at SciVerse ScienceDirect

Biomaterials 34 (2013) 2130e2141

Contents lists available

Biomaterials

journal homepage: www.elsevier .com/locate/biomateria ls

The neurotoxicity of gene vectors and its amelioration by packaging with collagenhollow spheres

Ben Newland a, Teresa C. Moloney a, Gianluca Fontana a, Shane Browne a, Mohammad T. Abu-Rub a,Eilís Dowd b, Abhay S. Pandit a,*aNetwork of Excellence for Functional Biomaterials (NFB), IDA Business Park, Dangan, National University of Ireland Galway, Galway, Irelandb The Department of Pharmacology & Therapeutics, National University of Ireland, Galway, Ireland

a r t i c l e i n f o

Article history:Received 3 October 2012Accepted 27 November 2012Available online 13 December 2012

Keywords:BrainApoptosisPolymer vectorsGene deliveryToxicityCollagen microspheres

* Corresponding author. Tel.: þ353 91 49 2758.E-mail address: abhay.pandit@nuigalway.ie (A.S. P

0142-9612/$ e see front matter � 2012 Elsevier Ltd.http://dx.doi.org/10.1016/j.biomaterials.2012.11.049

a b s t r a c t

Over the last twenty years there have been several reports on the use of nonviral vectors to facilitate genetransfer in the mammalian brain. Whilst a large emphasis has been placed on vector transfection effi-ciency, the study of the adverse effects upon the brain, caused by the vectors themselves, remainscompletely overshadowed. To this end, a study was undertaken to study the tissue response to threecommercially available transfection agents in the brain of adult Sprague Dawley rats. The response tothese transfection agents was compared to adeno-associated viral vector (AAV), PBS and naked DNA.Furthermore, the use of a collagen hollow sphere (CHS) sustained delivery system was analysed for itsability to reduce striatal toxicity of the most predominantly studied polymer vector, polyethyleneimine(PEI). The size of the gross tissue loss at the injection site was analysed after immunohistochemicalstaining and was used as an indication of acute toxicity. Polymeric vectors showed similar levels of acutebrain toxicity as seen with AAV, and CHS were able to significantly reduce the toxicity of the PEI vector. Inaddition; the host response to the vectors was measured in terms of reactive astrocytes and microglialcell recruitment. To understand whether this gross tissue loss was caused by the direct toxicity of thevectors themselves an in vitro study on primary astrocytes was conducted. All vectors reduced theviability of the cells which is brought about by direct necrosis and apoptosis. The CHS delivery systemreduced cell necrosis in the early stages of post administration. In conclusion, whilst polymeric genevectors cause acute necrosis, administration in the brain causes adverse effects no worse than that of anAAV vector. Furthermore, packaging the PEI vector with CHS reduces surface charge and direct toxicitywithout elevating the host response.

� 2012 Elsevier Ltd. All rights reserved.

1. Introduction

Despite extensive proof that growth factors show neuro-protective properties in animal models of disease [1e4], thedelivery of proteins to the brain has often failed to produce thedesired therapeutic effect [5e7]. Such therapeutic strategies areintrinsically limited by a short protein half-life in the intercellularspace, resulting in short term beneficial effects [8]. To circumventthis drawback, the use of genetic manipulation of the host cells bytransfection with specific nucleic acids can lead to longer termover-expression or knock down of a desired protein and hence havegreater therapeutic benefit. However to obtain greater efficacy,usually a vector is required for enhanced delivery and protection ofthe nucleic acid from degradation [9].

andit).

All rights reserved.

Viral vectors provide an efficient means of nucleic acid deliveryto the brain, and although generally less efficacious, polymericvectors have also received much attention for brain applications asan alternative which do not pose the same safety concerns. Whilstnumerous studies have examined the capability of liposome orpolymer vectors to transfect various regions of the mammalianbrain following intracerebral injection, very few studies have re-ported on the negative side effects of these vectors. Administrationof liposomes directly to the rodent brain for the purpose of genetransfer was reported as early as 1990, with studies using thecommercially available Lipofectin� transfection agent [10,11].Subsequent studies have been reported with other liposomalformulations such as lipospermine vectors [12]; however thetoxicity of these liposomal vectors used was not assessed. Thesubject of the vector toxicity is typically reflected upon by quali-tative observations such as lack of extensive necrosis/brain alter-ations [13]. However, semi-quantitative analysis of the toxic effectof the Lipofectamine transfection agent on a range of brain derived

B. Newland et al. / Biomaterials 34 (2013) 2130e2141 2131

cell types was performed in vitro via Trypan blue staining prior toan in vivo study [14]. This study showed that Lipofectamine causeddose dependant toxicity. It is noteworthy that this analysis wasconducted with the lipid solution alone, not with complexed pDNA.It is therefore likely that the solution analysed was more cationicthan that of the liposomal-nucleic acid complex (lipoplex) admin-istered to the brain, through the interaction with pDNA resulting ina reduction of charge.

Similarly, polymeric vectors used in the brain have not beenstudied for in vivo toxicity at the site of injection. Of particularlynote is polyethyleneimine (PEI), where the majority of studies havereported on vector efficacy [15e18]. Comparative studies of vectorsystems only assess in vivo efficacy or in vitro toxicity [19]. Morerecently, polyethylene glycol moieties to PEI have been added,either to tether neuron targeting ligands [20] or to reduce thetoxicity of the vectors themselves [21]. The measure of the reduc-tion in the toxicity was ascertained through in vitro assessment onNT2 cell cultures, but not measured in vivo [21].

The toxicity of a material implanted in the brain can be analysedby various parameters, in terms of overt toxicity by quantifyingtissue loss, or by measuring the adverse host response by way ofglial cell recruitment to the injury site [22]. Microglia and astro-cytes, the immune effector cells of the central nervous system [23],remain in a resting state until an insult occurs. Following an insult,such as a needle stick injury, the microglia are recruited to the siteof injury and can remain in the activated state for approximatelyone week [24]. For analysing the microglial response to implantedbiomaterials, immunohistochemical processing has been per-formed with CD56 or OX-42 antibodies, showing microglialrecruitment to the site of injection [24,25]. Astrocytes, another glialcell, involved with maintenance of neuronal sustenance, alsomigrate to the site of the injury to form a glial scar; a barrierbetween the affected region and the surrounding healthy tissue[26]. Antibodies specific to the glial fibrillary acidic protein (GFAP)can be used to identify astrocytes at the injury site, which typicallyappear to surround the injection site [27].

