Effect of integrin targeting and PEG shielding on polyplex ... - CiteSeerX

GENE DELIVERY

Journal of Controlled Release 156 (2011) 364–373

Contents lists available at ScienceDirect

Journal of Controlled Release j o u r n a l h o m e p a g e : w w w. e l s ev i e r. c o m / l o c a t e / j c o n r e l

Effect of integrin targeting and PEG shielding on polyplex micelle internalization studied by live-cell imaging F.M. Mickler a, b, Y. Vachutinsky c, g, M. Oba d, K. Miyata e, N. Nishiyama e, K. Kataoka c, e, f, C. Bräuchle a, b, N. Ruthardt a, b,⁎ a

Department of Chemistry and Biochemistry and Center for NanoScience (CeNS), Ludwig-Maximilians-Universität München, Butenandtstr. 5-13, D-81377 München, Germany Center for Integrated Protein Science Munich (CIPSM), 81377 München, Germany Department of Bioengineering, Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8656, Japan d Department of Clinical Vascular Regeneration, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan e Center for Disease Biology and Integrative Medicine, Graduate School of Medicine, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-0033, Japan f Department of Materials Engineering, Graduate School of Engineering, The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113–8656, Japan g InSightec Ltd.5 Nahum Heth Street, Tirat Carmel 39120, Israel b c

a r t i c l e

i n f o

Article history: Received 11 July 2011 Accepted 2 August 2011 Available online 6 August 2011 Keywords: Integrin targeting Polyplex micelles Live-cell imaging Shielding Gene therapy

a b s t r a c t αvβ3 and αvβ5 integrins are attractive target structures for cancer therapy as they are upregulated in tumor and tumor associated host cells and play a pivotal role for tumor growth and metastasis. Gene vectors such as polyplex micelles consisting of thiolated PEG-block-poly(lysine) copolymers complexed with plasmid DNA can be targeted to these specific integrins by equipment with a cyclic RGD peptide. In this study, we analyzed the effect of the RGD ligand on micelle endocytosis by comparing fluorescently labeled, targeted and untargeted micelles in live-cell imaging experiments with highly sensitive fluorescence microscopy and flow cytometry. Two micelle types with 12 kDa (PEG12) and 17 kDa (PEG17) PEG shell layers were examined to evaluate the influence of surface shielding on the internalization characteristics. Our results reveal three major effects: First, the RGD ligand accelerates the internalization of micelles into integrin expressing HeLa cells without changing the uptake pathway of the micelles. Both targeted as well as untargeted micelles are predominantly internalized via clathrin mediated endocytosis. Second, the PEG shielding of micelles has an important effect on their targeting specificity. At high PEG shielding selective endocytosis of integrin targeted micelles occurs, whereas at low PEG shielding targeted and untargeted micelles show comparable internalization. In addition, PEG17 RGD(+) micelles induce the highest reporter gene expression. Third, our data demonstrate a clear influence of the applied micelle dose on the internalization of integrin targeted micelles. We propose that PEG17 shielded micelles equipped with a cyclic RGD ligand are the favored system of choice for clinical therapy as they exhibit higher transgene expression, a higher specificity for integrindependent endocytosis compared to PEG12 shielded micelles, and are functional at low doses as well. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Gene therapy is increasingly recognized as a promising technique to treat cancer on a molecular level without severe side effects for the human body [1–4]. Non-viral, synthetic gene vectors enable the safe delivery of transgenes into the target tissue with reduced immunogenicity, and can be modified with custom-designed ligands to achieve specific tumor targeting [5,6]. Nevertheless, synthetic gene vectors are still far less effective in transfection compared to viral gene vectors [7]. Therefore, there is a growing demand to understand the ⁎ Corresponding author at: Department of Chemistry and Biochemistry and Center for NanoScience (CeNS), Ludwig-Maximilians-Universität München, Butenandtstr. 513, D-81377 München, Germany. Tel.: + 49 89 2180 77544; fax: + 49 89 2180 9977547. E-mail address: [email protected] (N. Ruthardt). 0168-3659/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.jconrel.2011.08.003

detailed mechanisms of targeting, cellular internalization, and intracellular processing of gene vectors in order to optimize their future design for in vivo applications. Polyplex micelles that are composed of cationic block copolymers complexed with plasmid DNA are synthetic gene vectors that have the potential for application in clinical therapy [3,8,9] (see Fig. 1). They possess a suitable size of approximately 100 nm for systemic delivery and are shielded with a biocompatible poly(ethylene glycol) (PEG)-shell layer to increase their stability in the serum and to reduce unspecific interactions with non-target components [10,11]. By incorporation of disulfide-crosslinks into the micellar core, their stability in extracellular fluids is additionally enhanced [12]. These redox-sensitive cross-links are disrupted in the reducing environment of the cytosol, triggering the controlled liberation of plasmid DNA after endosomal release. Oba et al. demonstrated that micelles can be equipped with a cyclic RGD peptide, enabling selective targeting of

365

specificity that is required for a specific treatment of cancer cells with low side effects. 2. Results 2.1. Colocalization analysis of RGD(+) and RGD(−) micelles

Fig. 1. Schematic illustration of micelle formation between plasmid DNA and c(RGDfK)PEG-p(Lys-SH)-polymer. Micelles are formed through polyion complex formation between positively charged polylysine segments and negatively charged DNA. The charged micellar core is shielded by a PEG shell layer to which a cyclic RGD-peptide is attached as a targeting ligand. Covalent cross-linking of polylysine segments by disulfide bonds causes high stability of micelles.

micelles to αvβ5 and αvβ3 integrins on tumor cells [12,13]. αvβ5 and αvβ3 integrins are highly investigated target structures for cancer therapy because they are overexpressed in solid tumors as well as the angiogenic tumor vasculature, which facilitates the accumulation and extravasation of the therapeutic gene vectors at the tumor site [14–20]. Recent in vivo studies revealed that the treatment of adenocarcinoma bearing mice with integrin targeted polyplex micelles that encode for an antiangiogenic, therapeutic protein successfully inhibits tumor growth [21,22]. However, the mechanism of integrin mediated internalization of micelles into cancer cells is not completely understood so far. Therefore in this study, we investigated the effect of integrin targeting on micelle internalization in more detail by live-cell imaging experiments with highly sensitive fluorescence microscopy. This powerful technique enables the detection of single micelles with a high spatial and temporal resolution [23–26]. This way, the cellular localization of different micelle types can be followed in real time. To directly visualize the effect of the RGD ligand, we coincubated targeted (RGD(+)) micelles and untargeted (RGD(−)) micelles with different fluorescent labels onto HeLa cells and analyzed their time-dependent cellular localization. To evaluate the influence of surface shielding on the selectivity of integrin targeting, two micelle types with differently sized PEG shell layers were compared. PEG12 micelles were equipped with a 12 kDa PEG, whereas PEG17 micelles contained an elongated 17 kDa PEG resulting in enhanced shielding of the positively charged micelles core [21]. Our results reveal that the RGD ligand alters the uptake kinetics of micelles, resulting in enhanced accumulation of RGD(+) micelles in perinuclear endosomes and retention of RGD(−) micelles on the cell membrane. In contrast, the endocytic pathway of micelles was not affected by the RGD ligand, as observed in colocalization experiments with pathway specific markers and inhibitor experiments. Targeted as well as untargeted micelles were predominantly internalized into clathrin coated vesicles. We showed for the first time that the selectivity of integrin targeting is strongly affected by the quality of PEG shielding. Whereas the RGD ligand had only marginal effect on the internalization of PEG12 micelles, the uptake of PEG17 micelles was significantly increased in the presence of the RGD ligand. Interestingly, we observed a strong effect of the applied micelle dose on the internalization behavior of micelles, especially in the case of less shielded PEG12 micelles. This suggests the involvement of ligand induced integrin clustering for effective micelle internalization. Finally, the in vitro reporter gene expression of cells treated with integrin targeted and untargeted PEG12 and PEG17 micelles was determined and compared to our single cell observations. RGD(+) micelles resulted in the highest transgene expression in the presence of the elongated PEG17 that can be correlated to the enhanced targeting specificity in the presence of proper shielding. Overall, our data demonstrate that PEG17 micelles contain favorable characteristics for clinical therapy as they combine high transgene expression with a high targeting

