Molecular Cell
Previews Demethylases Go Mental Mark A. Dawson1,2,3 and Andrew J. Bannister3,* 1Department
of Haematology, Cambridge Institute for Medical Research Hospital University of Cambridge, Cambridge CB2 0XY, UK 3Gurdon Institute and Department of Pathology, Tennis Court Road, Cambridge CB2 1QN, UK *Correspondence:
[email protected] DOI 10.1016/j.molcel.2010.04.002 2Addenbrooke’s
Histone-modifying enzymes require recruitment to specific genomic loci in order to elicit their effects. In this issue, Kleine-Kohlbrecher et al. reveal how a histone demethylase is targeted to specific genes, thereby providing mechanistic insight into recruitment and regulation. X-linked mental retardation (XLMR) is a diverse clinical syndrome caused by a myriad of genetic abnormalities. Thus far, more than 80 distinct XLMR genes, accounting for up to 11% of all X chromosome genes, have been implicated in the pathogenesis of this syndrome, and despite a prevalence as high as 1 in 600 males, mechanistic insights into this complex developmental abnormality are limited (Ge´cz et al., 2009). What is clear, however, is that many of the XLMR genes encode proteins that function in transcriptional regulation, either as DNA sequencespecific binding proteins or as regulators of chromatin structure. Like transcription factors (TFs), chromatin regulators play an integral role in coordinating gene expression. In the last few years, we have seen a marked increase in our basic understanding of chromatin-modifying enzymes, and this, in turn, has provided insight into the molecular pathogenesis of complex diseases. However, our knowledge of the mechanisms recruiting chromatin-modifying enzymes to specific genomic loci is far from complete. In this issue of Molecular Cell, Helin and colleagues functionally link three XLMRassociated proteins (Kleine-Kohlbrecher et al., 2010). Importantly, they demonstrate how one of these proteins, the PHF8 JmjC domain-containing protein, is specifically targeted to genes. In many proteins, JmjC domains possess lysine demethylase activity. Using a combination of immunofluorescence and in vitro biochemical assays, the authors show that PHF8 is a histone H3K9me2/me1 demethylase. Of interest, however, and in keeping with other characterized
histone demethylases, PHF8 has negligible activity on nucleosomes in vitro. This raises the intriguing possibility that an ancillary factor is required for full nucleosomal demethylase activity in vivo, a feature that may be relatively common for this class of enzymes. Furthermore, XLMR-associated PHF8 mutant proteins do not demethylate H3K9me2, suggesting that loss of this activity is linked to development of XLMR. Another feature often shared by demethylases is the presence of a plant homeodomain (PHD) finger (Lan et al., 2008). Helin and colleagues find that the PHF8 PHD finger binds to H3K4me3, a histone modification typically located at the 50 end of active genes. Genomic analyses using ChIP-Seq reveal a striking coincidence of H3K4me3 and PHF8 at 6301 genes (80% of all PHF8-bound genes). Thus, it seems reasonable to suggest that PHF8 is recruited to genes, at least in part, via binding of its PHD domain to H3K4me3 (Figure 1). Notwithstanding the above, approximately one-third of genes harboring H3K4me3 lack PHF8, an indication that H3K4me3 is not sufficient for PHF8 recruitment. To identify potential proteins involved in PHF8 recruitment, the authors affinity purified the PHF8 complex. They then used mass spectrometry and identified ZNF711, a TF also associated with XLMR, as a PHF8-binding partner. In vitro, ZNF711 binding mildly potentiates the enzymatic activity of PHF8, whereas in vivo, ZNF711 colocalizes with PHF8 and H3K4me3 at more than 800 genes. Surprisingly, however, although PHF8 is detected at several documented XLMR genes, the presence of both ZNF711 and
PHF8 is noted at only a single XLMR gene, JARID1C. This may be a reflection of the cell lines used for these experiments. Of interest, knockdown of ZNF711 leads to a decrease of PHF8 and H3K4me3 at the transcription start site of several target genes, with a concomitant reduction in gene expression but without a reciprocal increase of H3K9me2. Therefore, reduced levels of PHF8 and ZNF711 are not sufficient for recruitment of a H3K9 methyltransferase. Consequently, these data raise an important issue concerning the mechanism of transcriptional silencing at these loci in the absence of the PHF8 complex. On the other hand, they provide compelling evidence that histone demethylases can be targeted to specific genomic loci via association with specific DNA-binding TFs (Figure 1). Finally, the authors turn to the C. elegans model system in an attempt to gain functional insight into the role of PHF8. Here, in contrast to mammalian cells, perturbation of the PHF8 homolog F29B9.2 clearly leads to a global increase of H3K9me2 in vivo. Of interest, F29B9.2 is expressed neuronally, and its mutation results in a gross locomotion defect that is rescued by the wild-type protein expressed from a pan-neuronal, but not a muscle-specific, promoter. These are significant observations because human PHF8 is expressed in early brain development and inactivating mutations are tightly linked with mental retardation. Altogether, they point toward an important role for PHF8 in developmental neurobiology. Human PHF8 has recently also been identified as an important regulator of
Molecular Cell 38, April 23, 2010 ª2010 Elsevier Inc. 155
Molecular Cell
Previews Targeting histone modifying enzymes (eg PHF8) to specific genes
H3
K4
Histone modifying enzyme contains a domain (eg PHD) that recognizes/binds directly to a histone modification (eg H3K4me3). Binding may affect enzyme activity.
