Characterizing DNA Condensation by Structurally Different Chitosans

Apr 9, 2009 - the structure-function relationship have shown that both DNA condensation and gene transfer efficacy strongly depend on structural varia...
0 downloads 0 Views 3MB Size
1508

Biomacromolecules 2009, 10, 1508–1515

Characterizing DNA Condensation by Structurally Different Chitosans of Variable Gene Transfer Efficacy Nina K. Reitan,*,† Gjertrud Maurstad,† Catharina de Lange Davies,† and Sabina P. Strand‡ Departments of Physics and Biotechnology, The Norwegian University of Science and Technology, 7491 Trondheim, Norway Received January 26, 2009; Revised Manuscript Received March 17, 2009

Chitosan can be used as a nonviral gene delivery vector for which DNA condensation and transfection efficacy strongly depend on structural parameters. In this study, we characterized the condensation of DNA by three molecularly tailored chitosans, including linear, trisaccharide substituted-, and self-branched trisaccharide substituted chitosan oligomers. No significant differences could be detected in the hydrodynamic diameters formed by the various chitosans as analyzed by dynamic light scattering. However, atomic force microscopy revealed that selfbranched chitosan formed complexes with a higher ratio of globules to rods, and the heights of both globules and rods were larger than for complexes formed by the other chitosans. Using an amino/phosphate ratio of 10, fluorescence correlation spectroscopy measurements showed that self-branched chitosan exhibited a lower fraction (30%) of bound chitosan than the other chitosans. YOYO-1 was a superior fluorescent DNA-label compared to Cy5 and PicoGreen, since labeling with YOYO-1 had least effect on the size and structure of the complexes.

Introduction The success of gene therapy depends on efficient delivery of exogenous DNA and involves several steps, including penetration through the extracellular matrix to target cells, intracellular uptake, endosomal escape, trafficking to the nucleus, and efficient transcription of the transgene. The size and charge of the DNA molecule impede several of these transport processes. In addition, exogenous DNA is rapidly degraded by nucleases, which are present in both the intracellular and extracellular environment. Successful gene delivery and cell transfection therefore depend on vectors that condense DNA and protect it from degradation.1 Although viral gene delivery vectors have so far yielded higher transfection efficacy, safety concerns have drawn attention to nonviral vectors, such as cationic lipids or polymers. Polycations condense the negatively charged DNA by electrostatic interactions. However, most of these transfection reagents exhibit significant cytotoxicity.2,3 Chitosan, a biopolymer derived from chitin, has been shown to efficiently compact DNA and protect it from degradation by nucleases.4 Chitosan is a linear polysaccharide composed of (1 f 4)-β-linked N-acetyl-D-glucosamine (acetylated units) and D-glucosamine (de-N-acetylated units) residues. The interest in chitosan for gene delivery and biomedical research is steadily increasing5,6 due to its superior biocompatibility compared to other polycations, like poly(ethylenimine) (PEI) or poly-L-lysine.7,8 Chitosans may be regarded as a family of copolymers with highly variable composition and properties. This may explain the large differences in efficacy of chitosan-based gene delivery systems reported in the literature. Several studies addressing the structure-function relationship have shown that both DNA condensation and gene transfer efficacy strongly depend on structural variables, such as the fraction of acetylated units (FA), the degree of polymerization (DP), and polydispersity.7,9-13 * Corresponding author. E-mail: [email protected]. † Department of Physics. ‡ Department of Biotechnology.

Furthermore, the amino group of GlcN may be easily derivatized, and a broad range of modified chitosans with improved properties has been prepared.8,14,15 Recently, it was reported that glycosylated chitosan oligomers with a GlcNAc-containing trisaccharide denoted AAM (TCO ) trisaccharide substituted chitosan oligomer) improve gene delivery efficacy compared to unsubstituted oligomers.8,14 Further tailoring of the molecular architecture led to self-branched glycosylated oligomers (SBTCO) with improved functional properties and high transfection efficacy in vitro.8 However, the reasons for the increased gene transfer efficacy of SB-TCO still remain unclear. The purpose of the present study was to characterize and compare DNA complexation by three molecularly tailored chitosans showing large variations in transfection efficacy:8 linear chitosan oligomers (LCO), TCO, and SB-TCO. Dynamic light scattering (DLS) and atomic force microscopy (AFM) were used to characterize the size and morphology of the complexes. Fluorescence correlation spectroscopy (FCS) is a technique for measuring diffusion and molecular interactions in solution,16 intracellularly,17,18 or in the interstitium.19,20 In this study, FCS was used to study the mobility of the complexes, as well as interactions between DNA and the various chitosans in solution, and this method also determined the extent of unbound chitosan. To employ FCS, both DNA and chitosan must be labeled by suitable fluorophores that do not alter the properties of the complexes. The effects of different DNA labels (Cy5, PicoGreen and YOYO-1) on the size and structure of DNA-chitosan complexes were therefore evaluated.

