¹Technische Universität München, Institute of Experimental Oncology, Munich, Germany

²Chemicell, Berlin, Germany

Correspondence to: C Plank, Institute of Experimental Oncology, TU München, Ismaninger Str 22, D-81675 Munich, Germany

^aThe first two authors contributed equally to this work

Low efficiencies of nonviral gene vectors, the receptor-dependent host tropism of adenoviral or low titers of retroviral vectors limit their utility in gene therapy. To overcome these deficiencies, we associated gene vectors with superparamagnetic nanoparticles and targeted gene delivery by application of a magnetic field. This potentiated the efficacy of any vector up to several hundred-fold, allowed reduction of the duration of gene delivery to minutes, extended the host tropism of adenoviral vectors to nonpermissive cells and compensated for low retroviral titer. More importantly, the high transduction efficiency observed in vitro was reproduced in vivo with magnetic field-guided local transfection in the gastrointestinal tract and in blood vessels. Magnetofection provides a novel tool for high throughput gene screening in vitro and can help to overcome fundamental limitations to gene therapy in vivo.

Gene Therapy 2001 9, 102-109. DOI: 10.1038/sj/gt/3301624

magnetofection; gene vectors; gene delivery; magnetic nanoparticles; magnetic drug targeting

Introduction

The true benefits of gene therapy cannot be realized until current gene delivery systems are perfected or new vectors are developed. This is the opinion of many experts in the field. Moreover, assigning function to the recently decoded primary sequences of vertebrate genomes affords rapid and highly efficient gene transfer techniques in vitro amenable to high throughput automatization. Slow vector accumulation and consequently low vector concentration at target tissues have been identified as simple but strong barriers to effective gene delivery.¹ Gene vectors can accumulate in target tissues based on their natural host cell tropisms and on their biophysical properties. Engineering surface proteins of viral vectors,² or coupling targeting ligands to viral,³ as well as nonviral vectors⁴ can further improve tissue selectivity. In practise however, these modalities of targeting are often insufficient for rapid and specific accumulation of active vectors in target tissues and hence for achieving the sustained expression levels required by many applications. Therefore, we applied the principle of magnetic drug targeting⁵ to gene vectors and demonstrated its universal applicability to gene delivery with viral and nonviral vectors. Magnetic targeting exploits paramagnetic particles as drug carriers, guides their accumulation in target tissues with local strong magnetic fields, and has been used with some success in the treatment of cancer patients.⁶ Applying this principle to gene vectors, given the rationale, would set the basis for a method of automatizable high throughput transfection in vitro and, more importantly, improve their efficacy in vivo. Both goals would require rapid transfection kinetics, greatly improved dose-response characteristics and the possibility of vector targeting to a selected area.

Results

To achieve these goals, we associated gene vectors with superparamagnetic iron oxide nanoparticles which were coated with polyethylenimine for this purpose (transMAG^PEI). Vector association with transMAG^PEI was mediated by electrostatic interaction and salt-induced colloid aggregation, a phenomenon well known in colloid chemistry.⁷ In the presence of salt, vector-magnetic particle suspensions associated with linear kinetics, giving rise to particle sizes of 400 to 1000 nm (as determined by dynamic light scattering) at the time point of vector administration. Under the conditions used in gene transfer experiments, the major fraction of the DNA (or virus) dose was associated with magnetic particles (Figure 1). For example, at 2:1 weight ratios of DNA and transMAG^PEI, 92% (naked DNA), 72% (DNA + DOTAP/Chol) and 83% (DNA + polyethylenimine; PEI), respectively, of the ³²P-labeled DNA dose were associated with magnetic particles, while 75-97% of ¹²⁵I-labeled adenovirus (at MOIs of 200 to 25 per 3 g transMAG^PEI, respectively) were associated under the conditions of the gene delivery experiments presented here Figure 1.

Magnetofection in vitro

We constructed magnetic devices with neodymium-iron-boron (Nd-Fe-B) permanent magnets for in vitro gene delivery upon which cell culture plates were positioned during transfection. The magnetic gradient fields of the devices are strong enough to sediment the applied paramagnetic vectors on to, possibly even into cells within a few minutes. This had two immediate consequences which fulfill important prerequisites to practicable automatizable functional gene screening by transfection. Firstly, the largest fraction of the vector dose is available to the cells within a few minutes of incubation, greatly accelerating transfection kinetics. Peak transfection levels can be achieved with as little as 10-min incubation of cells with vector cocktails (Figure 2a), compared with 2 to 4 h required with standard protocols. Secondly, the strongly increased concentration of vectors at the cell surface warrants a dramatically improved dose-response profile Figure 2b.

