The genes that make up our genome are not static entities but undergo a complex modification process that results in certain genes being switched on or off, depending on the state of the cell. This so-called epigenetic regulation is now known in enough detail to allow its specific interference via targeted drugs, strengthening the fight against such human diseases as cancer and neurodegeneration with an epigenetic component.
Fueled by the expertise of a team of international specialist authors, this first reference on the booming topic covers everything a drug researcher needs to know about targeting epigenetic mechanisms of disease.
The first part of the book surveys current methodologies for finding and validating drug candidates that act via epigenetic mechanisms. The second part systematically surveys known and suspected drug targets within the epigenetic machinery, including coverage of the first successful drug candidates available for these novel targets.

Wolfgang Sippl is the head of Pharmaceutical and Medicinal Chemistry at the Martin-Luther University of Halle-Wittenberg (Germany). He obtained his Ph.D. in Pharmaceutical Chemistry at University of Düsseldorf and was a post-doctoral fellow at the Université Louis-Pasteur in Strasbourg (France), where he worked with Camille G. Wermuth. He then took a senior researcher position in Düsseldorf before moving to the University of Halle-Wittenberg as a full professor in 2003. His main interests are focussed on medicinal chemistry, computational chemistry and drug design

Manfred Jung is Professor for Pharmaceutical and Medicinal Chemistry at the Albert-Ludwigs University of Freiburg (Germany). He obtained a Ph. D. in Pharmaceutical Chemistry from the University of Marburg, working subsequently as a postdoctoral fellow with T. Durst at the University of Ottawa (Canada). In 1994 he started his own group at the Institute of Pharmaceutical Chemistry at the University of Münster and in 2003 he moved to Freiburg to his current position. His main research focus is the synthesis of enzyme inhibitors and in-vitro assay development in the field of histone modifying enzymes.

Epigenetic Targets in Drug Discovery

John Wiley & Sons

Copyright © 2009 WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim
All right reserved.
ISBN: 978-3-527-32355-5

Chapter One

New Frontiers in Epigenetic Modifications

Adam L. Garske and John M. Denu

1.1 Introduction

The basic packaging unit of the genome, the nucleosome, consists of ~146 bp of DNA wound around an octamer of histone proteins. The histone octamer is composed of an H3/H4 tetramer and two H2A/H2B dimers. The unstructured N-terminal regions (tails) of histones protrude outward from the nucleosomal core through superhelical gyres of DNA. While genetic information is encoded in the DNA sequence, processes such as transcription, recombination, DNA replication and DNA repair are controlled by the epigenome (epi is Greek for upon or in addition to). The epigenome is often characterized by heritable or long-term alteration in gene expression patterns that cannot be ascribed to changes in DNA sequence. At the molecular level, epigenetics involves the dynamic regulation of covalent modifications to DNA and the histone proteins. Epigenetics is implicated in processes such as gene expression and silencing, apoptosis, maintenance of stem cell pluripotency, X-chromosome inactivation and genomic imprinting. Therefore, epigenetics can be viewed as the conduit from genotype to phenotype. This chapter provides a framework for our current understanding of molecular epigenetics with particular emphasis on the histone code and it examines the utility of small molecule inhibitors of enzymes that modify DNA and histones.

1.2 DNA Methylation

In multicellular eukaryotes, DNA methylation is associated with transcriptional silencing. In these organisms, DNA methylation has been observed exclusively on the C5 position of the cytosine ring and is frequently found in CpG-rich regions. This process is attributed to the action of DNA methyltransferases (DNMTs), which utilize the cofactor, S-adenosyl-L-methionine. Approximately half of all human genes have CpG islands in their promoter regions but these stretches of DNA are typically hypomethylated and transcriptionally permissive. Methylation in the proximity of the transcription start site or within a gene is associated with transcriptional repression.