Materials and Methods Chitosan and DNA. Stock solutions of DNA pWizLuc plasmids (6.7 kb) (Aldevron, Fargo, ND) were prepared at a concentration of 0.5 mg/mL in sterile Milli-Q grade water. The chitosans used in this study were prepared in our laboratory as previously described.8,21 Briefly, LCO was prepared by nitrous acid degradation of completely deacetylated chitosan (FA < 0.002) to an average degree of polymerization (DPn) of 31. TCO was prepared by reductive amination of LCO with the trimer 2-acetamido-2-deoxy-D-glucopyranosyl-β-(1-4)-2-

10.1021/bm900102d CCC: $40.75  2009 American Chemical Society Published on Web 04/09/2009

DNA Condensation by Structurally Different Chitosans Table 1. Characteristics of Chitosansa chitosan

d.s.

Mn

Mw

PdI

LCO TCO SB-TCO

7.8 7.8

6300 7100 13000

8000 9800 21000

1.28 1.37 1.59

a The molecular weights, Mw (g/mol) and Mn (g/mol), and polydispersity index (PdI ) Mw/Mn) were analyzed by size-exclusion chromatography and multiangle laser scattering detector (SEC-MALLS). The degree of substitution (d.s.) of AAM was determined by 1H NMR.21

acetamido-2-deoxy-D-glucopyranosyl-β-(1-4)-2,5-anhydro-D-mannofuranose (AAM), as described previously.21 For SB-TCO, selfbranching of nonreduced linear oligomers was allowed to proceed along with AAM substitution. Properties of the chitosans used in this study are summarized in Table 1. Chitosans were dissolved at 1 mg/mL in Milli-Q grade water, sterile filtered (0.2 µm Millipore), and stored in aliquots at -20 °C. Preparation of DNA-Chitosan Complexes. Complexes were prepared by the self-assembly method to a final DNA concentration of 5 µg/mL. The required amount of DNA stock solution was diluted in Milli-Q grade water, and sterile filtered chitosan stock solution was added during intense stirring on a vortex mixer. The complexes were incubated for 30-40 min at room temperature. An amino/phosphate (A/P) ratio of 10 was used. Fluorescent Labeling. To label chitosans, Alexa Fluor 647 carboxylic acid, succinimidyl ester (Molecular Probes, Eugene, OR) was chosen due to its amine-reactive properties, high extinction coefficient, and photostability. Alexa Fluor 647 was not quenched in the proximity of DNA, as was observed for Alexa Fluor 633 (data not shown). Further experiments also confirmed that chitosans labeled with Alexa Fluor 647 retained the same ability to mediate transgene expression in HEK293 cells as their unlabeled counterparts (data not shown). The theoretical labeling degree has previously been optimized to approximately one molecule of dye per 200 GlcN residues. Unconjugated dye was removed by extensive dialysis and the labeled chitosans were freeze-dried. Three fluorescent dyes were tested for labeling of DNA. A Cy5 Nucleic Acid Labeling Kit (Mirus Bio Corp., WI) that covalently attaches Cy5 to DNA was used according to the protocol from the supplier. After incubation, the labeled sample was purified using microspin columns provided in the kit. Labeling of DNA with PicoGreen (Molecular Probes, Eugene, OR), which intercalates the DNA helix, was performed according to the protocol from the supplier. The DNA intercalating YOYO-1 (Molecular Probes, Eugene, OR) was used at a ratio of one dye molecule per 50 bp DNA. One molecule of YOYO-1 per 100 bp DNA gave insufficient fluorescence (data not shown). PicoGreen and YOYO-1 are high affinity intercalators that are practically nonfluorescent when not bound to DNA, and removal of any unbound dye was therefore not relevant. Dynamic Light Scattering. The size of DNA-chitosan complexes was determined by DLS on a Zetasizer Nano Zs (Malvern Instruments, U.K.). Three measurements with a measurement time of 400 s were acquired for each parallel sample at 25 °C. The size of the complexes was characterized by the mean diameter (z-average) and the polydispersity index (PdI), both obtained by cumulant analysis of correlation functions using the Dispersion Technology Software DTS 4.20 (Malvern Instruments, U.K.). In the cumulant analysis, the first cumulant (a1) is used to calculate the z-average diameter, and the polydispersity index is calculated by PdI ) 2a2/a12, where a2 is the second cumulant. The viscosity and refractive index of water were used in the calculations. The Zetasizer Nano Zs is equipped with a 633 nm laser line. To avoid absorption of laser light by Alexa Fluor 647, unlabeled chitosans were used in the DLS measurements. Cy5 did not absorb the 633 nm laser light to the same extent. Gel Electrophoresis. DNA was complexed with unlabeled chitosan and chitosan labeled with Alexa Fluor 647. The samples were run on a 0.8% agarose gel containing 0.5 µg/mL ethidium bromide (Sigma-

