|
Why Rational Pooling of siRNAs is SMART
Mike Straka, PhD, and Queta Boese, PhD
There has been much discussion and some degree of confusion surrounding the use of siRNA pools, as opposed to one or more individual siRNAs, in gene silencing research. This article discusses the issue within a historical context and describes the theoretical and experimental basis supporting the advantages of pooling.
RNAi in Nature
RNA interference (RNAi) is an evolutionarily conserved mechanism, which represents a unique form of post-transcriptional gene silencing (reviewed in 1-6). In nature, RNAi is initiated by a long double-stranded RNA (dsRNA) that can originate as transcripts from an invading virus, a mobilized transposon or other inappropriately transcribed endogenous sequence. Long dsRNAs are processed by Dicer into a mixture of overlapping siRNAs that interact with RISC (Figure 1).
|
|
| Figure 1 |
Pools of siRNAs generated by nature and by rational design.
In mammalian cells, Dicer digests long dsRNA into a pool of duplexes that interact with RISC to induce silencing. |
|
One obvious feature of RNAi in nature is that the siRNAs generated by Dicer represent a set, or pool of siRNAs that function together to silence gene expression. It was thus reasoned that by replicating the mechanism using pools of synthetic siRNAs, individual genes could be silenced in a specific manner.
In early attempts, conventionally designed siRNAs targeting multiple sequences within the same mRNA were used; however, gene silencing by pooled siRNAs was often observed to be lower when compared to potent individual siRNAs. The observed reduction in silencing by pools was attributed to saturation of the RISC with poor or non-functional siRNAs that compete with the more functional duplexes for access to the RNAi apparatus. Figure 2 shows that only 4 of the 12 conventionally designed siRNAs (4, 8, 10, 11) knock down stably transfected luciferase expression by 90% or more (F90). When these duplexes were pooled with the less effective duplexes (F65 or less), silencing by the pools was less effective compared to the potent siRNAs, demonstrating that the effectiveness of pooled siRNAs is dependent in part upon the potency of the constituent duplexes.
|
|
| Figure 2 |
Silencing of stably transfected luciferase in HEK293 cells: rationally designed duplexes and pools are more potent silencers than those conventionally designed.
Conventional Design, Green bars: 4 of 12 individual siRNAs display ~90% knockdown; pools (P) are poor silencers.
Rational Design, Blue bars: 12 of 12 of Dharmacon SMARTselection® rationally designed siRNAs display 80-90% knockdown, and pooled duplexes (P) silence at >90%.
Black bar: Non-specific Control siRNA. |
|
Based on these results, it appeared that the inability to silence genes with pools of conventionally designed siRNA would be a barrier to gene silencing. However, Dharmacon scientists discovered that pools of rationally designed siRNA (Dharmacon SMARTpoo® siRNA reagents) resulted in potent and consistent silencing.
Conventional vs. Rational Design
In their early work with cell lysates, Tuschl and colleagues developed a simple set of four guidelines for the selection of siRNAs (7,8). These initial studies focused on the structural attributes of siRNA (19-25mer duplexes with 3' dinucleotide overhangs) and resulted in a "conventional" design process for the generation of functional siRNA.
In contrast to the simplicity of conventional design, the sophistication of Dharmacon SMARTselection technology represents the state of the art in rational siRNA design; it is an in silico process based on a weighted algorithm which incorporates numerous sequence-specific and thermodynamic parameters as well as bioinformatic analysis to identify unique functional siRNAs. The basis for the Dharmacon rational design algorithm has been previously described by Reynolds et al. (9).
The algorithm utilizes the fundamental principles described by Reynolds and colleagues along with further refinements; the refinements include but are not limited to GC content, siRNA internal stability profiles, potential competing internal or foldback structures, and base preferences at key positions (9).
The data in Figure 2 show that conventional siRNA design is inconsistent and may require the investigator to test multiple duplexes to obtain effective silencing. In contrast, rational design consistently generates highly functional siRNA. When such highly functional siRNA duplexes are pooled (Figure 2, blue bars), silencing is consistent and highly effective. In fact, Dharmacon SMARTpool siRNA reagents are guaranteed to knock down gene expression by at least 75%; in many cases the level of silencing is significantly higher, in the range of 90-95%.
Pooling Reduces Off-Target Activity
We have shown that Dharmacon SMARTselection technology enables consistent and potent silencing by individual siRNAs; an extension of the same SMARTselection algorithm is used to create SMARTpool reagents which consist of four rationally designed duplexes, targeting different regions of the same mRNA, combined into one reagent.
A distinct advantage to SMARTpool reagents, aside from guaranteed target gene silencing, is the reduction of off-target effects. This is when an unintended gene target is silenced, either by entry of the non-targeting siRNA strand into RISC, or when either strand silences an mRNA that is not the intended target. Alternatively, an excessively high concentration of siRNA can also cause off-target activity. In either case, silencing of unintended target(s) can lead to incorrect or confusing experimental results.
Global off-target effects are evaluated by gene expression profiling. Using microarray analysis, the profile, or signature, of gene regulation by an siRNA can be displayed in terms of a gene being either up- or down-regulated by some standard magnitude, generally 2-fold. By convention, down-regulated genes are displayed as green and up-regulated genes as red. A magnified portion of such an analysis is shown in Figure 3.
Numerous observations of off-target activity indicate that individual siRNAs likely have unique signatures; that is, each duplex will silence its own specific set of off-target genes. Figure 3 shows the gene expression profile of four rationally designed siRNAs targeting human Cyclophilin B; these were used at 100 nM in duplicate experiments. The data show that the off-target signatures of the four duplexes differ from each other and also between the two experiments, representing variation within biological replicates.
|
|
| Figure 3 |
Microarray analysis of off-target effects by individual and SMARTpool siRNAs targeting human Cyclophilin B.
