INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

AML1 is one of the most frequent targets of chromosomal abnormalities in acute leukemias and is involved in multiple translocations. The t(8;21)(q22;q22) translocation fuses residues 1 to 177 of AML1, including the DNA binding domain, to ETO (MTG8), a gene of unknown function that is homologous to the Drosophila gene nervy (8, 9, 13, 41, 42, 44, 48). It is the second most common chromosomal abnormality in acute myeloblastic leukemias (AML) (42). A second translocation, t(3;21), is rare in de novo AML and is detected in therapy-related AML and during blast crisis of chronic myelogenous leukemias (40, 46, 49, 59). It fuses the first five or six exons of AML1 to three different exons of EviI, a gene encoding a transcription factor on chromosome 3 (43, 47, 60). A third translocation, t(12;21), fuses the first 333 amino acids of TEL, an ets-like protein (17), to nearly all of AML-1B, the largest AML1 product (38), and has been detected in approximately 30% of pediatric B-cell acute lymphoblastic leukemia cases (16, 57, 58, 63). Unlike AML-1-ETO, TEL-AML-1B contains the entire carboxy terminus of AML-1B, including the nuclear matrix targeting signal (NMTS) and transactivation (TA) domain (76). A fourth alteration, inv(16), is observed in AML of the M4Eo subtype (French-American-British classification) and indirectly targets AML-1B by fusing CBF-, the gene for the heterodimeric DNA binding partner of AML-1B (71), to the smooth muscle myosin heavy chain gene, MYH11 (35, 64).

AML-1B (also known as core binding factor 2 [CBF-2] and polyomavirus enhancer binding protein 2B1 [PEBP-2B1]) binds to and activates transcription from enhancer core motifs (TGT/cGGT), which are present in numerous myeloid promoters and lymphoid enhancers (e.g., granulocyte [G]-monocyte [M] colony-stimulating factor [CSF] receptor, M-CSF receptor, myeloperoxidase, neutrophil elastase [NE], interleukin-3 [IL-3], and T-cell receptors [TCR] , , and ) (4, 14, 19, 27, 36, 38, 45, 55, 66, 68, 78). The core binding motif is necessary but not sufficient for tissue-specific activation of myeloid promoters and lymphoid enhancers; therefore, it is possible that AML-1B functions as a promoter organizer. We previously showed that the t(8;21) and t(12;21) translocations convert AML-1B from a transcriptional activator to a repressor (25, 38). Because only one allele of AML1 is altered in leukemic cells expressing t(8;21) and because substoichiometric levels of AML-1-ETO efficiently repressed AML-1B-dependent transcriptional activation, we hypothesized that the t(8;21) product is a dominant inhibitor of AML-1B function (14, 36, 38). AML-1-ETO also repressed transcriptional activation induced by AML-2 (CBF-3 or PEBP-2C) and the murine homolog of AML-3 (mAML-3, CBF-1, or PEBP-2A) (1, 33, 39, 51). Thus, AML-1-ETO is able to repress transcription mediated by all core binding factors in hematopoietic tissues. Interestingly, AML1-ETO and inv(16) transgenic mice display a phenotype similar to that of AML-1 (CBF-)- and CBF--deficient mice, as they die during embryogenesis from central nervous system hemorrhages and exhibit severe blocks in fetal liver hematopoiesis (6, 52, 69, 70, 74). These data suggest that AML1 is an important regulator of hematopoiesis.

CCAAT enhancer binding protein  (C/EBP-) is a tissue-specific transcription factor that was originally described as a rat liver nuclear protein (28) but is also expressed in differentiating adipocytes and proliferating myelomonocytic cells (3, 5, 62). During myelopoiesis, C/EBP- expression is temporal, as its levels are high in dividing myelomonocytic cells but decrease during granulocyte differentiation (62). C/EBP- minimally activates several myeloid cell-specific promoters, including those for cytokine receptors (e.g., GM-CSF, G-CSF, and M-CSF receptors) (26, 65, 78, 79) and granule proteins (e.g., NE) (50). Moreover, C/EBP- cooperates with other myeloid transcription factors, including AML-1B and PU.1, to synergistically upregulate the expression of several myeloid cell-specific promoters (26, 50, 78). Unlike AML1, C/EBP- has not been identified as a target of chromosomal translocations in leukemias. However, the critical role of C/EBP- in hematopoiesis is underscored by the lack of G-CSF receptors on multipotential myeloid progenitors and the absence of neutrophils in C/EBP--deficient mice (80).

Defensins are 3- to 4-kDa antimicrobial cytotoxic peptides produced by neutrophils, some macrophages, and intestinal Paneth cells (30, 31). In human neutrophils, defensins constitute greater than 5% of total cellular protein (15). During myeloid differentiation, defensin mRNA levels are highest in promyelocytes and decrease during differentiation; however, mature defensin proteins are present in the primary granules of all neutrophils (75). Four defensins (or neutrophil proteins) have been identified thus far in rat (NP1-4) and human (HNP1-4) neutrophils (7, 75). The promoters of several of these defensin genes contain core binding motifs and CCAAT boxes (2, 34).

