Microbubble-enhanced ultrasound to deliver an antisense oligodeoxynucleotide targeting the human androgen receptor into prostate tumours

Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

Microbubble-enhanced ultrasound to deliver an antisense oligodeoxynucleotide targeting the human androgen receptor into prostate tumours夽 Petra Haag a , Ferdinand Frauscher b , Johann Gradl b , Alexander Seitz b , Georg Sch¨afer c , Jonathan R. Lindner d , Alexander L. Klibanov e , Georg Bartsch a , Helmut Klocker a , Iris E. Eder a,∗ a

Department of Urology, Innsbruck Medical University, A-6020 Innsbruck, Anichstraße 35, Austria b Department of Radiology II, Innsbruck Medical University, Austria c Department of Pathology, Innsbruck Medical University, Austria d Division of Cardiology, Oregon Health and Science University, Portland, OR, United States e Cardiovascular Division, University of Virginia, Charlottesville, VA, United States

Abstract We have shown recently that downregulation of the androgen receptor (AR), one of the key players in prostate tumor cells, with short antisense oligodeoxynucleotides (ODNs) results in inhibition of prostate tumor growth. Particularly with regard to an application of these antisense drugs in vivo, we now investigated the usefulness of microbubble-enhanced ultrasound to deliver these ODNs into prostate cancer cells. Our short antisense AR ODNs were loaded onto the lipid surface of cationic gas-filled microbubbles by ion charge binding, and delivered into the cells by bursting the loaded microbubbles with ultrasound. In vitro experiments were initially performed to show that this kind of delivery system works in principle. In fact, transfection of prostate tumor cells with antisense AR ODNs using microbubble-enhanced ultrasound resulted in 49% transfected cells, associated with a decrease in AR expression compared to untreated controls. In vivo, uptake of a digoxigenin-labelled ODN was found in prostate tumour xenografts in nude mice following intratumoral or intravenous injection of loaded microbubbles and subsequent exposure of the tumour to ultrasound, respectively. Our results show that ultrasound seems to be the driving force of this delivery system. Uptake of the ODN was also observed in tumors after treatment with ultrasound alone, with only minor differences compared to the combined use of microbubbles and ultrasound. © 2006 Elsevier Ltd. All rights reserved. Keywords: Antisense oligonucleotides; Androgen receptor; Microbubbles; Ultrasound; Prostate cancer

1. Introduction We have shown recently that the inhibition of AR expression with an antisense androgen receptor (AR) oligodeoxynucleotide (ODNasAR) results in prostate tumour growth 夽 Lecture presentation at the 17th International Symposium of the Journal of Steroid Biochemistry & Molecular Biology, ‘Recent Advances in Steroid Biochemistry and Molecular Biology’ (Seefeld, Tyrol, Austria, 31 May–3 June 2006). ∗ Corresponding author. Tel.: +43 512 504 24819; fax: +43 512 504 24817. E-mail address: [email protected] (I.E. Eder).

0960-0760/$ – see front matter © 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.jsbmb.2006.09.027

retardation in vitro [1] and in vivo [2]. Thus, eliminating the expression of AR in prostate tumour cells with antisense ODNs may be an innovative strategy to treat prostate cancer. One of the major limitations of using antisense ODNs in the clinic, however, is their safe and efficient delivery to the prostate. Therefore, a number of different delivery techniques have been proposed in the past to improve gene delivery into the prostate [3]. We investigated the usefulness of microbubble contrast agents in combination with ultrasound to transfer an antisense ODN into prostate tumours. The principle behind this technique is the use of contrast agent microbubbles as transport vehicles, which carry the ODN to the tumour tissue

104

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

upon intravenous injection. There, the loaded microbubbles are burst by ultrasound resulting in delivery of the ODN into the adjacent tissue by a mechanism called sonoporation [4,5]. One main advantage of this technique is that microbubbleenhanced ultrasound is already routinely used in patients for improved diagnostic imaging and is therefore known as safe and non-toxic. Moreover, the use of focused ultrasound treatment gives the possibility to release the drug specifically at the tumour site. The development of novel therapies for prostate cancer is an important issue due to the limited selection of efficient treatment modalities. Whereas organ-confined prostate cancer can be treated by surgery, androgen ablation is the main choice of treatment for patients with advanced prostate cancer [6,7]. However, although the majority of patients respond to this treatment, most tumours relapse and progress during therapy despite castrate levels of testosterone. One of the key molecules contributing to prostate cancer growth is the AR which is expressed in all stages of prostate cancer, including those which are resistant to androgen ablation treatment [8–10]. Since, increased AR expression and promiscuous inappropriate transcriptional AR activation is associated with hormone-refractory prostate cancer [11–18], the AR represents an interesting target for a novel therapy. The present study was therefore undertaken to investigate if microbubble-enhanced ultrasound can be used to deliver the ODNasAR into prostate tumours. We demonstrate here that the ODNasAR can be delivered into prostate tumor xenografts in nude mice through microbubble-enhanced ultrasound resulting in a significant uptake of the ODN in the tumour and tumour-associated stromal compartments as compared with untreated controls. Ultrasound is considered as the driving force of this system since there is also significant uptake of the ODN in tumors after delivery with ultrasound alone. 2. Materials and methods 2.1. Antisense oligonucleotides A 15-nucleotide short antisense phosphorothioate ODN was directed against the CAG polyglutamine region of the human AR gene which was tested for its activity previously [1] (5 -CTGCTGCTGCTGCTG-3 , ODNasAR, GenXpress, Vienna, Austria). To investigate tissue distribution in vivo, we used a digoxigenin-labelled ODN (5 -CTGCTGCTGCTGCTG-3 -digoxigenin, Eurogentec S.A., Belgium). A fluorescein-labelled ODN (5 -CTGCTGCTGCTGCTG-3 -fluorescein, GenXpress) was used to demonstrate charge-coupling to cationic microbubbles by flow cytometry. 2.2. Loading of the ODN to cationic microbubbles Positively charged microbubbles were produced as described previously [19]. Briefly, an aqueous solution of

