Previous Article | Next Article ![]()
Molecular and Cellular Biology, May 2006, p. 3902-3916, Vol. 26, No. 10
0270-7306/06/$08.00+0 doi:10.1128/MCB.26.10.3902-3916.2006
Copyright © 2006, American Society for Microbiology. All Rights Reserved.
Department of Pathology and Molecular Pathology Graduate Program, University of California at San Diego School of Medicine, 9500 Gilman Drive, La Jolla, California 92093,1 Biomedical Sciences Graduate Program, University of California at San Diego School of Medicine, 9500 Gilman Drive, La Jolla, California 920932
Received 16 September 2005/ Returned for modification 16 November 2005/ Accepted 21 February 2006
| ABSTRACT |
|---|
|
|
|---|
| INTRODUCTION |
|---|
|
|
|---|
The functions of Meis, Pbx, and Hox proteins are usurped in leukemogenesis. Originally, Meis1 was discovered as one of two genes activated by proviral integration in BXH2 murine acute myeloid leukemia (AML), the second being Hoxa9 or Hoxa7 (32, 33). Over 50% of human AML and virtually all acute lymphoblastic leukemias containing translocations of the mixed lineage leukemia (MLL) gene exhibit strong expression of MEIS1, HOXA9, and HOXA7 (2, 17, 30). In murine AML models, Meis1 dramatically accelerates leukemia initiated by Hoxa9, Hoxb6, or Hoxb3 (reduces latency from
200 days to
60 days) yet exhibits no independent ability to cause disease by itself (13, 28, 46). Dominant activating forms of the MEIS1-interacting protein PBX1 [E2A-PBX1, produced by the t(1;19) translocation in human pre-B leukemia (24)] and HOXA9 [NUP98-HOXA9, produced by the t(7;11) of human AML (8)] also induce AML in mouse models, suggesting that activation of the same subset of Meis1-Pbx-Hox target genes may underlie leukemic transformation.
Hox and Meis1 have distinct roles in progenitor immortalization and in regulating gene expression. Hoxa9 blocks differentiation of cultured myeloid progenitors that do not express endogenous Meis1/2/3 and do not cause leukemia when transplanted into syngeneic recipients (7, 47, 49). Coexpression of Meis1 plus Hoxa9 in cultured progenitors immortalizes an earlier myeloid progenitor that induces AML in syngeneic recipients. The leukemia stem cell (LSC) phenotype induced by Meis1 is paralleled expression of a set of leukemia-associated genes, including Cd34, Flt3, and Erg, which have been therefore defined as Meis1-related leukemic signature genes, and we hypothesized that these Meis1 signature genes are important for LSCs to survive and expand in marrow stem cell niches (49). Whether the expression of Meis1-related signature genes requires Hoxa9 function is unknown, but it is now strongly implicated by our results presented below.
Equally unclear is the type of transcriptional activity that Meis1-Pbx complexes use to induce the LSC phenotype. Whereas Pbx can mediate repression through its recruitment of corepressors such as histone deacetylases and mSin3, Meis1 contains a C-terminal domain (CTD) that can promote transactivation in response to cell signals such as those induced by protein kinase A (16, 40). By interacting with both Hox and non-Hox factors (e.g., MyoD [6] and Oct1 [38]), Meis1-Pbx complexes contribute to both gene activation (e.g., of Myogenin [6], Hoxb1 [16], Hoxb2 [21], and Pax6 [51]) and gene repression (e.g., of CYBB [5]). It is not clear whether activation, repression, or both functions mediate the role of Meis1 in leukemia.
Here, we ask whether activation or repression by Meis1 plays the more important role in myeloid leukemogenesis by fusing Meis1 to a dominant activation domain of Vp16 or a dominant repression domain of engrailed. This same strategy was used to identify transcriptional functions of the Drosophila melanogaster Meis1 homologue, Homothorax, that contribute to embryonic development (18). Strikingly, in the absence of coexpressed Hoxa9, Vp16-Meis1 alone was able to immortalize early hematopoietic progenitors in culture that caused myeloid leukemia in syngeneic mice. Neither the immortalized progenitors nor the leukemia cells derived from them expressed endogenous Hox genes. The transforming ability of Vp16-Meis1 required interaction with Pbx and binding to DNA but not the function of the Meis1 CTD, which is required for cooperative leukemogenesis by wild-type Meis1 with Hoxa9. Vp16-Meis1 progenitors expressed low to moderate levels of the same Meis1-related leukemic signature genes (e.g., Flt3, Cd34, Erg, and Msi2h). Surprisingly, retroviral expression of Hoxa9 or Hoxa7 in Vp16-Meis1-expressing progenitors strongly up-regulated these Meis1-associated signature genes, promoted cell division, and reduced disease latency to less than 30 days. This is the first evidence that Hox proteins contribute to transcriptional activation of Meis1-related signature genes. Hoxa7, which also cooperates with Meis1 to induce AML, induced the same cellular and genetic changes as Hoxa9. Strikingly, the N-terminal domain (NTD) of Hoxa9 (residues 1 to 138), which is essential for its transforming properties with Meis1, was unnecessary for promoting proliferation and transcription of Meis1 signature genes (such as Flt3) in Vp16-Meis1 progenitors. Taken together, these results indicate that transactivation represents the key function of Meis1 that contributes to leukemogenesis and that Meis, Hox, and Pbx proteins form a common regulatory complex that activates transcription through mechanisms dependent on the Hoxa9 NTD and the Meis1 CTD, functions that can be replaced by the Vp16 domain fused onto Meis1. Consistent with this hypothesis, chromatin immunoprecipitation (ChIP) revealed that Hoxa9 and Meis1 were both recruited to the Flt3 promoter.