Four individual SMARTselection designed duplexes (C1-C4) were used to target human Cyclophilin B in two experiments (A.C1-A.C4 and B.C1-B.C4). The duplexes were also pooled and tested in duplicate (A.pool and B.pool).
All siRNAs were transfected at 100 nM. |
|
The data in Figure 3 also demonstrate that, at the same concentration as the component duplexes, pooled siRNAs generally have less off-target activity. The explanation lies in the effective concentration of the duplexes in the pool: at a total siRNA concentration of 100 nM, the individual duplexes comprising the pool are at a much lower concentration, approximately 25 nM. Thus the off-target activity of each duplex is reduced by a concentration-dependent mechanism.
Pooling Reduces the Rate of False-Positive Results
Another class of undesirable observations in RNAi is false-positive results.
In a screen using a phenotypic assay such as cell proliferation, in which inhibition of cell growth is a positive response, a single siRNA, even if rationally designed, can indicate a false-positive and lead the investigator to consume precious time and resources tracking down a false hit.
In a given experiment, 64 rationally designed siRNAs were used to screen 16 genes purportedly involved in cell growth (four individual duplexes per target). Designating 60% survival as the cutoff (i.e. less than 60% survival indicates target involvement and thus is a potential positive), 12 positive "hits" were identified from the original 64 siRNAs screened. Two of the genes indicated as hits were then targeted for secondary screening with three additional individual siRNAs and the corresponding SMARTpool consisting of all four duplexes. Figure 4 shows the result of the secondary screen.
|
|
| Figure 4 |
Secondary screen of two "positives" (A and B). Pooling helps avoid false-positives.
Black Bars: target knockdown
Red Bars: cell survival using the individual duplex giving "positive" for two genes
Grey Bars: cell survival using remaining three siRNAs targeting the corresponding genes; these give the true phenotype, which is target knockdown with no effect on cell survival
Blue Bars: Pools give the true phenotype - gene knockdown but no effect on cell survival
Ctl: Controls |
|
The data in Figure 4 indicate all duplexes and pools are effective silencers (black bars). Grey and red bars indicate survivability, the screening phenotype, when silenced with individual siRNA duplexes. Red bars show survivability using two of the "positive" duplexes from the initial screen, and again these duplexes inhibit cell growth. However, the grey bars show survivability using the other three duplexes targeting the corresponding genes; these duplexes give the true phenotype (ie., no effect on cell growth). Thus, the two duplexes which inhibit growth (red bars) are revealed as false-positives.
The blue bars show survivability using a SMARTpool consisting of the same four duplexes: the pooled siRNAs give the true phenotype - no effect on cell growth. Thus, if the investigator had performed the initial screen using pooled siRNAs, the false-positive would have been avoided.
SUMMARY - Benefits of SMARTpool siRNAs
Synthetic siRNA-mediated gene silencing exploits a naturally occurring endogenous cellular pathway. The potency and specificity of silencing are highly dependent upon the particular siRNA sequence used, which in turn is the direct result of the technology used in its design. The Dharmacon design algorithm relies on and emulates the natural process, which selects for the most stable and potent silencers.
It has been asserted that pooled siRNAs are ineffective (10); however, the data presented to support that assertion were pools comprised of poorly-designed siRNAs which would clearly fall well below acceptable standards of functionality and would not be considered appropriate for use.
The experimental evidence presented here demonstrates that the functionality of individual duplexes and pools is directly related to the strength of the design algorithm – the more sophisticated and refined the algorithm, the better the performance. If an siRNA pool consists of poorly-designed duplexes which are inconsistent silencers, it should not be surprising that the pool’s gene knockdown is equally ineffective. In sharp contrast, rational design generates potent duplexes, and pools of highly functional duplexes demonstrate effective silencing: Dharmacon SMARTpool technologies represent the most advanced, potent and consistent gene silencing reagents available in the industry, and may additionally reduce false positives and off-target silencing activity. They enable rapid and reliable screening of potential targets with high confidence, saving time and experimental costs.
In conclusion, siRNA pools are only as SMART as the technology used in their design.
References
|
1
|
Zamore PD, RNA interference:listening to the sound of silence. Nat. Struct. Biol., 2001. 8(9):746-50. |
6
|
Hutvagner G and Zamore PD, RNAi: Nature abhors a double strand. Curr Opin. Genetics Dev., 2002. 12(2):225-32. |
|
2
|
Tuschl T, RNA interference and small interfering RNAs. Chembiochem., 2001. 2(4):239-45. |
7
|
Elbashir SM, Lendeckel W, and Tuschl T, RNA interference is mediated by 21- and 22-nucleotide RNAs. Genes Dev., 2001. 15(2):188-200. |
|
3
|
Sharp PA, RNA interference--2001. Genes Dev., 2001. 15(5):485-90. |
8
|
Elbashir SM, et al, Analysis of gene function in somatic mammalian cells using small interfering RNAs. Methods, 2002. 26(2):199-213. |
|
4
|
McManus MT and Sharp PA, Gene silencing in mammals by small interfering RNAs. Nature Rev. Genet., 2002. 3(10):737-47. |
9
|
Reynolds A, et al, Rational siRNA design for RNA interference. Nature Biotechnol., 2004. 22(3):326-30. |
|
5
|
Hannon GJ, RNA interference. Nature, 2002. 418(6894): 244-51. |
10
|
Brown, D, Byrom, M, Krebs, J, Kelnar, K, Jarvis, R, Campbell, A, Ford, L. Ambion TechNotes 12(1), February 2005 (http://www.ambion.com/techlib/
tn/121/11.html). |
|
|
|
|
|