In order to directly compare the internalization of integrin targeted (RGD(+)) and untargeted (RGD(−)) micelles without knowing details of their internalization pathway, we simultaneously applied both micelle types with different fluorescent labels onto HeLa cells and analyzed their cellular localization over time. Due to crosslinking of the polymer chains by disulfide bonds, micelles are highly stable and mixing of RGD-positive and RGD-negative polymer between different micelles should rarely occur [8]. Directly after application of the micelles and homogeneous distribution on the cell, little colocalization is expected. An increase in the colocalization degree should emerge in case that both micelle types are internalized into the cell and transported to the same endocytic compartment. As endosomal fusion is a statistical process, reference values for the time dependent degree of colocalization were obtained by control measurements with two identical micelle types differing only in the fluorescence label. Internalization of RGD(+) and RGD(−) micelles was then compared to the reference values. To evaluate the importance of surface shielding for effective receptor targeting, we compared two differently shielded micelle types in our experiments. PEG12 micelles were shielded with 12 kDa PEG, whereas PEG17 micelles were equipped with longer 17 kDa PEG resulting in enhanced shielding of the positively charged micelle core [21]. For the measurement, HeLa cells were incubated with a low dose of premixed micelles (2.5 ng of DNA per 10,000 cells) allowing the detection and subsequent quantification of single micelles on the cell surface by highly sensitive wide-field microscopy. We assume that at this low concentration, micelles are rarely taken up in clusters but are rather incorporated singly into distinct endosomes. To provide a defined starting point for micelle incubation, the cellular medium was removed leaving a thin fluid film on the cell surface prior to application. Micelle addition resulted in immediate contact of the micelles with the cells. One minute after micelle addition, the removed medium was resupplied. Short movies of single cells were recorded 0, 2, 4 and 6 h after micelle addition and analyzed for colocalization of simultaneously applied micelles. For evaluation of the colocalization degree exclusively data points from one measurement day were plotted to reduce the variability in between single cells and improve comparability. However, the effects described in the following text were reproducible and have been observed in over 40 cells at different measurement days. For reference measurements a mixture of Cy3 and Cy5 labeled RGD(−) PEG12 micelles was added to the cells (Fig. 2). As shown in the fluorescence overlay images in Fig. 2, separate, non-aggregated micelles were uniformly distributed all over the cell during the first minutes post micelle addition. After two hours, we observed a shift of micelle distribution from the cell periphery toward the central part of the cell. The total number of detected spots was reduced while their intensity was increased. In addition, colocalizing spots represented by the white label in the overlay images appeared, indicating incorporation of several micelles in intracellular compartments. Within two to six hours of incubation, the fraction of colocalizing spots and the accumulation in the nuclear proximity further increased. Unexpectedly, the same pattern was observed for coincubated RGD(+) and RGD(−) micelles with PEG12 shielding. We observed transport of both micelle types to the cell center comparable to the reference measurement and increasing colocalization over time. In contrast, after coincubation of RGD(+) and RGD(−) micelles with PEG17 shielding, we detected a separation in localization of targeted and untargeted PEG17 micelles during the

GENE DELIVERY

F.M. Mickler et al. / Journal of Controlled Release 156 (2011) 364–373

GENE DELIVERY

366


Fig. 2. Cellular localization of coincubated RGD(+) and RGD(−) micelles with different PEG shielding. Cy3 and Cy5 labeled micelles were simultaneously applied at low concentration (2.5 ng DNA per 10,000 cells) onto HeLa cells and imaged by wide-fieldmicroscopy after 0, 2, 4 and 6 h with alternating laser excitation. Fluorescence overlay images were obtained by superposition of the Cy3 and the Cy5 channels. Colocalizing endosomes appear in white. The region of the cell nucleus as well as the cell membrane is both marked with a dashed, yellow line. (A) Cy3 and Cy5 labeled PEG12 micelles without a targeting ligand (= reference cells) were transported into the same endocytic compartments, resulting in increasing colocalization over time. (B) RGD(+) (green) and RGD(−) (magenta) PEG12 micelles show a similar time-dependent localization compared to the reference. (C) Coincubation of PEG17 shielded RGD(+) (green) and RGD(−) (magenta) micelles resulted in a separated cellular distribution. RGD(+) micelles were transported to the nuclear proximity, whereas RGD(−) micelles were predominantly retained in the cell periphery. Scale bar: 10 μm.

Fig. 3. Quantification of the time-dependent colocalization degree of coincubated micelles. HeLa cells were coincubated with different combinations of fluorescently labeled micelles and imaged by wide-field fluorescence microscopy at the indicated time points. Obtained movies were analyzed for colocalizing endosomes using customwritten software. Each data point represents the calculated colocalization degree in one camera section corresponding to one or two HeLa cells. Plotted data points were approximated by linear regression (gray, dashed line). White circle: reference cells coincubated with Cy3 and Cy5 labeled PEG12 micelles without a targeting ligand for determination of normal distribution of colocalization, black diamond: cells coincubated with targeted (RGD(+)) and untargeted (RGD(−)) PEG12 micelles, gray square: cells coincubated with RGD(+) and RGD(−) PEG17 micelles.

detected spots in the overlay images indicates the enrichment of multiple micelles in endocytic compartments during incubation. Our results demonstrate that the RGD ligand significantly affects the subcellular distribution of micelles. However, this effect was strongly dependent on the employed PEG shielding and was only observed with the longer 17 kDa PEG. With the shorter 12 kDa PEG shielding cells were indistinguishable from reference measurements without RGD ligand.

2.2. Effect of dosage on micelles internalization measurement (Fig. 3C). After six hours, in the majority of the cells, PEG17 RGD(−) micelles were retained in peripheral section of the cell, whereas PEG17 RGD(+) micelles accumulated in the nuclear proximity. To quantify the degree of colocalization for the various micelle combinations, recorded movies were analyzed using customdesigned software. As displayed in Fig. 3, calculated colocalization values were plotted over time and were best approximated by linear regression. The reference measurements revealed a constant increase of the colocalization degree over time with an R-Squared value of 0.85 for the linear fit. In the first 30 min after addition, between 5% and 20% colocalizing micelles were detected. After four hours the colocalization value increased to 50% and reached 90% after six hours of incubation. Coincubated RGD(+) and RGD(−) micelles with PEG12 shielding showed a similar time-dependent progression of the colocalization degree as the reference measurement and were also best fitted by linear regression, giving a R-Squared value of 0.94. A different colocalization behavior was determined for simultaneously applied RGD(+) and RGD(−) micelles with PEG17 shielding. Compared to the reference, significantly lower colocalization values were reached after four to six hours of incubation. Furthermore, we observed a broad spread of the data points and a low R-Squared value of 0.2 for the linear regression of the plotted data. The obtained colocalization values coincide with the described cellular distribution of micelles in the overlay images. Cells showing a separation in micelle localization after coincubation with PEG17 RGD(+) and RGD(−) micelles possessed a low colocalization value, whereas increasing colocalization over time was observed for PEG12 RGD(+) and RGD(−) micelles with largely identical cellular distribution. The timedependent increase in colocalization as well as the intensity increase of

To examine the effect of dosage on the internalization behavior, the previously described colocalization experiment was repeated with a 53 fold increase in micelle concentration (132 ng DNA per 10,000 cells). To allow the application of a high number of micelles under physiological conditions as well as the fast sedimentation of micelles for a defined starting point, the micelles were diluted in a medium-sized volume (100 μl cell medium) and applied onto HeLa cells, where the total medium was removed. After one minute, 300 μl of medium was resupplied. Treated cells were imaged four to six hours post application by wide-field fluorescence microscopy. Again, as a reference, untargeted PEG12 micelles labeled either with Cy3 or Cy5, were simultaneously applied in a 1:1 mixture. In Fig. 4, images of representative cells are presented showing the two individual emission channels as well as the overlay images of both fluorescence channels. The images illustrate the accumulation of a subset of untargeted micelles in the nuclear proximity during four to six hours of incubation, whereas another subset of the initially applied untargeted micelles remained in the cell periphery. For coincubated PEG12 RGD(+) and RGD(−) micelles, the RGD(−) micelles showed a localization pattern comparable to the reference. In contrast, RGD(+) micelles were predominantly found in the nuclear proximity. As a consequence, in the overlay images a higher degree of green label encoding for RGD(+) micelles appears in the central section of the cell, whereas the magenta label, representing the RGD(−) micelles, is dominating in the periphery. This separated distribution of targeted and untargeted micelles was even more pronounced for coincubated PEG17 RGD(+) and RGD(−) micelles. Here, a strong accumulation of RGD(+) micelles was observed in the nuclear proximity, whereas untargeted micelles were almost completely retained in the cellular periphery (Fig. 4C).

