in vitro evaluation of a ‘stealth’ adenoviral vector for targeted gene delivery to adult...

THE JOURNAL OF GENE MEDICINE R E S E A R C H A R T I C L EJ Gene Med 2009; 11: 335–344.Published online 26 February 2009 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/jgm.1306

In vitro evaluation of a ‘stealth’ adenoviral vectorfor targeted gene delivery to adult mammalianneurones

Karen Sims1†

Zubair Ahmed1†*Martin L. Read1

Lisa Cooper-Charles1

Ana Maria Gonzalez1

Kerry D. Fisher2

Martin Berry1

Leonard W. Seymour2

Ann Logan1

1Molecular Neuroscience Group,School of Clinical and ExperimentalMedicine, University of Birmingham,Birmingham, UK2Department of ClinicalPharmacology, University of Oxford,Oxford, UK

*Correspondence to: Zubair Ahmed,Molecular Neuroscience Group,School of Clinical and ExperimentalMedicine, Room WX2.17 Institute ofBiomedical Research (West),University of Birmingham,Edgbaston, Birmingham B15 2TT,UK. E-mail: [email protected]

†Both investigators contributedequally and should be considered assenior authors.

Received: 7 November 2008Accepted: 8 January 2009

Abstract

Background Polymer coating of adenovirus type 5 (Ad5) particles producesa ‘stealth’ Ad5 (sAd5) that confers protection from immune recognition,blocks receptor-mediated uptake, and favours uptake into pinocytic cells.

Methods In mixed cultures of primary adult rat dorsal root ganglionneurones (DRGN), rat C6 glioma cells, A9 non-Coxsackie and Ad Receptor(CAR)- and CAR-expressing fibroblasts, reporter gene expression after sAd5pinocytotic uptake was monitored using the green fluorescent protein (gfp)gene, and viral particle trafficking and polymer coat dismantling was followedusing Yoyo-1 tagged Ad5 DNA and Texas Red (TR) to label the coat.

Results sAd5.gfp was pinocytosed by significantly higher proportions ofneurones, than other cells, but GFP was not expressed. The TR-labelledcoat remained co-localised with tagged viral DNA within transfected DRGN,showing that sAd5 did not uncoat and viral DNA did not traffic to thenucleus. Noncoated Ad5 transduced non-neuronal DRG cells more efficientlythan DRGN, whereas A9CAR cells were more significantly transduced thanany other cell type. Retargeting of the sAd5.gfp with either fibroblast growthfactor-2 or nerve growth factor (NGF) enhanced internalisation by DRGN intoendocytic vesicles allowing uncoating and thus GFP expression. Retargetingwith NGF resulted in significantly higher numbers of DRGN expressing GFPthan non-neuronal DRG cells.

Conclusions These findings indicate that DRGN pinocytose atropic geneticparticles at higher levels than non-neuronal DRG cells and the environment ofpinocytic vesicles is not conducive to sAd5 uncoating and capsid dismantling,requiring reformulation of sAd5 with either a neurone specific ligand or aself-dismantling coat to target sAd5 transgene expression to neurones.Copyright 2009 John Wiley & Sons, Ltd.

Keywords adenovirus 5; neurones; pinocytosis; retargeting;‘stealth’ modification

Introduction

Recombinant type 5 adenovirus (Ad5) is internalised into target cellsafter binding to cell surface receptors that include Coxsackie andAd Receptor (CAR), αv-integrins and heparin sulphate glycosamino-glycans [1]. Because these receptors are expressed by cells ofmany tissues, infection and transduction by Ad5 is promiscuous and,when coupled with the acquired and innate Ad5 immunogenicity ofthe host, the potential of Ad vectors for clinical use is compro-mised. This has led numerous investigators to evaluate strategies toincrease infection specificity and to decrease immunogenicity of Ad5.

Copyright 2009 John Wiley & Sons, Ltd.

336 K. Sims et al.

Historically, many different methods have been usedto mask Ad5 immunogenicity. For example, the cationiclipid GL-67 used with polyethylene glycol (PEG) as a‘stealth’ (s) coat successfully blocked antibody bindingin vitro [2]. This coated sAd complex, however, providedno appreciable protection against antibodies whendelivered to animals previously immunised againstAd5. PEG-coated Ad5 ablated in vitro but not in vivogene delivery while enhancing vector persistence andreducing innate immune responses [3–7]. Similarlyreduced immunogenic responses and protein interactionsin serum were blocked when viral particles were coatedwith a multivalent hydrophillic copolymer poly [N-(2-hydroxypropyl)methacrylamide] (pHPMA-ONp) [8,9].For example, pHPMA-ONp interacts with amino groupson the Ad5 surface to produce a retargetable sAd5 vectorthat escapes natural immune surveillance in mice [9,10].Coating Ad5 particles with this multivalent reactive formof pHPMA-ONp reduced titres of circulating antibodiesto viral proteins and abolished Ad5-dependant transgeneexpression by eliminating binding of native Ad5 ligands(e.g. the knob domain) to CAR on target A549 cells [9].