The majority of non-viral gene delivery systems used in thebrain have focused primarily on transfection efficacy whilstcomprehensive toxicity analyses of such vectorse bymeans of hostastrocyte and microglia response e remains sidelined. This studytherefore aims to specifically investigate the adverse effects ofnonviral vectors towards brain tissue and primary astrocytes.Secondly, this study explores whether a sustained deliveryapproach can ameliorate these effects, without inferring a largerhost response. We hypothesise that the direct cytotoxic effects ofnonviral vector systems can be reduced using collagen hollowspheres (CHS) by controlling the rate of vector delivery and byreducing the overall surface charge of the system.

2. Methods

2.1. Outline of the experimental design

The first objective of this study was to determine the size of the gross tissue loss,astroglial response and microglial activation caused by intrastriatal administrationof three commonly used and commercially available transfection agents: Super-Fect�, PEI and Lipofectin�. SuperFect�, a polyamidoamine (PAMAM) partiallydegraded dendrimer, and PEI are dendritic and branched polymer vectors respec-tively, while Lipofectin� is a cationic lipid. The polyplexes/lipoplexes formed usingthese vectors were analysed in the Sprague Dawley rat brain in comparison to PBS,naked pDNA and an adeno-associated virus encoding green fluorescent protein(AAV2/5-GFP). The second objective was to analyse whether protective hollowmicrospheres composed of collagen can reduce the cerebral toxicity caused by thepolyplexes in terms of the above stated parameters. The third objective of this studywas to ascertain the likely mechanisms of cell death that occur due to the exposureof the cells to the different nonviral vectors. An in vitro culture of primary astrocyteswas used to study the effects of the nonviral treatment groups on cellular metabolicactivity, membrane integrity, apoptosis, necrosis and cellular proliferation.

A total of 20 male Sprague Dawley rats were used, each weighing 280 � 30 g onthe day of surgery. Rats were randomly assigned to groups and received intrastriatalinjections of 4 ml of PBS, naked DNA (2 mg), SuperFect� polyplexes (with 2 mg DNA),Lipofectin� lipoplexes (with 2 mg DNA), PEI polyplexes (with 2 mg DNA) or collagenhollow spheres loaded with PEI polyplexes (with 2 mg DNA) (n¼ 3 injection sites pergroup). An additional group of animals were injected with 2 ml of AAV-GFP forcomparison (prepared as reported previously [28], titre ¼ 1.07 � 1010 drp/ml, n ¼ 2injection sites). At seven days post injection, the animals were sacrificed, andthe brains were explanted and processed for analysis of gross tissue loss, andassessment of tissue response was determined by quantifying microglial recruit-ment (via staining for themicrogliamarker OX42), and astrocyte reactivity (via GFAPstaining).

2.2. Materials

PEI (Sigma) was prepared in a solution of RNAse free water (Sigma) ata concentration of 3 mg/ml. SuperFect� (Invitrogen) and Lipofectin� (Sigma) wereused as received according to the manufacturers’ protocol. The plasmid expressinggreen fluorescent protein (pCMV-GFP) was obtained from New England BioLabs,U.K. The alamarBlue� Cell Viability Assay (Invitrogen), CellTiter-Glo� LuminescentCell Viability Assay (Promega), CytoTox 96� NonRadioactive Cytotoxicity Assay(Promega) and Apoptotic/Necrotic/Healthy Cell Indicator (Promokine) were usedaccording to manufacturer’s protocols.

2.3. Formation of polyplexes/lipoplexes

The polymer/DNA complexes (polyplexes) or liposome/DNA complexes (lip-oplexes) were formed at ratios determined previously [29,30] using a plasmidencoding GFP. Keeping the amount of DNA constant, the amount of polymer/lipo-some to DNAwas varied according to the following weight ratios: Naked DNA (0:1),SuperFect� (8:1), PEI (2:1), and Lipofectin� (10:1). To maintain constant volumes,a “top-up” solution of sterile PBS was required. Thus, polyplex/lipoplex formationwas performed simply through the addition of the polymer/liposome solution to theplasmid solution followed by the addition of the top-up solution and incubation atroom temperature for 30 min prior to use.

2.4. Characterisation of polyplexes/lipoplexes

Polyplexes or lipoplexes were prepared as stated above, and were characterisedby three techniques. Gel electrophoresis was performed using a 0.7% agarose gel inTris-borate-EDTA (TBE) buffer with 10 ml of SYBRsafe gel stain. 1 mg of control DNA orequivalent quantity of polyplexes/lipoplexes was placed within the wells of the gelwith an equal volume of loading dye. This gel was then subjected to a potentialdifference of 80 mV for 20 min and imaged using a UV filter and GeneSnap software(Thermo Scientific). Both size and charge analysis of the polyplexes/lipoplexes wascarried out using a Zetasiser (Malvern Instruments), by the addition of polyplexes/lipoplexes containing 5 mg of DNA dispersed in water to a disposable zeta cell, andmeasurement of a minimum of 20 readings per sample (n ¼ 4). A drop of thisdispersion was placed on a copper grid and allowed to dry for subsequent imagingby transmission electron microscopy (TEM) (Hitachi).

2.5. Synthesis and loading of collagen hollow spheres (CHS)

CHS were fabricated using a template coating method [31] as reported previ-ously [32,33]. Briefly, commercially available polystyrene beads (Gentaur, Chicago,Illinois) were sulfonated to increase their negative charge. Following sulphonation,beads were coated by incubation in a collagen solution at acidic pH. Subsequent tocoating, collagen was crosslinked with pentaerythritol poly(ethylene glycol) ethertetrasuccinimidyl glutarate (4S-PEG)(Sigma). To produce a hollow sphere, thepolystyrene core template was dissolved by washing the coated beads with tetra-hydrofuran (THF). The spheres were then washed at least three times with distilledwater. CHS were loaded with polyplexes following the previously reported method[33]. Polyplexes formed using PEI, were loaded in CHS at a concentration of 40 mg ofplasmid (in polyplexes) per mg of CHS. For both in vivo and in vitro studies, poly-plexes containing 40 mg of plasmid were added to 1 mg of CHS in 1 ml of dH2O.Following 8 h of incubation, the samples were centrifuged at 13,000 rpm for 10 minand washed once with dH2O to remove un-loaded polyplexes. After the secondcentrifugation, the loaded sphereswere re-suspended in 40 ml of sterile H2O (4 ml perinjection site, or 4 ml per well for in vitro studies).

