A) Schematic diagram of the full length MLL protein with key domains indicated. The Flag/HA tagged CxxC-RD2 MLL fragment used for immunoprecipitation is shown below. The first and last amino acids of the protein fragments are indicated. B) MigR1 and Flag/HA-tagged CxxC-RD2 was expressed in 293 cells and immunoprecipitated. Immunoprecipitates were analyzed by SDS-PAGE and visualized by silver staining. Arrows indicate bands at the predicted molecular weights of the PAF complex components. C) Immunoprecipitation and western blot of the PAF complex following immunoprecipitation of Myc tagged CxxC-RD2 or Myc tag control from transiently tranfected 293 cells. D) The experiment described in C was repeated after treatment of the lysate with benzonase to digest DNA indicating the PAFc interaction with CxxC-RD2 is DNA-independent. Myc tagged CxxC-RD2-AF9 but not the Myc tag alone interacts with PAFc subunits showing that PAFc binding to CxxC-RD2 is maintained in the presence of the AF9 translocation partner. E and F) A full length flag tagged MLL-AF9 fusion protein (E) or Flag tagged full length MLL (F) was expressed in 293 cells and immunoprecipitated. The PAF complex stably associated with full length MLL fusion protein and MLL as indicated by western blot. (See also Figure S1)

A) Schematic of Myc-tagged CxxC-RD2-AF9 constructs made with serial deletions of the RD2 region. The first and last MLL amino acid retained in the expression constructs are indicated. All constructs include the AF9 fusion partner at the C-terminus and Myc tag at the N-terminus. The Cx-1154-AF9 through Cx-1357-AF9 constructs are named according to the last MLL residue retained in the fusion protein. B) Myc-tagged RD2 deletion constructs described in A were expressed in 293 cells followed by immunoprecipitation. The Myc-tagged MLL deletion proteins and associated PAF complex components were detected by western blot with the indicated antibodies. C) Schematic of a second set of deletion constructs made through the RD2 region that also delete the pre-CxxC domain. Cx-del-1209-AF9 through Cx-del-1357-AF9 were made by deleting amino acids 1115 through 1179 of MLL. The amino acids included in each construct are indicated. D) Myc tagged constructs described in C were expressed in 293 cells, immunoprecipitated and detected as described in B.

A) Schematic diagram for MLL-AF9 and MLL-AF9 deletion colony forming assay. B) Real time qPCR using primers for the Flag tag on the MLL-AF9 fusion proteins. qPCR was performed on cDNA bone marrow cells after retroviral transduction. Expression levels are shown relative to MLL-AF9 transduced cells. Error bars indicate +/− SD. C) Western blot for the MSCV-based Flag tagged MLL-AF9 deletion constructs shown in D. Protein was extracted from transfected Plat-E cells and immunoprecipitated with M2 agarose beads. Proteins were detected with Flag antibodies. The loading control (LC) shows equal loading of protein. D) The constructs used for the retroviral infection and bone marrow colony assay are shown on the left. Final amino acids of the MLL deletions are shown. Primary, secondary and tertiary colony counts are shown for methylcellulose colony assays performed with the indicated MLL-AF9 fusion proteins. Error bars indicate standard deviation from duplicate experiments. One of more than three representative experiments is shown. E) Representative colony morphology is shown for each transduced MLL-AF9 fusion protein. Dense colonies are indicative of transformation while diffuse colonies indicate differentiation. Scale bar = 500μm F) Wright-Giemsa stained cytospins on cells isolated after the third round of methylcellulose plating. Scale bars indicate 50 μm. G) p-iodonitro tetrazolium violet (INT) stained colonies after three rounds of colony assay replating. Dense red colonies are visible from MLL-AF9, MLL-1299-AF9 and MLL-1357-AF9 transduced bone marrow. H) Schematic diagram for MLL-AF9 colony assay with shRNA mediated knock down of Cdc73 and Ctr9. Colony assays were performed as described in A except for an additional transduction after the second replating with shScram, shCdc73 and shCtr9 retrovises followed by plating in methylcellulose with puromycin selection. Colonies were scored following the third plating. I) Real Time qPCR was used to confirm knock down of Cdc73 and Ctr9 mRNA compared to shScram following transduction of 3T3 cells. Error bars indicate +/− SD. J) Protein lysate was collected from 3T3 cells transduced with shScram, shCdc73 and shCtr9 and separated by SDS-PAGE. Immunoblotting with Cdc73 and Ctr9 confirms knock down of the respective proteins. β-actin was probed as a loading control. K) Third round colony counts following transduction with shScram, shCdc73 or shCtr9. Error bars indicate +/− SD. L) Dense colonies are visible from colony assay plates containing shScram transduced cells. Significantly reduced colony numbers were visible from the shCdc73 and shCtr9 transduced cells. M) Cells collected after the second transduction with shRNA were grown in liquid culture in the presence of IL3, SCF and puromycin. A significant proliferative advantage was observed from MLL-AF9 cells transduced with shScram compared with shCdc73 or shCtr9. (See also Figure S4)

A) siRNA mediated knock down of individual PAF components was verified by western blotting. Antibodies used for western blotting are shown on the right and siRNA is indicated on top. B) Luciferase assays were performed with the Hoxa9-LUC reporter construct and MLL-AF9 in HeLa cells after transfection with the indicated siRNA. Luciferase units are shown relative to the MigR1 control transfected cells. C) Simultaneous siRNA mediated knock down of PAFc (siCTR9, siLEO1, siPAF1, siCDC73) is shown by western blotting for PAF components after siRNA transfection of HeLa cells. β-ACTIN was detected as a loading control. D) ChIP experiments were performed in HeLa cells following simultaneous knock down of the PAF components shown in C. PAF1, LEO1, CDC73, MLL^C, H3K4me3, H3K79me3 and H3 were immunoprecipitated from HeLa cells treated with siPAFc or non-targeting siRNA (Mock). Binding was assessed in both the promoter and coding region as indicated. A schematic of the HOXA9 locus and location of the primer-probe sets used for qPCR are shown. Error bars indicate +/− SD. One of more than three representative experiments is shown.

