Specific systemic nonviral gene delivery to human hepatocellular carcinoma xenografts in SCID mice

Specific Systemic Nonviral Gene Delivery to Human Hepatocellular Carcinoma Xenografts in SCID Mice 3 Matthew Allen,2 ¨ Markus F. Wolschek,1,2 Christiane Thallinger,1 Malgorzata Kursa,3 Vanessa Rossler, Cornelia Lichtenberger,2 Ralf Kircheis,3 Trevor Lucas,1 Martin Willheim,4 Walter Reinisch,2 Alfred Gangl,2 Ernst Wagner,3,5 and Burkhard Jansen1

Systemic tumor-targeted gene delivery is attracting increasing attention as a promising alternative to conventional therapeutical strategies. To be considered as a viable option, however, the respective transgene has to be administered with high tumor specificity. Here, we describe novel polyethylenimine (PEI)-based DNA complexes, shielded by covalent attachment of polyethylene glycol (PEG), that make use of epidermal growth factor (EGF) as a ligand for targeting gene delivery to EGF receptor-expressing human hepatocellular carcinoma (HCC) cells. In vitro transfection of luciferase reporter DNA resulted in high levels of gene expression in the human HCC cell lines Huh-7 and HepG2. An excess of free EGF during transfection clearly reduced expression levels, indicating a specific EGF receptor-mediated uptake of the DNA particles. Following intravenous injection into human HCC xenograft-bearing SCID mice, luciferase expression was predominantly found in the tumor, with levels up to 2 logs higher than in the liver, which was the highest expressing major organ. Histologic investigation showed reporter gene expression (␤-galactosidase) localized to tumor cells. Assessing DNA distribution within the tumor by immunofluorescence microscopy, rhodamine-labelled transgene DNA was found to be mainly associated with HCC cells. In the liver, DNA was taken up almost exclusively by Kupffer cells and, as indicated by the low expression, subsequently degraded. In conclusion, we have shown that intravenous injection of PEGylated EGF-containing DNA/PEI complexes allows for highly specific expression of a transgene in human HCC tumors. (HEPATOLOGY 2002;36:1106-1114.)

I

ncreased incidence of hepatocellular carcinoma (HCC), which represents the third leading cause of cancer mortality worldwide, has recently been reported.1,2 Because of advanced tumor stage or reduced

Abbreviations: HCC, hepatocellular carcinoma; PEI, polyethylenimine; PEG, polyethylene glycol; EGF, epidermal growth factor; SCID, severe combined immunodeficient; mEGF, mouse epidermal growth factor; MW, molecular weight; FCS, fetal calf serum; FITC, florescein isothiocyanate. From the 1Department of Clinical Pharmacology, Section of Experimental Oncology, and 2Department of Internal Medicine IV, Division of Gastroenterology and Hepatology, University of Vienna, Vienna; 3Boehringer Ingelheim Austria, Vienna; 4Department of Pathophysiology, University of Vienna, Vienna, Austria; and 5Pharmaceutical Biology-Biotechnology, Department of Pharmacy, Muenchen, Germany. Received March 13, 2002; accepted August 2, 2002. Supported by Boehringer Ingelheim Austria, the Austrian National Bank (Grant 7883), and the “Medizinisch-wissenschaftlicher Fonds des Buergermeisters der Bundeshauptstadt Wien” (Grant 1967). R.K.’s present address is Igeneon AG, Immunotherapy of Cancer, Brunner Str. 59, A-1230 Vienna, Austria. B.J.’s present address is Prostate Centre, University of British Columbia D9, 2733 Heather St., Vancouver, BC, Canada V5Z 3J5. Address reprint requests to: Markus Wolschek, Ph.D., Department of Clinical Pharmacology, Section of Experimental Oncology, University of Vienna, Waehringer Guertel 18-20, A-1090 Vienna, Austria. E-mail: markus.wolschek@ akh-wien.ac.at; fax: (43) 1 40400 2998. Copyright © 2002 by the American Association for the Study of Liver Diseases. 0270-9139/02/3605-0013$35.00/0 doi:10.1053/jhep.2002.36372 1106

organ function, only palliative treatment is applicable in more than half of patients at diagnosis.3 Novel therapeutic concepts are therefore warranted to improve the poor outcome in patients with advanced HCC. Tumor-targeted gene delivery represents an attractive alternative compared with conventional therapies. The tumor-specific expression of cytotoxic cytokines such as tumor necrosis factor-␣ or prodrug converting enzymes such as herpes simplex virus-thymidine kinase has been shown to be efficient for the treatment of malignancies in a variety of animal studies.4-6 By locally expressing the therapeutic gene, systemic toxicity such as that reported for tumor necrosis factor-␣7 can be drastically reduced. To date, all successful strategies for systemically applied hepatoma-targeted gene transfer have made use of viral vectors4,8,9 because of their ability to transfect a broad range of cells with high efficiency. However, several important issues, including safety, costs, and higher flexibility regarding the size of the delivered gene, make the development of nonviral vectors very attractive. Whereas nonviral gene delivery in vitro, at least to dividing cells, is not a major problem nowadays, systemic in vivo gene delivery still faces a variety of difficulties. First, the administered DNA has to be protected from serum nucleases. Second, transfection complexes have to be reasonably in-

HEPATOLOGY, Vol. 36, No. 5, 2002

ert against unspecific interactions with nontarget cells, the extracellular matrix, and biologic fluids, but should be able to bind specifically to the target cell. Complexes should also be small enough to pass physical barriers such as fenestrations in the tumor vasculature and, once taken up by the target cell, should have properties allowing endosomal escape to avoid degradation by lysosomal enzymes. Polyethylenimine (PEI)-based synthetic vectors have proven to allow efficient gene delivery to cells in culture as well as in vivo.10-16 Besides condensing and protecting the DNA to be delivered, PEI supports the endosomal escape of the DNA complexes.10,17 Although usually positive charged DNA/PEI particles are adequate for in vitro applications, additional modifications have to be introduced to add sufficient target cell specificity for systemic in vivo application. To reduce nonspecific binding caused by the positive charge and to prolong the circulation half-life of the complexes in the blood stream, hydrophilic polymers such as polyethylene glycol (PEG) can be attached covalently.14 To achieve cell-specific receptor-mediated endocytosis, a variety of ligands have been conjugated to PEI.18-23 Epidermal growth factor (EGF) is of special interest because its receptor is overexpressed in many human tumors, including HCC (for reviews see Arteaga24 and Kim et al.25 and references therein). Accordingly, the incorporation of EGF has been shown to increase the transfection efficacy of PEGylated DNA/PEI complexes on cultured tumor cells.26 In the current work, we describe novel and optimized PEGylated DNA/PEI complexes that employ EGF as a ligand for targeting to human HCC cells. We also show that systemic administration of these complexes allows highly specific transgene expression in human HCC tumors grown in severe combined immunodeficient (SCID) mice.

