CROSS-REFERENCE TO RELATED APPLICATIONS

This application is a divisional application of U.S. Ser. No. 12/806,570, filed Aug. 17, 2010, which is a divisional application of U.S. Ser. No. 12/653,402, filed Dec. 11, 2009, which is a divisional application of U.S. Ser. No. 11/978,457, filed Oct. 29, 2007, now abandoned, which is a continuation-in-part of U.S. Ser. No. 10/714,333, filed Nov. 14, 2003, which claims the benefit of U.S. Provisional Application No. 60/426,137, filed Nov. 14, 2002, and also claims the benefit of U.S. Provisional Application No. 60/502,050, filed Sep. 10, 2003; U.S. Ser. No. 11/978,457 is also a continuation-in-part of U.S. Ser. No. 10/940,892, filed Sep. 14, 2004, which is a continuation of PCT Application No. PCT/US 04/14885, international filing date May 12, 2004. The disclosures of the priority applications, including the sequence listings and tables submitted in electronic form in lieu of paper, are incorporated by reference into the instant specification.

SEQUENCE LISTING

The sequence listing for this application has been submitted in accordance with 37 CFR Â§1.52(e) and 37 CFR Â§1.821 on CD-ROM in lieu of paper on a disk containing the sequence listing file entitled â€œDHARMA_â€”2100-US153_CRF.txtâ€ created Jun. 21, 2011, 128 kb. Applicants hereby incorporate by reference the sequence listing provided on CD-ROM in lieu of paper into the instant specification.

FIELD OF INVENTION

The present invention relates to RNA interference (â€œRNAiâ€ ).

BACKGROUND OF THE INVENTION

Relatively recently, researchers observed that double stranded RNA (â€œdsRNAâ€ ) could be used to inhibit protein expression. This ability to silence a gene has broad potential for treating human diseases, and many researchers and commercial entities are currently investing considerable resources in developing therapies based on this technology.

Double stranded RNA induced gene silencing can occur on at least three different levels: (i) transcription inactivation, which refers to RNA guided DNA or histone methylation; (ii) siRNA induced mRNA degradation; and (iii) mRNA induced transcriptional attenuation.

It is generally considered that the major mechanism of RNA induced silencing (RNA interference, or RNAi) in mammalian cells is mRNA degradation. Initial attempts to use RNAi in mammalian cells focused on the use of long strands of dsRNA. However, these attempts to induce RNAi met with limited success, due in part to the induction of the interferon response, which results in a general, as opposed to a target-specific, inhibition of protein synthesis. Thus, long dsRNA is not a viable option for RNAi in mammalian systems.

More recently it has been shown that when short (18-30 bp) RNA duplexes are introduced into mammalian cells in culture, sequence-specific inhibition of target mRNA can be realized without inducing an interferon response. Certain of these short dsRNAs, referred to as small inhibitory RNAs (â€œsiRNAsâ€ ), can act catalytically at sub-molar concentrations to cleave greater than 95% of the target mRNA in the cell. A description of the mechanisms for siRNA activity, as well as some of its applications are described in Provost et al. (2002) Ribonuclease Activity and RNA Binding of Recombinant Human Dicer, EMBO J. 21(21): 5864-5874; Tabara et al. (2002) The dsRNA Binding Protein RDE-4 Interacts with RDE-1, DCR-1 and a DexH-box Helicase to Direct RNAi in C. elegans, Cell 109(7):861-71; Ketting et al. (2002) Dicer Functions in RNA Interference and in Synthesis of Small RNA Involved in Developmental Timing in C. elegans; Martinez et al., Single-Stranded Antisense siRNAs Guide Target RNA Cleavage in RNAi, Cell 110(5):563; Hutvagner & Zamore (2002) A microRNA in a multiple-turnover RNAi enzyme complex, Science 297:2056.

From a mechanistic perspective, introduction of long double stranded RNA into plants and invertebrate cells is broken down into siRNA by a Type III endonuclease known as Dicer. Sharp, RNA interferenceâ€”2001, Genes Dev. 2001, 15:485. Dicer, a ribonuclease-III-like enzyme, processes the dsRNA into 19-23 base pair short interfering RNAs with characteristic two base 3â€² overhangs. Bernstein, Caudy, Hammond, & Hannon (2001) Role for a bidentate ribonuclease in the initiation step of RNA interference, Nature 409:363. The siRNAs are then incorporated into an RNA-induced silencing complex (RISC) where one or more helicases unwind the siRNA duplex, enabling the complementary antisense strand to guide target recognition. Nykanen, Haley, & Zamore (2001) ATP requirements and small interfering RNA structure in the RNA interference pathway, Cell 107:309. Upon binding to the appropriate target mRNA, one or more endonucleases within the RISC cleaves the target to induce silencing. Elbashir, Lendeckel, & Tuschl (2001) RNA interference is mediated by 21- and 22-nucleotide RNAs, Genes Dev. 15:188, FIG. 1.

The interference effect can be long lasting and may be detectable after many cell divisions. Moreover, RNAi exhibits sequence specificity. Kisielow, M. et al. (2002) Isoform-specific knockdown and expression of adaptor protein ShcA using small interfering RNA, J. Biochem. 363:1-5. Thus, the RNAi machinery can specifically knock down one type of transcript, while not affecting closely related mRNA. These properties make siRNA a potentially valuable tool for inhibiting gene expression and studying gene function and drug target validation. Moreover, siRNAs are potentially useful as therapeutic agents against: (1) diseases that are caused by over-expression or misexpression of genes; and (2) diseases brought about by expression of genes that contain mutations.

Successful siRNA-dependent gene silencing depends on a number of factors. One of the most contentious issues in RNAi is the question of the necessity of siRNA design, i.e., considering the sequence of the siRNA used. Early work in C. elegans and plants circumvented the issue of design by introducing long dsRNA (see, for instance, Fire, A. et al. (1998) Nature 391:806-811). In this primitive organism, long dsRNA molecules are cleaved into siRNA by Dicer, thus generating a diverse population of duplexes that can potentially cover the entire transcript. While some fraction of these molecules are non-functional (i.e., induce little or no silencing) one or more have the potential to be highly functional, thereby silencing the gene of interest and alleviating the need for siRNA design. Unfortunately, due to the interferon response, this same approach is unavailable for mammalian systems. While this effect can be circumvented by bypassing the Dicer cleavage step and directly introducing siRNA, this tactic carries with it the risk that the chosen siRNA sequence may be non-functional or semi-functional.

A number of researches have expressed the view that siRNA design is not a crucial element of RNAi. On the other hand, others in the field have begun to explore the possibility that RNAi can be made more efficient by paying attention to the design of the siRNA. Unfortunately, none of the reported methods have provided a satisfactory scheme for reliably selecting siRNA with acceptable levels of functionality. Accordingly, there is a need to develop rational criteria by which to select siRNA with an acceptable level of functionality, and to identify siRNA that have this improved level of functionality, as well as to identify siRNAs that are hyperfunctional.

SUMMARY OF THE INVENTION

The present invention is directed to increasing the efficiency of RNAi, particularly in mammalian systems. Accordingly, the present invention provides kits, siRNAs and methods for increasing siRNA efficacy.

According to a second embodiment, the present invention provides a method for selecting an siRNA, said method comprising applying selection criteria to a set of potential siRNA that comprise 18-30 base pairs, wherein said selection criteria are non-target specific criteria, and said set comprises at least two siRNAs and each of said at least two siRNAs contains a sequence that is at least substantially complementary to a target gene; and determining the relative functionality of the at least two siRNAs.

According to a third embodiment, the present invention also provides a method for selecting an siRNA wherein said selection criteria are embodied in a formula comprising:

(âˆ’14)*G₁₃âˆ’13*A₁âˆ’12*U₇âˆ’11*U₂âˆ’10*A₁₁âˆ’10*U₄âˆ’10*C₃âˆ’10*C₅âˆ’10*C₆âˆ’9*A₁₀âˆ’9*U₉âˆ’9*C₁₈âˆ’8*G₁₀âˆ’7*U₁âˆ’7*U₁₆âˆ’7*C₁₇âˆ’7*C₁₉+7*U₁₇+8*A₂+8*A₄+8*A₅+8*C₄+9*G₈+10*A₇+10*U₁₈+11*A₁₉+11*C₉+15*G₁+18*A₃+19*U₁₀âˆ’Tmâˆ’3*(GC total)âˆ’6*(GC_15-19)âˆ’30*X; orâ€ƒâ€ƒFormula VIII:

