Targeted delivery of SiRNA to CD33-positive tumor cells with liposomal carrier systems

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 e v i e r. c o m / l o c a t e / j c o n r e l

Targeted delivery of SiRNA to CD33-positive tumor cells with liposomal carrier systems Miriam Rothdiener a, Dafne Müller a, Patricia Garrido Castro b, Anja Scholz b, Michael Schwemmlein c, Georg Fey c, Olaf Heidenreich b, Roland E. Kontermann a,⁎ a b c

Institut für Zellbiologie und Immunologie, Universität Stuttgart, Allmandring 31, 70569 Stuttgart, Germany Northern Institute for Cancer Research, Paul O'Gorman Building, Medical School, Framlington Place, Newcastle upon Tyne NE2 4HH, United Kingdom Lehrstuhl für Genetik, Universität Erlangen-Nürnberg, Erwin-Rommel-Strasse 3, 91058 Erlangen, Germany

a r t i c l e

i n f o

Article history: Received 23 October 2009 Accepted 16 February 2010 Available online 22 February 2010 Keywords: siRNA AML t(8;21) translocation Liposomes Immunoliposomes Silencing

a b s t r a c t SiRNA molecules represent promising therapeutic molecules, e.g. for cancer therapy. However, efficient delivery into tumor cells remains a major obstacle for treatment. Here, we describe a liposomal siRNA carrier system for targeted delivery of siRNA to CD33-positive acute myeloid leukemia cells. The siRNA is directed against the t(8;21) translocation resulting in the AML1/MTG8 fusion protein. The siRNA was encapsulated in free or polyethylene imine (PEI)-complexed form into PEGylated liposomes endowed subsequently with an anti-CD33 single-chain Fv fragment (scFv) for targeted delivery. The resulting siRNA-loaded immunoliposomes (IL) and immunolipoplexes (ILP) showed specific binding and internalization by CD33-expressing myeloid leukemia cell lines (SKNO-1, Kasumi-1). Targeted delivery of AML1/MTG8 siRNA, but not of mismatch control siRNA, reduced AML1/MTG8 mRNA and protein levels and decreased leukemic clonogenicity, a hallmark of leukemic self-renewal. Although this study revealed that further modifications are necessary to increase efficacy of siRNA delivery and silencing, we were able to establish a targeted liposomal siRNA delivery system combining recombinant antibody fragments for targeted delivery with tumor cell-specific siRNA molecules as therapeutic agents. © 2010 Elsevier B.V. All rights reserved.

1. Introduction Oncogenes such as leukemic fusion genes, which arise from chromosomal translocations, are promising targets for therapeutic approaches as they are exclusively expressed in premalignant and malignant tissues. Frequently, tumor cells are absolutely dependent on, i.e. “addicted” to such oncogenes. However, the majority of these translocations affect transcription factors, which are difficult to target by conventional small molecule-based approaches, consequently limiting their attractiveness as cancer-specific targets [1]. In this situation, short interfering RNAs (siRNAs) provide a promising option [2]. These small, double-stranded RNAs of 21 to 23 nucleotides in length exploit the RNA interference (RNAi) pathway by guiding the RNA-induced silencing complex (RISC) to complementary target sequences leading to the cleavage and subsequent degradation of the targeted transcript [3,4]. Since the design of siRNAs primarily requires the sequence of the target transcript, virtually any gene of interest including “undruggable” transcription factor genes can be suppressed.

Abbreviations: AML, acute myeloid leukemia; scFv, single-chain fragment variable; PEG, polyethylene glycol; siRNA, short interfering RNA; RNAi, RNA interference; IL, immunoliposome; ILP, immunopolyplex. ⁎ Corresponding author. Tel.: + 49 711 685 6698; fax: + 49 711 685 67484. E-mail address: [email protected] (R.E. Kontermann). 0168-3659/$ – see front matter © 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.jconrel.2010.02.020

Thus, siRNA approaches are commonly used in drug development programs for target validation. RNAi libraries are also employed for target identification. Finally, combinations of siRNA-mediated knockdown and small molecular drug screening have been successfully used to identify new treatment options in cancer [5]. However, the therapeutic potential of direct siRNA applications has not been realized yet because of their unfavorable pharmacokinetic properties such as low stability in body fluids and rapid renal clearance. Several promising approaches using siRNA-packaging particles with and without ligands have been described [6,7]. Advanced delivery systems, e.g. liposomal carrier systems, employ siRNA complexed to polycationic substances encapsulated into a sterically stabilized lipid bilayer. Furthermore, these particle surfaces have been modified to faciliate endosomal release, e.g. through incorporation of fusogenic peptides [8]. However, in the majority of the cases, siRNA accumulation was restricted to liver, lung, spleen and kidney, with only marginal amounts found in other organs such as brain and bone marrow. The translocation t(8;21) is the most frequent chromosomal rearrangement found in AML [9]. It replaces the C-terminal transactivation domain of AML1 (also known as RUNX1), a transcription factor vital for definitive hematopoiesis, with the almost complete open reading frame of MTG8 (also known as RUNX1T1 or ETO) [10,11]. The resultant AML1/MTG8 represents a chimeric transcription factor,

