Improved gene silencing methods

FIELD OF THE INVENTION

The invention relates to the field of agriculture, more particularly to the modification of plants by genetic engineering. Described are methods for modifying so-called gene silencing in plants or other eukaryotic organisms by modulating the functional level of enzymes with ribonuclease activity responsible for the generation of RNA intermediates in various gene silencing pathways. Also described are methods for modifying gene silencing in plant cells or plants through modification of genes that have an influence on the initiation or maintenance of gene silencing by the silencing RNA encoding chimeric genes, such as genes involved in RNA directed DNA methylatioti. Thus, methods and means are provided to modulate post-transcripliomtl gene silencing in eυkaryotes through the alteration of the functional level of proteins involved in transcriptional silencing of the silencing UNA encoding genes.

BACKGROUND TO THE INVENTION

Gene silencing is a common phenomenon in eukaryotcs, whereby the expression of particular genes is reduced or even abolished through a number of different mechanisms ranging from mRNA degradation (post transcriptional silencing) over repression of protein synthesis to chromatin remodeling (transcriptional silencing).

Tlic gene-silencing phenomenon has been quickly adopted to engineer the expression of different target molecules. Initially, two predominant methods for the modulation of gene expression in eukaiγotic organisms were known, which are referred to in the art as "antisense" downregulation or "sense" downregulation.

In the last decade, it has been demonstrated that the silencing efficiency could be greatly improved both on quantitative and qualitative level using chimeric constructs encoding

RNA capable of forming a double stranded RNA by basepatrtng between the antisense and sense RNA nucleotide sequences respectively complementary and homologous to the

target sequences. Such double stranded RNA (dsRNA) is also referred to as hairpin RNA (hpRNA).

The following references describe the use of such methods:

Fire et al., 1998 describe specific genetic interference by experimental introduction of double-stranded RNA in Caenorhabditis elegans.

WO 99/32619 provides a process of introducing an RNA into a living cell to inhibit gene expression of a target gene in that cell. The process may be practiced ex vivo or in vivo. The RNλ has a region with double-stranded structure. Inhibition is sequence-specific in that the nucleotide sequences of the duplex region of the RNA and or a portion of the target gene are identical.

Waterhousc ct al. 1998 describe that virus resistance and gene silencing in plants can be induced by simultaneous expression of sense and anti-sense RNA. The sense and antisense RNA may be located ill one transcript that has self-compleinentarity.

Hamilton et al. 1998 describes that a transgeπe with repeated DNA, i.e., inverted copies of tts 5' untranslated region, causes high frequency, posl-tπtπscriptional suppression of ACXv-oxidasc expression in tomato.

WO 98/53083 describes constructs and methods for enhancing the inhibition of a target gene wilhiπ an organism which involve inserting into the gene silencing vector an inverted repeat sequence of all or part of a polynucleotide region within the vector,

WO 99/53050 provides methods and means for reducing the phenol ypic expression of a nucleic acid of interest in eukaryolic cells, particularly in plant cells, by introducing chimeric genes encoding sense and antisense RNA molecules directed towards the target nucleic acid. Thesσ molecules are capable of foiiuiug a uυuuie sLiaπucu RNA region by base-pairing between the regions with the sense and anlisense nucleotide sequence or by .

introducing the KNA molecules themselves. Preferably, the RNA molecules comprise simultaneously both sense and anlisense nucleotide sequences.

WO 99/49029 relates generally to & method of modifying gene expression and to synthetic genes for modifying endogenous gene expression in a cell, tissue or organ of a transgenic organism, in particular to a transgenic animal or plant. Synthetic genes and genetic constructs, capable of forming a dsRNA which are capable of repressing, delaying or otherwise reducing the expression of an endogenous gene or a target gene in an organism when introduced thereto are also provided.

WO 99/61631 relates to methods to alter the expression of a target gene in a plant using sense and antisense RNA fragments of the gene. Tine sense and aπtisense RNA fragments are capable of pairing and forming a double-stranded RNA molecule, thereby altering the expression of the gene. The present invention also relates to plants, their progeny and seeds thereof obtained using these methods.

WO 00/01846 provides a method of identifying DNA responsible for conferring a particular phcnoiype in a cell which method comprises a) constructing a cDNA or genomic library of lhe DNA of the cell in a suitable vector in an orientation relative to (a) promoter(s) capable of initialing transcription of the cDNA or DNA to double stranded (ds) RNA upon binding of an appropriate transcription factor to the promotcr(s); b) introducing the library into one or more of cells comprising the transcription factor, and c) identifying and isolating a particular phcnotype of a cell comprising the library and identifying the DNA or cDNA fragment from the library responsible for conferring the phenotype. Using this technique, it is also possible to assign function to a known DNA sequence by a) identifying homologucs of the DNA sequence in a cell, b) isolating the relevant DNA homologue(s) or a fragment thereof from the cell, c) cloning the homologue or fragment thereof into an appropriate vector in an orientation relative to a suitable promoter capable of initiating transcription of dsRNA from said DNA hornoioguc or ffSgiucftl upon binding of an appiύp-iaie irziiiscrψiiuM facior to the

promoter and d) introducing the vector into the cell from step a) comprising the transcription factor.

WO 00/44914 also describes composition and methods for iπ vivo and in vitro attenuation of gene expression using double stranded RNA, particularly in zebrafϊsh.

WO 00/49035 discloses a method for silencing the expression of an endogenous gene in a cell, the method involving ovcrexpressing in the cell a nucleic acid molecule of the endogenous gene and an antisense molecule including a nucleic acid molecule complementary to the nucleic acid molecule of the endogenous gene, wherein the overexpression of the nucleic acid molecule of the endogenous gene and the anlisense molecule in the cell silences the expression of the endogenous gene.

Smith et ul., 2000 as well as WO 99/53050 described that intron containing dsRNA further increased the efficiency of silencing. Intron containing hairpin PLNA is often also referred to as ihpRNA.

Although gene silencing was initially thought of as a consequence of the introduction of aberrant RNA molecules, such as upon the introduction of transgcncs (transcribed to antisense sense or double stranded RNA molecules) it has recently become clear that these phenomena are not just experimental artifacts. RNA mediated gene silencing in eukaryotes appears to play an important role in diverse biological processes, such as spatial and temporal regulation of development, heterochromatin formation and antiviral defense.

A)I eukaryotes possess a mechanism that generates small RNAs which are then used to tcgulate gene expression at the transcriptional or post-transcriplional level. Various double sttatided RNA substrates are processed into small, 21-24 nucleotide long RNA molecules through the action of specific ribonucleases (Dicer or Dicer-Like (DCL) proteins). These small RNAs serve as guide molcc-uica fo. μiuicin complexes (RNA-

induced silencing complexes (RISC)) which lead Io the various effects achieved through gene silencing-

Small RNAs involved in repression of gene expression in cukaryoles through sequence specific interactions with RNA or DNA are generally subdivided in two classes: inkroRNAs (mi RNAs) and small interfering RNAs (siRNAs). These classes of small RNA molecules arc distinguished by the structure of their precursors and by their targets. raiRNAa are cleaved from the short, imperfectly paired stem of a much larger foldback transcript and regulate the expression of transcripts to which they may have limited similarity. siRNλs arise from a long double stranded RNA (dsRNA) and typically direct the cleavage of transcripts to which they are completely complementary, including the transcript from which they are derived (Yoshikawa et ah, 2005, Genes & Development, 19: 2164-2175).

The number of Dicer family members varies greatly among organisms. In humans and C. elβgans there is only one identified Dicer. In Drosøphila, Dicer- 1 and Dicer-2 are both required for small interfering RNA directed mRNA cleavage, whereas Dicer-1 but not Dicer-2 is essential for microRNA directed repression (Lee et ah, 2004, Pham et ah, 2004).

Plants, such as Ambidopsis, appear to have at. least four Dicer-like (DCL) proteins and it has been suggested in the scientific literature that these DCLs are functionally specialized (Qi et ah, 2005 Molecular Cell, 19, 421-428)

DCLl processes miRNAs from partially double-stranded stem-loop precursor RNAs transcribed from MlR genes (Kurihara and Watanabe, 2004, Proc. Natl. Acad. Scϊ. USA 101 : 12753-12758).

DCL3 processes endogenous repeat and intergenic-region-derived siRNAs that depend on RNA dependent RNA polymerase 2 and is involved in the acύumulaϋuπ «.>f ύm 24nt

siRNAs implicated in DNA and histone methylatioπ (Xie et al. _v 2004, PLosBiology,

2004, 2, 642-652).

DCL2 appears to function in lhe antiviral silencing response for some, but not all plant- viruses ((Xie et al., 2004, PLosBiology, 2004, 2, 642-652).

Several publications have ascribed a role to DCL4 in the production of trans-acting siRNAs (ta-siRNAs). ta-siRNAs are a special class of endogenous siRNAs encoded by three known families of genes, designated TASl, TAS2 and TAS3 in λrabidopsis thaliana. The biogenesis pathway for ta-siRNAs involves site-specific cleavage of primary transcripts guided by a miRNA. The processed transcript is then converted to dsRNA through the activities of RDR6 and SGS3. DCLA activity then catalyzes the conversion of the dsRNA into siRNA duplex formation in 21-nt increments {Xie ct al.

2005, Pfoc. Natl. Acad. Sci. USA 102, 12984-12989; Yoshikawa ct al., 2005, Genes & Development, 19: 2164-2175; Gasciolli et al ., ^J 2005 Current Biology, 15, 1494-1500).

As indicated ill Xie et al. 2005 (supra) whether DCL4 is necessary for transgene and antiviral silencing remains to be determined.

Dunoyer et al, 2005 (Nature Genetics, 37 (12) pp l 356 to 1360) describe that DCL4 is required for RNA interference and produces the 21-nucleotidc small interferening RNA component, of the plant cell-to-cell silencing signal.

WO2004/09G995 describes Dicer proteins from guar (Cyamopsis tetmgonoϊυba), corn (Zea mays), rice (Oryza sativa), soybean (Glycine max) and wheal (Triticum aestivum). Tlic patent application also suggests the construction of recombinant DNA constructs encoding all or portion of these Dicer proteins in sense or antisense orientation, wherein expression of the recombinant DNA construct results in production of altered levels of the Dicer in a transformed host cell,

Cao st a ^j . (2003) described the rc.'c of the DRM and CMTS meUryuiauslWuses in RNA directed DNA methylation. Neither dπn nor emi3 mutants affected the maintenance of

pre-established RNA directed CpG methylation. The methyltransfcrases were described as appearing to act downstream of the generation of siRNAs, since diml dmi2 cmt3 triple mutants showed a lack of πυn-CpG methylation but elevated levels of siRNAs.

None of the prior art documents describe the possibility of modulating the gene-silencing effect achieved by introduction of double stranded RNA molecules or the genes encoding such dsRNA through the modulation of the functional level of particular types of Dicer- like proteins or through the modulation of genes involved in transcriptional silencing of the silencing ItNA encoding chimeric genes in plants or other cukaryotic organisms,. These and other problems have been solved as hereinafter described in the different embodiment, examples and claims.

SUMMARY OF THE INVENTION tn. one embodiment, the current invention provides the use of a eukaryotic cell or non- human organism with a modified functional level of a Dicer protein, particularly a DCL3 or DCL4 protein, to reduce the expression of a gene of interest, wherein the gene of interest is silenced in said cell by providing said cell with a gene-silencing molecule. If the eukaryotie cell is a cell other than a plant cell, the modified functional level of DCL 3 or DCL4 protein is an increased level of activity, preferably of DCL4 activity.

In another embodiment, the current invention provides the use of a plant or plant cell wiLh a modified functional level of a protein involved in processing of artificially introduced double-stranded RNA (dsRNA) molecules in short interfering RNA (siRNA), preferably a dicer-like protein such as DCL3 or DCL 4, to modulate a gene-silencing effect achieved by the introduction of a gene-silencing chimeric gene. The gene-silencing chimeric gene may be a gene encoding a silencing RNA, the silencing RNA being selected from a sense RNA, an antisense RNA, an unpolyadeπylatcd sense or antiscnse RNA, a sense or antisense RNA further comprising a largely double stranded region, hairpin RNA (hpRNA) or micro-RN A (mi RNA).

In another embodiment, the invention relates to the use of a plant or plant cell with modified functional level of a Dicer-like 3 protein to modulate the gene-sileπciπg effect obtained by introduction of silencing RNA involving a double stranded RNA during the processing of the silencing RNA into siRNA, such as a dsRNA or hpRNA. The 5 modulation of the functional level of the Dicer-like 3 may be a decrease in the functional level, achieved e.g. by mutation of the Dicer-like 3 protein encoding endogenous gene and the gene-silencing effect obtained by introduction of the silencing RNA is increased when compared to a corresponding plant or cell -wherein the Diccr-like 3 protein level is not modified. Alternatively, the modulation of the functional level of the Dicer-like 3

10 may be an increase in the functional level, achieved e.g. by introduction into the plant cell of a chimeric gene comprising opcrably linked DNA regions such as a plant-expressible promoter, a DNA region encoding a DCL3 protein and a transcription termination and polyadenylation region functional in plant cells, and the gene-silencing effect obtained by introduction of the silencing RNA is decreased when compared to a corresponding plant

15 or cell wherein the Diccr-like 3 protein level is not modified. The silencing RNA may be a dsRNA molecule which is introduced in the plant, cell by transcription in the cell of a chimeric gene comprising a plant-expressible promoter, a DNA region which when transcribed yields an RNA molecule, the RNA molecule comprising a sense and aήtisense nucleotide sequence, the sense nucleotide sequence comprising about 19 contiguous

20 nucleotides having at least about 90% to about 100% sequence identity to a nucleotide sequence of about 19 contiguous, nucleotide sequences from the RNλ transcribed from a gene of interest comprised within the plant cell; the antisensc nucleotide sequence comprising about 19 contiguous nucleotides having at least about 90 to 100% sequence identity to the complement of a nucleotide sequence of aboul 19 contiguous nucleotide

25 sequence of the sense sequence; wherein the sense and antisensc nucleotide sequence are capable of forming a double stranded RNA by basepairing with each other. Preferably, the sense and antiscnse nucleotide sequences bascpair along their full length, i.e. they are fully complementary.

■><« «71 ysi sπOt.icf iiϊc iuVcπuCϊ. proviucs 5. rπctuϋiϊ ϊ*JI icuuciiig the expression of a gene of interest in a eukaryotic cell, the method comprising the step of providing a

silencing RNA molecule to the cell, wherein said cell comprises a functional level of Dicer protein, preferably DCL3 or DCL4, which is different from the level thereof in a corresponding wiJd-type cell. The silencing RNA molecule may be any silencing RNA moteculc as described herein.

In yet another embodiment, the invention provides a method for reducing the expression

. of a gene of interest in a eukaryotic ceJ), such as a plant cell, the method comprising the step of providing a silencing RNA molecule into the cell, such as the plant cell, wherein processing of The silencing RNA into siRNA comprises a phase involving dsRNA, characterized in that the cell comprises a functional level of Dicer-like 3 protein which is modified, preferably reduced, compared to the functional level of the Dicer-like 3 protein in a corresponding wild-type cell, Preferably, when the functional level of DCO protein is reduced in a plant cell, the target gene of interest whose expression is targeted by lhe silencing RNA molecule, is an endogenous gene or transgene. Preferably, when the functional level of DCL3 protein is increased in the cell, the silencing mechanism involved in virus resistance, particularly against a virus having a double stranded RNA intermediate molecule during its lite cycle, can be increased.

