In Vitro Synthesized Small Interfering RNAs Elicit RNA Interference in African Trypanosomes AN IN VITRO AND IN VIVO ANALYSIS* □ S

RNA interference (RNAi) describes an epigenetic gene silencing reaction by which gene-specific double-stranded RNA acts as a trigger to induce the ribonucleolytic degradation of homologous transcripts. RNAi in African trypanosomes has been shown to be involved in regulating the transcript abundance of retroposons, and the process currently represents the method of choice in gene function studies of the parasite. However, little is known concerning the mechanistic and structural aspects of the processing reaction. This is in part due to the absence of a trypanosome-specific RNAi in vitro system. Here we demonstrate that both the Dicer and the RNA-induced silencing complex steps of the RNAi reaction pathway can be monitored in vitro using cell-free trypanosome extracts. The two in vitro activities and the generated small interfering RNAs (siRNAs) are characterized by features known from other organisms, and we demonstrate that chemically as well as enzymatically synthesized siRNAs are functional in the parasite. Thus, the transfection of synthetic siRNAs can be used to rap-idly monitor gene knockdown phenotypes in Trypanosoma brucei , which should be helpful in genome-wide, RNAi-based screening experiments.

In animals, double-stranded RNA (dsRNA) 1 has been shown to induce a gene-specific silencing reaction known as RNA interference (RNAi) (2,3). The phenomenon is characterized by the ribonucleolytic degradation of the dsRNA into 21-26-ntlong RNA molecules, which have been termed small interfering RNAs (siRNAs). siRNAs represent trans-acting specificity determinants of the RNAi reaction pathway, which results in the degradation of homologous transcripts. The in vivo function of RNAi seems to be an adaptive genome defense reaction against viruses and mobile genetic elements in addition to functioning in the developmental regulation of gene expression (1)(2)(3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14). The reaction pathway has been shown to be widely conserved, and as a consequence, has become a powerful experimental tool to study gene knockdown or loss-of-function phenotypes in many organisms (3,(15)(16)(17).
The current knowledge of the mechanism and the participating molecular components of the RNAi pathway is mainly derived from cell-free RNAi in vitro systems (18 -21). It was shown that the dsRNA trigger is first processed by the nuclease Dicer, which is a member of the class 3 ribonuclease III superfamily. Dicer contains two RNaseIII domains (18), and each domain coordinates a Mg 2ϩ or Mn 2ϩ ion. The cations function as nucleophiles in the hydrolysis of the two RNA strands (22)(23)(24). The cleavage reaction generates the above described 21-26-nt siRNAs, which are double-stranded with 2-nt singlestranded 3Ј overhangs. The 5Ј termini are phosphorylated, whereas the 3Ј ends contain hydroxyl groups (13). For recombinant human Dicer, it was shown that the dsRNA cleavage is an endonucleolytic reaction (25).
The Dicer-generated siRNAs are assembled together with accessory proteins, such as Drosophila R2D2, into a so-called "RNA-induced silencing complex" (RISC) (1, 20, 23, 26 -29). For the sequence-specific mRNA recognition, siRNAs become unwound in an ATP-dependent reaction in the presence of AGO2 (30,31). The thermodynamic stability at either end of the siRNA determines which of the two strands of the symmetric molecule remains in the RISC complex and guides the recognition of the mRNA via base pairing (1,32). After the hybridization, the mRNA is endonucleolytically cleaved by the endonuclease Slicer (4,5), and the resulting mRNA fragments are likely subjected to nonspecific degradation processes in the cytoplasm.
RNAi has been identified in higher eukaryotic organisms as well as in lower eukaryotes such as Trypanosoma brucei, a protozoan parasite (33,34). This suggests that the reaction pathway emerged early during evolution, although not all protozoan organisms are RNAi-positive (33,35,36). For T. brucei, which is the causative agent of sleeping sickness in Africa, it was shown that siRNAs with a length of 24 -26 nt appear after transfection or expression of dsRNA in vivo. About 30% of these siRNAs encode sequences from retroposons, which indicates that RNAi, at least in part, is involved in the control of retroposon transcripts (37). However, in contrast to other organisms, only one RNAi component has so far been identified in the parasite. The polypeptide is a member of the AGO protein family and has been termed TbAGO1 (38,39). TbAGO1 is essential for RNAi in insect stage trypanosomes and was shown to be assembled in siRNA-associated ribonucleoprotein particles, which are important for the stabilization and/or generation of siRNAs (39). Surprisingly, although the T. brucei genome data base is near to completion, data base mining has up to now failed to identify a Dicer homologue (33). Furthermore, biochemical in vitro systems for the characterization of putative RNAi components have not been described, despite the fact that the RNAi pathway is the preferred technology to down-regulate gene expression in the parasite (for recent reviews, see Ullu et al. (33) and Motyka and Englund (6)).
