Human TRBP and PACT Directly Interact with Each Other and Associate with Dicer to Facilitate the Production of Small Interfering RNA*

Mammalian Dicer interacts with double-stranded RNA-binding protein TRBP or PACT to mediate RNA interference and micro-RNA processing. TRBP and PACT are structurally related but exert opposite regulatory activities on PKR. It is not understood whether TRBP and PACT are simultaneously required for Dicer. Here we show that TRBP directly interacts with PACT in vitro and in mammalian cells. TRBP and PACT form a triple complex with Dicer and facilitate the production of small interfering RNA (siRNA) by Dicer. Knockdown of both TRBP and PACT in cultured cells leads to significant inhibition of gene silencing mediated by short hairpin RNA but not by siRNA, suggesting that TRBP and PACT function primarily at the step of siRNA production. Taken together, these findings indicate that human TRBP and PACT directly interact with each other and associate with Dicer to stimulate the cleavage of double-stranded or short hairpin RNA to siRNA. Our work significantly alters the current model for the assembly and function of the Dicer-containing complex that generates siRNA and micro-RNA in human.

RNA interference (RNAi) 2 is an evolutionarily conserved mechanism for gene silencing mediated through small RNAs of ϳ22 nucleotides in length (1). At least two classes of small RNAs have been described in mammals, micro-RNAs (miRNA) produced from hairpin precursors and small interfering RNAs (siRNAs) derived from long double-stranded RNAs (dsRNAs) (2). Both miRNAs and siRNAs are generated by RNase III-type nuclease Dicer, and they are assembled into effector complex termed RNA-induced silencing complex (RISC) (3,4). Although two Dicer enzymes Dcr1 and Dcr2 have been found in fruit flies and are responsible for the generation of miRNAs and siRNAs, respectively, there exists only one single Dicer in humans, which produces both miRNAs and siRNAs (3). Dcr2 is known to associate with a dsRNA-binding protein (dsRBP) termed R2D2, which binds to siRNA and facilitates its passage from Dcr2 to RISC (5,6).
A new family of dsRBPs, which includes Loquacious in Drosophila as well as TRBP and PACT in humans, has been shown recently to interact with Dicer and be required for its function in RNAi (7)(8)(9)(10)(11)(12). Unlike its counterparts in Drosophila that have only one dsRBP partner, human Dicer appears to associate with two closely related dsRBPs, TRBP and PACT (10 -12). TRBP was originally identified and characterized by its high affinity for TAR, a hairpin RNA encoded by human immunodeficiency virus type 1 (13). PACT was initially cloned as a cellular protein activator of PKR kinase (14,15). Although both PACT and TRBP bind to PKR and have three similar dsRNA-binding domains (dsRBDs), TRBP exerts an inhibitory effect on PKR (16 -20).
Although both TRBP and PACT interact with and support the function of human Dicer in RNAi (10 -12), it is not understood whether their binding with Dicer is simultaneous or mutually exclusive. TRBP and PACT are closely related, and both are capable of homodimerization (21,22). In addition, they share two binding partners, PKR and Dicer (10 -12, 16 -20). This raises the possibility that TRBP and PACT might function as a protein complex in cells. In this study, we explored the direct interaction between TRBP and PACT as well as its impact on Dicer and RNA silencing.
Protein Purification and Protein Analysis-GST-TRBP, MBP-TRBP, His-PACT, and GST-PACT, as well as the truncated forms of His-TRBP and His-PACT, were expressed in Escherichia coli BL21 (DE3) strain. GST fusion proteins were purified by glutathione-Sepharose column (Amersham Biosciences) and eluted with 50 mM reduced glutathione. MBP-TRBP was purified through amylose resin (New England Biolabs) and eluted with 10 mM maltose solution. Polyhistidine-tagged proteins were purified through nickel-chelating column (Qiagen) and eluted with 200 mM imidazole. All fusion proteins were dialyzed overnight at 4°C. To remove contaminating RNA bound to TRBP/ PACT, protein purification was performed in the presence of 200 g of RNase A (Sigma). All proteins were purified to Ͼ90% homogeneity as verified by SDS-PAGE analysis. All preparations of purified proteins were free of contaminating RNA as measured by UV absorbance (A 260 ). GST pulldown, co-immunoprecipitation, and Western blot analysis were carried out as described (27,28).