The invention also provides a eukaryotic cell, preferably a plant cell comprising a silencing RNA molecule which has been introduced into the cell, wherein processing of the silencing RNA into siRNA comprises a phase involving dsRNA, characterized in that the cell further comprises a functional level of Dicer-like 3 protein which is different from the wild type functional level of Dicer-like 3 protein in a corresponding wild-type cell, The silencing RNA may be transcribed from a chimeric gene encoding the silencing RNA. The functional level of Dicer-like 3 protein may be decreased or increased, preferably increased when the cell is a cell other than a plant cell, and preferably decreased when the cell is a plant cell.

Yet another embodiment of the invention is a chimeric gene comprising the following operably linked DNA molecules: a. a eukaiyotic promoter, preferably a plant -expressible promoter

b. a DNA region encoding a Dicer-like 3 protein, preferably wherein the Dicer-like 3 protein is a protein comprising a double stranded binding domain of type 3, such as a double stranded binding domain comprising an amino acid sequence having at least 50% sequence identity to an amino acid sequence selected from lhe amino acid sequence of SEQ ID No.; 7

(At_DCX3) from the amino acid at position L436 to the amino acid at position 1563; the amino acid sequence of SEQ ID No.: 1 1 (OSJ)CL3) from the amino acid at position 1507 to the amino acid at position 1643; the amino acid sequence of SEQ ID No.: 13 (OSJDCL3b) from the amino . acid at position 1507 to the amino acid at position 1603; the amino acid sequence of SEQ ID No.: 9 (Pt_DCL3a from the amino acid at position 1561 to the amino acid at position 1669; and c. a termination transcription and polyadenylation signal which functions in a cell, preferably u plant cell. The DCL3 protein may have an amino acid sequence having at least 60% sequence identity with the amino acid sequence of SEQ ID Nos.: 7, 9, 11 or 13,

In yet another embodiment, a eukaryotio host cell, such as a plant cell, comprising a eliimeric DCL3 encoding gene as herein described is provided.

The invention also relates to the use of a plant or plant cell with modified functional level of a Dicer-like 4 protein to modulate lhe gene-silencing effect obtained by introduction of silencing RNA involving a double stranded RNA during the processing of the silencing RNA into siRNA, such as a dsRNA or hpRNA. The modulation of the functional level of the Diccr-like 4 may be decreased in the functional level (e.g. achieved by mutation of the Dicer-like 4 protein encoding endogenous gene) whereby the gene-silencing effect obtained by introduction of the silencing RNA will be decreased compared to & corresponding plant or call wherein the Dicer-like 4 protein level is not modified. Alternatively, the modulation of the functional level of lhe Dicer-lite 4 may be an increase in the ^' functional level, and wherein the gcnc-silencing effect obtained by introduction of the silencing RNA is increased compared to a plant wherein the Dicer-like

4 protein level is not modified. The increase in the functional level can be conveniently achieved by introduction into the plant cell of a chimeric gene comprising a. plant- expressible promoter opcrably linked to a DNA region encoding a DCL4 protein and a transcription termination and polyadenylacion region functional in plant cells. The mentioned silencing RNA may be a dsRNA molecule which is introduced in the plant cell by (lanscription in the cell of a. chimeric gene comprising a plant-expressible promoter; it DNA region which when transcribed yields an RNA molecule, the RNA molecule comprising a sense and aπtisense nucleotide sequence, the sense nucleotide sequence comprising about 19 contiguous nucleotides having at least about 90 to about 100% sequence identity to a nucleotide sequence of about 19 contiguous nucleotide sequences from the RNA transcribed from a gene of interest comprised within the plant cell; the antisense nucleotide sequence comprising about 19 contiguous nucleotides having at least about 90 to 100% sequence identity to the complement of a nucleotide sequence of about 19 contiguous nucleotide sequence of the sense sequence; wherein the sense and antisense nucleotide sequence are capable of forming a double stranded RNA by basepaiiϊng with each other. Preferably, the sense and antisense nucleotide sequences basepair along their full length, i.e. they are fully complementary.

It is also an embodiment of the invention to provide a method for reducing the expression of a gene, of interest in a cukaryotic cell, preferably ti plant cell, the method comprising the step of introducing a silencing RNA molecule into the cell, wherein processing of the silencing RNA into siRNA comprises a phase involving dsRNA, characterized in that the cell comprises a functional level of Dicer-like 4 protein which is modified compared to the functional level of the Dicer-like 4 protein in a corresponding wild-type celL

The invention also provides eukaryotic cells, preferably plant celJs comprising a silencing RNA molecule which has been introduced into the eell, wherein processing of the silencing RNA into siRNA comprises a phase involving dsRNA, characterized in that the cell further comprises a functional level of Dicer-like 4 protein which is different from the wild type functional Isvei of Dicar like 4 protein in a corresponding wild-type ύβll. Tlic functional level of Dtccr-like 4 protein may be decreased e.g. by mutation of the

endogenous gene encoding the Dicer-like 4 protein of a plant celf. The functional level of Dicer-Ijkc 4 protein may also be increased e.g. by expression of a chimeric gene encoding a DCM protein in a eukaryotic cell.

Yet another embodiment of the invention is a cliimeiic gene comprising the following operably linked DNA molecules: a. a eukaryotic promoter, preferably a plant -expressible promoter b. a DNA region encoding a Dicer-like 4 prolein, preferably wherein the Dicer-like 4 protein is a protein comprising a double stranded binding domain of type 4, such as a double stranded binding domain comprises an amino aciά sequence having at least 50% sequence identity to an amino acid sequence selected from the amino acid sequence of SEQ ID No.: t (At_DCL4) from the amino acid at position 1622 to the amino acid at position 1696; the amino acid sequence of SEQ ID No.: 5 (OS_DCL4) from the amino acid at position 1520 to the amino acid at position 1593; or the amino acid sequence of SEQ ID No.: 3 (Pl_DCL4) from the amino acid al position 15 (4 to the amino acid at position 1588; and c. a termination transcription and polyadenylation signal which functions in a cell, preferably a plant cell. The DCL4 protein may have an amino acid sequence having at least 60% sequence identity with the amino acid sequence of SEQ ID Nos.: 1, 3 or 15.

In yel another embodiment, a eukaryotic host cell, such as a plant, cell, comprising a chimeric DCL4 encoding gene as herein described is provided.

The invention also provides the use or ^" a eukaryotic cell with a modulated functional level of a Dicer protein to reduce the expression of a gene of interest, as well as eukaryotic cells with a modified functional level, particularly increased level, of a Dicer protein, particularly of DCI-3 or DCL4.

In yet another embodiment of the invention, a method is provided for modulating, preferably reducing the expression of a target gene in a eukaryotic cell or organism, through the introduction of a silencing RNA encoding chimeric gene into the eukaryotic cell, whereby the eukaryotic cell is modulated in genes that have an influence (e.g. through transcriptional silencing of the silencing RNA encoding chimeric genes) on the initiation υr maintenance of gene silencing by the silencing RNA encoding chimeric genes, particularly hairpin RNA encoding chimeric genes. As an example, the eukaryotic cetl may be modulated in a gene involved in RNA directed DNA methylau'on, e.g. mεthylation ai cytosines in CpG, in CpNpG or cytosincs in asymmetric context- such as the CMT3 methyltransferase or DRM mcthyltransferases in plants.

BRIEF DESCRIPTION OF THE FIGURES

Figure 1. The chromosome locations of DCL genes inArabidopsis, poplar and rice, Each chromosome is depicted approximately to scale, within a genome, with its pseudomolecule length in nucleotides provided. The number under each gene is the position on the pseudomolecule of the start of the gene. The regions shown in yellow on poplar chromosomes Vϊϊl and X represent, the large duplicated and transposed blocks that have been mapped to have been generated between 8 and 13 million years ago (Størck et cd. _f 2005).

Figure 2. Locations of domains in DCL and DCR proteins.

Schematic representation of the different domains within Diccr-Uke and Dicer genes. The linear arrangement of domains typically found in DCL or DCR proteins is depicted above the Figure. DExD: DEAD and DEAH box helicase domain; Helicase_C: Helicase C domain found in helicases and helicase related proteins; Duf283: domain of unknown function with 3 possible zinc ligaiids found in Dicer protein family; PAZ: Piwi Argonaut Zwille domain; RNAse Wl: signature of ribonudease IH proteins; dsKB: double stranded RNλ binding motif, table contains the locations, in amino acid residues, where the eight different domains can be found in a DCL or DCR molecule. Boxes that have been blacked out represent the absence or failure to detect die presence of tile domain in iuo appropriate DCL or DCR. The genes are named according to the species in which they

are found and their DCL or DC-R type. Tt: Tatrahymena thermaphUa; Cr: Chlamydomorum reinhardtiϊ, Nc: Nϋurospora crasstx', Hs: Homo sapiens; Dm: Drosophfla melanogasier; Au λrahidopsis thaliana; Qs: Oryza saliva', Pt; Populus trichocarpa. Plant gene IDs are indicated using the nomenclature in which the number preceding the "g" indicates lhe chromosome and the number after the "g" indicates the nucleotide position of the start of the coding region on the TAIR database, the JGI poplar chromosome pseudomolecules or TlOR build 3 for rice sequences, Spfl; spliceform J; Spf2: spHccfoπn 2.

Figure 3. Phylogcnelie analysis of riee, poplar and Arabidopsis.

Consensus phylogcnctie trees, constructed by neighbour-joining method with pairwise deletion, using the Dayhof matrix model for amino acid substitution, presented in radial format for [A] the entire DCL molecules and [Bl the C-terminal dsRBb domain. The colour coding shows the grouping of DCL types 1, 2, 3 and 4 based on clustering with the Arabidopsis type member. Branches with 100 percent consistence after 1000 bootstrap replications arc indicated with black dots.

Figure 4. Detection of 0sDCL2A and 0sDCL2B mjaponica and indica rice. PCR analysis of japυnica (lane I) and indica (lane 2) rice using a set of primers that should give a bund of 772 nt for the presence of OsDCLlA . and a bund of 577πt for the presence of OsDCLZB. The gel indicates that both rice subspecies contain both the 2A and 2B genes.

Figure 5. Detection of DCL3A and DCL3B genes in moπocots and their phylogenetic relationships.

[A] The phyiogcnclic analysis of the helicase- C domains of rice, maize _. , Arahidøpsis arid poplar £>CLJ-type genes, with the inclusion of their DCLl counterparts to root the tree. The analysis was done in a similar way Io that described in Fig. 2. (R] PCR analysis for the detection of DCL3A and DCISB genes in a range of monocots using A- and B^ specific primer pairs. The product from the 3£ primers were expected to be larger (~600nt) than lhe product from the detection of DCLJA (-50OnIs). Lanes 1 & 18:

markers; lanes 2, 4, 6, 10, 14 and 16 DCZJA-spccific primer pairs; lanes 3, 5, 7, 11, 15 and 17 DCZJ5-specϊtlc primer pairs. Lanes 8 and 12 negative control 3 A forward with 3Fi reverse primers; lanes 9 and 13 negative control 3B forward with 3A reverse primer pairs. Lanes 2 and 3 water control; lanes 4 and 5 rice DNA; lanes 6-9 Triticum DNA; 5 lanes 10-13 barley DNA; lanes 14 and 15 maize DNA and lanes 16 and 17 Arahidopsis DNA, The results show the detection of DCIJA and DCL3B in all of the monocots DNA tested.

Figure 6, Phylogenetic analysts of RNAse Ul domains of plants, insects and ciliates. The 1.0 analysis was done essentially as described in Figure 2. The coloured regions show that the N-terminal RNascIII domains from rice, Arabidopsia, poplar, C.eϊegans, Drosophila, and Tetrahymena all form one cluster while the Olerminal RNasclII domains show a similar counterpart cluster.

5 Figure 7. Proposed evolutionary tree of Dicer genes in plants.

The presence or absence of different DCL genes and the times of divergence of the different nodes are depicted on the currently accepted phylogenetic tree of species. Branch lengths are not to scale. The estimated large scale gcαc duplication events are depicted by blue ellipses. The numbers at the nodes and at the ellipses are estimated dales 0 in million years (my). These (lumbers are rounded to the nearest 5my, and for dates that have been previously estimated in ranges, the median of that range has been taken. The different plant DCL types are colour coded and the non-plant dicer genes are represented as while boxes. The duplication of a DCL gene is indicated by at» addition (+) sign. The phylogenetic tree with its times of divergence and large scale duplication events are based 5 on the calculations and phylogenetic trees of Blanc & Wolfe (2004) [20], Hedges et al., (2004) 1271 and Stcrck et al, (2005) [19].

Figure 8: Phenotypes of silencing achieved by a chimeric gene encoding a double stranded RNA molecule comprising complementary sense and antisense RNA targeted 0 towards phylccπc dcsaturasc (PDS-hρ) in Arabirfύjjάis seedlings of different generic backgrounds. WT: wild type A. thaliana (without PDS-hp); WT PDS-lrp: Wild type

A.thaliana. with PDS-hp gene, dcl2: mutant A. thaliana wherein Dicer like 2 gene is inactivated. DcB: mutant A. thaliana wherein Dicer like 3 gene is inactivated. Dcl4: mutant A. thaliana wherein Dicer like 4 gene is inactivated. The degree of bleaching is a measure of the degree of silencing.

Figure 9: The effect of CMT3 mutation on hpKNA-mediatcd EIN2 and CHS silencing. Left panel: The length of hypocotyls grown in the dark on ACC containing medium, is generally longer for the F3 hpElN2 plants with the homozygous cmt3 mutation than with the wild-type background (wt), indicating stronger EIN2 silencing in the cmt3 background. The transgenic plants inside the box have the mutant background, while the transgenic plants outside the box have the wild-type background.

Right panel: the seed coat color is significantly lighter for the hpCηS plants with lhe cmt3 background than with the wild-type background, indicative of stronger CHS silencing in the former transgenic plants.

Table 1. Variation within and between DCLs of rice, poplar and Arabidυpsis. The variations are given as number of amino acid changes (to the αearesl integer), and were calculated using MEGA 3.1 using the complete deletion option and assuming uniform rates among sites. The number in brackets indicates the standard error (to the nearest integer). The variability between DCLs is net variability.

Table 2. Pairwisc distances between DCLS of rice, poplar and Arabidopsis.

DETAILED DESCRIPTION OF THE INVENTION

The current, invention is based on the demonstration by the inventors that modulating the functional level of several types of Dicer-like proteins in eukaryotic cells, such as plants modulates the gcnc-silcncing effect achieved by the introduction of double stranded RNA molecules, particularly hairpin RNA into such cells. In another aspect, the invention is based on the demonstration by the inventors that ,thc gene-silencing effect achieved by silencing RNA-encoding chimeric genes, particularly hairpin RNA encoding chimeric

genes, can be modulated by modulating genes in eukaryolic cells which influence the initiation or maintenance of gene silencing.

In particular, it was demonstrated that gene-silencing achieved by chimeric genes encoding a double stranded RNA molecule (paiticulafly a hpRNA) in plant cells lacking functional DCL3 protein was unexpectedly enhanced. Further it was also found that gene- silencing achieved by chimeric genes encoding a double stranded RNA molecule, particularly a hpRJNA molecule, in plant cells lacking functional T)ClA protein was reduced leading to the realization thai increase in the functional level of DCL4 protein could lead to a stronger gene-silencing effect achieved by introduction of double-stranded RNA molecules into such plant cells. In addition, it was demonstrated that gene-silencing achieved by chimeric genes encoding a double stranded RNA molecule (particularly a hpRNA) in plant cells lacking functional CMT3 methyltransferase protein was unexpectedly enhanced.

