RESEARCH ARTICLE

Improved chitosan-mediated gene delivery basedon easily dissociated chitosan polyplexes of highlydefined chitosan oligomers

M Koping-Hoggard1, KM Varum2, M Issa1, S Danielsen3, BE Christensen2, BT Stokke3 and P Artursson1

1Department of Pharmacy, Uppsala University, Uppsala, Sweden1; 2Norwegian Biopolymer Laboratory (NOBIPOL), Department ofBiotechnology, Norwegian University of Science and Technology, Trondheim, Norway2; 3Norwegian Biopolymer Laboratory (NOBIPOL),Department of Physics, Norwegian University of Science and Technology, N-7491 Trondheim, Norway3

Nonviral gene delivery systems based on conventional high-molecular-weight chitosans are efficient after lung adminis-tration in vivo, but have poor physical properties such asaggregated shapes, low solubility at neutral pH, highviscosity at concentrations used for in vivo delivery and aslow dissociation and release of plasmid DNA, resulting in aslow onset of action. We therefore developed highly effectivenonviral gene delivery systems with improved physicalproperties from a series of chitosan oligomers, ranging inmolecular weight from 1.2 to 10 kDa. First, we establishedstructure–property relationships with regard to polyplexformation and in vivo efficiency after lung administration tomice. In a second step, we isolated chitosan oligomers froma preferred oligomer fraction to obtain fractions, ranging from10 to 50-mers, of more homogeneous size distributions withpolydispersities ranging from 1.01 to 1.09. Polyplexes basedon chitosan oligomers dissociated more easily than thoseof a high-molecular-weight ultrapure chitosan (UPC, approxi-

mately a 1000-mer), and released pDNA in the presenceof anionic heparin. The more easily dissociated polyplexesmediated a faster onset of action and gave a higher geneexpression both in 293 cells in vitro and after lungadministration in vivo as compared to the more stable UPCpolyplexes. Already 24 h after intratracheal administration, a120- to 260-fold higher luciferase gene expression wasobserved compared to UPC in the mouse lung in vivo. Thegene expression in the lung was comparable to that of PEI(respective AUCs of 27567710 and 33207871 pg lucifer-ase� days/mg of total lung protein). In conclusion, a majorimprovement of chitosan-mediated nonviral gene deliveryto the lung was obtained by using polyplexes of well-definedchitosan oligomers. Polyplexes of oligomer fractions also hadsuperior physicochemical properties to commonly used high-molecular-weight UPC.Gene Therapy (2004) 11, 1441–1452. doi:10.1038/sj.gt.3302312; Published online 22 July 2004

Keywords: nonviral gene therapy; nonviral gene delivery; chitosan; plasmid DNA; polyethylenimine

Introduction

Cationic polymers such as polyethylenimine (PEI)1,2 andpolyamidoamine dendrimers3,4 form complexes withpDNA (polyplexes) that are regarded as some of themost effective nonviral gene delivery systems.5 Thesepolymers are synthetic and not biocompatible and arough correlation between toxicity and efficiency hasbeen suggested for PEI.6 In particular, several recentstudies have addressed concerns about the toxicity ofconventional PEI,7–9 although PEIs of decreased mole-cular weights and degree of branching seem to have areduced toxicity.10,11 Therefore, the search for alternativebiocompatible cationic polymers that form efficient andnontoxic polyplexes is motivated.

Chitosans, a family of linear binary polysaccharidescomprised of beta (1–4) linked 2-amino-2-deoxy-b-D-glucose (GlcN; D-unit) and the N-acetylated analogue(GlcNAc; A-unit), have been proposed as biocompatiblealternative cationic polymers, suitable for nonviral genedelivery. Previous studies have shown that chitosanpolyplexes exhibit low cytotoxicity12–15 and are suitablefor gene delivery to mucosal tissues such as the intestineand the lung in vivo.12,16–21 In most of these studies,commercially available chitosans of high molecularweights (100–400 kDa) were used. These chitosans formextremely stable polyplexes with DNA, which delays therelease of the pDNA, and the physical shape of thepolyplexes is dominated by aggregates.12,17,20 Otherpharmaceutical drawbacks with high-molecular-weightchitosans include their low solubility at physiologicalpH22 and their viscosity enhancing properties at con-centrations suitable for in vivo gene therapy.12 Wespeculated that these drawbacks could be eliminated ifchitosans of lower molecular weights were used, that is,chitosans that were long enough to form stable poly-plexes, but sufficiently short to be soluble at neutral pH,give a reduced viscosity, form less aggregated shapes

Received 28 October 2003; accepted 25 April 2004; published online22 July 2004

1Project outline, formulation and all in vitro and in vivo studies withpolyplexes2Chitosan preparation and characterization3Atomic force microscopy studiesCorrespondence: Professor P Artursson, Department of Pharmacy,Uppsala University, Box 580, Uppsala SE-751 23, Sweden

Gene Therapy (2004) 11, 1441–1452& 2004 Nature Publishing Group All rights reserved 0969-7128/04 $30.00

www.nature.com/gt

more typical for multivalent ions, and form more easilydissociated polyplexes.23,24

Chitosan oligomers shorter than 14 monomer unitswere recently found to fulfill many of these pharma-ceutical requirements, but formed only weak complexeswith DNA, resulting in physically unstable polyplexesthat transfected cells at low efficiencies in vitro and invivo.20 A larger chitosan oligomer fraction (approxi-mately a 24-mer; 4.7 kDa) formed physically stablepolyplexes of comparable efficiency to very stablepolyplexes based on higher molecular weights of thechitosan. We therefore hypothesized that chitosanoligomers with chain lengths intermediate to thosethat form unstable (p14-mers) and more stable poly-plexes (X24-mers), that is, oligomers around 18–20monomer units, would give the best possible chitosansfor nonviral gene delivery. Furthermore, since thereseems to be a strong dependence between the chitosanoligomer length and the properties of the polyplexes,we also wanted to study more homogeneous chitosanoligomers.

For this purpose, we first produced a low-molecular-weight chitosan by random depolymerization, with anumber average degree of polymerization (DPn) of 18monomer units (DPn18), and with a well-defineddistribution of the oligomers. Initial characterizationswere performed of polyplexes based on DPn18 andchitosans of smaller and larger sizes to establishstructure–property relationships with regard to polyplexformation and in vivo efficiency after lung administrationto mice. In a second step, we isolated the chitosanoligomers from the broader DPn18 fraction to obtainfractions, ranging from 10- to 50-mers, of more homo-geneous size distributions. Polyplexes of these well-defined chitosan fractions were then investigated withregard to physical–chemical properties as well as thetransfection efficiency in vitro and after lung administra-tion to mice in vivo.

Results

Structure–property relationshipsA low-molecular-weight chitosan with a number averageDPn of 18 (denoted as DPn18) was fractionated andfound to be comprised of oligomers ranging from 6- to50-mers (Figure 1a). Thus, the DPn18 chitosan containsa mixture of shorter and longer oligomers that formphysically unstable (DPp14) and stable (DPX24)polyplexes, respectively.20 We first studied the stabilityof polyplexes between DPn18 and pDNA at increa-sing charge ratios (10:1–60:1 (+/�)) in the agarose gelretardation assay. A very high charge ratio of 60:1 (+/�)was required to form polyplexes that were sufficientlystable to retain completely the pDNA in the agarose gels(Figure 1b). The transfection efficiency of these poly-plexes was tested in the 293 cell line since our previousstudies have shown that there is a good qualitativerelationship between the transfection efficiency in thesecells and that in the mouse lung epithelium in vivo afterintratracheal administration.12,20 As expected, the physi-cally stable polyplexes at charge ratio 60:1 (+/�)mediated higher luciferase gene expression in 293 cellsin vitro than those that separated in the gel assay. Thus,a 100-fold higher in vitro transgene expression was

obtained at charge ratio 60:1 (+/�) than at 10:1 (+/�)(Figure 1c).

