WO2005038054A1 - METHOD OF MEASURING THE EFFICACY OF siRNA MOLECULES - Google Patents

Abstract

Only a small fraction of all siRNAs are effective in silencing their target genes, and siRNA efficacy can only be determined experimentally. Previously described reporter based siRNA validation methods all rely on the availability of physical cDNA clones, and this limits the high throughput applicability of the method. In the current application, we use short synthetic DNA fragment containing a siRNA targeting site to fuse with a reporter gene. When targeting such transcripts with different siRNAs, we found that such constructs can faithfully report the efficacy of the corresponding siRNAs in a sequence specific manner even when the inserted DNA fragment is essentially only long enough to cover the targeting site. The efficacy of both vector based siRNA and synthetic siRNA can be evaluated using this system. Since only readily available short synthetic DNA fragments are needed for forming the evaluation vector, this method provides an appealing way of validating siRNAs in high throughput.

Description

Method of measuring the efficacy of siRNA molecules .

BACKGROUND OF THE INVENTION

This invention relates to a novel method of identification of nucleic acid molecules which are useful for modulation of gene expression.

Since the discovery of the mechanisms underlying gene expression, specifically nucleic acid based transcription and translation, a great deal of effort has been placed on blocking or altering these processes for a variety of purposes, such as understanding biology, gene function, disease processes, and identifying novel therapeutic targets. Approaches involving nucleic acid molecules for modulating gene expression have gained popularity in recent years. RNA interference (RNAi) is a form of sequence-specific, post- transcriptional gene silencing in animals and plants, elicited by double-stranded RNA (dsRNA) that is homologous in sequence to the silenced gene [1-8]. dsRNA triggers the specific degradation of homologous RNAs, only which starts within the region of homology. The dsRNA is processed to 21- to 23-nucleotide fragments, sometimes called short interfering RNAs (siRNAs) which are believed to be the guide fragments for sequence- specific mRNA degradation. The processing of longer dsRNA to these short siRNA fragments is believed to be accomplished by RNase III.

RNAi and siRNA has become widely adapted for deciphering gene functions. Successful implementation of this method (in the form of siRNA produced chemically or from vectors) for gene silencing in mammalian cells represents an important step in approaching the daunting task of functional annotation of the human genomes, and it also opened new prospects for gene therapy [1-8].

Only a small fraction of all siRNAs are effective in silencing their target genes, and siRNA efficacy can only be determined experimentally. Previously described reporter based siRNA validation methods all rely on the availability of physical cDNA clones, and this limits the high throughput applicability of the method.

While perfectly suited for large-scale functional genomics, it is also true that only a small fraction of all siRNA designed so far are efficacious. People has started to uncover some mechanisms that might have impact on the efficacy of siRNA [9], but until such mechanisms are more thoroughly investigated, researchers have to rely on experimental tools to assess the efficacy of each siRNA. Due to the excessive labour and cost for carrying out large-scale siRNA validation using existing methods, the most feasible alternative is using redundant siRNAs without validating their silencing efficacies [10]. This would be translated into huge waste of time and resources. Despite the number of reports documenting siRNA-induced silencing, there remain significant challenges to the successful implementation of this technology for large-scale gene silencing which is essential for mammalian functional genomics, as only a limited fraction of siRNA appear capable of producing highly effective RNAi

Although several studies suggested that siRNA might have sequence preferences [11-17], and this has resulted in different algorithms that can help researchers to filter out siRNAs that might be active statistically, such software does not provide definite indication about the efficacy of any individual siRNA, not to mention the silencing strength of the reagent which is of critical importance for interpretation of knockdown experiments. In practice, three or more siRNAs are normally chosen (or designed by computer program) for each gene to be studied, and each siRNA is experimentally tested against its natural target (or fusion gene including full length cDNA) in order to identify a good siRNA. The suppression activities of siRNA are usually evaluated by methods including quantitative PCR, northern blot and western blot methods, which are still very laborious. Taking into consideration the unpredictability of various delivery methods and the distinct properties of different experimental systems, results are usually complicated and difficult to interpret. It will become even more complicated for animal-based experiments and large-scale gene function studies, when the experimental conditions needed to be optimized gene by gene. To validate siRNAs for each gene, the experimental process and conditions have to be optimized individually. This imposes a serious restraint on the throughput of any large-scale siRNA validation effort.

A new type of siRNA validation tool has been created to circumvent the difficulties of siRNA efficacy evaluation. In this approach, long target cDNA fragments were fused into different reporter genes [17-18]. It was found that siRNA attacking on the target could result in efficient silencing of the reporter gene by disrupting the open reading frame. The rationale behind using full length or partial cDNA clones in these approach was partially that the siRNA silencing on a target gene may be affected by the sequence or secondary structure of the mRNA around the targeting site. These elegant methods, however, are all subject to the limitations of the availability of physical clones of full- length (or even partial) cDNA. The high level of error rate of IMAGE clones calls for the sequencing of any clone obtained, which is time consuming and costly. Sequence-verified clones still remain to be prohibitively expensive for individual users, although the situation has been improved for mouse cDNAs. Additionally, such methods were configured to work only with siRNA targeting the coding region of mRNA. Non-coding regions can not be studied by these methods.

Thus, there exists a need in the art for a method for fast and simple validation of the gene silencing efficacy of a siRNA molecule that is applicable to coding and non-coding regions. FIELD OF THE INVENTION

The present invention relates to a method for validating the activity of a siRNA in a cell based assay and to means for carrying out this method.

SUMMARY OF THE INVENTION

To overcome the limitations found in the art, we have developed an oligonucleotide-based siRNA validation methodology by using the dual-luciferase reporter system. Taking advantage of enzymatic-based dual luciferase reporter system, our validation assay exhibits great accuracy and excellent internal control for various experimental factors, including variance of plated cells, transfection efficiencies and translation efficiencies.

In one aspect, this invention relates to a method of identification of nucleic acid molecules which are useful for modulation of gene expression.

In a further aspect, this invention relates to DNA molecules, such as plasmids, viral vectors and linear DNA molecules, that are used to evaluate a small interfering RNA molecule (siRNA) for its capacity to modulate gene expression through a variety of mechanisms.

In a further aspect, this invention relates to a cell carrying and transcribing a DNA molecule according to the second aspect

One embodiment of the invention is a method to evaluate the efficacy of a siRNA molecule to suppress a gene using plasmids with a synthetic nucleic acid fragment, substantially complementary to the siRNA, inserted in front of a reporter gene for measurement of siRNA attack on the transcripts from the plasmids. This siRNA attack can be visualized by measuring the activity of the reporter gene product.

A further embodiment of the invention is a method of measuring the gene silencing efficacy of a siRNA molecule comprising the steps bringing the siRNA molecule in contact with a cell expressing an RNA molecule comprising a siRNA target site of 19-38 nucleotides and a reporter gene, directly or indirectly measuring the translation of the reporter gene wherein a decrease in translation of reporter gene indicates the efficacy of the siRNA molecule.

A further embodiment of the current invention is a method comprising the use of a synthetic DNA fragment of 15-40, preferably 19-38, base pairs in length that has a sequence that is identical to the sequence of siRNA, inserted in-frame in the 5' end of coding sequence of the reporter gene in a plasmid, and the use of such a plasmid to measure the activity of the corresponding siRNA by measuring the activity of the reporter gene. The DNA fragment may also be inserted out-of- frame and/ or in the 3 '-end of the coding region.

In yet another embodiment of the current invention, when the synthetic DNA fragments of different length are inserted out-of-frame in the 5' end of the reporter, an in-frame stop codon and an in-frame ATP (translation start site) are inserted between the synthetic DNA fragment and the coding sequence of the reporter.

In one embodiment of the current invention, the reporter gene used is firefly luciferase. It is further recognized that other reporter genes such as Renilla luciferase, beta-galactosidase, phosphatase and various fluorescent proteins can also be used as the reporter gene in the current method. In one embodiment of the invention, a synthetic DNA fragment containing the targeting sites of a single siRNA is inserted into a plasmid in the same or a similar way as mentioned above. In this and all other embodiments, the term "targeting site" is used in singular or plural form to refer to a DNA fragment that contains a segment of the sequence that is identical to the sequence of a siRNA.

