The androgen receptor (AR) plays a critical role in the progression of prostate cancer. Silencing this protein using short-hairpin RNA (shRNA) has been correlated with tumor growth inhibition and decreases in serum prostate specific antigen (PSA). In our study, we have investigated the ability of lipid nanoparticle (LNP) formulations of small-interfering RNA (siRNA) to silence AR in human prostate tumor cell lines in vitro and in LNCaP xenograft tumors following intravenous (i.v.) injection. In vitro screening studies using a panel of cationic lipids showed that LNPs containing the ionizable cationic lipid 2,2-dilinoleyl-4-(2-dimethylaminoethyl)-[1,3]-dioxolane (DLin-KC2-DMA) exhibited the most potent AR silencing effects in LNCaP cells. This is attributed to an optimized ability of DLin-KC2-DMA-containing LNP to be taken up into cells and to release the siRNA into the cell cytoplasm following endocytotic uptake. DLin-KC2-DMA LNPs were also effective in silencing the AR in a wild-type AR expressing cell line, LAPC-4, and a variant AR expressing cell line, CWR22Rv1. Importantly, it is demonstrated that LNP AR-siRNA systems containing DLin-KC2-DMA can silence AR gene expression in distal LNCaP xenograft tumors and decrease serum PSA levels following i.v. injection. To our knowledge, this is the first report demonstrating the feasibility of LNP delivery of siRNA for silencing AR gene expression in vivo.
Prostate cancer is the most frequently diagnosed nonskin cancer and the second-leading cause of cancer-related deaths in males.1 It is projected that the number of new cases in the United States will exceed 200,000 in 2012. While often curable in its early stages by surgical or radiation ablation, the outcome for advanced or recurrent disease is less favorable, even if some form of androgen withdrawal therapy is employed. The cancer eventually progresses to a castration-resistant form that results in a median survival of 12–18 months.2, 3 Survival extensions of a few months can be achieved using docetaxel-based therapies4; however, more durable treatment options are clearly required.
A potentially new therapeutic strategy is to utilize RNA interference (RNAi) to silence critical genes involved in prostate cancer growth.5–7 The androgen receptor (AR) represents one such target as the inappropriate activation/dysfunction of this protein is believed to be responsible for the development of castration-resistant prostate cancer.6, 8 Silencing of the AR using small-interfering RNA (siRNA) has been shown to increase cellular apoptosis due to disruption of the Bcl-xL-mediated survival signaling pathway.9 Prior in vivo experiments using short-hairpin RNAs (shRNAs) against the AR have been shown to inhibit prostate tumor growth in mice and in castration-resistant disease and to cause tumor regression in 50% of the tumors.6
The design of lipid nanoparticle (LNP) systems for in vivo delivery of siRNA is complicated by the need to employ cationic lipids to achieve efficient encapsulation of negatively charged lipid polymers such as siRNA. The use of cationic lipids usually leads to a positive charge on the LNP carrier, which results in enhanced serum protein adsorption and rapid clearance from the circulation following intravenous (i.v.) injection, leading to lack of penetration to target tissue.10 We have addressed this problem by designing ionizable cationic lipids that have pKa values of 7 or below, enabling encapsulation of nucleic acid polymers at low pH values (e.g., pH 4) but also allowing the LNP to exhibit a near neutral surface charge at physiological pH values.11, 12 Such LNP systems are of increasing utility for the in vivo delivery of siRNA.13, 14
Recent work has shown that the in vivo gene silencing potencies of LNP siRNA systems following i.v. administration are sensitive to relatively small changes in the structure of the ionizable cationic lipids employed. For example, the in vivo (hepatocyte) gene silencing activity of LNP siRNA systems containing 2,2-dilinoleyl-4-(2-dimethylaminoethyl)-[1,3]-dioxolane (DLin-KC2-DMA), 1,2-dilinoleyloxy-ketal-N,N-dimethyl-3-aminopropane (DLinKDMA), 1,2-dilinoleyloxy-N,N-dimethyl-3-aminopropane (DLinDMA) and 1,2-dilineoyl-3-dimethylammonium-propane (DLinDAP) varies according to the relationship DLin-KC2-DMA > DLinKDMA > DLinDMA ≫ DLinDAP.14 In this work, we compare levels of AR gene silencing induced in vitro employing LNP siRNA systems containing DLin-KC2-DMA, DLinKDMA, DLinDMA and DLinDAP and show that LNP siRNA systems containing DLin-KC2-DMA are the most effective systems for achieving intracellular delivery and gene silencing of AR in the LNCaP cell lines. They are also effective for silencing AR in LAPC-4 and CWR22Rv1 human prostate cancer cell lines. Furthermore, it is shown that LNP AR-siRNA systems containing DLin-KC2-DMA effectively silence the AR gene in distal LNCaP xenograft tumors following i.v. injection.