The in vivo efficiency of the DPn18 chitosan wasstudied after intratracheal administration to mouselungs.12 First, the effect of the charge ratio of DPn18polyplexes on the level of gene expression was studied atthree different charge ratios, at 10:1 (+/�) whereunstable polyplexes were formed, at 30:1 (+/�) wherethe complexes were partially stable during agarose gelelectrophoresis, and at 60:1 (+/�) where the pDNA wascompletely retained in the agarose gel retardation assay,c.f. Figure 1b. The luciferase gene expression wasanalysed 72 h after administration since previous studiesusing chitosan polyplexes showed peak expression atthis time point.12 The luciferase gene expression inmouse lungs increased with increasing stability of thepolyplexes. Consistent with the in vitro results, the stablepolyplexes at charge ratio 60:1 (+/�) mediated a four-fold higher gene expression in comparison to the lessstable polyplexes at 10:1 (+/�) (Po0.001) (Figure 2a).

Figure 1 In vitro characterization of polyplexes based on low-molecular-weight chitosan. (a) A low-molecular-weight chitosan with a numberaverage DPn18 was fractionated with size-exclusion chromatography andfound to be comprised of oligomers between 6- and 50-mers. (b) TheDPn18 chitosan was complexed with 100 ng pDNA at increasing chargeratios (10:1�60:1 (+/�)) and tested for physical stability in the agarose gelretardation assay. A representative gel from three independent experimentsis shown. (c) The DPn18 polyplexes were also incubated with 293 cells andthe luciferase gene expression was investigated after 48 h. Physically stablepolyplexes at charge ratio of 60:1 (7) mediated higher gene expressionthan the less stable polyplexes obtained at lower charge ratios. Results areexpressed as mean values7s.d. (n¼ 4). Statistical differences are denotedas *Po0.05, **Po0.01 and ***Po0.001, respectively.

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1442

Gene Therapy

Notably, the gene expression obtained with less stableDPn18 polyplexes at charge ratio 10:1 (+/�) wascomparable to that obtained previously with an opti-mized composition of ultrapure (high-molecular-weight)chitosan (UPC).12

Dose titration studies at charge ratio 60:1 (+/�)showed that the efficiency of the polyplexes increasedwith an increase in the dose, from 45713 pg luciferase/mg of total lung protein at a dose of 5 mg pDNA to3407177 pg luciferase/mg of total lung protein at 25 mgpDNA, or seven-fold (Po0.01) (Figure 2b). Histologicalobservations of lung sections from mice given DPn18polyplexes (25 mg pDNA; 60:1 (+/�)) showed no signsof immune cell infiltration in the bronchiolar epitheliumor the alveolar regions (Figure 2c). Similarly, no acuteinflammatory response to the single administration ofPEI polyplexes was observed (data not shown). Based onthese results, we compared the in vivo efficiency ofpolyplexes (containing 25 mg pDNA) of different chit-osan oligomers that were both shorter and longer thanDPn18, at a fixed charge ratio of 60:1 (+/�) (Figure 2d).Polyplexes based on DPn18 mediated a four-fold higher

gene expression than those based on a slightly largerchitosan fraction, DPn25 (Po0.05), and a 50- to 100-foldhigher gene expression than those based on 12- and6-mer chitosan (Po0.001) (which formed unstablepolyplexes even at a charge ratio of 60:1 (+/�) ofcomparable efficiency to naked pDNA), supporting thehypothesis that the optimal size range for the formationof efficient chitosan polyplexes was in the vicinity of 20monomer units.

Polyplexes based on high-molecular-weight chitosansare known to be extremely stable against decomplexationwith other anionic molecules such as heparin17 and wereeven found to resist high salt and detergent concentra-tions.12 The high stability was suggested to be a majorrate limiting step for the intracellular release of pDNAfrom these polyplexes,17 and kinetic studies usingtransmission electron microscopy showed that chitosanpolyplexes remained in intact endosomes for 424 h.12

Since shorter oligomers (DPo14) show a weaker bindingwith DNA,20 we speculated that polyplexes of DPn18would release pDNA more easily than polyplexes ofhigh-molecular-weight chitosan. We therefore investi-gated the stability of pDNA polyplexes of DPn18 withthose of high-molecular-weight UPC (having a DP ofaround 1000) in the presence and absence of heparin. Thepolyplexes of the high-molecular-weight chitosan re-mained stable, while polyplexes of DPn18 dissociatedand released the pDNA upon incubation with heparin(Figure 3a). Also, the plasmid remained almost comple-tely supercoiled after dissociation from the low-molecu-lar-weight chitosan, indicating that the complexationprocedure with DPn18 did not alter the physical integrityof the plasmid (Figure 3a).

Figure 2 In vivo characterization of polyplexes based on low-molecular-weight chitosan after intratracheal administration to mouse lungs. (a) Theeffect of charge ratio (10:1, 30:1 and 60:1 (7)) on the luciferase geneexpression using 25 mg pDNA was studied with DPn18 polyplexes. Theluciferase gene expression was analysed 72 h after administration. (b) Theeffect of pDNA dose (5, 10 and 25 mg pDNA) on the luciferase geneexpression was studied with DPn18 polyplexes 60:1 (7) at different doses.The luciferase gene expression was analysed 72 h after administration. (c)Haematoxylin/eosin-stained lung sections (5 mm), obtained from mouselungs 24 h after the administration of DPn18 polyplexes (charge ratio 60:1(7), 25 mg pDNA), were examined for the presence of tissue damage andinflammatory cells (left column). No signs of tissue damage or immune cellinfiltration in the bronchiolar epithelium (top row) or the alveolar regions(bottom row) were observed. Lung sections from nontreated animals areshown in the right column. Original magnifications, � 300. (d) The invivo efficiencies of chitosan polyplexes using different chitosan oligomers(charge ratio 60:1 (7), 25 mg pDNA) were compared 72 h afteradministration to mouse lungs. The highest luciferase lung gene expressionwas obtained with polyplexes having a mean chain length of 18 monomerunits. Results are expressed as mean values7s.d. (n¼ 4). Statisticaldifferences are denoted as *Po0.05, **Po0.01 and ***Po0.001,respectively. NS¼no significant difference (P40.05).

Figure 3 Effect of heparin on the stability of polyplexes using chitosan ofvarying molecular weights. (a) Chitosan polyplexes of DPn18 at chargeratio 60:1 (7) and of the high molecular weight UPC at a previouslyoptimized charge ratio of 2.4:1 (7)12,20 were tested for stability in thepresence of heparin (+) using the agarose gel retardation assay. The arrowsindicate loading position, open circle (o.c.) and supercoiled (s.c.) forms ofpDNA. A representative gel from one of three independent experiments isshown. (b) The more easily dissociated polyplexes of DPn18 mediated afour- to 24-fold higher in vitro luciferase gene expression at 48 h comparedto the more stable polyplexes of UPC. Results are expressed as meanvalues7s.d. (n¼ 4). *Po0.05.

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1443

Gene Therapy

The efficiencies of these two chitosan systems werethen compared in three different cell lines, that is, kidneyepithelial 293 cells, airway epithelial Calu-3 cells and incervix epithelial HeLa cells. An early time point, 48 h,was chosen since we hypothesized that the differencein the release pattern of pDNA would impact early geneexpression levels. A four- to 24-fold higher geneexpression was detected with the low-molecular-weightchitosan DPn18 than with the very stable polyplexesformed with high-molecular-weight chitosan (Po0.05)(Figure 3b).

In summary, the low-molecular-weight chitosanDPn18 (that contains oligomers between 6- and 50-mers)formed physically stable polyplexes at high charge ratios(+/�) with pDNA, but where more easily dissociatedand of higher efficiency than conventional high-molecu-lar-weight chitosans. We reasoned that a fractionation ofmore monodisperse fractions of DPn18 would furtherimprove the efficiency of the oligomer-based polyplexes,and give polyplexes of better defined physicochemicalproperties. Therefore, narrower sized chitosan oligomerfractions were purified from the DPn18 fraction forfurther investigations.

