High efficiency transfection of porcine vascular cells in vitro with a synthetic vector system

RESEARCH ARTICLE

High efficiency transfection of porcine vascularcells in vitro with a synthetic vector system

Richard Parkes1

Qing-Hai Meng1

K. Elena Siapati1

Jean R. McEwan2

Stephen L. Hart1*

1Molecular Immunology Unit,Institute of Child Health, UniversityCollege London, London, UK2The Cardiovascular RepairRemodelling Group, The HatterInstitute, Royal Free and UniversityCollege London Medical School,London, UK

*Correspondence to: Dr Stephen L.Hart, Molecular Immunology Unit,Institute of Child Health, London,WC1N 1EH, UK.E-mail: [email protected]

Received: 30 October 2001

Revised: 19 December 2001

Accepted: 9 January 2002

Abstract

Background Gene therapy strategies for the treatment of vascular diseasesuch as the prevention of post-angioplasty restenosis require efficient, non-toxic transfection of vascular cells. In vitro studies in these cells contribute tovector development for in vivo use and for the evaluation of genes withtherapeutic potential. The aim of this project was to evaluate a novel syntheticvector consisting of a liposome (L), an integrin targeting peptide (I), andplasmid DNA (D), which combine to form the LID vector complex.

Methods Cultures of porcine smooth muscle cells and endothelial cells wereestablished and then transfected with the LID vector, using the reporter genesluciferase and green fluorescent protein and the metalloprotease inhibitorTIMP-1.

Results The LID vector system transfected primary porcine vascular smoothmuscle cells and porcine aortic endothelial cells with efficiency levels of40% and 35%, respectively. By increasing the relative DNA concentrationfour-fold, incubation periods as short as 30 min achieved the same levels ofluciferase transgene expression as 4 h incubations at lower DNA concentra-tions. The transfection did not affect cell viability as measured by theirproliferative potential. Serum levels of up to 20% in the transfection med-ium had no adverse affect on the efficiency of transfer and gene expressionin either cell type. Transfections with the cDNA for TIMP-1 produced pro-tein levels that peaked at 130 ng/ml per 24 h and persisted for 14 days at10 ng/ml per 24 h.

Conclusion This novel vector system has potential for studies involvinggene transfer to cardiovascular cells in vitro and in vivo. Copyright # 2002John Wiley & Sons, Ltd.

Keywords gene therapy; non-viral; smooth muscle cells; endothelial cells;cardiovascular disease

Introduction

Several viral and non-viral vectors have been investigated for gene transfer inthe vascular system, yet none satisfies all of the requirements of safety andefficacy [1]. Non-viral vectors that deliver plasmid DNA, such as cationicliposomes and DNA–polycation complexes, are safer generally than viralvectors and are easier to prepare and use. In addition, unlike viruses, thereare no nucleic acid packaging constraints, which enables them to deliverlarger DNA molecules containing genomic regulatory elements or multiplegenes, while their low level of immunogenicity is favourable for their usein vivo. The major shortcoming of synthetic vector systems has been their poor

THE JOURNAL OF GENE MEDICINEJ Gene Med 2002; 4: 292–299.Published online 10 April 2002 in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/ jgm.265

Copyright # 2002 John Wiley & Sons, Ltd.

efficiency of gene transfer relative to viral vectors,particularly in primary cell cultures and in vivo [2].Efforts to improve the efficiency of synthetic vectors haveincluded such techniques as ultrasonication of the cells[3], but have largely been directed towards modifica-tions of the lipid component of the vector. The in vivo

transfection efficiency of rat adventitial and medialcells with lipofectin/DNA complexes was only 0.05%, asopposed to 10% for an adenoviral vector, suggesting aneed for improvement in the formulation of syntheticvectors [4].

