Letters to the Editor 236 HOX regulate Rac1 activity in hematopoietic cells through control of Vav2 expression

Leukemia (2013) 27, 236–238; doi:10.1038/leu.2012.166 demonstrated to be an essential component of HOX-induced oncogenic pathways.8 There was a high concordance between the genes identified in our study (66% for upregulated and 72% for HOX homeobox transcription factors are not only master regu- downregulated targets) and those of Huang et al.4 Within the lators of segment identity but also instrumental for proper common HOXA1/Hoxa9 targets the known oncogene vav2 caught blood formation (reviewed in Argiropoulos and Humphries1 and our interest because Vav2 is a guanine-exchange factor for the Eklund2). The physiology of hematopoietic stem and progenitor small GTPase Rac1 and this factor has been implicated in MLL- cells (HSPC) is under the direct control of HOX . In rearranged leukemia9,10 where MLL fusions work at least in part particular, members of the ‘HOX-A’ and ‘HOX-B’ paralog groups through upregulation of HOX genes. are predominantly expressed in HSPC. Differentiation requires a Vav2 RNA and Vav2 were quantified by quantitative downregulation of HOX genes and mature blood cells usually do reverse transcriptase-PCR and western blotting in HOXA1/ not express HOX proteins any more. An abnormal HOX activity Hoxa9-transformed cells cultivated with TAM and 72 h after TAM and, concomitantly, a block of differentiation are frequently deprivation (Figures 1a and b). Vav2 transcripts were reduced to observed in malignant hematopoiesis, thus qualifying HOX approximately 20% of initial levels and also Vav2 protein dropped members as bona fide oncogenes. significantly in TAM-free medium. The genomic area upstream of Despite their importance in hematopoiesis, little is known about the vav2 transcription initiation site contains several potential the HOX-subordinate pathways that are crucial for HSPC biology Pbx/Hox- and Meis/Hox-binding sites between À 1 and À 3kb and leukemogenesis. A few studies have been published trying to (Figure 1c). ChIP analysis was performed with primer pairs flank- identify HOX target genes in hematopoietic cells.3–5 One recurrent ing these regions. An additional primer was placed into an problem is the multitude of genes that are under control of the X- satellite sequence as negative control. Significant HOX ‘master regulators’. It is difficult to discriminate the ‘driver’ enrichment of HOXA1-ER and Hoxa9-ER was detected at all genes with importance in hematopoiesis from ‘bystanders’ that potential binding sites. Owing to the technical resolution of the may be true HOX targets but have less impact on HSPC ChIP analysis (fragment size B500–800 bp), the primer pair development. In this study we tried to address this problem by at À 2 kb could also detect the occupation of neighboring Pbx/ identifying a common expression pattern shared between HOX sites. Indeed the highest HOX enrichment scores were HOXA1 and Hoxa9, two unrelated, yet both transforming HOX obtained for the region at À 2 kb. Anti-ER-specific antibodies proteins.6 precipitated up to 10-fold more DNA in HOXA1-ER-transformed As shown previously,6 the labial-type HOXA1, as well as the cells compared with the control experiments. In Hoxa9-ER abdominal-like HOXA9, transforms primary hematopoietic proge- populations a 20-fold enrichment of this genomic region was nitors into immortalized myeloblasts. Therefore we reasoned that detected, arguing strongly for a direct regulation of vav2 by HOX/ genes essential for myeloid transformation should be triggered by Pbx and/or HOX/Meis. both HOX proteins in a similar fashion and hence these genes Next we checked if the corresponding sequences confer HOX should be enriched within the common targets. To investigate this responsiveness to a heterologous gene. For that purpose the hypothesis we used tamoxifen (TAM)-inducible derivatives of putative vav2 promoter from position À 3081 to þ 1 relative to HOXA1 and Hoxa9 created by fusion of a modified estrogen– the transcription start was cloned into a luciferase reporter (pGL3). ligand binding domain (ER) as reported previously.6 (Hoxa9 shares The construct was introduced into the embryonal carcinoma cell 97% homology with human HOXA9; therefore we decided to use line P19 because these cells, in contrast to many other cell lines, an existent murine clone.) As expected, transduction7 of progeni- have been demonstrated to be permissive for HOX-mediated tor cells with both HOX derivatives leads to the immortalization transactivation.11 Expression plasmids for HOXA1, Hoxa9 and the of cytokine-dependent, myeloblastic populations that could be HOX cofactors Meis1 and Pbx1 were cotransfected individually or cultivated in medium supplemented with 100 ng/ml SCF, 10 ng/ml in combinations. Consistent with a promoter under HOX control, each of IL-3, IL-6, GM-CSF, and 100 nM 4-hydroxy-TAM. Importantly, the transfection of each HOX protein and/or cofactor had an these cells did not apoptose in the absence of TAM, but initiated a independent and cumulative effect on promoter output normal program of differentiation, terminating with the appearance (Figure 1d). Although total transactivation rates were moderate, of mature granulocytes and macrophages within approximately they were significant and in the range of what had been obtained 1 week (data not shown). Four individual cell lines were created in similar experiments with other HOX-dependent promoter/ for each HOX construct from different donor animals. RNA was enhancer elements.4,5 Therefore these results are consistent with a isolated with TAM present and 72 h after TAM withdrawal. This direct regulation of vav2 by HOXA1 and Hoxa9. schedule was chosen because first significant changes in Subsequently we tested if also Rac1 activity is controlled by expression of several differentiation genes became detectable at HOX activity. Active Rac-GTP can be specifically precipitated by a that time point. Comparison of the expression profiles (raw data GST-fusion with the Rac downstream partner PAK (p21 protein - deposited at ArrayExpress under accession number E-MEXP-3648) Cdc42/Rac-activated kinase) as described in Giehl et al.12 and revealed 948 and 374 genes that were significantly (change Figure 2a. Lysates of HOXA1-ER and Hoxa9-ER cells were prepared 41.7-fold, Po0.1) upregulated by HOXA1 and Hoxa9, respec- from cultures in TAM and 72 h after transfer to TAM-free media. tively. A similar number of transcripts with opposite behavior After incubation with GST-PAK and thorough washing, the bound could be detected (731 and 441 genes increased expression material was analyzed by a Rac1-specific immunoblot alongside after HOX inactivation in HOXA1 and Hoxa9 cells). Among all input controls (Figure 2b). Whereas total Rac1 levels did not HOX-responsive transcripts, a core group of 61 upregulated and change significantly, Rac1-GTP amounts were clearly correlated to 57 downregulated genes was shared by HOXA1 and Hoxa9. This HOX activity. Although HOXA1-ER-transformed cells contained included the myeloblastosis oncogene c-Myb that had been lower concentrations of active Rac1-GTP than their Hoxa9-ER