Materials and Methods Chemicals. PEI, branched, molecular weight 25 kd (PEI25), and linear PEI (22 kd; PEI22) were obtained from Sigma-Aldrich (Milwaukee, WI) and Euromedex (Souffelweyersheim, France), respectively. Mouse epidermal growth factor (mEGF) was purchased from Serotec (Oxford, United Kingdom); ␣-maleimide-␻-N-hydroxysuccinimide ester-polyethylene glycol (molecular weight [MW] 3,400), ␣-methoxy-␻-propionyl-N-hydroxysuccinimide ester-polyethylene glycol (MW 20,000), as well as branched ␣-methoxy-␻-propionyl-N-hydroxysuccinimide ester-polyethylene glycol (MW 40,000) from Shearwater Polymers (Birmingham, AL); cell culture media, fetal calf serum (FCS), and antibiotics from Life

WOLSCHEK ET AL.

1107

Technologies (Gaithersburg, MD); HEPES, Nonidet P-40, 1,4-dithio-L-threitol, Tween-20 from Sigma (St. Louis, MO); and succinimidyl 3-(2-pyridyldithio)-propionate from Fluka (Buchs, Switzerland). Synthesis of mEGF-PEG-PEI25 Conjugate. The synthesis of thiol-functionalized mEGF is described elsewhere.26 The amount of EGF was determined by measuring the absorbance at 280 nm. PEI25 as hydrochloride in water was subjected to gel filtration with 0.25 mol/L NaCl on a Sephadex G-25 superfine column (Pharmacia, Uppsala, Sweden; 300 ⫻ 10 mm). Forty milligrams (1.6 ␮mol) of PEI25 in 1.6 mL of 0.225 mol/L NaCl (the pH was adjusted to 4.4 by addition of 32% hydrochloric acid) was mixed with 21.8 mg (6.4 ␮mol) of ␣-maleimide-␻-N-hydroxysuccinimide ester-polyethylene glycol in 0.4 mL of water. After 1 hour at room temperature, the product was isolated by cation-exchange chromatography (Macro-prep High S; Bio-Rad, Hercules, CA; 100 ⫻ 10 mm column) with a salt gradient from 1.0 to 3.0 mol/L NaCl in 20 mmol/L sodium acetate (NaOAc), pH 4.5, to give 25 mg PEI25 (1 ␮mol; eluted at 3 mol/L salt), modified with PEG with 2 ␮mol of maleimide linker (molar ratio of 1:2). This PEI25 derivative at 4.8 mg (0.2 ␮mol) in 0.7 mL of 3 mol/L NaCl, 20 mmol/L NaOAc, pH 4.5, was mixed under argon with 2.3 mg (0.4 ␮mol) of mEGF modified with 0.4 ␮mol of mercaptopropionate linker in 5.5 mL of 20 mmol/L HEPES, pH 7.1/20% ethanol at a molar ratio of 1:1 (maleimide linker:SH group) and kept under argon at room temperature for 26 hours. The conjugate was isolated by cation-exchange chromatography (Macroprep; gradient elution 0.66-3 mol/L NaCl in 20 mmol/L HEPES, pH 7.1). Free EGF was eluted at 0.66 mol/L NaCl, whereas the product (8.8 mL) was eluted at approximately 2.4 to 3.0 mol/L NaCl. After dialysis against 2 liters of HEPES buffered saline (150 mmol/L NaCl, 20 mmol/L HEPES, pH 7.3), 10.3 mL of conjugate consisting of 1.25 mg (208 nmol) of mEGF modified through PEG bridges with 3.9 mg (156 nmol) of PEI25 at molar ratio of 1.3:1 was obtained. The yield based on PEI25 was 81%. The conjugate was aliquoted and stored at a concentration of 0.4 mg PEI/mL at ⫺20°C. Synthesis of PEG-PEI22 Conjugates. ␣-Methoxy␻-propionyl-N-hydroxysuccinimide ester-polyethylene glycols, (molecular weights 20,000 and 40,000) were coupled to linear PEI22 at a molar ratio of 0.9:1 and 0.3:1, respectively, to give the derivatives PEG20-PEI22 and PEG40-PEI22 as described elsewhere (Kursa M, Ro¨ ssler V, and Wagner E, manuscript in preparation). Plasmids. The plasmids pCMVluc27 coding for the Photinus pyralis luciferase gene and pCMV␤gal28 coding for ␤-galactosidase were purified using the Endofree Giga

1108

WOLSCHEK ET AL.