In another embodiment of the invention, multiple DNA fragments containing targeting sites of different siRNA are inserted into the plasmid in the same or a similar way as described above.

In one embodiment of the current invention, the synthetic DNA fragment containing the siRNA targeting site and the reporter gene are cloned within a plasmid. It is also recognized that these elements can be cloned into a viral vector or a linear DNA fragment (PCR product for example) to be used for the same purpose.

In all the embodiments listed above, the transcription of the synthetic DNA fragment containing the siRNA targeting site and the reporter gene is controlled by a promoter that can drive the transcription of genes in eukaryotic cells, and under the control of such a promoter the synthetic DNA fragment containing the siRNA targeting site and the reporter gene are transcribed as a single transcript. Such a transcript in turn can produce a functional reporter protein, the activity of which can be measured. A Cauliflower Mosaic Virus (CMV) promoter is one of the preferred promoters. Other RNAse II promoters may also be used.

Further embodiments are described in the detailed description and in the appended claims.

BRIEF DESCRIPTION OF THE FIGURES Figure 1.

The synthetic DNA fragment (hatched box) is cloned either at the 5' or the 3' end of the coding sequence of the reporter gene, firefly luciferase in this case, (black box). Dotted box depicts part of the same transcript that are not translated into proteins (5' and 3' UnTranslated Regions, UTR). Black lines attached to the 5' UTR and 3' UTR depicts the plasmid vector used for the propagation and expression of this fusion gene. A promoter (arrow) for driving gene expression in eukaryotic cells is placed close to the 5' UTR in the plasmid to drive the expression of the fusion gene. In Figure 1A, a ATG start codon is place immediately after the end of the 5' UTR. In case the siRNA targeting site (siRNA targeting sequence) is not in frame with the firefly luciferase coding sequence, a second in-frame ATG start codon was placed at the very beginning of the firefly coding sequence (Black box). In Figure IB, a single in-frame ATG was placed at the beginning of the firefly coding sequence.

Figure 2.

Example of two DNA fragments (19 and 38 base pairs respectively, in bold letters) that harbors a siRNA targeting site (marked with + at the bottom) that was cloned in frame with the firefly luciferase coding sequence siRNA targeting sites. The frame of the firefly luciferase was marked segregating the nucleotides in triplets. Between the first ATG and the inserted DNA fragment several nucleotides were included as cloning sites. Immediately after the inserted synthetic DNA fragment there are also four nucleotide that serve as cloning site.

Figure 3.

A: Two siRNA were used to treat cells transfected with plasmid vectors that have synthetic siRNA targeting site -containing DNA fragment of different length (19mer =19 base pairs and 38mer =38 base pairs) inserted in-frame at the 5' end of the firefly luciferase coding sequence. The luciferase activity is then compared with cells that are not treated with any siRNA. This result demonstrated that a siRNA targeting site does not need to be in its authentic location in order to serve as a targeting site of a siRNA. When a siRNA targeting site (no other sequences) is grafted from its authentic mRNA sequence into an artificial location, for example into a reporter gene, the site still serves as a targeting site for the same siRNA. This discovery eliminates the need of using any cDNA in making such siRNA activity reporter, and thus enabled the construction of reporter construct to measure the activity of siRNA of any given sequence by chemically synthesizing its targeting site and insert into reporter vector according to what is specified herein.

B: Transfection of HEK293 cells with expression vector containing 19mer target site of CD46-604 (target construct) and another 19mer fragment from the same gene (irrelevant control) and treatment with CD46-604 shows significant reduction of luciferase gene expression.

Figure 4.

Cross test if siRNA validation method using the vectors constructed by cloning synthetic DNA fragment fused to the coding sequence of reporter gene (firefly luciferase) (A) and using northern blot analysis using endogenous mRNA as the target (B) using the same siRNAs as gene silencing reagent. This result has demonstrated that the readout of the activity of a siRNA reflects the activity of the siRNA on authentic genes. In this experiment, two siRNA showed significant silencing activity on the reporter gene that is fused with a synthetic DNA fragment that contains the siRNA targeting site. These siRNAs do not have any gene silencing activity towards any fusion constructs that does not have sequence identity to the siRNA. One of the tested siRNA has very weak gene silencing activity on reporter gene that contain a fused targeting site. This siRNA efficacy assignment was confirmed nicely by using Northern blot analysis.

Figure 5 The siRNA validation system described in the current invention is sequence specific and one nucleotide mutation on the DNA fragment harboring the siRNA targeting site, which was inserted in frame with a reporter gene, can render the mutated site resistant to a siRNA that is effective to wild-type sequence. A, comparison of the wild-type sequence of the siRNA targeting site (upper underlined sequence Fas297, targeting site underlined) and the mutated sequence (lower underlined sequence Fas297m, mutated nucleotide marked with an arrow). B, the silencing efficacy of siRNA that has an identical sequence to the wild type targeting site on plasmids that have either the wild-type targeting site or the mutated targeting site fused with the reporter gene. The result indicated that this siRNA has potent silencing activity on the wild-type targeting site but does not have a significant silencing activity on the site that have just one nucleotide mutation compared to the wild-type targeting site.

Figure 6.

When the short synthetic DNA fragment is fused with the reporter gene, there is a choice to fuse them in-frame, so that the inserted short synthetic DNA fragment will be transcribed into part of the coding sequence of the resulting protein from the fusion gene, or it can be fused in such way that the inserted DNA fragment is transcribed into the un-translated region of the resulting transcript from the fusion gene. In part A of Figure 6, three constructs are shown. In the in-frame construct, translation starts from the first ATG-codon. The out-of-frame construct has a a C-→G mutation inserting a Stop codon TGA, making the transcription start at the second ATG-codon. The first two types of fusion constructs were treated with the same siRNA, and it was found that both types can be silenced by the siRNA but the siRNA has a stronger silencing effect on the reporter that harbors the in-frame construct. A frame-control vector was included to control for the change of nucleotide at the Stop codon location so as to confirm that it is really the stop codon, not the nucleotide change at that location that has changed the response of the fusion construct to the siRNA. This result testified that this method can be used to evaluate the activity of a siRNA even if its sequence properties will make it impossible to fuse the targeting site in-frame with the reporter gene.

Figure 7.

The siRNA validation method described in the current invention was also used to look into the activity of siRNA towards synthetic fragments that contains mismatches (mismatched targeting sites) to the siRNA sequence because natural occurring small interfering RNA like molecules, so called microRNA, are actually working in such a manner. A, in this figure synthetic DNA fragment containing a siRNA targeting site that is matching perfectly with the siRNA (wild-type) and three mutated DNA fragment containing different "mismatched targeting sites" were fused to the reporter gene, as examples, as shown in the diagram, B, additionally all 56 mismatched target sites with one mismatch to the wild-type siRNA targeting site were fused to the reporter gene in the same manner as shown in A, as additional examples, for the method to evaluate the siRNA effect on mismatched targeting sites. With these examples successfully assessed, we also have constructed and tested mismatched targeting sites that contain more than one mismatches with the siRNA targeting site that is perfectly matching the sequence of the siRNA. In the sequences, in figure 7B, V stands for "A, G, or C, but not T", B stands for "G, C, or T, but not A", D stands for "A, G, or T but not C", H stands for "A, C, or T, but not G".

Figure 8. The silencing activity of siRNA on reporter genes fused with its wild type and mutated targeting sites. The 19-mer nucleotide sequence of the wild type siRNA targeting site is listed below the x-axes. At each position of the sequence from 1 to 19, the site was mutated into all three additional nucleotides and the mutated targeting sites were fused with the reporter. Then the activity of the wild-type siRNA for silencing each mutated sites was measured. The activity of the siRNA is then plotted according to the nature of mutation (A, G, T, C) and position (1-19). The horizontal dimension of the diagram showed four- bar groups numbered from 1 to 19 according to the location of the site of introduced mismatches in the mismatched targeting sites, and in the vertical dimension the remaining reporter activities were plotted with 0,00 indicating complete silencing and 1,00 no silencing activity at all (or in other words, 100% of the reporter activity remains). This result demonstrated that the siRNA validation system described in the current invention can also be used to evaluate siRNA towards targets that have mismatches vis-a-vis the sequence of the siRNA.

