Similar articles

Abstract: polycomb group (PcG) complexes are epigenetic regulatory complexes that conduct transcriptional repression of target genes via modifying the chromatin. The two best characterized forms of PcG complexes, polycomb repressive complexes 1 and 2 (PRC1 and PRC2), are required for maintaining the stemness of embryonic stem cells and many types of adult stem cells. The spectra of target genes for PRCs are dynamically changing with cell differentiation, which is essential for proper decisions on cell fate during developmental processes. Chromobox (CBX) family proteins are canonical components in PRC1, responsible for targeting PRC1 to the chromatin. Recent studies highlight the function specifications among CBX family members in undifferentiated and differentiated stem cells, which reveal the interplay between compositional diversity and functional specificity of PRC1. In this review, we summarize the current knowledge about targeting and functional mechanisms of PRCs, emphasizing the recent breakthroughs related to CBX proteins under a number of physiological and pathological conditions.

多梳抑制复合体PRC1的CBX家族蛋白在表观遗传调控中的作用

研究目的：多梳蛋白家族（PcG）是一类染色质水平上通过表观遗传修饰调控靶基因的转录因子，其主要功能是使其靶基因转录受到抑制进而沉默。PcG通常以多梳蛋白复合体（PRC）的形式存在，目前研究的最多的是PRC1和PRC2。PRC1在PcG对其靶基因进行转录抑制发挥着主要作用。本综述主要论述了哺乳动物中PRC1核心成员CBX蛋白在多梳蛋白调控基因转录过程中发挥的作用及其对胚胎发育、细胞记忆、细胞周期、细胞增殖和肿瘤形成等过程的影响。
创新要点：现已有大量有关PcG在表观遗传水平对其靶基因进行修饰转录机制的综述报道，且以PRC1和PRC2为整体来介绍表观遗传调控机制的文章也屡见不鲜。然而，关于PRC1核心成员CBX蛋白在哺乳动中的同源蛋白CBX2、CBX4、CBX6、CBX7、CBX8对哺乳动物个体发育调节及肿瘤发生过程的分子机制并没有系统的论述。本综述主要将这五种CBX蛋白在转录分子水平上的所发挥的功能进行相关的介绍，并且总结了CBX2、CBX4、CBX6、CBX7、CBX8各自最新的研究进展，体现出五种CBX蛋白的共同功能、各自独特的功能及彼此间的相互联系。
重要结论：总结了在哺乳动物中的五种CBX蛋白在胚胎发育和肿瘤形成等过程中独特的功能调节机制以及整体的相互作用，发现CBX作为PRC1的核心组分在基因表观遗传调控中发挥着极其重要的作用。

关键词：多梳蛋白；多梳抑制复合体；CBX蛋白；表观遗传调控；癌症

1.  Compositions and functions of polycomb group (PcG) complexes

 PcG proteins were originally identified as a set of genes controlling proper body segmentation in Drosophila via repressing Hox genes (Lewis, 1978; Struhl, 1981). PcG proteins are now widely recognized in all metazoans for their roles in the variety of biological processes, such as cell cycle control, X-inactivation, maintenance of pluripotency and self-renewal in embryonic stem cells (ESCs), cell fate decisions, and developmental controls (Fig. 1c) (Muller and Verrijzer, 2009; Richly et al., 2011; Aloia et al., 2013). PcG proteins are assembled into a variety of multi-protein complexes, and the best characterized are polycomb repressive complexes 1 and 2 (PRC1 and PRC2). The PRC2 core complex of Drosophila includes the enhancer of zeste [E(z)], the suppressor of zeste [Su(z)], and extra sex combs (Esc). The mammalian PRC2 core components comprise EZH1 or EZH2 [a homolog of E(z)], SUZ12 [a homolog of Su(z)], and an Esc homolog called EED. As the only PRC2 subunit with enzymatic activity, EZH2 is responsible for catalyzing the di- and tri-methylation of Lys27 on histone H3 (H3K27me^{2 or 3}) (Margueron et al., 2008). The other core PRC2 components are necessary for complex assembly and for proper enzymatic activity (Pasini et al., 2004; Margueron et al., 2009). In Drosophila, PRC1 core complex consists of polycomb (Pc), polyhomeotic (Ph), posterior sex combs (Psc), and sex combs extra (Sce) (also known as RING) (Fig. 1a), and mammals have several homologs to each subunit (Fig. 1b). There are five Pc (CBX2, CBX4, CBX6, CBX7, and CBX8), two Sce (RING1/RING1A and RING2/RING1B), three Ph (PHC1, PHC2, and PHC3), and six Psc known collectively as polycomb group ring fingers (PCGFs) (Morey and Helin, 2010).