Fig. 4. Effect of dosage on micelle internalization. HeLa cells were coincubated with a high concentration (132 ng DNA/10,000 cells) of micelles and imaged after four to six hours by wide-field fluorescence microscopy. (A) Overlay images of reference cells coincubated with Cy3 and Cy5 labeled RGD(−) micelles reveal high degree of colocalizing endosomes. (B) Coincubated PEG12 RGD(+) (green) and RGD(−) (magenta) micelles show a different distribution compared to the reference. RGD(+) micelles accumulate in the inner part of the cell, whereas RGD(−) micelles remain in the outer cell region. (C) A more pronounced separated distribution is observed for coincubated PEG17 RGD(+) (green) and RGD(−) (magenta) micelles. Scale bar: 10 μm.

To define the exact localization of micelles in proximity to the nucleus with enhanced z-resolution, additional spinning disk confocal microscopy was performed (see Supplementary material). The obtained confocal z-stacks of single cells revealed that the majority of micelles was entrapped in endosomes in close distance to the nucleus but did not enter the nucleus yet. A quenching assay with untargeted PEG12 shielded micelles revealed that the peripheral fraction mainly consisted of extracellularly attached micelles, whereas the micelles in the nuclear proximity were in intracellular compartments (see Supplementary material). This suggests that the internalization of untargeted micelles is saturated at high concentration resulting in the retention of a certain fraction of micelles on the plasma membrane. In the microscopic images, we observed varying amounts of peripheral micelles between different cells, indicating that the cell population is heterogeneous in their level of internalization. To determine the colocalization degree of the micelle combinations at high concentration, the mean colocalization value from cells, incubated for four to six hours, was calculated and normalized to the reference measurement with untargeted micelles (Fig. 5). Interestingly, for coincubated PEG12 RGD(+) and RGD(−) micelles the colocalization degree was comparable to the reference, although a partly separated cellular localization of the micelles was observed in the respective microscopical images. This high colocalization value

Fig. 5. Quantification of micelle colocalization at high concentration. HeLa cells were coincubated with the indicated combinations of fluorescently labeled micelles and imaged by wide-field fluorescence microscopy after four to six hours. Obtained movies were analyzed for colocalizing endosomes using custom-written software. Mean colocalization values of the imaged cells were determined for the micelle combinations and normalized to reference cells that were coincubated with Cy3 and Cy5 labeled PEG12 micelles without a targeting ligand. The standard error of the mean (SEM) is represented by error bars (N= 24 for PEG12 RGD(−)RGD(−), N = 26 for PEG12 RGD(+)RGD(−), N = 32 for PEG17 RGD(+)RGD(−).*** P b 0.0001 for PEG17 RGD(+)RGD(−) compared to the reference).

367

indicates that a considerable amount of PEG12 RGD(−) micelles was still internalized and transported to the same cellular compartments as RGD(+) micelles. In contrast, for coincubated PEG17 RGD(+) and RGD(−) micelles, the colocalization was significantly reduced to 34% of the reference value. Together with the split localization of RGD(+) and RGD(−) micelles in the microscopical images, this result suggests that solely integrin targeted micelles were efficiently transported to the nuclear proximity in the presence of enhanced shielding within four to six hours of incubation. Our data revealed a strong effect of shielding as well as the applied concentration on the cellular distribution of micelles. Only at high micelle concentration, we observed the enhanced accumulation of PEG12 RGD(+) micelles in the cell center compared to untargeted micelles. In contrast, PEG17 shielded micelles showed the separated localization of targeted and untargeted micelles also at low concentration.

2.3. Quantification of micelle uptake by flow cytometry Live-cell imaging is a powerful tool to study the detailed mechanisms of particle internalization and to visualize the cellular localization in single cells. In the present study, the transfection experiments were performed with PEGylated polyplexes at extraordinarily low DNA concentrations, compared to conventional transfection conditions e.g. 2.5 ng/10,000 cells. Such low micelle concentrations cannot be detected in standard bulk experiments. However, as considerable heterogeneity between cells exists, the quantification of nanoparticle internalization is challenging with microscopical methods and should be verified by standard bulk experiments such as flow cytometric analysis under conventional transfection conditions. Therefore, we performed a cellular uptake study of PEG17 RGD(+) and RGD(−) micelles by a flow cytometric analysis at 1 μg DNA per 10,000 HeLa cells after 24 h of incubation (Fig. 6). The obtained result clearly revealed that the uptake of PEG17 RGD(+) micelles into HeLa cells was significantly increased compared to the RGD(−) ones. For PEG12 shielded micelles, the uptake was not significantly increased by introduction of the RGD ligand as revealed by a previous flow cytometry study under the same transfection conditions (1 μg DNA per 10,000 cells and 24 h of incubation) [12]. These results demonstrate the enhanced cellular uptake of micelles with RGD ligand when combined with the longer 17 kDa PEG, consistent with our results from the microscopic observation.

Fig. 6. Uptake efficiencies of PEG17 RGD(+) and RGD(−) micelles, obtained by flow cytometric analysis. Each micelle sample containing 1 μg of Cy5-labeled plasmid DNA was incubated with HeLa cells (10,000 cells) for 24 h. The standard error of the mean (SEM) is represented by error bars (N = 3).

GENE DELIVERY


GENE DELIVERY

368


2.4. Identification of the uptake pathway To specify the uptake pathway used by integrin targeted and untargeted micelles, we performed spinning disk confocal microscopy colocalization studies with pathway specific markers. RGD(+) micelles with PEG17 shielding were used to determine the pathway of targeted micelles, as the improved shielding reduced unspecific interactions with the cell membrane and therefore improved the targeting specificity. PEG12 RGD(−) micelles served as reference system for the internalization pathway of receptor-independent uptake. In a previous study it was suggested that integrin targeting with the cyclic RGD ligand leads to caveolin dependent internalization of micelles [12]. This conclusion was based on colocalization experiments with labeled cholera toxin B, a specific marker for caveolin dependent endocytosis in several cell types. Singh et al. reported that the uptake mechanism of cholera toxin B is highly cell type dependent and seems to occur predominantly via clathrin mediated endocytosis in HeLa cells [27]. Therefore we decided to use HeLa cells expressing caveolin-GFP as a more specific marker for caveosomes compared to cholera toxin B. As a marker for clathrin mediated endocytosis, transferrin 488 was applied to HeLa cells. According to the literature, transferrin is efficiently incorporated into clathrin coated vesicles, followed by transport to early, sorting, and recycling endosomes [28]. After one hour coincubation of untargeted micelles and transferrin on cells, most micelles were located in the cell periphery whereas transferrin was localized in the central part of the cell and little colocalization was detected. This observation indicates that no direct interaction of micelles and marker occurred in the cell medium (data not shown). Coincubation of both targeted as well as untargeted micelles with transferrin 488 for four hours resulted in a high colocalization degree as demonstrated in Fig. 7A by the distinct white signal in the fluorescence overlay image. Concerted movements of spots in both emission channels were observed, proving that transferrin and micelles were in fact entrapped in the same endosomal compartments. This result suggests that the uptake of integrin targeted as well as untargeted micelles occurs via clathrin mediated endocytosis resulting in transportation to early, sorting or recycling endosomes during the first four hours of incubation. In contrast, after application of micelles to caveolin-GFP expressing cells, only a low number of colocalizing caveosomes were detected for the targeted as well as untargeted micelles, indicating that caveolin dependent endocytosis is not the dominant internalization pathway for both micelle types (Fig. 7B). To verify the predominant uptake of targeted and untargeted micelles via clathrin-mediated endocytosis additional inhibitor experiments with chlorpromazine, a specific inhibitor of clathrinmediated endocytosis, were performed (Fig. 8). Inhibitor studies on the single cell level require careful controls [29]. Therefore we first proved the integrity of the cell membrane in the presence of 10 μg/ml