Locus-Specific Enzyme Activity
ZNF 711
m m e m e e
H3
m m e m e e
PHF8
K4
Specific transcription factor plus histone modification is required for recruitment (and/or activity) of histone modyifying enzyme. Distinct factors may recruit PHF8 to other genes with H3K4me3.
integration of histone modification and transcription factor binding defines enzyme recruitment and activity
ZNF 711
?
?
?
Specific transcription factor plus unknown component of chromatin is required for recruitment (and/or activity) of histone modifying enzyme. Distinct factors may recruit PHF8 to other genes lacking H3K4me3.
Unknown factor (eg protein or RNA) plus unknown component of chromatin is required for recruitment (and/or activity) of histone modifying enzyme.
Figure 1. Potential Recruitment Mechanisms for Histone-Modifying Enzymes Schematic depiction of nonmutually exclusive potential recruitment mechanisms for histone-modifying enzymes such as PHF8. Recruitment via binding to histone modifications and/or transcription factors may also directly modulate enzyme activity.
rRNA genes and cellular proliferation (Feng et al., 2010). Although this study was not performed in neural cells, we do know that PHF8 is highly expressed in specialized structures of the brain such as the hippocampus (Laumonnier et al., 2005). Consequently, a proliferative defect in neurogenesis may be detrimental to certain susceptible neuronal populations such as those involved in memory and learning, ultimately manifesting as XLMR. In this regard, it is noteworthy that many of the genes bound by PHF8 and ZNF711 are themselves required for cellular proliferation (KleineKohlbrecher et al., 2010). A number of important issues arise from this work. For instance, we need to understand how perturbation of a ubiquitously expressed chromatin-modifying enzyme, present on several thousand genetic loci, culminates in such a specific and subtle neurological defect. In addition,
as PHF8 is present on so many genes, its perturbation will presumably lead to other developmental problems. Indeed, this is the case; mutation of PHF8 in humans has also been linked to cleft lip and palate malformation, whereas ZNF711 has no such connection. Considering these issues together, it seems likely that different PHF8 complexes function in different cells at different times and that the consequences of PHF8 perturbation during development depend on spatial and temporal parameters (i.e., which target genes are deregulated in which specific cells at which specific developmental point). A key question in chromatin biology is how are the regulators recruited to, and regulated at, their sites of action? The findings of Kleine-Kohlbrecher et al. clearly indicate that multiple mechanisms exist for modifying enzymes. Direct recruitment via binding to pre-existing
156 Molecular Cell 38, April 23, 2010 ª2010 Elsevier Inc.
histone modifications is one possible way, whereas indirect recruitment via TF binding is another. PHF8 appears to employ both of these mechanisms. The present data also suggest other, but not necessarily mutually exclusive, mechanisms (Figure 1). The efficiency and stability of recruitment will depend on the interactions involved. Furthermore, peptides bearing H3K4me3 stimulate the demethylase activity of PHF8 (Feng et al., 2010; Horton et al., 2010), as does interaction with ZNF711 (Kleine-Kohlbrecher et al., 2010). It is therefore tempting to speculate that PHF8 can integrate, via regulation of its enzymatic activity, information derived from the recognition of modified histones and sequence-specific DNA-bound TFs. In this scenario, the presence of cell-type-specific histone modifications and/or TFs could regulate the demethylase activity of PHF8 at specific genes at specific times.
Molecular Cell
Previews Further investigation will undoubtedly provide insight into these recruitment mechanisms, and this, in turn, will lead to a more comprehensive understanding of how chromatin-modifying enzymes regulate gene expression. This will not only provide a clearer insight into the pathogenesis of disease, but will also offer greater opportunity for therapeutic intervention to curtail often devastating disorders such as XLMR.