To assess the percentage of polyplexes loaded, plasmid DNA was labelled usinga Cy5 Label-IT kit as described previously [34], and was used to form the polyplexesin the solution to be loaded in CHS. A 20 ml sample of this solution was removed forlater assessment of total fluorescence (control sample), CHS were then added to thesolution and incubated for 8 h. This solution was then centrifuged to pellet the CHS,and 20 ml of the supernatant was removed for subsequent fluorescence analysis(sample supernatant) alongside the control sample. These samples were thendiluted in 100 ml of dH2O in an opaque black 96-well plate and read with a platereader (Varioskan Flash, Thermo Scientific, Ireland) using the following excitationand emission wavelengths: ex ¼ 649 nm, em ¼ 670 nm. Thus the fluorescence

B. Newland et al. / Biomaterials 34 (2013) 2130e21412132

remaining in the supernatant (non-loaded polyplexes) minus the original fluores-cence (normalised to 100%) gave the percentage of loaded polyplexes.

2.6. Toxicity of gene vectors in vivo

2.6.1. SurgeryAll procedures were carried out in accordance with the European Communities

Council Directive (86/609/EEC), and were approved by the Animal Ethics Committeeof the National University of Ireland, Galway. Male Sprague Dawley rats (CharlesRivers,UK)wereused in all experiments. Ratsweredeeplyanesthetisedusinggaseousisoflurane (2e5% in oxygen), and mounted via teeth and ear bars. Bilateral injectionswere targeted at the striatum via the following sterotactic coordinates: Ante-roposterior 0.0mm,Mediolateral�3.7mm(fromBregma) andDorsoventral�5.0mm(from Dura). An injection volume of 4 ml was delivered to each injection site,controlledbya syringepumpat a rateof 1mg/minwithanadditional 2minof diffusiontime following injection before cannula removal.

2.6.2. Tissue processingRats were deeply anesthetised by intraperitoneal injection of pentobarbital

(100mg per kg bodyweight) and transcardially perfused using 100ml of heparinisedsaline followed immediately with 150 ml of cold phosphate buffered formaldehydesolution (4%). Removed brains were left in the fixative for a further 4 h before beingtransferred to a phosphate buffered sucrose solution (25%) for 24 h. A freezing stagesledge microtome (Bright, UK), was used to cut 40 mm serial coronal sections.Immunohistochemical analysis of astrocytosis and the recruitment of microglia atthe site of injectionwere performed by labelling astrocytes and microglia with GFAPantibodies (rabbit anti-GFAP, Dako, UK) and OX42 antibodies (mouse anti-OX42,Chemicon, Ireland) respectively. All quantitative immunohistochemistry was con-ducted in a blind manner to the treatment of the rats.

The freely floating sections were firstly treated to quench endogenous perox-idise activity (3% hydrogen peroxide/10%methanol in dH2O). Following this step andin-between all subsequent steps, a thrice performed 10minwash with Tris-bufferedsaline (TBS) was performed. All incubations were undertaken at room temperaturewith gentle agitation. Blocking of non-specific antibody binding was then performedfor 1 h (3% normal horse/goat serum, 0.2% Triton-X 100 in (TBS)). Primary antibodiesdiluted (1:2000 for GFAP, 1:400 for OX42) in 0.2% Triton-X 100 in TBS were incu-bated with the sections overnight. The following day, the sections were incubatedfor 3 h with biotinylated secondary antibodies (goat anti-rabbit for GFAP; Jackson,UK, or horse anti-mouse for OX42; Vector, UK) diluted 1:200 in 1% normal horseserum/goat serum in TBS. In order to develop the sections using diaminobenzidinetetrahydrochloride (DAB) staining, the sections were first incubated for 2 h withstreptavidinebiotinehorseradish peroxidase solution (ABC kit; Dako, UK). DABsolution was then added (5 min e until brown staining develops)(0.5% DAB, 0.3 ml/ml hydrogen peroxide in Tris buffer). After three washings in Tris buffer, sectionswere mounted on Superfrost� Plus microscope slides (Thermo Scientific, UK),dehydrated in ethanol, washed in xylene and cover slipped with DPX mountant(BDH chemicals, UK).

2.6.3. Analysis of astrocytosis, migroglial activation and gross tissue lossPhotomicrographs were taken of all sections surrounding the visible injection

site (Olympus BX51 Microscope). Image J software was used to quantify the opticaldensity of the immune staining of either GFAP or OX42 at the immediate vicinity ofthe injection site after subtraction of the background optical density. Image J soft-warewas also used tomeasure the volume of gross tissue loss by applying Cavalieri’sprinciple (applied previously for volumetric analysis [35]) to the gross tissue lossareas in each 40 mm section. For volumetric fraction analysis, the total striatalvolumewas measured by applying Cavalieri’s principle to striatal areas. Examples ofthe anterior, median and posterior sections are shown in SI Fig. 4 as outlined else-where [36].

2.7. In vitro analysis of gene vector toxicity

2.7.1. Cell culturePrimary astrocytes were extracted from newborn pups (day 3) and plated by a

combination of two protocols [37,38] to yield astrocytes of over 90% purity (analysedby GFAP staining e data not shown). All cells were used at sixth passage or lower,and cultured using standard sterile techniques in T75 flasks at 37� C in humidified 5%CO2 environment, using Dulbecco’s Modified Eagles Medium (Sigma) supplementedwith F12 Ham mixture (50%)(Sigma), foetal calf serum (10%) and penicillin/strep-tomycin (1%). Prior to analysis, cells were seeded at a density of 10,000 cells/well in96-well plates, or the xCELLigence� 16 well E-plates (Roche Diagnostics).

2.7.2. Analysis of cytotoxicityFormulations of equal volumes (20 ml) of controls/polyplexes/lipoplexes (con-

taining 1 mg of DNA per well) or polyplex loaded spheres were added to the wells24 h prior to assessment. For the analysis of polyplexes in comparison to loaded CHS,a larger dose was used (3 mg of complexed DNA), so as to assess the same quantitiesof DNA. This is because CHS are successfully loaded with 3 mg of complexed DNA, sofor these studies, the polyplex alone groups also contained 3 mg of DNA.

AlamarBlue�, CellTiter-Glo� and CytoTox 96� assays were all performed accordingto the manufacturer’s protocol using a multiwell plate reader (Variskan Flash e

Thermo Scientific�), taking into account the subtraction of blank readings for allassays, and normalising to control cells where appropriate. The real-time prolifer-ation/death data obtained with the xCELLigence� system was analysed using theRTCA software (Roche Diagnostics). This systemmeasures the attachment of cells tothe well plate surface that is covered with gold electrodes by a change in impedancevalues. These impedance values are stated as a “cell index” value e essentiallya measure of area of cell attachment. For necrotic and apoptotic cell imaging, theApoptotic/Necrotic/Healthy Cell Indicator (Promokine) kit was used which con-tained fluorescein isothiocyanate (FITC) labelled Annexin V, Ethidium Homodimerand 40 ,6-diamidino-2-phenylindole (DAPI) nuclear stain. However, the primaryastrocytes were seeded at the same cell density in 8-well glass chamber slides(Lab-Tek), previously coated with poly-L-lysine to enhance cell attachment. Poly-plexes/lipoplexes were added in the samemanner as described above, and 24 h afteraddition, the cells were subjected to the addition of the kit contents according toprotocol, then were fixed in a 4% formaldehyde solution made up in PBS (n ¼ 3).Quantification of necrotic and apoptotic cells was performed by taking a minimumof 10 images (average ¼ 14) at random across each well using a fluorescencemicroscope (Olympus BX51 Microscope) fitted with a 10� lens. Subsequently, usingImage J software, the images from the three filters were then made to binary,a watershed added to take overlapping nuclei into account and the number ofapoptotic and necrotic cells were counted as a percentage of total cells. 20� imageswere also taken for representative images.

