Leukemia (1999) 13, 1993–1999  1999 Stockton Press All rights reserved 0887-6924/99 $15.00 http://www.stockton-press.co.uk/leu Frequent co-expression of the HOXA9 and MEIS1 in human myeloid leukemias

HJ Lawrence1, S Rozenfeld1, C Cruz1, K Matsukuma1, A Kwong1,LKo¨mu¨ves1, AM Buchberg2 and C Largman1

1Division of Hematology and Medical Oncology, Department of Medicine, University of California VA Medical Center, San Francisco, CA; and 2Kimmel Center, Jefferson Medical College, Philadelphia, PA, USA

There is increasing evidence that HOX homeobox genes play these leukemias.12 Recent work has shown that the engine- a role in leukemogenesis. Recent studies have demonstrated ered co-expression of the Hoxa-9 and Meis1 genes in murine that enforced co-expression of HOXA9 and MEIS1 in murine 13 marrow leads to rapid development of myeloid leukemia, and hematopoietic cells leads to rapid development of AML, that these exhibit cooperative DNA binding. However, indicating a synergy between the two genes and suggesting it is unclear whether co-activation of HOXA9 and MEIS genes that co-activation of these genes is sufficient for transform- is a common occurrence in human leukemias. We surveyed ation. expression of HOXA9 and MEIS1 in 24 leukemic cell lines and We have recently demonstrated that HOXA9 and HOXA10 80 patient samples, using RNase protection analyses and proteins can interact with the MEIS1 and stabilize immunohistochemistry. We demonstrate that the expression of 14 HOXA9 and MEIS1 in leukemia cells is uniquely myeloid, and MEIS1-DNA interactions on consensus binding sites. These that these genes are commonly co-expressed in myeloid cell data support the notion that the leukemogenicity of the lines and in samples of acute myelogenous leukemia (AML) of HOXA9 and MEIS1 genes is related to the concerted actions all subtypes except in promyelocytic leukemia. While HOXA9 of their protein products on targets. A major question is expressed in most cases of chronic myelogenous leukemia, remains whether co-activation of these genes occurs in human MEIS1 is weakly expressed or not at all. Immunohistochemical leukemias. In this study we surveyed expression patterns of staining of selected AML samples showed moderate to high levels of HOXA9 protein, primarily cytoplasmic, in leukemic both genes in a large panel of human cell lines and primary myeloblasts, with weaker and primarily nuclear staining for samples of leukemia, and observed frequent lineage-specific MEIS1. These data support the concept that co-activation of co-expression of these two genes in AML. HOXA9 and MEIS1 is a common event in AML, and may rep- resent a common pathway of many different oncogenic mutations. Materials and methods Keywords: homeobox genes; myeloid leukemia; ; tran- scription factors Leukemic cell lines and patient samples

Twenty-four human leukemic cell lines were studied, as out- Introduction lined in Figure 1. All lines were maintained in media as pre- viously described.15 Cryopreserved cell samples from 80 There is increasing evidence that DNA-binding proteins enco- patients, gathered from six different centers, with various ded by the HOX family of homeobox genes play a role in forms of myeloid and lymphoid leukemias and myelodyspla- leukemic transformation. Several HOXA and HOXB genes are sia were studied and subclassified according to the French– expressed in myeloid leukemias,1–3 with HOXC genes American–British (FAB) nomenclature. The original samples expressed predominantly in lymphoid malignancies.4,5 Evi- contained a range of 30–100 × 106 cells each and in some dence that mutations of HOX genes occur in human leuke- cases were over 10 years old. In all cases, leukocytes were mias comes from characterization of the t(7;11) translocation, isolated by either a buffy-coat preparation or Ficoll–Hypaque seen in rare cases of acute myelogenous leukemia (AML),6 and density centrifugation before freezing in liquid nitrogen. Cryo- which results in the fusion of the HOXA9 gene with a nucleo- preserved samples were thawed quickly by shaking in a 37°C porin gene.7,8 However, these studies do not establish a causal water bath, diluted in warm RPMI with 10% fetal calf serum, role for HOX genes in leukemic transformation. In this