Biomacromolecules, Vol. 10, No. 6, 2009

1509

Aldrich, Saint Louis, Missouri) in 40 mM Tris-acetate-EDTA (TAE) buffer at pH 8 at 80 V for 60 min, and they were then visualized on a UV transilluminator. Atomic Force Microscopy. Samples were prepared by adding 10 µL DNA-chitosan solution at a DNA concentration of 5 µg/mL onto a mica disk. After 2 min, excess liquid was blown off with N2 gas, and the samples were incubated in a vacuum for 2 h. Tapping mode AFM topographs were acquired using a NanoScope IIIa (Digital Instruments, CA) equipped with an E-scanner. Tapping mode silicon cantilevers (Nanosensors, Switzerland) with nominal spring constants of 10-130 N/m were employed. Prior to extracting quantitative data, the topographs were flattened line by line after excluding the compacted structures. The asphericity index (A) was used as a shape factor for the complexes and was calculated from the AFM topographs by employing user-interactive software22 developed in the IDL language (Research System, Inc., Boulder, CO). Theoretical values of the asphericity index are A ) 0 for a globule, A ) 0.25 for a toroid, and A ) 1 for a rod.23 In the present study, visual inspection of AFM topographs confirmed the absence of toroids, and A < 0.4 and A > 0.4 represent ensembles of globules and rods, respectively. Heights of the linear and globular structures were analyzed using the “section” function in the NanoScope 6.13R1 software (Digital Instruments, CA). Fluorescence Correlation Spectroscopy. FCS measurements were performed using the LSM 510/Confocor2 (Carl Zeiss Jena GmbH, Germany) with a C-Apochromat 40×/1.2 W corr objective. YOYO-1 and PicoGreen were excited by a 488 nm Ar laser, and fluorescence was detected through a BP 530-600 (“blue channel”) filter and a pinhole of 70 µm. Cy5 was excited by a 633 HeNe laser and detected through an LP650 (“red channel”) filter and a pinhole of 90 µm. Laser powers were kept low to avoid artifacts like photobleaching and optical saturation.24 Photobleaching was further avoided by using a short measurement time (5 s). For quantitative measurements, statistics were improved by calculating the diffusion coefficient from the average correlation function of several (20) repetitive runs.25 For dual-labeled complexes, the prepared solutions of DNA-chitosan were further diluted (1:10) in Milli-Q grade water before the FCS measurements to obtain a sufficient signal from both Alexa Fluor 647 and YOYO-1. The basic principles of FCS have been previously described.26 In this study, fluorescence autocorrelation functions were fitted by the oneand two-component free diffusion model included in the Confocor2 software27 (see Supporting Information). The diffusion coefficient, D, is related to the measured characteristic diffusion time, τD, by

D)

2 ωxy 4τD

(2)

The radial dimension, ωxy, of the detection volume depends on the FCS instrumentation and was determined from eq 2 by extracting the diffusion time, τD, from FCS measurements of Rhodamine 6G (SigmaAldrich, St. Louis, MO; 488 nm excitation) and Cy5 (Amersham Biosciences, U.K.; 633 nm excitation), which have known diffusion coefficients of 2.8 × 10-6 and 3.2 × 10-6 cm2/s, respectively.27 Although the difference between the molecular weight of free chitosan and the complex is sufficient to differentiate their respective diffusion coefficients in an autocorrelation experiment, FCCS (fluorescence cross-correlation spectroscopy) was performed to confirm that YOYO-1 and Alexa Fluor 647 were colocalized and also that the additional diffusion coefficient for Alexa Fluor 647 indeed originated from the complex, not from clusters of free chitosan. The cross-talk contribution from the 488 nm laser was defined as the ratio of the average photon count rate in the red channel to the average photon count rate in the blue channel. Cross-talk was measured to less than 10% and was therefore neglected. To analyze FCCS data quantitatively, samples must be uniformly labeled with a defined number of fluorophores per DNA.28 Because this is not the case here, quantitative analysis was based on the autocorrelation functions. Statistics. A two-sample, two-tailed Student’s t test (Minitab, Minitab Inc., State College, PA), assuming nonequal variances, was used to

1510

Biomacromolecules, Vol. 10, No. 6, 2009

Reitan et al.

Table 2. DLS Measurements of the Mean Hydrodynamic Diameter (z-Average) and Polydispersity Index (PdI)a LCO DNA label

z-average (nm)

unlabeled Cy5 PicoGreen YOYO-1

69.4 ( 5.3 738.0 ( 112.0 51.3 ( 7.1 67.1 ( 14.8

TCO PdI 0.22 ( 0.11 0.35 ( 0.04 0.15 ( 0.03 0.22 ( 0.09

z-average (nm)

PdI

61.4 ( 4.1 185.6 ( 63.6 59.4 ( 10.3 63.0 ( 14.2

0.14 ( 0.06 0.18 ( 0.07 0.18 ( 0.05 0.17 ( 0.05

a Calculated from the cumulants analysis for complexes formed by unlabeled and fluorescently labeled DNA. Unlabeled chitosans (LCO and TCO) were used. Mean values and standard deviations are based on 3-6 parallel samples.

