In order to begin to understand how the misregulation of gene expression is caused by MLL-fusion proteins, we expressed some of the most common MLL fusion proteins, MLL-ELL1, MLL-ENL, MLL-AFF1 and MLL-AF9 in 293 cells with a Flag epitope tag under an inducible promoter, each integrated at the same site within the genome. Expression and purification of the MLL-N-terminal region most frequently found in MLL-fusion proteins, resulted in the isolation of Menin, which is known to associate with the N-terminus of MLL and LEDGF, an interactor of Menin (Supplemental Table 1
, Yokoyama and Cleary, 2008
). Following the biochemical isolation of MLL-ELL1, MLL-AFF1, MLL-AF9 and MLL-ENL (), these purified complexes were subjected to Multidimensional Protein Identification Technology (MudPIT) to carry out proteomic analyses for each complex. While the MLL-AF9 and MLL-ENL identified a few of the proteins previously described as ENL associated proteins, including AFF1, AFF4, and Dot1(Mueller et al., 2007
; Mueller et al., 2009
), the MLL-ELL1 and MLL-AFF1 chimera complexes included AFF4, but notably not Dot1 (Supplemental Table S1
). In fact, AFF4, which itself is a rare translocation partner of MLL, is a shared subunit for all of the purified MLL-chimeras (Supplementary Table 1
AFF4 is a shared subunit of several of the MLL-chimeras and associates with known RNA polymerase II elongation factors
Given the unexpected observation of finding this largely uncharacterized protein associating with the MLL chimeras, we therefore generated a cell line expressing Flag epitope-tagged AFF4 and identified associated proteins (). Surprisingly, all three ELL proteins were found in the isolated complexes. In turn, expressing and isolating Flag-tagged ELL1, ELL2 and ELL3 revealed AFF4 associating with each ELL (). Furthermore, the ELL and AFF4-containing complexes also consist of additional MLL partners, AFF1, ENL, and AF9 (). Another subunit of the AFF4 – ELL1 complex is the component of the Pol II C-terminal domain (CTD) kinase, P-TEFb, consisting of Cdk9 with cyclin T1, T2a or T2b (). We also detected the previously identified ELL-associated factors EAFs (Simone et al., 2003
; Simone et al., 2001
) in the AFF4, ELL2 and ELL3 complexes (). Since the purification of the ELL1 complexes were performed with ELL1 lacking the first 50 amino acids (missing in the MLL-ELL1 chimera and required for interactions with EAFs), we did not detect any of the EAFs in the ELL1 purifications. Given the fact that EAFs enhance the in vitro
transcription elongation properties of ELLs (Kong et al., 2005
), it is interesting that we also observe these factors with the AFF4-containing complexes.
The observed interactions between ELL1–3, AFF4, and the components of P-TEFb were also confirmed by Flag and endogenous co-immunoprecipitations (). In different preparations and with different tagged subunits, the relative amounts of some subunits in the isolated complexes can vary (e.g. see ELL1 and AFF4 levels in ). Therefore, it is important to tag and purify multiple subunits to get a clear picture of the complexes in vivo. Indeed, previous interpretations that Dot1, AFF1 and AFF4 exist in a single complex were primarily based on these proteins co-purifying with a single subunit, ENL (Mueller et al., 2007
; Mueller et al., 2009
). We also find that ENL associates with AFF1, AFF4 and Dot1, but importantly, we find that Dot1 is not associated with AFF1, AFF4 or the ELL complexes indicating that ENL is part of at least two distinct complexes (Supplemental Table 1
and data not shown).
To further characterize the AFF4-containing complexes, we analyzed nuclear extracts by their application to size exclusion chromatography, followed by SDS-PAGE and Western analysis of AFF4 and the components of the P-TEFb elongation complex (). These studies clearly indicate that a small portion of the P-TEFb co-purifies with the AFF4 large complex at about 1.5 MDa (, fraction 11–13), which we call the super elongation complex (SEC) due to the presence of multiple Pol II elongation factors. Overall, these studies reveal that many of the MLL partners found in leukemia, which have very little sequence or seemingly functional similarities, are found in large macromolecular complexes associated with the Pol II elongation factors ELL and P-TEFb.
