Despite the high potential of the technology its limits and caveats have to be understood in order to design proper experiments and in order to critically evaluate the results. Artifacts related to off-target effects have generated the need to set up stricter rules for acceptance of siRNA experiments as defined in many peer-reviewed journals. RNAi induced by long dsRNA was originally used to study gene function in plants, worms, and flies. However, when dsRNAs of more than 30 bp in length were used in mammalian cells it was noted that there was an inhibition of protein translation within the cell due to the activation of an interferon response.26 These nonspecific responses to dsRNA are not triggered by dsRNAs shorter than 30 base pairs, including the siRNA duplexes. Moreover, studies in worms and flies have clearly demonstrated that synthetic siRNAs can produce effects similar to those of the long dsRNAs.46'47 Based on these experiments, siRNAs are now being optimized for systematic exploration of gene function in a variety of organisms.
In the following sections some of the critical factors are summarized, which may influence the efficiency of RNAi or which should be considered when performing or planning an RNAi experiment.
3.09.4.1 Critical Factors for Small Interfering Ribonucleic Acid Experiments
Only a subset of siRNAs designed for different regions of a target mRNA sequence is sufficiently effective. Even though there exist several useful algorithms to increase the success rate, a number of siRNAs must usually be tested against different sites of a particular mRNA to get satisfactory downregulation of the mRNA. It is recommended to avoid the first 75-100 nucleotides of any mRNA as potential target sites, since they may contain protein-binding regulatory sequences (5' UTR) that could interfere with the action of siRNA. It is also important to ensure that the sequence is specific to the target gene by performing a BLAST database49 search against all possible genes in the species the siRNA is designed for, in order to avoid cross-reaction with unwanted genes. However, the best-scoring sequences are not necessarily the best choice, since they do not necessarily predict secondary targets and might have other sequence liabilities. Also genetic polymorphisms and alternative splicing of genes have to be taken in consideration to reduce the probability of choosing an inappropriate target site.
3.09.4.1.2 Small interfering ribonucleic acid design criteria
Based on the experience from many siRNA experiments, a few guidelines have been defined for the synthesis of siRNAs. A general rule is that the sequence of one strand should be AA(N19)TT, where N is any nucleotide and these siRNAs should have a two-nucleotide 3' overhang. Furthermore, siRNAs should have a 5'-phosphate and a 3'-hydroxyl group for best efficiency. A number of additional factors have been proposed to play a role in the activity of siRNAs, including base length, internal thermodynamic stability, and base composition of the siRNA duplex: the use of RNA duplexes longer than 30 nucleotides has been shown to induce a nonspecific degradation pathway of all mRNAs in mammals.50 A duplex of 19-21 nucleotides appears to be the optimal length for RNAi activity. Since only one strand of the siRNA duplex is incorporated into RISC*51 it would be desirable to design the siRNA duplex so that the antisense strand is the one which is used preferentially. It has been shown that when siRNA are thermodynamically unstable (AU-rich) or even contain mismatches toward the 5' end of the antisense strand, this strand is preferentially used by RISC resulting in more efficient silencing.30,52 Moreover, lower thermal stability around the central region of the duplex facilitates cleavage of mRNA by RISC and higher thermal stability at the 3' antisense end of the duplex prevents unwinding and incorporation of the sense strand.
To prevent intramolecular folding, palindromic sequences, tandem repeats, and GC stretches of > 7-8 nucleotides should be avoided, and a number of algorithms have been developed to assist researchers in the identification of efficient and utilizable siRNA sequences. A collection of free siRNA design tools is available online.53
In the majority of cases siRNAs are synthesized chemically. An alternative is the independent transcription of the sense and antisense RNA strands in vitro from DNA oligonucleotide templates. The sense or antisense strand can be transcribed for example by using the T7 phage RNA polymerase.54 This polymerase produces individual siRNA sense and antisense strands which, once annealed, form siRNAs. Extra nucleotides required by the T7 promoter are removed by RNase digestion.