regard, pelleted, washed with medium and repelleted for RNA iso- the enforced expression of HOXB8 together with IL-3 in lation by the guanidinium thiocyanate/acid phenol method.16 murine marrow cells leads to myeloid leukemia;9 similarly, Samples of normal marrow and peripheral blood leukocytes over-expression of Hoxa-10 can induce myeloid leukemia were prepared as previously described.4 after a long latency period.10 These findings suggest that mis- expression of HOX genes can contribute to transformation but additional genetic events are required. RNase protection analysis Support for a role of HOX genes in spontaneous leukemias arose from studies of the BXH-2 mice, which develop a high Templates were constructed by subcloning a 378-nt fragment incidence of myeloid leukemia. A novel non-HOX homeobox of the HOXA9 cDNA, extending from amino acid 147 to the gene, Meis1, is up-regulated as a consequence of retroviral end of the coding cDNA at amino acid 270; and a 394 nt integration in 15% of leukemias arising in BXH-2 mice.11 Sub- fragment of MEISI, extending from amino acid 287 to the end sequent studies have revealed frequent activating insertions of of the MEISIa cDNA, into the sk+ Bluescript vector using stan- the Hoxa-9 or Hoxa-7 genes together with the Meis1 gene in dard procedures. Total RNA (25–50 ␮g/samples) from individ- ual specimens was subjected to separate RNase protection Correspondence: HJ Lawrence, 1, rue Laurent Fries, BP 163, 67 404 assays for each gene, following previously published 17 Illkirch Codex, France methods. As an internal control for loading, a probe protect- Received 8 October 1998; accepted 26 July 1999 ing an 83-nt fragment of glyceraldehyde 3-phosphate Expression of HOXA9 and MEIS1 in myeloid leukemias HJ Lawrence et al 1994

Figure 1 Expression of HOXA9 and MEIS1 in human leukemic cell lines. Total RNA was isolated from 24 immortalized human leukemic lines and subjected to separate RNase protection assays for HOXA9 (top panel) and MEIS1 (middle panel), using a probe for glyceraldehyde- 3 phosphate dehydrogenase (G3PD, lower panel) as an internal control for RNA loading, as described in Materials and methods. A clear signal for HOXA9 and MEIS1 is seen in 6/8 and 4/7 myeloid cell lines, respectively, with isolated expression of HOXA9 in the MB02 and M70E lines, and isolated expression of MEIS1 in a few of the non-myeloid lines. The signals for G3PD represent short exposure times to more clearly demonstrate relative loading.

dehydrogenase (G3PD) was included in each assay. RNA iso- Burlingame, CA, USA). MEIS1 was detected with affinity-pur- lated from the KG-1 cell line was used as a positive control. ified guinea pig antibodies followed by biotin-conjugated anti- For patient samples, autoradiographs of RNase protection guinea pig rabbit IgG FAB fragment followed by detection experiments were developed for 1 day for visualization of the with ABC alkaline phosphatase. Specimens were coun- G3PD signal and for 3–7 days for HOXA9 and MEIS1. Densi- terstained with methyl green to visualize nuclei. Slides were tometric scanning of the RNase protection bands for HOXA9 mounted with Vectashield (Vector Labs), photographed and and MEIS1 were performed and normalized to the G3PD band scanned, and images were assembled and labeled on a Mac- from the same gel (to correct for differences in loading), and Intosh G3 computer with Adobe Photoshop. Protein then normalized between gels by comparing the signals for expression levels were scored on an arbitrary scale of +1to KG-1 cell RNA, as previously described.1 Expression levels for +4. both genes were then assigned a value of 0 to 4+.

