Background
HOX genes form a subset of the larger family of homeobox genes [
1], encoding transcription factors with a conserved 60 amino-acid, helix-turn-helix DNA-binding domain, known as homeodomain. Human HOX genes are organized on different chromosomes in four clusters A, B, C and D, consisting of nine to twelve tandem genes [
2]. Although firstly identified as morphogenetic regulators during embryonic development [
3], many evidences have shown that HOX containing genes play also a significant role in normal and leukemic haematopoiesis [
4]. In particular, in primitive CD34
+ populations HOXB cluster genes are coordinately transcribed during differentiation of myeloid, erythroid [
5,
6] and lymphoid cells [
7]. Also some HOXB genes have been associated with specific functions and stages of the hematopoietic maturation: overexpression of HOXB4 has been shown to favour self-renewal of more primitive populations over differentiation [
8], whereas HOXB6 expression is required for normal granulo- and monocytopoiesis and its deregulation associated with a maturation block [
9]. HOX genes as HOXA9, HOXC11 and HOXD13 have been implicated in chromosomal translocations associated with myeloid leukemia where they are fused with the nucleoporin gene NUP98 [
10]. Expression profiles of pediatric AMLs obtained by Real-time PCR arrays revealed a novel signature of HOX down regulated genes, including HOXB1 which results significantly repressed (mean values 23.5 in normal controls vs 0,8 in AMLs) [
11]. Even so the authors did not discuss its tumor suppressor role. Other HOX genes, as HOXA5 in breast cancer, have been described as tumor suppressor genes [
12,
13]. In addition HOXA5 loss of expression, due to promoter hypermethylation, has been also suggested to arrest normal differentiation in AML [
14]. Recently the first genome-wide survey of the DNA methylome performed in sporadic pituitary adenomas demonstrated the association between increased methylation of HOXB1 and its significantly reduced transcription [
15]. In the present study we showed that HOXB1 was expressed in normal lymphocytes, erythrocytes, granulocytes and monocytes as well as in human multipotent CD34+ cells purified from peripheral blood of healthy donors, whereas it was not detectable in a number of analyzed primary AML blasts and leukemic cell lines. The deficiency of HOXB1 in leukemic cells, in contrast with the reported wide spread expression of other HOXB genes in AMLs [
16], prompted us to investigate whether its enforced expression could restore any biological function pushing the leukemic blasts towards apoptosis and/or differentiation. Moreover, as it is known that epigenetic deregulation of critical genes can contribute to leukemogenesis [
17], we evaluated HOXB1 gene silencing as a consequence of promoter CpG island hypermethylation or histones acetylation in the HL60 cell line. Finally, trying to dissect the molecular pathways possibly triggered by HOXB1, we searched its downstream genes by using an Atlas Human Cancer macroarray.
Materials and methods
Cells and cell cultures
The leukemia cell lines, including promyelocytic HL60 and NB4, myeloblastic AML193, monocytic U937, erytroblastic K562 and the lymphoid T cell Peer and CCRF-CEM, were grown in RPMI 1640 medium (Gibco Invitrogen, Grand Island, NY), supplemented with heat-inactivated fetal bovine serum (FBS) (HyClone, Logan, Utah). HL60 cell line was also grown in the presence of differentiation factors: all trans retinoic acid (ATRA) (Sigma-Aldrich, St. Louis, MO) at 10-7 M and 1α,25 dihydroxyvitamin (VitD3) (Sigma-Aldrich, St. Louis, MO) at 10-8 M, over a period of 7 or 11 days of culture, respectively. When indicated HL60 cells were also treated with Z-Val-Ala-DL-Asp(OMe)-fluoromethylketone (z-VAD) (Bachem, Bubendorf, Switzerland) 25 μM alone or in combination with ATRA. The human teratocarcinoma (NT2D1) cell line, utilized as a positive control of HOXB1 expression, was grown in DMEM medium, 10% FBS supplemented and induced to differentiate by ATRA 10-7 M over a period of 9 days.