 For PRC1, recent advances indicate that the composition and patterns of assembly are far more biochemically diverse than originally estimated. Besides the canonical subunits listed above, there are many other PRC1 components and supporting factors, such as RING1/YY1-binding protein (RYBP) and its homolog YAF2, and the mammalian orthologs of the Drosophila sex comb on midleg (Scm). For details about the complexity and diversity of the PRC1 subcategories, please refer to some recent reports and reviews (Gao et al., 2012; Schwartz and Pirrotta, 2013). The compositions among different mammalian PRC1 subcategories are largely divergent while all of these PRC1 complexes contain at least one of the two RING proteins, which are known as an E3 ubiquitin ligase responsible for ubiquitination of histone H2A (H2AK119ub) (Wang et al., 2004). Several types of PRC1 have been shown to have none or low levels of chromobox (CBX) proteins. RYBP or YAF2 is known to stimulate the enzymatic activity of RING1B toward H2AK119ub, while CBX proteins may not exist simultaneously in this type of PRC1 (Gao et al., 2012).

2.  CBX proteins in PRC1

 There are at least eight members of the CBX proteins in both mouse and human genomes, each of which contains a single N-terminal chromodomain (Wotton and Merrill, 2007). They are all involved in the regulation of heterochromatin, gene expression, and developmental programs. They are further divided into two groups: (1) CBX1, CBX3, and CBX5, also known as heterochromatin protein 1β (HP1β), HP1γ, and HP1α, respectively; (2) CBX2, CBX4, CBX6, CBX7, and CBX8, all having a C-terminal polycomb repressor box, serve as canonical components in PRC1 (Wotton and Merrill, 2007). In this review, we will mainly discuss the roles of these five CBX proteins in the context of PRC1.

 The domain organization of CBX proteins is shown in Fig. 2. CBX proteins physically interact with H3K27me³ via their chromodomains, which helps the recruitment and stabilization of PRC1 to specific regions of the chromatin (Bernstein B.E. et al., 2006; Buchwald et al., 2006). It was reported that mammalian CBX proteins exhibited differential bindings to methylated histone tails. CBX2 and CBX7 bind to both H3K9Me³ and H3K27Me³ whereas CBX4 shows stronger affinity for H3K9Me³ (Bernstein E. et al., 2006). The C-terminal polycomb repressor box of CBX proteins is involved in transcriptional silencing and binding to other PRC1 components such as RING1B (Muller et al., 1995; Bezsonova et al., 2009). Adjacent to chromodomain, all these CBX homologs have a DNA binding motif, AT-hook (in CBX2) or an AT-hook like motif (in the other four CBX proteins) (Senthilkumar and Mishra, 2009). The less conserved sequences in the middle of the CBX proteins may play a role in specifically directing each CBX family member to distinct regions of the chromatin (Vincenz and Kerppola, 2008).

3.  Recruitment of PcG complexes

 PcG complexes are involved in repressing over several thousand genes in mammalian genomes and the pool of their target genes is distinct in different cell types and dynamically changing in various cell states (Simon and Kingston, 2013). Components in PRC1 and PRC2 generally do not have DNA binding properties. The targeting mechanisms of the mammalian PcG complexes are diverse and extremely complicated. Many sources of the ‘targeting command’ are comprehensively integrated, including chromatin signatures (such as histone modifications, histone variants, DNA sequences, and CpG islands), varieties of non-coding RNA (ncRNA) species, transcription factors and cofactors, and possibly the status of RNA polymerase II (Pol II).

