View metadata, citation and similar papers at core.ac.uk brought to you by CORE

provided by Elsevier - Publisher Connector

compounds targeting riboswitches Fuchs, R.T., Grundy, F.J., and Henkin, T.M. Nudler, E., and Mironov, A.S. (2004). Trends already exist, such as the TPP analog (2006). Nat. Struct. Mol. Biol. 13, 226–233. Biochem. Sci. 29, 11–17. pyrithiamine (Kubodera et al., 2003); Kubodera, T., Watanabe, M., Yoshiuchi, K., Serganov, A., Yuan, Y.-R., Pikovskaya, O., however, they are toxic to animals for Yamashita, N., Nishimura, A., Nakai, S., Gomi, Polonskaia, A., Malinina, L., Phan, A.T., Ho- reasons unrelated to riboswitches. K., and Hanamoto, H. (2003). FEBS Lett. 555, bartner, C., Micura, R., Breaker, R.R., and 516–520. Patel, D.J. (2004). Chem. Biol. 11, 1729– The new structures provide clear 1741. guidelines on how to design more Mandal, M., and Breaker, R.R. (2004). Nat. Rev. Mol. Cell Biol. 5, 451–463. Serganov, A., Polonskaia, A., Phan, A.T., specific and safer riboswitch-target- Breaker, R.R., and Patel, D.J. (2006). Nature, ing drugs. Mironov, A.S., Gusarov, I., Rafikov, R., Lo- in press. pez, L.E., Shatalin, K., Kreneva, R.A., Peru- mov, D.A., and Nudler, E. (2002). Cell 111, Thore, S., Leibundgut, M., and Ban, N. (2006). References 747–756. Science 312, 1208–1211.

Batey, R.T., Gilbert, S.D., and Montange, R.K. Montange, R.K., and Batey, R.T. (2006). Na- Winkler, W., Nahvi, A., and Breaker, R.R. (2004). Nature 432, 411–415. ture, in press. (2002). Nature 419, 952–956.

It Takes a PHD to Read the Code

Jane Mellor1,* 1Department of Biochemistry, South Parks Road, Oxford, OX1 3QU, UK *Contact: [email protected] DOI 10.1016/j.cell.2006.06.028

The pattern of histone modifications, called the histone code, influences transitions between states and the regulation of transcriptional activity. Four recent papers describe how plant homeodomain (PHD) finger read part of this code. The PHD finger may promote both expression and repression through interactions with trimethylated 4 on histone 3 (H3K4), a universal modification at the beginning of active .

In the eukaryotic nucleus, chromatin standing for how this code may be found at the 5′ region of active genes carries not only genetic information read. These authors show that the together with acetylated . By encoded in the DNA but also epige- plant homeodomain (PHD) finger is contrast, K36me3 generally accumu- netic information carried by histone a highly specialized methyl-lysine lates toward the 3′ region of active proteins in the form of reversible cov- binding domain that is found in a genes that is also associated with alent modifications. Many of these variety of proteins and that regulates deacetylated lysines. A key question modifications occur at the unstruc- gene expression (Figure 1; Li et al., is how these simple small chemical tured histone “tails” that are pre- 2006; Peña et al., 2006; Shi et al., modifications, found on relatively dicted to protrude between the gyres 2006; Wysocka et al., 2006). large histone proteins, make such a of nucleosomal DNA that encircle the In recent work, high-resolution big difference to nuclear processes, histone core. These modifications chromatin immunoprecipitation has particularly gene regulation. may regulate access to the DNA and revealed distinct distributions and Accumulating evidence sug- thus influence nuclear processes, associations for the different modi- gests that evolutionarily conserved such as transcription. Accumulating fications throughout the genome. domains within code-reader proteins evidence suggests that these modi- For example, methylated lysine 9 bind to certain histone modifications fications are part of a histone code (K9) or K27 on histone H3 are gen- with very high specificity, thereby and that they act as highly selective erally associated with genes whose distinguishing the same modifica- binding platforms for the association transcription is repressed, whereas tion at different residues, for exam- of specific regulatory proteins (the methylated K4, K36, and K79 are ple trimethylation at K4, K9, and K27. code readers). Four papers recently found in active chromatin. Moreover, Both the sequence environment sur- published in Nature from the Patel, “active” marks show distinct distri- rounding the methylated lysine and Kutatelade, Allis, and Gozani labo- butions over transcribed genes. The the distinctive folds in otherwise ratories have increased our under- trimethylated form of K4 (K4me3) is conserved domains on the reader

22 Cell 126, July 14, 2006 ©2006 Elsevier Inc. proteins appear to be major deter- minants of site discrimination at this level. However, how different states of modification at one residue, such as K4, K4me1, K4me2, or K4me3, are discriminated is far from clear. How do these domains discrimi- nate between these very small chem- ical differences? At the simplest level, different domains associate with different marks. For example, previous work has shown that the shows a high affinity for acetylated lysine, whereas the shows high affinity for methylated lysine. The chromo- domain containing proteins het- erochromatin 1 (HP1) and polycomb potentiate the formation of repressive chromatin environ- ments via interactions with methyl- ated K9 or K27, respectively. Even though lysines 9 and 27 are found in an identical local sequence envi- ronment (ARKS) (Figure 1A), swap- ping the of HP1 and polycomb switches the specificity of the lysine that is recognized. This suggests that the chromodomains Figure 1. Methyl-Lysine Recognition by the PHD Finger and Chromodomains are involved in both binding target (A) The local sequence environment of the amino-terminal region of histone H3 influences the sites and discriminating between specific binding of PHD fingers and chromodomains to K4, K9, K27, and K36. Electrostatic sur- face representations (red as negatively charged and blue as positively charged) of the PHD do- them. The basis of this discrimina- main of BPTF, the double chromodomain of CHD1, and the chromodomain of HP1 in complex tion is explained by the high-resolu- with methylated H3 peptides are shown. The invariant tryptophan residues separating the K4me3 tion structures of the polycomb and and R2 binding pockets in the PHD finger and double chromodomain are indicated. (B) Ribbon representation of the crystal structure of the BPTF PHD finger, the bromodomain, and HP1 chromodomains in complex the linker between them. The two bound zinc ions within the PHD fold are shown as balls. The with H3 peptides. These structures positions of the K4me3, R2, and acetlylated lysine (Kac) binding pockets are indicated. Images indicate that the chromodomain of courtesy of Haitao Li and Sepideh Khorasanizadeh. Figure modified from Li et al. (2006) and Flanagan et al. (2005). polycomb distinguishes K27 from K9 via an extended recognition groove that binds five additional residues main of CHD1 that direct H3 peptide binding to K36me3 but not K4me preceding the ARKS motif (Fischle et binding to a groove at the interchro- (Shi et al., 2006). Subtle differences al., 2003). modomain junction and block the site in key residues within otherwise con- Members of the chromodomain that the H3 peptide would occupy in served protein folds coupled with protein (CHD) family have two chro- HP1 and polycomb (Flanagan et al., the immediate sequence environ- modomain motifs. In contrast to HP1 2005). In addition, the specificity of ment of the methylated lysine appear and polycomb, CHD1 shows high CHD1 for K4me may be coupled to to determine site specificity for the affinity for methylated K4 on active a binding pocket for arginine 2 (R2) chromodomain. genes. Moreover, the way in which as no other site at which a lysine is However, the chromodomains the CHD1 chromodomains bind to methylated has an arginine at the n-2 appear unable to distinguish between methyl-lysine is different from HP1 position (Figure 1). Tryptophan 67 is the degree of at their and polycomb (Figure 1A). For HP1 intimately involved with the formation target lysine. Given that mono, di, and polycomb, there is a three-resi- of the R2 binding pocket in CHD1 and trimethylation states of K4 are due aromatic cage surrounding the (Figure 1A). Other human CHD iso- found in different regions of chroma- methyl-lysine, whereas CHD1 rec- forms and Chd1 in budding yeast tin (which implies that the different ognition involves two aromatic resi- lack tryptophan 67 and are unlikely states of methylation are function- dues. Discrimination between K4me to bind to K4me (Flanagan et al., ally important) other strategies or and K9me may result from unique 2005). Consistent with this expecta- protein folds for discriminating dif- insertions within the first chromodo- tion, CHD3 (Mi-2α) shows specific ferent methylation states must exist.

Cell 126, July 14, 2006 ©2006 Elsevier Inc. 23 This prediction has been born out in ferent domains (the PHD and bro- potentially, tumor suppression. In the four papers recently published modomains) on the code reader. In response to DNA damage, ING2, via in Nature (Li et al., 2006; Peña et al., this model, the NURF (nucleosome K4me3, stabilizes the binding of an 2006; Shi et al., 2006; Wysocka et al., remodeling factor) complex that mSin3-HDAC1 histone deacetylase 2006). This work reveals that the PHD contains BPTF might be targeted to complex at the promoters of genes finger, although structurally unrelated the beginning of active genes by the that stimulate proliferation, such to the CHD1 chromodomain, is a binding of the BPTF bromodomain as cyclin D1, resulting in histone highly specialized methyl-lysine bind- to acetylated lysines. In this way, the deacetylation and the repression of ing domain (Figure 1A). Moreover, the BPTF bromodomain could influence the active gene. mechanism for recognition of K4me the specificity of the interaction of Given that the PHD domain is only is likely to be conserved in CHD1 and the PHD finger with K4me3 because one of a cluster of new methyl-lysine the PHD finger. In both cases recog- K4me3, like acetylated lysine, is binding motifs that have recently nition involves a cage formed by two concentrated at the beginning of been reported (Huang et al., 2006; aromatic residues and an invariant active genes. The helical linker that Kim et al., 2006; Wysocka et al., tryptophan that separates the K4me separates the two domains could 2005) and that structures and specif- and R2 binding pockets (Li et al., act as a molecular ruler, linking icities of other folds found on chro- 2006; Peña et al., 2006). An exten- a particular combination of me3/ matin-associated proteins remain to sive network of hydrogen bonds and acetyl marks to chromatin remod- be determined, the rules that deter- complementary surface interactions eling by NURF (Figure 1B). Whether mine how the putative histone code are responsible for the unique recog- other multidomain proteins, with is written, read, and interpreted are nition of ARTK(me3)QT in the histone helical linkers of different lengths, likely to remain enigmatic for some peptide by the PHD finger. Mutational recognize other combinations of time. analysis supports the essential role methyl/acetyl marks remains to be of the residues in the PHD finger that determined, but it is a very attrac- References form the K4me3 and R2 binding pock- tive model for how different states ets for the histone peptide in vitro and, of methylation are discriminated by Fischle, W., Wang, Y., Jacobs, S.A., Kim, Y., Allis, C.D., and Khorasanizadeh, S. (2003). importantly, for the function in vivo of the code readers. Genes Dev. 17, 1870–1881. the proteins that contain these PHD The importance of the bromodo- Flanagan, J.F., Mi, L.Z., Chruszcz, M., Cym- fingers (Shi et al., 2006; Wysocka et main for suppressing loss of BPTF borowski, M., Clines, K.L., Kim, Y., Minor, W., al., 2006). function, particularly the compro- Rastinejad, F., and Khorasanizadeh, S. (2005). Although a K4me2 histone pep- mised spatial control of Hox gene Nature 438, 1181–1185. tide has lower affinity for the PHD expression in Xenopus oocytes, sug- Huang, Y., Fang, J., Bedford, M.T., Zhang, Y., domain than a similar peptide con- gests that the PHD finger and the bro- and Xu, R.M. (2006). Science 312, 748–751. taining K4me3, this alone does not modomain cooperate in mediating Kim, J., Daniel, J., Espejo, A., Lake, A., Krish- explain how specificity for K4me3 BPTF function in early development. na, M., Xia, L., Zhang, Y., and Bedford, M.T. is achieved in vivo, as is observed Given that the BPTF loss-of-function (2006). EMBO Rep. 7, 397–403. in inhibitor of growth 2 (ING2) and phenotype mimics loss of WDR5, a Li, H., Ilin, S., Wang, W., Duncan, E.M., Wysoc- bromodomain-proximal PHD finger WD40 repeat protein that controls ka, J., Allis, C.D., and Patel, D.J. (2006). Na- (BPTF). The PHD finger in BPTF (like global levels of K4me3 by the MLL1 ture. Published online May 21, 2006. 10.1038/ many PHD fingers) is found in close (Wysocka et al., nature04802. proximity to a bromodomain (Figure 2005), all BPTF function is likely to be Peña, P.V., Davrazou, F., Shi, X., Walter, K.L., 1B). Intriguingly, the histone code mediated via K4me3. It is clear, how- Verkhusha, V.V., Gozani, O., Zhao, R., and Ku- hypothesis predicts the existence ever, that the biological function is tateladze, T.G. (2006). Nature. Published on- line May 21, 2006. 10.1038/nature04814. of code-reader proteins with double determined not by the K4me3 mark recognition domains such as this per se, but by the nature of the code Shi, X., Hong, T., Walter, K.L., Ewalt, M., Michishita, E., Hung, T., Carney, D., Pena, P., PHD-bromodomain module with the readers that recognize the modifi- Lan, F., Kaadige, M.R., et al. (2006). Nature. potential to recognize combinatorial cation (Shi et al., 2006; Wysocka Published online May 21, 2006. 10.1038/na- marks such as trimethylation and et al., 2006). This is illustrated by ture04835. acetylation on one or multiple his- the aromatic cage PHD fingers of Wysocka, J., Swigut, T., Milne, T.A., Dou, Y., tone tails. As it may be too difficult the ING tumor suppressor proteins Zhang, X., Burlingame, A.L., Roeder, R.G., to discriminate between me2 and that, like the PHD finger of BPTF, Brivanlou, A.H., and Allis, C.D. (2005). Cell 121, 859–872. me3 using a single protein fold, a bind with specificity to K4me3 and simpler solution might be to discrim- K4me2 (Peña et al., 2006; Shi et al., Wysocka, J., Swigut, T., Xiao, H., Milne, T.A., inate using a combinatorial code (in 2006). Here the interaction between Kwon, S.Y., Landry, J., Kauer, M., Tackett, A.J., Chait, B.T., Badenhorst, P., et al. (2006). this case the recognition of K4me3 the PHD finger and K4me3 leads to Nature. Published online May 21, 2006. and acetylated lysine) and two dif- the repression of active genes and, 10.1038/nature04814.

24 Cell 126, July 14, 2006 ©2006 Elsevier Inc.