Fig. 8. Chlorpromazine inhibition of clathrin-mediated endocytosis. HeLa cells were preincubated with 10 μg/ml chlorpromazine for 30 min before addition of Cy5 labeled PEG12 RGD(−) (A) or PEG17 RGD(+) (B) micelles to the cell medium. Cells were evaluated 3.5 h after micelle addition by spinning disk confocal microscopy. Representative z-stacks of treated cells are shown together with the transmission light image of the cell. The uptake of targeted and untargeted micelles is inhibited by the chlorpromazine resulting in extracellular accumulation of micelles on the cell membrane. Transferrin 488 that was coincubated with the micelles as a control for effective clathrin inhibition was not detected in the cytoplasm as well. Scale bar: 10 μm.

chlorpromazine by applying a trypan blue exclusion assay (see Supplementary information, Fig. S3). Additionally, we tested the successful uptake of lactosylceramide, a marker for caveolin dependent endocytosis, into chlorpromazine treated cells to exclude inhibition effects caused by cell damage and unspecific inhibition effects. Significant amounts of lactosylceramide were internalized into HeLa cells at the applied chlorpromazine concentration. This indicates that the caveolin-dependent uptake pathway was not inhibited under the applied conditions (Supplementary information, Fig. S4). Only cells that exhibited moderate changes in their cell shape in the bright field image were considered for evaluation. As a positive control for effective clathrin inhibition, we simultaneously applied fluorescently labeled transferrin with the micelles. Three to four hours post application, transferrin was absent in the cytoplasm of chlorpromazine treated cells, whereas untreated cell showed efficient uptake of transferrin. This verifies the effective inhibition of clathrinmediated uptake by chlorpromazine addition. Incubation of the chlorpromazine treated cells with either PEG17 RGD (+) or PEG12 RGD(−) micelles resulted in a distinct accumulation of both micelle types on the cell membrane. Confocal z-slices of the treated cells revealed a narrow rim of micelles in the membrane region. Neither targeted nor untargeted micelles were detected in the central section of the cell. This result gives further evidence that in HeLa cells clathrin-mediated endocytosis is the predominant uptake pathway for integrin targeted as well as untargeted micelles (Fig. 8). 2.5. Luciferase reporter gene expression

Fig. 7. Colocalization analysis of micelles and endocytosis pathway specific markers. Single z-slices of representative cells, imaged by spinning disk confocal microscopy, are shown. The regions of the cell nucleus and the plasma membrane are marked in the transmission light (left side) and fluorescence image with a yellow, dashed line. (A) Coincubation of Cy5 labeled PEG12 RGD(−) or PEG17 RGD(+) micelles (magenta) with transferrin 488 (green) on HeLa cells for four hours. Both targeted as well as untargeted micelles show high colocalization with transferrin as indicated by white endosomes in the overlay image. (B) Incubation of Cy5 labeled PEG12 RGD(−) or PEG17 RGD(+) micelles (magenta) on caveolin-GFP expressing (green) HeLa cells for four hour results in a low colocalization degree. Scale bar: 10 μm.

After the detailed studies on micelle internalization, the influence of integrin targeting and PEG shielding on the reporter gene expression of micelle treated cells was determined by standard luciferase expression assay. HeLa cells were incubated with the different micelle types and luciferase expression was determined 24 h post incubation (Fig. 9). Integrin targeting of PEG17 shielded micelles resulted in a 27 fold increase in gene expression compared to untargeted PEG17 micelles. The effect of integrin targeting was less prominent for micelles with PEG12 shielding and resulted in only 8 fold increase of gene expression. Notably, the transfection efficiency of PEG17 RGD(+) micelles exceeded

Fig. 9. Luciferase expression. Cells were transfected with RGD(+) or RGD(−) micelles with PEG12 or PEG17 shielding, respectively. Each well was transfected with 1 μg of DNA and analyzed for luciferase expression after 48 h by photoluminescence detection. The experiment was performed in triplicates, the standard error of the mean (SEM) is represented by error bars. PEG17 RGD(+) micelles show enhanced reporter gene expression compared to PEG12 RGD(+) micelles.

the efficiency of PEG12 RGD(+) micelles by factor two. In addition, the transfection efficiency of untargeted micelles was reduced in the presence of enhanced PEG shielding by 40%. These results indicate that the PEG17 shielding reduces the probability for transgene delivery by unspecifically internalized untargeted micelles and enhances the ligand mediated transgene delivery induced by integrin targeted micelles. 3. Discussion Installation of selective targeting ligands to therapeutic nanocarriers is a promising strategy to achieve specific targeting of cancer cells in systemic application [5,18,30,31]. To avoid unspecific interaction or adhesion in the blood stream, shielding of nanocarriers is necessary and is typically obtained by a hydrophilic PEG shell layer [32,33]. In the present study we analyzed the effect of integrin targeting with an RGD ligand in combination with differently sized PEG shell layers (12 kDa or 17 kDa PEG) on the cellular internalization of polyplex micelles by means of highly sensitive fluorescence microscopy and flow cytometry. We show that the RGD ligand leads to accelerated and preferential internalization of micelles into HeLa cells when combined with proper 17 kDa PEG shielding. Simultaneous addition of fluorescently labeled targeted and untargeted PEG17 shielded micelles to HeLa cells resulted in specific accumulation of RGD(+) micelles in the nuclear proximity and retention of RGD(−) micelles in the cell periphery. As a consequence, low colocalization of PEG17 shielded RGD(+) and RGD(−) micelles was observed. In presence of 12 kDa PEG shielding, we observed internalization characteristics dependent on the applied micelle dose. At low dose application, targeted PEG12 shielded micelles revealed an internalization behavior comparable to untargeted micelles. At high dose application, we observed split cellular localization reminiscent of PEG17 shielded micelles. However, a high colocalization of targeted and untargeted micelles in the nuclear proximity was still maintained indicating the persistent internalization of untargeted micelles. Targeted as well as untargeted micelles were internalized by clathrin-mediated endocytosis portending that the RGD ligand alters the kinetics of micelle internalization without changing the uptake pathway. In addition, targeted micelles with PEG17 shielding induced the highest transgene expression. The superior internalization characteristics of targeted PEG17 shielded micelles at all concentrations demonstrate the positive effect of ligand installation on micelle uptake and emphasize the careful testing of proper shielding at several concentrations. Although PEG12 and PEG17 shielded micelles had a difference in zeta potential of only 1 mV in Tris-buffer (+1.5 mV for PEG12 micelles and+0.5 mV for PEG17 micelles), they showed significant differences in internalization. This observation indicates that additional micelle–cell interactions

