Utilising the endogenous RNAi gene silencing pathway to selectively target genes of interest has greatly improved the understanding of cell biology at the in vitro level. The mechanism also has great therapeutic potential, but many hurdles remain to be overcome.
by Dr Mark A. Behlke
Introduction
RNA interference (RNAi), the endogenous cellular pathway to knock-down or silence gene expression by removing the mRNA signal, is a very popular experimental technique in the study of gene function. Many studies have used such processes to greatly increase our understanding of how individual proteins as well as networks operate. The approach can also provide vital information on where to target new pharmaceuticals to provide better treatments than already exist, or to enable presently untreatable diseases to be controlled. There is a push therefore, to enable RNAi technology to be applied successfully in vivo and therapeutically, but there are a significant number of challenges that need to be addressed before robust applications can be provided.
Piggybacking a natural process
In mammalian cells, small interfering RNAs (siRNAs) are produced by the enzymatic processing of long double-stranded RNAs (dsRNAs) by the RNase III class endoribonuclease Dicer. Newly formed siRNAs (usually around 21 bases in length) associate with Dicer, the human immunodeficiency virus transactivating response RNA-binding protein (TRBP) and Argonaute 2 (Ago2) to form the RNA-induced silencing complex (RISC). Only a single strand of the RNA is needed and therefore once the RISC has been formed, one strand of the siRNA (the ‘passenger’ strand) is degraded or discarded, whereas the other strand (the ‘guide’ or antisense strand) is retained and directs the sequence specificity of gene silencing [Figure 1]. This natural pathway can be piggybacked by researchers as an experimental tool for the reduction in expression of selected target genes.
Silent research
In research, siRNAs have traditionally been chemically synthesised as 21-mer RNA duplexes (19-base central double-stranded domain with terminal 2-base 3’ overhangs) that mimic the products of natural Dicer processing and assemble directly into the RISC. Alternative approaches to trigger RNAi exist that may have certain advantages, since they enter the endogenous process earlier on than traditional siRNAs. For example, Dicer-substrate RNAs (as available from Integrated DNA Technologies) are chemically synthesised 27-mer RNA duplexes that are optimised for Dicer processing. These are cleaved by Dicer into active, well defined 21-mers. Dicer-substrate RNAi methods take advantage of the link between Dicer and RISC loading that occurs when RNAs are processed by Dicer, and can boost potency by tenfold or more compared with 21-mer siRNAs at the same site. A further advantage of using Dicer-substrate RNAs is that the direction of Dicer processing can favour loading of the antisense strand into RISC. Correct design of the 27-mers is therefore crucial to realise the full benefits of this approach. Heavily 2’-O-methylated blunt 19-mer and 25-mer duplexes have also shown good potential [1].
A bumpy road ahead
Moving from in vitro paradigms to the more complex in vivo and therapeutic environments brings with it added complexities. These include site selection, design and chemistry, controls, route of administration and whether to use a delivery vehicle.
Site selection will not be discussed here, except to note that a number of site selection algorithms are available both commercially and academically. The design and chemistry of the siRNAs used in vivo are the main areas under intense investigation, as efficacy is affected by many different parameters. Unmodified and ‘naked’ siRNAs are rapidly degraded by nucleases when applied intravenously using traditional low-pressure injection. This highlights one of the major problems experienced when applying many biological compounds in vivo and it must be overcome to make them successful therapeutically. Stabilising siRNA compounds in this harsh environment can be achieved by chemically modifying the oligonucleotide separately or in combination with carriers/delivery vehicles. Hydrodynamic delivery has also shown promise with both unmodified and modified compounds; however this method is limited to mice [2, 3].
Chemical ‘mods’
The modification of oligonucleotides has been widely investigated and is applied for many purposes. Many different ‘mods’ are known to confer nuclease resistance, and have been applied successfully in vivo to single-stranded antisense oligonucleotides and ribozymes. Boranophosphate and phosphorothioate (PS) internucleoside modifications have been shown to provide effective resistance. Boranophosphate modification, however, cannot be achieved in a site-selective way; enzymatic in vitro transcription base-substitution techniques are used instead. Furthermore, PS modification can only be used to a limited degree as extensive modification reduces potency and can lead to toxicity.