In the human genome more than half of the genes undergo alternative splicing, which adds additional complexity for siRNA design. Polymorphisms must also be taken into consideration when designing a siRNA and all these factors should be considered during the interpretation of experimental results. Single nucleotide polymorphisms (SNPs) can be found on average once every 300-500 bases in the human genome and it is estimated that the average gene contains around four coding sequence SNPs, for genes with allele frequencies of at least a few percent in the human population.55 All these factors can be the source of conflicting results when compared to other validation methods.
Another, often neglected property of eukaryotic cells is genetic redundancy. Many pathways can be triggered by different signaling proteins and when one mediator is lacking the intracellular protein network can compensate. This is most evident when considering genetic deletion of genes in mice and it is surprising that many genes originally judged essential for a specific function do not yield the expected phenotype upon deletion. In fact, there are many knockout mice strains for which up to now no obvious phenotype could be assigned, and in most cases this is probably due to genetic redundancy. For example, mice lacking single members of the cyclin-dependent kinase (CDK) family, which are considered essential components for cells to progress through the cell cycle, are viable with no obvious proliferation defects of cells and only combined knockout of several CDKs results in more apparent phenotypes, hinting at redundancy in the system.56
siRNA experiments have only just started to be used to examine gene redundancy and an example in Drosophila has been recently published, in which the functional redundancies of seven highly related genes belonging to the enhancer of split gene family were studied.57
3.09.4.2 Delivery into Cells
3.09.4.2.1 Getting small interfering ribonucleic acid into the cell
In order to understand a gene's function by using chemically synthesized siRNA it is necessary to optimize protocols for efficient delivery into cells. Several transfection reagents exist, the most commonly used being liposomal or amine-based. In some cases electroporation may be used, but cell toxicity can be high with this technique. Cell lines show varying responses to different transfection reagents, and it may be necessary to try more than one reagent or approach. Transfection efficiency is optimized by titrating cell density, transfection time, and the ratio of siRNA to transfection reagent. The cell passage number and the type of antibiotics used can also affect the efficiency of transfection.
An important parameter to be taken into account is the turnover rate of the protein to be targeted. Even when the mRNA level of a specific target is reduced, the protein might still be present and functional. siRNA-mediated RNAi lasts only for three to five cell-doubling times, probably due to gradual dilution of siRNA through cell division. Therefore, multiple transfections might be necessary in cases where the protein is particularly stable or the cells need to be grown for a long time to follow a delayed phenotype.
3.09.4.2.2 Intracellular expression of short hairpin ribonucleic acid
The high costs of synthesizing siRNAs when performing genome-wide screens and the need for more efficient delivery into different cell types has resulted in the development of alternative strategies to generate siRNAs.