Figure 1. Tapping mode AFM topographs of LCO (a-d) and TCO (e-h) complexed with unlabeled DNA (a,e) and DNA labeled with Cy5 (b,f), PicoGreen (c,g), and YOYO-1 (d,h). Arrows in (c) show uncomplexed DNA.

perform statistical comparison of the Gaussian-distributed data with a significance criterion of p e 0.05.

Results Influence of Fluorescent Labels on Structure of DNA-Chitosan Complexes. To study DNA-chitosan interactions using fluorescent techniques, DNA and chitosan must be labeled with two different and spectrally separable dyes. Such labeling may alter the properties of the complexes. To choose the most suitable DNA label, three fluorescent dyes (Cy5, PicoGreen, and YOYO-1) were compared with regard to their effect on size and structure of DNA complexed with LCO and TCO. Table 2 shows the mean hydrodynamic diameters of complexes formed by unlabeled chitosan and labeled and unlabeled DNA. Only DNA labeled with YOYO-1 formed complexes that were not significantly different in size from unlabeled complexes for both chitosans, whereas Cy5 increased the mean hydrodynamic diameter approximately 3- and 10-fold for TCO and LCO, respectively. The morphology of the complexes was investigated by AFM. AFM topographs showed that Cy5-labeled DNA complexed with chitosan (Figure 1b and f) formed larger structures than complexes that were unlabeled (Figure 1a and e) or labeled with either PicoGreen (Figure 1c and g) or YOYO-1 (Figure 1d and h). The Cy5-labeled complexes also tended to form clusters,

especially complexes formed by TCO, and a large excess of free chitosan was seen. PicoGreen-labeled DNA complexed with LCO and TCO formed structures resembling unlabeled complexes. However, uncomplexed DNA was observed in the AFM topographs, which was not the case for complexes of unlabeled DNA. For both chitosans, DNA labeled with YOYO-1 formed well-defined complexes with resembling structures and the same ratio of globules to rods (data not shown) as for unlabeled complexes. To further characterize the molecular size and aggregation of complexes labeled with the various dyes, FCS analysis was performed. For complexes formed by Cy5-labeled DNA, the fluorescence intensity time traces showed large peaks and indicated more severe aggregation for LCO compared to TCO (Figure 2a and d). Furthermore, FCS measurements revealed several disadvantages of PicoGreen-labeled complexes (Figure 2b and e), such as poor signal-to-noise ratios caused by low autocorrelation amplitudes. Furthermore, a significant amount of molecules were in the triplet state. There were also indications of photobleaching (data not shown), even at low laser powers. DNA labeled with YOYO-1 (Figure 2c and f), on the other hand, showed high autocorrelation amplitudes without severe aggregation for both LCO and TCO. Based on the comparison of unlabeled and fluorescently labeled complexes by the various techniques, YOYO-1 was chosen as the most appropriate DNA dye.

DNA Condensation by Structurally Different Chitosans

Biomacromolecules, Vol. 10, No. 6, 2009

1511

Figure 2. Intensity time traces (upper panels) and autocorrelation curves (lower panels) from FCS of LCO (a-c) and TCO (d-f) complexed with DNA labeled with Cy5 (a,d), PicoGreen (b,e), and YOYO-1 (c,f). Arrow in (a) shows a large peak caused by aggregation. Arrows in (e) show correlation amplitude (solid arrow) and triplet decay or other processes (dotted arrow).

Characterization of DNA-Chitosan Complexes Formed by Structurally Different Chitosans. For further characterization of DNA-chitosan complexes, including all three chitosan oligomers (LCO, TCO, and SB-TCO), DNA was fluorescently labeled with YOYO-1, and chitosan was labeled with Alexa Fluor 647. There was no significant difference between the z-average hydrodynamic diameters measured by DLS for SBTCO complexed with unlabeled DNA (z ) 76.7 ( 11.0 nm) and with DNA labeled with YOYO-1 (z ) 64.7 ( 0.6 nm). Further, the hydrodynamic diameter of SB-TCO complexes was not significantly different from that of complexes formed by LCO and TCO (Table 2). A gel retardation assay was performed to compare the stability of complexes formed by unlabeled chitosan and chitosan labeled with Alexa Fluor 647. The DNA retardation pattern was similar for labeled and unlabeled chitosans, but a slight decrease in stability of labeled complexes was observed (Figure 3). Both labeled and unlabeled SB-TCO formed more stable complexes than the other chitosans, whereas the TCO complexes released

Figure 3. Stability of DNA-chitosan complexes evaluated by gel retardation assay for DNA complexed with unlabeled chitosans (a) and chitosans labeled with Alexa Fluor 647 (b).

DNA most easily. The plasmid preparation (pWizLuc) contained two distinct conformations of DNA, a supercoil fraction (top band) and an open circular form (lower band).

1512

Biomacromolecules, Vol. 10, No. 6, 2009

Reitan et al. Table 3. Heights of Unlabeled DNA-Chitosan Complexes Measured from AFM Topographsa height (nm) structure

LCO

TCO

SB-TCO

rods globules

5.7 ( 2.3 10.9 ( 5.8

6.7 ( 2.3 12.5 ( 3.9

8.4 ( 2.2 18.0 ( 7.1

a

Figure 4. Normalized distributions of asphericity indices calculated from AFM topographs of unlabeled DNA-chitosan complexes formed by LCO, TCO, and SB-TCO. The total numbers of observations, n, are indicated.