We formulated state-of-the-art nonviral and viral gene vectors (PEI-DNA,⁸ AVET,⁹ Lipofectamine, GenePorter and DOTAP-Cholesterol,¹⁰ recombinant adenovirus and a retrovirus) as magnetofectins and compared reporter gene expression levels achieved with short-term incubation to the parent standard vector formulations. The data show that the formulation with paramagnetic nanoparticles alone enhanced transfection, while the application of a magnetic field raised reporter gene expression levels up to three orders of magnitude over those achieved with standard vectors under the same conditions (Figure 2a, nonviral vectors; Figure 3, adenoviral vector; Figure 4, retroviral vector).

Nonviral magnetofection

The rapid increase of vector concentration at the cell surface greatly improved the dose-response profile Figure 2b. As a consequence, a minimal DNA dose was sufficient to achieve high transfection levels. We confirmed the potency of magnetofection with calcium phosphate coprecipitation, naked DNA and polylysine-DNA (not shown). Compared with a naked DNA magnetofectin (which hardly yielded any transfection; data not shown), complexation with cationic lipids or PEI Figure 2a and b, as well as the incorporation of the endosomolytic peptide INF7¹¹ (not shown) or a chemically inactivated adenovirus⁹ Figure 2a and c enhanced transfection several thousand-fold. Therefore, the mechanisms of cellular uptake and vector processing are probably similar to those of the parent vectors, namely endocytotic. As a model for in vivo gene delivery, we tested whether gene vectors can be targeted to a selected area of a target tissue by magnetic force and carried out transfection with the LacZ reporter gene in six-well plates (Figure 2c, nonviral; Figure 3a, adenoviral). In fact, gene delivery was confined to an area defined by the shape of the applied magnet and its gradient field. In vitro, this confinement can be useful in characterizing the local actions of gene products, such as factors involved in cell differentiation processes, on neighboring untransfected cells within a single culture dish.

Adenoviral magnetofection

The transduction efficiency of adenoviral vectors is highly dependent on the Coxsackie and adenovirus receptor (CAR) status of target cells. Unfortunately, many important target tissues express variable, little or no CAR including the apical surface of lung epithelium¹² or tumor tissue,¹³ highlighting the importance of extending the host tropisms of adenoviral vectors for gene therapy. Therefore we tested the strength of magnetofection on cells producing little or no CAR (NIH3T3,¹⁴ K562¹⁵ and primary human peripheral blood lymphocytes, PBL¹⁶). The association of virus with paramagnetic particles alone was sufficient to mediate infection of these cell lines, directly dependent on the virus-to-particle ratio Figure 3a. This implies that the cationic magnetic particles mediate cell binding which is followed by internalization. A similar effect has been observed previously when adenovirus was associated with polycations, cationic lipids or particulate structures.¹⁷ However, in our experiments with cationic magnetic particles the observed effect was modest compared with the drastic increase in transfection efficiency contributed by the application of a magnetic field Figure 3. At constant concentration of magnetic nanoparticles, gene transfer and thus expression is increasing with increasing MOI if a magnetic field is applied Figure 3b-d, culminating in a 500-fold enhancement of beta-galactosidase expression as compared with standard infection with Ad in NIH3T3 cells Figure 3b. Transduction of K562 cells was only seen in the presence of magnet and magnetic particles, and was dependent on the MOI used Figure 3c. Stimulated PBL did not express -galactosidase above background upon conventional infection with AdLacZ at an MOI of 200. Expression was achieved by magnetofection, but required a high MOI Figure 3d. Under these conditions, the transduced fraction of cells produced high levels of reporter gene product (Figure 3e, center panel) as judged by the intense X-gal staining.

Retroviral magnetofection

Retroviral vector technology suffers from the difficulty of achieving high viral titers. To test whether magnetofection can compensate for this, transMAG^PEI was associated with an purposefully low titer ecotropic MuLV-based retroviral vector preparation and used to transduce NIH3T3 cells. In comparison to the standard transduction in the presence of polybrene, the association with transMAG^PEI alone enhanced vector efficacy two-fold. Additional application of a magnetic field resulted in a seven-fold increase in transduced cells. If polybrene was omitted from the transducing preparation, virtually no transduction was observed with virus alone. In contrast, omission of polybrene improved paramagnetic particle-guided transduction in the absence of a magnetic field four-fold over the standard transduction (virus + polybrene) and culminated in a 20-fold enhancement in the presence of a magnetic field Figure 4. These results demonstrate that magnetofection is applicable to retroviral gene delivery as well, suggesting that cationic nanoparticles are superior mediators of retroviral transduction compared with polybrene. Thus, magnetofection will significantly improve retroviral gene transfer technology.