Distearoylphosphatidylcholine (DSPC) and cholesterol (Chol) were purchased from Avanti Lipids, 1,1′-dilinoleyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) and 3,3′-dioctadecyl-5,5′-di(4-sulfophenyl) oxacarbocyanine (SP-DiOC18) was obtained from Invitrogen (Burlington, ON, Canada). The ionizable cationic lipids 2,2-dilinoleyl-4-(2-dimethylaminoethyl)-[1,3]-dioxolane (DLin-KC2-DMA), 1,2-dilinoleyloxy-keto-N,N-dimethyl-3-aminopropane (DLinKDMA), 1,2-dilinoleyloxy-N,N-dimethyl-3-aminopropane (DLinDMA) and 1,2-dilineoyl-3-dimethylammonium-propane (DLinDAP) were provided by Tekmira Pharmaceuticals (Burnaby, BC, Canada) or were synthesized using established methods.14 PEG-S-DMG (3-N-[(ω-methoxypoly(ethyleneglycol)2000)carbamoyl]-1,2-dimyristyloxy-propylamine) and PEG-C-DOMG (R-3-[(ω-methoxy-poly(ethyleneglycol)2000)carbamoyl)]-1,2-dimyristyloxypropyl-3-amine) were provided by Alnylam Pharmaceuticals (Boston, MA). All other chemicals were of reagent or analytical grade.
Cell culture, cell lines and reagents
The LNCaP, LNCaP-eGFP, LAPC-4 and CWR22Rv1 human prostate cancer cell lines were used in all in vitro experiments.15–18 LNCaP, LAPC-4 and CWR22Rv1 cells were originally obtained from ATCC and were not passaged beyond 6 months after receipt or resuscitation. Cells maintained in RPMI 1640 (Invitrogen, Burlington, ON, Canada) supplemented with 10% heat-inactivated fetal bovine serum (FBS) and incubated at 37°C and 5% CO2. The LNCaP and LNCaP-eGFP cell line was supplemented with 0.1 nM R1881 (Sigma-Aldrich, Oakville, ON, Canada).
Sequence of the human AR gene (Genbank accession no. NM_000044) was extracted from the NCBI Entrez nucleotide database. The 25-mer AR-siRNA was purchased as Stealth RNAi siRNA (Invitrogen): 5′-AGCACUGCUACUCUUCAGCAUUAUU-3′ (AR sense) and 5′-AAUAAUGCUGAAGAGUAGCAGUGCU-3′ (AR antisense). An additional scrambled (SC) LoGC negative control was purchased as a Stealth RNAi siRNA (Invitrogen, Burlington, ON, Canada). Luciferase (LUC) siRNA was a gift from Alnylam Pharmaceuticals (Boston, MA): 5′-cuuAcGcuGAGuAcuucGAdTsdT-3′ (LUC sense) and 5′-UCGAAGuACUcAGCGuAAGdTsdT-3′ (LUC antisense). Lower case letters = 2′-OMe; Upper case = RNA; s = phosphothioate backbone.