Accordingly, the invention provides a method for modulating the gcnc-silenciπg effect in a eukaryolic cell or organism achieved by introduction of a gene silencing molecule, such as a gene-silencing RNA preferably encoded by a gene-silencing chimeric gene, by modulation or alteration of the functional level of a Dicer protein, including a DCL protein, such as DCL3 or DCL4, which Dicer protein or DCL protein is involved, directly or indirectly, in processing of artificially introduced dsRNA molecules, particularly of hpRNA molecules, particularly long hpRNA molecules into short-interfering siRNA of 21-24 nt.

As used herein, "artificially introduced dsRNA molecule" refers to the direct introduction of dsRNA molecule, which may e.g. occur exogenυualy, i.e. after synthesis of the dsRNA outside of tile cell, or endogenously by transcription from a chimeric gene encoding such dsRNA molecule, however it does not refer to the conversion of a single stranded RNA molecule into a dsRNλ inside the eukaryotic cell or plant cell.

As used herein, a "Dicer protein" is a protein having ribonuclcase activity which is involved in the processing of double stranded RNA molecules into short interfering RNA (siRNA). The ribonuclcase activity is so-called ribonuclcase III activity, which predominantly or preferentially cleaves double stranded RNA substrates rather than siπglc-stxanded RNA molecules, thereby targeting the double stranded portion of a RNA molecule. Typically, the double stranded RNA subsfrate comprises a double stranded region having at least 19 contiguous basepajrs. Alternatively, the double stranded RNλ substrate may be a transcript which is processed to form a mi RNA. The term Dicer includes Dicer-like (DCL) proteins which are proteins that show a high degree of similarity to Dicers and which are presumed to be functional based on their expression in a cell. Such relationships may readily be identified by those skilled in the art. Dicer proteins are preferentially involved in processing the double-stranded KNA substrates into siRNA molecules of about 21 to 24 nucleotides in length.

As used herein "gene-silencing effect" refers to the reduction of expression of a target nucleic acid in a host cell, preferably a plant cell, which can be achieved by introduction of a silencing RNA. Such reduction may be the result of reduction of transcription, including via mcthylatiϋn and/or chromatin remodeling., or post-transcriptional modification of the RNA molecules, including via RNA degradation, or both. Gene- silencing should not necessarily be interpreted as an abolishing of the expression of the target nucleic acid or gene. It is sufficient that the level expression of the target nucleic acid in the presence of lhe silencing RNA is lower that in the absence thereof. The level of expression may be reduced by at least about 10% or at least, about 15% or at least about 20% or at least about 25% or at least about 30% or at least about 35% or at least about 40% or at least about 45% Or at least about 50% or at least about 55% or at least about 60% or at least about 65% or at least about 70% or at least about 75% or at least about 80% or at least about 85% or at least about 90% or at least about 95%. or at least about 100%. Target nucleic acids may include endogenous genes, transgcnes or viral genes or genes introduced by viral vectors. Target nucleic acid may further include genes which arc siabiy introduced in the host's ceil genome, preferably the !iυ-ύ cell's nuclear genome. Preferably, gene silencing is a sequence-specific effect, wherein expression of

the target nucleic acid is specifically reduced compared to other nucleic acids in the cell, although the target nucleic acid may represent a family of related sequences.

As used herein, "silencing RNA" or silencing RNA molecule refers to any RNA molecule which upon introduction into a host cell, preferably a plant cell, reduces the expression of a target gene. Such silencing RNλ may e.g. be so-called "antisense RNA", whereby the RNA molecule comprises a sequence of at least 20 consecutive nucleotides having at least 95% sequence identity to the complement of the sequence of the target nucleic acid, preferably lhe coding sequence of the target gene. However, antisense RNA may also be directed to regulatory sequences of target genes, including the promoter sequences and transcription termination and polyadcnylation signals. Silencing RNA further includes so-called "sense RNA" whereby the RNA molecule comprises a sequence of at least. 20 consecutive nucleotides having at least 95% sequence identity to the sequence of the target nucleic acid. Without intending to limit lhe invention to any particular mode of action, it is generally believed that single stranded silencing RNA such as antisense RNA or sense RNA is converted into a double stranded intermediate e.g. through lhe action of RNA dependent RNA polymerase, whereby the double stranded intermediate is processed to form 21-24 nt short interfering RNA molecules,

The mentioned sense or antisense RNA may of course be longer and be about 50 nt, about lOOnt, about 200 nt, about 300nt, about 500nl, about 1000 nt, about 2000 nt or even about 5000 nt or larger in length, each having an overall sequence identity of respectively about 40 %, 50%, 60 %, 70%, 80%. 90 % or 100% with the nucleotide sequence of the target nucleic acid (or its complement) The longer the sequence, the less stringent the requirement for the overall sequence identity. However, the longer sense or antisense RNA molecules with less overall sequence identity should at least comprise 20 consecutive nucleotides having al least 95% sequence identity to the sequence of the target nucleic acid or its complement.

Other silencing RNλ may be "uiipolyadeπylated RNA" cύmμusiiiB ui. ϊeasi 20 consecutive nucleotides having at least 95% sequence identity to the complement of lhe

sequence of the target nucleic acid, such as described in WO01/12824 or US6423885 (both documents herein incorporated by reference). Yet another type of silencing RNA is an RNA molecule as described in WO03/076619 or WO2005/026356 (both documents herein incorporated by reference) comprising at least 20 consecutive nucleotides having at Ieast95% sequence identity to the sequence of the target nucleic acid or the complement thereof, and further comprising a largely-double stranded region us described in WO03/07.6619 or WO2005/026356 (including largely double stranded regions comprising a nuclear localization signal from a virøtd of the Potato spindle tuber viroid-lype or comprising CUG trinucleotide repeats). Silencing RNA may also be double stranded RNA comprising a sense and antiscnse strand as herein defined, wherein the sense and antisense strand arc capable of base-pairing with each other to form a double stranded RNA region (preferably the said at least 20 consecutive nucleotides of die sense and antisense RNA are complementary to each other. The sense and antisense region may also be present within one RNA molecule such that a hairpin RNA (hpRNA) can be formed when the sense and antisense region form a double stranded RNA region. hpRNA is well-known within the art (see e.g WO99/53050, herein incorporated by reference). The hpRNA may be classified as long hpRNA, having long, sense and antisense regions which can be largely complementary, but need not be entirely complementary (typically larger than about 200 bp, ranging between 200-1000 bp), hpRNA can also be rather small ranging in size from about 30 to about 42 bp, but not much longer than 94 bp (sec WO04/073390, herein incorporated by reference). Silencing RNA molecules could also comprise so-called micrυRNA or synthetic or artificial micToRNA molecules or their precursors, as described e.g. in Schwab et al. 2006, Plant Cell 18(5): 1121-1 133,

Silencing RNA can be introduced directly into the host cell after synthesis outside of the cell, or indirectly through transcription of a "gene-silencing chimeric gene" introduced into the host cell such that expression of the chimeric gene from a promoter in the cell gives rise to the silencing R-NA. The gene-silencing chimeric gene may be introduced stably into the host cell's (such as a plant cell) genuine, preferably nuclear gcuυme, or it may be introduced transiently. The silencing RNA molecules are preferably introduced

into the host cell, Or heterologous silencing RNA molecules, or silencing RNA molecules non-πaturally occurring in the eukaryotie host ceil, or artificial silencing RNA molecules.

As used herein, "modulation of functional level" means cither an increase or decrease in the functional level of the concerned protein. "Functional level" should be understood to refer to the level υf active protein, in casu the level of protein capable of performing the ribonucleic III activity associated with Dicer or DCL. The functional level is a combination of the actual level of protein present in the host cell and the specific activity of the protein. Accordingly, the functional level may e.g. be modified by increasing or decreasing the actual protein concentration in the host cell. The functional level may also be modulating the specific activity of the protein. Such increase or decrease of the specific activity may be achieved by expressing a variant protein, such as a non-nalurally occurring or man-made variant with higher or lower specific activity (or by replacing the endogenous gene encoding the relevant. DCL protein with an allele encoding such a variant). Increase or decrease of the specific activity may also be achieved by expression of an effector molecule, such as e.g. an antibody directed towards such a DCL protein and which affects the binding of dsRN A molecules or the catalytic RNAse 111 activity.

Increase of DCL3 activity in a plant cell will lead to a reduced gene silencing effect achieved by silencing RNA, the processing of which involves a dsRNA molecule, including sense RNA, aπlisense RNA, unpolyadeπylated sense and antisense RNA, sense or antisense RNA having a largely doubled stranded RNA region, and double stranded

RNA comprising u sense and aniisense regions which are capable of forming a ds stranded RNA region, particularly silencing RNA targeted to reduce the expression of endogenous genes, or trangenes. In the case of virus resistance, particularly where the virus has a double-stranded RNA phase, the gene silencing effect may be enhanced.

Decrease of the DCL 3 activity will yield to an enhanced silencing effect achieved by silencing RNλ, particularly silencing RNA targeted towards endogenes or transgenes, but

^" may result in reduced gene silencing for viral nucleic acids. Inversely, increase of DCL4 actsv'ty i" a plant Cell will lcndcd to iiicicase the gene sll&iiύiiig cfleci achieved by the

silencing RNA, while decrease of DCL4 activity will yield a reduced gene silencing effect.

Increase of DCL activity can be conveniently achieved by overexpression, i.e. through 5 the introduction of a chimeric gene into the host cell or plant cell comprising a region DNA region coding for an appropriate DCL protein operably linked to a promoter region and transcription termination and polyadcnylation signals functional in the host cell or the plant cell. Increase can however also be achieved by mutagenesis and selection- identification of the individual host/plant cell, host/plant cell line or host/plant having a 10 higher activity of the DCT. protein than the starting material .

A decrease in DCL activity can be conveniently achieved by mutagenesis and selection- identification of the individual host/plant cell, host/plant ceil line or host/plant having a lower activity of the DCL protein than the starting material. A decrease in DCL activity

15 can also be achieved by gene-silencing whereby the targeted gene whose expression is to be? reduced is the gene encoding the DCL protein. In case of reduction of DCL3 gene expression through gene silencing the silencing RNA could be any silencing RNA which is processed into a dsRNA form during siRNA genesis. Downrcgulation of DCJ-4 gene expression however will require use of an alternative gene-silencing pathway such as use

20 of artificial uαicro-RNA molecules as described e.g. in WQ2005/052170, WO2005/047505 or US 2005/0144667 (all documents incorporated herein by reference)

As indicated above, "Dicer or Diccriike proteins involved in processing of artificially introduced dsRNA molecules" include DCI, 3 and DCL4 proteins. As used herein a 25 "plant dicer" or plant "dicer-Jike" protein is a protein having riboπuclease activity on double stranded RNA substrates (ribonuclease JlI activity) which is characterized by flic presence of at least the following domains: a DExD or DExH domain (DEAD/DEAH domain), a Helicase-C domain, preferably a Duf283 domain which may be absent, a PAZ domain, two RNAse HI domains and at least one and preferably 2 dsRB domains.

^■ 3λ

Uelicase C: The domain, which defines this group of proteins is found in a wide variety of helicases and heiicase related proteins. It may be that this is not an autonomously folding unit, but an integral part of the heiicase (PF00271; IPR001650)

PAZ domain: This domain is named after the proteins Piwi Argonaut and ZwiJJe. It is also found in the CAF protein from Arabidopsis thaltana. The function of the domain is unknown but has been found in the middle region of a number of members of the Argonaute proLein family, which also contain the Piwi domain in their C-teπninal region. Several members of this family have been implicated in the development and maintenance of stem cells through the RNA-mediated gene-quelling mechanisms associated with the protein Dicer. (PF02I70; IPR003100)

Diif283: This putative domain is found in members of the Dicer protein family. This protein is a dsRNλ nuclease that is involved in RNAi and related processes. This domain of about 100 amino acids has no known function, but does contain 3 possible zinc ligands.(PF03368, IPR005034).

DExD: Members of this family include the DEAD and DEλH box helicases. Helicascs are involved in unwinding nucleic acids. The DEAD box heiicases are involved in various aspects of RNA metabolism, including nuclear transcription, pre mRNA splicing, ribosotnc biogenesis, nucleocytoplasmic transport, translation, RNA decay and organellar gene expression (PF00270, TPROl 1545).

RNAse HI: signature of the ribonuclease IH proteins (PF00636, IPR000999)

DsRB (Double stranded RNA binding motif): Sequences gathered for seed by HMMUterativejtraining Putative inolif shared by proteins that bind to dsRNA. At Ieasl some DSRM proteins seem to bind to specific RNλ targets. Exemplified by vStaufcn, which is involved in localisation of at least five different mRNAs in the early Drosophila cniu»ryo. j uoC u»y in<.oricrϋij~ϊϊϊuϋccu. ρrθιCϋϊ iCiuαSC ϊϋ ijUϊϊiαπs, wnicu ϊS pcirt oi me cellular response io dsRNA (PF0O035, MlOO 1 159).

These domains can easily be recognized by computer based searches using e.g. PROSlTE profiles PD0C5082I (PAZ domain), PDOC00448 (RNase III domain), PDOC50137

(dsRB domain) and PDOC00039 (DExD/DexH domain) (PROSlTE is available at

5 www.expasy.ch/prosite). Alternatively, the BLOCKS database and algorithm

(blocks.fhcrc.org) may be used to identify PAZ(1PB003100) or DUF283(IPB005034) domains. Other databases and algorithms are also available (pFAM; http://www.sangcr.ac.ulc/Software/Pfiim/ INTERl ¹ RO: hup://www.cbi.ac.uk/interpro/; the above cited PF numbers refer Io pFAM database and algorithm and IPR numbers to

I O lhe INTRRPRO database and algorithm).

Typically, a DCL2 protein will process double stranded RNA into short interfering RNA molecules of about 22 nucleotides, a DCL3 protein will process double stranded RNA into short interfering RNA molecules of about 24 nucleotides, and DCL4 will process 15 double stranded RNA into short interfering RNA molecules of about 21 nucleotides.

As used herein a "Dicer-like 3 protein (DCL3)" is a plan! dicer-likc protein further characterized, in that it has two dsRB domains (dsRBii and dsRBb) wherein the dsRBb domain is of type 3. Preferably, dsRBb has an amino acid sequence having al least 50% 0 sequence identity to an amino acid sequence selected from the following sequences: the amino acid sequence of SEQ ID No.: 7 (At_DCL3) from the amino acid at position 1436 to the amino aeid at position 1563;

- the amino acid sequence of SEQ ID No.: 1 1 (0S_DCL3) from the amino acid at position 1507 to the amino acid at position 1643; 5 - tile amino acid sequence of SEQ ID No.: 13 (OS_DCL3b) tVora the amino acid at position 1507 to the amino acid at position 1603;

- the amino acid sequence of SEQ JLD No.: 9 (H-DCLSa) from the amino actd at position 1561 to the amino acid at position 1669.

0 The dsRBb domain may of course have a higher sequouυe identity to ihe cited dsRBb domains such as at least 55%, at least 60%, at least 65%, at least 70%, at least 75%, at

least 80%, at least 85%, at least 90%, at least 95% or be identical with the cited amino acid sequences.

Nucleotide sequences encoding Dicer-like 3 enzymes can also be identified as those nucleotide sequences encoding a Dicer-like enzyme and which upon PCR amplification with a set of DCL3 diagnostic primers such as primers having the nucleotide sequence of

SEQ ID No.: 31 and SEQ ID No.: 32 yields a DNA molecule of about 600 nt in length or upon PCR amplification with a set of DCL3 diagnostic primers such as primers having the nucleotide sequence of SEQ ID No.: 35 and SEQ ID No,: 36 yields a DNA molecule or upon PCR amplification with a set of DCL3 diagnostic primers such as primers having tJ\c nucleotide sequence of SEQ ID No.: 37 and SEQ ID No.: 38 yields a DNA molecule.