One of these strategies is to produce the siRNAs in the cells by RNA polymerase III promoter-based DNA plasmids or expression cassettes.58 The most commonly used are the U6 and the H1 promoters. These constructs produce small inverted repeats, separated by a spacer of three to nine nucleotides, termed short hairpin RNAs (shRNAs), which are processed by DICER into siRNAs.59 Transcription begins at a specific initiation sequence, determined by the promoter used and the transcripts terminate with a series of 3' uridine residues, a feature that seems to favor efficiency of RNAi.60 In a slightly different approach the sense and antisense strands are transcribed separately by two individual promoters. In this case the two strands are annealed within the cell to form a siRNA, but it has been reported that this method of expressing siRNA results in less potent RNAi effects than expression of a corresponding shRNA.62
Using an RNA polymerase II promoter instead of an RNA polymerase III could be advantageous when a more regulated expression of siRNA is required, and a variety of inducible promoters are available to control the timing of the expression. A commonly used promoter is for example the human cytomegalovirus early promoter.63
These vectors provide advantages over chemically synthesized siRNAs, but they also have numerous disadvantages, including the often low or variable transfection efficiency, which is highly dependent on the cell type. To overcome these problems and to have a more prolonged expression, viral vectors are employed to deliver shRNA expression cassettes. Retroviral vectors are most widely used and murine retrovirus-based vectors have been shown to be efficient in delivery of shRNA.64'65 Lentivirus-based vectors have been also tested and appear to be promising vehicles for RNAi because they are effective in infecting noncycling cells, stem cells, and zygotes.66 Adenoviral vectors are highly effective, but allow only transient expression of siRNA.67
Early studies suggested that due to the high specificity of RNAi, a siRNA with one or two nucleotide sequence mismatches could serve as a negative control. However, with the identification of off-target effects this rule might be too stringent for some sequences. For example, the position and the type of base pairing for the mismatch influence silencing efficiency.68 The use of scrambled oligonucleotides is not a good alternative since these might not be recognized by the RISC complex and are not sufficiently homologous to the target sequence to function as an adequate control. A more convincing functional control can be to demonstrate the rescue of the target gene function following artificial overexpression of the target gene, although nonphysiological overexpression of a protein might result in additional artifacts. Observation of the same effect(s) upon transfection of different siRNAs, which target different regions within the mRNA of interest, certainly can increase confidence in experimental results.
The effect of RNAi should be quantified at both the mRNA and the protein level. In fact, a reduction in protein levels not accompanied by a decrease in mRNA might indicate that other mechanisms are at work, such as the induction of an interferon response. Many different assays to measure mRNA levels, including Northern blot, reverse-transcriptase polymerase chain reaction (RT-PCR) based methods, and microarray analysis, can be utilized to examine the effects of RNAi on the transcript levels. In most cases, assays performed between 24 and 72 h after initiation of RNAi should give appropriate results. Unrelated proteins with known, stable expression are basic controls to be included.
Protein knockdowns are typically confirmed by Western blot analysis, immunofluorescence, or flow cytometry. Although, RNAi generally occurs within 24 h of transfection, both onset and duration of RNAi depend on the turnover rate of the protein of interest, as well as the rate of dilution and longevity of the siRNAs.
Off-target effects resulting in the inhibition of genes not targeted by the specific siRNA are related to several cellular mechanisms, and represent one of the key limitations of the technology, potentially giving rise to misleading results. One potential pitfall, for example, is that a transcript with a high sequence homology to the target gene is degraded. Additionally, when there are some mismatches with a close homolog, siRNAs can sometimes act via a less stringent translational repression mechanism. As discussed above, an interferon or stress response can result in more general suppression of gene expression.
Off-target effects have been described for siRNAs with low homology to the nontargeted sequence, sharing as few as 11 contiguous nucleotides.69 In this study, the effects on target expression of 16 different siRNAs designed against the same gene were compared by microarray analysis. Some of these siRNAs showed strong differences in the expression profiles of genes with uniquely regulated genes for individual siRNAs. However, on the other hand, it has been also reported that just one mismatch between a siRNA and its target mRNA can abrogate silencing.5 The position of the mismatch is certainly a critical factor, and secondary structures might also contribute to the different responses. A systematic evaluation of 10 siRNAs targeting the menin gene suggests that induction of a stress response may result in off-target effects. Some but not all siRNAs caused a change in p53 and p21 RNA and protein levels, which were unrelated to silencing of the target gene.70
Off-target effects are more likely to be seen at high doses of siRNA oligonucleotides (>100nM), so a careful titration is recommended for any new siRNA experiment. The use of minimal amounts of siRNA may reduce the probability of inducing a nonspecific dsRNA response, of which most mammalian cells are capable.
Was this article helpful?
For centuries, ever since the legendary Ponce de Leon went searching for the elusive Fountain of Youth, people have been looking for ways to slow down the aging process. Medical science has made great strides in keeping people alive longer by preventing and curing disease, and helping people to live healthier lives. Average life expectancy keeps increasing, and most of us can look forward to the chance to live much longer lives than our ancestors.