Results Immunohistochemical localization of homeobox proteins Myeloid-specific expression of HOXA9 expression in leukemic cell lines Previously unstained blood or smears from selected samples of acute leukemia were fixed in 4% formal- Total RNA was isolated from 24 human leukemic cell lines dehyde (freshly prepared from paraformaldehyde) in PBS for and analyzed for HOXA9 expression by RNase protection 24 h. For immunohistochemical localization, HOXA9 was analysis. As shown in Figure 1, HOXA9 was expressed exclus- detected with affinity-purified chicken antibodies followed by ively in myelomonocytic cells, with the sole exception of the biotin-conjugated anti-chicken rabbit IgG FAB fragment bi-phenotypic MBO2 erythroid/myeloid line and the M70E (Jackson Immunochemicals, West Grove, PA, USA), followed megakaryocytic line. Two of eight myeloid cell lines tested, by detection with ABC alkaline phosphatase (Vector Labs, TMM and HL60, had no detectable HOXA9 message. The Expression of HOXA9 and MEIS1 in myeloid leukemias HJ Lawrence et al 1995 lymphoid lines studied were uniformly negative for expression Table 1. The expression of HOXA9 in these primary samples of this gene. This pattern of expression is virtually identical to was essentially identical to that seen in our previous survey the pattern previously observed for the neighboring for HOXA10 expression, using many of the same samples.1 HOXA10 gene.1 HOXA9 shows an exclusively myeloid-specific expression with rare exceptions, namely one of three cases of T cell leu- MEIS1 expression in leukemic cell lines kemia involving the t(10:14) translocation. Of the five cases of mixed-lineage leukemias involving translocations of 11q When the myeloid cell lines were assayed for MEIS1 one was positive and two showed barely detectable expression by RNase protection analysis, transcripts were expression (Table 1). Otherwise, HOXA9 message was uni- detected in all but one (THP-1) of the lines which showed formly undetectable in 18 samples of B cell acute lymphocytic HOXA9 expression (Figure 1). Thus in four of seven myeloid leukemias and one case of chronic lymphocytic leukemia cell lines, there was co-expression of both genes. However (data not shown). When AML samples were broken down by MEIS1 also showed detectable expression in a few B cell (SU, FAB classification, it was of interest to note that HOXA9 mess- U266), T cell (MOLT4), erythroid (HEL, K562) and megakary- age was not found in any of five promyelocytic leukemic ocytic (Dami) cell lines which did not express HOXA9. samples (Table 1 and Figure 2). Otherwise HOXA9 was expressed in every case of FAB-M2 and FAB-M4 tested, all Myeloid-specific expression of HOXA9 and MEIS1 in six cases of secondary AML, all eight cases of myelodysplasia primary samples of human leukemia (MDS), one case of an undifferentiated myeloproliferative dis- order, and all four samples of stable-phase chronic myelogen- Total RNA from 80 archival samples representing a variety of ous leukemia (Figure 2 and data not shown). Curiously, subtypes of myeloid and lymphoid leukemias was subjected HOXA9 message was not seen in four of six samples of CML to RNase protection assays for expression of HOXA9 and in blast crisis, including two samples of lymphoid blast crisis. MEIS1. Although for the majority of cases both genes were MEIS1 displayed a similar myeloid-restricted pattern of assayed, material limitations resulted in assay of only one expression, with no detectable expression in any of 20 lymph- gene for a few cases. Representative data is depicted in Figure oid samples (Figure 2 and data not shown). As with HOXA9 2, and the findings for all the samples are summarized in and HOXA10,1 MEIS1 message was not seen in any of six

Figure 2 Co-expression of HOXA9 and MEIS1 in primary samples of myeloid leukemias. RNase protection assays were performed on total RNA isolated from frozen bone marrow samples, as described in Figure 1. Separate analyses were performed for HOXA9 and MEIS1 using non- homeobox containing probes to ensure specificity. Gels were autoradiographed for 3–5 days for HOXA9 and MEIS1 or overnight for G3PD, respectively. Both genes are clearly co-expressed in most cases of FAB-M0, -M1, -M2 and -M4. However, the samples of FAB-M3 are uniformly negative for the expression of both genes. HOXA9 is expressed in all of the secondary AMLs tested, while MEIS1 expression was detected in two of three cases (trace amount seen in sample No. 17 on longer exposure). While HOXA9 is expressed in nearly all cases of MDS and chronic phase CML, the expression of MEIS1 is much weaker in those samples. Neither gene is strongly expressed in blast crisis of CML (data not shown). Expression of both genes is uniformly negative in B and T cell leukemias. The smearing seen in most lanes is most likely RNA degradation related to prolonged storage of archival samples. Expression