Fragments of nucleotide sequences encoding Diccr-Iike 3 enzymes can further be amplified using primers comprising the nucleotide sequence of SEQ ID No.: 15 and SEQ ID No.: 16 or .the nucleotide sequence of SEQ ID No-: 17 and SEQ ID No.: 18 or the nucleotide sequence of SEQ ID No,: 19 and SEQ ID No.: 20 or the nucleotide sequence of SEQ ID No.:- 21 and SEQ ID No.: 22. The obtained fragments can be joined to each other using standard techniques. Accordingly, suitable DCL3 encoding nucleotide sequences may include a UNA nucleotide sequence amplifϊable with the primers of SEQ ID No.: 15 and S.ϊϊQ ID No.: 16 .: or with primers of SBQ ID No.: 17and SEQ ID No.: 18 or with primers of SEQ TD No,: 19 and SEQ ID No.: 20 or with primers of SEQ ID No.:2l and SEQ ID No.: 22.

Further suitable nucleotide sequences encoding Diccr-likc 3 enzymes are those which encode a protein comprising an amino acid sequence of at least about 60% or at least

^• about 65% or at least about 70% or at least about 75% or at least about 80% or at least about 85% or at least about 90% or at least about 95% sequence identity or being essentially identical with the proteins comprising an amino acid sequence of SEQ ID

Nos.: 7 or 9 or 11 or 13 or with the proteins having amino acid sequences available from databases with the following accession numbers: NP_J 89378.

Such nucleotide sequences include the nucleotide sequences of SEQ ID Nos.: 8 or 10 or 12 or 14 or nucleotide sequences with accession numbers: NM_l 14260 or nucleotide sequences encoding a dicer-like 3 protein, wherein the nucleotide sequences have at least about 60% or at least about 65% or at least about 70% or at least about 75% or at least about 80% or at least about 85% or at least about 90% or at least about 95% sequence identity to lhese sequences or being essentially identical thereto.

As used herein a "Diccr-likc 4 proiein (DClA)" is a plant dicer-like protein further characterized in that it has Wo dsRB domains (dsRBa and dsKJBb) wherein the dsRBb domain is of type 4. Preferably, dsRBb has an amino acid sequence having at least 50% sequence identity to an amino acid sequence selected from the following sequences: the amino acid sequence of SEQ ID No.: 1 (At_DCL4) from the amino acid at position 1622 to the amino acid at position 1696;

- the amino acid sequence of SEQ ID No.; 5 (OS_DCL4) from the amino acid at position 1520 To The amino acid at position 1593; or

- the amino acid sequence of SEQ ID No.: 3 (Pt_DCL4) from the amino acid at position 1514 to the amino acid at position 1588.

The dsRBb domain may of course have a higher sequence identity Io the cited dsRBb domains such as at least 55%, at least 60%, at least 65%, at least 70%, at least 75%, at least 80%, at leasr 85%, ac least 90%, at least 95% or be identical with the cited amino acid sequences.

Nucleotide sequences encoding Dicer-like 4 enzymes can also be identified as those nucleotide sequences encoding a Dicer-like enzyme and which upon PCR amplification with Ά set of DCL4 diagnostic primers such as primers having the nucleotide sequence of SKQ ID JNo.; 33 and SEQ ID No.: 34 yields a DNA molecule, preferably of about 920 bp or about 924 bp in length.

F"αgtϊ5ci«ts of nucleotide Sequences encoding Dic&r-iikc 4 enzymes c«n further be amplified using primers comprising the nucleotide Sequence of SHQ ID No.: 23 and SEQ

ID No.: 24 or the nucleotide sequence or SEQ ID No.; 25 and SEQ ID No.: 26 or the nucleotide sequence of SEQ ID No.: 27 and SEQ ID No.: 28 or the nucleotide sequence of SRQ ID No.: 29 and SEQ ID No.: 30, The obtained fragments can be joined to each other using standard techniques. Accordingly, suitable DCL4 encoding nucleotide sequences may include a DNA nucleotide sequence amplifiable with the primers of SEQ ID No.: 23 and SEQ ID No.: 24 or with primers of SEQ ID No.: 25 and SEQ ID No.: 26 or with primers of SEQ BD No.:27 and SEQ ID No.: 28 or with primers of SEQ ID No.: 29 and SEQ ED No.: 30.

Further suitable nucleotide sequences encoding Dicer-like 4 proteins are those which encode a protein comprising an amino acid sequence of at least aboul 60% or 65% or 70% or 75% or 80% or 85% or 90% υr 95% sequence identity or being essentially identical with the proteins comprising an amino acid sequence of SEQ ID Nos.: I or 3 or 5 or with the proteins having amino acid sequences available from databases with the following accession numbers: AλZ80387; P84634.

Such nucleotide sequences include lhe nucleotide sequences of SEQ ID Nos.: 2 or 4 or 6 or nucleotide sequences with accession numbers: NM_122039; DQl 18423 or nucleotide sequences encoding a dicer-like 4 protein, wherein the nucleotide sequences have at least about 60% or at least aboul 65% or at least about 70% or at least about 75% or at Jcast about 80% or at least about 85% or at least about 90% or at least about 95% sequence identity to these sequences or being essentially identical thereto.

For the purpose of this invention, the "sequence identity" of two related nucleotide or amino acid sequences, expressed as a percentage, refers to the number of positions in the two optimally aligned sequences which have identical residues (xlOO) divided by the number of positions compared, A gap, i.e., a position in an alignment where a residue is present in one sequence but not in the other is regarded as a position with non-identical residues. The alignment of the two sequences is performed by the Needlαmaπ and Wunsch algorithm (Nccdlemari arid Vfimsdi 3970) The computer-assisted sequence alignment above, can be conveniently performed using standard software program such

as CAP which is part of the Wisconsin Package Version 10.1 (Genetics Computer Group, Madϊsion, Wisconsin, USA) using (he default scoring matrix with a gap creation penally of 50 and a gap extension penally of 3. Sequences are indicated as "essentially similar" when such sequence have a sequence identity of at least about 75%, particularly at least about 80 %, more particularly at least about &5%, quite particularly about 90%, especially abo.ut 95%, more especially about 100%, quite especially are identical. It is clear than when RNA sequences are the to be essentially similar or have a certain degree of sequence identity with DNA sequences, thymine (T) in the DJslA sequence is considered equal to uracil (U) in the RNA sequence. Thus when it is stated in this application that a sequence of 19 consecutive nucleotides has at least 94% sequence identity to a sequence of 19 nucleotides, this means that at least IX of the 19 nucleotides of the first sequence ate identical to 18 of the 19 nucleotides of the second sequence.

In one embodiment of the invention, a method for reducing the expression of a nucleic acid of interest in a host cell, preferably a plant cell is provided, the method comprising the step of introducing a dsRNA molecule into a host cell, preferably plant cell, said dsRNA molecule comprising a sense and antisense nucleotide sequence, whereby the sense nucleotide sequence comprises about 19 contiguous nucleotides having at least about 90 to about 100% sequence identity to a nucleotide sequence of about 19 contiguous nucleotide sequences from the RNA transcribed (or replicated) from (he nucleic acid of interest and the antisense nucleotide sequence comprising about )9 contiguous nucleotides haying at least about 90%, such as about 94% to 100% sequence identity to the complement of a nucleotide sequence of about 19 contiguous nucleotide sequence of the sense sequence and wherein said sense and antisense nucleotide sequence are capable of forming a double stranded RNA by basepairing with each other, characterized in that the host cell, preferably a plant cell comprises a functional level of Dicer-like 4 protein which is modified compared to the functional of said Dicer-like 4 protein in a wild-type host cell, preferably a plant cell. The functional level Dicerlike 4 protein can be increased conveniently by introduction of a chimeric gene comprising a pi'ύiϊiύtcr i'OgrOu Slid « tuiiiSCiiμuυu ieiiπiiuύiυii and puiyaueiiyituioii signal υpeπibiy

linked to a DNA region coding for a DCL4 protein, the latter being as defined elsewhere in this application.

As used herein, the term "promoter" denotes any DNA which is recognized and bound (directly or indirectly) by a DNA-dependeπt RNλ-polyrncrase during initiation of transcription. A promoter includes the transcription initiation site, and binding sites for transcription initiation factors and UNA polymerase, and can comprise various other sites (e.g., enhancers), at which gene expression regulatory proteins may bind.

The term "regulatory region", as used herein, means any DNA, that is involved in driving transcription and controlling (i.e., regulating) the timing and level of transcription of a given DNA sequence, such as a UNA coding for a protein or polypeptide. For example, a 5' regulatory region (or "promoter region") is a DNA sequence located upstream (i.e., 5') of a coding sequence and which comprises the promoter and the 5'-uπtπιnsϊatcd leader sequence. A 3' regulatory region is a DNA sequence located downstream (i.e.. 3') υf the coding sequence and which comprises suitable transcription termination (and/or regulation) signals, which may include one or more polyadenylation signals.

In one embodiment of the invention the promoter is a constitutive promoter. In another embodiment of the invention, the promoter activity is enhanced by external or internal stimuli (inducible promoter), such as but not limited to hormones, chemical compounds, mechanical impulses, abiotic or biotic stress conditions. The activity of the promoter may also be regulated in a temporal or spatial manner (tissue-specific promoters; developmental^ regulated promoters). The promoter may be a viral promoter or derived from a viral genome.

In a particular embodiment of the invention, the promoter is a plant-expressible promoter. As used herein, the term "plant-expressible promoter" means a DNA sequence that is capable of controlling (initialing) transcription in a plant cell. This includes any promotϋr of plant origin, bι:! -also any promoter of πcn-piani o.-jμu whidi is capable of directing transcription in a plant cell, i.e., certain promoters of viral υr bacterial origin such as the

CaMV35S (Hapster et al., 1988), the subterranean clover virus promoter No 4 or No 7 (WO9606932). υr T-DNA gene promoters but also tissue-specific or organ-specific promoters including but not limited to seed-specific promoters (e.g., WQ89/03887), organ-primordia specific promoters (An ct al., 1996), stem-specific promoters (Keller et al., 1988), leaf specific promoters (Hudspelh el al., 1989), mesophyl -specific promoters (such as the lighl-inducible Rubisco promoters), root-specific promoters (Keller et al.,1989), luber-specific promoters (Keii ct al., 1989), vascular tissue specific promoters (Peleman et al., 1989), stamen-selective promoters (WO 89/10396, WC) 92/13956), dehiscence /one specific, promoters (WO 97/ 13865) and the like.

In another embodiment of the invention, a method for reducing the expression of a nucleic acid of interest in a host cell, preferably a plant cell is provided, the method comprising the step of introducing a dsRNA molecule into a host cell, preferably plant cell, said dsRNA molecule comprising a sense and anlisense nucleotide sequence, whereby the sense nucleotide sequence comprises about 19 contiguous nucleotides hiiYϊπg at least about 90%, such as at least 94%, to about 100% sequence identity to a nucleotide sequence of about 19 contiguous nucleotide sequences from the RNA transcribed (or replicated) from the nucleic acid of interest and the antisense nucleotide sequence comprising about 19 contiguous nucleotides having at least, about 90%, such as about 94% to aboul 100% sequence identity to the complement of a nucleotide sequence of about 19 contiguous nucleotide sequence of the sense sequence and wherein said sense and antisense nucleotide sequence are capable of forming a double stranded RNA by bascpairing with each other, characterized in that the host cell, preferably a plant cell comprises a functional level of Dicer-like 4 protein which is reduced compared to the functional level of said Dicer-like 4 protein in a corresponding wild-type host cell, preferably a plant cell. Such a reduction could be achieved by mutagenesis of host cells or plant cells, host cell lines or plant cell lines, hosts or plants Or seeds, followed by identification of those host cells or plant cells, host cell lines or plant cell lines, hosts or plants or seeds wherein the Dicer-like 4 activity has been reduced or abolished. Mutants having a deletion or other Jesion in the DCI- 4 euuϋJing gene can wπvcnienily be

recognized using e.g. a method named "Targeting induced local lesions IN genomes (TILLING)". Plant Physiol. 2000 Jun;123(2):439-42 .

Preferably, the sense and antisense nucleotide sequences of dsKNA molecules as described herein bascpair along their full length, i.e. they are fully complementary.

"Basepairing" as used herein includes G:U basepairs as well as A:U and G:C basepairs.

Alternatively, the dsRNA molecules may be a transcript which is processed to form a miRNA. Such molecules typically fold to form double stranded regions in which 70-95% of the nucleotides are basepaired, e.g. in a region of 20 contiguous nucleotides, 1-6 nucleotides may be non- basepaired.

In yet another embodiment of the invention, the use of a plant or plant cell with a modified functional level Of ¹ DCO protein is provided to modulate the gene silencing effect obtained by introduction of silencing RNA requiring a double stranded RJSfA phase during processing into siRNA such as e.g. dsRNA or hpRNA or genes encoding such silencing RNA. A preferred embodiment of the invention is the use of a plant or plant cell wilh a reduced level of DCL3 protein, particularly a plant or plant cell which does not contain functional DCL3 protein. Gene silencing using silencing RNA requiring a double stranded RNA phase during die processing into siRJMA is enhanced in such a genetic background.

In yet another embodiment of the invention, the use of a plant or plant cell with a modified functional level of DCL3 protein is provided to modulate virus resistance of such a plant cell. A preferred embodiment, of the invention is the use of a plant or plant cell with an increased level of DCL3 protein.

Although not intending to limit the invention to a particular mode of action, it may be that the enhanced gene-silencing effect for endogeπe or transgeπe silencing is due to reduced transcriptional silencing of the silencing RNA, particularly hpRNA, encoding transgenes

mutants where transcriptional silencing is relieved such as inpol iv and rdr2 background.

However, DCL3 may also cleave hpRNA stems compromising RNAi by removing substrate that wouid otherwise be processed by DCL2 and DCL4 into 21 and 22 nt siRNλ molecules. It has been demonstrated that silencing of the target gene by silencing RNλ, particularly hpRNA, encoding transgcncs by is enhanced in silencing deficient mutants where transcriptional silencing is relieved including rdτ2 and cmt3 background.

A deli genetic background in a plant cell, which is suitable for the methods according to the invention can be conveniently achieved by insertion mutagenesis (e.g. using a T-DNA or transposon insertion mutagenesis pathway, whereby insertions in the region of the endogenous DCL3 encoding gene are identified, according to methods well known in the art. Similar genetic dcl3 genetic background can be achieved using chemical mutagenesis whereby plants with a reduced level of DCL3 are identified. Plants with a lesion in the genome region of a DCJL3 encoding gene can be conveniently identified using the so- called TILLING methodology (supra).

DCL3 alleles can also be exchanged for less or non-functional DCL3 encoding alleles through homologous recombination methods using targeted double stranded break induction (e.g. with rare cleaving double stranded break inducing enzymes such as homing endonucl eases).

Preferred, less functional, mutant alleles are those having an insertion, substitution or deletion in a conserved domain such as the OϋxD, Helicase-C, Duf 283, PAZ, Rnaselll and dsRB domains whose location in the different identified DCL3 proteins is indicated in Figure 2.

The methods according to the invention can be used in various ways. One possible application is the restoration of weak silencing loci obtained by introduction of chimeric genes yielding silencing RNA, preferably hpRNA, into cells of a plant, by introduction of such weak silencing loci into a dcl3 genetic background (with reduced functional level of DCL3 or into DCL1 overexpressing background. Another utility of the methods of the invention is the reversion of progressive loss over generations of certain silencing loci

which can sometimes be observed, by introduction into a dcl3 background. The methods of the invention can thus be used Io increase the stability of silencing loci in host cells, particularly in plant cells.