2.8. Statistical analyses

All statistical analyses were performed using GraphPad Prism 5 software andP values < 0.05 were considered significantly different. For all in vivo experimentscomparing vector groups with PBS a KruskaleWallis test was performed coupledwith a Dunn’s Multiple Comparison Test to compare all groups with PBS. To comparethe polyplex alone group with polyplex þ spheres group, a Student T-test wasperformed. For comparison of vectors alone versus vectors loaded to CHS and foranalysis of all in vitro experiments a one-way ANOVA was performed using Tukey’spost hoc test to compare all groups.

3. Results

3.1. Characterisation of polyplexes/lipoplexes

Before assessing the toxicity of the polyplexes/lipoplexes,characterisation was performed to determine their size, sizedistribution and surface charge. Polyplexes/lipoplexes made usingthe cationic polymers SuperFect�, PEI, and Lipofectin� wereinitially screened using gel electrophoresis. Successful polyplex/lipoplex formation at the ratios used was shown by a gel retarda-tion assay (SI Fig. 1). The size and surface charge (zeta potential)were then analysed using a Zetasizer. The dynamic light scattering(DLS) results showed that polyplexes formed using SuperFect� andLipofectin� are of similar sizes (w140 nm average diameters e

Fig. 1). However, polyplexes formed with PEI were on averagesmaller than 100 nm in diameter, but with a larger distribution ofsizes by signal intensity (SI Fig. 2). These polyplexes were alsoshown to have the highest surface charge (average¼ 42mV), whichcombined with the small size indicates a high charge density.Analysis by transmission electron microscopy showed that thepolyplexes had a similar diameter to the results obtained by DLSwhen hydrated (Fig. 1).

3.2. CHS loading and characterisation

Polyplexes formed with PEI were loaded in the CHS by simpleaddition as described earlier. By analysing the remaining super-natant post loading and washing, it was seen that a loading capa-bility of w75% of the original DNA content can be achieved(SI Fig. 3), a value typically obtained, as demonstrated previously[32,33]. The size of these loaded spheres was in the region ofw2700 nm asmeasured by DLS, and had a charge of eightmVwhenloaded, thereby reducing the positive surface charge observed viathe polyplexes alone (Fig. 1).

Fig. 1. Polyplex characterisation by surface charge (a), diameter (b) with representative transmission electron microscopy images of the polyplexes formed with SuperFect� (e), PEI (f), and Lipofectin� (g) (25k times magnification, scalebars represent 500 nm). Loaded collagen hollow sphere (CHS) characterisation by surface charge (in comparison to the high dose of unloaded polyplexes (c)), and size (d and h) analysis showing the predominantly loaded nature of theCHS. Representative 40k times magnified TEM images are shown (i) of PEI polyplex loaded CHS (scale bars represent 500 nm).

B.New

landet

al./Biom

aterials34

(2013)2130

e2141

2133

B. Newland et al. / Biomaterials 34 (2013) 2130e21412134

3.3. Toxicity of gene vectors

These studies assessed the effect of the gene vectors on the hosttissue by threemethods. The first was tomeasure the volume of thegross tissue loss after immunohistochemical tissue processing.Fig. 2 shows that SuperFect� and AAV group cause a tissue lossvolume of 0.166 mm3 (�0.066) and 0.133 mm3 (�0.006) respec-tively which are statistically significantly different from the injec-tion of PBS or naked plasmid DNA. When expressed as a volumefraction of the striatum, w0.34% is a relatively low percentage oftissue loss. Indeed, considering the large size of the striatum andthat only a single injection is made, this effect is indeed pronounced(see representative images shown in Fig. 2). SuperFect� and theAAV treated brains had tissue loss with a maximum area of0.42mm2 (mean of 3 animals�0.11mm2) and 0.25mm2 (mean of 2animals�0.13mm2) respectively, which is 4.6% and 2.7% of the areafraction of the striatum respectively.

In addition to the measurement of the gross tissue loss causedby the gene vectors, the host response was measured in terms ofboth the reactivity of astrocytes and microglial recruitment adja-cent to the site of injection using GFAP and OX-42 immunohisto-chemical analysis, respectively. A greater density of GFAP staining,a hallmark of the activation of astrocytes [39], was determined atthe injection site along with optical density measurements of themicroglia marker OX-42. Where a tissue loss was seen the tissue inthe immediate vicinity was analysed. Fig. 2 shows that a greaterdensity of both astrocytes and microglia was present for all genevectors compared to PBS and naked pDNA, with a 2.1 fold increasebeing shown for the lipoplexes (P < 0.05). As the representativeimages show, there is a reaction to the needle stick injury withsterile PBS; however this is augmented by the addition of a genevector, indicating a specific response to the materials themselves.

3.4. Collagen hollow sphere (CHS) delivery system for reducedtoxicity

It was hypothesised that by using a system that implementedthe sustained release of polyplexes, a reduction in the gross tissueloss can be achieved by reducing initial concentrations of poly-plexes delivered directly to the local tissue. Previous studies by ourgroup have shown that hollow spheres composed of type I collagen,crosslinked with four arm star PEG can be loaded with polyplexesand effectively mediate sustained release and a reduction of poly-plex cytotoxicity in vitro [32,33]. This study used CHS that havea loading efficiency of 75% (SI Fig. 3). 4 mg of DNA per injectionsample was used and a total of 3 mg was administered to each site(spheres were rinsed and re-collected so the 25% of unbound pol-yplexes were not administered). Animals were sacrificed sevendays post injection and gross tissue damage was analysed alongwith astrocyte reactivity and microglia recruitment (Fig. 3). CHSmediated delivery did not reduce the density of GFAP or OX-42staining at the site of injection in comparison to that observedusing the polyplexes alone, however; the volume of tissue loss wassignificantly reduced e by 0.05 mm3 (almost no tissue loss).

