Screening with small interfering RNAs (siRNAs) is an important method for identifying genes involved in biological pathways
This article analyzes the effectiveness of two strategies--transfection of individual siRNAs and transfection of pools of siRNAs--for large-scale screening experiments. The increased effectiveness of testing at least two distinct, algorithm-derived siRNAs to the same target results in decreased use of resources required to screen false positive targets, and more importantly, minimizes the risk of missing valid targets.
The realization that potent siRNAs share sequence characteristics has led to siRNA design algorithms that significantly improve the percentage of effective siRNA sequences. For instance, the Cenix siRNA design algorithm--upon which Ambion's Silencer® Pre-designed siRNAs are based--has a success rate of >80%, with the majority of the siRNAs providing >85% reduction in target gene expression (Figure 1). Given the success rate of individual algorithm-designed siRNAs, the benefits of siRNA pooling have decreased (Figure 2). Furthermore, the recent observations of off-target effects by siRNAs  suggest that combining multiple siRNAs might actually increase the chances that genes other than the target are being affected.
Figure 1. Efficacy of Silencer® Pre-designed siRNAs. More than 1100 siRNAs targeting nearly 400 endogenously expressed human genes were tested for target mRNA reduction. More than 82% of the siRNAs reduced target gene expression by at least 70% and more than 61% reduced target gene expression by at least 85%. All siRNA designs were based on an algorithm developed by Cenix BioScience.
Figure 2. Target mRNA Reduction by Single siRNAs and siRNA Pools. Cells transfected with three individual Silencer® Pre-designed siRNAs (30 nM) or a pool of all three siRNAs (10 nM each) were monitored for target mRNA reduction by real-time RT-PCR 48 h post-transfection. Relative target mRNA expression was taken as a percentage of mRNA expressed in cells transfected with Silencer® Negative Control #1 siRNA (Ambion). All real-time RT-PCR data were normalized with 18S rRNA real-time RT-PCR.
Comparing siRNA Pools and Single siRNAs: Experimental Design and Results
Figure 3. Cell Proliferation--Single siRNAs vs. siRNA Pools. (A) Expression levels of 59 kinases were reduced in HeLa cells by transfection in 96 well plates of three individual kinase-specific siRNAs or a pool of the three siRNAs. All samples were performed in triplicate. To each well containing siPORT™ NeoFX™ Transfection Agent and either one of three siRNAs (30 nM) or the three pooled siRNAs (10 nM each), 8x103 HeLa cells were added in a process termed reverse transfection . Three days post-transfection, the cells were assayed using alamarBlue® (AccuMed International). The mean ± standard deviation is presented. The bars in the graph represent the average alamarBlue result for each siRNA or siRNA pool relative to a negative control (Silencer® Negative Control #1 siRNA; Ambion) transfected sample. (B) Representative results from specific genes are presented to exemplify the types of agreement and disagreement that occurred between expression results using the various siRNA combinations. Zero on the y axis is set to 2 standard deviations below the average signal from the negative control transfected samples. Bars that fall below zero are significantly different than the negative control and thus represent hits in our assay.
Figure 4. Caspase 3 Activation--Single siRNAs vs. siRNA Pools. (A) The expression levels of 59 kinases were reduced in HeLa cells by transfection of three single kinase-specific siRNAs or a pool of the siRNAs in 96 well plates. All samples were performed in triplicate. One day post-transfection, the cells were treated with etoposide to activate the apoptosis pathway. Three days post-transfection, the cells were counted and lysed. The caspase 3 activity in the lysates was assayed and normalized by cell number. The mean ± standard deviation of the three replicates is presented, where zero on the y axis is set to 2 standard deviations below the average signal from the negative control transfected samples. So bars that fall below zero are significantly different than the negative control and thus represents hits in our assay. The bars in the graph represent the mean caspase 3 activity relative to a negative control (Silencer® Negative Control #1 siRNA; Ambion) transfected sample. (B) Representative results from specific genes illustrate the types of agreement and disagreement that occurred between the various siRNAs.
