Epigenetic Regulations in Hematopoietic Hox Code

HHe1, X Hua2 and J Yan1

1Institute for Marine Biosystem and Neurosciences, Department of Hydrobiology, College of Fisheries and Life Sciences, Shanghai Ocean University, Lingang New City, Shanghai, PR China and 2Abramson Family Cancer Research Institute, Department of Cancer Biology, University of Pennsylvania, Philadelphia, PA, USA

Hox genes encode DNA-binding proteins that are tic cells (Magli et al., 1997; Owens and Hawley, 2002). deployed in overlapping domains along various body axes Recent discoveries revealed that the menin and during embryonic development. This sequential activation MLL-containing histone methyltransferase complex is of Hox genes in temporal and spatial mode, the Hox code, required for Hox gene-dependent proliferation and is critical for the proper positioning of segmented differentiation of hematopoietic progenitors and leuke- structures along those axes, which include the vertebrate, mia stem cells (Hess, 2004; Hughes et al., 2004; Milne limbs and, also digestive and reproductive tracts. et al., 2005; Yokoyama et al., 2005, 2010; Chen et al., It remains unknown how Hox genes are regulated to 2006; Thiel et al., 2010). These studies have given determine the identity of hematopoietic stem cells and emphasis to epigenetic regulations in Hox-dependent their derivatives, which migrate and express most hematopoiesis and leukemogenesis. Hox genes. The key questions are whether the hematopoietic system has an axis, how epigenetic mechanisms Homeobox genes and Hox code restrict expression of Hox genes to specific cell types and what role Hox genes play in leukemic transformation? As the first homeobox (Hox) genes were discovered in Taking in account these questions, we propose a Drosophila, over 1000 homeodomain proteins have combinatorial axial model of hematopoietic Hox code to been identified in several species. These proteins are predict the positional identity of the hematopoietic cells. evolutionarily conserved among metazoans (refer to: This model will provide new insight into epigenetic therapy Hox ProDB at http://www.iephb.nw.ru/labs/lab38/ in leukemia. spirov/hox_pro/hox-pro00.html) (Spirov et al., 2002). Oncogene (2011) 30, 379–388; doi:10.1038/onc.2010.484; Invertebrates, including Drosophila, have a single cluster published online 25 October 2010 of Hox genes, but genome duplication events during Keywords: Hox code; epigenetic; hematopoiesis; vertebrate evolution have produced a minimum of four leukemia; polycomb and trithorax groups clusters in jawed vertebrates and a seven-cluster organization for zebrafish (Danio rerio) (Amores et al., 1998). A total of 39 mammalian Hox genes have been identified. They are distributed at four separate chromosomal clusters designated HoxA, HoxB, HoxC and Introduction HoxD. In each cluster, 13 paralogue groups are arranged from position 1 to 13 in a 30–50 order. The The regulation of the patterning of the embryonic trunk same position represents their homology among clus- and tail is a basic function of the homeobox-containing ters, sharing high sequence identity and functional Hox gene family in Drosophila (Deschamps and van redundancy. The cluster feature distinguishes Hox Nes, 2005). However, various downstream targets of genes from those ‘dispersed’ Hox-like genes (ParaHox). Hox genes in the developing central nervous system The latter also encode a homeodomain similar to the suggest that Hox genes encoding DNA-binding tran- clustered Hox genes, but do not display a clustered scription factors not only play a role in patterning of the organization. These Hox-like genes are grouped to- central nervous system during early development, but gether into various families based on their functional also contribute to cell specification and identity (Akin and structural characteristics (Garcia-Fernandez, 2005), and Nazarali, 2005). Increasing evidence suggests that for example, Cdx and Pbx genes. Hox transcription factors play a definitive role in What makes Hox genes special among developmental normal proliferation and differentiation of hematopoie- regulators is not only that their clustered organization is complex in chromosomes, but also that the distinct Correspondence: Dr J Yan, Institute for Marine Biosystem and combination of Hox genes is expressed in a specific Neurosciences, College of Fisheries and Life Sciences, Shanghai Ocean anterior–posterior region (referred to as ‘Hox code’) University, 999 Hucheng Huan Road, Lingang New City, Shanghai (Gruss and Kessel, 1991; Kessel and Gruss, 1991; 201306, PR China. E-mail: [email protected] Knittel et al., 1995). In vertebrates, the Hox code is Received 16 July 2010; revised and accepted 14 September 2010; characterized by four unique features: (1) Spatial published online 25 October 2010 colinearity. There is a 30–50 colinear sequence in which Epigenetic regulations in hematopoietic Hox code HHeet al 380 genes at the extreme 30 end of the individual clusters more differentiated cell types (such as committed have the most anterior boundaries of expression, progenitors) (Sauvageau et al., 1994). During definitive whereas the anterior expression domains of genes at hematopoiesis, there is a loss of Hox gene expression the 50 end are localized more caudally (Garcia-Fernan- along differentiation; expression of Hox genes decreases dez, 2005). (2) Temporal colinearity. 30 Hox genes are along differentiation in a manner that seems to follow expressed first, whereas more 50 Hox genes are expressed their chromosomal position: that is, 30-positioned genes later and sequentially. (3) Posterior prevalence. The are downregulated earlier than 50-positioned genes. In products of more posterior genes in the cluster tend to the more differentiated bone marrow cells represented be prevalent over those of more anterior genes, and by the CD34À phenotype, expression levels of the Hox more caudal body levels tend to show an increasing genes are low or absent. In regard to CD34 þ represen- amount and diversity of Hox products (Duboule and tative of primitive HSCs (Guo et al., 2003), further study Morata, 1994). (4) Overlapping expression. Most para- demonstrates that Hox genes from clusters A and B are logous genes have identical or similar domains of preferentially expressed in the primitive HSC-enriched expression, indicating some degree of functional redun- Sca-1 þ LinÀ subpopulation. Their expression is de- dancy. There are no dramatic alterations in morphogen- creased in the HSC-low and depleted subpopulations esis caused by mutations of a single Hox gene in (Sca-1 þ Lin þ and Sca-1ÀLin þ ) (Pineault et al., 2002). vertebrate. For example, Hoxa3, Hoxb3, and Hoxd3 According to quantitative PCR analysis of expression of generally have very similar patterning, and gene target- all Hox genes in the cell populations from human ing has shown that members of paralogue group 3 peripheral blood, the highest level of expression of the functionally compensate for each other when one HoxA and HoxC genes is generally detected, with HoxD paralogous gene is disrupted (Manley and Capecchi, and HoxB genes generally exhibiting a 10- or 100-fold 1997, 1998; Gaufo et al., 2003). However, mice with lower expression, respectively (Morgan and Whiting, conditional knockout of all nine HoxC genes die at 2008). These results suggest that Hox gene expression is birth, suggesting that genes in each Hox cluster are preferentially restricted to undifferentiated or prolifera- collectively necessary for the development of mice tive progenitor cells in a predispositional manner (Suemori and Noguchi, 2000). (Figure 1a). Recent studies support that Hox genes have an Moreover, Hox genes are expressed in an apparent enduring role in maintaining positional identity lineage-specific fashion, with HoxB and some HoxC throughout the lifetime of an organism. For instance, cluster genes primarily expressed in cell lines with adult fibroblasts consistently express distinct patterns of erythroid features and HoxA cluster genes predomi- Hox genes that are able to predict the original position nantly expressed in cell lines with myeloid features of the cells along three developmental axes (Rinn et al., (Lawrence et al., 1996). Populations enriched for 2006), and these findings also indicate that the pattern or granulocyte/macrophage myeloid progenitors preferen- Hox code of these cells can be epigenetically inherited or tially express genes located at the 50 region of the cluster memorized. Differential expression of the HoxA and (Hoxa9, b9, a10). These 50 cluster genes are coordinately HoxD genes reflects the location of a fibroblast at the activated in blocks in myeloid leukemia cells (Celetti proximal–distal axis along the upper and lower limbs, et al., 1993). In contrast, only a few scattered Hox genes whereas differential expression of HoxC genes strongly are switched on in cell lines with B-or T-lymphoid correlates with anterior–posterior location along the potential and early erythroid progenitors. Comparisons trunk, and expression of HoxB genes is associated with of Hox gene expression profiles of different progenitors origin from internal organs rather than from skin (Rinn indicate that fetal liver and adult bone marrow stem et al., 2006; Wang et al., 2009). These studies have cells are similar in terms of Hox gene expression profile, greatly increased our understanding of trafficking and but significantly different from T-cell progenitors in homing of tissue-specific progenitors, as reflected by child thymus. Hoxb3 and Hoxc4 are expressed through- their Hox expression profiles. It is interesting to learn out thymocyte development with a sequential loss of 30 about how Hox genes define the positional identity of region HoxA cluster genes (Taghon et al., 2003). As the hematopoietic cells. most immature thymocytes are derived from immigrated fetal liver and adult bone marrow stem cells, this result suggests that Hox gene expression is not only closely associated with the origin and homing niche of the Expression patterns of Hox genes in hematopoietic cells progenitors, but also is involved in proliferation and differentiation of hematopoietic cells. All hematopoietic progenitors express Hox genes in a The biological implication of the Hox code in pattern characteristic of the lineage and stage of hematopoiesis (referred to as hematopoietic Hox code) differentiation of the cell (Figure 1a). In murine is extensive. On one hand, Hox genes are orderly CD34 þ cells (previously known as primary stem cells), activated in normal hematopoietic progenitors and at least 22 of the 39 Hox genes are expressed (Grier preleukemia cells (Kumar et al., 2004; Yan et al., et al., 2005). 30 Region of Hox genes (characterized by 2006a), in accordance with the colinear expression the numbers 1–6) are maximally expressed in the most pattern that is observed during embryonic development. primitive cells (that is, hematopoietic stem cells, aka One the other hand, the expression patterns of HSCs), whereas 50 region genes (Hox7–13) are present in paralogous Hox genes seen in embryonic development