Retargeting sAd5 with ligands attached to the pHPMA-ONp coat resulted in gene expression in A549 cells,increased cell specificity and transgene expression, andimproved immunogenic profiles when the targeting ligandhad high specificity for the target cellular phenotype [9].Retargeting of Ad5 with fibroblast growth factor (FGF)-2enhanced the nuclear delivery of transgenes to A549 cells[11], efficiently transduced FGF2 receptor-positive cellsin a model of intraperitoneal cancer both in vitro andin vivo, and transduced primary glioblastoma multiformeendothelial cells that lack the CAR receptor but havea high expression profile of FGF2 receptor [12]. Otherligands used for Ad5 retargeting include epidermal growthfactor (EGF) to selectively transduce EGF receptor positivecells in a xenograft model of human ovarian cancer [13]and chicken hepatic lectin to target gene transfer to theavian liver [14]. However, most high affinity receptorsare promiscuous and neurone-specific ligands/receptorsare rare, making exclusive ligand-mediated targeting ofAd5 to neurones difficult. The co-expression of receptorsby both neurones and glia limits the options for ligand-mediated vector targeting to neurones.

Neurones, however, are actively pinocytotic in vitro andin vivo, a property that can be exploited for targetingmolecules/particle uptake. Small particles (diameter:100–200 nm) are fluid-phase pinocytosed into vesiclesthat bud internally from the plasmalemma [15]. Wewondered whether neurones would preferentially uptakesAd5 by pinocytosis over non-neuronal cells, to provide atargeted delivery mechanism for therapeutic transgenes.In the present study, we evaluated in vitro the potentialfor exploiting the neuronal membrane for selectivepinocytotic uptake and green fluorescent protein (GFP)expression of atropic sAd5 and compared this withFGF-2 and nerve growth factor (NGF)-retargeted sAd5(sAd5FGF.gfp and sAd5NGF.gfp).

Materials and methods

Dorsal root ganglia neurone (DRGN)cultures

DRGN were enriched from L3-L6 DRG of adult maleSprague-Dawley rats (weighing 200–250 g) by dissocia-tion in Neurobasal-A (Invitrogen, Paisly, UK) containing0.1% collagenase (Sigma-Aldrich, Poole, UK) for 2 hat 37 ◦C in a humidified atmosphere containing 5%CO2. Cells were transferred to Neurobasal-A supple-mented with 0.02% B27 (Invitrogen), 0.0025% 200 mM

L-glutamine (Invitrogen) and 0.005% Gentimicin (Invit-rogen) and triturated several times. After centrifugationat 120 g for 8 min through a 15% bovine serum albumin(BSA) gradient, pellets were resuspended in 100 µl of sup-plemented Neurobasal-A and DRGN plated at a density of1500 per well on glass coverslips coated with 100 µg/ml ofpoly D-lysine (Sigma-Aldrich) and 100 µg/ml of laminin-1 (Sigma-Aldrich) in Neurobasal-A. DRG mixed culturescontain DRGN, together with satellite glia, Schwann cells,fibroblasts and endothelial cells, which are collectivelyreferred to as non-neuronal DRG cells (NNDRG) through-out the present study.

Cell lines

The following cell lines were used: (i) C6 glioma cells(ATCC, Middlesex, UK), which are glial cells originatingfrom rat glioma that have an adherent fibroblastoidmorphology, were maintained in F-12K nutrient medium(Invitrogen) supplemented with 15% normal horse serum(Invitrogen), 2.5% foetal calf serum (FCS) (Invitrogen)and 2.5% gentomycin (Invitrogen), and plated at 30 000cells/well. (ii) A9non−CAR (non-CAR expressing) andA9CAR (CAR expressing) cells (all from ATCC), which arefibroblasts originated from mouse connective tissue thatexhibit a fibroblast-like morphology, were maintainedin Dulbecco’s minimum essential medium (DMEM)(Invitrogen) supplemented with 10% adult bovine serum(Invitrogen). A9CAR cells, containing a bacterial geneticinresistance gene, were grown in the presence of G418geneticin (Invitrogen) at a concentration of 400 µg/ml toselect for CAR expressing A9 cells. Both the A9non−CAR andA9CAR cells were plated at 125 000 cells/well medium.(iii) human embryonic kidney (HEK; ATCC) 293 cellswere maintained in DMEM with 10% FCS, 1% glutamine(Invitrogen) and 1% antibiotic and antimycotic solution(Invitrogen); (4), human embryonic retinoblast (HER)911 cells were maintained in DMEM with 2% FCS and1% Penicillin/Streptomycin(Invitrogen).