Magnetofection in vivo

For magnetofection in vivo, we confronted one nonviral and an adenoviral formulation with the harshest conditions we could think of, by applying them in the ilea lumens of rats or in the stomach lumens of mice, respectively. The extreme conditions of pH, the abundance of degradative enzymes, the presence of degraded nutrition and bacteria render gene delivery and gene therapy in the guts and stomach a particular challenge. Nevertheless, the high frequency of malignancies in these organs make them important targets for gene therapy. For proof-of-principle, the experiments were carried out with LacZ as reporter gene and gene transfer efficiency with and without magnetic targeting was evaluated in histological sections (Figure 5). We found strong and consistent X-gal staining in tissues transfected under the influence of a magnetic field, only rare transfection events in the absence of a magnetic field and no background staining in untreated control tissue. Efficient transfection with nonviral magnetofectin was confined to the ileum lamina propria, while the adenoviral magnetofectin applied to the stomach produced strong staining limited to the crypts of fundic glands. For additional proof-of-principle, we carried out magnetofection in the ear veins of pigs as a model for gene delivery to endothelial cells. We infused transMAG^PEI + DNA + PEI (DNA dose 500 g) into the right and left ear veins and a Nd-Fe-B permanent magnet block was attached above the right veins proximal to the injection sites. No reporter gene expression (luciferase) was observed in the control blood vessels (left ears) and distal from the magnet positions (right ears), while reproducible, though variable luciferase expression (40.4 ± 37.7 pg luciferase/g tissue) was found in all vein samples which were under direct influence of the magnetic field. No luciferase signal (light emission) was found in samples of any other major organ. Hence, magnetofection can provide a useful complementation to current techniques of vascular gene delivery.¹⁸

Discussion

Magnetofection is an appropriate tool to overcome the strong barriers of slow vector accumulation and consequently low vector concentration at target tissues, a fundamental limitation to gene delivery outlined previously by Luo and Saltzman.¹ We made gene vectors susceptible to magnetic force by a simple physical association with superparamagnetic nanoparticles which were designed and provided with a polycation coating for this purpose. Importantly, this type of design was universally applicable to any type of gene vector we tested. The concept of magnetofection greatly profits from the fact that the individual modules of the system can be optimized independently and variants can be assembled in a combinatorial manner, thus facilitating optimization towards specific applications. The size and surface chemistry of magnetic particles can be tailored to meet specific demands on physical and biological characteristics, also the linkage between vector and magnetic particle can be designed accordingly. This conclusion is in accordance with parallel work by Hughes et al,¹⁹ who most recently attached retroviral vectors via biologic linkage to commercially available magnetic microbeads. These authors demonstrated a similarly encouraging improvement in vector targeting and efficacy by magnetic force as we found for any vector type in a less sophisticated approach. Concerning the mechanism of magnetofection, our results suggest that the actual uptake of vectors into the cells remains endocytotic, although an influence of the magnetic field beyond mere vector concentration at the target site (the cell surface) is conceivable. While such a function, resulting in enhanced tissue penetration, has been demonstrated in magnetic drug targeting,²⁰ additional experimentation is required to decide whether a similar effect applies to magnetofection.

There are several perspectives to the future use of magnetofection. For in vitro application, the three important features of magnetofection are (1) the drastically lowered vector dose; (2) the considerably reduced incubation time required to achieve high transfection/transduction efficiency; and (3) the possibility of gene delivery to otherwise nonpermissive cells (such as demonstrated here with adenoviral magnetofection; Figure 3). These characteristics can make magnetofection particularly useful in the transfection/transduction of difficult-to-transfect/transduce cells, for example in ex vivo gene therapy approaches. In addition, magnetofection will be an ideal research tool where the available vector dose, the required process time and the sustainable costs of the procedure are limiting factors. Also for in vivo gene- and nucleic acid-based therapies, magnetofection may become a strong choice where local treatment is required. Even with the simple set-up presented here, gene delivery to surgically accessible sites such as gut, stomach and vasculature can be greatly improved and further local applications can be envisaged. Obvious target diseases are cancer, cardiovascular or neuromuscular diseases, and potentially also genetic diseases. Specific and efficient magnetic targeting of interior body regions upon systemic vector administration will require the generation of focused strong magnetic fields, a challenge for physical and medical sciences. Combined with existing advanced concepts of particulate drug delivery and vector targeting, magnetofection may provide additional specificity and efficacy which are required in many gene therapy approaches.

Materials and methods

Reagents and polyethylenimine) (PEI; average molecular weight 25 kDa) were purchased from Sigma-Aldrich (Deisenhofen, Germany), unless otherwise stated. 1,2-dioleoyl-3-trimethylammoniumpropane (DOTAP) was purchased from Avanti Polar Lipids (Alabaster, AL, USA).

Superparamagnetic iron oxide nanoparticles with an average size of 200 nm (by dynamic light scattering), coated with PEI (800 kDa; transMAG^PEI) were custom made by Chemicell (Berlin, Germany) similarly as described.^21,22

Sintered Nd-Fe-B magnets (NeoDelta; remanence Br, 1080-1150 mT) were purchased from IBS Magnet (Berlin, Germany). Dimensions for magnetofection in six- and 24-well plate formats: 20 ´ 10 ´ 5 mm; for 96-well plate format: cylindrical; d = 6 mm, h = 5 mm, inserted in an acrylic glass template in 96-well plate format with strictly alternating polarization. The fields of the individual magnets influence each other such that the vector dose becomes concentrated in the centers of individual wells.