Oncogene Epigenetic regulations in hematopoietic Hox code HHeet al 381

Figure 1 Hox regulation and proposed hematopoietic Hox code. (a) Hox regulation in hematopoiesis and leukemogenesis. During hematopoiesis from HSCs to multilineage progenitors, and then, to differentiated effectors, menin-MLL-modulated chromatin modifications regulate the Cdx-Hox pathway and direct HSC differentiations (Top). However, the MLL–AF9 complex perturbs this process, but deregulates 50-Hoxa cluster genes and leads to leukemic transformation from HSC to pre-leukemia and leukemia cells (Bottom). The middle of the figure indicates that Hox genes are sequentially activated from 30 to 50, with some overlapping along the anterior–posterior (AP) axis. Note the differences in AP position, as well as the overlap differences in the primaxial versus abaxial expression patterns of 30–50 Hox cluster genes between hematopoiesis and leukemogenesis. (b) A tentative hematopoietic Hox code. A proposed diagram illustrates the cellular positions and the progenitor origins in a vertebrate body by the four quarters along AP, dorsal–distal (D–Ds), proximal (Px) and ventral–distal (V–Ds) axes. What the Hox code represents is a somewhat digital mechanism for regulating Hox axial patterning. Hox genes are arranged from 30 to 50 along the AP axis; A–D clusters are abaxially expressed in the bone marrow of head, limb and spinal bones; B–C clusters are primaxially expressed in internal tissues and organs. The cell positions can be predicted onto the four arbitrary quarters by integrating expression levels of all four Hox cluster genes and the origin of progenitors. The diagram described is at a rather coarse level, revealing broad chunks of the Hox regulatory scheme, but future work should refine the details and specific and finer aspects of positional regulation. are not obvious in hematopoietic progenitors and cell virtually integrates a combinatorial expression pattern lines, but rather whole clusters (or large regions of a of Hox cluster genes by quantification along various cluster) are turned on or off in a lineage-specific manner. body axes to define the cellular behaviors, including The redundant Hox activity is required for the homing and trafficking versus proliferation and differ- proliferative expansion of hematopoietic progenitor entiation (Figure 1b). cells, whereas subsequent differentiation is influenced by particular Hox genes (Ernst et al., 2004). Consistent with multiple Hox genes, no single growth factor or cytokine but certain combinations of two or more can Regulation of Hox gene transcription support expansion of HSCs ex vivo (Huynh et al., 2008). Thus, vertebrates have these parallel, overlapping sets Hox genes are transcribed in a segment-specific manner. of Hox genes so that phenotype could be a consequence The presence of multiple negative and positive regula- of a combinatorial expression of four cluster genes. tory elements in the promoter of mouse Hox genes, We conjecture that the cellular positional identity can which are required at different stages of development, be defined with more gradations by mixing up the suggests a biphasic mechanism of vertebrate Hox boundary of expression of each gene. Taking in account gene expression: the elements responsible for correct migration and related differentiation of hematopoietic initiation of Hox gene expression and maintenance cells, we propose that the hematopoietic Hox code elements controlling its proper maintenance (Gutman