Viral vectors

A replication deficient, E1 and E3 deleted, Ad5 vector,containing the gfp reporter gene under the control ofthe cytomegalovirus promoter (sAd5.gfp) was propagated

Copyright 2009 John Wiley & Sons, Ltd. J Gene Med 2009; 11: 335–344.DOI: 10.1002/jgm

In vitro evaluation of a ‘stealth’ adenoviral vector 337

[9] using HEK 293 cells and purified on a CsCl densitygradient. A pico green assay (Molecular Probes, Eugene,OR, USA) determined the frequency of sAd5.gfp/ml anda plaque assay with HER 911 cells determined Ad5.gfpinfectivity by measuring the frequency of particle formingunits. The sAd5.gfp vector was added to experimentalcultures at multiplicities of infection (MOI) of 100and 500.

Labelling Ad5.gfp DNA with Yoyo-1

To sAd5.gfp, 10 µM of Yoyo-1 (Molecular Probes) wasadded, incubated overnight on ice, and dialysed ×3 against phosphate-buffered saline (PBS) containingglycerol, CaCl2 and MgCl2, using a Slidalyser (Pierce,Rockford, IL, USA) overnight at 4 ◦C.

Coating Ad5.gfp coat with pHPMA-ONp

sAd5.gfp was coated with pHPMA-ONp, prepared asdescribed by Fisher et al. [9] at an initial concentrationof 50 mg/ml in sterile water. Briefly, 1011 particles/ml ofsAd5.gfp was added to 1M HEPES buffer (Sigma-Aldrich),pH 7.8, and sterile water. The solution was vortex mixedprior to adding pHPMA-ONp to make a final coatingconcentration of 20 mg/ml. The prepared ssAd5.gfp wasleft at room temperature for 1 h, filtered through a S400column (Amersham Biosciences, Bucks, UK) to removefree polymer, and essential salts were replaced by additionof 50 mM HEPES buffer.

Labelling pHPMA-ONp with Texas Red(TR)

sAd5.gfp was first coated with 20 mg/ml pHPMA-ONp, asdescribed above. TR cadaverine (Molecular Probes) wasadded to the ssAd5.gfp and sAd5FGF.gfp complexes ata concentration of 500 µg/ml. For labelling sAd5FGF.gfpwith TR, 100 µg/ml FGF-2 and 500 µg/ml TR were addedsimultaneously, incubated on ice overnight, filtered in amicrocon YM-50 filter device (Millipore, Billerica, MA,USA) at 13 400 r.p.m.10, 055 × g for 3 min to removefree TR, and collected by centrifugation at 13 400 r.p.m.for 3 min. The number of TR-labelled ssAd5.gfp particleswas calculated using transmission electron microscopy(Jeol, Welwyn Garden City, UK). The TR-labelled sAd, ina volume of 10 µl, was mixed with 10 µl of 2.14 × 1011

polylatex beads (Agar Scientific, Stansted, UK) and 5 µl ofuranyl acetate (Agar Scientific), and mounted onto carboncoated grids (Agar Scientific). The number of TR-labelledsAd5 particles was calculated by counting both 100 beadsand the number of virus particles simultaneously in fieldsof view. The number of virus particles/ml = the numberof virus particles counted/100 (the number of beadscounted) × 2.14 × 1011 (the total number of beads).

Retargeting sAd5.gfp with NGF andFGF-2

To produce sAd5NGF.gfp/sAd5FGF.gfp, 100 µg/ml ofeither NGF (Caltech Biosystems, Pasadena, CA, USA), orFGF-2 (a kind gift from Professor Andrew Baird, SelectiveGenetics, San Diego, CA, USA), were added to ssAd5.gfpand incubated on ice overnight and virus was prepared asdescribed previously [9].

Infection of cells with Ad5

Cells were infected at MOI of 100 and 500. The sAd5.gfp,ssAd5.gfp, sAd5FGF. gfp, and sAd5NGF. gfp were added tocells in 200 µl of 2% adult bovine serum (Sigma-Aldrich)in DMEM and incubated for 2 h at 37 ◦C, after whicha further 500 µl of 2% FBS in DMEM was added andthe cells incubated for 5 days at 37 ◦C, unless otherwisestated.