Jt will be clear that the invention also relates to modifying the gene-silencing effect achieved in eukaryotic cells such as plant, cells, by modifying the functional Jevel of more than one Dicer protein.

In one embodiment of the invention, eukaryotic cells are provided wherein the functional level of DCL 3 is decreased and the functional level of DCL4 is increased; in another embodiment eukaryotic cells are provided wherein lhe functional level of both DCL2 and DQJA are decreased or increased. Plant cells with a reduced level or functional level of DCL2 and DCL4 protein may be used to increase viral replication in such cells.

In another aspect of lhe invention, a method is provided for reducing the expression of a target gene in a cukaiyotic cell or organism, particularly in a plant cell or platit, comprising the introduction of a silencing RNA encoding chimeric gene, as herein defined, into said cell or organism, characterized in that the cell or organism is m<χlulated in the expression of genes or the functional level of proteins involved in the transcriptional silencing of said silencing RNA encoding chimeric gene.

One example of a class of genes involved in transcriptional silencing arc the mcthyltransfcrases controlling RNA-directed DNλ methylatdon, such as the MET class, the CMT class and the DRM class (Fπmegan and Kovac 2000 Plant MoI. Biol. 43, 189- 201, herein incorporated by reference), JVTETI in Ambidopsis, like its mammalian homolog Dnrntl (Bcstor et at. 19X8, J. MoI. Biol. 203, 971-983) or corresponding genes in other cells encodes a major CpG maintenance methyltransfeπise (Finnegan Ct. al. 1996, Proc. Natl. Acad- Scϊ. USA 93, 8449-8454; Ronemus et al. 1996, Science 273, 654-657; Kishimoto ct al. Plant MoI. Biol. 46, 171-183). CMT-like genes are specific to the plant kingdom ar.d encode rnsihyllransferaee prot≥itis containing a chromodomasn (Her.ikoff and Comai, 1998, Genetics 349, 307-318). The DRM genes share homology with

mammalian DπillG genes that encode de novo methyltransferases (Cao et al. 2(H)O, Proc; Nail. Acad. Sci. USA 97, 4979-4984).

Methods to reduce or inactivate die expression of mcthyltransfcrascs are as described elsewhere in this document concerning the Dicer-like proteins. The nucleotide sequences and amino acid sequences of mcthyltransfer&ses in plants are known and include NPJL7713S, AAK69756, AAK71870, AAK69757, NP_199727, NP_OOIO59O52 and others (herein incorporated by reference). Methods to identify the endogenous homologucs of the above mentioned specific methyltransferases and encoding genes are known in the art and may be used to identify nucleic acids encoding proteins having at least 50%, 60%, 70%, 80%, 90%, 95% sequence identity with the above mentioned amino acid sequences, variants thereof as well as mutant, less or non-functional variants thereof.

Another class of genes involved in transcriptional silencing includes the RDR2 (RNA dependent polymerase) genes and polIV (DNA polymerase IV) genes (also named NRPD l a/SDE4 and NRDP2a) (Elmayan et al. 2005, Current Biology 15, 1919-1925 and references therein). The amino acid sequences for these proteins are known and include NP_192851 and ABL68089 (herein incorporated by reference). Methods to identify the endogenous homologucs of the above mentioned specific polymerases and encoding genes arc known in the art and may be used to identify nucleic acids encoding proteins having at least 50%, 60%, 70%, 80%, 90%, 95% sequence identity with the above mentioned amino add sequences, variants thereof as well as mutant, Jess or nonfunctional variants thereof.

Having read the exemplified embodiments with hpRNλ silencing RNA, the skilled person will immediately realize that similar effeul can be achieved using other types of silencing RNA artificially introduced into a host cell/plant cell, whereby the processing in si RNA molecules involves a double stranded RNA phase, including conventional sense RNA, Sπtiscnsc RNλ, unpclyadenylated RNA, and RNA wherein the silencing RNA includes largely double stranded regions comprising a nuclear localization signal

from a viroid of the Potato spindle tuber viroid-lype or comprising CUG trinucleotide repeats as described e.g. in WO 03/076619 WO04/073390 Wθ ^£ .>9/53050 or WOOl/ 12824.

5 An enzymatic assay which can be used for detecting RNAse III enzymatic activity is described e.g; in Lamontagne et al,, MoI Cell Biol. 2000 February; 20(4): 1104-1 1 15. The resulting cleavage products can be further analyzed according to standard methods in the art for the generation of 21-24 nt siRNAs.

10 It is also an object of the invention to provide host cells, plant cells and plants containing the chimeric genes or mutant alleles according to the invention. Gametes, seeds, embryos, either zygotic or somatic, progeny or hybrids of plants comprising the chimeric genes or mutant alleles of the present invention, which are produced by traditional breeding methods are also included within the scope of the present invention. Also encompassed

15 by the invention are plant parts from the herein described plants, such as leaves, stems, roots, fruits, stamen, carpels, seeds, grains, flowers, petals, sepals, flower primordial, cultured tissues and the like.

The methods and means described herein are believed to be suitable for all plant cells and 20 plants, gymπosperms and angiospcπns, both dicotyledonous and monocotyledonous plant cells and plants including but not limited to Ambidυpsis, alfalfa, barley, bean, corn or maize, cotton, flax, oat. pea, rape, rice, rye, safflower, sorghum, soybean, sunflower, tobacco and other Nhυtiana species, including blicotiana benthamiana, wheat, asparagus, beet, broccoli, cabbage, carrot, cauliflower, celery, cucumber, eggplant, lettuce, onion, 25 oilseed rape such as canola or other Brassicas, pepper, potato, pumpkin, radish, spinach, squash, tomato, zucchini, almond, apple, apricot, banana, blackberry, blueberry, cacao, cherry, coconut, cranberry, date, grape, grapefrait, guava, Idwi, lemon, lime, mango, melon, nectarine, orange, papaya, passion fruit, peach, peanut, pear, pineapple, pistachio, plum, raspberry, strawberry, tangerine, walnut and watermelon, Brassica vegetables, ^ i rn~1πrϊ i π α tinH eπ πnrhppf Virtr cλmύ

embodiments of the invention, lhe plant cell could be a plant cell diffcrctit from an Arabidυpsis cell, or lhe plant could be different from Arabidopxis,

The methods according to the invention, particularly the increase of the functional level of DCL3 or DCL4 protein may also be applicable to other cukaryotic cells, e.g. by introduction of a chimeric gene expressing DCL4 into such eukaryotic cells. The eukaryotic cell or organism may also be a fungus, yeast or mold or an animal cell or organism such as a non-human mammal, fish, cattle, goat, pig, sheep, rodent, hamster, mouse, rat, guinea pig, rabbit, primate, nematode, shellfish, prawn, crab, lobster, insect, fruit fly, Coleopteran insect, Dϊpteran insect, ϊ-Cpidoptcrart insect or Homeopteran insect cell or organism, or a human cell. Eukaryotic cells according to Lhe invention may be isolated cells; cells in tissue culture; in vivo, ex vivo or in vitro cells; or cells in non- human eukaryotic organisms, λlso encompassed are non-human cukaryotic organisms which consist essentially of the eukaryotio cells according to the invention.

Introduction of chimeric genes (or RNA molecules) into the host cell can be accomplished by a variety of methods including calcium phosphate traπsfection, DEAE- dcxlran mediated transfection, electropomtion, microprojeutile bombardment, microinjection into nuclei and the tike.

Methods for the introduction of chimeric genes into plants are well known in the art and include .λg/'oέαcrm ^' Mλrt-medϊated transformation, particle gun delivery, microinjection, clcctroporalion of intact cells, polyethyleneglycol-mcdiated protoplast transformation, electroporation of protoplasts, liposome-mediated transformation, silicon-whiskers mediated transformation etc. The transformed cells obtained in this way may then be regenerated into mature fertile plants, and may be propagated to provide progeny, seeds, leaves, roots, stems, flowers or other plant parts comprising lhe chimeric genes.

A "transgenic plant", "transgenic cell" or variations thereof refers to a plant or cell that contains a chimeric gene {"tsnEgene") not found in a wild type plant or ccϊl of the same species. A "transgene" as referred to herein has the normal meaning in the- art of

biotechnology and includes a genetic sequence which has been produced or altered by recombinant DNA or RNA technology and. which has been introduced into the cell. The transgαne may include genetic sequences derived from the same species of cell. Typically, the transgene has been introduced into the plant by human manipulation such as, for example, by transformation but any method can be used a$ one of skill in the art recognizes.

Transgenic animals can be produced by the injection of the chimeric genes into the pronucleus of a fertilized oocyte, by transplantation of cells, preferably undifferentiated cells into a developing embryo to produce a chimeric embryo, transplantation of a nucleus from a recombinant cell into an enucleated embryo or activated oocyte and the like. Methods for the production of transgenic animals are welt established in the art and include US patent. 4, 873, 191 ; Rudolph el al. 1999 (Trends Biotechnology 17 :367- 374) ; Dalrymple el al. (1998) Biotechnol. Genet. Eng. Rev. 15 : 33-49 ; Colmaii (1998) Biocl). Soc. Symp. 63 : 141-147 ; Wilmut et al. (1997) Nature 385 : 810-813, Wilαiute et al. (1998) Reprod. Feitil, Dev. 10 : 639-643 ; Petty et a!. (1993) Transgenic Res. 2 : 125- 133 ; Hogan et al. Manipulating the Mouse Embryo, 2"' ¹ ed. CIoId Spring Harbor Laboratory press, 1994 and references cited therein,

Gametes, seeds, embryos, progeny, hybrids of plants or animals comprising the chimeric genes of the present invention, which are produced by traditional breeding methods are also included within the scope of the present invention.

As used herein, "the nucleotide sequence of gene of interest" usually refers to the nucleolide sequence of the DNA strand corresponding in sequence Io the nucleotide sequence of the RNA transcribed from such a gene of interest unless specified otherwise.

Mutants in Dicers or Dicer-like proteins, such as DCL3- or DCL4-encoding genes are usually recessive, accordingly it may advantageous to have such mutant genes in homozygous forra for ths puspcso of reducing the func-uOual level υf such Dicer proteins. However, it may also be advantageous to have the mutant, genes in heterozygous form.

Whenever reference is made to a "functional level which is modulated, or increased or decreased with regard ro the wild type level" typically, the wild type level refers to the functional or actual level of the corresponding protein in a corresponding organism which is isogenic to the organism in which lhe modulated functional level is assessed, but for the genetic variation, the latter including presence of a transgene or presence of a mutant allele. Preferably, the "wild type" level in terms of functional level or activity of an enzyme or of a protein refers to the average of the activity of the protein or enzyme in a collection of individuals of a species which arc generally recognized in the art as being wild type organisms. Preferably, the collection of individuals consists of at least 6 individuals, but may of course include more individuals such as at least 10, 20, 50, 100 or even 1000 individuals. With regard to an amino acid sequence of a polypeptide or protein, the "wild type" amino acid sequence is preferably considered as the most common sequence of that protein or polypeptide it) a collection of individuals of a species which are generally recognized in the art as being wild type organisms. Again preferably the collection of individuals consists of at least 6 individuals. A modulated functional level differs from the wild type functional level preferably by at leasl 5% or 10% or 15% or 20% or 25% or 30% or 40% or 50% or 60% or 70% or 80% or 90% or 95% or 99%. Tlic modulated functional level may even be a level of protein or enzyme activity which is non-existent or non-detcctabtc for practical purposes. A mutant protein cat) be considered as a protein which differs in at least one amino acid (e.g. insertion, deletion or substitution) from the wild type sequence as herein defined and which is preferably also altered in activity or function,

It will be clear that the methods as herein described when applied to animal or humans may encompass both therapeutic and non-therapeutic methods and that the chimeric nucleic acids as herein described may be used as medicaments for the purpose of the above mentioned therapeutic methods.

The following non-1 i mixing Examples describe methods and means for Jϊiodulatitsg dsRNA mediated silencing of the expression of a target gene in a plant cell by modulating

the functional level of proteins involved in processing in siRNA of artificially introduced dsRNA πioJecules such as DCL3 aπd _. DCL4.

Unless stated otherwise in The Examples, all recombinant DNA techniques are carried out according to standard protocols as described in Sambrook et al. (1989) Molecular Cloning: A Laboratory Manual, Second Edition, Cold Spring Harbor Laboratory Press, NY and in Volumes 1 and 2 of Ausubel et al. (J 994) Current Protocols in Molecular Biology, Current Protocols, USA. Standard materials and methods for plant molecular work arc described in Plant Molecular Biology Lahfax (1993) by R.D.D. Croy, jointly published by BIOS Scientific Publications Ltd (UK) and Blackwell Scientific Publications, IJK, Other references for standard molecular biology techniques include Sambrook and Russell (2001) Molecular Cloning: A Laboratory Manual. Third Edition, Cold Spring Harbor Laboratory Press, NY, Volumes I and II of Brown (1998) Molecular Biology Lab fax, Second Edition, Academic Press (UK)- Standard materials and methods for polymerase chain reactions can be found in Dieffcnbacli and Dveksier (1995) PCR Primer: A TMborutory Manual, Cold Spring Harbor Laboratory Press, and in McPhcrsoπ at al. (2000) PCR - Basics: From Background to Bench, First Edition, Springer Verlag, Germany.

Throughout the description and Examples, reference is made to the following sequences:

SEQ ID No.: 1: amino acid sequence of At_DCL4 (λrabidopais tlialianά).

SEQ ID No.: 2: nucleotide sequence encoding AtJDCIA

SEQ ID NO.: 3: amino acid sequence of PL.DCL4 (Pυpulus trichacarpa). SEQ TD No.: 4: nucleotide sequence encoding Pt_DCL4.

SEQ ID No.: 5: amino acid sequence of Os._DCL4 (Oryza .mtivti).

SEQ ID No.: 6: nucleotide sequence encoding Os_DCL4.

SEQ ID No.: 7: amino acid sequence of At_DCL3 (Arabidopsls thaliana).

SEQ ID No.: 8: nucleotide sequence encoding Al_DCL3. SEQ ID No.: 9: amino acid sequence of PtJDCuϊ (Popuϊus trichocarpa).

SEQ FD NO.: 10: nucleotide sequence encoding PtJDCL3.

SEQ ID No.: 11: amnio acid sequence of Os_DCL3a (Oryza saliva). SEQ ID No.: 12; nucleotide sequence encoding Os_DGL3a. SEQ ID No.: 13: amino acid sequence of Os_DCL3b {Oryza saliva). SEQ ID No.: 14: nucleotide sequence encoding Os_DCL3b. SBQ ID No.: 15: oligonucleotide primer for the amplification of fragment 1 of the coding sequence of DCL3. SEQ ID No.: 16: oligonucleotide primer for the amplification of fragment 1 of the coding sequence of DCL3-

SEQ ID No.: 17: oligonucleotide primer for the amplification of fragment 2 of the coding sequence of DCL3.

SEQ ID No.: 18: oligonucleotide primer for the amplification of fragment 2 of the coding sequence of DCL3. SEQ ID No.: 19: oligonucleotide primer for the amplification of fragment 3 of the coding sequence of DCL3. SEQ ID No,: 20: oligonucleotide primer for the amplification of fragment 3 of the coding sequence of DCL3. SEQ ID No.: 21.: oligonucleotide primer for the amplification of fragment 4 of the coding sequence of DCL3.

SEQ ID No.: 22: oligonucleotide primer for the amplification of fragment 4 of the coding seq ueπce of D<X3.