1 mg/mL polyethyleneglycol-40 stearate (Sigma Chemical Co., St. Louis, MO), 2 mg/mL distearoyl phosphatidylcholine (Avanti Polar Lipids, Alabaster, AL) and 0.4 mg/mL 1,2distearoyl-3-trimethylammoniumpropane (Avanti) was sonicated with decafluorobutane gas, creating microbubbles with a cationic lipid shell (schematic model in Fig. 1). Microbubble charge (zeta potential) was determined by laser Doppler velocimetry (Zetaplus, Brookhaven Instruments) of approximately 1 × 107 microbubbles dispersed in 1 mM KCl at pH 7.4. Zeta potential was +58 for cationic microbubbles. Mean size (3.18 ± 0.1 ␮m) and concentration of the microbubbles was determined with a Cell Counter and Analyzer System (CASY, Model TTC, Sch¨arfe System, Reutlingen, Germany). To demonstrate charge coupling, a fluorescein-labelled ODN (1 ␮M) was added to 2 × 107 microbubbles in 500 ␮L PBS and incubated for 10 min on ice. Microbubbles were centrifuged once at 1500 rpm for 1 min to get rid of unbound material and suspended in 500 ␮L fresh PBS. Fluorescence intensity of microbubbles was analyzed on a FACS Calibur (Becton Dickinson, San Jose, CA). To quantify the amount of ODN loaded to the lipid surface, the microbubble solution as well as the aqueous solution containing the unbound material were boiled (5 min and 90 ◦ C) to solubilize the microbubbles and thus to prevent background scattering. Optical density was measured at 260 nm using a spectrophotometer (U-2000, Hitachi). 2.3. In vitro transfection and analysis of transfection efficiency Human LNCaP prostate cancer cells were obtained from the American Type Culture Collection. These cells express high levels of AR and are a valuable in vitro model to study prostate cancer. They were maintained in RPMI supplemented with 10% fetal calf serum and penicillin/streptomycin (100 U/mL and 100 ␮g/mL) in a humidified atmosphere of 5% CO2 in air at 37 ◦ C. Passaging as well as cell harvesting was performed with trypsin/EDTA (0.05% and 0.02%). Exponentially growing cells were used for transfection experiments. 1 × 106 LNCaP cells were seeded in 10 mL RPMI with 10% fetal calf serum on one side of an opticell culture chamber (OptiCellTM , Westerville, OH) and incubated overnight. These cell culture chambers were used in vitro since they were found to be nearly acoustically transparent. All transfection experiments were carried out on subconfluent cell monolayers. About 2 × 107 microbubbles were charge-coupled with oligonucleotides (1 ␮M) in 100 ␮L PBS and incubated for 10 min on ice. Cell culture supernatant was removed and loaded microbubbles, suspended in 10 mL RPMI with 5% fetal calf serum and 20 mM HEPES, were added to the cells. Thus, final ODN concentration was 10 nM. Immediately after adding the bubbles, the sealed opticell cell culture chambers were immersed vertically in a water bath containing deionized water (room temperature). The submersed ultrasound probe (4C1, Acuson Sequoia 512, Siemens Medical

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

105

Fig. 1. Binding of an antisense ODN to cationic microbubbles. About 2 × 107 microbubbles were diluted in 500 ␮L PBS and analyzed by flow cytometry either alone (Mb) or after charge-coupling of 1 ␮M of a fluorescein-labelled ODN to the lipid surface (Mb + ODN). Unloaded (Mb) and loaded microbubbles (Mb + ODN) are depicted schematically above each measurement. Cartoons above each flow cytometric analysis describe the principle of binding.

Solutions, Mountain View, CA) was mounted in parallel 5 cm afar from the cell chamber and slowly moved up and down to guarantee consistent insonification of all cells (74.8 mm × 65 mm × 2 mm). In vitro, we used a colour Doppler frequency of 1.75 MHz and a mechanical index (MI) of 1.9. The MI reflects the ultrasound energy calculated by peak negative acoustic pressure divided by the square root of the frequency. One focal zone was placed at the area of the cell membrane and ultrasound exposure was performed until all microbubbles were destroyed (9 min). Destruction of the microbubbles was verified by microbubble signal decay. To determine transfection efficiency in vitro, LNCaP cells were treated with 1 ␮M of a fluorescein-labelled ODN charge-coupled to 2 × 107 microbubbles and subsequent bursting with ultrasound. Twenty-four hours after treatment, cells were washed with cold PBS and the percentage of fluorescently-labelled cells was evaluated by flow cytometry (FACS Calibur, Becton Dickinson, San Jose, CA). 2.4. Transfection of human prostate tumour xenografts in nude mice Animal experiments were performed in accordance with the regulations of the Austrian Federal Ministry for Education, Science and Culture. Male nude athymic balb-c mice (BALB/c-nu/AnN/Crl) were obtained from Charles River Laboratories (Sulzfeld, Germany) and housed under