Polylysine- fluorescein isothiocyanate(FITC)

After washing cells twice in PBS to remove serum,the polylysine-FITC (0.003–0.01 mol FITC/mol lysinemonomer) (Sigma-Aldrich) was added to cultures in500 µl of serum free medium to give a final concentrationof 20 µg/ml, and incubated at 4 ◦C for 10 min. Afterremoval of the medium, cells were washed twice in PBSto remove free polylysine-FITC, and examined undera fluorescence microscope to detect net cell surfacecharge. Negatively-charged cells bind the positively-charged polylysine-FITC.

Lysotracker

The green fluorescent acidotropic probe LysotrackerDND-26 (Molecular Probes) labels and tracks acidicorganelles in live cells and was used to detect acidicpinocytotic vesicles. After removal of the medium, cellswere washed twice with PBS to remove serum and 500 µlof 75 nm Lysotracker in serum-free medium were addedto each well, cells were incubated at 37 ◦C for 5 min, andevaluated microscopically for fluorescence at intervals ofbetween 5 and 30 min.

GFP+ cell counts

Numbers of GFP+ DRGN, C6, A9non−CAR and A9CAR cellswere counted using fluorescence microscopy (Axiovision,Zeiss, Gottingen, Germany). Each well was divided intosix equal segments and the numbers of GFP+ cellstogether with the total cell number were counted withina defined area in each segment. The average percentageof fluorescent cells/well was then calculated.


338 K. Sims et al.

Confocal microscopy

Yoyo-1, TR and GFP fluorescence of DRGN were examinedand photographed using the Zeiss LSM 510 confocalmicroscope (Zeiss).

Immunocytochemistry

DRGN cultures were either single or double stainedafter fixation with 4% paraformaldehyde in PBS for10 min at room temperature, washed three times withPBS, and then incubated with blocking solution (3%BSA, 0.1% Triton X-100 in PBS) for 10 min at roomtemperature. Primary antibodies added to fixed DRGcultures were 0.042 mg/ml mouse anti-ßIII tubulin (cloneSDL.3D10) (Sigma-Aldrich) to detect neurones, and0.09 mg/ml of goat anti-adenovirus polyclonal antibody(Chemicon International, Inc., Temecula, CA, USA) todetect Ad5, and incubated for 1 h in 3% BSA in PBS.The cells were washed three times in PBS and incubatedwith the appropriate secondary antibodies (0.02 mg/mlgoat anti-mouse Texas red; Molecular Probes) and0.02 mg/ml of rabbit anti-goat Alexa fluor 633 (MolecularProbes), respectively, for 1 h in the dark at roomtemperature. Coverslips were then mounted onto slidesusing either Fluorsave (Calbiochem, San Diego, CA, USA)or Vectashield with 4′-6-diamidino-2-phenylindole (DAPI)(to detect nuclei) (Vector Laboratories, Burlingame, CA,USA) and immunostained cultures were visualised undera fluorescent microscope (Zeiss).

Statistical analysis

Significant differences between sample means werecalculated using GraphPad Prism, version 4.0 (GraphPad

Software Inc., San Diego, CA, USA) by one-way ANOVA(analysis of variance) followed by post-hoc testing withDunnet’s method.

Results

Physical parameters affecting sAd5uptake and gene expression

DRGN and NNDRG cells were identified in cultureusing previously described morphological criteria [16].DRGN and C6 glioma cells were evaluated for cellsurface charge and pinocytotic vesicle acidity, whichfacilitates pinocytosis of charged particles and viral capsiddismantling, respectively [17]. The surface charge ofsAd5, as measured by ζ potential analysis was −25 mV[9], indicating a potentially biased attraction to positively-charged cells. Positively-charged polylysine FITC boundto 90% and 80% of C6 glioma cells and DRGN,respectively, suggesting that both are predominantlynegatively charged (Figure 1). Lysotracker DND-26, anacidotrophic probe that fluoresces green at <7.0 pH,detected uptake of label into acidic pinocytic vesicles inthe cytoplasm of all C6 glioma cells and DRGN (Figure 2).

Overview of sAd5 pinocytosisDRG, C6 glioma, A9non−CAR and A9CAR cell cultures wereinfected with sAd5.gfp at various levels. A similarly lowfrequency of A9non−CAR and A9CAR cells internalisedsAd5.gfp, confirming that uptake was mediated bypinocytosis and not by CAR (Figure 3A). There wasno background autofluorescence in any of the controluntransfected culture types (Figure 3A). None of thecultured cells that internalised TR-labelled sAd5.gfp

Figure 1. Cellular binding of polylysine FITC in C6 (A) and DRG cultures (B). Positively-charged polylysine FITC binds tonegatively-charged cell surfaces and is detected by its green fluorescence. This suggests that positively charged sAd5 would bind tonegatively charged cell surfaces (For interpretation of the references to colour in this figure legend, the reader is referred to theweb version of this article).



