ultra membranes 5000 MWCO, Millipore, Irland) using a centrifuge (11,000 rcf, 30 min) (Beckman CR412, Beckman Coulter, Spain). The association of siRNA and pDNA to the nanoparticles was determined by an agarose gel electrophoresis assay (see Supplementary Data) (n = 3). The % siRNA associated was quantified by Ribogreen1 (see Section 2.3.2). The drug loading content (L.C.) and the association efficiency (A.E.) were calculated using the following equations, respectively:

14

J.E. Parraga et al. / International Journal of Pharmaceutics 477 (2014) 12–20

L:C:ð%Þ ¼

Weight of associated bioactive molecule  100 Weight of nanoparticles

A:E:ð%Þ ¼ Weight of associated bioactive molecule  100 Weight of bioactive molecule used for nanoparticles preparation

2.3. Optimisation of the initial nanoparticle prototypes for nucleic acid delivery 2.3.1. Synthesis of cationized gelatine (CGsp) by amidation Gelatine was cationized as previously described (LopezCebral et al., 2011; Zorzi et al., 2011). Briefly, 1% (w/v) gelatine solution was prepared in 0.1 M phosphate buffer (PBS) (pH 5.3). The reaction was carried out for 100 mg of gelatine, by mixing with 1620 mg of spermine and 268 mg of EDC for 18 h at 37  C, under slow magnetic stirring. The resulting solution was dialysed for 48 h and was subsequently lyophilised. We evaluated the isolectric point (pI) of the cationized gelatin CGsp by the PAGI method (Sanchez et al., 2010). Briefly, a solution of 1% cationized gelatin (w/v) was incubated at 37  C during 30 min with a previously washed mix of acid and basic resins (1:2). The pH obtained from the solution indicates the pI of the cationized gelatin. We obtained a pI of 11 for the CGsp. 2.3.2. Design and characterisation of ABs:CGsp nanoparticles as a vehicle for siRNA delivery CGsp was incorporated to the nanoparticles components in order to obtain nanostructures able to associate genetic material. To this end, aqueous solutions of CGsp (0.75, 1.5, 1 or 2 mg/mL) were mixed with spermine solutions (0.1, 0.2, 0.25, 0.75 or 1 mg/mL) under slow magnetic stirring and the resulting solution was added over an aqueous solution of ChS (0.25, 0.35, 7.5 or 10 mg/mL) or HA (0.25, 0.35, 5 or 10 mg/mL) and siRNA (0.03 mg/mL). The volume ratio of the components was 10:20:5 AB: CGsp:SPM. For the in vivo studies, the nanoparticles were previously concentrated by centrifugation (Beckman CR412, Beckman Coulter) at 10,000 rcf for 30 min at 4  C with 0.1% glycerol (v/v). After the supernatants were discarded, the concentrated nanoparticles were resuspended in 5% glucose at a final concentration of 0.3 mg/mL of siRNA. The size and zeta potential of the developed nanoparticle suspension were determined as described in Section 2.3.1. The morphological analysis of these nanoparticles was performed by transmission electron microscopy (See Supplementary Data). The association efficacy was indirectly evaluated in the supernatant of samples collected upon centrifugation of the nanoparticles (10,000 rcf, 30 min, Avanti 30 Beckman, Barcelona, Spain) by using a commercial Ribogreen1 intercalation assay (Life Technologies, Spain) in a fluorimeter at lex = 480 nm and lem = 520 nm (LS 50B luminescence spectrometer, PerkinElmer, USA). 2.4. Biological evaluation in vitro 2.4.1. Cell culture Human corneal epithelial (HCE) cells were cultured in DMEM/F12 with 15% foetal bovine serum, penicillin–streptomycin (100 and 100 mg/mL, respectively), 0.5% DMSO (Sigma, Spain), and 10 ng/mL EGF (Invitrogen, Spain). The cells were maintained at 37  C in a 5% CO2 humidified atmosphere. Passages 20–22 were used for the experiments.

