|Home | About | Journals | Submit | Contact Us | Français|
The ability of small-interfering RNA (siRNA) to silence specific target genes not only offers a tool to study gene function but also represents a novel approach for the treatment of various human diseases. Its clinical use, however, has been severely hampered by the lack of delivery of these molecules to target cell populations in vivo due to their instability, inefficient cell entry, and poor pharmacokinetic profile. Various delivery vectors including liposomes, polymers, and nanoparticles have thus been developed in order to circumvent these problems. This review presents a comprehensive overview of the barriers and recent progress for both local and systemic delivery of therapeutic siRNA using lipidic vectors. Different strategies for formulating these siRNA-loaded lipid particles as well as the general concern about their safe use in vivo will also be discussed. Finally, current advances in the targeted delivery of siRNA and their impacts on the field of RNA interference (RNAi)-based therapy will be presented.
The discovery of RNA interference (RNAi) by Fire and Mello in late 1990s opened up an entirely new field of “gene” therapy. Previously, gene therapy had mainly concentrated on the concept of introducing new genes into cells to correct genetic defects and was mired by various technical issues, a lack of efficacy, and the vexing issue of unwelcome integration-induced changes in host gene expression that had, in rare instances, resulted in cancer (1). RNAi, in the form of dsRNA called short-interfering RNA (siRNA), provides a fresh approach to the field via the ability to turn off single target genes without genomic integration, thus avoiding some of the issues of gene therapy and offers more promising outcomes in an area pioneered by antisense RNA some 10 years earlier. It has been established that the introduction of siRNA into cells can efficiently trigger a naturally occurring gene silencing mechanism, thereby permitting its use as a pharmacological agent. These dsRNAs are generally 21–27 bp in length and work by binding to the mRNA of target genes via base-pair interactions, thereby ensuring their high degree of target specificity and subsequently inhibit gene expression by destruction of target mRNAs. This whole process is extremely efficient due to the fact that the siRNAs themselves are not destroyed by the process and can be used again and again (their major advantage over antisense technology).
To date, numerous siRNA targets have been identified in various disease models ranging from cancer (2) to infectious (3) or neurodegenerative diseases (4). The major issue confronting the therapeutic use of these siRNAs, however, is the inefficiency of delivering these molecules to target cell populations in vivo. This is due to the instability of these molecules as well as their poor cellular uptake and pharmacokinetic profiles in vivo (5). Much effort has, therefore, been devoted to the development of suitable in vivo siRNA delivery systems. Among these, the lipidic delivery vectors show great promise due to their favorable characteristics, such as biocompatibility and the ease of large-scale production. Their use in gene therapy is currently under investigation in several clinical trials for the treatment of diseases such as cancer and cystic fibrosis (6–10). This review presents an overview of the barriers and recent progress for both local and systemic siRNA delivery using lipidic vectors. Different strategies for formulating siRNA-loaded lipid particles as well as the general concern about their safe use in vivo will also be discussed in detail. Finally, the current advances in the targeted delivery of siRNA will be presented.
Local delivery of siRNA is ideal for diseases where the target sites are easily accessible, such as skin or mucosal surfaces. It has the advantage of circumventing any potential side effects resulting from systemic administration and avoids first-pass hepatic clearance making it more likely that the therapeutic concentration is reached at the target site. To date, local application of siRNA has been widely investigated in diseases such as age-related macular degeneration (AMD) and respiratory virus infections. Local delivery may also be applied to cancers where the tumors are easily accessible and successful intratumoral delivery of siRNA has been reported (11–15). In general, local delivery can be categorised into five main groups: mucosal (intranasal, intratracheal, intravaginal and intrarectal), intraocular, transdermal, intrathecal and intratumoral. Some examples of these applications are summarised in Table I.