 In Drosophila, specific DNA modules called polycomb response elements (PREs) are responsible for recruiting PcG complexes to their chromatin targeting regions (Simon et al., 1993). This process is mediated by sequence-specific DNA binding proteins such as the zinc finger protein PHO. Recent studies found that Drosophila PREs consist of binding sites for a complex array of DNA binding factors and the interplay among these regulatory factors is important for specifying the function of the PREs in a cell or tissue-specific fashion (Oktaba et al., 2008; Brown and Kassis, 2013). In mammals, it becomes much more difficult in searching and defining ‘PRE’. Transcription repressor protein YY1 is the mammalian homolog of the Drosophila PHO, which works together with YAF2 and the transcription corepressor C-terminal binding protein (CtBP) in recruiting PcG complexes to many of their target genes, including HOX genes (Atchison et al., 2003; Srinivasan and Atchison, 2004; Basu et al., 2014). Several potential human PREs have been found in HOXB, HOXC, and HOXD clusters to facilitate the recruitment of both PRC2 and PRC1 components to a reporter gene, yet the DNA binding factors mediating these interactions are variable (Woo et al., 2010; 2013). Another potential human PRE containing 25 repeats of YY1 binding motif was validated to be able to down-regulate reporter genes via PRC2 in both HEK cells and in Drosophila (Bengani et al., 2013). Besides YY1, many other transcription factors have been reported to act as recruiters for PcG complexes, such as neuronal inhibitor REST (Ren and Kerppola, 2011).

 In addition to potential PREs, chromatin regions enriched in CpG dinucleotides (CpG islands) may also facilitate targeting of PcG complexes in mammals. CpG islands that lack both 5-methylcytosine and activator-binding sites are largely overlapping with H3K27me³, PRC2, and PRC1, and notably, constructs, containing CpG island-like sequences, are found to be capable of recruiting PRC2 upon integration into a mouse genome (Ku et al., 2008; Mendenhall et al., 2010). One PRC1 component Kdm2b/Fbxl10 was recently identified as a binding factor towards CpG islands (Wu X. et al., 2013). Other signatures of the chromatin may also contribute to the recruitment of PcG complexes, such as histone variants H2AZ (Creyghton et al., 2008) and MacroH2A (Buschbeck et al., 2009).

 The accumulating evidence has emphasized another scheme involved in ncRNA for the recruitment of PcG complexes. In the model of mammalian X chromosome inactivation, the long non-coding RNA (lncRNA) Xist (more specifically its A repeat region called RepA) has been implicated in PRC2 recruitment (Plath et al., 2003; Zhao et al., 2008). Both Suz12 and Ezh2 are known to have RNA binding capability in PRC2. The chromatin recruitment of canonical PRC1 complex is mainly mediated by CBX proteins. As mentioned above, the histone modification mark H3K27me³ serves as a signal for CBX protein mediated PRC1 recruitment (Fig. 3a). CBX proteins are the only known RNA binding proteins in PRC1 and their chromodomains are required for RNA binding (Bernstein E. et al., 2006). At the Ink4a/ARF/Ink4b locus, the anti-sence lncRNA ANRIL has been demonstrated to recruit both PRC1 and PRC2 in a cis-manner, and this guidance plays a key role in repressing these target genes by PcG complexes (Yap et al., 2010; Kotake et al., 2011) (Fig. 3b). Other similar examples including H19 ncRNA and Kcnqtlot1 ncRNA are described in more detail in other reviews (Brockdorff, 2013; Simon and Kingston, 2013). ncRNA may also recruit PcG complexes in a trans-manner. The only reported example is the human HOTAIR ncRNA, which is transcribed from the HOXC loci while having influence on genes located in the HOXD loci (Tsai et al., 2010). However, deletion of the putative mouse HOTAIR did not exhibit any effects on PRC2 targeting (Schorderet and Duboule, 2011), and deleting various segments within the mouse HoxD loci did not disturb the overall patterns of PRC2 recruitment (Schorderet et al., 2013). Besides lncRNA species, genome-wide studies have revealed that promoter-associated short RNAs transcribed from CpG island promoters and other types of RNA species also potentially facilitate the recruitment of PRC2 (Kanhere et al., 2010; Zhao et al., 2010).