| MATERIALS AND METHODS |
|---|
|
|
|---|
Infection and culture of hematopoietic progenitors. Protocols for retroviral infection and culture of primary hematopoietic progenitors in granulocyte-macrophage colony-stimulating factor (GM-CSF), stem cell factor (SCF), or Flt3 ligand (FL) were described previously (7, 49). Briefly, 2 x 105 to 4 x 105 of enriched Sca+ Lin bone marrow progenitors from BALB/c mice were subjected to a single round of spinoculation-infection using 1 ml of retroviral supernatant (2 x 105 to 4 x 105/ml), followed by 3 days of drug resistance selection (0.6 to 1 mg/ml of G418 for MSCV-Neo or 1 µg/ml of puromycin for MSCV-Puro [drug selection is optional]), and cultured in medium containing either GM-SCF, SCF, or FL. Proliferation kinetics were evaluated by plating 2 x 105 to 4 x 105 drug-resistant Lin progenitors in a 12-well tissue culture plate (Corning, New York) and splitting every 3 to 4 days in fresh medium, using a protocol that keeps the progenitor number lower than 2 x 106 per well. For coexpression of Meis1 with Hoxa9 mutants, equal amounts of retrovirus encoding Flag-Meis1 (MSCV-Puro) and Hoxa9 (MSCV-Neo) were combined and used for infection, followed by dual drug selection and cultivation in SCF or FL. Wright-Giemsa staining and fluorescence-activated cell sorter (FACS) analysis were performed as described elsewhere (49).
Antibodies and immunoblotting. Ten microliters of total cellular protein (107 cells per ml) was used for immunoblotting. Anti-Flag, anti-Flt3, and anti-Hoxa9 were obtained and used as described previously (49). Anti-Pbx was obtained from Santa Cruz Biotechnology (C-20; sc-888).
Luciferase reporter assays. Cooperative transactivation by Vp16-Meis1 and Pbx1a was evaluated using 7xPRS-pGL3, a pGL3 luciferase derivative driven by seven tandem repeats of the Pbx1-Meis1-responsive sequence (PRS), TGATTGAT, which was first identified as an E2a-PBX1 binding site (27) and later demonstrated to be recognized by Meis1 plus Pbx1 (26). One microgram of 7xPRS-pGL3 was cotransfected with 1 µg of MSCV-Vp16-Meis1, 0.5 µg of pcDNA3-Pbx1a, and 0.2 µg of pRL-TK (internal control that expresses Renilla luciferase persistently) into 293T cells. At 48 h after transfection, cell lysate was prepared and quantified for firefly and Renilla luciferase activity using a dual luciferase reporter system (Promega) and LMaxII luminometer (Molecular Devices, California). Luciferase measurements were calculated as firefly luciferase units versus Renilla luciferase units. Transcriptional activation was compared to basal luciferase levels in cells cotransfected with 7xPRS-pGL3, empty MSCV, and Pbx1a.
Semiquantitative RT-PCR and real-time qPCR. Total RNA was extracted (49) from immortalized progenitors 6 to 8 weeks post-retroviral infection, when normal progenitors are no longer present and when immortalized progenitors exhibit stable proliferation, stable expression of surface markers, and little to no differentiation. First-strand total cDNA, used as a template for reverse transcription-PCR (RT-PCR) and quantitative PCR (qPCR), was generated with the SuperScript reverse transcription system (Invitrogen) by using 4 µg of total RNA and 50 ng of random hexamers. Real-time qPCR was performed using 4 µl of 1:10-diluted cDNA as a template, 400 nM of primers, and SYBR Green PCR master mix (Applied Biosystems) in a 15-µl reaction volume and an Mx3000P real-time PCR system (Stratagene). The housekeeping gene Gapdh was used as control for mass of cDNA. The specificity and relative signal intensity of real-time qPCR amplifications were analyzed according to the manufacturer's specifications (Stratagene). Primers covered at least one intron-exon boundary. Semiquantitative RT-PCR was performed as described elsewhere (49). The sequences of primers used for each murine gene were as follows: for Gapdh, 5'-ATGACATCAAGAAGGTGGTGAAG and 5'-TCCTTGGAGGCCATGTAGG; for Hoxa9, 5'-TGGTTCTCCTCCAGTTGATAG and 5'-AGAAACTCCTTCTCCAGTTCC; for Hoxa7, 5'-CGGGCTTATACAATGTCAACAG and 5'-AAATGGAATTCCTTCTCCAGTTC; for Hoxa5, 5'-GCAAGCTGCACATTAGTCAC and 5'-GCATGAGCTATTTCGATCCT; for Hoxa10, 5'-GGAAGGAGCGAGTCCTAGA and 5'-TTCACTTGTCTGTCCGTGAG; for Flt3, 5'-AAATCCCAGAAGGAGTTTGG and 5'-CGCATAGAAGGAGATGTTGTC; for Erg, 5'-GACCACAAATGAGCGCAGAG and 5'-CTTTGGACTGAGGGGTGAGG; for Cd34, 5'-ACAGCAGTAAGACCACACCAGC and 5'-AGACACTAGCACCAGCATCAGC; for Il7r, 5'-CTGACCTGAAAGTCGTTTATCGC and 5'-CATCCTCCTTGATTCTTGGGTTC; and for Msi2h, 5'-ACGATTGACCCAAAAGTTGC and 5'-ACAAAGCCAAACCCTCTGTG.
