Evidence of off-target effects associated with long dsRNAs in Drosophila cell-based assays

To evaluate the specificity of long dsRNAs used in high-throughput RNAi screens performed at the Drosophila RNAi Screening Center (DRSC), we performed a global analysis of their activity in 30 genome-wide screens completed at our facility. Surprisingly, our analysis predicts that dsRNAs containing ≥19 nucleotide perfect matches identified in silico to unintended targets may contribute to a significant false positive error rate arising from off-target effects. We confirmed experimentally that such sequences in dsRNAs lead to false positives and to the efficient knockdown of a cross-hybridizing transcript, raising a cautionary note when interpreting results based on the use of a single dsRNA per gene. Although a full appreciation of all causes of false positive errors remains to be determined, we suggest simple guidelines to help ensure high quality information from RNAi high-throughput screens Keywords: Specificity of long dsRNAs, Drosophila melanogaster, SL2 cell line, custom cDNA arrays, off-target effects, Drosophila RNAi Screening Center (DRSC) Custom cDNA microarrays were used to compare the transcriptional signature elicited by the PP2A-B’ dsRNA, DRSC16337 (D1), to those generated by 3 additional PP2A-B’ dsRNAs (C1, C2 and C3) in Drosophila melanogaster SL2 culture cells. Three independent experiments were set up for each dsRNA experiment. To avoid dye bias during labeling of cDNA we performed dye swaps. Cell culture and dsRNA experiments: dsRNA was generated and used essentially as previously described (Clemens, J. C. et al. Use of double-stranded RNA interference in Drosophila cell lines to dissect signal transduction pathways. Proc Natl Acad Sci U S A 97, 6499-503 (2000). Briefly, gene-specific primers were designed to include T7 promoter sequences used in vitro transcription of the PCR products (T7 MEGAscript Kit; Ambion) to generate both RNA strands in one reaction. The dsRNAs were purified with RNeasy columns (Qiagen) and quality controlled by agarose gel electrophoresis. Drosophila Schneider SL2 cells were cultured in Schneider’s medium (GibcoBRL) supplemented with 10% fetal bovine serum (JRH Bioscience) and Penicillin-Streptomycin (GibcoBRL) at 25 degree C, unless noted otherwise. 4x106 SL2 cells were seeded in each well of a 6-well cell culture dish (Falcon) in 1ml of serum-free media and treated with 15 µg of dsRNA at room temperature for 45 min. After 45 min, 2 ml of serum-supplemented Schneider’s medium was added and cells were further incubated for 72 hr. For expression profiling experiments three independent treatments were setup for each dsRNA. Total RNA was isolated by Trizol according to the manufacturer’s specifications (GibcoBRL), purified with RNeasy columns (Qiagen) and quality was assessed using the Agilent Bioanalyzer according to manufacturer’s specifications (Agilent). Custom cDNA-based Microarray: DNA probes to a set of approximately 3,000 strategically selected genes were generated by PCR. These probes were obtained from the PCR amplicons used to generate the dsRNA library 16. The 3,000 genes represented on the array include genes for all the Kinases (350) and Phosphatases (161) encoded in the Drosophila genome, known/canonical components and targets of the major signaling pathways such as Wg/Wnt, Hh, TGF-β, JAK/STAT, NFκB, Notch, RTK, JNK, and INR pathways; (http://genome.med.yale.edu/Lifecycle). Approximately 1,000 of the 3,000 genes were randomly selected to facilitate normalization of signal intensity across the array. Only PCR products showing a clear and strong bands were recovered using Montage PCR plates (Millipore). The DNA concentration was determined using PicoGreen reagent (Molecular Probes). The final concentration of DNA averaged 400 ng/microliter. Samples were transferred to 384 well plates (Genetix: x6004) containing DMSO to give a final concentration of 200 ng/microliter in 50% DMSO. The DNA probes were directly spotted onto Corning UltraGAPS coated glass slides using a Genetix QArray robot in the Biopolymer Facility at Harvard Medical School. Spots were arrayed in a 14 x 14 arrangement using 24 (4 x 6) Genetix 150 microm diameter tip solid microarray pins with a center-to-center spacing of 315 microm. A number of controls were prepared and printed on the glass slides with the probe DNAs. These controls include Negative controls (water, 50% DMSO, sonicated salmon sperm DNA, human Cot1, mouse Cot1, GFP, CFP, YFP, LacZ, and Luciferase) to assess the degree of nonspecific hybridization, and Spiked controls (Lucidea Universal Scorecard, Amersham Biosciences).