 In addition to the above mechanisms, the phosphorylation status of the serine residues within the carboxyl-terminal domain (CTD) of Pol II has also been shown to correlate with the occupancy of PcG complexes at certain groups of target genes. The phosphorylation at Ser5 normally occurs right after initiation (Ser5p⁺) while the phosphorylation at Ser2 and Ser7 positions are generally associated with productive elongation. The new genome-wide study discovered that in mouse ESC, the development-related PRC targets were generally associated with poised Pol II (Ser5p⁺, Ser2p⁻, Ser7p⁻) while PRC targets involved in metabolism were decorated with transcriptional active Pol II (Ser5p⁺, Ser2p⁺, Ser7p⁺) (Brookes et al., 2012). It remains highly debatable whether the chromatin-bound PcG complexes help establish the block of transcription elongation or the preset states of Pol II serve as signs for recruiting the PcG complex.

 Although the targeting of PRC1 to chromatin is mainly dependent on PRC2, differential recruitment of mammalian PRC1 vs. PRC2 has been observed in several cases (Ren et al., 2008; Vincenz and Kerppola, 2008; Tavares et al., 2012; Wu X. et al., 2013). The detailed mechanisms remain unclear. PRC1 may be initially recruited to chromatin regions distal to the target genes, searching for its targets; further recognition to H3K27me³ with or without the guidance of ncRNA helps direct and stabilize the complex to final destinations.

4.  Molecular functions of CBX proteins

4.1.  Functions in common

 The functions of CBX family proteins have been studied in many cell types. The Ink4a/ARF/Ink4b locus is one of the earliest identified and the most well-known targets for CBX-containing PRC1 complex (Gil et al., 2004). The three tumor suppressors encoded by this locus play a central role in cell-cycle inhibition, senescence, and stress-induced apoptosis. Consistently, knockdown of Cbx genes is generally associated with reduced cell proliferation, and in contrast, overexpression of Cbx genes in some primary cells normally results in the extension of life span or even cell immortalization (Gil et al., 2004) (Table 1).

4.2.  Interplay among CBX proteins

 In stem cells, the consequences upon changing expressions of various CBX family members are not very consistent because it seems that CBX/PRC1 can promote both self-renewal and differentiation. This apparent inconsistency can now be better explained by recent discoveries of an autoregulatory loop among CBX family members (Morey et al., 2012; O′Loghlen et al., 2012; Klauke et al., 2013) (Fig. 4a). Studies indicated that CBX7 was the primary CBX protein assembled in PRC1 in the mouse ESC, and notably, it directly participated in transcription repression of the genes encoding other CBX family members (including Cbx2, Cbx4, and Cbx8) by PcG complexes. Interestingly, there was a dynamic switch of the identity of CBX proteins in PRC1 upon differentiation. During ESC differentiation, the expression of Cbx7 was down-regulated and Cbx2, Cbx4, and Cbx8 were up-regulated simultaneously. The mechanism that triggers the onset of this switch remains elusive. Depletion of CBX7 induced differentiation of ESC and mainly ectoderm-lineage-associated genes were derepressed, while the ectopic expression of Cbx7 enhanced ESC self-renewal. In differentiating cells, CBX2 and CBX4 replaced the function of CBX7 in PRC1 targeting but they were found to have non-overlapping functions, repressing distinct target genes. The teratomas derived from CBX2- and CBX4-depleted ESC displayed abnormal differentiation phenotypes with increased number of cells expressing endodermal or mesodermal marker genes. More recent studies demonstrated that in mouse hematopoietic stem cells (HSCs), CBX7 was also required for their self-renewal, whereas CBX2, CBX4, and CBX8 were crucial for their proper differentiation (Klauke et al., 2013). Ectopically expressed CBX2, CBX4, or CBX8 in HSC was able to compete with CBX7 in PRC1 assembly while they shifted the targets of PRC1 and resulted in the differentiation of HSC. However, in a human study, the depletion of CBX2 resulted in decreased cell proliferation and increasing levels of apoptosis, emphasizing the requirement of CBX2 in the maintenance of HSC (van den Boom et al., 2013). Overall, in stem cells (including embryonic and adult stem cells), CBX proteins play a key role in maintenance of pluripotency or multipotency. The discrepant results in some of these studies suggest the specific roles each CBX protein might be variable in different species. Some data about manipulation of Cbx expression in cancer cell lines are also listed in Table 1, while the roles of CBX proteins in cancer will be mainly discussed later in this review.