The morphology of the complexes was analyzed quantitatively from AFM topographs. LCO, TCO, and SB-TCO formed linear and globular structures, and only a few toroids were seen. All three chitosans formed a majority of globular complexes with asphericity indices centered at A ∼ 0.25 (Figure 4). The asphericity distribution showed that SB-TCO formed a higher fraction (0.84) of globular structures than LCO (0.58) and TCO (0.61). AFM topographs also revealed that the average heights of SB-TCO were larger than for LCO and TCO for both globules and rods (Table 3). Interactions between chitosan and DNA, as well as the diffusive properties of both free components and complexes, were investigated by FCS and FCCS. Autocorrelation experiments were acquired for diffusion measurements of free chitosan and free DNA (Figure 5). As expected, the experimental autocorrelation functions for free chitosan were well fitted by

Mean values and standard deviations are based on 37-189 structures.

Figure 5. One-component free diffusion model (s) was successfully fitted to the experimental autocorrelation curve (×) for free SB-TCO (representative for all three chitosans) labeled with Alexa Fluor 647 (a). For free DNA labeled with YOYO-1 (b), the one-component model (----) failed, whereas a two-component model (s) gave a better fit to the experimental autocorrelation curve (×). Residuals are shown in lower panels.

the one-component free diffusion model (Figure 5a), and the calculated diffusion coefficients indicate that SB-TCO diffuses slower than the other two chitosans (Table 4). For measurements of free DNA, the one-component model failed (Figure 5b), whereas the two-component model gave a satisfactory fit, with a fast and a slow diffusion coefficient of 5.5 ( 1.4 × 10-8 and 0.4 ( 0.1 × 10-8 cm2/s, respectively. This is in accordance with results from gel electrophoresis, which showed two distinct bands for free DNA (Figure 3), indicating two populations of different sizes. The fraction of the fast component of free DNA measured with FCS was 0.60 ( 0.15. The dual-labeled complexes were analyzed by FCCS. Although a high excess of free chitosans was observed in all solutions, fluorescence intensity time traces from cross-correlation experiments (Figure 6) showed colocalized peaks in the red channel (Alexa Fluor 647) and the blue channel (YOYO-

DNA Condensation by Structurally Different Chitosans

Biomacromolecules, Vol. 10, No. 6, 2009

1513

Table 4. Diffusion Coefficients (10-8 cm2/s) for Free Chitosan and DNA-Chitosan Complexesa red channel sample

D1

D2

blue channel F2

D3

Free Chitosans LCO TCO SB-TCO

71.9 ( 0.1 61.1 ( 0.5 40.2 ( 0.2

DNA-Chitosan Complexes DNA-LCO 71.9b DNA-TCO 61.1b DNA-SB-TCO 40.2b

6.0 ( 1.5 0.51 ( 0.03 5.2 ( 0.8 0.42 ( 0.08 4.5 ( 0.9 0.30 ( 0.06

7.4 ( 3.0 4.9 ( 1.2 4.4 ( 0.9

a Chitosan was labeled with Alexa Fluor 647 (emission red channel) and DNA with YOYO-1 (emission blue channel). Mean values and standard deviations are based on 3–7 measurements. For the dual-labeled complex, a fast (D1) and slow (D2) diffusion coefficient were detected in the red channel. The fraction, F2, of the slow component indicates the fraction of bound chitosan. b D1 measured by the red channel for complexes was fixed to be equal to D1 for free chitosan.

Figure 6. Intensity time traces from FCCS experiments showing colocalized signals from Alexa Fluor 647 (red channel) and YOYO-1 (blue channel) bound to complexes. The figure is representative for all chitosans.

1), demonstrating that the two dyes both fluoresce when attached to the complexes. Autocorrelation curves for chitosan (red channel) and DNA (blue channel) forming complexes were obtained from the FCCS measurements. Best fits to the experimental autocorrelation curves for free and complexed chitosan and complexed DNA are shown in Figure 7. The one-component free diffusion model (not shown) was poorly fitted to the autocorrelation function for complexes detected in the red channel. A two-component free diffusion model was adequate, indicating two species diffusing with different rates. The fast component was recognized as free chitosan, as no significant difference between the diffusion coefficient of this component and free chitosans was found. We therefore fixed the diffusion coefficient of the fast species to be equal to the diffusion coefficient of free chitosan when calculating the diffusion coefficient of the slow species, D2, and the bound fraction, F2, of chitosan (Table 4). The DNA-chitosan complexes were also characterized by detecting YOYO-1 fluorescence in the blue channel. In this case, the one component free diffusion model was well fitted to the autocorrelation curve (Figure 7). This is in accordance with AFM images showing that free DNA was practically absent in samples with complexes. No significant difference could be found between the diffusion coefficient, D3, of the complex