NANOMEDICINE

Journal of Controlled Release 144 (2010) 251–258

NANOMEDICINE

252

M. Rothdiener et al. / Journal of Controlled Release 144 (2010) 251–258

which interferes with myeloid differentiation by recruiting histone deacetylases to AML1-regulated promoters as well as by sequestering transcription factors driving differentiation, thus converting an important transcriptional modulator to a constitutive repressor of gene expression [12]. Suppression of AML1/MTG8 with siRNAs relieves the differentiation block in leukemic cells and impairs leukemic growth both in cell culture and in animal models [13–17]. Tumor therapy involving antibodies or antibody-derived molecules offers the possibility for tumor targeting by directly reaching the tumor cells via surface antigen binding. The high internalization rate of CD33 [18], its restriction to the myeloid lineage and its expression on AML cells and leukemic stem cells make it a suitable antigen for targeted drug delivery in AML. CD33 is down regulated in mature granulocytes but retained in monocytes [19]. CD33 has already been used for clinical approaches of targeted drug delivery [20,21]. Experience with Gemtuzumab-ozogamycin (GO, Mylotarg), an antiCD33 monoclonal antibody-drug conjugate, has shown clinical relevance [22]. In another study, a CD33-specific immunotoxin was highly effective in killing human myeloid leukemia cells [23]. Thus, CD33 is an interesting surface antigen for targeted delivery. For instance, immunoliposomes, generated by coupling of antibodies to a liposomal surface, allow for an active tumor targeting, e.g. through binding to tumor cell-specific receptors, and thus present a promising approach for targeted drug delivery [24]. Here, we describe the generation of liposomal carrier systems for targeted delivery of anti-leukemic siRNA into CD33-positive myeloid tumor cells. As a targeting ligand we used an anti-CD33 single-chain Fv fragment [23], which was further modified with a C-terminal cysteine residue to allow for a site-directed conjugated to PEGylated lipids. The siRNA was encapsulated into liposomes either in free form (immunoliposomes; IL) or complexed with polyethylenimine (PEI) (immunolipoplexes, ILP). Specific binding and uptake into tumor cells was observed, which led to silencing of the AMG1/MTG8 fusion protein and a reduced growth in colony formation assays as compared to a control siRNA. 2. Materials and methods 2.1. Materials The human CD33-expressing acute myeloid leukemic cell lines SKNO-1 [25] and Kasumi-1 [26] (DSMZ No. ACC 220) as well as the acute T-cell leukemia line Jurkat [27] (DSMZ No. ACC 282) were grown in RPMI1640 containing 10% FCS, 2 mM glutamine or in case of SKNO-1 20% FCS + 7 ng/mL GM-CSF (ImmunoTools, Friesoythe, Germany). FITC-labeled anti-CD33 antibody was purchased from ImmunoTools (Friesoythe, Germany). Anti His-Tag unconjugated mAb IgG1 mouse DIA 900 from Dianova (Hamburg, Germany), goat anti-mouse IgG-R-Phycoerythrin (PE), P9287 and rabbit anti-mouse IgG-FITC, F9137 from Sigma-Aldrich (St. Louis, USA). Egg phosphatidylcholine (EPC) was purchased from Lipoid (Ludwigshafen, Germany) and cholesterol was purchased from Calbiochem (Merck, Darmstadt, Germany). All other lipids were purchased from Avanti Polar Lipids (USA). Human plasma (stabilized with citrate-phosphatedextrose solution) was kindly provided by the blood donation center of the Katharinenhospital (Stuttgart, Germany). His-probe (H-3) Horseradish Peroxidase (HRP)-conjugated mouse monoclonal antibody was purchased from Santa Cruz Biotechnology (SC8036, Santa Cruz, USA). AML1/RHD antibody was purchased from Calbiochem (Ab-2, PC285, Merck Chemicals, Nottingham, UK), goat anti-rabbit IgG peroxidase conjugate, A0545 from Sigma-Aldrich (St. Louis, USA). SiRNA was purchased from Eurofins MWG Operon (Ebersberg, Germany): siAGF1 5′- CCU CGA AAU CGU ACU GAG AAG -3′, siAGF6 5′- CCU CGA AUU CGU UCU GAG AAG -3′, siMA6 5′-AAG AAA AGC AGA CCU ACU CCA-3′. Cy3-labeled siRNA: siMAX Cy3-AML1/MTG8 5′(Cy3) CCU CGA AAU CGU ACU GAG A(dTdT) -3′. Polyethylenimine

(PEI 25) was purchased from BASF (Ludwigshafen, Germany). Cell tracker green-CMFDA was purchased from Invitrogen (San Diego, USA). Methylcellulose was purchased from Fluka (Buchs, Switzerland) and XTT assay from Sigma-Aldrich (St. Louis, USA). 2.2. Antibody production and characterization Antibody fragments were expressed in E. coli TG1 and purified by immobilized metal ion affinity chromatography (IMAC) as described [28]. Protein concentration was determined by measuring the absorbance at 280 nm. ScFv′ were analyzed by SDS-PAGE under reducing and non-reducing conditions and stained with Coomassie Brilliant Blue R250. The melting point of the scFv′ CD33 variants was determined with a ZetaSizer Nano ZS (Malvern, Herrenberg, Germany). Purified scFv′ (100 µg) was diluted in PBS to a total volume of 1 ml and sterile filtered into a quartz cuvette. Dynamic laser light scattering intensity was measured while the temperature was increased in 1 °C intervals from 30 to 70 °C with 2 min equilibration for each temperature step. The melting point was defined as the temperature at which the measured size dramatically increased. 2.3. Flow cytometry Approximately 250,000 cells were incubated with 10 µg/ml scFv′ in 100 µl PBS containing 2% FCS, 0.02% sodium azide (PBA) for 2 h at 4 °C. After washing cells three times with PBA buffer (4 °C), cells were incubated with anti His-Tag unconjugated mAb anti-mouse IgG1 for 1 h at 4 °C and goat anti-mouse IgG-PE for 30 min at 4 °C. Cells were resuspended in 500 µl PBA buffer and analyzed by flow cytometry (Cytomics FC 500, Beckmann-Coulter). Data were evaluated with WinMDI, version 2.9. For the detection of cellular antigen expression levels FITC-conjugated anti-CD33 mAb was used at a dilution of 1:20. 2.4. Preparation of liposomes A lipid composition of EPC : cholesterol : mPEG2000-DSPE in a molar ratio of 13:6:1 was used for the preparation of liposomes. The lipid formulation contained DiI as a fluorescent lipid marker at a molar concentration of 0.3 mol%. A thin lipid film was formed in a round bottom flask by dissolving the lipids in chloroform and removing the solvent in a rotary evaporator for 10 min at 42 °C. Subsequently the lipid film was dried completely in a vacuum drying oven for at least 1 h at room temperature. The lipid film was hydrated in 10 mM HEPES buffer, pH 6.7 and vortexed until all components were dissolved. The final lipid concentration was 10 mM. The lipid solution was then extruded 21 times through 50 nm pore size polycarbonate filter membrane using a LiposoFast extruder to obtain small unilamellar vesicles. 2.5. Preparation of immunoliposomes 100 µg scFv′ were reduced by adding 5 µl tris(2-carboxyethyl) phosphine (TCEP) (625 nmol TCEP per 1 nmol scFv′) (Pierce, Rockford, USA) and incubated under nitrogen atmosphere for 2 h at room temperature. TCEP was then removed by dialysis against deoxygenated coupling buffer (10 mM Na2HPO4/NaH2PO4 buffer, 0.2 mM EDTA, 30 mM NaCl, pH 6.7) overnight at 4 °C. For the postinsertion method [29], scFv′ were coupled to malPEG2000-DSPE micelles. For this purpose, 50 µl of mal-PEG2000-DSPE stock solution were transferred into a 1.5 ml test tube for preparation of maleimide-functionalized micelles. The solvent was evaporated in the open tube at room temperature until a lipid film became visible. The lipid film was dissolved in H2O to a final concentration of 4.2 mM for the formation of micelles and incubated for 5 min at 65 °C. Micellar lipid and reduced scFv′ were mixed at a molar ratio of 4.67:1, overlaid with nitrogen and incubated for 30 min at room temperature. To saturate