369

besides the receptor-mediated endocytosis were involved in the internalization of PEG12 shielded micelles. The single-cell experiments with PEG12 shielded micelles revealed interesting insights into ligand effects on internalization of polyplex micelles. We suppose that due to the different lengths of the PEG in the PEG12 and the PEG17 micelles, different interactions of the positively charged micelle core and the negatively charged plasma membrane may arise as the self-assembly of polymeric particles is considerably affected by the composition of the polymer [34–36]. A different arrangement of the PEG and the RGD ligand might appear on the micelle surface dependent on the length of the applied PEG. We propose that in the presence of the longer PEG, the flexibility of the RGD ligand might be enhanced, improving the receptor binding properties of the micelles. The fact that PEG12 micelles are well internalized also in absence of the RGD ligand suggests an insufficient shielding of the positively charged micelle core. Consistently, PEG17 shielding resulted in reduced internalization of untargeted micelles. Due to electrostatic interactions, the PEG12 RGD(−) micelles may bind to negatively charged plasma membrane components such as proteoglycans [23,37,38] resulting in receptor-independent micelle uptake. In case that the kinetics of this receptor-independent endocytosis is similar to the kinetics of integrin-dependent endocytosis, untargeted micelles and integrin-bound micelles would be endocytosed with comparable uptake efficiencies and may end in the same endosomal compartments, resulting in high colocalization values as observed in our colocalization studies at low concentration. We assume that the kinetics of integrin dependent endocytosis depends on the local concentration of RGD ligands that are available to activate receptor signaling and clustering. Sancey et al. demonstrated that multimeric RGD is required to induce efficient integrin clustering and fast endocytosis [39]. The specific binding of RGD(+) micelles to integrins may be hindered in the presence of electrostatic interactions with membrane components and could additionally be hampered by the small sample volume that was applied to the cells. The small volume may limit the free diffusion of micelles in the cell medium and thus promote forced receptor-independent interactions between micelles and the cell membrane resulting in receptor-independent endocytosis. Therefore, at the applied low micelle dose the local concentration of accessible RGD ligands of PEG12 shielded micelles might lie below the critical level that is required to induce efficient integrin mediated uptake. In this scenario, a certain portion of PEG12 RGD(+) micelles may also be internalized by other mechanisms than receptor-mediated endocytosis. We expect that at the applied low micelle concentration, combined endocytosis of multiple RGD(+) and RGD(−) micelles in clusters does not play a major role as micelles were applied in sufficiently low concentration to observe single micelles on the cell. Clustering effects would rather be expected at high concentration, however in this case a differing internalization behavior was observed for RGD(+) and RGD(−) micelles. PEG12 RGD(+) micelles preferentially accumulated in the nuclear proximity whereas untargeted PEG12 micelles were partly retained in the cell periphery. At high concentration of RGD(+) micelles, enhanced integrin clustering might occur, enforcing the integrin mediated endocytosis of the targeted micelles [39]. Additionally, negative charges on the cell membrane might be completely covered by positively charged micelles. In this scenario, excess micelles that diffuse in the cell medium, weakly interact with membrane components, resulting in preferential interaction of the RGD ligand with accessible integrins. The fact that untargeted PEG12 micelles were partly retained in the cellular periphery at high micelle dose, suggests the saturation of the receptor-independent micelle internalization at high concentration. Accordingly, a quenching assay revealed that the peripherical rim of RGD(−) micelles observed in the microscopic images, consisted mainly of extracellularly bound micelles that were not internalized yet. In contrast, PEG17 RGD(+) micelles were well internalized also at high concentration and saturation effects were not observed. Activated

GENE DELIVERY


GENE DELIVERY

370


integrins are recycled to the plasma membrane after their endocytosis [40–42]. They may then serve as receptors for further endocytosis of micelles and promote increased internalization after activation. The enhanced perinuclear accumulation of RGD(+) micelles at high concentration may be promoted by two mechanisms. First, the integrin mediated endocytosis may induce faster uptake kinetics, and second, an earlier onset of endosomal transportation of targeted micelles toward the cell nucleus may be possible. Previous studies on integrin uptake revealed that integrin heterodimers can be endocytosed via different internalization routes such as clathrin-dependent endocytosis [43,44], caveolin-dependent endocytosis [45,46] and macropinocytosis [47]. Our results from the colocalization experiments with markers for clathrinand caveolin-dependent endocytosis as well as inhibitor studies suggest that clathrin-dependent endocytosis plays the major role for the uptake of integrin targeted as well as untargeted micelles. The colocalization of RGD(−) and RGD(+) micelles at later time points demonstrates that untargeted micelles end up in the same cellular compartments as targeted micelles, but with slower kinetics. The RGD ligand therefore seems to accelerate the perinuclear accumulation of micelles without changing the uptake pathway. Lakadamyali et al. [48] demonstrated that different endocytic ligands for clathrin-mediated endocytosis are sorted into distinct populations of dynamic, rapidly maturing or static, slowly maturing early endosomes. Interestingly, the sorting already starts at the plasma membrane and is ligand dependent. As the maturation rate of endosomes is highly correlated with their mobility along microtubules [48], cargo in dynamic endosomes can reach the nuclear proximity faster than static ones and may explain the fast transfer of RGD equipped micelles to the nuclear proximity. Within six hours of microscopical observation of micelle internalization, we did not observe significant DNA release into the cytosol or nucleus. Confocal images of treated cells indicate that micelles were mainly entrapped in endosomes. However, successful luciferase expression after 24 h reveals that certain amounts of DNA have been released and reached the nucleus. As few plasmids are sufficient to induce significant levels of gene expression [49,50], the endosomal release of micelle DNA may have been below the detection limit of our microscopical set-up. Alternatively, the effective release may occur at later time points. The highly increased transgene expression of ligand equipped micelles with proper PEG17 shielding suggests that the RGD ligand also induces alterations in the intracellular processing in addition to accelerated uptake kinetics. Effects of the RGD ligand on intracellular processes have been reported. Shayakhmetov et al. revealed that the RGD motif triggers the enhanced endosomal release of adenovirus [51]. Chavez et al. described membrane destabilizing properties of the RGD ligand at low pH [52]. The RGD ligand may therefore account for enhanced transfection in later stages of transfection besides the accelerated and preferential internalization. To conclude, our results give mechanistical insights into the interplay of shielding and receptor targeting. Surface shielding of integrin targeted micelles has a significant effect on their targeting specificity. This effect is expected to be a general phenomenon for the targeting of charged polyplexes to all kinds of membrane receptors and has to our knowledge not been investigated on single cell level so far. Solely in the presence of proper shielding, integrin targeting had a significant effect on internalization and transgene expression, which is an important feature for the selective targeted therapy of cancer cells. Our results emphasize that highly sensitive microscopy on the single cell level provides additional information on the cellular localization which cannot be resolved in bulk experiments. Furthermore, as shown for the coincubation experiments at low micelle dose, microscopical observations can be performed at extraordinary low particle concentrations and on a single particle level which is not feasible with standard cytometric analysis. Previous studies revealed that micelles built from block copolymers are promising candidates for gene therapy because they have a uniform size and are stable over a long time period [12,13,21]. We propose that PEG17 shielded