Cell Proliferation Results
Cell number was measured using an alamarBlue® (AccuMed International) microplate-based assay. These results were confirmed by high-throughput microscopy. As shown in Figure 3A, siRNAs targeting the various kinases differentially affected cell number relative to negative control transfected samples. Different siRNAs targeting the same gene typically had similar effects on cell proliferation; however, it was common for one of the three single siRNAs to behave differently from the other two siRNAs. Similarly, siRNA pools often failed to display a phenotype consistent with the majority of the siRNAs that were individually transfected (Figure 3B).
Apoptosis Assay Results
During screens for kinases that influence apoptosis, both cell number and caspase 3 activity were measured following siRNA transfection and etoposide treatment. Assaying for cell number was necessary because some of the transfected siRNAs differentially altered cell growth or survival. Normalization of the caspase 3 activity data by cell number allowed us to express data as caspase 3 activity per cell. As expected, reducing expression of several kinases inhibited activation of caspase 3 by etoposide, suggesting that these kinases play a role in apoptosis (Figure 4). Consistent with the cell proliferation results, the single siRNAs for each gene had similar, though non-identical effects, on caspase 3 activity. Also consistent with the cell proliferation results, the siRNA pools only partially overlapped the results of the single siRNAs (Figure 5). Examples of single and pooled siRNA results are shown in Figure 4B.
Figure 5. Overall Performance of Single siRNAs and siRNA Pools. Using the definition of a hit as a gene for which two single siRNAs produce a phenotype that is significantly different than samples transfected with a negative control siRNA, the performance of each of the single siRNAs (siRNA #1, #2, or #3) as well as the siRNA pool of the three siRNAs was compared. The combined results of screening with two (#1+#2, #1+#3, and #2+#3) or three (#1+#2+#3) individual siRNAs illustrates the higher accuracy of this strategy. Accurate predictions of hits and inaccurate predictions of hits are shown in green and orange, respectively. False negatives are shown in grey.
Single siRNAs vs. siRNA Pools: Results
- False Positives. For both proliferation and caspase 3 assays, less than 60% of the hits predicted by the pooled siRNA results could be validated by the single siRNAs (False Positives, Figure 5).
- False Negatives. While the analysis of false positives wastes time and resources, of greater concern were the genes that failed to register as hits when the gene target did indeed participate in the pathway being analyzed. These occurrences, referred to as false
- negatives, were especially prevalent for the siRNA pools: Results from pooled siRNAs suggested that the gene was not involved in the apoptosis pathway, whereas results from two or more individual siRNAs to the same target suggested the gene was important for activation of the apoptosis pathway. In both screens, the pooled siRNA experiments had a 50% false negative rate, indicating that screening with siRNA pools can result in lost opportunities for target gene identification.
- Use of more than one individual siRNA. Though less problematic than experiments using siRNA pools, studies involving only a single gene-specific siRNA were prone to the relatively high rates of false positives and false negatives described above. The use of two, and preferably three, distinct siRNAs per gene significantly decreased the false negative rates of screening, making it possible to identify a more complete complement of interesting genes within the collection being analyzed (Figure 5).
Phenotype vs. siRNA Efficacy
Figure 6. Phenotype vs. mRNA Knockdown in Transfected Cells. HeLa cells were transfected with three different single siRNAs as well as an siRNA pool targeting each of the nine genes shown on the x-axis. Three days post-transfection, the wells were assayed for cell number (A) using alamarBlue® (AccuMed International). The RNA in the cells was isolated and assayed for target mRNA expression by real-time RT-PCR (B). The alamarBlue and the target mRNA quantification are presented as percentages of cells transfected with with Silencer® Negative Control #1 siRNA (Ambion).
David Brown, Mike Byrom, Joe Krebs, Kevin Kelnar, Rich Jarvis, Amanda Campbell, Lance Ford • Ambion, Inc.
2. (2004) Streamline Your siRNA Transfections. TechNotes 11(6):24-25.