Characterization of chitosan polyplexes basedon oligomers with narrow size distributionsIn order to obtain better defined chitosan oligomers,DPn18 was separated into four fractions using size-exclusion chromatography25 (Figures 1a and 4a). Thefractions were defined according to their DPn as DP10–14, DP15–21, DP22–35 and DP36–50. Chitosans shorterthan 10 monomer units were excluded since they formedunstable polyplexes of low in vivo efficiency (Figure 2d).Based on the initial studies, polyplexes of the differentfractions were formed at a charge ratio of 60:1 (+/�) andproperties such as the physical stability, size, morphol-ogy and the in vitro and in vivo efficiencies werecompared to those of DPn18. For transfection studiesin vitro and after lung administration in vivo, PEIpolyplexes (25 kDa) at a previously optimized chargeratio of 5:1 (+/�)2,12 were used as controls.

Physicochemical characterization. All the chitosanfractions formed colloidal particles of comparable sur-face charge (zeta potentials ranging from 29.571.1 to32.671.0 mV) when mixed with pDNA at a charge ratioof 60:1 (+/�) (Table 1). However, only fractions thatcontain chitosan oligomers with chain lengths longerthan 22 monomer units, that is, DP22–35 and DP36–50,formed physically stable polyplexes in the agarose gelretardation assay (Figure 4b). As shown in Figure 4c, allthe chitosan fractions formed nonaggregated polyplexesof rod-like, toroidal and spherical shapes when visua-lized by atomic force microscopy. The three differentmorphologies are apparent at larger magnifications,Figure 5. The distribution between the three shapesdepended on the chitosan fraction (Table 1). Theoligomer fractions DP15–21 and DP22–35 formedslightly, but significantly, higher proportions of spherescompared to the polydisperse DPn18 oligomer (Po0.05).

At higher pDNA concentrations (250 mg/ml), similarto those used for in vivo studies, the size of the DPn18and DP15–21 polyplexes differed significantly: 25476and 45713 nm, for DPn18 and DP15–21 polyplexes,

respectively (Po0.001) (Table 2). This suggests thatDP15–21 polyplexes are more stable towards aggregationat higher concentrations than those of DPn18, and aretherefore pharmaceutically more desirable.

In vitro gene expression and toxicity studies. Thein vitro transfection efficiencies of chitosan oligomer andPEI polyplexes were studied in the 293 cell line (Figure6). In general, physically stable polyplexes of DPn18 andDP36–50 gave the highest luciferase gene expression ofthe chitosan oligomers. A clear dependence of the pH of

Figure 4 Characteristics of chitosan oligomer fractions and of chitosanpolyplexes based on these fractions. (a) Chitosan oligomer fractionsseparated from the DPn18 chitosan had very narrow sized distributions:DP10–14, DP15–21, DP22–35 and DP36–50. (b) Physical stability ofpolyplexes formed with the respective oligomer fraction at a charge ratio of60:1 (7) as assayed by the agarose gel electrophoresis assay. Naked pDNAwas used as control. Representative gels from three independentexperiments are shown. (c) Representative AFM topographs showing thephysical shape of the polyplexes. The concentration of pDNA was 13.3 mg/ml and the charge ratio was 60:1 (7). Naked pDNA was used as control.The horizontal bars indicate 200 nm and the vertical bars the height of thepolyplexes. The arrows point at spherically shaped polyplexes.

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1444

Gene Therapy

the transfection medium on the transfection efficiencywas observed. At pH 7.4, PEI polyplexes gave high geneexpression already at 24 h (reaching a plateau of9637101 pg luciferase/mg of total cell protein untildeclining after 4 days), while the gene expression ofDPn18 and DP36–50 polyplexes increased with time up

to 120 h (expressing 258796 and 10577246 pg lucifer-ase/mg of total cell protein, respectively) (Figure 6a).

A more rapid onset and higher gene expression wereobtained with chitosan oligomer polyplexes when thetransfection studies were performed at pH 5.0 comparedto pH 7.4. However, in contrast to the chitosan oligomers,a lower level and a slower onset of gene expression wereobtained with PEI polyplexes at pH 5.0. At pH 5.0, bothDPn18 and DP36–50 showed the maximum expression at72 h compared to 120 h at pH 7.4 (Figure 6b). At this timepoint, chitosan polyplexes mediated a four-fold higherluciferase gene expression compared to PEI polyplexes(Figure 6c), corresponding to 0.2% of the total cell protein(E2000 pg luciferase/mg of total cell protein).

Interestingly, the higher in vitro efficiency of theoligomer polyplexes at pH 5 related to a more compactpolyplex structure, similar to PEI, as indicated byreduced intercalation of ethidium bromide with pDNAwhen the pH of the final agarose gel electrophoresisbuffer was reduced (Figure 6d). This agrees withpreviously reported data on chitosan oligomer-basedDNA complexes using a different technique, that is,video-enhanced fluorescence microscopy.20

No acute in vitro toxicity of the polyplexes wasobserved with the MTT assay (which measures theintracellular dehydrogenase activity) at concentrationsused for the transfection studies (0.33 mg pDNA, corre-sponding to 235 mg/ml chitosan oligomer and 4.5 mg/mlPEI). However, in contrast to the low-molecular-weightchitosans, PEI showed a dose-dependent toxicity atconcentrations above 20 mg/ml, giving IC50 values of81 and 108 mg/ml at pH 7.4 and 5.0, respectively,supporting the notion that this polymer has a dose-dependent toxicity.12,26 Chitosan oligomers showed notoxicity at pH 5.0 or pH 7.4 even at concentrations ashigh as 2.5 mg/ml.

Intratracheal administration to mouse lungs. Thein vivo efficiencies of the oligomeric chitosans werestudied at 72 h after intratracheal administration tomouse lungs.12 As controls, PEI and a chitosan fractionwith 25 as the number average DPn (DPn25) were used.

The highest luciferase gene expression was obtainedwith the fraction containing 15–21 chitosan monomerunits (Po0.05) (Figure 7). This was surprising since thesepolyplexes were physically unstable in the agarose gelretardation assay (c.f. Figure 4b) and mediated low geneexpression in vitro (c.f. Figure 6c). The higher efficiency ofthe fraction DP15–21 was also in contrast to the findingthat the in vivo efficiency of DPn18 polyplexes increasedwith increasing physical stability of the polyplexes, c.f.Figure 2a. DPn18 gave a lower luciferase gene expressionthan DP15–21, comparable to PEI (but a three- to four-fold higher expression than DPn25).

Stability of pDNA against degradation by DNase. Sincethe commonly accepted relationship between physicalstability and efficiency was violated by DP15–21, weinvestigated if the physically unstable, but efficientpolyplexes may have retained their capacity to protectpDNA against DNase degradation. Naked pDNA, used ascontrol, was completely degraded upon incubation withDNase (Figure 8a). In contrast, for DPn18 polyplexes, bothphysical stability and stability against degradation byDNase increased with increasing charge ratio (Figure 8a).

Table 1 Physical shape and size of chitosan polyplexes based ondifferent oligomer fractions

Chitosan fractiona Physical shapeb Size (nm)c

% Rods % Toroids % Spheres

DPn18 64 9 27 9178DP10–14 55 3 42 6879DP15–21 50 6 44d 78712DP22–35 48 4 48d 10476DP36–50 53 8 39 164711

aPolyplexes were formed at a charge ratio of 60:1 (7) and aconcentration of 13.3 mg/ml pDNA.bCalculation of shape distributions is based on more than 100individual polyplexes, collected from several AFM topographs.cSize of the polyplexes was measured by photon correlationspectroscopy.dFractions DP15–21 and DP22–35 formed significantly higher(Po0.05) proportions of spheres in comparison to DPn18.