micelles equipped with a cyclic RGD ligand are the favored system of choice for clinical therapy as they exhibit higher transfection efficiencies, a higher specificity for integrin-dependent endocytosis compared to PEG12 shielded micelles, and are functional at low doses as well. This gained knowledge enables the improved design of future gene vectors in order to maximize their therapeutic benefit for clinical application. 4. Materials and methods 4.1. Chemicals and reagents Cyclo[RGDfK (CX-)] (c(RGDfK)) peptide (X= 6-aminocaproic acid: ε-Acp) was purchased from Peptide Institute (Osaka, Japan). Thiolated block copolymers, poly(ethylene glycol)-block-poly(L-lysine-SH) (= PEG-p(Lys-SH)) for RGD(−) micelles and c(RGDfK)-poly(ethylene glycol)-block-poly(L-lysine-SH) (= c(RGDfK)-PEG-p(Lys-SH)) for RGD(+) micelles, were synthesized as previously reported [12,21]. Plasmid DNA encoding for luciferase (Luc) under the control of CAG promoter was provided by RIKEN Gene Bank (Tsukuba, Japan). The DNA was amplified in competent DH5α Escherichia coli and purified by the HiSpeed Plasmid Maxi Kit purchased from QIAGEN Sciences Co., Inc. (Germantown, MD). Luc DNA was labeled with Cy3 or Cy5 by the Label IT Nucleic Acid Labeling Kit (Mirus, Madison, WI) according to the manufacturer's protocol. Dulbecco's modified eagle's medium (DMEM), CO2 independent medium, fetal bovine serum (FBS), Transferrin Alexa 488 and Bodipy FL Lactosylceramide were obtained from Invitrogen (Karlsruhe, Germany). 0.4% Trypan Blue solution and chlorpromazine was purchased from Sigma (Munich, Germany). 4.2. Preparation of micelles Micelles were generated as previously reported [12,21]. Briefly, plasmid DNA was ion complexed with thiolated block copolymers (PEG-p(Lys-SH) for RGD(−) micelles or c(RGDfK)-p(Lys-SH) for RGD(+) micelles) at a molar N:P ratio of 2:1 (primary amine in lysine to phosphate in DNA) in 10 mM Tris–HCl buffer (pH 7.4) supplemented with 10% volume of 100 mM DTT. Disulfide linkages were formed during dialysis of micelles against 10 mM Tris–HCl buffer, for details see [12,21]. As determined by Ellman's method, more than 90% of thiol groups were converted to disulfide linkages. The block copolymers used for formation of PEG17 micelles consisted of a 17,000 g/mol PEG, a polylysine segment with a polymerization degree of 73 and a thiolation degree of 15%. Block copolymers of PEG12 micelles were equipped with a 12,000 g/mol PEG and a thiolation degree of 11%. The number of repeated lysine units and thiolation degree in PEG-p(LysSH) was determined from the peak intensity ratio of the assigned protons to PEG protons in 1H NMR spectra, according to a previous publication [13]. In detail, the lysine unit was calculated from the peak intensity ratio of beta-gamma-delta methylene protons in lysine to methylene protons in PEG. Also, the thiolation degree was calculated from the peak intensity ratio of methylene protons in 3-mercaptopropyonyl moiety to methylene protons in PEG. The narrow molecular weight distribution (Mw/Mn ~1.12) of the synthesized polymers was confirmed by gel permeation chromatography. Cumulant diameters analyzed by light scattering (Nano SZ zetasizer, ZEN3600, Malvern Instruments, Worcestershire, UK) were approximately 112 nm for PEG12 micelles and 104 nm for PEG17 micelles. Ζeta-Potentials measured in 10 mM Tris–HCl, pH 7.4 by laser Doppler electrophoresis using Nano ZS with a He–Ne laser (633 nm) were approximately + 1.5 mV for PEG12 + 0.5 mV for PEG17 micelles. 4.3. Cell culture HeLa cells were grown in Dulbecco's modified Eagle medium (DMEM), (Invitrogen) supplemented with 10% fetal bovine serum

(FBS) at 37 °C in 5% CO2 humidified atmosphere. For live cell-imaging, cells were seeded at a density of 1.0 × 10 4–2.0 × 10 4 per well into a collagen coated 8 well-chambered Lab-Tek slide (Nunc, Rochester, NY), 24–48 h before micelle addition. Cells were imaged in CO2 independent medium (Invitrogen) supplemented with FBS on a heated microscope stage at 37 °C. For the application of micelles, the CO2 independent cell medium was removed down to a thin fluid film to achieve immediate contact of micelles with the cell surface and their uniform distribution in the Lab-Tek chamber. One minute post micelle addition, the removed medium was resupplied. Standardized concentrations of DNA, equivalent to a defined amount of micelles, were applied for each measurement. For experiments at low micelle concentration 5 ng of DNA was added to the Lab-Tek chamber in a total volume of 3 μl (2.5 ng per micelle type in case of coincubation of two micelle species). For high micelle concentration, 264 ng of DNA was applied in a volume of 100 μl CO2 independent cell medium (132 ng per micelle type in case of coincubation studies). For colocalization with transferrin, Cy5 labeled micelles were coincubated with 2.5 μg/ml Alexa 488 labeled transferrin (Invitrogen) per well. Control experiments excluded direct interactions of transferrin and micelles in the cell medium. For colocalization with caveolin-GFP, Cy5 labeled micelles were added to stable caveolin-GFP transfected HeLa cells. Caveolin-GFP transfected HeLa cells were a kind gift of EJ Ungewickell (Medical School Hannover). 4.4. Wide-field fluorescence microscopy Wide-field fluorescence microscopy was performed at the indicated time-points on a custom built setup based on the Nikon Ti microscope equipped with a Plan Apo 60x, 1.49 NA oil immersion objective. Cy3 and Cy5 labeled micelles were excited with 532 nm and 633 nm laser light in alternating fashion (500 ms per frame). Fluorescence was collected in epifluorescence mode, split into two emission channels by a dichroic mirror (565 DCXR, Chroma) and passed through filter sets (Cy3 and 725/150, semrock). The green and red emission channels were projected onto two EMCCD cameras (DU-897 iXon+, Andor). Laser powers were optimized for each sample to receive maximum signal to noise ratio and low autofluorescence background. Overlay images were post-processed with ImageJ by applying a smoothing filter. To control the viability of cells, transmission light images of the treated cells were recorded. The region of the nucleus and the plasma membrane were identified in the transmission light images and transferred to the fluorescence overlay images. 4.5. Spinning disk confocal microscopy Spinning disk confocal microscopy was performed on a setup based on the Nikon TE2000E microscope and the Yokogawa spinning disk unit CSU10. The system was equipped with a Nikon 1.49 NA 100x Plan Apo oil immersion objective. For two color detection, cells were excited with alternating 488 nm and 633 nm laser light for 500 ms per frame. Confocal z-stacks of cells were imaged with a spacing of 166 nm between two planes. Image sequences were captured with an EM-CCD camera (iXon DV884; Andor). Displayed images were postprocessed with ImageJ by smoothing. 4.6. Colocalization analysis For the quantification of the colocalization degree in the recorded movies, the Cy3 and the Cy5 channel were geometrically calibrated according to control measurements with 0.17 μm tetraspec beads (Invitrogen). To enhance the signal to noise ratio in the images, median background subtraction was performed. Extracellular micelles and micelle containing endosomes were then counted in both channels with custom-built software, written in Labview and Matlab, using an intensity threshold criterion and a defined size restriction.

371

Micelles were determined as colocalized in case the mean fluorescence intensities in both fluorescent channels were above the defined threshold (manuscript in preparation containing a detailed description of the analysis software). Optimized parameters were kept constant for all evaluated images. To calculate the degree of colocalization, the number of colocalized spots was divided by the total number of spots in the reference channel. For low concentration experiments, the channel with the limiting number of detected spots was set as reference channel. In the experiments with high micelle concentration the Cy3 channel was set as reference, as the Cy3 labeled RGD(+) micelles were preferentially localized in intracellular endosomal compartments. For the histogram, the mean value of all analyzed cells was calculated and normalized to the reference measurement. P-values were determined by Student's t-test. 4.7. Fluorescence quenching Fluorescence of Cy-3 labeled polyplexes was quenched by resuspending 3 μl of 0.4% trypan blue solution (Sigma, Munich) into 400 μl medium of the observation chamber during image acquisition. As trypan blue is membrane-impermeable, only extracellular micelles were quenched, whereas internalized micelles remained fluorescent. 4.8. Inhibitor experiments HeLa cells were preincubated with 10 μg/ml chlorpromazine (Sigma) in CO2 independent medium (Invitrogen) supplemented with FBS for 30 min. Micelles were then added in the presence of chlorpromazine at a DNA concentration of 264 ng, together with 2.5 μg/ml Alexa Fluor 488 transferrin (Invitrogen) as a positive marker for clathrin inhibition. Treated cells were imaged 2–3 h post micelle addition by spinning disk confocal microscopy. Cell medium was changed to fresh, chlorpromazine containing medium before imaging to reduce background fluorescence by non-internalized fluorophores. To prove the membrane integrity of chlorpromazine treated cells, a trypan blue exclusion assay was performed. Briefly, cells were preincubated with 10 μg/ml chlorpromazine for 4 h before addition of 2 μl of trypan blue solution (Sigma, Munich). As a positive control for dead cell staining, ethanol (25% vol.) was added to a cell chamber with trypan blue treated cells. Cell staining was then detected with 633 nm laser excitation by wide-field fluorescence microscopy. To prove functional caveolin-dependent endocytosis in chlorpromazine treated cells, control cells were coincubated with 0.8 μM Bodipy FL lactosylceramide (Invitrogen) and 2.5 μg/ml Alexa Fluor 633 transferrin (Invitrogen) for 2.5 h, before imaging with spinning disk confocal microscopy using alternating laser excitation. 4.9. Quantification of polyplex internalization by flow cytometry HeLa cells were plated on a 24-well plate at a cell density of 10,000 cells/well in DMEM containing 10% FBS, followed by incubation for 24 h. After medium exchange with fresh DMEM containing 10% FBS, micelle samples prepared from Cy5 labeled DNA were applied to each well (1 μg DNA/well). After 24 h of incubation, cells were washed 3 times with 500 μl of PBS, treated with trypsin-EDTA solution, and suspended in PBS. The fluorescence intensity of Cy5 from the suspended cells was measured using a BD™ LSR II flow cytometer (BD Biosciences, Franklin Lakes, NJ) equipped with FACS Diva software (BD Biosciences). The results were expressed as mean fluorescence intensity obtained from three samples. 4.10. Luciferase assay HeLa cells were seeded on 24-well culture plates (10,000 cells/ well) and incubated for 24 h in 500 μl DMEM medium containing 10% FBS. Micelle solutions were then added at a concentration equivalent