DNA methyltransferases can be classified in two categories: (1) de novo DNMTs and (2) maintenance DNMTs. De novo DNMTs methylate previously unmodified cytosines in CpG islands, while maintenance DNMTs duplicate existing DNA methylation patterns onto newly synthesized DNA strands during replication. DNMT3a and 3b are examples of de novo DNMTs and are capable of methylating both unmethylated and hemimethylated (only one strand in the DNA duplex is methylated) sites of DNA. Another protein, DNMT3-like (DNMT3L), acts as a regulatory factor in de novo methylation of DNA despite lacking a catalytic domain. This protein is involved in genetic imprinting and methylation (indirectly) of retrotransposons in pre-meiotic spermatogonial stem cells. In the case of genomic imprinting, DNMT3L was found to collaborate with DNMT3a to achieve DNA methylation by localizing the latter to unmodified K4 of histone H3 via a plant homeodomain (PHD)-like domain. DNMT1, which has a catalytic preference for hemimethylated DNA, is an example of a maintenance DNA methyltransferase. DNMT1 is localized to replication foci by interaction with proliferating cell nuclear antigen (PCNA). Recently, an accessory protein, UHRF1, was shown to target DNMT1 to hemimethylated DNA during S phase. UHRF1 is known to bind methylated DNA in the context of CG, CXG (X A, T or C) or an asymmetrical sequence using a SET and RING associated (SRA) domain. Furthermore, UHRF1 was found to be required for stable association of DNMT1 with chromatin. Therefore, DNMT1 appears to regulate epigenetic inheritance in a mechanism that involves a complex with UHRF1 and PCNA in regions of replicating heterochromatin (tightly packed, transcriptionally repressive chromatin) that contain hemimethylated DNA.

Several other means for targeting DNMTs have been identified. In a sequence-dependent manner, DNMTs bind directly to DNAby virtue of a conserved proline and tryptophan (PWWP) domain. For both de novo methyltransferases, DNMT3a and DNMT3b, a PWWP domain is essential for chromatin targeting. Missense mutation of the PWWP domain in the DNMT3B gene triggers centromeric heterochromatin instability, pericentromeric instability and facial anomalies (ICF) syndrome. A second mechanism for DNMT targeting is through recruitment by site-specific transcriptional repressors. The oncogenic fusion protein, promyelocytic leukemia–retinoic acid receptor (PML-RAR), localizes methylation to specific genes in cancer cells by recruitment of DNMTs. More recently, the polycomb group protein, enhancer of Zeste homolog 2 (EZH2), was found to target DNMTs to EZH2-repressed genes. Finally, small RNA molecules have been implicated in DNMT targeting.

Transcriptional repression by DNA methylation is achieved by various modes of action. In one such mechanism, DNA methylation simply inhibits the binding of a transcription factor. By a more complex means of action, a number of DNA methyl-binding proteins potentiate transcriptional silencing. In some cases, binding is accompanied by the action of an associated histone-modifying enzyme. For example, one DNA methyl-binding protein, MBD1, associates with SETdomain bifurcated 1 (SETDB1), a histone methyltransferase (HMT). During DNA replication, SETDB1 associates with MBD1 in addition to chromatin assembly factor-1 and catalyzes methylation of histone H3 at lysine 9. Trimethylation at H3K9 is associated with heterochromatin. A second example that links DNA methylation to a histone modification state involves PML-RARα. MBD1 can form a repressor complex with N-CoR, DNMTs, HDAC3 and PML-RARa to silence PML-RARα-dependent promoters. This aberrant gene silencing is manifested by HDAC-mediated histone deacetylation, DNA methylation, as well as time-dependent spreading of MBD1 outside of the promoter region. The spreading of MBD1 along regions of methylated DNA is thought to recruit additional repressor enzymes.

1.3 Histone Modifications and the Histone Code Hypothesis

A vibrant area in epigenetic research involves covalent modifications of histones. The most commonly observed histone modifications include acetylation, methylation and phosphorylation. However, other modifications such as citrullination, ubiquitination, SUMOlyation, prolyl isomerization, ADP-ribosylation and biotinylation are being increasingly recognized. These modifications are primarily located on the unstructured N-terminal tails of histones (Figure 1.1), yet an increasing number of modifications in the α-helical histone fold domains have been documented. In this chapter, we limit our detailed discussion to histone acetylation and methylation. One interpretation of the epigenetic consequences of histone modification is the histone code hypothesis, which predicts that combinatorial histone modification states result in unique biological outcomes. This molecular code is regulated by alteration of histone–histone interactions, histone–DNA interactions and histone–nonhistone protein interactions. While there is some dispute whether each histone modification state results in a unique downstream function, it is generally agreed that histone modifications can lead to differences in binding and selectivity and that these differences can propagate a variety of epigenetic outcomes.