In vivo leukemogenesis assay. The leukemic potential of oncogenes was evaluated in sublethally irradiated syngeneic BALB/c mice (450-rad dosage) followed by tail vein injection with 2 x 106 immortalized progenitors cultured ex vivo for 4 to 8 weeks (49) or with 5 x 105 ScaI+ Lin progenitors that were prestimulated in a cytokine cocktail containing murine SCF (1:100 dilution of culture supernatant of SCF-secreting cell lines), 10 ng/ml of murine interleukin-3 (IL-3; PeproTech Inc., New Jersey), 10 ng/ml of IL-6 (PeproTech), IL-7 (1:100 dilution of culture supernatant of IL-7-secreting cell lines), and 5 ng/ml FL (Sigma); infected with retrovirus; subjected to drug resistance selection for 2 days; and cultured for 2 more days in nonselection medium containing the same cytokine cocktail. SCF- or IL-7-containing supernatant from cytokine-secreting cell lines (SCF-secreting CHO cells and IL-7-secreting J558 cells) was harvested every 24 h as 50 ml of total medium from 15-cm plates of highly confluent cells (45). Mice exhibiting a leukemia phenotype (scruffy fur, lethargy, paralysis, or splenomegaly) were sacrificed, and cells isolated from leukemic tissues were subjected to further analysis, including flow cytometry, Wright-Giemsa staining, secondary injection, immunoblotting, retroviral integration site analysis by Southern blotting, and RT-PCR (45).
Affymetrix microarray analysis. Total RNA from 5 x 107 to 10 x 107 progenitors cultured in SCF was purified using a Maxi RNA isolation kit (QIAGEN). Gene expression profiling was examined by using the Affymetrix GeneChip Mouse Genome 430 2.0 array probed with total RNA (MicroArray Core Facility, UCSD-VA Medical Center). After normalization of overall signal intensities among different hybridization experiments, levels of gene expression were calculated using d-CHIP software, taking into consideration both the perfect match (PM) and mismatch (MM) values for all 11 sets of oligonucleotide probes representing each gene (49). At signal intensities of less than 50, the PM versus MM ratio approached unity; therefore, any signal below 50 units was designated background. Because computation analysis methods will include false positives selected due to high-level expression indicated by only one or a few of the 11 pairs of diagnostic oligonucleotides, all d-CHIP results were confirmed by visual inspection of normalized chip raw data, and a subset of genes was confirmed by semiquantitative or real-time PCR.
Two approaches were used to characterize leukemogenesis genes in Vp16-Meis1 progenitors: (i) identification of genes up-regulated threefold or greater in Vp16-Meis1 progenitors by comparison to nonleukemic Hoxa9 progenitors, and (ii) identification of genes up-regulated threefold or greater by Hoxa9 in Vp16-Meis1 progenitors. Among 39,000 genes represented on the array, only 24 genes were up-regulated >3-fold in three of the Vp16-Meis1 progenitor lines by comparison to five of the Hoxa9 progenitor lines, and only 77 genes (which included 22 of those 24 genes induced by Vp16-Meis1) were up-regulated >3-fold by expression of Hoxa9 in each of the Vp16-Meis1 progenitor lines, demonstrating that in both cases, the transcriptional impact of the oncoprotein is specific and limited. Among those 24 genes induced by Vp16-Meis1 were 16 of the original 25 genes we reported as Meis1-related signature genes that were expressed specifically in leukemogenic Hoxa9-plus-Meis1 progenitors but not in Hoxa9 progenitors (i.e., Cd34, Flt3, Il7r, Crlr, Erg, Gpr56, Tilz1b, Nrip1/RIP140, C1qr1, Tmsb10, Kcna3, Ngn, Ptprcap/CD45AP, Satb1, Ngn, and Ly86/MD-1) (see Table 3, below). The other nine of the original Meis1 signature genes, expressed at extremely low levels in our original analysis of Hoxa9-plus-Meis1 progenitors, have been eliminated as candidate leukemogenesis genes based on their inconsistent activation in additional cell lines immortalized by Hoxa9 plus Meis1, lack of their expression in Vp16-Meis1 progenitors, and lack of activation by Hoxa9 (Tspan2, CD244, Gem, PLZF, Cd27, Itgb7, Oaf, CXXC5, and Ebf). The eight new genes activated by Vp16Meis1 (Msi2h, Sox4, Dbx4, P2x, Bcl2, and Sesn3, plus two unknown expressed sequence tags) were also expressed at significantly higher levels in Hoxa9-plus-Meis1 progenitors than in Hoxa9 progenitors and were added to the genes shown below in Table 3. The remaining 54 genes that were activated by Hoxa9 in Vp16-Meis1 progenitors (except c-kit, listed below in Table 3) were not expressed in Hoxa9 progenitors or Hoxa9-plus-Meis1 progenitors; therefore, they may be artificial targets of Hoxa9 in the context of Vp16-fused Meis1 or they may be real targets under different cellular contexts that were mimicked/activated by Vp16-Meis1. However, their activation is physiologically irrelevant to leukemogenesis by Hoxa9 plus Meis1 (they are not included in Table 3; information is available upon request).
|
EMSA. Heterodimerization of Hoxa9 with either Pbx1 or Meis1 on DNA was examined by electrophoretic mobility shift assay (EMSA), as described previously (7, 43).