Integrins are heterodimeric membrane proteins that areimportant for attachment of cells to the extracellularmatrix, cell–cell interactions, and signal transduction.They are exploited as receptors for cell attachment andentry by a number of pathogenic viruses, includingadenovirus, echovirus, and foot-and-mouth disease virus.An integrin-targeted polycation–DNA complex vectorsystem may thus mimic the infection of cells by viruses.We have developed the LID vector system, a modular,self-assembling complex containing a liposome, such aslipofectin (L), an integrin-binding peptide (I), and plas-mid DNA (D) [5–8]. The peptides contain an integrin-targeting domain and a 16-lysine tail for electrostaticcomplexing of DNA. These complexes form particles witha mean diameter of approximately 50 nm. In the opti-mal mixture of components, determined by transfectionefficiency, it was calculated that solutions containingapproximately 6000 peptide molecules are mixed witheach plasmid molecule [5], although it is not known if allpeptide molecules are associated with LID particles. TheLID vector transfects a range of cell lines with greater than50% efficiency in vitro [5], but transfection of primarycell types has not been reported previously. Primary cellsare usually much more difficult to transfect with non-viral vectors. Vascular smooth muscle cells (VSMCs) andendothelial cells express a5b1 and avb3 integrins thatmediate cell adhesion to extracellular matrix proteinsincluding fibronectin and vitronectin [9]. These cells aretherefore potential targets for transfection by the LIDvector system using peptides that target these integrins.

We aim to develop the LID vector as a tool forapplications to the study of cardiovascular disease and forpotential gene therapy of these diseases. Our initialstrategy was to optimize and evaluate the LID vectorsystem for transfection of cardiovascular target cellsin vitro, modelling obstacles that might be anticipatedin vivo. In this report, we demonstrate that the LID vectorcan be optimized to achieve transfection efficiencies of35% in primary porcine endothelial cells and 40% inprimary VSMCs, in vitro, with exposure conditionssufficiently short to be relevant to in vivo use.

Expression of a transgene encoding for human tissueinhibitor of metalloprotease-1 (hTIMP-1), introduced byan adenoviral vector, has been shown to be a potentiallyeffective approach for the prevention of restenosis afterballoon catheter angioplasty [10,11]. Here we havedemonstrated secretion of bioactive TIMP-1 from porcineVSMCs transfected in vitro with the LID vector containing

a cDNA encoding hTIMP-1, which persisted for at least 14days, at 10 ng/ml per 24 h, illustrating the potential ofthis non-viral strategy for gene therapy of restenosis.

Materials and methods

Plasmids

The luciferase reporter gene vector pCILux and pEGFP-N1(Clontech, Heidelberg, Germany) were used for in vitro

optimization of transfection. The human TIMP-1 (hTIMP-1) expression vector pGEMhTIMP-1 was generated bysubcloning a PCR fragment containing hTIMP-1 cDNAand BamH1 and HindIII cloning sites into pGEM2(Promega, Southampton, UK), which was kindly donatedby Professor Dylan Edwards, University of East Anglia.TIMP-1 was then subcloned into the pCI vector under thecontrol of the CMV-IE promoter (Promega, Southampton,UK) to generate pCIhTIMP. All plasmids were grown in E.

coli DH5a and purified on columns according to themanufacturer’s instructions (Hybaid, Ashford, UK). Theplasmid DNA pellets were washed in 70% ethanol andthen resuspended in sterile non-pyrogenic water (Baxter,Thetford, UK).

Cell culture

Porcine arterial endothelial cells (PAECs) were isolatedand purified from porcine thoracic aortas as describedpreviously [12] and maintained at 37uC in Medium 199supplemented with 5% fetal calf serum (Life Technolo-gies, Paisley, UK), 1% amphotericin B (Fungizone), and2% penicillin/streptomycin. Porcine vascular smoothmuscle cells (PVSMCs) were prepared using the explantmethod described previously [13] and identified by theirtypical ‘hill-and-valley’ growth pattern and by stainingwith an anti-smooth muscle actin antibody (Dako Ltd.,Ely, UK). All cells used in this study were from passages3–8 to reduce any affects of senescence on transfectionefficiency.

Transfection reagents

Peptide 6, [K16]GACRRETAWACG, contains an a5b1integrin-targeting motif, RRETAWA [5], while the control,Peptide 6J2, with the sequence [K16]GACATRWARECG,contained the same constituent amino acids as thetargeting motif of Peptide 6 but in a jumbled order.Peptides were made by Zinsser Analytic (Maidenhead,Berks, UK), dissolved at a concentration of 1 mg/mlin OptiMEM medium (Life Technologies, Paisley, UK),and cyclized by exposure to air overnight [5]. Peptidesolutions were then sterilized by filtration through 0.2 mmAcrodisc filters (Gelman Sciences, Ann Arbor, MI, USA).The cationic lipid used was lipofectin (Life Technologies,Paisley, UK).

