ORIGINAL ARTICLE

Fatty acid–spermine conjugates as DNA carriersfor nonviral in vivo gene delivery

JR Viola1, H Leijonmarck2, OE Simonson1, II Oprea1, R Frithiof3, P Purhonen2, PMD Moreno1,

KE Lundin1, R Stromberg2 and CIE Smith1

1Clinical Research Center, Department of Laboratory Medicine, Karolinska Institutet, Karolinska University Hospital Huddinge,Stockholm, Sweden; 2Department of Biosciences and Nutrition, Karolinska Institutet, Stockholm, Sweden and 3Department of Physiologyand Pharmacology, Karolinska Institutet, Stockholm, Sweden

The lack of efficient in vivo gene delivery is a well-knownshortcoming of nonviral delivery vectors, in particular ofchemical vectors. We developed a series of novel nonviralcarriers for plasmid-based in vivo gene delivery. This newtransport device is based on the assembly of DNA plasmidswith synthetic derivatives of naturally occurring molecules—fatty acid–spermine conjugates (or lipospermines). We testedthe ability of these fatty acid conjugates to interact withplasmid DNA (pDNA) and found that they formed DNAnanocomplexes, which are protected from DNase I degrada-tion. This protection was shown to directly correlate with thelength of the aliphatic component. However, this increase inthe length of the hydrocarbon chain resulted in increasedtoxicity. The cationic lipids used for transfection typically havea C16 and C18 hydrocarbon chain. Interestingly, toxicity

studies, together with further characterization studies, sug-gested that the two most suitable candidates for in vivodelivery are those with the shortest hydrocarbon chain,butanoyl- and decanoylspermine. Morphological characteriza-tion of DNA nanocomplexes resulting from these liposper-mines showed the formation of a homogenous population,with the diameter ranging approximately from 40 to 200 nm.Butanoylspermine was found to be the most promising carrierfrom this series, resulting in a significantly increased geneexpression, in relation to naked plasmid, in both tissues hereintargeted (dermis and M. tibialis anterior). Thus, we establisheda correlation between the in vitro properties of the ensuingDNA nanocarriers and their efficient in vivo gene expression.Gene Therapy advance online publication, 3 September 2009;doi:10.1038/gt.2009.108

Keywords: non-viral gene delivery; lipospermines; in vivo gene delivery

Introduction

Nonviral gene delivery systems are well known not onlyfor the biosafety they offer over viral carriers but also forthe fact that their efficiency is much at the rear comparedwith that of the latter. Physical methods have beenshown to be highly efficient in targeting several tissues,especially skeletal muscle or cancer tissues.1–4 Yet, all theelectrical equipments used in this method present someinconveniences. Chemical vectors are, therefore, analternative and are the focus of this work. Comparedwith physical methods, chemical systems have lowefficiency and, in spite of their safety, they frequentlyshow toxic properties because of the high concentrationsof carriers that are needed to obtain significant geneexpression.5–8 Thus, highly efficient nonviral gene

carriers are needed. For this purpose, DNA condensationis a recurrent requisite for this type of nonviraltherapeutic approach.5,8–12

Bloomfield13 has defined DNA condensation as thecollapse of extended DNA chains into compact, orderlyparticles containing only one or a few molecules. Typically,this process results from a neutralization of the negativecharges of the DNA phosphate groups with transition intothe ordered phase that occurs when 90% of charges areneutralized.13 Different systems have been proposed andtested for nonviral gene delivery based on DNA condensa-tion and protection.12,14,15 However, the drawback of thistherapeutic approach is often the lack of a successful in vivoapplication. There is clearly a need to develop highlyefficient, chemical, nonviral gene carriers that can protectDNA and persist under physiological in vivo conditions.

Taking this into consideration, polyamines wouldseem excellent candidates as one of the roles of thesenaturally occurring molecules is related to DNA con-densation.16 Nevertheless, and despite the fact that thespatial molecular arrangement in between their positivecharges fits perfectly with that in between phosphates inthe DNA molecule, suggesting a perfect interaction fit,spermine was shown to bind poorly to DNA, resulting inno DNA protection against nuclease degradation.17

DNA binding by the condensing agent can beimproved either by increasing the charge density of the

Received 19 November 2008; revised 25 July 2009; accepted 27 July2009

Correspondence: JR Viola or Professor CIE Smith, Clinical ResearchCenter, Department of Laboratory Medicine, Karolinska Institutet,Karolinska University Hospital Huddinge, KI at Novun, Plan 5, Lab514, Hosolvagen 7, 141 86 Huddinge, Stockholm, Sweden or ProfessorR Stromberg, Department of Biosciences and Nutrition, KarolinskaInstitutet, Novum, Halsovagen 7, 141 57 Huddinge, Stockholm,Sweden.E-mails: Joana.Viola@ki.se or Roger.Stromberg@ki.se, orE-mail: Edvard.Smith@ki.se

Gene Therapy (2009), 1–12& 2009 Macmillan Publishers Limited All rights reserved 0969-7128/09 $32.00

www.nature.com/gt

positively charged molecule or by introducing hydro-phobic contributions.18,19 Thus, the attachment of a shortlipidic tail to spermine could potentially improve thebinding of this molecule to DNA and thereby affect itsability to protect DNA on condensation. Geall andBlagbrough19 showed the binding affinity betweenplasmid DNA (pDNA) and oleic and palmitic liposper-mines. Yet, to our knowledge, a good characterization ofthese conjugates for delivery purposes is still missing.Oleic and palmitic lipospermines are two quite similarderivatives, and a good characterization of spermineconjugates with a more varied hydrocarbon chain shouldprovide larger differences in properties that could bebalanced or tuned for delivery purposes. Moreover, thereis frequently a lack of correlation with regard to thein vitro characterization properties of delivery vectorsand their in vivo expression efficiency.

Therefore, in this study, we introduce a series of sixamphiphilic molecules—fatty acid–spermine conjugates(lipospermines, Figure 1). Endogenous or naturallyoccurring lipids were chosen because no major toxicityor inflammation was expected to result from suchformulations or from their potential degradation pro-ducts. Carbon chains of different lengths (between 4 and18 carbons) and with different saturation levels (between0 and 3) were attached to spermine. Herein, we describeself-assembling lipospermine/DNA particles and intro-duce them for gene therapy purposes. The characteriza-tion of DNA nanocomplexes included physicochemicaland morphological studies. Our results show that,depending on the (+/�) (ammonium to phosphate)ratio, all compounds induce DNA condensation andthese nanoparticles can reach an average diameter

ranging from 40 to 200 nm. Among the differentlipospermines that were analyzed for their in vitroproperties, the results suggested decanoyl- and buta-noylspermine as the most suitable candidates to beevaluated as in vivo gene delivery vectors. Such DNAparticles protected DNA from DNase I degradation andresulted in the most homogenous and stable populationof particles. Most importantly, these studies allowed forthe selection of the most promising conjugate. Comparedwith pDNA, butanoylspermine significantly increasedgene expression, both in the muscle and dermis, inducinga higher in vivo gene expression by approximately10-fold.

Results

Toxicity of fatty acid–spermine conjugates increaseswith the length of the lipidic partFatty acid–spermine conjugates (Figure 1) were synthe-sized by reacting the partially protonated polyaminewith an acid chloride or a fatty acid activated byisobutylchloroformate. This is a new simplified proce-dure compared with that previously used for oleyl andpalmitoyl derivatives.19,20 Given that most of the studiedlipospermines are new, or have not been analyzed for thepurpose of in vivo gene delivery, we found it relevant tofirst evaluate their cytotoxic properties. The toxicologicalevaluation of lipospermines and lipospermine/DNAcomplexes was assessed using an in vitro cytotoxicityassay on the basis of the activity of mitochondrialdehydrogenases. Cell proliferation is directly correlatedto the amount of formazan dye formed, resulting from

N1-butanoylspermine

N1-decanoylspermine

N1-linolylspermine

N1-linolenoylspermine

spermine

N1-acetylspermine

N1-palmitoylspermine

N1-oleylspermine

NH

NH

NH

NH

NH

NH

O

O

O

NH

O

O

O

+H3N NH2+

+H3N

+H3N

+H3N

+H3N

+H3N

+H3N

+H3N

NH2+

NH2+

NH2+

NH2+

NH2+

NH2+

NH2+

H2+N

H2+N

H2+N

H2+N

H2+N

H2+N

H2+N

H2+N

O

NH3+

Figure 1 Fatty acid conjugates used in this study.

