The CTD of the largest subunit of Pol II serves as a structural binding platform for nuclear factors which, in return, positively or negatively influence the different steps of Pol II transcription and transcription-coupled RNA processing (Buratowski, 2009
). The timely recruitment of transcription/RNA processing factors during the transcription cycle is tied to dynamic phosphorylation and dephosphorylation of CTD repeats. Among the three main cotranscriptional phosphorylation events (Ser2/Ser5/Ser7), Ser7P is by far the most obscure. While Ser5P and Ser2P have been described to play key roles in promoter clearance/capping and splicing/elongation, respectively, Ser7P has no clear function in expression of protein-coding genes. So far, the only role for this modification is in expression of the subset of Pol II-transcribed genes encoding snRNAs (Egloff et al., 2007
). In this study, we have shown that RPAP2 interacts with the CTD phosphorylated at Ser7 alone, in vitro and in vivo. RPAP2 belongs to the RPAP family, a set of proteins occupying key positions in the protein interaction network of the human Pol II machinery (Jeronimo et al., 2007
). Our results indicate that RPAP2 specifically and directly interacts with the Pol II CTD phosphorylated at Ser7.
Importantly, siRNA-mediated knockdown of RPAP2 drastically affects expression of snRNA genes, reducing the transcription rate by Pol II and the efficiency of the 3′ end processing step. These defects are similar to the ones observed after mutation of Ser7 of the CTD (Egloff et al., 2007
), suggesting that the loss of RPAP2 may account for the defects observed when Ser7 is mutated. Accordingly, depletion of RPAP2 also causes a consequent loss of the Integrator complex from snRNA genes, as previously shown for Ser7 mutation (Egloff et al., 2007
). We confirmed the previously described interaction between RPAP2 and the Int4 and Int6 subunits of the Integrator complex (Jeronimo et al., 2007; Malovannaya et al., 2010
). Unexpectedly, the Int11 subunit, thought to carry the endonuclease activity responsible for the cleavage of the 3′ end of pre-snRNAs (Baillat et al., 2005
), does not stably associate with RPAP2. Importantly, Int11 binding requires Ser2P, which occurs later in the transcription cycle (Egloff et al., 2009
), in addition to Ser7P to efficiently interact with Pol II (Egloff et al., 2010
), whereas Int4 requires only Ser7P (Figures S4
A and S4B). Together, our data suggest that a subcomplex of Integrator, containing at least Int1, Int4, Int5, Int6, and Int7, and possibly Int3, since Jeronimo et al. identified an interaction between Int3 and RPAP2 (Jeronimo et al., 2007
), is loaded onto snRNA genes through RPAP2 and that the catalytic subunit is stably recruited later in the transcription cycle (D). This possibility is supported by the finding that Int11 is at its highest level nearest to the 3′ box (C and Figure S5
The finding that Ser7 is required for the recruitment of a Ser5P phosphatase also has important implications for Ser7 function during snRNA gene transcription, since it suggests that Ser7P tags Pol II for Ser5 dephosphorylation. Accordingly, the Ser5P level drops significantly downstream of the peak of Ser7P on snRNA genes (Egloff et al., 2009
). In yeast, abrogation of the two Ser5 phosphatase activities, Rtr1 and Ssu72, causes a transcription defect (Gibney et al., 2008; Mosley et al., 2009; Reyes-Reyes and Hampsey, 2007
). The defect in transcription of snRNA genes caused by RPAP2 knockdown may therefore be, at least in part, due to the failure to dephosphorylate Ser5. Ser5 hyperphosphorylation induced by the loss of Ser5 phosphatases has been proposed to either reduce the rate of transcription reinitiation by Pol II, or result in persistent association of the capping enzyme (Reyes-Reyes and Hampsey, 2007
), which has been reported to repress transcription (Myers et al., 2002
). In addition, early Ser5 dephosphorylation might be a prerequisite for further association of elongation/processing factors, as suggested by our previous observation that Ser5P has a negative impact on Int11 recruitment to the CTD (Egloff et al., 2010
). Thus, failure to dephosphorylate Ser5 may also contribute to the failure to efficiently 3′ process snRNA gene transcripts.
Based on these results, we propose a model in which recruitment of RPAP2 to snRNA genes through CTD Ser7 phosphorylation underpins a cascade of events critical for proper gene expression (D). Recruitment of RPAP2 results both in stable association of Integrator with snRNA genes and removal of Ser5P, generating the characteristic pattern of CTD phosphorylation found on actively transcribed snRNA genes.
Interestingly, this cascade appears to be specific for snRNA genes, as RPAP2 recruitment to a protein-coding gene is not affected by mutation of Ser7. RPAP2 recruitment to protein-coding genes must therefore involve either additional or entirely different interactions, perhaps reflecting the increased complexity of the transcription apparatus involved in expression of these generally more complex genes.
Finally, we identified the DUF408 domain of RPAP2 as a CTD-interacting domain (CID), raising the possibility that other DUF408-containing proteins may also recognize Ser7P of the Pol II CTD. So far, the DUF408 domain has only been reported to support the phosphatase activity of the yeast Rtr1 protein (Mosley et al., 2009
). Our results raise the possibility that the lack of phosphatase activity of the resulting protein could be due to its inability to properly associate with its substrate. Additional studies will therefore be required to precisely map the phosphatase domain and the CID of the human RPAP2.