Kit (Qiagen, Hilden, Germany). Rhodamine-labelled plasmid DNA (pGeneGrip) encoding GFP was obtained from Gene Therapy Systems (San Diego, CA). Cells and Cell Culture. Human Huh7 (JCRB 0403; Tokyo, Japan) and HepG2 hepatoma (ATCC, HB-8065; Rockville, MD) cells were cultured in DMEM/F12 medium containing 10% FCS and 50 ␮g/mL gentamycin in a fully humidified air atmosphere containing 5% CO2 at 37°C. Preparation of Transfection Complexes. DNA/PEI complexes were prepared by flash mixing of indicated amounts of plasmid DNA with either PEI or the indicated mixture of PEI derivatives at an N/P ratio of 6 (N/P ⫽ molar ratio of PEI nitrogen to DNA phosphate, N/P ⫽ 6: 40 ␮g PEI per 50 ␮g DNA). DNA/PEI complexes containing PEG20-PEI or PEG40-PEI were mixed in 20 mmol/L HEPES at a final DNA concentration of 200 ␮g/mL, snap frozen in liquid nitrogen, and subsequently stored at ⫺20°C or ⫺80°C. All other complexes were prepared in 0.5⫻ HBS at a final DNA concentration of 20 ␮g/mL without freezing. To ensure isoosmolarity, glucose was added from a 50% (wt/vol) stock solution to a final concentration of 5% (for complexes formed in 20 mmol/L HEPES) or 2.5% (for complexes formed in 0.5⫻ HBS). Complexes were kept at room temperature for 15 to 20 minutes after formation or thawing before use in physical characterization or transfection experiments. In Vitro Transfections. The day before transfection, 4 ⫻ 104 cells were seeded in 24-well dishes. Prior to transfection, medium was removed and replaced by 200 ␮L of fresh culture medium containing 10% FCS. Transfection complexes corresponding to 1 ␮g pCMVluc DNA were added to the cells. After 4-hour incubation, transfection medium was removed and replaced by 1 mL fresh culture medium. Twenty-four hours after transfection, cells were lysed in Reporter Lysis Buffer (Promega, Mannheim, Germany) supplemented with 1% (wt/vol) Nonidet P-40, and assayed for luciferase expression using the Luciferase Assay System (Promega). Protein was determined with a BCA Protein Assay Kit (Pierce, Rockford, IL) using bovine serum albumin as a standard. Measurement of Particle Size and Zeta Potential. Particle size of transfection complexes was measured by laser light scattering using a Malvern Zetasizer 3000 (Malvern Instruments, Worcester, United Kingdom) as previously described.14,29 For estimation of the surface charge, transfection complexes were diluted in 10 mmol/L NaCl, and the zeta potential was measured using a Malvern Zetasizer 3000. In Vivo Application of Transfection Complexes in Tumor-Bearing Mice. C.B-17-scid/scid (SCID) mice (7-9 weeks, female; Harlan Winkelmann, Borchen, Ger-

HEPATOLOGY, November 2002

many) were injected subcutaneously with human hepatoma cells (5 ⫻ 106, Huh-7; 10 ⫻ 106, HepG2) suspended in 100 ␮L PBS. When the tumors had reached a size of approximately 10 to 13 mm in diameter (2-5 weeks, depending on the cell line), transfection complexes (250 ␮L per mouse) were injected slowly into the tail vein. All animal studies were performed according to the local guidelines for animal care and protection. Luciferase Assay After In Vivo Application. To examine luciferase reporter expression, animals were injected with transfection complexes containing 50 ␮g of the plasmid pCMVluc and killed 48 hours after application by cervical dislocation. Tumors as well as major organs were resected, homogenized in 2 mL 250 mmol/L Tris-HCl buffer, pH 7.5, containing 2 mmol/L 1,4-dithio-L-threitol using an Ultra-Turrax (IKA, Staufen, Germany) homogenizer, frozen in liquid nitrogen, and stored at ⫺80°C. After thawing, cell lysates were centrifuged for 5 minutes at 16,000g, 4°C. Relative luciferase units were determined from 10 ␮L of the supernatant using an LB9507 luminometer (Berthold, Bad Wildbad, Germany) with 10-second integration after injection of 200 ␮L Luciferase Assay System solution (Promega) and values adjusted for background. Examination of ␤-Galactosidase Expression After In Vivo Application. For histologic examination of ␤-galactosidase expression, mice were injected with complexes containing 50 ␮g pCMV␤gal DNA or 50 ␮g pCMVluc DNA as a control. Forty-eight hours after application, tumors were resected, embedded in OCT medium (Sakura Finetek, Zoeterwoude, The Netherlands), and frozen in liquid nitrogen. Cryosections (⬃7 ␮m) were fixed for 5 minutes in 0.5% glutaraldehyde (Sigma), stained for 4 to 6 hours using a standard protocol (␤-Gal Staining Kit; Invitrogen, Groningen, The Netherlands), and, finally, counterstained with eosin. Histologic Examination of DNA Distribution After In Vivo Application. For histologic examination of DNA distribution, animals were injected with complexes containing 50 ␮g of rhodamine-labeled plasmid DNA (pGeneGrip). Four hours after injection, tumors and livers were resected and frozen in liquid nitrogen as described above. Immunofluorescence Staining. For immunofluorescence staining, cryosections (⬃7 ␮m) on microslides were fixed in cold acetone. To visualize cell nuclei, slides were incubated with Hoechst No. 33258 (Sigma). Staining of endothelial cells (CD31) was performed with a rat antimouse CD31 mAb (Pharmingen, San Diego, CA; 5 ␮g/ mL) and a mouse anti-rat IgG2a horseradish peroxidase conjugate (Zymed, San Francisco, CA; 1:100), using a Tyramide signal amplification Fluorescein System (NEN

HEPATOLOGY, Vol. 36, No. 5, 2002

WOLSCHEK ET AL.

Life Science Products, Boston, MA) according to the manufacturer’s instructions For specific staining of macrophages, slides were incubated with rat anti-mouse macrophage F4/80 mAb [Clone Cl:A3-1; Serotec; 10 ␮g/mL in I-Block (Tropix, Bedford, MA)] and subsequently with a F(ab)2 goat anti-rat IgG-FITC Ab (Serotec; 7 ␮g/mL in I-Block). Specific staining of hepatoma cells was performed by first incubating the slides with a rabbit antihuman ␣-fetoprotein Ab (NeoMarkers, Fremont, CA; 1:200 in I-Block), followed by incubation with a goat anti-rabbit IgG-FITC Ab (Santa Cruz Biotechnology, Heidelberg, Germany; 1:100 in I-Block). Slides were mounted with Fluoprep medium (bioMe´rieux, Lyon, France) and evaluated under an Axioplan 2 (Zeiss, Vienna, Austria) fluorescence microscope equipped with a Zeiss Axiocam (Zeiss) camera using Axivision 3.0 (Zeiss) software. Images obtained from the individual channels (FITC, Rhodamine, DAPI) were overlaid using Photoshop 5.0 software (Adobe Systems, Unterschleissheim, Germany).