One surprising finding has been that it is not always necessary to actually use a delivery vector for local delivery, although this is highly dependent on the target site (16–19). For example, several phase I and II clinical trials have investigated the intraocular delivery of naked siRNA for the treatment of wet AMD (reviewed in (20)). The target gene was the vascular endothelial growth factor (VEGF), of which the overexpression is well established as the basis of this disease. While results have been positive, it must be noted that some of these siRNAs have recently been shown to work in an unexpected way. Rather than specifically silencing the VEGF gene, these naked siRNAs function to reduce VEGF expression via activation of an innate immune response by binding to toll-like receptor 3, an effect that does not require cellular uptake (21). However, others have shown that local delivery of naked siRNA does work. Bitko and colleagues have demonstrated specific gene silencing using naked siRNA by intranasal delivery in the absence of any interferon response (22). Interestingly, they compared delivery of siRNA with and without a transfection reagent, in this case TransIT TKOTM, and noted only a marginal enhancement (20%) in the knockdown of the respiratory syncytial virus (RSV) target gene when lipid agent was used. It should be noted that the mechanism by which the cells take up these siRNA molecules remains unknown.
In contrast to those studies, Zhang and colleagues have reported inefficient uptake of naked siRNA into vaginal tissues after intravaginal administration and that the delivery efficiency can be dramatically improved with the use of LipofectamineTM (23). This was likely due to the rapid degradation and the inefficient mucosal uptake of naked siRNAs in the vaginal cavity. The difference between this study and those described above likely reflects the difference in the physical and biological environment between different application sites. Numerous other studies have also reported the benefit of using cationic lipidic vectors in the local delivery of siRNA for the treatment of diseases such as respiratory virus infections, cancer, or inflammation disorders (see Table I).
While these cationic lipidic systems can facilitate siRNA delivery due to their efficient interaction with cell membranes and nucleic acids, concerns regarding their safety use in vivo have been raised by a recent study performed by Wu and colleagues (24). In that study, it was shown that inflammation occurred in vaginal tissues following intravaginal administration of OligofectamineTM, a cationic lipidic transfecting reagent. Though the observation was likely to be due to the high concentration of OligofectamineTM used, it has been previously established that cationic lipids may provoke an inflammation response more than neutral ones (25). The level of this nonspecific effect is dependent on the dosage and type of lipids used, as well as nitrogen/phosphate (N/P) ratio employed (reviewed in (26) and (27)).
Despite the favorable biocompatibility profiles of neutral lipids, their use in gene delivery is generally limited by the lack of interaction with anionic nucleic acids. To overcome this issue, Soutschek et al. conjugated cholesterol directly to the sense-strand of siRNA duplexes (28). This strategy has been shown to significantly enhance their delivery in vivo while preserving the antisense activity of these molecules (28,29). Using the cholesterol-conjugated siRNA targeting herpes simplex virus (HSV), Wu and colleagues reported efficient silencing of the HSV-related genes after intravaginal administration without provoking inflammation or interferon response at the administrative site (24). The high dose of cholesterol-siRNA (2 nmol in 12 μL) required in that study, however, is likely to limit their therapeutic use. Thus, the recent development of more biocompatible cationic lipids such as cholesterol-based polyamine lipid N1-cholesteryloxycarbonyl-3,7-diazanonane-1,9-diamine (CDAN) (30), 3β[l-ornithinamide-carbamoyl] cholesterol (O-Chol) (31), or carbamate-linked polyamine cholesterol derivatives (32) may play a significant role in the future application of siRNA for treatment of local diseases.
Systemic administration is a feasible means to deliver siRNA molecules for the treatment of diseases such as cancer or metabolic disorders where the target sites are not easily accessible. This can be achieved via intravenous, intraperitoneal, or subcutaneous injections. Of these, intravenous administration is the most widely investigated delivery route to date, owing to the simplicity of the procedure as well as the fast distribution of particles to various tissue sites. The intraperitoneal route of administration has also been studied for treatment of diseases such as sepsis (33), Ebola virus infection (34), and cancer (35–37), although its clinical acceptability for repeated administrations may be limited due of the risk of infections from the catheter implant (38).