REFERENCES Feng, W., Yonezawa, M., Ye, J., Jenuwein, T., and Grummt, I. (2010). Nat. Struct. Mol. Biol. 17, 445–450.
Ge´cz, J., Shoubridge, C., and Corbett, M. (2009). Trends Genet. 25, 308–316.
Horton, J.R., Upadhyay, A.K., Qi, H.H., Zhang, X., Shi, Y., and Cheng, X. (2010). Nat. Struct. Mol. Biol. 17, 38–43.
Kleine-Kohlbrecher, D., Christensen, J., Vandamme, J., Abarrategui, I., Bak, M., Tommerup, N., Shi, X., Gozani, O., Rappsilber, J., Salcini, A.E., and Helin, K. (2010). Mol. Cell 38, this issue, 165–178. Lan, F., Nottke, A.C., and Shi, Y. (2008). Curr. Opin. Cell Biol. 20, 316–325. Laumonnier, F., Holbert, S., Ronce, N., Faravelli, F., Lenzner, S., Schwartz, C.E., Lespinasse, J., Van Esch, H., Lacombe, D., Goizet, C., et al. (2005). J. Med. Genet. 42, 780–786.
SUMO Weighs In on Polycomb-Dependent Gene Repression Grace Gill1,* 1Department of Anatomy and Cellular Biology, Tufts University School of Medicine, 136 Harrison Avenue, Boston, MA 02111, USA *Correspondence:
[email protected] DOI 10.1016/j.molcel.2010.04.006
In this issue of Molecular Cell, Kang et al. (2010) report that their studies to characterize the biological activity of a SUMO-specific protease, SENP2, have revealed an important role for reversible SUMO modification in regulating promoter occupancy and gene-specific repression by polycomb repressive complex 1 (PRC1). Polycomb group (PcG) proteins are key regulators of chromatin structure and gene expression in development, stem cell biology, and cancer (Bracken and Helin, 2009). The PcG proteins assemble into large multiprotein complexes called polycomb repressive complexes (PRC) that function to silence expression of numerous target genes. Notably, the two main complexes, PRC1 and PRC2, have associated histone-modifying activities that contribute to repression. PRC2 subunits catalyze trimethylation of H3K27. H3K27me3-modified histones contribute to recruitment of PRC1, which catalyzes ubiquitination of H2A. The combined activities of PRC2 and PRC1 present a barrier to transcription (Figure 1). In stem cells, targets of PcG-mediated silencing include key regulators of specific differentiation programs, such as the Hox genes, whereas, in differentiated cells, PcG proteins repress key regulators of alternative fates as well as some genes associated with stemness (Boyer et al., 2006; Lee et al., 2006). Thus, the identity of specific
gene targets for PcG-mediated repression is a critical determinant of cell fate, and this choice is dynamically regulated during lineage specification. The mechanisms by which gene-specific PcG binding and release are regulated during development are not fully understood. Polycomb group proteins are subject to diverse posttranslational modifications that may contribute to regulation of PcG binding and activity. Posttranslational modification by the small ubiquitin-related modifier, SUMO, has emerged as an important mechanism to regulate transcription and chromatin structure. Covalent attachment of SUMO to lysine residues in substrates is mediated by SUMO-specific E1 (AOS1/UBA2) and E2 (UBC 9) enzymes and can be stimulated by various E3 ligases. SUMOylation is reversible, and six SUMO-specific proteases (SENPs) have been identified in mammals. The finding that SUMO modification of many transcription factors and cofactors is correlated with transcriptional repression has been proposed to
depend, in part, on noncovalent interactions with corepressors bearing SUMO interaction motifs (SIMs) (Gill, 2004). In fact, SUMO has been shown to promote promoter-specific recruitment of diverse chromatin-modifying enzymes, including the demethylase LSD1 and deacetylases HDAC1 and 2 (Ouyang et al., 2009). Several PcG proteins can be posttranslationally modified by SUMO, and one, Pc2, has been reported to promote SUMOylation of other substrates such as CtBP (Kagey et al., 2003). Studies in C. elegans have supported a role for SUMOylation of the PcG protein SOP-2 in repression of Hox genes (Zhang et al., 2004). Little is known, however, about how SUMOylation affects PcG-dependent repression in mammals. In this issue of Molecular Cell, Kang et al. (2010) present data to suggest that SUMOylation and deSUMOylation of the PcG protein Pc2/CBX4 contributes to the regulated binding and release of PRC1 and thus affects expression of PcG target genes required for lineage specification.
Molecular Cell 38, April 23, 2010 ª2010 Elsevier Inc. 157
