Targeted RNAi of BIRC5/Survivin Using Antibody-Conjugated Poly(Propylene Imine)-Based Polyplexes Inhibits Growth of PSCA-Positive Tumors

Delivery of siRNAs for the treatment of tumors critically depends on the development of efficient nucleic acid carrier systems. The complexation of dendritic polymers (dendrimers) results in nanoparticles, called dendriplexes, that protect siRNA from degradation and mediate non-specific cellular uptake of siRNA. However, large siRNA doses are required for in vivo use due to accumulation of the nanoparticles in sinks such as the lung, liver, and spleen. This suggests the exploration of targeted nanoparticles for enhancing tumor cell specificity and achieving higher siRNA levels in tumors. In this work, we report on the targeted delivery of a therapeutic siRNA specific for BIRC5/Survivin in vitro and in vivo to tumor cells expressing the surface marker prostate stem cell antigen (PSCA). For this, polyplexes consisting of single-chain antibody fragments specific for PSCA conjugated to siRNA/maltose-modified poly(propylene imine) dendriplexes were used. These polyplexes were endocytosed by PSCA-positive 293TPSCA/ffLuc and PC3PSCA cells and caused knockdown of reporter gene firefly luciferase and Survivin expression, respectively. In a therapeutic study in PC3PSCA xenograft-bearing mice, significant anti-tumor effects were observed upon systemic administration of the targeted polyplexes. This indicates superior anti-tumor efficacy when employing targeted delivery of Survivin-specific siRNA, based on the additive effects of siRNA-mediated Survivin knockdown in combination with scFv-mediated PSCA inhibition.


Introduction
Small interfering RNAs (siRNAs) can selectively target and downregulate tumorrelevant genes, holding great promise in cancer treatment. The siRNA-mediated silencing (RNAi) of genes utilizes a conserved biological process among multicellular organisms that mediates resistance to both endogenous parasitic and exogenous pathogenic nucleic acids, and regulates the expression of protein-coding genes. It relies on 21 bp to 23 bp doublestranded, so-called small interfering RNAs (siRNAs), whose guide strand is integrated into the RNA-induced silencing complex (RISC), which, upon sequence-specific binding, atic adenocarcinoma [21], renal clear-cell carcinoma [22], transitional cell carcinoma [23], and glioblastoma [24]. Recently, we demonstrated successful delivery of TLR3 agonist Riboxxol-biotin conjugated to neutravidin/anti-PSCA exclusively to PSCA-positive target cells in vitro and in vivo [25]. We therefore considered PSCA as a relevant tumor-associated antigen which is a targetable structure for employing our polyplex approach for delivery of therapeutic siRNA.
In this study, we therefore investigate a novel polyplex system based on the combination of targeted delivery of therapeutic siRNA, siRNA-mediated Survivin knockdown and scFv-mediated PSCA inhibition. To specifically distinguish the extent to which each component contributes to the targeted siSurv delivery mediated by the PSCA-specific polyplexes, the in vivo anti-tumor effects were directly compared to those obtained with PSCA-specific polyplexes loaded with control siRNA (scFv-mediated PSCA inhibition only) and polyplexes loaded with siSurv but conjugated with a non-specific control antibody (Survivin knockdown only) [26]. Indeed, we found that the anti-tumor effects of the PSCA-targeted, siSurv-containing polyplexes significantly exceeded the anti-tumor effects of the single-component systems. Thus, the mPPI-based PSCA-targeted polyplexes provide an innovative platform for further exploration of targeted delivery of siRNAs and other nucleic acids for therapeutic application.