Biotinylated PEI was obtained upon reaction with 2 mol equivalents of NHS-LC-Biotin (Pierce, Rockford, IL, USA) followed by gel filtration (Sephadex G-25; Pharmacia, Freiburg, Germany).

Streptavidinylated transMAG^PEI was obtained by derivatization of streptavidin (Molecular Probes, Leiden, The Netherlands) with succinimidyl-pyridyl-dithiopropionate (SPDP; 3.5 mol equivalents) and reaction with thiolated transMAG^PEI which was obtained by reaction with SPDP (8 nmol/mg), followed by addition of -mercaptoethanol and washing with water.

Vector association studies with magnetic nanoparticles

Plasmid DNA was labeled with ³²P using the Nick Translation Kit from Amersham-Pharmacia with the protocol of the supplier modified such that the incubation time was 15 min at 15°C instead of 2 h. The labeled plasmid was purified using MicroSpin columns (Pharmacia) and the Promega Wizard PCR Preps DNA Purification System (Promega, Mannheim, Germany) for removal of unincorporated nucleotides and enzymes from the reaction mixture. The product DNA had the same size as the starting plasmid as confirmed by gel electrophoresis.

In two separate set-ups, 120 l each of DNA stock solution (124.8 g cold plasmid plus 1.56 ´ 10⁷ c.p.m. ³²P-labeled plasmid in 3120 l of water) were added to 120 l each of a dilution series of transMAG^PEI in water. The complexes were prepared in Eppendorf tubes and mixed by pipetting. After 15 min of incubation, the mixtures were added to either 120 l each of PEI stock solutions (41.7 g/ml in water) or of DOTAP-Cholesterol liposome stock suspensions (121.2 l 5 mM liposome stock per ml in water). This resulted in PEI:DNA N/P ratios of 8 or DOTAP:DNA charge ratios of 5. After further 15-min incubation, the complexes were added to 120 l each of 600 mM sodium chloride, initializing salt-induced aggregation. After 20 min of incubation, 120 l each of the resulting complexes were transferred to the wells of a U-bottom 96-well plate in triplicates. The plate was positioned upon the 96-well format magnetic plate. After 30 min of magnetic sedimentation, 80 l supernatants were removed and mixed with 125 l each of Microscint 40 (Canberra Packard, Dreieich, Germany) in an opaque 96-well plate. In the same manner, 80 l each of unsedimented samples were added to the plate as reference. The samples were counted using a Topcount instrument (Canberra Packard, count delay set to 10 min, count time in triplicates, 5 min each). In an analogous set-up, aliquots of a plasmid DNA stock solution were mixed with equal volumes of a transMAG^PEI dilution series followed by mixing with a volume aliquot each of 450 mM sodium chloride. Magnetic sedimentation and radioactivity determination in the supernatants was carried out as described above.

Binding was calculated as: %bound = 100 ´ c.p.m._sample/c.p.m._reference. Recombinant adenovirus (10 l stock, corresponding to 7.2 ´ 10¹⁰ viral particles or 6.6 ´ 10⁸ p.f.u. (plaque forming units) was diluted to 100 l with HBS buffer (8 g/l sodium chloride, 0.37 g/l potassium chloride, 0.27 g/l di-sodium hydrogen phosphate dihydrate, 1 g/l dextrose), mixed with 7.8 MBq ¹²⁵I (2 l; Amersham-Pharmacia) and incubated for 10 min at room temperature in an iodogen cap (Pierce). After addition of 200 l HBS buffer, the virus was separated from unbound label by gel filtration using a Pharmacia PD-10 column. Fraction 1, containing 2.61 ´ 10¹⁰ viral particles (determined by UV absorbance) and an activity of 485 kBq per ml was used for binding studies.

Thirty-six microliters of labeled adenovirus were mixed with 18 l aliquots of a transMAG^PEI dilution series in HBS resulting in ratios of viral particles (VP) per pg of transMAG^PEI of 8700, 4350, 2900, 2175, 1740, 1450, 1088, 870, 725, 621 and 544. After 20-min incubation at room temperature, the samples were filled up to 432 l with HBS. Aliquots of 120 l each were transferred to a U-bottom 96-well plate in triplicates which subsequently was positioned on the 96-well format magnetic plate. After 1-h magnetic sedimentation, 80 l of the supernatants each and of unsedimented samples were transferred to individual scintillation tubes (Polyvials V; Zinsser Analytic, Frankfurt, Germany) and counted using a gamma counter (Wallac, Turku, Finland). The binding isotherm was calculated as above.