Mouse monoclonal anti-V5 antibody was purchased from Invitrogen. Rabbit polyclonal anti-Myc serum was from Sigma. Mouse anti-Dicer and goat anti-T7 antibodies were from Abcam. A rabbit anti-TRBP serum was raised against purified recombinant GST-TRBP-A. Specificity of anti-TRBP antibodies was verified with lysate of TRBP-overexpressing HEK293T cells.
Fluorescence Resonance Energy Transfer (FRET) Imaging-Multicolor fluorescence microscopy was performed as described previously (29,30). HeLa cells were transfected with pcDNA-CFP-TRBP, pcDNA-YFP-PACT, or both. Cells were fixed with methanol/acetone (1:1, v/v) at 24 h post-transfection. Fixed cells were excited either at 820 nm (for CFP excitation) or 488 nm (for YFP excitation), and signals were recorded at either 435-485 nm (signal for CFP) or 535-590 nm (signal for YFP) on a Zeiss LSM510 META laser scanning microscope. FRET signal was captured at 535-590 nm during excitation at 820 nm. FRET signal between CFP and YFP was corrected using Zeiss LSM software.
Protein Fractionation Analysis-Two C57 mouse testes were dissected and homogenized in 1 ml of MTPBS buffer with protease inhibitor (Roche Applied Science) for 1 min on ice. Homogenized testes were centrifuged at 14,000 ϫ g for 30 min at 4°C. Protein lysates were treated with RNase A at 50 g/ml for 15 min at 4°C and then passed through a 0.22-m filter. Filtered lysates were loaded into Superdex 200 gel filtration column (Amersham Biosciences) and were fractionated at 0.3 ml/min in MTPBS buffer. Fractions (1 ml each) were lyophilized overnight and redissolved in 100 l of MTPBS buffer. Samples were analyzed by Western blotting with rabbit anti-TRBP, rabbit anti-PACT (Abcam), and mouse anti-Dicer (Abcam) antibodies.
In Vitro Dicer Cleavage Assay-dsRNA was transcribed in vitro using T7 RNA polymerase (Ambion) and purified with RNeasy reagents (Qiagen). Excess amount of dsRNA was incubated with recombinant human Dicer (Stratagene or Ambion) for 16 h at 37°C. Cleavage products were purified by passing through G-25 (Amersham Biosciences) and then YM-100 (Millipore) columns to remove salts and undigested dsRNA. Purified products were analyzed through urea-PAGE and stained with ethidium bromide. The purity of recombinant human DICER was verified by SDS-PAGE, and a single discrete band of Ͼ200 kDa was seen.
Northern Blot Analysis-RNA was isolated from cultured cells with mirVana miRNA isolation kit (Ambion). RNA on denaturing polyacrylamide gel was transferred onto ZetaProbe membrane (Bio-Rad) using a semi-dry Transblot apparatus (Hoefer). RNA on the membrane was cross-linked and then hybridized in ULTRAhyb-Oligo solution (Ambion) with 32 Plabeled RNA oligonucleotides that can hybridize to siRL.
Luciferase Assay-Renilla luciferase activity was determined as described (23,31) using the Dual-Luciferase reagents (Promega). Transfection efficiencies were normalized to a control plasmid expressing firefly luciferase.

TRBP and PACT Interact Directly with Each Other-Because
TRBP and PACT are related both structurally and functionally, we asked whether they bind to each other directly. We first performed GST pulldown experiments with recombinant proteins purified from E. coli. We found that a polyhistidine-tagged PACT (His-PACT) bound with high affinity to GST-TRBP but not to GST (Fig. 1A, lane 2 compared with lane 1 and lane 5 compared with lane 4). In agreement with the previous finding on homodimerization of PACT (22), we also observed the retention of His-PACT in the GST-PACT resin (Fig. 1A, lanes 3  and 6). To confirm the direct binding between TRBP and PACT, we employed another TRBP fusion protein MBP-TRBP. We noted that MBP-TRBP specifically bound to GST-PACT resin but not to GST alone (Fig. 1B, lane 4 compared with lane 2). In keeping with the concept of TRBP homodimerization (21), homophilic interaction between GST-TRBP and MBP-TRBP was also seen in this experiment (Fig. 1B, lane 6 compared with lane 2). Thus, our results from in vitro affinity binding assay consistently support that TRBP and PACT cannot only form homodimers but also interact directly with each other.