Encapsulation of siRNA into LNPs
Lipid stocks [DSPC, PEG-S-DMG, PEG-C-DOMG, cholesterol, SP-DiOC18 and a cationic lipid (DLin-KC2-DMA, DLinKDMA, DLinDMA and DLinDAP)] were co-dissolved in ethanol at appropriate molar ratios (10 mol % DSPC, 10 mol % PEG-S-DMG, 39.8 mol % Chol, 0.2 mol % SPDiOC18 or DiI and 40 mol % cationic lipid). In some cases, other molar ratios of lipids [10 mol % DSPC, 2.5 (or 10) mol % PEG-C-DOMG, 47.5 (or 40) mol % cholesterol and 40 mol % DLin-KC2-DMA] were constructed to examine the effect of PEG on in vitro transfection. Multilamellar vesicles (MLVs) were generated by adding the lipid in ethanol to an aqueous buffer (50 mM citrate, pH 4.0) to achieve a final concentration of 40% ethanol. Large unilamellar vesicles (LUVs) were generated by extruding the MLVs through two nuclepore polycarbonate filters (80 nm, 10 passes) using an extrusion device from Northern Lipids (Vancouver, BC, Canada) at ∼300 psi.
The siRNA was dissolved in 10 mM citrate, 30 mM NaCl, pH 6.0, quantified by measuring absorbance at OD260 and added to the LUV dispersion dropwise under constant vortex at an siRNA:lipid ratio of 1:10 (wt/wt). The resulting LNP siRNA systems were then dialyzed in PBS overnight to increase the pH to ∼7.4. The mean diameter of the vesicles was determined using a NICOMP370 particle sizer (Nicomp Particle sizing, Santa Barbara, CA) using the intensity mode. For LNP containing PEG-S-DMG diameters were 75 ± 12 nm (DLinDAP), 73 ± 6 nm (DLinDMA), 71 ± 7 nm (DLinKDMA) and 70 ± 8 nm (DLin-KC2-DMA). LNP containing PEG-C-DOMG had slightly larger diameters, 86 ± 9 nm (DLin-KC2-DMA, 2.5% PEG-C-DOMG) and 82 ± 10 nm (DLin-KC2-DMA, 10% PEG-C-DOMG). Lipid concentrations were measured by total cholesterol (T-CHO) using the cholesterol E enzymatic assay from Wako Chemicals (Richmond, VA). Free siRNA was removed using VivaPureD MiniH columns (Sartorius Stedim Biotech GmbH, Goettingen, Germany). The elutants were then dissolved in 75% ethanol and siRNA was quantified by measuring absorbance at OD260. The siRNA encapsulation efficiency was 80% or greater for all formulations.
Western blotting and immunofluorescence
LNCaP, LAPC-4 and CWR22Rv1 cells were plated in 12-well plates (2.0 × 105 cells per well). Cells were washed in PBS and lysed in RIPA buffer (1% NP-40, 0.25% deoxycholic acid) with protease inhibitor tablets (Roche Diagnostics, Indianapolis, IN). Total protein (10 μg) quantified by Bradford Assay was analyzed by immunoblotting. Antibodies to AR from Santa Cruz Biotechnology (AR-441) (Santa Cruz, CA). Antibodies to GAPDH from Abcam (Cambridge, MA). Antigen–antibody complexes were detected using Millipore Immobilon Western Chemiluminescent HRP Substrate (Billerica, MA).
LNCaP cells were seeded onto poly-D-lysine cover slips (2.0 × 105 cells) per well in a 12-well plate. Cells were treated with 5 μg/ml of AR-siRNA encapsulated in LNPs containing DLin-KC2-DMA, DLinKDMA, DLinDMA and DLinDAP for 48 hr. Cells were then fixed in 3% PFA with Hoechst's stain and examined under an Olympus FV1000 (Center Valley, PA) laser scanning microscope.
LNCaP cells were seeded (2.0 × 104 cells) per well in a 96-well plate. Cells were treated with 1 μg/ml of AR-siRNA encapsulated in LNPs containing DLin-KC2-DMA, DLinKDMA, DLinDMA and DLinDAP for 4, 12, 24 and 48 hr. Cells were fixed in 3% PFA with Hoechst's stain and examined using a Cellomics ArrayScan VTI HCS Reader (Thermo Scientific, Pittsburgh, PA).
