The recruitment of Cdk9/CyclinK to target genes does not involve RNA molecules 108

Our P-TEFb-tethering assays have shown that artificial recruitment of CyclinK to promoter DNA was not sufficient to direct transcription, suggesting that additional co-factors are required to allow the interaction of Cdk9 with the CTD. To date, many studies have pointed to the role of a variety of DNA- and RNA-binding proteins that recruit P-TEFb to its targets, most of them having been identified in human cells (Table 1). At the transition from transcription initiation to elongation, the production of nascent RNA that moves along the polymerase CTD offers an opportunity for Cdk9/CyclinK to bind the RNA and phosphorylate the CTD. To test this possibility, we examined the binding of CyclinK on polytene chromosomes following ribonuclease (RNAse) treatment of the salivary glands. This technique has been successfully used to investigate the involvement of RNA molecules in the recruitment of chromatin factors or the binding of proteins to RNA molecules at the chromatin level (Richter et al., 1996 ; Perrin et al., 1999 ; Akhtar et al., 2000 ; Buscaino et al., 2003). In this experiment, nascent transcripts were degraded by incubating the salivary

Figure 34. P-TEFb tethering assays using GBD:Cdk9 and GBD:Cdk9^DN. (A) Schematic drawings of the transfected proteins. GBD:Cdk9 and GBD:Cdk9^DN were obtained by fusion of the GBD (GAL4-binding domain) at the 5’ end of the Myc-tagged wild-type and mutant cdk9 constructs. (B) Luciferase assays. Transfections with the indicated plasmids were done as depicted in Fig. 33. Transfections were repeated 3 times independently and normalized to the levels of the constitutively expressed renilla luciferase reporter. The graph indicates the levels of luciferase activities monitored in each assay relative to the basal level (lane 1), and the errors bars indicate the standard deviations.

glands with the single-stranded RNA-specific RNase A before fixation. As shown in figure 35, the binding of CyclinK to chromosomes after RNase treatment remained unchanged compared to control chromosomes, even under extensive digestion conditions. This observation does not support the hypothesis that Cdk9/CyclinK associates with the transcripts in early elongation. However, it is still possible that higher-order structures between P-TEFb and the RNA are involved, making the transcripts insensitive to RNase digestion. This question could be further investigated by comparing the binding pattern of CyclinK following treatment with other RNase enzymes.

4.4. Overexpression of CyclinT rescues the larval phenotypes of Cdk9^DN.

An important remaining question concerns the role of the two cyclins in P-TEFb function. We have shown that both CyclinT and CyclinK are essential in developing Drosophila and do not display redundant activities. Moreover, our staining of polytene chromosomes indicated that both P-TEFb complexes are recruited to the same targets including the heat shock genes following their activation. It is therefore possible that both complexes are involved in the regulation of these genes by controlling separated transcriptional processes. Alternatively, although they are recruited to similar chromosomal sites the two cyclins might regulate the transcription of different subsets of target genes to regulate distinct cellular processes such as cell growth or cell differentiation. We attempted to discriminate between these possibilities by comparing the phenotypes caused by the co-expression of Cdk9^DN with one or the other cyclin. If both complexes are involved in the regulation of the same targets and/or cellular processes, we would expect that the co-expression of mutant cdk9 with cyclinT or cyclinK would produce similar modifications (or absence of modifications) of the phenotypes caused by cdk9^DN alone. In contrast, if the complexes are involved in distinct cellular tasks, some phenotypes should be more pronounced or even partially rescued depending on the overexpressed cyclin subunit.

We started this analysis by comparing the effects of the co-expression of Cdk9^DN with CyclinT or CyclinK during larval growth using the ubiquitous da-GAL4 driver. We were surprised to see that, while expression of both CyclinK and Cdk9^DN caused strong larval lethality prior to the second stage like Cdk9^DN alone, the co-expression of CyclinT with Cdk9^DN almost completely rescued the larval growth defect until the end of the third instar.

This observation raises the possibility that CyclinK has a critical function during the larval growth phase, possibly by promoting the endoreplication of the larval tissues. Alternatively,

Figure 35. CyclinK binding to chromosomes does not require RNA molecules. (A-D) Polytene chromosomes squashes from control (A,B) and RNase A treated salivary glands (C,D) expressing Flag:cyclinK and Myc:cdk9 under control of the AB1 driver were stained with an anti-Flag antibody (red) and DAPI (blue). Flag:CyclinK binding to chromatin is not modified following RNase A digestion (30 minutes in 1 mg/ml RNase A) compared to the control. (A-D) : UASp-Myc:cdk9/+ ; UASp-Flag:cyclinK/AB1.

CyclinT may well be required for normal growth and its overexpression could change the ratio of active P-TEFb over inactive complexes so that the absolute amount of active P-TEFb surpassed a threshold needed for cell growth and endoreplication. To determine whether CyclinT overexpression indeed rescues the endoreplication defect caused by Cdk9^DN, we compared the size of salivary gland nuclei from larvae expressing either wild-type or mutant cdk9 alone or together with cyclinT or cyclinK using the AB1 driver. As shown in figure 36A, nuclei expressing cdk9^DN alone or together with cyclinK failed to grow (B,D) while nuclei co-expressing cyclinT and cdk9^DN endoreplicated normally (compare E-F). To ensure that the rescue of the cdk9^DN endoreplication defect by cyclinT was not due to lower expression of the transgenes, we co-expressed the tagged proteins and compared their levels of accumulation in salivary gland nuclei by immunostaining (Fig. 36B). We found that the staining for both Myc:Cdk9/HA:CyclinT (G-I) and Myc:Cdk9^DN/HA:CyclinT (J-L) were roughly equivalent, suggesting that inactive P-TEFb does not strongly interfere with the expression of the transgenes and that the GAL4 driver is not in limiting amounts. In addition, Myc:Cdk9^DN conserved its nuclear localization indicating that the overexpressed HA:CyclinT does not trap the kinase in the cytoplasm. Taken together, these data may provide evidence for distinct cellular functions of P-TEFb depending on its association with one or the other cyclin.

4.5. Overexpression of CyclinT in imaginal discs partially rescues the Cdk9^DN