P-TEFb is a CTD kinase involved in the regulation of transcription elongation by Pol II and can exist in both active and inactive forms (Peterlin and Price, 2006
). To determine whether the purified ELL and AFF4-containing complexes contain active P-TEFb, we tested the kinase activity of these purified complexes towards the GST-Pol II C-terminal domain fusion protein (GST-CTD). The ELL2, ELL3, and AFF4 complexes were assayed in the presence and/or absence of ATP and the GST-CTD (). The resulting products were subjected to SDS-PAGE followed by Western analysis with antibodies specific to Pol II CTD either phosphorylated on serine 2 (pS2) or serine 5 (pS5) (). We also tested the autophosphorylation of CDK9 and the possible phosphorylation of AFF4 by P-TEFb (). From these studies, it appears that the purified ELL and AFF4-containing complexes are active as a Pol II CTD kinase. Our studies also suggest that AFF4, which sequence alignment demonstrates similarly bears repeated SP motifs (), is also phosphorylated by P-TEFb (), a phenomenon also observed previously for AFF1 (Bitoun et al., 2007
). Similar CTD kinase activities are found in ELL1 and MLL-fusion protein complexes (Supplemental Figure S1
AFF4 is required for the assembly and the enzymatic activity of SEC containing ELLs, P-TEFb and MLL partners
To determine which of the components of the AFF4 complex are required for complex stability and association with P-TEFb kinase, we reduced the levels of several components of the complex using RNAi (). We observed that the reduction of the AFF4 homologue AFF1 does not alter ELL1 and P-TEFb stability in these cells (, Supplementary Figure S2
). However, the loss of AFF4 results in the instability of ELL1 with no significant effect on the stability of the P-TEFb components (). Our studies so far indicate that the AFF4-containing complex associates with a small portion of the P-TEFb components either when the AFF4 complex is purified (), or when nuclear extracts were analyzed by size exclusion chromatography (). We, therefore, tested for the association of P-TEFb with the AFF4/ELL complex in the absence of AFF4 (). Nuclear extracts from cells treated with AFF4 RNAi were subjected to size exclusion chromatography and the fractions were analyzed by SDS-PAGE, followed by Western analyses using antibodies specific to Pol II, CDK9, Cyclin T1 and Hexim 1 (). This biochemical analysis demonstrated that reduction in AFF4 levels result in the loss of association of Cdk9 and Cyclin T1 with the large AFF4-containing complex ( fractions 11–13), but not the Hexim1-containing P-TEFb complexes ( fractions 15–17, see Peterlin and Price, 2006
for a review of known P-TEFb complexes). Together, these results demonstrate that AFF4 is a central component of the P-TEFb/ELL complexes, which we call SEC.
The first in vivo characterization of ELL as a transcription elongation factor was in Drosophila, where there is only one ELL-like protein, dELL (Eissenberg et al., 2002
; Gerber et al., 2001
). Indeed, the first hint of a connection between P-TEFb was the RNAi-mediated knockdown of Cdk9 and the loss of dELL from chromatin (Eissenberg et al., 2007
). Additionally, although it took years to identify the affected genes, dELL and the sole Drosophila homologue of AFF4 (see Supplemental Figure S3
) were part of a small set of genes isolated in a screen for Ras signaling components (Eissenberg et al., 2002
; Neufeld et al., 1998
; Su et al., 2001
; Tang et al., 2001
; Wittwer et al., 2001
). Based on our new findings, we were interested to extend these intriguing links between ELL, P-TEFb and AFF4 in Drosophila.