LNCaP-eGFP cells were treated with DiI-labeled LNPs encapsulating either AR- or LUC-siRNA. Cellular uptake and AR knockdown were then assessed using a LSRII flow cytometer and FACS Diva Software (BD Bioscience, Billerica, MA) by measuring the fluorescence intensity of DiI and GFP, respectively.
Assessing effect of siRNA to knockdown AR in vivo
LNCaP xenograft prostate tumors were established as described previously.6 When tumors became palpable, volumes were measured (width × length × depth × 0.5236) and blood was collected from the tail vein to assess serum prostate specific antigen (PSA) by ELISA (ClinPro International, Union City, CA). Once PSA values reached 50–100 ng/ml, animals were randomized into three groups. LNP containing AR-siRNA or LNP scrambled siRNA (SC-siRNA) and DLin-KC2-DMA were i.v. administered through the lateral tail vein. PBS was injected as a baseline control. Doses of LNP siRNA were administered three times on consecutive days at a dose level of 10-mg siRNA/kg of mouse body weight. Further injections were made an additional three times on Days 7, 9 and 11. Animals were sacrificed at Day 14. Serum PSA and protein levels of AR in the xenograft tumors were analyzed. All animal procedures were done according to the guidelines of the Canadian Council of Animal Care with appropriate institutional certification.
5′-RNA-linker-mediated rapid amplification of cDNA ends PCR
5′-RNA-linker-mediated rapid amplification of cDNA ends (5′-RLM-RACE) was performed using a GeneRacer kit according to the manufacturer's protocol (Invitrogen, Burlington, ON, Canada). LNCaP cells were transfected with 10 nM of AR-siRNA with oligofectamine.19 After 72 hr, LNCaP cells were harvested. Total RNA from LNCaP cells after siRNA treatment or from tumor tissues collected from mice treated with AR-siRNA, scrambled siRNA or PBS was isolated using the Trizol method (Invitrogen, Burlington, ON, Canada). Hundred nanograms of total RNA was linked to an RNA oligonucleotide with T4 RNA ligase. Purified RNA after DNAse I digestion were reverse transcribed. PCR was performed with primers specific for the RNA linker and for the targeted gene with the following conditions: 94°C, 2 min; 54°C, 1 min and 72°C, 1 min; then 35 cycles of 94°C, 30 sec; 54°C, 1 min and 72°C, 30 sec, and finally an extension period at 72°C, 7 min. Five microliters of this product was taken for nested PCR using the same conditions, but with the annealing temperature decreased to 50°C. To use as a control, GAPDH primers were used to amplify the cDNA transcribed with random hexamers. Ten microliters of each reaction was analyzed on a 2% agarose gel visualized with SybrSafe (Invitrogen, Burlington, ON, Canada). The band corresponding to the predicted size of the amplicon was excised and purified using QIAquick Gel Extraction kit (Qiagen, Mississauga, ON, Canada). Subsequently, the fragment was cloned using a TA Cloning kit supplied with the GeneRacer kit and sequenced.
All statistical analyses were performed using GraphPad. Initially, a one-way ANOVA was used to statistically evaluate the differences between treatment groups. In the case of statistically significant results, the differences between treatment groups were assessed through the use of the Tukey–Kramer multiple comparisons test unless otherwise stated. Probability (p) values less than 0.05 were considered significant.