A key site for stabilising- and nuclease resistance-modifications is the 2’ position on the ribose unit. Naturally occurring 2’-O-methyl modifications are seen in some rRNAs, snRNAs and tRNAs, though complete 2’-O-methyl modification seems to inactivate the RNAi trigger properties of the siRNA. Alternating the 2’-O-methyl modifications with unmodified nucleotides, however, retains potency and nuclease reistance. Substitution of the 2’-OH with fluorine (2’-F) and combining 2’-O-methyl (at purine bases) with 2’-F (at pyrimidine bases) have also shown good results in vivo [2, 4]. Locked nucleic acid (LNA) technology also involves the 2’-O but uses a methylene bridge to connect it with the 4’-C, essentially locking the ribose unit in the 3’-endo conformation. Incorporation of LNA bases in siRNA improves duplex stability and nuclease resistance, but placement is more restricted than for other 2’ modifications. As well as providing stability and nuclease resistance, chemical modifications placed at the ends of oligonucleotides can improve entry into cells as well as biodistribution. For example, conjugation to lipids (such as cholesterol) can improve binding to serum proteins and delivery to hepatocytes, and also results in better overall pharmacokinetics. Behlke [5] provides a complete overview of the use of siRNA modifications for in vivo paradigms.
Immune triggering
Innate immunity is a highly complex system of natural processes that protect species from a broad range of challenges. Long dsRNAs are known to induce the expression of one of the main response element families of the innate immune system – the interferons (IFN) – which in turn activate the interferon-stimulated gene (ISG) cascades. Short dsRNAs such as siRNA can also trigger IFN responses, which can be sequence specific as well as cell-type specific. Reviews of the numerous factors involved in this important consideration are available in Behlke [2] and Robbins et al [6]. This innate immune induction not only has an impact on the application of specific siRNA sequences in vivo and therapeutically, but also poses interesting questions for the use of negative control sequences in studies. As a result, it is important to consider monitoring IFN levels to check for any immune triggering.
Off-target effects
The precision of RNAi is due to highly specific base-pairing between the guide strand and the target sequence, yet off-target effects (OTEs) seem to be reasonably regular occurrences. Related to the degradative RNAi mechanism moderated by RISC, is a translational repression pathway using microRNA (miRNA) that may be regulated by a variant RISC utilising Ago1 (rather than Ago2). Conversely to the accuracy of RNAi, the miRNA pathway is surprisingly imperfect and exogenous siRNAs can function as miRNAs in this system. Furthermore, research has shown that these can occur with as little as 7-bp of homology between the guide strand and the off-target sequence. It is therefore possible that even the most carefully designed siRNA can produce OTEs due to this miRNA repression system. Another point to mention is that these systems are used for internal gene regulation, and exogenous siRNAs compete for processing, and if applied in large amounts can saturate the cellular machinery. This has a clear impact on the processing of endogenous molecules and could therefore have broader implications on gene regulation, significantly increasing the potential for OTEs [7].
Route of administration
Many of the points already discussed must also be considered in parallel with the route of administration. As with most pharmaceutical compounds, pharmacokinetics and bio-availability are dependent, in part, on the route of administration, and many of the same principles apply to siRNA. Systemic administration, such as intravenous injection, will require a larger dosage to achieve action at the desired location. In these situations it is very important to consider using modifications to prolong siRNA half-life and aid cellular uptake. However, as discussed briefly, large dosages of siRNA could have knock-on effects for global endogenous gene regulation, making such applications untenable for many desired effects or treatments. Furthermore, even if an siRNA has an acceptable half-life, plasma clearance can greatly reduce bio-availability very quickly. Alternative injection routes, such as intraperitoneal (ip) and subcutaneous (sc), can offer a more targeted route for specific conditions. For example, ip injections are very good for targeting the loading of siRNAs into macrophage populations, which can efficiently transport siRNA around the body. Where applicable, direct injection is a very good option since the desired tissue/component can be targeted with relatively small doses. For example, clinical trials are ongoing for a number of compounds that are injected intraocularly. Direct injection into the CNS also helps avoid the issues associated with delivery across the blood-brain barrier. Topical and inhalation applications have also shown some success [8].