2.4.2. Cell viability assay The XTT assay was employed to evaluate the viability of the HCE cells after the incubation of the nanoparticles. The cells were seeded in a density of 10,000 cell/well in a 96-well culture plate (Nunc, Denmark). After 24 h, the culture medium was replaced by a serial dilution of the nanoparticles associating siRNA (6.25–100 mg/cm2 of CGsp incorporated into the formulation) and incubated for 3 h. Afterwards, the nanoparticle suspension was removed and replaced by fresh culture medium. The cell viability was assayed after 48 h using the XTT-based Toxicology Kit (Sigma, Spain), according to the manufacturer’s protocol. 2.4.3. Transfection efficiency of siRNA against GAPDH associated with the developed nanoparticles in HCE cells The nanocarriers’ ability to transfect siRNA was evaluated in HCE cells. For this purpose, these cells were plated 24 h before the experiment at a density of 7000 cells per well into a 96-well plate (Becton Dickinson, France). Cells were washed and the culture medium replaced by non-supplemented DMEM/F-12 (without fetal bovine serum). Cells were incubated for 3 h with nanoparticles associating siRNA against GAPDH, non-specific siRNA against EGFP and the siRNA control at three different concentrations (35, 75 or 100 nM) After the incubation, the medium was replaced with fresh culture medium after washed the cells properly. Lipofectamine (Invitrogen, Spain) was used as a positive transfection control according to the manufacturers protocol. The transfection efficiency was evaluated two days post-transfection. The knock down of GAPDH was evaluated by KDalert GAPDH Assay Kit (Ambion, TX, USA) according to the manufacturers protocol. 2.4.4. Cellular uptake of the developed nanoparticles The uptake of the developed nanoparticles by the HCE cells was evaluated by confocal microscopy. For this purpose, the HA and ChS components of the nanoparticles were previously labelled with fluorescein–amine (fl–HA and fl–ChS), as described elsewhere (de Belder and Wik, 1975). GAPDH siRNA or GAPDH siRNA–Cy3 were associated with the nanoparticles, according to the preparation technique described in Section 2.3.2. The HCE cells were seeded at a density of 100,000 cells per well in a multi-chamber Labtek1. Twenty four hours later, the cells were incubated with the nanoparticles (200 nM siRNA per well). After 3 h, the cells were rinsed with PBS and fixed with 4% paraformaldehyde. The cells were permeabilised with 0.1% Triton-X100, the cell nuclei were stained with DAPI (Sigma, Spain) and the cytoplasmic protein F-actin was stained with BODIPY-phalloidin (Invitrogen, Spain). The cell fluorescence was analysed by confocal laser scanning microscopy (Leica TCS SP, Leica Microsystems, USA). Laser excitation wavelengths of 405, 488, 561 and 633 nm were used. Fluorescent emissions from DAPI (emission l = 415–470 nm), FITC (emission l = 495–540) BODIPY-phalloidin (emission l = 645–710 nm) and Cy3 (emission l = 580–625 nm) were collected using a sequential scan. 2.5. In vivo interaction of the developed nanoparticles associating siRNA with the rabbits ocular surface The in vivo study was performed in accordance to the guidelines of the association for research in vision and ophthalmology (ARVO) and protocols were approved by the USC Institutional Board. Healthy New Zealand rabbits (6–8 weeks old) were used and topical administration of the tested formulations was performed in both eyes in order to minimise the number of animals. Doublelabelled nanoparticles (polymer and siRNA) described in the previous section were used for this study. To evaluate the ability of the nanoparticles-associated siRNA to penetrate the corneal and conjunctival epithelia, normal conscious rabbits were maintained

J.E. Parraga et al. / International Journal of Pharmaceutics 477 (2014) 12–20

15

Table 1 Naturally-occurring polysaccharides used in the design of nanoparticles based on anionic biopolymers. Polymer

Repetitive units

Characteristics

Chondroitin sulfate (ChS)

Linear heteropolymeric glycosaminoglycan of the extracellular matrix composed of D-glucuronate and GalNAc

Polysialic or colominic acid (CA)

Linear homopolymer composed of N-acetylneuraminic acid

Dextran sulfate (DS)

Branched heteropolymer composed of sulfated glucose units

Heparan sulfate (HS)

Linear heteropolymeric glycosaminoglycan composed of iduronate-2-sulfate (D-glucuronate-2sulfate) and N-sulfo-D-glucosamine-6-sulfate.

Hyaluronic acid (HA)

Linear heteropolymeric glycosaminoglycan of the extracellular matrix composed of D-glucuronate and GlcNAc.

in an upright position using restraint cages. Then, 30 mL of concentrated nanoparticles suspension (0.3 mg/mL of siRNA) was administered topically into the cul-de-sac every 15 min to provide a final dose of 36 mg of siRNA. Naked siRNA was used as a control. At 2 h post-administration, the rabbits were euthanised. Corneal and conjunctival tissues were collected, mounted on glass slides and analysed by confocal laser scanning microscopy under the aforementioned conditions. 2.6. Statistical analysis The results, expressed as mean  standard deviation (SD) of three independent experiments, were statistically analysed by ANOVA, followed by the Tukey’s test.

3. Results and discussion 3.1. Development of versatile nanoparticle delivery platforms based on naturally-occurring biopolymers The main objective of the present work has been to elucidate the possibility of exploring our recent findings about the ability of endogenous polyamine spermine (SPM) to interact with the anionic polymers in a controllable way to provide a robust and versatile delivery platform based on naturally-occurring biopolymers, which would serve as the basis to develop custom-made nanosystems able to fulfil the association requirements of specific delicate bioactive molecules. Our first results in these studies indicated that, through a judicious selection of components and their relative proportions,

Table 2 Particle size and zeta potential of blank nanoparticles (without bioactive ingredients) developed due to the ionotropic interaction of spermine (SPM) and different anionic biopolymers (ABs) (hyaluronic acid (HA), chondroitin sulfate (ChS), heparan sulfate (HS), colominic acid (CA) and dextran sulfate (DS)) at different SPM/AB mass ratios (n = 3). ABs used

ABs (mg/mL)

SPM (mg/mL)

Size (nm)