(âˆ’8)*A1+(âˆ’1)*A2+(12)*A3+(7)*A4+(18)*A5+(12)*A6+(19)*A7+(6)*A8+(âˆ’4)*A9+(âˆ’5)*A10+(âˆ’2)*A11+(âˆ’5)*A12+(17)*A13+(âˆ’3)*A14+(4)*A15+(2)*A16+(8)*A17+(11)*A18+(30)*A19+(âˆ’13)*U1+(âˆ’10)*U2+(2)*U3+(âˆ’2)*U4+(âˆ’5)*U5+(5)*U6+(âˆ’2)*U7+(âˆ’10)*U8+(âˆ’5)*U9+(15)*U10+(âˆ’1)*U11+(0)*U12+(10)*U13+(âˆ’9)*U14+(âˆ’13)*U15+(âˆ’10)*U16+(3)*U17+(9)*U18+(9)*U19+(7)*C1+(3)*C2+(âˆ’21)*C3+(5)*C4+(âˆ’9)*C5+(âˆ’20)*C6+(âˆ’18)*C7+(âˆ’5)*C8+(5)*C9+(1)*C10+(2)*C11+(âˆ’5)*C12+(âˆ’3)*C13+(âˆ’6)*C14+(âˆ’2)*C15+(âˆ’5)*C16+(âˆ’3)*C17+(âˆ’12)*C18+(âˆ’18)*C19+(14)*G1+(8)*G2+(7)*G3+(âˆ’10)*G4+(âˆ’4)*G5+(2)*G6+(1)*G7+(9)*G8+(5)*G9+(âˆ’11)*G10+(1)*G11+(9)*G12+(âˆ’24)*G13+(18)*G14+(11)*G15+(13)*G16+(âˆ’7)*G17+(âˆ’9)*G18+(âˆ’22)*G19+6*(number of A+U in position 15-19)âˆ’3*(number of G+C in whole siRNA),â€ƒâ€ƒFormula X

wherein position numbering begins at the 5â€²-most position of a sense strand, and
A₁=1 if A is the base at position 1 of the sense strand, otherwise its value is 0;
A₂=1 if A is the base at position 2 of the sense strand, otherwise its value is 0;
A₃=1 if A is the base at position 3 of the sense strand, otherwise its value is 0;
A₄=1 if A is the base at position 4 of the sense strand, otherwise its value is 0;
A₅=1 if A is the base at position 5 of the sense strand, otherwise its value is 0;
A₆=1 if A is the base at position 6 of the sense strand, otherwise its value is 0;
A₇=1 if A is the base at position 7 of the sense strand, otherwise its value is 0;
A₁₀=1 if A is the base at position 10 of the sense strand, otherwise its value is 0;
A₁₁=1 if A is the base at position 11 of the sense strand, otherwise its value is 0;
A₁₃=1 if A is the base at position 13 of the sense strand, otherwise its value is 0;
A₁₉=1 if A is the base at position 19 of the sense strand, otherwise if another base is present or the sense strand is only 18 base pairs in length, its value is 0;
C₃=1 if C is the base at position 3 of the sense strand, otherwise its value is 0;
C₄=1 if C is the base at position 4 of the sense strand, otherwise its value is 0;
C₅=1 if C is the base at position 5 of the sense strand, otherwise its value is 0;
C₆=1 if C is the base at position 6 of the sense strand, otherwise its value is 0;
C₇=1 if C is the base at position 7 of the sense strand, otherwise its value is 0;
C₉=1 if C is the base at position 9 of the sense strand, otherwise its value is 0;
C₁₇=1 if C is the base at position 17 of the sense strand, otherwise its value is 0;
C₁₈=1 if C is the base at position 18 of the sense strand, otherwise its value is 0;
C₁₉=1 if C is the base at position 19 of the sense strand, otherwise if another base is present or the sense strand is only 18 base pairs in length, its value is 0;
G₁=1 if G is the base at position 1 on the sense strand, otherwise its value is 0;
G₂=1 if G is the base at position 2 of the sense strand, otherwise its value is 0;
G₈=1 if G is the base at position 8 on the sense strand, otherwise its value is 0;
G₁₀=1 if G is the base at position 10 on the sense strand, otherwise its value is 0;
G₁₃=1 if G is the base at position 13 on the sense strand, otherwise its value is 0;
G₁₉=1 if G is the base at position 19 of the sense strand, otherwise if another base is present or the sense strand is only 18 base pairs in length, its value is 0;
U₁=1 if U is the base at position 1 on the sense strand, otherwise its value is 0;
U₂=1 if U is the base at position 2 on the sense strand, otherwise its value is 0;
U₃=1 if U is the base at position 3 on the sense strand, otherwise its value is 0;
U₄=1 if U is the base at position 4 on the sense strand, otherwise its value is 0;
U₇=1 if U is the base at position 7 on the sense strand, otherwise its value is 0;
U₉=1 if U is the base at position 9 on the sense strand, otherwise its value is 0;
U₁₀=1 if U is the base at position 10 on the sense strand, otherwise its value is 0;
U₁₅=1 if U is the base at position 15 on the sense strand, otherwise its value is 0;
U₁₆=1 if U is the base at position 16 on the sense strand, otherwise its value is 0;
U₁₇=1 if U is the base at position 17 on the sense strand, otherwise its value is 0;
U₁₈=1 if U is the base at position 18 on the sense strand, otherwise its value is 0.
GC_15-19=the number of G and C bases within positions 15-19 of the sense strand, or within positions 15-18 if the sense strand is only 18 base pairs in length;
GC_total=the number of G and C bases in the sense strand;
Tm=100 if the siRNA oligo has the internal repeat longer then 4 base pairs, otherwise its value is 0; and
X=the number of times that the same nucleotide repeats four or more times in a row.

According to a fourth embodiment, the invention provides a method for developing an algorithm for selecting siRNA, said method comprising: (a) selecting a set of siRNA; (b) measuring gene silencing ability of each siRNA from said set; (c) determining relative functionality of each siRNA; (d) determining improved functionality by the presence or absence of at least one variable selected from the group consisting of the presence or absence of a particular nucleotide at a particular position, the total number of As and Us in positions 15-19, the number of times that the same nucleotide repeats within a given sequence, and the total number of Gs and Cs; and (e) developing an algorithm using the information of step (d).

According to a fifth embodiment, the present invention provides a kit, wherein said kit is comprised of at least two siRNAs, wherein said at least two siRNAs comprise a first optimized siRNA and a second optimized siRNA, wherein said first optimized siRNA and said second optimized siRNA are optimized according a formula comprising Formula X.

The present invention also provides a method for identifying a hyperfunctional siRNA, comprising applying selection criteria to a set of potential siRNA that comprise 18-30 base pairs, wherein said selection criteria are non-target specific criteria, and said set comprises at least two siRNAs and each of said at least two siRNAs contains a sequence that is at least substantially complementary to a target gene; determining the relative functionality of the at least two siRNAs and assigning each of the at least two siRNAs a functionality score; and selecting siRNAs from the at least two siRNAs that have a functionality score that reflects greater than 80 percent silencing at a concentration in the picomolar range, wherein said greater than 80 percent silencing endures for greater than 120 hours.

According to a sixth embodiment, the present invention provides a hyperfunctional siRNA that is capable of silencing Bcl2.

According to a seventh embodiment, the present invention provides a method for developing an siRNA algorithm for selecting functional and hyperfunctional siRNAs for a given sequence. The method comprises:

(a) selecting a set of siRNAs;

(b) measuring the gene silencing ability of each siRNA from said set;

(d) determining the amount of improved functionality by the presence or absence of at least one variable selected from the group consisting of the total GC content, melting temperature of the siRNA, GC content at positions 15-19, the presence or absence of a particular nucleotide at a particular position, relative thermodynamic stability at particular positions in a duplex, and the number of times that the same nucleotide repeats within a given sequence; and

(e) developing an algorithm using the information of step (d).

According to this embodiment, preferably the set of siRNAs comprises at least 90 siRNAs from at least one gene, more preferably at least 180 siRNAs from at least two different genes, and most preferably at least 270 and 360 siRNAs from at least three and four different genes, respectively. Additionally, in step (d) the determination is made with preferably at least two, more preferably at least three, even more preferably at least four, and most preferably all of the variables. The resulting algorithm is not target sequence specific.

In another embodiment, the present invention provides rationally designed siRNAs identified using the formulas above.

In yet another embodiment, the present invention is directed to hyperfunctional siRNA.

The ability to use the above algorithms, which are not sequence or species specific, allows for the cost-effective selection of optimized siRNAs for specific target sequences. Accordingly, there will be both greater efficiency and reliability in the use of siRNA technologies.

In various embodiments, siRNAs that target nucleotide sequences for beta secretase (BACE) are provided. In various embodiments, the siRNAs are rationally designed. In various embodiments, the siRNAs are functional or hyperfunctional.

In various embodiments, an siRNA that targets the nucleotide sequence for BACE is provided, wherein the siRNA is selected from the group consisting of various siRNA sequences targeting the nucleotide sequences for BACE that are disclosed herein. In various embodiments, the siRNA sequence is selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734.

In various embodiments, siRNA comprising a sense region and an antisense region are provided, said sense region and said antisense region together form a duplex region comprising 18-30 base pairs, and said sense region comprises a sequence that is at least 90% similar to a sequence selected from the group consisting of siRNA sequences targeting nucleotide sequences for BACE that are disclosed herein. In various embodiments, the siRNA sequence is selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734.