Production of scFv-P-BAP and Polyplexes
The biotinylated scFv(AM1)-P-BAP was expressed in transiently transfected 293T huBirA producer cells. The biotinylated scFv(MR1.1)-P-BAP was produced in 293T huBirA-scFv(MR1.1)-P-BAP cells. The single-chain antibodies were purified from the harvested cell culture supernatants using Ni 2+ -NTA affinity chromatography. Briefly, 50 mL of clarified supernatant was passed through a Ni 2+ -NTA spin column (Qiagen, Hilden, Germany) and washed with 1X PBS containing 150 mM NaCl and 10 mM/20 mM imidazole. Elution of bound scFvs was performed in 500 µL 1X PBS containing 150 mM NaCl and 350 mM imidazole per column. Eluted scFvs were dialyzed in 1x PBS twice for 2 h and additionally for 12 h at 4 • C. The biotinylated scFv-P-BAPs were further purified using avidin-biotin affinity chromatography with monomeric avidin columns (Thermo Fisher Scientific, Waltham, MA, USA) according to the protocol of the manufacturer. Eluted scFvs were dialyzed again as described previously. The recombinant proteins were subsequently concentrated using Amicon tubes Ultra-15 (Merck Millipore, Burlington, MA, USA) and were stored in aliquots at −20 • C. Polyplexes were generated as described previously [10]. Briefly, polyplexes were built by sequential mixing of neutravidin, mono-biotinylated scFv-P-BAP and mono-biotinylated mal19-PPI-biotin using molar ratios 1:2:1. Saturation of remaining free biotin binding sites of neutravidin was accomplished with 0.3 mM D-biotin. Complexation between siRNA and mal19-PPI was achieved using a molar ratio 1:4. The electrostatic interactions between mal19-PPI-biotin/neutravidin/scFv-P-BAP and mal19-PPI/siRNA intermediate conjugates resulted in scFv(AM1)-P-BAP-guided polyplexes. After 24 h incubation at 4 • C, the generated polyplexes were used for the experiments.

Multiparameter Nanoparticle Tracking Analysis
Multiparameter nanoparticle tracking analysis was performed using the ZetaView ®PMX120 (Particle Metrix GmbH, Inning am Ammersee, Germany) according to the manufacturer's instructions to determine the zeta potential and the particle size of scFv(AM1)-P-BAP-polyplexes containing a control siRNA (siLuc). Polyplexes were prepared as described above. Data were analyzed using the manufacturer's software (ZetaView 8.05.05).

RNA Interference and Western Blot Analysis
Produced scFv(AM1)-P-BAP and its conjugation to neutravidin was investigated using SDS-PAGE. For immunoblot analysis, 1 µg of scFv(AM1)-P-BAP was separated by SDS-PAGE (12% polyacrylamide gel) under reducing conditions. Proteins were transferred by semi-dry Western Blot to a PVDF membrane (Whatman). After blocking of the PVDF membrane with 5% non-fat dry milk in Tris-buffered saline containing 0.1% Tween 20 (TBS-T), the scFv-P-BAPs were detected using a primary monoclonal murine anti-c-myc antibody (1:5000, Invitrogen) and a secondary polyclonal rabbit anti-mouse IgG HRP conjugate (1:1000; Dako). Biotinylated scFv(AM1)-P-BAPs were detected by HRP-conjugated antibiotin antibody (1:3000, Sigma-Aldrich). For determination of conjugated scFv(AM1)-P-BAP to neutravidin, the two components were mixed at various molar ratios of scFv-P-BAPs to neutravidin from 8:1 to 0.25:1 in 1x PBS for 30 min at room temperature. The polyplex-mediated knockdown of Survivin in PC3 PSCA cells was investigated using SDS-PAGE. Therefore, PC3 PSCA cells were consecutively transfected for three times every 8 h with scFv(AM1)-P-BAP polyplexes containing siSurv. Polyplexes containing a control siRNA (siLuc) were included as control. Interferin-mediated transfection of siSurv into PC3 PSCA cells was included to demonstrate the feasibility of siSurv for RNAi. Treated cells were lysed in lysis buffer (10 mM Tris-HCl, pH 8.0; 140 mM NaCl; 1% Triton X-100) 72 h after transfection. Cell lysates were cleared by centrifugation. Equal amounts of protein were subjected to SDS-PAGE under reducing conditions and blotted on PVDF membrane using semi-dry Western Blotting. After blocking PVDF membrane with 5% BSA in TBS-T, Survivin was detected using a polyclonal rabbit anti-human Survivin antibody (1:1000, R&D Systems, Minneapolis, MN, USA), followed by HRP-conjugated anti-rabbit IgG secondary antibody (1:1000; Dako). To demonstrate equal loading, PVDF membranes were subsequently stained using an anti-α tubulin antibody (1:5000; Sigma-Aldrich), followed by a secondary polyclonal rabbit anti-mouse IgG HRP conjugate (1:1000; Dako, Glostrup, Denmark). Visual capturing of proteins was performed by Luminata Forte Western HRP substrate (Merck Millipore) and G:Box Chemi XX9 (VWR) gel doc system and analyzed by Fiji software (ImageJ 1.51k, National Institute of Health).