Cryopreserved cell samples obtained from a group of twelve patients with acute myeloid leukemia were studied and subclassified according to the FAB nomenclature (staged from M1 to M6) and cytogenetic analysis (7CN-AML lacking major translocation, 3INV16 and 2 t:15,17) [see Ref. 9]. The original samples (two for each group) contained a range of 20 to 500×10
6 cells and >80% of blastic infiltration. Leukocytes were isolated by Ficoll-Hypaque density centrifugation. Normal granulocytes, monocytes/macrophages, lymphocytes and erythroblasts were obtained from peripheral blood of healthy donors. CD34+ progenitor cells were purified from peripheral blood as reported [
18].
Retroviral gene transduction
The HOXB1 cDNA encompassing its complete coding sequence was cloned into the retroviral vector LXSN as LB1SN; the LXSN empty vector was always used as an internal control [
19]. AML193, U937, NB4 and HL60 cell lines were transduced with the LXSN empty vector and with LB1SN helper-free virus containing supernatants. Cells were treated twice for 4 hr with undiluted packaging cell supernatants in presence of 8 μg/ml of polybrene. Infected target cells were grown for 48 hr and then selected with G418 (0.8 mg/ml). As the ectopic expression of HOXB1 in AML193, U937 and NB4 cell lines was apparently lost in the first days after selection (see Additional file
1: Figure S1 and not shown), the subsequent functional studies were performed on the sole HL60 cell line.
RNA analysis
HOXB1 expression was evaluated either by traditional or Real-time RT-PCR. For the traditional technique relative quantifications were done by densitometric analysis after GAPDH samples normalization. When indicated PCR products were verified by southern blotting using an internal probe. Negative samples were confirmed after 40 amplification cycles.
Real-time RT–PCR was performed by the TaqMan technology, using the ABI PRISM 7700 DNA Sequence Detection System (Applied Biosystems, Foster City, CA) as reported [
19]. Commercial ready-to-use primers/probe mixes (Assays on Demand Products, Applied Biosystems) are listed: HOXB1: #Hs00157973_m1; early growth response 1 (EGR1): #Hs00152928_m1; fatty acid synthase (FASN): #Hs00188012_m1; mouse double minute 2 homolog (MDM2): #Hs00234760_m1; programmed cell death 10 (PDCD10): #Hs00200578_m1; caspase2 (CASP2): #Hs00154240_m1; non metastatic cells 1 protein (NME1): #Hs00264824_m1; secreted protein acidic and rich in cysteine (SPARC): #Hs00234160_m1, Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) #Hs4326317E.
cDNA expression array
Commercially available cDNA expression arrays (Atlas Human Cancer cDNA expression array 1.2, containing 1176 human genes involved in cancer, Clontech, Mountain View, CA) were used to compare gene expression of LXSN- and HOXB1-transduced HL60 cell line. Arrays, twice repeated, were screened according to the manufacturer’s protocol and as reported [
19]. The gene list of Table
1 was obtained by using 1.6 as cutoff value.