 Besides the mutually inhibitory mode described above, multiple CBX proteins can also be expressed and shown to be functioning in the same cell type. Through genome-wide analyses, Klauke et al. (2013) found that targets of CBX7 and CBX8 were largely overlapping in the mouse HSC. Meanwhile, they also identified about 200 genes that were differentially targeted. Interestingly, the genes uniquely occupied by CBX8 were actually highly expressed in HSC and later became repressed in progenitors, which was opposite to the expression pattern of the CBX7 targets. This suggests that CBX/PRC1 can be pre-deposited to the future targets and distinct CBX family members may help specify the ‘actively repressed’ vs. ‘ready-to-be repressed’ targets. A similar phenomenon was also observed by Pemberton et al. (2014). In this study, the genome-wide occupancy of several PRC1 components was analyzed and compared in human fibroblast cell lines. The results demonstrated that the target genes of CBX6, CBX7, and CBX8 were mostly overlapping, yet the detailed configuration of co-occupied PRC1 complexes remains unknown (Pemberton et al., 2014). Theoretically, PRC1 complexes containing each CBX protein respectively could be lined up or form oligomers at their common targets, or it is possible that various CBX proteins are integrated into a single unit of PRC1 complex (Fig. 4b). Notably, Pemberton et al. (2014) also identified that cell type-specific PRC1 binding sites and the occupancy of PRC1 at many loci did not correlate with an outcome of transcription repression.

4.3.  Phenotypes of Cbx knockout (KO) mice

 The roles of CBX proteins playing during embryogenesis have been investigated in KO mice or human patients with genetic mutations (Table 2). The observed phenotypes revealed the functional specification of these CBX family members. All known Cbx KO mice (Cbx2, Cbx4, and Cbx7) can survive the entire embryonic development. Homozygous mutants of Cbx2 or Cbx4 KO mice displayed postnatal lethality. The distinct features of Cbx2 KO mice include male-to-female sex reversal and defects related to skeletal development were observed in two separate KO lines. Some of these functions of CBX2 are apparently conserved across species: a combination of two point mutations in Cbx2 led to male-to-female sex reversal in humans without other obvious developmental defects. CBX2 plays a critical role in germ cell development, meiosis onset and homologous chromosome synapsis in the mammalian germ line (Baumann and de la Fuente, 2011). CBX4 has been recently found to play a key role in the development of the immune system: the proliferation of thymic epithelial cells and the maintenance of thymic epithelium were impaired in the KO mice. The Cbx7 KO mice were grossly normal in morphology and growth although these mice attempted to develop liver and lung adenomas and carcinomas at adulthood. Considering the dominant role it plays in mouse ESC compared to other CBX proteins, alternative pathways have to be applied to compensate for the loss of CBX7 in these mice.

4.4.  Unique roles of CBX proteins

 In the five CBX proteins, CBX4 is the only one known to have an enzymatic activity. It acts as a small ubiquitin-like modifier (SUMO) E3 ligase and this activity is dependent on its chromodomain and a C-terminal substrate binding domain (Kagey et al., 2003). SUMOylation has been implicated in the regulation of many cellular processes, including transcriptional repression, genome stability, chromatin organization, and DNA repair (Galanty et al., 2009; Morris et al., 2009; Dou et al., 2011). It has been well known that SUMO is covalently conjugated to lysine residues on its substrates while there are also specific motifs identified in SUMO E3 ligases that mediate non-covalent interactions with SUMO (Song et al., 2004). CBX4 indeed contains two such motifs that facilitate non-covalent SUMO binding and are required for full E3 ligase activity (Nacerddine et al., 2005). The known substrates for CBX4 are listed in Table 3. The CBX4-mediated SUMOylations of CCCTC-binding factor (CTCF) and homeo-domain interacting protein kinase 2 (HIPK2) are associated with transcriptional repression. The SUMOylations of heterogeneous nuclear ribonucleoprotein K (HnRNP-K) and BMI1 are essential for their stabilization or site-specific targeting, respectively, in response to DNA damage.