Although a few reports have demonstrated successful delivery of siRNA to various tissue sites after intravenous injection of naked siRNA (39–41), the use of a suitable delivery system can significantly improve its efficacy in vivo and thus has been an area of intense research in recent years. Despite the successes of cationic lipid vectors in delivering siRNA for local applications, the formation of aggregates resulting from the undesired interaction between these vectors and anionic serum proteins generally precludes their use in systemic delivery. These aggregates also often accumulate in first-pass organs such as lungs or livers, which severely hinder their delivery to other tissues (2,33,42). Strategies have, therefore, been developed to circumvent this problem. These include the use of polyethylene glycol (PEG) to shield the positive charge on the particle surface as well as the use of neutral lipids to deliver these siRNA molecules systemically. Due to the limited electrostatic interactions between these vectors and anionic siRNAs, however, the formulation of such systems often requires more sophisticated techniques. Examples of these formulation procedures are summarized in Fig. 1 and Table II. Factors such as the complexity of the procedures and stability of final products, as well as resultant particle size, should all be taken into consideration while choosing a formulation method. Our recent development of the HFDM method, for example, shows promise in formulating siRNA-loaded PEGylated lipid particles due to its simplicity as well as the superior stability of the final products (43). To date, these formulation procedures have been widely employed to prepare siRNA-loaded particles for the treatment of dyslipidemia, cancer, and viral infections (see Tables III and andIV).IV). Here, we will discuss some of these recent advances and how they impact on the field of RNAi therapy.
Direct conjugation of cholesterol to siRNA molecules was first demonstrated by Soutschek and colleagues to improve the delivery efficiency of siRNA targeting Apoliproprotein B (ApoB) in liver and jejunum after intravenous injection (28). This delivery strategy has also been recently adapted to deliver siRNA subcutaneously in the treatment of diabetic nephropathy in mice, with good success (44). Both of these studies, however, require the use of high doses of cholesterol-conjugated siRNA (50 mg/kg (28) or 400 μg/mouse (44)) which significantly limits their therapeutic applications in humans due to cost. In an effort to improve the delivery efficiency of this system, Wolfrum and colleagues showed that the interaction of these conjugates with the lipoprotein particles in the bloodstream is crucial for their cellular uptake (29). Preassembling of these conjugates with HDL was thus demonstrated to be five times more efficient in silencing ApoB expression in mice compared to equal amounts of cholesterol-siRNA conjugates. This, along with the ability of these particles to accumulate in a wide range of tissues after intravenous administration, dramatically improves the therapeutic potential of these cholesterol-conjugated siRNAs.
Apart from chemical conjugation, another strategy to deliver siRNA systemically is to entrap siRNA molecules in neutral lipid particles. These neutral vectors are generally deemed more favorable than the cationic ones as they are more biocompatible and also have superior pharmacokinetic profiles (45). The lack of interaction between neutral lipids and anionic polynucleotides, however, usually results in low entrapment efficiency (<10%) during the formulation process (46). To circumvent this problem, Semple and colleagues have utilized a pH-sensitive ionizable aminolipid, DODAP (pKa=6.6) to entrap antisense oligonucleotides using ethanol dialysis technique (46). While this lipid exhibits neutral charge at physiological pH, its cationic nature at acidic pH provides means to efficiently interact with nucleic acids with up to 70% entrapment efficiency being reported. Its use in siRNA delivery has also recently been demonstrated by Herringson and colleagues, although a small amount (2%) of cationic lipid, DOTAP, was reported to be essential in order to achieve high (>50%) entrapment efficiency in that study (47).
Alternatively, Landen and colleagues have developed a method of formulating dioleoylphosphatidylcholine (DOPC)-encapsulated siRNA liposomes which involves dissolving DOPC and siRNA in tert-butanol in the presence of Tween 20TM followed by lyophilization and rehydration (2). Although the mechanism by which the lipids interact with siRNA is unclear, this method of preparation was reported to result in the encapsulation efficiency of ~65%. Using this delivery system, 35–50% reduction in tumor growth was reported after intravenous or intraperitoneal administration of siRNA targeting EphA2 into mice bearing intraperitoneally implanted ovarian tumors (2,37). Their therapeutic potential in the treatment of other malignant diseases remains to be investigated.