Transfection of Primary Porcine Vascular Cells 293

Copyright # 2002 John Wiley & Sons, Ltd. J Gene Med 2002; 4: 292–299.

In vitro transfections

PVSMCs were seeded in a volume of 1 ml in 24-well platesat 5r104 per well 24 h prior to transfection. LID com-plexes were formed by mixing components in the follow-ing order and amounts: 100 ml of lipofectin (7.5 mg/ml inOptiMEM), 40 ml of integrin binding peptide (0.1 mg/mlin OptiMEM), and 100 ml of plasmid DNA (10 mg/ml inOptiMEM), giving a final charge ratio of 1 : 1, 3 : 1 or 7 : 1,calculated in terms of net positive charge on the peptide(N) to negatively charged phosphate groups (P) of theplasmid, as described previously [5]. Integrin-bindingpeptide/DNA (ID) complexes were prepared in the sameway as LID complexes but without lipofectin. Lipofectin/DNA (LD) complexes were formed by mixing lipofectinwith plasmid DNA at a weight ratio of 5 : 1. Com-plexes were incubated at room temperature for 10 minand then diluted to 0.5 ml in OptiMEM (range of DNA0.5–8 mg/well in 0.5 ml) and applied to cells afterremoving the medium and washing once with PBS(phosphate buffered saline). Transfection proceeded at37uC for 4 h, unless otherwise stated, after whichcomplexes were replaced with supplemented media.Plates were returned to 37uC incubator for 24 h, afterwhich they were assayed for reporter gene activity.

Luciferase activity was assayed after harvesting thecells in 100 ml of luciferase reporter passive lysis buffer(Promega, Southampton, UK), according to the manu-facturer’s instructions, using a Lucy 1 luminometer(Labtech, Uckfield, UK) for 10 s. Protein concentrationsof cell lysates were determined using Protein Assayreagent (BioRad, Hemel Hempstead, UK) and luciferaseactivity was expressed as reactive light units per milligramof protein (RLU/mg).

FACS analysis

The transfection efficiency of PVSMCs and PAECs withpEGFP was analysed by flow cytometry. Twenty-fourhours after transfection, cells were washed twice withPBS and detached from the plate by addition of 250 mlof 1r trypsin–EDTA (Life Technologies, Paisley, UK) for5–10 min at 37uC. Cells were subsequently washed in 1rPBS, fixed in 4% paraformaldehyde, and stored at 4uCprior to analysis using a FACS sorter (FACSCalibur,Becton Dickinson, Oxford, UK) and CellQuest software.

Integrin expression

Analysis of a5 and a4 integrin expression on PVSMCs andPAECs was performed using mouse anti-human CD49eand CD49d antibodies (Pharmingen, San Diego, CA,USA), which cross-react with porcine integrins. A goatanti-mouse FITC-labelled secondary antibody (DAKO,Ely, UK) was used to determine antibody binding by flowcytometry (FACSCalibur, Becton Dickinson, Oxford, UK).A mouse IgG antibody (Chemicon International Ltd.,Harrow, UK) was used as an isotype control.

In vitro proliferation of PVSMCs

PVSMCs were plated in 96-well plates at a density of6.3r103 cells per well and transfected the following daywith 1 mg of DNA per well as described above. Cellproliferation following LID transfection was estimatedby [3H]thymidine incorporation. One, two, and threedays after LID transfection, 0.5 mCi/well of [3H]thymidinein a total of 20 ml was added per well and the incor-porated tritium measured 16 h later.