Since many Drosophila Pol II elongation factors have been shown to associate with elongating Pol II on chromatin and relocalize to heat shock loci upon stress (Ardehali et al., 2009
; Eissenberg et al., 2002
; Gerber et al., 2001
; Gerber et al., 2005y
; Gerber et al., 2005b
; Smith et al., 2008
; Tenney et al., 2006
), we generated polyclonal sera to dAFF4 and performed colocalization studies of dAFF4 with dELL and the elongating form of Pol II (). dELL and the elongating form of Pol II colocalize extensively with dAFF4 (), not seen with preimmune sera (data not shown). The Hsp70
loci in Drosophila have been used as a model system for studying transcription elongation. The Hsp70
loci contain poised polymerase, which upon heat shock is phosphorylated at serine 2 in the CTD repeats by P-TEFb, allowing productive transcription (Lis et al., 2000
). We assayed the presence of dAFF4 at Hsp70
after heat shock and observe that indeed dAFF4 colocalizes with dELL and the elongating form of Pol II on polytene chromosomes at major heat shock loci, including the Hsp70
genes at 87A and 87C (, Supplemental Figure S4
). Chromatin immunoprecipitation of dAFF4 shows that it becomes associated with Hsp70 upon heat shock, and is present throughout the transcribed unit (), similar to what was previously observed for P-TEFb (Boehm et al., 2003
The Drosophila ortholog of AFF4 colocalizes with ELL and the elongating form of Pol II on Drosophila polytene chromosomes
We next asked if AFF4 was similarly recruited to the human HSP70
gene. Using chromatin immunoprecipitation (ChIP), AFF4 levels were measured across the human HSP70
gene before and after heat shock in HeLa cells (, Supplemental Figure S5
). Upon heat shock, AFF4 is found at the HSP70
promoter and throughout the transcribed region along with RNA Pol II. Interestingly, ELL2 and ELL3 are also recruited to the 5′ end of HSP70
, but are not significantly enriched as far into the body of the gene as AFF4. This could reflect different sensitivities of the antibodies employed, or conceivably to differential usage of the elongation factors in this complex at distinct steps of transcription elongation. ELL1 antibodies did not work in any of our ChIP assays, which we believe reflects the relative higher quality of ELL2 and ELL3 antibodies, and not the differential usage of these factors on chromatin.
AFF4 is required for HSP70 induction and is recruited to MLL chimera target genes in leukemic cells
The effect of AFF4 recruitment to HSP70
was assessed by siRNA-mediated knockdown of AFF4. Knockdown of AFF4 leads to a defective heat shock response, showing reduced induction of HSP70
compared to control siRNA-treated cells (). While HSP70
is used as a model gene for studying transcription elongation, we recognize that other factors, not known to directly stimulate transcription elongation, also travel with the polymerase, such as components of the exosome (Andrulis et al., 2002
). However, based upon the proven in vitro stimulation of transcription elongation by ELL1, the requirement for AFF4 in the stability of the P-TEFb-AFF4-ELL complex, the association of AFF4 with HSP70
upon heat shock and its requirement for full expression of HSP70
, we propose that AFF4 is a central component of the SEC complex.
To begin to investigate the role of AFF4 as a common component of complexes formed by MLL chimeras, we assessed the recruitment of AFF4 to HOXA9
loci in the MV4–11 cell line from a patient with a MLL-AFF1 translocation. As with many MLL translocations, HOXA9
are up-regulated in these cells. Indeed, chromatin immunoprecipitation with antibodies corresponding to the C-terminal portion of AFF1, which is contained in this MLL chimera, shows recruitment to HOXA9 in the MV4–11 cells (), as well as in another MLL-AFF1 fusion cell line SEM, but not in an unrelated leukemia cell line, REH (Supplemental Figure S6
). Interestingly, AFF4 is also recruited to HOXA9
in the MV4–11 cells, despite the fact that it is the related AFF1 gene that is involved in the MLL translocation in these cells. The antibody to AFF4 was raised against an amino-terminal portion not found in MLL chimeras, ruling out cross-reaction with the related AFF1 protein that is part of the MLL chimera. To assess the functional significance of AFF4 recruitment to MLL-AFF1 target genes, we performed lentiviral delivery of shRNA to the MV4–11 cells. Significant reductions of Hoxa9 and Hoxa10 are observed upon knockdown of AFF4 in these cells (). Together, these findings lend support to our hypothesis that AFF4, a very rare translocation partner of MLL is nonetheless a component of many MLL-fusion protein complexes and participates in leukemogenesis. This finding is supported by our initial biochemical purifications of the MLL-chimeras () and now with the analysis of the MLL-target genes from cell lines generated from a MLL-translocation patient.