Here we present two in vitro assays that recapitulate the dsRNA processing and mRNA degradation steps of the RNAi pathway in T. brucei cell-free extracts. The dsRNA processing reaction shows Dicer-like characteristics, and the in vitro-generated siRNAs are characterized by structural features known from other organisms (13,25,40,41). We further demonstrate that synthetic siRNAs are capable of mediating a sequencespecific mRNA degradation reaction in vitro and that the transfection of siRNAs results in a sequence-specific knockdown in vivo.

Dicer in Vitro Assay
T. brucei cell-free extracts were centrifuged at 4°C for 20 min at 200,000 ϫ g. The supernatant (50 l, protein concentration 2-3 mg/ml) was mixed with 60 pM radioactively labeled ␣-tubulin dsRNA in 50 l of 25 mM HEPES/KOH, pH 7.5, 150 mM KCl, 100 mM sucrose, 5 mM MgCl 2 , 2 mM ATP and incubated for 60 min at 28°C. The reaction was terminated by the addition of phenol, and a 5Ј radioactively labeled 18-nt RNA molecule was added as an internal standard. After phenol extraction and ethanol precipitation, the RNA was analyzed in urea-containing polyacrylamide gels. Radioactive signals were detected by phosphorimaging and analyzed using the software Image Gauge 3.41 (Fuji).

RISC in Vitro Assay
T. brucei cell-free extracts were centrifuged at 4°C for 12 min at 100,000 ϫ g. The supernatant (50 l, protein concentration Յ4 mg/ml) was mixed with 10 pM radioactively labeled ␣-tubulin sense or antisense RNA and 10 -100 nM siRNA si-315 in 50 l of 25 mM HEPES/ KOH, pH 7.5, 150 mM KCl, 100 mM sucrose, 5 mM MgCl 2 , 2 mM ATP, 500 M CTP, GTP, and UTP, 0.2 units/l RNasin (Promega). After incubation (28°C, 60 min), the reaction was terminated by the addition of phenol and further treated as described for the Dicer in vitro assay.
pZJM-␣TubTrunc-The coding region of ␣-tubulin (position 299 -414) was amplified using primers 5Ј-CCCAAGCTTCGGCCAACAAC-TACGCTCG-3Ј and 5Ј-CCGCTCGAGATACACGAGGAAGCCCTGAAG-3Ј. After digestion with XhoI and HinDIII, the PCR product was inserted into pZJM (46), replacing the ␣-tubulin coding region. Plasmid pRNAi-Timer contains a constitutive expression site for Fluorescent Timer (45) and a tetracycline-inducible RNAi site, consisting of two T7 head-to-head promoters flanking the Fluorescent Timer coding region. For the construction of pRNAi-Timer, the vector pZJM-Timer was digested with BamHI and KpnI. The resulting 1280-bp RNAi site was ligated into the EcoRI restriction site of vector pLew82-Timer⌬Op, which is an operator-minus derivative of pLew82 (47). To eliminate the operator, pLew82 was digested with BglII and religated. The coding region of Fluorescent Timer was amplified as described for pZJM-Timer with the exception of the timer-antisense primer. The luciferase gene in pLew82 was substituted by the PCR product, using the BamHI and HinDIII restriction sites.