253

the unconjugated coupling groups, L-cysteine was added to a final concentration of 1 mM to the scFv′-coupled micelles as well as to the control micelles, and incubated for at least 10 min at room temperature. The scFv′-coupled micelles were then inserted into preformed PEGylated liposomes at an adequate molar ratio (between 0.1 and 10 mol% mal-PEG-DSPE of total lipid) by incubation for 30 min at 55 °C in a water bath. Uncoupled scFv′-molecules were removed from the immunoliposome preparation by gel filtration using a Sepharose CL4B column equilibrated with 10 mM HEPES buffer pH 7.4. Alternatively, uncoupled scFv′ were separated by ultra-centrifugation at 300,000 g for 1 h at 4 °C. Immunoliposomes were analyzed by western blot under reducing conditions. 2 µg of SDS-PAGE separated proteins were blotted for 30 min per gel at 10 V onto a nitrocellulose membrane. The nitrocellulose membrane was incubated in western blot PBS containing 5% milk, 0.05% Tween 20 for 1 h at room temperature. For detection of scFv′ and lipid-coupled scFv′ the membrane was then incubated with a HRP-conjugated mouse antihis6-tag antibody (Santa Cruz Biotechnology Santa Cruz, USA) diluted 1:1000 in PBS containing 5% milk, 0.05% Tween 20 for 1 h at room temperature or o/n at 4 °C. After washing the membrane three times with PBS, 0.05% Tween and one time with PBS for 5 min, the blot was developed with ECL substrate solution for 1 min. For the determination of binding activity of immunoliposomes, 25 nmol immunoliposomes in 100 µl RPMI 1640, 10% FCS were tested on 250,000 cells in flow cytometry as described above. Immunoliposomes were detected directly by incorporation of DiI in the liposomal membrane. Cell binding was analyzed by flow cytometry after incubating cells with 25 nmol of the immunoliposomal formulations for 1 h at 4 °C. Mean fluorescence intensities (MFI) were determined with the help of WinMDI subtracting the MFI of untreated SKNO-1 cells. 2.6. In vitro plasma stability To analyze plasma stability, 25 nmol of scFv′ CD33 VHVL HC4 (see Fig. 1) immunoliposomes were preincubated in the presence of PBS or 50% human plasma for up to 10 days at 37 °C in a total volume of 50 µl. Subsequently, binding of immunoliposomes to SKNO-1 cells was analyzed by flow cytometry as described above. To study siRNA stability, liposomal siRNA was incubated in 50% FCS. Aliquots were withdrawn at the indicated times and siRNAs were purified using the miRNeasy Kit (Qiagen, Hilden, Germany) followed by separation on native 20% polyacrylamide-TBE gels and visualization by ethidium bromide staining. 2.7. Passive encapsulation of free siRNA into liposomes For siRNA encapsulation a less leaky liposome formulation was chosen. It contained the more rigid hydrogenated soy phosphatidylcholine (HSPC) instead of egg phosphatidylcholine (EPC) and the lipid concentration was increased to 100 mM. An AML1/MTG8-specific siRNA (siAGF1) as well as a mismatch siRNA (siAGF6) were used [15]. SiRNA encapsulation into liposomes was performed passively during liposome formation. In a further step, siRNA-loaded immunoliposomes were prepared by the postinsertion method as described before. 50 µM siRNA siAGF1 or siAGF6 in 10 mM HEPES pH 6.7 were added to a dry lipid film in a round bottom flask. During resolubilization of the lipid film by vortexing the siRNA in the aqueous phase was passively encapsulated into the liposomes. Non-encapsulated siRNA was then separated from the liposomes by ultracentrifugation at 300,000 g for 1 h at 4 °C. The liposomal pellet was taken up in 10 mM HEPES pH 6.7 for postinsertion of scFv′ coupled micelles. 2.8. Passive encapsulation of polyethylene imine (PEI)-complexed siRNA To enhance the encapsulation efficiency siRNA was complexed to polyethyleneimine. For optimal complex formation by electrostatic

Fig. 1. a) Schematic structure of scFv′ CD33 variants. Two scFv′ molecules differing in the order of VH and VL were each elongated in their peptide spacers by two different sequences (HC2 and HC4) (L, leader peptide; VH, variable heavy chain; VL, variable light chain; SH, sulfhydryl residue of the added cysteine residue). b) Binding monoclonal antiCD33 antibody to leukemic cell lines SKNO-1, Kasumi-1, and Jurkat measured by flow cytometry (grey, cells alone; black line, monoclonal antibody). c) Binding of scFv′ CD33 variants (1, scFvHL-HC2; 2, scFvHL-HC4; 3, scFvLH-HC2; 4, scFvLH-HC4; 360 nM each) to leukemic cell lines SKNO-1, Kasumi-1 and Jurkat determined by flow cytometry (grey, cells alone; black, scFv′).

interactions, a 25 kDa branched PEI was chosen and complexed to 50 µM siRNA at an N/P ratio of 5, representing the ratio of PEI nitrogen to siRNA phosphate ions. The inserted mass of PEI (mPEI), corresponding to a constant amount of siRNA (n siRNA ), was calculated as follows: mPEI = nsiRNA × NP, siRNA × MPEI × QN/P (NP represents the number of phosphate residues of the siRNA of 42, MPEI the molecular mass of one protonable nitrogen unit of PEI of 43 g/mol and QN/P the N/P ratio of 5). In the process, PEI was added to siRNA while both components were diluted in the same buffer volume. Complexation took place in 10 mM HEPES pH 7.4, where PEI was protonated. The mixture was pipetted up and down 10 times, vortexed and incubated for 20 min at room temperature. For the determination of encapsulation efficiency, 20 µM Cy3-labeled siRNA was used. Cy3 fluorescence was excited at a wavelength of 488 nm and its emitted light was detected at 570 nm in a TECAN microplate reader. After separation of the liposomes, Cy3 signal intensity of the supernatant was determined and compared to the signal intensity of the siRNA solution before encapsulation. To determine the size of siRNA-loaded liposomes and ILs, lipoplexes and ILPs, the formulations were diluted 1: 100 in PBS in a low volume disposable cuvette and size was measured by dynamic light scattering using a ZetaSizer Nano ZS.