Oncogene Epigenetic regulations in hematopoietic Hox code HHeet al 382 et al., 1994). Between the initiation and maintenance et al., 2006a). Overexpression of Cdx4 can activate a phases, a transition phase is proposed to establish a reporter gene driven by Cdx-protein-binding sites within proper Hox expression domain in vertebrates. In a Hoxa9 gene. However, in a hematopoietic cell line, Drosophila, regulation of Hox genes is divided into Cdx4-Hoxa9 biding is abolished by ablation of menin three phases: initiation, modulation and maintenance (Yan et al., 2006a), which is a cofactor of the histone (Vasanthi and Mishra, 2008). methyltransferase–MLL complex (Yu et al., 1998; Yokoyama et al., 2004, 2005). These results suggest that an active chromatin domain is a prerequisite for Initiation of Hox expression formation of the Cdx–Hox transcription initiating The precise mechanisms of initial Hox transcription are complex. not well understood. Several upstream regulatory A two-step model has been postulated to explain how factors have been suggested in modulating Hox gene chromatin structure regulates transcription of the initial expression during its anteriorward spreading, such as domain of Hox clusters (Chambeyron and Bickmore, Wnt, retinoic acid (RA), fibroblast growth factor (Fgf), 2004). According to this model, a locus-wide change in Krox20 and Kreisler (Deschamps and van Nes, 2005). high-order chromatin structure induced by histone The other important upstream regulators of Hox gene modifications and chromatin decondensation may first expression are Cdx proteins. The vertebrate Cdx genes create a transcriptionally poised state. Then a progres- (Cdx1,Cdx2 and Cdx4 in the mouse) encode home- sive 30–50 change in high-order chromatin structure, odomain transcription factors related to the Drosophila which is manifested as the movement outside of caudal genes. Numerous studies indicate that Cdx genes chromatin territories, is necessary to allow for the are targets of Wnt, RA and Fgf signaling. They can programmed expression of genes along the cluster. integrate multiple signaling pathways, such as Fgfs, Wnt In parallel with Hox colinearity, a sequential opening and RA, to control Hox expression (Bel-Vialar et al., of chromatin, starting at the 30 end of a cluster and 2002; Lohnes, 2003). Importantly, Cdx4 induces wide- moving successively toward 50, leads to the release of spread alterations in Hox expression level, including silencing first at the 30 end, and sequentially, allowing Hoxb4, Hoxb3, Hoxb8 and Hoxa9, all of which have the expression of more 50 genes (Roelen et al., 2002; been implicated in maintenance of HSCs or expansion Kmita and Duboule, 2003; Duboule and Deschamps, of immature progenitors (Davidson et al., 2003). 2004). In hematopoietic cells, menin–MLL complexes Accordingly, the Fgf, Wnt, RA-Cdx4-Hoxa9 pathway modulate the chromatin accessibility to upstream is proposed to direct the differentiation of the hemato- regulators of Hox genes (Figure 1a). poietic progenitors (Figure 1a). To correctly implement the early activation process of An obvious question is how these transcription the cluster genes, neighboring Hox genes are coordi- factors or regulators are sequentially recruited to the nately regulated by local and global enhancer sequences. target Hox genes. Generally, recruitment of specific Local enhancer functions are shared among subsets of DNA-binding proteins to the promoter sequence is neighboring genes. They ultimately provide distinct essential for transcriptional initiation. A variety of expression features through unequal partitioning of regulatory elements, including general and tissue-specific their activity on the genes they control (Sharpe et al., regulatory elements, have been identified in several Hox 1998). In contrast, a global enhancer is located outside genes: Hoxb3 (Kwan et al., 2001), Hoxb4 (Gutman the clusters, which controls a series of contiguous genes et al., 1994), Hoxa5 (Nowling et al., 1999), Hoxa7 at once in a relatively promoter-unspecific manner (Spitz (Puschel et al., 1991; Knittel et al., 1995), Hoxa9 (Yan et al., 2003). Use of the auto-, cross- or para-regulatory et al., 2006a) and Hoxa10 (Akbas et al., 2004). These loop between Hox genes, paraHox genes and other regulatory elements are much conserved among cluster factors (Manzanares et al., 2001) is an exceptional genes and even among interspecies of Hox genes from mechanism to control Hox expression level. For human and mouse (Knittel et al., 1995; Kwan et al., instance, human Hoxd4 gene product functionally 2001; Lehoczky et al., 2004). Displacing the regulatory interacts with its upstream sequence. This upstream elements outside of the Hox complex can completely or sequence is strikingly homologous to the sequence partially recapitulate endogenous expression pattern of within the human Hoxd4 spatial enhancer (Popperl a Hox gene in transgenic mice. and Featherstone, 1992) and functions as an autoregu- The most well-characterized regulatory elements are latory element. functional retinoic acid response elements, which have been identified from Hoxa1, Hoxb1, Hoxb4, Hoxd4, Hoxb5, Hoxb6 and Hoxb8 genes (Oosterveen et al., Establishment of Hox expression domain 2003a, b; Kobrossy et al., 2006). Similarly, the transcrip- The other question is how the limit of expression tional stimulation of the Hox genes by Cdx proteins boundaries and tissue restriction of Hox genes links to occurs at the Cdx-binding sites (TTTATG), which are the genomic organization of the genes. Eukaryote often found in clusters throughout the Hox complexes chromosomes are subdivided into a series of discrete (Gaunt et al., 2004; Tabaries et al., 2005). Two Cdx4- higher-order chromatin domains. Each domain has one binding sites have been identified in the 50 regulatory or more transcription units, a locus control region that region of mouse Hoxa9 gene: one contains a tandem is required for initiation of transcription and a matrix repeat unit and the other a single Cdx-binding site (Yan attachment site that attaches the domain to the nuclear

Oncogene Epigenetic regulations in hematopoietic Hox code HHeet al 383 matrix. The autonomy of each chromatin domain is dose-dependent manner, and Cdx gradients might serve delimited by the special cis-acting elements called as instructional (morphogen) gradients for setting Hox boundaries or insulators (West et al., 2002; Felsenfeld expression patterns. In this model, temporal colinearity et al., 2004; Zhou and Berger, 2004). In accordance with and Cdx gradients operate together in the establishment chromatin domain, 50 HoxA cluster genes are sequen- of Hox boundaries (Gaunt et al., 2004, 2008). Cdx tially activated in MLL-AF9-transduced hematopoietic protein concentration may regulate the rate and extent cells. All 50 HoxA cluster genes (Hoxa5–13) are of sequential Hox gene activation (temporal colinearity). hyperexpressed in MLL-AF9 leukemia cells, whereas The instructional gradient model could explain why only Hoxa5–9 genes are activated in MLL-AF9 multiple Hox genes are regulated in dose-dependent preleukemic cells (Kumar et al., 2004; Yokoyama activation in primitive progenitors and downregulation et al., 2005; Yan et al., 2006a). These data suggest that at the time of differentiation (Figure 1a). In this regard, 50 HoxA cluster genes are divided into at least two environmental cues supplied by the microenvironment chromatin domains, whereas Hoxa5–9 genes are tran- may play a primary role in lineage determination (Wei scriptionally delimited from Hoxa10–13 by a speculated et al., 2008). insulator. A similar insulator, polar silencer, is located between Hoxd12 and Hoxd13 in mice. Expression of Maintenance of Hox expression pattern genes in the 50 HoxD region is driven in appropriate The established expression patterns of some Hox genes tissues by the digit enhancer, the hernia enhancer, or must be precisely and clonally maintained throughout both. The digit and hernia enhancers are located at development. The Polycomb group (PcG) and Trithorax either end of the cluster. The polar silencer could block group (TrxG) genes have been recognized as essential one of the enhancers from activating the neighboring regulatory factors that are involved in epigenetic genes beyond the chromatin domain (Kondo et al., maintenance (Dejardin and Cavalli, 2004). The PcG/ 1998; Kmita et al., 2000; Monge et al., 2003). TrxG system and PRE/TRE-binding elements are The chromatin boundary can also be delimited by detailed in the Supplementary data. CpG methylation. Unmethylated CpG-island regions Distinct methylation patterns of specific lysines in are generally flanked by regions of less-CpG-rich DNA, histone H3 and H4 produce heritable marks that can which is heavily methylated (Jones and Baylin, 2002). continue recruiting the PcG complex to repressor- The separation of unmethylated promoter CpG-island binding sites (Cao et al., 2002; Muller et al., 2002; Cao regions from immediately flanking areas of methylation and Zhang, 2004a, b) or the TrxG complex to active seem to involve a functional boundary (Jones and promoters (Milne et al., 2002; Nakamura et al., 2002). A Baylin, 2002). Currently, there is no report of this sophisticated pathway is proposed to sequentially mechanism for formation of Hox gene boundary. recruit PcG/TrX proteins to chromatin maintenance The establishment of ordered domains of Hox gene elements along HoxA cluster genes (Figure 2). Once the expression is thought to be controlled by gradients of core factors are recruited to PRE/TRE sites of the target various putative morphogens, such as RA (Faiella et al., Hox gene, multimeric PcG and TrxG protein complexes 1994; Maves and Kimmel, 2005) and Cdx proteins. Hox modulate the target chromatin structure and accessi- expression boundaries respond to Cdx proteins in a bility via deposition of specific posttranslational histone

Figure 2 The depicted models summarize graphically Hox epigenetic initiation and maintenance. Here, we take MLL-regulating Hox gene expression as an example. The Cdx4 containing initiation complex initiates Hox activation, whereas the PcG complex remodels chromatin structure and delimits the chromatin domain. To maintain the Hox active state, both PcG and TrxG complexes (that is, MLL super complex) are recruited to guarantee a dynamic equilibrium between active and repressed histone modiﬁcations, such as H3K4 di-/tri methylation (H3K4M2/M3), H3K9M2/M3 and H3K27M2/M3 (a). (b) To silence Hox expression, the PcG protein Ezh2 and methylated H3K9 and H3K27 facilitate target DNA methylation. TF, tissue-speciﬁc transcription factor. An assumed chromatin domain isolator is indicated.