Results Synthesis of mEGF-PEG-PEI25 Conjugate. Previously, we have demonstrated that EGF is more accessible and therefore more efficient as a ligand when conjugated to preformed PEGylated DNA/PEI complexes via a PEG linker.26 To facilitate the process of complex formation, we synthesized a derivative of PEI25 that contains mEGF coupled via a PEG moiety. This conjugate was generated by first modifying PEI25 with heterobifunctional PEG (␣-maleimide-␻-N-hydroxysuccinimide ester-polyethylene glycol) that is highly reactive toward the amino groups of PEI25 through the NHS ester group. The maleimide group at the distal end of the PEG-PEI25 copolymer was subsequently allowed to react with thiol-functionalized mEGF. The resulting EGF-PEG-PEI25 polymer contained 1.3 mEGF peptides on average per PEI molecule. To allow a titration of mEGF molecules incorporated into the DNA/PEI complexes, we synthesized 2 deriva-

1109

tives of linear PEI22 that contained 20 kd PEG (PEG20PEI22) or 40 kd PEG (PEG40-PEI22) at a molar ratio of 0.9:1 or 0.3:1, respectively. Measurement of Surface Charge and Particle Size. We next investigated the influence of the ratios of the PEI components on both the size and surface charge of the resulting DNA/PEI complexes. Generally, complexes were formed at a PEI-nitrogen to DNA-phosphate ratio of 6 (N/P ⫽ 6) and contained PEI22 as a main compacting agent, EGF-PEG-PEI25 to provide the mEGF ligand, and either PEG20-PEI22 or PEG40-PEI22 as a shielding agent to reduce the surface charge (zeta potential). Whereas unshielded DNA/PEI22 complexes displayed a high surface charge (⫹28-30 mV), inclusion of only 10% EGF-PEG-PEI25 reduced the zeta potential to 12 to 14 mV (Table 1). Addition of either PEG20-PEI22 or PEG40-PEI22 caused a further decrease to 0.5 to 7 mV for the optimized complexes composed of PEI22, EGFPEG-PEI25, and PEG20-PEI or PEG40-PEI22, respectively (ratio 6.5:1:2.5). To produce larger quantities of DNA complexes in advance for storage until further use, we measured the size of the PEGylated complexes after formation and subsequently after 1 freeze-thaw cycle. Notably, the size increased after 1 freeze-thaw cycle and was found to be 800 to 1,200 nm for the main population (85%-90%) and 2,500 to 3,000 nm for the remaining subset (Table 1). Complexes different from the optimized composition either formed large, visible aggregates (when less PEG-PEI22 was used) or were significantly smaller (when the relative amount of PEG-PEI22 was increased) and, therefore, exhibited a reduced transfection potential (Kursa M., unpublished data). For convenience, DNA particles generated with the optimized ratio of PEI derivatives will subsequently be called EP20P (containing PEG20-PEI) and EP40P (PEG40-PEI). In Vitro Transfection of Human HCC Cell Lines. We next analyzed the transfection properties of the optimized EP20P and EP40P complexes in the human hepatoma cell lines Huh-7 and HepG2 by employing plasmid

Table 1. Biophysical Characterization of DNA/PEI Complexes After Complex Formation

After Thawing

Composition (Final DNA Concentration; Buffer)

Size (nm)

Zeta (mV)

Size (nm)

Zeta (mV)

PEI22 (20 ␮g/mL; 0.5⫻ HBS) PEI22:EGF-PEG-PEI25 ⫽ 9:1 (20 ␮g/mL; 0.5⫻ HBS) PEI22:EGF-PEG-PEI25:PEG20-PEI22 ⫽ 6.5:1:2.5 (200 ␮g/mL; 20 mmol/L HEPES) PEI22:EGF-PEG-PEI25:PEG40-PEI22 ⫽ 6.5:1:2.5 (200 ␮g/mL; 20 mmol/L HEPES)

900-1,100 (60%) ⬎ 3,000 (40%) 2,500-3,000 190-210

⫹28 to ⫹30 ⫹12 to ⫹14 ND

ND ND ⫹0.5 to ⫹3.0

480-550

ND

ND ND 850-1,200 (90%) 2,500-3,000 (10%) 810-870 (85%) 2,500-3,000 (15%)

⫹3.0 to ⫹7.0

NOTE. Complexes were prepared with the indicated PEI derivatives (ratios are calculated on the basis of PEI N) using an N/P ratio of 6 at a final DNA concentration of 20 or 200 ␮g/mL in either 20 mmol/L HEPES or 0.5 ⫻ HBS as specified. Size (diameter) and zeta potential were measured 20 minutes after formation and thawing, respectively, at a DNA concentration of 20 ␮g/mL. Three measurements were carried out per sample.

1110

WOLSCHEK ET AL.

HEPATOLOGY, November 2002

Gene Delivery After Systemic Application. To investigate whether the optimized EGF-containing complexes are suitable to transfect human HCCs after systemic in vivo application, we implanted human HCC tumors (Huh-7 or HepG2) subcutaneously into SCID mice. Luciferase-encoding plasmid DNA (pCMVluc) complexed with either the EP20P or the EP40P formulation was injected intravenously into the tail vein, and, 48 hours after injection, luciferase expression in the tumor as well as in the major organs was analyzed (Fig. 3). In both models, EP20P as well as EP40P complexes efficiently targeted transgene expression to the tumor, with values up to 2 logs higher than the highest expressing major organ, the liver. Luciferase expression in other organs was minimal (⬍0.5% of levels found in the tumors). Notably, no acute toxicity was observed in any of the animals. Application of EGF-free 20P complexes resulted in clearly reduced expression levels in Huh-7 tumors (approximately 9%; P ⫽ .0006) when compared with EP20P complexes, whereas levels in the other organs were not significantly altered (Fig. 4). Fig. 1. In vitro transfection efficiency of DNA/PEI complexes in human HCC cells. (A) Huh-7 cells and (B) HepG2 cells were transfected in 24-well dishes with the indicated complexes containing 1 ␮g pCMVluc DNA and analyzed for luciferase expression 24 hours after transfection. Relative light units were normalized to protein and are the mean ⫾ SD of 2 independent experiments (n ⫽ 4 wells) expressed on a logarithmic scale. *Statistical significance at P ⬍ .002 (t test).