In this report, we demonstrate that AML family members and C/EBP- independently and synergistically activated the rat NP-3 promoter. The t(8;21) product, AML-1-ETO, associated with C/EBP- and inhibited C/EBP--dependent transactivation and synergistic activation by C/EBP- and AML-1B. AML-1-ETO also blocked G-CSF-induced differentiation of myeloid progenitors. These results demonstrate for the first time that AML-1-ETO may disrupt the organization and normal activity of multiple transcription factors on myeloid promoters, thereby blocking differentiation and contributing to leukemogenesis.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Cell culture. C33A and COS-7 cells were cultured in Dulbecco's modified Eagle medium (BioWhittaker Inc., Walkersville, Md.) containing 10% heat-inactivated fetal calf serum (FCS), 50 U of penicillin per ml, 50 µg of streptomycin per ml, and 2 mM L-glutamine (all from BioWhittaker). 32D.3 cells were maintained in RPMI 1640 medium (BioWhittaker) containing 10% FCS, antibiotics, L-glutamine, and 15 U of IL-3 per ml.

Construction of plasmids. Luciferase constructs containing rat NP-3 5'-flanking sequences were synthesized by PCR with sequence-specific sense and antisense primers containing MluI and XhoI sites as described elsewhere (73). The PCR products were subcloned into these restriction sites upstream of the luciferase gene in the pGL2-basic vector (Promega, Madison, Wis.). The AML-1B(1-275/314-480) deletion mutation was constructed by use of PCR to generate a fragment consisting of nucleotides 1 to 825 of the AML-1B sequence (38). The primers used were 5'-GTCGAATTCATGGCTTCAGACAGCATA-3' and 5'-CATCTGCAGATGGTTGGATCTGCCTTGTATC-3'. This fragment was cloned into the pCMV5-AML-1B expression plasmid (38) at the EcoRI and PstI sites. The pCMV5-C/EBP- expression plasmid was produced by subcloning the EcoRI-HindIII fragment from MSV-C/EBP- (a kind gift from Alan Friedman) into pCMV5. The remaining CMV5 expression plasmids were previously described (32, 36, 38, 39, 54). The sequences of all PCR products were confirmed by the dideoxy chain termination method. The sizes of the deletion mutant proteins were confirmed by Western blot analysis (data not shown).

Electrophoretic mobility shift assays. COS-7 cells were transiently transfected with 3 µg of supercoiled CMV5 plasmids expressing AML-1B, AML-2, AML-3, or C/EBP- by the DEAE-dextran method (38, 39). Whole-cell extracts were prepared 40 h later by washing the cells with phosphate-buffered saline (PBS) (pH 7.4) prior to resuspension in microextraction buffer (20 mM HEPES [pH 7.4], 450 mM NaCl, 0.2 mM EDTA, 0.5 mM dithiothreitol, 25% glycerol, 100 µg of phenylmethylsulfonyl fluoride per ml, 10 µg of aprotinin per ml, 100 µM sodium orthovanadate) and sonication. Lysates were precleared by high-speed centrifugation, and protein concentrations were determined with Bradford reagent (Bio-Rad Laboratories, Hercules, Calif.). Protein (10 µg) was added to DNA binding reaction mixtures, which were previously described (21, 36). Annealed oligomers containing the AML binding site (36) or MluI/XhoI NP-3 promoter fragments were labeled with [-³²P]dATP (Amersham Life Science, Inc., Arlington Heights, Ill.) in a standard Klenow reaction mixture. For competition studies, 100 ng of unlabeled, annealed oligomers containing the wild-type (TGTGGT) or mutated (TGTTAG) AML binding site (36), the C/EBP- binding site (5'-CATGAATTCTGCAGATTGCGCAATCTGCAGGATCCT-3' or 5'-ATAGGATCCTGCAGATTGCGCAATCTGCAGAATTCA-3'), or the indicated NP-3 promoter region (Fig. 1) was added to the DNA binding reaction mixtures. For supershift analyses, 1 µg of C/EBP- antiserum (Santa Cruz Biotechnology, Inc., Santa Cruz, Calif.) was added to the binding reaction mixtures.