3.5. In vitro analysis of gene vector toxicity

In order to study the mechanism of vector toxicity, in vitrostudies analysing the cytotoxicity profiles of the different genevectors or polyplex loaded CHS using primary astrocytes, extractedfrom the striatum of three day old rat pups were carried out as seenin Figs. 4 and 5. The cytotoxicity profile of all three gene vectors wassimilar. For instance, when the reduction of the resazurin compo-nent of alamarBlue� solution was normalised to 100% for controlcells, a reduction in cell viability tow50% was observed for all three

polyplexes (Fig. 4a). All the polyplex groups caused a reduction inintracellular ATP activity, and had a change in cell membraneintegrity (w13% change in extracellular LDH detection), (Fig. 4band c). The polyplexes formed with SuperFect� showed the highestnecrotic effects (70.8%) and PEI polyplexes treated group resulted inthe highest proportion of apoptotic cells (48.3%) (Fig. 4d,e and f).

Whilst the in vivo data for all the polymeric vectors indicatedthat either an increased gross tissue loss or host response occurredabove that of a PBS injection, the effect of the polyplex inducedtoxicity was much more pronounced in vitro. However this wassignificantly ablated by the use of CHS (Fig. 5). It must be noted thatfor these studies, the polyplex alone and polyplex loaded CHSsamples contained 3 mg of DNA. This represents the amount of DNAloaded to the spheres (75% of 4 mg added during the loading phase),hence the difference in viability values for PEI with 1 mg of DNA(48%e Fig. 4a) compared to 3 mg (34%e Fig. 5a). At these high dosesof polyplexes, cell viability was ameliorated to 81% by using thespheres loaded as opposed to the PEI polyplexes alone. Whilst thesphere mediated delivery of polyplexes caused a slight increase inmembrane disruption (Fig. 5c), the effect on intracellular ATPactivity was pronounced and the percentage of necrotic andapoptotic cells decreased by 47% and 24% respectively.

To analyse the effect that these groups had on astrocyte viabilityin real time, the xCELLigence� system was then used to takereadings every 15 min during the early stages (Fig. 6b and d), andover a period of two days (Fig. 6a and c). This system effectivelyuses cellular adhesion as a marker of viability, which is calculatedby taking surface impedance measurements. This method of anal-ysis was particularly desired to interpret the gross tissue loss dataobtained in in vivo studies. Cells previously untreated and attachedto the well bottom, once killed, will often become detached;a change that can be detected by the system. All data obtained fromthis system uses the cell index value, which can be subsequentlyplotted and analysed. Fig. 6a shows the cell index values obtainedduring the first 48 h following the addition of the samples. PBS andnaked pDNA have similar cell index profiles, as do the polyplexesformed with SuperFect� and PEI. However the cellular response tolipoplexes differs from the polyplexes by a much quicker reductionin the cell index rise, but equally a slower subsequent decline. Thiscan be easily observed in comparison to the other polymers inFig. 6b, where the 2 h “slope” of cell index is measured (simply asDy/Dx) and plotted as a function of time. For both the PEI andSuperFect� polyplexes, the cell proliferation appears to continueunaltered for the first 2 h, followed by a steady decrease in slope.The reduction in slope occurs quicker but does not turn toa decrease in cell index over these first 24 h analysed. This differ-ential in cell index profile, coupled with the matching cell viabilityplot, sheds light on why Lipofectin� caused the largest hostresponse but the lowest gross tissue loss.

The CHS xCELLigence� data showed the same trend, that theCHS reduced the toxic effects of the polyplexes, which was signif-icantly different after 24 h Fig. 6d shows that the initial decrease incell index slope is almost halved by the use of this sustained releasedelivery system. Although the toxic effects are still observed, itmust be noted that these doses are very high at 3 mg of DNA per well(6 mg of PEI per well). However, the combination of these in vitrodata suggests that sustained delivery systems may be able to servea dual purpose of sustained delivery and a reduction of vectortoxicity, both of significance for many CNS applications.

4. Discussion

This aim of the study was to analyse the toxicity and hostresponse to gene vectors and evaluate whether loading polyplexesto a sustained delivery system would reduce their associated

Fig. 2. Quantification of (a) the hole volume left by gross tissue loss, (b) the optical density of DAB staining for astrocyte reactivity (GFAP) and (c) microglia (OX-42) adjacent to the injection site, with representative images shown inpanel (d), 7 days post injection. n ¼ 3, error bars indicate � standard error of the mean, asterisks mark statistically significant differences from control groups by Dunn’s Multiple Comparison Test, P < 0.05, scale bars ¼ 1 mm.

B.New

landet

al./Biom

aterials34

(2013)2130

e2141

2135

Fig. 3. Analysis of the effect of using collagen hollow spheres to deliver polyplexes formed with PEI in comparison to data from injections of PBS or those polyplexes alone.Quantification of (a) the hole volume left by gross tissue loss, (b) the optical density of DAB staining for astrocyte reactivity (GFAP) and (c) microglia (OX-42) adjacent to the injectionsite, with representative images shown in panel (d), 7 days post injection. n ¼ 3, error bars indicate � standard error of the mean, asterisk marks statistically significant differencesbetween polyplex alone and polyplex þ spheres group (Student T-test P < 0.05 scale bar ¼ 1 mm).

B. Newland et al. / Biomaterials 34 (2013) 2130e21412136

toxicity. Three commercial non-viral vectors were used, of whichPEI was the only one where the chemical structure is known (theexact structure of SuperFect� and Lipofectin� is not divulged by themanufacturers). PEI has a high charge density arising through everythird atom being a protonisable nitrogen atom (primary, secondaryand also tertiary in the branched PEI used in these studies). The 2:1w/w ratio used for these studies correlates to approximately a 15:1nitrogen to phosphate (N:P) ratio. As the structure of SuperFect� isunknown, a conversion of the 8:1 w/w ratio to an N:P ratio cannotbe performed. What is known is that the structure of SuperFect� isbased on a partially degraded polyamidoamine, therefore the lowersurface charge of these polyplexes can potentially be extrapolatedto the lower charge density on the amidoamine monomer.Although the polyplexes formed with the three different vectorsshowed different physical properties, these properties did notresult in an overall attributable trend seen in their toxicity profilein vivo or in vitro. There was, for instance, no clear link betweenpolyplex size and toxicity.

The striatum was used to assess the in vivo toxic effects of thethree polymeric gene vectors in comparison to an adeno-associatedvirus with PBS and naked pDNA as controls. The striatum providesa target with clinical relevance for gene therapy applications inneurological disorders such as Parkinson’s disease. Dopaminergic

neurons that terminate in the caudate and putamen are an idealtarget for growth factor mediated protection via gene delivery tothe striatum. In addition, the presence of astrocytes and thepermissiveness for microglia cell recruitment allows the identifi-cation of various host responses to different gene vectors.