of HOXA9 and MEIS1 in myeloid leukemias HJ Lawrence et al 1996 Table 1 HOXA9 and MEIS1 expression in myeloid leukemic promyelocytic leukemia samples tested, but was expressed in samples 13 of 19 (68%) other AML samples tested (Figure 2 and Table 1). MEIS1 was expressed in five of eight cases of myelodyspla- Sample No. Diagnosis HOXA9 MEIS1 sia, although expression levels in some cases of myelodyspla- sia were weak. MEIS1 was expressed weakly or not at all in 1 AML-M0 +3 +4 2 AML-M0 +2 +2 samples of CML, regardless of the stage of disease. HOXA9 3 AML-M1 ND +2 and MEIS1 were frequently co-expressed in AML, and less 4 AML-M1 +3 +3 often and less intensely in MDS and CML. In those cases of 5 AML-M1 neg neg primary and secondary AML (excluding promyelocytic + 6 AML-M2 2 neg leukemia) for which both genes could be assayed in the same 7 AML-M2 +3 +3 8 AML-M2 +3 neg sample, 12 of 18 cases (67%) expressing HOXA9 also 9 AML-M2 +4 +2 expressed MEIS1, and in no case studied was MEIS1 10 AML-M2 +4 +4 expression detected in the absence of HOXA9. 11 AML-M2 +3ND 12 AML-M2, −7 +4 +4 13 AML-M3 neg neg 14 AML-M3 neg neg Expression of HOXA9 and MEIS1 RNA in normal 15 AML-M3 neg neg blood cells 16 AML-M3 neg neg 17 AML-M3 ND neg 18 AML-M3 neg neg Total RNA was isolated from normal unfractionated bone mar- 19 AML-M4 +4 +2 row, and purified populations of peripheral blood granulo- 20 AML-M4 +3 neg cytes, monocytes and lymphocytes and analyzed by RNase 21 AML-M4 +2 +1 protection. Both HOXA9 and MEIS1 were detectably 22 AML-M5, −7 ND neg 23 2° AML, −7 +3 neg expressed in normal marrow but not any of the mature cell 24 2° AML +4 +4 fractions from the peripheral blood (data not shown), results 25 2° AML +2 +1 similar to those reported for HOXA10.1 The findings for 26 2° AML +2ND HOXA9 are also consistent with previous studies showing that 27 2° AML +3 +4 ° + HOXA9 expression is down-regulated as marrow cells leave 28 2 AML 2ND + 18 29 RA +2 +4 the CD34 compartment. 30 RAEB +2ND 31 RAEB +3ND 32 RAEB ND +3 Immunohistochemical staining of leukemic blasts 33 RAEBIT +4 +1 34 CMML +1 neg 35 CMML +2 +1 To confirm that both proteins were co-expressed in leukemic 36 CMML +2 +1 cells, replicate blood and marrow smears from selected 37 CMML +2 neg samples were stained with specific antibodies against HOXA9 38 CML-CP +2 +1 39 CML-CP +2 +1 and MEIS1. Moderate to high levels of HOXA9 protein, 40 CML-CP +2 +3 localized primarily in the cytoplasm (black arrows) but also 41 CML-CP +3 neg in a speckled pattern over the nucleus (white arrows in insert), 42 CML-AP +2 +2 were seen in the leukemic blasts from most cases of AML (see 43 CML-BC +3ND 44 CML-BC +4ND panels a and c, Figure 3 and Table 2), but not in the lympho- 45 CML-BC neg neg blasts of two cases of ALL (panel e, Figure 3) or in one sample 46 CML-BC neg neg of promyelocytic leukemia with the classic t(15;17) translo- 47 CML-BC-ALL neg neg cation (data not shown). Weak to moderate MEIS1 staining 48 CML-BC-ALL neg neg was demonstrable in blasts of most of the same myeloid leuke- 49 MPD +3 neg 50–54 Common ALL 0/5 0/4 mia samples that showed positive signals for HOXA9. MEIS1 56–59 Pre-B ALL 0/2 0/3 protein showed both nuclear (white arrows in insert of panel 60–63 Pre-B ALL t(1:19) 0/3 0/2 b, Figure 3) and cytoplasmic localization (black arrows in 64–68 B-ALL 0/5 0/4 panel d, Figure 3), and was not demonstrable in two ALL 69–72 T-ALL 0/3 0/4 73–75 T-ALL t(10:14) 1/2 (+1) 0/1 samples (see panel f, Figure 3) or one promyelocytic leukemia 76 AL t(9:11) 1/1 (+1) ND (not shown). This pattern of cytoplasmic and speckled nuclear 77 AL t(4:11) 0/1 ND staining is reminiscent of that seen with HOXA9 and MEIS1 78–80 AL t(11:19) 2/3 (trace) ND in myeloid leukemic cell lines.19