ChIP analysis. ChIP analysis was performed according to a modified protocol based on ChIP protocols of Upstate Biotechnology (New York) and Active Motif (California). Briefly, 2 x 107 to 3 x 107 myeloid progenitors were collected and subjected to DNA-protein cross-linking in 10 ml of phosphate-buffered saline plus 1% formaldehyde for 10 to 15 min at room temperature, followed by a 5-minute treatment in 5 ml of 0.125 M glycine (supplemented with 1x Complete protease inhibitor cocktail [PIC; Roche]) to stop the cross-linking reaction. After washing with cold phosphate-buffered saline-1x PIC, the cell pellet was suspended in 200 µl of sodium dodecyl sulfate lysis buffer (Upstate; 1x PIC added), incubated on ice for 15 min, and then subjected to three repeats of 12- to 13-second sonication (Sonifier 450 [VMR Scientific]). Sheared DNA was 400 bp to 1.5 kbp. The sheared sample was centrifuged at 14,000 rpm for 15 min at 4°C, and the supernatant was diluted 1:10 with ChIP dilution buffer (Upstate; 1x PIC added). The chromatin was immuno-precleared with 60 µl salmon sperm DNA-protein A-agarose (Upstate) per 2 ml chromatin, followed by a 1-h rotation (4°C) and removal of the agarose by 2 min of centrifugation at 1,500 rpm. Five hundred microliters of chromatin was used for overnight immunoprecipitation by control antibody (immunoglobulin G [IgG]) or specific antibody with rotation at 4°C. Sixty microliters of protein A-agarose was added to each tube followed by rotation for 2 h and sequential washing with low-salt wash buffer, high-salt wash buffer, LiCl wash buffer, and Tris-EDTA buffer as described elsewhere (Upstate; 1x PIC added to all buffers). Chromatin was eluted twice with 100 µl of freshly made 1% sodium dodecyl sulfate plus 0.1 M NaHCO3 with 15 min of vigorous shaking on a vortex mixer. A 200-µl aliquot of eluted chromatin-protein was subjected to reverse cross-linking, removal of RNA (by addition of 8 µl of 5 M NaCl and 1 µl of RNase A [10 µg/µl] and incubation at 65°C for 6 to 16 h), and then removal of protein (by subsequent addition of 4 µl of 0.5 M EDTA, 8 µl of 1 M Tris-HCl [pH 6.5], and 1 µl of 20-mg/ml protease K, with incubation at 45°C for 1 to 2 h). DNA was recovered using a QIAQuick spin column and suspended in 100 µl of elution buffer (QIAGEN). PCR amplification was performed using 1 U of Platinum Taq polymerase (Invitrogen) and 2 to 4 µl of eluted DNA.
Antibodies for ChIP (amount added per 500 µl of chromatin) included M2 anti-Flag (8 µg; Sigma), anti-Hoxa9 HD (10 µg), anti-Meis1 HD (10 µg), anti-Meis1 N terminal (10 µg; gift of Mark Featherstone), anti-acetyl-histone H3 (1 µg; Upstate 06-599) and anti-dimethylated histone H3 Lys4 (anti-H3K4Me2 [1 µg]; Upstate 07-030).
The murine Flt3 genomic structure was deduced using information provided by the UCSC Genome Browser (http://genome.ucsc.edu), and the putative transcription start site was defined using in silico mapping of common 5' expressed sequence tag sequences onto the genomic sequence. The ChIP primers used to amplify regions of the Flt3 locus were P3 (5'-GGACAAATGGCTCAGGAGGG and 5'-CATCTGGTAGATCTGCATTGTGG), P4 (5'-CTAGATGTGGCTCTGGGAACC and 5'-AGACAGATAGACAGAGTGCCAGC), P5 (5'-AGTCAGAAGGGACTGGCTCC and 5'-GAGTGCTGCTTAGCAGATTACC), and P6 (5'-CCTCAGAAGTGAACTCAGTTTCC and 5'-CTGCATCCAGACCATGAAGG).
| RESULTS |
|---|
|
|
|---|
|
|
Vp16-Meis1 immortalizes hematopoietic progenitors in the absence of coexpressed Hox genes. To identify the transcriptional function of Meis1 that is important for cooperation with Hoxa9 and Hoxa7 in establishing the leukemic stem cell phenotype, dominant transactivating or transrepressing forms of Meis1 were produced by fusing it to the activation domain of herpes simplex viral protein Vp16 or to the repression domain of Drosophila engrailed (en) (Fig. 2A). Sca1+ Lin-enriched marrow progenitors were infected with retrovirus encoding empty vector, Meis1, Vp16-Meis1, or en-Meis1 and cultured in medium containing SCF or FL as the sole cytokine. With either cytokine, expression of Vp16-Meis1 itself was able to immortalize progenitors, while cultures infected with Meis1 or en-Meis1 retrovirus followed the same decline in progenitors as uninfected cultures (Fig. 2B). Vp16-Meis1 had no effect on NIH 3T3 fibroblasts, indicating that the differentiation arrest phenotype was not accompanied by a cell proliferation phenotype, as is the case with E2A-PBX1, a transcriptionally activated form of the MEIS1 partner PBX1 (25). To test whether immortalization by Vp16-Meis1 might be mediated by expression of endogenous Hox genes, the expression of Hoxa5, Hoxa7, Hoxa9, and Hoxa10 was measured by semiquantitative RT-PCR (Fig. 2C) and quantitative real-time PCR (data not shown, except for Hoxa9 in Fig. 4D, below). None of these four Hox genes was expressed in two of the Vp16-Meis1-immortalized progenitors, while they were all expressed strongly in progenitors immortalized by the Hox locus transactivators MLL-ENL (50) or Nup98-Hoxa9 (8). A global view of Hox gene expression assessed by using Affymetrix mouse gene arrays demonstrated no expression of any of 31 Hox genes present on the array chip (Hoxa1 to -5, -a7, -a10, -a11, -a13, -b1, -b3 to -9, -b13, -c4 to -6, -c8, -c9, -c13, -d1, -d3, -d4, or -d10 to -13) among five independent Vp16-Meis1 progenitor lines (19), while each of two Nup98-Hoxa9-immortalized progenitor lines exhibited strong expression of eight Hox genes (Hoxa5, -a7, -a9, -a10, -c4 to -6, and -c8). Therefore, Hox genes are not expressed in Vp16-Meis1 progenitors. Affymetrix arrays also revealed strong expression of Pbx1a, weak expression of Pbx2a, and no expression of Pbx1b, Pbx2b, or Pbx3 in Vp16-Meis1 progenitors (Fig. 2E). Pbx1a expression was verified by Western blotting (Fig. 2D). Vp16-Meis1 progenitors were cytokine dependent, proliferating as progenitors in SCF or FL, and a subset (5 to 25%) proliferated as undifferentiated progenitors in IL-7, GM-CSF, or G-CSF (Fig. 2F and G), a behavior similar to progenitors coexpressing Hoxa9 plus Meis1 (49). Thus, Vp16-Meis1 mimics the combined activities of Hoxa9 plus Meis1, immortalizing progenitors capable of responding to both lymphoid and myeloid cytokines. The ability of a dominant activating form of Meis1 to replace the immortalizing function of Hox proteins raises the possibility that Meis1 binds the same promoters as Hoxa9 and Hoxa7 and that the histone acetyltransferase (HAT) activity recruited by Vp16 (31) replaces an intrinsic Hox-dependent activation function.