DNA coding for luciferase (pCMVluc). As an internal reference, complexes prepared in 0.5⫻ HBS with only PEI25 or PEI22 were used. Both cell lines showed high levels of luciferase expression after transfection with EP20P as well as EP40P complexes (Fig. 1). No significant differences were seen whether the 20 kd or the 40 kd PEG molecule was incorporated. Notably, expression levels were even slightly higher (although not statistically significant) than those obtained with the strongly positive charged DNA/PEI22 complexes. To see whether the incorporated mEGF indeed serves as a ligand, competition assays were performed. Addition of free mEGF (2 ␮g ⫽ 80 molar equivalents excess) to the cell culture medium during transfection of Huh-7 cells reduced expression levels to 8% and 13% of original values for the EP20P and EP40P formulations, respectively (Fig. 2A), that were found to be in the same range as for EGF-free PEGylated complexes formed under identical conditions (20P; PEI22:PEI25:PEG20-PEI ⫽ 6.5:1:2.5; Fig. 2B). Increasing the amount of free mEGF to 10 ␮g caused only a minimal further reduction. In contrast, transfection efficiencies of EGF-free DNA/PEI22 and 20P complexes were not influenced by the addition of free mEGF.

Fig. 2. Competitive inhibition of EGF receptor-mediated gene delivery. Huh-7 cells were transfected in 24-well dishes with (A and B) EP20P, (A) EP40P, (A) EGF-free PEI22, or (B; PEI22:PEI25:PEG20-PEI ⫽ 6.5:1:2.5, mixed under the same conditions as EP20P) EGF-free PEGylated 20P complexes containing 1 ␮g pCMVluc DNA in the presence of 2 or 10 ␮g/well free mEGF as indicated. Luciferase activity was normalized to protein and is given as mean ⫾ SD of 2 independent experiments (n ⫽ 4 [A], 6 [B] wells). *Statistical significance at P ⬍ .0053 (t test).

HEPATOLOGY, Vol. 36, No. 5, 2002

WOLSCHEK ET AL.

1111

Fig. 3. Gene expression in vivo after systemic application of PEGylated DNA complexes. (A and C) EP20P or (B and D) EP40P complexes containing 50 ␮g pCMVluc DNA were injected into the tail vein of SCID mice bearing (A and B) subcutaneous Huh-7 or (C and D) HepG2 tumors. Transgene expression at 48 hours after application was determined by luciferase assay. Luciferase activities are given as relative light units per organ on a logarithmic scale and are the mean ⫾ SD of at least 2 independent experiments (n ⫽ 10 [A], 14 [B], 9 [C], 8 [D] mice). *Statistical significance at P ⬍ .0001 (t test).

To assess transgene expression in the tumors histologically, a plasmid coding for ␤-galactosidase complexed with the 40-kd PEG formulation was injected intravenously into Huh-7 tumor-bearing animals. Reporter gene expression was found throughout the tumor, although the pattern was rather heterogeneous, showing larger areas without any significant expression. Transfected tumor cells mainly appeared clustered, often in the vicinity of vessel-like structures (Fig. 5). No staining was found in control animals that received complexes containing pCMVluc DNA. We further investigated the biodistribution of EP40P transfection complexes 4 hours after systemic application within the liver and within the tumor by using rhodamine-labeled plasmid DNA (pGeneGrip).30 In the liver, labeled DNA was found to be homogeneously distributed (Fig. 6). Staining with an antibody specific for macrophages (F4/80) revealed that DNA uptake was restricted to Kupffer cells (Fig. 6C and D). No DNA was found to

Fig. 4. Gene expression in vivo after systemic application of PEGylated DNA complexes. Application of EP20P or EGF-free 20P complexes as well as analysis for luciferase expression was performed as described in Fig. 3 (n ⫽ 5 mice). *Statistical significance at P ⬍ .0006 (t test).

be associated with endothelial cells as assessed by antiCD31 staining (Fig. 6A and B) or hepatocytes (cells negative for F4/80 and CD31), respectively. In contrast to the liver, distribution within the tumor was clearly more heterogeneous, showing transgene DNA localized to focal areas. Large regions were without any significant uptake. Most of the DNA was associated with tumor cells as indicated by staining with ␣-fetoprotein antibody specific for human hepatoma cells (Fig. 7A). To a lesser extent, DNA was taken up by macrophages (Fig. 7B). Sometimes, clusters of rhodamine fluorescence were found in close proximity to blood vessels (Fig. 7C and D).

Discussion Tumor targeted gene delivery via the systemic route is one of the major challenges in gene therapy. Reports on

Fig. 5. ␤-Galactosidase expression in Huh-7 tumors. EP40P complexes containing 50 ␮g pCMV␤gal DNA were injected into the tail vein of Huh-7 tumor bearing SCID mice. ␤-Galactosidase expression at 48 hours after application was visualized by X-gal staining and evaluated under a light microscope. (Original magnification ⫻200.)

1112

WOLSCHEK ET AL.