DEFINITIONS

The term "siRNA" as used herein, refers to small interfering RNA, which is double stranded RNA or RNA analogs, normally 19-21 base pairs in length but could be as long as 30 base pairs, and may or may not have single stranded overhangs of up to 5 nucleotides on its ends, that are formed either by chemical enzymatic synthesis out of cells or from different plasmid or viral vectors by cellular enzymes inside the cells. Some of the siRNA have a stronger activity in eliciting degradation of RNA than similar or identical sequences while some other siRNA have weaker or none such activities.

The terms "siRNA targeting site" and "siRNA target region" as used herein, refers to that portion/ region of a RNA that can be recognized, bound, and cleaved by a siRNA molecule. The siRNA targeting site is 15-50 nucleotide in length. Usually, the siRNA targeting site has about 19-30 nucleotides base pairs identical to the actual siRNA. There may however be some mismatches between the siRNA and the siRNA targeting site, as described below. The terms "silencing efficacy", "efficacy" and "silentivity" are used interchangeably in this paper and mean the capability of a siRNA molecule to silence the production of a gene product.

The term "in-frame" is intended to mean inside the Open Reading Frame, ORF. That is, a sequence positioned in-frame is within an ORF. Analogously, "out-of-frame" or "out-of-ORF" means that the sequence is not inside an ORF.

The term "microRNA" as used herein, refers to a molecule comprising two or more nucleotides. "miRNA" is a short form for microRNA.

The term "RNA" as used herein, refers to a molecule comprising at least one ribonucleotide residue. By "ribonucleotide" or "2'-OH" is meant a nucleotide with a hydroxyl group at the 2' position of a β-D- ribo-furanose moiety.

The term "nucleic acid molecule" as used herein, refers to a molecule having nucleotides. The nucleic acid can be single, double, or multiple stranded and can comprise modified or unmodified nucleotides or non-nucleotides or various mixtures and combinations thereof.

The term "gene" as used herein, refers to a nucleic acid that encodes an RNA, for example, nucleic acid sequences including but not limited to structural genes encoding a polypeptide.

The term "mismatched sequences" means that the sequences are similar but not identical. Typically, the sequences differ in 1, 2, 3 or 4 positions.

DETAILED DESCRIPTION OF THE INVENTION The current invention present a method, where a short RNA fragment of 15 to 50 nt is fused to the mRNA of a marker protein firefly luciferase by introducing a synthetic DNA duplex of substantially the same length to the 5' side of the translation start codon of the firefly luciferase mRNA, that uses such a chimeric construct to demonstrate the activity of the siRNA. Difference of the current method from the prior art is that the current method eliminated the need for cDNA or in-frame fusion, thus made it possible to use synthetic DNA oligos as the sole input to get an assessment of effectiveness of the siRNA.

The current invention involves the fusion of a synthetic DNA duplex 15-50 bp in length upstream of a functional translation start codon of the firefly luciferase cDNA and place the fusion construct down stream of an eukaryotic transcription promoter in a plasmid vector. Upon transfection of the plasmid into eukaryotic cells, a mRNA covering both the firefly luciferase part and the synthetic region will be produced in cells, and functional luciferase protein can be produced in turn from the mRNA. If a cleavage is introduced in the mRNA part covering the synthetic oligos, for example by siRNA induced cleavage, the translation start codon of the firefly luciferase will be quickly destroyed by the internal RNA clearance system due to the closeness of the translation start codon to the cleavage site. In such a case, no functional luciferase will be produced.

In such a way, by measuring the level of luciferase activity, we can precisely measure the extent by which a siRNA can degrade its target mRNA.

Compared the existing methods, this method will not need to have a full or partial cDNA prepared for the target gene, a short synthetic duplex will be enough. Further, since the method relies not on the interruption of the elongation process of the translation, but the translation initiation by destroying the translation start codon, the inserted sequence does not need to be fused to the marker gene in- frame. Rather, the synthetic sequence can be in any frames, and even with stop codons. This advantage make the method suitable for assess the activities of siRNA targeting not only the coding sequences, but also the un-translated regions. This provides great flexibility and robustness to the method.

The method is validated initially with firefly luciferase as the marker gene, but the mechanism should be applicable to any marker gene.

Proof of concepts 1) Insertion of a duplex with a in-frame stop codon in front of the firefly luciferase cDNA translation start codon does not interference with the luciferase activity of the plasmid as showed in transfection assay. 2) insertion of a 50mer duplex in front of the firefly luciferase cDNA translation start codon does not interference with the luciferase activity of the plasmid as showed in transfection assay. 3) insertion of a 36mer duplex in front of the firefly luciferase cDNA translation start codon does not interference with the luciferase activity of the plasmid as showed in transfection assay. 4) insertion of a 23 mer duplex infront of the firefly luciferase cDNA translation start codon does not interference with the luciferase activity of the plasmid as showed in transfection assay. 5) insertion of a 19mer duplex in front of the firefly luciferase cDNA translation start codon does not interference with the luciferase activity of the plasmid as showed in transfection assay. 6) Effective siRNA, for which the efficacy has been demonstrated in on authentic genes epitopically and endogenously expressed, targeting the corresponding sequence in the inserted region can efficiently knock-down the luciferase activity. 7) Non-effective siRNA, for which the lack of efficacy has been demonstrated in on authentic genes epitopically and endogenously expressed, targeting the corresponding sequence in the inserted region does not induce efficient knock-down the luciferse activity. 8) Irrelevant siRNA does not inhibit the activity of the luciferase. 9) Experiment showed that when a siRNA is effective in targeting the corresponding sequence in the inserted region, it will in the same time induce the degradation of the translation start codon of the luciferase region. 10) Effective siRNA, for which the efficacy has been demonstrated by targeting the corresponding sequence in the inserted region and measuring the knock-down of luciferase activity, were validated on authentic genes epitopically and endogenously expressed and find to be effective.

11) Non-effective siRNA, for which the lack of efficacy has been demonstrated by targeting the corresponding sequence in the inserted region and measuring the knockdown of luciferase activity, were validated on authentic genes epitopically and endogenously expressed and find to be non-effective.

The above results demonstrated that the activities of siRNA against their authentic gene epitopically expressed, or endogenously expressed, can be reflected precisely by their activities demonstrated by targeting a short fragment of 19-50 bp, which contains the targeting site of the siRNA, inserted to the 5' side of the translation start codon of a marker gene, firefly luciferase. With this method, then we can precisely predict the activity of an individual siRNA on its authentic target by the activity it demonstrated when targeting a site formed by short synthetic oligonucleotides. In the art, complementary DNA (cDNA) of length that are much longer than the length of the targeting sites of a siRNA (normallyl5-40 bp), generated in a living organism or generated by enzymatic reaction, has been fused with reporter genes for serving as the target of siRNA (17- 18). In contrast, the current invention discloses the integration of chemically synthesized oligonucleotides containing essentially the targeting sites of a siRNA for the purpose of measuring siRNA activity.

Compared to the art, the current invention is novel in the following aspects: 1) the current invention demonstrated that when the targeting site of a siRNA is isolated from its authentic sequence environment (or secondary structure), and inserted into a new environment (or secondary structure), it can still be found and silenced by the original siRNA. In contrast, the insertion of cDNA into the reporter gene does not introduce secondary change into the targeting site and its immediate nucleotide neighbours (normally 50- 100 base bases around a site). 2) The current invention further demonstrated that the silencing efficacy of siRNA on such targeting site of a siRNA, that is isolated from its authentic sequence environment (or secondary structure), and inserted into a new environment (or secondary structure), is indicative of the siRNA efficacy on the target gene that contain the same targeting site. Because of these novel aspects, it has for the first time enabled people to incorporate a short synthetic DNA fragment as short as 19 base pairs into a reporter and use the reporter to assess the effectiveness of siRNA targeting sites with identical sequences.