PDI

z potential (mV)

HA

1.5 1 2 2.5 1 1.2 2 1.5 1 1.5

0.4 0.3 0.3 0.7 2 1.5 0.75 0.5 0.5 0.8

572  34 422  32 173  6 178  4 682  20 582  21 150  13 125  17 151  8 161  3

0.3 0.3 0.1 0.2 0.3 0.3 0.1 0.1 0.1 0.1

-18  5 -15  3 -31  1 -22  1 -10  3 -10  2 -25  4 -23  5 -25  2 -19  2

ChS CA HS DS

16

J.E. Parraga et al. / International Journal of Pharmaceutics 477 (2014) 12–20

Table 3 Particle size, zeta potential, bioactive molecules association efficacy (A.E.) and loading content (L.C.) of nanoparticles developed by ionotropic interaction of spermine (SPM) and different anionic biopolymers (ABs) (hyaluronic acid (HA), chondroitin sulfate (ChS), colominic acid (CA) and dextran sulfate (DS)) at different SPM/AB mass ratios (n = 3). (N.A.: no association). Bioactive molecule

Nanosystem composition

Mass ratio

Size(nm)

Albumin

ChS:SPM DS:SPM ChS:SPM HA:SPM ChS:SPM DS:SPM DS:SPM HA:SPM ChS:SPM HA:SPM

40:8 20:5 8:1 8:1 40:8 30:5 2:0.75 40:7.5 1:0.25 1:0.375

161 145 209 472 183 120 171 532 192 126

Kinetin bFGF pDNA siRNA

a b

         

5 4 6 20 4 3 1 21 1 1

z potential (mV)

A.E. (%)

L.C. (% w/w)


19  1
18  1
18  1
18  1
25  2
27  3
13  1
21  1
21  1
22  1

74  5 28  6 >99a 41  6 97  10 94  12 N.A. N.A. N.A. N.A.

24.05 5.64 9.60a 4.35 0.015 0.013 0b 0b 0b 0b

Bellow quantification limit. Complete migration in electrophoresis gel.

some biodegradable blank nanostructures (without bioactive ingredient associated) can be obtained. Table 1 summarises the main characteristics of the initial components selected for the design of our first nanoparticle prototypes, which were anionic biopolymers (ABs) with high biological value. Using these biopolymers, a broad range of different nanoparticles have been developed, on the basis of the cross-linking properties of SPM and the appropriate selection of component mass ratios. Table 2 shows the investigated compositions and ratios of these components, as well as the particle size and zeta potential of the resulting formulations. In addition, we can appreciate how polymers that include sulfate groups in their structures, like ChS, HS and DS, led to the formation of nanoparticles of smaller size and polydispersion indexes than those developed using polymers that carry functional groups such as hydroxyl and carboxylic groups (i.e. HA and CA). The pKa for carboxylic groups is around 2.6 for CA (Vimr et al., 2004) and around 3 for HA (Lapcík et al., 1998), whereas for ChS carboxylic groups is around 4.5 (Larsson et al., 1981). These values are higher than those attributed to the sulphate groups, (presented in both DS and ChS) which vary from 1.5 to 2 (Larsson et al., 1981). The results suggest that sulfate groups could promote significant interaction of polymers bearing such groups with spermine, that would result in a higher ionotropic cross-linking effect, thus leading to more compact nanostructures as compared to nanoparticles based on polymers with other groups (i.e. hydroxyl or carboxylic groups). Indeed, if we consider the fact that the sulfate groups have a lower pKa than carboxylic groups, it is possible to expect a higher number of ionized groups in the polymer and, therefore, hypothesize a higher avidity of this polymer bearing sulfate groups to interact with protonated groups such as the amine groups of SPM. On the other hand, these results also confirm the versatility of the developed nanoparticles in terms of composition and possibility to associate different bioactive molecules and, in addition, they reinforce the importance of the components selection as a strategy to modulate the nanoparticles properties. This versatility is a very interesting feature of the designed nanosystems, because it is well known that the development of

numerous nanoparticulate systems is subjected to critical formulation parameters and limited composition variations. As a consequence, slight modification of these parameters or components can result in inappropriate characteristics, instability of the developed systems or even the inability to produce them (Howard et al., 2000; Liu et al., 2007). As an example, the delicate ionic balance established in chitosan based nanosystems between this positively charged semisynthetic polymer and the corresponding counter ions clearly difficult the incorporation of additional ingredients and/or the association of different bioactive molecules (Sanchez et al., 2010), whereas these limitations are circumvent with the delivery platforms herein described. 3.2. Association of different labile bioactive molecules to the developed nanoparticles In a second step, the above-mentioned initial prototype formulations were tested for their ability to associate different labile bioactive molecules that were incorporated during their preparation. As can be observed in Table 3, the developed nanosystems demonstrated an efficient association of some bioactive hydrophilic molecules with very different physicochemical properties (molecular weight, conformation and charge), including low molecular weight drugs and proteins. The high association efficiency for the model protein albumin and the growth factor FGFb can be explained on the basis of their moderate charge and the previously described ability of the functional groups of proteins to interact through hydrogen bonds and hydrophobic domains with ABs (Kreuger et al., 2006; Raman et al., 2005; Sasisekharan et al., 2006), which could facilitate their integration in the nanostructures. On the contrary, these initial nanosystems were not able to associate nucleic acids such as pDNA (pEGFP) and siRNA (siGAPDH) (see Supplementary Data), which can be attributed to the establishment of high repulsive forces of these highly negatively charged bioactive molecules and the negatively charged biopolymers used to prepare the nanoparticles.