Figure 2. Pinocytic vesicle acidity in (A) C6 cells and (B) DRGN, as detected by Lysotracker (For interpretation of the referencesto colour in this figure legend, the reader is referred to the web version of this article).

expressed GFP (Figure 3B). In all cultures, except thoseof A9non−CAR cells, significantly more cells (p < 0.001)expressed GFP after transfection with noncoated Ad5.gfp(Figure 3B) than with sAd5.gfp (p < 0.001) (Figure 3B).

A9non−CAR and A9CAR cellsAs might be predicted, significantly more A9CAR thanA9non−CAR cells expressed GFP at a MOI of 500 wheninfected with uncoated Ad5.gfp (p < 0.001) (Figure 3B).Internalisation of TR-tagged sAD5.gfp occurred in a smallnumber of A9non−CAR and A9CAR cells, but only at a MOI of500 (Figure 3A), suggesting that the stealth coat blockednormal Ad5 knob domain binding to CAR. Despite thelow frequency uptake, no GFP expression was seen inA9non−CAR and A9CAR cells infected with sAd5.gfp.

C6 glioma cellsAlthough occasional C6 glioma cells pinocytosed TR-labelled sAd5.gfp at a MOI of 500 (Figure 3A), no GFPexpression was observed (Figure 3B). Even the noncoatedAd5 induced only low levels of GFP expression in C6 cells.

DRG culturesOnly 7% of cultured DRGN expressed GFP after infectionwith a MOI of 500 sAd5.gfp, despite 70% pinocytosisof TR-labelled sAd5.gfp (p < 0.001) (Figure 3A). TR-labelled sAd5.gfp was also taken up by NNDRG cells butGFP was not expressed (Figure 4A). GFP expression wasonly apparent in DRG cultures infected with noncoatedAd5.gfp (Figure 4B).

Confocal microscopy confirmed that TR-labelledsAd5.gfp (Figure 5, red) was pinocytosed into DRGN,and incorporated into cytoplasmic pinocytotic vesicles(Figure 5, merged), where it was exclusively co-localisedwith Yoyo-1-labelled Ad5 DNA (Figure 5) and not within

the DAPI stained nuclei of DRGN, indicating cytoplasmicvesicular entrapment of the coated viral DNA (Figure 5,white). More DRGN contained TR-labelled sAd5.gfp intheir cytoplasm than NNDRG (p < 0.001); however, GFPwas not expressed.

Retargeting sAd5.gfp with FGF-2 andNGF leads to GFP expression in DRGN

After transfection of DRG cultures with sAd5NGF.gfp ata MOI of 500 (Figure 6), 66.6% of DRGN and 16.4%of NNDRG cells expressed GFP, which are frequenciesthat did not differ significantly from values achievedwith Ad5.gfp (Figure 6), indicating that NGF-retargetedsAd5.gfp and noncoated Ad5.gfp both bias uptake toDRGN and are equally efficient as gene delivery vectorsfor DRGN and NNDRG cells. By contrast, infection withsAd5FGF.gfp gave equal frequencies of GFP expressingDRGN and NNDRG cells.

Localisation of coat and caspid of thesAd5.gfp and sAd5FGF.gfp in DRGN

The fate of the internalised coat and the capsid of sAd5.gfpand sAd5FGF.gfp in DRG cell cultures was studied bytracking the TR-labelled pHPMA-ONp coat and Ad5 hexonimmunostaining with anti-hexon antibody (Figure 7).DRGN transfected with sAd5.gfp did not express GFPbut co-localised the pHPMA-ONp coat and the Ad5 hexonin cytoplasmic pinocytic vesicles (Figure 7A). However,after infection of DRG cell cultures with sAd5FGF.gfp, thepHPMA-ONp coat had disassembled from the Ad5 viralcapsid in GFP-expressing DRGN (Figure 7B) (i.e. virusand stealth coat had physically dissociated). The resultsdemonstrate that retargeting sAd5 with either NGF orFGF-2 facilitated particle uncoating and expression of


340 K. Sims et al.

Figure 3. A comparison of the frequency of virus pinocytosisand resultant GFP expression in cells infected with Ad5.gfp andsA5.gfp. The untransfected cultures determined the backgroundlevels of autofluorescence in cells. (A) Comparison of TR-labelledsAd5.gfp pinocytosis at a MOI of 100 and 500 in DRGN andNNDRG cells in the same DRG cultures, and also in cultures ofC6, A9non−CAR and A9CAR cells. DRGN show significantly higherlevels of pinocytosis of TR-labelled sAd5.gfp at a MOI of 100and 500 compared to NNDRG cells (∗∗∗p < 0.0001, ANOVA).(B) Comparison of the percentage GFP expression betweenAd5.gfp and sAd5.gfp at a MOI of 100 and 500 in DRGN andNNDRG in the same DRG cultures, and also in cultures ofC6, A9non−CAR and A9CAR cells. Levels of GFP expression aresignificantly higher with Ad5.gfp compared to sAd5.gfp at a MOIof 500 in all cell types (∗∗∗p < 0.0001; ∗p < 0.05, ANOVA, n = 18for samples, n = 8 for controls)

GFP. Furthermore, retargeting the sAd5 with NGF, but notFGF-2, resulted in fewer NNDRG cells expressing GFP.