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

2

Gene Therapy

the cleavage of tetrazolium salts using cellular enzymes.A visual inspection of the cells correlated well with theabsorbance reading. As expected, toxicity was shown tobe concentration dependent and related to the length anddegree of saturation of the fatty acid in this class ofmolecules (Figure 2). In general, toxicity resulting frombutanoyl- (4 carbons), decanoyl- (10 carbons), linolyl- (18carbons, 2C¼C) and linolenoylspermines (18 carbons,3C¼C) was low or nonexistent. For the last twocompounds, the cells tolerated concentrations up to80 mM with a low occurrence of cell death (Figure 2a). Incontrast, oleylspermine (containing 18 carbons and oneC¼C bond) and palmitoylspermine (16 carbons) showedtoxicity at concentrations above 10–20 mM (Figure 2a). Forseveral of the tested compounds, formulation withpDNA resulted in a higher cytotoxicity when comparedwith that of lipospermines alone (Figure 2b). Whenformulated, oleyl-, palmitoyl-, decanoyl- and linolylsper-mines showed toxicity at a lower concentration thanwithout pDNA. However, oleyl- and palmitoylspermineremained the most toxic compounds. The cytotoxicity ofthe different fatty acid–spermine conjugates followed thesame tendency in a different cell line—baby hamsterkidney (see Supplementary data, Supplementary FigureS1). From in vitro assays, it seemed that most of the newconjugates were worth exploring in an in vivo setting.

Interaction of DNA with lipospermines is influencedby the length of the fatty acid chainWe next studied the ability of fatty acid–spermineconjugates to bind to DNA using a gel retardation assay.DNA complexes were formulated at different charge

ratios (NH3+/PO4

�) and were analyzed using agarose gelelectrophoresis. An interaction between conjugates andpDNA is seen here as a retardation of the pDNAmigration. This retardation is typical for the formationof larger particles as compared with pDNA alone. Alllipospermine conjugates were able to interfere with DNAmigration in agarose gel, confirming an interactionbetween them and nucleic acid (Figure 3). DNAretardation decreased with decreasing length of the lipidtail. This effect became more evident as the concentrationof lipospermine molecules increased. As a consequence,neutralization of DNA charge and the formation ofheavier DNA particles occurred, decreasing the DNAmobility in the gel until it eventually became retained inthe wells. At very high lipospermine concentrations, dueto a high condensation state, DNA is no longer availablefor intercalation by ethidium bromide and was thus notdetected in the gel (that is, lanes 5–8 in Figure 3c). TheDNase I protection assay shows that linolenoyl-, linolyl-and palmitoylspermines (containing 18 and 16 carbons)are able to protect DNA when present in relatively lowconcentrations (corresponding to 0.3:1 (+/�) ratio)compared with decanoyl- and butanoylspermines, inwhich at least double concentration is needed to observea similar DNA protection (Figure 4). Nevertheless, alllipid conjugates were able to protect DNA fromdegradation, depending on their concentration.

Different saturation levels are shown to influenceDNA formulation.21 Comparing all the unsaturatedderivatives, DNA retardation is observed at lowerconcentrations with complexes formulated with linolyl-spermine. The DNase I protection assay, on the otherhand, is not very conclusive, as bands do not seem tofollow any trend. Particularly intriguing and inconclu-sive is the lack of stronger bands at higher ratios. Thisfact is likely to be an artifact, owing to the inability ofDNA complexes to dissociate (see Supplementary data,Supplementary Figure S2), a phenomenon reportedearlier.22 It is also likely that the pDNA purification step(through a column) used in this assay contributed to lossof material. Thus, it is very possible that this procedureimpacted on the pDNA band intensity found in theagarose gel in the DNase I protection assay. Interestingly,after the initial studies, we found that repeated handlingand exposure to air of linolyl and linolenoylspermineconjugates caused oxidation of these derivatives.Although this does not necessarily disqualify them fromuse, we chose to exclude these compounds from furthercharacterization for now.

The population of DNA nanocomplexes formedby shorter spermine conjugates shows a narrower sizedistribution and a smaller average diameterTo better characterize the complex formed betweenlipospermine and pDNA, different lipospermine/DNApreparations were analyzed using dynamic light scatter-ing (DLS). Initially, 0.6:1, 1:1 and 2:1 (+/�) charge ratioswere chosen, because we found it to be of particularinterest to characterize DNA complexes that wereexpected to differ significantly in the theoretical totalnet charge. In addition, if possible, we intended toestablish a correlation between these theoretical condi-tions and the resulting particles. In all clear andtransparent preparations, intensity values resulting from

Figure 2 In vitro evaluation of the toxicity of lipospermines. Thecytotoxic properties of lipid-spermines were assessed using WST-8assay for in vitro conditions with HepG2 cells. Concentrations of thelipid-spermine that were chosen for this assay are the same as in thedifferent charge ratios ranging from 0.3:1 to 3:1. Absorbance wasmeasured at 450 l and sample values were divided by those ofuntreated cells. The figure shows the impact of different lipid-spermines, (a) and lipid-spermine/DNA complexes (b) in theproliferation of HepG2 cells. Each point represents a mean±s.d. ofthree independent experiments.

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

3

Gene Therapy

light scattered by particles with diameters larger than1 mm were neglected, as this light scatter most probablyresults from dust particles.18,23 Most of the formulationspresented a relatively broad population distribution,except for decanoylspermine/DNA 1:1 (Figure 5). Anincrease in the concentration of butanoylspermine in theformulation of DNA complexes resulted in the formation

of more stable colloidal solutions, as evidenced bya decrease in the average particle diameter, together withan increase in light intensity (Figure 5). No comparabletendency or particular trend could be observed in theDNA preparations with oleyl- and palmitoylspermine,and both resulted in formulations with a high hetero-geneity and polidispersity index. However, unlikeoleylspermine, palmitoylspermine/DNA formulated ata 2:1 charge ratio resulted in a turbid solution, composedof large aggregates (41 mm). Interestingly, the quality ofdecanoylspermine preparations varied significantlywithin a narrow concentration window. At the 0.6:1ratio, the population distribution was broad, with anaverage diameter of approximately 200 nm, whereasdecanoylspermine/DNA formulated at a 1:1 ratio re-sulted in a clearly improved homogeneity and thepopulation distribution with the smallest average dia-meter (80 nm) and with the highest stability over time.An increase up to the concentration present at 2:1 (+/�)ratio, however, induced colloidal instability, resulting inthe precipitation of particles (data not shown). Bycomparing all four fatty acid conjugates/DNA formula-tions, at charge ratios of 0.6:1 and 1:1, it is possible to seethat the populations with a lower average size are thoseformed by conjugates with shorter fatty acids. Thebehavior of butanoyl- and decanoylspermine formula-tions could be because of the fact that the relatively shorthydrocarbon chains result in particle–particle interac-tions that are weaker. Thus, initially formed DNAcomplexes would be less prone to aggregate into largerparticles through hydrophobic interactions.

Apart from size, morphology is also an interestingproperty for a drug carrier system and is likely toinfluence the efficiency of drug uptake. To bettercharacterize this type of nanoparticles, morphologicalstudies of DNA complexes resulting from condensationinduced by butanoyl- and decanoylspermines werecarried out, either by negative staining or by cryo-electron microscopy (cryo-EM) (Figure 6). The choice ofshorter fatty acid derivatives is based on their lowercytotoxicity and on DLS studies, which showed thatthose formulations were the most homogenous andcomposed of particles with the lowest average diameter.