Figure 5 Spherical, toroidal and rod shapes of polyplexes based onfractionated chitosan oligomers. The polyplexes were formed at aconcentration of 13.3 mg/ml and a charge ratio of 60:1 (7) and examinedin the atomic force microscope. The horizontal bars indicate 100 nm(spheres and toroids) and 200 nm (rods). The vertical bar gives the heightof the polyplexes.

Table 2 Size of DPn18 and DP15–21 polyplexes at charge ratio 60:1(7) as a function of the increase in pDNA concentration using theSpeedVac centrifugation technique

Concentration (mg/ml) Particle sizea (nm)

DPn18 DP15-21

25 41715 3771050 41712 3279250 25476 45713b

500 463712 119727b

aSize of the polyplexes was measured by photon correlationspectroscopy.bSize of the DP15–21 polyplexes was significantly lower (Po0.001)than the DPn18 polyplexes at pDNA concentrations of 250 and500 mg/ml, respectively.

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1445

Gene Therapy

This agrees with the observed increase in efficiency withan increased charge ratio of these polyplexes in vivo(Figure 2a). Interestingly, polyplexes of DP15–21 at chargeratio 60:1 (+/�) that were classified as physically unstable

in the agarose gel retardation assay also protected pDNAagainst degradation by DNase (Figure 8b). This suggeststhat protection against degradation by DNase is alsodependent on the amount of chitosan and/or the chargeof the polyplexes.

In an attempt to investigate whether the DP15–21oligomers themselves have any inhibitory effects on theDNase activity, we dissociated the DP15–21 polyplexeswith heparin before incubating the decomplexed poly-plexes with DNase. However, no conclusions could bedrawn since the presence of heparin inhibited the DNaseactivity (data not shown), which is in agreement with aprevious report.27 Nevertheless, preliminary purificationstudies using ultrafiltration showed that the removal ofexcess unbound DP15–21 oligomers from the DP15–21polyplex solution (around 60–70% of the total amountof oligomers) at the charge ratio 60:1 (+/�) drasticallyreduced the protection of pDNA against DNase degra-dation (Figure 8c), and resulted in a four-fold lower lunggene expression (data not shown).

Kinetics and location of luciferase gene expressionin mouse lungs. Polyplexes based on low-molecular-weight chitosan released pDNA after incubation withheparin, in contrast to those based on high-molecular-weight chitosan (Figure 3a). We hypothesized that poly-plexes that are more easily dissociated would show peakexpression, Cmax, after lung administration at earlier time

Figure 6 In vitro transfection efficiency of polyplexes based on fractionated chitosan oligomers. (a and b) Time-course studies of luciferase gene expressionin 293 cells at (a) pH 7.4 and (b) pH 5.0 after incubation with different chitosan oligomer polyplexes and PEI polyplexes at charge ratios of 60:1 (7) and 5:1(7), respectively. (c) Comparison of luciferase gene expression in 293 cells between chitosan oligomers 60:1 (7) and PEI 5:1 (7) polyplexes 72 h aftertransfection at pH 5.0. Results are expressed as mean values7s.d. (n¼ 4). ***Po0.001. (d) Physical stability of DPn18 and PEI polyplexes at charge ratiosof 60:1 (7) and 5:1 (7), respectively, at varying pHs of the agarose gel electrophoresis buffer (standard pH of 8.5, and pH 7.4 and 5.0). Representative gelsfrom three independent experiments are shown.

Figure 7 In vivo efficiency of polyplexes based on fractionated chitosanoligomers after intratracheal administration to mouse lungs. Polyplexes ofthe different oligomer fractions (DP10–14, DP15–21, DP22–35, DP36–50, DPn18 and DPn25) were formed at charge ratio 60:1 (7) and a dose of25 mg pDNA. PEI polyplexes at a previously optimized charge ratio of 5:1(7)2,12 were used as controls. The luciferase gene expression in mouselungs was analysed 72 h after administration. Results are expressed asmean values7s.d. (n¼ 4–6). *Po0.05, **Po0.01 and NS¼ no signifi-cant difference (P40.05).

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1446

Gene Therapy

points than the 72 h observed previously.12 Indeed, theluciferase gene expression peaked already at 24 h afterthe administration of DPn18 and DP15–21 polyplexes,expressing 695798 and 13647285 pg luciferase/mg oftotal lung protein, respectively, at 24 h compared to3517125 and 8607374 pg luciferase/mg of total lungprotein, respectively, at 72 h. The DP15–21 polyplexeswere as efficient as PEI polyplexes, based on AUC (areaunder the curve) calculations of the luciferase geneexpression profile 0–72 h (Table 3).

We also compared the distribution of the polyplexesof DPn18, DP15–21 and PEI, 24 h after administration. Tolocalize the distribution of gene expression in mouselungs after intratracheal administration, the luciferasegene expression was analysed in the trachea, bronchialtree and parenchyma. Independent of the deliverysystem, 490% of the luciferase gene expression waslocated in the bronchial tree (Figure 9). In this part of thelung, PEI polyplexes mediated a two-fold higher geneexpression than DP15–21 polyplexes (Po0.05) and afour-fold higher gene expression than DPn18 polyplexes(Po0.01), 24 h after administration.

Discussion

We developed more efficient chitosan-based gene deliv-ery systems with improved physical properties by usingnarrow size-distributed chitosan oligomers as complex-ing agents of plasmid DNA. The size range of thechitosan oligomers was selected from a previous studythat identified chitosan polyplexes based on oligomersof around 20 monomer units as promising gene deliverysystems.20 Indeed, our preliminary in vivo studies afterlung administration to mouse showed a maximal effi-ciency with polyplexes of a number average chain lengthof 18 (DPn18) (Figure 2d).

We found that a charge ratio as high as 60:1 (+/�) wasrequired for the formation of physically stable polyplexesbetween the polydisperse DPn18 and pDNA thatefficiently transfected cells in vitro and after lungadministration in vivo. The high charge excess agreeswith a recently reported charge ratio of 67:1 (+/�) foroptimized polyplexes of low-molecular-weight PEI.10,11,28

The necessity of such an excess of positive charges for thelow-molecular-weight oligomers, in comparison tocharge ratios p5:1 (+/�) used for high-molecular-weight

Figure 8 Polyplex stability against DNase I degradation. (a) Effect ofDPn18 polyplex charge ratio (10:1, 30:1 and 60:1 (7)) on the protectionagainst DNase I degradation. Naked pDNA was used as control. (b)Stability of DP15–21 polyplexes (60:1 (7) against degradation by DNaseI). Naked pDNA and DPn18 polyplexes (60:1 (7)) were used as controls.(c) Effect of free DP15–21 chitosan oligomers on the protection againstDNase I degradation. Removal of free DP15–21 oligomers from thepolyplex solution (filtered) resulted in a lower protection against DNasedegradation. After incubation with DNase I, the polyplexes weredissociated with heparin and 100 ng pDNA was loaded for agarose gelretardation assays. The addition of DNase I and heparin, respectively, isindicated with positive signs. Arrows indicate the loading position, opencircle (o.c.) and supercoiled (s.c.) forms of pDNA. Representative gels fromthree independent experiments are shown.