LNP AR-siRNA systems formulated with DLin-KC2-DMA exhibit maximum levels of AR gene silencing in LNCaP cells
Previous studies have shown that upregulation of the AR occurs at both the mRNA and protein levels and causes hypersensitivity to androgens even at small concentrations.20 We selected an siRNA sequence that has been demonstrated to effectively silence AR and halt tumor progression in a study using inducible plasmid shRNA to knockdown AR expression in vivo.6 The AR-siRNA and SC-siRNA was encapsulated as indicated in Material and Methods section using four species of LNPs containing DLin-KC2-DMA, DLinKDMA, DLinDMA or DLinDAP (Fig. 1). Essentially complete AR knockdown was seen in cells incubated with DLin-KC2-DMA LNP encapsulating AR-siRNA at a concentration of 5 μg/ml (Fig. 2a). Significant AR knockdown was also detected when LNP AR-siRNA systems containing DLinKDMA or DLinDMA were used; however, low amounts of protein still remained (Figs. 2a–2c). Cells treated with LNP containing SC-siRNA showed normal expression levels of AR.
The results shown in Figure 2a suggest that LNP AR-siRNA containing DLin-KC2-DMA are more potent than LNP containing DLinKDMA, DLinDMA or DLinDAP. However, the nearly complete knockdown of AR for the DLin-KC2-DMA system implies that saturating levels of LNP-siRNA are present, which makes comparative estimates of relative potency difficult. When the experiments were repeated using siRNA at a concentration of 1 μg/ml, there again was substantial knockdown of AR with the LNP AR-siRNA system containing DLin-KC2-DMA, with LNP systems containing DLinKDMA or DLinDMA being less effective (Fig. 2b). Quantification of the residual AR normalized to GAPDH expression and relative to the scrambled control (Fig. 2c) revealed a dose-dependent drop in residual AR levels for DLin-KC2-DMA and DLinKDMA. The relative potency of the LNP formulations was DLin-KC2-DMA > DLinDMA > DLinKDMA ≫ DLinDAP (Fig. 2c).
The potency of LNP siRNA systems containing DLin-KC2-DMA can be attributed to improved siRNA uptake and endosome release properties
The superior potency of LNP AR-siRNA systems containing DLin-KC2-DMA could arise due to increased uptake into cells and/or improved release from endosomes following uptake. To better understand how these factors contribute, we first compared the potency of these lipids with their uptake into cells. Higher positive surface charges would be expected to lead to greater uptake into cells due to enhanced potential for association with the negatively charged cell exterior. In this regard, LNPs containing DLin-KC2-DMA, DLinKDMA, DLinDMA or DLinDAP will exhibit different surface charges due to the differing pKa values of these ionizable cationic lipids. Previous work14 has shown that DLin-KC2-DMA, DLinKDMA, DLinDMA or DLinDAP exhibit pKa values of 6.7, 5.9, 6.8 and 6.2, respectively. These results suggest that LNP containing DLin-KC2-DMA or DLinDMA should exhibit higher levels of uptake than those observed for LNPs containing DLinKDMA or DLinDAP, which have lower pKa values. Uptake experiments employed LNCaP cells incubated with AR-siRNA LNP containing DLin-KC2-DMA, DLinKDMA, DLinDMA or DLinDAP, as well as the SPDiOC18 fluorescent lipid label, for 4, 12, 24 and 48 hr. Uptake was measured as the mean fluorescence intensity per cell as determined employing the Cellomics ArrayScan apparatus, which averaged fluorescent intensity of over ∼400 LNCaP cells. LNP containing DLin-KC2-DMA exhibited the highest level of uptake at 48 hr, followed closely by DLinDMA (Fig. 3b). Confocal microscopy images taken at 48 hr are consistent with increased uptake in LNCaP cells when DLin-KC2-DMA is used in the LNP formulation (Fig. 3a). The cellular uptake data correlates with the extent of AR knockdown (Figs. 2b and 2c) and suggests that uptake is a major contributing factor to the differential knockdown profiles observed for the four cationic lipids (Fig. 3).