Oncogene Epigenetic regulations in hematopoietic Hox code HHeet al 384 modification marks. The E(z)–Esc (PRC2) complex is H3K9 and H3K27 methylations can direct DNA responsible for catalyzing repressive histone methyla- methylation (Tamaru and Selker, 2001; Jackson et al., tion. E(z) contains a SET domain (named after the three 2002, 2004; Mathieu et al., 2005). The PcG SET domain proteins: Su(var)3–9, E(z) and Trithorax). The SET- containing protein EZH2 interacts with DNA methyl- domain proteins possess histone lysine methyltransfer- transferases (DNMT1, DNMT3A and DNMT3B) and ase activity (Dillon et al., 2005). The E(z) SET domain is required for DNA methylation of EZH2 target preferentially methylates histone K27 to a minor extent promoters (Vire et al., 2006). and also K9 of H3 (Muller et al., 2002; Cao and Zhang, Second, CpG-binding protein (CXXC finger 2004b), whereas the Su(var)3–9 SET domain selectively protein 1, CFP1) is a component of a mammalian methylates H3K9. In contrast, PRC1 does not contain SET1 histone H3K4 methyltransferase and acts as a any enzymatic activity, but can be recruited to critical epigenetic regulator of both cytosine and histone chromatin by the Pc chromodomain-mediated recogni- methylations (Lee and Skalnik, 2005). MLL, a mamma- tion of H3K27M3 mark, which is deposited by PRC2 lian trithorax protein, contains multiple domains, (Wang et al., 2004). Thus, the PRC1 complex is including a DNA methyltransferase homology domain hypothesized to create a stably repressed chromatin (DNMT1) in the N terminus and a highly conserved C- structure through recognition of the E(z)-mediated terminal SET domain. Loss of the SET domain is H3K27 trimethylation. In addition, PRC1 inhibits associated with a dramatic reduction in histone H3 chromatin remodeling by SWI–SNF chromatin-remo- lysine 4 monomethylation and DNA methylation deling complexes (Shao et al., 1999; Saurin et al., 2001; defects at the same Hox loci (Terranova et al., 2006). Francis et al., 2004) and maintains a repressive Although its DNMT1 domain binds in vitro to chromatin structure. The presence of TATA-binding unmethylated CpG-rich DNA (Fuks et al., 2000), Mll protein Associated Factors in PRC1 suggests that its binds to specific clusters of CpG residues within the components might prevent the transition from recruit- Hoxa9 locus and protects the CpG clusters from DNA ment to the promoter to initiate transcription (Dellino methylation in murine embryonic fibroblast cells (Er- et al., 2004). furth et al., 2008), providing another mechanism to The induced state of H3K4 methylation seems not to maintain Hox gene activation. persist indefinitely through yeast cell divisions but is Third, methyl-CpG-binding proteins and DNA maintained for a significant portion of an individual cell methyltransferases are associated with histone deacety- cycle after transcriptional inactivation and SET1 dis- lases (HDACs) (Esteller, 2006). Both DNMT1 and sociation. This phenomenon is termed as ‘short-term DNMT3a have been shown to interact through several memory’ of recent transcription (Ng et al., 2003). methyl-CpG-binding proteins (MECPs) that read DNA- However, new evidence has indicated that mammalian methylation patterns. For instance, MECP2 forms a methylated histone H3K4 is subjected to continuous complex with HDACs and a corepressor protein, Sin3a, dynamic turnover of acetylation (Hazzalin and Maha- to repress transcription in a methylation-dependent devan, 2005). Hypermethylation and hyperacetylation manner. Another methyl-CpG-binding protein MBD2 are segregated to the isolated subregions of mitotic forms a complex with the multi-subunit NuRD complex chromatin and generate a stable epigenetic mark for (previously known as MECP1), which contains an ATP- active chromatin (Kouskouti and Talianidis, 2005). dependent chromatin-remodeling protein, Mi-2, and Acetylation-based epigenetic code may operate through HDACs. The MECP2-Sin3a-HDACs and MBD2– both mitosis and meiosis (Turner, 2000). The most long- NuRD complexes provide a mechanistic link between standing and highly conserved example of lysine-specific DNA hypermethylation and histone deacetylation in histone acetylation is the di-acetylation of newly transcriptional repression. synthesized H4 at lysine 5 and 12, in which form it is The molecular links between histone modifications deposited on newly replicated DNA during chromatin and DNA methylation can strengthen the stability of assembly (Sobel et al., 1995). two covalent modifications and provide highly stable cellular memory. Stable memory is biologically important for Hox genes to specify the identity of embryonic Molecular links between DNA methylation and histone tissues (Ringrose and Paro, 2004), and epigenetic modifications silencing can restrict the expression of Hox genes to specific cell types in the developing embryo. An Recent advances in the field of chromatin structure are important issue in studies of epigenetically mediated beginning to link the histone code with the DNA gene silencing is to understand whether DNA methyla- ‘cytosine-methylation code’, which indicates that these tion is the initial silencing event or whether it is targeted two epigenetic processes are intimately linked (Jones to the region by earlier chromatin-remodeling events. and Baylin, 2002; Lachner and Jenuwein, 2002). First, A variety of tumor suppressor genes has been shown to H3K9 and CpG methylations show complex interplay in be silenced in human cancer, and DNA methylation has which each mark can influence the other (Fuks et al., often been interpreted to be causally involved in this 2003a, b). Several experiments have indicated that silencing. However, histone modifications associated methylated DNA influences the methylation of H3K9 with the silencing of a tumor suppressor gene, p16INK4a, during DNA replication, resulting in the silencing of occur before DNA methylation (Bachman et al., 2003). gene expression (Martin and Zhang, 2005). Conversely, Similarly, DNA methylation associated with silencing of