PEGylated cationic lipid particles have been widely employed to deliver siRNA systemically due to their superior pharmacokinetic profiles, including the enhanced circulatory half-life compared to their non-PEGylated counterparts (48,49). The presence of cationic lipids in these systems also ensures the efficient interaction between lipids and anionic nucleic acids, thereby resulting in higher entrapment efficiencies than formulations made using neutral lipids (90% vs 65%, see Table II). Despite this, the presence of PEG in the formulations has also been shown to reduce the gene transfer efficiency to target cells (50). This is likely contributed by PEG’s interference with cellular uptake and the release of nucleotides from the endosomal compartment (51). A few strategies have, therefore, been developed to overcome this problem, and these are summarized in Fig. 2. These strategies can be incorporated into the existing PEGylated lipidic systems to optimize their delivery efficiency. Here, we will discuss two most successful PEGylated cationic lipidic systems to date for the systemic delivery of siRNA: (a) stable nucleic acid-lipid particles (SNALP) and (b) lipid-protamine-DNA/hyaluronic acid (LPD/LPH) nanoparticles.
In 2005, Jeffs et al. developed a novel “spontaneous vesicle formation” method for preparation of nucleotide-entrapped cationic PEGylated liposomes (52). In this formulation method, lipid solution was first prepared in 90% (v/v) ethanol and then mixed with an aqueous solution of DNA in a controlled manner using a T-connector. The mixing resulted in the ethanol concentration dropping below the value required to support lipid solubility. This led to the precipitation of solubilized lipid, and spontaneous liposomes were formed with entrapped DNA inside. The liposomes were then stabilized by further dilution, and finally, ethanol was removed by dialysis. The controlled, stepwise, mixing process employed in this method ensures the reproducibility of particle formation, and this procedure was subsequently adapted by Morrissey and colleagues to encapsulate siRNA (53). The resulting particles are usually termed SNALP, for stable nucleic acid-lipid particles. Using this SNALP system, Morrissey and colleagues demonstrated a 10-fold decrease in serum hepatitis B virus (HBV) DNA level in mice following intravenous injection of SNALP-containing siRNAs targeting HBV. The success of this delivery system led to the first systemic siRNA delivery study in nonhuman primates in 2006 (54). Using a dose as low as 1–2.5 mg/kg, Zimmermann and colleagues showed an 80–90% decrease in ApoB expression in the liver and a dramatic reduction in serum ApoB protein, cholesterol, and low-density lipoprotein levels in cynomolgus monkeys. This silencing potency was reported to be 100-fold greater than that achieved by cholesterol-conjugated siApoB.
It is important to note that ApoB is a hepatocyte-expressed gene, and the silencing of this gene indicated the effective delivery of SNALP to hepatocytes instead of Kupffer cells (specialized macrophages located in the liver which form part of the reticuloendothelial system (RES)). This is significant as it suggests that these PEGylation particles can target tissues outside of the RES and thus indicates their potential application in other clinical settings. Indeed, Judge et al. have recently demonstrated successful delivery of siRNA directed against polo-like kinase 1 (PLK1) to solid tumors in mice using SNALP, resulting in 75% reduction in subcutaneous tumor size (55). In addition, intraperitoneally administered SNALP containing siRNA targeting the polymerase L gene of Ebola virus (EBOV) has also been shown to protect guinea pigs against viremia and death following EBOV challenge (34). Overall, the development of the SNALP delivery system shows great promise in the systemic application of RNAi therapies.
Recent studies designed to improve SNALP have concentrated on the development of novel chemical methods to allow rapid synthesis of a large library of lipid-like delivery molecules, termed lipidoids, and testing their efficacy in siRNA delivery. SNALP formulations, which contain promising lipidoid molecules, have been recently demonstrated by Akinc and colleagues to achieve 75–90% reduction in ApoB or FVII factor expression in hepatocytes in nonhuman primates or mice, with more than 50% of silencing still observed after 14 days (56). This modified system was reported to reduce the total mass of delivery material relative to siRNA by 66%, compared to the original SNALP formulation, resulting in reduced toxicity as demonstrated by decreased elevation of ALT or AST enzyme after administration, a feature that is favorable for their clinical applications.