TIMP-1 expression analysis by westernblotting and reverse zymography

Expression of recombinant human TIMP-1 by PVSMCsfollowing transfection with LID complexes carryingpCIhTIMP-1 (LID.TIMP) was analysed by western blot-ting of conditioned media, while functional activity ofTIMP was monitored by reverse zymography. Cells werewashed 24 h after transfection and QBSF 51 (Sigma,Poole, UK) serum-free medium was added to the cells.One day later, conditioned medium was harvested andconcentrated using Centricon-3 filter units (Amicon, Inc.,Stonehouse, UK). Equal amounts of protein were loadedand electrophoresed on a 12% SDS-polyacrylamide geland transferred onto nitrocellulose filters by semi-dryelectroblotting with a BioRad Transblot (BioRad, HemelHempstead, UK). Filters were probed for 2 h with a mouseanti-human TIMP-1 monoclonal antibody (ChemiconInternational, Inc., Temecula, CA, USA) diluted 1 : 200,followed by a horseradish peroxidase-conjugated rabbitanti-mouse monoclonal antibody (Dako, Ely, UK) diluted1 : 2000. Immunoreactive proteins were visualizedusing the ECL system (Amersham International, LittleChalfont, UK). Conditioned serum-free medium fromdexamethasone-treated human fibrocarcinoma (HT1080)cells was used as a source of human TIMP-1 positivecontrol [14]. TIMP activity was assayed by reversezymography as described previously [15]. Supernatantsamples in non-reducing Laemlli buffer were run on a12% SDS-polyacrylamide gel containing 2 mg/ml gelatin(Sigma, Poole, UK) plus a source of gelatinase (7% v/vconditioned media from hamster BHK cells). Gels werewashed in 2.5% Triton X-100 for 15 min; followed by2.5% Triton X-100, 50 mM Tris–HCl (pH 7.5), 5 mMCaCl2 for 2 h, and then incubated at 37uC in 50 mMTris-HCl (pH 7.5), 5 mM CaCl2 for 16 h before stainingwith Gelcode Blue stain reagent (Pierce, Rockford, IL,USA). Functional TIMP activity was identified as zones ofinhibition of gelatinase activity on a partially digestedgelatin background.

TIMP-1 expression analysis by ELISA

PVSMCs (2r105) were cultured in 25 cm2 flasks andtransfections performed with the LID vector carryingpCIhTIMP or with the empty control pCI as above, exceptthat the complexes contained 12 mg of plasmid in 3 ml ofbuffer. The quantitative production of human TIMP-1 was

294 R. Parkes et al.


immunoassayed using a Quantikine ELISA kit (R&DSystems, Oxon, UK) according to the manufacturer’sinstruction, using triplicate samples of 50 ml of mediumsupernatant or hTIMP-1 standard, the recombinant humanTIMP-1 provided in the kit. Horseradish peroxidase-conjugated polyclonal antibody against hTIMP-1 wasadded to each well and hydrogen peroxide was used todevelop the colour reaction. The reaction was terminatedby adding 50 ml of stop solution and the optical density ofthe solution in each well was determined at 450 nm usinga microplate reader.

Statistical analysis

All experiments were carried out in triplicate and theresults are shown as meanststandard deviation. Thestatistical significance of results was assessed using theStudent t-test and results were taken to be significantwhen p<0.05.

Results

Integrin-targeted transfection

PVSMCs and PAECs were labelled with a5 and a4integrin-specific monoclonal antibodies and integrin

expression was determined by FACS analysis (data notshown). Both PVSMCs and PAECs expressed the a5integrin subunit (93.3% and 94.1%, respectively), whilethe a4 integrin subunit was not identified on either celltype. Peptide 6 was used in LID vector formulations(LID-6), since this peptide contains an RRETAWA motifflanked by two cysteine residues, which targets the a5b1integrin [5,16]. LID complexes were generated by mixingthe peptide component with lipofectin and plasmidDNA in a ratio of 0.75 : 4 : 1 (L : I : D). These complexeshave been shown to be particulate [5], although theprecise alignment and association of vector componentsare presently unknown. The diagram in Figure 1a is arepresentation of one possible structure. LID vector trans-fection efficiency was significantly higher than lipofectin/DNA in PVSMCs (three- to four-fold) and PVSMCs(six- to seven-fold), while integrin-targeted peptide/DNA (ID) complexes lacking lipid displayed extremelylow transfection levels (Figure 1b). Flow cytometricanalysis of PVSMCs and PAECs transfected with pEGFPindicated transfection efficiencies of 40% in PVSMCs(Figure 1c) and 35% in PAECs (Figure 1d).