Transcriptional analysis. C33A cells were transiently transfected in 10-cm dishes by adding calcium phosphate precipitates containing 5 µg of pGL2-NP-3-luciferase plasmid (NP-3-Luc) and various amounts of control pCMV5 or CMV5 expression plasmid(s) and/or control pMSV or MSV-C/EBP- expression plasmid dropwise to cell cultures as previously described (18, 21, 38). Rous sarcoma virus (RSV) long terminal repeat (LTR)-chloramphenicol acetyltransferase (CAT) plasmid (RSV-CAT plasmid) (0.5 µg) or 5 µg of RSV-secreted alkaline phosphatase (SEAP) plasmid (RSV-SEAP plasmid) was also added as an internal control for transfection efficiency. Sonicated salmon sperm DNA (Sigma) was added to bring the total amount of DNA per transfection to 25 µg. After 40 to 48 h, the cells were washed twice with PBS and lysed in 350 µl of reporter lysis buffer (Promega). Luciferase activity in 20 µl of lysate was determined by measuring the relative light units (RLU) produced within 10 s after the addition of 100 µl of luciferase assay reagent (Promega). RLU were normalized with respect to CAT or SEAP activity, which was measured as previously described (18, 22, 38).

Immunoprecipitations. COS-7 cells were transiently transfected with 3 µg of supercoiled CMV5 expression plasmids by use of DEAE-dextran. After 40 h, the cells were washed with PBS, incubated for 30 min with methionine- and cysteine-free Dulbecco's modified Eagle medium containing 2% dialyzed FCS, and then metabolically labeled for 3 h in the same medium containing [³⁵S]methionine and [³⁵S]cysteine (PROMIX; Amersham). Cells were resuspended in extraction buffer (PBS [pH 7.4], 0.5% Triton X-100, 10 µg of aprotinin per ml, 5 µg of leupeptin per ml, 100 µg of phenylmethylsulfonyl fluoride per ml, 100 µM sodium orthovanadate) and sonicated. Lysates were precleared at 4°C for 30 min with Immunoprecipitin (formalin-fixed staphylococcal protein A membranes; GIBCO-BRL, Gaithersburg, Md.) and then immunoprecipitated for 16 to 20 h with affinity-purified rabbit AML-N (36), ETO (38), or C/EBP- (Santa Cruz Biotechnology) antisera. Immunoprecipitates were collected with protein A-Sepharose beads (Pharmacia Biotech, Uppsala, Sweden), washed three times with extraction buffer, and analyzed by sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis. Gels were fixed in 45% methanol-10% acetic acid for 30 min and then incubated in Amplify (Amersham) for 30 min before being dried and exposed to film.

32D.3 differentiation. Parental 32D.3 cells were electroporated with pMTCB6⁺-AML/ETO plasmids and selected in G-418. The pMTCB6⁺ vector was a kind gift from Ismail Kola. AML-ETO-expressing pools and single-cell clones were identified by Western blot analysis with anti-ETO antibodies as described previously (24, 38). Asynchronously growing cell lines were washed twice with PBS to remove IL-3 and resuspended in RPMI 1640 medium supplemented with 15% fetal bovine serum, 1% L-glutamine, 1% penicillin and streptomycin, and 25 ng of G-CSF (Neupogen; Amgen Biologicals Inc., Thousand Oaks, Calif.) per ml. The morphology of differentiating cells was determined by cytocentrifugation of 5 × 10⁴ cells onto a glass slide and staining with Wright stain.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

AML family members and C/EBP- bind to rat NP-3 promoter sequences. The rat NP-3 promoter contains several potential binding sites for myeloid transcription factors, including ones for AML and C/EBP family members (Fig. 1A) (2). To determine if AML family members could bind to sites in the NP-3 promoter, lysates from COS-7 cells overexpressing AML-1B, AML-2, or mAML-3 were incubated with a ³²P-labeled AML consensus binding site oligonucleotide probe, and specific complexes competed with NP-3 promoter regions or unlabeled oligonucleotides. AML-1B, AML-2, and mAML-3 specific complexes were reduced by the addition of the unlabeled wild-type oligonucleotide but not the mutated binding site oligonucleotide for binding to the probe (Fig. 1B to D). No endogenous core binding proteins from COS-7 lysates were observed. Furthermore, the sizes of the AML isoforms were distinguished by their relative mobilities and supershift analysis (data not shown). Three NP-3 promoter sequences extending from the first exon to 187, 137, or 87 also competed with the labeled oligonucleotide probe for protein binding. In addition, AML-1B, AML-2, and mAML-3 specifically associated with labeled NP-3 promoter fragments (data not shown).

