(A) Clonal cell lines expressing flag-tagged MLL-ELL1, MLL-ENL, MLL-AFF1 and MLL-AF9 were generated in 293 cells and the resulting protein complexes were purified using the FLAG-affinity purification method and analyzed by SDS-PAGE, silver staining and mass spectrometry. Arrows indicate the position of the Flag-tagged proteins. (B-C) Purification of Pol II elongation factors ELL1, ELL2, ELL3 and AFF4. Clonal cell lines expressing flag-tagged ELL1–3 were generated in 293 cells and the resulting protein complexes were purified and analyzed as in (A). Arrows in (B) indicate the position of the Flag-tagged subunit. (C) All three human ELL paralogs were identified in the Flag-AFF4 purification. ELL1 and its paralogs ELL2 and ELL3 were separately purified and demonstrated a similar set of associated proteins as found in the AFF4 purification. The Flag-ELL1 construct corresponded to the portion of ELL1 in MLL-ELL1 chimeras and lacks the N-terminal Eaf1/Eaf2 interaction domain, explaining the failure to identify these two proteins in this purification. AFF4 was found in all of the Flag-ELLs purifications indicating that it is a component of a novel RNA polymerase II elongation complex. (D–F) Confirmation of an interaction of AFF4 with the MLL-chimeras and components of the P-TEFb elongation factor by Flag and/or endogenous immunoprecipitations. Arrowheads show the position of the protein probed by Western analysis. (D) Flag immunoprecipitations of MLL-chimeras demonstrate an association of AFF4 with all chimeras, but not with a Flag-tagged MLL-N-terminal domain common to all chimeras. (E) Western blot analyses of ELL1, ELL2, ELL3, AF9, ENL and AFF4 immunoprecipitations confirm the observed interactions of Cyclin T1 and CDK9 with these factors. (F) Immunoprecipitations from 293 cells show the endogenous association of P-TEFb with AFF4 and ELL1. (G) Size exclusion chromatography of HeLa nuclear extracts demonstrated that AFF4-containing complexes elute at ~ 1.5 MDa (fractions 11–14, underlined in red) and contain a small portion of the CDK9 and CycT1-containing (P-TEFb) complexes from the nuclear extracts. Fractions corresponding to 1.5 MDa and 670 kDa are indicated with arrowheads. The 1.5 MDa complex is referred to as the super elongation complex (SEC).

(A and B) Polytene chromosome preparations from 3rd instar larval salivary glands were probed with antibodies to dELL (A, red) or the H5 monoclonal antibody recognizing the Ser-2 phosphorylated (P-Ser2), elongating form of Pol II (B, red). Both antibodies show substantial colocalization with dAFF4 (A and B, green). Chromosomes were counterstained with DAPI (A and B, blue). (C and D) Polytene chromosomes were prepared from heat shocked 3rd instar larvae and stained as in (A and B). Phase contrast images show positions of the 87A, 87C and 93D major heat shock loci. dAFF4 is recruited along with dELL at these loci after heat shock, associated with the P-Ser2 form of RNA Pol II. See Figure S4 for additional images. (E) Chromatin immunoprecipitation of dAFF4 and RNA Pol II large subunit (Rpb1) at Hsp70 before and after 10 minutes of heat shock at 37° C. While Rpb1 is present at Hsp70 prior to heat shock, dAFF4 can only be detected at Hsp70 after heat shock, where it is found throughout the transcription unit along with Pol II. Hsp70 primers have been previously described (Boehm et al., 2003). Error bars represent standard deviations.