Oncogene Epigenetic regulations in hematopoietic Hox code HHeet al 385 genes on the inactive X chromosome occurs only after X consequence of carcinogenesis. For example, Hoxa9 is chromosome inactivation. It seems that methylation at highly expressed in MLL-AF9 leukemia cells, and yet promoters does not lead to silenced transcription until Hoxa9 knockout mice still develop leukemia (Kumar chromatin proteins are recruited to the region, which et al., 2004). However, in human leukemia cell lines, mediate gene silencing (Bachman et al., 2003). No Hoxa9 knockout reduces proliferation and increases matter whether DNA methylation directly or indirectly apoptosis and differentiation (Faber et al., 2009). A key initiates the process that results in loss of transcription, problem to the acceptance of Hox dysregulation as a DNA methylation is crucial to ‘locking in’ gene cause rather than a consequence of cancer is the lack of silencing. There is growing evidence that aberrant gene an obvious relationship between the transcriptional expression in leukemia is linked to epigenetic deregula- properties of Hox genes and their oncogenic potential. tion, such as promoter DNA methylation in CpG In many types of cancer cells, Hox genes that are islands (Bullinger and Armstrong, 2010). normally expressed in undifferentiated cells are upregu- lated, whereas those that are normally expressed in differentiated tissues are downregulated (Abate-Shen, Failure of maintenance of Hox genes in leukemia cells 2002). It may be fair to conclude that Hox genes do not individually function as classic tumor suppressors nor as Overexpression of several Hox genes induces leukemo- bona fide oncogenes; instead the combinatorial expres- genesis, which could also result from chromosomal sion profile of multiple Hox genes correlates with the cell translocations (Linggi et al., 2005). The Mll gene is phenotype and behavior in a dose-dependent manner frequently targeted by chromosomal translocations in (Figure 1). In that case, early stages of leukemia or acute lymphoid and myeloid leukemias (ALL and AML, preleukemia could be more treatable by epigenetic respectively) (Ayton and Cleary, 2001). There are reversal of Hox gene deregulation. various MLL fusion proteins with the N-terminal portion of MLL fused in frame with 1 of over 60 different potential fusion partners (Hess, 2004; Popovic Conclusions and outlook and Zeleznik-Le, 2005). Both human and murine leukemias triggered by MLL-AF9 are most frequently Epigenetic mechanism plays a crucial role in Hox gene myeloid in phenotype (Sorensen et al., 1994; Look, regulation during embryogenesis and carcinogenesis. 1997; Dobson et al., 1999). It is easily concluded that Understanding the biological significance of misexpres- chromosomal translocation results in development of sion of Hox cluster genes in hematopoietic progenitors is MLL fusion oncogenes and accordingly, mis-expression particularly apposite as Hox code study moves into of Hox genes. However, numerous studies have shown mobile cells. As the first step, it is required to know that overexpression of MLL fusion oncoproteins in whether and how combinatory expression patterns of normal ES and bone marrow cells first promotes non- Hox genes influence migration and homing of HSCs and malignant expansion of myeloid precursors and Hox their derivatives? So far, there is no direct evidence to deregulation, and then is followed by accumulation of support this possibility. However, many studies have malignant cells (Dobson et al., 1999; Kumar et al., 2004; indicated that Hox genes (mostly from cluster A) Yokoyama et al., 2005; Chen et al., 2006). The long control cell adhesion and migration in hematopoietic latency in vivo and monoclonal nature of the leukemias cells (Leroy et al., 2004), epithelial cells (Mace et al., suggest that secondary genetic transformation is 2005; Arderiu et al., 2007) and pontine neurons(Geisen required for the development of leukemia (Dobson et al., 2008). If Hox expression patterns define the cell et al., 1999; Johnson et al., 2003). Further studies positional identity and behavior, how menin-MLL or indicate that the wild-type MLL allele is required for RA-Fgf-Wnt signaling manipulate Hox networks and MLL-AF9-induced leukemogenesis. Wild-type MLL how Hox networks regulate cell–cell and cell–extra- controls both H3K4 methylation and MLL-AF9- cellular matrix interactions? induced H3K79 methylation (Thiel et al., 2010). PcG and TrxG proteins are best known for their Aggressive histone modifications in 50 Hoxa cluster gene epigenetic regulators of Hox gene clusters. The idea that expression (Yan et al., 2006a) and promoter DNA the PcG–TrxG system maintains Hox expression is neat methylation (Bullinger and Armstrong, 2010) substan- and simple but it leaves several open questions. For tially contribute to leukemic cell phenotypes. Based on example, how are Hox cluster genes presented in the these results, it would be reasonable to propose a context of PcG and TrxG systems and why are Hox hypothetical sequence of events in which MLL fusion genes highly expressed in undifferentiated stem cells? preferentially increases H3K4 methylation, distorts Although many components of both PcG and TrxG chromatin domain along the target Hox cluster genes proteins have been characterized, yet their relationship and eventually causes genome instability. with other chromatin-remodeling factors remains to be It appears that genomic translocation, MLL fusion elucidated. Forkhead transcription factor, Foxi1, stably oncoprotein, and epigenetic alterations constitute a binds to chromatin through the cell cycle and generally cause–effect triangle relationship in the development of inhibits the accessibility of nucleases to chromatin MLL-associated leukemia. In this triangular model, (Yan et al., 2006b). Given that a cascade of Fox factors Hox genes exert downstream effects on cell fate. It is have been found to endow progenitor cells with the disputable whether Hox dysregulation is the cause or the competence to activate genes in response to tissue-

Oncogene Epigenetic regulations in hematopoietic Hox code HHeet al 386 inductive signals (Zaret et al., 2008), an attractive Acknowledgements possibility is that the pioneer Fox proteins may be involved in PcG/TrxG complex-mediated chromatin We apologize to authors whose work has not been cited modifications. here owing to the stringent limit of space and our knowledge. We thank Shivani Sethi’s assistance in editing the manuscript. This work was supported by the International Conflict of interest Cooperation Initiative Program of Shanghai Ocean University (A2302100002) and the Shanghai Leading Academic Disci- The authors declare no conflict of interest. pline Project (S30701).