Delivery vehicles
Another very active area of research is into the vehicles used to deliver oligonucleotides. This not only involves getting the molecules into the cells, but also in some way targeting the right cells and the right compartments within cells. Again, Behlke [2] offers a more in-depth review of the current landscape than is offered here.
With nucleic acids possessing a high negative charge density, cationic lipids have long been the molecules of choice for delivery vehicles. For example, liposomal formulations have been approved by the FDA for small-molecule drug delivery, and can be modified to further improve delivery and help target specific cells/tissues. Another important delivery molecule is polyethylenimine (PEI), which is very flexible in terms of enabling site targeting through various modifications and conjugations. Many cationic lipids and PEI derivatives though, posses in vivo toxicity, and therefore formulations have to be tested very carefully.
A promising group of delivery vehicles based on cationic peptides (e.g. MPG, Penetratin and transportin) has enabled researchers to exploit native cell-penetrating and cell-targeting capabilities. Eguchi et al recently described efficient siRNA delivery using peptide transduction domain-dsRNA binding domain fusion protein [9]. Other delivery systems utilising native capabilities include the bacteriophage phi29 DNA-packaging motor as well as various modified virus envelopes. A novel study into the inhibition of HIV-1 using two different aptamer siRNA delivery systems has shown great promise, not only in targeting HIV infected cells but also enabling rapid siRNA adaptation to avert viral resistance [10].
Physical and mechanical delivery systems are also regularly used in studies, such as the previously mentioned hydrodynamic delivery (HD), electroporation, ultrasound and ‘gene-gun’ systems. A novel ex vivo system that avoids many of the problems associated with these in vivo delivery mechanisms can also be used. Here a cell line can be transfected in vitro using standard methods and then implanted into the target area.
Summary
Piggybacking the endogenous RNAi gene silencing pathway to selectively target genes of interest has greatly improved our understanding of cell biology at the in vitro level and to some extent at the in vivo level. The mechanism also has great therapeutic potential, but the hurdles to providing consistent and robust silencing without significant side-effects (or OTEs) are numerous and not all fully understood. A number of compounds are undergoing clinical trials at present, but the recent failings of some of these highlight the barriers that still need to be overcome. For example, clinical trials of Opko Health’s ‘bevasiranib’ and Allergan’s ‘AGN-745’, siRNA developed to treat wet age-related macular degeneration via direct injection intraocularly have been halted. Data for both compounds show that their ‘therapeutic effects’ were due to the activation of the innate immune system (Toll-like Receptor 3 – TLR3) and not through any RNAi effects, raising questions about the misinterpretation of putative siRNA therapeutic results [11].
While these lead compound failures are obvious set-backs, if the information from them is used in combination with the ongoing research a lot can be learnt about the intricate interactions involved and the reasons why these compounds have failed. As a result, siRNA applications can move forward at all levels and eventually robust therapies may become available for a broad spectrum of disorders.
References
1. Kim DH et al. Nat Biotechnol 2005; 23: 222-226.
2. Behlke MA. Mol Therapy 2006; 13(4): 644-670.
3. Peek AS, Behlke MA. Curr Opin Mol Ther 2007; 9: 110-118.
4. Amarzguioui M et al. Nucleic Acids Res 2003; 31: 589-595.
5. Behlke MA. Oligonucleotides 2008; 18: 305-320.
6. Robbins M, Judge A, MacLachlan I. Oligonucleotides 2009; 19: 89-102.
7. Kim V N. Nat Rev Mol Cell Biol 2005; 6: 376-385.
8. Howard KA et al. Molecular Therapy 2009; 17: 162-168.
9. Eguchi A et al. Nat Biotechnol 2009; 27: 567-571.
10. Zhou J et al. Nucleic Acids Res 2009; 37: 3094-3109.
11. Editorial. As it matures, RNAi sees failures of key clinical candidates. RNAi News, June 11, 2009.
The author
Mark A. Behlke, M.D., Ph.D.,
Integrated DNA Technologies
Coralville, IA, USA
&
Integrated DNA Technologies, BVBA
Interleuvenlaan 12A,
B-3001 Leuven, Belgium