Table 4 Particle size, zeta potential and association efficacy (A.E.) of nanoparticles developed by ionotropic interaction of spermine (SPM), gelatine cationised with spermine (CGsp) and different anionic biopolymers (ABs) (hyaluronic acid (HA) or chondroitin sulfate (ChS)) at different mass ratios (siRNA loading: 3%) (n = 3). ABs used

Mass ratio AB:CGsp:SPM

Size (nm)

PDI

z potential (mV)

A.E. (%)

HA

100:15:1.25 50:15:1.25 3.5:30:1 2.5:30:0.5

340  24 270  27 220  15 215  21

0.2 0.2 0.2 0.1

-35  2 -33  1 +26  2 +28  3

83  6 85  9 90  7 87  6

ChS

2.5:30:1 3.5:20:1.25 100:40:5 75:30:3.75

250  30 283  24 150  15 182  25

0.2 0.1 0.2 0.1

+20  3 +21  2 - 35  6 -32  3

85  6 83  8 89  6 87  7

J.E. Parraga et al. / International Journal of Pharmaceutics 477 (2014) 12–20

3.3. Customising the initial prototype formulations to associate nucleic acids In order to facilitate the association of nucleic acid derivatives to the developed drug delivery systems, while maintaining simple and fast preparation techniques and structural properties, two main strategies could be used. The first one is based on the modification of the initial components of the formulations, and the second one is based on the incorporation of additional components with a higher affinity for these bioactive molecules. The strategy based on glycosaminoglycans and polysaccharides modification has been widely developed for gene delivery applications (Raemdonck et al., 2013). However, many good features of these polymers such as their biological properties, targeting ability and biocompatibility might be modified as well (Jiang et al., 2008). Therefore, we decided to avoid such modifications in order to exploit the high biological value and properties of these biopolymers in their natural state. As a consequence, the option of selecting an additional component was chosen. In the selection of potential components we took into account that, as has been pointed out, the initial nanosystem platforms showed a high capacity to associate proteins. This feature prompted us to include proteins with an isoelectric point >7 (displaying a positive charge at physiological pH) in these initial prototype nanoparticles, such as gelatine, which has a natural isoelectric point of 9. However, preliminary experiments evidenced that the inherent positive charge of these molecules is not high enough to facilitate the association of pDNA or siRNA to the nanosystems (data not shown). For this reason, our strategy was focused on the use of proteins that were previously modified to provide them with suitable characteristics for nucleic acid physical interactions, taking advantage of the presence of functional groups in proteins such as carboxylic groups, which would allow to perform modifications in a controlled way. Concerning this strategy, it is necessary to clarify that the use of cationized proteins as biomaterials for the development of nanoparticles has been reported during the last few years (Kushibiki et al., 2006; Obata et al., 2012). However, this should be considered a critical limitation to the clinical exploitation of delivery systems based on these modified proteins so far developed, a limitation that is related to the use of organic solvents and chemical cross-linkers such as glutaraldehyde for the production of micro- and nanoparticles, which should be avoided due to its well-known toxicity and its potential to inactivate bioactive molecules (i.e. the associated bioactive molecules and/or additional natural components of these systems) (Jayakrishnan and Jameela, 1996; Zeiger et al., 2005). On the basis of these considerations, our aim was to develop modified proteins through a very simple reaction based on the use of endogenous compounds such as SPM, with a double objective. Firstly, the modification should result in a compound that is able to interact with the pDNA and siRNA molecules, as has already been demonstrated for other cationic proteins (Kushibiki et al., 2006; Saito and Tabata, 2012). Secondly, the modified proteins should be able to be incorporated to the initial prototype nanoparticles composition without affecting the aforementioned ionic cross-linking process. This is a critical aspect, taking into account the limited variations in the charge ratios balance observed in the preparation of numerous nanoparticulate systems. Therefore, the modification of the crosslinking agent or the incorporation of a highly charged high molecular weight moiety such as a charged protein is a difficult task, which can result in inappropriate characteristics, instability of the developed systems or even an inability to produce them. Only if the nanosystem platform in which these components should be incorporated is versatile enough can we envisage an incorporation success. From the results included in Table 4, it can be concluded that this is the situation described in the present