(A) Pol II C-terminal domain (CTD) Kinase assays with the ELL2, ELL3 and AFF4-containing complexes were performed with GST-Pol II C-terminal domain fusion protein (GST-CTD). ELL2, ELL3, or AFF4 complexes were assayed in the presence of ATP and/or the GST-CTD and subjected to Western blot analyses with antibodies specific to Pol II CTD phosphoserine 2 (pS2) and phosphoserine 5 (pS5), CDK9 and AFF4. Consistent with previous observations, serine 2 and serine 5 of Pol II CTD are good substrates for the P-TEFb complexes in vitro. CDK9, itself, is also known to be autophosphorylated, resulting in a shift in gel migration in SDS- PAGE (indicated by asterisk, while an arrow indicates the faster migrating unphosphorylated form). AFF4 shows a similar gel mobility shift as CDK9, also indicated by an asterisk, suggesting that it is a substrate for P-TEFb as well. See Figure S1 for additional kinase assays. (B) Sequence alignment of a potential site for multiple phosphorylation of AFF4-related proteins bearing SP motifs favored by P-TEFb. (C–D) AFF4, and not AFF1, is required for stability of the SEC containing ELL1, P-TEFb and MLL-partners in HeLa cells. Western blot analysis of ELL1, CDK9 and Cyclin T1 was performed in the presence and absence of AFF1 or AFF4. Nuclear extracts from the siRNA-mediated for AFF4 or AFF1 were analyzed by SDS-PAGE and Western blot analysis. Arrows indicate increasing protein loads. Bulk protein levels of ELL1 are reduced in AFF4, but not AFF1 knockdown in these cells. Bulk protein levels of P-TEFb were not affected by AFF1 or AFF4 RNAi. Global H3K4 and H3K79 methylation levels were not affected by AFF4 knockdown. Tubulin serves as a loading control. (E) Gel filtration analyses of nuclear extracts from control and AFF4-directed siRNA treated cells. Larger P-TEFb-containing complexes, SEC (fractions 10–14 also seen in Figure 1G and indicated by underlining in red) are reduced in AFF4 knockdown cells, indicating that the presence of AFF4 is required for the assembly of SEC.

(A–G) HeLa cells were heat shocked by incubation at 42° C for 2 hours. Non-heat shocked and heat-shocked cells were used in chromatin immunoprecipitation-quantitative PCR (ChIP-qPCR) assays with AFF4, ELL2, ELL3, general Pol II and the H14 monoclonal antibody recognizing Serine 5 phosphorylated form of Pol II (B–F). (A) Position/primer pairs used for QPCR along the HSP70 gene are indicated. (B–F) AFF4 is recruited to the HSP70 gene after heat shock along with ELLs and RNA polymerase II. (G) Knockdown of AFF4 in HeLa cells by RNAi inhibits HSP70 induction. Control and AFF4 siRNA-treated cells were heat shocked as in (A) and HSP70 mRNA levels were assessed by quantitative RT-PCR and normalized to GAPDH mRNA levels. Non-heat shock control and AFF4 siRNA-treated cells are shown for comparison. (H-J) Recruitment of AFF4 to genes induced by the MLL-AFF1 chimera in MV4–11 cells. (H) Antibodies to the C-terminal domain of AFF1 found in the MLL chimera and antibodies raised against the N-terminal domain of AFF4 were used in ChIP-qPCR assays at HOXA9 and HOXA10 loci, known targets of the MLL-chimera found in human leukemia. As expected AFF1, shows recruitment to HOXA9 and HOXA10 in the leukemic MV4–11 cells. AFF4 is also recruited to HOXA9 and HOXA10, consistent with its co-purification with the MLL-AFF1 chimera in Figure 1A and Supplementary Table 1. Similar findings were observed with sera from two immunized rabbits (data not shown). The beta globin gene (Hemo), which is not expressed in HeLa or MV4–11 cells, is used as a negative control in (B–F and H). Note that experiments in Figure 4B and 4H were done with crude 4 week serum or affinity purified from 10 week serum, respectively and were normalized with different amounts of input DNA. See Figure S6 for additional ChIP experiments. (I) Knockdown of AFF4 in the leukemic MV4–11 cells by retroviral introduction of a shRNA targeting AFF4. (J) Reduction of HOXA9 and HOXA10 expression in MV4–11 cells after AFF4 knockdown. GFP shRNA is used as a non-targeting control shRNA. Expression levels were measured by quantitative RT-PCR and normalized to 18S rRNA. Error bars represent standard deviations.