Accepted article preview online 20 June 2012; advance online publication, 13 July 2012

Leukemia (2013) 233 – 258 & 2013 Macmillan Publishers Limited Letters to the Editor 237

Figure 1. HOX-dependent regulation of vav2.(a) Quantitative reverse transcriptase PCR measuring vav2 transcripts in RNA isolated from HOXA1-ER and Hoxa9-ER-transformed cells in the presence and 72 h after removal of TAM as indicated. Expression in fully induced HOX- Figure 2. Rac1 activity is HOX-dependent and is controlled by Vav2. transformed cells was set arbitrarily to one unit. Average values and (a) Schematic principle of Rac assay. A GST fusion with a Rac s.d.s of PCR triplicates are indicated by error bars. (b) Western blot downstream mediator, the PAK protein, permits specific precipita- analysis detecting Vav2 protein levels in total cellular extracts of the tion of Rac-GTP. (b) Rac1 activity in HOXA1-ER- and Hoxa9-ER- cells as given in (a). Blotting was done with an anti-mouse Vav2 transformed cells in the presence and 72 h after removal of TAM. antibody. Actin served as loading control. In Hoxa9-ER cells a Cellular extracts were incubated with GST-PAK and bound Rac1-GTP truncated version of Vav2 is labeled with an asterisk. (c) Enrichment was analyzed by Rac1-specific immunoblotting alongside with of HOX proteins in the vav2 upstream region was detected by ChIP. respective input samples. The figure shows a typical example out Crosslinked chromatin from HOXA1-ER and Hoxa9-ER cells was of two replicates. (c) Typical example of a replating assay performed precipitated by an anti-ER antibody and analyzed by quantitative either with wild-type or vav2 knock-out bone marrow progenitor PCR employing primers spanning the HOX/Meis consensus sites in cells. The cells were transduced with HOXA9 and replated in the putative vav2 promoter. The structure of the vav2 upstream methocel.7 No significant difference in colony formation was seen in region including the location and sequence of the potential HOX/ four independent experiments. The lower panel shows a Vav2- Meis and HOX/Pbx sites is depicted in the top panel. Anti-mouse IgG specific western blot done on extracts from HOXA9-transformed and a primer pair (primer X) amplifying a X-chromosomal satellite wild-type, vav2 À / À , and knock-out cells complemented with VAV2, sequence were used as a negative controls. Given are relative as indicated. (d) Upper panel: Rac activity in wild-type and vav2 À / À enrichment levels above IgG control divided by background cells. Rac-GTP levels were determined analogous to (b) in cellular enrichment at the X-satellite as mean and s.d. of triplicates. extracts of HOXA9-transformed cells from wild-type or knockout (d) Reporter gene analysis of HOX-dependent promoter activity. cells. The panel shows a typical example for the results obtained Approximately 3 kb of genomic sequence upstream of the vav2 with three independent knockout lines. Lower panel: Reintroduction transcriptional start (reaching to the EcoRI site at À 3081 bp) were of human VAV2 rescues Rac1 activity. HOXA9-transformed vav2 À / À inserted into the luciferase vector pGL3. The reporter was cells were transduced with the cDNA for human VAV2 and Rac1 cotransfected together with expression plasmids for HOXA1 or activity was detected as described above. Hoxa9 and their protein cooperation partners Meis1 and/or Pbx1 into P19 embryonic carcinoma cells. This cell type had been shown before11 to be permissive for HOX-mediated transactivation. independently derived knockout lines revealed a reduction of Luciferase activities were recorded from three independent Rac1-GTP to 18±10% (mean±s.d., n ¼ 3) of wild-type levels (data transfections and plotted as mean and s.d. of relative values with not shown). Active Rac1 concentrations could be largely restored background levels set to one unit. Significance was determined by by complementation with VAV2 (the human version with 95% a t-test with **Po0.01 and *Po0.05. homolog to the mouse protein was used because the correspond- ing mouse cDNA could not be obtained from a cDNA repository). counterparts, control of active Rac1 by HOXA1 was still discernible. Therefore Vav2 was likely the predominant determinant of Rac1 As HOX-dependent Rac1 activity might be induced also by other activity in this cell type. HOX targets, we wanted to determine the specific contribution of Finally, transplantation experiments were set up to find out vav2 towards this phenomenon. For this purpose vav2 knockout if leukemia biology is affected by HOX-mediated control of vav2. mice were obtained.13 Vav2 deletion is not associated with For this purpose wild-type and vav2 À / À bone marrow progeni- any obvious phenotype. This was also reflected by the fact tors were simultaneously transduced with HOXA9 and the that vav2 À / À bone marrow cells were indistinguishable from HOX cofactor Meis1 or in addition also with VAV2 as control. wild-type cells in transformation experiments (Figure 2c). Cell As expected, Rac1 activity was dependent on Vav2 in HOX/Meis lines could be derived from HOXA9-transformed vav2 À / À double-transduced cells. Rac1-GTP was below detection limit in progenitors although Vav2 protein was undetectable in immuno- vav2 À / À cells and complementation with VAV2 rescued part of blots. Vav2 À / À lines, however, contained significantly lower levels this activity (data not shown). In spite of this, there was of Rac1-GTP (Figure 2d). Densitometric quantification in three no significant difference in disease latency regardless of the