NANOMEDICINE

M. Rothdiener et al. / Journal of Controlled Release 144 (2010) 251–258

NANOMEDICINE

254

M. Rothdiener et al. / Journal of Controlled Release 144 (2010) 251–258

2.9. Binding and internalization of siRNA-loaded IL to SKNO-1 cells Free Cy3-siRNA was used for the preparation of siRNA-loaded liposomes and ILs in order to investigate their binding and uptake into SKNO-1 cells. CD33-negative Jurkat cells incubated with ILs were taken as a negative control besides unconjugated liposomes incubated on CD33-positive SKNO-1 and Kasumi-1 cells. Binding and uptake were investigated by fluorescence microscopy. For this purpose, 250,000 cells were stained with 3.6 nM cell tracker green for 30 min at 37 °C and incubated with 400 nmol liposomes and ILs for 1 to 8 h at 37 °C in a total volume of 100 µl. Cells were fixed in 2% paraformaldehyde, mounted in mowiol and viewed on glass slides in a fluorescence microscope. Images were prepared with Zeiss Axiovision software. 2.10. Transfection with siRNA-loaded liposomal formulations 5 mM siRNA-loaded immunoliposomes were incubated with 500,000 cells in 100 µl cell culture medium for 8 h at 37 °C. Afterwards, the wells were completely filled with medium to a total volume of 400 µl. After 3 days of incubation the cells were harvested and analyzed by qRT-PCR and immunoblotting as described [13,16]. 2.11. Colony formation assay For colony formation assays 20,000 Kasumi-1 cells were transfected with 250 µM siRNA-loaded liposomal formulations in 100 µl RPMI1640 as described above. After 8 h of incubation, the cells were seeded in 0.5 ml semisolid RPMI1640 containing 10% FCS and 0.5% methocel in 24-well tissue culture plates. Colonies consisting of approximately 20 cells and more were counted after 7 days of incubation. For each sample within one experiment, colonies from 4 wells were counted. The results represent mean values of 3 independent experiments.

further confirmed by flow cytometric analysis of 250 µM immunoliposomes of an anti-endoglin scFv′ [30] on CD33 positive SKNO-1 cells, which showed no binding (Fig. 3f). In order to determine the optimal ratio of scFv′-coupled micelles to total lipid, immunoliposomes with different ratios (0.01 to 10 mol%) of micellar lipid were prepared by the postinsertion technique. An optimum with regard to binding activity was found by flow cytometry at a concentration of 0.3 mol% micellar lipid (Fig. 4a). In vivo stability of binding activity of the anti-CD33 immunoliposomes was determined by incubating liposomes in 50% human plasma at 37 °C for up to 10 days. Binding activity was analyzed by flow cytometry and compared to CD33 immunoliposomes incubated in PBS under the same conditions (Fig. 4b). As a comparison immunoliposomes which were stored at 4 °C were tested (day 0). An initial loss of binding activity (approximately 30 to 40%) was observed after one day of incubation in human plasma, without further reduction for up to day 10. Incubation of the immunoliposomes in PBS had only marginal effects on binding activity. 4.2. Stability of siRNA-loaded liposomes We examined siRNA stability as a surrogate marker for liposome integrity. For that, either free siRNA or liposomes loaded with uncomplexed siRNA were incubated for up to three days in 50% fetal calf serum. Free siRNA started to decline after 20 min and was completely degraded within three hours (not shown). In contrast, siRNA encapsulated into liposomes remained stable for at least one hour and was still visible after three days of incubation (Supplemental Fig. 1). 4.3. Binding, internalization of siRNA-loaded immunoliposomes Binding and internalization studies were performed with ILs loaded with uncomplexed Cy3-labeled siRNA at 37 °C to allow for receptor-mediated endocytosis. Binding and uptake into cells labeled

3. Statistics For statistical analysis, a non-parametric paired t-test was performed with the help of GraphPad Prism 4 software. In a paired test, values in each population represent paired observations. Two tailed p-values and confidence intervals of 95% were chosen. 4. Results 4.1. Anti-CD33 scFv′ variants Four variants of scFv′ CD33 differing in the order of VH and VL and in the length of the peptide spacer between the hexa-histidine-tag and the C-terminal cysteine (Fig. 1a) were produced in E. coli TG1 and purified by IMAC. Production yields varied between 0.2 and 0.7 mg/L bacterial culture for the different constructs. CD33-positive human leukemic cell lines SKNO-1 and Kasumi-1 and CD33-negative Jurkat cells were analyzed by flow cytometry for binding of the different scFv′ variants. Using an anti-CD33 monoclonal antibody strong expression of CD33 on SKNO-1 and Kasumi-1 as well as absence of CD33 on Jurkat cells was confirmed (Fig. 1b). All scFv′ variants showed specific binding but differed in their binding strength (Fig. 1c). An increased binding was observed for scFv′HL-HC2 and -HC4 as compared to the scFv′LH constructs, especially on Kasumi-1 cells. The anti-CD33 immunoliposomes generated from scFv′CD33 VHVL HC4 were further investigated for their antigen-specific binding to target cells (Fig. 2). In a blocking experiment, preincubation of SKNO1 cells with 3 µM free scFv′ resulted in a loss of binding for the antiCD33 immunoliposomes (Fig. 3c, d), in contrast to preincubation with a carcinoembryonic antigen (CEA)-specific scFv′, which had no effect (Fig. 3d). Specificity of immunoliposome binding to target cells was

Fig. 2. a) Coupling of scFv′ CD33 variants to malPEG2000-DSPE micelles and insertion into preformed liposomes (postinsertion) or coupling to preformed malPEG2000-DSPE liposomes (conventional). ScFv′ molecules were analyzed by 15% SDS-PAGE and blotted on a nitrocellulose membrane. Bands were detected by a HRP-labeled anti-HisTag antibody (1, purified scFv′ CD33 variant (2 µg); 2, scFv′-coupled micelles; 3, ILs, postinsertion; 4, ILs, conventional). Coupling of the lipids is indicated by a reduced mobility. b) Binding of CD33 ILs (250 µM lipid) to leukemic cell lines SKNO-1, Kasumi-1 and Jurkat (250000 cells) determined by flow cytometry (grey, cells alone; black line, anti-CD33 immunoliposomes).

255

Fig. 4. a) Binding activity of anti-CD33 VHVL HC4 ILs (250 µM lipid) prepared from different amounts of scFv-coupled malPEG2000-DSPE micelles. MFIs plotted against molar ratios of malPEG2000-DSPE to total lipid. b) Plasma stability of anti-CD33 ILs. Binding of the anti-CD33 ILs to SKNO-1 cells was analyzed by flow cytometry after incubated ILs for up to 10 days in 50% human plasma, at 37 °C. Plain mPEG liposomes served as negative control. Shown are the MFIs in percentage of cells incubated with untreated ILs.

with cell tracker could then be visualized due to the fluorescencelabeled siRNA (Cy3) within the ILs by fluorescence microscopy (Fig. 5). Antigen-specific uptake into SKNO-1 and, albeit more attenuated, into Kasumi-1 cells was observed. Approximately 30 to 50% of SKNO-1 and Kasumi-1 cells stained positive for Cy3siRNAencapsulated ILs. CD33-negative Jurkat cells served as a negative control and only showed weak binding of ILs. In addition, only very weak binding of unconjugated liposomes to SKNO-1 and Kasumi-1 cells was observed. 4.4. Silencing of AML1/MTG8 by siRNA-loaded IL and ILP

Fig. 3. a) Analysis of purified scFv′ CD33 VHVL HC4 variants by 15% SDS-PAGE, Coomassie stained (2 µg scFv′ per lane). b) Melting point analysis of scFv′ CD33 VHVL HC4 by dynamic light sacttering c-f) Analysis of specificity of binding of anti-CD33 IL to SKNO-1 cells: c) Binding of anti-CD33 immunoliposome, and d) Loss of binding of antiCD33 immunoliposome after preincubation with 3 µM scFv′ CD33, e) Binding of antiCD33 immunoliposome is not affected by preincubation with 3 µM control scFv′ CEA, and f) No binding to SKNO-1 cells is observed for anti-endoglin control immunoliposomes (grey, cells alone; black, CD33 immunoliposomes).