In order to carry out siRNA validation more robustly and cost effectively, we developed a reporter-based siRNA validation system that eliminated the need for cDNA clones. In this system, short synthetic DNA oligonucleotides fragments corresponding to the target sites of tested siRNA were inserted into a reporter plasmid referred to as a siQuant construct, and the fusion construct is used directly for validating the siRNA efficacy. When targeting such transcripts with different siRNAs, we demonstrated that siRNA silencing efficacy on the fusion constructs reflects the efficacy on its endogenous target genes siRNAs in a sequence specific manner even when the inserted DNA fragment is essentially only long enough to cover the targeting site. This paves the way for genome-wide siRNA validation in a high throughput manner in any molecular biology laboratory. Since only readily available short synthetic DNA fragments are needed for forming the evaluation vector, this method provides an appealing way of validating siRNAs in high throughput .

Examples

Fusion of short synthetic DNA fragment and reporter gene and use of the fusion constructs for validation of the siRNA efficacy

Rationale of the efficacy assay

It has been well elucidated that RNAi is a sequence-dependent expression-silencing phenomenon. In theory, two types of factors could contribute to siRNA activity: a) sequence properties of the siRNA (the intrinsic gene silencing properties of the siRNA), and b) the environmental factors. Although several reports suggested that the secondary structures of mRNA and mRNA binding proteins might interfere with the target site accessibility for RISC complex [10], it is generally believed that mRNAs could adopt more flexible secondary structures than their rRNA and tRNA counterparts, whose functions are more structure-based. Data from our recent experiments and other studies, however, showed that the efficacy of siRNA might to a large extent be secondary structure- independent [19,20]. If this hypothesis is true, then it could be possible to assess the efficacy of a particular siRNA in an sequence context that is unlikely to recruit any of the authentic environmental factors. With this thought in mind, we started to explore whether the siRNA efficacy can be analyzed out of its original sequence (secondary structure) context, and how faithful is such data in reflecting the silencing efficacy of the siRNA on the endogenous target genes.

A modified firefly luciferase expression vector (Promega, USA) was used to incorporate a short synthetic double-stranded DNA fragment immediately after the translation start codon (ATG) of the luciferase gene (Figure 1). The short DNA fragment was designed to harbor the siRNA target sites and was chosen to be 38 bp and 19 bp respectively, a length that could be dealt with easily by chemical synthesis (Figure 2). They were designed mostly to be incorporated in-frame with the luciferase open reading frame. Out of frame fusions were also tested in order to understand whether sequence containing un-avoidable stop- codons could be analyzed using this system. After siRNA target sites were cloned into this expression vector, they were co-transfected into HEK293 cells, together with respective cognate siRNA and Renilla luciferase expression vector (pRL-TK) [17] as the control. The dual of tested siRNAs were measured by dual-luciferase assay.

Effective siRNA can silence the expression luciferase gene fused with a short siRNA target site

The first question we tried to answer was whether a proven efficacious siRNA could successfully silence its target site if such a site was grafted out of its authentic mRNA (sequence and or structural) context. Towards this end, an effective siRNA targeting an endogenous gene (CD46) was chosen from our earlier study [20], and an expression vector containing its 19mer target site was constructed as described in Materials and Methods. Another 19mer fragment from the same gene was used as irrelevant sequence control. The effective siRNA (CD46- 604) was then tested against both the artificial target and the irrelevant control. As shown in Figure 3B, HEK293 cells transfected with the target construct exhibited a significant reduction (92% knockdown) of luciferase gene expression compared to cells transfected with the irrelevant control construct. No difference of Renilla luciferase expression, which was used as an internal control for the transfection experiment, was observed. No luciferase silencing was observed when the siRNA was used to treat the firefly luciferase without the fused target site, suggesting that the silencing effect of the CD46 siRNA was achieved by attacking the inserted target site in a sequence- specific manner. The same experiments were performed with several effective siRNAs from other genes including: Cyclin Gl, Dsup6 and NPY. Oligos as outlined in table 1 were inserted into the same cloning site, then the recombinant plasmids were analyzed for fusion gene expression and the impact of cognate siRNAs. All tested siRNAs could knockdown the expression of respective fusion luciferase activity through attacking the inserted cognate target sites. These data showed that an efficacious siRNA does have the capacity to find and cleave its target site even if such a targeting is grafted out of its authentic sequence context and inserted into a reporter gene such as luciferase.

Optimization of siRNA target size

The next question we explored was the optimal length of the siRNA target site for this system. Desirably the shorter the region, the better if such a length could work as expected. To address this issue, four plasmids containing respective 19mer and 38mer fragments covering the target regions of two siRNAs were constructed and assessed for their capabilities to mediate siRNA-induced knockdown of the fusion target genes (Table 1). As shown in figure 3 A, no significant difference of inhibition has been observed among the plasmids containing 19mer and 38mer target sites for each siRNA, implying that the cognate sequence of siRNA, instead of the flanking sequence, was enough to serve as target site for the siRNA to accomplish the suppressive mission. Thus, the sequence of the target site may be shorter than 19 base pairs, e.g. 15 base pairs. It is also possible to insert a longer fragment than absolutely necessary, e.g. 25, 30, 35, 40, 45 or even 50 base pairs.

Since the above results suggested that the 19mer target site might contain most, if not all, necessary information for determining whether a siRNA worked or not, all subsequent experiments were carried out with the siQuant constructs containing 19mer siRNA target sites.

Table 1.

Names of recombinant Synthetic short DNA fragments inserted in each siQuant plasmid siQuant plasmid

CD46-siteA-38mer GATCTATTTTTCACTTATTGGAGAGAGCACGATTTATTGTGGTGGGCC (SEQ ID NO: 1) ATAAAAAGTGAATAACCTCTCTCGTGCTAAATAACACCAC

CD46-siteA-19mer GATCTCACTTATTGGAGAGAGCACGATTGGGCC (SEQ ID NO: 2) AGTGAATAACCTCTCTCGTGCTAAC

out-of-frame construct GATCTGACTTATTGGAGAGAGCACGATTGGGCC (SEQ ID NO: 3) ACTGAATAACCTCTCTCGTGCTAAC

frame-control construct GATCTTACTTATTGGAGAGAGCACGATTGGGCC (SEQ ID NO: 4) AATGAATAACCTCTCTCGTGCTAAC

CD46-siteB-38mer GATCTCTACTTACAAGCCTCCAGTCTCAAATTATCCAGGATATGGGCC (SEQ ID NO: 5) AGATGAATGTTCGGAGGTCAGAGTTTAATAGGTCCTATAC

CD46-siteB-19mer GATCTACAAGCCTCCAGTCTCAAATTATGGGCC (SEQ ID NO: 6) ATGTTCGGAGGTCAGAGTTTAATAC

Fas297 GATCTCAGGGAAGGAGTACATGGACACCGGGCC

(SEQ ID NO: 7) AGTCCCTTCCTCATGTACCTGTGGC

FaS297m GATCTCAGGGAAGGAATACATGGACACCGGGCC (SEQ ID NO: 8) AGTCCCTTCCTTATGTACCTGTGGC

Fas679 GATCTCAACCATACCAATGAATGCCTCCGGGCC

(SEQ ID NO: 9) AGTTGGTATGGTTACTTACGGAGGC Fas873 GATCTCATCTCATGGGAAGAGTGATGCCGGGCC (SEQ ID NO: 10) AGTAGAGTACCCTTCTCACTACGGC

Cyclin Gl GATCTCAGATCTACTTAGTCTAACTCCCGGGCC (SEQ ID NO: 11) AGTCTAGATGAATCAGATTGAGGGC

Duspβ GATCTCAGAGTTTGGCATCAAGTACACCGGGCC (SEQ ID NO: 12) AGTCTCAAACCGTAGTTCATGTGGC

NPY GATCTCAATGAGAGAAAGCACAGAAAACGGGCC (SEQ ID NO: 13) AGTTACTCTCTTTCGTGTCTTTTGC

Table 1: Sequences of synthetic DNA oligonucleotides used in this study to construct different siQuant plasmids for carrying out the efficacy assay. For making each siQuant plasmids two oligos were synthesized to form the short duplex as shown in the right panel. Note that all such duplexes have universal left and right over-hangs. Sequences disclosed in the sequence listing are the longer ones.