Discussion

Coating Ad5 with pHPMA-ONp attenuates both theimmune response and promiscuous infection, by mask-ing surface epitopes and CAR binding sites [9,18]. Inthe present study, we show that, although pinocytosisbiased uptake of sAd5 to neurones in mixed DRG cul-tures, once internalised into pinocytotic vesicles, sAd5 did

not uncoat, preventing dismantling of the capsid, nucleartranslocation of viral DNA and expression of the trans-gene. However, after retargeting the pHPMA-ONp coatwith FGF-2 and NGF so that virus was taken up into endo-cytic vesicles, particle uncoating occurred and transgeneexpression was facilitated, suggesting that re-formulationof the sAd5 coat is required in order to exploit neuronalpinocytotic mechanisms for targeted gene expression.

The importance of cell charge

Both sAd5 and the surface of most DRGN werenegatively charged, which might be expected to reduceelectrostatic neuronal binding of sAd5.gfp. However, ahigh percentage of DRGN did pinocytose TR-labelledsAd5gfp and thus plasmalemma charge is not a primarydeterminant of neuronal pinocytotic uptake. C6 gliomacells and NNDRG cells were also negatively chargedand pinocytosed sAd5.gfp, although with very lowefficiency. The significantly higher frequency of DRGNthat pinocytosed sAd5.gfp compared to NNDRG cellsreflects the high rates of active pinocytosis observed inneuronal somata and axon growth cones compared to thelow pinocytosis rates of non-neuronal cells [19].

Intracellular viral processing

Although pinocytosis significantly biased uptake of TR-labelled atropic sAd5.gfp into DRGN compared to NNDRGcells, no GFP expression was observed. Colocalisation ofTR-labelled pHPMA coat with Yoyo-1-labelled sAd5.gfpDNA in cytoplasmic pinocytotic vesicles of GFP-negativeDRGN demonstrated that the stealth coat remainedassociated with the Ad5 capsid and thus prevented Ad5escape and capsid dismantling. Lysotracker-labelledpinocytic vesicles of cultured DRGN and showed themto be acidic. A low pH environment is necessaryfor Ad5 capsid dismantling and escape of viral DNAinto the cytoplasm [17], which are prerequisites fornuclear trafficking and transgene expression. Althoughneuronal pinocytic vesicle were acidic and would favourcapsid dismantling, it appears that the pinocytic vesicleenvironment was unfavourable for particle uncoating, sothat capsid dismantling, release of viral DNA and nucleartrafficking could not occur.

In DRGN that did express the transgene afterretargeting the sAd5.gfp with NGF or FGF-2, the capsidwas seen within DRGN dissociated from the pHPMA-ONpstealth coat, in direct contrast to DRGN infected withuntargeted sAd5.gfp, where consistent co-localisation ofcoat and capsid in cytoplasmic vesicles occurred. Thebiased uptake of sAd5 into neurones suggests a potentialfor pinocytic targeting of neurones if the stealth coatcan be stripped from the virus after internalisation by,for example, formulating a coat that dismantles moreeasily after internalisation into acidic pinocytotic vesicles.Exploiting pinocytosis remains a potential means oftargeting sAd5 to highly pinocytic cells like neurones.



Figure 4. TR-labelled sAd5.gfp (red) uptake and GFP expression (green) observed after infection of DRG cultures with TR-labelledsAd5.gfp. TR labelling was entrapped in perinuclear cytoplasmic vesicles (A), whereas noncoated Ad5.gfp led to expression of GFPin both DRGN and NNDRG cells (B). White arrows, DRGN; black arrowheads, NNDRG cells (For interpretation of the references tocolour in this figure legend, the reader is referred to the web version of this article).

Although particle uncoating may require more thanjust an acidic environment [20], others have used labile,redox-labile and other intracellularly triggered linkageswithin the pHPMA coat that may help the deliveredgenetic particle to uncoat. For example, covalent cross-linking of primary amines in poly L-lysine/DNA complexeswith a cross-linking agent that are cleaved intracellularlyby reduction facilitated efficient transgene expression in

Xenopus oocytes and the cytoplasm and nucleus of Rat-1fibroblasts [21,22].