SHQ ID No.: 23: oligonucleotide primer for the amplification of fragment 1 of the coding sequence of DCL4. S1ϊQ ID No.: 24: oligonucleotide primer for the amplification of fragment 1 of the coding sequence of DCL4. SEQ ID No-: 25: oligonucleotide primer for lhe amplification of fragment 2 of the coding sequence of DCL4. SEQ ID No.: 26: oligonucleotide primer for the amplification of fragment 2 υf ths coding sequence of DCL4.

SEQ ID No.: 27: oligonucleotide primer for the amplification of fragment 3 of the coding sequence of DCiA.

SEQ ID No.: 28: oligonucleotide primer for the amplification of fragment 3 of the coding sequence of DCI-4. SEQ ID No.: 29: oligonucleotide primer for the amplification of fragment 4 of the coding sequence of DCL4, SHQ ID No.: 30: oligonucleotide primer for lhe amplification of fragment 4 of the coding sequence of DCL4. SEQ ID No.: 31 : forward oligonucleotide primer for diagnostic PCR amplification of

DCL3.

SEQ ID No.: 32; reverse oligonucleotide primer for diagnostic PCR amplification of DCL3.

SEQ ID No.: 33: forward oligonucleotide primer for diagnostic PCR amplification of

DCL4. SEQ ID No.; 34: reverse oligonucleotide primer for diagnostic I ¹ CR amplification of

DCL4. SEQ ID No.: 35: forward oligonucleotide primer for diagnostic PCR amplification of

DCL3λ. SBQ ID No,: 36: reverse oligonucleotide primer for diagnostic PCR amplification of

DCL3A.

SEQ ID No.: 37-" forward oligonucleotide primer for diagnostic PCR amplification of DCL3B.

SEQ ID No.: 38: reverse oligonucleotide primer for diagnostic PCR amplification of

DCL3B.

REFERENCES

An ct al., 1996 The Plant Cell 8, 15-30

Blanc, G. & Wolfe, K.H. (2004) Plant Cell 16, 1679-1691.

Colman (1998) Bioch. Soc. Symp. 63 : 141-147 Dalrymple et al. (1998) Biςrtechnol, Gcαet. Eπg. Rev. 15 : 33-49

Fire ct al., 1998 Nature 39L 806-81 1

Gasciolli et al., 2005 Current Biology, 15, 1494-1500).

Hamilton ct al. 1998 Plant J. 15: 737-746

Hapster ct al., 1988 MoJ. Gen. Genet. 212, 182-190 Hausmann, 1976 Current Topics in Microbiology and Immunology, 75: 77-109

Hedges, S.B, Blair, J.E., Venturi, M.L. & Shoe, J.L. BMC EDOL Biol (2004) 4:2 1471- 2148/4/2

ϊTcπtkoff et al. Plant Physiol. 2000 Jun;.l23(2):439-42 .

Hogan ct. al. Manipulating the Mouse Embryo, 2"' ¹ ed. Cold Spring Harbor Laboratory press, 1994 and references cited therein.

Hudspeth et al., 1989 Plant MoI Biol 12: 579-589

Kβii et al., 1989 EMBO J. 8: 1323-1330

Keller et a!., 19KX EMBO J. 7: 3625-3633

Keller et al., 1989 Genes Devcl. 3: 1639-1646 Kurihara and Watanabe, 2004, Prac. Natl. Acad. Sci. USA 101 : 12753- 12758),

Lamontagnc ct al. MoI Cell Biol. 2000 February; 20(4): 1104-1115

Lee et al., 2004 Cell 75:843-854

Needleman and Wunsch 1970

Pclcman el al., 1989 Gene 84: 359-369 Perry ct. al. (1993) Transgenic Res. 2 : 125- 133

Phani et al., 2004 Cell 117 : 83-94.

Qi et at., 2005 Molecular Cell, 19, 421-428

Rudolph et ai. 1999 (Trends Biotechnology 17 :367-374)

Smith et al., 2000 Nature 407: 319-320

Stcrck, L., Rombaute, S.» Jansson, S., Stεrky, F., Rouzc, P. & Van de Peer, Y. (2005)

New Phytol. 167, l65-l70Watcrhouse et al. 1998 Proc. Natl. Acad. Sci. USA 95: 13959-

13964.

Wilmut et al. (1997) Nature 385 : 810-813

Wilmute et al. (1998) Rcprod. Ferlil. Dev. 10 : 639-643

Xic et al.. 2004, 1'LosBiology, 2004, 2, 642-652).

Yυshikawa et al,, 2005, Genes Sc Development, 19: 2164-2175).

EXAMPLES

Example 1. Identification of different dicer types in plants

1.1 Introduction

5 Eukaryotes possess a mechanism thai generates small RNAs and uses them lυ regulate gene expression at the transcriptional or post-transcπpliomu ¹ level (1), These 2l-24nt small RNAs are defined as micro (mi) RNAs ₁ which are produced from partially self- complernent-iry precursor RNAs. or small interfering (si) RNλs, which arc generated from double stranded (ds) RNAs (U 2). The large RNase IH-like enzymes that cleave

10 these templates into small RNλs are called Dicer or Dicer-like (IXTL) proteins (3). Humans, mice and nematodes each possess only one Dicer gene, yet regulate their development through miRNAs, modify lheir chromatin Slate through siRNλs, and are competent to enact siRNA-mediatcd RNA interference (RNAi) (1, 4). Insects, such as Drosophiid mdunogastei\ and fungi, such as Neura.ψora crassa and Magnaporthe

\5 oryzae, each possess Iwυ Dicer genes (4, 5). In Drosophαu, the two Dicers have related but different roles: one processes miRNAs and the other is necessary for RNAi (6). In plants, the picture is not clear. Il has been reported that rice {Oryza sativa) has two DCL genes, although this was before the complete rice genome had been sequenced, while Arάhidopsis ihaliana has four (4). Analysis of insertion mutants of the four A. thaliana 0 DCL (AtDCL) genes has revealed that the role of a small RNA appears to he governed by the type of DCI, enzyme that generated it; AtOChI generates miRNAs, AtOCUl generates SiRNAs associated with virus defense, AtDCLi generates siRNAs that guide chromatin modification, and AiDClA generates trans-acting siRNAs that regulate vegetative phase change (7-10). Fn this study, we sought to identify whether most plants 5 were like rice, fungi and insects in having two Dicers, or were like Arabidøpsis with multiple divergent Dicers. We found evidence suggesting that it is advantageous for plants to have a set of four Dicer types, and that these have evolved by gene duplication after the divergence of animals from plants. The number of Dicer-like genes has continued to increase in plants over evolutionary time, whereas in mammals, the number 0 has decreased. These opposite trends are probably a reflection of the differing threats and defence strategies that apply to plant ¹ ? and mammals. Mammals have immune, interferon

and ADAR systems to protect them against invaders, and may only need a Dicer to process miRNAs. Plants have none of these defence systems and, therefore, rely on Dicers to not only regulate (heir development Through miRNAs, but also to defend them against a multitude of viruses and transposons,

1.2 Materials and Methods

1.2.1 Plant Material, PCR Amplification and Sequencing

RNA was extracted from leaf material of the Columbia ecotypc of Arabidopsis thaliana using the TRLsol reagent (Invitrogen), reverse transcribed, amplified and cloned into pGEM-T Easy using llle OneStep RT-PCR Kit (Quiagen) and pGEM-T Easy vector system L kit (Promcga). The inserts were sequenccd using BigDye terminator cycle sequencing ready reaction kits (PE Applied Biosystcms, CA, USA). Amplification

' reaction conditions for detection of orthologous genes were 35 cycles at 95 ⁰ C for 30 sec,

52"C for 30 sec and 72°C for I minute. DNA samples of rice, maize, cotton, lupin, barley and Trificum tauchii were kind gifts from Narayaπa Upadhyaya, Qϊng Liu and Evans Lagudah. PCR products were separated on a 1.3% agarose geJ.

1.2.2 Data Collection

The sequences of Arabidopsis, rice, maize, poplar, Chlamydomonas reinhardtϋ and Tetrahymeita genes were accessed via the Arabidopsis Information Resource (TλIR) database Qi»Bi//w.\yw-4m/>Mfop.v»A\org/indexjsp). the Institute for Genomic Research (TTGR) rice and maize databases (http://www.tiμr.org/tigr- scrip.ts/osa I _web/gbro wee/rice; http://tifirblast.tigr.orft/tp]' maize/index.cgi), and the JGl Eiikaryotic Genomics databases flntp://gcnoms.j gi- psf.org/Pophα/Foptrl Jrøme,html),http;//genome.igi-p,sf.θfg/chire2/chire2.ho]r� �e.htm]. and the Tctrahymena genome database hup://seq x-iliate.org/cfii-bin/blast-t.gtl .pi.

1.2.3 Sequence Alignment and Phylogenetlc Analysis.

Coding secμiences of predicted genes were determined by using tBlastn and manual comparison of clustalW-alϊgncd genomic sequences, cDNλ secμiences and predicted coding sequences ((2DS). All protein sequence alignments were made using the program Clusial-W (11). Phylogenctic and molecular evolutionary analyses were conducted using MEGA Version 3.1 (12). Trees were generated using the following parameters: complete deletion. Poisson correction, neighbor-joining, Dayhof matrix model for amino acid substitution, and bootstrap with 1000 replications. Protein domains were analysed by scanning protein sequences against the InterPro protein signature database (hUp://www.ebi.ac.uk/InterProSoan) with the IntCrProScan program (13). Unless otherwise stated, domains were defined according to pFAM predictions (http://www.sanger.ac.uk/Softwarc/Pfam/)

1.3 Results and Discussion

1.3.1 Identification of Dicer-like Genes in Arabidopsis, Poplar and Rice

The amino acid sequence of AtDCLl (AtIgOHWO) has been determined previously by sequencing of dDNAs generated from the gene's mRNA (14). However, the sequences of λtDCL2 (At3gO33OO), λtDCL3 (At3g43920) and λrϋCL4 (At5g20320) have previously been inferred from the chromosomal DMA sequences determined by the λrάhidopςis Genome Project (TAIR) using mRNA splicing prediction programs. To obtain more accurate sequences of these proteins, eDNAs were generated from the appropriate Arαhldαpm mRNAs, cloned into plasmids and their nucleotide sequences determined. Analysis of these sequences (Genbank accession numbers NM_111200, NM_114260, and NM_122039) showed that the inferred amino add sequences of λ/DCL2, 3 and 4 were largely but not completely correct: at least one exon/intron region has been miscalled for each gene and two different spHcetbrms of AlDCL2 mRNA were identified. Interrogation of the λrαhidopsi.y genome with the tBLASTn algorithm, using amino acid sequences of each of (he. DCl , κ«]i.ico.cc-$, identified no further Dicer-like gcties. Repeating essentially the same procedure on the recently completed sequences of flic

whole yenoπics of poplar (Poputus tricliocarpά) and rice {Oryzu stttivά) revealed five υCL-likv genes in poplar (FtO2g 14226280, PtOδgl 1470720, Pt08g4686890, Ptl0gl6358340, Ptl8g3481550; using the nomenclature in which the number preceding the "g" indicates the chromosome and the number after the "g" indicates the nucleotide position of the start of the coding region on the JdI poplar chromosome pseudαmoleculcs) and six genes in rice (Os01g68120, Os04g43050, Os03g02970, Os03g38740, Os09gl46I0, Osl()g34430; TlGR build 3 nomenclature). The location of these genes on the genome maps of poplar and rice is shown in Figure 1.

Phylogcnetic analysis, using the PλM-Dayhof matrix model, JTT matrix model, minimum evolution methods and neighbour-joining methods in MEGA 3.1, all showed that the inferred amino acid sequence of each of the rice and poplar DCL proteins strongly aligned with the sequence of an individual member of the four Arubklopsis DCL proteins (Fig. 2A, and pairwise distances in Table 2). With the diversity represented by these plants, from small alpine plant to large tree, and from monocot to dicot, this result suggests that these four types of Dicer are present in all angiosperms and quite possibly all multi-cellular plants. This was further supported by detection of all four genes in barley, maize, cotton and lupin by PCR assays, using primers designed to conserved type- specific sequences (data not shown). We interpreted these groupings to be indicators of orthologous genes, showing that, in poplar, there are single orthologs of AtDCLl, AtDCLS and AtDCLA and a pair of orthologs of AtDCLl, and that in rice, there are single orthologs of AiDCLl and AtDCLA and pairs of orthologs of AtDCLT. and AtDClZ. Each gene was named to reflect the species in which it is present, using the prefix Pt or Os, and the number of its Arabidυpsii oitholog e.g. PtDCLl . Members of a pair of orthologs were designated A or B with the gene termed A having greater sequence identity to the Arahidopsia ovthnlog- For all DCL types, the poplar and Arabidυpsia orthologs are more similar to each other than to the rice ortholog, as might be expected given that the first two arc dicots and rice is a monocot. The Ambidnpsis, poplar and rice DCLl genes group most tightly together, and the second tightest cluster is formed by the DClA genes. The DCLI and DCLHs genes form more expansive clusters showing that they have a higher ύcgree of divergence, and the gene that is the racsi divergent from the others within the group is OsDCLSB,

1.3,2 Correlation of Dicer Type with Domain Variation

Six domain types are present in animal, fungal and plant DCR or DCL proteins, collectively, although many individual proteins lack one or more of them (Tablel). These six types are the DEXH-helicase, helicase-C, Duf283, PAZ, RNaselll and double stranded RNA-binding (dsRB) domains (4, 15, 16 and references therein). The DEXH and — C domains are found towards the N-teππinal and C-terminal regions of the helicase region, respectively. There are always two RNAseIII domains (termed a and b) in a. Dicer protein, and the Duf283 is a domain of unknown function but which, is strongly conserved among Dicers. The role of the dsRB domain in human Dicer is generally thought to mediate unspeciftc reactions with dsRNA, with the PAZ, RNaseIIla and RNascTIIb domains being crucial for the recognition and spatial cleavage of dsRNAs into si or miRNA (16). In organisms with only one Dicer, this enzyme, with its associated proteins, is presumably the only generator of si and mi RNAs. In organisms with two or more Dicers, there is probably a division of labour. ' Each of the inferred amino acid sequences of the Arabidopsis, poplar and iice

DCL proteins, along with examples of ciliate, algal, fungal, mammalian and insect DCRs (from previously published information or identified by tBLASTn interrogation of available databases) were analysed using the lnterpro suite of algoritbms. All six domain types were identified and located (Figure 2) in all of the plant DCL sequences, except for AφCL3 and CλsDCL2B, which were partially lacking the Du£283 domain. The two most striking results from this analysis were that all of the DCLl, 3 and 4 types in plants have a second dsRB (dsRBn) domain which is completely lacking in non-plant DCRs, and that the PAZ domain is absent in the ciliate, fungal and algal DCRs but detectable in all of the plant DCLs, including all three DCL4s, despite previous reports that this domain is missing in AtT)CfA (4, 15), Tt has been suggested that the absence of a PAZ domain may play an important role in discriminating which accessory profcins a DCR Or DCL interacts with, thereby guiding the recognition of its template (18). The correlation between the absence of mϊRNAs and presence of only a PAZ-frec Dicer in Shizosaccharomycyex pornbe, has also led to the suggestion that the PAZ domain may play an impurUuil role in measuring the length of nuRNAs. However, the presence of the

PAZ domain in all plant Dicer types seems to rule out its presence or absence dictating the function of a DCL in plants. The DUF283 domain is absent in some ciliate and fungal DCRs and in λrDCL3. However, it is present in all the other-plant Dicers, including the DCL3-lypes in rice and poplar. This, similarly, suggests that its presence or absence docs not characterize a Dicer-type or Us function in plants.