|
|
64-202), a five-residue neutral mutation in the M2 domain (M2
LRF/LELL) that disrupts heterodimerization with Pbx1 (49), or an HD mutation (HDN51S/
RRR) that virtually eliminates binding of Meis1-Pbx1 heterodimers to DNA (Fig. 3A) each reduced activation of a reporter construct driven by Pbx1-Meis1 binding motifs to levels 20% or less of those induced by wild-type Vp16-Meis1 (Fig. 3C), and each of these mutants was incapable of immortalizing progenitors in either SCF or FL (Fig. 3B). By contrast, an N-terminal deletion preceding the Pbx interaction domain (
1-64), a three-residue neutral mutation in M2 (M2
LELL), and an internal deletion between the M2 domain and HD (
202-260), all of which retained active Pbx-Meis1 heterodimerization potential on DNA (49), maintained more robust activation of the reporter construct and immortalized FL-responsive progenitors actively (Fig. 3B and C). The Meis1 CTD, which is essential for the ability of wild-type Meis1 to immortalize Flt3+ progenitors in cooperation with Hoxa9, was dispensable for transactivation by Vp16-Meis1 and was also unnecessary for Vp16-Meis1-mediated immortalization of Flt3+ progenitors. Immunoblot analysis confirmed expression of all retrovirus-encoded Vp16-Meis1 proteins in infected progenitors subjected to drug selection (Fig. 3D). The fact that the Vp16 transactivation domain can replace the function of the Meis1 CTD in immortalization of FL-responsive progenitors (16) supports the notion that gene activation is the essential biochemical function performed by the Meis1 CTD in establishing leukemia with Hoxa9.
|
|
LRF/LELL, did not develop leukemia over 370 days. Leukemia cells were factor dependent, expressed retroviral Vp16-Meis1 (Fig. 4C), and contained the same retroviral integration sites as those within the oligoclonal parental progenitors (Fig. 4E), indicating that they arose from injected cells. FACS analysis demonstrated that the majority of leukemic cells were myeloid (Mac-I+ B220 Cd19 Thy1.2 c-Kit+) (Fig. 4F and Table 2). Mice exhibited two forms of disease. The first (10 of 18 examined animals) was AML, in which 30 to 95% of leukemic cells stained as immature myeloblasts. The second (in 8 out of 18) was a myeloproliferative disease (MPD) in which 70 to 90% of leukemia cells exhibited significant neutrophil maturation (Table 2 and Fig. 4B). In both diseases, mice presented with 3- to 12-fold enlarged spleens, 10- to 50-fold enlarged lymph nodes, and 4- to 40-fold increases in circulating myeloid cells (Table 2). Comparable amounts of Vp16-Meis1 protein were present in myeloid cells from both diseases (data not shown), discounting the possibility of a reactive granulocytosis. Over 50% of leukemic progenitors expressed Flt3 (FACS in Fig. 4F and Table 2 and immunoblot in Fig. 4C). Thus, Vp16-Meis1 causes myeloid leukemia with 100% penetrance, and leukemic myeloblasts retain a low level of Flt3 and Cd34 expression, similar to the parental cells.
|
3% of the Hoxa9 level expressed in MLL-ENL progenitors) (sample sc1.2), also expressed Hoxa5, Hoxa7, and Hoxa10 at levels 1 to 5% of that expressed in progenitors immortalized by MLL-ENL. Thus, the large majority of myeloid leukemias induced by Vp16-Meis1 did not acquire activation of endogenous Hoxa7 or Hoxa9 genes as secondary cooperating events. Expression of Hoxa9 or Hoxa7 in Vp16-Meis1 progenitors induces further activation of Meis1 signature genes, stimulates proliferation, and accelerates disease progression to the level of bona fide AML myeloblasts. An important unresolved question concerning the mechanism of cooperativity by Hox plus Meis1 is whether Hoxa7 or Hoxa9 assists Meis1 in the activation of Meis1-related signature genes or Meis1 functions independent of coexpressed Hox proteins. Because Vp16-Meis1 progenitors expressed low levels of Meis1-related signature genes and expressed no endogenous Hox genes, they represented a good cell model to test whether Hoxa9 or Hoxa7 could further activate expression of Meis1-related signature genes. Strikingly, retroviral expression of Hoxa9 or Hoxa7 in Vp16-Meis1 progenitors induced strong proliferation (Fig. 5B), strong up-regulation of Flt3 and Cd34 proteins (Fig. 3F and 5A), and 3- to 20-fold activation of other Meis1 signature genes (Table 3 and Fig. 5D). The reliability of the Affymetrix array analysis was verified using semiquantitative RT-PCR (data not shown) and real-time qPCR (Fig. 5E). Six newly identified genes (Msi2h, Sox4, Dbx4, and P2rx3, plus two unknown genes) were activated by Hoxa9 in Vp16-Meis1 progenitors and expressed at much higher levels in Hoxa9-plus-Meis1 progenitors than in Hoxa9 progenitors, and they have been added to our list of Meis1 signature genes (Table 3).