Self-assembling LPD nanoparticles were first developed in the Huang laboratory to deliver DNA plasmids for vaccine delivery (57). This system was subsequently modified to deliver siRNA, with the siRNA and DNA mixture being first condensed by a cationic polypeptide, protamine. This condensed core was then wrapped within cationic lipid membranes to facilitate cellular uptake. PEG-lipid moieties, with or without targeting ligands, were subsequently post-inserted onto the particle surface, providing surface protection and targeting specificity (58). The inclusion of calf-thymus DNA in the formulation overcomes the common issue of incomplete condensation of siRNA by lipidic vectors, thereby providing enhanced protection of siRNA from nuclease degradation (59). The resultant particles were demonstrated to achieve a significantly higher level (70–80%) of tumor localization (60) following systemic administration compared to the SNALP system (61), highlighting their potential use in cancer therapies. The circulatory half-life of these particles was found to be 20.5 h, which is longer than that reported for the SNALP system (12.4 h) (53). This difference could, however, be partially explained by the differences in doses, administrative and detection techniques used in these studies. Using these LPD particles, Li and colleagues showed a 40% reduction in subcutaneous tumor growth (60) or a 70–80% reduction in lung metastasis (62) after two to three doses of LPD particles containing siRNA targeting VEGF and/or MDM2 or c-myc at doses of 0.45–1.2 mg/kg. It is important to note that this dosing regimen is significantly lower than that used by Judge and colleagues for the treatment of subcutaneous tumors using SNALP system (55) (see Table III).
Despite these successes, dose-dependent elevation of various inflammatory or immune cytokines, including IL-6, IL-12, and IFN-α, were reported after systemic administration of these LPD nanoparticles (62). This nonspecific immunotoxicity was later found to be overcome by the replacement of calf-thymus DNA with hyaluronic acid (a high MW, anionic polysaccharide) (63). Similar to calf-thymus DNA, hyaluronic acid efficiently facilitates the condensation of siRNA in the presence of protamine but results in much lower immunotoxicity due to the lack of immunostimulatory CpG motifs. The resultant formulation was termed LPH (Lipid-protamine-hyaluronic acid) nanoparticles. Using this delivery system, Chono and colleagues showed an 80% reduction in luciferase activity in luciferase+ve B16F10 tumors in the lungs of the mice following a single intravenous injection of siRNA targeting luciferase (0.15 mg/kg) (63). Due to the significant decrease in the nonspecific inflammatory side effects compared to LPD nanoparticles, the LPH system presents as a more clinically acceptable siRNA delivery vector.
One of the major issues confronting the clinical use of lipidic systems for siRNA delivery is their toxicity or side-effect profile. Toxicity of the lipidic delivery systems generally depends on the type of lipids and the lipid/siRNA ratio used, with, for example, formulations containing DOPE typically displaying poorer toxicity profiles (64). Some concerns have also been raised about nonspecific activation of inflammatory cytokines and interferon responses by lipidic vectors. Ma and colleagues, for example, reported potent induction of both type I and type II interferon responses as well as activation of STAT 1 following intravenous administration of siRNA-containing DOTAP lipoplexes (65). Judge and colleagues have also reported similar results with DODMA-containing siRNA-loaded SNALP (66). Some of this immunostimulatory effect is likely to result from the introduction into cells of siRNA itself, exposing the siRNA to Toll-like receptors within the endosomes of these cells (65). It has been shown that this effect is somewhat sequence-dependent, with siRNAs containing 5′-GUCCUUCAA-3′ (67) or 5′-UGUGU-3′ (66) being highly immunostimulatory compared to other siRNAs when delivered using lipidic vectors. Several studies also indicate the possible involvement of other sequence motifs or factors for this phenomenon (65,68).
Apart from chemical modifications of siRNA molecules, such as substitution or methylation of uridine or guanosine residues (68,69), the optimization of the delivery system itself can also play a role in reducing this nonspecific effect. Hu-Lieskovan and colleagues, for example, showed that in contrast to lipidic vectors, systemic delivery of highly immunostimulatory siRNA using cyclodextran did not induce an interferon response (70). It was speculated that the endosomal buffering capacity of the cyclodextran delivery system contributed to this observation (71), as it has been shown by Sioud et al. (72) that the endolysosomal acidification process is crucial for the siRNA-mediated immunostimulatory phenomenon. It remains to be investigated whether the incorporation of polymers which contain high level of histidine (pKa~6) residues or secondary/tertiary amine moieties in liposomal formulations are able to reduce this effect due to the “proton sponge” mechanism (reviewed in (73)).
