We assayed Tat interaction with Dicer by transfecting 293T cells with myc-tagged Dicer (pDicer-myc) in the absence or presence of flag-tagged Tat (pTat-flag) (Figure 1). Cell extracts were immunoprecipitated with anti-myc beads, and analyzed by Western blotting. As shown in figure 1, Tat co-immunoprecipitated (co-IP) with Dicer (lane 2). To assess better Tat/Dicer interaction, we conducted the co-IP using two Tat point-mutants. The TatK51A mutant previously was found to have little suppressive effect on Dicer activity while being proficient for viral transactivation; the TatK41A mutant did moderate Dicer activity while being deficient in Tat's transcriptional transactivation activity.

We expressed Tat, TatK51A and TatK41A comparably (Figure 1A, lower panel), and we also expressed myc-Dicer equally in each of the transfections (Figure 1A, upper panel). When Dicer was immunoprecipitated, we found that the recovery of the various Tat proteins was different. Tat K41A and Tat co-immunoprecipitated similarly with Dicer (Figure 1A, lanes 2 and 4); however, Tat K51A reproducibly co-immunoprecipitated less effectively (Figure 1A, lane 3). These results suggest that the association between Tat and Dicer as assayed by co-IP correlates with the ability of the former to moderate the activity of the latter.

Because Tat and Dicer are both RNA-binding proteins, we wondered next if their interaction required RNA. To address this question, lysates from cells transfected with myc-Dicer and Tat proteins were divided into two groups prior to immunoprecipitation. One group was treated with RNase A while the other group was not (Figure 1B). Without RNase treatment (Figure 1B, lanes 1 to 4), Tat and TatK41A interacted well with Dicer while TatK51A did less well; however, after RNase treatment, none of the Tat proteins was able to co-immunoprecipitate with Dicer (Figure 1B, lanes 6, 7 and 8). As a control, the amounts of the Tat proteins in the lysates were verified to be unchanged after RNase treatment (Figure 1B, right lower panel). Furthermore, TRBP, whose interaction with Dicer is RNA independent 23, co-immunoprecipitated with Dicer comparably regardless of RNase treatment (compare anti-TRBP, Figure 1B left to right). Hence, Tat and TRBP interact differently with Dicer; the former requires RNA while the latter does not. It remains not known whether a specific form of RNA (i.e. pre-miRNA) or general cellular RNAs suffice to mediate Dicer and Tat interaction. This requirement needs to be investigated further.

We next characterized the region in Dicer needed for Tat interaction. Co-immunoprecipation assays were performed using flag-tagged Dicer mutants deleted progressively from the N-terminus to encompass the DEAD domain (ΔDEAD), the Helicase domain (ΔHelicase), the Domain of Unknown Function 283 (ΔDUF), and the PAZ domain (ΔPAZ) (Figure 2A, B) 23. Each of the mutants expressed well after co-transfection with Tat into 293T cells, and all were immunoprecipitated equivalently using anti-flag beads (IP: anti-flag; lanes 1–6, top panel, Figure 2A). By contrast, when co-immunoprecipitation of Tat was assessed, only wt Dicer and ΔDEAD Dicer mutant (IB: anti-Tat; Figure 2A, middle panel; lanes 2 and 3), but not ΔHelicase, ΔDUF nor ΔPAZ mutants (Figure 2A, lanes 4, 5, 6), associated with Tat. These results suggest that removal of Dicer's helicase domain abolished its ability to co-immunoprecipitate Tat.

We performed two controls for the above experiment. First, we checked that Tat was equally expressed in the lysates of all the transfections (Figure 2A, lanes 8 – 12). Second, we verified that Dicer co-IP'd Ago2. Dicer- Ago2 interaction is dependent on Dicer's RNase III domain located in its C-terminus 24; and in our experiments, Ago2 co-immunoprecipitated wt Dicer and all the Dicer RNase III domain-containing mutants (Figure 2A; bottom panel, lanes 1 – 6).

We noted with interest that while the interaction of Tat and Dicer is RNA dependent (Figure 1), the presence of Dicer's C-terminal dsRNA binding domain in the above Dicer mutants was insufficient for Dicer to co-immunoprecipitate Tat. Intriguingly, Dicer's helicase domain was previously found to be required to interact with both TRBP and PACT 23. One interpretation of the collective results is that rather than a simple protein-RNA-protein bridging interaction, there are additional protein-protein contact points between Tat and the helicase region of Dicer which specifies association inside a cell. That Tat, TRBP and PACT all impinge at Dicer's helicase region raises a possibility that these factors may interfere and compete with each other functionally for limiting contact at this locale. Potential competition between Tat and TRBP or Tat and PACT, two key components of miRNA pathway, remains to be further characterized. While under our current experimental conditions no decrease in TRBP recovery was observed after Tat co-IP with Dicer (Figure 1B), whether more notable competition could be seen upon escalated titration of Tat expression remains to be evaluated.

We next characterized the region in Tat required to association with Dicer. We performed GST-pull down assays since we had access to a large number of GST-Tat deletion mutants and because our immunoprecipitation of Tat proteins was uneven with differently deleted Tat mutants. Using GST-Tat mutants that included Tat's transactivation domain (Tat 1–45), or Tat's basic region (Tat 1–60), or GST-Tat mutants that were deleted in their transactivation domain but retained their middle regions (Tat 20–72, Tat 30–72; Figure 3B), we assessed pull-down of Dicer using purified GST-Tat and four GST-Tat-deletion mutants. Control GST did not capture Dicer, while GST-Tat (Figure 3A, lane 1), GST-Tat 1–60, GST-Tat 20–72 and GST-Tat 30–72 did pull down Dicer (Figure 3A, lanes 4–6). Interestingly, GST-Tat 1–45 did not pull-down Dicer. These results agree with previous findings that the trans-activation domain of Tat does not account for physical and function interplay with Dicer 25.

Here we have characterized some of the requirements for Tat-Dicer physical association. We found that Tat-Dicer interaction requires RNA, although simple protein-protein bridging by RNA does not seem to be a sufficient explanation. Dicer-Tat interaction also requires Dicer's helicase domain and a portion of Tat's 30–72 amino acids. Whether the latter requirements imply direct protein-protein contact remains to be established.