Polycomb Group Genes: Keeping Stem Cell Activity in Balance

Overexpression of Polycomb group genes is often associated with cancer development, whereas complete deletion results in loss of stem cell activity. New studies show that partial loss of function of Polycomb group genes enhances the activity of blood stem/progenitor cells.

site for the recruitment of the PRC1 complex, resulting in uH2AK119. This covalent modification likely prevents full access to other chromatin remodeling factors or the transcription machinery and facilitates chromatin compaction (see Figure 1) [10][11][12][13].
In mice, loss of function of all core PRC2 components studied to date is embryonic lethal due to severe defects at the implantation and early post-implantation stages (see Table  1). Recently, it was found that embryonic stem (ES) cells mutant for PRC2 genes lose the ability to maintain themselves in an undifferentiated state [14,15]. With the exception of mice mutant for Ring1b, which is essential for the survival of early embryos, homozygous null mutant mice for other PRC1 genes (i.e., Bmi1, Mel18, Cbx2, or Phc1) survive to birth, but all display homeotic transformations and die perinatally (see Table 1). Functional redundancy and compensation by paralogous genes may explain the milder phenotypes found with most PRC1 versus PRC2 homozygous null mutant mice.
Both PRC1 and PRC2 genes are implicated in regulation of stem cell self-renewal and in cancer development (reviewed in Sparmann et al. [10] and Rajasekhar et al. [16]). Bmi1 was first discovered as an oncogene overexpressed in lymphomas and cooperating with c-Myc [17]. It was found to regulate proliferation and senescence mainly through repression of the Ink4a locus [18]. In addition, Bmi1 is overexpressed in human leukemias and different types of solid cancers [10,16]. This gene also represents an essential regulator of self-renewal for both normal and leukemic hematopoietic stem cells (HSCs), since both of these cell types eventually disappear in its absence [19]. Similar phenotypes were observed with the loss of function of Phc1, another PRC1 gene [20].
The PRC2 genes EZH2 and SUZ12 are also overexpressed in a broad spectrum of human cancers [10,16]. Notably, EZH2 is known as a marker for "aggressiveness" in prostate and breast cancer [21,22]. Moreover, recent studies indicate that overexpression of the Ezh2 protein in mouse HSCs preserves self-renewal activity in serial passages, a condition never observed in unmanipulated HSCs and sometimes referred to as "HSC senescence" [23]. This type of activity may be exploited by tumor cells that overexpress these genes. Although activity of Ezh2 and Ezh1 homozygous null HSCs remains undescribed, the data with Ezh2 overexpression are reminiscent of those recently observed with Bmi1, potentially indicating that similar molecular bases (e.g., H3K27 trimethylation; H2A mono-ubiquitination) underlie PRC1 and PRC2 function in HSCs. However, contrasting with Ezh2 overexpression, partial loss of function and hypomorphic alleles of its PRC2 partner, Eed, restricts the proliferation of lymphoid and myeloid progenitors and antagonizes PRC1 function [24]. Two independent studies have also demonstrated that Eed possesses tumor-suppressive activity in the hematopoietic system [25,26]. Therefore, it seems that adequate PcG protein levels and activity are important and greatly affect the ability of cells to excessively self-renew (the result of high PcG levels) or to become transformed (the result of low levels).