In various embodiments, an siRNA comprising a sense region and an antisense region is provided, said sense region and said antisense region together form a duplex region comprising 18-30 base pairs, and said sense region comprises a sequence that is identical to a contiguous stretch of at least 18 bases of a sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734. In various embodiments, the duplex region is 19-30 base pairs, and the sense region comprises a sequence that is identical to a sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734.

In various embodiments, a pool of at least two siRNAs is provided, wherein said pool comprises a first siRNA and a second siRNA, said first siRNA comprising a duplex region of length 18-30 base pairs that has a first sense region that is at least 90% similar to 18 bases of a first sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734, and said second siRNA comprises a duplex region of length 18-30 base pairs that has a second sense region that is at least 90% similar to 18 bases of a second sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734, wherein said first sense region and said second sense region are not identical.

In various embodiments, the first sense region comprises a sequence that is identical to at least 18 bases of a sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734, and said second sense region comprises a sequence that is identical to at least 18 bases of a sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734. In various embodiments, the duplex of said first siRNA is 19-30 base pairs, and said first sense region comprises a sequence that is at least 90% similar to a sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734, and said duplex of said second siRNA is 19-30 base pairs and comprises a sequence that is at least 90% similar to a sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734.

In various embodiments, the duplex of said first siRNA is 19-30 base pairs and said first sense region comprises a sequence that is identical to at least 18 bases of a sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734, and said duplex of said second siRNA is 19-30 base pairs and said second region comprises a sequence that is identical to a sequence selected from the group consisting of SEQ ID NO. 438 to SEQ ID NO. 734.

For a better understanding of the present invention together with other and further advantages and embodiments, reference is made to the following description taken in conjunction with the examples, the scope of which is set forth in the appended claims.

BRIEF DESCRIPTION OF THE FIGURES

FIG. 1 shows a model for siRNA-RISC interactions. RISC has the ability to interact with either end of the siRNA or miRNA molecule. Following binding, the duplex is unwound, and the relevant target is identified, cleaved, and released.

FIG. 2 is a representation of the functionality of two hundred and seventy siRNA duplexes that were generated to target human cyclophilin, human diazepam-binding inhibitor (DB), and firefly luciferase.

FIG. 3a is a representation of the silencing effect of 30 siRNAs in three different cells lines, HEK293, DU145, and Hela. FIG. 3b shows the frequency of different functional groups (>95% silencing (black), >80% silencing (gray), >50% silencing (dark gray), and <50% silencing (white)) based on GC content. In cases where a given bar is absent from a particular GC percentage, no siRNA were identified for that particular group. FIG. 3c shows the frequency of different functional groups based on melting temperature (Tm).

FIG. 4 is a representation of a statistical analysis that revealed correlations between silencing and five sequence-related properties of siRNA: (A) an A at position 19 of the sense strand, (B) an A at position 3 of the sense strand, (C) a U at position 10 of the sense strand, (D) a base other than G at position 13 of the sense strand, and (E) a base other than C at position 19 of the sense strand. All variables were correlated with siRNA silencing of firefly luciferase and human cyclophilin. siRNAs satisfying the criterion are grouped on the left (Selected) while those that do not, are grouped on the right (Eliminated). Y-axis is â€œ% Silencing of Control.â€ Each position on the X-axis represents a unique siRNA.

FIGS. 5A and 5B are representations of firefly luciferase and cyclophilin siRNA panels sorted according to functionality and predicted values using Formula VIII. The siRNA found within the circle represent those that have Formula VIII values (SMARTSCORESâ„¢, or siRNA rank) above zero. siRNA outside the indicated area have calculated Formula VIII values that are below zero. Y-axis is â€œExpression (% Control).â€ Each position on the X-axis represents a unique siRNA.

FIG. 6A is a representation of the average internal stability profile (AISP) derived from 270 siRNAs taken from three separate genes (cyclophilin B, DBI and firefly luciferase). Graphs represent AISP values of highly functional, functional, and non-functional siRNA. FIG. 6B is a comparison between the AISP of naturally derived GFP siRNA (filled squares) and the AISP of siRNA from cyclophilin B, DBI, and luciferase having >90% silencing properties (no fill) for the antisense strand. â€œDGâ€ is the symbol for Î”G, free energy.

FIG. 7 is a histogram showing the differences in duplex functionality upon introduction of base pair mismatches. The X-axis shows the mismatch introduced in the siRNA and the position it is introduced (e.g., 8C>A reveals that position 8 (which normally has a C) has been changed to an A). The Y-axis is â€œ% Silencing (Normalized to Control).â€ The samples on the X-axis represent siRNAs at 100 nM and are, reading from left to right: 1A to C, 1A to G, 1A to U; 2A to C, 2A to G, 2A to U; 3A to C, 3A to G, 3A to U; 4G to A, 4G to C; 4G to U; 5U to A, 5U to C, 5U to G; 6U to A, 6U to C, 6U to G; 7G to A, 7G to C, 7G to U; 8C to A, 8C to G, 8C to U; 9G to A, 9G to C, 9G to U; 10C to A, 10C to G, 10C to U; 11G to A, 11G to C, 11G to U; 12G to A, 12G to C, 12G to U; 13A to C, 13A to G, 13A to U; 14G to A, 14G to C, 14G to U; 15G to A, 15G to C, 15G to U; 16A to C, 16A to G, 16A to U; 17G to A, 17G to C, 17G to U; 18U to A, 18U to C, 18U to G; 19U to A, 19U to C, 19U to G; 20 wt; Control.

FIG. 8 is histogram that shows the effects of 5â€² sense and antisense strand modification with 2â€²-O-methylation on functionality.

FIG. 9 shows a graph of SMARTSCORESâ„¢, or siRNA rank, versus RNAi silencing values for more than 360 siRNA directed against 30 different genes. SiRNA to the right of the vertical bar represent those siRNA that have desirable SMARTSCORESâ„¢, or siRNA rank.

FIGS. 10A-E compare the RNAi of five different genes (SEAP, DBI, PLK, Firefly Luciferase, and Renilla Luciferase) by varying numbers of randomly selected siRNA and four rationally designed (SMART-selected) siRNA chosen using the algorithm described in Formula VIII. In addition, RNAi induced by a pool of the four SMART-selected siRNA is reported at two different concentrations (100 and 400 nM). 10F is a comparison between a pool of randomly selected EGFR siRNA (Pool 1) and a pool of SMART-selected EGFR siRNA (Pool 2). Pool 1, S1-S4 and Pool 2 S1-S4 represent the individual members that made up each respective pool. Note that numbers for random siRNAs represent the position of the 5â€² end of the sense strand of the duplex. The X-axis indicates the duplex that was applied. The Y-axis represents the % expression of the control(s).

FIG. 11 shows the Western blot results from cells treated with siRNA directed against twelve different genes involved in the clathrin-dependent endocytosis pathway (CHC, DynII, CALM, CLCa, CLCb, Eps15, Eps15R, Rab5a, Rab5b, Rab5c, Î²2 subunit of AP-2 and EEA.1). siRNA were selected using Formula VIII. â€œPoolâ€ represents a mixture of duplexes 1-4. Total concentration of each siRNA in the pool is 25 nM. Total concentration=4Ã—25=100 nM.

FIG. 12 is a representation of the gene silencing capabilities of rationally-selected siRNA directed against ten different genes (human and mouse cyclophilin, C-myc, human lamin A/C, QB (ubiquinol-cytochrome c reductase core protein I), MEK1 and MEK2, ATE1 (arginyl-tRNA protein transferase), GAPDH, and Eg5). The Y-axis is the percent expression of the control. Numbers 1, 2, 3 and 4 represent individual rationally selected siRNA. â€œPoolâ€ represents a mixture of the four individual siRNA.

FIG. 13 is the sequence of the top ten Bcl2 siRNAs as determined by Formula VIII. Sequences are listed 5â€² to 3â€².

FIG. 14 is the knockdown by the top ten Bcl2 siRNAs at 100 nM concentrations. The Y-axis represents the amount of expression relative to the non-specific (ns) and transfection mixture control.

FIG. 15 represents a functional walk where siRNA beginning on every other base pair of a region of the luciferase gene are tested for the ability to silence the luciferase gene. The Y-axis represents the percent expression relative to a control. The X-axis represents the position of each individual siRNA. Reading from left to right across the X-axis, the position designations are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, and Plasmid.

FIGS. 16A and 16B are histograms demonstrating the inhibition of target gene expression by pools of 2 (16A) and 3 (16B) siRNA duplexes taken from the walk described in FIG. 15. The Y-axis in each represents the percent expression relative to control. The X-axis in each represents the position of the first siRNA in paired pools, or trios of siRNAs. For instance, the first paired pool contains siRNAs 1 and 3. The second paired pool contains siRNAs 3 and 5. Pool 3 (of paired pools) contains siRNAs 5 and 7, and so on. For each of 16A and 16B, the X-axis from left to right designations are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, and Plasmid.