17

work, because optimised nanoparticulate systems have been successfully developed, including cationized gelatine in their composition, and providing an effective association capacity for siRNA that initial prototype platforms did not. In addition, the incorporation of CGsp contributed to improve the physicochemical characteristics of the initial HA:SPM nanoparticle prototypes, substantially diminishing their particle size and PDI (see also Supplementary Data), which renders these customised nanostructures more attractive for delivery applications. Another interesting result from the customisation strategy developed for these systems is the additional possibility that it can modulate their surface charge from negative to positive values by selecting appropriate mass ratios of their components or, in other words, by modifying the presence of CGsp (see Table 4). Once more, such a possibility indicates the efficacy of the optimisation process and the versatility of the resulting formulations. In order to verify the siRNA association efficacy we used a conventional electrophoretic assay (see Supplementary Data). This assay revealed the absence of the migration bands characteristic of free nucleic acids. We corroborated these results by quantifying the siRNA association efficacy by using the commercial Ribogreen1 kit. Table 4 shows that the customised nanoparticles show a high capacity (>80%) to associate siRNA regardless of their composition. 3.4. Biological evaluation of the customised nanoparticles After the customisation process of the initial prototype formulations previously described, we decided to select two different formulations from those described in Table 4, in order to perform a biological evaluation. The selected systems were HA:CGsp:SPM (3.5:30:1) and ChS:CGsp:SPM (2.5:30:1) that, due to their positive surface charge and composition, could have the potential to provide a significant interaction with the ocular surface structures in which we decided to perform the biological evaluation on the basis of our previous experience in gene therapy. For this purpose we selected a human cell line of cornea. It should be point out that we avoided the conventional cell lines used in most transfection studies. Indeed, few studies are addressed as this one in specific human cell lines with specific difficulties to be transfected, thus simulating the limitations found in an in vivo scenario, such as, in this case, those inherent to corneal cells and related to specific cellular membrane features and the production of a mucus barrier (Gipson and Inatomi, 1997).

Fig. 1. Percentage of cell viability in human corneal epithelial (HCE) cells after incubation of ABs:CGsp:SPM nanoparticles formulated with HA and ChS and associating siRNA against GAPDH (n = 3). (Data are mean  SD, *p > 0.05 ChS:CGsp: SPM vs. HA:CGsp:SPM).

18

J.E. Parraga et al. / International Journal of Pharmaceutics 477 (2014) 12–20

Fig. 2. Confocal fluorescence images in human corneal epithelial cells after the incubation with nanoparticles labelled with FITC. (A) fl–ChS:CGsp:SPM (2.5:30:1); (B) fl–HA: CGsp:SPM (3.5:30:1). The images show, from the left to the right, the three separate channels, the overlay of the images and the cross-section x–y,y–z and x–z axis, respectively.

For such evaluation the nanoparticles were not centrifuged and concentrated. As stated before, the yield of nanoparticle formation ranges between 40 and 70%. Therefore and taking into account that in the described preparation technique the crosslinker (spermine) is the limiting reactant in the nanoparticle formation, minimum or negligible amounts of free crosslinking agent should be expected. Fig. 1 shows the viability of human corneal epithelial (HCE) cells after incubation with these nanosystems. Taking into account that cytotoxicity is directly related to the positively charged moieties in non-viral gene therapy formulations, we focused our attention in these studies to the amount of CGsp used in the nanoparticles composition. As we can see in Fig. 1, the viability remains high at concentrations that would be subsequently used for transfection assays (10–30 mg/cm2) regardless of the nanoparticles composition. However, at 100 mg/cm2, the nanosystems showed some slight differences and the HA/CGsp/SPM formulation appeared to elicit higher cytotoxicity compared with those that incorporated ChS. In this respect, although is well known that the inclusion of glycosaminoglycans among the formulation components diminishes the toxicity induced by cationic polymers (Zorzi et al., 2011), we can see how this effect depends on the composition or specific glycosaminoglycan incorporated, being more favourable in our case when incorporating ChS. We already know from our previous studies that nanoparticles with a high proportion of ABs in their composition show a better biocompatibility than those with a lower concentration and, in addition, we also evidenced that that this tendency is composition-dependent (Parraga et al., 2013). In order to perform cellular internalisation studies of the developed formulations, we previously labelled the nanoparticles with a fluorescent tag. For this purpose, we modified HA and ChS by the covalent attachment of fluorescein-amine (FITC) in their backbone structure. Fig. 2 shows the differences in the performance of the different nanosystems. Nanoparticles formulated with HA appear to disintegrate more rapidly after the internalisation process, under the studied conditions, in comparison to those formulated with ChS. It is widely acknowledged that polymeric nanoparticles are often internalised by endocytic pathways (Torchilin, 2006) and lysosomes are responsible for

degrading the active molecule if the nanoparticles cannot escape from early or late endosomes (endo-lysosomes) (Panyam et al., 2002). In this context, the different behaviours of the formulations HA:CGsp:SPM and ChS:CGsp:SPM at the intracellular level, could be related to the aforementioned higher or stronger ability of ChS to interact with the cross-linker agent and thus generate more compact and robust or stronger nanostructures than using HA. Here, HA-based nanoparticles may suffer a very fast destabilisation at the acidic pH present into the endo-lysosomes or lysosomes, which prompted the breakdown of these nanoparticles. Therefore, this may facilitate easier access to and the subsequent degradation of HA by the hydrolases present in the lysosomes than ChS. In order to evaluate whether the ChS:CGsp:SPM nanoparticles are able to deliver the siRNA to the target site (the cytosolic compartment), we designed nanoparticles with a double label. In addition to the inclusion of fl–ChS in the nanostructure, siRNA