Table 3 Efficiency of different chitosan polyplexes and PEIpolyplexes in mouse lung after intratracheal administration

Polyplex AUCa P-valueb Ratio chitosan/PEI

DP15-21c 27567710 NS 0.83DPn18c 11327212 o0.01 0.34UPCd 133761 o0.001 0.04PEIe 33207871 — 1.00

aAUC calculations (pg luciferase�days/mg of total lung protein)are based on the luciferase gene expression in mouse lungs 0 to 3days after intratracheal administration.bLevel of significant difference compared to PEI polyplexes (NSindicate no significant difference, P40.05).cDP15–21 and DPn18 polyplexes were used at charge ratio 60:1 (7).dUPC polyplexes were used at an optimized charge ratio of 3:1 (7)and data are obtained from Koping-Hoggard et al.12

ePEI polyplexes were used at an optimized charge ratio of 5:1 (7)2,12

and data presented are mean values of data obtained in this studyand data obtained from Koping-Hoggard et al.12

Figure 9 Location of luciferase gene expression in mouse lungs afterintratracheal administration of chitosan oligomer polyplexes. Polyplexes ofDP15–21 and DPn18 were formed at charge ratio 60:1 (7) and a dose of10 mg pDNA. PEI polyplexes at charge ratio 5:1(7) were used as control.At 24 h after intratracheal administration, lung tissues were dissected intoparenchyma, bronchial tree and trachea, and the luciferase gene expressionin the different parts was determined. Results are expressed as meanvalues7s.d. (n¼ 4–6). Statistical differences are denoted as *Po0.05 and**Po0.01, respectively. See also Table 3 for the comparison of efficiencyduring the 0–72 h time interval (expressed as AUC).

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1447

Gene Therapy

counterparts,12,17,29 can partly be explained by the lowerefficiency of shorter chain polycations to complex withthe negatively charged phosphate groups of DNA.20,30,31

Furthermore, the DPn18 oligomer is polydisperse (6–50mers) and contains oligomers of chain lengths that formboth physically unstable and stable polyplexes.20 Thissuggests that the charge ratio of this mixture of chitosanoligomers should be regarded as an apparent chargeratio and that a free fraction of chitosan oligomers ispresent in the polyplex solutions. In fact, preliminarypurification studies indicate that 60–70% of the chitosanoligomers were unbound in the chitosan polyplexsolutions (Issa et al, unpublished results).

The high excess of low–molecular-weight chitosan inthe polyplexes may imply an increased toxicity of thedelivery system. However, in agreement with polyplexesof low-molecular-weight PEI at a charge ratio of 67:1(+/�),11 no acute toxicity of chitosan oligomer poly-plexes at charge ratio 60:1 (+/�) could be detected in theMTT assay in vitro. Furthermore, histological investiga-tions of the mouse lung showed no structural damage orimmune cell infiltration. This is in agreement withprevious studies, which showed that the intratrachealadministration of PEI polyplexes at a charge ratio of5:1 (+/�) did not result in an inflammatory response,while higher charge ratios did.32 We therefore concludethat the toxicity of the oligomers is not an issue in thepresent study.

In order to obtain better defined oligomer polyplexes,we separated the DPn18 oligomer into more narrowsized fractions, that is, DP10–14, DP15–21, DP22–35 andDP36–50.

In agreement with previous findings, polyplexes thatwere physically stable in the agarose gel retardationassay mediated higher gene expression in 293 cellsin vitro compared to those that separated in the gelassay.5,10,12,13,20,33 However, in contrast to the qualitativerelationship obtained previously between the transfec-tion efficiency in vitro in 293 cells and in vivo in mouselung,12,20 no clear relationship between the in vitro and invivo efficiency was obtained with the chitosan oligomersin the present study.

The most likely reason for the discrepancy betweenthe in vitro and in vivo efficiency of the less stable DP15–21 polyplexes may be that in vitro transfection actuallyrequires relatively more stable polyplexes for efficientpresentation to the cell surface in comparison to the invivo situation where the more concentrated polyplexsuspension is delivered directly to the epithelial liningof the airways. This is in agreement with the reportedmolecular-weight-dependent efficiency of PEI in vitroand in vivo, where more stable polyplexes of highermolecular weights are preferred in vitro,34 but a lowermolecular weight in vivo.4,35

Obviously, the high in vivo efficiency of the DP15–21polyplexes has to be explained by factors other than thephysical stability in the gel assay. Thus, we tested thehypothesis that the in vivo efficiency was directly relatedto the ability of the chitosan to protect pDNA againstdegradation by DNase. The impact of protecting DNAagainst degradation by extracellular DNase present inthe lung fluid, on the efficiency of gene expression afterintratracheal administration, was recently found toincrease the efficiency of naked DNA 50- to 80-fold.36,37

We therefore studied the stability of DPn18 and DP15–21

polyplexes against DNase at a DNase concentration(1.0 U) corresponding to those found in the mouse lung(o0.6 U).36 Indeed, our finding that the luciferase geneexpression in mouse lung increased with increasingstability of polyplexes against degradation by DNase(Figures 2a and 8a), confirmed that DNA degradation byDNase is a major barrier to efficient in vivo geneexpression in the lung, and indicated that chitosanoligomers provided complete protection against thisdegradation at the optimal charge ratio 60:1 (+/�).Interestingly, we found a decreased protection againstDNase when the excess of unbound chitosan oligomersin the DP15–21 polyplex solution was removed, suggest-ing that the protection against DNase degradationdepends more on the amount of chitosan and/or thecharge of the polyplexes than the physical stability asdetermined by the gel retardation assay.

Both our in vitro and in vivo transfection data supportthe hypothesis that dissociation of the oligomer poly-plexes is a faster process for the intracellular release ofpDNA from the polyplexes and the subsequent onset ofaction compared to a biodegradation-dependent release.First, the weaker binding of the chitosan oligomers topDNA resulted in dissociation of the polyplexes andrelease of pDNA upon incubation with anionic heparin(Figure 3a). This is in sharp contrast to the behaviour ofchitosan polyplexes of higher molecular weights, whichonly release pDNA after enzymatic degradation of thehigh-molecular-weight chitosan into shorter oligomersby chitosanase.12,33 Therefore, the high polyplex stabilityshown by high–molecular-weight chitosans has beenconcluded to be a major rate limiting step for theintracellular release of pDNA from these polyplexes,leading to a later onset of gene expression.12,17 Second,the in vivo luciferase gene expression reached a Cmax ofaround 700 and 1500 pg/mg of total lung protein alreadyat 24 h after the intratracheal administration of DPn18and DP15–21 polyplexes to the mouse lung, respectively.This is 120- and 260-fold higher than the valuespreviously reported for high-molecular-weight chitosanat this early time point.12 Overall, the oligomer poly-plexes in this study showed a 10- to 20-fold higher in vivoefficiency than high-molecular-weight chitosan reportedpreviously (Table 3).12

The relationship between chitosan polyplex formula-tion, DNase protection and release of pDNA on the invivo efficiency of chitosan-based pDNA complexes afterlung administration is schematically summarized inFigure 10.

From the molecular weight of the chitosans, it can becalculated (using Huggins equation38) that the low-molecular-weight chitosans used in this study display a10-fold reduced viscosity compared to the high-molecu-lar-weight UPC at concentrations used for in vivo studies.A further improvement of oligomer-based polyplexescompared to those based on higher molecular weights isexpected when these delivery systems are administratedas aerosols. The 10-fold lower viscosity of the concen-trated polyplex solutions of the oligomers compared tohigher molecular weight counterparts results in a morepreferable size range of the generated aerosol droplets(Koping-Hoggard et al, unpublished results).

The importance of dissociation of the oligomerpolyplexes for the release of pDNA, compared todegradation, finds further support in the fact that the

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1448

Gene Therapy

enzymatic degradation rate of chitosan by lysozyme ishighly dependent on the degree of acetylation (DA) asthis rate increases proportionally to DA in the fourthpower.39,40 Since the chitosan oligomers in the presentstudy are fully deacetylated (DAo0.001) the degradationrate is negligible, which strongly agrees with ourproposal that dissociation is more important thandegradation for the release of pDNA from thesecomplexes. Indeed, using the assumptions applied byZabner et al,41 we calculated that complete dissociation ofthe oligomer polyplexes would increase the endosomalosmolarity to 4 M, which is far above that required forendosomal rupture and release of the endosomal cargointo the cytoplasm.