 Like other CBX proteins, CBX8 was found in diploid human and mouse fibroblasts to regulate premature senescence through controlling the transcription of the Ink4a/Arf locus (Dietrich et al., 2007). Recently, Zhou et al. (2013) discovered that the DN oxidative damage caused up-regulation of Cbx8 expression and blocking this response by silencing Cbx8 led to more serious DNA damage, indicating that CBX8 played an important role in DNA repair. Notably, some recent studies uncovered a novel inhibitory mechanism for PRC1 activity. CBX8 acts as an essential cofactor required for mixed lineage leukemia (MLL)-AF9 fusion or MLL-ENL fusion-induced transcriptional activation and leukemic transformation, which is contrary to its role as a transcriptional repressor in PRC1 (Tan et al., 2011; Maethner et al., 2013). The physical interaction between ENL and CBX8 eliminated the transcription repression activity of PRC1, which was required for efficient transformation of hematopoietic cells (Maethner et al., 2013). Along with that, CBX8 competed with H3K79 methyltransferase disruptor of telomeric silencing 1-like (DOT1L) for binding to AF9, resulting in up-regulation of the ENaCα gene, a known target of AF9-DOT1L (Malik and Hemenway, 2013). In this case, it was known that CBX8 worked in the context of PRC1 since the other PRC1 components including RING1B and BIM1 were found together with CBX8. Although essential for MLL-AF9 induced leukemic transformation, analyses of the Cbx8 conditional-depletion mice on hematopoietic steady-state conditions revealed that CBX8 is not required for steady-state hematopoiesis, long-term HSC maintenance, or stem and progenitor cell function (Tan et al., 2011).

5.  Regulations of CBX proteins

 Besides the auto-regulation among CBX family members at the transcriptional level discussed above, CBX proteins are also subjected to various post-translational regulations. The reported information about identified post-translational modifications is listed in Table 4, which involves phosphorylation, SUMOylation/de-SUMOylation, and methylation/de-methylation. Although phosphorylation events have been detected in all five CBX proteins, it is not clear whether they share any common functional mechanisms. Many phosphorylation sites were detected in mass spectrometry analyses without further validation. The most recent discovery was that phosphorylation of CBX7 by mitogen-activated protein kinase (MAPK) helped strengthen its association with PRC1 upon mitogen stimulation. As a SUMO E3 ligase, CBX4 itself is subject to the reversible regulation of SUMOylation and de-SUMOylation, and SUMOylation at Lys492 facilitates its interaction with H3K27me³ (Kang et al., 2010). A recent work has demonstrated that the unmethylated and methylated forms of CBX4 specifically bind to distinct ncRNA named MALAT1/NEAT2 and TUG1, respectively (Yang et al., 2011). The switch between these interactions correlates with the subnuclear relocation of growth-control genes in response to growth signals. The binding of NEAT2 to unmethylated CBX4 promotes E2F1 SUMOylation, resulting in activation of the growth-control genes. Some key points remain unclear, such as whether the enzymatic activity of CBX4 in this transcription activation event relies on the presence of PRC1 complex, and whether the outcome of transcription activation is achieved by somehow inhibiting the PRC1 function.

6.  Misregulation of CBX proteins and cancer

 Misregulation of PcG proteins has been associated with many cancer types. In regard to CBX proteins, there is increasing evidence that they play an important role in tumor initiation, progression, and development by blocking differentiation and promoting self-renewal of cancer stem cells. The reports about misregulation of CBX proteins in various cancers are listed in Table 5. The findings are categorized based on the direction of expression changes for these CBX proteins in a given cancer type. CBX7 is the most characterized CBX protein in cancer-associated studies. The level of CBX7 is elevated in prostate, lymphoma, and gastric cancer; conversely loss of CBX7 has been detected in a prevalence of other cancer types, such as thyroid, colon, pancreatic, and breast cancer. The apparent dual activities of CBX proteins as either oncogenes or tumor suppressors in distinct cancer types may mirror their specialized function during embryonic development and tissue specification. As the discovery about mouse HSC discussed above, CBX7 was the primary CBX protein in the maintenance of the stemness and elevated level of CBX7 promoted T cell lymphomas genesis. However, in the same system, over-expression of other ‘non-dominant’ CBX proteins including CBX2, CBX4, and CBX8 actually triggered cell differentiation and decreased cell proliferation (Klauke et al., 2013).