Confocal Laser Scanning Microscopy
For visualization of cellular siRNA uptake, 6 × 10 5 293T PSCA/ffLuc cells grown on a coverslip were incubated with scFv(AM1)-P-BAP-polyplexes loaded with Cy3-labelled siRNA for 24 h at 37 • C. After fixation of cells with 4% paraformaldehyde in PBS, cell membranes and nuclei were stained with Alexa Fluor 647 conjugated Wheat Germ Agglutinin (WGA, Life Technologies, Carlsbad, CA, USA) and Hoechst 33,342 (Invitrogen, Carlsbad, CA, USA) according to the manufacturers protocols. The coverslips were placed upside down in a drop of mounting medium (Vector Laboratories) on a microscope slide. Cy3-labelled polyplexes conjugated with control scFv(MR1.1)-P-BAP, which cannot bind to 293T PSCA/ffLuc cells or without conjugation of scFv-P-BAP were included as negative controls. The images were captured by a confocal laser scanning microscope (Leica SP5, Leica, Wetzlar, Germany) and analyzed by Fiji software (ImageJ 1.51k, National Institute of Health).

Delivery of PSCA-Specific Polyplexes to Mice Bearing Human Prostate Cancer Xenografts
The animal experiments were approved by the Landesdirektion Sachsen (animal permission number: TVV 21/13). The methods were carried out in accordance with the approved guidelines and regulations. 2 × 10 6 PC3 PSCA cells were subcutaneously implanted into the right and left flank of athymic nude mice (NMRI Foxn1nu/Foxn1nu ). After reaching an average tumor diameter of 2.5 mm, mice were injected intraperitoneally with PSCA-specific polyplexes containing 10 µg siSurv, 80.8 µg mal19-PPI, 15.6 µg mal19-bPPI, 43.8 µg neutravidin, 35.9 µg scFv(AM1)-P-BAP and 1.8 µg D-biotin every three days for 17 days. Mice injected with PSCA-specific polyplexes loaded with negative control siRNA specific for luciferase (siLuc), with EGFRvIII-specific polyplexes containing siSurv or with EGFRvIII-specific polyplexes containing siLuc were included as negative controls. Tumor sizes were determined by measuring all three dimensions.

Statistical Analysis
All experiments were performed at least three times. Differences between groups were examined for statistical significance using Student's t-test. Values of p < 0.05 were considered as statistically significant: * p < 0.05, ** p < 0.01, and *** p < 0.001.

Synthesis and Characterization of scFv(AM1)-P-BAP Molecules
Here, we explored a biotin-neutravidin-based, so-called polyplex modular system for anti-PSCA scFv antibody-mediated delivery of poly(propylene imine) carrier molecules containing Survivin-specific siRNA to PSCA-positive tumor cells. For this, we established a PSCA-specific scFv(AM1), fused to a biotin acceptor peptide sequence (P-BAP), and an similarly structured EGFRvIII-specific scFv(MR1.1)-P-BAP as negative control [10]. The structures of the scFv(AM1)-P-BAP and scFv(MR1.1)-P-BAP are shown in Figure 1A. All antibody constructs contained an N-terminal Igκ chain leader sequence for the extracellular secretion as well as a C-terminal c-myc-epitope and a 6x histidine (His)-tag for detection and purification, respectively. The biotinylated scFv-P-BAPs were purified from clarified cell culture supernatant of antibody-expressing 293T huBirA cells using Ni 2+ -NTA affinity chromatography, followed by biotin affinity chromatography to obtain only biotinylated scFv-P-BAPs. In addition to the detection of the c-myc-epitope in Western Blot analysis, a biotin-specific antibody demonstrated the successful biotinylation of scFv(AM1)-P-BAP and the control antibody scFv(MR1.1)-P-BAP ( Figure 1B). The observed bands match the calculated molecular mass of roughly 55 kDa. The Coomassie Brilliant Blue stained polyacrylamide gel shown in Figure 1C confirmed the purity and successful expression of the full-length proteins ( Figure 1C). The binding properties of the scFv(AM1)-P-BAP were characterized by flow cytometry ( Figure 1D). Results clearly demonstrated highly efficient binding of the scFv(AM1)-P-BAP to 293T PSCA and PC3 PCSA cells, while as expected, the PSCA-specific single-chain antibody failed to bind to 293T and PC3 wild-type cells. Since a biotin-specific antibody was used as a secondary antibody, we were able to show that the C-terminal biotin residue was accessible under native conditions. As expected, the EGFRvIII-specific control antibody did not bind PSCA-positive 293T cells or PSCA-positive PC3 cells ( Figure 1D).