Table 1
Differentially expressed genes evaluated by macroarray in HL60/HOXB1 vs HL60/LXSN
BRCA2
| U43746 | Oncogenes & tumor | 0,10 | |
CCNI
| D50310 | Cell cycle regulators | 0,34 | |
EGR1
| X52541 | Transcriptional regulators | 0,21 | 0,02 |
FASN
| S80437 | Fatty acid/Lipid metabolism | 0,13 | 0,25 |
FBP2
| U69126 | Hydrolases/carbohydrate biosynthesis | 0,50 | |
MDM2
| Z12020 | Oncogenes/Apoptosis assoc. proteins | 0,31 | 0,5 |
OAS1
| M11810 | NucleotidylTransferases | 0,06 | |
SKY
| D17517 | Protein kinases receptors | 0,58 | |
SOD1
| K000454 | Antioxidant/Oxidoreductases | 0,65 | 0,7 |
TNFRSF1A
| L41690 | Growth factor receptors | 0,06 | |
AKAP1
| X97335 | Mitochondrial targeting of proteins | 1,94 | |
CASP2
| U13021 | Cysteine proteases/Caspases | 1,63 | 1,4 |
CCND3
| M92287 | Cell cycle regulators | 4,54 | |
CDC37
| U63131 | Cell cycle regulators | 7,03 | |
CRM1
| Y08614 | Transporter proteins | 2,43 | |
MAPRE1
| U24166 | Cytoskeleton Regulators | 1,62 | |
EIF3B
| U78525 | Initiation of translation factors | 2,49 | |
ERBB3
| M29366 | Receptor tyrosine kinases | 3,00 | |
JNK2
| L31951 | Intracellular kinase network members | 1,93 | |
KPNB1
| L38951 | Transporter proteins | 1,59 | |
NME1
| X17620 | Kinases/ Transferases | 2,62 | 1,4 |
PDCD10
| AF022385 | Apoptosis-associated proteins | 3,50 | 1,4 |
PTP4A1
| U48296 | Protein tyrosine phosphatases | 2,00 | |
RPS5
| U14970 | Ribosomal proteins | 3,00 | |
SPARC
| J03040 | Matrix-associated proteins | 2,85 | 1,5 |
ST13
| U28918 | HSC70-interacting proteins | 2,00 | |
TRAM
| X63679 | Secreted protein translocation | 2,55 | |
Western Blotting
Protein analysis was performed by immunoblot according to standard procedures. The primary antibodies used were: rabbit polyclonal anti-HOXB1 (Covance Research Products, Berkeley, CA); anti-apoptotic peptidase activating factor 1 (APAF1) and anti-BCL2-associated X protein (BAX) (BD, San Jose, CA); anti-histone deacetylase 4 (HDAC-4) and anti-caspase3 (CASP3) (Cell Signaling Technology, Beverly, MA); anti-B-cell CLL/lymphoma 2 (BCL2) and anti-myeloid cell leukemia1 (MCL-1) (Santa Cruz Biotechnology, Dallas, TX) and mouse monoclonal anti-actin (actin) (Calbiochem, La Jolla, CA).
In vitro growth and cell cycle assays
The proliferative rate of LXSN- and HOXB1-transduced cells was evaluated by a XTT-based colorimetric assay (Roche Molecular Biochemicals, Mannheim, Germany) [
19] and the Trypan-Blue exclusion dye test. Cell cycle analysis was performed using a CycleTEST™ PLUS Kit (BD, San Josè, CA) on HL60 cells, transduced or not with HOXB1.
Apoptosis assay
For each sample 105 cells were incubated and stained according to standard procedures (TACS™ AnnexinV-FITC apoptosis detection Kit) (R&D Systems Inc, Minneapolis, MN). Results were expressed as total absolute percentages of AnnexinV+, Annexin+/PI+and PI+ gated cells.
Apoptosis was also evaluated by the ApoONE Homogenous Caspase 3/7 Assay. A spectrofluorometer 96 wells plate reader (Wallac VICTOR2, Turku, Finland) was used for measuring the fluorescence of 5×104 cells/well of both HL60/LXSN and HL60/HOXB1. Cells were kept in 1% FBS or in 10% FBS. As a control, cells were grown in the presence of staurosporine at 200nM for 1 hr.
Cell surface markers and morphological analysis
To evaluate the granulocytic and monocytic differentiation capacities, LXSN- and HOXB1- transduced HL60 cells were grown in vitro up to 7 or 11 days in the presence of 10-7 M ATRA or 10-8 M VitD3, respectively. Cells were then analyzed for cell surface markers and morphology. Specifically, the cells were labelled with anti-CD11b and anti-G-CSF receptor (G-CSFR) (for G-lineage differentiation), double stained with anti-CD14/anti-CD11b (for M-lineage differentiation) (Pharmingen, San Diego, CA) and subjected to FACS analysis (FACS Scan Becton Dickinson, San Diego, CA).