Expression was detected by RNase protection analysis as shown in Figure 2. Bands for HOXA9 and MEISI were quantitated by densi- Discussion tometric scanning of autoradiographs, using an arbitrary scale from trace to +4. neg, no detectable expression; ND, not done; AML, acute myelo- In this survey, we find that co-activation of HOXA9 and MEIS1 genous leukemia, 2° AML, secondary AML; RA, refractory anemia; occurs exclusively in AML and is a common event in those RAEB, RA with excess blasts; RAEBIT, RA with excess blasts in cases. This work has been confirmed by one study showing transition; CMML, chronic myelogenous leukemia; CML-CP, chronic expression of HOXA9 in a smaller series of AML samples,20 myelogenous leukemia in chronic phase; CML-AP, CML in acute and another study which demonstrated MEIS1 expression in phase; CML-BC, CML in blast crisis; ALL, acute lymphoblastic leu- 21 kemia; AL, acute leukemia (mixed lineage involving 11q). 50% of cases of AML and rarely in ALL. Our study also dem- onstrates that HOXA9 and MEIS1 proteins are present in leu- kemic myeloblasts, both in the cytoplasm and the nucleus. We have recently demonstrated, that in myeloid leukemic cell Expression of HOXA9 and MEIS1 in myeloid leukemias HJ Lawrence et al 1997 a b

c d

e f

g h

Figure 3 Immunohistochemical localization of HOXA9 and MEIS1 in leukemic myeloblasts. HOXA9 (a, c and e) or MEIS1 (b, d and f) were localized using affinity purified antisera as described in Materials and methods. (g and h) are negative controls for HOXA9 or MEIS1 staining, respectively, in which sections were stained with all reagents but the primary antibody was omitted. (a and b) HOXA9 and MEIS1 (brown signals) in blasts of a patient with AML-M1. (c and d) HOXA9 and MEIS1 staining in blasts of a patient with AML-M2. (e and f) Absence of HOXA9 and MEIS1 staining of lymphoblasts from a patient with ALL. (g and h) Negative controls for the AML-M2 sample shown in panels (c and d). Note strong and predominantly cytoplasmic HOXA9 staining with weaker speckled nuclear signals (white arrows in insert) for the myeloblasts in patients with AML (panels a and c). MEIS1 staining is weaker and more clearly localized to nuclear speckles (white arrows in insert) with fainter cytoplasmic staining (black arrows) in the same samples (panels b and d). The ALL sample stains for neither protein (panels e and f). lines, there is significant co-localization of these two proteins genes, one member of which is involved in the t(1;19) translo- by confocal microscopy.19 The Meis1 gene was initially cation found in pre-B ALL.22,23 There are two other known discovered as a common site of retroviral insertions in members of the murine MEIS subfamily, Meis2 and Meis3, myeloid leukemias of BXH-2 mice.11 The subsequent dis- and in a number of BXH-2-related murine leukemias in which covery that activating insertions adjacent to the Hoxa9 and Meis1 is inactive, both Meis2 and Meis3 are expressed.24 A Hoxa7 genes often occurred in the same BXH-2 leukemic preliminary report indicates that MEIS2 is also active in human cells suggested that Hox and Meis genes could operate in con- myeloid cell lines.21 Thus it seems likely that activation of cert to produce murine myeloid leukemias.12 MEIS genes together with HOXA9 and HOXA10 occur in the The MEIS1 gene belongs to the TALE subfamily of non-HOX large majority of cases of AML, with the exception of promye- homeobox genes, characterized by a unique 3-amino acid locytic leukemia. insert. Other members of the TALE family include the PBX The uniform lack of expression of HOXA9, HOXA10 and Expression of HOXA9 and MEIS1 in myeloid leukemias HJ Lawrence et al 1998 Table 2 Immunohistochemical staining for HOXA9 and MEIS1 transplanted with FDCP1 cells engineered to over-express protein expression in myeloid leukemia samples both HOXA9 and MEIS1 all succumbed to a fatal in less than 2 weeks.13 In experiments util- FAB Patient No. HOXA9 Meis1 izing normal murine marrow over-expressing both genes, leu- kemia arose in all the transplanted animals in 2 months. In + + M0 06–08 3 2 contrast, mice transplanted with cells expressing either 06–09 3+ 1+ HOXA9 or MEIS1 alone, or HOXA9 together with PBX1a, M1 06–04 3+ 2+ developed leukemias much more slowly or not at all. The rap- 13–01 1+ 1+ idity and frequency of leukemias in this model suggests that M2 06–10 3+ 2+ co-activation of HOXA9 and MEIS1 is sufficient to induce 13–03 2+ 1+ malignant transformation, and may require no other genetic 13–04 1+ 1+ events. + 13–05 1 0 There is now substantial genetic and biochemical evidence M3 13–09 0 0 that the actions of HOX proteins are mediated by interactions M4 06–05 4+ 2+ with the non-HOX homeodomain proteins, PBX and MEIS. A 06–07 4+ 3+ large amount of data indicates that HOX proteins from paralog 