View larger version (11K): [in this window] [in a new window]   View larger version (47K): [in this window] [in a new window]   FIG. 1.   AML-1B, AML-2, AML-3, and C/EBP- bind to NP-3 promoter sequences. (A) Schematic of the rat NP-3 promoter showing locations of core binding motifs (underlined), C/EBP binding sites, and reporter constructs used in this study. (B to D) Binding of AML family members to NP-3 promoter sequences determined by electrophoretic mobility shift assays. A 32P-labeled annealed oligonucleotide probe containing the consensus core site, TGTGGT, was incubated with 3 µg of lysates from COS-7 cells overexpressing AML-1B (B), AML-2 (C), or mAML-3 (D) in the presence or absence of the indicated unlabeled competitor. The NP-3 promoter competitors were generated by PCR. (E and F) Binding of C/EBP- to NP-3 promoter sequences. NP-3 promoter sequences extending 137 (E) and 87 (F) nucleotides from the transcription start site were labeled with [-32P]dATP and incubated with 3 µg of C/EBP--overexpressing COS-7 cell lysates in the presence or absence of an unlabeled oligonucleotide containing the consensus C/EBP binding site. Supershift assays were performed with 1 µg of C/EBP- antiserum. The location of the C/EBP--DNA complex is denoted by the arrows, and the supershifted complex is adjacent to the asterisks.

Members of the AML family of transcription factors activate the NP-3 promoter. AML-1B activates transcription from numerous myeloid and lymphoid cell-specific promoters or enhancers (4, 14, 19, 27, 36, 38, 45, 55, 66, 68, 78). To determine if AML-1B could activate the NP-3 promoter, various regions of the promoter were linked to the luciferase gene in the pGL2-basic vector (Fig. 1A). The NP-3-Luc plasmid and the control RSV-CAT or RSV-SEAP reporter plasmid were cotransfected with AML-1B expression plasmids into a human cervical carcinoma cell line, C33A. These cells contain low levels of endogenous AML family members but high levels of CBF- (38). As shown in Fig. 2A, AML-1B activated all four NP-3 promoter constructs. NP-3(700), which contains four potential AML binding sites, and NP-3(187) and NP-3(137), which each have three sites, were activated to levels five- to sevenfold higher than the background. The NP-3(87) region contains two potential AML binding sites and was activated approximately three- to fourfold. AML-1B did not significantly alter CAT or SEAP expression from the RSV LTR (data not shown but used to control transfection efficiency). These results suggest that the three AML binding sites within NP-3(137) are required for maximal activation of the promoter by AML-1B.

View larger version (40K): [in this window] [in a new window]   FIG. 2.   Activation of the NP-3 promoter by AML (CBF) transcription factors. (A) C33A cells were transfected with 2 to 5 µg of the indicated NP-3-Luc reporter construct, either 5 µg of RSV-SEAP or 0.5 µg of RSV-CAT as an internal control, and 1 µg of pCMV5 or pCMV5-AML-1B expression plasmid. RLU were normalized with respect to CAT or SEAP activity. Fold activation represents the normalized promoter activity from cells transfected with AML-1B relative to activity from cells transfected with NP-3-Luc alone. The basal activity was normalized to 1 for each promoter construct. (B) Effects of 1 µg of pCMV5-AML-1B, -AML-2, and -mAML-3 on NP(137) activity were determined as described for panel A. Results represent the mean ± SD for triplicate experiments.

C/EBP- activates the rat NP-3 promoter. Promoters of several myeloid cell-specific genes are regulated by C/EBP-, including those of the primary granule proteins, myeloperoxidase and NE (50, 79). Because the NP-3 promoter contains two potential C/EBP- binding sites (Fig. 1A), we tested the effects of C/EBP- on NP-3 promoter activity in C33A cells. We did not detect endogenous C/EBP- in these cells using Western blot analysis (data not shown). C/EBP- activated NP-3(137) and NP-3(187) approximately sixfold over background levels (Fig. 3). C/EBP- also activated NP-3(700) approximately fourfold. By contrast, C/EBP- minimally activated NP-3(87), even though it bound to this fragment in electrophoretic mobility shift assays (Fig. 1F). Because the NP-3(137) construct was the shortest sequence that was maximally activated by both AML-1B and C/EBP-, it was used in all subsequent studies.

View larger version (43K): [in this window] [in a new window]   FIG. 3.   C/EBP- activates the NP-3 promoter. C33A cells were transfected with 2 to 5 µg of the indicated NP-3-Luc reporter construct, either 5 µg of RSV-SEAP or 0.5 µg of RSV-CAT as an internal control, and 0.5 µg of pMSV or pMSV-C/EBP- expression plasmid. RLU were normalized with respect to CAT or SEAP activity. Fold activation represents the normalized promoter activity from cells transfected with C/EBP- relative to activity from cells transfected with NP-3-Luc alone. The basal activity was normalized to 1 for each promoter construct. Results represent the mean ± SD for triplicate experiments.