A) Coomassie blue staining of bacterially purified His-MOCR tagged CxxC-RD2, RD2, PAF1, LEO1, and His-MBP tagged CDC73, CTR9-N and CTR9-C. The amino acids of CTR9-N and CTR9-C are indicated. B) Schematic diagram of MLL and bacterially purified CxxC-RD2 and RD2 regions. Starting and ending amino acids are indicated. C) Immunoprecipitations performed with bacterially purified recombinant CxxC-RD2 or RD2 and PAF complex components. Individual PAF components were incubated with either CxxC-RD2 or RD2 and immunoprecipitated with MLL antibodies. PAF components and immunoprecipitated MLL fragments were detected with the indicated antibodies by western blot. Asterisk denotes detection of the IgG heavy chain. (See also Figure S2)

A) Luciferase assays were performed with the Hoxa9-LUC reporter construct and increasing doses of full length MLL (lanes 1–4) (0 – 0.6 μg) or PAFc (lanes 5–8) (0 – 0.6 μg). PAFc includes equal amounts of PAF1, LEO1, CDC73, CTR9 and WDR61. Lanes 9–12 show constant MLL (0.6 μg) with increasing doses of PAFc (0 – 0.6 μg). All changes are shown relative to lane 1 which includes Hoxa9-LUC and an empty expression vector. Error bars indicate +/− SD. Results of one of more than three representative experiments performed are shown. B) Experiment was performed as described in A except increasing doses of MLL-AF9 (0 – 0.6 μg) were used in lanes 1–4. Lanes 5–8 show increasing doses of PAFc (0 – 0.6 μg). Lanes 9–12 show increasing doses of PAFc (0 – 0.6 μg) in the presence of constant MLL-AF9 (0.6 μg). Error bars indicate +/− SD. One of three representative experiments is shown. C) Luciferase assay performed in transfected 293 cells using the Hoxa9-LUC reporter construct and individual PAFc components in lanes 1–6, the PAF complex (lane 7), MLL-AF9 alone (lane 8), MLL-AF9 plus individual PAF components (lanes 9–13) and MLL-AF9 with the PAF complex (lane 14). Error bars indicate +/− SD. Results of one of more than three representative experiments are shown. (See also Figure S3)

A) qPCR for Hoxa9 and Meis1 expression in MLL-ENL and E2A-HLF cell lines. Error bars indicate +/− SD. B) ChIP experiment performed in mouse bone marrow cell lines established with the MLL-ENL or E2A-HLF fusion protein. MLL-ENL IPs are shown in black and H2A-HLF IPs are shown in gray. Solid lines indicate the binding pattern of the component or histone modification listed to the right. The dotted lines indicate control IgG IPs for each cell line. The Hoxa9 locus is shown schematically at the bottom. C) Same experiment as described in B, but binding was determined on the Meis1 locus. The MLL-ENL IPs are shown in black (left) and the E2A-HLF IPs are shown in gray (right). (See also Figure S5)

A) Heat map generated from expression array data collected from differentiation of the Hoxa9ER cell line. Data is shown in triplicate for each time point following tamoxifen withdrawal. Labels indicate components of the PAF complex, MLL and MLL targets, and genes associated with myeloid cell differentiation. B) Expression of PAF components were verified using qPCR by separately differentiating the Hoxa9ER cell line by tamoxifen withdrawal. Time points indicate hours post tamoxifen withdrawal that RNA was collected. Expression for each component is shown relative to 0 hours. Error bars indicate +/− SD. C) THP-1 cells were transfected with an equal mixture of PAFc expression vectors or empty vectors along with a GFP vector at a 5:1 ratio (PAFc/Empty:GFP). Half were treated with PMA to induce differentiation. Surface expression of CD11b was monitored by FACS in the GFP positive gated cell population to track differentiation. Mean fluorescence values are shown for each sample. D) Structure of the CxxC domain and flanking sequences (PDB code: 2JYI). Solution structure of MLL CxxC region shows the flanking sequences of the CxxC domain brought into close juxtaposition creating a binding surface for both PAF1 and CTR9 of the PAF complex. E) Model for myeloid cell differentiation showing (top) PAFc dependent recruitment of MLL to target genes in hematopoietic progenitors (HSC/myeloblasts). PAFc and MLL recruitment promotes H2B mono-ubiquitination and H3K4 and H3K79 methylation resulting in transcriptional activation. (Middle) More differentiated myeloid cells down regulate PAFc expression resulting in decreased recruitment and transactivation by MLL. (Bottom) In leukemic cells harboring MLL fusion proteins the fusion protein synergizes with PAFc resulting in robust transcriptional activation of target genes. This is associated with increased H2B mono-ubiquitination and histone H3K79 methylation by DOT1L recruited through the MLL fusion partner. (See also Figure S6)