Previous studies have provided evidence for links among different MLL translocation partners. ENL, AF9 and AF10 have been linked to the histone methyltransferase Dot1; and it was suggested that a common mechanism of MLL-translocation-based leukemia was through H3K79 methylation by Dot1 (Bitoun et al., 2007
; Krivtsov et al., 2008
; Mueller et al., 2007
; Mueller et al., 2009
; Okada et al., 2005
). However, the most common translocation partner of MLL is AFF1, which our present studies show does not associate with Dot1. Other studies suggesting a physical interaction between Dot1 and AFF1 were based on the isolation of these two proteins in ENL immunoprecipitates and through building a network of 2-hybrid and other interactions. However, we have determined the proteins associated with some of the commonly occurring MLL chimeras found in infant leukemias. MLL-AFF1 does not physically associate with Dot1, so a role for Dot1 at genes upregulated in MLL-AFF1 leukemias may be subsequent to gene activation by this MLL chimera. We have recently shown that Dot1-mediated H3K79 methylation is linked to cell cycle control in yeast (Schulze et al., 2009
) and methylation by Dot1 could also have some role in transcriptional enhancement in leukemogenesis. In contrast, MLL-AFF1 co-purified the SEC complex containing ELL1 and P-TEFb, two proven transcription elongation factors in vitro and in vivo, each with demonstrated abilities to activate transcription through transcription elongation. It will be interesting and essential to further determine the mechanisms and regulation of the SEC complex in transcriptional control in normal development and leukemogenesis. Both AFF1 and AFF4 copurify with the ELL proteins and another AFF protein, AFF3, which is also a rare translocation partner with MLL (von Bergh et al., 2002
). The related AFF2 gene (FMR2) is silenced in a form of mental retardation (Knight et al., 1993
), thus implicating all members of this family in human diseases (Bitoun and Davies, 2009
). P-TEFb itself is involved in a number of malignancies and developmental diseases (Romano and Giordano, 2008
), and it will be intriguing to determine which of these processes involve SEC or SEC-like complexes.
We originally proposed that ELL1, which we had identified as an activity from rat liver extracts that stimulated transcription elongation in vitro, and is homologous to the ELL
gene in humans involved in chromosomal translocations in leukemia, is linked to transcription elongation by Pol II to leukemogenesis (Shilatifard et al., 1996
). Recently, other groups have continued to propose a role for MLL-translocation partners with enhancing transcription elongation of the target genes of MLL chimeras (Bitoun et al., 2007
; Krivtsov et al., 2008
; Mueller et al., 2007
; Mueller et al., 2009
). These recent studies have focused primarily on a proposed Dot1-AFF1 interaction, which we now believe was the result of the isolation of ENL. Our evidence indicates that ENL participates in two distinct complexes, one with Dot1 and one within the SEC. An important area for future investigation is to define the relative contributions of these two types of complexes to leukemogenesis. This will require extensive conventional biochemical studies in hematopoietic cells bearing the relevant MLL chimeras.
In this study, we have demonstrated the following: 1) Many of the MLL fusion partners appear to interact with each other within a large macromolecular complex named SEC. 2) These large macromolecular complexes containing the MLL fusion partners also consist of several of the known Pol II elongation factors including ELL1, ELL2, ELL3 and the components of the P-TEFb kinase, Cdk9 and cyclin T1, T2a and T2b. 3) AFF4, which itself is a rare translocation partner of MLL, is a common factor associated with the purified MLL chimeras and Pol II elongation complexes. 4) AFF4 is required for the proper assembly and activity of large elongation complexes and association with P-TEFb. 5) AFF4 is a core component of the Pol II elongation complex containing ELLs and P-TEFb, associates with the elongating Pol II and relocalizes to heat shock puff sites upon stress, and is required for proper heat shock gene expression by Pol II. 6) We have demonstrated that AFF4 relocalizes to the MLL target genes HOXA9 and HOXA10 in cells bearing an MLL-AFF1 fusion, where it is required for the expression of these genes.
Collectively, the results of this study identify AFF4 as a component of the Pol II elongation complexes consisting of ELLs, P-TEFb and several of the common MLL fusion partners. These findings could prove critical for understanding the etiology of MLL-translocation based leukemias and for identifying additional targets for the treatment of the hematological malignancies resulting from these translocations, as well as for understanding fundamental aspects of transcription elongation control in development.