C/EBP- and AML family members synergistically activate the rat NP-3 promoter. C/EBP- and AML-1B were previously shown to physically interact and cooperatively activate the myeloid cell-specific, M-CSF receptor promoter (78). Because C/EBP- and AML-1B independently activated the NP-3 promoter, we tested their combined effects. Consistent with earlier results, AML-1B and C/EBP- each activated NP-3(137) approximately sixfold (Fig. 4). When cotransfected, however, C/EBP- and AML-1B synergistically activated the promoter to levels more than 60-fold higher than the background. This result translates into a fivefold synergistic effect (calculated by dividing the observed fold activation by the expected additive response). C/EBP- and AML-1B also synergistically activated the NP-3(700) and NP-3(187) promoter constructs (data not shown). However, AML-1B and C/EBP- together only activated the shorter, NP-3(87), promoter by 1.9-fold (data not shown). Thus, the sequences between 137 and 87 of the NP-3 promoter are required for maximal synergy of AML-1B and C/EBP-.

View larger version (40K): [in this window] [in a new window]   FIG. 4.   C/EBP- and AML family members cooperatively activate the NP-3 promoter. C33A cells were transfected with 5 µg of NP-3(137); 0.5 µg of RSV-CAT; 1 µg of pCMV5 or pCMV5-AML-1B, -AML-2, or -mAML-3; and 0.5 µg of pMSV (open bars) or MSV-C/EBP- (shaded bars). RLU were normalized with respect to CAT activity. Results represent the mean ± SD for triplicate experiments.

The C terminus of AML-1B is required for synergistic activation with C/EBP-. The DNA binding domain (also known as the runt homology domain [rhd]) of AML-1B and the b-Zip region of C/EBP- physically interact (78). To determine if the C terminus of AML-1B is important for synergistic activation with C/EBP-, we tested deletion mutants that lack various functional regions of AML-1B. The deletion mutants used are illustrated in Fig. 5A. AML-1B(1-381) lacks the last 99 amino acids of AML-1B, including the transactivation domain. AML-1B(1-290) is a truncated version of AML-1B and lacks the NMTS and the TA domain (76). It differs slightly from AML-1, which lacks the N-terminal residues of AML-1B and has a unique C-terminal 9-amino-acid segment encoded by alternative splicing of exon 7A (41). AML-1B(1-290/351-381) and AML-1B(1-290/432-480) restore the NMTS and the TA domain of AML-1B, respectively, to the AML-1B(1-290) mutant. AML-1B(1-275/314-480) contains both the NMTS and the TA domain of AML-1B but lacks the PST-rich region, which contains four serine or theonine residues that are potential phosphorylation sites for the extracellular signal-regulated kinase pathway (67). The AML-1B(1-290) mutant and its variants lack three of these residues but retain the major phosphorylation site, S276.

View larger version (22K): [in this window] [in a new window]   View larger version (22K): [in this window] [in a new window]   FIG. 5.   Effects of AML-1B C-terminal deletion mutants on C/EBP- synergism. (A) Schematic of AML1, AML-1B, and AML-1B mutants used in this study. (B to D) Synergistic effects of C/EBP- and AML-1B or AML-1B C-terminal deletion mutants (B and C), C/EBP- and AML-1 (D), or C/EBP- and an AML-1B mutant lacking the ERK phosphorylation sites (E) on NP-3(137) activity determined in C33A cells as described in the legend to Fig. 4. Fold synergy is the quotient of the actual synergistic activation and the expected additive response. Results represent the mean ± SD for triplicate experiments.

The t(8;21) product, AML-1-ETO, blocks AML-1B- and C/EBP--dependent activation of the NP-3 promoter. The t(8;21) translocation fuses the amino terminus and rhd of AML-1B to nearly all of ETO (Fig. 6A). AML-1-ETO represses AML-1B- and AML-2-dependent transcription from the TCR- enhancer and the GM-CSF receptor and IL-3 promoters (14, 38, 39, 68); however, it has also been reported that AML-1-ETO activates the M-CSF receptor and bcl-2 promoters (29, 56). To determine whether AML-1-ETO affected activation from the NP-3 promoter, we coexpressed AML-1-ETO with AML-1B and/or C/EBP- in C33A cells. AML-1-ETO was without effect on the basal activity of the NP-3(137) promoter (Fig. 6B). Once again, AML-1B and C/EBP- each activated transcription to levels approximately fivefold higher than the background and AML-1B plus C/EBP- synergistically activated the promoter approximately 60-fold. AML-1-ETO repressed AML-1B-dependent activation of NP-3(137) approximately 35%; however, it completely inhibited C/EBP--dependent activation. AML-1-ETO also repressed synergistic activation by AML-1B and C/EBP-. Synergistic activation by C/EBP- and AML-2 or mAML-3 was inhibited by AML-1-ETO as well (72). Although substoichiometric amounts of AML-1-ETO were sufficient to inhibit AML-1B- and AML-2-dependent activation of the TCR- enhancer (38, 39), we found that larger amounts of AML-1-ETO were necessary for complete repression of the AML-1B-C/EBP- synergy (data not shown). Larger amounts of AML-1-ETO, however, did not affect transcriptional activity from the control or test reporter plasmid. A point mutation (L148D) (32) that prevented AML-1-ETO DNA binding also eliminated repression mediated by AML-1-ETO (Fig. 6C). These results suggest that the t(8;21) fusion product not only interferes with AML family member activity but also, when placed in proximity, can inhibit other hematopoietic transcription factors, including C/EBP-.