pathogen-free conditions. Mice were anesthetized with ketasol (80–100 mg/kg) and xylasol (5–10 mg/kg) administered intraperitoneally. For our in vivo experiments, we used an androgenhypersensitive LNCaP subline called LNCaPabl, which has been developed in our laboratory by long-term androgen ablation of LNCaP cells [20]. These cells represent the features of castration-resistant tumour cells following androgen ablation therapy. They have increased AR expression and activity over LNCaP wild type cells and are able to grow in castrated mice. LNCaPabl tumours were grown by subcutaneous implantation of a 0.1 mL suspension of 1.5 × 106 LNCaPabl cells mixed with 0.1 mL matrigel (BD Biosciences, Bedford, MA) into both, the right and left flank of castrated mice, respectively. Castration was performed by orchiectomy of anesthetized animals. After approximately 7 weeks, tumours were randomly distributed into the different treatment groups. Microbubbles (5 × 107 ) were loaded with 100 ␮g digoxigenin-labelled antisense ODN (ODNasAR) in 100 ␮L 0.9% sodium chloride for 10 min on ice. They were injected either directly into the tumour (50% of tumour volume) or were administered intravenously (100 ␮L) via the mouse-tail vein. If no microbubbles were used, the ODN was diluted to 100 ␮g in 100 ␮L 0.9% sodium chloride. Ultrasound was focused to the tumour site with a 4C1 or a 15L8 transducer (Acuson Sequoia 512, Siemens Medical Solutions, Mountain View, CA), respectively, using Color Doppler ultrasound or

106

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

intermittent contrast pulse sequencing technology (CPS) with a mechanical index (MI) of 1.9 over 9 min. The animals were sacrificed and tumours and organs were collected 24 h after treatment. 2.5. Immunohistochemical detection of a digoxigenin-labelled ODN Tissues were fixed in 3.5% formaldehyde overnight at 4 ◦ C, washed with PBS, embedded in paraffin and subsequently cut into 2 ␮m sections followed by deparaffinization and rehydration. The digoxigenin antigen was enzymatically retrieved by 10 min incubation with 0.05 U Pronase (Nexes Protease, Ventana Medical Systems). Endogenous peroxidase activity and unspecific binding was inhibited with 3% H2 O2 in methanol followed by blocking in 5% nonfat dried milk in TBS buffer including 0.05% Tween-20. Sections were incubated in primary anti-digoxigenin-POD antibody (diluted 1:80, Roche), followed by detection with DAB (diaminobenzidine) using Zymed Polymer Detection System. After DAB staining, slides were counterstained with hematoxylin for 5 min. As controls, slides were incubated without primary antibody or were incubated with DAB substrate alone. TissueQuest software (TissueGnostics, Vienna) was used to determine digoxigenin staining intensity after immunofluorescent staining with a rhodamine-labelled antidigoxigenin antibody (Roche). 2.6. Western blot analysis of AR expression Cells were harvested with trypsin/EDTA and lysed in a 1× PAGE sample buffer (62 mM Tris–HCl, pH 6.8, 10% glycerol, 5% ␤-mercaptoethanol, 1 mg bromphenolblue, 2% sodium dodecylsulfate). Proteins were separated on 3–8% Tris–acetate polyacrylamide gels and subsequently transferred onto polyvinylidene difluoride (PVDF, Invitrogen Corporation, Carlsbad, CA) membranes. Blots were blocked with Blocking Buffer (Pierce) for 1 h at room temperature and then incubated at 4 ◦ C overnight with 1:500 diluted mouse monoclonal antihuman AR antibody (BioGenex Laboratories) and a 1:10,000 mouse monoclonal antihuman actin antibody (Chemicon International), both diluted in PBS with 0.2% Tween-20. Blots were washed and then incubated with a 1:6000 diluted fluorescently-labelled secondary antibody (Molecular Probes, Eugene, Oregon) for 1 h at room temperature. After washing, protein expression was quantified with an Infrared Imaging System (Odyssey). 2.7. Real time PCR to analyze AR mRNA expression in prostate tumours To isolate total RNA, the tumours were sliced into small pieces on ice, shock-frozen in liquid nitrogen, and extracted in 300 ␮L Tri reagent (Molecular Research Centre, Cincinnati, OH). The samples were sonicated followed by extraction of RNA through agitation at room temperature (RT) for

20 min. Afterwards, the samples were centrifuged to remove undissolved tissue. Another 700 ␮L Tri reagent was added together with 2 ␮L pellet paint (Novagen). After 5 min incubation at RT, 100 ␮L BCP (1-brom-3-chlor-propan) were added to the precipitated RNA and centrifuged. The aqueous RNA phase was precipitated by adding isopropanol. The pellet was washed with ethanol and subsequently dissolved in diethylpyrocarbonate-treated water. Concentration and integrity of the RNA were determined by spectrophotometric analysis at A260 and A280 . About 500 ng total RNA was reverse transcribed with Superscript III (Invitrogen) using N6 random primers. About 1 ␮L cDNA was mixed with 2× Mastermix (Applied Biosystems), 818 nM forward and reverse primer, respectively, and 136 nM Taqman probe (GenXpress, Vienna, Austria) and run in a 11 ␮L real time PCR reaction on an Applied Biosystems 7500 FAST machine using the standard program. The primers were selected to specifically recognize the human AR (sense: 5 -TCC AGG ATG CTC TAC TTC GCC-3 , antisense: 5 -TGA GAG AGG TGC CTC ATT CGG-3 , and Taqman probe: 5 -FAM-TGG TTT TCA ATG AGT ACC GCA TGC ACA-Tamra-3 ). The house keeping gene TATA box binding protein (TBP), with human specific primers, served as control (sense: 5 -CAC GAA CCA CGG CAC TGA TT-3 , antisense: 5 -TTT TCT TGC TGC CAG TCT GGA C-3 , and Taqman probe: 5 -Cy5-TCT TCA CTC TTG GCT CCT GTG CAC A-3 -BHQ2; black hole quencher2). Primers and Taqman probes were designed using Primer Express 1.0 software (PE Biosystems). Ct values were evaluated and the cycle numbers were normalized to the endogenous TBP. The relative amounts were calculated using the formula 2−Ct , where Ct = Ctsamples − Ctendogenous control . 2.8. Statistical analyses Data are expressed as mean values ± S.D. calculated from at least three independent experiments. Student’s t-test was used to evaluate statistically significant differences, which were considered at p < 0.05 (two-sided).