Targeted Delivery of siRNA in PSCA-Positive Tumor Cells
For characterization of targeted siRNA delivery to tumor cells, 293T PSCA/ffLuc were treated with scFv(AM1)-P-BAP-guided polyplexes. As negative controls, non-specific polyplexes containing the EGFRvIII-specific scFv(MR1.1)-P-BAP or dendriplexes without scFv-P-BAP were included. All poly-and dendriplexes contained a Cy3-labelled siRNA (Cy3-siLuc) in order to allow detection of siRNA delivery into cells. Flow cytometry analysis, shown in Figure 3A, demonstrated a scFv(AM1)-P-BAP-mediated internalization of the polyplexes by 293T PSCA/ffLuc cells, whereas no Cy3 signals were detectable after treatment with non-specific dendriplexes devoid of scFv-P-BAP or polyplexes conjugated with scFv(MR1.1)-P-BAP, respectively. This indicates that the siRNA delivery is enabled by receptor-mediated endocytosis. The right histogram of Figure 3A shows a moderate unspecific internalization of dendriplexes (devoid of scFv-P-BAP) which is most likely mediated by electrostatic interaction between polyplexes and the cell membrane. Confocal laser scanning microscopy studies, shown in Figure 3B, supported the data obtained from flow cytometry analysis. The tumor-specific scFv(AM1)-P-BAP-Cy3-siLucpolyplexes were found to be internalized by 293T PSCA/ffLuc cells. Well recognizable, the Cy3-labelled siRNA was located intracellularly in the cytoplasm, but not on the outer membrane, thus confirming cell internalization. In contrast, the scFv(MR1.1)-P-BAP-guided polyplexes as well as the scFv-P-BAP-free dendriplexes show essentially no nanoparticle uptake. The endosomal release and knockdown potential of polyplex-mediated luciferasespecific siRNA in 293T PSCA/ffLuc cells was determined by a luciferase assay ( Figure 3C). Of note, PSCA-positive target cells treated with scFv(AM1)-P-BAP-guided siLuc-containing polyplexes revealed a significant 39% decrease in luciferase activity, while luciferase knockdown of control polyplexes conjugated to non-binding scFv(MR1.1)-BAP was less than 3%. The low knockdown activities of the non-specific control polyplexes thus coincided with the data from flow cytometry and confocal laser scanning microscopy analysis shown above. To analyze the receptor mediated uptake of scFv(AM1)-P-BAP-guided polyplexes in more detail, the same experiments were conducted in the presence of inhibitors of endocytosis. In general, endocytosis of human cells can be differentiated between clathrindependent and clathrin-independent endocytosis. Clathrin-dependent endocytosis, which can be blocked by chlorpromazine [30], is characterized by the formation of clathrincoated vesicles from clathrin-coated pits, which then enter the endosomal pathway [31]. In contrast, clathrin-independent endocytosis involves caveolae-dependent and lipid raftmediated mechanisms that are inhibited by filipin III [32]. Clearly, the filipin III treatment of 293T PSCA/ffLuc cells during incubation with scFv(AM1)-P-BAP-guided polyplexes abolished RNAi-mediated luciferase knockdown, while endocytosis and gene knockdown were not inhibited by treatment with chlorpromazine. This identifies clathrin-independent endocytosis as the uptake mechanism.