Cell culture and magnetofection

Cells were grown and transfected in ATCC-recommended culture medium containing 10% FCS. PBL were isolated using a Ficoll-Hypaque gradient²³ (Biocoll, Biochrom, Germany), cultured and stimulated with phytohemagglutinin-M (10 g/ml, Roche, Germany) for 3 days, and with recombinant human IL-2 (10 units/ml, Roche). During magnetofection experiments, K562 and PBL were kept in polylysine-coated culture dishes. Unless otherwise stated, vector formulations were added to cells kept in 150 l or 1.5 ml fresh medium, respectively (96-well plate and six-well plate formats). During the incubation times specified below, pairs of culture dishes with identical vector formulations were kept either under standard conditions or placed upon magnets. Subsequently, the cells were washed once with fresh medium, grown for 24 or 48 h and subjected to reporter gene assays as described elsewhere.^24,25

Nonviral magnetofection

Plasmids p55pCMV-IVS-luc+, coding for the firefly luciferase, and pCMV--gal were kindly provided by Andrew Baker (Bayer, USA) and by Walter Schmidt (Intercell, Vienna, Austria), respectively. Plasmids were purified by cesium chloride gradient. A psoralene-inactivated, biotinylated adenovirus, used as endosomolytic agent in the Adenovirus-Enhanced-Transfection system (AVET⁹), was kindly provided by Ernst Wagner (Vienna University Biocenter, Austria).

PEI

Equal volumes of stock solutions in water containing DNA (40 g/ml), transMAG^PEI (0 and 80 g/ml, respectively), PEI (25 kDa from Aldrich; 41.7 g/ml) and sodium chloride (600 mM) were mixed sequentially at 15-min intervals and a final 30-min incubation step (salt-induced aggregation). Cells (2 ´ 10⁴ NIH3T3 or CHO-K1/96-well, seeded the day before transfection and freshly supplemented with 150 l fresh serum-containing medium) were incubated for 10 min and 4 h, respectively, upon addition of 50 l aliquots per well in triplicates.

AVET

Equal volumes of stock solutions in HBS (20 mM HEPES pH 7.4, 150 mM sodium chloride) containing DNA (40 g/ml), PEI (41.7 g/ml), transMAG^PEI (0 and 40 g/ml, respectively), and inactivated adenovirus (5.74 ´ 10¹⁰ virus particles per ml) were mixed sequentially with 15-min incubation steps. The resulting solution was diluted 1:1 with HBS. Cells (as above) were incubated for 10 min and 4 h, respectively, upon addition of 50 l aliquots per well in triplicates.

GenePorter and Lipofectamine

GenePorter (Gene Therapy Systems, La Jolla, CA, USA) and Lipofectamine (Life Technologies, Karlsruhe, Germany): one volume equivalent of a DNA stock solution (20 g/ml) was mixed with equal volumes of transMAG^PEI stock solutions (0 and 40 g/ml, respectively). This was followed by immediate mixing with 2 vol equivalents of GenePorter (50 l/ml) or Lipofectamine (40 l/ml) stock solutions, respectively. All stock solutions were in serum-free DMEM. After 20-min incubation, the resulting complexes were diluted 2.5-fold with serum-free DMEM. Cells (2 ´ 10⁴ NIH3T3 or CHO-K1/96-well, seeded the day before transfection and freshly supplemented with 50 l serum-free DMEM) were incubated for 10 min and 4 h, respectively, with 50 l aliquots in triplicates, followed by washing and cultivation overnight with serum-containing medium.

DOTAP-Cholesterol

DOTAP/Cholesterol (1:1 mol/mol; 5 mM) liposomes were prepared in water essentially as described.²⁶ Equal volumes of stock solutions in HBS containing DNA (30 g/ml), transMAG^PEI (30 g/ml) and DOTAP/Chol liposomes (455 M) were mixed sequentially with 15-min incubation after the first step and 30 min after the second. The resulting mixture was subjected to serial 1:1 dilution in HBS and cells (3 ´ 10⁴ NIH3T3/96-well in 150 l serum-containing DMEM) were incubated upon addition of 50 l aliquots in triplicates for 10 min.

AVET (six-well plate)

Equal volumes of stock solutions in HBS containing plasmid (48 g/ml), PEI^biotin (50 g/ml), inactivated adenovirus (6.9 ´ 10¹⁰ virus particles per ml) and transMAG^PEI-Stav (57.6 g/ml) were mixed sequentially with 15-min incubation steps. Aliquots of 500 l were added to the cells (3.5 ´ 10⁵ NIH3T3 per well supplied with 1.5 ml serum-containing DMEM) and incubated for 15 min followed by medium change. A magnet remained attached under one well for 30 min. After 24 h, the cells were washed with PBS and subjected to X-gal staining for 45 min.

Adenoviral magnetofection

AdLacZ, a replication-defective Ad5 expressing LacZ under control of the HCMV promoter replacing the E1 region in the right to left orientation, was constructed and purified by double cesium chloride gradient according to standard methods²⁷ and titred on 293 cells. MOI is given as p.f.u., unless otherwise stated. Complexes of AdLacZ and transMAG^PEI were produced by incubating the specified virus quantities with the indicated amounts of transMAG^PEI in total volumes of 250 l (six-well plate format) or 100 l HBS (24- and 96-well plate formats) for 20 min at room temperature.