RESULTS
Dicer in Vitro Assay-siRNA transfection has been demonstrated as a fast and efficient method to knock down gene In Vitro RNAi in Cell-free Extracts of African Trypanosomes expression in higher eukaryotic organisms (15,49). Although siRNAs have been identified in African trypanosomes (37), the technique has so far not been used in the protozoan parasite. In addition, structural and mechanistic aspects of the RNAi reaction in trypanosomes are only poorly understood. This is in part due to the absence of an RNAi in vitro system to monitor the individual steps of the RNA processing reaction. siRNAs have been shown to be generated by the ribonucleolytic action of the RNaseIII-type enzyme Dicer (18,25). Thus, we tried to identify a Dicer-like activity in cell-free extracts of trypanosomes. Fig.  1A shows a representative example of the Dicer in vitro assay. Incubation of a 131-bp internally 32 P-labeled dsRNA with a cell-free protein extract of insect-stage T. brucei resulted in the processing of the dsRNA into a population of siRNA-like molecules with a length of about 25 nt. The extent of cleavage was dependent on the extract concentration and on the duration of the incubation. siRNA-like molecules appeared after 15 min and reached a maximal level around 90 min. Two dsRNA processing intermediates with lengths diagnostic of a gradual removal of one or two siRNA units appeared with a similar kinetic. At a time point of 180 min, Ͼ95% of the input dsRNA was processed.
Titration experiments (Fig. 1, B and C) revealed that the Dicer-like activity is dependent on the presence of optimally 1 mM ATP and that it requires divalent cations such as Mg 2ϩ or Mn 2ϩ but not Ca 2ϩ . ATP can be substituted by ADP and by the ATP analogues ATP␥S and AMP-CPP, which contain only one hydrolyzable phosphoanhydride bond (data not shown). The Dicer-like activity further showed a pH optimum of pH 8 and a temperature optimum of 27°C, which represents the optimal growth temperature of the insect life cycle stage of the parasite (data not shown).
Hydrolysis of the in vitro-generated siRNAs with the single strand-specific RNase T1 verified that the short RNAs are mostly double-stranded ( Fig. 2A). However, an increase in the electrophoretic mobility suggested the presence of short singlestranded overhangs of about 1-2 nt, which was confirmed by nuclease S1 digestion (data not shown). Further experiments demonstrated that both strands of the input dsRNA were present in the siRNA-like molecules (Fig. 2C).
Since 5Ј-phosphate groups are important for the recruitment of siRNAs into the RISC-catalyzed mRNA cleavage reaction (1, 31), we addressed the phosphorylation state of the generated siRNAs. The RNAs were incubated with either alkaline phosphatase or T4 PNK, which contains a 3Ј-phosphatase activity. High resolution PAGE at denaturing conditions was used to detect differences in the electrophoretic mobility. Although alkaline phosphatase-treated siRNA-like molecules migrated with a reduced mobility of about 0.5 nt, PNK-treated siRNAs showed no difference (Fig. 2B). Thus, the in vitro-generated siRNAs are 5Ј-monophosphorylated. The data further demonstrated that the majority (Ն75%) of the generated siRNAs have a length of 24 -26 nt, which is in agreement with published data from in vivo studies (37). However, shorter RNAs with a length of 22 and 23 nt as well as molecules of 27 nt were also detected. To verify whether the long siRNAs might act as precursors for the generation of shorter siRNAs, we chemically synthesized a 26-nt siRNA (si-315-26) and incubated the molecule with a Dicer-competent T. brucei extract. No processing was observed over a period of up to 90 min (see supplemental figure).
Design of siRNAs and Knockdown Experiments-The described experiments established that cytosolic extracts of T. brucei contain a Dicer-like activity capable of generating short RNAs that resemble the siRNAs in higher eukaryotes (15,18,25,41). To test whether chemically or in vitro transcribed siRNAs can be used as reagents to knock down gene expression in trypanosomes, we designed several ␣-tubulin- specific siRNAs (Fig. 3A). ␣-Tubulin was chosen because its gene silencing has been shown to cause a well defined morphologic phenotype known as FAT cells (17). The design of the siRNAs was based on the recently determined correlation between the internal thermodynamic stability of siRNAs and their silencing efficiency in higher eukaryotic organisms (32,44). siRNAs si-315 and si-956 (Fig. 3A) were derived from the ␣-tubulin coding region starting at position 315 and 956. The two siRNAs show a low thermodynamic stability at the 3Ј ends of the sense strands and within their central regions. The stability at the 5Ј ends is increased (Fig. 3B). siRNA si-329 was designed with inverse characteristics. Its thermodynamic stability profile resembled the features of a non-functional or low efficiency siRNA in higher eukaryotes (44). Lastly, we synthesized pre-si-315. The RNA has a length of 50 nt and folds into a hairpin structure, which represents a precursor of siRNA si-315. The individual RNA strands of si-956 and si-329 were synthesized by solid phase RNA synthesis using 2Ј-t-butyldimethylsilyl-protected phosphoramidites. si-315 and pre-si-315 were generated enzymatically by in vitro transcription.