Figure 7. Best fits to the autocorrelation functions for free chitosan (SB-TCO, representative for all chitosans used in this study) and complexes of SB-TCO and DNA; free chitosan labeled with Alexa Fluor 647 (dotted line, one-component fit), dual-labeled complex detected in the red channel (dashed line, two-component fit), and dual-labeled complex detected in the blue channel (solid line, onecomponent fit). For comparison, the curves were normalized to the autocorrelation curve for free SB-TCO.

measured in the blue channel and the slow component, D2, detected in the red channel (Table 4). Compared to free DNA at the same DNA concentration, the average fluorescence for complexes was halved, whereas the brightness (photon counts per molecule) and the correlation amplitude increased 3- and 10-fold, respectively. To investigate the properties of DNA-chitosan complexes in a more relevant physiological environment, the complexes prepared in Milli-Q grade water were further diluted in Hank’s balanced salt solution pH 7.4 (HBSS) to obtain an isotonic formulation suitable for use in cells in vitro. FCS measurements performed three minutes after adding the media revealed severe aggregation, as detected by large peaks in the fluorescence intensity time traces (Figure 8). The aggregation was detected in all three chitosan formulations, and no significant differences for the various chitosans could be detected.

Discussion Studies of nonviral gene delivery strive to establish a correlation between a vector’s structure and its performance in order to allow rational design of novel gene delivery systems. In agreement with this concept, three structurally distinct chitosans were chosen to study their ability to condense DNA using DLS, AFM, and FCS. LCO, TCO, and SB-TCO represent three generations of molecularly designed chitosan oligomers that have been gradually improved to increase their gene delivery efficacy.8 Labeling of DNA and chitosan with fluorophores may affect the condensation of DNA, thereby affecting the intracellular fate of the complexes. To label DNA, high affinity intercalators offer several advantages over covalently linked dyes, as they do not chemically modify the DNA and are simple to use. Intercalation of these dyes into the DNA helix results in large fluorescence enhancements.29 However, upon DNA condensation, the intercalating dyes are expelled from the DNA helix, causing a decrease in the fluorescence intensity. For this reason, intercalating dyes are frequently used to measure the degree of compaction of DNA by condensing agents.12,13,30 In our case, YOYO-1 showed sufficient fluorescence after condensation with chitosans.

1514

Biomacromolecules, Vol. 10, No. 6, 2009

Figure 8. Count rate traces detected in the red channel (a) and the blue channel (b and c) for DNA-chitosan complexes formed by SBTCO (representative for all chitosans used in this study) diluted in Milli-Q grade water (gray) and HBSS (black). A change in photon count rate scale of (b) is shown in (c) for better comparison of average fluorescence baselines.

In the present study, the effect that various fluorescent dyes (Cy5, PicoGreen, and YOYO-1) had on the DNA-chitosan complexes was evaluated. The size and morphology of the complexes were least altered by YOYO-1, which was chosen as the fluorophore for labeling DNA in further measurements. To obtain a good spectral separation of the dyes used for duallabeling of the complex, Alexa Fluor 647 was chosen to label chitosan. AFM topographs showed no clear difference between structures formed by unlabeled chitosan and chitosan labeled with Alexa Fluor 647. The gel retardation assay revealed that labeling chitosans with Alexa Fluor 647 led to less stable complexes, but this was expected considering the relatively large molecular weight of the conjugated dye and the low molecular weight of the chitosans used.

Reitan et al.

Following the choice of labels, the size, morphology, and mobility of DNA-chitosan complexes formed by LCO, TCO, and SB-TCO were compared. Both DLS and AFM showed that all three chitosans formed complexes of similar size and shape when reconstituted in Milli-Q grade water. It has been reported that condensation of DNA by chitosan induces the formation of rods, toroids, and globules,11,12 and the rod has been found to be the most thermodynamically stable structure for chitosan with FA ) 0.15.30 The fraction of the different structures reportedly depends on the particular chitosan used to compact DNA, with sensitivity to both charge density and chain length of the chitosan, solution parameters such as ionic strength and pH, and A/P ratio.11,12,31 In this study, all of the chitosans compacted DNA into rods or globules, and practically no toroids were observed. This is in accordance with other studies where a number of low molecular weight chitosans produced either a very small fraction of toroids or no toroids at all upon complexation with DNA.9,12,31 The different shapes of the complexes may play a role in intracellular trafficking of polyplexes. A previous study has suggested that the toroid seems to be an unfavorable structure for transfection purposes.12 It has also been shown that two fractionated LCOs with low polydispersity mediating the highest gene transfer to mouse lung in vivo formed significantly higher proportions of globules.9 Of the three chitosans compared in this study, SB-TCO formed the highest fraction of globules and it has been shown to have the highest gene transfection efficiency.8 It may thus be hypothesized that the globular structure is the more favorable for gene delivery by chitosans. The condensation of DNA is known to depend on both the polycation valence and its structure.32 We therefore propose that the increase in heights of the DNA-chitosan complexes is due to the modifications of the chitosan structure. The increased branching is suggested to lead to stearic hindrance of the interaction between the DNA and chitosan, preventing strong condensation of DNA and thus giving rise to complexes with larger heights. This is consistent with reports on DNA-PEI complexes, where it has been found that PEI condenses DNA less tightly than smaller condensing agents16 and that the condensation of DNA with branched PEI gives rise to complexes with larger heights than when DNA is compacted by linear PEI.33 The lower diffusion coefficient (Table 4) for free SBTCO further demonstrates that substitution and self-branching changes linear chitosan into a branched polymer that occupies a larger volume. The diffusion coefficients obtained for the DNA-chitosan complexes were in the same size order as previously found for DNA-PEI complexes.16 Fluorescence autocorrelation curves for free DNA labeled with YOYO-1 were fitted by the two-component free diffusion model. This is probably attributed to the presence of supercoiled and open circular forms of DNA, as also evident from gel electrophoresis. No open structure conformation was detected, confirming the integrity of the DNA preparation. Another explanation for the two-component fit may be that the large, elongated structure of free DNA causes violation of the Gaussian approximation, whereas condensed DNA diffuses through the detection volume as a compact and more well-defined particle than free DNA. Both the autocorrelation amplitude, which is inversely proportional to the number of fluorescent molecules, and the brightness of these molecules increased when free DNA was condensed by chitosans. This indicates that each complex was formed by more than one plasmid, giving rise to brighter structures containing a higher number of fluorophores.