The reporter-based assay showed similar efficacy as demonstrated by targeting endogenous transcripts

Another critical issue was whether the efficacy observed using a siRNA on a reporter gene fused with its target sites could be used to predict the efficacy of the siRNA on its endogenous target gene. We designed three siRNAs against Fas mRNA (Table 1) and did a double-blinded test where the siRNA silencing efficacies were evaluated by one researcher using the fusion luciferase system described above, and the same siRNAs were used to target Fas mRNA by another researcher. Efficacy assay showed siRNA Fas679 and Fas297 could suppress the reporter gene expression by targeting their fusion sites, and the luciferase activities were knocked down to 12% and 14% of un-treated controls respectively, whereas Fas873 siRNA did not affect the expression of fusion gene (Figure 4A). As internal control, the activities of Renilla luciferase were not altered as judged from the raw data. In parallel, Northern blot analysis was performed in AML12 cell line using the same siRNAs to silence endogenous Fas gene. The suppression profiles observed for the authentic Fas gene expression closely matched the corresponding profiles obtained from efficacy assays where the fusion reporter genes were used (Figure 4B). The expression of β-actin gene, which was used as internal control for Northern blot assay, was not affected by these siRNAs.

The specificity of the efficacy assay was further demonstrated by a mismatch test

Several studies have shown that mismatches between the siRNA and its target site could affect the suppression activities of various siRNAs [21-24]. Therefore, we went further to test whether the efficacy assay system we set up had similar sensitivities to mismatches. We modified the target site of Fas297 siRNA and tested whether the introduced mismatch could alter the suppression activity of this otherwise proven active siRNA in cultured cells. A single nucleotide mutation (G->C) was thus constructed at the position 9 of the sense target strand in fusion luciferase gene (Figure 5A) . Both the original and modified constructs were tested using the same siRNA. The result showed that this single mismatch in the target site rendered the effective siRNA inactive for its mutated target (Figure 5B) . Since we knew that the siRNA can be readily integrated into the RNAi pathway, this experiment suggested that the change of the silencing effect for the reporter gene was caused by the point mutation. This result thus further demonstrated that our efficacy assay was dependent on the siRNA target sites in a highly sequence specific manner.

In-frame υs. out-of-ORF fusion of the siRNA target site After demonstrated that in-frame fusion of the siRNA target site appeared to be well suited for efficacy assay, we went further to investigate whether siRNA target sites harbouring stop codons could be used in such an assay as well. Normally one can avoid such complications by shifting frames slightly, but we thought that the out- of-ORF cases would not only be useful for assessing the sites with unavoidable stop codons, but should also be very helpful for target selection within or out of the open reading frames. In this experiment, we used the same siRNA and target sites as we used for the in-frame studies except that an artificial in-frame stop codon was introduced to exclude the target site from the ORF (Figure 6) . It was observed that a strong luciferase activity could still be observed from such a fusion construct, suggesting that the ribosomes might have used the ATG immediately after the insertion site to synthesize a full functional enzyme. After titrating plasmid amount to get comparable luciferase activities, we compared the silencing effect of the same siRNA on sites that was in the ORF and that was excluded from the ORF by a stop codon. It was observed that the siRNA could also induce efficient silencing of the reporter gene for the out-of- frame construct although the silencing efficiency seemed to be slightly reduced in comparison to inhibition of the in-frame fusion construct. The difference of silencing efficacy in correlation to the frame situation could be due to two reasons: a) in an out-of-ORF targeting event, the cleaved target mRNA has to be further degraded to remove the down stream ATG site. b) the insertion of the stop codon might have impacts on ribosome entry or the recruitments of other protein factors that could result in the change of accessibility in an otherwise accessible site.

In summary, we established a siRNA efficacy validation system where the efficacy of a siRNA could be read out by measuring the activity of a reporter enzyme, whose mRNA was tagged by a short sequence representing the target site of the tested siRNA. Our data demonstrated that siRNAs could sequence-dependently knockdown their target-site-coupled luciferase gene expressions and the suppression profiles against these artificial targets could be verified by targeting their natural target genes. In comparison to similar systems where the tagging sequences have to be large cDNA fragments, the current method presents a significant potential for high throughput applications due to the fact that the tagging sequences can be readily available through high throughput oligonucleotide synthesis, thus avoiding the restraint imposed by the availability of correct cDNA. Other advantages of the current efficacy assay system include a) the universal cloning scheme makes it much simpler than the individual frame adjustment for cDNA; b) sequences with or without stop codon can be assessed using the same scheme; c) this system is internally controlled. In addition to making the results more consisting and reliable, internal control eliminates variability that may arise as a consequence of indirect normalization, differences of mRNA abundance and difference of translation efficacies. The read-out from the efficacy assay would provide a relative guide on the efficacies of different siRNAs targeting the same gene. We have attempted on-chip synthesis of the tagging sequences for large-scale siRNA validation with promising initial results. This makes it highly feasible to validate siRNAs against all human and mouse genes in a short period of time.

Oligonucleotides

All DNA oligonucleotides used for target sites cloning (Table 1) were obtained from MWG Biotech AG (Ebersberg, Germany). RNA oligonucleotides were obtained from Dharmacon Research (Lafayette, CO, USA) and GENSET SA (Paris, France). siRNA duplex was prepared by mixing complementary sense-strand RNA and antisense- strand RNA at equal concentration at 50 DM in MQ water. The mixture was incubated in boiling water for 1 min and cooled down over night to allow the efficient formation of siRNA duplex. The quality of the RNA duplexes was assessed on a 15% PAGE gel.

Plasmid construction and cloning of siRNA target site The luciferase expression plasmid (pTRE_PSKH l_Luc) was kindly provided by Dr. Hans Prydz [17]. A cloning site was introduced immediately after the start codon of firefly luciferase gene by polymerase chain reaction (PCR) using the site forward primers 5'- TGCTAGATCTCACAGCCCATGGTGCGGAT-3' (SEQ ID NO: 14) and site reverse primer 5'-CATGGGCCCGGCGCCATGGAAGACGCCA-3' (SEQ ID NO: 15). The design of the primer pair incorporated a Bgl II and an Apa I restriction sites to facilitate cloning. The list of oligonucleotides used for target sites cloning is summarized in table 1. Oligo sets, designed with the appropriate restriction sites, were annealed and subcloned into the Bgl II - Apa I sites of constructed plasmid. Sequencing was performed to verify the inserted sequences using the Big- DYE terminator Method (Perkin Elmer Corp). The tested siRNAs are listed in table 2.

Transfection and dual-luciferase assay

To investigate the suppression effect of these siRNA on their respective constructs, we carried out the dual luciferase reporter assay. Human embryonic kidney (HEK293) cells were maintained in DMEM medium (Life Technologies, GIBCO) and seeded in 24-well multi-dishes (0.5 ml medium/well) to reach about 50% confluence at transfection. The cells were grown for 24 h and the culture medium was changed to OPTIMEM (GIBICO), 0.5 ml/well, before transfection. The cells were co-transfected with plasmids and siRNA duplex in the presence of

Lipofectamine 2000 (Invitrogen, USA) at a final concentration of 0.17% (total transfection volume, 0.6 ml). For each well, 0.17 g of recombination plasmid and 0.017 g of pRL-TK was used. The final concentration of siRNA is 13nM. The transfection medium was changed to culture medium (1 ml) after 4 h. All experiments with were performed in triplicates and repeated at least twice. Cells were harvested in 24 hours by passive cell lysis and the dual-luciferase assay was conducted. Luciferase activities were determined with 10 DL of cell lysate using the Dual-Luciferase Assay System (Promega) by NOVOStar (BMG Labtechnologies GmbH, Germany). The firefly /renilla activity ratio was generated for each well, and the inhibition efficiency of each siRNA was calculated by normalizing to respective buffer control. Transfection and northern blot assay

The mouse AML12 cell line (ATCC; CRL-254) was cultured in DMEM/F12 (Life Technologies, GIBCO) supplemented with 40 ng/ml dexamethasone (SIGMA), 10% fetal bovine serum (Life Technologies, GIBCO), lx Insulin-Transferrin-Selenium-X (Life Technologies, GIBCO), and lx Penicillin- Streptomycin- Glutamine (Life Technologies, GIBCO). For transfection, the medium described above was used without antibiotics. Transfection was done in accordance with the manufacturer's manual with 0.32% v/v of LipofectAmine2000, on cells between passages 5-18. Briefly, cells were counted and seeded at around 1.000.000 cells / 25cm² flask. Typically after 24 h, when the cell cultures were around 90% confluent, liposome- plasmids/ siRNA complexes formed in OptiMem were added to the cultures. Twenty- four hours after transfection, the cells were harvested in 1 ml of RNAwiz solution (Ambion) according to the manufacturer's instruction. Total RNA, 10-12 μg, was separated by electrophoresis in an ethidium bromide-containing agarose-formaldehyde gel [15]. The intensity of the 18S and 28S rRNA bands under UV light was checked to verify that all samples were equally loaded and that no RNA degradation had occurred. The β-actin cDNA probe used was a fragment described earlier [16] and the Fas cDNA was an EcoR I digest of plasmid pMFI (17). The probes were labeled with [γ-32P] dCTP with Ready -To-Go DNA labeling beads (Amersham Pharmacia Biotech). Hybridization and stringency wash was performed as previously described [15], and the signal was detected by a BAS 1500 scanner (Fuji).