& 2013 Macmillan Publishers Limited Leukemia (2013) 233 – 258 Letters to the Editor 238 Vav2/ Rac1 status after transplantation into syngenic, lethally 3Institute of Biochemistry and Cell Biology and irradiated recipients (median survival 38±3 days). However, that Department of Functional Genomics and Medical Toponomics, was concordant with a very recent study14 published during University Magdeburg, Magdeburg, Germany preparation of this manuscript that showed that Rac2 rather than E-mail: [email protected] Rac1 influences leukemia latency. Indeed Rac2-GTP levels in HOX-transformed cells did not correlate with the vav2 status (not shown). Only the spleen weight of the diseased animals was REFERENCES significantly different in vav2 À / À transplants (270±64 mg, 1 Argiropoulos B, Humphries RK. Hox genes in hematopoiesis and leukemogenesis. mean±s.d., n ¼ 7) compared with wild-type (349±83 mg, n ¼ 7, ± Oncogene 2007; 26: 6766–6776. P ¼ 0.04) and VAV2-complemented cells (399 70 mg, n ¼ 5, 2 Eklund E. The role of hox proteins in leukemogenesis: insights into key regulatory P ¼ 0.004), indicating potentially altered engraftment behavior of events in hematopoiesis. Crit Rev Oncog 2011; 16: 65–76. À / À vav2 cells. 3 Ferrell CM, Dorsam ST, Ohta H, Humphries RK, Derynck MK, Haqq C et al. In summary, here we describe a potential method how to Activation of stem-cell specific genes by HOXA9 and HOXA10 homeodomain prioritize the large numbers of HOX-controlled genes in order to proteins in CD34 þ human cord blood cells. Stem Cells 2005; 23: 644–655. search for targets with increased biological relevance in HSPC. 4 Huang Y, Sitwala K, Bronstein J, Sanders D, Dandekar M, Collins C et al. We validated this approach by identifying Vav2 and therefore Identification and characterization of Hoxa9 binding sites in hematopoietic cells. Rac1 activity to be under HOX control. Rac1, a member of the Blood 2012; 119: 388–398. Rho superfamily of small GTPases, is widely involved in 5 Shah CA, Wang H, Bei L, Platanias LC, Eklund EA. HoxA10 regulates transcription of the gene encoding transforming growth factor beta2 (TGFbeta2) in myeloid cells. hematopoietic development and it is particularly important for J Biol Chem 2011; 286: 3161–3176. control of HSPC-specific properties such as engraftment and 6 Bach C, Buhl S, Mueller D, Garcia-Cuellar MP, Maethner E, Slany RK. Leukemogenic homing, as well as for proper interaction with the transformation by HOXA cluster genes. Blood 2010; 115: 2910–2918. 15 microenvironment. 7 Lavau C, Szilvassy SJ, Slany R, Cleary ML. Immortalization and leukemic transfor- mation of a myelomonocytic precursor by retrovirally transduced HRX-ENL. EMBO J 1997; 16: 4226–4237. CONFLICT OF INTEREST 8 Hess JL, Bittner CB, Zeisig DT, Bach C, Fuchs U, Borkhardt A et al. c-Myb is an essential downstream target for homeobox-mediated transformation of hema- The authors declare no conflict of interest. topoietic cells. Blood 2006; 108: 297–304. 