DNA Condensation by Structurally Different Chitosans

The formulation of SB-TCO was found to contain the largest fraction of free chitosan compared to LCO and TCO. This may be attributed to a higher condensation efficiency resulting from the higher molecular weight of the chitosan due to a higher degree of cooperativity of DNA-chitosan interactions. Therefore, longer chitosan chains exhibit more efficient binding, and fewer molecules are necessary to condense DNA. Previous characterization of PEI-DNA complexes at a nitrogen/ phosphate (N/P) ratio of 10 (equivalent to A/P 10) indicated more than 80% free polycation,16 which is a concern due to the toxicity of PEI. On the other hand, a high fraction of free polycations may improve the transgene delivery due to enhanced buffering capacity leading to endosomal escape of the complexes. According to the “proton sponge” hypothesis, the H+buffering capacity of the polyamines causes accumulation of Cl-, thereby inducing swelling and lysis of the endosome/ lysosome.34 Consistent with these findings, removal of free PEI and chitosan has been shown to reduce the transfection efficacy.9,35 Thus, the high fraction of free SB-TCO may be advantageous for endosomal release, thereby contributing to the higher transfection efficacy compared to LCO and TCO. FCS measurements performed in HBSS demonstrated that complexes formed by all chitosans equally aggregated when pH and osmomolality were adjusted to physiological levels. Complexes formed by SB-TCO have been previously shown to be colloidaly stable in PBS,8 but this effect was observed at higher A/P of 30. Despite the difference in colloidal stability at A/P 10 and 30, SB-TCO exhibits similar transfection efficacy at both A/P 10 and 30.8 This implies that aggregation of complexes does not impair the gene delivery in vitro.

Conclusions Three structurally different tailored chitosans exhibiting large differences in gene transfer efficacy were used to study condensation of DNA. SB-TCO has previously demonstrated a higher transfection efficacy than LCO and TCO,8 and comparing complexes formed by LCO, TCO, and SB-TCO is a necessary step toward understanding chitosan-mediated gene transfer. This study revealed that, besides differences in the stability of complexes, SB-TCO and DNA formed structures with a larger height and a larger fraction of globular structures compared to the other chitosans. Also, complexes formed by SB-TCO contained a larger fraction of unbound chitosan, which may lead to increased transfection. As a next step, we will use the same characterized complexes to investigate uptake and intracellular trafficking in human cells to identify intracellular barriers of efficient transgene delivery, and we intend to correlate such barriers with the other features of the chitosans. Acknowledgment. This work is supported by The Norwegian Research Council (Grant Nos. 166794/V30 and 182695/I40). We gratefully acknowledge Edrun Andrea Schnell, Florian Mumm, and Kristin Sæterbø, Department of Physics, NTNU, Norway, for assistance in performing experiments. Supporting Information Available. FCS multicomponent free diffusion fitting model. This material is available free of charge via the Internet at http://pubs.acs.org.