References

Abate-Shen C. (2002). Deregulated homeobox gene expression in Dillon SC, Zhang X, Trievel RC, Cheng X. (2005). The SET-domain cancer: cause or consequence? Nat Rev Cancer 2: 777–785. protein superfamily: protein lysine methyltransferases. Genome Biol Akbas GE, Song J, Taylor HS. (2004). A HOXA10 estrogen response 6: 227. element (ERE) is differentially regulated by 17 beta-estradiol and Dobson CL, Warren AJ, Pannell R, Forster A, Lavenir I, Corral J diethylstilbestrol (DES). J Mol Biol 340: 1013–1023. et al. (1999). The mll-AF9 gene fusion in mice controls myelopro- Akin ZN, Nazarali AJ. (2005). Hox genes and their candidate liferation and specifies acute myeloid leukaemogenesis. Embo J 18: downstream targets in the developing central nervous system. Cell 3564–3574. Mol Neurobiol 25: 697–741. Duboule D, Deschamps J. (2004). Colinearity loops out. Dev Cell 6: Amores A, Force A, Yan YL, Joly L, Amemiya C, Fritz A et al. 738–740. (1998). Zebrafish hox clusters and vertebrate genome evolution. Duboule D, Morata G. (1994). Colinearity and functional hierarchy Science 282: 1711–1714. among genes of the homeotic complexes. Trends Genet 10: 358–364. Arderiu G, Cuevas I, Chen A, Carrio M, East L, Boudreau NJ. (2007). Erfurth FE, Popovic R, Grembecka J, Cierpicki T, Theisler C, Xia ZB HoxA5 stabilizes adherens junctions via increased Akt1. Cell Adh et al. (2008). MLL protects CpG clusters from methylation within Migr 1: 185–195. the Hoxa9 gene, maintaining transcript expression. Proc Natl Acad Ayton PM, Cleary ML. (2001). Molecular mechanisms of leukemo- Sci USA 105: 7517–7522. genesis mediated by MLL fusion proteins. Oncogene 20: 5695–5707. Ernst P, Mabon M, Davidson AJ, Zon LI, Korsmeyer SJ. (2004). An Bachman KE, Park BH, Rhee I, Rajagopalan H, Herman JG, Baylin Mll-dependent Hox program drives hematopoietic progenitor SB et al. (2003). Histone modifications and silencing prior to DNA expansion. Curr Biol 14: 2063–2069. methylation of a tumor suppressor gene. Cancer Cell 3: 89–95. Esteller M. (2006). The necessity of a human epigenome project. Bel-Vialar S, Itasaki N, Krumlauf R. (2002). Initiating Hox gene Carcinogenesis 27: 1121–1125. expression: in the early chick neural tube differential sensitivity to Faber J, Krivtsov AV, Stubbs MC, Wright R, Davis TN, van den FGF and RA signaling subdivides the HoxB genes in two distinct Heuvel-Eibrink M et al. (2009). HOXA9 is required for survival in groups. Development 129: 5103–5115. human MLL-rearranged acute leukemias. Blood 113: 2375–2385. Bullinger L, Armstrong SA. (2010). HELP for AML: methylation Faiella A, Zappavigna V, Mavilio F, Boncinelli E. (1994). Inhibition of profiling opens new avenues. Cancer Cell 17: 1–3. retinoic acid-induced activation of 30 human HOXB genes by Cao R, Wang L, Wang H, Xia L, Erdjument-Bromage H, Tempst P antisense oligonucleotides affects sequential activation of genes et al. (2002). Role of histone H3 lysine 27 methylation in Polycomb- located upstream in the four HOX clusters. Proc Natl Acad Sci USA group silencing. Science 298: 1039–1043. 91: 5335–5339. Cao R, Zhang Y. (2004a). SUZ12 is required for both the histone Felsenfeld G, Burgess-Beusse B, Farrell C, Gaszner M, Ghirlando R, methyltransferase activity and the silencing function of the Huang S et al. (2004). Chromatin boundaries and chromatin EED-EZH2 complex. Mol Cell 15: 57–67. domains. Cold Spring Harb Symp Quant Biol 69: 245–250. Cao R, Zhang Y. (2004b). The functions of E(Z)/EZH2-mediated Francis NJ, Kingston RE, Woodcock CL. (2004). Chromatin methylation of lysine 27 in histone H3. Curr Opin Genet Dev 14: compaction by a polycomb group protein complex. Science 306: 155–164. 1574–1577. Celetti A, Barba P, Cillo C, Rotoli B, Boncinelli E, Magli MC. (1993). Fuks F, Burgers WA, Brehm A, Hughes-Davies L, Kouzarides T. Characteristic patterns of HOX gene expression in different types of (2000). DNA methyltransferase Dnmt1 associates with histone human leukemia. Int J Cancer 53: 237–244. deacetylase activity. Nat Genet 24: 88–91. Chambeyron S, Bickmore WA. (2004). Chromatin decondensation and Fuks F, Hurd PJ, Deplus R, Kouzarides T. (2003a). The DNA nuclear reorganization of the HoxB locus upon induction of methyltransferases associate with HP1 and the SUV39H1 histone transcription. Genes Dev 18: 1119–1130. methyltransferase. Nucleic Acids Res 31: 2305–2312. Chen YX, Yan J, Keeshan K, Tubbs AT, Wang H, Silva A et al. Fuks F, Hurd PJ, Wolf D, Nan X, Bird AP, Kouzarides T. (2003b). (2006). The tumor suppressor menin regulates hematopoiesis and The methyl-CpG-binding protein MeCP2 links DNA methylation myeloid transformation by influencing Hox gene expression. Proc to histone methylation. J Biol Chem 278: 4035–4040. Natl Acad Sci USA 103: 1018–1023. Garcia-Fernandez J. (2005). Hox, ParaHox, ProtoHox: facts and Davidson AJ, Ernst P, Wang Y, Dekens MP, Kingsley PD, Palis J guesses. Heredity 94: 145–152. et al. (2003). cdx4 mutants fail to specify blood progenitors and can Gaufo GO, Thomas KR, Capecchi MR. (2003). Hox3 genes be rescued by multiple hox genes. Nature 425: 300–306. coordinate mechanisms of genetic suppression and activation in Dejardin J, Cavalli G. (2004). Chromatin inheritance upon Zeste- the generation of branchial and somatic motoneurons. Development mediated Brahma recruitment at a minimal cellular memory 130: 5191–5201. module. Embo J 23: 857–868. Gaunt SJ, Cockley A, Drage D. (2004). Additional enhancer copies, Dellino GI, Schwartz YB, Farkas G, McCabe D, Elgin SC, Pirrotta V. with intact cdx binding sites, anteriorize Hoxa-7/lacZ expression in (2004). Polycomb silencing blocks transcription initiation. Mol Cell mouse embryos: evidence in keeping with an instructional cdx 13: 887–893. gradient. Int J Dev Biol 48: 613–622. Deschamps J, van Nes J. (2005). Developmental regulation of the Hox Gaunt SJ, Drage D, Trubshaw RC. (2008). Increased Cdx protein dose genes during axial morphogenesis in the mouse. Development 132: effects upon axial patterning in transgenic lines of mice. Develop- 2931–2942. ment 135: 2511–2520.