In Arabidopsis, and probably all plants, the four different Dicer lypes produce small RNAs that play different roles. Each different type requires specificity in recognising its substrate RNA and the ability to pass the small (s) RNA thai il generates to the correct effector complex. Unlike all of the other domains, the dsRBb domain, by its presence, absence or type, is a good candidate for regulating substrate specificity •and/or the interaction with associated proteins to direct processed sRNAs to the appropriate effector complex. DCL2 proteins are different from the other Dicer-types by their lack of a dsRBb domain and, with the exception of the variation between the dsRBa domains of DCLl and 3, the net variation between the pair-wise combinations of Dicer- types 1, 3 and 4 is mosL variable in this domain (Figure 2 and Table 1). There is good evidence that dsRB domains not only bind to dsRNA but also function as protein-protein interaction domains (21, 22, 23). Indeed, it has been shown that fusion proteins containing both the dsRBa and dsRBb domains of AtωCLl, A/DCL3 and λrDCL.4 can bind to members of the ITYL1/DRB family of proteins that are probably associated with sRNA pathways in Arabidopsis (23). The simplest model seems be that the dsRRa domain along with the PλZ and RNaseϊϊT a and b domains recognize and process the substrate RNA, while the dsRBb domain specifically interacts with one or two of the different HYL1/DRB members to direct the newly generated sRNλs to their appropriate RNA-clcaving or DNA-methylating/histonu-modifying effector complexes (24).

1.3.3 DCL Paralqgs in Poplar and Rice and Other Gramineae

In both poplar and rice, the DCL2 gene has been duplicated. The paralogs in poplar,

PtDCL2λ and PtDCL2B, have 85% sequence similarity at the amino acid level and are located on chromosomes 8 and 10, respectively. They arc within large duplicated bl<κ:ks (Fig. I) that are predicted to have formed during a large scale gene duplication event 8 to

13 million years ago (mya) (19, 25). The timing for this duplication OϊDCL2 in poplar is consistent with the lack of a DCL2B in ArabldopsLs, since the common ancestor of λrahidop.iis and poplar is estimated to have existed about 90mya (20).

The paralogs, OsDCL2A and OsDCL2B, in rice have almost identical sequences (99% sequence similarity at the amino acid level), except for a ~200bp deletion, largely within an intron, but also deleting part of the Duf 283 domain in OsDCL2B, which may possibly abolish or impair the protein's function. Apart from lhis deletion, there are less than lOOnt variations in a genomic sequence of 14.5 kb. This suggests that the gene duplication occurred relatively recently. Applying the unsophisticated approach of using the rate of amino acid changes that occurred between PtDCL2A and PsDCL2B during the ~ 10 million years (my) since their duplication as a measure .of time (~ 20 aa changes/my), the ~15 amino acid difference between OsDCL2A and OsDCLlB suggest that this duplication occurred about 1 mya. It has been estimated that ttic rice subspecies indica and japonica fast shared a common ancestor -0.44jivya (26). To test whether the duplication event occurred before or after this divergence, DNA extracted from japonica and indica was assayed by PCR using primers, flanking the OsDCTJZB deletion. The assay (Fig. 3) showed that bolh 0sDCL2A and OsDCLZB are present in both subspecies, hence placing the duplication event that created them before this time. Examination of the regions surrounding these genes on rice chromosomes 3 and 9 suggest that the duplication was of a relatively small region of chromatin (50-lOOkb).

The DCL3 paraJogs, OsDCUA and OxDCTJB, in rice arc highly divergent, showing about 57% similarity at the amino acid level. Therefore, the duplication event which created them probably occurred before the generation of PϊDCL2A and PtDCL2B in poplar (~lθmya). However, there is no pair of DCL3 paralogs in either poplar or λrabidopsis, suggesting that the event that produced the OsDCLS paralog pair occurred after the divergence of monocotyledonous plants from dicotyledonous plants (about 200mya). In an attempt to refine the estimation of the date wl)cn the OsDCL3 paralogs were generated, we sought to determine if they existed before the divergence of maize and rice (~ 50 mya). Therefore, the TIGR Release 4.0 of assembled Zea mays (AZM) and singleton sequences was searched for both OsDCUA-lϊke and OsDCL3B-likv sequences.

Three sequences were identified, two of which (AZM4JJ7726 and PUDDE51TD) have greater similarity to OaDCFJB and one (AZM4_120ό75) which has greater similarity to OsDCUA. Fortunately, one of the OsDCL3B-\ike clones (AZM4_67726) covered the same helicase-C domain region as the OsDCL3A-liIce clone. Phylogenetic analysis (Fig. 4A) showed that these clones grouped as orthologs of OsDCUA and OsDCLξB _^ strongly suggesting that the duplication event that generated the DCU paralogs occurred before the divergence of maize from rice. Examination of the aligned hclicase-C sequences of all of the Arabidopsis, poplar, and rice DCL gene sequences and the two maize clones allowed two SeLs of primers to he designed that, when used in PC-R assays with maize or rice DNA, should discriminate between the DCTJA and DCUB paralogs in either species and may also he similarly effective in other cereals. Fortunately, the polymorphisms that allowed the design of these discriminating primers are in sequences that flank an intron that is -.mailer in the OsDCL3A gene than in the OsDCTJiB gene (but not in the equivalent genes in maize), thus providing a visible control for the specificity of the amplification products. Using these primer pairs on DNA from rice, maize, and two olher diploid cereals, barley (Hvrdeum vulgare) and Trhlcwn tauchii, a progenitor of wheat, (Fig. 4B), showed that orthologs of both OsDCTJA and 0&DCL3B could be detected ill all of these species. The PCR products from barley and T. tauchii were cloned and sequenccd, which were then compared wtth the DCL3 HeI-C sequences represented in Fig. 4A. The sequences amplified from barley and T. tauchii with the 3A-spccific primers clustered with the OsDCIJA and AZm467726 sequences, and the sequences amplified with the

^" 3B-speci+lc primers clustered with OsDCUB and AZm467726 (data not shown). This demonstrates that the DCT3 duplication occurred not only before the common ancestor of maize and rice, but also before the common ancestor of barley and rice (~ 60mya).

1 ,3.4 A Fifth Dicer Type in Monocots

The 0sDCL3B gene in rice is transcribed, as we could detect its sequence in EST clones (RSTCEK_13981 and CK062710), and has no premature stop codons, suggesting that it is translated into a functional protein. However, this protein has 57% amino acid sequence identity with that of 0sDCL3A, showing that the gene has diverged significantly from its paralog, although it has retained lhe landmark amino acids that give it the domain

hallmarks of a functional Dicer. Furthermore, its dsRB domain, which probably governs the role of the small RNAs that the enzytne generates, is highly divergent from all of the other Dicers, showing no pbylogcnctic grouping with any of them (Fig. 3B). λs the DCUB gene is present in all of the monocots thai we tested, and probably has a specificity different from that of its paralog OsDCLJA, which groups well with PtDCU and AtDCLS, we suggest that it lias probably evolved to perform a different function. The highly divergent dsRBb may allow H to interact with proteins other than those interacting with the other four Dicer types. Alternatively, this peptide region may be non-functional and thereby give the protein a characteristic similar to the DCL2s. If so, it is possible that it is a case of convergent evolution that increases the plant's ability to combat viruses. Whatever its function, OsDCLSB and its counterparts in other monocots have been retained for over 60my suggesting that they confer advantage. We suggest that since the gene is highly likely to have a different function to other DCL3 types, it and its counterparts should be considered a different form of Dicer, DCL5.

1.3.5 The Origin of Plant Dicers

Examination of the genome of the green algae, Chlamydamonas reinhardtU, which diverged from plants ~955mya (27), revealed a single DCR-like gene (C_130110 ehk©2/scaffold_l 3:93930-105880) encoding a protein with single helicase-C, a DUF283 and dsRB domains, and iwυ RNλselH domains. This initially suggested that the four DCL types in plants have evolved from a single common gene that was present in the common ancestor of algae and plants. However, examining the genome of the ciliate, Tetrahymena thβrmophlla, which shared a last common ancestor with plants ~2 biJiion years ago (27), revealed that there are two DCR-like genes (AB 182479 and AB I 82480 and (ref 28» which both possess hclicasc domains and Iwυ RNase III domains (Figure 2). Searching the available gcnoracs of Aϊchacbacieria and Eubacterkt, we were unable to identify any protein containing two adjacent RNAseTTI domains. In an attempt to discover whether one (and which one) or both of the Tetrahymena genes were the progenitors of animal and plant Dicers, the two RNAselll domains of both these genes were compared With the RNaseIIIa and b domains of DCRs or DCLs of a nematode, an insect and three plant species. The result (Fig. 5) shows that with the exception of the Tetrahymena

domains, all RNasellla domains cluster together and all RNAseϊJIb domains cluster together. However, the Tetrahymena RNaseIII a and b domains from DCRl and DCR2 are more similar to themselves than Lo either of the RNAseIIIa or RNAseffib domain groupings of plants, nematodes and insects. This is an interesting dichotomy of conservation. Insects, nematodes and plants shared a common ancestor about ~ 1.6 billion years ago and the phylogenctic tree in Figure 6 suggests that duplication and distinction into RNAseIIIa and b domains had been well established at this point., and that these differences have been largely conserved since then. Unfortunately, because the Tetrahymena KNAseIIla and b domains, form an out-group from the domains of the other species, it does not shed light on which one (or both) of the Tetrahymena DCR-I ike genes is the modern day representative of the progenitor of plant and animal Dicers. However, the simplest model is that the Tetrahymena DCR-like genes were derived from a very

' ancient duplication, that this pair has been maintained in some fungi and insects, and that in plants the pair has undergone a further duplication. In nematodes, mammals, and other organisms which possess only one Dicer, it appears that they have lost one of the original progenitor genes. Figure 7 presents a summary of the different Dicer-like genes described in this study, in die context of the evolutionary history of plants, algae, fungi and animals, and predicted events of large scale gene duplication that have occurred in plants. It seems likely that the gene duplication from two to four plant DCL genes that occurred between 955 and 200mya, tho generation of OsDCLiB between 200 and 60inya, and the generation of λDCL2B, occurred during the large scale gene duplication events that have been mapped to - 270, -70 and -lOmya, respectively (20).

1.4 RKPERENGES 1. Finnegan, E.J. & Matzke, M.λ. (2003) J. Cell ScL 116, 4689-4693.

2. Bartβl, D. (2004) Cell 1 16, 281-297.

3. Bernstein, K, Caudy, A-, Hammond, S. & Hannon, G. (2001) Nature 409, 363- 366.

4. Sehauer. S., Jacobseπ, S _M Meinke, D. & Ray, A. (2002) Trends Plant Sci, 7, 487- 491.

5. Catalanotto, C, Pallotta, M., ReFaIo, I ¹ ., Sachs, M.S., Vayssie, L., Macino, C. & Cogoπi, C. (2004) MoL Cell. Biol. 24, 2536-2545.

6. Lee, Y.S., Nakahara, K., Pham, J.W., Kirn, K., He, Z., South rner, EJ. & Carthew, R-W- (2004) Cell 117, 69-81. • 7. Park, W., Li, J,, Song, R., Messing, J. & Chen, X. (2002) Curr. Biol. 12, 1484- 1495.

8. Xiβ, Z., Johansen, L.K., Guatafson, A.M., Kasschau, K.D., Lellis, A.D., Zilbennan, D., Jacobsen, S-E. & Cairiπglon, J.C. (2004) PLoS Biol. 2, E 104.

9. Gasciolli, V., Mailory, A.C., Bartel, D.P. & Vaucheret, H. (2005) Curr. Biol. 15, 1494-1500,

10. Xie, Z., Allen, E., Wilken, A. & Cam ngton, J.C. (2005) Proc. Natl. Acad. ScL USA 102, 12984-12989.

11. Thompson, J. D., Higgins, D.G. & Gibson, T.J. (1994) Mud. Acids Res. 22, 4673- 4680. 12. Kumar, S., Tamura, K. and Nci, M. (2004) Bioinformatics 5, 150463.

13. Zdobnov, E.M. & Apweiler, R. (200.1) Bioinformatics 17, 847-848.

14. Golden, T.A., Schauer, S.E., Lang, J.D., Pien, S., Mushegian, A.R., Grossxiiklaas, U., Meϊnkc, IXW. & Ray, A. (2002) Plant. Physiol. 130, 808-822.

15. Finnegah, EJ., Margis, R. & Waterhousc, P.M. (2003) Curr. Biol. 13, 236-240. 16. Zhang, H., KoIb, RA., Jaskiewicz, L., Westhot; E. & Filipυwicz, W. (2004) Cell

118, 57-68. 17. Liu, Q., Rand, T.λ., Kalidas, S., Du, R, Kim, R.E., Smith, D.P. & Wang, X.

(2003) Science 3OJ, 1921-1925.

^' 18. Caimell, M.A. & Harmon, GJ (2004) Nat. Struct. MoI, Biol. 11, 214-218. 19. StCfCk, L., Rornbauts, S., Jansscm, S., Sterky, F., Rouzc, P. & Van de Peer, Y,

(2005) New Phytυh 167, 165-170.

20. Blanc, G. & Wolfe, K.H. (2004) Plant Cell 16, 1679-1691. 21. Consentino, GP, Veiikatesan, S., Serluca, FC, Green, SR, Matthews, MB, &

Sonenberg, N (1995) Proc. Nad. Acad. ScL USA 92, 9445-9449. 22. FaIeI, RC, otaiiiςjii, F, McMillan, NM, W'iiiiams, BR & Sen GC (1995) Proc.

Natl. Acad, ScL USA 92, 8283-8287.

23. Hiraguri A, Itoh R, Kondo N, Nomura Y, Aizawa D, Murai Y. Koiwa H, Seki M ₁ Shinozaki K, & Fukuhara T {2005) Plant Mot Biol. 57 173-88.

24. Meisier G. & Tuschl T- (2004) Nature 431, 343-349.

25. Slerck, L., Rombauts, S.,Rouze, P. & Van de Peer, Y. (2005) http://bioinfQrmatics.p5b.ugcnt.be/pd-yJiste BBC 2005.pdf

26. Ma, J. & Beπnetzen J.L. (2004) Prυc. Natl. Acad. ScL USA 101, 12404-12410.

27. Hedges, SJS ₁ Blair, J.E., Vcnturi, M.L. & Shoe, J.L. BMC Evol Biol. (2004) 4:2 14-71-2148/4/2

28. Mochizuki, K. & Gorovsky, M.A. (2005) Genes and Development 19, 77-89.

Example 2. Demonstration of the involvement of DCL3 and DCL4 in transgene encoded hpRNA mediated silencing

A chimeric gene encoding a dsRNA molecule targeted to silence the expression of the phytoeπe dcsaturasc in λrabidopsis thaliana (PDS-hp) (according to WO99/53050) was introduced into A. thaliana plants with different genetic background, respectively wild- type, homozygous mutants for DCL2, DCL3 or DCL4. Silencing of the PDS gene expression results in photoblcaching.

The results of this experiment are shown in Figure 8. Silencing by the hpRNA encoding transgene of PDS expression was unimpaired in DCL2 or DCL3 mutant background compared to the silencing of PDS gene expression in a wild-type background, but was significantly reduced in a DCL4 mutant background. Unexpectedly, silencing in mutant DCI-3 background was significantly increased.

Example 3. Overexpressiαn of DCL4 fn A. thaliana and effect on the silencing of different silencing loci

Using standard recombinant DNA techniques, a chimeric gene is constructed COinprising the following operably Jinked DNA fragments;

• a CaMV 35S promoter region

• a DNA region encoding DCI-4 from A. thaliana (SHQ DO 1).

• A fragment of lhe 3' untranslated end from the octopine synthetase gene from Agrobacterium tumefaciens.