View larger version (40K): [in this window] [in a new window]   View larger version (32K): [in this window] [in a new window]   FIG. 6.   Effects of AML-1-ETO, AML-1-ETO truncation mutants, and inv(16) on C/EBP-- and AML-1B-induced synergistic activation of the NP-3 promoter. (A) Schematic of the AML-1-ETO and inv(16) proteins used in this study. (B to D) Transfection of C33A cells with 5 µg of NP-3(137), 0.5 µg of RSV-CAT, 1 µg of pCMV5 or CMV5-AML-1B expression plasmid, 0.5 µg of pMSV or MSV-C/EBP- expression plasmid, or both CMV5-AML-1B and MSV-C/EBP- in the presence of 10 µg of pCMV5 (control) or pCMV5-AML-1-ETO (B), pCMV5-AML-1-ETO-L148D (C), or pCMV5-inv(16) (D). RLU were normalized with respect to CAT activity. Results represent the mean ± SD for triplicate experiments.

CBF-

AML-1-ETO physically interacts with C/EBP- in vivo. C/EBP- was previously shown to interact with the rhd of AML-1B (78) in vitro. Physical interactions between AML-1-ETO and C/EBP-, however, have not been defined. To determine whether ETO sequences affected interactions between AML-1B and C/EBP-, we attempted to coimmunoprecipitate these factors from metabolically labeled COS-7 cells expressing these proteins. AML-N antiserum specifically immunoprecipitated AML-1, AML-1B, AML-1-ETO, and AML-1-ETO469 but not C/EBP- (Fig. 7A, lanes 1 to 5). Analogously, C/EBP- antiserum immunoprecipitated C/EBP- but not the AML proteins (Fig. 7B, lanes 1 to 5). Incubation of lysates from COS-7 cells overexpressing C/EBP- and an rhd-containing protein (AML-1, AML-1B, AML-1-ETO, or AML-1-ETO469) with AML-N antiserum (Fig. 7A, lanes 6 to 9) or C/EBP- antiserum (Fig. 7B, lanes 6 to 9) revealed an association between C/EBP- and each of these AML proteins. C/EBP- also associated with AML-2 and mAML-3 (72). To confirm the association between AML-1-ETO and C/EBP-, we also coimmunoprecipitated lysates overexpressing these proteins with ETO antiserum. As shown in Fig. 7C, ETO antiserum immunoprecipitated AML-1-ETO and ETO but not C/EBP- unless it was coexpressed in vivo with AML-1-ETO. ETO was not immunoprecipitated by C/EBP- antiserum when expressed in COS-7 cells in either the presence or the absence of C/EBP- (Fig. 7D). Because ETO did not associate with C/EBP- and the deletion of ETO sequences from AML-1-ETO did not affect the interaction, we conclude that AML sequences, most likely the rhd, mediate the interaction with AML-1-ETO.

View larger version (53K): [in this window] [in a new window]   FIG. 7.   C/EBP- physically associates with AML-1-ETO but not ETO in vivo. COS-7 cells were transfected with the indicated pCMV5 expression plasmid(s) and metabolically labeled for 3 h with [35S]methionine. Lysates were immunoprecipitated with antiserum to the N terminus of AML1 (A), C/EBP- (B and D), or ETO (C) and analyzed by denaturing sodium dodecyl sulfate-10% polyacrylamide gel electrophoresis. Numbers at the left indicate the migration of molecular weight standards.

ETO sequences are required for repression of C/EBP- by AML-1-ETO. Because the fusion protein physically interacts with C/EBP-, we tested whether AML-1-ETO simply titrates C/EBP- or whether ETO sequences are required for repression of the synergistic activation of the NP-3 promoter. AML-1-ETO469 lacks the carboxy-terminal 283 amino acids of AML-1-ETO and two conserved regions that are potential protein interaction domains: a hydrophobic heptad repeat (HHR) and two zinc finger domains. AML-1-ETO540 retains the HHR but lacks the zinc finger domains (Fig. 6A) (32). AML-1-ETO469 does not repress AML-1B-dependent activation of the TCR- enhancer, whereas AML-1-ETO540 represses this activation nearly as well as wild-type AML-1-ETO (32). Neither of these mutants had an effect on the basal activity of NP-3(137) (Fig. 8); however, each of these proteins inhibited AML-1B-dependent activation. Thus, unlike what was previously observed with the TCR- enhancer, regions within the first 469 amino acids of AML-1-ETO may be required for repression on some promoters. By contrast, C/EBP--induced activation of NP-3(137) was completely inhibited by AML-1-ETO and AML-1-ETO540 but not by AML-1-ETO469. Similar repression patterns were observed for AML-1B and C/EBP- synergistic activity. Wild-type AML-1-ETO and AML-1-ETO540 repressed 80 to 90% of the synergism, whereas AML-1-ETO469 repressed it by approximately 40%. Overexpression of ETO did not repress AML-1B and C/EBP- synergistic activity, nor did it reverse AML-1-ETO-mediated repression (data not shown). These results suggest that ETO sequences alone are not sufficient to inhibit transcriptional activity; however, when fused to AML sequences, they are required for repression.