SiRNA-loaded immunoliposomes or immunolipolexes were generated by using either free or PEI-complexed siRNAs. To control for siRNA specificity, we used in addition to the active AML1/MTG8 siRNA (siAGF1) a mismatch control siRNA (siAGF6) [15]. In some experiments we used a second control siRNA (siMA6) targeting the MLL/AF4 fusion gene, which is neither expressed in Kasumi-1 nor in SKNO-1 cells, and which did not affect Kasumi-1 clonogenicity [31]. Anti-CD33 scFv was subsequently inserted by the postinsertion method. The efficiency of siRNA loading was determined with Cy3-labeled siRNA and depended on lipid concentration and siRNA complexation (Supplemental Table 1). Without complexation, approximately 25% of the free siRNA was encapsulated using a lipid concentration of 100 mM, corresponding to approximately 50 to 125 µmol siRNA/mol lipid depending on the concentration of the siRNA (20–50 µM) used for encapsulation. In contrast, siRNA encapsulation close to 100% was achieved with PEI-complexed siRNA and a lipid concentration of 100 mM (corresponding to 500 µmol siRNA/mol lipid using siRNA at 50 µM for encapsulation). Sizes of immunoliposomes varied between 120 and 140 nm as determined by dynamic light scattering (Table 1). Incubation of SKNO-1 cells for 3 days with increasing liposome concentrations resulted in the case of CD33 scFv-containing liposomes

NANOMEDICINE

M. Rothdiener et al. / Journal of Controlled Release 144 (2010) 251–258

NANOMEDICINE

256

M. Rothdiener et al. / Journal of Controlled Release 144 (2010) 251–258

Fig. 6. a) Anti-AML1 immunoblot analysis of Kasumi-1 cells three days after transfection with siRNA-loaded liposomal preparations (5 mM lipid). SiRNA concentration in siRNA ILs was 600 nM, in siRNA-PEI ILPs 2.5 µM. Lysates were separated by 10% SDS-PAGE (-, untreated cells; L, liposomes; IL, immunoliposomes; siAGF1, antiAML1/MTG8 siRNA; siAGF8, irrelevant control siRNA; black arrow, AML1/MTG8; white arrow, AML1). b) Colony formation of 20,000 Kasumi-1 cells seven days after transfection with empty and siRNA-loaded ILs and ILPs in a concentration of 250 µM lipid. Free scFv′ CD33 molecules approximately corresponding to the amount present in the immunoliposomes were included. SiRNA concentration in siRNA-ILs was 30 nM, in siRNA-PEI ILPs 125 nM (-, untreated cells; L, liposomes; IL, immunoliposomes) (n = 3).

Fig. 5. Analysis of uptake of liposomal preparations (4 mM) liposomes and anti-CD33 ILs into target cells after incubation for 1 h at 37 °C. Cytoplasm of cells was labeled with cell tracker, siRNA with Cy3. a) anti-CD33 IL on SKNO-1 cells, b) non-targeted liposomes on SKNO-1 cells, c) anti-CD33 IL on Kasumi-1 cells, d) non-targeted liposomes on Kasumi-1 cells, and e) anti-CD33 IL on Jurkat cells. Images of representative cells are shown.

in a 50% reduction of AML1/MTG8 transcript levels at a lipid concentration of 5 mM, which corresponds to siRNA concentrations of 600 nM for free siRNA and 2.5 µM for PEI-complexed siRNA (Supplemental Fig. 2). Incubation of the AML1/MTG8-expressing cell line Kasumi-1 with siAGF1-loaded IL and ILP caused a strong reduction of fusion protein with AML1/MTG8 siRNA, but not with mismatch control siRNA, thus proving siRNA specificity (Fig. 6A and Supplemental Fig. 3). Furthermore, as in the case of SKNO-1, this knockdown was only observed with immunoliposomes targeting CD33, but not with non-targeted liposomes. Since electroporation of Kasumi-1 with AML1/MTG8 siRNA impairs leukemic clonogenicity [16], we examined the effects of

Table 1 Hydrodynamic diameters of siRNA-loaded liposomes and ILs(mean values and standard deviations of n = 3 measurements).

siAGF1 siAGF6 siAGF1-PEI siAGF6-PEI

Liposome [nm]

Immunoliposome [nm]

117 ± 27 135 ± 2 131 ± 1 137 ± 3

143 ± 4 131 ± 1 135 ± 2 129 ± 6

siRNA-loaded liposomes on Kasumi-1 colony formation in semi-solid medium. In these experiments, we lowered the lipid concentration to 250 µM resulting in an siRNA concentration of 30 nM free siRNA and 125 nM PEI-complexed siRNA, respectively. Notably, colony formation was affected in an siRNA- and scFv-dependent fashion: only immunoliposomes containing AML1/MTG8 siRNA significantly diminished clonogenicity almost twofold compared to vectors with mismatch control siRNA. Untargeted liposomes containing any siRNA did not affect colony formation. However, we observed a minor increase in colony formation by control immunoliposomes. Interestingly, a similarly increased colony formation was obtained with empty immunoliposomes, but not with CD33 scFv alone or with immunoliposomes containing a control scFv′ against fibroblast activating protein (FAP) (not shown). 5. Discussion In this study, two different liposomal carrier systems were generated for the targeted delivery of therapeutic siRNA and compared with regard to their target cell-specific binding and uptake, and their silencing efficiency in vitro. As a targeting moiety for the carrier systems, the scFv′ antibody format was chosen. Different antibody formats such as whole antibodies [32], Fab′ fragments [33] and scFv fragments [32,34,35] have been used for the generation of carrier systems. However, whole antibodies have been shown to be immunogenic and are rapidly cleared from circulation through Fcmediated uptake by macrophages, e.g. Kupffer cells of the liver [36]. These disadvantages can be circumvented using Fab′ or scFv molecules as ligands. They can be easily modified through genetic