Eighteen hours after the siRNA transfection, we determined the fraction of FAT cells within the growing parasite population. To account for dead cells after the electroporation, we used the T. brucei cell line pRNAi-Timer. T. brucei pRNAi-Timer constitutively expresses the encoded reporter protein Fluorescent Timer (45). Fluorescent Timer is characterized by the time-dependent appearance of two conformations of the polypeptide, which have different fluorescence absorption and emission maxima. In vivo synthesized Fluorescent Timer shows an initial green fluorescence, which shifts to a red fluorescence after ϳ24 h (45). Consequently, only parasite cells with an unaltered green fluorescence were counted. As shown in Fig. 4, A and B, both chemically and enzymatically synthesized siRNAs were functional, and all siRNAs with the proper thermodynamic stability profile caused a FAT phenotype. The highest knockdown value was achieved with siRNA si-315 (85%), which was nearly as efficient as a 113-bp ␣-tubulinspecific control dsRNA (93%). For si-956, a value of 70% was determined, and transfection of si-329 resulted in 1.3% FAT cells. The latter value was comparable with control cells treated with a Fluorescent Timer-specific dsRNA (0.8%) or with mock-treated parasites (0.3%). Transfection of pre-si-315 RNA resulted in about 70% FAT cells, indicating that the RNA hairpin was processed into a functional siRNA as expected (7). The siRNA-induced silencing phenotype was concentration-dependent; 25 g (per 10 8 parasite cells) of si-315 resulted in a knockdown level of 32%. Increasing the si-315 amount to 60 g raised the silencing efficiency to 85%.
To correlate the phenotypic silencing values with steady state ␣-tubulin mRNA concentrations, we performed a Northern hybridization analysis 18 h after the transfection. The results are shown in Fig. 4C. As expected, ␣-tubulin mRNA levels were significantly decreased in the ␣-tubulin dsRNAtreated cell line as well as in si-315-and si-956-treated parasites. For the si-329 transfected trypanosomes, only a small reduction in the ␣-tubulin mRNA level was detected, in line with functioning as a low efficiency siRNA.
siRNA-mediated, Site-specific mRNA Cleavage in Vitro-Finally, we addressed the question whether the above described siRNA si-315 was able to mediate the site-specific cleavage of ␣-tubulin mRNA in vitro. For that, we established a RISC in vitro assay. A truncated version of internally radiolabeled ␣-tubulin mRNA (233 nt) was incubated with a T. brucei cytosolic extract and analyzed by denaturing PAGE. Fig. 5 shows a representative time course experiment. In the absence of si-315, no mRNA cleavage was observed over a period of 2 h. In the presence of si-315, however, the input mRNA was ribonucleolytically hydrolyzed after 30 min. An RNA fragment with a length of 145 nt was detected, which corresponds to the expected RISC-mediated 5Ј cleavage fragment of the ␣-tubulin mRNA. On the contrary, no corresponding 3Ј fragment (88 nt) was identified. This is identical to the situation in Drosophila in which stable 3Ј cleavage fragments could only be identified in chemical modification experiments (8). The RISC-mediated mRNA cleavage reaction reached a maximal yield of roughly 5% at an incubation time of 90 min.
