Technologies that mediate targeted delivery of small interfering RNAs (siRNAs) are needed to improve their therapeutic efficacy and safety. Therefore, we have developed aptamer-siRNA chimeric RNAs capable of cell type-specific binding and delivery of functional siRNAs into cells. The aptamer portion of the chimeras mediates binding to PSMA, a cell-surface receptor overexpressed in prostate cancer cells and tumor vascular endothelium, whereas the siRNA portion targets the expression of survival

more »

... enes. When applied to cells expressing PSMA, these RNAs are internalized and processed by Dicer, resulting in depletion of the siRNA target proteins and cell death. In contrast, the chimeras do not bind to or function in cells that do not express PSMA. These reagents also specifically inhibit tumor growth and mediate tumor regression in a xenograft model of prostate cancer. These studies demonstrate an approach for targeted delivery of siRNAs with numerous potential applications, including cancer therapeutics. First described in Caenorhabditis elegans, RNA interference (RNAi) is a cellular mechanism by which 21-to 23-nucleotide RNA duplexes trigger the degradation of cognate mRNAs 1 . The promise of RNAibased therapeutic applications has been apparent since the demonstration that exogenous small interfering RNAs (siRNAs) can silence gene expression via the RNAi pathway in mammalian cells 2 . The properties of RNAi that are attractive for therapeutics include (i) stringent target-gene specificity, (ii) relatively low immunogenicity of siRNAs and (iii) simplicity of design and testing of siRNAs. A critical technical hurdle for RNAi-based clinical applications is the delivery of siRNAs across the plasma membrane of cells in vivo. A number of solutions for this problem have been described, including cationic lipids 3 , viral vectors 4-6 , high-pressure injection 7 and modifications of the siRNAs (e.g., chemical, lipid, steroid, protein) 8-13 . However, most of the approaches described to date have the disadvantage of delivering siRNAs to cells nonspecifically, without regard to the cell type. For in vivo use, one would like to target therapeutic siRNA reagents to particular cell types (e.g., cancer cells), thereby limiting side effects that result from nonspecific delivery as well as reducing the quantity of siRNA necessary for treatment, an important cost consideration. One recent study described a promising approach in which antibodies that bind cell type-specific cell-surface receptors were fused to protamine and used to deliver siRNAs to cells via endocytosis 14 . Similarly, another study described a method to specifically deliver anti-ews-fli1 siRNAs to transferrin receptor-expressing tumors in mice using a cyclodextrin-containing polycation bearing transferrin as a targeting ligand 15 . Here, we describe a completely RNA-based approach for specific delivery of siRNAs. A number of groups have identified structured RNAs capable of binding a variety of proteins with high affinity and specificity with SELEX (systematic evolution of ligands by exponential enrichment). We decided to exploit the structural potential of RNA to target siRNAs to a cell-surface receptor specific to a particular cell type. Our approach relies on RNAs that include both a targeting moiety, the aptamer, and an RNA-silencing moiety, the siRNA, which is recognized and processed by Dicer in a manner similar to the processing of microRNAs (Fig. 1a) . We generated and tested aptamer-siRNA chimeric RNAs for their ability to (i) specifically bind prostate cancer cells expressing the cellsurface receptor PSMA using an RNA aptamer previously selected against human PSMA (A10) 16 and (ii) deliver therapeutic siRNAs that target polo-like kinase 1 (PLK1) 17 and BCL2 (ref. 3), two survival genes overexpressed in most human tumors 18-20 . Because Dicer acts upon the chimeric RNAs, the siRNAs are directed into the RNAi pathway and silence their cognate mRNAs (Fig. 1a) . In contrast to most described delivery methods, this approach involves only RNA (that is, an RNA aptamer linked to an siRNA), an important advantage given the various side effects associated with many commonly used reagents such as proteins. Indeed, aptamer-siRNA chimeras present several advantages for in vivo applications. Aptamers and siRNAs have low immunogenicity. They can easily be synthesized in large quantities at a relatively low cost and are amenable to a variety of chemical modifications that confer both resistance to degradation and improved pharmacokinetics in vivo. The smaller size of aptamers compared with that of antibodies (o15 kDa versus