This chimeric gene is introduced in a T-DNA vector, between the left and right border sequences from the T-DNA, together with a selectable marker gene providing resistance to e.g. (he herbicide phosphinotricin. The T-DNA vector is introduced into Agrobacterium lumefaeiens comprising a helper Ti-piasmid. The resulting A, tumefaciens strain is used to introduce the chimeric JDCL4 gene in A. thaliana plants using standard A. thaliana transformation techniques.

Plants with different existing gene-silencing loci, particularly weaker silencing loci are crossed with the transgenic plant comprising the chimeric DCL4 gene and progeny is selected comprising bυth the gene-silencing locus and the chimeric DCL4 gene.

The following gene-silencing loci comprising the following silencing KNA encoding chimeric genes are introduced:

35S-hpCHS: a chimeric gene under control of a CaMV35S promoter which upon transcription yields a hairpin dsRNA construct comprising long complementary sense and antisense regions of the Chalcone synthase coding region (as described in WO 03/076620 )

35S-hpEIN2; a chimeric gene under control of a CaMV35S promoter which upon Lmnscriptiυn yields a hairpin dsRNA construct comprising long complementary sense and antisense regions of the ethylene insensitive 2 coding region (as described in WO 03/076620.)

5S-GUShp93: a chimeric gene under control of a CaMV35S promoter which upon transcription yields a hairpin dsRNA construct comprising short cuinplcmenLiiTy SCπϊC HπU αϊϊtϊScfiSc regions of the CJ'.Jo coding i'Ogioύ (as described in WO 2004/073390).

AtU6+20-GUShp93: a chimeric gene under control of a PoIIIl type promoter which upon transcription yields a hairpin dsRNA construct comprising short complementary sense and antisense regions of the OUS coding region (as described in WO2004/073390)

35S-GUS : a conventional GUS co-suppression construct (note that one of the lines used is a promo Ler-cυsuppressed GFP line).

35S-asETN2-PSTVd: a chimeric gene under control of a CaMV35S promoter which upon transcription yields an RNA molecule comprising a long aniisense region of the ethylene insensitive 2 coding region and further comprising a PTSVd nuclear localization signal (as described in WC) 03/076619)

The progeny plants exhibit a stronger sileniririg of the expression of the respective target gene in the presence of the chimeric DCL4 gene than in the absence thereof.

Example 4: Introduction of different silencing loci in a dcl3 genetic background

The gene silencing loci mentioned in Example 2 are introduced into A. thalina dd3 by crossing. The progeny plants exhibit a stronger silencing of the expression of the respective target gene in the absence of a functional PCL3 protein than in the presence thereof.

Example 5: RNAMnduclng hairpin RNAs in plants act through the viral defence pathway

The plant species, Arαhidopsis ύiαliαnα., has four Dicer-like proteins that produce differently-sized small RNAs, which direct a suite of gcπc-silcncing pathways. DCLl produces miRNλs ⁴ , DCL2 generates both stress-related natural antisense transcript

SiRNAs and siKNAs against at least one virus ⁶ , DCL3 makes ~24nt siRNAs that direct helerυchromatin formation ⁶ , and DCL4 generates both rrø/ι.τ-actiπg siRNAs which regulate some aspects of developmental timing, and siRNAs involved in RNAi ^7"9 . To obtain further detail of the pathways involved in RNAi and virus defence, we examined the size and efficacy/function of small RNλs engendered by a number of RNAi-inducing hpRNλs, two distinct viruses, and a viral satellite RNA in different single and multiple DfZ-rnutant Arabiώψsis backgrounds. Examination of siRNA profiles from more than 30 different hpRNλ constructs in wϋd-type (Wt) Arabidopsis, targeting either endogenes or transgcnes, revealed that the predominant size class is usually ~21nt with a smaller proportion of ~24nt RNAs. However, the 21/24nt ratio can vary depending on the construct. To examine hpRNA-derived siRNAs in Del mutants, a hpRNA construct (hpPDS), regulated by the rubisco small subunit (SSU) promoter, was made that targeted The phytoene desaturase gene (Pd\); silencing Pds causes a photoWeachcd phenotype in plants ³ . This construct was transformed into Wl plants and into plants that were homozygous mutant for Dcl2, DcB or DcU. The primary Wt and dcl2 transfomianls showed similar degrees of photobleachiπg, dcl3 transformants exhibited extreme photobleaching, and dc!4 Lransfαrmants were mildly photϋbleached (Fig M). The mild silencing in dcl4 indicates that. DCL4 activity is important, but not essential, for RNAi. To further test this, the dcl4 line (dcl4-l) and a different mutant line (dcl4-2) were transformed with an hpRNA construct targeting the chalcoπe synthase (Chs) gene. CHS is required for anlhocyamn production; silencing the gene reduces the production of red/brown pigment in the hypocotyls of young seedlings and in the seed coat ³ . Approximately 30% of the dcl4-l and 20% of the dcl4-2 plant lines Iran-formed with hpCHS had green hypocotyls and yielded pale seed, affirming that DCL4 activity is not essential for RNλi. In dcti ^' plants, hpPDS produced stronger photobleaching than in Wt, showing that OCL3 activity is not required for RNAi. Indeed, its absence appears to enhance silencing. Therefore, we investigated whether DCL2 was processing hpRNA into RNAi-mediating siRNAs in the absence of DCL4.

λ construct (hpGFP), containing a green fluorescent protein (GFF) goiiύ and an iψRNA transgcnc against GF1\ was transformed into dcl4-l and dcl4-l/ dcl2 lines. No primary

hpGbV/dcl4-I transformants showed any GFP expression but 5 primary hpGFP/A:/4- Ildcl2 traivjformanls expressed GFP. This suggested lhat RNAi can operate in the absence DCl/4, but not in the absence of both DClA and DCL2. To examine this further, a crossing strategy was undertaken. A hpPDS/r/c72 line was crossed with dcl4-2 Io produce a double heterozygous plant which had also inherited hpPDS. This was self- pollinated to produce progeny that were germinated on media, selective for inheritance of the hpPDS construct, and monitored for symptoms of photoblcaching. Most of lhe seedlings exhibited phυtobieaching, but a few were unbleached. Genotyping the unbleached seedlings revealed that they were double homozygous dcϊ2ldcl4-2. Seedlings with any of the other possible genotype combinations exhibited a degree of photoblcaching similar to that of the parental hpPDS/dcl2 line, except for a small number which had slightly less severe photobleachitϊg and were homozygous dcl4-2 in combination with either heterozygous Dcl2 or wild-type. The levels of Pds mRNA and lipPDS siRNA profiles were examined in the different genotypes. There were 21 and 24nt siRNAs in both Wt and dcl2, 22 and 24nt siRNAs iα dcU-2 and only 24nt siRNAs

• detectable in dcl2-dcl4-2. These results suggest that the 24nl siRNAs have no role in directing mRNA degradation, that 21nt siRNAs are produced by DCL4 and are the major component directing the mRNA degradation, and lhat DCL2 (especially in the absence of

DVXA) produces 22nt siRNAs that can also direct mRJM A degradation.

To examine the roles of the differently-sized siRNAs in defending plants against viruses, the range of Del mutants was challenged with Turnip mosaic virus (TuMV) and Cucumber mosaic virus (CMV), with or without its satellite RNA (Sat). About 18 days post inoculation (dpi), siRNAs derived from C]MV or Sat were readily delectable in Wt Arabidopsis plants. Analysing the Del mutants at 18 after infection with CIMV, CMV+Sal, or TuMV revealed essentially the same siRNA/Dd-mutant profiles as were obtained for the hpPDS/Dcl-mutants. Furthermore, the .steady-state levels of CMV and Sat genomic RNAs were higher in dcl2-dcl4 than in Wt plants. These results suggested that, in planls, hpRNAs arc processed into siRNAs and arc used to target RNA degradation by the same enzymes and co factors ϊlϊat ύic ^r used to recognise and restrain viruses. However, when a triple dd2-dcl3-dcl4-2 mutant was similarly infected, no

siRNAs were detectable and the CMV and Sat genomic RNA levels were even higher. This implies that PCU plays a role in restricting viral replication and/or accumulation, and contrasts with the increased, rather than decreased, silencing observed for the hpPDS in dcl3 mutants. To investigate this, dcl3 plants were infected with CMV-Sat and the resulting siRNA profile was compared to that in hpPDS/tfcW. In both cases, the production of 24nt siRNAs was abolished. This similarity in -24 siRNA production, but dichotomou.s consequences, may be explained by DCL3 cleaving the transient double- stranded replicative form of viral RNA to directly reduce its steady-state level, whereas cleavage of hpRNA stems by DCL3 compromises RNAi by removing substrate that would otherwise be processed by DCL2 and DCL4 iπlo 21 and 22nl siRNAs, respectively.

If hpRNAs arc processed like dsRNA from an invading virus, they may also evoke other virus-like characteristics. It has been well demonstrated that virus-infected cells in a plant are able to generate and transmit a long-distance specific signal to uninfected cells thereby triggering a silencing- like response which defends against virus spread . It has also been shown that viruses contain suppressor proteins that suppress the virus defence response ¹⁰ . Therefore, we conducted grafting experiments to test whether hpRNAs arc processed to produce such a signal, and whether RNAi directed by hpRNAs could be prevented by the transgenic expression of the viral suppressor protein HC-Pro" ^"12 . Scions from a tobacco plant expressing a GUS reporter gene were grafted onto rootstocks from plants' transformed with an anli-GUS hpRNA construct, and scions from Arabidυpm plants expressing GFP were grafted onto rootstocks transformed with an anti-GFP hpRNA construct. In both systems, the reporter gene in the newly-developing tissues of the scion was silenced. Tobacco plants containing an anti-Potato virus Y construct (hpPVY) and sibling plants also expressing HC-Pro were analysed for their response to inocukition with PVY. The plants containing hpPVY weie protected against PVY whereas plants containing the same construct in the He-Pro background were susceptible to the virus. Both sets of results further show that hpRNAs are processed by

References for Example 5

1. Vaucheret, H- (20Qd) Post-transcriptional small RNA pathways in plants: mechanisms and regulations. Genes & Development 20 759-771.

2. Paddison, PJ., Suva, J.M., Conklin, D.S., Schlabach, M., Li, M., Aruleba, S-, Balija, V., O'Shaughncssy, A., Cmoj, L., Scobie, K., Chang, K., Westbrook, T., Cleary, M., Sachidaπandam, R.. McC:ombie, W.R., Ellcdge, SJ. and Hannon, GJ. (2004) A resource for large-scale RNA-interfcrcncc-based screens in mammals. Nature 428, 427-431.

3/Wcslcy, S.V., Hclliwcll, C, Smith, N.λ., Wang, M-B., Rouse, D., Liu, Q., Gooding, P. _; Singh, S., Abbott, D-, Stoutjcsdijk, P,, Robinson, S., Gleave A., Green, A. and Wuterhouse, P.M. (2001) Constructs for Efficient, Effective and High Throughput Gene Silencing in Plants. Plata J. 27, 581-590-

4. Park, W, Ii, J, Song, R, Messing, J, Chen, X: (2002) CARPEL FACTORY, a Dicer honiolog, and HENl, a novel protein, act in microRNA metabolism in Arabidopsis thaliana. CurrBiol. 12, 1484-1495.

5. Borsani O, ZIm J, Vcrslucs PE, Sunkar R, Zhu JK. (2005) Endogenous siRNAs derived from a pair of natural cis-antisense transcripts regulate salt tolerance in Axabidopsis. Cell 123, 1279-91.

6. Xie, Z., Johansen, L.K., Gustafson, A.M., Kasschaυ, K.D., Lellis, A.D., Zilberman, D., Jacobsen, S.E. and Caπϊngton, J.C (2004) Gctictic and functional diversification of small

RNA pathways in plants. PLoS BioL B, E 104

7. Gasciolli, V., Mtdlory, A.C., Bartel, D.P. and Vaucheret, TI, (2005) Partially redundant functions of Arabidopsis Dicer-like enzymes and a role for DCL4 in producing trans- acting SiRNAi;. Curr. Biol 15, 1494-1500.

8. Xie, Z., Allen, E., Wilken, A. and Carrington, J.C. (2005) Diccr-LIKE 4 functions in trans-acting small interfering RNλ biogenesis and vegetative phase change in Arabidopsis thaliana. Proc. Natlλcad, Sci. USA 102, 12984-12989.

9. Dunoyer P, Himber C, Voinnet O. (2006). Dicer-LIKE 4 is required for RNA interference and produces the 2i-nucleotide small interfering RNA component of the plant ccll-lo-cell silencing signal. Nature Genet 37,1356-1360,

10. Voinnet, O. (2005) Induction and suppression of RNA silencing; insights from viral infections. Nature Rev Genet. 6, 206-220.

1 1. Mallory, A. K., Ely, L., Smith, T. H., Maralhe, R., Anandalakshmi, R., Fagard,

M., Vaucherel, H., Pruss, G. _; Bowman, L. & Vance, V. B. (200I )HC-PiO suppression of transgene silencing eliminates the small RNAs but not tπtnsgene mcthylation or the mobile signal. Plant Celt 13, 571-583.

12. Anandalakshmi, R., l*russ, G. J., Marathe, R., Mallory, A. C, Smith, T. H. & Vance, V. B. (1998) A viral suppressor of gene silencing in plants. Proc. Natl Acad, ScL USA 95, 13079-13084.

13. Waterhouse, P.M., Wang, M-B & Lough T. (2001) Gene silencing as an adaptive defence against viruses. Nature 411, 834-842.

14. Rccd, J. W., Nagataαi, A., Elich, T. D., Fagan, M. and Chory, J. (1994) Phytodirome A and phyiochrome B have overlapping but distinct functions in Arabidopsis development. Plant Physiol. ϊ04, 1139-1149.

Example 6: Effect of mutations affecting transcriptional gene silencing on the post-transcriptlonal gene silencing achieved by introduced silencing RNA encoding chimeric genes

Transgenic Atabidopsis plant? which when transcribed yield hpRNA comprising EIN2, CHS or PDLS specific dsRNA regions were crossed with Arabidopsis lines a having background comprising a mutation in lhe CMT3 encoding gene and offspring comprising both the transgeπe and the background mutation have been selected. Alternatively. Arabidopsis plants comprising a background having a mutation in RDR2 were transformed through floral dipping with the above mentioned hpRNA encoding chimeric genes.

Figure 9 shows the effect of CMT3 mutation on hpRNA-mediated E1N2 and CHS silencing. The length of hypocotyls grown in the dark on ACX ¹ containing medium, is generally longer for flic F3 hpEIN2 plants with the homozygous cmt3 mutation than with the wild-type background (wt), indicating stronger EIN2 silencing in the ont3 background. The transgenic plants inside the box have the mutant background, while tilts transgenic plants outside the box have the wild-type background. In hpCHS containing plants, the seed coat color is significantly lighter for the hpCHS plants with the cmt3 background than with the wild-type background, indicative of stronger CHS silencing in the former transgenic plants.

λrabidosis plants comprising a 35S-hpl ^J DS transgene and a mutation in RDR2 exhibited more cotyledon and leaf bleaching were significantly more silenced than plants comprising only the 35S-hpPDS transgene.

Both lines of experimentation indicate thai a relief of transcriptional silencing through reduction of the functional level of proteins involved in transcriptional silencing enhance the post-transcriplional silencing of the target genes such as E1N2, CHS or PDS,

mediated through the introduction of dsRNA encoding chimeric genes targeted to these genes.

Table 1. Variation within and between DCLs of Rice, Poplar and Arabidopsis

Ln

Variability : Amino acid substitutions/100 sites

^* AtDCL3A was removed from group as deletion in this domain meant thai its inclusion would drastically reduce the number of sites analysed