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8

1 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8

Figure 3 Gel electrophoresis assay. DNA complexes were formulated at different charge ratios in 5.45% mannitol and analyzed usingagarose gel for DNA mobility. (a) N1-linolenoylspermine (b) N1-linolylspermine (c) N1-oleylspermine (d) N1-palmitoylspermine (e) N1-decanoylspermine (f) N1-butanoylspermine. Lane 1—plasmid DNA. Lanes 2–8 correspond to different charge ratios: 0.3:1, 0.6:1, 1:1, 2:1, 3:1,5:1 and 10:1, respectively. The figures are representatives of two independent experiments.

Figure 4 DNase I protection assay. DNA complexes wereformulated at different charge ratios in 5.45% mannitol andincubated at room temperature for 30 min before digestion withDNase I for 15 min. After purification, DNA was analyzed usingagarose gel for DNA mobility and protection. Controls were run inparallel in each experiment. 1—plasmid DNA (pDNA); 2—pDNAdigested with DNase I; 3—pDNA purified. Lanes 1–3 arerepresentative figures. (a) N1-linolenoylspermine (b) N1-linolylsper-mine (c) N1-oleylspermine (d) N1-palmitoylspermine (e) N1-decanoylspermine (f) N1-butanoylspermine. Lanes 4–10 correspondto different charge ratios: 0.3:1, 0.6:1, 1:1, 2:1, 3:1, 5:1 and 10:1,respectively. The figures are representatives of two independentexperiments.

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

4

Gene Therapy

The average diameter of particles, as shown by themicroscopic techniques used here, was in general smallerthan that measured by DLS. This has been described

before in atomic force microscopy studies, and someexplanations have been suggested. Among them is thechange in the three-dimensional configuration of the

Figure 5 Dynamic light scattering studies. Representative curves of the intensity distribution of different DNA formulations. Samples werefreshly prepared and left at room temperature for 10 min before each measurement. The curves are representative of three independentexperiments.

Figure 6 Electron microscopy studies—morphological characterization of decanoylspermine/DNA complexes was carried out using cryo-electron microscopy. This showed the formation of single nanoparticles with an average diameter of approximately 30 nm for (a)decanoylspermine/DNA 1:1 charge ratio and also the (b) presence of some aggregates of approximately 100 nm diameter. Black arrows showan agglomerate of DNA molecules. The DNA complexes resulting from the interaction of DNA with butanoylspermine were characterizedusing negative staining for 2:1 charge ratio and cryo-electron microscopy for 1:1 charge ratio. Figures (c) and (d) show particles that wereformed at charge ratios of 2:1 and 1:1, respectively. The figures are representative of two independent experiments. Bar: 50 nm.

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

5

Gene Therapy

sample because of interaction with a solid phase,imaging in a dried state or, in the case of cryoEM, thethickness of the vitrified ice layer with embeddedsample.24–26 The imaging of DNA complexes studiedhere was not possible using atomic force microscopy, asit resulted in a complete aggregation of the samplestested (data not shown). Alle et al.27 have reported asimilar phenomenon, and depending on the order ofaddition of the DNA and condensing agent (protamine,in this case) to the mica surface, the shape and size of theresulting particles vary. For decanoylspermine/DNAcomplexes formulated at 1:1 ratio, cryo-EM revealedthe formation of relatively small particles, with adiameter of roughly 40 nm (Figure 6a), which, byaggregation, formed larger particles (Figure 6b). Electronmicroscopy of butanoylspermine/DNA nanocomplexesformed at the charge ratio (+/�) of 2:1 did not show theformation of well-defined particles (Figure 6c). At thecharge ratio of 1:1, the average diameter seen isextremely small, of approximately 15–20 nm (Figure6d). The discrepancy with DLS measurements (averagediameter of 200 nm) is quite substantial and we cannotexclude that the negative staining procedure used tovisualize these complexes could cause a collapse ofinitially formed larger particles. In such a case, thesestructures most likely consist of pure core particles,whereas pDNA partially exists in a free and unboundform in the surrounding medium.

Butanoylspermine formulations result in a significantlyincreased gene expression on in vivo gene deliveryTo address and characterize the in vivo gene expressionof lipospermine/DNA complexes, we injected DNAnanocondensates locally in the M. tibialis anterior and inthe dermis of mice. Gene expression was observed for 22days by imaging the action of the luciferase. Differentlipospermine/pDNA formulations to be used in thisassay were ranked by parameters, such as their low or nodetectable cytotoxicity, as well as the homogeneity andstability of the DNA formulation. Therefore, butanoyl-and decanoylspermine were chosen as pDNA carriers.For initial screening, all three ratios (0.6:1, 1:1 and 2:1)were considered and their ensuing gene expression wasobserved for 8 days (data not shown). The mostpromising formulations were thereafter chosen andfurther analyzed and are reported here. Gene expressionfollowed consistent patterns in both muscle and dermis,although it was more persistent in the muscle (Figure 7,compare a with b). Butanoylspermine was formulatedwith pDNA at two different charge ratios, both resultingin a significant increase in gene expression comparedwith naked pDNA, depending on the targeted organ(Figures 7a and b). Although DNA complexed withbutanoylspermine at the ratio of 0.6:1 seemed to be moresuitable for gene delivery into the muscle (Figure 7a),butanoylspermine/DNA at the ratio of 1:1 significantlyincreased gene expression in skin (Figures 7a and b).

Figure 7 In vivo studies—bioluminescence. Fatty acid–spermine/DNA nanocomplexes were freshly prepared and injected intradermallyand intramuscularly in mice. Gene expression was assessed using imaging of the firefly luciferase. The figure shows luciferase geneexpression resulting from butanoylspermine/plasmid DNA. Luciferase expression was followed over time (a and b). Luminescence units areshown here as an average of photon counts per second. Each point represents a mean±s.d. of at least seven samples. Statistical analyses wereperformed in relation to pDNA. M. tibialis anterior: butanoylspermine/DNA 0.6:1 P¼ 0.02, butanoylspermine/DNA 1:1 P¼ 0.09; dermis:butanoylspermine/DNA 0.6:1 P¼ 0.95, butanoylspermine/DNA 1:1: P¼ 0.0004. (c) Luminescence intensity reading: comparison betweenintradermal expression of pDNA and different butanoylspermine/DNA formulations. Mice were intradermally injected with pDNA orbutanoylspermine/DNA complexes formulated at 0.6:1 and 1:1 N/P , or (+/�), ratios. D-Luciferin was injected intraperitoneally andluminescence was measured after 5 min. During the procedure, the mice were kept under anesthesia. Here, 1, 2 and 3 represent threeindependent experiments.

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

6

Gene Therapy

The DNA particles formed by decanoylspermine atthe ratio of 1:1 were only able to induce a higher geneexpression than the naked plasmid in six out of eightobservation points. Despite the occurrence of an increase,this was not statistically significant (data not shown).With regard to the possible toxicity of the used dosage,histological analyses of muscles (injected with decanoyl-spermine/DNA nanoparticles) were conducted. Ingeneral, if observed at all, pathological changes wereconsidered as mild and no severe pathology wasobserved in any of the muscle sections (data not shown).No difference could be seen between the musclesinjected with the buffer solution and treated muscles,thereby suggesting that the small changes that were seencould be due to the injection itself. Thus, this newlydeveloped system emerges as a promising, safe tool forin vivo gene therapy.

Discussion

Relatively few reports correlate in vitro studies of newcompounds for gene delivery with in vivo gene expres-sion data. In this study, we developed a series of fattyacid–spermine conjugates and showed that one of theselected derivatives enhances in vivo gene delivery.Furthermore, we analyzed the in vitro properties offormulated DNA containing particles, and in vivotransfection efficiency. Polyamines have been extensivelyused for the development of nonviral vectors and thisincludes studies in which lipospermines typically eithercontain double or have longer hydrocarbon chains.18,21,28–31

It has previously been suggested that DNA-bindingaffinity is a function of both charge and hydrophobi-city.19,32 However, to what degree this affinity isbeneficial for gene delivery is not clear. Some degree ofaffinity is, certainly, a prerequisite, but it is likely that thishas to be balanced towards other properties of thedelivery agents and the particles formed on theirinteraction with DNA.