Cell morphology was evaluated on May-Grünwald-Giemsa stained slides according to standard criteria. Classification includes blasts, promonocytes and promyelocytes as intermediate cells, and monocytes, myelocytes and beyond as mature cells. Three separate experiments were analyzed by two independent blind observers.
Epigenetic analysis of HOXB1 promoter
The methylation status of CpG islands of HOXB1 promoter was evaluated by the SABiosciencesEpiTect Methyl DNA Restriction kit (Qiagen, Gaithersburg, MD) [
20]. HOXB1 CpG island location was Chr17:46607804–46608390. Related RefSeq ID: NM_002144 (HOXB1). Briefly, 250 ng of DNA-RNA free, extracted by the DNeasy blood and tissue KIT (Qiagen), were digested in four equal reactions with no enzymes, methylation-sensitive enzyme, methylation-dependent enzyme, or both enzymes according to the manual instructions (EpiTect® Methyl qPCR Assay Handbook,
http://www.qiagen.com). To determine the relative amounts of hypermethylated (HM), intermediately methylated (IM) and unmethylated (UM) DNAs, the products of these reactions were amplified by SABiosiences EpiTect Methyl qPCR primer assay for human HOXB1 (MePH22204-2A). To analyze the effects of demethylation on HOXB1 gene expression, we treated HL60 cells (0,5×10
6/ml) for 1 up to 5 days with the demethylating agent 5-Azacytidine (5-AzaC) at 1 μM and 5 μM concentrations (Sigma-Aldrich, Saint Louis, MO), replacing medium and adding new 5-AzaC every 48 hrs. Moreover, to evaluate HOXB1 epigenetic regulation by the histones acetylation-deacetylation mechanisms, we treated the HL60 cells (0,5×10
6/ml) with 100 or 600 ng of the histone deacetylase inhibitor Trichostatin A (TSA) (Sigma-Aldrich) for 48 and 72 hr [
21]. Following all the above mentioned treatments, we searched for HOXB1 mRNA re-expression in HL60 cells by RT-PCR.
Statistical analysis
All the experiments were repeated at least three times, unless otherwise stated. Reported values represent mean ± standard errors (S.E). The significance of differences between experimental variables was determined using parametric Student’s t-test with P < 0.05 deemed statistically significant. P-values relative to HOXB1-transduced cells were always referred to LXSN-transduced cells.
Discussion
Numerous reports have catalogued differences in HOX genes expression between normal and neoplastic cells, but their functional relationship with the malignant phenotype in many cases remained elusive [
22]. HOX genes are currently under evaluation in order to correlate specific HOX alterations with changes in cellular processes such as cell proliferation, differentiation and apoptosis. Other than HOX overexpression, also HOX downregulation has been associated with different malignancies, including leukemia. Examples of tumor suppressors are the homeodomain protein NKX3.1 and HOXD10 commonly down-regulated in human prostate cancer [
23], breast tumor cells and gastric carcinogenesis [
24,
25]. In addition HOXA5 expression is lost in breast tumors [
12] and HOXA genes, normally playing suppressor roles in leukemia development, are frequent targets for gene inactivation [
26]. Accordingly, expression studies indicated a set of seven downregulated HOX genes (HOXA3, A4, A5, A7, B1, B9, C9) as significantly clustered in pediatric AMLs [
11].
In this study we propose HOXB1 as an additional member of the HOX family with tumor suppressor properties. HOXB1 is expressed in terminally differentiated blood cells (erythrocytes, granulocytes, monocytes and lymphocytes) and in CD34+ progenitors from peripheral blood, but not in primary blasts from M1 to M5 and myeloid cell lines. Our results indicate a mechanism of CpG island promoter hypermethylation at the basis of HOXB1 silencing in AML as demonstrated by the higher amount of the hypermethylated DNA fraction in HL60 cells compared to normal cells. Accordingly, the demethylating agent 5-AzaC was able to reactivate HOXB1 expression in HL60 cells, whereas treatment with the histone deacetylase inhibitor TSA had no effect.