A number of strategies have emerged for characterizing histone modifications and understanding their significance. The workhorses of histone modification mapping have been modification-specific antibodies and mass spectrometry (see the requisite chapters in this book for more details). In the ChIP on chip approach, DNA is crosslinked to DNA-binding proteins, digested and immunoprecipitated with an antibody for the histone modification of interest. Following PCR amplification of the associated DNA, the histone modification can be linked to particular gene regions using microarray technology. This methodology was used extensively in a large-scale human and mouse epigenome study. While this method was extremely powerful for mapping locations of histone modification on genes, it is limited by the specificity of available antibodies and cannot be used to determine the modification status of histones within the same nucleosome. Mass spectrometry, in contrast, is capable of determining the modification status of individual histone proteins. A recent study identified 74 unique histone H4 modification patterns in differentiating human embryonic stem cells. Several chemically driven strategies have found utility in studying the effects of histone posttranslational modifications. For example, native chemical ligation/expressed protein ligation, chemical incorporation of methyl-lysine mimetics and genetic incorporation of modified amino acids (such as acetyl lysine) have enabled the generation of site-specifically modified histones. In one study employing semisynthetic histones, acetylation of H4K16 was shown to modulate chromatin compaction and its ability to form cross-fiber interactions. Recently, we developed a method for assaying the specificity of enzymes and proteins that read the histone modification patterns using combinatorial peptide libraries based on the modification patterns of N-terminal histone tails [29].

Reversible lysine acetylation of histones is regulated by the opposing activity of histone acetyltransferases (HATs) and histone deacetylases (HDACs). Acetylation of the lysine e-amino group results in neutralization of a positive charge, while deacetylation reestablishes the presence of a primary amine. Hypoacetylation typically facilitates formation of highly condensed chromatin (heterochromatin) and transcriptional repression, while hyperacetylation tracks with chromatin that is more loosely associated with DNA (euchromatin) and transcriptional activity. HATs employ an acetyl-CoA cofactor, operate in large multiprotein complexes and can be classified as members of the GNAT, MYST or CBP/p300 families of enzymes. However, a recently identified HAT in yeast, Rtt109, appears to belong to an entirely different family due to lack of homology with other HATs. HDACs are responsible for the removal of acetyl groups from histone lysines. These protein deacetylases are categorized as class I, II, III and IV HDACs. Class I, II and IV HDACs are metal-dependent acetyl hydrolases that yield acetate as a product. Class III HDACs, or sirtuins, utilize an NAD⁺ cofactor and couple deacetylation with the formation of O-acetyl-ADP-ribose (OAADPr) and nicotinamide. Sirtuins have been postulated to link epigenetics to metabolic processes and act on a number of nonhistone proteins.

Reversible histone methylation is a highly specific process that is catalyzed by the action of histone methyltransferases (HMTs) and histone demethylases on lysine and arginine residues. Like DNMTs, HMTs employ a SAM cofactor. Lysine can be mono-, di- or trimethylated and arginine can be monomethylated and symmetrically or asymmetrically dimethylated. The consequences of histone methylation appear to be largely context-dependent. For example, trimethylation of H3K4 is a mark of transcriptionally active chromatin, while trimethylation of H3K9 tracks with transcriptionally repressive chromatin. Lysine HMTs are divided into the SET and Dot1 families. Members of the SET family contain a SET domain and representative examples include mixed-lineage leukemia (MLL) and EZH2 proteins, which are specific for methylation of H3K4 and H3K27, respectively. Dot1 lacks a SET domain and is responsible for mono- and dimethylation of H3K79. Protein arginine methyltransferases (PRMTs) usually act on glycine-arginine-rich regions within their substrates. Following N^ω-monomethylation, type I PRMTs can catalyze a subsequent methylation on the same atom (asymmetric dimethylation), while type II PRMTs methylate the other w-nitrogen (symmetric dimethylation).

The discovery of LSD1, the first histone demethylase to be characterized, invalidated the notion that histone methylation was a permanent mark. LSD1 is an FAD-dependent amine oxidase that catalyzes mono- and di- demethylation of H3K4. Interestingly, a byproduct of LSD1 catalyzed demethylation, hydrogen peroxide, is thought to link local DNA oxidation to estrogen-induced gene expression. The Jumonji C (JmjC) hydroxylases, another class of histone demethylases, require iron and α-ketoglutarate and produce formaldehyde as a byproduct. While LSD1 only demethylates mono- and dimethyl lysine residues in proteins, members of the JmjC family such as JMJD2A and JARID1A are specific for di- and trimethyl lysines. JMJD6 demethylates both symmetric and asymmetric arginine to their monomethylated counterparts at H3R2 and H4R3. Protein arginine deiminases (PADs) have been suggested to catalyze methylated arginines demethylimination to furnish citrulline, yet in vitro evidence for this transformation suggests that methylarginine is a very poor substrate, if at all. Further studies are needed to resolve these issues. Deimination of arginine to citrulline, however, has been widely characterized. Currently, there is no evidence that protein deimination is a reversible modification. See the relevant chapters of this book for more information on reversible histone aceylation and methylation.