Receptor-mediated transfection

The integrin targeting specificity of the vector in PAECsand PVSMCs was assessed by comparing the transfection

Figure 1. Integrin specificity of transfection. (a) Diagram of the formation of the LID vector complex. (b) Transfection effi-ciency of the LID vector compared with LD and ID formulations in PVSMCs (solid bars) and PAECs (open bars) in luciferasetransfections. Expression levels are plotted as a percentage of the expression achieved with the LID formulation. PVSMCs (c)and PAECs (d) were transfected with the LID vector containing the plasmid pEGFP. The percentage of green fluorescing cellswas determined by flow cytometry



efficiency of LID-6 with LID complexes containing thecontrol Peptide 6J2 (LID-6J2). LID-6 and LID-6J2 com-plexes were made at three different charge ratios, 1 : 1,3 : 1, and 7 : 1, and luciferase activity was assessed. Thetransfection efficiency in PAECs of LID-6 was significantlygreater than that of LID-6J2 at the lower charge ratios of1 : 1 (2350t430 RLU/mg for LID-6 vs. 30t26 RLU/mgfor LID-6J2) and 3 : 1 (t2-fold difference), indicatingenhancement of transfection by receptor targeting of thecomplex, while at the higher charge ratio of 7 : 1 bothvectors displayed similar levels of efficiency (Figure 2a).In PVSMCs, Peptide 6 also generated greater luciferaseexpression at the 3 : 1 charge ratio, but this was notstatistically significant compared with Peptide 6J2(Figure 2b). Calculations for significance were performedby a Student t-test, with significance accepted wherep<0.05.

Optimization of DNA dose andtransfection incubation period

The transfection conditions for LID complexes wereoptimised in PAECs and PVSMCs using the luciferasereporter gene (Figures 2c and 2d). In endothelial cells,increasing the duration of the transfection incubationfrom 30 min to 4 h increased luciferase expression forDNA doses up to 4 mg/ml. However, at the highest dose ofDNA of 8 mg/ml, luciferase expression levels were muchlower with the 4 h incubation than with the 30 minincubation (Figure 2c). Expression levels achieved at the30 min incubation with plasmid DNA at 8 mg/ml were ashigh as those achieved with a 4 h transfection incubationat 4 mg/ml. In PVSMCs the 4 h transfection incubation

led to higher levels of luciferase expression at allDNA doses, with peak expression achieved at 2 mg/mlDNA with a 4 h incubation. A plateau of expressionwas observed at 4 mg/ml with the 30 min incubation(Figure 2d).

Vector toxicity assessment

LID vector toxicity in PVSMCs was assessed by the abilityof cells to continue proliferating in the first 3 days aftertransfection in vitro by incorporation of tritiated thymi-dine. Cells transfected with the LID vector containingeither pCILux or pCI-TIMP1 were significantly less activethan cells maintained in complete medium, although evencells treated with OptiMEM alone showed a decline intritium incorporation. By 3 days post-transfection, allcells displayed similar levels of tritium incorporation(Figure 3a).

Serum tolerance of the LID vector

Transfections of PVSMCs were performed with the LIDvector in concentrations of fetal calf serum up to 20% ofthe transfection medium for 4 hours. Transfected luci-ferase expression levels were not significantly reducedcompared with transfections performed in the absence ofserum (Figure 3b).

Expression of human TIMP1 in PVSMCsby LID transfection

LID.TIMP-1 transfected cells produced a 28 kD band onwestern blots, characteristic of human TIMP-1, while

Figure 2. (a) PAECs and (b) PVSMCs were transfected with pCILux by the LID vector containing the integrin targeting Peptide6 (solid bars) or the control Peptide 6J2 (open bars) at three different ratios, calculated by charge. Statistical analysis was per-formed by the Student t-test, with * indicating p<0.05. The effects of DNA concentration and transfection incubation periodon transfection efficiency with luciferase were studied (c, d). PAECs (c) and PVSMCs (d) were transfected by the LID vectorwith 0.5, 1, 2, 4 or 8 mg of plasmid pCILux per well. All transfection complexes were formulated at a 7 : 1 charge ratio. Trans-fection incubations were performed for 30 min (open bars) or 4 h (solid bars)



untransfected cells did not (Figure 4a). The biologicalactivity of TIMP-1 was demonstrated by reverse zymo-graphy, in which human TIMP-1 produced by HT1080control cells and LID.TIMP-1 transfected PVSMCs wasvisualized as a band produced by the TIMP-1 prevent-ing degradation of the gelatin in the gel by gelatinase(Figure 4b).