3. Results 3.1. Binding of ODNasAR to cationic microbubbles Initially, we analyzed whether our 15 nucleotides-short ODNasAR can be loaded onto the lipid surface of cationic microbubbles by ion charge binding. The principle of this process is schematically depicted in Fig. 1. Using spectrophotometry we found that by incubating 1 ␮M of the ODNasAR with 2 × 107 microbubbles 42% of the ODNasAR were bound to the lipid surface whereas the residual amount of the ODN was not bound to the microbubbles and recovered from the aqueous phase. Loading of the ODN to microbubbles was further confirmed by flow cytometric analysis of a fluorescently-labelled ODNasAR to microbubbles (Fig. 1).

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

3.2. Transfection of prostate cancer cells with the ODNasAR using microbubble-enhanced ultrasound In the next step, we established an in vitro model to examine whether the amount of ODNasAR that we managed to load to microbubbles was sufficient to transfect LNCaP prostate cancer cells with ultrasound. In vitro transfection was performed as described in Section 2. The mean transfection efficiency for LNCaP prostate cancer cells transfected with the ODNasAR by microbubble-enhanced ultrasound (49.43 ± 25.4%) was significantly increased as compared with untreated controls (1.26 ± 0.8%, p = 0.03, Fig. 2A). By contrast, less than 1% of cells were transfected by passive diffusion with the naked ODN and only a minute amount of cells showed uptake of the ODN after incubation with ODN-loaded microbubbles without ultrasound treatment. On the other hand, ultrasound treatment alone without the use of microbubbles was sufficient to transfer the ODN into 39.22 ± 21.8% of cells. Even if this transfection effi-

Fig. 2. (A) Transfection efficiency evaluated by flow cytometry. A fluorescently-labelled ODN targeting the human AR (ODNasAR, 10 nM) was delivered into LNCaP prostate cancer cells either alone (ODN), chargecoupled to microbubbles (Mb + ODN), with ultrasound alone (ODN + US), or with microbubbles and ultrasound (Mb + ODN + US). Sham-treated cells were incubated with microbubbles alone (Mb) or remained untreated (control). Statistically significant differences were indicated by an asterisk (* p < 0.05). (B) Effect on AR expression. AR protein levels were determined by Western blotting following delivery of 10 nM of the ODNasAR with ultrasound alone (ODN + US) or with ultrasound and microbubbles (Mb + ODN + US) and were compared with untreated controls (Co). Bands were visualized with an Infrared Imaging System (Odyssey) and normalized to actin expression for densitometric analysis. There is one representative Western blot depicted (M indicates the size marker).

107

ciency was not significantly different from untreated controls (p > 0.05), ultrasound seemed to be the driving mechanism of this delivery system. To further ascertain that the antisense ODN was taken up by the cells and that it was still active, we determined the effect of treatment on AR expression. As shown in Fig. 2B, AR protein levels were significantly decreased to 63.77% (n = 8) in cells treated with 10 nM of the ODNasAR delivered with microbubble-enhanced ultrasound as compared with untreated controls (set 100%, n = 6; p = 0.001). Corresponding with transfection efficiencies, however, we also observed downregulation of the AR (79.65%, n = 5; p = 0.268) when the ODNasAR was delivered into the cells with ultrasound alone. 3.3. Uptake of a digoxigenin-labelled ODN into prostate tumour xenografts For analysis of ODN uptake with microbubble-enhanced ultrasound into prostate tumours in vivo we used a human prostate tumour xenograft model representing advanced therapy-resistant prostate cancer. LNCaPabl cells were established in our laboratory by long-term culture of LNCaP wild type cells in an androgen-deprived medium. These cells form tumours after subcutaneous injection into castrated male nude mice, thus mimicking castration-resistant prostate cancer. A digoxigenin-labelled ODN targeting the human AR (ODNasAR) was used to analyze distribution of the ODN in xenografted prostate tumours as well as in mouse liver by immunohistochemistry, respectively. The ODN (100 ␮g) was either loaded to microbubbles or delivered alone by direct injection into the tumour (intratumorally) or by intravenous administration, respectively. Subsequently, ultrasound treatment was performed at the tumour site using a 15L8 transducer with 7 MHz and a mechanical index of 1.9 over 9 min. Animals were killed 24 h after treatment. As shown in Fig. 3, there was efficient uptake of the ODN in prostate tumour xenografts following delivery with microbubbles and ultrasound after intratumoral as well as after intravenous injection of loaded microbubbles. The ODNasAR was distributed throughout the tumours but digoxigenin staining was most prominent around blood vessels and stromal tissue compartments. Uptake of the ODNasAR was also found in the liver of treated mice with predominant accumulation in Kupffer cells. Similar to our findings in vitro, we also observed ODN uptake in prostate tumours when the naked ODN was delivered with ultrasound alone (Fig. 4). There was strong digoxigenin staining after intratumoral injection as well as after intravenous administration of naked ODN and subsequent treatment of the tumour with ultrasound. Again, there was also strong staining of digoxigenin in the liver of mice treated with ODNasAR and ultrasound. These data let us consider that ultrasound is the driving force of this delivery system. We therefore reasoned whether it would be possible to focus delivery of the ODNasAR to