In conclusion, polyplexes based on well-defined andnarrow size-distributed chitosan oligomers fulfill manyof the requirements of an efficient gene delivery systemto the lung tissue. Their superior efficiency in compar-ison to conventional high-molecular-weight chitosan-based polyplexes is a result of their weaker associationwith pDNA and the retained capacity of the chitosanoligomers to protect pDNA against DNase degradation,which enables an efficient release and delivery of theintact transgene. The results of this study forwardoligomeric chitosans as biocompatible nonviral genedelivery systems and present a platform for furtheroptimization studies of chitosan-based gene deliverysystems, for example, with regard to steric stabilizationand targeting.

Materials and methods

MaterialsA plasmid (gWizTMLuc) of GMP grade containing acytomegalovirus promoter (CMV) and a firefly luciferasereporter gene (pCMV-Luc) was purchased from Aldev-ron, Fargo, ND, USA. UPC, Protasan UPG 210, batchnumber 902-572-05, was obtained from Pronova Bio-

polymer, Oslo, Norway. PEI (25 kDa) was obtained fromAldrich Sweden, Stockholm, Sweden.

CellsThe human embryonic kidney epithelial cell line 293 andthe human cervix epithelial cell line HeLa were obtainedfrom ATCC, Rockville, MD, USA, and the human airwayepithelial cell line Calu-3 was a kind gift from Dr UrsulaHultkvist-Bengtsson, AstraZeneca R&D Lund, Sweden.The cells were regularly assayed and found to be freefrom mycoplasm infections and were maintained accord-ing to the supplier’s recommendations.

Preparation and characterization of chitosan oligomersA fully deacetylated chitosan (FAo0.001) was depoly-merized with nitrous acid42 to obtain samples withnumber average DPn of 25 and 18, respectively, asdetermined by 13C-NMR.43 The DPn18 sample wassubsequently fractionated by Superdex 30 gel filtrationas described previously,25 and fractions corresponding toDP intervals of 10–14, 15–21, 22–35 and 36–50 werepooled (Figure 1a). The chain length distributions wereanalysed by size-exclusion chromatography with amultiangle laser light scattering detector (SEC-MALLS).The results are summarized in Table 4. The resultsdemonstrated that the Superdex fractions had number(DPn) and weight average (DPw) degrees of polymeriza-tion (residues per chain) that were very close to thosecalculated for the corresponding subpopulations (DPintervals) of a Kuhn distribution with DPn¼ 25. Theexperimentally determined polydispersities wereslightly higher than the theoretical values, which weattributed to some peak overlap with oligomers elutingjust before and after the peaks selected for analysis. Wealso note that 13C-NMR gave lower DPn values thanSEC-MALLS, probably reflecting the difficulty of accu-rately determining the area of minor NMR signals. Thegood agreement between theoretical and experimentalDP values obtained by SEC-MALLS adds credibility tothis method.

Fractionated, monodisperse 6- and 12-mer chitosans(denoted as DP6 and DP12, respectively) were obtainedas described previously.20

Formulation of polyplexesChitosan stock solutions (2 mg/ml) were prepared bydissolving chitosan in sterile deionized MilliQ water, pH6.270.1, and then filtering the solutions under sterileconditions. Chitosan polyplexes were formulated byadding chitosan and then pDNA stock solutions to thesolvent under intense stirring on a vortex mixer(Heidolph REAX 2000, level 4, Kebo Lab, Spanga,Sweden). The following amounts of the differentchitosans were used per mg pDNA to prepare chitosanpolyplexes at a charge ratio of 1:1 (+/�):0.60 mg of DP10–14, DP15–21, DP22–35, DP36–50, DPn18 and DPn25 aschloride salts, 0.68 mg of DP6 and DP12 as acetate saltsand 1.05 mg of UPC as glutamate salts. The charge ratiowas defined as the ratio between the maximum numberof protonable primary amines in chitosan and thenumber of negative phosphates on pDNA.44

For in vitro studies (ie gel retardation and stabilityassays, analysis of particle size and morphology, andtransfection experiments), different charge ratios of

Figure 10 A schematic contour plot presentation of the relationshipbetween chitosan polyplex formulation, DNase protection and release ofpDNA on the in vivo efficiency of chitosan-based pDNA complexes afterlung administration. The efficiency is viewed as the product of the releaseof pDNA and the protection against DNase degradation as evaluated inthe agarose gel retardation assay. Each curve corresponds to the same levelof efficiency. Polyplex formulations that at the same time exhibit highrelease and high protection of pDNA, for example, DP15–21 polyplexes atcharge ratio 60:1 (7), give the highest gene expression in vivo after lungadministration.

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1449

Gene Therapy

chitosan polyplexes were produced at a constant pDNAconcentration (13.3 mg/ml).

Chitosan polyplexes were prepared at pDNA concen-trations of 5, 10 and 25 mg/ml in 1 ml sterile MilliQ waterfor use in in vivo studies. The polyplexes wereconcentrated in a SpeedVac Plus centrifuge (SavantInstruments, Holbrook, NY, USA) at 1400 rpm forapproximately 90 min to obtain DNA concentrationsof around 50, 100 and 250 mg/ml, respectively.12 PEI(25 kDa) stock solutions (10 mM) and PEI polyplexeswere prepared as described previously.1 An optimalcharge ratio of 5:1 (+/�) was used throughout.2,12

Physicochemical characterization of chitosanpolyplexesThe size of the polyplexes was determined by photoncorrelation spectroscopy (Zetasizer 4000, Malvern Instru-ments, Malvern, UK) and their zeta potentials (whichindicate the surface charge) were determined by electro-phoretic light scattering (Zetasizer 2000, Malvern Instru-ments). The physical stability of the complexes wasstudied using the agarose gel retardation assay (0.8%agarose in 40 mM TAE buffer, pH 8.5) as describedpreviously.12 In some agarose gel retardation experi-ments, indicated in the text, TAE buffer at pH 5.0 and 7.4was used. The stability of the polyplexes was alsostudied in the gel retardation assay after incubating thepolyplexes with 5 mg/ml heparin (Sigma, St Louise, MO,USA) for 2 h at room temperature. Protection ofcomplexed pDNA against DNase degradation wasstudied after incubating polyplexes in the presence orabsence of 1 U DNase I (Ambion, Austin, TX, USA) for15 min as described previously.45 After incubation,polyplexes were dissociated with heparin (5 mg/ml)and the integrity of pDNA was examined using theagarose gel retardation assay. pDNA obtained from thestock solution was used as control.

Morphological characterization of chitosan polyplexesChitosan polyplexes were imaged by tapping modeatomic force microscopy, employing a Digital InstrumentMultimode IIIa equipped with an E scanner (maximum

xy range B15 mm, Digital Instruments, Santa Barbara,CA, USA). Dried specimens of the samples to be imagedwere prepared by deposition of a 10 ml aliquot of theaqueous solution of polyplexes on a freshly cleaved5 mm diameter mica surface and incubated for 2 min.The samples were then dried with a stream of N2 gas(P¼ 1.5 bar), followed by vacuum drying (P¼ 10�9 bar)for at least 2 h. For the immobilization of uncomplexedDNA on mica, 0.5 mM ZnCl2 was added to the solutionimmediately before transferring the solution to themica.46 Tapping mode silicon nitride cantilevers TESP(Digital Instruments), with nominal spring constants of20–100 N/m and nominal resonant frequencies of 200–400 kHz, were used for the imaging. The drive frequencyused was less than the free oscillation resonancefrequency of the cantilever, yielding 90% of the ampli-tude at the resonance frequency. Topographs wereobtained with scan sizes in the range 1.5–2.0 mm (datacollection at 512� 512 pixels) and a scan speed ofapproximately 1.5 Hz. Quantitative information wasextracted from the AFM topographs by image analysis.Based on the calculation of a shape factor for eachpolyplex structure, reflecting the symmetry of thestructure about the three axis of rotation, the polyplexeswere sorted into three different classes (toroidal, rod-likeand spherical shapes).47,48