FIGS. 17A and 17B are histograms demonstrating the inhibition of target gene expression by pools of 4 (17A) and 5 (17B) siRNA duplexes. The Y-axis in each represents the percent expression relative to control. The X-axis in each represents the position of the first siRNA in each pool. For each of 17A and 17B, the X-axis from left to right is 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, and Plasmid.

FIGS. 18A and 18B are histograms demonstrating the inhibition of target gene expression by siRNAs that are ten (18A) and twenty (18B) base pairs base pairs apart. The Y-axis represents the percent expression relative to a control. The X-axis represents the position of the first siRNA in each pool. For 18A, the X-axis from left to right designations areâ€”1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65 and Plasmid. For 18B, the X-axis from left to right designations are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, and Plasmid.

FIG. 19 shows that pools of siRNAs (dark gray bar) work as well (or better) than the best siRNA in the pool (light gray bar). The Y-axis represents the percent expression relative to a control. The X-axis from left to right designations are 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64, 65, 66, 67, 68, 69, 70, 71, 72, 73, 74, 75, 76, 77, 78, 79, 80, 81, 82, 83, 84, 85, 86, 87, 88, 89, 90, and Plasmid

FIG. 20 shows that the combination of several semifunctional siRNAs (dark gray) result in a significant improvement of gene expression inhibition over individual (semi-functional; light gray) siRNA. The X-axis represents the position of the individual siRNAs that were measured for effect on expression. The Y-axis represents the percent expression relative to a control.

FIGS. 21A, 21B and 21C show both pools (Library, Lib) and individual siRNAs in inhibition of gene expression of Beta-Galactosidase, Renilla Luciferase and SEAP (alkaline phosphatase). Numbers on the X-axis indicate the position of the 5â€²-most nucleotide of the sense strand of the duplex. The Y-axis represents the percent expression of each gene relative to a control. Libraries contain siRNAs that begin at the following nucleotides: FIG. 21Bâ€”Seap: Lib 1: 206, 766, 812, 923, Lib 2: 1117, 1280, 1300, 1487, Lib 3: 206, 766, 812, 923, 1117, 1280, 1300, 1487, Lib 4: 206, 812, 1117, 1300, Lib 5: 766, 923, 1280, 1487, Lib 6: 206, 1487; FIG. 21Aâ€”Bga1: Lib 1 (denoted as I on the figure): 979, 1339, 2029, 2590, Lib 2 (denoted as II on the figure): 1087, 1783, 2399, 3257, Lib 3 (denoted as III on the figure): 979, 1783, 2590, 3257, Lib 4 (denoted as IV on the figure): 979, 1087, 1339, 1783, 2029, 2399, 2590, 3257, Lib 5 (denoted as V on the figure): 979, 1087, 1339, 1783, Lib 6 (denoted as I on the figure): 2029, 2399, 2590, 3257; FIG. 21Câ€”Renilla: Lib 1: 174, 300, 432, 568, Lib 2: 592, 633, 729, 867, Lib 3: 174, 300, 432, 568, 592, 633, 729, 867, Lib 4: 174, 432, 592, 729, Lib 5: 300, 568, 633, 867, Lib 6: 592, 568.

FIG. 22 shows the results of an EGFR and TfnR internalization assay when single gene knockdowns are performed. The Y-axis represents percent internalization relative to control.

FIG. 23 shows the results of an EGFR and TfnR internalization assay when multiple genes are knocked down (e.g., Rab5a, b, c). The Y-axis represents the percent internalization relative to control.

FIG. 24 shows the simultaneous knockdown of four different genes. siRNAs directed against G6PD, GAPDH, PLK, and UQC were simultaneously introduced into cells. Twenty-four hours later, cultures were harvested and assayed for mRNA target levels for each of the four genes. A comparison is made between cells transfected with individual siRNAs vs. a pool of siRNAs directed against all four genes.

FIG. 25 shows the functionality of ten siRNAs at 0.3 nM concentrations.

DETAILED DESCRIPTIONDefinitions

Unless stated otherwise, the following terms and phrases have the meanings provided below:

Complementary

The term â€œcomplementaryâ€ refers to the ability of polynucleotides to form base pairs with one another. Base pairs are typically formed by hydrogen bonds between nucleotide units in antiparallel polynucleotide strands. Complementary polynucleotide strands can base pair in the Watson-Crick manner (e.g., A to T, A to U, C to G), or in any other manner that allows for the formation of duplexes. As persons skilled in the art are aware, when using RNA as opposed to DNA, uracil rather than thymine is the base that is considered to be complementary to adenosine. However, when a U is denoted in the context of the present invention, the ability to substitute a T is implied, unless otherwise stated.

Perfect complementarity or 100% complementarity refers to the situation in which each nucleotide unit of one polynucleotide strand can hydrogen bond with a nucleotide unit of a second polynucleotide strand. Less than perfect complementarity refers to the situation in which some, but not all, nucleotide units of two strands can hydrogen bond with each other. For example, for two 20-mers, if only two base pairs on each strand can hydrogen bond with each other, the polynucleotide strands exhibit 10% complementarity. In the same example, if 18 base pairs on each strand can hydrogen bond with each other, the polynucleotide strands exhibit 90% complementarity.

Deoxynucleotide

The term â€œdeoxynucleotideâ€ refers to a nucleotide or polynucleotide lacking a hydroxyl group (OH group) at the 2â€² and/or 3â€² position of a sugar moiety. Instead, it has a hydrogen bonded to the 2â€² and/or 3â€² carbon. Within an RNA molecule that comprises one or more deoxynucleotides, â€œdeoxynucleotideâ€ refers to the lack of an OH group at the 2â€² position of the sugar moiety, having instead a hydrogen bonded directly to the 2â€² carbon.

Deoxyribonucleotide

The terms â€œdeoxyribonucleotideâ€ and â€œDNAâ€ refer to a nucleotide or polynucleotide comprising at least one sugar moiety that has an H, rather than an OH, at its 2â€² and/or 3â€² position.

Duplex Region

The phrase â€œduplex regionâ€ refers to the region in two complementary or substantially complementary polynucleotides that form base pairs with one another, either by Watson-Crick base pairing or any other manner that allows for a stabilized duplex between polynucleotide strands that are complementary or substantially complementary. For example, a polynucleotide strand having 21 nucleotide units can base pair with another polynucleotide of 21 nucleotide units, yet only 19 bases on each strand are complementary or substantially complementary, such that the â€œduplex regionâ€ has 19 base pairs. The remaining bases may, for example, exist as 5â€² and 3â€² overhangs. Further, within the duplex region, 100% complementarity is not required; substantial complementarity is allowable within a duplex region. Substantial complementarity refers to 79% or greater complementarity. For example, a mismatch in a duplex region consisting of 19 base pairs results in 94.7% complementarity, rendering the duplex region substantially complementary.

Filters

The term â€œfilterâ€ refers to one or more procedures that are performed on sequences that are identified by the algorithm. In some instances, filtering includes in silico procedures where sequences identified by the algorithm can be screened to identify duplexes carrying desirable or undesirable motifs. Sequences carrying such motifs can be selected for, or selected against, to obtain a final set with the preferred properties. In other instances, filtering includes wet lab experiments. For instance, sequences identified by one or more versions of the algorithm can be screened using any one of a number of procedures to identify duplexes that have hyperfunctional traits (e.g., they exhibit a high degree of silencing at subnanomolar concentrations and/or exhibit high degrees of silencing longevity).

Gene Silencing

The phrase â€œgene silencingâ€ refers to a process by which the expression of a specific gene product is lessened or attenuated. Gene silencing can take place by a variety of pathways. Unless specified otherwise, as used herein, gene silencing refers to decreases in gene product expression that results from RNA interference (RNAi), a defined, though partially characterized pathway whereby small inhibitory RNA (siRNA) act in concert with host proteins (e.g., the RNA induced silencing complex, RISC) to degrade messenger RNA (mRNA) in a sequence-dependent fashion. The level of gene silencing can be measured by a variety of means, including, but not limited to, measurement of transcript levels by Northern Blot Analysis, B-DNA techniques, transcription-sensitive reporter constructs, expression profiling (e.g., DNA chips), and related technologies. Alternatively, the level of silencing can be measured by assessing the level of the protein encoded by a specific gene. This can be accomplished by performing a number of studies including Western Analysis, measuring the levels of expression of a reporter protein that has e.g., fluorescent properties (e.g., GFP) or enzymatic activity (e.g., alkaline phosphatases), or several other procedures.

miRNA

The term â€œmiRNAâ€ refers to microRNA.

Nucleotide

The term â€œnucleotideâ€ refers to a ribonucleotide or a deoxyribonucleotide or modified form thereof, as well as an analog thereof. Nucleotides include species that comprise purines, e.g., adenine, hypoxanthine, guanine, and their derivatives and analogs, as well as pyrimidines, e.g., cytosine, uracil, thymine, and their derivatives and analogs.

