RNA interference in specific gene silencing ('knockdown')

1. RNA interference in specific gene silencing ('knockdown') Christopher V. Jones Jason Carter

2. RNA Interference • mRNA transcribed from DNA encodes for a protein expressed by a certain gene • The presence of certain double-stranded RNA (dsRNA) interferes with expression of a gene by interfering w/ the translation of its mRNA • dsRNAs direct the creation of small interfering RNAs (siRNAs) which target RNA-degrading enzymes (RNAses) to destroy mRNA transcripts complementary to the siRNAs

3. dsRNA (usually 21-nt) with 2-nt overhangs on either end, including a 5' phosphate group and a 3' hydroxy (-OH) group dsRNA enters RNAi pathway via enzyme Dicer producing siRNA siRNA molecules associate with a group of proteins termed the RNA-induced silencing complex (RISC), and directs the RISC to the target mRNA Small interfering RNA (siRNA)

5. Applications Typically, a single mRNA translates about 5,000 protein copies • RNAi can be used experimentally to "knockout" genes in organisms to help determine gene function • dsRNAs that trigger RNAi may be usable as drugs to treat genetic disorders or cancers • dsRNA can repress essential genes in pathogens or viruses that are dissimilar from any host genes

6. Advantages • Broad Applicability — Diseases for which abnormal gene function is a cause or a contributing factor are potentially treatable with RNA interference • Therapeutic Precision — Side effects associated with traditional drugs may be reduced or avoided by using RNAi-based drugs designed to inhibit expression of only a targeted gene and no others • Target RNA Destruction — Most drugs only temporarily prevent targeted protein function, RNAi-based drugs are designed to destroy the target RNA stopping undesirable protein production required for disease progression

7. Treatable diseases —Macular Degeneration • Eye disease caused by the growth of excess blood vessels • Caused by protein VEGF that promotes blood vessel growth • Vessels leak, clouding vision • dsRNAs can be delivered locally via injection • clinical trial of two dozen patients in 2004 • In two months: ¼ improved, ¾ stabilized

8. Treatable diseases —HIV • In 2002, scientists at MIT accounted they could interrupt various steps in the HIV life cycle using RNAi in cell cultures • Mutates and evolves resistance too rapidly for any single target mRNA • Molecular biologists at Colorado State University have engineered RNAi therapy aiming at multiple HIV genes • Clinical trials may start as early as 2006

9. Treatable diseases — Cancer • involves mutant genes that promote uncontrolled cell growth • researchers have silenced more than a dozen known cancer-causing genes with RNAi in cell cultures • delivery poses the key challenge for RNAi therapies: how to reach and penetrate tumors • Could stop production of P-glycoprotein which purges existing chemotherapy drugs from tumors, thus enhancing existing treatments

10. siRNA Prediction Given a target gene, how to design an siRNA to knock it down? • Select a candidate subsequence from the target gene • Not all subsequences are recognizable by Dicer • Arbitrary subsequence may knockdown unrelated gene(s) • Identify siRNA patterns that are effective through experimentation • Search entire genome to eliminate subsequences with off-target specificity

11. siRNA Prediction Method from:siDirect: highly effective target-specific siRNA design software for mammalian RNA interference, (Naito, Yamada, Ui-Tei, Morishita, Saigo, 2004) Studies of several genes led to these heuristics: • A/U at the 5' end of the antisense strand • G/C at the 5' end of the sense strand • AU richness in the 5' terminal 1/3rd of the antisense strand • the absence of any G/C stretch exceeding 9 bp in length

12. siRNA Prediction Method from:Rational siRNA design for RNA interference (Reynolds, Leake, Boese, Scaringe, Marshall, Khvorova, 2004) At least 7 points are required to be scored as effective siRNA • 30%-52% GC content – Add 1 point • Three or more A/Us at positions 15-19 (sense) - Add 1 point for each A/U for a total up to 5 points. At least 3 points are required. • A at position 19 (sense) - Add 1 point • A at position 3 (sense) - Add 1 point • U at position 10 (sense) - Add 1 point • No G/C at position 19 (sense) - Subtract 1 point for not satisfying this criterion. • No G at position 13 (sense) - Subtract 1 point for not satisfying this criterion.

13. Filtering out off-target hits • Once we have predicted potentially effective candidate siRNAs, we must search the entire genome for off-target matches • Exhaustive search is expensive, but accurate: Smith-Waterman algorithm • Approximate search: BLAST algorithm • Genes have introns that are spliced out of the mRNA • Alternative-splicing means exons are spliced several ways – we must search these areas also

14. Exhaustive vs. Approximate search • The human genome contains ~3B nt • Only 1.5% encodes proteins as genes • Must search ~45M nt, exon overlap sites, and alternative exon overlaps • Must repeat search for each candidate siRNA • Exhaustive search is O(nm) time and space complexity Smith-Waterman is a dynamic algorithm that finds optimal local alignment using a scoring system, a substitution matrix, and gap-scoring • Approximate search BLAST can run ~50 times faster using heuristic approach

15. Approximate Search -Basic Local Alignment Search Tool • BLAST breaks a search into stages • Searches for short matches of fixed length W between query and database • If there is a matching word W, performs an ungapped alignment between the query and database sequence, extending the match in each direction • High-scoring matches then subjected to a gapped alignment between the query sequence and the database sequence using a variation of the Smith-Waterman algorithm • Statistically significant matches are returned • Potential matches may get discarded due to heuristics

16. siRNA specificity • siRNA matches to any other gene of as few as 11 residues can lead to off-target silencing • High specificity has been observed with siRNAs that have at least 3 mismatches to all other genes • Would be considered to have a mismatch tolerance of 3 • Higher mismatch tolerance indicates higher specificity • Provides means to rank resulting siRNA candidates for study

17. Conclusions • Hundreds of successful experiments in cell cultures, and dozens in lab animals • siRNA delivery methods major hurdle • siRNA design will mature through competing prediction heuristics and better characterization of the RNAi machinery • As RNAi databases mature, novel biocomputing approaches are likely • Optimistic many RNAi therapies will enter clinical trials in next five years • Possible FDA approvals within the next decade

18. WebTools • siDirect: http://design.rnai.jp/ • Whitehead Institute siRNA: http://jura.wi.mit.edu/bioc/siRNAext/ • Wistar Bioinformatics Gene-specific siRNA selector: http://bioinfo.wistar.upenn.edu/siRNA/siRNA.htm • Ambion siRNA design and databases: http://www.ambion.com/techlib/misc/siRNA_tools.html Web RNAi databases http://www.rnainterference.org/ http://nematoda.bio.nyu.edu/cgi-bin/rnai/index.cgi

19. Bibliography • Review: Gene Silencing in mammals by small interfering RNAs, (McManus, Sharp) Genetics Vol. 3 Oct. 2002, 737-747 • Rational siRNA design for RNA interference (Reynolds, Leake, Boese, Scaringe, Marshall, Khvorova) Nature Biotechnology Vol. 22:3 Mar. 2004, 326-330. • siDirect: highly effective target-specific siRNA design software for mammalian RNA interference, (Naito, Yamada, Ui-Tei, Morishita, Saigo) Nucleic Acids Research Vol. 32 2004, 124-129. • Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference, (Ui-Tei, Naito, Takahashi, Haraguchi, Okhi-Hamazaki, Juni, Ueda, Saigo, 2004) Nucleic Acids Research Vol. 32:3 2004 • Potent and Persistent in-vivo anti-HBV activity of chemically modified siRNAs, (Morrisey, Lockridge, et. al.) Nature Biotechnology July 2004