GENE DELIVERY


GENE DELIVERY

372


to 1 μg of DNA per well and cells were incubated for 48 h. Following this incubation period, the luciferase gene expression was evaluated by the Luciferase Assay Kit (Promega, Madison, WI), and the intensity of photoluminescence was measured by the Mithras LB 940 (Berthold Tech.). The amount of protein in each well was determined by the Micro BCA™ Protein Assay Reagent Kit. Acknowledgments This work was performed on the basis of a collaboration agreement between the University of Tokyo and the Ludwig-Maximilians University (LMU) of Munich. Support from the Nano Initiative Munich (NIM) and the Center for Integrated Protein Science Munich (CIPSM) excellence clusters as well as the Core Research Program for Evolutional Science and Technology (CREST) from the Japan Science and Technology Agency (JST) and Funding Program for World-Leading Innovative R&D on Science and Technology (FIRST Program) from the Japan Society for the Promotion of Science (JSPS) is gratefully acknowledged. F.M.M thanks the Elitenetzwerk Bayern for funding. The authors gratefully thank Monika Franke (LMU Munich) for her technical assistance with the cell culture and V. Kudryavtsev (LMU Munich) for providing the data analysis software. Appendix A. Supplementary data Supplementary data to this article can be found online at doi:10. 1016/j.jconrel.2011.08.003. References [1] R. Duncan, The dawning era of polymer therapeutics, Nat. Rev. Drug. Discov. 2 (5) (2003) 347–360. [2] D. Schaffert, E. Wagner, Gene therapy progress and prospects: synthetic polymerbased systems, Gene Ther. 15 (16) (2008) 1131–1138. [3] K. Itaka, K. Kataoka, Recent development of nonviral gene delivery systems with virus-like structures and mechanisms, Eur. J. Pharm. Biopharm. 71 (3) (2009) 475–483. [4] Y. Lu, C.O. Madu, Viral-based gene delivery and regulated gene expression for targeted cancer therapy, Expert Opin. Drug Deliv. 7 (1) (2010) 19–35. [5] F. Danhier, O. Feron, V. Préat, To exploit the tumor microenvironment: passive and active tumor targeting of nanocarriers for anti-cancer drug delivery, J. Control. Release 148 (2) (2010) 135–146. [6] M. Günther, E. Wagner, M. Ogris, Specific targets in tumor tissue for the delivery of therapeutic genes, Curr. Med. Chem. Anticancer Agents 5 (2005) 157–171. [7] K.L. Douglas, Toward development of artificial viruses for gene therapy: a comparative evaluation of viral and non-viral transfection, Biotechnol. Progr. 24 (4) (2008) 871–883. [8] K. Miyata, Y. Kakizawa, N. Nishiyama, A. Harada, Y. Yamasaki, H. Koyama, K. Kataoka, Block catiomer polyplexes with regulated densities of charge and disulfide cross-linking directed to enhance gene expression, J. Am. Chem. Soc. 126 (8) (2004) 2355–2361. [9] K. Miyata, M. Oba, M. Kano, S. Fukushima, Y. Vachutinsky, M. Han, H. Koyama, K. Miyazono, N. Nishiyama, K. Kataoka, Polyplex micelles from triblock copolymers composed of tandemly aligned segments with biocompatible, endosomal escaping, and DNA-condensing functions for systemic gene delivery to pancreatic tumor tissue, Pharm. Res. 25 (12) (2008) 2924–2936. [10] S. Mishra, P. Webster, M.E. Davis, PEGylation significantly affects cellular uptake and intracellular trafficking of non-viral gene delivery particles, Eur. J. Cell Biol. 83 (3) (2004) 97–111. [11] X. Zhang, S.R. Pan, H.M. Hu, G.F. Wu, M. Feng, W. Zhang, X. Luo, Poly(ethylene glycol)-block-polyethylenimine copolymers as carriers for gene delivery: effects of PEG molecular weight and PEGylation degree, J. Biomed. Mater. Res. A 84A (3) (2008) 795–804. [12] M. Oba, K. Aoyagi, K. Miyata, Y. Matsumoto, K. Itaka, N. Nishiyama, Y. Yamasaki, H. Koyama, K. Kataoka, Polyplex micelles with cyclic RGD peptide ligands and disulfide cross-links directing to the enhanced transfection via controlled intracellular trafficking, Mol. Pharm. 5 (6) (2008) 1080–1092. [13] M. Oba, S. Fukushima, N. Kanayama, K. Aoyagi, N. Nishiyama, H. Koyama, K. Kataoka, Cyclic RGD peptide-conjugated polyplex micelles as a targetable gene delivery system directed to cells possessing alphavbeta3 and alphavbeta5 integrins, Bioconjug. Chem. 18 (5) (2007) 1415–1423. [14] W. Arap, R. Pasqualini, E. Ruoslahti, Cancer treatment by targeted drug delivery to tumor vasculature in a mouse model, Science 279 (5349) (1998) 377–380. [15] C. Rader, M. Popkov, J.A. Neves, C.F. Barbas, Integrin avb3-targeted therapy for Kaposi's sarcoma with an in vitro-evolved antibody, FASEB J. 16 (14) (2002) 2000–2002.