TRBP and PACT contain three dsRBDs ( Fig. 1C) that have differential activities in regulating PKR (17,18) and in the formation of homodimers (21,22). To determine the roles of the three dsRBDs in mediating TRBP-PACT interaction, we constructed two sets of truncated mutants and interrogated them for binding activities with TRBP or PACT using GST pulldown assay. We found that PACT-A containing the first dsRBD was capable of binding to GST-TRBP (Fig. 1C, lane 1), whereas the interaction with GST-PACT was mediated by TRBP-A and TRBP-B corresponding to the first and second dsRBDs, respectively (Fig. 1C, lanes 4 and 5).
Next we carried out co-immunoprecipitation experiment to further investigate the interaction of TRBP with PACT in human cells. Immunoprecipitation and Western blotting were performed reciprocally with antibodies recognizing Myctagged TRBP (Myc-TRBP) and V5-tagged PACT (V5-PACT). Because both Myc-TRBP and V5-PACT were found in the precipitates (Fig. 1D, lanes 4 and 8 compared with lanes 3 and 7, respectively), the two entities plausibly formed a protein complex inside human cells in culture. To rule out the possibility that TRBP and PACT associate as a result of lysate preparation, Myc-TRBP and V5-PACT were also expressed in separate cells. Precipitation was then performed immediately after mixing the lysates of different cells. Because similar results were also obtained in this setting (data not shown), the interaction of TRBP and PACT was unlikely an artifact generated during the preparation of cell lysates.
In a third approach, we used FRET imaging to analyze the interaction between TRBP and PACT. FRET imaging can detect the interaction between two proteins differentially fused with fluorescent tags. The FRET signal is emitted only when the two proteins are within a distance of less than 10 nm (32). In our experiment, both CFP-TRBP and YFP-PACT were found to localize in the cytoplasm (Fig. 2, panels 1 and 5). No FRET signal was measurable when CFP-TRBP alone or YFP-PACT alone was expressed (Fig. 2, panels 3 and 6). In contrast, when cells co-expressing CFP-TRBP and YFP-PACT were recorded simultaneously in the CFP, YFP, and FRET channels, a signifi-cant FRET signal was observed (Fig.  2, panel 9). These results lent further support to the notion that TRBP and PACT interact directly with each other within human cells.

Formation of TRBP-PACT-Dicer Complex in Vitro and in Cultured
Cells-The direct interaction between TRBP and PACT ( Fig. 1 and Fig. 2) as well as the association of TRBP/PACT with Dicer (7-12) prompted us to ask whether TRBP, PACT, and Dicer might function as a protein complex. In this regard, although the roles of TRBP and PACT in RNAi have been well described (10 -12), two important questions remain unanswered. First, because the association of PACT with Dicer has been demonstrated by co-fractionation and coimmunoprecipitation only (12), it is not known whether PACT directly interacts with Dicer or through TRBP. Second, it is not understood whether TRBP and PACT bind to Dicer simultaneously or exclusively. To address these questions, we performed in vitro protein affinity binding assays using bacterially produced TRBP, PACT, and Dicer ( In this experiment, GST-PACT was first saturated with an excess of MBP-TRBP, and the unbound MBP-TRBP was then removed by extensive washing (Fig. 3A, lane 2). Next, Dicer was added and found to be retained in the resin (Fig. 3A, lane 3). Likewise, when we saturated GST-TRBP resin with excess His-PACT, washed away all free His-PACT, and incubated the GST-TRBP-His-PACT-bound resin with Dicer, Dicer was also retained in the resin (Fig. 3A, lane 6 compared with lane 5). Reciprocally, MBP-TRBP or His-PACT was found in the resin if the GST-PACT or GST-TRBP resin was first saturated with Dicer (data not shown). Collectively, our results are consistent with the formation of a stable triple complex of TRBP, PACT, and Dicer.
To verify that a protein complex containing TRBP, PACT, and Dicer is also formed in cultured human cells, we performed reciprocal immunoprecipitation and Western blotting (Fig. 3B). Although both Dicer and V5-PACT were found in the anti-Myc precipitate that contained Myc-TRBP (Fig. 3B, lane 4 compared with lanes 1-3), both Dicer and Myc-TRBP were also detected in the precipitate containing V5-PACT (Fig. 3B, lane 8  compared with lanes 5-7). Thus, TRBP and PACT likely bound with Dicer as a functional complex.