108

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

Fig. 3. Uptake of ODNasAR in prostate tumours and mouse liver after delivery with microbubbles and ultrasound: a comparison between intratumoral and intravenous injection. Prostate tumour bearing mice were injected either intratumorally or intravenously with microbubbles which were loaded with 100 ␮g of a digoxigenin-labelled ODN. Tumours were then exposed to ultrasound. Organs were collected 24 h after treatment. Images show representative immunohistochemical digoxigenin staining in tumours (left panel) and liver (right panel) (A + B) untreated controls (C + D) intravenous administration of the ODNasAR with microbubbles and subsequent exposure of the tumor to ultrasound (E + F) delivery of the ODNasAR with microbubble-enhanced ultrasound after intratumoral administration.

the tumour site by ultrasound. To answer this question, we established two tumours in each animal at the left and the right flank, respectively, injected the ODN-loaded microbubbles intravenously, and then exposed only one of the tumours to ultrasound. To avoid scattering of ultrasound waves over the entire animal, mice were covered with a sterile air-filled thin synthetic sponge with a punched hole of the size of the tumour. The collateral tumour was not directly exposed to ultrasound and thus served as untreated control. As shown on two representative tumours in Fig. 5, digoxigenin staining intensity was significantly stronger in treated tumours

(16–49%) that were exposed to ultrasound as compared with the untreated collateral control tumours (2–18%). We thus concluded that focused ultrasound treatment results in enhanced uptake of the ODN at the tumour site. Since, in vitro experiments revealed best transfection efficiencies with low ultrasound frequencies (<2 MHz), we also addressed the issue of how much ultrasound frequency influences uptake of the ODN into prostate tumours with microbubble-enhanced ultrasound. We therefore compared two different ultrasound probes, the 4C1 transducer at low frequencies (1.5 and 2.5 MHz) and the 15L8 transducer with

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

109

Fig. 4. Uptake of the ODNasAR into prostate tumours after delivery with ultrasound alone. A digoxigenin-labelled ODNasAR was injected without microbubbles either intravenously or intratumorally followed by exposure of the tumour to ultrasound. Twenty-four hours later, digoxigenin staining was evaluated in tumours (A + C) and liver (B + D) of treated mice. (A) Tumour, intravenous injection; (B) liver, intravenous injection; (C) tumour, intratumoral injection; (D) liver, intratumoral injection.

a medium frequency (7 MHz), both at a constantly high mechanical index of 1.9, respectively. Immunohistochemical staining of digoxigenin showed that uptake of the ODN in prostate tumours was achieved with both ultrasound probes and all three ultrasound frequencies tested (Fig. 5). 3.4. Effect of treatment with the ODNasAR on AR expression in prostate tumours We next investigated whether delivery of the ODNasAR had any effect on AR expression. AR mRNA levels were analyzed in prostate tumours by real time PCR 24 h after treatment. Animals bearing tumours on both, the left and the right flank, were injected intravenously with microbubbles loaded with 100 ␮g of the digoxigenin-labelled ODNasAR. Subsequently, one tumour was exposed to ultrasound (7, 2.5, and 1.5 MHz frequency), whereas the collateral tumour on the other flank was not directly exposed to ultrasound. Although digoxigenin staining clearly demonstrated stronger uptake of the ODN in “treated” than in “collateral untreated” tumours as described above, there was no downregulation of AR mRNA expression. However, since the aim of this in vivo study, was to investigate the principle usability of microbubbleenhanced ultrasound to deliver the ODNasAR into prostate tumours, an effect on AR expression was unlikely expected with this experimental design (Fig. 6).

4. Discussion Previous studies in our lab demonstrated that downregulation of the AR with short antisense ODNs is a powerful strategy to inhibit prostate cancer cell growth in vitro as well as in vivo [1,2]. Although in vitro, these ODNs can be easily and efficiently delivered into prostate cancer cells either by electroporation or by the use of lipid-based transfection reagents, their delivery into prostate tumours in vivo still represents one of the major limiting factors for a clinical use. Several clinical trials in the past have demonstrated that systemic administration of “naked” ODNs is hardly successful since tremendous drug amounts are required to yield a sufficient response. The use of viral vectors undoubtedly results in sufficient transfection efficiencies in vivo, however, the use of viruses in humans may be associated with toxic side effects and severe immunogenic reactions [21]. The development of non-viral approaches for drug delivery therefore was subject of various studies in the past few years. An ideal in vivo transfection system has to fulfil several important features. These include sufficient stability of the drug during administration, high accumulation in the target tissue and tolerable adverse side effects of both, the drug as well as the delivery system, in the human body. To our mind, the use of microbubble contrast agents in combination with ultrasound may fulfil some of these important features. The main

110

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

Fig. 5. Comparison of different ultrasound frequencies to perform focused ultrasound treatment of tumour xenografts. We established two tumours in each animal by subcutaneous injection of tumour cells into the left and right flank, respectively. A digoxigenin-labelled ODNasAR (100 ␮g) charge-coupled to microbubbles was injected intravenously. Only one of the tumours was exposed to ultrasound (right panel: B, D and F) as described under results using different ultrasound frequencies whereas the collateral tumour on the other flank remained untreated (left panel: A, C and E). Digoxigenin staining in tumours was performed 24 h afterwards. (A + B) 15L8 transducer, 7 MHz; (C + D) 4C1 transducer, 2.5 MHz; (E + F) 4C1 transducer, 1.5 MHz.