[16] O.M. Merkel, O. Germershaus, C.K. Wada, P.J. Tarcha, T. Merdan, T. Kissel, Integrin [alpha] v[beta]3 targeted gene delivery using RGD peptidomimetic conjugates with copolymers of PEGylated poly(ethylene imine), Bioconjug. Chem. 20 (6) (2009) 1270–1280. [17] J.S. Desgrosellier, D.A. Cheresh, Integrins in cancer: biological implications and therapeutic opportunities, Nat. Rev. Cancer 10 (1) (2010) 9–22. [18] S. Meng, B. Su, W. Li, Y. Ding, L. Tang, W. Zhou, Y. Song, Z. Caicun, Integrin-targeted paclitaxel nanoliposomes for tumor therapy, Med. Oncol. (2010) 1–8. [19] K. Temming, R.M. Schiffelers, G. Molema, R.J. Kok, RGD-based strategies for selective delivery of therapeutics and imaging agents to the tumour vasculature, Drug Resist. Updat. 8 (6) (2005) 381–402. [20] G. Zuber, M. Dontenwill, J.-P. Behr, Synthetic viruslike particles for targeted gene delivery to [alpha]v[beta]3 integrin-presenting endothelial cells, Mol. Pharm. 6 (5) (2009) 1544–1552. [21] Y. Vachutinsky, M. Oba, K. Miyata, S. Hiki, M.R. Kano, N. Nishiyama, H. Koyama, K. Miyazono, K. Kataoka, Antiangiogenic gene therapy of experimental pancreatic tumor by sFlt-1 plasmid DNA carried by RGD-modified crosslinked polyplex micelles, J. Control. Release 149 (1) (2011) 51–57. [22] M. Oba, Y. Vachutinsky, K. Miyata, M.R. Kano, S. Ikeda, N. Nishiyama, K. Itaka, K. Miyazono, H. Koyama, K. Kataoka, Antiangiogenic gene therapy of solid tumor by systemic injection of polyplex micelles loading plasmid DNA encoding soluble Flt1, Mol. Pharm. 7 (2) (2010) 501–509. [23] K. de Bruin, N. Ruthardt, K. von Gersdorff, R. Bausinger, E. Wagner, M. Ogris, C. Brauchle, Cellular dynamics of EGF receptor-targeted synthetic viruses, Mol. Ther. 15 (7) (2007) 1297–1305. [24] A.M. Sauer, K.G. de Bruin, N. Ruthardt, O. Mykhaylyk, C. Plank, C. Bräuchle, Dynamics of magnetic lipoplexes studied by single particle tracking in living cells, J. Control. Release 137 (2) (2009) 136–145. [25] K.G. de Bruin, C. Fella, M. Ogris, E. Wagner, N. Ruthardt, C. Bräuchle, Dynamics of photoinduced endosomal release of polyplexes, J. Control. Release 130 (2) (2008) 175–182. [26] G. Seisenberger, M.U. Ried, T. Endre, H. Buning, M. Hallek, C. Braeuchle, Real-time single-molecule imaging of the infection pathway of an adeno-associated virus, Science 294 (5548) (2001) 1929–1932. [27] R.D. Singh, V. Puri, J.T. Valiyaveettil, D.L. Marks, R. Bittman, R.E. Pagano, Selective caveolin-1-dependent endocytosis of glycosphingolipids, Mol. Biol. Cell 14 (8) (2003) 3254–3265. [28] A. Dautry-Varsat, Receptor-mediated endocytosis: the intracellular journey of transferrin and its receptor, Biochimie 68 (1986) 375–381. [29] D. Vercauteren, R.E. Vandenbroucke, A.T. Jones, J. Rejman, J. Demeester, S.C. De Smedt, N.N. Sanders, K. Braeckmans, The Use of Inhibitors to Study Endocytic Pathways of Gene Carriers: Optimization and Pitfalls, Mol Ther 18 (3) (2010) 561–569. [30] M. Ogris, G. Walker, T. Blessing, R. Kircheis, M. Wolschek, E. Wagner, Tumor-targeted gene therapy: strategies for the preparation of ligand-polyethylene glycolpolyethylenimine/DNA complexes, J. Control. Release 91 (1–2) (2003) 173–181. [31] O. Veiseh, F.M. Kievit, J.W. Gunn, B.D. Ratner, M. Zhang, A ligand-mediated nanovector for targeted gene delivery and transfection in cancer cells, Biomaterials 30 (4) (2009) 649–657. [32] L.M. Kaminskas, B.J. Boyd, P. Karellas, G.Y. Krippner, R. Lessene, B. Kelly, C.J.H. Porter, The impact of molecular weight and PEG chain length on the systemic pharmacokinetics of PEGylated poly l-lysine dendrimers, Mol. Pharm. 5 (3) (2008) 449–463. [33] L. van Vlerken, T. Vyas, M. Amiji, Poly(ethylene glycol)-modified nanocarriers for tumor-targeted and intracellular delivery, Pharm. Res. 24 (8) (2007) 1405–1414. [34] R.B. Cheyne, M.G. Moffitt, Self-assembly of polystyrene-block-poly(ethylene oxide) copolymers at the Air water interface: is dewetting the genesis of surface aggregate formation? Langmuir 22 (20) (2006) 8387–8396. [35] D. Hoenig, J. DeRouchey, R. Jungmann, C. Koch, C. Plank, J.O. Raedler, Biophysical Characterization of Copolymer-Protected Gene Vectors, Biomacromolecules 11 (7) (2010) 1802–1809. [36] Y. Kim, J. Pyun, J.M.J. Frechet, C.J. Hawker, C.W. Frank, The dramatic effect of architecture on the self-assembly of block copolymers at interfaces, Langmuir 21 (23) (2005) 10444–10458. [37] A. Kichler, Gene transfer with modified polyethylenimines, J. Genet Med. 6 (S1) (2004) S3–S10. [38] M. Piest, J.F.J. Engbersen, Effects of charge density and hydrophobicity of poly(amido amine)s for non-viral gene delivery, J. Control. Release 148 (1) (2010) 83–90. [39] L. Sancey, E. Garanger, S. Foillard, G. Schoehn, A. Hurbin, C. Albiges-Rizo, D. Boturyn, C. Souchier, A. Grichine, P. Dumy, J.-L. Coll, Clustering and internalization of integrin [alpha]v[beta]3 with a tetrameric RGD-synthetic peptide, Mol. Ther. 17 (5) (2009) 837–843. [40] P.T. Caswell, S. Vadrevu, J.C. Norman, Integrins: masters and slaves of endocytic transport, Nat. Rev. Mol. Cell Biol. 10 (12) (2009) 843–853. [41] M.S. Roberts, A.J. Woods, T.C. Dale, P. van der Sluijs, J.C. Norman, Protein kinase B/Akt acts via glycogen synthase kinase 3 to regulate recycling of {alpha}v{beta}3 and {alpha}5{beta}1 integrins, Mol. Cell. Biol. 24 (4) (2004) 1505–1515. [42] A.J. Woods, D.P. White, P.T. Caswell, J.C. Norman, PKD1/PKC promotes [alpha]v [beta]3 integrin recycling and delivery to nascent focal adhesions, EMBO J. 23 (13) (2004) 2531–2543. [43] E.J. Ezratty, C. Bertaux, E.E. Marcantonio, G.G. Gundersen, Clathrin mediates integrin endocytosis for focal adhesion disassembly in migrating cells, J. Cell Biol. 187 (5) (2009) 733–747. [44] T. Nishimura, K. Kaibuchi, Numb controls integrin endocytosis for directional cell migration with aPKC and PAR-3, Dev. Cell 13 (1) (2007) 15–28. [45] B.G. Galvez, S. Matias-Roman, M. Yanez-Mo, M. Vicente-Manzanares, F. SanchezMadrid, A.G. Arroyo, Caveolae are a novel pathway for membrane-type 1 matrix metalloproteinase traffic in human endothelial cells, Mol. Biol. Cell 15 (2) (2004) 678–687.

[46] F. Shi, J. Sottile, Caveolin-1-dependent beta1 integrin endocytosis is a critical regulator of fibronectin turnover, J. Cell Sci. 121 (14) (2008) 2360–2371. [47] Z. Gu, E.H. Noss, V.W. Hsu, M.B. Brenner, Integrins traffic rapidly via circular dorsal ruffles and macropinocytosis during stimulated cell migration, J.Cell Biol. 193 (1) (2011) 61–70. [48] M. Lakadamyali, M.J. Rust, X. Zhuang, Ligands for clathrin-mediated endocytosis are differentially sorted into distinct populations of early endosomes, Cell 124 (5) (2006) 997–1009. [49] R.N. Cohen, M.A.E.M. van der Aa, N. Macaraeg, A.P. Lee, F.C. Szoka Jr., Quantification of plasmid DNA copies in the nucleus after lipoplex and polyplex transfection, J. Control. Release 135 (2) (2009) 166–174.

373

[50] G. Schwake, S. Youssef, J.-T. Kuhr, S. Gude, M.P. David, E. Mendoza, E. Frey, J.O. Rädler, Predictive modeling of non-viral gene transfer, Biotechnol. Bioeng. 105 (4) (2010) 805–813. [51] D.M. Shayakhmetov, A.M. Eberly, Z.-Y. Li, A. Lieber, Deletion of penton RGD motifs affects the efficiency of both the internalization and the endosome escape of viral particles containing adenovirus serotype 5 or 35 fiber knobs, J. Virol. 79 (2) (2005) 1053–1061. [52] A. Chávez, M. Pujol, I. Haro, M.A. Alsina, Y. Cajal, Membrane fusion by an RGDcontaining sequence from the core protein VP3 of hepatitis A virus and the RGAanalogue: implications for viral infection, Biopolymers 58 (1) (2001) 63–77.

GENE DELIVERY