Retroviral magnetofection

NIH 3T3 cells (1 ´ 10⁵ cells plated on 35-mm dishes 24 h before infection) were incubated for 3 h with 1 ml aliquots of 24-h supernatants from low titer MuLV producing ecotropic packaging cells (subclone A6.LT of GP86-NA.6; approximately 1-5 ´ 10³ X-gal CFU/ml²⁸). These supernatants were applied untreated or treated with transMAG^PEI (3 g/ml for 20 min) and/or polybrene (8 g/ml immediately before infection). Magnets were applied to specified groups for 1 h. After 48 h, the cells were stained with X-gal, and blue nuclei were counted.

Magnetofection in vivo

After laparatomy of anesthetized Wistar rats in the linea alba region, ileum and caecum were exposed and the guts was clamped off 8 cm in the oral direction of the ileo-caecal junction. Ingested material was carefully rinsed towards the caecum by application of 1 ml of isotonic saline. Then, a second clamp was placed 3 cm aborally from the first clamp. The vector preparation (200 g DNA/400 g transMAG^PEI in 1 ml 5% glucose) was injected with a 20-G needle adjacent to the first clamp. The injection site was closed with surgical suture while a sterile magnet block (20 ´ 10 ´ 5 mm) was placed under the clamped-off section. Five min after injection, both clamps were removed. The magnet was left for a total of 20 min. Subsequently, the guts were returned carefully into the abdominal cavity which was closed with surgical suture. The animals were killed after 48 h. The treated section of the guts and adjacent areas were isolated, rinsed exhaustively with PBS and fixed for 30 min with 2% formaldehyde and 0.2% glutaraldehyde in PBS. The tissue was rinsed again with PBS followed by 4 h X-gal staining at 37°C. Subsequently, the tissue was again rinsed exhaustively with PBS and stored overnight at 4°C in 2% formaldehyde/PBS followed by embedding for paraffin and cryosections. Sections were stained with eosin.

Laparotomy of anesthetized Bl6/Cico mice was performed at the left costal margin to expose the stomach and duodenum. Magnetofectin (5 ´ 10⁸ p.f.u. of AdLacZ mixed with 25 g of transMAG^PEI in a total volume of 200 l HBS) was injected into the greater curvature of the stomach via the duodenum using a 27-G needle. Magnets (6 mm diameter) were positioned for 20 min under the exposed stomach covering the area of the gastric fundus. Control animals were treated with the same vector without exposure to the magnetic field. Four days after injection, animals were killed, organs excised, rinsed extensively with PBS, fixed for 1 h with 2% formaldehyde/0.2% glutaraldehyde, washed extensively with PBS, then stained for 4 h at 37°C with X-gal followed by overnight incubation at 4°C. After rinsing with PBS, organs were kept in 2% formaldehyde for 24 h, and subsequently paraffin embedded. Thin sections were stained with nuclear fast red.

Acknowledgements

We thank P Swaan, E Wagner and J-S Rémy for helpful discussions and Ursula Putz, Sieglinde Wegerer and Katja Honert for technical assistance. This work was supported in part by the Deutsche Forschungsgemeinschaft and the BMBF.

1 Luo D., Saltzman W.M. Enhancement of transfection by physical concentration of DNA at the cell surface. Nat Biotechnol 2000; 18: 893, 893. MEDLINE

2 Kasahara N., Dozy A.M., Kan Y.W. Tissue-specific targeting of retroviral vectors through ligand-receptor interactions. Science 1994; 266: 1373, 1373. MEDLINE

3 Curiel D.T. Strategies to adapt adenoviral vectors for targeted delivery. Ann NY Acad Sci 1999; 886: 158, 158. MEDLINE

4 Wu G.Y., Wu C.H. Receptor-mediated in vitro gene transformation by a soluble DNA carrier system. J Biol Chem 1987; 262: 4429, 4429. MEDLINE

5 Lübbe A.S., Bergemann C. Magnetically controlled drug targeting. Cancer J 1998; 11: 104, 104.

6 Lübbe A.S.et al. . Clinical experiences with magnetic drug targeting: a phase I study with 4'-epidoxorubicin in 14 patients with advanced solid tumors. Cancer Res 1996; 56: 4686, 4686. MEDLINE

7 Hiemenz P.C. Principles of Colloid and Surface Chemistry 2nd edn Marcel Dekker, 1986.

8 Boussif O.et al. . A versatile vector for gene and oligonucleotide transfer into cells in culture and in vivo - polyethylenimine. Proc Natl Acad Sci USA 1995; 92: 7297, 7297. MEDLINE

9 Cotten M.et al. . Psoralen treatment of adenovirus particles eliminates virus replication and transcription while maintaining the endosomolytic activity of the virus capsid. Virology 1994; 205: 254, 254. Article MEDLINE