PSCA-Specific Polyplexes Delivering Survivin siRNA Exert Anti-Tumor Effects in PC3 PSCA Xenograft-Bearing Mice
For the exploration of our system in a more relevant therapeutic in vivo situation, we investigated the efficacy of targeted RNAi-mediated Survivin knockdown in PC3 PSCA xenografts in NMRI Fox1nu/Fox1nu mice. The efficient knockdown of Survivin expression upon successful delivery of siSurv into PC3 PSCA cells by PSCA-specific polyplexes was first proven in vitro, with Survivin knockdown almost reaching efficiencies of interferin as classical reagent for in vitro siRNA transfection (Supplementary Figure S2). This also demonstrates the efficacy of the selected siRNA. For in vivo therapy, the scFv(AM1)-P-BAPguided polyplexes containing the Survivin-specific siRNA (siSurv) were injected intraperitoneally (i.p.) every third day for 17 days into PC3 PSCA tumor-bearing mice ( Figure 4A). Remarkably, this treatment with siSurv-loaded scFv(AM1)-P-BAP-guided polyplexes led to a profound,~50% inhibition of subcutaneous tumor growth when compared to the doublenegative control group, i.e., mice treated in the same way with polyplexes containing the non-binding scFv(MR1.1)-P-BAP and negative control siLuc ( Figure 4B). Notably, however, anti-tumor effects were also observed in mice treated with polyplexes containing the PSCA targeting antibody scFv(AM1)-P-BAP plus negative control siRNA (siLuc) or treated with non-targeted, scFv(MR1.1)-P-BAP-guided polyplexes loaded with siSurv. While these therapeutic effects reached statistical significance as well, they were less profound compared to the double-specific treatment group. To exclude anti-tumor activity of polyplexes due to siRNA-mediated activation of Toll-like receptor 3 (TLR3) [33,34], PSCA-engineered HEK-Blue hTLR3/PSCA reporter cells were treated with polyplexes loaded with siSurv. As shown in Supplementary Figure S3, the potent TLR3 ligand poly(I:C) induced SEAP expression in the reporter cells, whereas internalized scFv(AM1)-P-BAP polyplexes and the non-internalized scFv(MR.1.1)-P-BAP-polyplexes, both loaded with siSurv, did not induce an innate cellular immune response. Thus, it can be concluded that the siRNAs and their complexation with mPPI did not interfere with TLR3 signaling and that the moderate anti-tumor effects observed after systemic administration of polyplexes containing the control antibody in PC3 PSCA tumor-bearing mice are based on the non-targeted delivery of siSurv. Likewise, the results of this study demonstrate moderate anti-tumor activity of PSCA antibodies, which is in line with previous reports [18] and contributes to the anti-tumor effects of our system.

Discussion
Since the discovery of RNAi as a powerful method for gene silencing [35], RN fector molecules (e.g., synthetic siRNA) have been applied in numerous clinical trial which has more recently led to the approval of the first siRNA therapeutics (Pati Givosiran). However, before RNAi-based therapeutic approaches can be applied, si cant hurdles still need to be overcome, such as the development of efficient deliver tems. Recently, we established a novel tumor-specific siRNA delivery system, cons of maltose-modified-PPI/PPI-biotin, neutravidin and a biotinylated scFv for tar siRNA delivery to tumor cells [10]. Here, we investigated this modular platform tec ogy with regard to the combination of a therapeutic siRNA for knockdown of a rel target (Survivin) with a PSCA-specific scFv that simultaneously functions as a targ moiety and PSCA inhibitor. PSCA was chosen since it is found on a broad range of t entities [20][21][22][23]. Recent studies from our group have established PSCA as a targetab tigen, which after receptor crosslinking by anti-PSCA immunoconjugates induces cytosis [25]. In general, two main pathways for receptor-mediated endocytosis can b tinguished: clathrin-dependent and caveolae-dependent and lipid raft-mediated en tosis. In this study, we identify caveolae/lipid raft-mediated endocytosis, but not cla mediated endocytosis as the mechanism of PSCA-specific polyplex internalization. findings are in line with previous studies, that demonstrated involvement of cave dependent endocytosis in the internalization of other GPI-anchored proteins  Yet, these results are in contrast to our recent report which revealed a mixed clathrin