Here, we tried to evaluate this balance by conducting acomparative study with well-defined and well-charac-terized lipospermines that differ in the length andsaturation of the alkyl chain. We showed that an increasein the length of the hydrocarbon chain resulted in anenhanced gel retardation of DNA, which could indicatehigher affinity. However, this is more complex, as thenewly formed particles will become larger as themolecular weight of spermine conjugates increases.Results from the DNase I protection assay suggest thatthe protection of DNA, at equal concentrations, wasmore prominent with lipospermines having longerhydrocarbon chains and that the degree of saturationaffects these properties. In fact, Abbasi et al.22 see asimilar trend in the studies that they conducted withlipid-substituted poly-L-lysine , in which they claim thatDNA complex stability increases with the ratio of –CH2–groups per poly-L-lysine molecule. One can also spec-ulate about the contribution of hydrophobic interactionsto the formulation of DNA nanoparticles on the basis ofz-potential studies (see Supplementary Figure S2).Regardless of the fact that all lipospermine conjugateshave the same formal charge (+3), characterization of thez-potential of pDNA complexes showed a fairly largerange of potentials for the different nanocomplexes

formulated at the same (+/�) charge ratio. The variationfound could definitely be attributed to the differences inthe carbon chain for butanoyl- and decanoylspermines,with the longer chain giving a more negative potential.Although this trend holds for oleyl and palmitoylderivatives, it is not excluded that this particular effectis also a consequence of the size of the resulting particles,as these were larger than those formed by butanoyl- anddecanoylspermine.

An increase in the length of the hydrocarbon chainwas accompanied by an increase in toxicity. This is notunexpected, because a longer aliphatic chain grants themolecules more detergent-like properties and could,thus, increase cell membrane disruption and inducesubsequent cell death. On the other hand, inclusion oflipid moieties in the gene carrier could be advantageousfor facilitating cell entry. Therefore, a balance betweenuptake properties and toxicity must be found. In light ofthe toxicity and DLS studies, we suggest the existence ofan optimal length of lipidic moiety among the studiedlipospermines that generates the most stable and efficientDNA formulation for the proposed purpose of in vivogene delivery. A balance between hydrophobic andhydrophilic contribution has been shown to be criticalfor the functioning of many amphiphilic compounds andas a determining parameter for their transfectionefficiency, for example, for the pluronics blockcopolymers.33 In this study, we establish a parallelbetween the importance of this equilibrium and that ofthe symmetry between the contributions to DNA bindingby the two different components of lipospermines. Thus,interaction of butanoylspermine with pDNA is close toa pure cationic interaction. This can be seen in thex-potential of butanoylspermine, which presents valuesclose to neutralization with no significant changes withinthe range of studied concentrations (see SupplementaryFigure S3). Also, the behavior expressed by butanoyl-spermine/DNA nanoparticles in DLS studies is typicalof DNA complexes formed by cationic peptides, that is,the improvement in the quality of DNA formulationwithin a short range of charge ratios tested (from 0.6 to2:1), as a consequence of the increase in concentration.

Although not clearly verified, a more sizable lipidicportion could potentially increase DNA-binding affinity,and it certainly should increase hydrophobic interactionbetween initially formed DNA nanocomplexes. Decan-oylspermine-derived particles showed a concentration-dependent behavior, suggesting a more balancedcontribution from both charged and hydrophobic com-ponents toward the properties of nanoparticles. Theconcentration range used in this study is very narrow(between 20 and 60 mM). Therefore, the concentrationwindow to tune what could be the most optimal DNAnanoparticle formulation that ensues from this series isequally narrow. Interestingly, this fine-tuning also meansthat the concentrations that are required are low. Thus,this avoids possible toxicity hurdles when used in vivo.In fact, butanoyl- and decanoylspermine were among theconjugates that showed the lowest in vitro cytotoxicity.More importantly, the doses used in vivo did not induceany histological changes per se, underlining the safety ofthis system and supporting its further development anduse in gene therapy. However, considering the nonsigni-ficant relative increase in gene expression of decanoyl-spermine/DNA nanoparticles, the study of fatty acid

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

7

Gene Therapy

conjugates with lipid tails, sized between 4 and 10carbons, could potentially improve the design of thistype of carriers.

Morphological studies showed that DNA nano-particles resulting from the interaction betweendecanoyl- and butanoylspermines form a relativelyhomogenous population with an average diameterranging approximately from 40 to 200 nm. Previousstudies with a chemically different set of sperminederivatives used for DNA condensation have shownthe formation of complexes, but not as well defined orwith as regular a shape as some of the DNA particlesshown in this investigation, particularly with decanoyl-spermine/DNA.18,34 Influence of imaging techniques onmorphological characterization of DNA complexes doesnot seem to be uncommon and cryo-EM emerges as thatwhich may interfere the least. 24–27,35 The morphology ofthese lipospermines is likely to influence their in vivotransfection efficiency, but the correlation is not clear.Moreover, the conditions in which the microscopictechniques were conducted are not physiological. Never-theless, from the electron micrographs, the pDNAcomplexes emerge as spherical nanoparticles. It is notprobable that these particles correspond to micelles, as itis unlikely that butanoyl- and decanoylspermine canform micelles. Figure 6b suggests the existence of DNAmolecules on the surface of these particles. This could,then, possibly explain the aggregation phenomenon: theformation of larger particles as a result of common DNAmolecules that interact with the surface of severalindividual nanoparticles. The structure of liposper-mines/DNA nanoparticles is very much dependent onthe nature of the fatty acid. Hence, the morphologiesherein described do not, necessarily, apply to DNAcomplexes formed by other lipospermines.

One should emphasize that few delivery systems havebeen successful in vivo without the use or introduction ofa polymer, such as poly-ethylene-glycol or others thatsuit the same purpose of steric hindrance.36–38 The mostsuccessful cases occur after intranasal or intratrachealadministration, when targeting airway epithelialcells.3,5,9,39 Tissues containing nondividing cells (as inmuscle) or slow mitotic cells (as in the liver) are veryattractive targets because pDNA can remain for a longtime inside the cells.40 In relation to the relatively fastkinetics seen here, we do not exclude the possibility thatthe decrease in expression after 5–7 days may beindicative of toxicity. However, if this is the case, thistoxicity should be associated with pDNA, or with itsgenomic information—such as the exogenous reporterprotein—as pDNA alone results in a similar expressionprofile to that of the formulations, with a paralleldecrease. Either way, this ‘early decrease’ in expressionhas been previously reported. Chang et al., althoughtargeting M. tibialis anterior in rat, showed a similardecrease between days 5 and 10, whereas Brooks et al.41

noticed the decrease in mouse M. tibialis anterior afterday.7,39 Furthermore, Brooks et al. saw a dose-dependentsilencing of the expression of the transgene: the higherthe initial gene expression the quicker it was silenced.41

In the study by Chang et al.,38 pDNA was formulatedwith the block copolymer, PEG-PLGA-PEG.The resultingpDNA complexes induced a significant increase inluciferase expression by threefold, as compared withnaked DNA. On the contrary, pDNA complexed with

these short lipospermines increased luciferase activity bya factor of 10 over that of naked pDNA. These results arecomparable with some of the research carried out withpluronics block copolymers, which have been, andcontinue to be, extensively explored.34,42,43 Similar tothese, as well as to other vehicles, in vitro transfectionactivity of lipospermines was poor (data not shown).38,42

Delivery systems that are efficient in the skeletalmuscle or skin may be used for the purpose of DNAvaccine development. In addition, in the muscle, theycan also be important for the treatment of disorderscaused by the lack of an active protein, which has asystemic effect and can be produced in the muscle andthereafter secreted; for example, the Anderson–Fabrydisease. Lavigne et al.44 have designed a chimeric vectorto target skeletal muscle. They developed a devicecombining the nuclear targeting activity of TAT peptide,together with the specific interactions between themultisubunit DNA-binding protein (M2S) and theirtherapeutic plasmid (encoding for a-galatosidase Aprotein, AGA). With this vector, they increased proteinactivity up to eightfold as compared with unformulatedplasmid. However, the dose of DNA used was twice asmuch as that used in this study. In view of the fact thatpDNA alone has been used for intramuscular genedelivery and has proven to be very efficient, in atherapeutic setup, a 10-fold enhancement of the proteinexpression, similar to what is seen with butanoylsper-mine/DNA formulations, could mean a substantiallyreduced administration dose.40 Moreover, because theselipospermines are readily varied, a further tweaking ofproperties can be investigated. Optimizing particleproperties through mixing of different liposperminesmay also be a possible way of improving these systems.Moreover, by varying the composition and formulationof DNA nanoparticles, it may be possible to favordelivery into different tissues, similar to what wasshown here for butanoylspermine/DNA particlesformed at the 0.6:1 ratio. Additional ratios may also beexplored.