Fig. 3. Confocal fluorescence microscopy images in human corneal epithelial cells of double-labelled nanoparticles fl–ChS:CGsp:SPM (2.5:30:1) associating siGAPDHCy3. The cell nuclei were stained with DAPI (blue light), the F-actin filaments of the cytoskeleton were stained with phalloidin (dark blue) and the nanoparticles were labelled with FITC (green). The images show the four separate channels, the overlay of the images (A) and the cross-section x–y, y–z and x–z axis (B). Colour composition: nanoparticles + siGAPDH–Cy3 (orange–yellow), empty nanoparticles (green), free siRNA–Cy3 (red). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

J.E. Parraga et al. / International Journal of Pharmaceutics 477 (2014) 12–20

19

Fig. 4. Inhibition of GAPDH expression in human corneal epithelial cells after incubation with nanoparticles associating siRNA at different concentrations (positive control: lipofectamine; negative control: non-specific siRNA).

labelled with Cy3 was also associated with the systems. Fig. 3 confirms that the nanoparticles are able to deliver the siRNA in the cytosol. The free siRNA is indicated by red colour, while the encapsulated siRNA displays orange or yellow tonality depending on the diffusion of the siRNA to the surface of the particles. Indeed, we hypothesised that the orange particle colour is caused by the Cy3–siRNA being located close to or on the surface of the nanosystems, while the yellow colour would be caused by deeply-associated siRNA. On the other hand, blank nanoparticles are recognised because they can only show the green colour that displays the fl–ChS. To further corroborate whether the internalised nanoparticles associating siRNA are able to release the genetic material properly, we evaluated their gene-interfering potential, using GAPDH as a model target gene. In such experiments, the observation of a silencing effect can only be interpreted as a concluding proof-ofconcept of effective targeted delivery of the associated siRNA. As shown in Fig. 4, the developed nanosystems were able to induce the knock-down of the targeted gene with a comparable efficiency (60% knock down) to that of a commercially available silencing tool, Lipofectamine 2000, which was used as a positive silencing control at the ratio siRNA:Lipofectamine recommended by the manufacturer’s instructions and the amount of lipofectamine was adjusted according to the cytotoxicity showed by the cell line (35 nM). On the other hand, despite the different intracellular behaviours of the nanosystems formulated with HA or ChS, they inhibited the expression of the model protein in the same proportion, indicating that the optimised nanoparticles, independent of their composition, are able to protect the siRNA from the degradation by enzymes, to escape from the endosomes and to release the siRNA in its active form on the site of action, thus enabling the silencing effect observed.

In addition to these positive results, two additional potentials or added values of the developed nanosystems should be taken into account: (i) the advantages from the toxicological point of view that represents the incorporation of HA or ChS into the nanostructures; and (ii) their well-known affinity for specific receptors such as CD44, which are expressed in several epithelial cells, for instance ocular epithelial cells or cancer cells (Contreras-Ruiz et al., 2011; Marhaba and Zoller, 2004), which could open up applications of these nanosystems to achieve active targeting of several kinds of hydrophilic therapeutic molecules such as proteins, peptides or genetic material to specific tissues in the frame of specific treatments. Finally, in order to evaluate the in vivo ocular penetration of the developed nanosystems, we selected the fl–ChS/CGsp/SPM formulation to associate fluorescently-labelled siRNA–Cy3. After their topical ocular instillation to rabbits, corneal and conjunctival tissues were analysed. The results obtained revealed a different behaviour at these two levels. Thus, in Fig. 5 we can appreciate a higher number of green-coloured particles into the conjunctival epithelium compared with the corneal epithelium, where more yellow-coloured particles were detected. These observations suggest an easier and/or faster siRNA release from the nanoparticles in the conjunctival tissues than in the corneal epithelium. This is in concordance with previous experiments performed in our research group with particles which include ChS for the delivery of pDNA, which also provided higher transfection levels in conjunctiva cells than in cornea cells (Konat Zorzi et al., 2011). However, deeper studies should be required to establish which biological conditions displayed by the different ocular tissues could affect the ocular performance of these nanosystems in terms of the delivery of associated active molecules.

Fig. 5. Confocal fluorescence microscopy images of the rabbit conjunctiva (A) and cornea (B) excised after 3 h post-instillation of nanoparticles fl–ChS:CGsp:SPM (2.5:30:1). C-1 and C-2 show the images obtained after the instillation of a control siRNA–Cy3 + fl–ChS solution in cornea and conjunctiva, respectively.