13–07 0 0 groups 1–10, including HOXA9 and HOXA10, show cooper- ative DNA binding with the PBX proteins.28–32 In a similar Smears of peripheral blood and/or bone marrow from patients with fashion, we have recently demonstrated that HOXA9 and various FAB subtypes of myeloid leukemia were stained as HOXA10 proteins can physically interact with the MEIS1 pro- described in Materials and methods and as depicted in Figure 3. tein, and that these proteins can cooperatively bind DNA on Intensity of staining of leukemic blasts was assessed semiquantit- + + a TGACAGT(T/A)A(T/C)GAC consensus site, in which the atively on a scale from 0 to 4 , with 4 representing the staining 14 intensity of U937 cells, which show high levels of both proteins. TGACAG site is bound by MEIS1. Interestingly only those Lymphocytes on the smears were consistently negative for both proteins encoded by the more 5Ј HOX genes, the so-called HOXA9 and MEIS1. Furthermore, negative control slides, for which AbdB members from paralog groups 9–13, can interact with the primary antibody was omitted, showed no staining. MEIS1. Thus among the 39 HOX proteins only those from par- alog groups 9 and 10, including HOXA9 and HOXA10, can interact with both PBX and MEIS1. Since HOXA9 and MEIS1 in the FAB-M3 subtype of AML is intriguing and unex- HOXA10 appear to be the only paralog 9 and 10 genes plained. It is possible that the malignant promyelocytes, which expressed in blood cells, these two genes may be uniquely are typically CD34−, are beyond the developmental stage at positioned as potential oncogenes in myeloid cells. Key issues which these genes are normally expressed in blood cells.25 include determining the identity of downstream target genes Given that retinoic acid receptors appear to regulate the acti- that are controlled through such regulatory elements, the poss- vation of HOX genes,26 it is conceivable that the PML-RAR␣ ible competition between MEIS1 and PBX proteins for binding fusion protein suppresses the expression of these genes.27 Both with HOXA9 and HOXA10, and regulation of the sub-cellular HOXA9 and MEIS1 are strongly expressed in all the secondary localization of these proteins. AMLs tested, including cases progressing from a previous AML is characterized by a large array of somatic mutations, myelodysplastic disorder. The expression of HOXA9 in sec- no one of which occurs in a majority of cases; these findings ondary leukemias is of interest since most of these cases dem- indicate that no single genetic event triggers AML (reviewed onstrated loss of one copy of 7, the site of the in Ref. 33). Although the samples used in this study were not HOXA cluster, indicating that the remaining allele remains directly tested for mutations in either the HOXA9 or MEIS1 active. The weak expression of MEIS1 relative to HOXA9 in genes, we believe it is unlikely that either gene is frequently myelodysplasias suggests that, in these preleukemic disorders, mutated in human AML. If this supposition is correct, the co- the progression to AML may be a sequential process with activation of HOXA9 and/or HOXA10 along with MEIS1 in initial activation of HOXA9 and then later MEIS1. The low AML must be a common downstream event which level of activation of MEIS1 in any phase of CML may indicate accompanies oncogenic mutations in other genes. Examples that MEIS1 is not involved in the induction of that disorder or of such mutations include translocations involving the MLL its progression. gene on chromosome 11q. Genetic and molecular biologic The fact that HOXA and MEIS1 genes are expressed in evidence strongly indicates that the MLL gene product directly myeloid leukemias could simply reflect the normal expression binds to regulatory regions within clusters and of these genes in primitive myeloid cells. Evidence against this maintains those genes in an activated configuration.34,35 It is hypothesis comes from recent studies comparing the noteworthy that we found activation of HOXA9 in samples of expression of HOXA9, HOXA10 and MEIS1 in sorted CD34+ leukemias involving 11q translocations. It has been proposed and CD34− populations from normal and leukemic human that many other mutations in AML involve genes which bone marrow. Both HOXA9 and HOXA10 were noted to