Figure 1.
Core structure of the EAP complex.
Schematic representation of the EAP core and stabilizing protein interactions. The composition of the EAP core, protein interaction motifs, and the enzymatic reactions catalyzed by EAP are depicted. Please note the individual protein orientation with N- and C-termini labeled appropriately. For clarity, only one RNA Pol II repeat sequence (YSPTSPS) is shown. Solid lines mark protein–protein interaction domains. The exact N-terminal extent of the CyclinT2 binding site in Dot1l could not be delineated, and therefore, this boundary is represented by a dashed line. The interaction of histone H3 with the YEATS domain of ENL has been described previously [20]. The calculated molecular weights from the respective reference sequences are given for each protein. Numbers denote the respective amino acid residues of the individual interaction domains. Abbreviations are used as follows: AF4, ALL1 fused gene on Chromosome 4 (also known as MLLT1 or AFF1); ALF, AF4-LAF4-FMR2 homology region; CDK9, cyclin dependent kinase 9; CHD, C-terminal homology domain; cyc box, cyclin box; CYCT, Cyclin T2; H, histidine rich; Dot1l, disruptor of telomeric silencing like; ENL, eleven nineteen leukemia; H3, histone H3; H3K79met, Histone H3 Lysine 79 methyltransferase domain; hydp, hydrophobic region; L, leucine rich; KRM, lysine-arginine-rich motif; NHD, N-terminal homology domain; NLS, nuclear localization signal; P, proline rich region; Ser, serine rich domain; YEATS, Yaf9, ENL, AF9, Taf14, and Sas5 conserved domain.
Figure 2.
Elongation stimulation by EAP.
(A) Mechanism of RNA tethering assay. A luciferase reporter driven by a modified HIV LTR promoter is used to detect elongation stimulation. The Rev RNA recognition site (IIb stem-loop) is grafted onto the TAR Tat-recognition loop. Proteins tethered to RNA via Rev will release the stalled RNA Pol II and create luciferase activity only if they can recruit active pTEFb (dimer of cyclin T1 or T2 and CDK9). (B) Expression of Rev fusion proteins. ENL, AF4, AF5, and derivatives thereof were fused to Rev and expressed in 293T cells. Cell lysates were probed with an antibody specific for Rev. For detection of Rev-MLL-fusion proteins, an anti-MLL antibody was employed. (C) Results of RNA tethering assays with MLL fusion partners. Rev or Rev fusions as indicated were expressed together with the TAR-IIb reporter and luciferase activity was determined. Boxes inside the graphical representation correspond to the protein–protein interaction domains from Figure 1. Values represent average and standard deviation of triplicate experiments and are expressed relative to background with Rev alone. (D) Elongation activity of MLL fusion proteins. Chimeras of Rev with MLL fusion proteins as depicted were tested in RNA tethering assays as described for (C).
Figure 3.
Incorporation of MLL fusion proteins into EAP.
(A) Coimmunoprecipitation of MLL-ENL with EAP components. MLL-ENL (MLL-ENL), a MLL-ENL mutant lacking the last 15 amino acids of ENL (MLL-ENL1–544), or MLL without fusion partner (MLL) were expressed either alone or together with HA-tagged proteins AF4, AF5, or Dot1l. A schematic overview of the EAP core structure in the presence of MLL-ENL including the expected protein–protein interactions (double-headed arrows) is depicted in the upper-left panel. The presence of HA-tagged proteins and endogenous CDK9 in anti-MLL precipitates was probed alongside with a sample of the input (inp, 5%) by immunoblots as indicated. As a control, the successful precipitation of the MLL-ENL derivatives was confirmed by an anti-MLL blot. (B) Interaction of MLL-AF4 and MLL-AF5 with endogenous (endo) ENL and CDK9. MLL-AF4 (MLLAF4758–1210) and MLL-AF5 (MLLAF5731–1163) proteins analogous to patient-derived fusions were expressed in 293T cells. Note that the N-terminal CYCT interaction domain is missing in leukemogenic MLL-AF4/5 fusions as depicted in the upper-left and -right panels. MLL-AF4/5 derivatives deleting the ENL binding domain (MLLAF41023–1210, MLLAF5991–1163) served as controls. The coprecipitation of endogenous ENL and CDK9 was detected by immunoblot as indicated next to a sample of 5% input (inp). The presence of MLL-AF4/AF5 in the lysates was controlled as above.