To further assess the correlation between knockdown, LNP uptake and endosomal release, flow cytometry studies were performed using an LNCaP cell line that stably expresses an AR-responsive promoter linked to an eGFP reporter. As shown elsewhere, in the absence of AR expression no GFP expression is observed18 in this cell line. Cells were incubated (1-μg siRNA/ml for 48 hr) with LNP containing DLin-KC2-DMA, DLinKDMA, DLinDMA or DLinDAP and the fluorescent lipid analogue (DiI). These LNP contained either AR-siRNA or control (Luc)-siRNA. Uptake was measured by quantifying DiI (red) fluorescence, and GFP expression was used as a surrogate for AR expression. The GFP expression was normalized to cellular uptake (i.e., the ratio of GFP fluorescence to DiI fluorescence) for both AR and control-siRNA sequences. The ratio of normalized GFP expression for LNP containing AR siRNA to normalized GFP expression for LNP containing control siRNA then gives a measure of the relative potencies of the different cationic lipids for inducing endosomal escape. As shown in Figure 3c, DLin-KC2-DMA showed the highest potency, followed closely by DLinDMA, indicating that DLin-KC2-DMA and DLinDMA have an increased ability to facilitate endosomal release of siRNA as compared to DLinKDMA or DLinDAP.
The species of PEG–lipid influences LNP siRNA-induced AR silencing
The LNP siRNA systems used above contain 10 mol % PEG-S-DMG, which contains labile ester linkages between the acyl chains and the headgroup. To achieve more chemically stable systems on storage, and to make in vivo results more comparable with recent in vivo studies using related LNP systems,14 we employed PEG-C-DOMG in place of PEG-S-DMG for in vivo studies. PEG-C-DOMG is more chemically stable than PEG-S-DMG due to the presence of ether bonds linking the acyl chains to the headgroup (Fig. 1b). This substitution required crossover studies to determine the relative effects of PEG-C-DOMG and PEG-S-DMG on LNP potency.
As shown in Figure 4a, LNP AR-siRNA systems containing 10 mol % PEG-C-DOMG exhibited reduced silencing of the AR as compared to systems containing 10 mol % PEG-S-DMG. These results could be explained by a slower rate of release of PEG-C-DOMG from the LNP surface as compared to PEG-S-DMG. As discussed elsewhere, the PEG–lipids have been engineered to exhibit rapid leaving rates from the LNP surface following i.v. injection as the presence of PEG–lipids can inhibit interaction with target cells.21 The inhibitory effects of a slower dissociation rate of PEG-C-DOMG from the LNP systems can be most easily corrected by starting with a lower PEG–lipid content in the LNP siRNA system. Thus, the in vitro AR silencing properties of AR-siRNA systems containing PEG-C-DOMG were tested at 2.5 mol % as opposed to 10 mol %. As shown in Figure 4a, substantial AR silencing is observed for systems containing 2.5 mol % PEG-C-DOMG. LNP that contained 2.5 mol % PEG-C-DOMG and the most potent cationic lipid, DLin-KC2-DMA, were used in all subsequent LNP formulations.
LNP AR-siRNA induces AR silencing in vitro in LAPC-4 and CWR22Rv1 cell lines
LNCaP cells contain a point-mutated AR within the ligand-binding domain. It is important to examine the potency of LNP AR-siRNA in other prostate cancer cell lines such as LAPC-4 (expressing wild-type AR), and CWR22Rv1 (expressing a variant AR).16, 17, 22 The potency of LNP AR-siRNA systems (40 mol % DLin-KC2-DMA and 2.5 mol % PEG-C-DOMG) for silencing the AR in LNCaP, LAPC-4 and CWR22Rv1 cell lines is shown in Figure 4b. Substantial knockdown is observed in both LNCaP and LAPC-4 cell lines as indicated by the reduced intensity in the band at 110 kDa when using LNP AR-siRNA compared to the LNP SC-siRNA and untreated controls. In the case of the CWR22Rv1 cell line,17 the full-length AR band is located at 114 kDa and significant AR knockdown is also observed for this band in addition to some knockdown at 80 kDa, representing the truncated AR lacking the ligand-binding domain (ARΔLBD) (Fig. 4b).