1.4 Origins of Specificity in Histone Binding Proteins/Modifying Enzymes

The primary readers of the histone code are histone binding domains (HBDs). HBDs are often found in histone-modifying enzymes, ATP-dependent chromatin remodeling factors and transcription factors. In this chapter, we focus on domains with the ability to recognize acetylation and methylation. The observation that bromodomains function as histone acetyl-lysine binding modules set the stage for discovery of a number of specialized HBDs. The specificity of the these binding modules appear to be dictated by both the type of modification and the context of the modified amino acid. For example, although both heterochromatin protein 1 (HP1) and polycomb protein PC2 both contain chromodomains (methyl-lysine binding modules), the former binds trimethylated H3K9 while the latter binds trimethylated H3K27 despite an identical sequence surrounding each lysine (ARKS). The binding specificity is achieved by means of an extended recognition groove in polycomb that recognizes residues 20–24. Another mechanism for discriminating among potential binding partners is by the extent of a particular modification state (e.g. mono-, dior trimethylation).

Bromodomains, which are composed of approximately 110 amino acids, fold into a left-handed antiparallel four-helix bundle. They contain a hydrophobic tunnel, which accommodates binding of acetylated lysine. Relative to other histone binding domains, bromodomains are promiscuous with regard to the sequence to which they bind and typically have dissociation constants in the ~50–350 μM range. These targeting modules are frequently found in HATs. It has been suggested that bromodomains enable a mechanism whereby HATs can propagate acetylation along a histone or a nucleosome. Other proteins that harbor bromodomains include members of the HMT family and ATP-dependent remodeling enzymes. One such example, the SWI/SNF ATP-dependent remodeling complex employs a bromodomain to associate with acetylated promoter nucleosomes. Using the energy liberated from ATP hydrolysis, the SWI/SNF complex mobilizes nucleosomes. The bromodomain binding activity of SWI/SNFis essential for displacing the SAGA HAT complex, as well as facilitating octamer transfer on H3-acetylated nucleosomes.

The tudor domain royal superfamily includes the chromodomain, double chromodomain, chromo barrel, tudor, double/tandem tudor and MBTdomains. Tudor and chromo domains bind to di- and trimethylated lysine while MBT domains prefer mono- and dimethylated lysine. Tudor domains are also capable of recognizing symmetrically dimethylated arginine. Increasing lysine methylation results in reduced hydrogen bonding potential and solvation properties, as well as greater hydrophobicity and a larger cation radius. Because many methyl binding modules engage in cation-p interactions, the degree of methylation is critical in conferring specificity. Domains that preferentially bind to higher lysine methylation states can accommodate a diffuse and hydrophobic cation. In contrast, hydrogen bonding and steric exclusion are dominant factors in regulating specificity for lower methylation states (or unmethylated lysine). It was noted that the chromodomain of HP1 binds specifically to di- and trimethylated H3K9. HP1 recognizes methylation catalyzed by the action of the suppressor of variegation 3–9 homologue 1 (SUV39H1) methyltransferase. Structural analysis revealed that a H3K9 trimethylated histone peptide was immobilized in an induced β-strand sandwich with the protein. Since the original HP1 structures, a similar induced β-strand fit was observed for the interaction between other methyl-recognizing proteins and their methylated peptide binding partners. An example of a double tudor domain is that of the histone demethylase and transcriptional repressor, JMJD2A. The JMJD2A double tudor domain preferentially binds to di- and trimethylated H3K4 and H4K20.

(Continues…)

Excerpted from Epigenetic Targets in Drug Discovery Copyright © 2009 by WILEY-VCH Verlag GmbH & Co. KGaA, Weinheim. Excerpted by permission of John Wiley & Sons. All rights reserved. No part of this excerpt may be reproduced or reprinted without permission in writing from the publisher.
Excerpts are provided by Dial-A-Book Inc. solely for the personal use of visitors to this web site.