20

J.E. Parraga et al. / International Journal of Pharmaceutics 477 (2014) 12–20

4. Conclusions The special properties of the endogenous polyamine spermine together with a judicious selection of anionic natural polymers allowed us to design extremely versatile nanoparticulate delivery platforms. Indeed, the designed nanocarriers can be easily tuned to effectively associate labile biomolecules with very different properties, such as proteins, peptides and nucleic acids. Concretely, the introduction among their components of a modified protein such as gelatine cationized with spermine is a successful strategy to associate nucleic acids. Moreover, these delivery platforms can be conveniently modulated in respect to the most important properties responsible for their interactions with living cells and tissues (i.e. surface charge and composition). The potential of these nanoparticles as delivery systems has been explored for siRNA and the obtained results confirmed that these systems are able to penetrate in vivo into the cornea and conjunctiva tissues, and to provide an effective siRNA internalisation and transfection in an ocular in vitro corneal model, thus resulting in a significant gene silencing of similar magnitude than that obtained with transfection laboratory reagents (i.e. Lipofectamine). Therefore, we can conclude that these nanosystems constitute robust and versatile delivery platforms that would serve as the basis to develop custom-made nanosystems able to fulfil the association requirements of specific delicate bioactive molecules. Acknowledgements This work was supported by grants of the UE (Seventh Framework Programme-ERA-NET  MATERA+, MATERA/ BBM-1856 10TMT203011PR), the Ministry of Education and Science, Spain(MAT2010-20452-C03-01 and -02) and Xunta de Galicia (Competitive Reference Groups, Ref. 2010/18, FEDER Funds). J. Párraga acknowledges the Ministry of Economy and Competitiveness, Spain for the FPI scholarship (BES-2008-001998) and G. Zorzi acknowledges the Programme AlBan the European Union Programme of High Level Scholarships for Latin America (Scholarship No. E07D402978BR). Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j. ijpharm.2014.09.049. References Blow, N., 2007. Small RNAs: delivering the future. Nature 450, 1117–1120. Byrne, J.D., Betancourt, T., Brannon-Peppas, L., 2008. Active targeting schemes for nanoparticle systems in cancer therapeutics. Adv. Drug Deliv. Rev. 60, 1615–1626. Contreras-Ruiz, L., de la Fuente, M., Parraga, J.E., Lopez-Garcia, A., Fernandez, I., Seijo, B., Sanchez, A., Calonge, M., Diebold, Y., 2011. Intracellular trafficking of hyaluronic acid–chitosan oligomer-based nanoparticles in cultured human ocular surface cells. Mol. Vision 17, 279–290. de Belder, A.N., Wik, K.O., 1975. Preparation and properties of fluorescein-labelled hyaluronate. Carbohydr. Res. 44, 251–257. de Fougerolles, A., Vornlocher, H.P., Maraganore, J., Lieberman, J., 2007. Interfering with disease: a progress report on siRNA-based therapeutics. Nat. Rev. Drug Discov. 6, 443–453. Farokhzad, O.C., Langer, R., 2009. Impact of nanotechnology on drug delivery. ACS Nano 3, 16–20. Gipson, I.K., Inatomi, T., 1997. Mucin genes expressed by the ocular surface epithelium. Prog. Retina Eye Res. 16 (1), 81–98. Howard, K.A., Dash, P.R., Read, M.L., Ward, K., Tomkins, L.M., Nazarova, O., Ulbrich, K., Seymour, L.W., 2000. Influence of hydrophilicity of cationic polymers on the