Intravenous administration of LNP AR-siRNA can reduce serum PSA levels in mice bearing LNCaP tumors
The results summarized to this point indicate that an LNP siRNA system containing DLin-KC2-DMA and 2.5 mol % PEG-C-DOMG exhibits optimized activity in vitro. However, these systems are primarily designed to enable the long circulation lifetimes required to result in accumulation at tumour sites following i.v. injection (diameter < 100 nm and little positive surface charge). It is therefore of interest to determine the effectiveness of this formulation for in vivo AR knockdown. The LNCaP xenograft tumor model generated in athymic nude mice has previously been shown to provide a reproducible in vivo experimental system for monitoring antitumor therapeutics.6, 7, 23–25 Although androgen ablation does not generally result in significant tumor regression in this model, it does result in a drop in serum PSA, which can serve as an index for impacting androgen signaling. Consistent with this, we have previously reported that AR knockdown in these tumors is correlated with the extent that serum PSA is reduced.6, 7
A few weeks after inoculation with LNCaP cells, when serum PSA levels reached 50–75 ng/ml, mice were treated i.v. with 10-mg siRNA/kg of DLin-KC2-DMA LNP AR-siRNA or LNP SC-siRNA on Days 1, 2 and 3 (Fig. 5b) and then again on Days 7, 9 and 11. A second set of injections was made on Days 7, 9 and 11. Serum PSA levels were measured daily to gauge the effectiveness of these treatments.
While the serum PSA continued to increase in both control (PBS) and LNP SC-siRNA-treated mice, mice treated with LNP AR-siRNA showed no increase in serum PSA over levels observed at the initiation of treatment (Day 0) and at least 40% lower than that measured in the controls by Day 7 (p < 0.05) (Fig. 5a). Significantly, the second round of treatment with LNP-AR-siRNA on Days 7, 9 and 11 was sufficient to maintain suppression of serum PSA whereas PSA levels continued to rise in animals treated with either LNP SC-siRNA or the PBS (Fig. 5a). This result suggests that repeated LNP AR-siRNA injections could be used for long-term treatment. No obvious toxic side effects of such treatment were observed. Immunoblotting analyses revealed that protein levels of AR dropped in xenograft tumor tissues collected from mice treated with LNP AR-siRNA, whereas in tumor tissues from control mice treated with LNP SC-siRNA or PBS, the AR levels remained relatively unchanged (Fig. 5b).
AR-siRNA-induced specific cleavage in AR mRNA
To confirm that knockdown of the AR was RNAi mediated, a RLM RACE-PCR was conducted to amplify the mRNA cleaved by AR-siRNA in LNCaP cells treated with free AR-siRNA and in tumor tissues collected from mice treated with LNP AR-siRNA. As shown in Figure 6, there were amplifications of the cleaved mRNA with the predicted size in LNCaP cells treated with free AR-siRNA and also in tumor tissues collected from mice treated with LNP AR-siRNA. Sequencing of cDNAs reverse transcribed from cleaved mRNA confirmed that those cDNAs are authentic to the cDNA encoding AR. Further analysis demonstrated that the cleaved sites were within the sequence recognized by the siRNA against the AR (Supporting Information Fig. S1).
To our knowledge, the results presented here constitute the first demonstration of the feasibility and efficacy of using a LNP siRNA delivery system to knockdown the AR in a human prostate tumor in vivo and thereby significantly inhibit any subsequent increase in serum PSA. Three points of interest concern the reasons why the LNP containing DLin-KC2-DMA are the most active in vitro, comparison to other systems for silencing AR in vivo and ways in which the potency of these LNP siRNA systems may be improved.