View larger version (29K): [in this window] [in a new window]   FIG. 8.   ETO sequences are required for C/EBP- inhibition by AML-1-ETO. C33A cells were transfected with 2.5 µg of the indicated pCMV5-AML-1-ETO construct, 5 µg of NP-3(137), 0.5 µg of RSV-CAT, 1 µg of pCMV5 or CMV5-AML-1B expression plasmid, and 0.5 µg of pMSV or MSV-C/EBP- expression plasmid. Data were corrected as described in the legend to Fig. 6. Results represent the mean ± SD for triplicate experiments.

AML-1-ETO blocks granulocyte differentiation. AML1 is required during development for fetal liver hematopoiesis (52, 69), and C/EBP--deficient mice lack granulocytes (80), suggesting that the activity of these factors is required for myeloid cell differentiation. To test whether the t(8;21) fusion protein inhibits granulocyte differentiation, we expressed AML-1-ETO from the zinc sulfate-inducible sheep metallothionein promoter in 32D.3 murine myeloid progenitor cells. As shown in Fig. 9A, zinc sulfate induced AML-1-ETO expression in 32D.3 clones (A/E.6 and A/E.17). When control, G-418-resistant, cells were cultured with G-CSF and zinc sulfate (Fig. 9B), they differentiated into mature granulocytes (note the segmented nuclei characteristic of polymorphonucleated cells at days 6 and 12). By contrast, 32D.3 cells expressing AML-1-ETO did not fully differentiate, even after 12 days of culturing with G-CSF, with the majority of cells resembling band neutrophils (Fig. 9B, A/E.6 and A/E.17, days 3, 6, and 12). Moreover, cells expressing AML-1-ETO (clone A/E.17) failed to express the granulocyte differentiation marker GR-1, indicating a differentiation blockade (data not shown). However, these cells did not become growth factor independent, as G-CSF was required for their continued growth (data not shown).

View larger version (60K): [in this window] [in a new window]   FIG. 9.   The t(8;21) fusion protein blocks granulocyte differentiation. (A) AML-1-ETO expression in 32D.3 clones containing empty vector (lane 1, MT) or expressing AML-1-ETO (lanes 2 to 5, clones 6 and 17) before and after 6 h of incubation in 400 µM zinc sulfate was determined by immunoblot analysis with ETO-specific antisera. (B) G-418-resistant control (pMT) and two clonal cell lines expressing AML-1-ETO (A/E) were cultured in the presence of G-CSF. At the indicated times, cells were cytocentrifuged and Wright stained. Note that at day 6, the control but not the cells expressing AML-1-ETO contained multilobed nuclei, which are characteristic of granulocyte maturation. By day 12, few viable cells remained in the control cell cultures, whereas cells expressing the fusion protein maintained an immature morphology.

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

Hematopoiesis is a highly regulated process in which the growth and differentiation of pluripotent stem cells into specific cell lineages are controlled by the coordinated regulation of gene expression. Leukemias result when either the growth pathways become uncontrollable or when cells lose their ability to differentiate (61). How the chimeric genes that are formed at chromosomal translocation breakpoints disrupt normal cellular proliferation and differentiation is a central question in determining the mechanism(s) of leukemogenesis. The t(8;21) fusion protein, AML-1-ETO, acts as a dominant inhibitor of core binding factors on the TCR- enhancer and the IL-3 and GM-CSF receptor promoters (14, 38, 68). AML-1-ETO also blocks AML-2- and mAML-3-induced activation of the TCR- enhancer, suggesting that AML-1-ETO interferes with the activity of all core binding factors (23, 39). Because substoichiometric amounts of AML-1-ETO were sufficient to inhibit AML-1B-induced activation, AML-1-ETO did not simply compete for DNA binding sites but rather acted as a dominant inhibitor. These results were corroborated by the generation of AML-1-ETO "knock-in" mice, which displayed a phenotype similar to that of AML1 "knock-out" mice (i.e., embryonic lethality, central nervous system hemorrhaging, and lack of fetal liver hematopoiesis) (52, 70, 74).