Vp16-Meis1 progenitors expressing Hoxa9 or Hoxa7 contained the same Vp16-Meis1 retroviral integration site as parental progenitors (Fig. 5C), ruling out the possibility that these dramatic transcriptional changes arose from selective expansion of an unrelated Cd34+ Flt3+ stem cell subpopulation that might still exist in cultures. Expression of Hoxa9 in Vp16-Meis1 progenitors reduced AML latency to an average of 24 days (Fig. 4A) and generated AML that was more similar to the myeloblastic phenotype (Fig. 4B, bottom panels), comprised of Lin and MacI+ B220 subpopulations with high levels of Flt3 and Cd34 (Fig. 4F, bottom panels). This suggests that Hoxa9 and Hoxa7 participate in gene coactivation with Meis1.
The Hoxa9 N-terminal domain, which is essential for all transforming functions of Hoxa9, is dispensable for activating Flt3 expression and transforming properties in Vp16-Meis1-expressing progenitors, suggesting that it performs an activation function replaced by Vp16 on promoters co-occupied by Vp16-Meis1 and Hoxa9.
The fact that fusion of Vp16 to Meis1 replaces Hox function, combined with the fact that both Meis1/Vp16-Meis1 and Hoxa9/a7 activate transcription of the same Meis1-related signature genes, favors a model in which Meis1, Hox, and Pbx directly bind these signature gene promoters that mediate leukemogenesis. If this model were correct, then Hoxa9 transactivation domains that are essential for activation of the Meis1 signature genes in the context of wild-type Meis1 might be dispensable in the context of Vp16-Meis1. Indeed, the Hoxa9 NTD (residues 1 to 138), which has been reported to contain a transactivation domain and Meis1 interaction motif (41), which is essential for immortalization by Hoxa9 alone in GM-CSF (Fig. 6A) or in SCF (not shown), is essential for immortalization of SCF- or FL-dependent progenitors with Meis1 (Fig. 6B), and is essential for Meis1-dependent production of Flt3 in infected ScaI+ Lin marrow progenitors (Fig. 6C, lanes 2 to 5 versus lane 1), was dispensable for Hoxa9-induced proliferation (Fig. 5B), for Flt3 up-regulation (Fig. 6F, lane 4 versus lane 2) in Vp16-Meis1 progenitors, or for increasing the aggressiveness of leukemias produced by Vp16-Meis1 progenitors (Fig. 6G). By contrast, the DNA binding and Pbx interaction mutants Hoxa9-HDN51S and Hoxa9-PIMAAAA (7) exhibited no activation or minimal activation (25% of wild type) of Flt3 transcription (Fig. 6F, lanes 7 and 8 versus lane 2). Hoxa9
1-138 retained its ability to interact with Pbx1 on DNA (Fig. 6D, lane 3 versus lane 4) and lost its heterodimerization abilities with Meis1 on DNA (Fig. 6E, lane 4 versus lane 8). The fact that the NTD of Hoxa9 is dispensable for Hoxa9 transforming functions in Vp16-Meis1 progenitors suggests that its transactivation function is replaced by the persistent transactivation function of a Vp16-Meis1 complex bound to an adjacent DNA element.
|
|
| DISCUSSION |
|---|
|
|
|---|
Four transcriptional principles related to Meis1 and its interacting cofactors in hematopoietic progenitors are suggested by our analysis. First, the fact that Vp16-Meis1-Pbx complexes enforce self-renewal of leukemic stem cells in vivo suggests that Meis1-Pbx complexes regulate progenitor expansion during normal hematopoiesis by activating target gene transcription. Second, the fact that Hoxa7 and Hoxa9 augment expression of Meis1 signature genes in Vp16-Meis1 progenitors suggests that Hox proteins actively participate in transactivation of Meis1-related signature genes in normal hematopoiesis. Prior to this report, there was no evidence that Hoxa7 or Hoxa9 controlled expression of Meis1 target genes in leukemia. Third, the mutational analysis of Hoxa9 and Meis1 domains critical for their functions in Vp16-Meis1 progenitors suggests a model in which all three homeodomain proteinsPbx1/2, Meis1, and Hoxa7/a9cooperate on single promoters critical for establishing the leukemogenic stem cell phenotype (Fig. 8). Specifically, the Meis1 CTD and the Hoxa9 NTD may recruit cofactors harboring HAT activity (Fig. 8), functions replaced by the Vp16 domain in Vp16-Meis1. Fourth, the ability of the C-terminal half of Hoxa9 to activate transcription of Flt3 in Vp16-Meis1 progenitors in a manner as robust as that of wild-type Hoxa9 suggests that this region possesses an additional transcriptional function which might be, for example, recruiting histone methyltransferase or chromatin remodeling factors (Fig. 8) that complements those provided by the Meis1 CTD and Hoxa9 NTD. The biochemical activities responsible for the N-terminal and C-terminal transactivating functions of Hoxa9 remain to be determined. This hypothesis, which is based on the genetic and cellular properties of leukemic progenitors, is consistent with physical interaction studies, which also found Meis1, Hoxa9, and Pbx2 in single nuclear complexes in myeloid cells (44), and reporter studies, which delineated cooperating functions of Meis1, Pbx1, and Hoxa1 proteins on regulation of the Hoxb1/b2 enhancer (16, 21).