 Piles of evidence indicate that elevated expression of CBX proteins (CBX2, CBX7, and CBX8), together with other PRC1 subunits such as BMI1, directly causes transcription repression of the Ink4a/Arf locus. The locus encodes two independent tumor suppressors, INK4a and ARF, which activate the retinoblastoma (Rb) and p53 tumor suppressor pathways, respectively. Regulation of the expression of this locus by CBX protein-containing PRC1 modulates the balance between cell proliferation and senescence (Bernard et al., 2005; Scott et al., 2007; Maertens et al., 2009; Zhang et al., 2010).

 Loss of Cbx7 expression may be strictly correlated with the acquisition of invasiveness accompanied in a process called the epithelial mesenchymal transition (EMT), which has been seen in several types of cancer, such as thyroid cancer, lung cancer, colon cancer, and pancreatic cancer. CBX7 was found to control the cell cycle and cell proliferation via regulation of the expression of the encoding cyclin E 1 (CCNE1). CBX7 directly competed with the transcription activator HMGA1 for binding to the promoter of CCNE1, therefore causing down-regulation of the CCNE1. Lack of the CBX7 protein in human lung cancer carcinomas correlates with over-expression of CCNE1 (Forzati et al., 2012a; 2012b).

 CBX7 and EZH2 are also known to physically associate with DNA methyltransferases (DNMTs), responsible for programming the altered DNA methylation profiles observed in multiple cancer types (Mohammad et al., 2009). This interplay uncovered another mechanism by which multiple sources of epigenetic regulators were integrated to silence their common targets. CBX7 was also known to influence early breast cancer development by controlling microRNA expression (Hannafon et al., 2011), while in breast cancer, CBX8 cooperated with SIRT1 for suppressing p53 acetylation induced by sirtinol and etoposide/TSA (Lee et al., 2013). Upon ectopic expression, CBX8 or SIRT1 repressed the expression of p21 by inhibiting p53 binding to the promoter.

 The most studied components of PcG complexes in cancer research are EZH2 and BMI1, and they both are often over-expressed in cancers (Mills, 2010). They have been shown to be essential in controlling proliferation and maintaining self-renewal of cancer stem cells in various models (Crea et al., 2012). Some PRC2 inhibitors, such as DZNeP (a non-specific inhibitor of histone methylation), have been tested on cancer models and exhibited effectiveness in abolishing cancer stem cells self-renewal and tumorigenicity (Crea et al., 2012). The association between CBX proteins in cancer cells and stem cell proliferation makes them become the focus of ongoing investigations in cancer research. The functional specification among CBX family members provides a reasonable hypothesis that each CBX protein may play very unique roles in various lineages and alter the behavior of normal or transformed stem cells in specific contexts.

7.  Concluding remarks

 Epigenetic regulation by PcG complexes plays a crucial role in maintaining stemness and determining stem cell fates. The diversity of the composition of PcG complexes, particularly PRC1 has been found to be far more complicated than initially estimated. Recent studies uncovered the function specification of the PRC1 components, such as CBX family members, in fine-tuning the cell fate of the different lineages. These discoveries may help explain the reason why both aberrant up- and down-regulation of PcG proteins have been detected in distinct types of cancer. Further exploration of the specified functions of PcG proteins and the detailed molecular consequences triggered by abnormally acting PcG complexes would be extremely meaningful. At the same time, a lot more remains to be investigated and discovered relating to the interplay among PcG functions and other transcription regulatory mechanisms including other epigenetic regulators, ncRNAs, transcription machinery, and transcription factors. In addition, there is a plethora of post-translational modifications of PcG proteins identified in various cell types and tissues with little known mechanisms. Altogether, increasing knowledge about the roles of PcG proteins in stem cell biology and oncology can help to explore these epigenetic modifiers as potential biomarkers for cancer and further develop therapeutic strategies for early diagnosis, prognosis, and treatment of cancer.

---

* Project supported by the Fundamental Research Funds for the Central Universities from Lanzhou University (No. lzujbky-2014-87), ChinaAuthors contributions Rong-gang MA, Yang ZHANG, and Ting-ting SUN collected all the references, made all the figures and tables, and wrote a draft. Bo CHENG wrote the manuscript.Compliance with ethics guidelines Rong-gang MA, Yang ZHANG, Ting-ting SUN, and Bo CHENG declare that they have no conflict of interest.