Nucleotide analogs include nucleotides having modifications in the chemical structure of the base, sugar and/or phosphate, including, but not limited to, 5-position pyrimidine modifications, 8-position purine modifications, modifications at cytosine exocyclic amines, and substitution of 5-bromo-uracil; and 2â€²-position sugar modifications, including but not limited to, sugar-modified ribonucleotides in which the 2â€²-OH is replaced by a group such as an H, OR, R, halo, SH, SR, NH₂, NHR, NR₂, or CN, wherein R is an alkyl moiety. Nucleotide analogs are also meant to include nucleotides with bases such as inosine, queuosine, xanthine, sugars such as 2â€²-methyl ribose, non-natural phosphodiester linkages such as methylphosphonates, phosphorothioates and peptides.

Modified bases refer to nucleotide bases such as, for example, adenine, guanine, cytosine, thymine, uracil, xanthine, inosine, and queuosine that have been modified by the replacement or addition of one or more atoms or groups. Some examples of types of modifications that can comprise nucleotides that are modified with respect to the base moieties include but are not limited to, alkylated, halogenated, thiolated, aminated, amidated, or acetylated bases, individually or in combination. More specific examples include, for example, 5-propynyluridine, 5-propynylcytidine, 6-methyladenine, 6-methylguanine, N,N,-dimethyladenine, 2-propyladenine, 2-propylguanine, 2-aminoadenine, 1-methylinosine, 3-methyluridine, 5-methylcytidine, 5-methyluridine and other nucleotides having a modification at the 5 position, 5-(2-amino)propyl uridine, 5-halocytidine, 5-halouridine, 4-acetylcytidine, 1-methyladenosine, 2-methyladenosine, 3-methylcytidine, 6-methyluridine, 2-methylguanosine, 7-methylguanosine, 2,2-dimethylguanosine, 5-methylaminoethyluridine, 5-methyloxyuridine, deazanucleotides such as 7-deaza-adenosine, 6-azouridine, 6-azocytidine, 6-azothymidine, 5-methyl-2-thiouridine, other thio bases such as 2-thiouridine and 4-thiouridine and 2-thiocytidine, dihydrouridine, pseudouridine, queuosine, archaeosine, naphthyl and substituted naphthyl groups, any O- and N-alkylated purines and pyrimidines such as N6-methyladenosine, 5-methylcarbonylmethyluridine, uridine 5-oxyacetic acid, pyridine-4-one, pyridine-2-one, phenyl and modified phenyl groups such as aminophenol or 2,4,6-trimethoxy benzene, modified cytosines that act as G-clamp nucleotides, 8-substituted adenines and guanines, 5-substituted uracils and thymines, azapyrimidines, carboxyhydroxyalkyl nucleotides, carboxyalkylaminoalkyl nucleotides, and alkylcarbonylalkylated nucleotides. Modified nucleotides also include those nucleotides that are modified with respect to the sugar moiety, as well as nucleotides having sugars or analogs thereof that are not ribosyl. For example, the sugar moieties may be, or be based on, mannoses, arabinoses, glucopyranoses, galactopyranoses, 4â€²-thioribose, and other sugars, heterocycles, or carbocycles.

The term nucleotide is also meant to include what are known in the art as universal bases. By way of example, universal bases include but are not limited to 3-nitropyrrole, 5-nitroindole, or nebularine. The term â€œnucleotideâ€ is also meant to include the N3â€² to P5â€² phosphoramidate, resulting from the substitution of a ribosyl 3â€² oxygen with an amine group.

Further, the term nucleotide also includes those species that have a detectable label, such as for example a radioactive or fluorescent moiety, or mass label attached to the nucleotide.

Off-Target Silencing and Off-Target Interference

The phrases â€œoff-target silencingâ€ and â€œoff-target interferenceâ€ are defined as degradation of mRNA other than the intended target mRNA due to overlapping and/or partial homology with secondary mRNA messages.

Polynucleotide

The term â€œpolynucleotideâ€ refers to polymers of nucleotides, and includes but is not limited to DNA, RNA, DNA/RNA hybrids including polynucleotide chains of regularly and/or irregularly alternating deoxyribosyl moieties and ribosyl moieties (i.e., wherein alternate nucleotide units have an â€”OH, then and â€”H, then an â€”OH, then an â€”H, and so on at the 2â€² position of a sugar moiety), and modifications of these kinds of polynucleotides, wherein the attachment of various entities or moieties to the nucleotide units at any position are included.

Polyribonucleotide

The term â€œpolyribonucleotideâ€ refers to a polynucleotide comprising two or more modified or unmodified ribonucleotides and/or their analogs. The term â€œpolyribonucleotideâ€ is used interchangeably with the term â€œoligoribonucleotide.â€

Ribonucleotide and Ribonucleic Acid

The term â€œribonucleotideâ€ and the phrase â€œribonucleic acidâ€ (RNA), refer to a modified or unmodified nucleotide or polynucleotide comprising at least one ribonucleotide unit. A ribonucleotide unit comprises an hydroxyl group attached to the 2â€² position of a ribosyl moiety that has a nitrogenous base attached in N-glycosidic linkage at the 1â€² position of a ribosyl moiety, and a moiety that either allows for linkage to another nucleotide or precludes linkage.

siRNA

The term â€œsiRNAâ€ refers to small inhibitory RNA duplexes that induce the RNA interference (RNAi) pathway. These molecules can vary in length (generally 18-30 base pairs) and contain varying degrees of complementarity to their target mRNA in the antisense strand. Some, but not all, siRNA have unpaired overhanging bases on the 5â€² or 3â€² end of the sense strand and/or the antisense strand. The term â€œsiRNAâ€ includes duplexes of two separate strands, as well as single strands that can form hairpin structures comprising a duplex region.

siRNA may be divided into five (5) groups (non-functional, semi-functional, functional, highly functional, and hyper-functional) based on the level or degree of silencing that they induce in cultured cell lines. As used herein, these definitions are based on a set of conditions where the siRNA is transfected into said cell line at a concentration of 100 nM and the level of silencing is tested at a time of roughly 24 hours after transfection, and not exceeding 72 hours after transfection. In this context, â€œnon-functional siRNAâ€ are defined as those siRNA that induce less than 50% (<50%) target silencing. â€œSemi-functional siRNAâ€ induce 50-79% target silencing. â€œFunctional siRNAâ€ are molecules that induce 80-95% gene silencing. â€œHighly-functional siRNAâ€ are molecules that induce greater than 95% gene silencing. â€œHyperfunctional siRNAâ€ are a special class of molecules. For purposes of this document, hyperfunctional siRNA are defined as those molecules that: (1) induce greater than 95% silencing of a specific target when they are transfected at subnanomolar concentrations (i.e., less than one nanomolar); and/or (2) induce functional (or better) levels of silencing for greater than 96 hours. These relative functionalities (though not intended to be absolutes) may be used to compare siRNAs to a particular target for applications such as functional genomics, target identification and therapeutics.

SMARTSCOREâ„¢, or siRNA Rank

The term â€œSMARTSCOREâ„¢â€ , or â€œsiRNA rankâ€ refers to a number determined by applying any of the formulas to a given siRNA sequence. The term â€œSMART-selectedâ€ or â€œrationally selectedâ€ or â€œrational selectionâ€ refers to siRNA that have been selected on the basis of their SMARTSCORESâ„¢, or siRNA ranking.

Substantially Similar

The phrase â€œsubstantially similarâ€ refers to a similarity of at least 90% with respect to the identity of the bases of the sequence.

Target

The term â€œtargetâ€ is used in a variety of different forms throughout this document and is defined by the context in which it is used. â€œTarget mRNAâ€ refers to a messenger RNA to which a given siRNA can be directed against. â€œTarget sequenceâ€ and â€œtarget siteâ€ refer to a sequence within the mRNA to which the sense strand of an siRNA shows varying degrees of homology and the antisense strand exhibits varying degrees of complementarity. The phrase â€œsiRNA targetâ€ can refer to the gene, mRNA, or protein against which an siRNA is directed. Similarly, â€œtarget silencingâ€ can refer to the state of a gene, or the corresponding mRNA or protein.

Transfection

The term â€œtransfectionâ€ refers to a process by which agents are introduced into a cell. The list of agents that can be transfected is large and includes, but is not limited to, siRNA, sense and/or anti-sense sequences, DNA encoding one or more genes and organized into an expression plasmid, proteins, protein fragments, and more. There are multiple methods for transfecting agents into a cell including, but not limited to, electroporation, calcium phosphate-based transfections, DEAE-dextran-based transfections, lipid-based transfections, molecular conjugate-based transfections (e.g., polylysine-DNA conjugates), microinjection and others.

The present invention is directed to improving the efficiency of gene silencing by siRNA. Through the inclusion of multiple siRNA sequences that are targeted to a particular gene and/or selecting an siRNA sequence based on certain defined criteria, improved efficiency may be achieved.

The present invention will now be described in connection with preferred embodiments. These embodiments are presented in order to aid in an understanding of the present invention and are not intended, and should not be construed, to limit the invention in any way. All alternatives, modifications and equivalents that may become apparent to those of ordinary skill upon reading this disclosure are included within the spirit and scope of the present invention.