While efforts must be made to minimize the disturbance of the physiology of the subject receiving siRNA treatment, it is important to note that the nonspecific immunostimulatory responses may be of therapeutic benefit in certain clinical scenarios. A recent study reported by Poeck and colleagues clearly indicates the potential added benefit of the bifunctional siRNA in a melanoma cancer mouse model (74). The activation of the innate immune system was demonstrated to synergistically promote tumor cell apoptosis when an immunostimulatory siRNA targeting Bcl2 was administered intravenously using a linear polyethylenimine delivery vector. Whether this observation can be translated to other cancer models or infectious diseases remains to be investigated.
The attachment of targeting moieties on the surface of delivery vectors has been shown to enhance the delivery of siRNA to target cell population in vivo and thus improve therapeutic outcomes (12,60,75–77). It must be noted, however, that the presence of these targeting ligands generally do not affect the overall pharmacokinetics profiles or biodistribution of the delivery vectors (59,78–82). One biodistribution study performed using positron emission tomography and bioluminescent imaging, for example, clearly revealed the lack of correlation between the presence of transferring-targeting ligands and the level of tumor localization for stealth siRNA-containing nanoparticles (78). Instead, the improved therapeutic outcomes observed in those studies were attributed to the enhanced cellular uptake via receptor-mediated endocytosis when targeting ligands are present in the formulations. Ligand-targeting strategies are thus most beneficial for the systemic delivery of PEGylated particles where the presence of PEG interferes with cell entry or for the delivery of siRNA to cell populations which do not passively take up siRNA-containing particles readily (see Table V).
The targeting ligands investigated to date can generally be categorized into three main groups: glycosylated molecules, peptides or proteins (including antibodies; reviewed in (83)). The choice of ligands depends on the target cell population, with transferrin (Tf) and arginine-glycine-aspartic acid (RGD) moieties being widely applied in cancer therapies as receptors for these ligands are highly upregulated in various malignant tissues or cells. Hu-Lieskovan and colleagues, for example, have demonstrated significant inhibition of tumor growth in a murine model of metastatic Ewing’s sarcoma using EWS-FLI1 siRNA-entrapped Tf-targeted stealth cyclodextran nanoparticles (70). In contrast, siRNA entrapped in the corresponding nontargeted nanoparticles did not show any antitumor effect due to the lack of cellular uptake. This delivery system subsequently formed the basis of the first clinical trial on systemic targeted delivery of siRNA for the treatment of solid tumors (commenced in 2008) (84).
Apart from transferrin, attachment of folic acid or RGD moieties to siRNA delivery systems has also been reported in the treatment of cancer (see Table V). However, while these targeting moieties are easy to prepare and handle, they also bind to nontargeted tissues/cells and compete for binding with native molecules in the body (83). In contrast, antibody-mediated cell targeting is more specific although its application in cancer therapy is limited due to the heterogeneous nature of cancer cells (85,86) as well as the lack of identification of the suitable antibodies in different cancer types. Nevertheless, successful targeting of both leukocytes or human immunodeficiency virus (HIV)-infected cells have been reported using this antibody targeting strategy for siRNA delivery (76,87,88).
The technology of RNA interference shows great promise, and in less than a decade, its application has progressed from the bench to clinical trials. Systemic application of the siRNA molecules is, however, much more challenging. Nevertheless, a few delivery technologies such as SNALP or LPH systems have emerged as promising candidates. These systems display favorable pharmacokinetics profiles and result in accumulation of siRNA in target tissues following administration. Despite these successes, efforts in the future need to concentrate on developing more effective and safer delivery systems with better specificity for target sites. The effective targeting of novel sites such as microtumors or the central nervous systems will be one of the next challenges. Moreover, as the exclusive delivery of siRNA to specific tissue site is not possible following systemic administration, it is important to ensure the specificity of the siRNA for its gene target as well as having a complete understanding of its function in disease and normal tissues. It is envisaged that with the increasing development of suitable formulations and reliable preparation procedures, RNAi-based therapy will play a significant role in the treatment of various human diseases in the near future.