|
1-138) suggest that it interacts with Pbx-Meis1 complexes via Pbx on target promoters as others have hypothesized previously (21). The promoter co-occupancy model would also account for leukemic cooperativity between Meis1 and Hoxb6/b3, which are also not predicted to bind Meis1 directly. We predict that coexpression of Meis1 and Hoxb6/b3 in SCF-dependent immortalized progenitors would also result in transcriptional activation of Flt3 and other Meis1 signature genes. The Hox/Meis1-related signature genes that are coactivated by Meis1 plus Hox are involved in normal HSC biology and implicated in leukemogenesis. Cd34 and Flt3 are prominent HSC markers (1). Flt3 catalytic function is activated by mutations in over one-third of human AML, and we propose that Flt3 expression permits expansion of LSCs in response to FL in stem cell niches. Gpr56 (20), Msi2h (34), Erg (48), C1qR1 (10, 48), and Wbscr5 (48) are enriched in HSC or other tissue-type stem cells. The Msi2h-related gene Msi1 is crucial for maintenance of neuronal stem cells (39). The Ets-related transcription factor Erg is fused to TLS/FUS and EWS in leukemia and Ewing sarcoma, respectively (14, 35). Il7r is important for lymphoid progenitor expansion, and Gpr56 is overexpressed in brain tumors and mutated in a genetic disorder involving brain cortical malformation (36, 42). It is possible that up-regulation of the Grb2-interacting adaptor Wbscr5/LAB (22) contributes to specific signaling pathways in HSCs. Understanding functions of these genes will cast additional light on how they contribute to the LSC phenotype.
Previous studies reported over 200 genes regulated by Hoxa9 in leukemia cell lines or in CD34+ HSCs (11, 12), but with the exception of Erg, these genes do not include the Hox/Meis1-activated signature genes examined herein. It is possible that studies using leukemia cell lines do not identify these signature genes because such genes are already expressed at high basal levels in the control parental progenitors. Alternatively, these Hox/Meis1 signature genes may not be activated by introduction of exogenous Hoxa9 if progenitors do not express essential cofactors, such as Meis1. Indeed, Hoxa9 alone does not activate any of the signature genes, as exhibited by Hoxa9-immortalized progenitors. When exogenous Hoxa9 is expressed in normal CD34+ HSCs, the expression of endogenous Hox and Meis1 may preclude significant further activation of these Hox/Meis1 signature genes. The heterogeneity of this population may also dilute the strong response of responsive subsets that express required cofactors. By contrast, because Vp16Meis1 progenitors lack Hox gene expression, they may be particularly sensitive to transcriptional regulatory mechanisms induced by Hox and, because they are uniform populations, the impact of Hox genes is maximized. Thus, Vp16-Meis1 progenitors may represent a unique tool for determining both the specificity and biochemical mechanisms of Hoxa7 and Hoxa9 gene activation in AML.
| ACKNOWLEDGMENTS |
|---|
We thank Dennis Young for assistance with FACS analysis and Joe Aguilera for assistance with mouse radiation (UCSD Cancer Center Shared Resources). We give special thanks to Mark Featherstone, Corey Largman, Robert Slany, and Adi Salzberg for providing reagents for work related to this paper.
| FOOTNOTES |
|---|
| REFERENCES |
|---|
|
|
|---|
2. Afonja, O., J. E. Smith, Jr., D. M. Cheng, A. S. Goldenberg, E. Amorosi, T. Shimamoto, S. Nakamura, K. Ohyashiki, J. Ohyashiki, K. Toyama, and K. Takeshita. 2000. MEIS1 and HOXA7 genes in human acute myeloid leukemia. Leuk. Res. 24:849-855.[CrossRef][Medline]
3. Ayton, P. M., and M. L. Cleary. 2003. Transformation of myeloid progenitors by MLL oncoproteins is dependent on Hoxa7 and Hoxa9. Genes Dev. 17:2298-2307.
4. Azcoitia, V., M. Aracil, A. C. Martinez, and M. Torres. 2005. The homeodomain protein Meis1 is essential for definitive hematopoiesis and vascular patterning in the mouse embryo. Dev. Biol. 280:307-320.[CrossRef][Medline]
5. Bei, L., Y. Lu, and E. A. Eklund. 2005. HOXA9 activates transcription of the gene encoding gp91Phox during myeloid differentiation. J. Biol. Chem. 280:12359-12370.
6. Berkes, C. A., D. A. Bergstrom, B. H. Penn, K. J. Seaver, P. S. Knoepfler, and S. J. Tapscott. 2004. Pbx marks genes for activation by MyoD indicating a role for a homeodomain protein in establishing myogenic potential. Mol. Cell 14:465-477.[CrossRef][Medline]
7. Calvo, K. R., D. B. Sykes, M. Pasillas, and M. P. Kamps. 2000. Hoxa9 immortalizes a granulocyte-macrophage colony-stimulating factor-dependent promyelocyte capable of biphenotypic differentiation to neutrophils or macrophages, independent of enforced Meis expression. Mol. Cell. Biol. 20:3274-3285.