Furthermore, this disclosure is not a primer on RNA interference. Basic concepts known to persons skilled in the art have not been set forth in detail.

According to a first embodiment, the present invention provides a kit for gene silencing, wherein said kit is comprised of a pool of at least two siRNA duplexes, each of which is comprised of a sequence that is complementary to a portion of the sequence of one or more target messenger RNA, and each of which is selected using non-target specific criteria. Each of the at least two siRNA duplexes of the kit complementary to a portion of the sequence of one or more target mRNAs is preferably selected using Formula X.

In one embodiment, the present invention also provides a method wherein said selection criteria are embodied in a formula comprising:

(âˆ’14)*G₁₃âˆ’13*A₁âˆ’12*U₇âˆ’11*U₂âˆ’10*A₁₁âˆ’10*U₄âˆ’10*C₃âˆ’10*C₅âˆ’10*C₆âˆ’9*A₁₀âˆ’9*U₉âˆ’9*C₁₈âˆ’8*G₁₀âˆ’7*U₁âˆ’7*U₁₆âˆ’7*C₁₇âˆ’7*C₁₉+7*U₁₇+8*A₂+8*A₄+8*A₅+8*C₄+9*G₈+10*A₇+10*U₁₈+11*A₁₉+11*C₉+15*G₁+18*A₃+19*U₁₀âˆ’Tmâˆ’3*(GC_total)âˆ’6*(GC_15-19)âˆ’30*X; orâ€ƒâ€ƒFormula VIII:

wherein position numbering begins at the 5â€²-most position of a sense strand, and

A₁=1 if A is the base at position 1 of the sense strand, otherwise its value is 0;

A₂=1 if A is the base at position 2 of the sense strand, otherwise its value is 0;

A₃=1 if A is the base at position 3 of the sense strand, otherwise its value is 0;

A₄=1 if A is the base at position 4 of the sense strand, otherwise its value is 0;

A₅=1 if A is the base at position 5 of the sense strand, otherwise its value is 0;

A₆=1 if A is the base at position 6 of the sense strand, otherwise its value is 0;

A₇=1 if A is the base at position 7 of the sense strand, otherwise its value is 0;

A₁₀=1 if A is the base at position 10 of the sense strand, otherwise its value is 0;

A₁₁=1 if A is the base at position 11 of the sense strand, otherwise its value is 0;

A₁₃=1 if A is the base at position 13 of the sense strand, otherwise its value is 0;

A₁₉=1 if A is the base at position 19 of the sense strand, otherwise if another base is present or the sense strand is only 18 base pairs in length, its value is 0;

C₃=1 if C is the base at position 3 of the sense strand, otherwise its value is 0;

C₄=1 if C is the base at position 4 of the sense strand, otherwise its value is 0;

C₅=1 if C is the base at position 5 of the sense strand, otherwise its value is 0;

C₆=1 if C is the base at position 6 of the sense strand, otherwise its value is 0;

C₇=1 if C is the base at position 7 of the sense strand, otherwise its value is 0;

C₉=1 if C is the base at position 9 of the sense strand, otherwise its value is 0;

C₁₇=1 if C is the base at position 17 of the sense strand, otherwise its value is 0;

C₁₈=1 if C is the base at position 18 of the sense strand, otherwise its value is 0;

C₁₉=1 if C is the base at position 19 of the sense strand, otherwise if another base is present or the sense strand is only 18 base pairs in length, its value is 0;

G₁=1 if G is the base at position 1 on the sense strand, otherwise its value is 0;

G₂=1 if G is the base at position 2 of the sense strand, otherwise its value is 0;

G₈=1 if G is the base at position 8 on the sense strand, otherwise its value is 0;

G₁₀=1 if G is the base at position 10 on the sense strand, otherwise its value is 0;

G₁₃=1 if G is the base at position 13 on the sense strand, otherwise its value is 0;

G₁₉=1 if G is the base at position 19 of the sense strand, otherwise if another base is present or the sense strand is only 18 base pairs in length, its value is 0;

U₁=1 if U is the base at position 1 on the sense strand, otherwise its value is 0;

U₂=1 if U is the base at position 2 on the sense strand, otherwise its value is 0;

U₃=1 if U is the base at position 3 on the sense strand, otherwise its value is 0;

U₄=1 if U is the base at position 4 on the sense strand, otherwise its value is 0;

U₇=1 if U is the base at position 7 on the sense strand, otherwise its value is 0;

U₉=1 if U is the base at position 9 on the sense strand, otherwise its value is 0;

U₁₀=1 if U is the base at position 10 on the sense strand, otherwise its value is 0;

U₁₅=1 if U is the base at position 15 on the sense strand, otherwise its value is 0;

U₁₆=1 if U is the base at position 16 on the sense strand, otherwise its value is 0;

U₁₇=1 if U is the base at position 17 on the sense strand, otherwise its value is 0;

U₁₈=1 if U is the base at position 18 on the sense strand, otherwise its value is 0.

GC_15-19=the number of G and C bases within positions 15-19 of the sense strand, or within positions 15-18 if the sense strand is only 18 base pairs in length;

GC_total=the number of G and C bases in the sense strand;

Tm=100 if the siRNA oligo has the internal repeat longer then 4 base pairs, otherwise its value is 0; and

X=the number of times that the same nucleotide repeats four or more times in a row.

Any of the methods of selecting siRNA in accordance with the invention can further comprise comparing the internal stability profiles of the siRNAs to be selected, and selecting those siRNAs with the most favorable internal stability profiles. Any of the methods of selecting siRNA can further comprise selecting either for or against sequences that contain motifs that induce cellular stress. Such motifs include, for example, toxicity motifs. Any of the methods of selecting siRNA can further comprise either selecting for or selecting against sequences that comprise stability motifs.

In another embodiment, the present invention provides a method of gene silencing, comprising introducing into a cell at least one siRNA selected according to any of the methods of the present invention. The siRNA can be introduced by allowing passive uptake of siRNA, or through the use of a vector.

According to a third embodiment, the invention provides a method for developing an algorithm for selecting siRNA, said method comprising: (a) selecting a set of siRNA; (b) measuring gene silencing ability of each siRNA from said set; (c) determining relative functionality of each siRNA; (d) determining improved functionality by the presence or absence of at least one variable selected from the group consisting of the presence or absence of a particular nucleotide at a particular position, the total number of As and Us in positions 15-19, the number of times that the same nucleotide repeats within a given sequence, and the total number of Gs and Cs; and (e) developing an algorithm using the information of step (d).

In another embodiment, the invention provides a method for selecting an siRNA with improved functionality, comprising using the above-mentioned algorithm to identify an siRNA of improved functionality.

According to a fourth embodiment, the present invention provides a kit, wherein said kit is comprised of at least two siRNAs, wherein said at least two siRNAs comprise a first optimized siRNA and a second optimized siRNA, wherein said first optimized siRNA and said second optimized siRNA are optimized according a formula comprising Formula X.

According to a fifth embodiment, the present invention provides a method for identifying a hyperfunctional siRNA, comprising applying selection criteria to a set of potential siRNA that comprise 18-30 base pairs, wherein said selection criteria are non-target specific criteria, and said set comprises at least two siRNAs and each of said at least two siRNAs contains a sequence that is at least substantially complementary to a target gene; determining the relative functionality of the at least two siRNAs and assigning each of the at least two siRNAs a functionality score; and selecting siRNAs from the at least two siRNAs that have a functionality score that reflects greater than 80 percent silencing at a concentration in the picomolar range, wherein said greater than 80 percent silencing endures for greater than 120 hours.

In other embodiments, the invention provides kits and/or methods wherein the siRNA are comprised of two separate polynucleotide strands; wherein the siRNA are comprised of a single contiguous molecule such as, for example, a unimolecular siRNA (comprising, for example, either a nucleotide or non-nucleotide loop); wherein the siRNA are expressed from one or more vectors; and wherein two or more genes are silenced by a single administration of siRNA.

According to a sixth embodiment, the present invention provides a hyperfunctional siRNA that is capable of silencing Bcl2.

(a) selecting a set of siRNAs;

(b) measuring the gene silencing ability of each siRNA from said set;

(e) developing an algorithm using the information of step (d).

In another embodiment, the present invention provides rationally designed siRNAs identified using the formulas above.

In yet another embodiment, the present invention is directed to hyperfunctional siRNA.

The methods disclosed herein can be used in conjunction with comparing internal stability profiles of selected siRNAs, and designing an siRNA with a desirable internal stability profile; and/or in conjunction with a selection either for or against sequences that contain motifs that induce cellular stress, for example, cellular toxicity.

Any of the methods disclosed herein can be used to silence one or more genes by introducing an siRNA selected, or designed, in accordance with any of the methods disclosed herein. The siRNA(s) can be introduced into the cell by any method known in the art, including passive uptake or through the use of one or more vectors.

Any of the methods and kits disclosed herein can employ either unimolecular siRNAs, siRNAs comprised of two separate polynucleotide strands, or combinations thereof. Any of the methods disclosed herein can be used in gene silencing, where two or more genes are silenced by a single administration of siRNA(s). The siRNA(s) can be directed against two or more target genes, and administered in a single dose or single transfection, as the case may be.