Construction of siRNA validation reporter using short synthetic DNA fragment and use of the reporter for evaluation of siRNA efficacy on targets that contain mismatches to the siRNA The discovery that siRNA can silence target genes through sequence- specific cleavage of the cognate RNA transcript has led to the rapid adoption of this technology for analysis of gene function in mammalian cells with the expectation of identifying novel drug targets and developing more specific siRNA therapeutics [1-7, 19]. Target recognition by siRNA was initially thought to be a highly sequence- specific process mediated by the siRNA complementary to the target mRNA [1-8]. A single base mismatch with respect to the target was found to abolish gene- silencing effect in some cases [21-24]. This view was further strengthened by the assessment done on the specificity of RNAi silencing using genome-wide expression profilings [25]. The rosy picture was, however, challenged when significant off-target hit was found for some siRNA in another carefully designed microarray experiment, in which a number of genes that only have partial sequence similarity to the siRNA was also down-regulated significantly in a manner that can only be explained by direct siRNA attack on their transcripts [26-27]. While off-target hits of siRNA have become widely aware of concern in the siRNA applications, systematic analysis of off- target hits by siRNA is missing. The art stopped short of demonstrating whether the down-regulation of those un-intended "target genes" were directly silenced by the siRNA or if they are silenced as a secondary events. Limited mutational analysis has been conducted to explore the specificity of siRNA, and it was found that some of the terminal nucleotides does little to change the efficacy of some siRNA but some other mutations do abolish the silencing activity of a siRNA [1-7, 21-27]. In these cases, the siRNA sequences, instead of the target sequences, were mutated. As we now know, the efficacy of siRNA is actually governed by at least two factors, its ability to enter the RISC complex and its ability to recognize its targets (wild-type or mutated) [9]. In the cases where the siRNA sequence is mutated, it becomes uncertain whether loss of silencing activity is the impact of the mutation on the target recognition step or at the RISC entry step. Currently available study of impact of mismatch on the efficacy of siRNA is also complicated by that fact that it was the siRNA, not the target, that was mutated in these studies

Understanding of the impact of off-target hits is important for design of siRNA and interpretation of research results of siRNA experiments, but it is especially important for the future development of siRNA drugs. We choose to investigate the silencing effects of a proven functional siRNA on all 75 permutations of target sites that each has a single mismatch with this siRNA in order to provide the first quantitative measurement of the impact of mismatch positions and nucleotide composition on the target recognition of siRNA. The results demonstrated that targets with single mismatches with a siRNA could be silenced to different extent by the siRNA in correlation to the position and composition of the mismatch.

Results

Organization of fusion luciferase construct. To study the effect of sequence mismatch on siRNA silencing activity and the rules governing siRNA:mRNA interaction, we introduced single-nucleotide changes at every position of a siRNA target site, and evaluated the gene silencing efficacies in HEK293 cells. A functional siRNA(CD46i) targeting human endogenous gene CD46 was previously validated and utilized in our present study [19]. The sense strand sequence of this siRNA is (5'-CTTATTGGAGAGAGCACGA-3') (SEQ ID NO: 16) located from position 604 to 622 relative to the first nucleotide of the start codon in human CD46 gene (XM_036622). Starting from a modified luciferase expression vector [15], we cloned the systematically mutated target sites of CD46i into firefly luciferase gene by polymerase chain reaction. A schematic map depicting the configuration of this fusion luciferase gene is shown in figure 1. Modified constructs express fusion luciferase genes incorporating various siRNA target sites. A total of 57 constructs were prepared for silencing analysis, with three different constructs for each position on the siRNA target site. Each modified construct incorporated a mutated target site differing from the wild-type site by only one nucleotide. All of the modified target sites and the wild-type control were validated by sequencing. Constructs were named according to the mutated position and the mutated nucleotide. For example, construct siteO l-G consists of a C to G change at position 1 of the mRNA target site relative to the 5' end of siRNA sense strand. When the expressed transcript base pairs with CD46i, the siRNA antisense strand forms a G/G mismatch at position 1 of the modified target site, while the wild-type transcript matches perfectly with the antisense strand, providing a positive control for the transfection assays. CD46i cause extensive off-target effects on single mismatch target. When the systematically mutated target sites were tested in HEK293 cells, the fusion transcripts were targeted by the co-transfected siRNA molecules, so that the firefly luciferase signal could be brought down depending on the silencing efficacy. The silencing effects were measured by dual-luciferase assay, using pRL-TK vector for normalization control. Figure 8 showed the suppression efficacies of CD46i on various single mismatch target sites. HEK293 cells transfected with the wild-type target construct exhibited a significant reduction (92% knockdown) of firefly luciferase signal, which was presented as the gray bar for every site. Of total 57 single mismatch target constructs (Figure 7), firefly luciferase expression of 23 constructs (40% of 57 tested constructs) were knock-downed >70%, 19 constructs (33%) were knock-downed between 40% to 70%, and only 15 constructs (26%) were knock- downed less than 40%. In summary, 73% of tested constructs were significantly silenced because of the off- target effects (Figure 8).

Mismatch tolerances of CD46i are position-dependent. According to the silencing efficacies of tested constructs, siRNA target sites can be classified into a low tolerance group, a medium tolerance group and a high tolerance group. The low tolerance group comprises target sites from position 5 to position 11 on the siRNA sense strand. Mismatches on these sites abolished most of the suppression activities. Consisting with earlier reports, central mismatches between siRNA and target sites were critical for the silencing activities, rendering these sites low tolerance, or high sensitivity to mismatch. However, our data further extended the low tolerance region to the 5' end of siRNA sense strand, covering sites from position 5 to position 11. This observation contrasted to those observed by others, wherein a core sequence was proposed to be localized towards the 3' end of siRNA sense strand. Another interesting finding was that even at these low tolerance sites, some mismatches could still significantly knockdown the fusion luciferase expression, such as Site05-C (83% knockdown), Site06-C (43% knockdown), Site07-A (65% knockdown), Site08-A (68% knockdown), Site09-G (54% knockdown) and Sitel l-G (66% knockdown) . These high tolerance mismatches at low tolerance sites could be very important for RNAi mechanism study and its applications. The medium tolerance group contains site 3, site 4 and sites from position 12 to position 17, where mismatches are well tolerated. All mismatches of these sites could significantly reduce the target gene expression. Surprisingly, wild-type like silencing efficacies were observed for some mismatches at site 13 (sitel3-T) and site 14 (sitel4- A), indicating that it was not only the position of mismatches but also the nature of the mismatch that could determine the silencing efficacies. The high tolerance group consists of site 1, site 2, site 18 and site 19, basically the two sites on both ends. Mismatches of these sites almost did not affect the silencing activities.

Mismatch tolerances of CD46i are alternation-dependent. Different mismatches of the same site cause remarkably diverging luciferase signal reduction, and varies from 65% to 0% knockdown for site 7, revealed a correlation between the silencing efficiency and the nature of the mismatch of a given mismatch. Further analysis showed that the suppression profiles of target sites paired with the same nucleotide on the siRNA antisense strand was specific and reproducible, suggesting that there were discernible rules governing mismatch interaction which might be generally applicable.