In vitro gene transferMost in vitro transfection studies were performed in theepithelial human embryonic kidney cell line 293. Thecells (45 000 cells/cm2) were seeded at 70% confluence in96-well tissue culture plates (Costar, Cambridge, UK)24 h before transfection. For transfection studies per-formed at pH 7.4, serum-free medium (Optimem, Gibco/BRL Life Technologies AB, Taby, Sweden) was added topolyplexes formulated in MilliQ water (having a final pHof around 5). For transfection studies performed at pH 5,acetate buffer (25 mM, pH 5) was used.20 Isotonicity(300 mOsm/kg) was obtained by the addition of manni-tol. The cells were washed in preheated Optimem and50 ml of the complex formulations (corresponding to0.33 mg pDNA) was added per well. After 5 h incubation,

Table 4 Characterization of chitosan oligomers in the present study

Sample code Preparation method 13C-NMRa SEC-MALLSb

DPn DPw DPn DPw/DPn

DPn25c d 25 78 33 2.36DPn18e d 18 31 25 1.22DP10–14 f 14 (12.1) 13 (11.9) 1.04 (1.01)DP15–21 f 20 (18.1) 19 (17.8) 1.03 (1.01)DP22–35 f 27 (28.4) 26 (27.8) 1.03 (1.02)DP36–50 f 44 (42.7) 40 (42.2) 1.09 (1.01)

aNumber average degree of polymerization (DPn) as determined by 13C-NMR.43

bWeight average (DPw) and the number average (DPn) were analysed by size-exclusion chromatography with a multiangle laser lightscattering detector (SEC-MALLS). The ratio DPw/DPn indicates the polydispersity.cDPn25 chitosan oligomer fraction was found to be comprised of oligomers ranging from 13- to 280-mers.dA fully deacetylated chitosan (FAo0.001) was depolymerized by nitrous acid42 to obtain samples with number average DPn of 25 and 18,respectively, as determined by 13C-NMR.43

eDPn18 chitosan oligomer fraction was found to be comprised of oligomers ranging from 6- to 50-mers.fDPn18 was subsequently fractionated by Superdex 30 gel filtration as described previously,25 and fractions corresponding to DP intervalsof 10–14, 15–21, 22–35 and 36–50 were pooled. The figures within parentheses are the theoretical values for corresponding subfractions ina Kuhn distribution with DPn¼ 25.

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1450

Gene Therapy

the formulations were removed and 0.2 ml of the freshculture medium was added. The medium was changedevery second day for experiments that exceeded 2 days.At the time points indicated, the cells were washed withphosphate-buffered saline (PBS) (pH 7.4), lysed withluciferase lysis buffer (Promega, Madison, WI, USA) andthe emitted light over 8 s was measured with aluminometer (Mediators PhL, Vienna, Austria).12 Theamount of luciferase expressed was determined from astandard curve prepared with firefly luciferase (Sigma, StLouis, MO, USA) and total cell protein was measuredusing the bichinchoninic acid test (Pierce, Rockford, IL,USA).

Intracellular dehydrogenase activityThe effect of polyplexes on intracellular dehydrogenaseactivity (a measure of cellular toxicity) in 293 cells wasdetermined by the MTT method as described pre-viously.12,49

Intratracheal administration to mouse lungsThe animal experiments were approved by The SwedishNational Board for Laboratory Animals (the local ethicalcommittee in Uppsala, Sweden). BALB/c mice aged 6–8weeks were anaesthetized with ketamin/xylazine (5/20 vol%, 0.1 ml/10 g of body weight), and the tracheawas surgically exposed by a 0.5 cm long skin incision inthe neck. Polyplexes in a volume of 100 ml were slowlyinjected into the trachea with a 28 G needle in two 50 mlportions. The mice were killed (CO2) 24, 48 and 72 h later,the lungs were removed, washed in ice-cold PBS,homogenized in a Beadbeater (Biospec Products, Bartles-ville, OK, USA) for 1 min in ice-cold luciferase lysisbuffer (Promega) with a protease inhibitor cocktail(Complete, Boehringer Mannheim Scandinavia AB,Bromma, Sweden), centrifuged at 15 000 rpm at 41C,mixed with luciferase reagent (Promega) and theluciferase gene expression was determined as describedfor the in vitro expression analysis above.

In order to determine the distribution of the luciferasegene expression in mouse lungs, lungs were removed,placed on ice and dissected into parenchyma, bronchialtree and trachea. A blunt scalpel was used to carefullyscrape off the lung parenchyma, leaving a network ofbronchi and bronchioles. Samples were then processedand assayed for luciferase gene expression as describedabove.

HistologyThe lung histology was evaluated in blind observationsfor the appearance of inflammatory cells and structuraldamage as described previously.50 Mice were killed 24 hafter the intratracheal administration of polyplexes, andafter cardiac perfusion with PBS and 3% paraformalde-hyde in PBS, the lungs were removed, rinsed in PBS andstored overnight in 3% paraformaldehyde. The lungswere frozen the next day in OCT Embedding Medium(Sakura Finetek Europe, Zoeterwoude, The Nether-lands). Cryosections (5 mm) were cut in a Leica JungCM 3000 cryostat (Leica Instruments GmbH, Nussloch,Germany), stained briefly with haematoxylin and eosin,mounted and examined under a light microscope.

StatisticsThe experiments were performed on a minimum of twooccasions using quadruplicate samples each time. Thegene expression is presented as the amount of theexpressed transgene per mg (in vitro) or mg (in vivo) oftissue protein. All data are expressed as mean values71standard deviation (s.d.). Statistical differences betweenmean values were investigated using ANOVA. Statisticaldifferences between the physical shapes of the poly-plexes were investigated using contingency tablesanalysis. Statistical differences are denoted as *Po0.05,**Po0.01 and ***Po0.001, respectively.

Acknowledgements

We thank Dr Kristoffer Tommeraas for skilful technicalassistance with the preparation of the chitosan oligomerfractions. This work was supported by the SwedishBoard for Technical Development (NUTEK) Grant no.p11381/1, by the Norwegian Research Council Grantnos. 134674/140, 129104/420 and 1218941420, andthrough a PhD grant to MI from the Egyptian State.

References

1 Boussif O et al. A versatile vector for gene and oligonucleotidetransfer into cells in culture and in vivo: polyetylenimine. ProcNatl Acad Sci USA 1995; 92: 7297–7301.

2 Bragonzi A et al. Biodistribution and transgene expression withnonviral cationic vector/DNA complexes in the lungs. GeneTherapy 2000; 7: 1753–1760.

3 Tang MX, Redemann CT, Szoka Jr FC. In vitro gene delivery bydegraded polyamidoamine dendrimers. Bioconjug Chem 1996; 7:703–714.

4 Turunen MP et al. Efficient adventitial gene delivery to rabbitcarotid artery with cationic polymer–plasmid complexes. GeneTherapy 1999; 6: 6–11.

5 Gebhart CL, Kabanov AV. Evaluation of polyplexes as genetransfer agents. J Control Rel 2001; 73: 401–416.

6 Luo D, Saltzman WM. Synthetic DNA delivery systems. NatBiotechnol 2000; 18: 33–37.

7 Godbey WT, Wu KK, Mikos AG. Poly(ethylenimine)-mediatedgene delivery affects endothelial cell function and viability.Biomaterials 2001; 22: 471–480.

8 Putnam D, Gentry CA, Pack DW, Langer R. Polymer-based genedelivery with low cytotoxicity by a unique balance of side-chaintermini. Proc Natl Acad Sci USA 2001; 98: 1200–1205.