We next compared the binding activities of truncated TRBP and PACT mutants (Fig. 3C). Although the first two dsRBDs of TRBP (TRBP-A and TRBP-B) were responsible for interaction with TRBP and PACT, the third dsRBD (TRBP-C) was accounted for binding with Dicer. This is generally consistent with previous findings (8,11). On the other hand, the first dsRBD of PACT (PACT-A) was necessary for binding to TRBP and PACT, whereas the second and third dsRBDs (PACT-B and PACT-C) were required for interaction with Dicer. Thus, TRBP and PACT used different domains to mediate their interaction with partner proteins. Interestingly, all truncated forms of TRBP and PACT were found to associate with Dicer in cultured cells, suggesting that some mutants incapable of binding with Dicer directly in vitro might interact with it indirectly through its TRBP or PACT partner. Moreover, when we pre-saturated recombinant Dicer with excess TRBP-C or PACT-C, other forms of TRBP or PACT was no longer able to bind with Dicer (data not shown), suggesting that TRBP and PACT likely compete for the same binding site(s) in Dicer.
Formation of TRBP-PACT-Dicer Complex in Mouse Testicular Tissue-Consistent with a previous report (33), we found that TRBP and PACT were abundantly expressed in male germ cells (data not shown). Thus, we investigated the in vivo formation of TRBP-PACT-Dicer complex in mouse testicular tissue. As a first step, the specificity of polyclonal antibodies against endogenous TRBP, PACT, and Dicer, which we prepared or purchased, was verified with human and mouse tissues as well as transfected cells. All three antibodies reacted with target proteins in human and mouse tissues and cells in a highly specific manner (data not shown). We then fractionated extracts of mouse testes through a Superdex 200 gel filtration column and analyzed the fractions by Western blotting with anti-TRBP, anti-PAC, and anti-Dicer antibodies (Fig. 4A). TRBP, PACT, and Dicer were detected in fractions 40 -43, consistent with the presence of a TRBP-PACT-Dicer complex displaying a molecular mass of more than 440 kDa. The detection of multiple TRBP or PACT species in mammalian cells was consistent with previous reports (10, 11, 34). , and CFP-TRBP ϩ YFP-PACT (panels 7-9). In CFP channel, cells were excited at 820 nm, and emission was recorded at 435-485 nm (panels 1, 4, and 7). In YFP channel, cells were excited at 488 nm, and emission was recorded at 535-590 nm (panels 2, 5, and 8). When cells were excited at 820 nm, corrected FRET signal was simultaneously recorded at 535-590 nm (panels 3, 6, and 9). Results were representative of 100 transfected cells. Bar, 20 m. . Target proteins are highlighted with "*", "#," and "‹". Arrows point to the heavy chain of immunoglobulins. Similar results were obtained when Myc-T, V5-P, and T7-Dicer proteins were expressed in separate cells, and the lysates of different cells were mixed immediately before immunoprecipitation was carried out (data not shown). C, summary of protein-protein interactions. All interactions were based on in vitro GST pulldown assay except that the binding to Dicer in cells was defined by co-immunoprecipitation of proteins from extracts of transfected HeLa cells.
In keeping with the co-fractionation results, we also detected PACT and Dicer in the protein complex precipitated with anti-TRBP (Fig. 4B, lanes 3 and 7). Thus, our results suggested that endogenous TRBP, PACT, and Dicer proteins associate to form a triple complex in mouse testicular tissue.
Roles of TRBP and PACT in Dicer-dependent Production of siRNA-The formation of a stable TRBP-PACT-Dicer complex led us to explore the influence of TRBP and PACT on Dicer function. In previous studies, comparison of the siRNA/ miRNA-producing activities of TRBP-Dicer or PACT-Dicer with Dicer alone suggests that neither TRBP nor PACT is required for the RNA cleavage reaction catalyzed by Dicer (10,12). However, a Drosophila homolog of TRBP and PACT named Loquacious can stimulate the specific pre-miRNA processing activity of Dicer-1 (8,35). Moreover, a dsRBP partner of Caenorhabditis elegans Dicer termed RDE-4 also functions in concert with Dicer at the step of siRNA production (36). To clarify whether human TRBP, PACT, or TRBP-PACT complex could facilitate RNA processing by Dicer, we monitored the cleavage of dsRL566, a 566-bp-long dsRNA corresponding to a fragment of Renilla luciferase mRNA. In reactions containing the same amount of recombinant Dicer and escalating amounts of GST-TRBP, His-PACT, or GST-TRBP plus His-PACT, a significant increase in the production of siRNA could be appreciated (Fig. 5). Notably, the increase in siRNA yield attributed to GST-TRBP plus His-PACT is comparable with the elevation ascribed to the same amount of GST-TRBP or His-PACT (Fig. 5, lanes 10 -12  compared with lanes 4 -6 and 7-9).