Receptor retargeting of sAd5

A previous study by Fisher et al. (2001) showed successfulsAd5 retargeting and reporter gene expression in A549cells with sAd5FGF.gfp. To test whether retargeted


342 K. Sims et al.

Figure 5. Uptake of labelled sAd5.gfp into DRGN. Pseudo-coloured confocal microscopy of DRGN in culture showing co-localisation(purple) of Yoyo-1-labelled Ad5.gfp (blue) with the TR-labelled pHPMA-ONp coat (red) in DRGN cytoplasm. Nuclei (white) werestained with DAPI (For interpretation of the references to colour in this figure legend, the reader is referred to the web version ofthis article).

Figure 6. Frequency of GFP expression in DRG cultures afterinfection with sAd5FGF.gfp and sAd5NGF.gfp (MOI = 500).The histogram shows the levels of GFP expression in DRGNand NNDRG cells demonstrating no specific ligand-mediatedtargeting to either cell type, although targeting with NGF biaseduptake towards neurones. NS, not significant

sAd5.gfp could infect and transduce DRGN leadingto transgene expression, sAd5.gfp was retargeted withFGF-2 and NGF. Both sAd5FGF.gfp and sAd5NGF.gfp

transfected both neuronal and NNDRG cells and expressedsignificant levels of GFP. The GFP expression of sAd5gfpafter ligand retargeting may be explained by the capsidretargeted particle exploiting a similar receptor-mediatedendosomal pathway to that used by Ad5 becauselocal receptor binding activates αv-integrins that triggerclathrin-mediated viral endocytosis thereby facilitatingAd5 endosomal escape [20,22–24]. Receptor-mediateduptake through the FGF-2 and NGF receptor pathwaysmay similarly trigger particle uncoating and escape fromthe endosome resulting in GFP expression. Importantly,retargeted sAd5FGF.gfp has a lower systemic toxicitythan wild-type Ad5, even when a higher concentrationof virus is delivered [10,18]. Hence, these resultssuggest that retargeting of the stealth coat with aneuronal ligand facilitates sAd5 uncoating and endosomalescape.

Conclusions

The results of our in vitro studies with primary culturesof mammalian neurones and sAd5gfp suggest that addinga pHPMA-ONp stealth coat to Ad5 inhibits knob domainligand binding to CAR, preventing widespread indiscrim-inate gene delivery and biases viral uptake to activelypinocytic cells, such as neurones. Moreover, the formu-lated sAd5 has reduced systemic toxicity [7,10,18–25].



Figure 7. Localisation of sAd5.gfp (A) and sAd5FGF.gfp (B) in DRGN at 3 days after infection. In cultures treated with sAd5.gfp (A),the pHPMA coat (red) and Ad5 hexon (blue) are co-localised (purple) to the cytoplasm of cells that do not express GFP (green).Note the presence of extracellular pHPMA coat. By contrast, after infection of cultures with sAd5FGF.gfp in (B), the Ad5 hexon(blue) is visible distinct from the pHPMA coat in GFP expressing DRGN. Note the presence of extracellular uncoated Ad5. Whitearrows, DRGN (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of thisarticle).

However, when internalised into pinocytotic vescicles, the‘stealth’ coat blocked capsid dismantling, vesicular escape,nuclear trafficking and transgene expression, thus limit-ing the use of this pathway for targeting sAd5 uptake.The untargeted stealth coat may be useful if a coatdesign can be made to facilitate particle dismantling after

uptake into pinocytotic vesicles. Targeting the ‘stealth’coat with a ligand such as NGF enables sAd5 to uncoat andgene expression to occur in DRGN. To further improveneuronal retargeting of sAd5, exploiting a more neu-rone specific receptor, not so far identified, would berequired.


344 K. Sims et al.

Acknowledgements

We would like to thank the BBSRC (grant numbersBB/C50466X/1 and E18035) for generously funding this work.All authors declare that there are no competing financial inter-ests.

References

1. Dechecchi MC, Melotti P, Bonizzato A, Santacatterina M,Chilosi M, Cabrini G. Heparan sulfate glycosaminoglycans arereceptors sufficient to mediate the initial binding of adenovirustypes 2 and 5. J Virol 2001; 75: 8772–8780.

2. Chillon M, Lee JH, Fasbender A, Welsh MJ. Adenoviruscomplexed with polyethylene glycol and cationic lipid isshielded from neutralizing antibodies in vitro. Gene Ther 1998;5: 995–1002.

3. Mok H, Palmer DJ, Ng P, Barry MA. Evaluation of polyethyleneglycol modification of first-generation and helper-dependentadenoviral vectors to reduce innate immune responses. MolTher 2005; 11: 66–79.