Fig. 6. Uptake of plasmid DNA in the liver visualized by immunofluorescence microscopy. EP40P complexes containing 50 ␮g of rhodamine-labeled plasmid DNA (pGeneGrip) were injected into the tail vein of SCID mice bearing subcutaneous Huh-7 tumors. Plasmid DNA distribution is indicated by the red fluorescence of pGeneGrip. Nuclei are visualized by Hoechst staining (blue). Endothelial cells were identified with anti-CD31 antibody (FITC, green; A and B), and macrophages were stained with anti-F40/80 antibody (FITC, green; C and D). Slides were evaluated under a fluorescence microscope. (Original magnification [A] ⫻100, [B] ⫻200, and [C and D] ⫻400.)

specific systemic gene delivery to HCC are particularly sparse. Adenoviral vectors, despite their potential to transfect cells, have the clear drawback of predominantly transfecting hepatocytes upon intravenous application.31-33 Although administration via the portal vein or the hepatic artery improves gene transfer into hepatic HCC nodules, high-expression levels are nevertheless found in the surrounding hepatocytes.34,35 Strategies to tackle these problems include the use of tumor-specific promoters allowing transcriptional targeting of recombinant adenoviruses to HCC tumors.4,8,9 Alternatively, adenoviral vectors have been retargeted to EGF receptor-expressing tumor cells by means of bispecific antibodies.36,37 There is, however, an urgent need to further develop improved methods of tumor-specific targeting to HCC. Here, we describe the specific targeting of gene delivery to subcutaneous xenotransplants of human HCC by systemic administration of PEGylated DNA/PEI complexes that carry mEGF as a ligand. In a recent study, we demonstrated the successive conjugation of PEG and mEGF to preformed PEI/DNA complexes, which yielded highexpression levels upon in vitro transfection of EGF receptor-positive cells.26 The current study extends this strategy to the effective targeting of HCC in vitro and in vivo and presents a novel method of complex production. To streamline the experimental process and design, both the PEG and the EGF moiety were covalently linked

HEPATOLOGY, November 2002

to PEI in advance of DNA particle preparation. Following the chemical syntheses of PEG20-PEI22 and PEG40PEI22 as well as EGF-PEG-PEI25, zeta-potential measurements showed that the surface charge of DNA particles, indeed, was strongly reduced by including PEGylated PEI derivatives. Besides having a low zeta potential to reduce nonspecific interactions following systemic administration, DNA/PEI complexes should be at least several hundred nanometers in size to allow efficient escape from endosomes following cellular endocytosis.29 Laser light-scattering measurements demonstrated that the main particle population of both EP20P and EP40P complexes range in size from 800 to 1,200 nm. Therefore, the size distribution is similar to DNA/PEI22 complexes, which are known to have high-transfection potential.26,38-40 Accordingly, transfection experiments in vitro revealed high expression levels in the 2 human HCC cell lines tested. Competition with free mEGF clearly reduced the transfection levels in Huh-7 cells, indicating that gene delivery is indeed mediated by mEGF binding. In contrast to a previous report,41 PEG grafting of PEI did not interfere with particle formation and subsequent transfection. This may be explained by the different PEI backbone used in this study as well as the fact that the main part (65%) of PEI used for DNA compaction was free PEI22.

Fig. 7. Uptake of plasmid DNA in Huh-7 tumors visualized by immunofluorescence microscopy. EP40P complexes containing 50 ␮g of rhodamine-labeled plasmid DNA (pGeneGrip) were injected into the tail vein of SCID mice bearing subcutaneous Huh-7 tumors. Plasmid DNA distribution is indicated by the red fluorescence of pGeneGrip. Nuclei are visualized by Hoechst staining (blue). Tumor cells were positive for anti-human ␣-fetoprotein antibody (FITC, green; A), endothelial cells were identified with anti-CD31 antibody (FITC, green; C and D), and macrophages were stained with anti-F40/80 antibody (FITC, green; B). Slides were evaluated under a fluorescence microscope. (Original magnification A, C, and D, ⫻ 400; B, ⫻200 magnification.)

HEPATOLOGY, Vol. 36, No. 5, 2002

EP20P and EP40P complexes were shown to be highly active in cell culture transfection experiments. Even more importantly, the low surface charge of complexes made them suitable for systemic use in vivo. This hypothesis was further explored in 2 xenotransplantation models for human HCC. Following intravenous application in SCID mice, analysis of luciferase distribution in both models showed that expression levels within the tumors were up to 2 logs higher than levels found in the liver, which was the highest expressing major organ, with expression in the tumor accounting for 97% to 99% of the total transgene expression measured in vivo. Notably, no acute toxicity, as reported for positively charged DNA/PEI particles,13,14 was observed. Besides low surface charge, which enables a prolonged circulation by reducing nonspecific interactions with blood components and nontarget cell,14 the size of the complexes may be another parameter that contributes to the high tumor specificity observed. Assuming relatively large pores within the tumor vasculature (200-1,200 nm)42 as compared with normal tissue (e.g., 100-120 nm for the liver and even less for other organs), DNA complexes of 800 to 1,200 nm in size as used in this study may simply be too large to extravasate into normal tissue and, therefore, predominantly may infiltrate tumor tissue (passive targeting). Furthermore, transfection efficiency with nonviral vectors is strongly dependent on the breakdown of the nuclear membrane during cell division,43 which is rather low in normal tissues compared with highly proliferating tumor cells. Both effects may also explain the expression in the tumor, which is found, although to a much lesser degree (than with targeted complexes), even after administration of nontargeted complexes. However, the in vitro competition studies as well as the significantly lower transfection levels in the tumors observed following systemic application of EGF-free complexes indicate that the interaction between EGF receptors on the surface of HCC cells with the EGF ligand displayed by the particles is another integral part of the tumor targeting observed. In situ staining of transgene expression in the tumor following injection of a ␤-galactosidase encoding plasmid showed focal areas of highly expressing cells, which were clearly identified as tumor cells based on their morphology. Accordingly, DNA uptake was restricted to focal areas, therein being associated with tumor cells and to a lesser extent with macrophages, as assessed by immunofluorescence microscopy. Large areas within the tumor remained nontransfected and were without significant DNA uptake. Similarly, smaller long-circulating liposomes (100-200 nm) were reported to extravasate rather diffusely, whereas larger particles (380-780 nm) escaped from the tumor vessels in a focal manner.42 Keeping a

WOLSCHEK ET AL.