Biomacromolecules, Vol. 10, No. 6, 2009

1515

References and Notes (1) Nguyen, D. N.; Green, J. J.; Chan, J. M.; Langer, R.; Anderson, D. G. AdV. Mater. 2008, 20, 1–21. (2) Hunter, A. C. AdV. Drug DeliVery ReV. 2006, 58, 1523–1531. (3) Lv, H. T.; Zhang, S. B.; Wang, B.; Cui, S. H.; Yan, J. J. Controlled Release 2006, 114, 100–109. (4) Richardson, S. C. W.; Kolbe, H. V. J.; Duncan, R. Int. J. Pharm. 1999, 178, 231–243. (5) Prabaharan, M. J. Biomater. Appl. 2008, 23, 5–36. (6) Issa, M. M.; Ko¨ping-Ho¨ggård, M.; Artursson, P. Drug DiscoVery Today: Technol. 2005, 2, 1–6. (7) Ko¨ping-Ho¨ggård, M.; Tubulekas, I.; Guan, H.; Edwards, K.; Nilsson, M.; Vårum, K. M.; Artursson, P. Gene Ther. 2001, 8, 1108–1121. (8) Strand, S. P.; Issa, M. M.; Christensen, B. E.; Vårum, K. M.; Artursson, P. Biomacromolecules 2008, 9, 3268–3276. (9) Ko¨ping-Ho¨ggård, M.; Vårum, K. M.; Issa, M.; Danielsen, S.; Christensen, B. E.; Stokke, B. T.; Artursson, P. Gene Ther. 2004, 11, 1441– 1452. (10) Lavertu, M.; Me´thot, S.; Tran-Khanh, N.; Buschmann, M. D. Biomaterials 2006, 27, 4815–4824. (11) Danielsen, S.; Vårum, K. M.; Stokke, B. T. Biomacromolecules 2004, 5, 928–936. (12) Danielsen, S.; Strand, S.; Davies, C. de L.; Stokke, B. T. Biochim. Biophys. Acta, Gen. Subj. 2005, 1721, 44–54. (13) Strand, S. P.; Danielsen, S.; Christensen, B. E.; Vårum, K. M. Biomacromolecules 2005, 6, 3357–3366. (14) Issa, M. M.; Ko¨ping-Ho¨ggård, M.; Tømmeraas, K.; Vårum, K. M.; Christensen, B. E.; Strand, S. P.; Artursson, P. J. Controlled Release 2006, 115, 103–112. (15) Park, I. K.; Ihm, J. E.; Park, Y. H.; Choi, Y. J.; Kim, S. I.; Kim, W. J.; Akaike, T.; Cho, C. S. J Controlled Release 2003, 86, 349– 359. (16) Clamme, J. P.; Azoulay, J.; Me´ly, Y. Biophys. J. 2003, 84, 1960– 1968. (17) Schwille, P.; Haupts, U.; Maiti, S.; Webb, W. W. Biophys. J. 1999, 77, 2251–2265. (18) Clamme, J. P.; Krishnamoorthy, G.; Me´ly, Y. Biochim. Biophys. Acta, Biomembr. 2003, 1617, 52–61. (19) Alexandrakis, G.; Brown, E. B.; Tong, R. T.; McKee, T. D.; Campbell, R. B.; Boucher, Y.; Jain, R. K. Nat. Med. 2004, 10, 203–207. (20) Reitan, N. K.; Juthajan, A.; Lindmo, T.; Davies, C. de L. J. Biomed. Opt. 2008, 13, 054040. (21) Tømmeraas, K.; Ko¨ping-Ho¨ggård, M.; Vårum, K. M.; Christensen, B. E.; Artursson, P.; Smidsrød, O. Carbohydr. Res. 2002, 337, 2455– 2462. (22) Maurstad, G.; Danielsen, S.; Stokke, B. T. J. Phys. Chem. 2003, 107, 8172–8180. (23) Noguchi, H.; Yoshikawa, K. J. Chem. Phys. 1998, 109, 5070–5077. (24) Gregor, I.; Patra, D.; Enderlein, J. ChemPhysChem 2005, 6, 164–170. (25) Krichevsky, O.; Bonnet, G. Rep. Prog. Phys. 2002, 65, 251–297. (26) Magde, D.; Elson, E.; Webb, W. W. Phys. ReV. Lett. 1972, 29, 705– 708. (27) Weisshart, K.; Jungel, V.; Briddon, S. J. Curr. Pharm. Biotechnol. 2004, 5 (2), 135–154. (28) Bacia, K.; Schwille, P. Nat. Protoc. 2007, 2, 2842–2856. (29) Glazer, A. N.; Rye, H. S. Nature 1992, 359, 859–861. (30) Danielsen, S.; Maurstad, G.; Stokke, B. T. Biopolymers 2005, 77, 86– 97. (31) Maurstad, G.; Danielsen, S.; Stokke, B. T. Biomacromolecules 2007, 8, 1124–1130. (32) Plum, G. E.; Arscott, P. G.; Bloomfield, V. A. Biopolymers 1990, 30, 631–643. (33) Itaka, K.; Harada, A.; Yamasaki, Y.; Nakamura, K.; Kawaguchi, H.; Kataoka, K. J. Gene Med. 2004, 6, 76–84. (34) Sonawane, N. D.; Szoka, F. C.; Verkman, A. S. J. Biol. Chem. 2003, 278, 44826–44831. (35) Boeckle, S.; von Gersdorff, K.; van der Piepen, S.; Culmsee, C.; Wagner, E.; Ogris, M. J. Gene Med. 2004, 6, 1102–1111.

BM900102D