RNAi Joins the “Singles Club”

One of our greatest opportunities in modern medicine is to develop therapies by targeting the body's very own genetic material. For illnesses characterized by loss-of-function phenotypes, we are tasked with increasing gene expression. Other conditions stem from aberrant or excessive gene expression, for which we would like to reduce messenger RNA (mRNA) or protein levels, or both, depending on the mechanism of disease. With these, the issue is twofold: how do we repress the expression of unwanted genes, and how do we formulate the therapeutic to be both safe and effective? Much progress has been made in this arena, resulting in the development of strategies employing antisense oligonucleotides (ASOs) or short interfering RNAs (siRNAs) for knockdown of undesirable genes. These techniques are a promising step forward but are not without caveats. Traditional nucleic acid approaches frequently encounter problems with metabolic stability, allele specificity, and tissue delivery.

The recently reported work that we highlight in this Commentary addresses these and other concerns. The paired studies of Lima,¹ Yu,² and their colleagues published in Cell demonstrate that potent single-stranded siRNAs (ss-siRNAs) can be designed to selectively target the unwanted mRNA and to inhibit gene expression through the RNA interference (RNAi) mechanism both in cultured mammalian cells and in vivo.

RNA-based methods of gene silencing have been generating interest for some time, both as research tools and for their therapeutic potential. Most of these techniques rely on short nucleotide sequences that work through base complementarity. ASOs are single-stranded molecules approximately 12–25 nt long; their resistance to degradation by nucleases and ability to bind mRNA are commonly enhanced through substitution of phosphodiester (PO) bonds with a phosphorothioate (PS) backbone and 2′-O-methyl- (2′-OMe-), 2′-O-methoxyethyl- (2′-MOE-), or locked nucleic acid–modified bases at the 3′ and 5′ ends.³ The unmodified “gap” region allows for cleavage of the target mRNA by RNase H, thereby blocking its translation into protein. The aforementioned modifications, in combination or with others, can be optimized for ASOs as well as for other RNA therapeutics to produce highly potent and efficient molecules. However, the foremost advantage of PS-modified ASOs thus far has been their propensity to be taken up by cells in vivo and to distribute throughout tissues because of the increased solubility afforded by this backbone, while retaining their high target specificity and susceptibility to RNase H.^4,5 The RNAi pathway begins with double-stranded RNA, but the 21- to 22-nt-long siRNA fragments lose their sense strand upon entry into the RNA-induced silencing complex (RISC). At this point the antisense, or “guide,” strand forms a complex with its target, causing cleavage of this mRNA by the argonaute-2 (AGO2) protein. Unmodified, siRNA can be quickly degraded by circulating nucleases, thus decreasing its utility as a systemically administered therapeutic.

Efforts have been made to enhance stability of siRNAs using chemical modifications such as those discussed above, but it is apparent that fewer changes can be incorporated without hindering the recruitment of the RISC complex to these siRNAs.³ If stability can indeed be achieved, some suggest that siRNAs are among the more potent oligonucleotides (albeit with the potential for adverse side effects) because of their efficient recognition of target mRNAs through RISC.⁴ Though we have come a long way in our understanding of what makes an effective RNA-induced gene knockdown strategy, key questions have yet to be answered before these become mainstream therapeutics.

Aiming to combine the strengths of single-stranded ASOs and double-stranded siRNAs, Lima et al. designed ss-siRNA molecules with maximal metabolic stability; these molecules nonetheless interacted with the RISC pathway to exert a potent silencing effect. The authors found that substitution of PS linkages uniformly throughout the ss-siRNA resulted in inactive molecules, but that ss-siRNAs with seven continuous PS substitutions at the 3′ end were active for mediating target mRNA cleavage by AGO2. Addition of a 2′-MOE adenosine dinucleotide to the 3′ end, alternating 2′-fluororibose (2′-F) and 2′-OMe at the 5′ end, alternating PS and PO motifs at the 5′ pole, a metabolically stable 5′ phosphonate analogue (for enhanced potency in vitro and in vivo), and, finally, a lipophilic C16 modification (for efficient cellular uptake and tissue distribution), yielded the optimally stable and active ss-siRNA upon either subcutaneous or intravenous administration.¹ In addition to chemical modifications, sequence length has been shown to affect activity. Of interest is the finding that in some cases, well-designed double-stranded sequences as short as 15 or 16 nt can induce more potent RNAi than a classical 19-nt siRNA.^6,7 Further experiments or perhaps future applications of ss-siRNA technology could implement these strategies to maximize silencing of a desired target.