4. Hofherr SE, Mok H, Gushiken FC, Lopez JA, Barry MA.Polyethylene glycol modification of adenovirus reduces plateletactivation, endothelial cell activation, and thrombocytopenia.Hum Gene Ther 2007; 18: 837–848.

5. Hofherr SE, Shashkova EV, Weaver EA, Khare R, Barry MA.Modification of adenoviral vectors with polyethylene glycolmodulates in vivo tissue tropism and gene expression. Mol Ther2008; 16: 1276–1282.

6. Weaver EA, Barry MA. Effects of shielding adenoviral vectorswith polyethylene glycol (PEG) on vector-specific and vaccine-mediated immune responses. Hum Gene Ther 2008.

7. Crowther C, Ely A, Hornby J, et al. Efficient inhibition ofhepatitis B virus replication in vivo using PEG-modifiedadenovirus vectors. Hum Gene Ther 2008.

8. Carlisle RC, Benjamin R, Briggs SS, et al. Coating of adeno-associated virus with reactive polymers can ablate virus tropism,enable retargeting and provide resistance to neutralisingantisera. J Gene Med 2008; 10: 400–411.

9. Fisher KD, Stallwood Y, Green NK, Ulbrich K, Mautner V,Seymour LW. Polymer-coated adenovirus permits efficientretargeting and evades neutralising antibodies. Gene Ther 2001;8: 341–348.

10. Green NK, Herbert CW, Hale SJ, et al. Extended plasmacirculation time and decreased toxicity of polymer-coatedadenovirus. Gene Ther 2004; 11: 1256–1263.

11. Hoganson DK, Sosnowski BA, Pierce GF, Doukas J. Uptake ofadenoviral vectors via fibroblast growth factor receptors involvesintracellular pathways that differ from the targeting ligand. MolTher 2001; 3: 105–112.

12. Gupta V, Wang W, Sosnowski BA, Hofman FM, Chen TC.Fibroblast growth factor-2-retargeted adenoviral vector forselective transduction of primary glioblastoma multiformeendothelial cells. Neurosurg Focus 2006; 20: E26.

13. Morrison J, Briggs SS, Green N, et al. Virotherapy of ovariancancer with polymer-cloaked adenovirus retargeted to theepidermal growth factor receptor. Mol Ther 2008; 16: 244–251.

14. Thoma C, Wieland S, Moradpour D, et al. Ligand-mediatedretargeting of recombinant adenovirus for gene transfer in vivo.Gene Ther 2000; 7: 1039–1045.

15. Griffiths G, Back R, Marsh M. A quantitative analysis of theendocytic pathway in baby hamster kidney cells. J Cell Biol1989; 109: 2703–2720.

16. Sims K, Ahmed Z, Gonzalez AM, et al. Targeting adenoviraltransgene expression to neurons. Mol Cell Neurosci 2008; 39:411–417.

17. Greber UF, Willetts M, Webster P, Helenius A. Stepwisedismantling of adenovirus 2 during entry into cells. Cell 1993;75: 477–486.

18. Green NK, Morrison J, Hale S, et al. Retargeting polymer-coatedadenovirus to the FGF receptor allows productive infection andmediates efficacy in a peritoneal model of human ovarian cancer.Journal of Gene Medicine 2008; 10: 280–289.

19. Tischner K. Uptake of horseradish peroxidase by sensory nervefibres in vitro. J Anat 1977; 124: 83–97.

20. Meier O, Greber UF. Adenovirus endocytosis. J Gene Med 2004;6(Suppl 1): S152–S163.

21. Oupicky D, Carlisle RC, Seymour LW. Triggered intracellularactivation of disulfide crosslinked polyelectrolyte gene deliverycomplexes with extended systemic circulation in vivo. Gene Ther2001; 8: 713–724.

22. Carlisle RC, Etrych T, Briggs SS, Preece JA, Ulbrich K, Sey-mour LW. Polymer-coated polyethylenimine/DNA complexesdesigned for triggered activation by intracellular reduction.J Gene Med 2004; 6: 337–344.

23. Svensson U. Role of vesicles during adenovirus 2 internalizationinto HeLa cells. J Virol 1985; 55: 442–449.

24. Meier O, Gastaldelli M, Boucke K, Hemmi S, Greber UF. Earlysteps of clathrin-mediated endocytosis involved in phagosomalescape of Fcgamma receptor-targeted adenovirus. J Virol 2005;79: 2604–2613.

25. Fisher KD, Green NK, Hale A, Subr V, Ulbrich K, Seymour LW.Passive tumour targeting of polymer-coated adenovirus forcancer gene therapy. J Drug Target 2007; 15: 546–551.


in vitro evaluation of a ‘stealth’ adenoviral vector for targeted gene delivery to adult...

Documents