These results indicate the facilitation of Dicer-mediated production of siRNA by TRBP and PACT. In addition, a protein complex consisting of TRBP and PACT is at least equally active in facilitating the production of siRNA by Dicer when compared with TRBP or PACT alone.
To assess the roles of endogenous TRBP and PACT in RNA silencing in human cells, we knocked down the expression of TRBP or PACT effectively and specifically using short hairpin RNAs (shRNAs) in HEK293T cells (Fig. 6, A and B). We them performed Northern blot analysis to confirm that the production of siRNAs corresponding to the Renilla luciferase mRNA (siRL) was significantly diminished in TRBPor PACT-depleted cells (Fig. 6C,  lanes 3 and 4 compared with lane 2). Notably, the combined effect of shTRBP and shPACT was even more pronounced than that of shTRBP or shPACT alone (Fig. 6C,    3 and 4). Consistent with these results and with previous reports (10 -12), depletion of TRBP or PACT by shRNAs also led to substantial inhibition of the gene-silencing activity of an shRNA targeting Renilla luciferase mRNA (shRL), which has been shown (23) to be highly effective and specific in blocking the expression of Renilla luciferase (Fig. 6D, columns 3-5 compared with columns 1  and 2, and columns 8 -10 compared with columns 6 and 7). The observed effect was unlikely because of nonspecific saturation of the RNAi machinery by shRNA, because shGFP at the highest dose had no influence on shRL activity (Fig. 6D,  column 16). The effectiveness of this shGFP in silencing GFP expression has been shown previously (23), and its abundant expression in HEK293T cells was confirmed by Western blotting (data not shown). Again, the combined effect of shTRBP and shPACT was greater than that of shTRBP or shPACT alone (Fig. 6D, columns 13-15 compared with columns 3-5 and 8 -10). This was most evident at the lowest dose of shRNAs. In fact, the inhibitory activity of shTRBP ϩ shPACT at the lowest dose was significantly higher than that of shTRBP alone or shPACT alone at the same dose (Fig. 6D, columns with # and *; p Ͻ 0.005 by t test). Thus, simultaneous knockdown of TRBP and PACT in cultured human cells had a significant impact on RNAi. In other words, TRBP and PACT likely cooperate with each other in facilitating the execution of RNA silencing. To address the concern of other nonspecific off-target effects, we performed additional control experiments by adding back TRBP or PACT. Because enforced re-expression of TRBP was able to reverse the effect induced by shTRBP at the highest dose (Fig. 6D, column 17 compared with column 5), the inhibition of shRL-induced silencing by shTRBP was specific. Likewise, reversal of shPACT-induced inhibition by PACT (Fig. 6D, column 18 compared with column 10) verified the specificity of effect.
In contrast to the above results obtained with shRL, depletion of endogenous TRBP and/or PACT in HEK293T by shTRBP/shPACT had no influence on the silencing of Renilla luciferase expression induced by siRL, a synthetic siRNA targeting the same sequence on Renilla luciferase mRNA as shRL (Fig. 6E). As a positive control, depletion of Dicer by siRNA was effective in rescuing the siRL activity significantly (Fig.  6E, column 16), as demonstrated previously by others (26). Our results are surprising because TRBP and PACT have been shown previously to be required for siRNA-mediated silencing (10,12). Nevertheless, by taking into account the effects of shTRBP/shPACT on shRL (Fig. 6D) and siRL (Fig.  6E), we argued that TRBP and PACT function primarily at the step of siRNA production.

DISCUSSION
In this study, we established the direct interaction between human dsRBPs TRBP and PACT by using in vitro affinity binding assay (Fig. 1). This interaction was further verified by coimmunoprecipitation and FRET imaging ( Fig. 1 and Fig. 2). In addition, we demonstrated the direct interaction of PACT with Dicer and the formation of a TRBP-PACT-Dicer complex in vitro, in cultured human cells, and in mouse testicular tissue ( Fig. 3 and Fig. 4). Finally, we provided the first evidence that human Dicer associated with TRBP and PACT is more active than Dicer alone in processing long dsRNA into siRNAs (Fig. 5). Accordingly, depletion of endogenous TRBP and PACT resulted in an inhibition of shRNA-induced gene silencing primarily at the step of siRNA production (Fig. 6). Our findings suggest a new model for the assembly and function of human Dicer complex in which TRBP and PACT directly interact with each other and bind simultaneously to Dicer to facilitate the production of siRNA (Fig. 7).