The kinetics of TIMP-1 expression in pCIhTIMP-1-transfected PVSMCs was monitored by ELISA and foundto be produced in the range 10–130 ng/ml per day. TIMPexpression from transfected cells showed a peak ofexpression at 3 days after transfection, although expres-sion continued at a lower level for up to 14 days(Figure 4c).

Discussion

Cardiovascular disease is the major cause of mortality andmorbidity in the western world. Heart muscle disease(hereditary or acquired through ischaemia), atherosclero-tic vascular disease, and the restenosis that frequentlyfollows vascular interventions are all potential targets for

gene transfer. Somatic genetic manipulation may lead tofurther insights into the disease processes, but genetherapies are limited by the lack of suitably efficient andnon-toxic vector systems. While adenoviral vectors havebeen the main focus of research in cardiovascular disease,there remains a need for a synthetic, non-immunogenicvector with potential for targeting [17]. We aim todevelop such an efficient synthetic vector system for genetherapy of cardiovascular disease. As a first step towardsthat goal, we report on the in vitro optimization oftransfection of primary pig aortic endothelial cells and pigvascular smooth muscle cells. The LID vector transfectsprimary porcine vascular smooth muscle cells (PVSMCs)in vitro with over 40% transfection efficiency in and over35% transfection of endothelial cells. The toxicity of thevector was demonstrated to be low by the minimal effecton cellular proliferation and protein levels from trans-fected cells. The ability of cells to continue proliferating

Figure 3. (a) PVSMCs were transfected with pCILux (n), pCIh-TIMP (#), treated with OptiMEM for the same transfectionperiod (&), or were maintained in complete medium (2).Incubations were performed in the presence of tritiated thy-midine and then samples were analysed at 24 h intervals for3 days after transfection. (b) Transfections of PVSMCs wereperformed with the LID vector in serum with final propor-tions in the transfection medium of 2, 5, 10, 15, and 20%.All transfection complexes were formulated at a 7 : 1 chargeratio. Luciferase expression levels were determined 48 h aftertransfection

Figure 4. Production of bioactive hTIMP-1 by PVSMCs trans-fected with LID.hTIMP-1. Analysis of human TIMP-1 produc-tion by western blotting (a), reverse zymography (b), andELISA (c). (a, b) Concentrated, conditioned media wereequalized for protein content from (1) dexamethasone-treatedhuman fibrocarcinoma (HT1080) cells, (2) PVSMCs trans-fected with LID complexes containing pCIhTIMP-1, and (3)untransfected PVSMCs before analysis. (c) hTIMP-1 produc-tion measured by ELISA of pCIhTIMP-transfected cells at 1, 3,5, 7, and 14 days after transfection, compared with hTIMP-1standards. All transfection complexes were formulated at a7 : 1 charge ratio



after transfection will permit use of the LID system in theevaluation of anti-proliferative gene therapy strategies forcardiovascular disease [1].