Since the thermodynamic stability at the 5Ј and 3Ј ends of the si-315 sense strand differed by about 4.4 kcal/mol, we also compared sense and antisense RNA cleavage. For this, we used either sense or antisense ␣-tubulin RNA containing a targeting site for si-315. As shown in Fig. 6, only the sense strand was hydrolyzed by the RISC in vitro activity, and no cleavage of the antisense RNA, even at 2 h of incubation, was detected. The specificity of the si-315-programmed RISC cleavage reaction was finally confirmed in control experiments with siRNA si-956. Incubation of si-956 in the in vitro RISC assay induced no cleavage reaction, but the addition of increasing amounts of si-956 in the presence of si-315 resulted in a competition of the mRNA cleavage activity (data not shown). DISCUSSION RNAi-mediated down-regulation of gene expression has become the method of choice to study the function of genes in African trypanosomes (17,46,50,51). RNAi has been shown to FIG. 2. In vitro-generated siRNAs are double-stranded and 5phosphorylated. A, RNaseT1 (T1) hydrolysis and thermal denaturation (95°C) of in vitro-generated siRNAs. Electrophoresis was performed in urea-containing 6% (w/v) polyacrylamide gels. ssRNA represents a radiolabeled, single-stranded RNA control oligonucleotide of 25 nt. Relative signal intensities (si) were derived as si(siRNA)/si(IS). M, radiolabeled 18-and 25-nt marker RNAs. B, characterization of the 5Ј-and 3Ј-terminal ends of in vitro-generated siRNAs. siRNAs were incubated with either alkaline phosphatase (AP) or T4 PNK and separated in denaturing 18% (w/v) polyacrylamide gels. OH, alkaline hydrolysis ladder of a 5Ј-radiolabeled RNA oligonucleotide (35 nt). IS, internal RNA standard of 18 nt length. An oligoribonucleotide of 25 nt was used as a control and was either 5Ј-phosphorylated (5ЈP) using [␥-32 P]ATP or 3Ј-radiolabeled (3ЈP) using [5Ј-32 P]5Ј-3Ј-cytidine diphosphate. C, in vitro-generated siRNAs derived from internally radiolabeled sense (s) or antisense (as) strands of a ␣-tubulin dsRNA.
regulate the transcript abundance of retroposons in the parasite (37,39,52); however, only little is known about the mechanistic details and the participating molecular components of the reaction. Although the sequence analysis of the T. brucei genome is near to completion, only one RNAi component has so far been identified (38,39); TbAGO1 is a member of the AGO protein family, and it is essential for RNAi in insect stage trypanosomes (39). Importantly, no Dicer homologue has been identified to date, perhaps suggesting a highly divergent gene sequence (33). As a consequence, we aimed at establishing two cell-free assays to analyze the Dicer-catalyzed initiation step and the RISC-mediated effector step in vitro.
Using T. brucei cell-free protein extracts, we were able to demonstrate that dsRNA can be processed into siRNA-like molecules by a Dicer-like ribonucleolytic activity. The biochemical characteristics of the reaction are similar to features known from other Dicer in vitro assays (13, 18, 19, 25, 40, 53), and the generated siRNAs are double-stranded; they contain 1-2 nt 3Ј overhangs and are 5Ј-monophosphorylated. The RNAs could be produced from endogenous as well as exogenous dsRNA and showed a length distribution of predominantly 24 -26 nt. This is in line with data from cloning experiments of T. brucei siRNAs, which identified an identical length distribution (37). Thirty percent of these siRNAs contained retroposon sequences, suggesting a housekeeping function for RNAi in the parasite to control the abundance of retroposon transcripts (37). However, a minor amount of 21-23-nt-long siRNAs was also produced in the in vitro assay, and the long siRNAs did not function as precursors for the generation of the short siRNAs. Whether that indicates that the parasite relies on two siRNA size classes similar to the situation in Arabidopsis (54) remains uncertain. Although the 21-23-nt siRNAs in Arabidopsis were shown to be involved in the posttranscriptional gene silencing reaction of mRNAs, the long siRNAs (24 -26 nt) play a separate role in transcriptional gene silencing processes such as systemic signaling of RNA silencing and RNA-directed DNA methylation (54). This is probably due to two separate Dicer activities: a nuclear enzyme that generates the 24 -26-nt siRNAs and  C). The ␣-tubulin mRNA amount of untransfected parasites (ϪRNA) was set to 100%. C, Northern hybridization of parasite cells 18 h after transfection with synthetic siRNAs. The abundance of Fluorescent Timer-specific transcripts (timer) was used as an internal control. ds ␣-tubulin represents a double-stranded ␣-tubulin-specific control RNA (113 bp), and ds timer represents a Fluorescent Timer-specific RNA of ϳ700 bp. a cytoplasmic Dicer producing the 21-23-nt siRNAs. Since posttranscriptional gene silencing as well as transcriptional gene silencing type of reactions have been identified in T. brucei (17,37,55) and since 21-23-nt-long siRNAs are clearly functional in the parasite, it is tempting to speculate that the described in vitro activity is due to two different Dicer activities. The activity that generates the 24 -26-nt siRNAs might be enriched or perhaps more active in the generated cell-free extracts, thereby creating the described siRNA length distribution. However, no genetic evidence is currently available to support such a scenario. As already mentioned above, this holds true for any candidate gene of Dicer as well as for additional genes of the Argonaute protein family. Aside from TbAGO1, no homologous sequences have been identified in the genome data base of T. brucei.