As a sequence- specific gene silencing mechanism, RNA interference is thought to require near-identity between the siRNA molecules and its cognate mRNA target. However, different reports show varying degrees of tolerance for mismatches in siRNA-mediated silencing [1-7, 21-27]. The identification of a region with generally increased sensitivity to mismatches within siRNA target site, would be of great importance for the potential applications of siRNAs to specifically target transcripts of disease-associated alleles in various dominant- negative disorders. In an earlier examination of sequence specificity required for target recognition, Elbashir et al. introduced sequence changes in the siRNA duplexes by inverting short segments of 3 or 4 nt [7]. Duplexes of siRNAs with inverted sequence segments showed dramatically reduced ability for targeting the firefly luciferase reporter. They found that the sequence changes located between the 3' end and the middle of the antisense siRNA completely abolished target RNA recognition, but mutations near the 5' end of the antisense siRNA exhibited a small degree of silencing. Recently, conflicting data was obtained by Prydz laboratory when they explored the tolerance to mutations and chemical modifications with a siRNA targeting the blood clotting initiator Tissue Factor [17]. Using single-mutant and double-mutant siRNAs, their experiments showed that the mutations in the 5 'end of siRNA sense strand were well tolerated, exhibiting essentially the same activity as the wild type. However, mutations localized towards the 3' end of the siRNA sense strand, and up to the approximate midpoint, were impaired in their activities, demonstrating a lower tolerance to mutations in the 3' end of tested siRNA. Consistent with this result, a genome-wide profiling revealed the similar bias using microarrays [22-27]. A core sequence encompassing the 3'end of the siRNA sense strand was showed to be important for transcript cleavage. Other genome-wide specificity studies on the mRNA level have been published but the results are conflicting . Up to now, mismatch tolerance of RNAi still remains as an open question. However, it seems clear that central mismatches between the siRNA and the target mRNA are more critical, and more likely to abolish silencing than mismatches at the ends. Our experiments demonstrated that the mismatch tolerance of RNAi is both position- and alternation-dependent. The data showed that different alternations could have different effects even at the same position, and the same alternations could have different effects on target silencing if they were at different positions. Without systematic study, the work on mismatch tolerance could easily lead to a biased conclusion, thus resulting conflicting reports. In contrast to the observation that that near perfect complementarity between the first 9 nt (from 5'-end) of a microRNA and its cognate miRNA-recognization element is required for miRNA function (28,29), our data showed that the complementarity between the 3' end to the midpoint of CD46i antisense strand and its mRNA target was more important, suggesting different functional mechanisms for siRNA and miRNA. To the best of our knowledge, all of the earlier off-target studies employed the modified siRNA instead of modified target site. Usually the nucleotides at specific positions of the siRNA sense and/ or antisense strands were substituted and the effects of the altered siRNA sequence were determined on the authentic mRNA target. As the suppression characteristics of a specific siRNA is correlated closely with its sequence property, modification of siRNA sequence could alternate the suppression efficacy even to its cognate target site, which would translate into compromised conclusion. Another disadvantage of using alternated siRNA sequence is that the modification could potentially affect the incorporation of siRNA duplex into RISC complex. To eliminate many of these variables, in the present study we evaluated the gene silencing effects with systematically mutated target sites instead of alternated siRNAs, as we believed that modifying the siRNA target site is a better way to mimic the binding characteristics between siRNA and the nonspecific target.

Besides the standard Watson-Crick base pairs, a large collection of specific base-base interactions (non-canonical base pairs) have been enumerated and frequently observed in crystal and NMR structures of RNA molecules [30-34]. It is now widely accepted that such non- standard interactions stabilize the secondary as well as the tertiary structures of RNA. However, general principles governing their associations are still emerging. It is anticipated that certain non- standard interactions will frequently be associated with particular local environments. A recent study demonstrated tolerance for G:U wobble pairing between the siRNA oliogonucleotide and the mRNA target [30-34]. Because of its geometric property, G:U wobble pairs fit very smoothly in a regular A-form helix. Although G:U wobble base pair was suggested like authentic Watson-Crick base pair in antisense RNA oliogonucleotide:mRNA duplex interaction, our experiment indicated that this kind of wobble base pair could significantly affect the suppression efficacy of target gene in a less degree compared with other mismatches on the same site. In addition to the reported G:U wobble, we repeatedly observed that A:C wobbles are well tolerated for siRNA:mRNA interaction. Our data demonstrated that both A:C and G:U wobble base pairs are were tolerated.

The G:U wobble base pairing was showed to have a comparable thermodynamic stability with standard Watson-Crick base pairs [30- 34], so that hydrogen bonding was taken as an obvious explanation for the tolerance for G:U wobble. Our observation further suggested that the A:C and G:U wobbles could provide the necessary orientation for siRNA and mRNA interaction, and the configuration, or three- dimension compatibility could play a significant role in determining the observed tolerance. So we hypothesized that siRNA:mRNA interactions might be guided by two factors. One of them is the high- affinity interactions, based on hydrogen bonding energies. Another factor is the configuration compatibility between siRNA and mRNA target.

Materials and methods

Plasmid construction and siRNA target site modification. A modified luciferase expression vector (pTRE_PSKH l_Luc) was kindly provided by Dr. Hans Prydz [17]. This vector was further modified in our lab to introduce an in-frame ATG start codon before luciferase gene. Various mutated target sites of CD46i were cloned between the introduced start codon and the original start codon of firefly luciferase gene by polymerase chain reaction. Degenerate primers used for target sites construction were summarized in table 1. After PCR amplification of template plasmid, the products were restricted by Bgl II, self-ligated, and then transformed into DH5a competent cells. Resulting clones were screened by sequencing. All DNA oligonucleotides used for target sites construction were purchased from biomers.net GmbH (Geschaftsleitung, Germany). RNA oligonucleotides were obtained from Dharmacon Research (Lafayette, CO, USA). siRNA duplex was prepared by mixing complementary sense and antisense strand RNA at equal concentration of 50 mM in water. The mixture was incubated in boiling water for 1 min and cooled down over night to allow the efficient formation of siRNA duplex. The quality of the RNA duplexes was assessed on 15% PAGE gel.

Transfection and dual-luciferase assay. Human embryonic kidney (HEK293) cells were maintained in DMEM medium (Life Technologies, GIBCO) and seeded into 24-well plates ( ~ 1 ^'105 cells/ well ) one day before the transfection. The cells is at about 50% confluence at the time of transfection. The fusion constructs (0.17 mg) bearing mutated siRNA target sites were co-transfected with pRL-TK normalization plasmid (0.017 mg) and CD46i using Lipofectamine 2000 (Invitrogen). The final concentration of CD46i was 13nM. Cells were harvested in 24 hours by passive cell lysis and the dual-luciferase assay (Promega) was performed using NOVOStar (BMG Labtechnologies GmbH, Germany). The firefly / Renilla activity ratio was generated for each well, and the silencing efficacy of each construct was calculated by normalizing to respective siRNA-untreated control. All experiments were performed in triplicates and repeated at least three times.

REFERENCES

1. S.M. Elbashir, J. Harborth, W. Lendeckel, A. Yalcin, K. Weber, T. Tuschl Duplexes of 21 -nucleotide RNAs mediate RNA interference in cultured mammalian cells. Nature 411 (2001) 494-498.

2. Zamore, P.D., Tuschl, T., Sharp, P.A., Bartel, D.P. RNAi: double-stranded RNA directs the ATP-dependent cleavage of mRNA at 21 to 23 nucleotide intervals. Cell 101, 25-33 (2000).

3. Q. Ge, M.T. McManus, T. Nguyen, CH. Shen, P.A. Sharp, H.N. Eisen, J. Chen RNA interference of influenza virus production by directly targeting mRNA for degradation and indirectly inhibiting all viral RNA transcription. Proc Natl Acad Sci USA 100 (2003) 2718-2723.

4. Fire, A. et al. Potent and specific genetic interference by double-stranded RNA in Caenorhabditis elegans. Nature 391, 806-81 1 (1998).

5. Bernstein, E., Caudy, A.A., Hammond, S.M., Hannon, G.J. Role for a bidentate ribonuclease in the initiation step of RNA interference. Nature 409, 363-366 (2001). 6. Parrish, S., Fleenor, J., Xu, S., Mello, C, Fire, A. Functional anatomy of a dsRNA trigger: differential requirement for the two trigger strands in RNA interference. Mol. Cell. 6, 1077-1087 (2000).