Figure 4.
Association of MLL fusion proteins with EAP in leukemia cells.
(A) Coimmunoprecipitation (IP) of MLL-AF4 with endogenous ENL and CDK9 in patient-derived SEM cells. The MLL-AF4 fusion protein was precipitated by anti-MLL antibodies from extracts of the t(4;11) B-lymphocytic leukemia line SEM (upper panels). To control for precipitation with wild-type MLL, a B-ALL line of different etiology (REH) was included (lower panels). Precipitates were analyzed alongside a sample of input and probed for ENL and CDK9. Mock precipitations without antibody served as additional controls (contr). Note that MLL-AF4 corresponds in size to the N-terminal part of MLL that is produced after posttranslational cleavage of wild-type MLL. Therefore, MLL-AF4 and MLL-N comigrate as single band. In SEM as well as in REH, a longer splice variant [58] of CDK9 (labeled by an asterisk [*]) also was prominently detectable. (B) Chromatin immunoprecipitation and HOXA9 expression in SEM and REH cells. Left panel: ENL specific ChIP was performed across the human HOXA9 locus in SEM and REH cells. Precipitation efficiencies relative to nonenriched input samples were determined for five locations across the human HOXA9 region by quantitative PCR analysis of ENL bound chromatin in SEM (brown squares) and REH (grey diamonds). Values are plotted as relative enrichment with results from REH normalized to one unit. The experiment was performed three times with one typical example shown. Given are averages and standard deviations of triplicate PCR measurements. Right panel: HOXA9 expression levels were determined by qRT-PCR from SEM and REH total RNA, normalized to actin, and plotted with REH as calibrator representing one unit.
Figure 5.
Correlation of MLL-ENL binding and Hox gene expression.
Murine cell lines derived from primary hematopoietic cells by transformation with flag-MLL-ENL were subjected to a flag-specific ChIP reaction. Precipitates were amplified and hybridized (ChIP-chip) to a commercial promoter array that tiles 2 kb upstream and 0.5 kb downstream of known transcription start sites. In addition, the expression level of every Hoxa gene was determined by qRT-PCR with calibrated primers. Values were normalized to actin. The results are included as a heat map in the upper line (array set-up/expression level) of the figure. The heat map legend is shown below the figure. Each color-coded rectangle corresponds to 2.5 kb of genomic sequence represented on the promoter array. The second line shows known transcripts within the Hoxa locus. All Hoxa transcripts (green arrows drawn to scale) are transcribed in centromeric direction. The blue arrow denotes an annotated cDNA (5730596B20RIK-201 in Ensembl) transcribed oppositely to the Hox genes. The complete locus as shown encompasses 95 kb of genomic DNA. The third line represents a diagram of flag-MLL-ENL binding. The bar height indicates the log2 ratio of relative enrichment compared to a control precipitation. The bar color indicates a false discovery probability of <0.05 (red), <0.1 (orange), and <0.2 (yellow). The graph is a composite image compiled from two independent experiments. The genomic environment between Hoxa9 and the gene for microRNA196b is shown enlarged. Arrows represent known transcript start sites, and double arrows indicate the location of the primer pairs used for the kinetic analysis described in Figure 6. n.d. = not detected.
Figure 6.
Dynamic modification of the Hox locus by MLL fusion-recruited EAP.