An obvious limitation of the Dicer in vitro reaction is the processivity of the siRNA formation. The processing reaction was limited to one or two processing steps, although Ͼ95% of the input dsRNA was ribonucleolytically hydrolyzed after an incubation period of 3 h. Whether that reflects a stability or degradation problem of an essential component of the reaction pathway or whether it is due to the lack of a processivity factor is unknown at present.
Since the in vitro-generated siRNAs showed no obvious difference to siRNAs characterized in other systems, we per-formed transfection experiments with chemically and enzymatically synthesized siRNAs. Both types of RNA molecules were functional, and knockdown values of Ն80% could be achieved. Within the context of the nearly completed sequencing of the T. brucei genome, this opens up the possibility of genome-wide, RNAi-mediated loss-of-function screens using synthetic siRNAs similar to what has been published in Drosophila, Caenorhabditis elegans, and mammals (56 -60). Furthermore, a comparison of the knockdown efficiencies of si-315 (85%) and si-956 (70%) suggests that a high thermodynamic stability at the 5Ј sense terminus in comparison with the 3Ј end of the sense strand may be a critical determinant for a high efficiency siRNA. The calculated ⌬G difference between the two siRNA ends is almost twice as high for si-315 (4.4 kcal/mol) as it is for si-956 (2.4 kcal/mol), which is consistent with a comparison of the thermodynamic stabilities between functional and non-functional siRNAs in higher eukaryotes (32,44). Furthermore, it has been shown that the relative thermodynamic stability at the ends of siRNAs determines which of the two strands enters the RISC. Decreasing the number of hydrogen bonds at the 3Ј end of the sense strand results in a preferential incorporation of antisense strands into the RISC (1, 32). Thus, the lower stability at the 3Ј end of si-315 sense strand might be responsible for the more efficient mRNA cleavage reaction. This was further sup- FIG. 5. RISC in vitro assay. Internally radiolabeled ␣-tubulin mRNA was incubated with a cytosolic protein extract in the absence (ϪsiRNA) or presence (ϩsiRNA) of si-315. At the indicated time points (in min), RNA reaction products were separated in denaturing 10% (w/v) polyacrylamide gels. Arrowheads mark the electrophoretic positions of the input ␣-tubulin mRNA, the 5Ј cleavage fragment, and the expected position of the 3Ј cleavage fragment. B, buffer/mock control. IS represents a radiolabeled, 18-nt-long RNA oligonucleotide, which was used as an internal standard. Internally radiolabeled ␣-tubulin sense (s) or antisense RNA (as) was incubated with a cytosolic protein extract in the presence (ϩsiRNA) and absence (ϪsiRNA) of si-315. At indicated time points (in min), RNA reaction products were separated in ureacontaining 10% (w/v) polyacrylamide gels. Arrowheads mark the electrophoretic position of the input sense and antisense RNAs and the 5Ј cleavage fragment, which can only be seen in the case of the sense RNA. B, buffer/mock control. IS represents a radiolabeled, 18-nt RNA oligonucleotide, which was used as an internal standard. M, RNA size marker with a length of 18 and 88 nt. ported in our RISC in vitro assay, in which si-315 was able to induce the cleavage of the sense RNA but not of the antisense RNA. The cleavage of internally radiolabeled mRNA in the assay showed that the 5Ј cleavage fragment but not the 3Ј fragment could be detected. The reason for this difference is at present not understood but likely reflects an RNA stability problem. However, the absence of a 3Ј cleavage fragment in vitro has also been reported in Drosophila, and it was only recently identified in chemical modification experiments using the cysteine-alkylating reagent N-ethylmaleimide in the assay (8).
Together, we have demonstrated that the transfection of siRNAs can be used as a rapid method for transient knockdown experiments in T. brucei. The different knockdown efficiencies of the various siRNAs suggest a base composition with a high G/C content at the 5Ј end and a high A/U content at the 3Ј end of the siRNA sense strand. The thermodynamic stability within in the central region should be low (44). The Dicer in vitro assays demonstrated that both the Dicer activity and the siRNA structures resemble features known from other organisms, and thus, it is very likely that the non-identified T. brucei Dicer is an RNaseIII-type endonuclease. Considering the early divergence of trypanosomes from the main eukaryotic lineage, our results indicate that the characteristics of RNAi are highly conserved throughout the entire kingdom of animals.