In conclusion, we synthesized a series of fatty acidspermines for potential use in gene delivery. Theproperties of their complexes with pDNA have beencharacterized and there is a clear effect from the fattyacid chain on toxicity as well as on the stability, size andhomogeneity of DNA nanocomplexes. Furthermore, inthis setup, the fatty acid conjugate with the shortest chain(four carbons long) is the most optimal carrier, inducingthe formation of noncytotoxic, homogenous and well-defined DNA/lipospermine nanoparticles, capable ofefficient gene delivery in vivo. Thus, the results of thisstudy present a novel platform based on fattyacid–spermine conjugates that can be readily tunedfurther and optimized for gene delivery purposes.

Materials and methods

Synthesis of fatty acid–spermine conjugatesMaterials. All reagents for synthesis were of commer-cial grade (Sigma Aldrich and Fluka brands werepurchased from Sigma Aldrich, Sweden) and were usedas received; solvents were of p.a. grade. Thin layerchromatography analysis was carried out using pre-coated Silica Gel 60 F254 (Merck product, purchased from

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

8

Gene Therapy

VWR, Sweden), with detection by ultraviolet light.Nuclear magnetic resonance (NMR) spectra wererecorded on a Bruker AVANCE DRX-400 instrument(Bruker Biospin AG, Zurich, Switzerland) (400.13 MHzfor 1H, 162.00 MHz for 31P and 100.62 MHz for 13C). Massspectra (time-of-flight mass spectrometry (TOF-MS) andelectrospray (ES)) were obtained using a MicromassLCT ES-TOF instrument and the MAXENT program(Micromass Ltd, UK) for calculation of masses frommultiple charged ions.

N1-butanoylspermine trihydrochloride salt. Butanoicanhydride (1.64 ml, 10.0 mmol) in dichloromethane(100 ml) was added dropwise for 25 min to a solutionof spermine (6.07 g, 30.0 mmol) and trifluoroacetic acid(TFA; 6.93 ml, 90.0 mmol) in methanol–dichloromethane(5:1, 600 ml) at 0 1C. The ice bath was removed andstirring continued at ambient temperature overnight. Themixture was concentrated under reduced pressure andpartitioned between 1 M NaOH ((aq.), 500 ml) andchloroform (CHCl3)–methanol (9:1, 500 ml). The aqueouslayer was separated and back extracted twice withCHCl3–methanol (9:1, 500 ml). The combined organicextracts were washed with 600 ml 1 M aqueous NaOH(containing 120 g NaCl) and then with 700 ml methanol–1 M NaOH aq. (1:6), containing120 g NaCl. The organiclayer was dried with Na2SO4 and concentrated in vacuo,yielding 2.15 g of crude amine. Crystallization as thetrihydrochloride salt from methanol yielded 2.08 g (55%)of product. 1H NMR (D3COD), 0.94 (t; 3H, J¼ 7.4 Hz;CH3), 1.64 (sextet, 2H, J¼ 7.4 Hz, CH3CH2CH2CO), 1.76–1.87 (m; 4H, 2�CH2), 1.91 (p, 2H, J¼ 7.0 Hz; CH2), 2.12(p, 2H, J¼ 7.8 Hz; CH2), 2.22 (t, 2H, J¼ 7.4 Hz; CH2CO),2.97–3.19 (m, 10H, 5�CH2N), 3.30 (t, 2H, J¼ 6.7 Hz;CH2N). 13C NMR (D2O), 15.5 (CH3), 21.6 (CH2), 25.5(CH2), 25.6 (CH2), 26.7 (CH2), 28.4 (CH2), 38.7 (CH2), 39.5(CH2), 40.4 (CH2), 47.3 (CH2), 47.8 (CH2), 49.6 (CH2), 49.8(CH2), 180.3 (CO). MS (ESI-TOF), m/z, calculated:273.2654; found: 273.2600.

N1-decanoylspermine tritrifluoroacetate salt. Sper-mine (9.11 g, 45 mmol) and TFA (10.40 ml, 135 mmol)were dissolved in methanol (500 ml) and dichloro-methane (100 ml) at 0 1C. Decanoyl chloride (3.11 ml,15 mmol) in dichloromethane (100 ml) was added drop-wise for over 15 min. After an additional hour at 0 1C, theice bath was removed and stirring was continued for24 h. The mixture was then concentrated in vacuo andpartitioned between NaOH (aq., 1 M, 200 ml) and chloro-form (CHCl3)–methanol (9:1, 400 ml). The aqueous layerwas back extracted with CHCl3–methanol (9:1, 100 ml).The combined organic layers were washed with NaOH(aq., 1 M, 200 ml) and again back extracted with CHCl3–methanol (9:1, 100 ml). They were then dried withNa2SO4 and evaporated in vacuo, yielding 4.90 g of crudeamine. Crystallization from ethanol as the tri-TFA salt(by addition of a slight excess of TFA) yielded 4.81 g (46%) of product. 1H NMR (D3COD), 0.90 (t; 3H, J¼ 6.6 Hz,CH3), 1.21–1.40 (m, 12H, 6�CH2), 1.54–1.67 (m, 2H,CH2CH2CO), 1.74–1.85 (m, 4H, 2�CH2), 1.88 (p, 2H,J¼ 6.8 Hz, CH2), 2.09 (p, 2H, J¼ 7.7 Hz, CH2), 2.22 (t, 2H,J¼ 7.6 Hz, CH2CO), 2.93–3.20 (m, 10H, 5�CH2N), 3.25–3.33 (m, 2H, CH2N). MS (ESI-TOF), m/z, calculated:357.3593; found: 357.3574.

N1-palmitoylspermine tritrifluoroacetate salt. Sper-mine (3.81 g, 18.8 mmol) was dissolved in methanol(222 ml) at 0 1C and trifluoroacetic acid (4.36 ml,56.6 mmol) was added. Palmitoyl chloride (576 ml,1.89 mmol) in dichloromethane (49 ml) was added forover 1.5 h. The reaction was left on a melting ice bath,which turned to room temperature overnight. Thesample was concentrated and then partitioned betweendichloromethane (200 ml) and NaOH (aq., 1 M, 200 ml).The aqueous layer was separated and reextracted withdichloromethane (2� 200 ml). The combined organicextracts were dried with Na2SO4 and evaporated invacuo, yielding 0.91 g of crude amine. Crystallizationfrom CHCl3–ethanol as the tri-TFA salt (by addition of aslight excess of TFA) yielded 762 mg (52 %) of product.1H NMR (D3COD), 0.90 (t, 3H, J¼ 6.9 Hz, CH3), 1.21–1.39(m, 24H, 12�CH2), 1.55–1.67 (m, 2H, CH2CH2CO), 1.74–1.86 (m, 4H, 2�CH2), 1.87 (t, 2H, J¼ 6.8 Hz 7.0 Hz, CH2),2.09 (p, 2H, J¼ 7.8 Hz, CH2), 2.22 (t, 2H, J¼ 7.6 Hz,CH2CO), 2.95–3.18 (m, 10H, 5�CH2N), 3.29 (t, 2H,J¼ 6.6 Hz, CH2N). MS (ESI-TOF), m/z, calculated:441.4532; found: 441.4553.