Optimizing siRNA

According to one embodiment, the present invention provides a method for improving the effectiveness of gene silencing for use to silence a particular gene through the selection of an optimal siRNA. An siRNA selected according to this method may be used individually, or in conjunction with the first embodiment, i.e., with one or more other siRNAs, each of which may or may not be selected by this criteria in order to maximize their efficiency.

The degree to which it is possible to select an siRNA for a given mRNA that maximizes these criteria will depend on the sequence of the mRNA itself. However, the selection criteria will be independent of the target sequence. According to this method, an siRNA is selected for a given gene by using a rational design. That said, rational design can be described in a variety of ways. Rational design is, in simplest terms, the application of a proven set of criteria that enhance the probability of identifying a functional or hyperfunctional siRNA. In one method, rationally designed siRNA can be identified by maximizing one or more of the following criteria:

(1) A low GC content, preferably between about 30-52%.

(2) At least 2, preferably at least 3 A or U bases at positions 15-19 of the siRNA on the sense strand.

(3) An A base at position 19 of the sense strand.

(4) An A base at position 3 of the sense strand.

(5) A U base at position 10 of the sense strand.

(6) An A base at position 14 of the sense strand.

(7) A base other than C at position 19 of the sense strand.

(8) A base other than G at position 13 of the sense strand.

(9) A Tm, which refers to the character of the internal repeat that results in inter- or intramolecular structures for one strand of the duplex, that is preferably not stable at greater than 50Â° C., more preferably not stable at greater than 37Â° C., even more preferably not stable at greater than 30Â° C. and most preferably not stable at greater than 20Â° C.

(10) A base other than U at position 5 of the sense strand.

(11) A base other than A at position 11 of the sense strand.

(12) A base other than an A at position 1 of the sense strand.

(13) A base other than an A at position 2 of the sense strand.

(14) An A base at position 4 of the sense strand.

(15) An A base at position 5 of the sense strand.

(16) An A base at position 6 of the sense strand.

(17) An A base at position 7 of the sense strand.

(18) An A base at position 8 of the sense strand.

(19) A base other than an A at position 9 of the sense strand.

(20) A base other than an A at position 10 of the sense strand.

(21) A base other than an A at position 11 of the sense strand.

(22) A base other than an A at position 12 of the sense strand.

(23) An A base at position 13 of the sense strand.

(24) A base other than an A at position 14 of the sense strand.

(25) An A base at position 15 of the sense strand

(26) An A base at position 16 of the sense strand.

(27) An A base at position 17 of the sense strand.

(28) An A base at position 18 of the sense strand.

(29) A base other than a U at position 1 of the sense strand.

(30) A base other than a U at position 2 of the sense strand.

(31) A U base at position 3 of the sense strand.

(32) A base other than a U at position 4 of the sense strand.

(33) A base other than a U at position 5 of the sense strand.

(34) A U base at position 6 of the sense strand.

(35) A base other than a U at position 7 of the sense strand.

(36) A base other than a U at position 8 of the sense strand.

(37) A base other than a U at position 9 of the sense strand.

(38) A base other than a U at position 11 of the sense strand.

(39) A U base at position 13 of the sense strand.

(40) A base other than a U at position 14 of the sense strand.

(41) A base other than a U at position 15 of the sense strand.

(42) A base other than a U at position 16 of the sense strand.

(43) A U base at position 17 of the sense strand.

(44) A U base at position 18 of the sense strand.

(45) A U base at position 19 of the sense strand.

(46) A C base at position 1 of the sense strand.

(47) A C base at position 2 of the sense strand.

(48) A base other than a C at position 3 of the sense strand.

(49) A C base at position 4 of the sense strand.

(50) A base other than a C at position 5 of the sense strand.

(51) A base other than a C at position 6 of the sense strand.

(52) A base other than a C at position 7 of the sense strand.

(53) A base other than a C at position 8 of the sense strand.

(54) A C base at position 9 of the sense strand.

(55) A C base at position 10 of the sense strand.

(56) A C base at position 11 of the sense strand.

(57) A base other than a C at position 12 of the sense strand.

(58) A base other than a C at position 13 of the sense strand.

(59) A base other than a C at position 14 of the sense strand.

(60) A base other than a C at position 15 of the sense strand.

(61) A base other than a C at position 16 of the sense strand.

(62) A base other than a C at position 17 of the sense strand.

(63) A base other than a C at position 18 of the sense strand.

(64) A G base at position 1 of the sense strand.

(65) A G base at position 2 of the sense strand.

(66) A G base at position 3 of the sense strand.

(67) A base other than a G at position 4 of the sense strand.

(68) A base other than a G at position 5 of the sense strand.

(69) A G base at position 6 of the sense strand.

(70) A G base at position 7 of the sense strand.

(71) A G base at position 8 of the sense strand.

(72) A G base at position 9 of the sense strand.

(73) A base other than a G at position 10 of the sense strand.

(74) A G base at position 11 of the sense strand.

(75) A G base at position 12 of the sense strand.

(76) A G base at position 14 of the sense strand.

(77) A G base at position 15 of the sense strand.

(78) A G base at position 16 of the sense strand.

(79) A base other than a G at position 17 of the sense strand.

(80) A base other than a G at position 18 of the sense strand.

(81) A base other than a G at position 19 of the sense strand.

The importance of various criteria can vary greatly. For instance, a C base at position 10 of the sense strand makes a minor contribution to duplex functionality. In contrast, the absence of a C at position 3 of the sense strand is very important. Accordingly, preferably an siRNA will satisfy as many of the aforementioned criteria as possible.

With respect to the criteria, GC content, as well as a high number of AU in positions 15-19 of the sense strand, may be important for easement of the unwinding of double stranded siRNA duplex. Duplex unwinding has been shown to be crucial for siRNA functionality in vivo.

With respect to criterion 9, the internal structure is measured in terms of the melting temperature of the single strand of siRNA, which is the temperature at which 50% of the molecules will become denatured. With respect to criteria 2-8 and 10-11, the positions refer to sequence positions on the sense strand, which is the strand that is identical to the mRNA.

In one preferred embodiment, at least criteria 1 and 8 are satisfied. In another preferred embodiment, at least criteria 7 and 8 are satisfied. In still another preferred embodiment, at least criteria 1, 8 and 9 are satisfied.

It should be noted that all of the aforementioned criteria regarding sequence position specifics are with respect to the 5â€² end of the sense strand. Reference is made to the sense strand, because most databases contain information that describes the information of the mRNA. Because according to the present invention a chain can be from 18 to 30 bases in length, and the aforementioned criteria assumes a chain 19 base pairs in length, it is important to keep the aforementioned criteria applicable to the correct bases.

When there are only 18 bases, the base pair that is not present is the base pair that is located at the 3â€² of the sense strand. When there are twenty to thirty bases present, then additional bases are added at the 5â€² end of the sense chain and occupy positions ^âˆ’1 to ^âˆ’11. Accordingly, with respect to SEQ. ID NO. 0001 NNANANNNNUCNAANNNNA and SEQ. ID NO. 0028 GUCNNANANNNNUCNAANNNNA, both would have A at position 3, A at position 5, U at position 10, C at position 11, A and position 13, A and position 14 and A at position 19. However, SEQ. ID NO. 0028 would also have C at position âˆ’1, U at position âˆ’2 and G at position âˆ’3.

For a 19 base pair siRNA, an optimal sequence of one of the strands may be represented below, where N is any base, A, C, G, or U:

SEQ.â€ƒIDâ€ƒNO.â€ƒ0001.NNANANNNNUCNAANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0002.NNANANNNNUGNAANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0003.NNANANNNNUUNAANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0004.NNANANNNNUCNCANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0005.NNANANNNNUGNCANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0006.NNANANNNNUUNCANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0007.NNANANNNNUCNUANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0008.NNANANNNNUGNUANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0009.NNANANNNNUUNUANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0010.NNANCNNNNUCNAANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0011.NNANCNNNNUGNAANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0012.NNANCNNNNUUNAANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0013.NNANCNNNNUCNCANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0014.NNANCNNNNUGNCANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0015.NNANCNNNNUUNCANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0016.NNANCNNNNUCNUANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0017.NNANCNNNNUGNUANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0018.NNANCNNNNUUNUANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0019.NNANGNNNNUCNAANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0020.NNANGNNNNUGNAANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0021.NNANGNNNNUUNAANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0022.NNANGNNNNUCNCANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0023.NNANGNNNNUGNCANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0024.NNANGNNNNUUNCANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0025.NNANGNNNNUCNUANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0026.NNANGNNNNUGNUANNNNA SEQ.â€ƒIDâ€ƒNO.â€ƒ0027.NNANGNNNNNUNUANNNNA

In one embodiment, the sequence used as an siRNA is selected by choosing the siRNA that score highest according to one of the following seven algorithms that are represented by Formulas I-VII:

Relative functionality of siRNA=âˆ’(GC/3)+(AU_15-19)âˆ’(Tm_{20Â° C.})*3âˆ’(G₁₃)*3âˆ’(C₁₉)+(A₁₉)*2+(A₃)+(U₁₀)+(A₁₄)âˆ’(U₅)âˆ’(A₁₁)â€ƒâ€ƒFormula I

Relative functionality of siRNA=âˆ’(GC/3)âˆ’(AU_15-19)*3âˆ’(G₁₃)*3âˆ’(C₁₉)+(A₁₉)*2+(A₃)â€ƒâ€ƒFormula II

Relative functionality of siRNA=âˆ’(GC/3)+(AU_15-19)âˆ’(Tm_{20Â° C.})*3â€ƒâ€ƒFormula III

Relative functionality of siRNA=âˆ’GC/2+(AU_15-19)/2âˆ’(Tm_{20Â° C.})*2âˆ’(G₁₃)*3âˆ’(C₁₉)+(A₁₉)*2+(A₃)+(U₁₀)+(A₁₄)âˆ’(U₅)âˆ’(A₁₁)â€ƒâ€ƒFormula IV

Relative functionality of siRNA=âˆ’(G₁₃)*3âˆ’(C₁₉)+(A₁₉)*2+(A₃)+(U₁₀)+(A₁₄)âˆ’(U₅)âˆ’(A₁₁)â€ƒâ€ƒFormula V

Relative functionality of siRNA=âˆ’(G₁₃)*3âˆ’(C₁₉)+(A₁₉)*2+(A₃)â€ƒâ€ƒFormula VI

Relative functionality of siRNA=âˆ’(GC/2)+(AU_15-19)/2âˆ’(Tm_{20Â° C.})*1âˆ’(G₁₃)*3âˆ’(C₁₉)+(A₁₉)*3+(A₃)*3+(U₁₀)/2+(A₁₄)/2âˆ’(U₅)/2âˆ’(A₁₁)/2â€ƒâ€ƒFormula VII

In Formulas I-VII:

wherein A₁₉=1 if A is the base at position 19 on the sense strand, otherwise its value is 0,

AU_15-19=0-5 depending on the number of A or U bases on the sense strand at positions 15-19;

G₁₃=1 if G is the base at position 13 on the sense strand, otherwise its value is 0;

C₁₉=1 if C is the base at position 19 of the sense strand, otherwise its value is 0;

GC=the number of G and C bases in the entire sense strand;

Tm_{20Â° C.}=1 if the Tm is greater than 20Â° C.;

A₃=1 if A is the base at position 3 on the sense strand, otherwise its value is 0;

U₁₀=1 if U is the base at position 10 on the sense strand, otherwise its value is 0;

A₁₄=1 if A is the base at position 14 on the sense strand, otherwise its value is 0;

U₅=1 if U is the base at position 5 on the sense strand, otherwise its value is 0; and

A₁₁=1 if A is the base at position 11 of the sense strand, otherwise its value is 0.

Formulas I-VII provide relative information regarding functionality. When the values for two sequences are compared for a given formula, the relative functionality is ascertained; a higher positive number indicates a greater functionality. For example, in many applications a value of 5 or greater is beneficial.

Additionally, in many applications, more than one of these formulas would provide useful information as to the relative functionality of potential siRNA sequences. However, it is beneficial to have more than one type of formula, because not every formula will be able to help to differentiate among potential siRNA sequences. For example, in particularly high GC mRNAs, formulas that take that parameter into account would not be useful and application of formulas that lack GC elements (e.g., formulas V and VI) might provide greater insights into duplex functionality. Similarly, formula II might by used in situations where hairpin structures are not observed in duplexes, and formula IV might be applicable for sequences that have higher AU content. Thus, one may consider a particular sequence in light of more than one or even all of these algorithms to obtain the best differentiation among sequences. In some instances, application of a given algorithm may identify an unusually large number of potential siRNA sequences, and in those cases, it may be appropriate to re-analyze that sequence with a second algorithm that is, for instance, more stringent. Alternatively, it is conceivable that analysis of a sequence with a given formula yields no acceptable siRNA sequences (i.e. low SMARTSCORESâ„¢, or siRNA ranking). In this instance, it may be appropriate to re-analyze that sequences with a second algorithm that is, for instance, less stringent. In still other instances, analysis of a single sequence with two separate formulas may give rise to conflicting results (i.e. one formula generates a set of siRNA with high SMARTSCORESâ„¢, or siRNA ranking, while the other formula identifies a set of siRNA with low SMARTSCORESâ„¢, or siRNA ranking). In these instances, it may be necessary to determine which weighted factor(s) (e.g. GC content) are contributing to the discrepancy and assessing the sequence to decide whether these factors should or should not be included. Alternatively, the sequence could be analyzed by a third, fourth, or fifth algorithm to identify a set of rationally designed siRNA.

The above-referenced criteria are particularly advantageous when used in combination with pooling techniques as depicted in Table I:

TABLE IFUNCTIONAL PROBABILITYOLIGOSPOOLSCRITERIA>95%>80%<70%>95%>80%<70%CURRENT33.050.023.079.597.30.3NEW50.088.58.093.899.980.005(GC)28.058.936.072.897.11.6

The term â€œcurrentâ€ used in Table I refers to Tuschl's conventional siRNA parameters (Elbashir, S. M. et al. (2002) â€œAnalysis of gene function in somatic mammalian cells using small interfering RNAsâ€ Methods 26: 199-213). â€œNewâ€ refers to the design parameters described in Formulas I-VII. â€œGCâ€ refers to criteria that select siRNA solely on the basis of GC content.

As Table I indicates, when more functional siRNA duplexes are chosen, siRNAs that produce <70% silencing drops from 23% to 8% and the number of siRNA duplexes that produce >80% silencing rises from 50% to 88.5%. Further, of the siRNA duplexes with >80% silencing, a larger portion of these siRNAs actually silence>95% of the target expression (the new criteria increases the portion from 33% to 50%). Using this new criteria in pooled siRNAs, shows that, with pooling, the amount of silencing>95% increases from 79.5% to 93.8% and essentially eliminates any siRNA pool from silencing less than 70%.

Table II similarly shows the particularly beneficial results of pooling in combination with the aforementioned criteria. However, Table II, which takes into account each of the aforementioned variables, demonstrates even a greater degree of improvement in functionality.

TABLE IIFUNCTIONAL PROBABILITYOLIGOSPOOLSNON-NON-FUNCTIONALAVERAGEFUNCTIONALFUNCTIONALAVERAGEFUNCTIONALRANDOM20405067973CRITERIA 152990.197930.0040CRITERIA 489990.199990.0000

The terms â€œfunctional,â€ â€œAverage,â€ and â€œNon-functionalâ€ used in Table II, refer to siRNA that exhibit >80%, >50%, and <50% functionality, respectively. Criteria 1 and 4 refer to specific criteria described above.

The above-described algorithms may be used with or without a computer program that allows for the inputting of the sequence of the mRNA and automatically outputs the optimal siRNA. The computer program may, for example, be accessible from a local terminal or personal computer, over an internal network or over the Internet.

In addition to the formulas above, more detailed algorithms may be used for selecting siRNA. Preferably, at least one RNA duplex of 18-30 base pairs is selected such that it is optimized according a formula selected from:

(âˆ’14)*G₁₃âˆ’13*A₁âˆ’12*U₇âˆ’11*U₂âˆ’10*A₁₁âˆ’10*U₄âˆ’10*C₃âˆ’10*C₅âˆ’10*C₆âˆ’9*A₁₀âˆ’9*U₉âˆ’9*C₁₈âˆ’8*G₁₀âˆ’7*U₁âˆ’7*U₁₆âˆ’7*C₁₇âˆ’7*C₁₉+7*U₁₇+8*A₂+8*A₄+8*A₅+8*C₄+9*G₈+10*A₇+10*U₁₈+11*A₁₉+11*C₉+15*G₁+18*A₃+19*U₁₀âˆ’Tmâˆ’3*(GC_total)âˆ’6*(GC_15-19)âˆ’30*X; andâ€ƒâ€ƒFormula VIII:

(14.1)*A₃+(14.9)*A₆+(17.6)*A₁₃+(24.7)*A₁₉+(14.2)*U₁₀+(10.5)*C₉+(23.9)*G₁+(16.3)*G₂+(âˆ’12.3)*A₁₁+(âˆ’19.3)*U₁+(âˆ’12.1)*U₂+(âˆ’11)*U₃+(âˆ’15.2)*U₁₅+(âˆ’11.3)*U₁₆+(âˆ’11.8)*C₃+(âˆ’17.4)*C₆+(âˆ’10.5)*C₇+(âˆ’13.7)*G₁₃+(âˆ’25.9)*G₁₉âˆ’Tmâˆ’3*(GC_total)âˆ’6*(G_15-19)âˆ’30*X; andâ€ƒâ€ƒFormula IX:

wherein