1113

possible contribution of particle size to tumor specificity in mind, a size reduction of EP20P and EP40P complexes might, therefore, improve their diffusion within the tumor. However, similar transfection levels were found to be sufficient to cause significant tumor regression in various murine tumor models following systemic application of a tumor necrosis factor-␣-encoding plasmid complexed with transferrin PEI.44,45 A considerable part of the injected DNA was found to be distributed evenly within the liver, where it was associated almost exclusively with Kupffer cells. This is in accordance with the low luciferase levels found in the liver because Kupffer cells are known to function in the uptake and degradation of foreign particles. A similar DNA distribution within the liver was previously reported upon intravenous injection of transferrin-shielded DNA/PEI complexes16 and DNA/poly-L-lysine complexes.46 In conclusion, we have shown that intravenous injection of EP20P and EP40P DNA complexes allows for highly specific expression of a transgene in human HCC tumors in vivo without signs of acute toxicity. There are certainly a number of hurdles remaining to be overcome before the use of this strategy can be considered as a treatment option for patients with HCC. However, our findings clearly support the potential of nonviral gene delivery strategies for the treatment of this malignancy.

References 1. El-Serag HB, Mason AC. Rising incidence of hepatocellular carcinoma in the United States. N Engl J Med 1999;340:745-750. 2. Murray CJ, Lopez AD. Mortality by cause for eight regions of the world: Global Burden of Disease Study. Lancet 1997;349:1269-1276. 3. Schafer DF, Sorrell MF. Hepatocellular carcinoma. Lancet 1999;353: 1253-1257. 4. Bilbao R, Gerolami R, Bralet MP, Qian C, Tran PL, Tennant B, Prieto J, et al. Transduction efficacy, antitumoral effect, and toxicity of adenovirusmediated herpes simplex virus thymidine kinase/ganciclovir therapy of hepatocellular carcinoma: the woodchuck animal model. Cancer Gene Ther 2000;7:657-662. 5. Gerolami R, Cardoso J, Lewin M, Bralet MP, Sa Cunha A, Clement O, Brechot C, et al. Evaluation of HSV-tk gene therapy in a rat model of chemically induced hepatocellular carcinoma by intratumoral and intrahepatic artery routes. Cancer Res 2000;60:993-1001. 6. Wright P, Zheng C, Moyana T, Xiang J. Intratumoral vaccination of adenoviruses expressing fusion protein RM4/tumor necrosis factor (TNF)-␣ induces significant tumor regression. Cancer Gene Ther 1998; 5:371-379. 7. Mueller H. Tumor necrosis factor as an antineoplastic agent: pitfalls and promises. Cell Mol Life Sci 1998;54:1291-1298. 8. Li Y, Yu DC, Chen Y, Amin P, Zhang H, Nguyen N, Henderson DR. A hepatocellular carcinoma-specific adenovirus variant, CV890, eliminates distant human liver tumors in combination with doxorubicin. Cancer Res 2001;61:6428-6436. 9. Ohashi M, Kanai F, Tateishi K, Taniguchi H, Marignani PA, Yoshida Y, Shiratori Y, et al. Target gene therapy for ␣-fetoprotein-producing hepatocellular carcinoma by E1B55k-attenuated adenovirus. Biochem Biophys Res Commun 2001;282:529-535. 10. Boussif O, Lezoualc’h F, Zanta MA, Mergny MD, Scherman D, Demeneix B, Behr JP. A versatile vector for gene and oligonucleotide transfer into

1114

11.

12.

13.

14.

15.

16.

17. 18.

19.

20.

21.

22.

23.

24.

25. 26.

27.

28. 29.

WOLSCHEK ET AL.

cells in culture and in vivo: polyethylenimine. Proc Natl Acad Sci U S A 1995;92:7297-7301. Abdallah B, Hassan A, Benoist C, Goula D, Behr JP, Demeneix BA. A powerful nonviral vector for in vivo gene transfer into the adult mammalian brain: polyethylenimine. Hum Gene Ther 1996;7:1947-1954. Coll JL, Chollet P, Brambilla E, Desplanques D, Behr JP, Favrot M. In vivo delivery to tumors of DNA complexed with linear polyethylenimine. Hum Gene Ther 1999;10:1659-1666. Kircheis R, Schuller S, Brunner S, Ogris M, Heider KH, Zauner W, Wagner E. Polycation-based DNA complexes for tumor-targeted gene delivery in vivo. J Gene Med 1999;1:111-120. Ogris M, Brunner S, Schuller S, Kircheis R, Wagner E. PEGylated DNA/ transferrin-PEI complexes: reduced interaction with blood components, extended circulation in blood and potential for systemic gene delivery. Gene Ther 1999;6:595-605. Zou SM, Erbacher P, Remy JS, Behr JP. Systemic linear polyethylenimine (L-PEI)-mediated gene delivery in the mouse. J Gene Med 2000;2:128134. Kircheis R, Wightman L, Schreiber A, Robitza B, Rossler V, Kursa M, Wagner E. Polyethylenimine/DNA complexes shielded by transferrin target gene expression to tumors after systemic application. Gene Ther 2001; 8:28-40. Kichler A, Leborgne C, Coeytaux E, Danos O. Polyethylenimine-mediated gene delivery: a mechanistic study. J Gene Med 2001;3:135-144. Kircheis R, Kichler A, Wallner G, Kursa M, Ogris M, Felzmann T, Buchberger M, et al. Coupling of cell-binding ligands to polyethylenimine for targeted gene delivery. Gene Ther 1997;4:409-418. Zanta MA, Boussif O, Adib A, Behr JP. In vitro gene delivery to hepatocytes with galactosylated polyethylenimine. Bioconjug Chem 1997;8:839844. Bettinger T, Remy JS, Erbacher P. Size reduction of galactosylated PEI/ DNA complexes improves lectin- mediated gene transfer into hepatocytes. Bioconjug Chem 1999;10:558-561. Diebold SS, Lehrmann H, Kursa M, Wagner E, Cotten M, Zenke M. Efficient gene delivery into human dendritic cells by adenovirus polyethylenimine and mannose polyethylenimine transfection. Hum Gene Ther 1999;10:775-786. Diebold SS, Kursa M, Wagner E, Cotten M, Zenke M. Mannose polyethylenimine conjugates for targeted DNA delivery into dendritic cells. J Biol Chem 1999;274:19087-19094. Erbacher P, Remy JS, Behr JP. Gene transfer with synthetic virus-like particles via the integrin-mediated endocytosis pathway. Gene Ther 1999; 6:138-145. Arteaga CL. The epidermal growth factor receptor: from mutant oncogene in nonhuman cancers to therapeutic target in human neoplasia. J Clin Oncol 2001;19:32S-40S. Kim ES, Khuri FR, Herbst RS. Epidermal growth factor receptor biology (IMC-C225). Curr Opin Oncol 2001;13:506-513. Blessing T, Kursa M, Holzhauser R, Kircheis R, Wagner E. Different strategies for formation of pegylated EGF-conjugated PEI/DNA complexes for targeted gene delivery. Bioconjug Chem 2001;12:529-537. Plank C, Zatloukal K, Cotten M, Mechtler K, Wagner E. Gene transfer into hepatocytes using asialoglycoprotein receptor mediated endocytosis of DNA complexed with an artificial tetra-antennary galactose ligand. Bioconjug Chem 1992;3:533-539. MacGregor GR, Caskey CT. Construction of plasmids that express E. coli ␤-galactosidase in mammalian cells. Nucleic Acids Res 1989;17:2365. Ogris M, Steinlein P, Kursa M, Mechtler K, Kircheis R, Wagner E. The size of DNA/transferrin-PEI complexes is an important factor for gene expression in cultured cells. Gene Ther 1998;5:1425-1433.