9 Somervaille TC, Cleary ML. Identification and characterization of leukemia stem cells in murine MLL-AF9 acute myeloid leukemia. Cancer Cell 2006; 10: 257–268. ACKNOWLEDGEMENTS 10 Wei J, Wunderlich M, Fox C, Alvarez S, Cigudosa JC, Wilhelm JS et al. Micro- We thank Renate Zimmermann for technical assistance. This work was supported by environment determines lineage fate in a human model of MLL-AF9 leukemia. research funding from DFG (Deutsche Forschungsgemeinschaft) Grant SL27/6-3 to Cancer Cell 2008; 13: 483–495. RKS and SCHA965/6-1 to AS, as well as SFB854/TP11 to KDF. 11 Schnabel CA, Jacobs Y, Cleary ML. HoxA9-mediated immortalization of myeloid progenitors requires functional interactions with TALE cofactors Pbx and Meis. Oncogene 2000; 19: 608–616. AUTHOR CONTRIBUTIONS 12 Giehl K, Skripczynski B, Mansard A, Menke A, Gierschik P. Growth factor-depen- dent activation of the Ras-Raf-MEK-MAPK pathway in the human pancreatic CB, EM, MPGC and RKS performed and analyzed the experiments, KDF and carcinoma cell line PANC-1 carrying activated K-ras: implications for cell pro- À / À KS provided vav2 animals and advice on Vav proteins, AS provided liferation and cell migration. Oncogene 2000; 19: 2930–2942. essential reagents and advice for the Rac1 assays, and RKS designed the 13 Tedford K, Nitschke L, Girkontaite I, Charlesworth A, Chan G, Sakk V et al. Com- experiments and wrote the manuscript. pensation between Vav-1 and Vav-2 in B cell development and antigen receptor signaling. Nat Immunol 2001; 2: 548–555. C Breitinger1, E Maethner1, M-P Garcia-Cuellar1, A Schambony2, 14 Mizukawa B, Wei J, Shrestha M, Wunderlich M, Chou FS, Griesinger A et al. Inhi- bition of Rac GTPase signaling and downstream prosurvival Bcl-2 proteins as KD Fischer3, K Schilling3 and RK Slany1 1 combination targeted therapy in MLL-AF9 leukemia. Blood 2011; 118: 5235–5245. Department of Genetics, University Erlangen, Erlangen, Germany; 15 Gu Y, Filippi MD, Cancelas JA, Siefring JE, Williams EP, Jasti AC et al. Hematopoietic 2 Department of Developmental Biology, University Erlangen, cell regulation by Rac1 and Rac2 guanosine triphosphatases. Science 2003; 302: Erlangen, Germany and 445–449.

Mutations in the DNMT3A exon 23 independently predict poor outcome in older patients with acute myeloid leukemia: a SWOG report

Leukemia (2013) 27, 238–241; doi:10.1038/leu.2012.168 DNA cytosine-5-methyltransferase 3A (DNMT3A) encodes an enzyme that catalyzes the addition of a methyl group to the cytosine residue of CpG dinucleotides. By regulating the methyla- Acute myeloid leukemia (AML) represents a heterogeneous group tion of clusters of CpG, this enzyme mediates the downregulation of malignancies with great variability in clinical course and response to therapy. Currently, cytogenetics is the single most of downstream genes. Recent studies using large-scale array- important prognostic factor in this disease. In recent years, an based genomic resequencing and whole-genome sequencing of increasing number of molecular features with prognostic sig- human leukemia uncovered recurrent mutations in this gene in nificance in AML have been identified.1–3 AML patients.4,5 Subsequent studies demonstrated that these

Accepted article preview online 22 June 2012; advance online publication, 13 July 2012

Leukemia (2013) 233 – 258 & 2013 Macmillan Publishers Limited