advantage of microbubble-enhanced ultrasound is certainly its safe and non-toxic use in humans, as demonstrated by its routine use for diagnostic imaging. Microbubbles tend to accumulate in neo-vascularized tissue with increased blood flow, a phenomenon that also helps to facilitate the detection of prostate tumour tissue by ultrasound imaging [22]. Moreover, the chemistry of contrast agent microbubbles provides the possibility to use them as transport carriers for short oligonucleotides and thus to direct the load to the tumour site. There oligonucleotides can be released site-specifically by bursting the bubbles with ultrasound, a process that can be simultaneously followed on the screen. Microbubbles and ultrasound have already been used to transfer plasmid DNA

into various mammalian cells and tissues [19,23,24], as well as to deliver antisense ODNs into coronary endothelial cells in vitro [25] and into rats and human saphenous veins in vivo [26,27]. In the present study, we tested the usability of microbubble-enhanced ultrasound to deliver an antisense ODN targeting the human AR (ODNasAR) into prostate tumours. We therefore initially demonstrated the possibility to load the ODNs to cationic microbubbles and to transfer them to prostate cancer cells in vitro by bursting the loaded microbubbles with ultrasound. Transfection efficiency in our in vitro experiments was around 49%. These results are similar to data from Oberle et al., who found a 12–53% trans-

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

Fig. 6. AR mRNA expression in prostate tumours. Twenty-four hours after intravenous injection of the digoxigenin-labelled ODNasAR loaded to microbubbles, ARmRNA levels were assessed in tumours exposed to ultrasound (treated) and in their untreated collateral counterparts (untreated) using real time PCR. The graph shows AR mRNA levels normalized with the house keeping gene TATA box binding protein (TBP) for each individual animal.

fection efficiency following delivery of a GFP-containing plasmid entrapped into a lipid complex formulation with ultrasound with strong dependency on the cell type [28]. Best results in our in vitro transfection experiments were obtained with a low ultrasound frequency (1.75 MHz) and a high mechanical index (MI 1.9), resulting in extensive microbubble destruction associated with an intense transient nonlinear backscatter. Sufficient uptake of ODNs into prostate tumour cells was demonstrated by flow cytometric analyses of a fluorescently-labelled ODN as well as by downregulation of AR expression. The mechanism by which the ODN may be transferred to the cells is hypothesized to be due to “sonoporation”, defined as the transient formation of small pores in the cell membrane through which macromolecules are able to enter the cells. This phenomenon was previously associated with the effect of ultrasound used for cell transfection [28,29]. On the other hand, not only the effect of ultrasound but also the microbubbles themselves may contribute to cell transfection through the energy, which is released by bursting the microbubbles [30]. Thus, the microbubbles may help to increase transfection efficiency over the use of ultrasound alone, a phenomenon that we also observed in our in vitro experiments. Moreover, the microbubbles may serve as transport carriers for the ODN, an advantage which is over all important for systemic delivery in vivo. In a prostate tumour xenograft model in nude mice we were able to detect the ODNasAR in the tumour tissue not only after intratumoral injection but also after intravenous administration of microbubbles loaded with the ODN and subsequent treatment of the tumour with ultrasound. Accumulation of the ODN was strongest in stromal compartments of the tumours and around blood vessels. In comparison to our results, Huber and Pfisterer found that transfection of plasmid DNA into Dunning prostate tumours with ultrasound was only effective when the DNA was injected directly into the tumour [31]. Thus,

111

the microbubbles may help to target the ODN to the tumour tissue when delivered systemically. In view of using this delivery system in patients, it has to be considered that transrectal ultrasound probes which are used for imaging of the prostate are not yet available at very low frequencies. Best transfection efficiencies in vitro, however, were obtained with ultrasound frequencies of 1.75 MHz. We therefore investigated whether uptake of the ODN into prostate tumours in vivo may also be possible with higher ultrasound frequencies. Using our human prostate tumour xenograft model, we tested different ultrasound probes with low (1.5, 2.5 MHz, 4C1) to medium (7 MHz, 15L8) ultrasound frequencies. In all treatment groups, intravenous injection of microbubbles charge-coupled with a digoxigeninlabelled ODNasAR and subsequent exposure of the tumour to ultrasound resulted in a significant increase in digoxigenin staining in prostate tumours as compared with untreated collateral control tumours. Interestingly, transfection of the ODNasAR with ultrasound alone without using microbubbles as transport carriers also resulted in efficient uptake of the ODN into xenografted prostate tumours. Digoxigenin staining was found irrespective of intratumoral or intravenous injection of the ODN, respectively. This result is in accordance with our in vitro data, where cell transfection with the ODNasAR was also achieved with ultrasound alone. This finding is also in concordance with other reports. Huber et al. [32], for instance, reported on enhanced expression of a reporter plasmid after ultrasound exposure alone. However, they also reported on a further increase in reporter plasmid expression by combining ultrasound with contrast agent microbubbles as carrier molecules for the DNA. Unfortunately, in our in vivo studies, we were not able to demonstrate a difference in ODN uptake between delivery with ultrasound alone or with the combined use of microbubbles. One of the advantages of using ultrasound for ODN delivery is the possibility to focus treatment to the tumour site. To test this, we injected mice with prostate tumour cells on both, the left and the right flank, in order to obtain two tumours in each animal. Following intravenous administration of the ODNasAR with or without microbubbles, only one of the tumours was exposed to ultrasound whereas the collateral tumour remained “untreated”. Although this experimental design is rather difficult in mice, we managed to strongly focus ultrasound treatment to only one tumour, resulting in strong ODN uptake in treated tumours but negative to very weak digoxigenin staining in untreated tumours. Nevertheless, it has to be noted that irrespective of intravenous administration or intratumoral injection of the ODN with or without microbubbles, there was always digoxigenin staining in the liver of treated animals whereas there was no staining in the liver of untreated animals. This accumulation of the ODN in the liver was mainly associated with Kupffer cells. Similar data have already been described by others, reporting on uptake of microbubble contrast agents by Kupffer cells in the liver of rats through phagocytosis [33].