Although more than one Dicer have been found in Drosophila and plants (1,37,38), there is one single Dicer in humans, and it is distinct from other members in the family of Dicer-like proteins by associating with two dsRBPs, TRBP and PACT (10 -12). Our demonstration of the direct interaction between TRBP and PACT suggests another level of complexity in the regulation of RNAi in human cells. This is the first example of an interaction between two Dicer-associated dsRBPs. TRBP and PACT are two members in the family of dsRBPs that have diverse cellular functions (39). Several other dsRBPs have also been known to interact with Dicer-like proteins or with other RNase III endonucleases such as Drosha (5-8, 35, 40). Thus, it will be of interest to determine whether any of these dsRBPs might be binding partners. We demonstrated that TRBP and PACT use different domains to interact with each other and with Dicer (Fig. 3C). Notably, all truncated mutants of TRBP and PACT can directly or indirectly bind with Dicer within cultured cells. Because there exist multiple naturally occurring isoforms of TRBP and PACT that contain only the first one or two dsRBDs (8,11), further characterization of these isoforms for influence on Dicer function is warranted. In addition, because TRBP and PACT use the same dsRBD(s) to mediate homodimerization and interaction with each other (Fig. 3C), it is possible that TRBP and PACT function as a stable heterodimer in vitro and in vivo. Dicer has been shown to form an intramolecular pseudodimer (41). Thus, additional biochemical and biophysical analyses are required to elucidate the protein stoichiometry and structure of the TRBP-PACT-Dicer complex. In this regard, it will also be of importance to compare the specific activities of TRBP and PACT homo-and heterodimers in terms of dsRNA binding, interaction with Dicer, and modulation of PKR.
Both TRBP and PACT have been shown to interact with PKR and to regulate its activity (14 -20). Moreover, their interaction with PKR has functional implications in inflammation, stress response, viral infection, and oncogenesis (20,34,(42)(43)(44). Although TRBP and PACT directly interact with each other (Fig. 1), it remains to be understood whether they associate in vivo with PKR as a TRBP-PACT heterodimer and how a TRBP-PACT complex might influence PKR activity. TRBP and/or PACT are also critically involved in human immunodeficiency virus biology (13,21,43,45), transcriptional gene silencing (46), translational control (33,44), and ear development (47). Plausibly, the interaction between TRBP and PACT may provide a platform or a common regulatory point for multiple pathways, including RNA silencing and PKR signaling.
Although TRBP and PACT have been shown to be necessary for RNA silencing in mammalian cells (10 -12), their exact roles in Dicer and RISC function are obscure. Although one report has suggested that depletion of TRBP by siRNAs reduces the level of mature miRNA (10), in another study this reduction has not been observed, and TRBP depletion has been demonstrated to affect pre-miRNA processing in vitro (11). We showed that recombinant Dicer associated with TRBP and/or PACT is more active than Dicer alone in cleaving dsRNA into siRNAs (Fig. 5). Consistent with these results, we observed that knockdown of TRBP and PACT suppressed siRNA production and shRNAmediated gene silencing in cultured cells, but it had no influence on siRNA-induced silencing (Fig. 6). Although we do not exclude the possibilities that TRBP and PACT are required for other aspects of Dicer function and RISC assembly as shown previously by others (10 -12), our data suggest that they function primarily in facilitating the production of siRNAs. Although the knockdown of Dicer could effectively rescue the siRNA-induced silencing effect in our experiment (Fig. 6), at this point we still do not understand fully whether the use of a highly effective siRNA and a different cell line could possibly account for the discrepancies between our results and those from other groups (10 -12, 48, 49). However, our finding that purified TRBP and PACT proteins facilitate the cleavage of dsRNA by Dicer (Fig. 5) is generally consistent with the previous observations that depletion of TRBP inhibits pre-miRNA processing in vitro (11) and that Loquacious in Drosophila stimulates pre-miRNA processing by Dicer-1 (7,35). This function of TRBP and PACT is also reminiscent of the role of Dicerassociated RDE-4 in C. elegans (36,50). We are currently in the process of comparing the pre-miRNA-cleaving activities of Dicer and TRBP-PACT-Dicer. This and other experiments will help clarify the roles of TRBP and PACT in siRNA and miRNA production.
Our in vitro Dicer cleavage experiments did not show cooperation between TRBP and PACT (Fig. 5), whereas the knockdown experiments indicated a synergistic effect (Fig.  6). One possibility is that other proteins or factors in the cell could be required for the cooperation of TRBP with PACT. Alternatively, the protein concentrations used in the reconstituted cleavage reaction might not be optimal for the cooperation to occur. Further experiments are required to clarify the discrepancy.