N1-oleylspermine tritrifluoroacetate salt. Spermine(6.90 g, 34.09 mmol) and trifluoroacetic acid (TFA,7.88 ml, 102 mmol) were dissolved in methanol (455 ml)at 0 1C. Oleic acid (3.37 g, 11.93 mmol), activated at 0 1Cwith isobutyl chloroformate (1.49 ml, 11.36 mmol) andtriethylamine (3.15 ml, 22.7 mmol), in dichloromethane(91 ml) was added for 1 h and 10 min. After anotherhour, the ice bath was removed and stirring wascontinued at room temperature for 19 h. The solventwas removed under reduced pressure and the mixturewas partitioned between dichloromethane (400 ml) and1 M NaOH (aq., 400 ml). The aqueous layer wasseparated and reextracted with dichloromethane(2� 400 ml). The combined organic layers were washedwith NaOH (aq., 0.5 M, 2� 500 ml), dried with Na2SO4

and evaporated in vacuo, yielding 4.77 g of crude amine.Crystallization in CHCl3–ethanol (1:1) as the tri-TFA salt(by addition of a slight excess of TFA) yielded 5.44 g (59%) of product. 1H NMR (D3COD), 0.90 (t; 3H, J¼ 6.8 Hz;CH3), 1.23–1.43 (m, 20H, 10�CH2), 1.55–1.67 (m, 2H,CH2), 1.73–1.83 (m, 2H, CH2), 1.87 (p, 2H, J¼ 6.8 Hz;CH2), 1.96–2.08 (m, 4H, 2�CH2), 2.08 (p, 2H, J¼ 8.2 Hz;CH2), 2.22 (t, 2H, J¼ 7.6 Hz; CH2), 2.95–3.19 (m, 10H,5�CH2), 3.29 (t, 2H, J¼ 6.4 Hz; CH2), 5.29–5.42 (m, 2H,2�CH). MS (ESI-TOF), m/z, calculated: 467.4689; found:467.4684.

N1-linolylspermine tritrifluoroacetate salt. Spermine(6.90 g, 34.1 mmol) and TFA (7.88 ml, 102 mmol) weredissolved in methanol (455 ml) at 0 1C. Linoleic acid(3.35 g, 11.93 mmol), activated at 0 1C with isobutylchloroformate (1.49 ml, 11.36 mmol) and triethylamine(3.15 ml, 22.72 mmol), in dichloromethane (91 ml) wasadded for 5 min. After an additional 30 min, the ice bathwas removed and stirring was continued at roomtemperature for 7 h. The sample was concentrated andthen partitioned between dichloromethane (600 ml) andNaOH (aq., 1 M, 600 ml). The aqueous layer wasseparated and reextracted with dichloromethane(3� 400 ml). The combined organic extracts were washedwith a solution of NaOH (aq., 1 M, 500 ml) saturated with

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

9

Gene Therapy

NaCl, which was also back extracted with dichloro-methane (600 ml). The combined organic extracts weredried with Na2SO4, filtered and evaporated in vacuo,yielding 4.97 g of crude amine. Crystallization in CHCl3-ethanol (1:1) as the tri-TFA salt (by addition of a slightexcess of TFA) yielded 6.81 g (74 %) of product. 1H NMR(D3COD), 0.90 (t; 3H, J¼ 6.8 Hz, CH3), 1.25–1.44 (m, 14H,7�CH2), 1.55–1.68 (m, 2H, CH2), 1.73–1.85 (m, 4H,2�CH2), 1.88 (p, 2H, J¼ 6.8 Hz, CH2), 2.01–2.15 (m, 6H,3�CH2), 2.22 (t, 2H, J¼ 7.6 Hz, CH2CO), 2.77 (t, 2H,J¼ 6.1 Hz, CHCH2CH), 2.95–3.20 (m, 10H, 5�CH2N),3.29 (t, 2H, J¼ 6.6 Hz, CH2N), 5.27–5.42 (m, 2H, 2�CH).MS (ESI-TOF), m/z, calculated: 465.4532; found:465.4536.

N1-linolenoylspermine tritrifluoroacetate salt. Sper-mine (6.07 g, 30.0 mmol) and TFA (6.93 ml, 90.0 mmol)were dissolved in methanol (400 ml) at 0 1C. Linolenicacid (2.92 g, 10.5 mmol), activated at 0 1C with isobutylchloroformate (1.31 ml, 10.0 mmol) and triethylamine(2.77 ml, 20.0 mmol), in dichloromethane (80 ml) wasadded for over 1 h and 20 min. After an additional hour,the ice bath was removed and stirring was continued atroom temperature overnight. The sample was concen-trated and then partitioned between dichloromethane(400 ml) and NaOH (aq., 1 M, 400 ml, with ca. 5% NaCl).The aqueous layer was separated and reextracted withdichloromethane (3� 200 ml). The combined organicextracts were washed with NaOH (aq., 0.2 M,2� 1000 ml) in which the aqueous layers were backextracted with dichloromethane (2� 300 ml). The com-bined organic extracts were dried with Na2SO4 andevaporated in vacuo, yielding 4.71 g of crude amine.Crystallization from CHCl3-ethanol as the tri-TFA salt(by addition of a slight excess of TFA) yielded 4.88 g (61%) of product. 1H NMR (D3COD), 0.97 (t, 3H, J¼ 7.5 Hz,CH3), 1.27-1.42 (m, 8H, 4�CH2), 1.55–1.68 (m, 2H, CH2),1.75–1.85 (m, 4H, 2�CH2), 1.88 (p, 2H, J¼ 6.8 Hz, CH2),2.02–2.15 (m, 4H, 2�CH2), 2.22 (t, 2H, J¼ 7.6 Hz, CH2),2.81 (t, 2H, J¼ 5.9 Hz, 2�CHCH2CH), 2.96–3.17 (m, 10H,5�CH2N), 3.29 (t, 2H, J¼ 6.6 Hz, CH2N), 5.23–5.43 (m,6H, 6�CH). MS (ESI-TOF), m/z, calculated: 463.4384;found: 463.4384.

Formulation of DNA nanoparticlesMaterials. Plasmids: A 6.7-kbp DNA plasmid, basedon the pEGFPLuc plasmid (Clonetech, BD Bioscience,CA, USA) and modified as described previously, wasused.45 It was isolated from grown DH5-a cells accordingto the Qiagen kit (Qiagen, Chatsworth, CA, USA)protocol and its concentration was estimated usingultraviolet-visible spectroscopy.

Formulations of DNA nanoparticles were prepared byequivolumetric mixing of DNA at the desired concentra-tion with fatty acid–spermine conjugates at variousconcentrations corresponding to different charge ratios(NH3+/PO4�), ranging from 0.3:1 to 10:1. DNA com-plexes were prepared in iso-osmotic mannitol solutions(5.45%). DNA complexes were freshly prepared beforeeach assay. When mentioned, the preparation methodinvolved filtration—using a cellulose acetate filter, 0.2 mmpore size (Whatman Ltd, Maidstone, UK)—after mixingDNA with lipid–spermine compounds.