biophysical properties of polyelectrolyte complexes formed by self-assembly with DNA. Biochim. Biophys. Acta 1475, 245–255. Jayakrishnan, A., Jameela, S.R., 1996. Glutaraldehyde as a fixative in bioprostheses and drug delivery matrices. Biomaterials 17, 471–484. Jiang, G., Park, K., Kim, J., Kim, K.S., Oh, E.J., Kang, H., Han, S.E., Oh, Y.K., Park, T.G., Kwang Hahn, S., 2008. Hyaluronic acid-polyethyleneimine conjugate for target specific intracellular delivery of siRNA. Biopolymers 89, 635–642. Kamaly, N., Xiao, Z., Valencia, P.M., Radovic-Moreno, A.F., Farokhzad, O.C., 2012. Targeted polymeric therapeutic nanoparticles: design, development and clinical translation. Chem. Soc. Rev. 41, 2971–3010. Kang, H.C., Huh, K.M., Bae, Y.H., 2012. Polymeric nucleic acid carriers: current issues and novel design approaches. J. Control. Release 164, 256–264. Konat Zorzi, G., Contreras-Ruiz, L., Parraga, J.E., Lopez-Garcia, A., Romero Bello, R., Diebold, Y., Seijo, B., Sanchez, A., 2011. Expression of MUC5AC in ocular surface epithelial cells using cationized gelatine nanoparticles. Mol. Pharm. 8, 1783–1788. Kreuger, J., Spillmann, D., Li, J.P., Lindahl, U., 2006. Interactions between heparan sulfate and proteins: the concept of specificity. J. Cell Biol. 174, 323–327. Kushibiki, T., Tomoshige, R., Iwanaga, K., Kakemi, M., Tabata, Y., 2006. In vitro transfection of plasmid DNA by cationized gelatine prepared from different amine compounds. J. Biomater. Sci. 17, 645–658. Lapcík Jr., L., Lapcík, L., De Smedt, S., Demeester, J., Chabrecek, P., 1998. Hyaluronan: preparation, structure, properties, and applications. Chem. Rev. 98, 2663–2684. Larsson, B., Nilsson, M., Tjälve, H., 1981. The binding of inorganic and organic cations and H+ to cartilage in vitro. Biochem. Pharmacol. 30, 2963–2970. Li, W., Szoka Jr., F.C., 2007. Lipid-based nanoparticles for nucleic acid delivery. Pharm. Res. 24, 438–449. Liu, X., Howard, K.A., Dong, M., Andersen, M.O., Rahbek, U.L., Johnsen, M.G., Hansen, O.C., Besenbacher, F., Kjems, J., 2007. The influence of polymeric properties on chitosan/siRNA nanoparticle formulation and gene silencing. Biomaterials 28, 1280–1288. Lopez-Cebral, R., Martin-Pastor, M., Parraga, J.E., Zorzi, G.K., Seijo, B., Sanchez, A., 2011. Chemically modified gelatin as biomaterial in the design of new nanomedicines. Med. Chem. 7, 145–154. Maham, A., Tang, Z., Wu, H., Wang, J., Lin, Y., 2009. Protein-based nanomedicine platforms for drug delivery. Small 5, 1706–1721. Marhaba, R., Zoller, M., 2004. CD44 in cancer progression: adhesion, migration and growth regulation. J. Mol. Histol. 35, 211–231. Obata, Y., Nishino, T., Kushibiki, T., Tomoshige, R., Xia, Z., Miyazaki, M., Abe, K., Koji, T., Tabata, Y., Kohno, S., 2012. HSP47 siRNA conjugated with cationized gelatine microspheres suppresses peritoneal fibrosis in mice. Acta Biomater. 8, 2688–2696. Panyam, J., Zhou, W.Z., Prabha, S., Sahoo, S.K., Labhasetwar, V., 2002. Rapid endolysosomal escape of poly(DL-lactide-co-glycolide) nanoparticles: implications for drug and gene delivery. FASEB J. 16, 1217–1226. Parraga J.E., P.A., Seijo B., Sanchez A., 2013. Glycosaminoglycans/cationic gelatine based nanoparticles for siAkt delivery in breast cancer cells. New Trends in Pharmaceutical Sciences. III Congress of the Portuguese Society of Pharmaceutical Sciences and IX Spanish-Portuguese Conference on Controlled drug delivery, Oporto, Portugal. October 13–15, 2011. (Eds.). Revista Portuguesa de Farmácia. Ordem dos Farmacêuticos, Oporto, 2011, Abstract 28, 139–140. Peppas, N.A., Langer, R.,1994. New challenges in biomaterials. Science 263,1715–1720. Phillips, M.A., Gran, M.L., Peppas, N.A., 2010. Targeted nanodelivery of drugs and diagnostics. Nano Today 5, 143–159. Pirollo, K.F., Chang, E.H., 2008. Does a targeting ligand influence nanoparticle tumour localisation or uptake? Trends Biotechnol. 26, 552–558. Raemdonck, K., Martens, T.F., Braeckmans, K., Demeester, J., De Smedt, S.C., 2013. Polysaccharide-based nucleic acid nanoformulations. Adv. Drug Deliv. Rev. 65, 1123–1147. Raman, R., Sasisekharan, V., Sasisekharan, R., 2005. Structural insights into biological roles of protein–glycosaminoglycan interactions. Chem. Biol. 12, 267–277. Saito, T., Tabata, Y., 2012. Preparation of gelatine hydrogels incorporating small interfering RNA for the controlled release. J. Drug Target. 20, 864–872. Sanchez A., Seijo B., Zorzi K., Paolicelli P., Parraga J.E., University of Santiago de Compostela. Nanoparticulate systems prepared from anionic polymers, WO 2010/049562 A, May 6, 2010. Sasisekharan, R., Raman, R., Prabhakar, V., 2006. Glycomics approach to structure– function relationships of glycosaminoglycans. Ann. Rev. Biomed. Eng. 8, 181–231. Torchilin, V.P., 2006. Recent approaches to intracellular delivery of drugs and DNA and organelle targeting. Ann. Rev. Biomed. Eng. 8, 343–375. Vimr, E.R., Kalivoda, K.A., Deszo, E.L., Steenbergen, S.M., 2004. Diversity of microbial sialic acid metabolism. Microbiol. Mol. Biol. Rev. 68, 132–153. Whitehead, K.A., Langer, R., Anderson, D.G., 2009. Knocking down barriers: advances in siRNA delivery. Nat. Rev. Drug Discov. 8, 129–138. Zeiger, E., Gollapudi, B., Spencer, P., 2005. Genetic toxicity and carcinogenicity studies of glutaraldehyde–a review. Mutat. Res. 589, 136–151. Zorzi, G.K., Parraga, J.E., Seijo, B., Sanchez, A., 2011. Hybrid nanoparticle design based on cationized gelatine and the polyanions dextran sulfate and chondroitin sulfate for ocular gene therapy. Macromol. Biosci. 11, 905–913.