112

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113

Unfortunately, we were not able to demonstrate that uptake of the ODN into prostate tumours was associated with a reduction in AR expression in vivo. Since, the therapeutic effect of the ODNasAR has already been shown in previous studies and the aim of this study was only to show delivery of the ODN into prostate tumours with microbubbleenhanced ultrasound, treatment was performed only once and tumours were harvested 24 h afterwards. Previous in vivo studies with the ODNasAR however revealed that an effect on AR expression as well as on tumour growth can only be seen after long-term treatment of several weeks. Considering short-term treatment and a very low concentration of the ODN (100 ␮g) in this study, a therapeutic effect was hardly expected. Moreover, xenografted LNCaPabl prostate tumours grow very slowly and heterogeneously and thus, to observe any effect on tumour growth, the tumours will probably have to be treated over a longer period of time. Therefore, to investigate an effect on tumour growth, ODN concentration and the time of treatment have to be optimized. In summary, this study demonstrates that microbubbleenhanced ultrasound can be used to deliver the ODNasAR into prostate tumours. According to our results, ultrasound is considered as the driving force of this delivery system since uptake of the ODN was also observed in tumors after treatment with ultrasound alone with only minor differences compared to the combined use of microbubbles and ultrasound. Acknowledgements We want to thank Reinhold Ramoner for help with flow cytometry, Hannes Steiner and Andreas P. Berger for their technical assistance with animal experiments. Moreover, we want to thank Radu Rogojanu for his help with TissueQuest software. This work was supported by the Austrian Research Foundation (FWF P16882-B13) and the EU FP6 program (FP6-504587). References [1] I.E. Eder, Z. Culig, R. Ramoner, M. Thurnher, T. Putz, C. NesslerMenardi, M. Tiefenthaler, G. Bartsch, H. Klocker, Inhibition of LNCaP prostate cancer cells by means of androgen receptor antisense oligonucleotides, Cancer Gene Ther. 7 (7) (2000) 997–1007. [2] I.E. Eder, J. Hoffmann, H. Rogatsch, G. Sch¨afer, D. Zopf, G. Bartsch, H. Klocker, Inhibition of LNCaP prostate tumor growth in vivo by an antisense oligonucleotide directed against the human androgen receptor, Cancer Gene Ther. 9 (2002) 117–125. [3] A. Ghosh, Different approaches of gene therapy used in prostate cancer, Cell. Mol. Biol. (Noisy-le-grand) 51 (1) (2005) 103–111. [4] S. Bao, B.D. Thrall, D.L. Miller, Transfection of a reporter plasmid into cultured cells by sonoporation in vitro, Ultrasound Med. Biol. 23 (6) (1997) 953–959. [5] K. Anwer, G. Kao, B. Proctor, I. Anscombe, V. Florack, R. Earls, E. Wilson, T. McCreery, E. Unger, A. Rolland, S.M. Sullivan, Ultrasound enhancement of cationic lipid-mediated gene transfer to primary tumors following systemic administration, Gene Ther. 7 (21) (2000) 1833–1839.