Gel retardation and DNase I protection assay withlipid–spermine/DNA nanoparticles. Condensation ofDNA was first characterized using an electrophoretic assay(0.8% agarose, in 40 mM Tris-acetate, 20 mM sodiumacetate and 1 mM ethylenediaminetetraacetic acid—Tris-acetate-ethylenediaminetetraacetic acid buffer—pH 7.8).DNA nanoparticles were prepared in 5.45% mannitolsolution at a DNA concentration of 0.1 mg ml�1. DNAcomplexes were prepared at different charge ratios byvarying the concentrations of the fatty acid spermineconjugate. The degree of DNA protection was analyzedusing a DNase I protection assay. In brief, DNA complexeswere first incubated with 5 U DNase I per mg of DNA(Roche, Germany) for 15 min. The reaction was stoppedby the addition of ethylenediaminetetraacetic acid up to0.05 M, pH 8.0. Trypsin (Invitrogen, Karlsruhe, Ger-many) was added to a final concentration of 85 mg l�1

and samples were incubated for 1 h at 37 1C. Sodiumdodecyl sulfate was then added, up to 0.1% (w v�1)concentration, and DNA was further purified using aPCR purification kit (Qiagen) before being subjected toelectrophoresis in agarose gel (0.8% agarose, Tris-acetate-ethylenediaminetetraacetic acid, pH 7.8) containingethidium bromide.

Dynamic light scattering (DLS) studies. The stabilityand hydrodynamic mean diameter of the DNA nano-particles were determined by dynamic light scatteringstudies using a zetasizer Nano ZS apparatus (MalvernInstruments, UK). Mannitol solution (5.45%) was intro-duced into the system as a complex solvent andmeasurements were carried out using a refractive indexof 1.338.

Morphological characterization of lipospermine/DNAnanoparticles. The morphological properties of theensuing DNA complexes were assessed using negativestain and cryo-EM. In brief, 3 ml of freshly preparedlipospermine/DNA complexes was placed on coatedelectron microscopy grids and negatively stained with1% uranyl acetate. The grids were checked using PhilipsCM120 transmission electron microscope (FEI Company,Eindhoven, The Netherlands). For cryo-EM, 2.5 ml ofsample solution was placed on Quantifoil R2/4 holeycarbon grids that were blotted and plunge frozen intoethane with a Vitrobot (FEI Company, Eindhoven, TheNetherlands). Frozen samples were stored under nitro-gen and transferred to a JEOL JEM-2100F (JEOL Ltd,Tokyo, Japan) electron microscope using a Gatan 626cryo-holder. Data were collected under low-dose condi-tions on Kodak SO-163 electron micrograph films. Anominal magnification of 50 000 was used. The filmswere digitized with a Zeiss Scai scanner (Oberkochen,Germany).

Toxicological evaluation. Cytotoxicity of the fatty acid-spermine compounds was assessed using WST-8 assay(Roche, Germany), according to the protocol provided bythe manufacturer. In brief, human hepatic (HepG2) andbaby hamster kidney (BHK) cells were seeded at adensity of 1�105 and 1�104 cells per well, respectively,in 96-well plates and maintained in Dulbecco’s modifiedEagles’ medium (Invitrogen) supplemented with 10%fetal bovine serum (Invitrogen) 24 h before transfection.

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

10

Gene Therapy

The culture medium was replaced with serum-freeDulbecco’s modified Eagles’ medium, including variousconcentrations of lipospermine, ranging from 9 to 92 mM

(corresponding to concentrations of the DNA conden-sates with charge ratios from 0.3:1 to 3:1), or DNAcomplexes formulated as described above. After 3 hincubation at 37 1C, the cell medium was replaced withDulbecco’s modified Eagles’ medium supplementedwith 10% fetal bovine serum. The cells were furtherincubated for 21 h. The number of surviving cells wasdetermined using WST-8 assay. Cell proliferation wasexpressed as the ratio of the A450 of treated cells to that ofuntreated cells.

In vivo studies. The animal experiments were ap-proved by the Swedish local board for laboratoryanimals. Male BALB/c mice aged 10–13 weeks werefirst anesthetized with isoflurane gas (400 ml air flow and4% isoflurane) and kept under anesthesia (220 ml airflow and 2.2% isoflurane) during the administrationprocedure. Fifty microliters containing 5 mg DNA, eitherof pDNA or pDNA-nanocomplexes, were injectedintradermally or intramuscularly. Nine replicates of eachformulation were used for both intramuscular andintradermal studies, except for butanoylspermine for-mulated at a 1:1 ratio, in which the number of sampleswas seven. Gene expression was assessed using imagingof the reporter gene (firefly luciferase) expression. Ondays 1, 8, 16 and 22 after injections, mice wereanesthetized and injected intraperitoneally with 150 mgper kg (B3 mg per mouse) of D-Luciferin (Xenogen,Alameda, CA, USA). Light signal (CCD) images wereobtained using a cooled IVIS CCD camera (Xenogen),and were analyzed using IGOR-PRO Living Imagesoftware, which generates a pseudoimage with anadjustable color scale. The maximum photon per secondof acquisition per cm2pixel per steridian was determinedwithin a region of interest to be the most consistentmeasure for comparative analysis. In general, acquisitiontimes ranged from 3 to 5 min. The results were plotted asthe average photon counts per second over time.

Statistical analysis. All statistical calculations wereperformed using Statistica 7.1 (Statsoft Inc., Uppsala,Sweden). The main effects of different fatty acid–spermine/DNA nanocomplexes on gene expressionwere analyzed using a two-way analysis of variance,including all treatments and using time as a repeatingvariable. To achieve normal distribution, the logarithm ofraw data was used. If the overall F-ratio for treatmentwas significant, pairwise Bonferroni-corrected analysesof variance was performed for comparing the differenttreatments. A Pp0.05 was considered statisticallysignificant.

Conflict of interest

The authors declare no conflict of interest.

Acknowledgements

We thank Alex Peralvarez-Marın for his invaluable helpin dynamic light scatter measurements. We also thank

the Organic Chemistry Department and Professor AstridGraslund from the Department of Biochemistry andByophysics at Stockholm University for providing accessto the zetasizer equipment. This project was financiallysupported by the Swedish Research Council, TheSwedish Foundation for Strategic Research Bio-Xgrant,KI Faculty funds for funding of postgraduate students,Marie Curie fellowship programme—Eurogendistraining site, Portuguese Foundation for Science andTechnology (SFRH-BD-16757-2004), the European UnionFP6 grant EuroPharmacoGene (FP6-2005-037283) andSynthe Gene Delivery (LSHB-CT-2005-018716).

References

1 Gehl J. Electroporation for drug and gene delivery in the clinic:doctors go electric. Methods Mol Biol 2008; 423: 351–359.

2 Lu X, Sankin G, Pua EC, Madden J, Zhong P. Activation oftransgene expression in skeletal muscle by focused ultrasound.Biochem Biophys Res Commun 2009; 379: 428–433.

3 Mir LM, Moller PH, Andre F, Gehl J. Electric pulse-mediatedgene delivery to various animal tissues. Adv Genet 2005; 54:83–114.

4 van Drunen Littel-van den Hurk S, Luxembourg A, Ellefsen B,Wilson D, Ubach A, Hannaman D et al. Electroporation-basedDNA transfer enhances gene expression and immune responsesto DNA vaccines in cattle. Vaccine 2008; 26: 5503–5509.

5 Ziady AG, Gedeon CR, Muhammad O, Stillwell V, Oette SM,Fink TL et al. Minimal toxicity of stabilized compacted DNAnanoparticles in the murine lung. Mol Ther 2003; 8: 948–956.

6 Freimark BD, Blezinger HP, Florack VJ, Nordstrom JL, Long SD,Deshpande DS et al. Cationic lipids enhance cytokine and cellinflux levels in the lung following administration of plasmid:cationic lipid complexes. J Immunol 1998; 160: 4580–4586.

7 Fischer D, Li Y, Ahlemeyer B, Krieglstein J, Kissel T. In vitrocytotoxicity testing of polycations: influence of polymerstructure on cell viability and hemolysis. Biomaterials 2003; 24:1121–1131.

8 Zintchenko A, Philipp A, Dehshahri A, Wagner E. Simplemodifications of branched PEI lead to highly efficient siRNAcarriers with low toxicity. Bioconjug Chem 2008; 19: 1448–1455;epub ahead of print 14 June 2008.