10 Leventis R., Silvius J.R. Interactions of mammalian cells with lipid dispersions containing novel metabolizable cationic amphiphiles. Biochim Biophys Acta 1990; 1023: 124, 124. MEDLINE

11 Plank C.et al. . The influence of endosome-disruptive peptides on gene transfer using synthetic virus-like gene transfer systems. J Biol Chem 1994; 269: 12918, 12918. MEDLINE

12 Pickles R.J.et al. . Retargeting the coxsackievirus and adenovirus receptor to the apical surface of polarized epithelial cells reveals the glycocalyx as a barrier to adenovirus-mediated gene transfer. J Virol 2000; 74: 6050, 6050. MEDLINE

13 Li D., Duan L., Freimuth P., O'Malley B.W. Jr. Variability of adenovirus receptor density influences gene transfer efficiency and therapeutic response in head and neck cancer. Clin Cancer Res 1999; 5: 4175, 4175. MEDLINE

14 Tomko R.P., Xu R., Philipson L. HCAR and MCAR: the human and mouse cellular receptors for subgroup C adenoviruses and group B coxsackieviruses. Proc Natl Acad Sci USA 1997; 94: 3352, 3352. Article MEDLINE

15 Shayakhmetov D.M., Papayannopoulou T., Stamatoyannopoulos G., Lieber A. Efficient gene transfer into human CD34(+) cells by a retargeted adenovirus vector. J Virol 2000; 74: 2567, 2567. MEDLINE

16 Leon R.P.et al. . Adenoviral-mediated gene transfer in lymphocytes. Proc Natl Acad Sci USA 1998; 95: 13159, 13159. Article MEDLINE

17 Fasbender A.et al. . Incorporation of adenovirus in calcium phosphate precipitates enhances gene transfer to airway epithelia in vitro and in vivo. J Clin Invest 1998; 102: 184, 184. MEDLINE

18 Isner J.M.et al. . Arterial gene transfer for therapeutic angiogenesis in patients with peripheral artery disease. Hum Gene Ther 1996; 20: 959, 959.

19 Hughes C., Galea-Lauri J., Farzaneh F., Darling D. Streptavidin paramagnetic particles provide a choice of three affinity-based capture and magnetic concentration strategies for retroviral vectors. Mol Ther 2001; 3: 623, 623. Article MEDLINE

20 Widder K.J.et al. . Selective targeting of magnetic albumin microspheres to the Yoshida sarcoma: ultrastructural evaluation of microsphere disposition. Eur J Cancer Clin Oncol 1983; 19: 141, 141. MEDLINE

21 Mendenhall G.D., Geng Y., Hwang J. Optimization of long-term stability of magnetic fluids from magnetite and synthetic polyelectrolytes. J Colloid Interface Sci 1996; 184: 519, 519. Article MEDLINE

22 Povey A.C., Bartsch H., Nixon J.R., O'Neill I.K. Trapping of chemical carcinogens with magnetic polyethyleneimine microcapsules: I. Microcapsule preparation and in vitro reactivity of encapsulated nucleophiles. J Pharm Sci 1986; 75: 831, 831. MEDLINE

23 Kanof M.E., Smith P.D., Zola H. Preparation of human mononuclear cell populations and subpopulations. Coligan JE (eds); Current Protocols in Immunology John Wiley, 1999, 7 pp.

24 Finsinger D.et al. . Protective copolymers for nonviral gene vectors: synthesis, vector characterization and application in gene delivery. Gene Therapy 2000; 7: 1183, 1183. MEDLINE

25 Plank C., Tang M., Wolfe A., Szoka F.C. Branched cationic peptides for gene delivery. Role of type and number of cationic residues in formation and in vitro activity of DNA polyplexes. Hum Gene Ther 1999; 10: 319, 319. Article MEDLINE

26 Meyer K.B.et al. . Intratracheal gene delivery to the mouse airway: characterization of plasmid DNA expression and pharmacokinetics. Gene Therapy 1995; 2: 450, 450. MEDLINE

27 Hitt M. Techniques for human adenovirus vector construction and characterization. Adolph KW (eds); Viral Gene Techniques Academic Press, 1995, pp 13-30.

28 Krüger A., Schirrmacher V., von Hoegen P. Scattered micrometastases visualized at the single-cell level: detection and re-isolation of lacZ-labeled metastasized lymphoma cells. Int J Cancer 1994; 58: 275, 275. MEDLINE

Figure 1 Vector association with magnetic particles. Top, nonviral vectors: Complexes were prepared in water followed by salt-induced aggregation upon adjusting to 150 mM sodium chloride. After magnetic sedimentation, radio-labeled DNA was quantified in the supernatants of triplicates. The figure demonstrates that saturation binding is achieved at a transMAG^PEI:DNA ratio of 2 (w/w) for naked DNA and of 4 for PEI-DNA (N/P = 8) and DOTAP-Cholesterol-DNA (+/- = 5). Under the conditions used for gene transfer experiments, >70% of the DNA dose were associated with magnetic particles. The data points are averages of triplicates ± standard deviation (error bars too small to be seen). Bottom: Iodine-125-labeled adenovirus and magnetic particles were mixed under conditions and covering the ratios applied in gene transfer experiments. The mixtures were subjected to magnetic sedimentation. Unbound virus was determined in supernatants of triplicates using a gamma counter. The figure shows that under the conditions used in the gene transfer experiments, 70+% percent of virus are associated with magnetic particles.