The LID vector demonstrated receptor-specific enhan-cement of transfection in PAECs but not PVSMCs inexperiments where LID complexes were compared con-taining either an a5b1 integrin targeting peptide motifor a scrambled control. Transfection of PAECs with LIDwas enhanced by the integrin-targeted Peptide 6, with ahigher degree of specificity at lower charge ratios (1 : 1>3 : 1>7 : 1). The efficiency of transfection at differentcharge ratios was in inverse proportion to the specificitydemonstrated (7 : 1>3 : 1>1 : 1), so that the highesttransfection levels were achieved without integrin speci-ficity. This suggests that increasing charge density onthe surface of the LID complexes becomes increasinglyimportant for cell targeting, eventually swamping theintegrin-targeting properties. This interpretation is borneout by zeta potential difference measurements (unpub-lished data, manuscript in preparation). These resultswere consistent with transfections of human endothelialcells with the LID vector system [5]. The lack of receptorspecificity in PVSMCs was surprising, since both PVSMCsand PAECs were shown by flow cytometry to displaya5b1 integrin abundantly. It is possible that the cationic-binding properties of the two cell types differ, overridingintegrin binding events in PVSMCs. The overall efficiencyof the vector, therefore, is a consequence of both thecationic properties of the vector complex and its integrin-targeting properties. In most transfections performed inthis study, the overall charge ratio of components wasapproximately 7 : 1 (+/x), at which ratio the integrintargeting properties of the LID complex in PAECs werenot exploited. However, for in vivo applications, targetedtransfection would be desirable and the receptor-targeting properties of the LID complex may be improvedby using new peptides with higher cellular affinities,possibly selected from phage display libraries.

Transfection conditions were optimized with a viewto modelling in vivo conditions and requirements forshorter transfection periods. LID vector transfection effici-ency was unaffected by serum and expression levels forshorter, 30 min, incubations could be increased by in-creasing the DNA concentration. The consequences ofthese observations are relevant to transfection of cathe-terized blood vessels in vivo or isolated graft vesselsex vivo, where incubation times will need to be minimizedand expression levels maximized. For in vivo applications,where incubation times will be necessarily short, theexpected loss of transfection efficiency may be offset byincreasing the DNA concentration. In addition, the toler-ance of the vector towards exposure to serum proteinsaugurs well for the efficiency of gene transfer in theanticipated environment in vivo.

Genetic manipulation of matrix metabolism has beenidentified as a potentially useful therapeutic strategy forvascular disease. There is evidence from studies usingadenoviral-mediated human TIMP-1, -2 and -3 genetransfer [11,18] and from transplant of VSMCs secreting

human TIMP-1 [19] of modulation of responses to injurysuch as angioplasty or vein grafting. Cell migration,intimal hyperplasia, and remodelling of the vessel wallare all reduced by the increased expression of TIMPs. Inpreparation for evaluation of the LID vector in vivo, wehave transfected PVSMCs with a hTIMP-1 cDNA underoptimal conditions with the LID vector and demonstratedthe production of bioactive hTIMP-1 by western blottingand reverse zymography. Quantification of productionby ELISA over time revealed continuous secretion ofhTIMP-1 for at least 14 days at about 10 ng/ml. Thepeak of expression occurred 3 days post-transfectionat 130 ng/ml per 24 h. This value corresponds to19.5 ng/24 h per 104 cells, calculated to be equivalentto a multiplicity of infection of approximately 17 plaque-forming units per cell transduced by adenoviral vectorsencoding hTIMP-1 [20].In summary, we have performed in vitro transfections

of primary cultures of porcine vascular cells. The vectorhas displayed desirable properties, including high trans-fection efficiency, serum tolerance, and lack of toxicity.Shorter incubation times were sufficient for efficienttransfection, provided that DNA concentrations wereincreased. In addition, cells transfected with LID com-plexes containing TIMP-1 produced high levels of hTIMP-1. The LID vector system has potential therapeutic use ingene transfer to cardiovascular cells and we now aim toprogress to evaluation of this strategy for the preventionof restenosis in relevant model systems in vivo. It alsohas great potential as a tool for the study of the cellularbasis of cardiovascular disease involving gene transferexperiments.

Acknowledgements

We thank Ms Jo Buddle for technical assistance with flow

cytometry. This work was supported by the British Heart

Foundation (PG99049).

References

1. Baek S, March KL. Gene therapy for restenosis: getting nearerthe heart of the matter. Circ Res 1998; 82: 295–305.

2. Pickering JG, Jekanowski J, Weir L, et al. Liposome-mediatedgene transfer into human vascular smooth muscle cells.Circulation 1994; 89: 13–21.

3. Lawrie A, Brisken AF, Francis SE, et al. Ultrasound enhancesreporter gene expression after transfection of vascular cellsin vitro. Circulation 1999; 99: 2617–2620.