9 Regnstrom K et al. PEI-a potent, but not harmless, mucosalimmuno-stimulator of mixed T-helper cell response and FasL-mediated cell death in mice. Gene Therapy 2003; 10: 1575–1583.

10 Fischer D et al. A novel non-viral vector for DNA delivery basedon low molecular weight, branched polyethylenimine: effect ofmolecular weight on transfection efficiency and cytotoxicity.Pharm Res 1999; 16: 1273–1279.

11 Kunath K et al. Low-molecular weight polyethylenimine as anon-viral vector for DNA delivery: comparison of physico-chemical properties, transfection efficiency and in vivo distributionwith high-molecular-weight polyethylenimine. J Control Rel 2003;89: 113–125.

12 Koping-Hoggard M et al. Chitosan as a nonviral gene deliverysystem. Structure–property relationships and characteristicscompared with polyethylenimine in vitro and after lungadministration in vivo. Gene Therapy 2001; 8: 1108–1121.

13 Lee M et al. Water-soluble and low molecular weight chitosan-based plasmid DNA delivery. Pharm Res 2001; 18: 427–431.

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1451

Gene Therapy

14 Thanou M et al. Quaternized chitosan oligomers as novel genedelivery vectors in epithelial cell lines. Biomaterials 2002; 23: 153–159.

15 Corsi K, Chellat F, Yahia L, Fernandes JC. Mesenchymal stemcells, MG63 and HEK293 transfection using chitosan–DNAnanoparticles. Biomaterials 2003; 24: 1255–1264.

16 Koping-Hoggard M et al. Chitosan–pDNA polyplex: in vivo geneexpression after tracheal, nasal and oral administration to mice.PharmSci 1998; 1: S–278.

17 MacLaughlin FC et al. Chitosan and depolymerized chitosanoligomers as condensing carriers for in vivo plasmid delivery.J Control Rel 1998; 56: 259–272.

18 Roy K, Mao HQ, Huang SK, Leong KW. Oral gene delivery withchitosan–DNA nanoparticles generates immunologic protectionin a murine model of peanut allergy. Nat Med 1999; 4: 387–391.

19 Kumar M et al. Intranasal gene transfer by chitosan–DNANanospheres protects BALB/c mice against acute respiratorysyncytial virus infection. Hum Gene Ther 2002; 13: 1415–1425.

20 Koping-Hoggard M et al. Relationship between the physicalshape and the efficiency of oligomeric chitosan as a genedelivery system in vitro and in vivo. J Gene Med 2003; 5: 130–141.

21 Iqbal M et al. Nasal delivery of chitosan–DNA plasmidexpressing epitopes of respiratory syncytial virus (RSV)induces protective CTL responses in BALB/c mice. Vaccine2003; 21: 1478–1485.

22 Varum KM, Ottoy MH, Smidsrod O. Water-solubility of partiallyN-acetylated chitosans as a function of pH: effect of chemicalcomposition and depolymerisation. Carbohydr Polym 1994; 25:65–70.

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

24 Schaffer DV, Fidelman NA, Dan N, Lauffenburger DA. Vectorunpacking as a potential barrier for receptor-mediated polyplexgene delivery. Biotechnol Bioeng 2000; 67: 598–606.

25 Tommeraas K, Varum KM, Christensen BE, Smidsrod O.Preparation and characterisation of oligosaccharides producedby nitrous acid depolymerisation of chitosans. Carbohydr Res2001; 333: 137–144.

26 Pack DW, Putnam D, Langer R. Design of imidazole-containingendosomolytic biopolymers for gene delivery. Biotechnol Bioeng2000; 67: 217–223.

27 Guo X, Han IS, Yang VC, Meyerhoff ME. Homogeneous enzyme-based binding assay for studying glycosaminoglycaninteractions with macromolecules and peptides. Anal Biochem1996; 235: 153–160.

28 Bieber T, Elsasser HP. Preparation of a low molecular weightpolyethylenimine for efficient cell transfection. Biotechniques2001; 30: 74–77; 80–81.

29 Erbacher P et al. Chitosan-based vector/DNA complexes forgene delivery: biophysical characteristics and transfectionability. Pharm Res 1998; 15: 1332–1339.

30 Tsuchida E. Formation of polyelectrolyte complexes and theirstructures. JMS-Pure Appl Chem 1994; A31: 1–15.

31 Kabanov VA, Kabanov AV. Supramolecular devices for targetingDNA into cells: fundamentals and perspectives. Macromol Symp1995; 98: 601–613.

32 Ferrari S et al. ExGen 500 is an efficient vector for gene deliveryto lung epithelial cells in vitro and in vivo. Gene Therapy 1997; 4:1100–1106.

33 Mao HQ et al. Chitosan–DNA nanoparticles as gene carriers:synthesis, characterization and transfection efficiency. J ControlRel 2001; 70: 399–421.

34 Godbey WT, Wu KK, Mikos AG. Size matters: molecular weightaffects the efficiency of poly(ethylenimine) as a gene deliveryvehicle. J Biomed Mater Res 1999; 45: 268–275.

35 Abdallah B et al. A powerful nonviral vector for in vivo genetransfer into the adult mammalian brain: polyethylenimine.Hum Gene Ther 1996; 7: 1947–1954.

36 Glasspool-Malone J, Malone RW. Marked enhancement of directrespiratory tissue transfection by aurintricarboxylic acid. HumGene Ther 1999; 10: 1703–1713.

37 Glasspool-Malone J et al. DNA transfection of macaque andmurine respiratory tissue is greatly enhanced by use of anuclease inhibitor. J Gene Med 2002; 4: 323–332.

38 Brandrup J, Immergut EH. Polymer Handbook, 2nd edn. JohnWiley & Sons: New York, 1975, p 1315.

39 Nordtveit RJ, Varum KM, Smidsrod O. Degradation of fullywater-soluble, partially N-acetylated chitosans with lysozyme.Carbohydr Polym 1994; 23: 253–260.

40 Nordtveit RJ, Varum KM, Smidsrød O. Degradation of partiallyN-acetylated chitosans with hen egg white and humanlysozyme. Carbohydr Polym 1996; 29: 163–167.

41 Zabner J et al. Cellular and molecular barriers to gene transfer bya cationic lipid. J Biol Chem 1995; 270: 18997–19007.

42 Allan GG, Peyron M. Molecular weight manipulation of chitosanI: kinetics of depolymerization by nitrous acid. Carbohydr Res1995; 277: 257–272.

43 Varum KM et al. Determination of enzymatic hydrolysisspecificity of partially N-acetylated chitosans. Biochim BiophysActa 1996; 1291: 5–15.

44 Felgner PL et al. Nomenclature for synthetic gene deliverysystems. Hum Gene Ther 1997; 8: 511–512.

45 Gebhart CL et al. Design and formulation of polyplexes based onpluronic–polyethylenimine conjugates for gene transfer.Bioconjug Chem 2002; 13: 937–944.

46 Hansma HG, Laney DE. DNA binding to mica correlates withcationic radius: assay by atomic force microscopy. Biophys J 1996;70: 1933–1939.

47 Noguchi H, Yoshikawa K. Folding path in a semiflexiblehomopolymer chain: a Brownian dynamics simulation. J ChemPhys 2000; 113: 854–862.

48 Maurstad G, Danielsen S, Stokke B. Analysis of compactedsemiflexible polyanions visualized by atomic force microscopy.Influence of chain stiffness on the morphologies ofpolyelectrolyte complexes. J Phys Chem B 2003; 107: 8172–8180.

49 Lappalainen K et al. Comparison of cell proliferation andtoxicity assays using two cationic liposomes. Pharm Res 1994;11: 1127–1131.

50 Rudolph C et al. In vivo gene delivery to the lung usingpolyethylenimine and fractured polyamidoamine dendrimers.J Gene Med 2000; 2: 269–278.

Chitosan oligomers as nonviral gene delivery systemsM Koping-Hoggard et al

1452

Gene Therapy