engineering, e.g. by insertion of an additional cysteine residue (scFv′). This permits a very defined and site-directed coupling to reactive groups of carriers. Four different scFv′ CD33 variants were created, differing in the order of their light and heavy variable chain and the length of the peptide spacer between the antibody domains and the Cterminal cysteine. In scFv′ binding studies, VHVL variants showed superior binding activity on SKNO-1 and Kasumi-1 cells compared to VLVH variants. These results indicate a potential influence of domain order on binding activity of these particular antibody fragments. Binding studies of ILs on SKNO-1 and Kasumi-1 cells confirmed, compared to all the other variants, superior binding activity of the scFv′ CD33 VHVL HC4 variant displayed on a PEGylated liposomal surface. An influence of the peptide spacer length on antibody binding activity has been described by Nobs et al. [37]. The authors claimed that the use of an extended spacer arm can reduce steric hindrance, making the binding site better accessible. An optimum binding to cells was observed at 0.3 mol% inserted micellar lipid corresponding to approximately 30 scFv′ molecules per liposome. These findings are in overall accordance with the literature. For example, maximal binding for anti-Her2-scFv ILs was observed with 30 to 40 attached scFv per liposome [38]. Kirpotin et al. [39] reported appropriate IL binding with 35 to 70 scFv molecules per liposome. An encapsulation efficiency of approximately 25% was determined for free siRNA yielding a molar siRNA/lipid ratio of 1.2 × 10−4. Passive encapsulation of siRNA during liposome formation has not been expected to reach high loading efficiencies, because neutral lipids were used, which have no ability to decoy negatively charged nucleic acids. Thus, encapsulation efficiencies between 3 and 20% were described for oligonucleotides and siRNA into neutral lipid formulations [40–42]. One strategy to improve the efficiency was the application of very high lipid concentrations. Here, we were able to confirm that siRNA loading efficiency is enhanced with increasing lipid concentrations. Other authors also showed that the addition of 2 to 5 mol% PEG-DSPE into neutral liposomes doubles the incorporation efficiency [40]. Binding and internalization of siRNA-loaded ILs to leukemic cell lines were investigated by fluorescence microscopy. For this purpose, liposomes loaded with free Cy3labeled siRNA were used. Similar to non-loaded liposomes, a highly antigen-specific uptake into CD33- positive SKNO-1 and Kasumi-1 cells was observed. The reason, why Kasumi-1 cells showed less distinct results compared to SKNO-1 cells, remains unclear, but confirmed the observations from non-encapsulated ILs in flow cytometry. SiRNA complexation with PEI enhanced loading efficiency to almost 100% achieving a molar siRNA/lipid ratio of 5 × 10−4. An N/P ratio of 5 was chosen for the siRNA complexation to PEI, which has also been confirmed by Shim and Kwon as a suitable ratio [43]. Furthermore, PEI can facilitate endosomal escape of siRNA by the proton sponge effect, which is provoked by the acidic milieu in the endosome leading to protonation of the PEI molecules and influx of chloride ions. This then results in osmotic swelling and physical rupture of the endosome [44]. Nevertheless, siRNA release from PEI after endosomal escape may be inefficient. Another important advantage of PEI complexation is the ability of PEI to release siRNA from the complex in the cytosol. Brunk et al. suggested that extension of the polymer network during endosomal acidification or by PEI interactions with endosomal or cytoplasmic constituents may support cytoplasmic siRNA release [45]. However, intact PEI-plasmid particles were found in cell nuclei, indicating that complexes might be too stable to release plasmid into the cytosol [46]. Various approaches to improve endosomal release of siRNA from carrier systems have been analyzed recently, including the incorporation of pH-sensitive fusogenic peptides [8]. However, PEG chains required for sterical stabilization have been found to impede a pH-triggered release. This might be circumvented introducing cleavable PEG-chains, e.g. through cleavage by proteases [47].

257

We have previously shown that electroporation of leukemic cells with AML1/MTG8 siRNAs, but not with mismatch control siRNAs, suppressed the AML1/MTG8 fusion gene and thereby facilitated differentiation, impaired leukemic proliferation and diminished leukemic clonogenicity [13,15,16]. In the present study, we observed comparably distinct and specific silencing effects of siRNA ILs and siRNA-PEI ILPs. Transfection with mismatch control siRNA showed neither unspecific silencing of the fusion gene nor any impairment of leukemic clonogenicity. However, in spite of an approximately fourfold higher siRNA concentration delivered, PEI complexation hardly improved AML1/MTG8 siRNA-mediated target suppression and inhibition of colony formation. Significantly reduced colony numbers of anti-CD33 siAGF1 ILs were obtained with and without PEI complexation in comparison to siAGF6 controls, but only marginal differences from untreated cells were observed. This may have been caused by a possible stimulating effect of the ILs and ILPs on the target cells, which was confirmed by an enhanced colony formation in both cell lines incubated with empty ILs, but not with free scFv′. In summary, we have established a liposomal siRNA carrier system for targeted siRNA delivery to tumor cells combining antibodydependent delivery of the carrier systems to CD33-positive leukemic cells and sequence-specific silencing of the leukemic AML1/MTG8 fusion gene by RNAi. However, our results also revealed that further improvements of the carrier systems are necessary, e.g. concerning lipid composition and complexing reagents for efficient siRNA-release within the cell [48]. Finally, further studies need to be performed to investigate the anti-tumor activity of our delivery system in relevant in vivo models such as xenotransplanted immunodeficient mice [17]. Acknowledgement This work was supported by grants from the Deutsche Krebshilfe (107026), the North of England Children's Cancer Research Fund and the Kay Kendall Leukaemia Fund. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.jconrel.2010.02.020. References [1] M. Thomas, J. Greil, O. Heidenreich, Targeting leukemic fusion proteins with small interfering RNAs: recent advances and therapeutic potentials, Acta Pharmacol. Sin. 27 (3) (2006) 273–281. [2] K.A. Whitehead, R. Langer, D.G. Anderson, Knocking down barriers: advances in siRNA delivery, Nat. Rev. Drug Discov. 8 (2) (2009) 129–138. [3] N.J. Caplen, S. Parrish, F. Imani, A. Fire, R.A. Morgan, Specific inhibition of gene expression by small double-stranded RNAs in invertebrate and vertebrate systems, Proc. Natl. Acad Sci. U.S.A. 98 (17) (2001) 9742–9747. [4] S.M. Elbashir, J. Harborth, W. Lendeckel, A. Yalcin, K. Weber, T. Tuschl, Duplexes of 21-nucleotide RNAs mediate RNA interference in cultured mammalian cells, Nature 411 (6836) (2001) 494–498. [5] S.M. Corsello, G. Roti, K.N. Ross, K.T. Chow, I. Galinsky, D.J. DeAngelo, R.M. Stone, A.L. Kung, T.R. Golub, K. Stegmaier, Identification of AML1-ETO modulators by chemical genomics, Blood 113 (24) (2009) 6193–6205. [6] Y.C. Tseng, S. Mozumdar, L. Huang, Lipid-based systemic delivery of siRNA, Adv. Drug Deliv. Rev. 61 (9) (2009) 721–731. [7] P. Kumar, H.S. Ban, S.S. Kim, H. Wu, T. Pearson, D.L. Greiner, A. Laouar, J. Yao, V. Haridas, K. Habiro, Y.G. Yang, J.H. Jeong, K.Y. Lee, Y.H. Kim, S.W. Kim, M. Peipp, G.H. Fey, N. Manjunath, L.D. Shultz, S.K. Lee, P. Shankar, T cell-specific siRNA delivery suppresses HIV-1 infection in humanized mice, Cell 134 (4) (2008) 577–586. [8] H. Hatakeyama, E. Itho, H. Akita, M. Oishi, Y. Nagasaki, S. Futaki, H. Harashima, A pH-sensitive fusogenic peptide facilitates endosomal escape and greatly enhances the gene silencing of siRNA-containing nanoparticles in vitro and in vivo, J. Control Release 139 (2) (2009) 117–132. [9] A.T. Look, Oncogenic transcription factors in the human acute leukemias, Science 278 (5340) (1997) 1059–1064. [10] P. Erickson, J. Gao, K.S. Chang, T. Look, E. Whisenant, S. Raimondi, R. Lasher, J. Trujillo, J. Rowley, H. Drabkin, Identification of breakpoints in t(8;21) acute myelogenous leukemia and isolation of a fusion transcript, AML1/ETO, with similarity to Drosophila segmentation gene, runt, Blood 80 (7) (1992) 1825–1831.