Oncogene Epigenetic regulations in hematopoietic Hox code HHeet al 387 Geisen MJ, Di Meglio T, Pasqualetti M, Ducret S, Brunet JF, Lawrence HJ, Sauvageau G, Humphries RK, Largman C. (1996). The Chedotal A et al. (2008). Hox paralog group 2 genes control the role of HOX homeobox genes in normal and leukemic hematopoi- migration of mouse pontine neurons through slit-robo signaling. esis. Stem Cells 14: 281–291. PLoS Biol 6: e142. Lee JH, Skalnik DG. (2005). CpG-binding protein (CXXC finger Grier DG, Thompson A, Kwasniewska A, McGonigle GJ, Halliday protein 1) is a component of the mammalian Set1 histone H3-Lys4 HL, Lappin TR. (2005). The pathophysiology of HOX genes and methyltransferase complex, the analogue of the yeast Set1/COM- their role in cancer. J Pathol 205: 154–171. PASS complex. J Biol Chem 280: 41725–41731. Gruss P, Kessel M. (1991). Axial specification in higher vertebrates. Lehoczky JA, Williams ME, Innis JW. (2004). Conserved expression Curr Opin Genet Dev 1: 204–210. domains for genes upstream and within the HoxA and HoxD Guo Y, Lubbert M, Engelhardt M. (2003). CD34- hematopoietic stem clusters suggests a long-range enhancer existed before cluster cells: current concepts and controversies. Stem Cells 21: 15–20. duplication. Evol Dev 6: 423–430. Gutman A, Gilthorpe J, Rigby PW. (1994). Multiple positive and Leroy P, Berto F, Bourget I, Rossi B. (2004). Down-regulation of Hox negative regulatory elements in the promoter of the mouse A7 is required for cell adhesion and migration on fibronectin during homeobox gene Hoxb-4. Mol Cell Biol 14: 8143–8154. early HL-60 monocytic differentiation. J Leukoc Biol 75: 680–688. Hazzalin CA, Mahadevan LC. (2005). Dynamic acetylation of all Linggi BE, Brandt SJ, Sun ZW, Hiebert SW. (2005). Translating the lysine 4-methylated histone H3 in the mouse nucleus: analysis at c- histone code into leukemia. J Cell Biochem 96: 938–950. fos and c-jun. PLoS Biol 3: e393. Lohnes D. (2003). The Cdx1 homeodomain protein: an integrator of Hess JL. (2004). MLL: a histone methyltransferase disrupted in posterior signaling in the mouse. Bioessays 25: 971–980. leukemia. Trends Mol Med 10: 500–507. Look AT. (1997). Oncogenic transcription factors in the human acute Hughes CM, Rozenblatt-Rosen O, Milne TA, Copeland TD, Levine leukemias. Science 278: 1059–1064. SS, Lee JC et al. (2004). Menin associates with a trithorax family Mace KA, Hansen SL, Myers C, Young DM, Boudreau N. (2005). histone methyltransferase complex and with the hoxc8 locus. Mol HOXA3 induces cell migration in endothelial and epithelial Cell 13: 587–597. cells promoting angiogenesis and wound repair. J Cell Sci 118: Huynh H, Iizuka S, Kaba M, Kirak O, Zheng J, Lodish HF et al. 2567–2577. (2008). Insulin-like growth factor-binding protein 2 secreted by a Magli MC, Largman C, Lawrence HJ. (1997). Effects of HOX tumorigenic cell line supports ex vivo expansion of mouse homeobox genes in blood cell differentiation. J Cell Physiol 173: hematopoietic stem cells. Stem Cells 26: 1628–1635. 168–177. Jackson JP, Johnson L, Jasencakova Z, Zhang X, PerezBurgos L, Manley NR, Capecchi MR. (1997). Hox group 3 paralogous genes act Singh PB et al. (2004). Dimethylation of histone H3 lysine 9 is a synergistically in the formation of somitic and neural crest-derived critical mark for DNA methylation and gene silencing in structures. Dev Biol 192: 274–288. Arabidopsis thaliana. Chromosoma 112: 308–315. Manley NR, Capecchi MR. (1998). Hox group 3 paralogs regulate the Jackson JP, Lindroth AM, Cao X, Jacobsen SE. (2002). Control of development and migration of the thymus, thyroid, and parathyroid CpNpG DNA methylation by the KRYPTONITE histone H3 glands. Dev Biol 195: 1–15. methyltransferase. Nature 416: 556–560. Manzanares M, Bel-Vialar S, Ariza-McNaughton L, Ferretti E, Johnson JJ, Chen W, Hudson W, Yao Q, Taylor M, Rabbitts TH et al. Marshall H, Maconochie MM et al. (2001). Independent regulation (2003). Prenatal and postnatal myeloid cells demonstrate stepwise of initiation and maintenance phases of Hoxa3 expression in the progression in the pathogenesis of MLL fusion gene leukemia. vertebrate hindbrain involve auto- and cross-regulatory mechan- Blood 101: 3229–3235. isms. Development 128: 3595–3607. Jones PA, Baylin SB. (2002). The fundamental role of epigenetic events Martin C, Zhang Y. (2005). The diverse functions of histone lysine in cancer. Nat Rev Genet 3: 415–428. methylation. Nat Rev Mol Cell Biol 6: 838–849. Kessel M, Gruss P. (1991). Homeotic transformations of murine Mathieu O, Probst AV, Paszkowski J. (2005). Distinct regulation of vertebrae and concomitant alteration of Hox codes induced by histone H3 methylation at lysines 27 and 9 by CpG methylation in retinoic acid. Cell 67: 89–104. Arabidopsis. Embo J 24: 2783–2791. Kmita M, Duboule D. (2003). Organizing axes in time and space; 25 Maves L, Kimmel CB. (2005). Dynamic and sequential patterning years of colinear tinkering. Science 301: 331–333. of the zebrafish posterior hindbrain by retinoic acid. Dev Biol 285: Kmita M, Kondo T, Duboule D. (2000). Targeted inversion of a polar 593–605. silencer within the HoxD complex re-allocates domains of enhancer Milne TA, Briggs SD, Brock HW, Martin ME, Gibbs D, Allis CD sharing. Nat Genet 26: 451–454. et al. (2002). MLL targets SET domain methyltransferase activity to Knittel T, Kessel M, Kim MH, Gruss P. (1995). A conserved enhancer Hox gene promoters. Mol Cell 10: 1107–1117. of the human and murine Hoxa-7 gene specifies the anterior Milne TA, Hughes CM, Lloyd R, Yang Z, Rozenblatt-Rosen O, Dou boundary of expression during embryonal development. Develop- Y et al. (2005). Menin and MLL cooperatively regulate expression ment 121: 1077–1088. of cyclin-dependent kinase inhibitors. Proc Natl Acad Sci USA 102: Kobrossy L, Rastegar M, Featherstone M. (2006). Interplay 749–754. between chromatin and trans-acting factors regulating the Monge I, Kondo T, Duboule D. (2003). An enhancer-titration effect Hoxd4 promoter during neural differentiation. J Biol Chem 281: induces digit-specific regulatory alleles of the HoxD cluster. Dev Biol 25926–25939. 256: 212–220. Kondo T, Zakany J, Duboule D. (1998). Control of colinearity in Morgan R, Whiting K. (2008). Differential expression of HOX genes AbdB genes of the mouse HoxD complex. Mol Cell 1: 289–300. upon activation of leukocyte sub-populations. Int J Hematol 87: Kouskouti A, Talianidis I. (2005). Histone modifications defining 246–249. active genes persist after transcriptional and mitotic inactivation. Muller J, Hart CM, Francis NJ, Vargas ML, Sengupta A, Wild B et al. Embo J 24: 347–357. (2002). Histone methyltransferase activity of a Drosophila Poly- Kumar AR, Hudson WA, Chen W, Nishiuchi R, Yao Q, Kersey JH. comb group repressor complex. Cell 111: 197–208. (2004). Hoxa9 influences the phenotype but not the incidence of Nakamura T, Mori T, Tada S, Krajewski W, Rozovskaia T, Wassell R Mll-AF9 fusion gene leukemia. Blood 103: 1823–1828. et al. (2002). ALL-1 is a histone methyltransferase that assembles a Kwan CT, Tsang SL, Krumlauf R, Sham MH. (2001). Regulatory supercomplex of proteins involved in transcriptional regulation. analysis of the mouse Hoxb3 gene: multiple elements work in Mol Cell 10: 1119–1128. concert to direct temporal and spatial patterns of expression. Dev Ng HH, Robert F, Young RA, Struhl K. (2003). Targeted recruitment Biol 232: 176–190. of Set1 histone methylase by elongating Pol II provides a localized Lachner M, Jenuwein T. (2002). The many faces of histone lysine mark and memory of recent transcriptional activity. Mol Cell 11: methylation. Curr Opin Cell Biol 14: 286–298. 709–719.