HEPATOLOGY, November 2002

30. Zelphati O, Liang X, Hobart P, Felgner PL. Gene chemistry: functionally and conformationally intact fluorescent plasmid DNA. Hum Gene Ther 1999;10:15-24. 31. Yoon SK, Armentano D, Wands JR, Mohr L. Adenovirus-mediated gene transfer to orthotopic hepatocellular carcinomas in athymic nude mice. Cancer Gene Ther 2001;8:573-579. 32. Herz J, Gerard RD. Adenovirus-mediated transfer of low-density lipoprotein receptor gene acutely accelerates cholesterol clearance in normal mice. Proc Natl Acad Sci U S A 1993;90:2812-2816. 33. Wood M, Perrotte P, Onishi E, Harper ME, Dinney C, Pagliaro L, Wilson DR. Biodistribution of an adenoviral vector carrying the luciferase reporter gene following intravesical or intravenous administration to a mouse. Cancer Gene Ther 1999;6:367-372. 34. Bilbao R, Bustos M, Alzuguren P, Pajares MJ, Drozdzik M, Qian C, Prieto J. A blood-tumor barrier limits gene transfer to experimental liver cancer: the effect of vasoactive compounds. Gene Ther 2000;7:1824-1832. 35. Gerolami R, Cardoso J, Bralet MP, Cuenod CA, Clement O, Tran PL, Brechot C. Enhanced in vivo adenovirus-mediated gene transfer to rat hepatocarcinomas by selective administration into the hepatic artery. Gene Ther 1998;5:896-904. 36. Haisma HJ, Grill J, Curiel DT, Hoogeland S, van Beusechem VW, Pinedo HM, Gerritsen WR. Targeting of adenoviral vectors through a bispecific single-chain antibody. Cancer Gene Ther 2000;7:901-904. 37. Grill J, Van Beusechem VW, Van Der Valk P, Dirven CM, Leonhart A, Pherai DS, Haisma HJ, et al. Combined targeting of adenoviruses to integrins and epidermal growth factor receptors increases gene transfer into primary glioma cells and spheroids. Clin Cancer Res 2001;7:641-650. 38. Ferrari S, Moro E, Pettenazzo A, Behr JP, Zacchello F, Scarpa M. ExGen 500 is an efficient vector for gene delivery to lung epithelial cells in vitro and in vivo. Gene Ther 1997;4:1100-1106. 39. Chemin I, Moradpour D, Wieland S, Offensperger WB, Walter E, Behr JP, Blum HE. Liver-directed gene transfer: a linear polyethlenimine derivative mediates highly efficient DNA delivery to primary hepatocytes in vitro and in vivo. J Viral Hepat 1998;5:369-375. 40. Wightman L, Kircheis R, Rossler V, Carotta S, Ruzicka R, Kursa M, Wagner E. Different behavior of branched and linear polyethylenimine for gene delivery in vitro and in vivo. J Gene Med 2001;3:362-372. 41. Erbacher P, Bettinger T, Belguise-Valladier P, Zou S, Coll JL, Behr JP, Remy JS. Transfection and physical properties of various saccharide, poly (ethylene glycol), and antibody-derivatized polyethylenimines (PEI). J Gene Med 1999;1:210-222. 42. Hobbs SK, Monsky WL, Yuan F, Roberts WG, Griffith L, Torchilin VP, Jain RK. Regulation of transport pathways in tumor vessels: role of tumor type and microenvironment. Proc Natl Acad Sci U S A 1998;95:46074612. 43. Brunner S, Sauer T, Carotta S, Cotten M, Saltik M, Wagner E. Cell cycle dependence of gene transfer by lipoplex, polyplex, and recombinant adenovirus. Gene Ther 2000;7:401-407. 44. Kircheis R, Ostermann E, Wolschek MF, Lichtenberger C, Magin-Lachmann C, Wightman L, Kursa M, et al. Tumor-targeted gene delivery of tumor necrosis factor-␣ induces tumor necrosis and tumor regression without systemic toxicity. Cancer Gene Ther 2002;9:673-680. 45. Kircheis R, Wightman L, Kursa M, Ostermann E, Wagner E. Tumortargeted gene delivery: an attractive strategy to use highly active effector molecules in cancer treatment. Gene Ther 2002;9:731-735. 46. Ward CM, Read ML, Seymour LW. Systemic circulation of poly (Llysine)/DNA vectors is influenced by polycation molecular weight and type of DNA: differential circulation in mice and rats and the implications for human gene therapy. Blood 2001;97:2221-2229.