Results obtained by HOXB1 gene transduction in HL60, in agreement with the rapid counter-selection of the ectopic HOXB1 in AML193, U937 and NB4 cell lines (Additional file
1: Figure S1), point to the contribution of HOXB1 abnormal silencing to the survival of myeloid leukemic cells.
In HL60, HOXB1 restored expression was per se able to induce apoptosis and, in the presence of ATRA or VitD3, to favour maturation towards granulocytic and monocytic differentiation pathways, respectively. Of note, the HOXB1 induced differentiation, visible in ATRA-treated cells, does not appear associated with the apoptotic process, as shown by ATRA + z-VAD treatment.
According to our Atlas macroarray analysis, we identified a number of HOXB1 dependent up- and down-modulated genes. Specifically, we observed the up-regulation of some apoptosis-related genes as CASP2, JNK2, PDCD10, SPARC and heat-shock protein 70 kD-interacting protein (ST13). In particular CASP2, JNK2, PDCD10, and ST13 have been associated with mitochondrial permeabilization [
27‐
30] and with the induction of the apoptotic process, while SPARC overexpression seems to play a tumor suppressor function in some low expressing SPARC AMLs [
31,
32]. As in HOXB1-transduced cells we also observed a significant enhancement of APAF1 (Figure
2e), we suggest the involvement of HOXB1 in triggering the mitochondrial as well as caspase dependent apoptotic pathways [
33], as indicated by the activation of caspase 3/7 (Figure
2d,e). Accordingly we also detected a HOXB1-dependent regulation of the BCL-2 family of proteins playing a major role in the control of apoptosis. In particular, the proapoptotic role of HOXB1 was sustained by the induction of BAX and the downregulation of MCL1 proteins. Moreover the BAX/BCL2 ratio, doubled by HOXB1, was indicative to increased cell susceptibility to apoptosis [
34]
.
In addition, the macroarray analysis showed the HOXB1-dependent downregulation of some antiapoptotic genes as MDM2, FASN, the antioxidant enzyme superoxidedismutase (SOD1) and the breast cancer susceptibility gene 2 (BRCA2). As the knockdown of MDM2 in p53 mutant non-small cell lung cancer [
35,
36], the FASN reduced expression in HepG2 cells [
37,
38] or the SOD1 downregulation in AMLs [
39,
40] can induce apoptosis, we might suggest a HOXB1 related anticancer activity. Nonetheless, as p53 is not expressed in HL60 cells, we should consider the involvement of other members of the p53 family, as p63 and p73 expressed in HL60 cells [
41]. Specifically p63 has been described to be activated by PBX cofactors [
42] and in HL60 cells we observed a HOXB1-related induction of PBX2 (data not shown), thus possibly suggesting the effectiveness of p63 downstream to HOXB1.
Finally, EGR1 displayed a striking downregulation. Although deserving further studies due to its complex and somehow divergent activities, its reduction was in agreement with the lower tumorigenicity of HL60 cells overexpressing HOXB1. In fact EGR1 has been reported to play a role in prostate tumor growth and survival [
43] and its abnormal expression has been recently associated with tumor invasion and metastasis in gastric cancer [
44]. In addition, a higher level of EGR1 has been associated with relapsing AML respect to AML at diagnosis with a direct correlation with increased proliferation and enhanced RAF/MEK/ERK1/2 activation [
45].
In conclusion our results indicate an antineoplastic role for HOXB1 in AMLs through its functional involvement in promoting apoptosis and powering ATRA-induced differentiation. Considering the presence of two RARE elements at the 5′ and 3′ ends of HOXB1 [
46], we might suggest a role for HOXB1 in ATRA-mediated anticancer activity. In this view a HOXB1/ATRA combination might represent a possible future therapeutic strategy in AML [
47,
48].
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
Conceived and designed the experiments: MP, FF and AC. Performed the experiments MP, FF, LB, MCE, OM, AB and ADF. Wrote the paper: MP, FF and AC. All authors read and approved the final manuscript.