Oncogene Epigenetic regulations in hematopoietic Hox code HHeet al 388 Nowling T, Zhou W, Krieger KE, Larochelle C, Nguyen-Huu MC, sequences contributes to Hoxa5 regional expression along the axial Jeannotte L et al. (1999). Hoxa5 gene regulation: a gradient of skeleton. Mol Cell Biol 25: 1389–1401. binding activity to a brachial spinal cord element. Dev Biol 208: Taghon T, Thys K, De Smedt M, Weerkamp F, Staal FJ, Plum J et al. 134–146. (2003). Homeobox gene expression profile in human hematopoietic Oosterveen T, Niederreither K, Dolle P, Chambon P, Meijlink F, multipotent stem cells and T-cell progenitors: implications for Deschamps J. (2003a). Retinoids regulate the anterior expression human T-cell development. Leukemia 17: 1157–1163. boundaries of 50 Hoxb genes in posterior hindbrain. Embo J 22: Tamaru H, Selker EU. (2001). A histone H3 methyltransferase 262–269. controls DNA methylation in Neurospora crassa. Nature 414: Oosterveen T, van Vliet P, Deschamps J, Meijlink F. (2003b). The 277–283. direct context of a hox retinoic acid response element is crucial for Terranova R, Agherbi H, Boned A, Meresse S, Djabali M. (2006). its activity. J Biol Chem 278: 24103–24107. Histone and DNA methylation defects at Hox genes in mice Owens BM, Hawley RG. (2002). HOX and non-HOX homeobox expressing a SET domain-truncated form of Mll. Proc Natl Acad Sci genes in leukemic hematopoiesis. Stem Cells 20: 364–379. USA 103: 6629–6634. Pineault N, Helgason CD, Lawrence HJ, Humphries RK. (2002). Thiel AT, Blessington P, Zou T, Feather D, Wu X, Yan J et al. (2010). Differential expression of Hox, Meis1, and Pbx1 genes in primitive MLL-AF9-induced leukemogenesis requires coexpression of the cells throughout murine hematopoietic ontogeny. Exp Hematol 30: wild-type Mll allele. Cancer Cell 17: 148–159. 49–57. Turner BM. (2000). Histone acetylation and an epigenetic code. Popovic R, Zeleznik-Le NJ. (2005). MLL: how complex does it get? Bioessays 22: 836–845. J Cell Biochem 95: 234–242. Vasanthi D, Mishra RK. (2008). Epigenetic regulation of genes during Popperl H, Featherstone MS. (1992). An autoregulatory element of the development: a conserved theme from flies to mammals. J Genet murine Hox-4.2 gene. Embo J 11: 3673–3680. Genomics 35: 413–429. Puschel AW, Balling R, Gruss P. (1991). Separate elements cause Vire E, Brenner C, Deplus R, Blanchon L, Fraga M, Didelot C et al. lineage restriction and specify boundaries of Hox-1.1 expression. (2006). The Polycomb group protein EZH2 directly controls DNA Development 112: 279–287. methylation. Nature 439: 871–874. Ringrose L, Paro R. (2004). Epigenetic regulation of cellular memory Wang KC, Helms JA, Chang HY. (2009). Regeneration, repair and by the Polycomb and Trithorax group proteins. Annu Rev Genet 38: remembering identity: the three Rs of Hox gene expression. Trends 413–443. Cell Biol 19: 268–275. Rinn JL, Bondre C, Gladstone HB, Brown PO, Chang HY. (2006). Wang L, Brown JL, Cao R, Zhang Y, Kassis JA, Jones RS. (2004). Anatomic demarcation by positional variation in fibroblast gene Hierarchical recruitment of polycomb group silencing complexes. expression programs. PLoS Genet 2: e119. Mol Cell 14: 637–646. Roelen BA, de Graaff W, Forlani S, Deschamps J. (2002). Hox cluster Wei J, Wunderlich M, Fox C, Alvarez S, Cigudosa JC, Wilhelm JS polarity in early transcriptional availability: a high order regulatory et al. (2008). Microenvironment determines lineage fate in a human level of clustered Hox genes in the mouse. Mech Dev 119: 81–90. model of MLL-AF9 leukemia. Cancer Cell 13: 483–495. Saurin AJ, Shao Z, Erdjument-Bromage H, Tempst P, Kingston RE. West AG, Gaszner M, Felsenfeld G. (2002). Insulators: many (2001). A Drosophila Polycomb group complex includes Zeste and functions, many mechanisms. Genes Dev 16: 271–288. dTAFII proteins. Nature 412: 655–660. Yan J, Chen YX, Desmond A, Silva A, Yang Y, Wang H et al. Sauvageau G, Lansdorp PM, Eaves CJ, Hogge DE, Dragowska WH, (2006a). Cdx4 and menin co-regulate hoxa9 expression in hemato- Reid DS et al. (1994). Differential expression of homeobox genes in poietic cells. PLoS ONE 1: e47. functionally distinct CD34+ subpopulations of human bone Yan J, Xu L, Crawford G, Wang Z, Burgess SM. (2006b). The marrow cells. Proc Natl Acad Sci USA 91: 12223–12227. forkhead transcription factor foxi1 remains bound to condensed Shao Z, Raible F, Mollaaghababa R, Guyon JR, Wu CT, Bender W mitotic chromosomes and stably remodels chromatin structure. Mol et al. (1999). Stabilization of chromatin structure by PRC1, a Cell Biol 26: 155–168. Polycomb complex. Cell 98: 37–46. Yokoyama A, Lin M, Naresh A, Kitabayashi I, Cleary ML. (2010). A Sharpe J, Nonchev S, Gould A, Whiting J, Krumlauf R. (1998). higher-order complex containing AF4 and ENL family proteins Selectivity, sharing and competitive interactions in the regulation of with P-TEFb facilitates oncogenic and physiologic MLL-dependent Hoxb genes. Embo J 17: 1788–1798. transcription. Cancer Cell 17: 198–212. Sobel RE, Cook RG, Perry CA, Annunziato AT, Allis CD. (1995). Yokoyama A, Somervaille TC, Smith KS, Rozenblatt-Rosen O, Conservation of deposition-related acetylation sites in newly Meyerson M, Cleary ML. (2005). The menin tumor suppressor synthesized histones H3 and H4. Proc Natl Acad Sci USA 92: protein is an essential oncogenic cofactor for MLL-associated 1237–1241. leukemogenesis. Cell 123: 207–218. Sorensen PH, Chen CS, Smith FO, Arthur DC, Domer PH, Bernstein Yokoyama A, Wang Z, Wysocka J, Sanyal M, Aufiero DJ, ID et al. (1994). Molecular rearrangements of the MLL gene are Kitabayashi I et al. (2004). Leukemia proto-oncoprotein MLL present in most cases of infant acute myeloid leukemia and are forms a SET1-like histone methyltransferase complex with strongly correlated with monocytic or myelomonocytic phenotypes. menin to regulate Hox gene expression. Mol Cell Biol 24: J Clin Invest 93: 429–437. 5639–5649. Spirov AV, Borovsky M, Spirova OA. (2002). HOX Pro DB: the Yu BD, Hanson RD, Hess JL, Horning SE, Korsmeyer SJ. (1998). functional genomics of hox ensembles. Nucleic Acids Res 30: 351–353. MLL, a mammalian trithorax-group gene, functions as a transcrip- Spitz F, Gonzalez F, Duboule D. (2003). A global control region tional maintenance factor in morphogenesis. Proc Natl Acad Sci defines a chromosomal regulatory landscape containing the HoxD USA 95: 10632–10636. cluster. Cell 113: 405–417. Zaret KS, Watts J, Xu J, Wandzioch E, Smale ST, Sekiya T. (2008). Suemori H, Noguchi S. (2000). Hox C cluster genes are dispensable for Pioneer factors, genetic competence, and inductive signaling: overall body plan of mouse embryonic development. Dev Biol 220: programming liver and pancreas progenitors from the endoderm. 333–342. Cold Spring Harb Symp Quant Biol 73: 119–126. Tabaries S, Lapointe J, Besch T, Carter M, Woollard J, Tuggle CK Zhou J, Berger SL. (2004). Good fences make good neighbors: barrier et al. (2005). Cdx protein interaction with Hoxa5 regulatory elements and genomic regulation. Mol Cell 16: 500–502.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene