Molecular Cell Previews

Histone Demethylation and Timely DNA Replication

Erica L. Gerace1 and Danesh Moazed1,2,* 1Department of Cell Biology 2Howard Hughes Medical Institute Harvard Medical School, Boston, MA 02115, USA *Correspondence: [email protected] DOI 10.1016/j.molcel.2010.11.036

It is well-established that silent regions of the genome replicate late during S phase. In this issue of Molecular Cell, Black et al. (2010) uncover a conserved role for the JMJD2 family of histone demethylases in promoting replication within silent chromatin regions that contain histone H3 lysine 9 methylation and HP1.

The faithful replication of the genome JMJD2 family was found to catalyze the level of initiation and/or fork during each cell division is a vital process the demethylation of trimethylated H3K9, progression. that is highly coordinated and tightly H3K36, and H1.4K26 (Fodor et al., 2006; Analysis of jmjd-2À/À worms revealed controlled. In eukaryotic cells, origins of Klose et al., 2006; Trojer et al., 2009; a decreased number of cells within the DNA replication, the sites where replica- Whetstine et al., 2006). Depletion of mitotic zone and an increased number of tion begins, are found throughout the JMJD-2 in C. elegans results in the activa- RAD-51 foci, an indication of stalled or genome and fire at different times tion of a DNA damage-induced apoptosis collapsed replication forks. Furthermore, throughout S phase. This timing is influ- (Whetstine et al., 2006), but the molecular direct in vivo visualization of DNA replica- enced by local chromatin structure and basis of this event had not been tion using pulse-chase labeling experi- correlates with cell type-specific differ- established. ments with cy3-dUTP/ALEXA-488-dUTP ences in gene expression and chromatin In this issue of Molecular Cell, Black in the adult germlines of jmjd-2À/À worms structure. Although there are exceptions, et al. (2010) describe a role for histone showed that replication was slowed, sug- in general, transcriptionally active por- demethylation in cell cycle progression gesting that in vivo, JMJD-2 likely affects tions of the genome are replicated earlier and DNA replication. They report that the replication timing. This observation also in S phase than those found in silent, expression of JMJD2A is cell cycle regu- explains the previously observed in- heterochromatic regions (MacAlpine and lated with peak expression at the G1/S creased apoptosis in jmjd-2 mutants Bell, 2005). Histone H3 lysine 9 (H3K9) transition dropping off in S phase and (Whetstine et al., 2006), a phenotype that methylation is a conserved modification lowest at G2/M. The overexpression of Black et al. (2010) also show can be that is found in heterochromatin from JMJD2A, but not catalytically inactive rescued by knockdown of the p53 and fission yeast to human, and defines a JMJD2A, which has a mutation at a key ATR homologs, CEP-1 and ATL-1, re- binding site for Heterochromatin Protein metal-coordinating residue required for spectively. However, the mutation of the 1 (HP1) family members, which help demethylase activity (Whetstine et al., ATM homolog did not provide a rescue make heterochromatin inaccessible. Re- 2006), in human cell lines led to faster of the observed phenotype, indicating cent studies suggest that heterochro- progression through S phase. In addition, that the increased apoptosis is mediated matin is more dynamic than previously the overexpression of JMJD2A resulted through the ATR pathway, which is acti- appreciated and can be broached by in both the early replication of a late- vated after replication stress, but not the machineries that transcribe and repair replicating satellite region, sat2 of chro- ATM pathway, which is typically activated DNA (Kwon and Workman, 2008). How- mosome 1 (Chr1 sat2), as well as faster in response to double-strand breaks. ever, how access to heterochromatin or recovery of these cells after treatment How does JMJD2A-mediated histone other silent regions is regulated is poorly with hydroxyurea (HU), which causes demethylation contribute to DNA rep- understood. replication stress. Similarly, mutating or lication timing? The overexpression of The discovery of histone lysine deme- knocking down the gene for the worm JMJD2A causes the redistribution of thylases suggested new mechanisms for homolog, jmjd-2, resulted in increased HP1 proteins (Klose et al., 2006) likely by regulation of chromatin structure and, sensitivity to HU. These results suggest removing the chromatin mark to which indeed, these enzymes have already that expression of JMJD2A/JMJD-2 dur- HP1 proteins bind, trimethylated H3K9. been identified with numerous roles in ing S phase facilitates DNA replication. So, potentially JMJD2A may regulate regulation of transcription. Histone lysine Consistently, Black et al. also demon- chromatin structure and replication timing demethylation is catalyzed by two distinct strated that more single-stranded DNA via effects on HP1 localization. Indeed, classes of evolutionarily conserved en- was observed during S phase in cells using micrococcal nuclease digestions zymes (Shi and Whetstine, 2007). JmjC ectopically expressing JMJD2A, which of isolated nuclei, Black et al. (2010) domain proteins are metalloenzymes suggests increased presence of replica- observed increased DNA accessibility in that catalyze oxidative reactions. Several tion forks. However, it remains unknown cells overexpressing catalytically active years ago, a JmjC domain protein of the whether the faster S phase effects are at JMJD2A, specifically during S phase. In

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 683 Molecular Cell Previews

addition, in these cells, the (1) wild-type regions and the mating type chromatin structure of the locus (Hayashi et al., 2009). late-replicating locus Chr1 JMJD2A Surprisingly, Black et al. sat2 was found to be more HP1γ HP1γ (2010) found that deletion of open corresponding to its H3K9me3 the worm HP1g homolog observed earlier replication. (HPL-2) had the same de- These data support the no- ori layed replication phenotypes tion that altering chromatin as jmjd-2À/À cells, suggesting accessibility can regulate (2) JMJD-2 depletion that intricacies in the balance cell cycle progression and between the two worm HP1 replication timing, providing proteins are also important another way to regulate cell for proper replication timing. HP1γ HP1γ HP1γ cycle that is independent or We can therefore look for- Me Me Me Me Me complimented by altered ward to future studies that transcriptional programs. ori ori will explore possible direct Interestingly, it appears that links between different HP1 JMJD2A/JMJD-2 act by proteins and activation or antagonizing a specific HP1 silencing of DNA replication. isoform. Most cells contain (3) JMJD2A overexpression two or more HP1 homologs, HP1a,-b, and HP1g in human, JMJD2A HP1γ REFERENCES JMJD2A and HPL-1 and -2 in worms. HP1γ Black, J.C., Allen, A., Capucine, The overexpression of HP1g, Me V.R., Forbes, E., Longworth, M., but not HP1a or HP1b, sup- Tscho¨ p, K., Rinehart, C., Quiton, pressed the JMJD2A-depen- ori J., Walsh, R., Smallwood, A., et al. (2010). Mol. Cell 40, this issue, dent increased number of 736–748. cells in late S phase, an effect that depended on Figure 1. Human and C. elegans JMJD2A/JMJD-2 Antagonize Fodor, B.D., Kubicek, S., HP1g/HPL-2 and Promote DNAReplication Yonezawa, M., O’Sullivan, R.J., both the chromodomain and (1) HP1g binds trimethylated histone H3 lysine 9 (H3K9me3), maintaining Sengupta, R., Perez-Burgos, L., chromoshadow domain of a closed chromatin structure, which inhibits the firing of origins of replication Opravil, S., Mechtler, K., Schotta, HP1g. Also, at the Chr1 sat2 or slows the progress of replication forks. Demethylation of H3K9me3 results G., and Jenuwein, T. (2006). Genes in dissociation of HP1g, thus promoting DNA replication and cell cycle Dev. 20, 1557–1562. locus, the overexpression of progression. (2) In JMJD-2-depleted germline nuclei in C. elegans, H3K9me3 JMJD2A decreased HP1g levels and HPL-2/HP1g association increase, leading to delayed replication Hayashi, M.T., Takahashi, T.S., origin firing. Potential spreading of HPL-2/HP1g may also prevent the firing Nakagawa, T., Nakayama, J., and localization. In worms, the Masukata, H. (2009). Nat. Cell Biol. À/À of distally located origins of replication. (3) During S phase or when JMJD2A jmjd-2 phenotypes of in- 11, 357–362. is overexpressed, demethylation of H3K9me3 and HP1g displacement allow creased RAD-51 foci, de- origin firing resulting in faster progression through S phase. Klose, R.J., Yamane, K., Bae, Y., creased number of nuclei in Zhang, D., Erdjument-Bromage, the mitotic zone, slowed replication, and at specific heterochromatic regions, anal- H., Tempst, P., Wong, J., and Zhang, Y. (2006). increased apoptosis are all rescued by ogous to human satellite repeats. The Nature 442, 312–316. the depletion of HP1 homolog HPL-2, findings also add to an emerging body of Kwon, S.H., and Workman, J.L. (2008). Mol. Cells establishing a conserved antagonistic evidence that suggests distinct roles for 26, 217–227. relationship between these two proteins. different HP1 proteins in regulation of MacAlpine, D.M., and Bell, S.P. (2005). Chromo- Together the results suggest that DNA replication and other heterochro- some Res. 13, 309–326. JMJD2A/JMJD-2 modulate replication matin functions, which may relate to Motamedi, M.R., Hong, E.J., Li, X., Gerber, S., timing by opposing reduced DNA acces- protein interactions and/or genomic loca- Denison, C., Gygi, S., and Moazed, D. (2008). sibility to the replication machinery in tion. For example, the two fission yeast Mol. Cell 32, 778–790. DNA domains that are decorated by homologs, Swi6 and Chp2, play distinct Shi,Y., and Whetstine, J.R. (2007). Mol.Cell25,1–14. H3K9 trimethylation and HP1g (Figure 1). roles in replication of heterochromatin It remains to be determined whether the and gene silencing (Hayashi et al., 2009; Trojer, P., Zhang, J., Yonezawa, M., Schmidt, A., HP1 Zheng, H., Jenuwein, T., and Reinberg, D. (2009). activation of DNA damage checkpoint Motamedi et al., 2008). Swi6 specifi- J. Biol. Chem. 284, 8395–8405. and apoptosis in jmjd-2À/À cells result cally recruits the Cdc7 kinase to promote Whetstine, J.R., Nottke, A., Lan, F., Huarte, M., from a widespread failure to complete origin firing and early replication of het- Smolikov, S., Chen, Z., Spooner, E., Li, E., Zhang, DNA replication or from a failure to do so erochromatin at pericentromeric DNA G., Colaiacovo, M., et al. (2006). Cell 125, 467–481.

684 Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell Previews

Hunting for Alternative Disulfide Bond Formation Pathways: Endoplasmic Reticulum Janitor Turns Professor and Teaches a Lesson

Deborah Fass1,* 1Department of Structural Biology, Weizmann Institute of Science, Rehovot, Israel 76100 *Correspondence: [email protected] DOI 10.1016/j.molcel.2010.11.034

In this issue of Molecular Cell, Ron and colleagues (Zito et al., 2010b) show that an enzyme responsible for cleaning up hydrogen peroxide in the endoplasmic reticulum can contribute productively to disulfide bond formation.

Since the discovery of the sulfhydryl to explain the viability of the Ero1 cysteine to attack the interprotein disul- oxidase enzyme Ero1 in yeast more than double-knockout mice. Some of these fide bond and liberate newly oxidized a decade ago, our notion of oxidative enzymes are conveniently ER localized PDI. In wild-type cells, both Ero1a and protein folding in the endoplasmic retic- and have already been shown to interface Ero1b were trapped in complex with PDI ulum (ER) has been the following: Ero1, with ER oxidoreductases. For example, as expected (Zito et al., 2010b). Another by reducing molecular oxygen to hy- the membrane-embedded protein vitamin major hit was peroxiredoxin IV (PRDX4). drogen peroxide, generates disulfide K epoxide reductase (VKOR) reduces PRDX4 was previously shown to be ER bonds de novo. Ero1 then oxidizes protein vitamin K by concomitant oxidation of localized and, in its janitorial role, to break disulfide isomerase (PDI), and PDI in turn PDI family members in the ER (Wajih down hydrogen peroxide generated by oxidizes diverse ER substrates. Mammals et al., 2007). Though VKOR or a VKOR pa- Ero1a activity (Tavender et al., 2008; have two Ero1 paralogs, Ero1a and Ero1b, ralog may in principle take some of the Tavender and Bulleid, 2010). After hook- and approximately 20 PDI family oxidore- reduced thiol load off an Ero1-depleted ing PRDX4 with the PDI mutant, though, ductases, many of which may be oxidized ER, it was recently shown that VKOR Zito et al. (2010b) asked whether cells by Ero1 enzymes (Schulman et al., 2010). interacts specifically with membrane- derived from the Ero1 double-knockout RNA transcripts corresponding to both anchored PDI family members and much mice might be hypersensitive to depletion Ero1a and Ero1b are distributed broadly less with luminal PDI-like proteins such of PRDX4. If the sole role of PRDX4 were across human tissue types, with Ero1b as PDI itself (Schulman et al., 2010). to clean up peroxide produced by Ero1 expressed at particularly high levels in Another candidate disulfide catalyst is activity, then Ero1 double-knockout cells the pancreas (Pagani et al., 2000). The quiescin sulfhydryl oxidase (QSOX). should be less dependent on PRDX4. presence of Ero1 in many cell types is QSOX is primarily localized to the Golgi However, if PRDX4 contributes produc- consistent with a fundamental role for apparatus (Mairet-Coello et al., 2004), tively to disulfide formation in cells Ero1 in oxidative protein folding in mam- but it presumably traverses the ER to get lacking Ero1 enzymes, then these cells mals, as has been demonstrated for there. In vitro, QSOX is a poor catalyst of should be more sensitive to depleting yeast. PDI-family protein oxidation but an effec- PRDX4 activity. Indeed, by a number of The recent observation that viable mice tive oxidant of unfolded proteins (Kodali measures, depletion of PRDX4 exacer- can be produced when both Ero1a and and Thorpe, 2010). Subtle changes in bated the modest defects in disulfide Ero1b are disrupted thus came as a great QSOX levels or trafficking might rescue formation seen in the Ero1a/Ero1b surprise (Zito et al., 2010a). These mice an underoxidizing ER. knockout cells, precisely the effect one are diabetic due to defects in insulin Rather than examining known disulfide would expect if PRDX4 is an important production and secretion, but numerous catalysts, Zito et al. (2010b) instead went disulfide catalyst in this context (Figure 1). other essential functions requiring disul- ‘‘fishing.’’ Using a PDI mutant with a single Participation of PRDX4 in disulfide forma- fide bond formation seem to proceed active-site cysteine as bait, they angled tion via PDI is supported by a study in apace. With the finding that the best- for new potential PDI oxidants, which press demonstrating rapid oxidation of known catalysts of disulfide bond forma- they identified by mass spectrometry. PDI family proteins by PRDX4 in vitro tion in the mammalian ER are less The idea behind this experiment, which (Tavender et al., 2010). essential than anticipated, the search has been used to identify other dithiol/di- In wild-type cells, the ability of a partner- was on for alternative mechanisms to sulfide exchange pathways in the past, ship between Ero1 and PRDX4 to get two keep oxidative protein folding going in is that disulfide-bonded complexes can disulfides out of a single oxygen molecule the absence of the Ero1 enzymes. form between mutant PDI and its potential seems a paradigm of efficiency. Ero1 In fact, several other known enzymes oxidants, but the complexes have trouble generates one disulfide and a molecule that generate disulfides are candidates disassembling without a second PDI of hydrogen peroxide, and PRDX4 uses

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 685 Molecular Cell Previews

occurs or is regulated, and a thorough analysis of the levels of reduction- oxidation-active small molecules in the Ero1 knockout mice remains to be per- formed. Finally, enzymes such as VKOR and QSOX, which do not directly interact with PDI, may nevertheless affect the PDI redox state indirectly. They may channel oxidizing equivalents to the glutathione pool through other PDI family proteins, through re-reduction of nonna- tive disulfides in folding proteins, or through as yet undiscovered pathways.

REFERENCES

Kodali, V.K., and Thorpe, C. (2010). Antioxid. Redox Signal. 13, 1217–1230.

Mairet-Coello, G., Tury, A., Esnard-Feve, A., Fell- mann, D., Risold, P.Y., and Griffond, B. (2004). J. Comp. Neurol. 473, 334–363. Figure 1. Allegory for the Newly Discovered Activity of PRDX4 in Promoting Disulfide Pagani, M., Fabbri, M., Benedetti, C., Fassio, A., Formation in Cells Lacking Ero1 Pilati, S., Bulleid, N.J., Cabibbo, A., and Sitia, R. The sulfhydryl oxidase Ero1 (sloppy professor) reduces oxygen to hydrogen peroxide to drive oxidation of (2000). J. Biol. Chem. 275, 23685–23692. PDI. Ero1 thereby generates a potential mess of reactive oxygen species as a byproduct of disulfide bond formation in the secretory pathway. PRDX4 (perceptive janitor) was previously identified as an ER- Schulman, S., Wang, B., Li, W., and Rapoport, T.A. localized peroxiredoxin, an enzyme that cleans up hydrogen peroxide by further reducing it to water. (2010). Proc. Natl. Acad. Sci. USA 107, 15027– Such peroxireductase activity by PRDX4 requires a source of electrons. PRDX4 could gain those electrons 15032. by oxidizing PDI and thus contribute to an Ero1-independent pathway for oxidizing secretory proteins— albeit requiring another source of hydrogen peroxide. Tavender, T.J., and Bulleid, N.J. (2010). J. Cell Sci. 123, 2672–2679.

Tavender, T.J., Sheppard, A.M., and Bulleid, N.J. the peroxide to generate a second disul- assumes a central and direct role for (2008). Biochem. J. 411, 191–199. fide. When PRDX4 functions as a backup PDI. Might there be other contributors to Tavender, T.J., Springate, J.J., and Bulleid, N.J. disulfide catalyst in the absence of protein oxidation in the ER that do not op- (2010). EMBO J., in press. Published online Ero1a and Ero1b, however, it has to use erate through PDI? If so, the strategy used November 5, 2010. 10.1038/emboj.2010.273. another source of peroxide. Identifying in this study would miss them. One Wajih, N., Hutson, S.M., and Wallin, R. (2007). J. that source was beyond the scope of the possible mechanism to compensate for Biol. Chem. 282, 2626–2635. current study, leaving a future challenge defects in processing a reducing load on Zito, E., Chin, K.T., Blais, J., Harding, H.P., and for cell biologists. the ER would be to decrease glutathione Ron, D. (2010a). J. Cell Biol. 188, 821–832. Although in some senses the method synthesis or to curtail glutathione import Zito, E., Melo, E.P., Yang, Y., Wahlander, A˚ ., used by Zito et al (2010b) to find alterna- into the ER lumen. Currently, little is Neubert, T.A., and Ron, D. (2010b). Mol. Cell 40, tive ER oxidants was unbiased, it still known about how glutathione import this issue, 787–797.

686 Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell Previews

Noxa: A Sweet Twist to Survival and More

Alfredo Gimenez-Cassina1,2 and Nika N. Danial1,2,* 1Department of Cancer Biology and Division of Metabolism and Chronic Disease, Dana-Farber Cancer Institute, Boston, MA 02115, USA 2Department of Pathology, Harvard Medical School, Boston, MA 02115, USA *Correspondence: [email protected] DOI 10.1016/j.molcel.2010.11.037

The BCL-2 family member Noxa induces apoptosis by antagonizing the prosurvival protein MCL-1. In this issue of Molecular Cell, Lowman et al. (2010) uncover a glucose-dependent phosphoregulatory mechanism that inactivates Noxa’s apoptotic function and triggers its capacity to modulate glucose metabolism.

Mounting evidence in recent years indi- by DNA damage, ischemia, and glucose ling of glucose to the pentose phosphate cates that the homeostatic crosstalk deprivation. Moreover, in light of the pathway (PPP). This idea would be best between nutrient metabolism and apop- significant species-specific differences validated through metabolic flux analysis tosis is governed by shared regulatory in Noxa protein sequence, species- using glucose tracers, which can also components. For example, metabolic by- specific regulatory mechanisms may help determine the exact branch of PPP products of nutrient breakdown can serve modulate Noxa’s function. Lowman et al. activated by Noxa. Furthermore, identifi- as signaling messengers that modulate (2010) provide evidence for posttransla- cation of potential metabolic enzymes survival pathways, and proteins with tional regulation of human Noxa (hNoxa) that may associate with phosphorylated canonical roles in cell death/survival may in leukemia cell lines and primary acti- Noxa in the macromolecular ‘‘survival’’ have additional physiologic functions in vated T cells. Initial clues that a potent complex may shed mechanistic insights nutrient metabolism (Sengupta et al., inhibitor of the atypical cyclin-dependent into Noxa’s metabolic function. 2010; Vander Heiden et al., 2009). These kinase-5 (Cdk5) induces apoptosis in Phosphorylation-dependent modifica- cross-regulatory mechanisms ultimately a Noxa-dependent manner prompted tion of another BH3-only protein, BAD, guarantee that cellular survival, division, the authors to test Noxa as a direct target has also been previously linked to glucose and repair are attuned to cellular meta- of this kinase. In pursuit of both the regu- metabolism (Danial, 2008), albeit different bolic flux and energy charge. The report latory mechanisms and functional conse- from Noxa in multiple aspects. BAD by Lowman et al. in this issue describes quences of this modification, the authors toggles between metabolism and apop- a glucose-sensitive signaling pathway uncovered a link to glucose metabolism tosis through phosphorylation of a that phosphorylates proapoptotic Noxa and signaling. Specifically, they found conserved serine residue within its BH3 and imparts cytoprotective effects by in- that glucose is required for Cdk5 phos- domain that neutralizes its apoptotic activating its apoptotic function and phorylation of hNoxa on Ser13 upstream function while stimulating glucose phos- enhancing glucose utilization (Lowman of its BH3 domain, and this modification phorylation and oxidation (Danial et al., et al., 2010). in turn imparts dual benefits by inactivat- 2008). In endocrine tissues such as The complex and selective networks of ing Noxa’s apoptotic function while trig- pancreas and liver, this is mediated interactions among the pro- and antia- gering a previously unknown effect on through activation of glucokinase (hexoki- poptotic members of the BCL-2 family glucose metabolism. Glucose-mediated nase IV) and plays an important role in are central to the regulation of the mito- control of Noxa is also in line with its systemic glucose homeostasis. BAD’s chondrial pathway of apoptosis (Chipuk selective role in glucose deprivation- effect on glucose phosphorylation is likely et al., 2010). The BH3-only subclass of induced death in leukemia cell lines, sug- not restricted to liver and pancreatic proapoptotic proteins share sequence gesting that changes in Noxa phosphory- b cells (Deng et al., 2008); however, its homology only within the BCL-2 ho- lation in this setting may inform cells precise metabolic role in nonendocrine mology (BH) 3 domain, which serves as of glucose availability and determine cells remains to be determined (Danial, a minimal death domain required for their whether the cells will undergo apoptosis 2008). The different mode of molecular interaction with other BCL-2 family or utilize glucose and survive. Initial engagements and the distinct aspect of members. The apoptotic activity of BH3- biochemical data suggest that Noxa’s glucose metabolism regulated by Noxa only molecules is regulated by tissue- dual functions are dictated by its dynamic and BAD reinforce both the specialization and signal-specific mechanisms. The recruitment to two distinct macromolec- and complexity of networks linking BH3-only protein Noxa was identified as ular complexes, the constituents of which apoptosis and metabolism. a phorbol myristate acetate (PMA) are being identified. The precise aspect of A potential role for Noxa in regulation of response gene in adult T cell leukemia glucose metabolism per se that is quanti- flux through PPP may have physiologic and later discovered as a p53-inducible tatively altered by Noxa modification is yet implications within the context of both gene in response to g irradiation (re- to be determined. However, initial studies activated primary T lymphocytes and viewed in Ploner et al., 2008). Noxa’s rule out changes in glycolysis and point to leukemic cells. Beyond provision of ribose apoptotic function is selectively induced a possible effect on preferential channel- sugars for DNA synthesis, intermediates

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 687 Molecular Cell Previews

of PPP can influence protein glycosylation interactions within these complexes, the REFERENCES or generation of complex oligosaccha- involvement of potential phosphatases rides. These carbohydrate-based modifi- that counteract Noxa phosphorylation, Chipuk, J.E., Moldoveanu, T., Llambi, F., Parsons, M.J., and Green, D.R. (2010). Mol. Cell 37, cations may be especially pertinent to and the nature of nutrient cues that may 299–310. specialized lymphocyte function, in- superimpose on this cross-regulation will cluding migration and homing. In addition, provide important insights into dynamic Danial, N.N. (2008). Oncogene 27 (Suppl 1 ), S53–S70. NADPH production through PPP can be regulation of Noxa’s bifunctional activities used for fatty acid and cholesterol in apoptosis and metabolism. In addition, Danial, N.N., Walensky, L.D., Zhang, C.Y., Choi, C.S., Fisher, J.K., Molina, A.J., Datta, S.R., Pitter, synthesis. Cholesterol-rich lipid rafts are how and if recruitment of MCL-1L to the K.L., Bird, G.H., Wikstrom, J.D., et al. (2008). Nat. implicated in T cell receptor signaling. ‘‘survival’’ complex synergizes with Med. 14, 144–153. NADPH levels can also influence the Noxa’s metabolic role and whether this Deng, H., Yu, F., Chen, J., Zhao, Y., Xiang, J., and pool of reduced glutathione and cellular may have implications for the control of Lin, A. (2008). J. Biol. Chem. 283, 20754–20760. ROS detoxification capacity (Schafer MCL-1’s stability, which is sensitive to et al., 2009). glucose metabolism (Zhao et al., 2007), Lowman, X.H., McDonnell, M.A., Koloske, A., Odu- made, O.A., Jenness, C., Karim, C.B., Jemmerson, One of the intriguing findings in the pose exciting questions for future R., and Kelekar, A. (2010). Mol. Cell 40, this issue, report by Lowman et al. (2010) is the pres- investigation. 823–833. ence of MCL-1L in both of the macromo- Given the critical role of BCL-2 proteins Ploner, C., Kofler, R., and Villunger, A. (2008). lecular complexes that mediate Noxa’s in apoptosis, it is perhaps not surprising Oncogene 27 (Suppl 1 ), S84–S92. functions. The specific interaction mode that a wide spectrum of regulatory mech- Schafer, Z.T., Grassian, A.R., Song, L., Jiang, Z., of Noxa and MCL-1L in the ‘‘apoptotic’’ anisms regulate their function. However, Gerhart-Hines, Z., Irie, H.Y., Gao, S., Puigserver, versus ‘‘survival’’ complexes is likely much remains to be learned about how P., and Brugge, J.S. (2009). Nature 461, 109–113. distinct. While Noxa’s apoptotic activity cellular survival/death decisions receive Sengupta, S., Peterson, T.R., and Sabatini, D.M. involves direct engagement of its BH3 input from other homeostatic pathways. (2010). Mol. Cell 40, 310–322. domain with MCL-1L, the BH3 domain is The findings of Lowman et al. (2010) are not required for Noxa phosphorylation, yet another indication that the expanding Vander Heiden, M.G., Cantley, L.C., and Thomp- son, C.B. (2009). Science 324, 1029–1033. and the interaction of phosphorylated functional interaction networks of BCL-2 Zhao, Y., Altman, B.J., Coloff, J.L., Herman, C.E., Noxa with MCL-1L within the ‘‘survival’’ proteins, which may include cell-death- Jacobs, S.R., Wieman, H.L., Wofford, J.A., Dimas- complex appears to be indirect. The exact and non-cell-death-related partners, cio, L.N., Ilkayeva, O., Kelekar, A., et al. (2007). mechanisms underlying Noxa’s binding hold powerful clues. Mol. Cell. Biol. 27, 4328–4339.

688 Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell Review

The Chromatin Signaling Pathway: Diverse Mechanisms of Recruitment of Histone-Modifying Enzymes and Varied Biological Outcomes

Edwin Smith1 and Ali Shilatifard1,* 1Stowers Institute for Medical Research, 1000 East 50th Street, Kansas City, MO 64110, USA *Correspondence: [email protected] DOI 10.1016/j.molcel.2010.11.031

Posttranslational modifications of histones are coupled in the regulation of the cellular processes involving chromatin, such as transcription, replication, repair, and genome stability. Recent biochemical and genetic studies have clearly demonstrated that many aspects of chromatin, in addition to posttranslational modifica- tions of histones, provide surfaces that can interact with effectors and the modifying machineries in a context-dependent manner, all as a part of the ‘‘chromatin signaling pathway.’’ Here, we have reviewed recent findings on the molecular basis for the recruitment of the chromatin-modifying machineries and their diverse and varied biological outcomes.

Introduction For example, HP1 can bind to the histone H3 tail methylated Although every cell within our body bears the same genetic infor- on lysine 9 as part of the process of stable maintenance of mation and the same set of genes, only a small subset of genes is heterochromatic silencing. Polycomb (Pc) and Eed proteins transcribed in a given cell at a given time. The molecular mech- can bind to histone H3K27-methylated tails at different stages anism underlying this cell-/stage-specific transcriptional control of the cell cycle to help maintain the silencing of developmental has been the subject of intense study for many years. The genes during differentiation (Hansen et al., 2008; Kouzarides, genetic information encoded in our DNA is packaged within 2007; Margueron et al., 2009). However, histone modifications nucleosomal arrays forming what is referred to as chromatin. A are not always epigenetic. The addition and removal of histone nucleosome contains 146 bp of DNA, wrapped twice around acetylation and phosphorylation can be transient events that an octamer composed of two copies of histones H3 and H4 are associated with the initiation and repression of genes. For and two copies of histones H2A and H2B. Initial structural example, histone H2B monoubiquitination is added and then studies using electron microscopy demonstrated that nucleo- quickly removed during the process of gene activation (Henry somes are found in arrays forming a series of ‘‘beads on a string,’’ et al., 2003). Histone H3K4 methylation is associated with coun- with the ‘‘beads’’ being the individual nucleosomes and the teracting Pc silencing of developmental genes when imple- ‘‘string’’ the linker DNA (Kornberg, 1974; Kornberg and Lorch, mented by Trithorax, but is also found on many housekeeping 1999). Histone proteins contain a flexible amino-terminal tail genes, where its function is unknown. Thus, the same modifica- and a characteristic histone fold, a globular domain that medi- tion can have both epigenetic and nonepigenetic functions. ates substantial interactions between histones to form the nucle- Recent studies have demonstrated that the posttranslational osome. High-resolution X-ray crystallography demonstrated that modifications of histones do not represent a code and are no histone N-terminal tails protrude outward from the nucleosome different than the posttranslational modifications associated (Luger et al., 1997), and biochemical studies have confirmed with any other proteins in the cell (Lee et al., 2010; Schreiber that such histone tails are extensively posttranslationally modi- and Bernstein, 2002; Sims and Reinberg, 2008). The posttransla- fied (Bhaumik et al., 2007). tional modifications of histones are part of signaling pathways, Reported posttranslational modifications of histones so far and their readout is context dependent, with the biological include acetylation, phosphorylation, methylation, monoubiquiti- outcomes dictated by many variables. In this review, we analyze nation, sumoylation, and ADP ribosylation (Berger, 2007; Bern- recent reports on the different molecular mechanisms of recruit- stein et al., 2007; Campos and Reinberg, 2009; Kouzarides, ment and biological outcomes for histone-modifying machin- 2007; Shilatifard, 2006; Weake and Workman, 2008; Workman eries. These studies demonstrate that a combination of many and Kingston, 1998). For almost every modification identified, factors, including DNA elements, protein-protein interactions, there are also machineries involved in their removal (Bhaumik stage and origin of cells, and posttranslational modifications on et al., 2007). Histone modifications can have a variety of func- transcription factors and histones, regulate the diverse biological tions; they can change the charge of a residue to disrupt outcome associated within the ‘‘chromatin signaling pathway.’’ protein-DNA, protein-protein, and nucleosome-nucleosome interactions, and they can form binding surfaces for a variety of The Coactivator and Corepressor Models proteins (Campos and Reinberg, 2009; Kouzarides, 2007). of Recruitment Histone modifications that form binding surfaces for other Although it was known for decades that histone acetylation was proteins have been implicated in several epigenetic processes. associated with actively transcribed genes, it was not known if

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 689 Molecular Cell Review

histone acetylation was a consequence of transcription or was involved in the activation process. Purification of a histone ace- tyltransferase from the transcriptionally active macronucleus of the ciliated protozoan Tetrahymena thermophila identified a protein homologous to yeast Gcn5, previously described as a coactivator of transcription (Berger et al., 1992; Brownell et al., 1996; Marcus et al., 1994). Coactivators were postulated to be factors required for bridging sequence-specific binding proteins and the basal transcriptional machinery (Berger et al., 1990; Pugh and Tjian, 1990). Subsequently, other previously known coactivators were shown to be histone acetyltransferases or to participate in complexes with histone acetyltransferases, including CBP/p300, SRC-1, and ACTR (Bannister and Kouzar- ides, 1996; Chen et al., 1997; Ogryzko et al., 1996; Spencer et al., 1997). Both Gcn5, as part of the SAGA histone H3 acetyltransferase complex, and Esa1, as part of the NuA4 histone H4 acetyltransferase complex, were shown to be recruited by certain acidic activators (Utley et al., 1998). Impor- tantly, another Gcn5-containing complex, the ADA complex, did not interact with these activators, demonstrating specificity in the recruitment of different histone-modifying activities by distinct activators. Concurrent with the discovery of a histone acetyltransferase as a transcriptional activator was the discovery of a known tran- scriptional repressor, Rpd3, as a histone deacetylase (Taunton et al., 1996). Rpd3 and related enzymes are found to act as core- pressors with sequence-specific transcriptional binding factors such as nuclear hormone receptors that can recruit SMRT- NCOR histone deacetylase complexes (Alland et al., 1997; Hassig et al., 1997; Heinzel et al., 1997; Kadosh and Struhl, 1997; Nagy et al., 1997). The same repressor can recruit different Figure 1. Coactivators and Corepressors Are Histone-Modifying deacetylase complexes. For example, IKAROS is a critical regu- Enzymes (A) The histone acetyltransferase, Gcn5, as part of the SAGA complex, is lator of hematopoiesis that can interact with both the SIN3A and recruited to genes by a variety of transcriptional activators such as the Mi-2-NuRD deacetylase complexes to repress lineage-specific sequence-specific binding protein Gal4. Acetylation of histone H3 through genes (Kim et al., 1999; Koipally et al., 1999). Together, these recruitment of Gcn5 by Gal4 synergizes with nucleosome remodeling activi- ties, leading to the recruitment of RNA Pol II. studies not only demonstrated a functional role for histone-modi- (B) The histone deacetylase HDAC1 can be recruited to genes as part of the fying activities as key regulators of transcription, but also SMRT-NCOR complexes by sequence-specific DNA-binding proteins such provided the first paradigm of how histone-modifying activities as retinoic acid receptors (RXR/RAR). Deacetylation of histones facilitates could be recruited to chromatin (Figure 1). reassembly of compact chromatin for transcriptional repression.

Recruitment of Histone Modifiers by RNA Polymerase II While histone acetylation and other histone modifications can et al., 2003a; Ng et al., 2003a; Wood et al., 2003). The Paf1 function at the promoter during transcriptional activation, complex is associated with elongating Pol II. It was found in histone modifications can also have roles in the body of genes. a global proteomic analysis in S. cerevisiae (GPS) to be required The trimethylation of histone H3 on lysine 36 is mediated by for proper H3K4 methylation (Krogan et al., 2003a; Wood et al., the Set2 enzyme in yeast and animals, and this modification 2003). Paf1 directly interacts with COMPASS (complex of peaks in the middle and the 30 end of genes associated with proteins associated with Set1), the sole H3K4 methyltransferase the elongating RNA polymerase II (Krogan et al., 2003b; Shilati- in yeast, and Paf1 is required for the recruitment of COMPASS to fard, 2006). Purification of the Set2 enzyme from yeast identified chromatin (Wood et al., 2003). These early studies in yeast set the large subunit of Pol II, Rpb1, as an interactor of Set2 (Krogan the paradigm that the Paf1 complex plays a role as a ‘‘platform’’ et al., 2003b; Xiao et al., 2003). This interaction depends on the for the requirement of histone-modifying machinery to the phosphorylation of Rpb1’s C-terminal repeat domain on serine elongating Pol II (Gerber and Shilatifard, 2003)(Figure 2). We 2, a marker of elongating Pol II. Thus, Set2 is recruited to gene now know that the Drosophila and mammalian Paf1 complexes bodies through an interaction with elongating Pol II (Figure 2). also function as a platform for the recruitment of Set1/TRX/ In contrast to the direct physical interaction between Set2 and MLL-containing complexes in Drosophila and human cells RNA Pol II, other histone-modifying activities can be recruited (Tenney et al., 2006; Wang et al., 2008). subsequently to transcription initiation through the interaction In addition to the recruitment of histone-modifying activities to with the polymerase-associated factor (Paf1) complex (Krogan chromatin, complexes can be recruited to chromatin, but remain

690 Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell Review

Figure 3. Recruitment of Histone-Modifying Activities by Pre-existing Histone Modifications Methylation of histone H3K36 in the body of actively transcribed genes (see Figure 2. Recruitment of Histone-Modifying Activities by RNA Pol II Figure 2) can be bound by the chromodomain-containing protein, Eaf3, as The histone H3K4 methyltransferase, Set1, as part of COMPASS, is recruited part of the Rpd3S deacetylase complex. This process helps to reposition to genes through interaction with the Polymerase-associated factor (Paf1) nucleosomes in the wake of transcribing RNA Pol II. In the absence of complex (1) (Krogan et al., 2003a; Wood et al., 2003). H3K4 trimethylation is Rpd3S, nucleosomes are misplaced, resulting in the exposure of cryptic start found at the start sites of the transcription of active genes, as well as ‘‘poised’’ sites that can lead to the formation of aberrant transcription. Thus, recruitment genes with stalled RNA Pol II (Gilmour, 2009). Upon release of RNA Pol II by of Rpd3S by a histone modification placed by passing RNA polymerase phosphorylation of its C-terminal domain (CTD) at serine 2 (P-Ser2), numerous complexes can maintain transcriptional fidelity. It has recently become factors associate with the phosphorylated C-terminal domain, including apparent that at most genes, Rpd3S can be recruited directly through the the histone H3K36 methyltransferase Set2 (Xiao et al., 2003) (2). Set2 can be interaction with the serine 5 phosphorylated form of the RNA Pol II CTD (Go- recruited through interactions with both the CTD of Pol II and the Paf1 complex vind et al., 2010; Drouin et al., 2010). These findings indicate that the relative (Krogan et al., 2003b; Xiao et al., 2003). contribution of H3K36 methylation and CTD phosphorylation on the recruit- ment of Rpd3S can be gene specific and context dependent. in a relatively inactive state. The Paf1 complex appears to regu- late several activities in this manner. Histone H2B can be mono- transcriptionally active genes that require the chromodomain- ubiquitinated by the Rad6-Bre1 E2/E3 ubiquitin ligase complex. containing protein Eaf3. Eaf3 preferentially binds to H3K36 This monoubiquitination is needed for the higher methylation di- and trimethylated states. While Eaf3 is a component of both states of H3K4 and H3K79, and Paf1 is required for H2B mono- histone acetyltransferase and deacetylase complexes, loss of ubiquitination (Krogan et al., 2003a; Wood et al., 2003). However, Eaf3 leads to an increase in acetylation in the body of genes, both Rad6-Bre1 and Dot1 appear to be recruited independently which is proposed to have the effect of opening up the chromatin of Paf1 (Wood et al., 2003). Paf1 stimulates the enzymatic activity structure to allow ‘‘cryptic’’ transcription (Carrozza et al., 2005; of Rad6-Bre1, perhaps after recruitment of these complexes to Keogh et al., 2005)(Figure 3). chromatin. The subsequent monoubiquitination of H2B stimu- An important issue is the relative contribution of histone modi- lates the activity of Set1/COMPASS and Dot1 to generate the fications in recruitment of other factors to chromatin. Since Eaf3 trimethylation of H3K4 and H3K79 (Dover et al., 2002; Krogan is also a component of the NuA4 histone acetyltransferase et al., 2003a; Wood et al., 2003). Several recent mammalian complex, one might expect both of the Eaf3 complexes to com- studies have also confirmed the generality of these early obser- pete for binding to H3K36-methylated nucleosomes. However, vations made in yeast regarding the role of the Paf1 complex in Rco1, a component of Rpd3S but not of NuA4, interacts with the regulation of histone H2B monoubiquitination by Rad6- nucleosomes (with or without histone modifications) and is Bre1, H3K4 methylation by Set1/COMPASS, and H3K79 methyl- required for the interaction of Rpd3S with nucleosomes methyl- ation by Dot1 (Kim et al., 2009; McGinty et al., 2008; Pavri et al., ated at H3K36, indicating that both histone modification-depen- 2006; Zhu et al., 2005). dent and independent mechanisms are important for targeting Rpd3S to the coding regions of transcribed genes (Li et al., Recruitment by the Preceding Histone Modifications 2007). Recently, Rpd3S has been demonstrated to be recruited Many of the posttranslational modifications of histones can to genes through interactions between the Rco1 subunit and the enhance chromatin binding by other proteins through a variety serine 5 phosphorylated form of the CTD of RNA Pol II. This of protein domains: bromodomains, found in several transcrip- finding suggests that the interaction with H3K36 methylated tional activators, can preferentially bind peptides with acetylated nucleosomes by Eaf3-Rco1 is subsequent to the initial recruit- lysines; 14-3-3 and forkhead domains can bind phosphorylated ment of the complex by the CTD of RNA Pol II (Govind et al., serines and threonines; chromodomains, MBT repeats, and PHD 2010). Supporting these observations, genome-wide profiling fingers can discriminate among lysines that are mono-, di-, or of Rpd3S in a Set2 deletion by Robert and colleagues demon- trimethylated; and tudor domains can recognize methylated strated that H3K36 methylation had a modest effect on the arginine or lysine residues (Maurer-Stroh et al., 2003; Taverna recruitment of Rpd3S to transcribed regions (Drouin et al., 2010). et al., 2007; Yaffe and Elia, 2001). The importance of the interac- Animals have at least two homologs of Eaf3: its ortholog, tion between histone modifications and proteins containing MRG15, and a paralog, MSL3. MRG15 participates in the these modules is exemplified by the occurrence of mutations NuA4-like Tip60 complex as well as the histone deacetylase within these domains in human disease (Matthews et al., 2007; complexes (Kusch et al., 2004; Lee et al., 2009; Moshkin et al., Pen˜ a et al., 2006). Modifications can help form a landing platform 2009; Spain et al., 2010). MSL3 is part of an H4K16-specific for proteins and their complexes to aid in recruitment to chro- histone acetyltransferase complex. MSL3 can interact with matin through recognition of the modified residue. One example H3K36-methylated nucleosomes and mediates acetylation in of this is the recruitment of deacetylase complexes to the body of the ORFs of transcribed genes (Sural et al., 2008). Although

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 691 Molecular Cell Review

a balance of histone acetylation/deacetylation in ORFs is prob- Posterior sex combs (PSC, or Bmi-1 in mammals) and Polyho- ably important for such processes as enhancing transcription meotic (PH). A third complex, the PHO-repressive complex, elongation and maintaining transcription fidelity, how the consists of Pleiohomeotic (PHO, similar to YY1 in mammals) MRG15 and MSL3 complexes are precisely targeted to the and Sfmbt (MBTD1 in mammals), a protein that can recognize promoters or gene bodies likely involves factors other than just H4K20 mono- and dimethylated states. Of the known PcG H3K36me3 binding. Rpd3S has not been extensively character- proteins, PHO, which bears several zinc fingers, is clearly a ized in metazoans, although complexes with MRG15 and histone sequence-specific DNA-binding protein and is required for deacetylases have been reported (Lee et al., 2009; Tominaga recruitment of PRC2 to chromatin in Drosophila. However, et al., 2003; Yochum and Ayer, 2002). Importantly, the metazoan PHO binding is not sufficient for recruitment, and other factors homologs of Rco1, PHF12 in mammals and CG3815 in have been implicated in PRC2 localization in Drosophila. While Drosophila, are found in Rpd3S-like complexes, although the statistically significant enrichment of sequences such as the relative contribution of this subunit to the recruitment of Rpd3S PHO/YY1 consensus can be found in several PREs, a se- by direct interaction with RNA Pol II or indirect interaction quence-based definition of PREs has remained elusive (Ringrose with the H3K36 methylation deposited by Set2 remains to be et al., 2003). determined. One outstanding question is whether PREs exist in mammals. Although PRC2 is conserved from flies to mammals, YY1 Direct Recruitment by DNA Sequences appears to be the only mammalian homolog of the Drosophila In the coactivator model, the DNA-bound transcription factors transcription factors implicated in sequence-specific binding to recruit histone-modifying complexes. However, the most direct elements commonly found within Drosophila PREs. However, way to recruit a histone-modifying activity to particular genes JARID2, a protein with very weak DNA-binding activity, was is through recognition of specific DNA sequences by the recently identified as a component of PRC2 that is required for histone-modifying complexes. This process is perhaps at the recruitment to chromatin. The mechanism by which JARID2 heart of, and the basis for, all recruitment to chromatin and can recruits PRC2 is not known (Landeira et al., 2010; Li et al., be considered the main step in the epigenetic regulation, as 2010; Pasini et al., 2010; Peng et al., 2009; for review, please well (Berger et al., 2009). Here, we review the role of DNA see Herz and Shilatifard, 2010). Unlike other PRC2 components, elements that form target sites for histone-modifying complexes JARID2 shows obvious tissue-specific differences in expression, that extend sequence-dependent recruitment mechanisms being particularly highly expressed in embryonic stem cells. beyond the simple coactivator/corepressor model. Additionally, JARID2 seems to have different effects on different genes, perhaps reflecting the diversity in the PRC2 complexes The PRE in Polycomb-Mediated Repression (Herz and Shilatifard, 2010). A full understanding of the role of The identification of the Polycomb group of genes (PcG) (Ring- JARID2 in the regulation of PRC2 activity and gene expression rose and Paro, 2004) was made by the discovery of mutations will require further detailed and comprehensive biochemical in several genes in Drosophila that led to the ectopic expression and in vivo studies using organisms amenable to genetic of the Hox genes and a transformation of segmental identity. The manipulations. PcG genes are required for the repression of Hox genes in Recently, Kingston and colleagues identified a 1.8 kb region regions where these genes have not been activated. The repres- that conferred PcG responsiveness at the HoxD locus on a sion can be stably maintained for several cell generations after reporter gene (Woo et al., 2010). This region contained YY1 the transcription factors that initially determined cell identity binding sites (Wang et al., 2004) as well as an evolutionarily are no longer present, constituting a form of epigenetic memory conserved sequence within the HoxD locus (Beckers and Du- (Berger et al., 2009). Cloned regions near the Hox genes were boule, 1998). In addition to these two features, this putative shown to confer proper repression of reporter genes in a PcG- mammalian PRE contains a CpG island (Illingworth and Bird, dependent manner, thus being known as Polycomb response 2009). In mammals, PRC2 binding is highly correlated with the elements (PREs). PREs are like enhancers in that they can func- presence of CpG islands, and CpG islands are largely predictive tion from large distances from genes and are sometimes found in of the presence of PRC2, suggesting that CpG islands behave as introns of the genes they regulate. However, unlike canonical PREs in mammalian cells (Ku et al., 2008). CpG islands are enhancer-binding proteins, PcG proteins are expressed ubiqui- regions enriched for CpG dinucleotides that are associated tously and are downstream of the decision of whether a target with promoters of 60%–70% of genes (Illingworth and Bird, gene is to be activated or repressed (Ringrose and Paro, 2004). 2009). Most CpG dinucleotides in the mammalian genome are PcG proteins are found in at least three distinct complexes cytosine methylated, but when they are sufficiently clustered (Schuettengruber et al., 2007; Schwartz and Pirrotta, 2007; together in islands, they are typically unmethylated, except Simon and Kingston, 2009; Wang et al., 2004). Polycomb during transcriptional silencing associated with imprinting, X repressive complex 2 (PRC2) contains the Suppressor of zeste inactivation, or silencing of tumor suppressor genes during 12 (Su[z]12) and Extra sex combs (ESC), as well as the histone oncogenesis. Genes that are silenced during development are methyltransferase Enhancer of zeste (E[z]) (EZH1 and EZH2 in not associated with the CpG methylation of their promoters, mammals) that implements H3K27 methylation. H3K27 methyla- even if associated with CpG islands (Baylin and Bestor, 2002; tion has been shown to form a binding site for the Polycomb Illingworth and Bird, 2009). Therefore, it is likely that PcG func- protein (Pc), a component of the Polycomb repressive complex tion, without DNA methylation, is the major repressor of many 1 (PRC1). PRC1 also includes the E3 ubiquitin ligase RING, developmentally regulated genes, perhaps through recognition

692 Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell Review

Figure 4. Recruitment by DNA Sequences (A–C) In humans, CpG islands are found near the promoters of 60%–70% of genes (Illingworth and Bird, 2009). While cytosines within CpG dinucleo- tides are frequently methylated, in CpG islands, cytosines are largely unmethylated. A number of proteins bearing a zinc finger of the CXXC class preferentially bind unmethylated CpG islands (A). These proteins are associated with a variety of histone-modifying activities, including the H3K4 methyltransferases, MLL and Set1/COMPASS, and the H3K36 demethylase, KDM2A (Clouaire and Stancheva, 2008). MBD1 has two zinc fingers that bind methylated DNA and a third of the CXXC class that binds unmethylated DNA. MBD1 is associated with H3K9 methyltransferase activities and is thought to help coordinate silencing by DNA and histone methylations (Sarraf and Stancheva, 2004). A recent study demonstrated that CpG islands were major sites of recruitment of CXXC1 (Cfp1) and H3K4 methylation via mammalian COMPASS, independent of the transcriptional status of the nearby gene (B) (Thomson et al., 2010). In another recent manuscript, KDM2A, an H3K36 demethylase, was found to be targeted to CpG islands via its CXXC motif (C). Since H3K36 methylation can recruit histone deacetylases (Figure 3), removal of H3K36 methylation in CpG islands could maintain CpG islands in an open, permissive state (Blackledge et al., 2010). of CpG islands; however, PcG proteins can also synergize with tance of the CpG islands in recruiting histone-modifying activi- DNA methylation machineries during silencing associated with ties to chromatin while also raising the question of which other imprinting and X inactivation (Illingworth and Bird, 2009). How features determine the specificity demonstrated for these is PRC2 targeted to CpG islands? CpG-rich sequences can be diverse activities. recognized by proteins bearing a CXXC domain, a type of zinc finger that can recognize CpG sequences. However, CXXC CpG Islands Recruit Multiple Modifiers domains are not found within PRC2 components, suggesting The CXXC domain binding of CpG was originally discovered in that a novel domain recognizes these sequences or that other MBD1, a protein isolated for its homology to the methylated motifs found within these islands help recruit PRC2. CpG DNA-binding protein, MeCP2, but also having three zinc fingers of the CXXC type (Cross et al., 1997). Subsequently, it PRE versus TRE was determined that while two of the CXXC domains within While Hox genes can be silenced with PcG complexes, they can MBD1 bind methylated DNA, many CXXC domains are specific be activated with the help of the trithorax group of proteins (trxG) for unmethylated CpG sequences (Birke et al., 2002; Lee et al., (Eissenberg and Shilatifard, 2010). Trithorax, like its mammalian 2001), including a third CXXC domain within MBD1 (Jørgensen counterpart MLL, is an H3K4 methyltransferase. H3K4- and et al., 2004) and the DNA methyltransferase DNMT1 (Pradhan H3K27-methylated nucleosomes are usually found to occur in et al., 2008). Thus, multiple chromatin modifiers could potentially a mutually exclusive pattern throughout the genome. This can be recruited by CpG-rich sequences (Figure 4A). Two recent perhaps be explained by the competition for the same recruit- studies have tested this idea and found a critical role for CXXC ment sites by the enzyme complexes that implement these domains in recruiting histone-modifying activities to chromatin modifications. Experimental evidence indicates that PREs and (Blackledge et al., 2010; Thomson et al., 2010). Trithorax response elements (TREs) substantially overlap (Ring- In one test of the role of the CXXC domains in targeting CpG rose and Paro, 2007). DNA-binding factors that are important for islands, Bird and colleagues studied the recruitment of CXXC1 recognizing TREs include the GAGA factor (also known as Trl), (Cfp1), a component of mammalian Set1/COMPASS, the major which recognizes GA-rich sequences (Farkas et al., 1994), and H3K4me3 in mammals (Lee and Skalnik, 2005; Miller et al., PHO and Zeste, although other combinations of factors are likely 2001; Thomson et al., 2010; Wu et al., 2008). Bird and colleagues to be important for recruitment (Schwartz et al., 2006). In performed genome-wide profiling of CXXC1 in mouse brain. mammals, the trithorax homologs MLL1 and MLL2 have CXXC They found CXXC1 to be localized at 80% of the CpG islands, domains within their polypeptide, suggesting that the CpG char- 90% of which were enriched for H3K4me3. Half of the CXXC1- acter within a PRE may constitute a distinct sequence that helps negative CpG islands were previously reported to be sites of recruit the competing trxG activity to the PREs/TREs. Interest- Polycomb and H3K27me3 occupancy. Comparing CXXC1 ingly, recent work from two laboratories has shown that the occupancy at Xist, a gene transcribed on only one of two X chro- CXXC domains are involved in targeting distinct types of chro- mosomes in females, CXXC1 was found to associate exclusively matin-modifying activities to chromatin, confirming the impor- with the transcribed, unmethylated CpG copy. Together, these

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 693 Molecular Cell Review

findings suggest that CpG islands could help recruit mammalian differences in the size or complexity of the genome in which Set1/COMPASS through CXXC1’s affinity for unmethylated CpG they are found. sequences (Figure 4B). Another class of histone-modifying enzymes bearing a CXXC To experimentally test the role of CpG islands in recruiting domain is KDM2A/B. KDM2A/B is a histone demethylase that CXXC1, Bird and colleagues created two embryonic stem cell preferentially uses H3K36me2 as a substrate (Tsukada et al., lines carrying a promoterless construct with eGFP and puro- 2006). H3K36me2 is a modification that has previously been mycin sequences that contain CpG dinucleotide densities similar linked to gene silencing (Bender et al., 2006) and can recruit to those found in CpG islands. In one cell line, the artificial CpG histone deacetylases (Li et al., 2009; Youdell et al., 2008), sug- island was targeted to the 30 end of Nanog, while in the other, the gesting that removal of H3K36me2 could facilitate the formation construct was targeted to the 30 end of Mecp2, an X-linked gene. of open chromatin. Klose and colleagues tested the role of the When located adjacent to Nanog, the cassette remained free of CXXC domain in targeting KDM2A. First, they tested the DNA- CpG methylation, despite lacking detectable Pol II. H3K4me3 binding specificity of KDM2A’s CXXC domain and found that it and CXXC1 occupancy tracked CpG density within the cassette preferentially binds to unmethylated CpG sequences (Black- and around the insertion site. Interestingly, the cassette inserted ledge et al., 2010). Genome-wide profiling demonstrated that next to the Mecp2 gene showed two-thirds CpG methylation, but KDM2A was highly enriched at annotated CpG islands. Major still was found to be bound by CXXC1. Bisulfite sequencing of peaks of KDM2A binding not corresponding to CpG islands the CXXC1 and H3K4me3 chromatin immunoprecipitation were shown to have a high CpG content with little DNA methyl- (ChIP) DNA demonstrated that only a third of the immunoprecip- ation, as assessed by bisulfite sequencing. Since most CpGs itated cassette was CpG methylated, strongly suggesting that outside of the CpG islands are unmethylated, Klose and unmethylated CpG sequences are recruitment sites for mamma- colleagues likely found novel CpG islands that were previously lian Set1/COMPASS. However, point mutations within the CXXC unnoticed due to the statistical criteria used in the annotation domain of CXXC1 would be required to demonstrate the direct process. Sites of strong KDM2A binding were also found to be role of this protein in binding CpG and the ensuing regulation depleted for H3K36me2, suggesting that KDM2A localization of H3K4 methylation by mammalian Set1/COMPASS. to CpG islands results in active demethylation of H3K36me2 While Set1/COMPASS is the major H3K4 trimethylase in (Figure 4C). Indeed, knockdown of KDM2A results in increased mammalian cells (Wu et al., 2008), in other studies it has been H3K36me2 at some CpG islands; however, very little alteration demonstrated that the loss of CXXC1 leads to an increase in in transcription was reported when KMD2A levels were reduced global H3K4 trimethylation levels in embryonic stem cells (Tate by RNAi, suggesting that H3K36 dimethylation at CpG islands at et al., 2009). Based on this observation, Skalnik and colleagues the promoters does not have a major transcriptional regulatory have suggested that CXXC1 functions by restricting Set1/ role. KDM2A and the highly related KDM2B have been shown COMPASS methyltransferase activity. In contrast, Bird and to be highly concentrated in nucleoli of cells where they can colleagues find that H3K4me3 levels are reduced at the CpG repress ribosomal RNA transcription (Frescas et al., 2007; islands in the absence of CXXC1. It will be interesting to learn Tanaka et al., 2010). Interestingly, rDNA accounts for 20% of where in the genome H3K4me3 is increasing upon loss of the unmethylated CpG sequences in the mouse genome (Bird CXXC1. et al., 1985), suggesting that KDM2A and KDM2B are targeted An important finding by Bird and colleagues is that H3K4 in part through CpG recognition for rDNA transcription by RNA methylation implemented by mammalian Set1/COMPASS is Pol I. independent of Pol II and transcription, while findings in yeast The studies by the Bird and Klose groups were both in agree- have shown that transcription is required for proper H3K4 trime- ment that recruitment to unmethylated CpG islands via the CXXC thylation (Krogan et al., 2003a; Ng et al., 2003b; Shilatifard, domain was uncorrelated with transcriptional activity, suggest- 2006). One possible explanation for this apparent difference is ing that unmethylated CpG content is sufficient for recruitment. that the interaction of Pol II with CpG islands might be transient Since MBD1 and the DNA methyltransferase DNMT1 also have and not as easily detectable as the product of the process, CXXC fingers that recognize unmethylated CpG, it will be impor- histone H3K4 trimethylation. Therefore, sensitive RNA-seq tant to ask how various CXXC-associated activities compete or methods, such as global run-on sequencing (GRO-seq) (Core coexist with each other on the same site and how this is regu- et al., 2008), could reveal transcription within these CpG islands, lated during development. It would not be surprising if future thus explaining the association of CXXC1 and mammalian studies in this regard will find that the interactions of the CXXC COMPASS at these CpG islands. finger-containing proteins with their target site are context Interestingly, the yeast homolog of CXXC1, Cps40, has a PHD dependent and require other cellular signals for proper function. finger in common, but lacks a CXXC domain. However, the The studies by the Bird and Klose groups should stimulate inves- Drosophila homolog of CXXC1, CG17446, contains both the tigations into the role of CpG islands in transcription and the PHD and CXXC domains. CpG islands have not been studied function of histone-modifying activities in this process. in Drosophila, although their existence has been predicted (Takai and Jones, 2002). It is notable that Drosophila Trithorax, unlike its Same Modification, Different Complexes, mammalian counterpart, MLL, lacks a CXXC domain, further and Distinct Outcomes demonstrating that although these H3K4 methyltransferase One of the best-studied histone modifications is the methylation complexes are largely conserved in composition and function, of H3 on lysine 4 (H3K4). The first methyltransferase complex for some aspects of their recruitment can differ, perhaps reflecting H3K4 to be identified was COMPASS in yeast (Krogan et al.,

694 Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell Review

2002; Miller et al., 2001; Roguev et al., 2001; Shilatifard, 2006). The original interest in purifying COMPASS was based on the similarity between yeast Set1 and the human MLL protein involved in leukemia, and indeed, human MLL forms a COMPASS-like complex with methyltransferase activity for H3K4 (Eissenberg and Shilatifard, 2010; Hughes et al., 2004; Miller et al., 2001; Shilatifard, 2006). While yeast has just one methyltransferase for H3K4, humans have at least six methyl- transferases that can methylate H3K4, including MLL1-4 and Set1A/B, which are found in COMPASS-like complexes (Cho et al., 2007b; Eissenberg and Shilatifard, 2010; Hughes et al., 2004; Lee and Skalnik, 2008; Lee et al., 2007a; Shilatifard, 2008; Wu et al., 2008). COMPASS-like complexes are defined as having a Set1 or MLL-related protein, and the core subunits Cps60/ASH2, Cps30/WDR5, and Cps50/RBBP5. In addition to the common subunits, the Set1 and MLL complexes each have unique subunits. MLL1 and MLL2, which are related to Drosophila Trithorax (TRX), interact with the tumor suppressor Menin, which helps target MLL1/2 to Hox and other Figure 5. The Same Histone Modification Implemented by Different targets for transcription activation (Hughes et al., 2004; Wang Complexes Can Have Different Consequences et al., 2009). MLL3 and MLL4, which show high similarity to (A) Set1/COMPASS is recruited to chromatin of transcriptionally active genes Drosophila Trithorax-related (TRR), contain NCOA6, PTIP, and through interaction with the Paf1 complex (1) (Krogan et al., 2003a), where it is only capable of implementing H3K4 monomethylation. Upon H2B monoubi- PA-1, which may help target MLL3/4 to hormone-responsive quitination (H2Bub) by Rad6/Bre1 (2), WDR82 (Cps35 in yeast) interacts genes, together with UTX, an H3K27 histone demethylase (Cho with chromatin and associates with COMPASS (3), enabling the complex to et al., 2007b; Issaeva et al., 2007; Patel et al., 2007). Two trimethylate H3K4 (4), whose presence at promoters is a mark of actively transcribed genes (see Figure 2). subunits found in the SETD1A and SETD1B complexes, but (B) MLL is part of a COMPASS-like complex that shares many subunits but, not in the MLL1-4 complexes, are CXXC1 and WDR82 (Lee importantly, lacks WDR82, and therefore, its methyltransferase activity is likely and Skalnik, 2008; Lee et al., 2007a; Wu et al., 2008). These to be independent of H2Bub. MLL, like its Drosophila homolog trithorax, is well subunits are homologous to the COMPASS subunits Cps40 known for its role in activating Hox genes. Recruitment of MLL via Menin and sequence-specific binding factors could help activate transcription of its target and Cps35 (Miller et al., 2001). Cps35/WDR82 has a unique loci by mediating trimethylation of H3K4 (1), which can form a binding site for role in mediating the crosstalk between histone H2B ubiquitina- TAF3, a component of TFIID in the preinitiation complexes (PIC) (2), thereby tion and H3K4 methylation (Lee et al., 2007b; Wu et al., 2008; mediating recruitment or stabilization and the basal machinery (PIC) and RNA Pol II (3) (Vermeulen et al., 2007; Wang et al., 2009). Thus, H3K4 trimethy- Zheng et al., 2010). lation implemented by the MLL complexes is instructive for transcription The COMPASS and MLL1-4 complexes can be recruited by (Wang et al., 2009), while H3K4 trimethylation implemented by Set1/ different mechanisms to chromatin, with varying functional COMPASS may be recruited subsequent to initiation of transcription by PIC and RNA Pol II through the H2Bub-WDR82 pathway (Wu et al., 2008; Wang consequences associated with H3K4 methylation. COMPASS et al., 2009). can be recruited to the actively transcribed genes through its interaction with the Paf1 complex and RNA Pol II, which is suffi- cient for H3K4 monomethylation (Gerber and Shilatifard, 2003; activate genes through recruitment of H3K4me3-binding pro- Krogan et al., 2003a; Wood et al., 2003). The Paf1 complex teins (Figure 5B). For example, the basal transcription factor may also be involved in recruiting Rad6-Bre1 through a direct can be directly recruited by H3K4me3 via the Taf3 subunit’s Paf1-Bre1 interaction, leading to ubiquitinated H2B within this PHD finger (Vermeulen et al., 2007). Alternatively, H3K4me3 at region (Wood et al., 2003). Cps35/WDR82 interacts with the promoter could help recruit histone acetyltransferase COMPASS in a histone H2B monoubiquitination-dependent complexes or nucleosome remodeling complexes, which harbor manner, and this interaction on chromatin converts COMPASS subunits with PHD fingers capable of recognizing the H3K4me3 to a trimethylation-competent complex (Lee et al., 2007b; Zheng state (Kouzarides, 2007; Ruthenburg et al., 2007). Thus, the et al., 2010). Deletion of COMPASS subunits or mutation of H3K4 timing and mechanism of recruitment to a gene can influence to alanine has little effect on transcription levels in yeast (Miller the biological readout of H3K4 methylation (Wang et al., 2009). et al., 2001), consistent with the recruitment and stimulation of COMPASS activity following gene activation (Krogan et al., Noncoding RNAs as Recruitment Vehicles 2003a; Ng et al., 2003b)(Figure 5A). for Histone Modifiers In contrast to COMPASS, the MLL complexes can function as In addition to protein and DNA-based recruitment of histone transcriptional activators. Loss of MLL leads to loss of H3K4 modifiers, noncoding RNAs have recently been shown to be methylation and activation of transcription at Hox and other major regulators of chromatin and transcription. While RNAs loci (Hughes et al., 2004; Wang et al., 2009). MLL1-4 complexes are well-known integral components of ribosomes and spliceo- appear to be coactivators of nuclear receptors (Cho et al., 2007a; somes for the translation and splicing of mRNAs, the role of Dreijerink et al., 2009; Eissenberg and Shilatifard, 2010; Goo RNA molecules in transcriptional regulatory complexes has et al., 2003). H3K4 methylation by the MLL complexes could only recently gained widespread notice. One of the first

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 695 Molecular Cell Review

noncoding RNAs associated with transcriptional regulation was G9a, while Kcnq1ot1 is also associated with PRC2 (Nagano the Xist RNA, whose transcription is required for somatic et al., 2008; Pandey et al., 2008). Although Xist, Air, and silencing of an X chromosome in mammalian females, part of Kcnq1ot1 all work to silence genes in cis, it is becoming apparent a process of dosage compensation to equalize expression that noncoding RNAs can silence genes on other chromosomes. from X-linked genes between XY males and XX females (Brock- HOTAIR, a noncoding RNA transcribed from the HOXC cluster, dorff et al., 1992; Brown et al., 1991). Inactivation of the X chro- recruits PRC2 to silence genes in the HOXD cluster (Rinn et al., mosome is associated with accumulation of repressive histone 2007). Separate regions of HOTAIR interact with the PRC2 and marks and the compaction of the X chromosome into the Barr LSD1/CoREST complexes (Tsai et al., 2010). LSD1 is an H3K4 body (Chow and Heard, 2009; Senner and Brockdorff, 2009). demethylase that can demethylate H3K4me2, but not Silencing initiates and spreads in cis from the site of Xist tran- H3K4me3, and associates with the CoREST histone deacetylase scription to encompass most of the X chromosome. The ability complex (Lee et al., 2005; Shi et al., 2004). In addition to a role in of certain forms of Xist RNA to interact with PRC2 complexes developmental regulation, HOTAIR is also implicated in provides a possible molecular explanation for the requirement promoting metastasis. HOTAIR is frequently upregulated in of Xist for H3K27 methylation and gene silencing (Zhao et al., metastatic tumors, and its high expression is associated with 2008). a poor prognosis (Gupta et al., 2010). Importantly, PRC2 compo- Equalization of transcription between males and females of nents are required for the matrix invasiveness of cells ectopically Drosophila also involves noncoding RNAs. roX1 and roX2 are expressing HOTAIR, suggesting that HOTAIR is mistargeting transcribed from the X chromosome in males and are required PRC2 components when overexpressed (Gupta et al., 2010). for the 2-fold increased expression of the male X chromosome The potential regulatory scope of noncoding RNAs is just to equalize transcription with the two X chromosomes from being realized. Genome-wide analyses have identified over females. Unlike Xist, roX RNAs can act in trans (Kelley et al., 1000 ‘‘large intergenic noncoding RNAs’’ (lincRNAs) based on 1999). They form a ribonucleoprotein complex with the MSL signatures of H3K4me3 at transcription start sites and proteins that includes the histone acetyltransferase MOF. MOF H3K36me3 in transcribed regions (Guttman et al., 2009). RNA mediates H4K16 acetylation in the transcribed region of genes, immunoprecipitation with PRC2 antibodies identified 24% of which is proposed to facilitate the decompaction of the X chro- the expressed noncoding RNAs as interactors (Khalil et al., mosome to facilitate higher rates of transcription elongation by 2009). RNAi knockdown of six of these lincRNAs showed signif- Pol II (Smith et al., 2001). The targeting of the roX-MSL complex icant overlap between genes upregulated by the knockdown of to its X-linked targets is comprised of two steps: (1) recognition PRC2 components, but no significant overlap among the six of high-affinity sites for the MSL1-MSL2 subunits and (2) noncoding RNAs (Khalil et al., 2009). One of these, TUG1, is spreading to nearby actively transcribed genes (Alekseyenko induced upon DNA damage in a P53-dependent manner and is et al., 2008). The nature of the high-affinity sites is unknown, required for the repression of a set of genes involved in cell-cycle perhaps consisting of low-affinity sites within a specific context regulation (Guttman et al., 2009; Khalil et al., 2009). Subsequent (Alekseyenko et al., 2008; Fauth et al., 2010; Straub et al., studies aimed at finding P53-induced noncoding RNAs identified 2008). Spreading from high-affinity sites to nearby genes could lincRNA-p21 as a gene required for repression of genes in the be facilitated by the MLE subunit, an RNA helicase whose P53 pathway (Huarte et al., 2010). While lincRNA-p21 is located closest mammalian homolog has been shown to interact with near p21, which is also a repressor in the P53 pathway, their RNA Pol II (Nakajima et al., 1997). MLE’s presence in the gene targets do not significantly overlap. LincRNA-p21 interacts roX-MSL complex is dependent on the integrity of the RNA with PRC2 similarly to many other lincRNAs, but it also interacts component (Ilik and Akhtar, 2009; Smith et al., 2000). Thus, roX with hnRNP-K (Huarte et al., 2010). hnRNP-K, when interacting RNAs could function in MSL spreading by interacting with the with P53, can activate genes. When associating with lincRNA- RNA helicase MLE, which in turn associates with RNA Pol II at p21, hnRNP-K mediates repression of genes; hnRNP-K localiza- active genes, allowing the roX-MSL complex to spread from tion to these targets requires intact lincRNA-p21 (Huarte et al., one active gene to another. The enrichment of the MSL com- 2010). The corepression of genes by hnRNP-K and lincRNA- plexes in the middle and the 30 end of the transcribed genes p21 provides a hint as to why so many different noncoding could be facilitated by interactions between the chromodomain RNAs associate with PRC2 for transcriptional repression. of the MSL3 subunit and H3K36me3 implemented by Pol II-as- Distinct noncoding RNAs could form unique scaffolds for sociated Set2 (Sural et al., 2008). Placing high-affinity binding histone-modifying complexes to associate with gene-specific sites for MSL1-MSL2 on autosomes allows a similar spreading targeting factors. P53 alone is involved in the upregulation of to nearby autosomal genes, which becomes even more pro- approximately 30 noncoding RNAs under a variety of conditions nounced if particular spliced forms of roX RNAs are ectopically and cell types, indicating that the gene regulatory potential for expressed (Park et al., 2005). these large noncoding RNAs is enormous (Huarte et al., 2010). While the roX RNAs are associated with the upregulation of Recent work suggests that short RNA transcripts from the transcription, many noncoding RNAs recruit repressive PRC2 target genes can form short hairpin structures that directly histone-modifying activities similarly to Xist. For example, Air recruit PRC2 (Kanhere et al., 2010), and small RNAs transcribed and Kcnq1ot1 are noncoding RNAs expressed from the paternal from LINE elements on the mammalian X chromosome could chromosome that are required for silencing of neighboring genes play a key role in X inactivation (Chow et al., 2010), indicating in cis (Mancini-Dinardo et al., 2006; Sleutels et al., 2002). Both that both large and small RNAs contribute to these silencing have been shown to associate with the H3K9 methyltransferase, events.

696 Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell Review

Multiple Targeting Modes within a Complex signaling pathway.’’ A better understanding of the multiple For many of the complexes we’ve discussed here, several modes of recruitment of these histone-modifying activities is mechanisms exist for recruiting the same complex. PRC2, for therefore essential for understanding the gene regulatory pro- example, can be recruited by its previously deposited cesses in which they are engaged. The growing number of non- H3K27me3 mark for epigenetic memory, it can be recruited coding RNAs involved in this process provides a challenging yet through DNA sequences such as PREs, and it can be recruited promising avenue for future research. Identification of the factors through interacting with noncoding RNAs. The opposing activity that bind to these RNAs could help us better understand how of PRC2 is mediated in part by MLL, which also displays multiple histone-modifying enzymes are targeted for gene activation or modes of targeting to chromatin. For example, MLL complexes repression. can be recruited through MLL’s CXXC domain to CpG islands (Ayton et al., 2004; Cierpicki et al., 2010). Translocation of the ACKNOWLEDGMENTS MLL gene with a variety of other genes results in chimeric proteins comprised of an N-terminal portion of MLL that includes We thank Laura Shilatifard for editorial assistance. The studies in the Shilatifard laboratory are supported in part by grants from the National Institute of Health: the CXXC domain. Mutations in MLL’s CXXC that prevent DNA R01GM069905, R01CA150265, and R01CA89455. binding caused an increase in DNA methylation at the Hoxa9 locus, reduced Hoxa9 expression, and reduced transformation REFERENCES of bone marrow cells by the MLL-AF9 chimera (Ayton et al., 2004; Bach et al., 2009; Cierpicki et al., 2010). It has also recently Alekseyenko, A.A., Peng, S., Larschan, E., Gorchakov, A.A., Lee, O.K., Kharchenko, P., McGrath, S.D., Wang, C.I., Mardis, E.R., Park, P.J., and been proposed that MLL can be recruited to genes through Kuroda, M.I. (2008). A sequence motif within chromatin entry sites directs a direct interaction between its CXXC domain and the PAF MSL establishment on the Drosophila X chromosome. Cell 134, 599–609. complex. Remarkably, it was found that only MLL1, and not Alland, L., Muhle, R., Hou, H., Jr., Potes, J., Chin, L., Schreiber-Agus, N., and other Set1-related proteins, could interact with the Paf1 complex DePinho, R.A. (1997). Role for N-CoR and histone deacetylase in Sin3- (Milne et al., 2010; Muntean et al., 2010). At this time, it is unclear mediated transcriptional repression. Nature 387, 49–55. how these data can be reconciled with demonstrated functional Ayton, P.M., Chen, E.H., and Cleary, M.L. (2004). Binding to nonmethylated differences for the Set1 and MLL complexes in mammals (Wang CpG DNA is essential for target recognition, transactivation, and myeloid et al., 2009; Wu et al., 2008). transformation by an MLL oncoprotein. Mol. Cell. Biol. 24, 10470–10478. Another domain that may be important for targeting MLL to Bach, C., Mueller, D., Buhl, S., Garcia-Cuellar, M.P., and Slany, R.K. (2009). chromatin is its third PHD finger (PHD3), which specifically Alterations of the CxxC domain preclude oncogenic activation of mixed- lineage leukemia 2. Oncogene 28, 815–823. recognizes H3K4me3 (Chang et al., 2010). MLL also contains AT hooks, which are found in proteins that bind AT-rich se- Bannister, A.J., and Kouzarides, T. (1996). The CBP co-activator is a histone acetyltransferase. Nature 384, 641–643. quences (Reeves and Nissen, 1990), although their requirement for recruitment of MLL to Hox or other loci isn’t known. Aside Baylin, S., and Bestor, T.H. (2002). Altered methylation patterns in cancer cell from domains that bind DNA sequences or histone modifica- genomes: cause or consequence? Cancer Cell 1, 299–305. tions, MLL could also be recruited to genes through the coacti- Beckers, J., and Duboule, D. (1998). Genetic analysis of a conserved sequence vator model due to its interaction with Menin-LEDGF and their in the HoxD complex: regulatory redundancy or limitations of the transgenic approach? Dev. Dyn. 213, 1–11. recruitment by nuclear hormone receptors (Dreijerink et al., 2009; Eissenberg and Shilatifard, 2010). Bender, L.B., Suh, J., Carroll, C.R., Fong, Y., Fingerman, I.M., Briggs, S.D., Cao, R., Zhang, Y., Reinke, V., and Strome, S. (2006). MES-4: an autosome- associated histone methyltransferase that participates in silencing the Summary and Future Perspectives X chromosomes in the C. elegans germ line. Development 133, 3907–3917.

Histone modifications play an indisputably important role in Berger, S.L. (2007). The complex language of chromatin regulation during transcription and other DNA-templated processes. However, transcription. Nature 447, 407–412. the identity of the enzyme and the mechanism of recruitment Berger, S.L., Cress, W.D., Cress, A., Triezenberg, S.J., and Guarente, L. can influence the effect of the modification. This is exemplified (1990). Selective inhibition of activated but not basal transcription by the acidic by the diverse functions of the highly related Set1/COMPASS activation domain of VP16: evidence for transcriptional adaptors. Cell 61, and MLL COMPASS-like complexes, whose biological roles 1199–1208. are reflected in part by their mode of recruitment. Another Berger, S.L., Pin˜ a, B., Silverman, N., Marcus, G.A., Agapite, J., Regier, J.L., example is the differing role of H3S10P at the FOSL1 gene, which Triezenberg, S.J., and Guarente, L. (1992). Genetic isolation of ADA2: a poten- tial transcriptional adaptor required for function of certain acidic activation when mediated early during activation by MSK1 at the promoter, domains. Cell 70, 251–265. or later at the enhancer by PIM1, leads to distinct downstream Berger, S.L., Kouzarides, T., Shiekhattar, R., and Shilatifard, A. (2009). events (Zippo et al., 2009). Furthermore, histone-modifying An operational definition of epigenetics. Genes Dev. 23, 781–783. enzymes can themselves be multifaceted transcription factors, with only one aspect being the ability to modify histones. For Bernstein, B.E., Meissner, A., and Lander, E.S. (2007). The mammalian epigenome. Cell 128, 669–681. example, distinct phenotypes are found for loss of MLL and loss of just the histone-modifying Set domain within MLL, the Bhaumik, S.R., Smith, E., and Shilatifard, A. (2007). Covalent modifications of histones during development and disease pathogenesis. Nat. Struct. Mol. latter of which yields viable offspring (Terranova et al., 2006). Biol. 14, 1008–1016. Thus, rather than simply ‘‘writing or erasing’’ a code that is Bird, A., Taggart, M., Frommer, M., Miller, O.J., and Macleod, D. (1985). waiting to be ‘‘read,’’ histone-modifying activities are integral A fraction of the mouse genome that is derived from islands of nonmethylated, components of gene regulatory networks in a larger ‘‘chromatin CpG-rich DNA. Cell 40, 91–99.

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 697 Molecular Cell Review

Birke, M., Schreiner, S., Garcı´a-Cue´ llar, M.P., Mahr, K., Titgemeyer, F., and Dreijerink, K.M., Lips, C.J., and Timmers, H.T. (2009). Multiple endocrine Slany, R.K. (2002). The MT domain of the proto-oncoprotein MLL binds to neoplasia type 1: a chromatin writer’s block. J. Intern. Med. 266, 53–59. CpG-containing DNA and discriminates against methylation. Nucleic Acids Res. 30, 958–965. Drouin, S., Larame´ e, L., Jacques, P.E´ ., Forest, A., Bergeron, M., and Robert, F. (2010). DSIF and RNA Polymerase II CTD Phosphorylation Coordinate the Blackledge, N.P., Zhou, J.C., Tolstorukov, M.Y., Farcas, A.M., Park, P.J., and Recruitment of Rpd3S to Actively Transcribed Genes. PLoS Genet. 6, Klose, R.J. (2010). CpG islands recruit a histone H3 lysine 36 demethylase. e1001173. Mol. Cell 38, 179–190. Eissenberg, J.C., and Shilatifard, A. (2010). Histone H3 lysine 4 (H3K4) Brockdorff, N., Ashworth, A., Kay, G.F., McCabe, V.M., Norris, D.P., Cooper, methylation in development and differentiation. Dev. Biol. 339, 240–249. P.J., Swift, S., and Rastan, S. (1992). The product of the mouse Xist gene is a 15 kb inactive X-specific transcript containing no conserved ORF and Farkas, G., Gausz, J., Galloni, M., Reuter, G., Gyurkovics, H., and Karch, F. located in the nucleus. Cell 71, 515–526. (1994). The Trithorax-like gene encodes the Drosophila GAGA factor. Nature 371, 806–808. Brown, C.J., Ballabio, A., Rupert, J.L., Lafreniere, R.G., Grompe, M., Tonlorenzi, R., and Willard, H.F. (1991). A gene from the region of the human Fauth, T., Mu¨ ller-Planitz, F., Ko¨ nig, C., Straub, T., and Becker, P.B. (2010). X inactivation centre is expressed exclusively from the inactive X chromo- The DNA binding CXC domain of MSL2 is required for faithful targeting the some. Nature 349, 38–44. Dosage Compensation Complex to the X chromosome. Nucleic Acids Res. 38, 3209–3221. Brownell, J.E., Zhou, J., Ranalli, T., Kobayashi, R., Edmondson, D.G., Roth, S.Y., and Allis, C.D. (1996). Tetrahymena histone acetyltransferase A: Frescas, D., Guardavaccaro, D., Bassermann, F., Koyama-Nasu, R., and a homolog to yeast Gcn5p linking histone acetylation to gene activation. Cell Pagano, M. (2007). JHDM1B/FBXL10 is a nucleolar protein that represses 84, 843–851. transcription of ribosomal RNA genes. Nature 450, 309–313.

Campos, E.I., and Reinberg, D. (2009). Histones: annotating chromatin. Annu. Gerber, M., and Shilatifard, A. (2003). Transcriptional elongation by RNA Rev. Genet. 43, 559–599. polymerase II and histone methylation. J. Biol. Chem. 278, 26303–26306.

Carrozza, M.J., Li, B., Florens, L., Suganuma, T., Swanson, S.K., Lee, K.K., Gilmour, D.S. (2009). Promoter proximal pausing on genes in metazoans. Shia, W.J., Anderson, S., Yates, J., Washburn, M.P., and Workman, J.L. Chromosoma 118, 1–10. (2005). Histone H3 methylation by Set2 directs deacetylation of coding regions Goo, Y.H., Sohn, Y.C., Kim, D.H., Kim, S.W., Kang, M.J., Jung, D.J., Kwak, E., by Rpd3S to suppress spurious intragenic transcription. Cell 123, 581–592. Barlev, N.A., Berger, S.L., Chow, V.T., et al. (2003). Activating signal cointegra- tor 2 belongs to a novel steady-state complex that contains a subset of Chang, P.Y., Hom, R.A., Musselman, C.A., Zhu, L., Kuo, A., Gozani, O., trithorax group proteins. Mol. Cell. Biol. 23, 140–149. Kutateladze, T.G., and Cleary, M.L. (2010). Binding of the MLL PHD3 finger to histone H3K4me3 is required for MLL-dependent gene transcription. Govind, C.K., Qiu, H., Ginsburg, D.S., Ruan, C., Hofmeyer, K., Hu, C., J. Mol. Biol. 400, 137–144. Swaminathan, V., Workman, J.L., Li, B., and Hinnebusch, A.G. (2010). Phosphorylated Pol II CTD recruits multiple HDACs, including Rpd3C(S), for Chen, H., Lin, R.J., Schiltz, R.L., Chakravarti, D., Nash, A., Nagy, L., Privalsky, methylation-dependent deacetylation of ORF nucleosomes. Mol. Cell 39, M.L., Nakatani, Y., and Evans, R.M. (1997). Nuclear receptor coactivator ACTR 234–246. is a novel histone acetyltransferase and forms a multimeric activation complex with P/CAF and CBP/p300. Cell 90, 569–580. Gupta, R.A., Shah, N., Wang, K.C., Kim, J., Horlings, H.M., Wong, D.J., Tsai, M.C., Hung, T., Argani, P., Rinn, J.L., et al. (2010). Long non-coding RNA Cho, Y.W., Hong, T., Hong, S., Guo, H., Yu, H., Kim, D., Guszczynski, T., HOTAIR reprograms chromatin state to promote cancer metastasis. Nature Dressler, G.R., Copeland, T.D., Kalkum, M., and Ge, K. (2007a). PTIP associ- 464, 1071–1076. ates with MLL3- and MLL4-containing histone H3 lysine 4 methyltransferase complex. J. Biol. Chem. 282, 20395–20406. Guttman, M., Amit, I., Garber, M., French, C., Lin, M.F., Feldser, D., Huarte, M., Zuk, O., Carey, B.W., Cassady, J.P., et al. (2009). Chromatin signature reveals Cho, Y.W., Hong, T., Hong, S., Guo, H., Yu, H., Kim, D., Guszczynski, T., over a thousand highly conserved large non-coding RNAs in mammals. Nature Dressler, G.R., Copeland, T.D., Kalkum, M., and Ge, K. (2007b). PTIP associ- 458, 223–227. ates with MLL3- and MLL4-containing histone H3 lysine 4 methyltransferase complex. J. Biol. Chem. 282, 20395–20406. Hansen, K.H., Bracken, A.P., Pasini, D., Dietrich, N., Gehani, S.S., Monrad, A., Rappsilber, J., Lerdrup, M., and Helin, K. (2008). A model for transmission of Chow, J., and Heard, E. (2009). X inactivation and the complexities of silencing the H3K27me3 epigenetic mark. Nat. Cell Biol. 10, 1291–1300. a sex chromosome. Curr. Opin. Cell Biol. 21, 359–366. Hassig, C.A., Fleischer, T.C., Billin, A.N., Schreiber, S.L., and Ayer, D.E. (1997). Chow, J.C., Ciaudo, C., Fazzari, M.J., Mise, N., Servant, N., Glass, J.L., Histone deacetylase activity is required for full transcriptional repression by Attreed, M., Avner, P., Wutz, A., Barillot, E., et al. (2010). LINE-1 activity in mSin3A. Cell 89, 341–347. facultative heterochromatin formation during X chromosome inactivation. Cell 141, 956–969. Heinzel, T., Lavinsky, R.M., Mullen, T.M., So¨ derstrom, M., Laherty, C.D., Torchia, J., Yang, W.M., Brard, G., Ngo, S.D., Davie, J.R., et al. (1997). Cierpicki, T., Risner, L.E., Grembecka, J., Lukasik, S.M., Popovic, R., A complex containing N-CoR, mSin3 and histone deacetylase mediates Omonkowska, M., Shultis, D.D., Zeleznik-Le, N.J., and Bushweller, J.H. transcriptional repression. Nature 387, 43–48. (2010). Structure of the MLL CXXC domain-DNA complex and its functional role in MLL-AF9 leukemia. Nat. Struct. Mol. Biol. 17, 62–68. Henry, K.W., Wyce, A., Lo, W.S., Duggan, L.J., Emre, N.C., Kao, C.F., Pillus, L., Shilatifard, A., Osley, M.A., and Berger, S.L. (2003). Transcriptional activation Clouaire, T., and Stancheva, I. (2008). Methyl-CpG binding proteins: special- via sequential histone H2B ubiquitylation and deubiquitylation, mediated by ized transcriptional repressors or structural components of chromatin? Cell. SAGA-associated Ubp8. Genes Dev. 17, 2648–2663. Mol. Life Sci. 65, 1509–1522. Herz, H.M., and Shilatifard, A. (2010). The JARID2-PRC2 duality. Genes Dev. Core, L.J., Waterfall, J.J., and Lis, J.T. (2008). Nascent RNA sequencing 24, 857–861. reveals widespread pausing and divergent initiation at human promoters. Science 322, 1845–1848. Huarte, M., Guttman, M., Feldser, D., Garber, M., Koziol, M.J., Kenzelmann- Broz, D., Khalil, A.M., Zuk, O., Amit, I., Rabani, M., et al. (2010). A large inter- Cross, S.H., Meehan, R.R., Nan, X., and Bird, A. (1997). A component of the genic noncoding RNA induced by p53 mediates global gene repression in the transcriptional repressor MeCP1 shares a motif with DNA methyltransferase p53 response. Cell 142, 409–419. and HRX proteins. Nat. Genet. 16, 256–259. Hughes, C.M., Rozenblatt-Rosen, O., Milne, T.A., Copeland, T.D., Levine, Dover, J., Schneider, J., Tawiah-Boateng, M.A., Wood, A., Dean, K., Johnston, S.S., Lee, J.C., Hayes, D.N., Shanmugam, K.S., Bhattacharjee, A., Biondi, M., and Shilatifard, A. (2002). Methylation of histone H3 by COMPASS requires C.A., et al. (2004). Menin associates with a trithorax family histone methyltrans- ubiquitination of histone H2B by Rad6. J. Biol. Chem. 277, 28368–28371. ferase complex and with the hoxc8 locus. Mol. Cell 13, 587–597.

698 Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell Review

Ilik, I., and Akhtar, A. (2009). roX RNAs: non-coding regulators of the male X Kusch, T., Florens, L., Macdonald, W.H., Swanson, S.K., Glaser, R.L., Yates, chromosome in flies. RNA Biol. 6, 113–121. J.R., 3rd, Abmayr, S.M., Washburn, M.P., and Workman, J.L. (2004). Acetylation by Tip60 is required for selective histone variant exchange at Illingworth, R.S., and Bird, A.P. (2009). CpG islands—‘a rough guide’. FEBS DNA lesions. Science 306, 2084–2087. Lett. 583, 1713–1720. Landeira, D., Sauer, S., Poot, R., Dvorkina, M., Mazzarella, L., Jørgensen, H.F., Issaeva, I., Zonis, Y., Rozovskaia, T., Orlovsky, K., Croce, C.M., Nakamura, T., Pereira, C.F., Leleu, M., Piccolo, F.M., Spivakov, M., et al. (2010). Jarid2 is Mazo, A., Eisenbach, L., and Canaani, E. (2007). Knockdown of ALR (MLL2) a PRC2 component in embryonic stem cells required for multi-lineage differen- reveals ALR target genes and leads to alterations in cell adhesion and growth. tiation and recruitment of PRC1 and RNA Polymerase II to developmental Mol. Cell. Biol. 27, 1889–1903. regulators. Nat. Cell Biol. 12, 618–624.

Jørgensen, H.F., Ben-Porath, I., and Bird, A.P. (2004). Mbd1 is recruited to Lee, J.H., and Skalnik, D.G. (2005). CpG-binding protein (CXXC finger protein both methylated and nonmethylated CpGs via distinct DNA binding domains. 1) is a component of the mammalian Set1 histone H3-Lys4 methyltransferase Mol. Cell. Biol. 24, 3387–3395. complex, the analogue of the yeast Set1/COMPASS complex. J. Biol. Chem. 280, 41725–41731. Kadosh, D., and Struhl, K. (1997). Repression by Ume6 involves recruitment of a complex containing Sin3 corepressor and Rpd3 histone deacetylase to Lee, J.H., and Skalnik, D.G. (2008). Wdr82 is a C-terminal domain-binding target promoters. Cell 89, 365–371. protein that recruits the Setd1A Histone H3-Lys4 methyltransferase complex to transcription start sites of transcribed human genes. Mol. Cell. Biol. 28, Kanhere, A., Viiri, K., Arau´ jo, C.C., Rasaiyaah, J., Bouwman, R.D., Whyte, 609–618. W.A., Pereira, C.F., Brookes, E., Walker, K., Bell, G.W., et al. (2010). Short RNAs are transcribed from repressed polycomb target genes and interact Lee, J.H., Voo, K.S., and Skalnik, D.G. (2001). Identification and characteriza- with polycomb repressive complex-2. Mol. Cell 38, 675–688. tion of the DNA binding domain of CpG-binding protein. J. Biol. Chem. 276, Kelley, R.L., Meller, V.H., Gordadze, P.R., Roman, G., Davis, R.L., and Kuroda, 44669–44676. M.I. (1999). Epigenetic spreading of the Drosophila dosage compensation complex from roX RNA genes into flanking chromatin. Cell 98, 513–522. Lee, M.G., Wynder, C., Cooch, N., and Shiekhattar, R. (2005). An essential role for CoREST in nucleosomal histone 3 lysine 4 demethylation. Nature Keogh, M.C., Kurdistani, S.K., Morris, S.A., Ahn, S.H., Podolny, V., Collins, 437, 432–435. S.R., Schuldiner, M., Chin, K., Punna, T., Thompson, N.J., et al. (2005). Cotranscriptional set2 methylation of histone H3 lysine 36 recruits a repressive Lee, J.H., Tate, C.M., You, J.S., and Skalnik, D.G. (2007a). Identification and Rpd3 complex. Cell 123, 593–605. characterization of the human Set1B histone H3-Lys4 methyltransferase complex. J. Biol. Chem. 282, 13419–13428. Khalil, A.M., Guttman, M., Huarte, M., Garber, M., Raj, A., Rivea Morales, D., Thomas, K., Presser, A., Bernstein, B.E., van Oudenaarden, A., et al. (2009). Lee, J.S., Shukla, A., Schneider, J., Swanson, S.K., Washburn, M.P., Florens, Many human large intergenic noncoding RNAs associate with chromatin- L., Bhaumik, S.R., and Shilatifard, A. (2007b). Histone crosstalk between H2B modifying complexes and affect gene expression. Proc. Natl. Acad. Sci. monoubiquitination and H3 methylation mediated by COMPASS. Cell 131, USA 106, 11667–11672. 1084–1096.

Kim, J., Sif, S., Jones, B., Jackson, A., Koipally, J., Heller, E., Winandy, S., Viel, Lee, N., Erdjument-Bromage, H., Tempst, P., Jones, R.S., and Zhang, Y. A., Sawyer, A., Ikeda, T., et al. (1999). Ikaros DNA-binding proteins direct (2009). The H3K4 demethylase lid associates with and inhibits histone deace- formation of chromatin remodeling complexes in lymphocytes. Immunity 10, tylase Rpd3. Mol. Cell. Biol. 29, 1401–1410. 345–355. Lee, J.S., Smith, E., and Shilatifard, A. (2010). The language of histone cross- Kim, J., Guermah, M., McGinty, R.K., Lee, J.S., Tang, Z., Milne, T.A., talk. Cell 142, 682–685. Shilatifard, A., Muir, T.W., and Roeder, R.G. (2009). RAD6-Mediated transcrip- tion-coupled H2B ubiquitylation directly stimulates H3K4 methylation in Li, B., Gogol, M., Carey, M., Lee, D., Seidel, C., and Workman, J.L. (2007). human cells. Cell 137, 459–471. Combined action of PHD and chromo domains directs the Rpd3S HDAC to transcribed chromatin. Science 316, 1050–1054. Koipally, J., Renold, A., Kim, J., and Georgopoulos, K. (1999). Repression by Ikaros and Aiolos is mediated through histone deacetylase complexes. Li, B., Jackson, J., Simon, M.D., Fleharty, B., Gogol, M., Seidel, C., Workman, EMBO J. 18, 3090–3100. J.L., and Shilatifard, A. (2009). Histone H3 lysine 36 dimethylation (H3K36me2) is sufficient to recruit the Rpd3s histone deacetylase complex and to repress Kornberg, R.D. (1974). Chromatin structure: a repeating unit of histones and spurious transcription. J. Biol. Chem. 284, 7970–7976. DNA. Science 184, 868–871. Li, G., Margueron, R., Ku, M., Chambon, P., Bernstein, B.E., and Reinberg, D. Kornberg, R.D., and Lorch, Y. (1999). Twenty-five years of the nucleosome, (2010). Jarid2 and PRC2, partners in regulating gene expression. Genes Dev. fundamental particle of the eukaryote chromosome. Cell 98, 285–294. 24, 368–380. Kouzarides, T. (2007). Chromatin modifications and their function. Cell 128, Luger, K., Ma¨ der, A.W., Richmond, R.K., Sargent, D.F., and Richmond, T.J. 693–705. (1997). Crystal structure of the nucleosome core particle at 2.8 A resolution. Krogan, N.J., Dover, J., Khorrami, S., Greenblatt, J.F., Schneider, J., Johnston, Nature 389, 251–260. M., and Shilatifard, A. (2002). COMPASS, a histone H3 (Lysine 4) methyltrans- ferase required for telomeric silencing of gene expression. J. Biol. Chem. 277, Mancini-Dinardo, D., Steele, S.J., Levorse, J.M., Ingram, R.S., and Tilghman, 10753–10755. S.M. (2006). Elongation of the Kcnq1ot1 transcript is required for genomic imprinting of neighboring genes. Genes Dev. 20, 1268–1282. Krogan, N.J., Dover, J., Wood, A., Schneider, J., Heidt, J., Boateng, M.A., Dean, K., Ryan, O.W., Golshani, A., Johnston, M., et al. (2003a). The Paf1 Marcus, G.A., Silverman, N., Berger, S.L., Horiuchi, J., and Guarente, L. (1994). complex is required for histone H3 methylation by COMPASS and Dot1p: link- Functional similarity and physical association between GCN5 and ADA2: ing transcriptional elongation to histone methylation. Mol. Cell 11, 721–729. putative transcriptional adaptors. EMBO J. 13, 4807–4815.

Krogan, N.J., Kim, M., Tong, A., Golshani, A., Cagney, G., Canadien, V., Margueron, R., Justin, N., Ohno, K., Sharpe, M.L., Son, J., Drury, W.J., 3rd, Richards, D.P., Beattie, B.K., Emili, A., Boone, C., et al. (2003b). Methylation Voigt, P., Martin, S.R., Taylor, W.R., De Marco, V., et al. (2009). Role of the of histone H3 by Set2 in Saccharomyces cerevisiae is linked to transcriptional polycomb protein EED in the propagation of repressive histone marks. elongation by RNA polymerase II. Mol. Cell. Biol. 23, 4207–4218. Nature 461, 762–767.

Ku, M., Koche, R.P., Rheinbay, E., Mendenhall, E.M., Endoh, M., Mikkelsen, Matthews, A.G., Kuo, A.J., Ramo´ n-Maiques, S., Han, S., Champagne, K.S., T.S., Presser, A., Nusbaum, C., Xie, X., Chi, A.S., et al. (2008). Genomewide Ivanov, D., Gallardo, M., Carney, D., Cheung, P., Ciccone, D.N., et al. (2007). analysis of PRC1 and PRC2 occupancy identifies two classes of bivalent RAG2 PHD finger couples histone H3 lysine 4 trimethylation with V(D)J domains. PLoS Genet. 4, e1000242. recombination. Nature 450, 1106–1110.

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 699 Molecular Cell Review

Maurer-Stroh, S., Dickens, N.J., Hughes-Davies, L., Kouzarides, T., Peng, J.C., Valouev, A., Swigut, T., Zhang, J., Zhao, Y., Sidow, A., and Eisenhaber, F., and Ponting, C.P. (2003). The Tudor domain ‘Royal Family’: Wysocka, J. (2009). Jarid2/Jumonji coordinates control of PRC2 enzymatic Tudor, plant Agenet, Chromo, PWWP and MBT domains. Trends Biochem. activity and target gene occupancy in pluripotent cells. Cell 139, 1290–1302. Sci. 28, 69–74. Pradhan, M., Este` ve, P.O., Chin, H.G., Samaranayke, M., Kim, G.D., and McGinty, R.K., Kim, J., Chatterjee, C., Roeder, R.G., and Muir, T.W. (2008). Pradhan, S. (2008). CXXC domain of human DNMT1 is essential for enzymatic Chemically ubiquitylated histone H2B stimulates hDot1L-mediated intranu- activity. Biochemistry 47, 10000–10009. cleosomal methylation. Nature 453, 812–816. Pugh, B.F., and Tjian, R. (1990). Mechanism of transcriptional activation by Miller, T., Krogan, N.J., Dover, J., Erdjument-Bromage, H., Tempst, P., Sp1: evidence for coactivators. Cell 61, 1187–1197. Johnston, M., Greenblatt, J.F., and Shilatifard, A. (2001). COMPASS: a complex of proteins associated with a trithorax-related SET domain protein. Reeves, R., and Nissen, M.S. (1990). The A.T-DNA-binding domain of Proc. Natl. Acad. Sci. USA 98, 12902–12907. mammalian high mobility group I chromosomal proteins. A novel peptide motif for recognizing DNA structure. J. Biol. Chem. 265, 8573–8582. Milne, T.A., Kim, J., Wang, G.G., Stadler, S.C., Basrur, V., Whitcomb, S.J., Wang, Z., Ruthenburg, A.J., Elenitoba-Johnson, K.S., Roeder, R.G., and Ringrose, L., and Paro, R. (2004). Epigenetic regulation of cellular memory by Allis, C.D. (2010). Multiple interactions recruit MLL1 and MLL1 fusion proteins the Polycomb and Trithorax group proteins. Annu. Rev. Genet. 38, 413–443. to the HOXA9 locus in leukemogenesis. Mol. Cell 38, 853–863. Ringrose, L., and Paro, R. (2007). Polycomb/Trithorax response elements and Moshkin, Y.M., Kan, T.W., Goodfellow, H., Bezstarosti, K., Maeda, R.K., epigenetic memory of cell identity. Development 134, 223–232. Pilyugin, M., Karch, F., Bray, S.J., Demmers, J.A., and Verrijzer, C.P. (2009). Histone chaperones ASF1 and NAP1 differentially modulate removal of active Ringrose, L., Rehmsmeier, M., Dura, J.M., and Paro, R. (2003). Genome-wide histone marks by LID-RPD3 complexes during NOTCH silencing. Mol. Cell 35, prediction of Polycomb/Trithorax response elements in Drosophila mela- 782–793. nogaster. Dev. Cell 5, 759–771.

Muntean, A.G., Tan, J., Sitwala, K., Huang, Y., Bronstein, J., Connelly, J.A., Rinn, J.L., Kertesz, M., Wang, J.K., Squazzo, S.L., Xu, X., Brugmann, S.A., Basrur, V., Elenitoba-Johnson, K.S., and Hess, J.L. (2010). The PAF complex Goodnough, L.H., Helms, J.A., Farnham, P.J., Segal, E., and Chang, H.Y. synergizes with MLL fusion proteins at HOX loci to promote leukemogenesis. (2007). Functional demarcation of active and silent chromatin domains in Cancer Cell 17, 609–621. human HOX loci by noncoding RNAs. Cell 129, 1311–1323.

Nagano, T., Mitchell, J.A., Sanz, L.A., Pauler, F.M., Ferguson-Smith, A.C., Feil, Roguev, A., Schaft, D., Shevchenko, A., Pijnappel, W.W., Wilm, M., Aasland, R., and Fraser, P. (2008). The Air noncoding RNA epigenetically silences R., and Stewart, A.F. (2001). The Saccharomyces cerevisiae Set1 complex transcription by targeting G9a to chromatin. Science 322, 1717–1720. includes an Ash2 homologue and methylates histone 3 lysine 4. EMBO J. 20, 7137–7148. Nagy, L., Kao, H.Y., Chakravarti, D., Lin, R.J., Hassig, C.A., Ayer, D.E., Schreiber, S.L., and Evans, R.M. (1997). Nuclear receptor repression mediated Ruthenburg, A.J., Allis, C.D., and Wysocka, J. (2007). Methylation of lysine 4 on by a complex containing SMRT, mSin3A, and histone deacetylase. Cell 89, histone H3: intricacy of writing and reading a single epigenetic mark. Mol. Cell 373–380. 25, 15–30.

Nakajima, T., Uchida, C., Anderson, S.F., Lee, C.G., Hurwitz, J., Parvin, J.D., Sarraf, S.A., and Stancheva, I. (2004). Methyl-CpG binding protein MBD1 and Montminy, M. (1997). RNA helicase A mediates association of CBP with couples histone H3 methylation at lysine 9 by SETDB1 to DNA replication RNA polymerase II. Cell 90, 1107–1112. and chromatin assembly. Mol. Cell 15, 595–605.

Ng, H.H., Dole, S., and Struhl, K. (2003a). The Rtf1 component of the Paf1 Schreiber, S.L., and Bernstein, B.E. (2002). Signaling network model of transcriptional elongation complex is required for ubiquitination of histone chromatin. Cell 111, 771–778. H2B. J. Biol. Chem. 278, 33625–33628. Schuettengruber, B., Chourrout, D., Vervoort, M., Leblanc, B., and Cavalli, G. Ng, H.H., Robert, F., Young, R.A., and Struhl, K. (2003b). Targeted recruitment (2007). Genome regulation by polycomb and trithorax proteins. Cell 128, of Set1 histone methylase by elongating Pol II provides a localized mark and 735–745. memory of recent transcriptional activity. Mol. Cell 11, 709–719. Schwartz, Y.B., and Pirrotta, V. (2007). Polycomb silencing mechanisms and Ogryzko, V.V., Schiltz, R.L., Russanova, V., Howard, B.H., and Nakatani, Y. the management of genomic programmes. Nat. Rev. Genet. 8, 9–22. (1996). The transcriptional coactivators p300 and CBP are histone acetyltrans- ferases. Cell 87, 953–959. Schwartz, Y.B., Kahn, T.G., Nix, D.A., Li, X.Y., Bourgon, R., Biggin, M., and Pirrotta, V. (2006). Genome-wide analysis of Polycomb targets in Drosophila Pandey, R.R., Mondal, T., Mohammad, F., Enroth, S., Redrup, L., Komorowski, melanogaster. Nat. Genet. 38, 700–705. J., Nagano, T., Mancini-Dinardo, D., and Kanduri, C. (2008). Kcnq1ot1 anti- sense noncoding RNA mediates lineage-specific transcriptional silencing Senner, C.E., and Brockdorff, N. (2009). Xist gene regulation at the onset of X through chromatin-level regulation. Mol. Cell 32, 232–246. inactivation. Curr. Opin. Genet. Dev. 19, 122–126.

Park, Y., Oh, H., Meller, V.H., and Kuroda, M.I. (2005). Variable splicing of non- Shi, Y., Lan, F., Matson, C., Mulligan, P., Whetstine, J.R., Cole, P.A., Casero, coding roX2 RNAs influences targeting of MSL dosage compensation R.A., and Shi, Y. (2004). Histone demethylation mediated by the nuclear amine complexes in Drosophila. RNA Biol. 2, 157–164. oxidase homolog LSD1. Cell 119, 941–953.

Pasini, D., Cloos, P.A., Walfridsson, J., Olsson, L., Bukowski, J.P., Johansen, Shilatifard, A. (2006). Chromatin modifications by methylation and ubiquitina- J.V., Bak, M., Tommerup, N., Rappsilber, J., and Helin, K. (2010). JARID2 tion: implications in the regulation of gene expression. Annu. Rev. Biochem. regulates binding of the Polycomb repressive complex 2 to target genes in 75, 243–269. ES cells. Nature 464, 306–310. Shilatifard, A. (2008). Molecular implementation and physiological roles for Patel, S.R., Kim, D., Levitan, I., and Dressler, G.R. (2007). The BRCT-domain histone H3 lysine 4 (H3K4) methylation. Curr. Opin. Cell Biol. 20, 341–348. containing protein PTIP links PAX2 to a histone H3, lysine 4 methyltransferase complex. Dev. Cell 13, 580–592. Simon, J.A., and Kingston, R.E. (2009). Mechanisms of polycomb gene silencing: knowns and unknowns. Nat. Rev. Mol. Cell Biol. 10, 697–708. Pavri, R., Zhu, B., Li, G., Trojer, P., Mandal, S., Shilatifard, A., and Reinberg, D. (2006). Histone H2B monoubiquitination functions cooperatively with FACT to Sims, R.J., 3rd, and Reinberg, D. (2008). Is there a code embedded in proteins regulate elongation by RNA polymerase II. Cell 125, 703–717. that is based on post-translational modifications? Nat. Rev. Mol. Cell Biol. 9, 815–820. Pen˜ a, P.V., Davrazou, F., Shi, X., Walter, K.L., Verkhusha, V.V., Gozani, O., Zhao, R., and Kutateladze, T.G. (2006). Molecular mechanism of histone Sleutels, F., Zwart, R., and Barlow, D.P. (2002). The non-coding Air RNA is H3K4me3 recognition by plant homeodomain of ING2. Nature 442, 100–103. required for silencing autosomal imprinted genes. Nature 415, 810–813.

700 Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell Review

Smith, E.R., Pannuti, A., Gu, W., Steurnagel, A., Cook, R.G., Allis, C.D., and Vermeulen, M., Mulder, K.W., Denissov, S., Pijnappel, W.W., van Schaik, F.M., Lucchesi, J.C. (2000). The drosophila MSL complex acetylates histone H4 at Varier, R.A., Baltissen, M.P., Stunnenberg, H.G., Mann, M., and Timmers, H.T. lysine 16, a chromatin modification linked to dosage compensation. Mol. (2007). Selective anchoring of TFIID to nucleosomes by trimethylation of Cell. Biol. 20, 312–318. histone H3 lysine 4. Cell 131, 58–69.

Smith, E.R., Allis, C.D., and Lucchesi, J.C. (2001). Linking global histone Wang, L., Brown, J.L., Cao, R., Zhang, Y., Kassis, J.A., and Jones, R.S. (2004). acetylation to the transcription enhancement of X-chromosomal genes in Hierarchical recruitment of polycomb group silencing complexes. Mol. Cell 14, Drosophila males. J. Biol. Chem. 276, 31483–31486. 637–646.

Spain, M.M., Caruso, J.A., Swaminathan, A., and Pile, L.A. (2010). Drosophila Wang, P., Bowl, M.R., Bender, S., Peng, J., Farber, L., Chen, J., Ali, A., Zhang, SIN3 isoforms interact with distinct proteins and have unique biological Z., Alberts, A.S., Thakker, R.V., et al. (2008). Parafibromin, a component of the functions. J. Biol. Chem. 285, 27457–27467. human PAF complex, regulates growth factors and is required for embryonic development and survival in adult mice. Mol. Cell. Biol. 28, 2930–2940. Spencer, T.E., Jenster, G., Burcin, M.M., Allis, C.D., Zhou, J., Mizzen, C.A., McKenna, N.J., Onate, S.A., Tsai, S.Y., Tsai, M.J., and O’Malley, B.W. Wang, P., Lin, C., Smith, E.R., Guo, H., Sanderson, B.W., Wu, M., Gogol, M., (1997). Steroid receptor coactivator-1 is a histone acetyltransferase. Nature Alexander, T., Seidel, C., Wiedemann, L.M., et al. (2009). Global analysis of 389, 194–198. H3K4 methylation defines MLL family member targets and points to a role for MLL1-mediated H3K4 methylation in the regulation of transcriptional Straub, T., Grimaud, C., Gilfillan, G.D., Mitterweger, A., and Becker, P.B. initiation by RNA polymerase II. Mol. Cell. Biol. 29, 6074–6085. (2008). The chromosomal high-affinity binding sites for the Drosophila dosage compensation complex. PLoS Genet. 4, e1000302. Weake, V.M., and Workman, J.L. (2008). Histone ubiquitination: triggering Sural, T.H., Peng, S., Li, B., Workman, J.L., Park, P.J., and Kuroda, M.I. (2008). gene activity. Mol. Cell 29, 653–663. The MSL3 chromodomain directs a key targeting step for dosage compensa- tion of the Drosophila melanogaster X chromosome. Nat. Struct. Mol. Biol. 15, Woo, C.J., Kharchenko, P.V., Daheron, L., Park, P.J., and Kingston, R.E. 1318–1325. (2010). A region of the human HOXD cluster that confers polycomb-group responsiveness. Cell 140, 99–110. Takai, D., and Jones, P.A. (2002). Comprehensive analysis of CpG islands in human chromosomes 21 and 22. Proc. Natl. Acad. Sci. USA 99, 3740–3745. Wood, A., Schneider, J., Dover, J., Johnston, M., and Shilatifard, A. (2003). The Paf1 complex is essential for histone monoubiquitination by the Rad6-Bre1 Tanaka, Y., Okamoto, K., Teye, K., Umata, T., Yamagiwa, N., Suto, Y., Zhang, complex, which signals for histone methylation by COMPASS and Dot1p. J. Y., and Tsuneoka, M. (2010). JmjC enzyme KDM2A is a regulator of rRNA Biol. Chem. 278, 34739–34742. transcription in response to starvation. EMBO J. 29, 1510–1522. Workman, J.L., and Kingston, R.E. (1998). Alteration of nucleosome structure Tate, C.M., Lee, J.H., and Skalnik, D.G. (2009). CXXC finger protein 1 contains as a mechanism of transcriptional regulation. Annu. Rev. Biochem. 67, redundant functional domains that support embryonic stem cell cytosine 545–579. methylation, histone methylation, and differentiation. Mol. Cell. Biol. 29, 3817–3831. Wu, M., Wang, P.F., Lee, J.S., Martin-Brown, S., Florens, L., Washburn, M., and Shilatifard, A. (2008). Molecular regulation of H3K4 trimethylation by Taunton, J., Hassig, C.A., and Schreiber, S.L. (1996). A mammalian histone Wdr82, a component of human Set1/COMPASS. Mol. Cell. Biol. 28, 7337– deacetylase related to the yeast transcriptional regulator Rpd3p. Science 7344. 272, 408–411. Xiao, T., Hall, H., Kizer, K.O., Shibata, Y., Hall, M.C., Borchers, C.H., and Taverna, S.D., Li, H., Ruthenburg, A.J., Allis, C.D., and Patel, D.J. (2007). How Strahl, B.D. (2003). Phosphorylation of RNA polymerase II CTD regulates H3 chromatin-binding modules interpret histone modifications: lessons from methylation in yeast. Genes Dev. 17, 654–663. professional pocket pickers. Nat. Struct. Mol. Biol. 14, 1025–1040. Yaffe, M.B., and Elia, A.E. (2001). Phosphoserine/threonine-binding domains. Tenney, K., Gerber, M., Ilvarsonn, A., Schneider, J., Gause, M., Dorsett, D., Curr. Opin. Cell Biol. 13, 131–138. Eissenberg, J.C., and Shilatifard, A. (2006). Drosophila Rtf1 functions in histone methylation, gene expression, and Notch signaling. Proc. Natl. Yochum, G.S., and Ayer, D.E. (2002). Role for the mortality factors MORF4, Acad. Sci. USA 103, 11970–11974. MRGX, and MRG15 in transcriptional repression via associations with Pf1, mSin3A, and Transducin-Like Enhancer of Split. Mol. Cell. Biol. 22, 7868– Terranova, R., Agherbi, H., Boned, A., Meresse, S., and Djabali, M. (2006). 7876. Histone and DNA methylation defects at Hox genes in mice expressing a SET domain-truncated form of Mll. Proc. Natl. Acad. Sci. USA 103, 6629– Youdell, M.L., Kizer, K.O., Kisseleva-Romanova, E., Fuchs, S.M., Duro, E., 6634. Strahl, B.D., and Mellor, J. (2008). Roles for Ctk1 and Spt6 in regulating the different methylation states of histone H3 lysine 36. Mol. Cell. Biol. 28, Thomson, J.P., Skene, P.J., Selfridge, J., Clouaire, T., Guy, J., Webb, S., Kerr, 4915–4926. A.R., Deaton, A., Andrews, R., James, K.D., et al. (2010). CpG islands influence chromatin structure via the CpG-binding protein Cfp1. Nature 464, 1082–1086. Zhao, J., Sun, B.K., Erwin, J.A., Song, J.J., and Lee, J.T. (2008). Polycomb Tominaga, K., Leung, J.K., Rookard, P., Echigo, J., Smith, J.R., and Pereira- proteins targeted by a short repeat RNA to the mouse X chromosome. Smith, O.M. (2003). MRGX is a novel transcriptional regulator that exhibits Science 322, 750–756. activation or repression of the B-myb promoter in a cell type-dependent manner. J. Biol. Chem. 278, 49618–49624. Zheng, S., Wyrick, J.J., and Reese, J.C. (2010). Novel trans-tail regulation of H2B ubiquitylation and H3K4 methylation by the N terminus of histone H2A. Tsai, M.C., Manor, O., Wan, Y., Mosammaparast, N., Wang, J.K., Lan, F., Shi, Mol. Cell. Biol. 30, 3635–3645. Y., Segal, E., and Chang, H.Y. (2010). Long noncoding RNA as modular scaffold of histone modification complexes. Science 329, 689–693. Zhu, B., Zheng, Y., Pham, A.D., Mandal, S.S., Erdjument-Bromage, H., Tempst, P., and Reinberg, D. (2005). Monoubiquitination of human histone Tsukada, Y., Fang, J., Erdjument-Bromage, H., Warren, M.E., Borchers, C.H., H2B: the factors involved and their roles in HOX gene regulation. Mol. Cell Tempst, P., and Zhang, Y. (2006). Histone demethylation by a family of JmjC 20, 601–611. domain-containing proteins. Nature 439, 811–816. Zippo, A., Serafini, R., Rocchigiani, M., Pennacchini, S., Krepelova, A., and Utley, R.T., Ikeda, K., Grant, P.A., Coˆ te´ , J., Steger, D.J., Eberharter, A., John, Oliviero, S. (2009). Histone crosstalk between H3S10ph and H4K16ac S., and Workman, J.L. (1998). Transcriptional activators direct histone generates a histone code that mediates transcription elongation. Cell 138, acetyltransferase complexes to nucleosomes. Nature 394, 498–502. 1122–1136.

Molecular Cell 40, December 10, 2010 ª2010 Elsevier Inc. 701 Molecular Cell Article

Epigenetic Instability due to Defective Replication of Structured DNA

Peter Sarkies,1 Charlie Reams,2 Laura J. Simpson,1 and Julian E. Sale1,* 1Medical Research Council Laboratory of Molecular Biology, Hills Road, Cambridge CB2 0QH, UK 2University of Cambridge Computer Laboratory, William Gates Building, 15, J.J. Thomson Avenue, Cambridge CB3 0FD, UK *Correspondence: [email protected] DOI 10.1016/j.molcel.2010.11.009

SUMMARY DNA replication is susceptible to interruptions caused by, for example, DNA damage. In turn, replication arrest interrupts The accurate propagation of histone marks during histone recycling (Jasencakova et al., 2010). The Y family DNA chromosomal replication is proposed to rely on the polymerase REV1 plays an important role in vertebrates in main- tight coupling of replication with the recycling of taining replication fork progression on damaged DNA templates parental histones to the daughter strands. Here, we (Edmunds et al., 2008; Jansen et al., 2009), a role it fulfils by show in the avian cell line DT40 that REV1, a key coordinating other specialized polymerases that are able to regulator of DNA translesion synthesis at the replica- bypass DNA damage directly in a process known as translesion synthesis (Guo et al., 2003; Ross et al., 2005). In REV1-deficient tion fork, is required for the maintenance of repres- cells, this results in damage bypass taking place predominantly sive chromatin marks and gene silencing in the in postreplicative gaps (Edmunds et al., 2008), which form vicinity of DNA capable of forming G-quadruplex when replication restarts downstream of a block. Importantly, (G4) structures. We demonstrate a previously unap- the DNA synthesis associated with the filling of these gaps, which preciated requirement for REV1 in replication of G4 have been estimated to be between about 400 and 3000 bp in forming sequences and show that transplanting length and visualized to persist up to 20 kb behind the fork (Leh- a G4 forming sequence into a silent locus leads to mann, 1972; Lopes et al., 2006), will be uncoupled from bulk DNA its derepression in REV1-deficient cells. Together, replication and therefore from the replicative helicase and histone our observations support a model in which failure recycling. Indeed, in budding yeast, gap filling can be deferred to to maintain processive DNA replication at G4 DNA G2 (Daigaku et al., 2010; Karras and Jentsch, 2010). Thus, chro- in REV1-deficient cells leads to uncoupling of DNA matinization of these regions will be likely to exhibit a bias toward the deposition of newly synthesized histones and therefore result synthesis from histone recycling, resulting in local- in the formation of a tract of nucleosomes lacking key epigenetic ized loss of repressive chromatin through biased marks present in the parental strands. incorporation of newly synthesized histones. We postulated that REV1-deficient cells would be more liable to loss of coordination between the replicative helicase and DNA INTRODUCTION synthesis, and that this might lead to loss of chromatin modifica- tions through the preferential incorporation of new, unmodified Multicellular organisms must maintain gene expression states, histones during gap filling. To test this hypothesis, we have taken and therefore cell identity, epigenetically through cell division advantage of the chicken b-globin locus, in which the histone (Corpet and Almouzni, 2009). It is proposed that this is achieved modifications associated with developmentally regulated through posttranslational modification of the histone proteins expression have been extensively studied (reviewed in Felsen- around which the DNA is wrapped in chromatin. During replica- feld, 1993; Felsenfeld et al., 2004) and the genetic tractability tion, histones are displaced by the replicative helicase and then of the chicken cell line DT40 (Buerstedde and Takeda, 1991). randomly distributed to the nascent daughter DNA strands in We therefore set out to ask initially whether loss of REV1 had a process coordinated by histone chaperones, notably Asf1 any impact on the ability of DT40 to maintain repression of the and Caf1 (reviewed in De Koning et al., 2007). To avoid b-globin locus. a reduction in nucleosome density, recycled histones are combined with newly synthesized histones. The modifications RESULTS on the parental histones can then be copied to the new histones (Bannister et al., 2001; Hansen et al., 2008; Lachner et al., 2001; Loss of H3K9 Dimethylation in the b-Globin Locus Margueron and Reinberg, 2010). In order for this process to be of rev1 Cells Is Associated with an Increase in Marks viable as a mechanism of maintaining gene expression states, of New Histone Deposition it is essential that there be coordination between histone eviction To test the hypothesis that rev1 cells might lose epigenetic infor- by the replicative helicase and the synthesis of new DNA. mation, we used chromatin immunoprecipitation (ChIP) to Without this coordination, parental histones will not be deposited examine the histone modifications at the b-globin locus (Fig- near to their original locations, resulting in loss of the epigenetic ure 1A), which has been previously shown to be silent in noner- information carried by their posttranslational modifications. ythroid cells (Litt et al., 2001a), and specifically in DT40 (Litt et al.,

Molecular Cell 40, 703–713, December 10, 2010 ª2010 Elsevier Inc. 703 Molecular Cell REV1 Maintains Epigenetic Stability at G4 DNA

constitutively Figure 1. Epigenetic Dysfunction in the b-Globin A FR condensed β-globin domain gene chromatin Locus of rev1 DT40 HSA HS4 ρ (A) Map of the region of the chicken b-globin locus studied in this paper (Litt et al., 2001b) between the folate receptor (FR) gene and first of the b-globin genes, r. HSA and HS4 are DNase hypersensitive sites that correspond with 0 5 10 15 20 25 30 35 chromatin domain insulator sequences (Felsenfeld et al., kb 2004). The distance markers and location of the ChIP primers ChIP primers are indicated below the diagram (see also

8.9910.35 13.19 15.85 17.7619.37 21.3722.19 32.00 Table S2). B 6 (B) H3K9 dimethylation (H3K9me2) in the WT (solid line) WT 5 and pcnaK164R (dashed line). In all cases, the specific pcnaK164R ChIP signal was normalized to total H3 then to the signal 4 at HS4 (21.37). Error bars represent the standard error of 3 the mean. 2 (C) H3K9 dimethylation (H3K9me2) in rev1 cells derived in H3K9me2 our laboratory (solid line, ‘‘rev1 Cambridge’’) (Simpson and relative to HS4

Fold enrichment 1 Sale, 2003) and independently the laboratory of Jean- 0 0 5 10 15 20 25 30 35 Marie Buerstedde (dashed line, ‘‘rev1 Munich’’) (Arakawa C 6 kb et al., 2006). 5 (D) Increased levels of acetylation of the N terminus of H4 rev1 Cambridge in rev1 cells. 4 rev1 Munich (E) Acetylation of H3 at K9 and K14 in WT and rev1 cells. 3 (F) Trimethylation of H3 at K4 in WT and rev1 cells. (G) Loss of DNA methylation at the r-globin promoter in H3K9me2 2

relative to HS4 rev1 cells. Loss of DNA methylation renders the r-globin Fold enrichment 1 promoter sensitive to restriction by HpaII. Amplification 0 0 5 10 15 20 25 30 35 of the r-globin promoter by qPCR after HpaII restriction al- kb lowed the fraction of DNA remaining uncleaved, and there- fore methylated, to be determined. Amplification was normalized to BamHI digested genomic DNA, then further D E normalized to set the WT level at 1. The amplified region 1.2 WT 1.2 WT does not contain any BamHI sites. 1.0 rev1 1.0 rev1 Error bars represent the standard deviation. See also Fig- 0.8 0.8 ure S1. 0.6 0.6 0.4 0.4 H3K9/K14ac

0.2 relative to HS4 0.2 Fold enrichment relative to HS4 Fold enrichment Associated with loss of H3K9me2, there was

H4 N-terminal tail ac 0 0

8.98 8.98 enrichment in acetylation of the H4 N-terminal 13.19 21.37 32.00 13.19 21.37 32.00 ChIP primer pair (kb) ChIP primer pair (kb) tail in rev1 compared to WT DT40 (Figure 1D). F G As the increase in H4 N-terminal acetylation 1.2 WT 1.2 was not associated with enrichment of other 1.0 rev1 II 1.0 marks of canonical transcriptional activation, 0.8

Hpa 0.8 H3K9/14ac and H3K4me3, previously observed 0.6 0.6 at this locus (Litt et al., 2001a; Litt et al., 2001b) 0.4 0.4 H3K4me3 protection (Figures 1E and 1F), it is consistent with enrich- Relative relative to HS4 0.2 0.2 Fold enrichment ment of newly synthesized histones in rev1 cells 0 0 WT rev1 relative to the WT (Lande-Diner et al., 2009; 8.98 13.19 21.37 32.00 Sobel et al., 1995). Interestingly, we also ChIP primer pair (kb) observed an increase in H3K56 acetylation (Fig- ure S1 available online), which is a robust marker of newly synthesized H3 in yeast 2001b). Consistent with this, we found an enrichment of the (Li et al., 2008) but whose significance in vertebrates remains repressive H3K9me2 modification in wild-type (WT) DT40 cells a subject of debate. Despite not observing histone marks asso- across the constitutively condensed chromatin domain and at ciated with transcriptional activation, we did observe significant the promoter of the nearby r-globin gene (Figure 1B). Compa- loss of DNA methylation at the r-globin promoter in rev1 cells rable enrichment was evident in a cell line, pcnaK164R, in which (Figure 1G). PCNA cannot be ubiquitinated and which is defective in postre- plicative gap filling (Arakawa et al., 2006; Edmunds et al., 2008) Derepression of the r-Globin Locus in REV1-Deficient (Figure 1B). In contrast, we observed a greatly reduced enrich- Cells ment of H3K9me2 in two independently generated rev1 lines Enrichment of H3K9me2 across the promoter of the r-globin (Arakawa et al., 2006; Simpson and Sale, 2003)(Figure 1C). gene is associated with its silencing (Litt et al., 2001a). Therefore,

704 Molecular Cell 40, 703–713, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell REV1 Maintains Epigenetic Stability at G4 DNA

A 600 Spontaneous DNA Damage Is Unlikely to Explain the Loss of Repressive Histone Marks in REV1-Deficient 500 Cells 400 Taken together, these data are consistent with the hypothesis 300 that gap-filling modes of DNA replication predominate in rev1 200 DT40 cells, leading to the replacement of parental modified 100 histones with newly synthesized histones devoid of repressive mRNA relative to WT

Fold increase in ρ -globin 0 marks. However, the effect appears without the introduction of exogenous DNA damaging agents. We therefore asked whether WT polη rev3 K164Rrev1/ xrcc3 spontaneous DNA damage could be sufficient to lead to this rev1 #162rev1 #217 rev1 Munichpcna phenotype. To address this question theoretically, we developed pcnaK164R a computer model simulating inheritance of histone modifica- tions across cell division through copying of modifications from B 200 parental histones to newly synthesized histones (see the Exper- imental Procedures). We also simulated replication fork stalling, occurring with a defined probability per cell division, that leads to 100 gap-filling DNA synthesis accompanied by a tract of newly synthesized histones with length equal to the length of the gap. We could vary the probability of replication fork stalling and the

mRNA relative to WT length of the gap to examine the effect these parameters might

Fold increase in ρ -globin 0 i ii iii iv v have on epigenetic stability. Our model showed that it was possible to obtain loss of histone modifications with random Figure 2. Derepression of r-Globin Expression in rev1 Cells replication fork stalling using a gap length consistent with that (A) Derepression of r-globin expression in rev1 cells. Comparison of r-globin observed in vivo, but only at a very high frequency of stalling expression in different DT40 mutants. Expression, monitored by qRT-PCR (one stall every 10 kb) (Figure 3A). Even assuming stalling at with primers RhoExpF and R (Table S2), is given as the fold increase over the maximum level possible from estimates of the frequency of the WT level, which is set at 1. rev1 #162 and #217 are two independent spontaneous damage, approximately one lesion per 60 kb, rev1 clones derived in our lab (‘‘rev1 Cambridge’’). Error bars show the range. (B) Effect of complementation with human REV1 on r-globin derepression. (Lindahl, 1996), the mean length of gap would have to be greater Increase in expression of r-globin in rev1 cells, and rev1 cells complemented than 8 kb (approximately 40 nucleosomes) in length to achieve with hREV1, relative to WT: i, rev1 cells cultured for >3 months; ii, rev1 comple- a 40% loss of histone modifications in our model (Figure 3B). mented with hREV1 at 4 weeks, which is as soon as practically possible, and This is considerably greater than the largest current estimate then cultured for >3 months; iii, rev1 cells from i, with established derepression for the length of postreplicative gaps (Lopes et al., 2006). of r-globin, complemented with hREV1 and cultured for 3 weeks; iv,asiii, but We therefore considered the possibility that replication forks cultured for 3.5 weeks; and v,asiii, but cultured for 5 weeks. Five weeks in may stall more frequently at specific sites within the b-globin culture corresponds conservatively with 70 cell divisions. locus. Such sites exist widely in all genomes and are frequently found where the DNA sequence can form secondary structures we predicted that loss of this modification in rev1 cells would lead (reviewed in Mirkin and Mirkin, 2007). Indeed, the chicken to increased expression of the gene. Quantitative PCR revealed r-globin gene has been previously shown to contain a region an approximately 100-fold increase in the expression of the in the second intron in which replication forks are slowed or r-globin gene in rev1 cells relative to WT cells (Figure 2A). Consis- blocked (Prioleau et al., 2003). We therefore simulated the effect tent with the enrichment of H3K9me2, no increase in r-globin of a fixed stall and found that, with a probability of stalling of expression was seen in the pcnaK164R line. Moreover, a mutant 0.4 or above per cell division, a stable tract of lost histone modi- lacking XRCC3, defective in homologous recombination, also fications approximately equal to the length of the gap had devel- showed no increase in r-globin expression. Interestingly, oped after 30 cell divisions (Figure 3C). mutants defective in the translesion polymerases Polh and Polz (REV3) showed only small increases in r-globin expression, G Quadruplex Formation by the r-Globin Second Intron which correlates with their mild phenotypes when assessing Sequence the progression of replication forks on damaged DNA templates We therefore examined the sequence in the second intron of (Edmunds et al., 2008; Jansen et al., 2009). Reintroduction of r-globin and noted, at the site identified by Prioleau et al. a human REV1 complementary DNA (cDNA), which we have (2003), a sequence that corresponds to a consensus for a G qua- previously shown to complement all phenotypes of the chicken druplex (G4) DNA (Figure 4A). Indeed, the region of the chicken rev1 line (Edmunds et al., 2008; Ross et al., 2005), was unable b-globin locus we studied in this work contains three G4 to reverse the r-globin derepression once established over the sequences, the other two residing in the constitutively course of 5 weeks in culture (conservatively 70 cell divisions), condensed region (Figure 4A). G4 DNA, of the general sequence suggesting that the kinetics of restoration of repressive marks G3-5-L1-7-G3-5-L1-7-G3-5-L1-7-G3-5 (where L can be any base), is, at best, slow (Figure 2B). However, a rev1 line in which can form a variety of secondary structures at physiological salt hREV1 had been reintroduced at an early stage but then cultured concentrations, whose stability exceeds that of duplex DNA, for several months did not exhibit r-globin derepression. both in vitro and in vivo (Lipps and Rhodes, 2009; Maizels,

Molecular Cell 40, 703–713, December 10, 2010 ª2010 Elsevier Inc. 705 Molecular Cell REV1 Maintains Epigenetic Stability at G4 DNA

A 40 et al., 1989). In vitro, the 29bp G4 sequence from the r-globin intron forms a K+-dependent quadruplex structure as shown by circular dichroism spectroscopy, with diagnostic positive 30 peaks at 210 nm and 265 nm (Kypr et al., 2009)(Figure 4B). This structure is dependent on four G-rich blocks in the oligonu- 20 cleotide (Figure 4B). Marks lost (%) 10 REV1 Is Required for Efficient Replication of G Quadruplex-Forming DNA on the Leading-Strand

0 Template 0 20 40 60 Translesion synthesis has been previously implicated in the repli- Cell divisions cation of G4 DNA (Be´ tous et al., 2009; Youds et al., 2006). We B therefore asked whether REV1 assists in replication of this

60 specific G4 DNA sequence. To do this, we took advantage of a replicating plasmid assay (Szu¨ ts et al., 2008), which can 50 measure the efficiency of replication as a change in the number 40 of ampicillin resistant colonies recovered normalized to a control kanamycin-resistant plasmid not containing the G4 sequence. 30 We found a striking reduction in the efficiency of replication of 20 the plasmid when the G4 sequence was placed on the leading- strand template, but not on the lagging-strand template (Figures 10 4C and 4D), implying a role for REV1 in the replication of this 0 sequence. This observation is consistent with the presence of 0 Kb / stall for 40% modification loss 0 1000 2000 3000 4000 5000 two strong origins mapped 3 of the r-globin G4, which would Gap Length (bp) place the G4 on the leading-strand template (Prioleau et al., 2003). The colonies that we recovered from transfection of C 8 rev1 cells did not show loss of the G4 sequence. This suggests that other mechanisms are able to compensate for the loss of 6 REV1 but that in rev1 cells these are not able to efficiently coun- Gap length (nucleosomes) teract loss of the plasmid. 4 4 Using this assay, we were able to dissect the contribution of 5 the different domains of REV1 to the replication of the G4 6 2 7 containing plasmid. REV1 has two principal activities. It is a deox- ycytidyl transferase (Nelson et al., 1996) and also has a noncata- in which marks are lost

Length of nucleosome tract lytic function in which its C terminus plays a crucial role in the 0 0.0 0.2 0.4 0.6 0.8 1.0 coordination of TLS by other polymerases (Guo et al., 2003; probability of stall at fixed site Ross et al., 2005). Full-length human REV1 complemented the G4 replication defect completely, as it does other rev1 pheno- Figure 3. Computational Simulation of Loss of Histone Modifica- types (Edmunds et al., 2008; Ross et al., 2005)(Figure 4E). A tions in Response to Formation of Postreplicative Gaps DT40 line harboring a REV1 BRCT domain deletion (Figure S2) (A) Percentage of marks lost as a function of time with a fixed stall probability. also exhibited no defect. However, consistent with our observa- The graph shows a representative time course for histone modification loss. tions on replication fork progression (Edmunds et al., 2008), For this simulation, the postreplicative gap length was set at 1 kb, the proba- REV1 lacking its C-terminal 100 amino acids did not comple- bility of two place copy at 0.25 and the probability of fork stalling at 0.025 per ment, implicating the polymerase-interacting region in the nucleosome. This corresponds to one stall every 8 kb. Each data point repre- sents an average of 30 simulations. replication of G4 DNA. Interestingly, a catalytically dead mutant (B) Postreplicative gap length necessary to produce 40% loss of histone marks complemented only to about 50% of full-length REV1, suggest- in 30 generations. A lower estimate for spontaneous stalling intervals of ing that the catalytic activity also plays a role in G4 DNA replica- 60–100kb would clearly place the necessary postreplicative gap length tion, perhaps by the incorporation of a nontemplated C opposite much higher than any current in vivo estimate (see the main text). The x axis G, leading to disruption of the G4 structure (Figure S3). scale assumes an internucleosome distance of 200 bp. (C) Length of nucleosome tract in which marks are lost as a function of the Loss of Transcriptional Repression in rev1 Cells probability of stalling per replication cycle at a fixed point. Data is shown for Is Associated with G4-Forming DNA a postreplicative gap length of four to seven nucleosomes, with the variance in length set to 0. In the light of the requirement for REV1 in replication of this specific G4 sequence, we speculated that the phenotype observed at the r-globin gene could represent a more general 2006). These structures are characterized by stacks of planar loss of repression in the vicinity of sequences with the potential arrays of four Hoogsteen bonded dG bases coordinated by to form G4 structures. In order to test this, we used a microarray a monovalent metal ion (Sundquist and Klug, 1989; Williamson to identify targets significantly upregulated in rev1 cells relative

706 Molecular Cell 40, 703–713, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell REV1 Maintains Epigenetic Stability at G4 DNA

A C p<0.005

constitutively 25 n=4 FR gene HSA condensed HS4 β-globin locus chromatin ρ 20 r 15 /Kan r 10

Amp 5 n=6 G4 on leading strand template 0 ggggagtaaaagggagcggggtgctgggg WT rev1 B D

7 + 12 with K n=3 n=3 6 without K+ 10 5 r G3 8 4 /Kan 3 r 6 2 4 G4 on lagging strand template Amp 1 2 CD (mdeg) 0 λ (nm) 220 240 260 280 300 320 0 -1 WT rev1 -2 -3

p<0.002

E 30 p<0.007

25 n=4 n=4 n=4

r 20

n=3 Kan /

r 15

n=4 Amp 10

n=4 5 n=6

0

WT rev1 rev1: K164R hREV1 ΔBRCT [1-1137] rev1 pcna [D570AE571A]rev1: hREV1 rev1: hREV1

Figure 4. REV1 Is Required for Replication of G Quadruplex-Forming DNA on the Leading-Strand Template (A) Position of G4 DNAs (ovals) in the region of the b-globin locus studied in this work. The sequence of the r-globin G4 DNA is shown. (B) Circular dichroism spectroscopy of the r-globin G quadruplex forming sequence. Renaturation of the minimal 29 bp G4 oligonucleotide (GQCDG4) in the presence (black line) or absence (light gray line) of K+ ions. Renaturation of a truncated r-globin G4 sequence (GQCDG3) in the presence of K+ ions (mid gray line). (C and D)Replication efficiency, shown as the ratio of Ampr to Kanr E. coli colonies, for the r-globin G4 DNA on the leading- (C) and lagging- (D) strand template of pQ (Szu¨ ts et al., 2008). Error bars represent standard error of the mean. p values were calculated using the unpaired t test (two-tailed). (E) Replication efficiency of the leading-strand template G4 in rev1 mutants. Complementation is with full-length, catalytically inactive (D570AE571A) and C-termi- nally truncated (1-1137) human REV1. The BRCT mutant is an endogenous deletion of amino acids 69–116 of REV1 (Figure S2). The WT and rev1 data from Figure 4C are shown again for comparison. Error bars represent standard error of the mean. See also Figures S2 and S3. to the WT. Having validated the upregulation of four transcripts in ria et al., 2006). As a control, we analyzed a set of genes with simi- two further independent rev1 lines by qPCR (Figure S4), we larly low levels of expression in WT but that were not upregulated selected genes from the array that were expressed at a low level in rev1 cells. At least one G4 sequence within the 3 kb window was in WT cells, but that were increased more than 1.4-fold in rev1 with found in 38% of the control set (Table S1), a similar figure to that a t test p value of <0.075, and examined the sequence 1500 bp obtained in previous analyses of G4 sequences near promoters either side of the annotated transcriptional start site for the pres- in the chicken (Du et al., 2007) and human genomes (Eddy ence of G4-forming sequences, using the Quadfinder server (Sca- and Maizels, 2008; Huppert and Balasubramanian, 2007).

Molecular Cell 40, 703–713, December 10, 2010 ª2010 Elsevier Inc. 707 Molecular Cell REV1 Maintains Epigenetic Stability at G4 DNA

Figure 5. Derepression of Silenced Loci in rev1 A Sa Ex2 C LYSC Ex1 Sp Ex3 Ex4 B Cells Is Associated with G4 DNA (A) Introduction of the r-globin G4 DNA into the LYSC Sthn probe locus. The genomic locus was amplified as a SalI (Sa)- B B ClaI (C) fragment with primers LYSCSalF and LYSCClaR.

loxP Cassette Seln. loxP A linker DNA (G4TplantSph1) containing a BamHI (B) site and the G4 sequence was introduced into the SphI (Sp) site in the first intron of the LYSC gene, so as to be on the feature strand. Correct insertion of the G4 DNA was Ex1 Ex2 Ex3 Ex4 confirmed by sequencing with primer LYSCG4seq. A bidi- 1 kb rectional origin has been demonstrated in the CpG island at the 30 end of the gene (Phi-van and Stra¨ tling, 1999) B C D meaning that the introduced G4 structure will form on 40 the leading-strand template. A puromycin-resistance selection cassette was inserted into the BamHI site.

WT lysc :G4 bp 30 This was then removed by transient expression of Cre re- kbp WT 1000 lysc :G4 combinase. 20 (B) Southern blot of BamHI-digested DNA showing target- >20 ing of one allele of LYSC producing at 5 kb band. 500 5.3 10 (C) Confirmation of the presence of the G4 sequence by LYSC expression PCR using primers (LYSCG4F and R) annealing either 0 side of the expected insertion of the G4 DNA (plus the remnants of the loxP recombination sites). (fold change relative to wild type) WT rev1 (D) qRT-PCR for LYSC expression in clones of WT and +/G4 +/G4 lysc lysc rev1 harboring the r-globin G4 DNA in the LYSC locus expressed as the fold change relative to unmanipulated WT DT40. E E E S LYSC GAS41 (E) Map of the chicken LYSC and GAS41 loci. E, transcrip- tion enhancer element; S, transcription suppressor element; MAR, matrix attachment region (adapted from 5’ MAR 3’ MAR CpG island Myers et al., 2003). The positions of the three pairs of ChIP primers are indicated. 015 015 20 (F) H3K9 dimethylation (H3K9me2) at the LYSC locus kb ChIP primers normalized to that at the constitutively active GAS41 U1 promoter in WT and rev1 cells harboring the r-globin G4 +/G4 prLYSC DNA in one allele of the LYSC locus (lysc ). Error bars prGAS41 represent standard error of the mean. (G) H4 N terminal acetylation at the LYSC locus in WT and +/G4 +/G4 F WT lysc G WT lysc rev1 lysc+/G4 cells, normalized to the GAS41 promoter. rev1lysc+/G4 rev1lysc+/G4 1.6 4 1.2 3 0.8 locus into the developmentally regulated lyso- 2 zyme C gene. LYSC, another well-studied locus H3K9Me2 1 0.4 (Myers et al., 2006) that is silent in chicken Fold enrichment Fold enrichment relative to prGAS41 H4 N-terminal tail Ac relative to prGAS41 lymphocytes, lacks any endogenous G4 0 0 sequence and is unaffected by loss of REV1 in U1 U1 DT40. Like the r-globin gene, it has a strong prLYSC prLYSC prGAS41 prGAS41 origin of replication at its 30 end (Phi-van and ChIP primer pair ChIP primer pair Stra¨ tling, 1999). Using homologous recombina- tion, we inserted the G4-forming sequence and a puromycin resistance selection cassette into Contrastingly, 71% of the targets upregulated in rev1 cells con- the first intron of LYSC approximately 500 bp from the transcrip- tained at least one predicted G4-forming sequence in the 3 kb tional start site, a position equivalent to that seen in the r-globin window. Therefore, there is a statistically significant (p < 0.001) locus (Figures 5A–5C). Having screened for successful inte- association between G4 sequences in the vicinity of the promoter grants in both WT and rev1 backgrounds, we removed the selec- and increased gene expression in rev1 cells. tion cassette by Cre-loxP recombination and cultured the clones for 4 weeks. We detected a marked increase in expression of Introduction of the 29 bp r-Globin G4 DNA into a Silent lysozyme in 7 of 14 clones of rev1 cells carrying the same G4 Locus Confers Susceptibility to Derepression in rev1 DNA integration (Figure 5D). Such an effect was not observed Cells in any of 12 WT clones harboring the same integration of the r- In order to test our hypothesis more directly, we conducted an globin G4 DNA in the LYSC locus (Figure 5D). To examine experiment to transplant the G4 sequence from the r-globin whether this increase in expression was correlated with the

708 Molecular Cell 40, 703–713, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell REV1 Maintains Epigenetic Stability at G4 DNA

recycled Figure 6. A Model for Loss of Repressive histones Histone Marks at Sites with G4-Forming Potential in rev1 Cells Replication is depicted arresting at a G4 DNA on the leading-strand template. Parental histones are shown as light-gray circles, with repressive new histones epigenetic marks represented as gray bars. New histones are shown in black. If REV1 is present, G4 unwinding + REV1 G4 unwinding + PCNA-ubiquitin the fork can replicate through the G4 DNA, main- at the fork or HR-dependent gap filling taining processive DNA synthesis and histone deposition. It is not clear whether the presence of REV1 prevents the formation of the structure or assists in its unwinding (see also Figure S2). In the absence of REV1 the fork remains arrested at the G4 DNA, resulting in a postreplicative gap. The DNA synthesis associated with the resolution of this gap and of the G4 DNA is accompanied by copying of preexisting marks to adjacent nucleosomes new histone incorporation resulting in a tract of chromatin lacking the parental epigenetic marks.

epigenetic changes predicted by our model, we examined the provides further evidence that these sequences pose a challenge H3K9 dimethylation and H4 N-terminal tail acetylation at the to the replicative machinery, even in normal cells. LYSC promoter in a rev1 lysc+/G4 clone exhibiting a >30-fold TLS has been previously implicated in the replication of G/C increase in lysozyme messenger RNA (mRNA). Compared to tracts in both C. elegans and human cell lines. Deletion of either aWTlysc+/G4, the level of H3K9me2 is decreased, while the level Polk or Polh in a dog-1 worm (deficient in the FANCJ helicase) of H4 N-terminal acetylation is increased. Thus, the r-globin G4 resulted in an increased frequency of small deletions in G/C sequence can trigger loss of repression when inserted into tracts (Youds et al., 2006). More recently, RNA interference a silenced locus in rev1, but not WT, cells. (RNAi)-mediated knockdown of Pols h, i, and k has been shown to sensitize cells to the G4-stabilizing compound telomestatin DISCUSSION and to result in elevated DNA damage associated with the human c-MYC promoter, which contains a G-rich sequence capable of forming G4 structures (Be´ tous et al., 2009). In this work, we demonstrate a link between two important facets However, the precise role of TLS in replication of structures of chromosomal replication, the replication of structured DNA that actually contain no damaged DNA remains unclear. In and the faithful maintenance of a repressive chromatin environ- particular, it remains to be shown whether REV1 collaborates ment. Our observations suggest a model (Figure 6) in which with the helicases that have been demonstrated to unwind G4 the absence of REV1 leads to uncoupling of histone recycling DNA, such as FANCJ, BLM, and WRN (Fry and Loeb, 1999; from DNA synthesis at sites capable of forming G4 structures. London et al., 2008; Sun et al., 1998; Wu et al., 2008), and In turn, this results in the repeated loading of newly synthesized whether similar epigenetic instability at G4 DNA is triggered histones that ultimately leads to a permanent loss of repressive by loss of these helicases. It is noteworthy that cells lacking epigenetic marks. either BLM or WRN exhibit altered expression of genes harboring sequences with G4-forming potential (Johnson The role of REV1 in Replication of G4 DNA et al., 2010), although this effect was mechanistically ascribed Sequences capable of forming G4 DNA are abundant throughout to regulation of transcription. the vertebrate genome but are highly enriched at telomeres, the A potential clue to the role of REV1 may come from our obser- immunoglobulin gene switch regions, and the vicinity of the tran- vation that not only the C-terminal polymerase-binding region, scription start site of genes. There is also increasing evidence for but also the catalytic activity of the enzyme is required for fully the formation of such structures in vivo (reviewed in Lipps and effective replication of the r-globin G4 DNA. REV1 is a deoxycy- Rhodes, 2009; Maizels, 2006). G4 DNA can block replicative tidyl transferase (Nelson et al., 1996) but also a template DNA polymerases in vitro (Woodford et al., 1994), and there is G-dependent DNA polymerase (Haracska et al., 2002). This now good evidence, particularly from the study of telomeres, suggests a possible model (Figure S3) in which the ability of that they form and can slow or block replication in vivo (Schaffit- REV1 to produce a tract of dC bases with minimal reference to zel et al., 2001; Sfeir et al., 2009). The exact correlation between the template may destabilize the G4 structure through base the replication slow zone in the r-globin second intron identified pairing between newly synthesized dC and template dG. The by Prioleau et al. (2003) and a robust G4-forming sequence C terminus of the protein could then coordinate the handoff to

Molecular Cell 40, 703–713, December 10, 2010 ª2010 Elsevier Inc. 709 Molecular Cell REV1 Maintains Epigenetic Stability at G4 DNA

other TLS polymerases, allowing extension of this dC-rich primer EXPERIMENTAL PROCEDURES and replication of the G4 sequence. DT40 Strains, Culture, and Transfection Replication Impediments and Epigenetic Stability DT40 cells were propagated and transfected as previously described (Simp- son and Sale, 2003). DT40 mutants used in this work have also been described Recycling of parental histones is likely to play a key role in the previously (Arakawa et al., 2006; Edmunds et al., 2008; Kawamoto et al., 2005; propagation of epigenetic memory but requires tight coupling Ross et al., 2005; Simpson and Sale, 2003; Takata et al., 2001). between histone displacement and redeposition in order that the register between histone marks and underlying DNA Chromatin Immunoprecipitation sequence is not lost. Such coupling is likely to be mediated by Chromatin immunoprecipitation (ChIP) was performed as described (Aparicio et al., 2005) using formaldehyde (FA) crosslinking to trap protein-DNA histone chaperones, notably Asf1 (reviewed in Annunziato, complexes, with minor modifications detailed in the Supplemental Experi- 2005; Groth et al., 2007; Margueron and Reinberg, 2010). mental Procedures. PCR primers for ChIP qPCR are listed in Table S2. Indeed, very recent evidence suggests that Asf1 can buffer histones displaced by hydroxyurea-induced replication arrest, Antibodies for ChIP leading to the suggestion that replication stress may jeopardize The following antibodies were used: anti-H3K9me2, Millipore ChIPAb+ catalog proper chromatin restoration and thereby trigger epigenetic number 17-648; anti-H3K9/K14ac, Millipore ChIPAb+ catalog number 17-615; changes in daughter cells (Jasencakova et al., 2010). Here, we anti-H3K4me3, Cell Signaling Technology catalog number 9727; anti-H3, Cell Signaling Technology catalog number 2650; and anti-acetylH4, Millipore provide evidence that replication impediments can indeed lead ChIPAb+ catalog number 17-630. This polyclonal antibody recognizes acety- to epigenetic change, although in our model it is failure to use lation of H4K5, 8, 12, and 16 and has been previously used (as Upstate catalog recycled histones during gap filling rather than unscheduled number 06-598) to monitor acetylation of H4 during histone deposition (Lande- deposition that underlies a loss of epigenetic information. Diner et al., 2009). A negative control for ChIP was provided by normal rabbit Our computer model predicts that histone mark propagation is IgG (Millipore). likely to be sufficiently robust to deal with such gaps occurring ChIP qPCR and Data Analysis sporadically, as would be caused by DNA damage. At levels of Quantitative PCR was performed in real time with SYBR green. ChIP DNA damage compatible with cell survival, the model suggests that (2.5 ml) was used in each reaction, with 400 nM primer mix and 12.5 ml spreading of histone methylation back into the demethylated 2xSYBR-green qPCR ready-mix (Invitrogen). The reaction was carried out on gap would result in ‘‘healing’’ of the repressive chromatin envi- a ABI Prism real-time cycler with the following program: 50C for 2 min, 90C ronment. However, for G4 DNA, it seems likely that the repeated for 10 min, and 40 cycles of 90C for 15 s (denaturation), 60C for 1 min (anneal- deposition of newly synthesized histones swamps this ability to ing and extension). Each reaction was performed in duplicate. ChIP results restore the pre-existing chromatin environment and ultimately were normalized to the positive control anti-H3 antibody with the formula 2–(Ct(Ab)-Ct(H3)). In control experiments to initially validate the protocol, immuno- leads to derepression. It is also conceivable that spreading of precipitation with the normal rabbit IgG antibody recovered extremely low histone demethylation can occur if the tract of demethylated amounts of material (less than 0.1% of the H3 signal and less than 0.05% of histones is sufficiently long. It seems likely that this spreading input). Data from the b-globin locus was further normalized to the hypersensi- would be limited by chromatin domain insulators, such as those tive site (HS4) of the b-globin locus, which has been previously shown to contain marked by HSA and HS4 (Figure 1A), resulting in switching of the high levels of H4 and H3 N-terminal acetylation (Litt et al., 2001b). This was histone methylation state of the whole domain. Such behavior found to allow reproducible comparison between different extracts. For the LYSC locus, normalization of the specific ChIP signal was to total H3, then to has been proposed on theoretical grounds (Dodd et al., 2007) the promoter of the adjacent, constitutively active GAS41 locus. Absolute and may explain how only two identifiable G4 DNA sequences enrichment relative to total H3 of H3K9me2 and H4 N-terminal acetylation at in the condensed chromatin region can nonetheless lead to the GAS41 promoter was found to be similar in WT and rev1 cells. loss of H3K9 dimethylation across the whole 15 kb domain. It is noteworthy that derepression in the absence of REV1 is qRT-PCR not seen in all silenced loci. Notably, the expression of HOX RNA was extracted with Trizol (Invitrogen) according to the manufacturers’ instructions. cDNA was prepared with 5 mg mRNA with Super RT (HT Biotech- genes, known to be under the control of the polycomb repressive nology, Cambridge, UK) and oligodT primer in a final volume of 40 ml. qPCR complexes, does not appear to be affected by loss of REV1 (data reactions were performed as described above, with 2.5 ml cDNA per reaction. not shown). This may reflect different mechanistic approaches to Quantitation was relative to b-actin (cDNA diluted 1/100) with the exception of the generation and maintenance of particular forms of silencing, LYSC, which was relative to the adjacent GAS41 gene (see Table S2 for primer which may include specific DNA signals, the use of RNA interfer- sequences). The efficiency of amplification was verified to be close to 1 (i.e., a ence, and histone recycling. Conversely, the heterochromatin Ct change of 1 reported a 2-fold change in concentration of template) for the region in the b-globin locus does appear to be affected, even control primers with a standard curve of cDNA dilutions. though it has recently been proposed that RNAi plays a role in DNA Methylation Analysis repression of this sequence (Giles et al., 2010). It will therefore Five micrograms of genomic DNA, quantified by nanodrop spectrophotom- be interesting to explore the relationship between histone eter, was cut with 5 units HpaII, BamHI, or MspI for 6 hr at 37C. After recycling and mechanisms such as RNAi in the initiation and phenol/chloroform extraction, DNA was precipitated overnight at –20C with maintenance of gene repression, possibly using REV1 deficiency sodium acetate and ethanol before analysis with qPCR. HpaII and MspI are as a tool. isoschizomers, but HpaII is blocked by CpG methylation whereas MspI is not; therefore, we could verify the assay by amplifying from the hypersensitive Finally, dysregulation of gene expression is common in many site showing no HpaII enrichment relative to MspI. Data from the r-globin cancers. Our observations suggest one possible mechanism promoter was normalized to BamHI digested DNA to control for differences by which a single mutation in a pathway promoting genetic in genomic DNA preparations (there are no BamHI sites within the expected stability could lead to more widespread epigenetic instability. amplicon).

710 Molecular Cell 40, 703–713, December 10, 2010 ª2010 Elsevier Inc. Molecular Cell REV1 Maintains Epigenetic Stability at G4 DNA

Circular Dichroism Spectroscopy ACKNOWLEDGMENTS Oligonucleotides corresponding to the full-length r-globin G4 and the shorter sequence lacking the first run of Gs (Figure 4A) were synthesized and purified We would like to thank Ian McFarlane and his team at the University of by desaltion (Sigma), resuspended in TE buffer, and diluted to a final concen- Cambridge School of Clinical Medicine Microarray Facility for carrying out tration of 10 mM before heating to 95C for 5 min to denature secondary struc- the microarray hybridization and guidance on analysis of the results, ture. At this point, either 2M KCl was added to a final concentration of 100 mM Guilherme Santos for help with the CD spectroscopy, and Shunichi Takeda or an equivalent volume of nuclease-free H2O was added. The oligos were then and Jean-Marie Buerstedde for sharing DT40 lines. We also thank Daniela left to cool overnight at room temperature. Spectroscopy was performed on Rhodes, Cristina Rada, K.J. Patel, and members of the Sale lab for helpful a Jasco J-810 spectrometer at room temperature with a bandwidth of 2 nm, discussions and critical comments. Work in the laboratory is supported by a response of 1 s, a data pitch of 0.2 nm and a scanning speed of 50 nm/min. the Medical Research Council and Association for International Cancer Scans were performed over the range 200 to 320 nm. Curves for each oligo Research. were processed by subtracting the trace produced by TE buffer (with or without KCl) and smoothing with the software provided by the manufacturer. Received: May 14, 2010 Revised: July 30, 2010 Replicating Plasmid Assay Accepted: September 10, 2010 For creation of the G4-containing replicating plasmid, oligonucleotides coding Published: December 9, 2010 the r-globin G4 sequence (RGG4LeadF and RGG4LeadR) were ligated into pQ1 (Szu¨ ts et al., 2008) as an EcoRI fragment. Sequencing was used to select REFERENCES plasmids where the G4 DNA was on the leading-strand template relative to the Gal origin (Figure 5). A second plasmid placing the G4 on the lagging-strand Annunziato, A.T. (2005). Split decision: what happens to nucleosomes during template was made with oligonucleotides with EcoRI and PstI cohesive DNA replication? J. Biol. Chem. 280, 12065–12068. ends (RGG4LagF and RGG4LagR). Oligonucleotide sequences are given in Aparicio, O., Geisberg, J.V., Sekinger, E., Yang, A., Moqtaderi, Z., and Struhl, Table S2. The plasmid replication assay was performed as described (Szu¨ ts K. (2005). Chromatin immunoprecipitation for determining the association of et al., 2008) with minor modifications: 1 mg control G4-free pQ2 plasmid proteins with specific genomic sequences in vivo. Curr. Protoc. Mol. Biol. (conferring kanamycin resistance) was used per transfection along with 1 mg Chapter 21, Unit 21 23. G4-containing plasmid, (conferring ampicillin resistance); DpnI-digested Arakawa, H., Moldovan, G.L., Saribasak, H., Saribasak, N.N., Jentsch, S., and plasmid was used to transform Invitrogen E-shot electrocompetent cells Buerstedde, J.M. (2006). A role for PCNA ubiquitination in immunoglobulin (catalog number 18290-015); and an equal amount of cells was plated out hypermutation. PLoS Biol. 4, e366. onto kanamycin and ampicillin plates. Bannister, A.J., Zegerman, P., Partridge, J.F., Miska, E.A., Thomas, J.O., Computer Modeling Allshire, R.C., and Kouzarides, T. (2001). Selective recognition of methylated The computer simulation ‘‘Zippee’’ is a Java applet that will run on most web lysine 9 on histone H3 by the HP1 chromo domain. Nature 410, 120–124. browsers with the latest implementations of Java. (We have tested it on Be´ tous, R., Rey, L., Wang, G., Pillaire, M.J., Puget, N., Selves, J., Biard, D.S., Internet Explorer 7, Firefox 3, and Safari 4). It can be found at http://www.cl. Shin-ya, K., Vasquez, K.M., Cazaux, C., and Hoffmann, J.S. (2009). Role of TLS cam.ac.uk/calr3/zippee/. The Java code is available on request. A detailed DNA polymerases h and k in processing naturally occurring structured DNA in description of the algorithm can be found in Supplemental Information. human cells. Mol. Carcinog. 48, 369–378. Buerstedde, J.M., and Takeda, S. (1991). Increased ratio of targeted to Microarray random integration after transfection of chicken B cell lines. Cell 67, 179–188. Microarray analysis was performed on three independent rev1 and WT lines Corpet, A., and Almouzni, G. (2009). Making copies of chromatin: the challenge simultaneously. RNA was extracted with Trizol (Invitrogen). The cDNA labeling of nucleosomal organization and epigenetic information. Trends Cell Biol. 19, and microarray hybridization was carried out by the Cambridge University 29–41. School of Clinical Medicine Department of Metabolic Science with the Affyme- Daigaku, Y., Davies, A.A., and Ulrich, H.D. (2010). Ubiquitin-dependent DNA trix Chicken Genome array. For analysis of the data, targets showing increased damage bypass is separable from genome replication. Nature 465, 951–955. mean probe intensity in rev1 relative to the WT were sorted according to t value. Targets were selected whose initial mean probe intensity value was De Koning, L., Corpet, A., Haber, J.E., and Almouzni, G. (2007). Histone chap- below 7 and whose upregulation was at least 40% and significant at erones: an escort network regulating histone traffic. Nat. Struct. Mol. Biol. 14, p < 0.075. We sorted the list by decreasing change in expression relative to 997–1007. WT and downloaded 1500 kb either side of the transcriptional start site as Dodd, I.B., Micheelsen, M.A., Sneppen, K., and Thon, G. (2007). Theoretical annotated in Ensembl (as at 31.3.2010), discarding targets that either had analysis of epigenetic cell memory by nucleosome modification. Cell 129, significant unsequenced regions or could not be found in the Ensembl data- 813–822. base. For our control set, we took the targets with the highest t value (i.e., Du, Z., Kong, P., Gao, Y., and Li, N. (2007). Enrichment of G4 DNA motif in no change between WT and rev1) and selected for those that were expressed transcriptional regulatory region of chicken genome. Biochem. Biophys. at a level below a mean WT probe intensity level of 7. From this list, targets Res. Commun. 354, 1067–1070. were selected at random and analyzed in the same way as the upregulated Eddy, J., and Maizels, N. (2008). Conserved elements with potential to form set. The sequence data from both sets was then analyzed with the Quadfinder polymorphic G-quadruplex structures in the first intron of human genes. server (http://miracle.igib.res.in/quadfinder/)(Scaria et al., 2006) to search for Nucleic Acids Res. 36, 1321–1333. potential G4 forming motifs on either strand according to the consensus

G(3-5)L(1-7)G(3-5)L(1-7)G(3-5)L(1-7)G(3-5). We used Fisher’s Exact Test to determine Edmunds, C.E., Simpson, L.J., and Sale, J.E. (2008). PCNA ubiquitination and whether there was a significant difference between the numbers of G4 motifs REV1 define temporally distinct mechanisms for controlling translesion found in the upregulated and the control sets. synthesis in the avian cell line DT40. Mol. Cell 30, 519–529. Felsenfeld, G. (1993). Chromatin structure and the expression of SUPPLEMENTAL INFORMATION globin-encoding genes. Gene. 135, 119–124. Felsenfeld, G., Burgess-Beusse, B., Farrell, C., Gaszner, M., Ghirlando, R., Supplemental Information includes Supplemental Experimental Procedures, Huang, S., Jin, C., Litt, M., Magdinier, F., Mutskov, V., et al. (2004). four figures, and two tables and can be found with this article online at Chromatin boundaries and chromatin domains. Cold Spring Harb. Symp. doi:10.1016/j.molcel.2010.11.009. Quant. Biol. 69, 245–250.

Molecular Cell 40, 703–713, December 10, 2010 ª2010 Elsevier Inc. 711 Molecular Cell REV1 Maintains Epigenetic Stability at G4 DNA

Fry, M., and Loeb, L.A. (1999). Human werner syndrome DNA helicase London, T.B., Barber, L.J., Mosedale, G., Kelly, G.P., Balasubramanian, S., unwinds tetrahelical structures of the fragile X syndrome repeat sequence d Hickson, I.D., Boulton, S.J., and Hiom, K. (2008). FANCJ is a structure-specific (CGG)n. J. Biol. Chem. 274, 12797–12802. DNA helicase associated with the maintenance of genomic G/C tracts. J. Biol. Giles, K.E., Ghirlando, R., and Felsenfeld, G. (2010). Maintenance of a consti- Chem. 283, 36132–36139. tutive heterochromatin domain in vertebrates by a Dicer-dependent mecha- Lopes, M., Foiani, M., and Sogo, J.M. (2006). Multiple mechanisms control nism. Nat. Cell. Biol. 12, 94–99, sup pp 91-96. chromosome integrity after replication fork uncoupling and restart at irrepa- Groth, A., Rocha, W., Verreault, A., and Almouzni, G. (2007). Chromatin chal- rable UV lesions. Mol. Cell 21, 15–27. lenges during DNA replication and repair. Cell 128, 721–733. Maizels, N. (2006). Dynamic roles for G4 DNA in the biology of eukaryotic cells. Guo, C., Fischhaber, P.L., Luk-Paszyc, M.J., Masuda, Y., Zhou, J., Kamiya, K., Nat. Struct. Mol. Biol. 13, 1055–1059. Kisker, C., and Friedberg, E.C. (2003). Mouse Rev1 protein interacts with Margueron, R., and Reinberg, D. (2010). Chromatin structure and the inheri- multiple DNA polymerases involved in translesion DNA synthesis. EMBO J. tance of epigenetic information. Nat. Rev. Genet. 11, 285–296. 22, 6621–6630. Mirkin, E.V., and Mirkin, S.M. (2007). Replication fork stalling at natural imped- Hansen, K.H., Bracken, A.P., Pasini, D., Dietrich, N., Gehani, S.S., Monrad, A., iments. Microbiol. Mol. Biol. Rev. 71, 13–35. Rappsilber, J., Lerdrup, M., and Helin, K. (2008). A model for transmission of the H3K27me3 epigenetic mark. Nat. Cell Biol. 10, 1291–1300. Myers, F.A., Chong, W., Evans, D.R., Thorne, A.W., and Crane-Robinson, C. (2003). Acetylation of histone H2B mirrors that of H4 and H3 at the chicken Haracska, L., Prakash, S., and Prakash, L. (2002). Yeast Rev1 protein is a G b-globin locus but not at housekeeping genes. J. Biol. Chem. 278, 36315– template-specific DNA polymerase. J. Biol. Chem. 277, 15546–15551. 36322. Huppert, J.L., and Balasubramanian, S. (2007). G-quadruplexes in promoters Myers, F.A., Lefevre, P., Mantouvalou, E., Bruce, K., Lacroix, C., Bonifer, C., throughout the human genome. Nucleic Acids Res. 35, 406–413. Thorne, A.W., and Crane-Robinson, C. (2006). Developmental activation of Jansen, J.G., Tsaalbi-Shtylik, A., Hendriks, G., Gali, H., Hendel, A., Johansson, the lysozyme gene in chicken macrophage cells is linked to core histone acet- F., Erixon, K., Livneh, Z., Mullenders, L.H., Haracska, L., and de Wind, N. ylation at its enhancer elements. Nucleic Acids Res. 34, 4025–4035. (2009). Separate domains of Rev1 mediate two modes of DNA damage bypass Nelson, J.R., Lawrence, C.W., and Hinkle, D.C. (1996). Deoxycytidyl trans- in mammalian cells. Mol. Cell. Biol. 29, 3113–3123. ferase activity of yeast REV1 protein. Nature 382, 729–731. Jasencakova, Z., Scharf, A.N., Ask, K., Corpet, A., Imhof, A., Almouzni, G., and Phi-van, L., and Stra¨ tling, W.H. (1999). An origin of bidirectional DNA replica- Groth, A. (2010). Replication stress interferes with histone recycling and prede- tion is located within a CpG island at the 300 end of the chicken lysozyme position marking of new histones. Mol. Cell 37, 736–743. gene. Nucleic Acids Res. 27, 3009–3017. Johnson, J.E., Cao, K., Ryvkin, P., Wang, L.S.,