NANOMEDICINE

M. Rothdiener et al. / Journal of Controlled Release 144 (2010) 251–258

NANOMEDICINE

258

M. Rothdiener et al. / Journal of Controlled Release 144 (2010) 251–258

[11] H. Miyoshi, T. Kozu, K. Shimizu, K. Enomoto, N. Maseki, Y. Kaneko, N. Kamada, M. Ohki, The t(8;21) translocation in acute myeloid leukemia results in production of an AML1-MTG8 fusion transcript, Embo J 12 (7) (1993) 2715–2721. [12] J.R. Downing, The AML1-ETO chimaeric transcription factor in acute myeloid leukaemia: biology and clinical significance, Br. J. Haematol. 106 (2) (1999) 296–308. [13] J. Dunne, C. Cullmann, M. Ritter, N.M. Soria, B. Drescher, S. Debernardi, S. Skoulakis, O. Hartmann, M. Krause, J. Krauter, A. Neubauer, B.D. Young, O. Heidenreich, siRNA-mediated AML1/MTG8 depletion affects differentiation and proliferation-associated gene expression in t(8;21)-positive cell lines and primary AML blasts, Oncogene 25 (45) (2006) 6067–6078. [14] F. Fazi, G. Zardo, V. Gelmetti, L. Travaglini, A. Ciolfi, L. Di Croce, A. Rosa, I. Bozzoni, F. Grignani, F. Lo-Coco, P.G. Pelicci, C. Nervi, Heterochromatic gene repression of the retinoic acid pathway in acute myeloid leukemia, Blood 109 (10) (2007) 4432–4440. [15] O. Heidenreich, J. Krauter, H. Riehle, P. Hadwiger, M. John, G. Heil, H.P. Vornlocher, A. Nordheim, AML1/MTG8 oncogene suppression by small interfering RNAs supports myeloid differentiation of t(8;21)-positive leukemic cells, Blood 101 (8) (2003) 3157–3163. [16] N. Martinez, B. Drescher, H. Riehle, C. Cullmann, H.P. Vornlocher, A. Ganser, G. Heil, A. Nordheim, J. Krauter, O. Heidenreich, The oncogenic fusion protein RUNX1-CBFA2T1 supports proliferation and inhibits senescence in t(8;21)positive leukaemic cells, BMC Cancer 4 (2004) 44. [17] N. Martinez Soria, R. Tussiwand, P. Ziegler, M.G. Manz, O. Heidenreich, Transient depletion of RUNX1/RUNX1T1 by RNA interference delays tumour formation in vivo, Leukemia 23 (1) (2009) 188–190. [18] V.H. van Der Velden, J.G. te Marvelde, P.G. Hoogeveen, I.D. Bernstein, A.B. Houtsmuller, M.S. Berger, J.J. van Dongen, Targeting of the CD33-calicheamicin immunoconjugate Mylotarg (CMA-676) in acute myeloid leukemia: in vivo and in vitro saturation and internalization by leukemic and normal myeloid cells, Blood 97 (10) (2001) 3197–3204. [19] P.R. Crocker, A. Varki, Siglecs, sialic acids and innate immunity, Trends Immunol. 22 (6) (2001) 337–342. [20] P.C. Caron, J.G. Jurcic, A.M. Scott, R.D. Finn, C.R. Divgi, M.C. Graham, I.M. Jureidini, G. Sgouros, D. Tyson, L.J. Old, et al., A phase 1B trial of humanized monoclonal antibody M195 (anti-CD33) in myeloid leukemia: specific targeting without immunogenicity, Blood 83 (7) (1994) 1760–1768. [21] J.G. Jurcic, S.M. Larson, G. Sgouros, M.R. McDevitt, R.D. Finn, C.R. Divgi, A.M. Ballangrud, K.A. Hamacher, D. Ma, J.L. Humm, M.W. Brechbiel, R. Molinet, D.A. Scheinberg, Targeted alpha particle immunotherapy for myeloid leukemia, Blood 100 (4) (2002) 1233–1239. [22] E.L. Sievers, F.R. Appelbaum, R.T. Spielberger, S.J. Forman, D. Flowers, F.O. Smith, K. Shannon-Dorcy, M.S. Berger, I.D. Bernstein, Selective ablation of acute myeloid leukemia using antibody-targeted chemotherapy: a phase I study of an anti-CD33 calicheamicin immunoconjugate, Blood 93 (11) (1999) 3678–3684. [23] M. Schwemmlein, M. Peipp, K. Barbin, D. Saul, B. Stockmeyer, R. Repp, J. Birkmann, F. Oduncu, B. Emmerich, G.H. Fey, A CD33-specific single-chain immunotoxin mediates potent apoptosis of cultured human myeloid leukaemia cells, Br. J. Haematol. 133 (2) (2006) 141–151. [24] R.E. Kontermann, Immunoliposomes for cancer therapy, Curr Opin Mol Ther 8 (1) (2006) 39–45. [25] S. Matozaki, T. Nakagawa, R. Kawaguchi, R. Aozaki, M. Tsutsumi, T. Murayama, T. Koizumi, R. Nishimura, T. Isobe, K. Chihara, Establishment of a myeloid leukaemic cell line (SKNO-1) from a patient with t(8;21) who acquired monosomy 17 during disease progression, Br. J. Haematol. 89 (4) (1995) 805–811. [26] H. Asou, S. Tashiro, K. Hamamoto, A. Otsuji, K. Kita, N. Kamada, Establishment of a human acute myeloid leukemia cell line (Kasumi-1) with 8;21 chromosome translocation, Blood 77 (9) (1991) 2031–2036. [27] U. Schneider, H.U. Schwenk, G. Bornkamm, Characterization of EBV-genome negative “null” and “T” cell lines derived from children with acute lymphoblastic leukemia and leukemic transformed non-Hodgkin lymphoma, Int. J. Cancer 19 (5) (1977) 621–626. [28] S.K. Messerschmidt, A. Kolbe, D. Muller, M. Knoll, J. Pleiss, R.E. Kontermann, Novel single-chain Fv′ formats for the generation of immunoliposomes by site-directed coupling, Bioconjug. Chem. 19 (1) (2008) 362–369. [29] D.L. Iden, T.M. Allen, In vitro and in vivo comparison of immunoliposomes made by conventional coupling techniques with those made by a new post-insertion approach, Biochim. Biophys. Acta 1513 (2) (2001) 207–216.