To establish an RNAi-dependent mechanism of action for these ss-siRNAs, this study used 5′ rapid amplification of complementary DNA ends to identify the site of cleavage induced by their ss-siRNA targeting the PTEN gene both in vitro and in vivo. They found that it was consistent with that of AGO2; however, this does not rule out the possibility of an additional mechanism. Perhaps there is some role for endonucleolytic cleavage as in the case with ASOs, or for nonspecific degradation, and these questions could be addressed in future studies. However, in mouse embryonic fibroblast cells lacking Ago2, treatment with ss-siRNA induced no reduction in PTEN mRNA compared to wild type. These results suggest that with careful, substantial chemical modification, siRNAs can be designed that do not require the sense strand for activation of RNAi and repression of a specific mRNA.

For a therapeutic application, however, it is important to consider our capacity to target only undesired genes and/or alleles for silencing. This is important, for example, in Huntington's disease (HD), which is caused by a dominant heterozygous trinucleotide (CAG) repeat expansion in the huntingtin (HTT) gene. In this case, an inhibitor specific for only the mutant allele is desirable, leaving the wild-type allele intact. Yu et al.² address this issue in their study using ss-siRNAs containing some of the modifications outlined above, but introducing mismatched bases in the central region. Among the ss-siRNAs screened in an HD patient-derived cell line with 69 CAG repeats, the lowest IC₅₀ (3.3 nM) and greatest selectivity (>30-fold) were achieved by an oligonucleotide with three mismatched bases.² To better represent the average HD patient, an ss-siRNA was shown to inhibit mutant HTT expression with >100-fold selectivity in a cell line from a patient carrying 44 repeats on the mutant allele. Surprisingly, the authors observed no inhibition of other CAG repeat-containing genes, even though such genes contain potential targeting sites with identical sequences. Although the observed selectivity could be due to increased availability of binding sites on the HD expanded allele, each of the control genes examined encoded a number of CAG repeats in the range of wild-type HTT (the exception being FOXP2, whose 40 repeats are a mix of CAG and CAA). Alternatively, it is possible that expanded HTT alleles form structures that are stable, but less resistant to the RISC complex than those formed by wild-type HTT RNA.⁸

In a mouse model of HD with one mutant expanded allele (Q150) and one wild-type (Q7), intracerebroventricular delivery of a 5′-(E)-vinylphosphonate ss-siRNA for 4 weeks caused a decrease in expression of mutant HTT protein in a number of brain regions as measured by western blot, but no decrease in HTT mRNA by quantitative polymerase chain reaction. As predicted, use of siRNAs with mismatched bases results in in vivo inhibition of HTT by blockade of protein translation; this is consistent with in vitro results demonstrating no decrease in mRNA levels and lack of AGO2 or RNase H cleavage. However, because AGO2 was shown to be necessary for ss-siRNA silencing and because HTT mRNA was recovered from ss-siRNA-treated cells by immunoprecipitation of AGO2, the authors suggest that in this case, the protein is involved in recruitment of the RISC complex so as to block translation through a microRNA-like mechanism. By using a customized ss-siRNA to recognize only the mutant HD allele, these results support the feasibility of harnessing RNAi in this manner to attack a specific disease process.

The potential for using single-stranded RNA molecules to induce gene knockdown via RNAi has been explored in the field for the better part of a decade. Early reports described antisense siRNA with silencing efficiency comparable to that of double-stranded siRNA,^9,10 but they also suggested that duplex siRNAs may have a structural advantage.¹¹ Since then, innovative steps have been taken toward the design of stable, efficient ss-siRNAs that can be used in vivo as therapeutics. The ss-siRNAs employed in these most recent studies were systemically administered without the lipid formulations that are commonly required, and did not cause elevation of transaminases, bilirubin, cholesterol, or triglycerides in mice. However, siRNA and its delivery vehicles are known to cause undesirable immune stimulation,¹² the consequences of which may be more subtle than can be measured by spleen weight. Considering the potentially long-term use of ss-siRNAs as treatments for chronic and/or genetic illnesses, it would be beneficial to further characterize their systemic effects.

Nonetheless, this is a substantial development in the pursuit of therapies for devastating genetic conditions such as HD. To date, treatments for HD are symptomatic and largely ineffective, and attempts at slowing disease progression have been unsuccessful.¹³ Selective inhibition of mutant HTT expression is perhaps our best chance to address the root cause of this disorder and, coupled with proper screening, may even present the possibility for prevention. There are many other, currently irreversible, conditions that are caused by expression of an aberrant allele; in these cases silencing with an effective, nontoxic ss-siRNA may prove to be the definitive answer.

Thus far, many siRNA therapies have entered clinical trials, but few have been successful enough for release onto the market. Specific reasons for this vary for each drug, but in general RNA-based therapies have wanted for a proper balance between the desired silencing, delivery to affected tissues, and off-target effects.¹⁴ Basic science and clinical trials have given us much insight into the types of adjustments that would improve the clinical utility of RNAi. The ss-siRNAs developed in the recently published studies have incorporated many of these key ideas. Though more thorough scrutiny of the off-target effects on a whole-genome scale is advised early on for the ss-siRNA, like what was done for conventional siRNA almost a decade ago, we are cautiously optimistic that they will become promising therapeutics.