(A) Morphology of cells transformed by a conditional derivative of MLL-ENL. Cytospin samples of cells transformed by tamoxifen-inducible MLL-ENL (MLL-ENL-ER as described in [8]) were prepared in the presence of tamoxifen (0 days) (+TAM) and 2 wk after withdrawal of inductor (−TAM). Slides were stained with May-Grünwald-Giemsa, and microphotographs were taken at room temperature with a Nikon Digital Sight DS-Fi1 electronic camera attached to a Zeiss Axioskop with a Zeiss Neofluar 63×/1.25 objective and processed with CorelDraw software (Corel) without any image enhancements. (B) ChIP experiments. MLL-ENL-ER cells as in (A) were used to record the histone H3 methylation status at lysines 9 (dimethylation, black line), lysine 27 (di- and trimethylation, brown line), and lysine 79 (dimethylation, blue line) as well as RNA Pol II presence (Ser2 phosphorylated polymerase, magenta line). ChIP samples were analyzed for two loci within the first exon of Hoxa9 and upstream of the gene for microRNA196b. Samples were drawn at day 0 (active MLL-ENL) as well as 3, 7, 10 and 14 d after inactivation of MLL-ENL by inductor withdrawal. Precipitation efficiencies were normalized to an input sample, and values were plotted relative to the modification levels at day 0. For absolute comparison, H3K9 trimethylation and H3K79 dimethylation were also determined for a nontranscribed X-chromosomal satellite repeat sequence (indicated in upper-left panel). In addition, changes in Hoxa9 mRNA after MLL-ENL-ER shutdown were quantified by qRT-PCR and plotted alongside the ChIP results (red line). The chart shows average values and standard deviations of a triplicate qPCR evaluation, and represents one typical of three independent experiments.
Figure 7.
MLL-ENL-introduced chromatin modifications resist differentiation stimuli.
(A) Left panel: cells transformed by constitutive MLL-ENL. FACS analysis of gr-1 differentiation marker in cells grown in IL-3 and in G-CSF for 7 d. Right panel: cells transformed by conditional MLL-ENL-ER. Gr-1 levels before and 3 d after inactivation of MLL-ENL-ER by removal of tamoxifen (TAM). (B) Left panel: qRT-PCR analysis of Hoxa9 expression in cells transformed by constitutive MLL-ENL cultured in IL-3 or G-CSF for 7 d (light columns) and in cells transformed by conditional MLL-ENL-ER before and 3 d after tamoxifen removal (dark columns, labeled “on” and “off”). Right panel: comparison of H3K79 dimethylation status of the Hoxa9 locus in cells treated as before. Values are averages and standard deviations of a triplicate, and the experiment was done twice with comparable results.
Figure 8.
Sensitivity of MLL cells to CDK inhibition.
(A) Effect of flavopiridol. Top panel: seven MLL cell lines (red lines) and four control leukemia lines of different etiology (black lines) were treated with increasing concentrations of flavopiridol as indicated. Proliferation was photometrically assessed in triplicate samples by MTT reduction. For better comparison, all optical density at 550 nm (OD550) readings were normalized to yield one unit in the absence of inhibitor. Mean values are plotted semilogarithmically against flavopiridol concentrations. To avoid clutter, standard deviations are indicated only for drug concentrations where MLL and control cell lines diverge in response to treatment. (B). Effect of alsterpaullone. Alsterpaullone was tested like described for (A). (C) Effect of flavopiridol and alsterpaullone on primary hematopoietic cells. Triplicate samples of mouse bone marrow cells enriched in early precursors by 5-fluorouracil treatment of the donors were cultured in liquid medium supplemented with IL-3, GM-CSF, IL-6, SCF, and the indicated amount of flavopiridol or alsterpaullone (black line). Proliferation was measured by MTT reduction. For comparison, precursor cells were retrovirally transduced with MLL-ENL, and the resulting lines were tested again for drug sensitivity under identical culture conditions (red line). Given are mean values and standard deviations of triplicates.
Figure 9.
The molecular mechanism of transcriptional activation by MLL fusion proteins.
MLL fusions recruit EAP core components and cause H3K79 methylation as well as Pol II CTD phosphorylation at serine 2. As a consequence, target genes are continuously expressed, inhibitory H3K27 and H3K9 methylation does not take place, and hematopoietic precursors are blocked for differentiation. The amino acid sequence of a single repeat unit in the CTD is given in the one letter code.