Discussion
Since the discovery of RNAi as a powerful method for gene silencing [35], RNAi effector molecules (e.g., synthetic siRNA) have been applied in numerous clinical trials [36], which has more recently led to the approval of the first siRNA therapeutics (Patisiran, Givosiran). However, before RNAi-based therapeutic approaches can be applied, significant hurdles still need to be overcome, such as the development of efficient delivery systems. Recently, we established a novel tumor-specific siRNA delivery system, consisting of maltose-modified-PPI/PPI-biotin, neutravidin and a biotinylated scFv for targeted siRNA delivery to tumor cells [10]. Here, we investigated this modular platform technology with regard to the combination of a therapeutic siRNA for knockdown of a relevant target (Survivin) with a PSCA-specific scFv that simultaneously functions as a targeting moiety and PSCA inhibitor. PSCA was chosen since it is found on a broad range of tumor entities [20][21][22][23]. Recent studies from our group have established PSCA as a targetable antigen, which after receptor crosslinking by anti-PSCA immunoconjugates induces endocytosis [25]. In general, two main pathways for receptor-mediated endocytosis can be distinguished: clathrin-dependent and caveolae-dependent and lipid raft-mediated endocytosis. In this study, we identify caveolae/lipid raft-mediated endocytosis, but not clathrin-mediated endocytosis as the mechanism of PSCA-specific polyplex internalization. These findings are in line with previous studies, that demonstrated involvement of caveolae-dependent endocytosis in the internalization of other GPI-anchored proteins [16][17][18][19]37]. Yet, these results are in contrast to our recent report which revealed a mixed clathrin-and caveolaemediated uptake of nanoparticle-like immunoconjugates comprising mono-biotinylated anti-PSCA-scFv conjugated via neutravidin to mono-biotinylated TLR3 agonist [25]. This discrepancy may be due to differences in nanoparticle sizes, since the nanoparticle-like immunoconjugates were estimated to be in the range of 22-55 nm in size, whereas the particle sizes of polyplexes were determined at 84-220 nm [23]. It remains to be investigated how the size of nanoparticles will indeed affect the mode of receptor-mediated internalization.
It is unanimously recognized that endosomal release of siRNA into cytoplasm is one of the key hurdles for most siRNA and nucleic acid delivery systems, often limiting their therapeutic effects. Notably, we demonstrated RNAi-mediated luciferase knockdown in 293T PSCA/ffLuc reporter cells after targeted delivery of siLuc using PSCA-specific polyplexes, indicating efficient endosomal escape of siRNA molecules. Concomitantly, PSCA-specific polyplexes delivering siSurv showed profound downregulation of Survivin in PC3 PSCA prostate cancer cells in vitro. In line with this, our in vivo administration of PSCA-specific polyplexes in human PC3 PSCA xenograft-bearing mice led to a significant inhibition of tumor growth. Of note, PSCA-targeted polyplexes containing siLuc control siRNA resulted in moderate anti-tumor effects as well. This is in line with previous studies reporting that anti-PSCA monoclonal antibodies derived from hybridoma 1G8 significantly inhibited the growth of established orthotopic tumors and prolonged the survival of mice bearing human prostate cancer xenografts [18,38]. Although our scFv was derived from a different hybridoma (7F5) [39], it is conceivable that the scFv(AM-1)-P-BAP antibody derivative when embedded in a polyplex might induce some inhibitory effects on PSCA-expressing cells. Yet, treatment of PC3 PSCA cells with PSCA-specific polyplexes loaded with siLuc had no effects on apoptosis or clonal survival of PC3 PSCA cells in vitro (Supplementary Figures S4 and S5). This also indicates that classical 2D cell culture systems may only insufficiently monitor tumor-inhibitory effects of anti-PSCA antibodies, which can also rely on the modulation of other processes critical for tumor growth like tumor cell differentiation [40], modulation of angiogenesis factors [41] or disruption of cell-cell or cell-matrix interaction [42]. Moderate anti-tumor effects were also observed after systemic administration of control polyplexes loaded with Survivin-siRNA. This may well be due to the non-targeted delivery of the siRNA, probably facilitated by the enhanced permeability and retention (EPR) effect [43], that is often observed in tumors and can support the accumulation of nanoparticles in tumor tissue.

Conclusions
In summary, we successfully validated a modular platform technology for the targeted delivery of therapeutic siRNA to tumor cells. In succeeding experiments, it is mandatory to investigate potential beneficial effects but also limitations of tumor targeting polyplexes to pave the way for translation into the clinics. In particular, studies using syngeneic tumors models are envisioned to investigate potential immune responses due to facilitated release of tumor antigens after treatment with polyplexes delivering therapeutic siRNAs.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author.