7. Elbashir, S.M. , Martinez, J., Patkaniowska, A., Lendeckel, W., Tuschl, T. Functional anatomy of siRNAs for mediating efficient RNAi in Drosophila melanogaster embryo lysate. EMBO J. 20, 6877-6888 (2001).

8. Hamilton, A.J., Baulcombe, D.C. A species of small antisense RNA in posttranscriptional gene silencing in plants. Science 286, 950-952 (1999).

9. D.S. Schwarz, G. Hutvagner, T. Du, Z. Xu, N. Aronin, P.D. Zamore. Asymmetry in the assembly of the RNAi enzyme complex. Cell 115 (2003), 199- 208.

10. G. Sen, T.S. Wehrman, J.W. Myers, H.M. Blau. Restriction enzyme -generated siRNA (REGS) vectors and libraries. Nat Genet. 36 (2004) 183- 189.

11. . Ui-Tei, Y. Naito, F. Takahashi, T. Haraguchi, H. Ohki-Hamazaki, A. Juni, R. Ueda, K. Saigo. Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference. Nucleic Acids Res. 32 (2004) 936-948.

12. A. Khvorova, A. Reynolds, S.D. Jayasena Functional siRNAs and miRNAs exhibit strand bias. Cell, 115 (2003), 209-216. 13. Saetrom, P., Snove, O. Jr. A comparison of siRNA efficacy predictors. Biochem. Biophys. Res. Commun. 321, 247-253 (2004). 301-309 (2002).

14. F.R. Kretschmer-Kazemi, G. Sczakiel The activity of siRNA in mammalian cells is related to structural target accessibility: a comparison with antisense oligonucleotides. Nucleic Acids Res. 31 (2003) 4417-4424. 15. M. Amarzguioui, G. Brede, E. Babaie, M. Grotli, B. Sproat, H. Prydz Secondary structure prediction and in vitro accessibility of mRNA as tools in the selection of target sites for ribozymes. Nucleic Acids Res. 28 (2000) 4113-4124.

16. J. Harborth, S.M. Elbashir, K. Vandenburgh, H. Manninga, S.A. Scaringe, K. Weber, T. Tuschl Sequence, chemical, and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing. Antisense Nucleic Acid Drug Dev. 13 (2003) 83-105.

17. T. Holen, M. Amarzguioui, M.T. Wiiger, E. Babaie, H. Prydz. Positional effects of short interfering RNAs targeting the human coagulation trigger Tissue Factor. Nucleic Acids Res. 30 (2002) 1757-1766.

18. P. Nagy, D.J. Arndt-Jovin, T.M. Jovin. Small interfering RNAs suppress the expression of endogenous and GFP-fused epidermal growth factor receptor (erbBl) and induce apoptosis in erbBl-overexpressing cells. Exp Cell Res. 285 (2003) 39-49. 19. Y. Xu, H.Y. Zhang, D. Thormeyer, O. Larsson, Q. Du, J. Elmen, C. Wahlestedt, Z. Liang Effective small interfering RNAs and phosphorothioate antisense DNAs have different preferences for target sites in the luciferase mRNAs. Biochem Biophys Res Commun. 306 (2003) 712-717.

20. Y. Xu, A. Linde, O. Larsson, D. Thormeyer, J. Elmen, C. Wahlestedt, Z. Liang Functional comparison of single- and double-stranded siRNAs in mammalian cells. Biochem Biophys Res Commun. 316 (2004) 680-687.

21. Amarzguioui, M., Holen, T., Babaie, E., Prydz, H. Tolerance for mutations and chemical modifications in a siRNA. Nucleic Acids Res. 31, 589-595 (2003).

22. Saxena, S., Jonsson, Z.O., Dutta, A. Small RNAs with imperfect match to endogenous mRNA repress translation. Implications for off-target activity of small inhibitory RNA in mammalian cells. J. Biol. Chem. 278, 44312-44319 (2003).

23. Boutla, A., Delidakis, C, Livadaras, I., Tsagris, M., Tabler, M. Short 5'- phosphorylated double-stranded RNAs induce RNA interference in Drosophila. Curr. Biol. 11, 1776- 1780 (2001). 24. Hamada, M. et al. Effects on RNA interference in gene expression (RNAi) in cultured mammalian cells of mismatches and the introduction of chemical modifications at the 3'-ends of siRNAs. Antisense Nucleic Acid Drug Dev. 12,

25. Semizarov, D. et al. Specificity of short interfering RNA determined through gene expression signatures. Proc. Natl. Acad. Sci. U S A 100, 6347-6352 (2003).

26. Chi, J.T. et al. Genomewide view of gene silencing by small interfering RNAs. Proc. Natl. Acad. Sci. U S A 100, 6343-6346 (2003). 27. Jackson, A.L. et al. Expression profiling reveals off-target gene regulation by RNAi. Nat. Biotechnol. 21, 635-637 (1997).

28. Zeng, Y., Cullen, B.R. Sequence requirements for micro RNA processing and function in human cells. RNA 9, 112- 123 (2003).

29. Kiriakidou, M. et al. A combined computational-experimental approach predicts human microRNA targets. Genes Dev. 18, 1165-1178 (2004). 30. Nagaswamy, U., Voss, N., Zhang, Z., Fox, G.E. Database of non-canonical base pairs found in known RNA structures. Nucleic Acids Res. 28, 375-376 (2000).

31. Hermann, T., Patel, D.J. Stitching together RNA tertiary architectures. J. Mo.l Biol. 294, 829-849 (1999).

32. Nagaswamy, U. et al. NCIR: a database of non-canonical interactions in known RNA structures. Nucleic Acids Res. 30, 395-397 (2002). 33. Gautheret, D., Konings, D., Gutell, R.R. G.U base pairing motifs in ribosomal RNA. RNA 1, 807-814 (1995).

34. Masquida, B., Westhof, E. On the wobble GoU and related pairs. RNA 6, 9-15 (2000).

Claims

1(3)CLAIMS

1. A method of measuring the gene silencing efficacy of a putative siRNA molecule comprising the steps i) bringing the siRNA molecule in contact with a cell expressing an RNA molecule comprising a siRNA targeting site of 15-50 nucleotides and a reporter gene, ii) directly or indirectly measuring the translation of the reporter gene wherein a decrease in translation of reporter gene indicates the gene silencing efficacy of the putative siRNA molecule.

2. The method according to claim 1, wherein the siRNA targeting site is positioned upstream or downstream of the coding sequence of the reporter gene.

3. The method according to claim 1 or 2, wherein the siRNA targeting site is positioned in the reading frame of the coding sequence of the reporter gene.

4. The method according to claim 1 or 2, wherein the siRNA targeting site is not positioned in the reading frame of the coding sequence of the reporter gene.

5. The method according to any of claims 1-4, wherein the translation of the reporter gene is measured by measuring the amount or activity of the reporter gene product. 2(3)

6. A method according to any of claims 1-4, wherein the RNA molecule is transcribed from a plasmid in an expression system.

7. A method according to any of claims 1-6, wherein the siRNA and siRNA targeting site have identical sequences.

8. A method according to any of claims 1-6, wherein the siRNA and siRNA targeting site have mismatched sequences.

9. A method according to any of claims 1-8, wherein the siRNA is a microRNA.

10. A DNA-construct comprising a siRNA targeting site of 15-50 nucleotides fused to a reporter gene.

11. The construct according to claim 10 wherein the siRNA targeting site is fused in-frame with the reporter gene.

12. The construct according to claim 10 wherein the siRNA targeting site is fused out-of-frame with the reporter gene.

13. A DNA molecule capable of being transcribed in a cell, containing the construct according to any of claims 10- 12.

14. The DNA molecule according to claim 13, being a plasmid, a viral vector or a linear DNA fragment. 3(3)

15. The DNA molecule according to claim 13 or 14 further comprising a promoter for control of the transcription of the DNA construct comprising the siRNA target site and the reporter gene.

16. A cell comprising the DNA molecule according to any of claims 10- 15.