[6] N. Kyprianou, H.F. English, J.T. Isaacs, Programmed cell death during regression of PC-82 human prostate cancer following androgen ablation, Cancer Res. 50 (1990) 3748–3753. [7] A.G. Rosenberg, A.G. von Eschenbach, Hormonal therapy for prostate cancer, Semin. Surg. Oncol. 6 (1990) 71–76. [8] A. Hobisch, Z. Culig, C. Radmayr, G. Bartsch, H. Klocker, A. Hittmair, Androgen receptor status of lymph node metastases from prostate cancer, Prostate 28 (2) (1996) 129–135. [9] A. Hobisch, Z. Culig, C. Radmayr, G. Bartsch, H. Klocker, A. Hittmair, Distant metastases from prostatic carcinoma express androgen receptor protein, Cancer Res. 55 (1995) 3068–3072. [10] T.H. Van der Kwast, J. Schalken, J.A. Ruizeveld de Winter, C.C.J. Van Vroonhoven, E. Mulder, W. Boersma, J. Trapman, Androgen receptors in endocrine-therapy-resistant human prostate cancer, Int. J. Cancer 48 (1991) 189–193. [11] C.D. Chen, D.S. Welsbie, C. Tran, S.H. Baek, R. Chen, R. Vessella, M.G. Rosenfeld, C.L. Sawyers, Molecular determinants of resistance to antiandrogen therapy, Nat. Med. 10 (1) (2004) 33–39. [12] Z. Culig, A. Hobisch, M.V. Cronauer, A.C.B. Cato, A. Hittmair, C. Radmayr, J. Eberle, G. Bartsch, H. Klocker, Mutant androgen receptor detected in an advanced-stage prostatic carcinoma is activated by adrenal androgens and progesterone, Mol. Endocrinol. 7 (12) (1993) 1541–1550. [13] Z. Culig, A. Hobisch, M.V. Cronauer, C. Radmayr, J. Trapman, A. Hittmair, G. Bartsch, H. Klocker, Androgen receptor activation in prostatic tumor cell lines by insulin-like growth factor-I, keratinocyte growth factor and epidermal growth factor, Cancer Res. 54 (1994) 5474–5478. [14] Z. Culig, A. Hobisch, M. Cronauer, C. Radmayr, A. Hittmair, J. Zhang, M. Thurnher, G. Bartsch, H. Klocker, Regulation of prostatic growth and function by peptide growth factors, Prostate 28 (1996) 329–405. [15] J. Veldscholte, C. Ris-Stalpers, G.G.J.M. Kuiper, G. Jenster, C. Berrevoets, E. Claassen, H.C.J. van Rooij, J. Trapman, A.O. Brinkmann, E. Mulder, A mutation in the ligand binding domain of the androgen receptor of human LNCaP cells affects steroid binding characteristics and response to anti-androgens, Biochem. Biophys. Res. Commun. 17 (2) (1990) 534–540. [16] J. Veltscholte, C.A. Berrevoets, N.D. Zegers, T.H. van der Kvast, J.A. Grootegoed, E. Mulder, Hormone-induced dissociation of the androgen receptor-heat-shock protein complex. Use of a new monoclonal antibody to distinguish transformed from nontransformed receptors, Biochemistry 31 (1992) 7422–7430. [17] X.-Y. Zhao, P.J. Malloy, A.V. Krishnan, S. Swami, N.M. Navone, D.M. Peehl, D. Feldman, Glucocorticoids can promote androgenindependent growth of prostate cancer cells through a mutated androgen receptor, Nat. Med. 6 (6) (2000) 703–706. [18] L. Yang, L. Wang, H.K. Lin, P.Y. Kan, S. Xie, M.Y. Tsai, P.H. Wang, Y.T. Chen, C. Chang, Interleukin-6 differentially regulates androgen receptor transactivation via PI3K-Akt, STAT3, and MAPK, three distinct signal pathways in prostate cancer cells, Biochem. Biophys. Res. Commun. 305 (3) (2003) 462–469. [19] J.P. Christiansen, B.A. French, A.L. Klibanov, S. Kaul, J.R. Lindner, Targeted tissue transfection with ultrasound destruction of plasmidbearing cationic microbubbles, Ultrasound Med. Biol. 29 (12) (2003) 1759–1767. [20] Z. Culig, J. Hoffmann, M. Erdel, I.E. Eder, A. Hobisch, A. Hittmair, G. Bartsch, G. Utermann, M.R. Schneider, K. Parczyk, H. Klocker, Switch from antagonist to agonist of the androgen receptor blocker bicalutamide is associated with prostate tumour progression in a new model system, Br. J. Cancer 81 (1999) 242–251. [21] J.M. Isner, Myocardial gene therapy, Nature 415 (6868) (2002) 234–239. [22] F. Frauscher, A. Klauser, E.J. Halpern, W. Horninger, G. Bartsch, Detection of prostate cancer with a microbubble ultrasound contrast agent, Lancet 357 (2001) 1849–1850. [23] E.C. Unger, E. Hersh, M. Vannan, T. McCreery, Gene delivery using ultrasound contrast agents, Echocardiography 18 (4) (2001) 355–361.

P. Haag et al. / Journal of Steroid Biochemistry & Molecular Biology 102 (2006) 103–113 [24] R.J. Price, D.M. Skyba, S. Kaul, T.C. Skalak, Delivery of colloidal particles and red blood cells to tissue through microvessel ruptures created by targeted microbubble destruction with ultrasound, Circulation 98 (13) (1998) 1264–1267. [25] S. Miura, K. Tachibana, T. Okamoto, K. Saku, In vitro transfer of antisense oligodeoxynucleotides into coronary endothelial cells by ultrasound, Biochem. Biophys. Res. Commun. 298 (4) (2002) 587–590. [26] J.M. Erikson, G.L. Freeman, B. Chandrasekar, Ultrasound-targeted antisense oligonucleotide attenuates ischemia/reperfusion-induced myocardial tumor necrosis factor-alpha, J. Mol. Cell. Cardiol. 35 (1) (2003) 119–130. [27] T. Kodama, P.H. Tan, I. Offiah, T. Partridge, T. Cook, A.J. George, M.J. Blomley, Delivery of oligodeoxynucleotides into human saphenous veins and the adjunct effect of ultrasound and microbubbles, Ultrasound Med. Biol. 31 (12) (2005) 1683–1691. [28] V. Oberle, G. de Jong, J.I. Drayer, D. Hoekstra, Efficient transfer of chromosome-based DNA constructs into mammalian cells, Biochim. Biophys. Acta 1676 (3) (2004) 223–230.

113

[29] J.D. Moroz, P. Nelson, Dynamically stabilized pores in bilayer membranes, Biophys. J. 72 (5) (1997) 2211–2216. [30] S. Chen, R.V. Shohet, R. Bekeredjian, P. Frenkel, P.A. Grayburn, Optimization of ultrasound parameters for cardiac gene delivery of adenoviral or plasmid deoxyribonucleic acid by ultrasound-targeted microbubble destruction, J. Am. Coll. Cardiol. 42 (2) (2003) 301– 308. [31] P.E. Huber, P. Pfisterer, In vitro and in vivo transfection of plasmid DNA in the Dunning prostate tumor R3327-AT1 is enhanced by focused ultrasound, Gene Ther. 7 (17) (2000) 1516–1525. [32] P.E. Huber, M.J. Mann, L.G. Melo, A. Ehsan, D. Kong, L. Zhang, M. Rezvani, P. Peschke, F. Jolesz, V.J. Dzau, K. Hynynen, Focused ultrasound (HIFU) induces localized enhancement of reporter gene expression in rabbit carotid artery, Gene Ther. 10 (18) (2003) 1600– 1607. [33] K. Shigeta, K. Itoh, S. Ookawara, N. Taniguchi, K. Omoto, The effects of Levovist and DD-723 in activating platelets and damaging hepatic cells of rats, J. Ultrasound Med. 24 (7) (2005) 967–974.