In this issue of PLoS Biology, a study by Ian J. Majewski et al. [27] further strengthens the notion that PRC2 restricts cellular proliferation. In their study, the authors provide evidence that Suz12 is sensitive to gene dosage in the hematopoietic compartment and that reduction in Suz12 levels enhances the activity of certain hematopoietic cells. By using ENU (N-ethyl-N-nitrosourea) mutagenesis and positional cloning experiments, Majewski et al. [27] identified an inactivating point mutation in Suz12, called Plt8, which is embryonic lethal in the homozygous state. More importantly, the study showed that heterozygote Suz12 Plt8/+ mice are viable and display increased numbers of platelets, megakaryocytes, lymphoid cells, and certain progenitors. Interestingly, the Plt8 mutation partly rescues the hematopoietic phenotype observed in mice lacking the thrombopoietin receptor c-Mpl. Moreover, the authors show that Suz12 Plt8/+ bone marrow cells are more competitive than wild-type counterparts, suggesting a negative regulatory role for Suz12 in HSC activity. The phenotype described in Suz12 Plt8/+ mice was reproduced by partial knockdown of Suz12 using RNA interference, confirming that the mutant phenotype is a result of decreased Suz12 expression. The authors also showed that Ezh2 levels are reduced in Suz12 Plt8/+ cells and that heterozygotic mutation of Ezh2 rescues defects seen in c-Mpl -/mice similarly to Suz12 Plt8/+ mutants. Although further experiments are needed, this suggests that Ezh2 is also haploinsufficient and that low levels enhance hematopoietic activity. The study by Majewski et al. [27] is clearly reminiscent of the results seen in partial loss of function of Eed. It indicates that complete loss of PRC2 components is detrimental to cells and produces unviable embryos, but that partial reduction in their levels has the opposite effect and enhances HSC and progenitor cell activity. In the case of Eed null/+ and homozygous hypomorph mutants, this reduction eventually leads to leukemia development [24][25][26]. Although the authors did not observe any leukemia in Suz12 Plt8/+ mice, oncogenic insults and additional mutagenic events may be required for full transformation of Suz12 Plt8/+ cells. This hypothesis could also be true for Ezh2 and should be tested. Interestingly, the human chromosomal locations of EED, EZH2, and SUZ12 are all found in regions of recurrent chromosomal deletions and aberrations. EED is particularly interesting because it is located in close proximity to ATM and MLL, two genes frequently involved in hematopoietic malignancies. Irradiation or carcinogen treatment of Suz12 Plt8/+ or Ezh2 +/cells may thus reveal a similar tumor-suppressive function as observed with Eed mutant mice.
Together with the current knowledge on Polycomb group genes and their role in self-renewal and cancer, the study by Majewski et al. [27] provides further evidence for a delicate balance and tight regulation of the PRC2 complex levels for proper function of stem and progenitor cells. This leads to a gene dosage model where up-regulation or modest downregulation of the PRC2 complex tips the balance toward enhanced HSC activity and increased chances of developing tumors, whereas complete knockout results in stem cell loss (see Figure 2). This model raises many questions regarding the function of Polycomb group genes in stem cell self-renewal and cancer. First, is there a similar dosage effect for PRC1 genes? Human PHC1 is located on Chromosome 12p13, a region frequently associated with loss of heterozygosity in acute lymphoblastic leukemia [28]. Studies on compound Bmi1 and Mel18 mutant mice seem to suggest that these genes are sensitive to dosage variations [29]. Careful analysis of stem cell activity and sensitivity to transformation in heterozygous mice would be of great interest. The mechanisms through which PcG haploinsufficiency versus overexpression leads to cancer are also yet to be defined. Do the results observed occur through similar or distinct pathways? This question is especially relevant now that we know that PcG proteins interact with    [31]. The histone octamer (in grey) is complexed with 146 base pairs of DNA (red). The histone H3 lysine 27 (blue) located on the N-terminal tail is tri-methylated (H3K27me3) by the PRC2 complex. The histone H2A lysine 119 (green), which is mono-ubiquitylated (uH2AK119) by the PRC1 complex, is located near the entry and exit point of DNA on the histone octamer. (B) In accordance with recent studies, the nucleosome structure shows that, because of their location at the entry and exit point of DNA, ubiquitin molecules (beige) bound to H2AK119 could maintain genes in a repressed state by limiting the access of the RNA polII to chromatin [11]. Interestingly, ubiquitinated H2AK119 is also located at the linker-histone H1 binding region. Studies have shown that uH2AK119 enhances histone H1 interaction with the nucleosome [12,13], suggesting that this epigenetic modification is important for maintaining the compacted chromatin structure. multiple other proteins and potentially have non-histone substrates, suggesting as yet unknown functions for both PRC1 and PRC2 complexes.
Taking into account that most cancers are derived from a single cell (clonal), it can be difficult to compare PcG gene expression levels in the rare normal cells in which transformation occurs to that in the cancer stem cells. Tools and knowledge are becoming available to resolve this important issue. Likewise, it is still not clear if PRC2 and/or PRC1 activity is enhanced as a result of PcG gene deregulation in these normal or tumor stem cells. Although a pattern of PcG-mediated histone modifications was recently ascribed to certain stem cells [30], its implication in selfrenewal remains difficult to assess. Such an endeavor would require the generation of histone mutants, a technical challenge in vertebrates considering the multiple variants and genes coding for all four nucleosomal subunits. In addition, evidence that PcG proteins also display non-chromatinrelated activity raises a fundamental issue about the targets (i.e., nucleosomes versus others) that control self-renewal in cancer and normal stem cells.
Finally, since very little is known about the transcriptional and post-transcriptional regulation of PcG genes, it becomes important to elucidate the pathways that determine the cellular levels of these proteins in order to prevent stem cell loss and cancer development.