Although rarely described, the tissue loss at the site of injectionappears to be an adverse effect that is not limited to gene vectorsalone. Indeed, of the numerous convection enhanced deliverystudies delivering recombinant proteins to the brains of rodents,non-human primates and humans, one study identifies tissue lossin the parkinsonian monkey brain post immunohistochemicalprocessing [40]. The authors report that the high flow rate requiredfor convection enhanced delivery from the single port catheter tipmay be the cause of this tissue loss. The flow rate of 1 mg/min usedin the present study resulted in no gross tissue loss for PBS or nakedDNA injections, but isw22 times slower than that of the maximumrate used for the convection enhanced delivery. It is thereforereasonable to assume that in this study the gross tissue loss is dueto the toxicity of the vector causing local cell death and loss ofstriatal parenchyma. Although not measured directly, large areasof tissue destruction were also observed when biodegradablepoly(methylidene malonate 2.1.2)-based microspheres wereadministered to the rat brain [27]. Images taken of the brain during

Fig. 4. In vitro comparison of the adverse effects of the different polymeric vectors on primary astrocytes. The alamarBlue� reduction analysis method was used to determine overall viability by normalising the cell metabolic activity tothat of the PBS group (a). The effects on intracellular ATP activity (b) and membrane integrity (c) were analysed using the CellTitre-Glo� and CytoTox 96� cytotoxicity assays respectively. Necrotic (d) and apoptotic cells (e) weredetermined by staining using Ethidium homodimer III and Annexin-V respectively, followed by fluorescence imaging and quantification. 20� representative images of the groups are shown (f) with necrotic cells (red), apoptotic cells(green) and cell nuclei counterstained with DAPI (blue). n ¼ 4 or higher for all studies, error bars indicate � standard error of the mean, asterisks mark statistically significant differences from control groups (one way ANOVA, P < 0.05scale bars ¼ 50 mm). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

B.New

landet

al./Biom

aterials34

(2013)2130

e2141

2137

Fig. 5. The use of collagen hollow spheres as a means of sustained polyplex delivery reduced toxicity observed with in vitro cultures of rat astrocytes, as observed using thealamarBlue� reduction assay (a) and CellTitre-Glo� assay (b). Membrane integrity (CytoTox 96�) (c) was slightly more compromised, but overall evaluation of necrotic cells(d) (Ethidium homodimer III) and apoptotic cells (e) (Annexin-V) showed CHS to reduce the toxicity of the polyplexes formed with PEI. Part (f) shows 20� representative images ofPEI polyplexes alone or polyplex loaded CHS showing necrotic cells (red) and apoptotic cells (green) both of which are counterstained to visualise the nuclei with DAPI (blue). n ¼ 4or higher for all studies, error bars indicate � standard error of the mean, asterisks mark statistically significant differences from control groups (one way ANOVA, P < 0.05 scalebars ¼ 50 mm). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

B. Newland et al. / Biomaterials 34 (2013) 2130e21412138

the sectioning process showed large destruction of the striatum,along with an overall gross morphology change due to tissuedestruction. Although not quantified, this destruction was deemedirreversible as it was present up to 18 months post administration.

The gene vectors used for these studies have been extensivelystudied in vitro and for a variety of applications; however to the bestof the authors’ knowledge, a specific study into the toxicity asso-ciated with these vectors for brain/central nervous system appli-cations has not been undertaken. The purpose of this study was toestablish not only the differences in toxicity and tissue responsecaused by these gene vectors, but whether the sustained deliverysystem based on collagen hollow spheres can potentially abate thistoxicity in astrocytes, a predominant cell type of the striatum. Asmany neurological disorders of the brain such as Alzheimersdisease, Parkinson’s disease, Amyotrophic lateral sclerosis, Hun-tington’s disease and multiple sclerosis are progressive in natureand occur over many years, a system of sustained gene delivery isdesirable. However, a delivery system that reduced the overall highsurface charge of the polyplexes can potentially aid in reducing theassociated toxicity. Collagen hollow spheres, previously syn-thesised by our group, offer an attractive means of sustained pol-yplex delivery, by loading several polyplexes to individual collagenspheres [32,33]. However this study investigated whether the

packaging of polyplexes by this extracellular matrix protein depotsystem could reduce the acute toxicity of polyplexes delivered tothe brain.

Whilst some of the polyplexes/lipoplexes delivered to the brainwithout a sustained delivery vector resulted in acute toxicity or anelevated host response, it must be noted that this response wassimilar to as seen in the adeno-associated virus treated group. Infact the low dose (1 mg of DNA) of PEI polyplexes produced a medialhost response in terms of reactivity of astrocytes/microglia and holeformation. For this reason, coupled with its history of use, PEI wasa good candidate to study with the CHS. However, with the use ofthe CHS delivery system, this small amount of tissue loss can beavoided. CHS were also loaded with 4 mg of complexed DNA,relating to a final amount of 3 mg (75% loading efficiency), soa higher dose was administered than the polyplex alone groups(2 mg), but this treatment group ameliorated the negative effects.

To understandwhether the tissue lost in the brainwas a result ofdirect toxicity induced by the vectors, or caused by subsequentphagocytosis/host response, an in vitro study was conducted withisolated primary astrocytes. This study was conducted so as toindicate possible mechanisms of the toxicity via measurement ofthe intracellular ATP activity, membrane integrity and resultingcaspase activation. Also, to further ascertain whether the CHS

Fig. 6. Real-time analysis of area of cell attachment via the xCELLigence system, showing the comparison of polymer vectors over the first 2 days following treatment of polymeric vectors (a), or vectors alone vs vectors loaded to CHS(c). Panels (b) and (d) show a breakdown of the cell index gradient by 2 h intervals for the first 24 h following addition of the vectors (b) or comparison with the use of CHS (d). For all groups n ¼ 3, average of each plotted for each timepoint.

B.New

landet

al./Biom

aterials34

(2013)2130

e2141

2139

B. Newland et al. / Biomaterials 34 (2013) 2130e21412140

reduced the toxic effects of the polyplexes, the in vitro studies weredesigned so that the quantity of administered complexed DNA wasequal for both groups. This resulted in the two separate sets ofstudies having different doses. The polyplex alone studies con-tained the “low dose” of 1 mg per experimental well, which wassufficient to see the toxic effects of the different polyplexes asoutlined by different methods in Fig. 4. The second set of studies,shown in Fig. 5, contained the “high dose” of 3 mg of DNA per well.These studies illustrate the protective effects of the CHS in terms ofcellular metabolic activity and intracellular ATP content and thepercentage of necrotic and apoptotic cells. The effect on cellmembrane integrity was slightly worsened by the use of CHS. Asseen in previous studies with macrophage cells, spheres over1000 nm showed minimal cellular uptake so as to act as an inter-stitial delivery system to amultitude of cells [33]. In light of this, theslight elevation in extracellular LDH may be due to the prolongedlocalisation of the loaded CHS in the vicinity of the cell membrane(Fig. 5f (ii)).