Figure 2 Nonviral magnetofection. (a) Efficacy of magnetofectins (magnetic particle containing vectors) in NIH3T3 and CHO cells upon short-time (10 min) incubation in the presence (black bars) and in the absence (light gray bars) of a magnetic field compared with standard transfections with the parent vectors (dark gray bars, 4-h incubation; white bars, 10-min incubation). Transfections were carried out in 96-well plates as specified in the experimental protocol. Cells were washed and supplemented with fresh medium containing 10% FCS after the respective incubation times. The table below the graph specifies the enhancements that were achieved upon the influence of the magnetic field on paramagnetic vectors compared with transfections in the absence of the magnetic field with paramagnetic vectors or the parent standard vectors which did not contain transMAG^PEI. The data demonstrate that magnetofection can strongly enhance transfection efficiencies over standard transfection protocols. The relative enhancements are dependent on vector type, cell line and incubation time under consideration. (b) Dose-response relationship for magnetofection of NIH3T3 cells with DOTAP/Cholesterol-DNA in the presence () and in the absence () of a magnetic field. Numbers in the graph specify the fold enhancement over transfection in the absence of a magnetic field. Similar dose-response profiles were obtained with Lipofectamine and GenePorter in CHO-K1 cells (not shown). (a, b) Luciferase activity after 24 h was normalized to cellular protein content. Shown are averages ± standard deviation of triplicates or quadruples. (c) Localization of gene delivery by a magnetic field. NIH3T3 cells, incubation for 15 min with an adenovirus-enhanced PEIbiotin-DNA:transMAG^PEISta formulation. X-gal staining after 24 h.

Figure 3 Adenoviral magnetofection (incubation for 30 min). (a) Dependence on transMAG^PEI dose at constant MOI. NIH3T3 cells (3 ´ 10⁵/six-well) incubated for 20 min with formulations of 200 p.f.u./cell of AdLacZ containing 3, 6 and 12 g transMAG^PEI. Upper panel (3+ etc) with magnet attached under the wells, lower panel without magnets (3- etc). Incubation with 6 g transMAG^PEI in PBS (PBS/6+) and a standard infection (200/0 -) served as controls. X-gal staining was performed for 1 h only, demonstrating that high reporter gene expression levels were achieved. (b-d) Dependence on MOI at constant transMAG^PEI (3 or 6 g). (b) NIH3T3 cells (3 ´ 10⁵/six-well), (c) K562 cells (3 ´ 10⁵/24-well) and (d) PBLs (3 ´ 10⁵/96-well; in polylysine-coated wells). Quantitative assay with CPRG performed 24 h (b) and (c) or 72 h (d) after magnetofection. Controls were buffer alone (PBS), transMAG^PEI in PBS and standard infection (200/0). The apparent background activity of control samples in (d) is an artifact from the calculation of the standard curve and is in fact not above the background noise of the blank in the standard curve. (e) PBLs. +/- indicate presence or absence of magnet. X-gal staining performed overnight.

Figure 4 Retroviral magnetofection. NIH3T3 cells were incubated with MuLV-containing supernatants treated as described. X-gal staining was carried out after 48 h and blue cells were counted. Results are expressed as transduction efficiency relative to the efficiency of a standard transduction (virus + polybrene).

Figure 5 Magnetofection in vivo. (a, b) DNA-transMAG^PEI was applied to the ilea of rats in the absence (a) and under the influence of a magnetic field for 20 min (b). X-gal staining performed 48 h after gene delivery reveals efficient gene delivery only in the presence of magnet (b), both on the macroscopic level (upper panel) and on the microscopic level (lower panel). Upper panel: intestinal tubes after X-gal stain. Inserts: cross-sections of tubes embedded in paraffin. Lower panel: Paraffin sections counter-stained with eosin, 400´ magnification. X-gal staining is found in the lamina propria. (c, d) AdLacZ-transMag^PEI complexes were applied to the lumen of the stomach. Gene delivery in the absence of a magnetic field (c) yields only a few transfected cells, while exposure to a magnet for 20 min produces strong and widespread X-gal staining in the crypts of the fundic glands 4 days after gene delivery (d). L, lumen; LP, lamina propria; F, fundic glands; S, submucosa; M, muscularis. 40´ magnification (upper panel), 400´ magnification (lower panel).