be encode transcription factors such as retinoic acid sharply down-regulated when normal hematopoietic cells alpha (RAR␣) and AML-1, which may perturb the expression leave the CD34+ compartment. In contrast, in myeloid leuke- and/or function of HOX genes.33 If this model is correct, ther- mias, both HOX genes as well as MEIS1 remain activated in apies designed to disrupt the interactions between HOX and the CD34− population,21 strongly supporting the notion that MEIS proteins and their DNA targets could have a therapeutic the developmental regulation of HOX and MEIS genes in impact in myeloid leukemias. myeloid leukemic cells is perturbed. The frequent retroviral activation of Hoxa-9 and Meis1 in Acknowledgements BXH-2 murine leukemias strongly suggests that there is a bio- logic interaction between the two genes. Direct evidence for We are grateful to Michael Link, Michelle LeBeau, Scott this interaction comes from recent experiments in which mice Kogan and Alfredo Lopez for furnishing patient samples for Expression of HOXA9 and MEIS1 in myeloid leukemias HJ Lawrence et al 1999 study, and to Stephen Fong for his capable technical assist- 15 Mathews CHE, Detmer K, Boncinelli E, Lawrence HJ, Largman C. ance. This work was supported in part by grants from the Erythroid-restricted expression of homeobox genes of the human National Institutes of Health (No. DK48642 to HJL; and No. HOX 2 . Blood 1991; 78: 2248–2252. 16 Chomczynski P, Sacchi N. Single-step method of RNA isolation by CA80029 to CL) and by grants from the Department of Vet- acid guanidinium thiocyanate-phenol-chloroform extraction. Anal erans Affairs (HJL and CL). HJL is a recipient of a VA Career Biochem 1987; 162: 156–159. Development Award. 17 Kreig PA, Melton DA. In vitro RNA synthesis with SP6 RNA poly- merase. Meth Enzymol 1987; 155: 397–415. 18 Lawrence HJ, Helgason CD, Sauvageau G, Fong ST, Izon DJ, Humphries RK, Largman C. Mice bearing a targeted disruption of the homeobox gene HOXA9 have defects in myeloid, erythroid References and lymphoid hematopoiesis. Blood 1997; 89: 1922–1930. 19 Shen W-F, Rozenfeld S, Kwong A, Komuves LG, Lawrence HJ, Largman C. HOXA9 forms triple complexes with PBX2 and MEIS1 1 Lawrence HJ, Sauvageau G, Ahmadi A, Lau T, Lopez AR, Le Beau in myeloid cells. Mol Cell Biol 1999; 19: 3051–3061. MM, Link M, Largman C. Stage and lineage-specific expression of 20 Kawagoe H, Humphries K, Blair A, Sutherland H, Hogge D. the HOXA10 homeobox gene in normal and leukemic hemato- Expression of HOX genes, HOX cofactors and MLL in pheno- poietic cells. Exp Hematol 1995; 23: 1160–1166. typically and functionally defined subpopulations of leukemic and 2 Magli MC, Barba P, Celetti A, De Vita G, Cillo C, Boncinelli E. normal human hematopoietic cells. Leukemia 1999;13: 687–698. Coordinate regulation of HOX genes in human hematopoietic 21 Afonja O, Smith J, Ohyashiki K, Ohyashiki J, Toyama K, Inghirami cells. Proc Natl Acad Sci USA 1991; 88: 6348–6352. G, Takeshita K. MEIS gene family in human acute myeloid leuke- 3 Celetti A, Barba P, Cillo C, Rotoli B, Boncinelli E, Magli MC. mia. Blood 1997; 90: 67a. Characteristic patterns of HOX in different types 22 Nourse J, Mellentin JD, Galili N, Wilkinson J, Stanbridge E, Smith of human leukemia. Int J Cancer 1993; 53: 237–244. SD, Cleary MC. Chromosomal translocation t(1;19) results in syn- 4 Lawrence HJ, Stage KM, Mathews CHE, Detmer K, Scibenski R, thesis of a homeobox fusion mRNA that codes for a potential MacKenzie M, Magliaccio E, Boncinelli E, Largman C. Expression chimeric . Cell 1990; 60: 535–545. of HOX C homeobox genes in lymphoid cells. Cell Growth Differ 23 Kamps MP, Murre C, Sun X-H, Baltimore D. A new homeobox 1993; 4: 665–669. gene contributes the DNA binding domain of the t(1;19) translo- 5 Bijl J, van Oostveen JW, Kreike M, Rieger E, van der Raaij-Helmer cation protein in Pre-B ALL. Cell 1990; 60: 547–555. LMH, Walboomers JMM, Corte G, Boncinelli E, van den Brule 24 Nakamura T, Jenkins NA, Copeland NG. Identification of a new AJC, Meijer CJLM. Expression of HOXC4, HOXC5, and HOXC6 in family of Pbx-related homeobox genes. 1996; 13: human lymphoid cell lines, leukemias, and benign and malignant 2235–2242. lymphoid tissue. Blood 1996; 87: 1737–1745. 