9 Ziady AG, Gedeon CR, Miller T, Quan W, Payne JM, Hyatt SLet al. Transfection of airway epithelium by stable PEGylatedpoly-L-lysine DNA nanoparticles in vivo. Mol Ther 2003; 8:936–947.

10 Luten J, van Nostrum CF, De Smedt SC, Hennink WE.Biodegradable polymers as non-viral carriers for plasmid DNAdelivery. J Control Release 2008; 126: 97–110.

11 Mann A, Richa R, Ganguli M. DNA condensation by poly-L-lysine at the single molecule level: role of DNA concentrationand polymer length. J Control Release 2008; 125: 252–262.

12 Mann A, Thakur G, Shukla V, Ganguli M. Peptides in DNAdelivery: current insights and future directions. Drug DiscovToday 2008; 13: 152–160.

13 Bloomfield VA. DNA condensation by multivalent cations.Biopolymers 1997; 44: 269–282.

14 Martin ME, Rice KG. Peptide-guided gene delivery. AAPS J 2007;9: E18–E29.

15 Davies OR, Head L, Armitage D, Pearson EA, Davies MC,Marlow M et al. Surface modification of microspheres with stericstabilizing and cationic polymers for gene delivery. Langmuir2008; 24: 7138–7146.

16 Childs AC, Mehta DJ, Gerner EW. Polyamine-dependent geneexpression. Cell Mol Life Sci 2003; 60: 1394–1406.

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

11

Gene Therapy

17 D0Agostino L, di Pietro M, Di Luccia A. Nuclear aggregates ofpolyamines are supramolecular structures that play a crucialrole in genomic DNA protection and conformation. FEBS J 2005;272: 3777–3787.

18 Patel MM, Anchordoquy TJ. Contribution of hydrophobicity tothermodynamics of ligand-DNA binding and DNA collapse.Biophys J 2005; 88: 2089–2103.

19 Geall AJ, Blagbrough IS. Homologation of polyamines inthe rapid synthesis of lipospermine conjugates and relatedlipoplexes. Tetrahedron 2000; 56: 2449–2460.

20 Geall A, Blagbrough IS. Homologation of polyamines in thesynthesis of lipo-spermine conjugates and related lipoplexes.Tetrahedron Lett 1998; 39: 443–446.

21 Ahmed OA, Pourzand C, Blagbrough IS. Varying the unsatura-tion in N4,N9-dioctadecanoyl spermines: nonviral lipopolya-mine vectors for more efficient plasmid DNA formulation.Pharm Res 2006; 23: 31–40.

22 Abbasi M, Uludag H, Incani V, Hsu CY, Jeffery A. Furtherinvestigation of lipid-substituted poly(L-lysine) polymers fortransfection of human skin fibroblasts. Biomacromolecules 2008; 9:1618–1630.

23 Arscott PG, Ma C, Wenner JR, Bloomfield VA. DNA condensationby cobalt hexaammine (III) in alcohol-water mixtures: dielectricconstant and other solvent effects. Biopolymers 1995; 36: 345–364.

24 Cherny DI, Jovin TM. Electron and scanning force microscopystudies of alterations in supercoiled DNA tertiary structure.J Mol Biol 2001; 313: 295–307.

25 Hud NV, Allen MJ, Downing KH, Lee J, Balhorn R. Identificationof the elemental packing unit of DNA in mammalian sperm cellsby atomic force microscopy. Biochem Biophys Res Commun 1993;193: 1347–1354.

26 Talelli M, Pispas S. Complexes of cationic block copolymermicelles with DNA: histone/DNA complex mimetics. MacromolBiosci 2008; 8: 960–967.

27 Allen MJ, Bradbury EM, Balhorn R. AFM analysis of DNA-protamine complexes bound to mica. Nucleic Acids Res 1997; 25:2221–2226.

28 Blagbrough IS, Geall AJ, Neal AP. Polyamines and novelpolyamine conjugates interact with DNA in ways that can beexploited in non-viral gene therapy. Biochem Soc Trans 2003; 31:397–406.

29 Gaucheron J, Santaella C, Vierling P. Highly fluorinatedlipospermines for gene transfer: synthesis and evaluation of theirin vitro transfection efficiency. Bioconjug Chem 2001; 12: 114–128.

30 Hosseinkhani H, Tabata Y. Self assembly of DNA nanoparticleswith polycations for the delivery of genetic materials into cells.J Nanosci Nanotechnol 2006; 6: 2320–2328.

31 Kichler A, Zauner W, Ogris M, Wagner E. Influence of the DNAcomplexation medium on the transfection efficiency of lipos-permine/DNA particles. Gene Therapy 1998; 5: 855–860.

32 Geall AJ, Al-Hadithi D, Blagbrough IS. Efficient calf thymusDNA condensation upon binding with novel bile acidpolyamine amides. Bioconjug Chem 2002; 13: 481–490.

33 Batrakova EV, Kabanov AV. Pluronic block copolymers:evolution of drug delivery concept from inert nanocarriersto biological response modifiers. J Control Release 2008; 130:98–106.

34 Bello-Roufai M, Lambert O, Pitard B. Relationships between thephysicochemical properties of an amphiphilic triblock copoly-mers/DNA complexes and their intramuscular transfectionefficiency. Nucleic Acids Res 2007; 35: 728–739.

35 Dunlap DD, Maggi A, Soria MR, Monaco L. Nanoscopicstructure of DNA condensed for gene delivery. Nucleic AcidsRes 1997; 25: 3095–3101.

36 Avgoustakis K. Pegylated poly(lactide) and poly(lactide-co-glycolide) nanoparticles: preparation, properties and possibleapplications in drug delivery. Curr Drug Deliv 2004; 1: 321–333.

37 van Vlerken LE, Vyas TK, Amiji MM. Poly(ethylene glycol)-modified nanocarriers for tumor-targeted and intracellulardelivery. Pharm Res 2007; 24: 1405–1414.

38 Chang CW, Choi D, Kim WJ, Yockman JW, Christensen LV, KimYH et al. Non-ionic amphiphilic biodegradable PEG-PLGA-PEGcopolymer enhances gene delivery efficiency in rat skeletalmuscle. J Control Release 2007; 118: 245–253.

39 Gautam A, Densmore CL, Golunski E, Xu B, Waldrep JC.Transgene expression in mouse airway epithelium byaerosol gene therapy with PEI-DNA complexes. Mol Ther 2001;3: 551–556.

40 Wolff JA, Budker V. The mechanism of naked DNA uptake andexpression. Adv Genet 2005; 54: 3–20.

41 Brooks AR, Wu F, Liu P, Qian HS, Wang P, Gibson H et al. 1078Silencing of transgene expression following plasmid baseddelivery to murine skeletal muscle is dose dependent. Mol Ther2006; 13: S413.

42 Pomel C, Leborgne C, Cheradame H, Scherman D, Kichler A,Guegan P. Synthesis and evaluation of amphiphilic poly(tetra-hydrofuran-b-ethylene oxide) copolymers for DNA delivery intoskeletal muscle. Pharm Res 2008; 25: 2963–2971.

43 Gaymalov ZZ, Yang Z, Pisarev VM, Alakhov VY, Kabanov AV.The effect of the nonionic block copolymer pluronic P85 on geneexpression in mouse muscle and antigen-presenting cells.Biomaterials 2009; 30: 1232–1245.

44 Lavigne MD, Yates L, Coxhead P, Gorecki DC. Nuclear-targetedchimeric vector enhancing nonviral gene transfer intoskeletal muscle of Fabry mice in vivo. FASEB J 2008; 22:2097–2107.

45 Lundin KE, Ge R, Svahn MG, Tornquist E, Leijon M, Branden LJet al. Cooperative strand invasion of supercoiled plasmid DNAby mixed linear PNA and PNA-peptide chimeras. Biomol Eng2004; 21: 51–59.

Supplementary Information accompanies the paper on Gene Therapy website (http://www.nature.com/gt)

Fatty acid–spermine conjugates as DNA carriersJR Viola et al

12

Gene Therapy