[30] D. Muller, G. Trunk, A. Sichelstiel, K.A. Zettlitz, M. Quintanilla, R.E. Kontermann, Murine endoglin-specific single-chain Fv fragments for the analysis of vascular targeting strategies in mice, J. Immunol. Methods 339 (1) (2008) 90–98. [31] M. Thomas, A. Gessner, H.P. Vornlocher, P. Hadwiger, J. Greil, O. Heidenreich, Targeting MLL-AF4 with short interfering RNAs inhibits clonogenicity and engraftment of t(4;11)-positive human leukemic cells, Blood 106 (10) (2005) 3559–3566. [32] J.W. Park, K. Hong, D.B. Kirpotin, G. Colbern, R. Shalaby, J. Baselga, Y. Shao, U.B. Nielsen, J.D. Marks, D. Moore, D. Papahadjopoulos, C.C. Benz, Anti-HER2 immunoliposomes: enhanced efficacy attributable to targeted delivery, Clin. Cancer Res. 8 (4) (2002) 1172–1181. [33] P. Sapra, E.H. Moase, J. Ma, T.M. Allen, Improved therapeutic responses in a xenograft model of human B lymphoma (Namalwa) for liposomal vincristine versus liposomal doxorubicin targeted via anti-CD19 IgG2a or Fab′ fragments, Clin. Cancer Res. 10 (3) (2004) 1100–1111. [34] C. Marty, P. Scheidegger, K. Ballmer-Hofer, R. Klemenz, R.A. Schwendener, Production of functionalized single-chain Fv antibody fragments binding to the ED-B domain of the B-isoform of fibronectin in Pichia pastoris, Protein Expr. Purif. 21 (1) (2001) 156–164. [35] T. Volkel, P. Holig, T. Merdan, R. Muller, R.E. Kontermann, Targeting of immunoliposomes to endothelial cells using a single-chain Fv fragment directed against human endoglin (CD105), Biochim. Biophys. Acta 1663 (1–2) (2004) 158–166. [36] G.A. Koning, H.W. Morselt, A. Gorter, T.M. Allen, S. Zalipsky, G.L. Scherphof, J.A. Kamps, Interaction of differently designed immunoliposomes with colon cancer cells and Kupffer cells. An in vitro comparison, Pharm. Res. 20 (8) (2003) 1249–1257. [37] L. Nobs, F. Buchegger, R. Gurny, E. Allemann, Biodegradable nanoparticles for direct or two-step tumor immunotargeting, Bioconjug. Chem. 17 (1) (2006) 139–145. [38] U.B. Nielsen, D.B. Kirpotin, E.M. Pickering, K. Hong, J.W. Park, M. Refaat Shalaby, Y. Shao, C.C. Benz, J.D. Marks, Therapeutic efficacy of anti-ErbB2 immunoliposomes targeted by a phage antibody selected for cellular endocytosis, Biochim. Biophys. Acta 1591 (1–3) (2002) 109–118. [39] D. Kirpotin, J.W. Park, K. Hong, S. Zalipsky, W.L. Li, P. Carter, C.C. Benz, D. Papahadjopoulos, Sterically stabilized anti-HER2 immunoliposomes: design and targeting to human breast cancer cells in vitro, Biochem. 36 (1) (1997) 66–75. [40] D.D. Stuart, T.M. Allen, A new liposomal formulation for antisense oligodeoxynucleotides with small size, high incorporation efficiency and good stability, Biochim. Biophys. Acta 1463 (2) (2000) 219–229. [41] D.T. Auguste, K. Furman, A. Wong, J. Fuller, S.P. Armes, T.J. Deming, R. Langer, Triggered release of siRNA from poly(ethylene glycol)-protected, pH-dependent liposomes, J. Control Release 130 (3) (2008) 266–274. [42] X. Zheng, C. Vladau, X. Zhang, M. Suzuki, T.E. Ichim, Z.X. Zhang, M. Li, E. Carrier, B. Garcia, A.M. Jevnikar, W.P. Min, A novel in vivo siRNA delivery system specifically targeting dendritic cells and silencing CD40 genes for immunomodulation, Blood 113 (12) (2009) 2646–2654. [43] M.S. Shim, Y.J. Kwon, Controlled delivery of plasmid DNA and siRNA to intracellular targets using ketalized polyethylenimine, Biomacromolecules 9 (2) (2008) 444–455. [44] A. Akinc, M. Thomas, A.M. Klibanov, R. Langer, Exploring polyethyleniminemediated DNA transfection and the proton sponge hypothesis, J. Gene. Med. 7 (5) (2005) 657–663. [45] U.T. Brunk, H. Dalen, K. Roberg, H.B. Hellquist, Photo-oxidative disruption of lysosomal membranes causes apoptosis of cultured human fibroblasts, Free Radic. Biol. Med. 23 (4) (1997) 616–626. [46] I. Honore, S. Grosse, N. Frison, F. Favatier, M. Monsigny, I. Fajac, Transcription of plasmid DNA: influence of plasmid DNA/polyethylenimine complex formation, J. Control Release 107 (3) (2005) 537–546. [47] Y. Sakurai, H. Hatakeyama, H. Akita, M. Oishi, Y. Nagasaki, S. Futaki, H. Harashima, Efficient short interference RNA delivery to tumor cells using a combination of octaarginine, GALA and tumor-specific, cleavable polyethylene glycol system, Biol. Pharm. Bull. 32 (5) (2009) 928–932. [48] P. Simard, J.C. Leroux, pH-sensitive immunoliposomes specific to the CD33 cell surface antigen of leukemic cells, Int. J. Pharm. 381 (2) (2009) 86–96.