25 Grignani F, Fagioli M, Alcaly M, Longo L, Pandolfi PP, Donti E, 6 Sato Y, Abe S, Mise K, Sasaki M, Kamada N, Kouda K, Musashi Biondi A, Lo Coco F, Grignani F, Pelicci PG. Acute promyelocytic M, Saburi Y, Horikoshi A, Minami Y, Miyakuni T, Yokoyama Y, leukemia: from genetics to treatment. Blood 1994; 83: 10–25. Ishihara Y, Miura Y. Reciprocal translocation involving the short 26 Simeone A, Acampora D, Arcioni L, Andrews PW, Boncinelli E, − + arms of 7 and 11, t(7p :11p ), associated with Mavilio F. Sequential activation of HOX 2 homeobox genes by myeloid leukemia with maturation. Blood 1987; 70: 1654–1658. retinoic acid in human embryonal carcinoma cells. Nature 1990; 7 Nakamura T, Largaespada DA, Lee MP, Johnson LA, Ohyashiki K, 346: 763–766. Toyama K, Chen SJ, Willman CL, Chen I-M, Feinberg AP, Jenkins 27 Kakizuka A, Miller WH, Umesono K, Warrell RPJ, Frankel SR, NA, Copeland NG, Shaughnessy JD Jr. Fusion of the Murty VVVS, Dmitrovsky E, Evans RM. Chromosomal translo- gene NUP98 to HOXA9 by the chromosome translocation cation t(15;17) in human acute promyelocytic leukemia fuses t(7;11)(p15;p15) in human myeloid leukaemia. Nat Genet 1996; RAR␣ with a novel putative transcription factor, PML. Cell 1991; 12: 154–158. 66: 663–674. 8 Borrow J, Shearman AM, Stanton VPJ, Becher R, Collins T, Willi- 28 Chang C-P, Shen W-F, Rozenfeld S, Lawrence HJ, Largman C, ams AJ, Dube I, Katz F, Kwong YL, Morris C, Ohyashiki K, Toyama Cleary ML. Pbx proteins display hexapeptide-dependent cooperat- K, Rowley J, Housman DE. The t(7;11)(p15;p15) translocation in ive DNA binding with a subset of Hox proteins. Genes Dev 1995; acute myeloid leukemia fuses the genes for nucleoporin NUP98 9: 663–674. and class I homeoprotein HOXA9. Nat Genet 1996; 12: 159–167. 29 Phelan ML, Rambaldi I, Featherstone MS. Cooperative interactions 9 Perkins A, Kong Suwan K, Visvader J, Adams JM, Cory S. Homeo- between HOX and PBX proteins mediated by a conserved peptide box gene expression plus autocrine growth factor production elic- motif. Mol Cell Biol 1995; 15: 3989–3997. its myeloid leukemia. Proc Natl Acad Sci USA 1990; 87: 8398– 8402. 30 Lu Q, Knoepfler PS, Scheele J, Wright DD, Kamps MP. Both Pbx1 10 Thorsteinsdottir U, Sauvageau G, Hough MR, Dragowska W, and E2A-Pbx1 bind the DNA motif ATCAATCAA cooperatively Lansdorp PM, Lawrence HJ, Largman C, Humphries RK. Over- with the products of multiple murine Hox genes, some of which expression of HOXA10 in murine hematopoietic cells perturbs are themselves oncogenes. Mol Cell Biol 1995; 15: 3786–3795. both myeloid and lymphoid differentiation and leads to acute 31 Popperl H, Bienz M, Studer M, Chan S-K, Aparicio S, Brenner S, myeloid leukemia. Mol Cell Biol 1997; 17: 495–505. Mann RS, Krumlauf R. Segmental expression of Hoxb-1 is con- 11 Moskow JJ, Bullrich F, Huebner K, Daar IO, Buchberg AM. Meis1, trolled by a highly conserved autoregulatory loop dependent upon a PBX1-related homeobox gene involved in myeloid leukemia in exd/pbx. Cell 1995; 81: 1031–1042. BXH-2 mice. Mol Cell Biol 1995; 15: 5434–5443. 32 Neuteboom S, Peltenburg L, van Dijk M, Murre C. The hexapep- 12 Nakamura T, Largaespada DA, Shaughnessy JDJ, Jenkins NA, tide LFPWMR in Hoxb-8 is required for cooperative DNA binding Copeland NG. Cooperative activation of Hoxa and Pbx1-related with Pbx1 and Pbx2 proteins. Proc Natl Acad Sci USA 1995; 92: genes in murine myeloid leukaemias. Nat Genet 1996; 12: 9166–9170. 149–153. 33 Look AT. Oncogenic transcription factors in the human acute leu- 13 Kroon E, Krosl J, Thorsteinsdottir U, Baban S, Buchberg AM, Sau- kemias. Science 1997; 278: 1059–1064. vageau G. Hoxa9 transforms primary bone marrow cells through 34 Yu BD, Hess JL, Horning SE, Brown GAJ, Korsmeyer SJ. Altered specific collaboration with Meis1a but not Pbx 1b. EMBO J 1998; Hox expression and segmental identity in Mll-mutant mice. Nat- 17: 3714–3725. ure 1995; 378: 505–508. 14 Shen W-F, Montgomery JC, Rozenfeld S, Lawrence HJ, Buchberg 35 Yu BD, Hanson RD, Hess JL, Horning SE, Korsmeyer SJ. MLL, a A, Largman C. The Abd-B-like Hox proteins stabilize DNA binding mammalian trithorax-group gene, functions as a transcriptional by the Meis1 homeodomain proteins. Mol Cell Biol 1997; 17: maintenance factor in morphogenesis. Proc Natl Acad Sci USA 6448–6458. 1998; 95: 10632–10636.