8. Calvo, K. R., D. B. Sykes, M. P. Pasillas, and M. P. Kamps. 2002. Nup98-HoxA9 immortalizes myeloid progenitors, enforces expression of Hoxa9, Hoxa7, and Meis1, and alters cytokine-specific responses in a manner similar to that induced by retroviral co-expression of Hoxa9 and Meis1. Oncogene 21:4247-4256.[CrossRef][Medline]
9. Chang, C. P., Y. Jacobs, T. Nakamura, N. A. Jenkins, N. G. Copeland, and M. L. Cleary. 1997. Meis proteins are major in vivo DNA binding partners for wild-type but not chimeric Pbx proteins. Mol. Cell. Biol. 17:5679-5687.[Abstract]
10. Danet, G. H., J. L. Luongo, G. Butler, M. M. Lu, A. J. Tenner, M. C. Simon, and D. A. Bonnet. 2002. C1qRp defines a new human stem cell population with hematopoietic and hepatic potential. Proc. Natl. Acad. Sci. USA 99:10441-10445.
11. Dorsam, S. T., C. M. Ferrell, G. P. Dorsam, M. K. Derynck, U. Vijapurkar, D. Khodabakhsh, B. Pau, H. Bernstein, C. M. Haqq, C. Largman, and H. J. Lawrence. 2004. The transcriptome of the leukemogenic homeoprotein HOXA9 in human hematopoietic cells. Blood 103:1676-1684.
12. Ferrell, C. M., S. T. Dorsam, H. Ohta, R. K. Humphries, M. K. Derynck, C. Haqq, C. Largman, and H. J. Lawrence. 2005. Activation of stem-cell specific genes by HOXA9 and HOXA10 homeodomain proteins in CD34+ human cord blood cells. Stem Cells 23:644-655.
13. Fischbach, N. A., S. Rozenfeld, W. Shen, S. Fong, D. Chrobak, D. Ginzinger, S. C. Kogan, A. Radhakrishnan, M. M. Le Beau, C. Largman, and H. J. Lawrence. 2005. HOXB6 overexpression in murine bone marrow immortalizes a myelomonocytic precursor in vitro and causes hematopoietic stem cell expansion and acute myeloid leukemia in vivo. Blood 105:1456-1466.
14. Giovannini, M., J. A. Biegel, M. Serra, J. Y. Wang, Y. H. Wei, L. Nycum, B. S. Emanuel, and G. A. Evans. 1994. EWS-erg and EWS-Fli1 fusion transcripts in Ewing's sarcoma and primitive neuroectodermal tumors with variant translocations. J. Clin. Investig. 94:489-496.[Medline]
15. Hisa, T., S. E. Spence, R. A. Rachel, M. Fujita, T. Nakamura, J. M. Ward, D. E. Devor-Henneman, Y. Saiki, H. Kutsuna, L. Tessarollo, N. A. Jenkins, and N. G. Copeland. 2004. Hematopoietic, angiogenic and eye defects in Meis1 mutant animals. EMBO J. 23:450-459.[CrossRef][Medline]
16. Huang, H., M. Rastegar, C. Bodner, S. L. Goh, I. Rambaldi, and M. Featherstone. 2005. MEIS C termini harbor transcriptional activation domains that respond to cell signaling. J. Biol. Chem. 280:10119-10127.
17. Imamura, T., A. Morimoto, M. Takanashi, S. Hibi, T. Sugimoto, E. Ishii, and S. Imashuku. 2002. Frequent co-expression of HoxA9 and Meis1 genes in infant acute lymphoblastic leukaemia with MLL rearrangement. Br. J. Haematol. 119:119-121.[CrossRef][Medline]
18. Inbal, A., N. Halachmi, C. Dibner, D. Frank, and A. Salzberg. 2001. Genetic evidence for the transcriptional-activating function of Homothorax during adult fly development. Development 128:3405-3413.[Medline]
19. Irizarry, R. A., B. Hobbs, F. Collin, Y. D. Beazer-Barclay, K. J. Antonellis, U. Scherf, and T. P. Speed. 2003. Exploration, normalization, and summaries of high density oligonucleotide array probe level data. Biostatistics 4:249-264.[Abstract]
20. Ivanova, N. B., J. T. Dimos, C. Schaniel, J. A. Hackney, K. A. Moore, and I. R. Lemischka. 2002. A stem cell molecular signature. Science 298:601-604.
21. Jacobs, Y., C. A. Schnabel, and M. L. Cleary. 1999. Trimeric association of Hox and TALE homeodomain proteins mediates Hoxb2 hindbrain enhancer activity. Mol. Cell. Biol. 19:5134-5142.
22. Janssen, E., M. Zhu, W. Zhang, and S. Koonpaew. 2003. LAB: a new membrane-associated adaptor molecule in B cell activation. Nat. Immunol. 4:117-123.[CrossRef][Medline]
23. Joosten, M., Y. Vankan-Berkhoudt, M. Tas, M. Lunghi, Y. Jenniskens, E. Parganas, P. J. Valk, B. Lowenberg, E. van den Akker, and R. Delwel. 2002. Large-scale identification of novel potential disease loci in mouse leukemia applying an improved strategy for cloning common virus integration sites. Oncogene 21:7247-7255.[CrossRef][Medline]
24. Kamps, M. P., C. Murre, X. H. Sun, and D. Baltimore. 1990. A new homeobox gene contributes the DNA binding domain of the t(1;19) translocation protein in pre-B ALL. Cell 60:547-555.[CrossRef][Medline]
25. Kamps, M. P., D. D. Wright, and Q. Lu. 1996. DNA-binding by oncoprotein E2a-Pbx1 is important for blocking differentiation but dispensable for fibroblast transformation. Oncogene 12:19-30.[Medline]
26. Knoepfler, P. S., K. R. Calvo, H. Chen, S. E. Antonarakis, and M. P. Kamps. 1997. Meis1 and pKnox1 bind DNA cooperatively with Pbx1 ut