LNP containing DLin-KC2-DMA gave rise to the greatest level of gene silencing in human prostate cancer cells (Fig. 2) of the four cationic lipids tested (Fig. 1). To determine whether the superior activity was due to increased uptake of LNPs, uptake experiments were performed in which LNCaP cells were incubated with LNP AR-siRNA containing DLin-KC2-DMA, DLinKDMA, DLinDMA or DLinDAP as well as a fluorescent label at 4, 12, 24 and 48 hr time points (Fig. 3b). As expected on the basis of surface charge, LNP containing DLin-KC2-DMA or DLinDMA exhibited higher levels of uptake (Fig. 3b). DLin-KC2-DMA and DLinDMA were also found to have increased AR knockdown compared to DLinKDMA and DLinDAP when normalized to cell uptake (Fig. 3c). As suggested elsewhere,26 the ability of cationic lipids to facilitate intracellular delivery of nucleic acid polymers can be attributed to an ability to form ion pairs with endogenous anionic lipids in the endosome following uptake. These ion pairs promote endosomolytic nonbilayer hexagonal (HII) phase structures, potentially resulting in the cytoplasmic release of the RNA or DNA. An enhanced ability to disrupt endosomes would be consistent with a greater ability of DLin-KC2-DMA and DLinDMA to induce HII phase structure in the presence of anionic lipids as compared to DLinKDMA or DLinDAP as demonstrated elsewhere.14
Intravenous administration of the LNP AR-siRNA formulation containing DLin-KC2-DMA is effective in knocking down the AR in distal LNCaP xenograft tumors, which is also reflected by reduced serum PSA levels as compared to controls (Fig. 5). This provides a proof-in-principle for developing a therapeutic strategy to treat advanced prostate cancers. Furthermore, our results indicate that AR knockdown and inhibition of serum PSA increases is sustainable with repeat i.v. injections of these DLin-KC2-DMA siRNA formulations (Fig. 5). These results are consistent with previous studies using a LNCaP-derived cell line engineered to express an inducible shRNA targeting the AR, which showed that extensive knockdown of the AR in all the cells inhibits tumor growth and decreases serum PSA levels.6 Moreover, in a castration-resistant (hormone refractory) prostate cancer model, shRNA silencing of the AR was equally effective and even caused regression in 50% of the tumors,7 suggesting that anti-AR therapy could be effective against both androgen-dependent and independent prostate cancers. To achieve a comparable impact on both tumor volume and serum PSA levels may require a substantial increase in the potency of in vivo LNP siRNA delivery systems.
It has been reported27 that free siRNA (daily i.p. injections of 125 μg/kg AR-siRNA) can lower AR expression and inhibit prostate tumor growth. We were unable to repeat these results following their exact protocols (Supporting Information Materials and Methods and Supporting Information Fig. S2). This was not unexpected since in vivo therapy with free siRNA is problematic due to rapid clearance of siRNA from the bloodstream, degradation by serum nucleases, poor distribution to the target tumor tissue and an inability to penetrate target cell membranes.28
While the potential utility of the LNP AR-siRNA systems as agents to treat prostate cancer is supported by the clinical use of lipid-based delivery systems for small molecule applications29 and the early stage clinical development of LNP siRNA systems aimed at hepatocyte targets (www.alnylam.com), improvements in in vivo potency are likely required before such systems can become viable therapies for prostate cancer therapy. That such improvements are possible is indicated by the potency of LNP siRNA systems developed for silencing FVII in hepatocytes, where dose levels of approximately 30-μg siRNA/kg body weight are required to achieve 50% gene silencing.14 This contrasts with the 10-mg siRNA/kg body weight doses employed here. The remarkable potency of the hepatocyte LNP siRNA system is a consequence of the targeting effects of ApoE, which becomes associated with the LNP following i.v. administration30 and leads to uptake into hepatocytes via the LDL receptor, the scavenging receptor and the “LDL-like” receptor.31 Likewise, other in vivo studies have revealed that rHDL nanoparticles incorporated with siRNA are readily taken up into ovarian and colorectal cancer cells via the scavenger receptor type B1 (SR-B1) and are very effective at silencing cancer promoting genes at dose levels of 0.2 mg/kg.32 This clearly suggests that strategies to improve targeting to and uptake of LNP AR-siRNA systems into prostate cancer cells in vivo should lead to more efficient AR knockdown and potentially better therapeutic control of advanced prostate cancer.
We thank Fariba Ghaidi, Jessica Firus, Anthony Chan (Vancouver Prostate Centre) and Sam Chen (University of British Columbia) for technical assistance. P.R.C. has a financial interest in Tekmira and received grant support from Alnylam.