It was also desired to see if a correlation was observed betweenthe results obtained in vivo and those in two-dimensional in vitro cellculture. Whilst a seven day time point was chosen for the in vivostudies to take into account the prolonged polyplex delivery via CHS,a 24 h period of incubation was chosen for in vitro study to give aninsight into different mechanisms or causes of cell death. Compari-sons between in vivo and in vitro data showed a similar pattern ofresults. For instance, all the polymeric vectors showed increasedtoxicity over PBS and naked pDNA groups, and the use of CHSresulted in a reduction in polyplex mediated toxicity. SuperFect�

caused thegreatest tissue loss volumebutwasnot significantlymoretoxic as analysed in vitroby the alamarBlue� reduction.However, theproportion of necrotic cellswas higher for the SuperFect� group thatthe other vectors. Also the Lipofectin� vector, that caused the mostgross tissue loss in the striatum, caused the lowest proportion ofnecrotic and apoptotic cells, thus indicating that observations madein vitro may potentially reflect prior in vivo data.

The use of real time analysis of cell viability has been usedpreviously for neuronal cell cultures and was used here to explorethe onset time of vector induced cell death [41]. The results servedto highlight the onset of damaging effects that high doses of genevectors can have on cells in culture. The SuperFect� and PEI poly-plexes at a dose of 1 ml DNA per well showed a slow change in cellindex values that decreases over 14 h (indicating cell death anddetachment). A 3 ml DNA dose per well of PEI polyplexes causeda reduction in cell index values within the first 2 h of incubation.The use of CHS reduced this immediate cell death rate. The sus-tained delivery of the polyplexes (lower initial amount delivered)together with the reduced surface charge of the polyplex loadedCHS at 8 mV, as opposed to 42 mV of polyplexes alone, is a likelyreason for this reduced cell death. The combination of in vivo andin vitro data indicates that delivery of polyplexes througha biomaterial delivery system can reduce the initial/direct cytotoxiceffects of PEI polyplexes.

5. Conclusions

Genetic manipulation of the CNS offers an attractive therapeuticavenue for the treatment of several neurodegenerative diseases.However, as with all therapeutic strategies, the benefits of inter-vention must out-weigh the negative effects of the therapy process.In this study, the analysis of the toxicity of a variety of viral and non-viral gene delivery vector systems in the rat brain was conducted.By performing a direct comparison, it was observed that the singleadministration of polymeric or liposomal vectors caused no moreacute toxicity than that observed with a viral vector. Furthermoreby loading polyplexes in a micron sized type I collagen hollow

spheres, the toxic effects of the polyplexes can be reduced. In vitroassays highlighted that this reductionwas due to a decrease in boththe necrotic and apoptotic proportion of cells, and reduces celldeath immediately post administration. A biomaterials approach togene delivery may provide a more suitable delivery platform thatenables polyplex release and also reduces acute toxicity.

Acknowledgements

The authors would like to thank Science Foundation Ireland,Strategic Research Cluster (SRC) (grant no. 07/SRC/B1163) for thefinancial support of this research.

Appendix A. Supplementary data

Supplementary data related to this article can be found at http://dx.doi.org/10.1016/j.biomaterials.2012.11.049.

References

[1] Kearns CM, Cass WA, Smoot K, Kryscio R, Gash DM. GDNF protection against6-OHDA: time dependence and requirement for protein synthesis. J Neurosci1997;17:7111e8.

[2] Sauer H, Rosenblad C, Björklund A. Glial cell line-derived neurotrophic factorbut not transforming growth factor beta 3 prevents delayed degeneration ofnigral dopaminergic neurons following striatal 6-hydroxydopamine lesion.Proc Natl Acad Sci U S A 1995;92:8935e9.

[3] Sullivan AM, Opacka-Juffry J, Blunt SB. Long-term protection of the ratnigrostriatal dopaminergic system by glial cell line-derived neurotrophicfactor against 6-hydroxydopamine in vivo. Eur J Neurosci 1998;10(1):57e63.

[4] Kirik D, Rosenblad C, Björklund A. Preservation of a functional nigrostriataldopamine pathway by GDNF in the intrastriatal 6-OHDA lesion modeldepends on the site of administration of the trophic factor. Eur J Neurosci2000;12:3871e82.

[5] Gill SS, Patel NK, Hotton GR, O’Sullivan K, McCarter R, Bunnage M, et al. Directbrain infusion of glial cell line-derived neurotrophic factor in Parkinsondisease. Nat Med 2003;9:589e95.

[6] Nutt JG, Burchiel KJ, Comella CL, Jankovic J, Lang AE, Laws ER, et al.Randomized, double-blind trial of glial cell line-derived neurotrophic factor(GDNF) in PD. Neurology 2003;60:69e73.

[7] Lang AE, Gill S, Patel NK, Lozano A, Nutt JG, Penn R, et al. Randomizedcontrolled trial of intraputamenal glial cell lineederived neurotrophic factorinfusion in Parkinson disease. Ann Neurol 2006;59:459e66.

[8] Kearns CM, Gash DM. GDNF protects nigral dopamine neurons against6-hydroxydopamine in vivo. Brain Res 1995;672:104e11.

[9] Gardlik R, Palffy R, Hodosy J, Lukacs J, Turna J, Celec P. Vectors and deliverysystems in gene therapy. Med Sci Monit 2005;11:110e21.

[10] Ono T, Fujino Y, Tsuchiya T, Tsuda M. Plasmid DNAs directly injected intomouse brain with lipofectin can be incorporated and expressed by brain cells.Neurosci Lett 1990;117:259e63.

[11] Li-Xin Z, Min W, Ji-Sheng H. Suppression of audiogenic epileptic seizures byintracerebral injection of a CCK gene vector. NeuroRep 1992;3:700e2.

[12] Schwartz B, Benoist C, Abdallah B, Scherman D, Behr JP, Demeneix BA. Lip-ospermine-based gene transfer into the newborn mouse brain is optimized bya low lipospermine/DNA charge ratio. Hum Gene Ther 1995;6:1515e24.

[13] Roessler BJ, Davidson BL. Direct plasmid mediated transfection of adult murinebrain cells in vivo using cationic liposomes. Neurosci Lett 1994;167:5e10.

[14] Kofler P, Wiesenhofer B, Rehrl C, Baier G, Stockhammer G, Humpel C.Liposome-mediated gene transfer into established CNS cell lines, primary glialcells, and in vivo. Cell Transplant 1998;7:175e85.