4. Laitinen M, Pakkanen T, Donetti E, et al. Gene transfer into thecarotid artery using an adventitial collar: comparison of theeffectiveness of the plasmid–liposome complexes, retroviruses,pseudotyped retroviruses, and adenoviruses. Hum Gene Ther1997; 8: 1645–1650.

5. Hart SL, Arancibia-Carcamo CV, Wolfert MA, et al. Lipid-mediated enhancement of transfection by a nonviral integrin-targeting vector. Hum Gene Ther 1998; 9: 575–585.

6. Hart SL, Collins L, Gustafsson K, Fabre JW. Integrin-mediatedtransfection with peptides containing arginine–glycine–asparticacid domains. Gene Ther 1997; 4: 1225–1230.

7. Hart SL, Harbottle RP, Cooper R, et al. Gene delivery andexpression mediated by an integrin-binding peptide. Gene Ther1995; 2: 552–554.

8. Hart SL, Knight AM, Harbottle RP, et al. Cell binding and



internalization by filamentous phage displaying a cyclic Arg–Gly–Asp-containing peptide. J Biol Chem 1994; 269: 12468–12474.

9. Itoh H, Nelson PR, Mureebe L, et al. The role of integrins insaphenous vein vascular smooth muscle cell migration. J VascSurg 1997; 25: 1061–1069.

10. George SJ, Johnson JL, Angelini GD, et al. Adenovirus-mediatedgene transfer of the human TIMP-1 gene inhibits smooth musclecell migration and neointimal formation in human saphenousvein. Hum Gene Ther 1998; 9: 867–877.

11. Dollery CM, Humphries SE, McClelland A, et al. Expression oftissue inhibitor of matrix metalloproteinases 1 by use of anadenoviral vector inhibits smooth muscle cell migration andreduces neointimal hyperplasia in the rat model of vascularballoon injury. Circulation 1999; 99: 3199–3205.

12. Pearson JD, Carleton JS, Hutchings A. Prostacyclin releasestimulated by thrombin or bradykinin in porcine endothelialcells cultured from aorta and umbilical vein. Thromb Res 1983;29: 115–124.

13. Ross R. The smooth muscle cell. II. Growth of smooth musclein culture and formation of elastin fibres. J Cell Biol 1971;50: 172–186.

14. Tanaka K, Iwamoto Y, Ito Y, et al. Cyclic AMP-regulatedsynthesis of the tissue inhibitors of metalloproteinasessuppresses the invasive potential of the human fibrosarcomacell line HT1080. Cancer Res 1995; 55: 2927–2935.

15. Webb KE, Henney AM, Anglin S, et al. Expression of matrixmetalloproteinases and their inhibitor TIMP-1 in the rat carotidartery after balloon injury. Arterioscler Thromb Vasc Biol 1997;17: 1837–1844.

16. Koivunen E, Wang B, Ruoslahti E. Isolation of a highly specificligand for the a5b1 integrin from a phage display library. J CellBiol 1994; 124: 373–380.

17. O’SullivanM, Bennett MR. Gene therapy for coronary restenosis:is the enthusiasm justified? Heart 2001; 86: 491–493.

18. Baker AH, Zaltsman AB, George SJ, Newby AC. Divergent effectsof tissue inhibitor of metalloproteinase-1, -2, or -3 overexpres-sion on rat vascular smooth muscle cell invasion, proliferation,and death in vitro. TIMP-3 promotes apoptosis. J Clin Invest1998; 101: 1478–1487.

19. Forough R, Koyama N, Hasenstab D, et al. Overexpression oftissue inhibitor of matrix metalloproteinase-1 inhibits vascularsmooth muscle cell functions in vitro and in vivo. Circ Res 1996;79: 812–820.

20. Baker AH, Wilkinson GW, Hembry RM, et al. Developmentof recombinant adenoviruses that drive high level expressionof the human metalloproteinase-9 and tissue inhibitor ofmetalloproteinase-1 and -2 genes: characterization of theirinfection into rabbit smooth muscle cells and human MCF-7adenocarcinoma cells. Matrix Biol 1996; 15: 383–395.



Documents

High efficiency transfection of porcine vascular cells in vitro with a synthetic vector system