To further probe the mechanism of transcriptional interference mediated by AML-1-ETO, we tested the ability of the fusion protein to inhibit the action of surrounding positively acting factors on the myeloid cell-specific NP-3 promoter. Like other myeloid cell-specific promoters, NP-3 contains core binding sites adjacent to C/EBP binding sites. Potential PU.1 and c-myb sites are also present in this region (2). The relatively low (e.g., fivefold) activation of NP-3 by AML-1B or C/EBP- alone is consistent with the activation of other promoters by these factors. The fivefold synergism seen for the NP-3 promoter when these factors are coexpressed is also similar to the cooperative activation of other myeloid cell-specific promoters (e.g., M-CSF receptor) (78). In this report, we show for the first time that AML-1-ETO can inhibit the activity of other transcription factors (C/EBP-) as well as synergistic activation by AML-1B and C/EBP-. AML-1-ETO also can inhibit synergistic activation by AML-2-C/EBP- and AML-3-C/EBP- (72). Although substoichiometric amounts of AML-1-ETO were sufficient to block C/EBP--dependent activation, equal amounts of AML-1-ETO were required for complete repression of AML-1B-C/EBP- synergistic activation. Interestingly, granulopoiesis in C/EBP--deficient mice is arrested at the myeloblastic stage (80). t(8;21)-positive leukemic cells also have a myeloblastic phenotype, although they show some maturation. It is interesting to speculate that the inhibition of C/EBP- by the t(8;21) product, AML-1-ETO, may contribute to the leukemic phenotype.

The mechanism by which AML-1-ETO blocks C/EBP- activation and the mechanism of its synergistic activity with the core binding factors are unknown. Physical interactions between AML family members and C/EBP- may be responsible for the synergy. Although AML-1-ETO also physically interacts with C/EBP-, this interaction is not sufficient to inhibit C/EBP- function. AML-1-ETO469 retains the DNA binding domain and associates with C/EBP- (Fig. 7) (32) but fails to inhibit C/EBP--dependent transactivation (Fig. 8). Thus, it is unlikely that AML-1-ETO simply titrates or sterically blocks C/EBP- from interacting with the promoter. Moreover, because AML-1-ETO does not affect the basal activity of the NP-3 promoter, it probably does not inhibit the basal transcriptional machinery. These results lead us to hypothesize that the HHR and possibly the zinc finger domains of AML-1-ETO recruit a second protein(s) that acts as a corepressor.

ETO is a nuclear protein expressed at high levels in the central nervous system and hematopoietic tissues (10, 13). It contains several domains that may mediate protein-protein interactions (Fig. 6A). The first domain is homologous to TAF110, an Sp-1 coactivator (13). The second domain, the HHR, may form an amphipathic -helix (32, 37). The third domain contains two zinc finger motifs that are conserved in several other proteins, including Drosophila Nervy and DEAF-1; a Deformed cofactor on homeotic response elements; and RP-8, a cell death-associated protein (11, 13, 20, 41, 53). The requirement of both the zinc finger and HHR regions for AML-1-ETO-mediated repression (32) (Fig. 8) leads us to speculate that ETO normally acts as a transcriptional repressor.

In some circumstances, AML-1-ETO may be a positive regulator of transcription and may contribute to leukemogenesis by upregulating the expression of genes (e.g., M-CSF receptor and BCL-2) (29, 56). Rhoades et al. (56) showed that coexpression of equal amounts of AML-1B and AML-1-ETO expression vectors in the presence or absence of C/EBP- resulted in modest activation of the M-CSF receptor; however, when added at higher levels, AML-1-ETO inhibited AML-1B-dependent activation of the M-CSF receptor. Because there is only one AML-1 binding site in the M-CSF receptor promoter (77), this "synergistic" activation may be the result of AML-1-ETO titrating a negatively acting factor.

The translocations that target AML-1B appear to create promoter-specific repressors. AML-1-ETO represses the AML-1B-C/EBP- synergistic activation of the NP-3 promoter. Consistent with the hypothesis that the inv(16) product titrates AML-1B through physical interactions (35), CBF--MYH11 represses AML-1B-dependent activation but has only minimal effects on C/EBP-. The finding that the inv(16) fusion protein had no effect on AML-1B-C/EBP- synergism (Fig. 6D) may indicate that it is inefficient in sequestering all of the available AML-1B expressed from the cytomegalovirus immediate-early promoter. The t(12;21) product, TEL-AML-1B, had effects similar to those of inv(16) on activation by AML-1B and C/EBP- but inhibited AML-1B-C/EBP- synergistic activation of the NP-3 promoter by approximately 50% (72). In other assays, AML-1-ETO failed to inhibit the basal activity of the TCR- enhancer linked to the simian virus 40 early promoter, whereas TEL-AML-1B efficiently repressed expression from this construct (23). TEL-AML-1B also inhibited AML-1B and C/EBP- synergistic activation of the M-CSF receptor promoter (12). We conclude that disruption of promoter-specific core binding factor complexes is likely to affect the growth and differentiation pathways of hematopoietic cells at discrete stages, leading to distinct leukemic phenotypes.