[15] Abdallah B, Hassan A, Benoist C, Goula D, Behr JP, Demeneix BA. A powerfulnonviral vector for in vivo gene transfer into the adult mammalian brain:polyethylenimine. Hum Gene Ther 1996;7:1947e54.

[16] Martres M-P, Demeneix B, Hanoun N, Hamon M, Giros B. Up- and down-expression of the dopamine transporter by plasmid DNA transfer in the ratbrain. Eur J Neurosci 1998;10:3607e16.

[17] Goula D, Remy JS, Erbacher P, Wasowicz M, Levi G, Abdallah B, et al. Size,diffusibility and transfection performance of linear PEI/DNA complexes in themouse central nervous system. Gene Ther 1998;5:712e7.

[18] Lemkine GF, Mantero S, Migné C, Raji A, Goula D, Normandie P, et al. Pref-erential transfection of adult mouse neural stem cells and their immediateprogeny in vivo with polyethylenimine. Mol Cell Neurosci 2002;19:165e74.

[19] Boussif O, Lezoualc’h F, Zanta MA, Mergny MD, Scherman D, Demeneix B, et al.A versatile vector for gene and oligonucleotide transfer into cells in cultureand in vivo: polyethylenimine. Proc Natl Acad Sci U S A 1995;92:7297e301.

[20] Kwon EJ, Lasiene J, Jacobson BE, Park I-K, Horner PJ, Pun SH. Targeted nonviraldelivery vehicles to neural progenitor cells in the mouse subventricular zone.Biomaterials 2010;31:2417e24.

B. Newland et al. / Biomaterials 34 (2013) 2130e2141 2141

[21] Tang GP, Zeng JM, Gao SJ, Ma YX, Shi L, Li Y, et al. Polyethylene glycol modifiedpolyethylenimine for improved CNS gene transfer: effects of PEGylationextent. Biomaterials 2003;24:2351e62.

[22] Fournier E, Passirani C, Montero-Menei CN, Benoit JP. Biocompatibility ofimplantable synthetic polymeric drug carriers: focus on brain biocompati-bility. Biomaterials 2003;24:3311e31.

[23] Aloisi F, Ria F, Adorini L. Regulation of T-cell responses by CNS antigen-presenting cells: different roles for microglia and astrocytes. ImmunolToday 2000;21:141e7.

[24] Bjugstad KB, Lampe K, Kern DS, Mahoney M. Biocompatibility of poly(ethyleneglycol)-based hydrogels in the brain: an analysis of the glial response acrossspace and time. J Biomed Mater Res Part A 2010;95:79e91.

[25] Menei P, Daniel V, Montero-Menei C, Brouillard M, Pouplard-Barthelaix A,Benoit JP. Biodegradation and brain tissue reaction to poly(D, L-lactide-co-glycolide) microspheres. Biomaterials 1993;14:470e8.

[26] Hatten ME, Liem RKH, Shelanski ML, Mason CA. Astroglia in CNS injury. Glia1991;4:233e43.

[27] Fournier E, Passirani C, Colin N, Sagodira S, Menei P, Benoit J-P, et al. The braintissue response to biodegradable poly(methylidene malonate 2.1.2)-basedmicrospheres in the rat. Biomaterials 2006;27(28):4963e74.

[28] Mulcahy P, O’Doherty A, Paucard A, O’Brien T, Kirik D, Dowd E. Developmentand characterisation of a novel rat model of Parkinson’s disease induced bysequential intranigral administration of AAV-a-synuclein and the pesticide,rotenone. Neuroscientist 2012;203:170e9.

[29] Newland B, Zheng Y, Jin Y, Abu-Rub M, Cao H, Wang W, et al. Single cyclizedmolecule versus single branched molecule: a simple and efficient 3D “Knot”polymer structure for nonviral gene delivery. J Am Chem Soc 2012;134(10):4782e9.

[30] Newland B, Tai H, Zheng Y, Velasco D, Di Luca A, Howdle SM, et al. A highlyeffective gene delivery vectorehyperbranched poly(2-(dimethylamino)ethylmethacrylate) from in situ deactivation enhanced ATRP. Chem Commun 2010;46:4698e700.

[31] Dash BC, Réthoré G, Monaghan M, Fitzgerald K, Gallagher W, Pandit A. Theinfluence of size and charge of chitosan/polyglutamic acid hollow spheres oncellular internalization, viability and blood compatibility. Biomaterials 2010;31:8188e97.

[32] Helary C, Browne S, Mathew A, Wang W, Pandit A. Transfection of macro-phages by collagen hollow spheres loaded with polyplexes: a step towardsmodulating inflammation. Acta Biomater 2012;8:4208e14.

[33] Browne S, Fontana G, Rodriguez B, Pandit A. A protective extracellular matrix-based gene delivery reservoir fabricated by electrostatic charge. Mol Pharmcol2012;9:3099e106.

[34] Holladay C, Keeney M, Newland B, Mathew A, Wang W, Pandit A. A reliablemethod for detecting complexed DNA in vitro. Nanoscale 2010;2:2718e23.

[35] Moloney TC, Dockery P, Windebank AJ, Barry FP, Howard L, Dowd E. Survivaland immunogenicity of mesenchymal stem cells from the green fluorescentprotein transgenic rat in the adult rat brain. Neurorehabil Neural Repair 2010;24:645e56.

[36] Andersson C, Hamer RM, Lawler CP, Mailman RB, Lieberman JA. Striatalvolume changes in the rat following long-term administration of typical andatypical antipsychotic Drugs. Neuropsychopharmacology 2002;27:143e51.

[37] Allen JW,Mutkus LA, AschnerM. Isolation of neonatal rat cortical astrocytes forprimary cultures. Current protocols in toxicology. JohnWiley & Sons, Inc; 2001.

[38] Albuquerque C, Joseph DJ, Choudhury P, MacDermott AB. Dissection, plating,and maintenance of cortical astrocyte cultures. Cold Spring Harb ProtocAugust 2009;(8) [pdb.prot5273].

[39] Ridet JL, Privat A, Malhotra SK, Gage FH. Reactive astrocytes: cellular andmolecular cues to biological function. Trends Neurosci 1997;20:570e7.

[40] Gash DM, Zhang Z, Ai Y, Grondin R, Coffey R, Gerhardt GA. Trophic factordistribution predicts functional recovery in parkinsonian monkeys. AnnNeurol 2005;58:224e33.

[41] Diemert S, Dolga AM, Tobaben S, Grohm J, Pfeifer S, Oexler E, et al. Impedancemeasurement for real time detection of neuronal cell death. J NeurosciMethods 2012;203:69e77.