Plant and Insect Epigenetics

Epigenetic gene regulatory mechanisms are important in a variety of phenomena throughout the eukaryotes including the silencing of transposons, sex chromosome dosage compensation, imprinting, and the inappropriate gene expression that occurs in cancer cells. In addition, recent findings suggest that epigenetic mechanisms play an unforeseen and expanding role in the normal regulation of genes during development. This review will focus on advances made in our understanding of epigenetic gene regulation in insect and plants.

The understanding of epigenetic gene regulation in insects and plants has been driven by discoveries in two model organisms, the common fruit fly Drosophila melanogaster and the small experimental plant Arabidopsis thaliana. The regulatory systems show many striking commonalities between organisms in these two different kingdoms of life, uncovering basic properties of epigenetic gene regulation that are conserved over billions of years and therefore likely found in the vast majority of eukaryotes. There are also interesting differences between plant and insect systems, and even important differences between different insect or plant species, suggesting that the evolution of epigenetic systems is a driving force behind the diversity of eukaryotic organisms.

Note: If you’d like to print out a hard copy of these reviews, you can find it at Zymo Research’s website in their Publications.

Epigenetic Gene Regulation In Insects

Drosophila was one of the first well studied model genetic organisms and early genetic studies in Drosophila have shed light not only on epigenetic mechanisms within insects, but also revealed highly conserved paradigms for the activation or silencing of genes controlled by chromatin structure within other eukaryotes. Chromatin is the physiological template for gene transcription and is composed of DNA, histones, and other non-histone proteins. The basic repeating unit of chromatin is the nucleosome octamer, which is composed of the core histones H3, H4, H2A, and H2B, wrapped around 147 base pairs of DNA1. The histone N-terminal tails, which protrude out from the core octamer, have been the topic of recent intense study. Histone tails are subject to a host of post-translational modifications including methylation, acetylation, phosphorylation, ubiquitination, and ADP-ribosylation. The “histone code” hypothesis2-4 posits that histone modifications direct the binding of specific proteins that mediate chromatin function, and thus gene regulation. Genomes are organized into two different general types of chromatin: heterochromatin and euchromatin. Heterochromatin is mostly located near centromeres and telomeres and corresponds to repeat rich areas of the genome. Heterochromatin is usually repressive to gene transcription, and is composed of regular arrays of nucleosomes with particular configurations of histone modifications including methylation of lysine 9 of histone H3 (H3K9), methylation of lysine 20 of histone H4 (H4K20), and low levels of acetylation of a number of sites on both histone H3 and H4. Euchromatin on the other hand generally constitutes most of the gene coding portions of genomes, and is more generally permissive to gene transcription. Euchromatin is composed of less regularly spaced nucleosomes and the histones are more frequently acetylated on H3 and H4 as well as methylated on lysine 4 of H3 (H3K4).

Mechanisms for the inheritance of constitutive heterochromatin discovered through the genetics of position effect variegation: Position effect variegation (PEV) is a phenomenon in which a normally active gene that is present at the boundary between euchromatin and heterochromatin shows a variable or mosaic pattern of gene expression due to the silencing or activation of gene expression in different cells. Screens for mutations that alter PEV in Drosophila have uncovered important factors involved in the processes of gene silencing or gene activation. One of the most important was Suppressor of position variegation effect3-9Su(var)3-95, which encodes a SET domain-containing histone methyltransferase enzyme6. Su(var)3-9 is specific for methylating H3K9 and is the main factor maintaining this mark in heterochromatin. The function of Su(var)3-9 is also conserved in other organisms. For example, in the yeast Schizosaccharomyces pombe, the Su(var)3-9 homolog CLR4 is involved in silencing of the mating type region and centromeric heterochromatin7, and loss of the mammalian Suv39h histone methyltransferases impairs heterochromatin and genome stability8. After the discovery of the function of the Su(var)3-9 SET domain, many other SET domains have been shown to methylate histones at other positions including H3K4, H3K27, H3K36, and H4K20 and to be critical in epigenetic gene regulation9. Methylation of lysine residues of histone tails can either be associated with transcriptional activation or transcriptional repression. In addition, lysines can accept three methyl groups, and can therefore be monomethylated, dimethylated, or trimethylated (hereafter denoted as m, m2, and m3), and abundant evidence shows functional differences between these methylation states.

A second factor affecting PEV was Su(var)2-5, HETEROCHROMATIN PROTEIN1 (HP1), a conserved protein and a major constituent of heterochromatin10. HP1 contains a chromodomain which specifically binds to methylated H3K9 [the site created by Su(var)3-9], and this binding is essential for heterochromatin formation in vivo7,11-13. HP1 serves as a platform for the recruitment of other proteins. For instance, in the yeast S. pombe the HP1 homolog SWI6 is known to recruit both silencing factors and antisilencing factors to effect an equilibrium between various chromatin modifications that confers the steady state level of gene activity14.

The Polycomb and Trithorax systems: Drosophila genetics also uncovered two very important and conserved systems for the long term silencing or activation of gene expression, the Polycomb and Trithorax systems. A group of proteins called the Polycomb group proteins (PcG) were initially discovered by mutations in genes required to prevent the inappropriate expression of a group transcription factors important for early development called the homeotic (Hox) genes15. Two groups of PcG proteins form two protein complexes called PRC1 and PRC2. The key component of the PRC2 complex is Enhancer of zeste [E(z)], which encodes a SET domain histone H3 methyltransferase protein specific for trimethylation of H3K27 (H3K27m3). The presence of H3K27m3 generally causes gene repression. A key component of PRC1 is the protein Polycomb, which contains a chromodomain that specifically binds to H3K27m316-18. Chromatin immunoprecipitation studies show that PRC1 and PRC2 are stably localized to the Hox genes, and both are required for Hox gene silencing. Although it is not entirely clear how the Polycomb complexes cause gene silencing, it is known that the PRC1 complex contains the component dRING, which causes the ubiquitylation of histone H2A at lysine 119, and this modification is required for gene silencing. Furthermore, two Jumonji domain proteins, UTX and JMJD3, act as specific histone demethylases for H3K27, and likely work in opposition to the E(z) histone methyltransferase to strike the right balance of H3K27 methylation levels15.

The Trithorax group (TrxG) of proteins were also discovered through Drosophila genetics, as suppressors of the PcG mutant phenotypes10. Trithorax proteins are localized to the same loci as the PcG proteins, and act in opposition to PcG by stimulating transcription. The key TrxG protein is Trithorax, which encodes a SET domain containing histone methyltransferase protein specific for H3K4 trimethylation, a mark of active chromatin. Both PcG and TrxG group proteins are recruited to specific DNA sequences called Polycomb Response Elements (PREs). While precisely how they are recruited is a major unanswered question in the field, it is most likely that a variety of different DNA binding transcription factors are involved15. In mammals it even less clear since no identifiable PREs have been found.

The paradigm of Hox gene regulation suggested a model in which silencing established during early development is then maintained by PcG proteins throughout the remainder of development. However, later work has shown exceptions to this general rule, and suggests a more dynamic interaction between Polycomb silencing, Trithorax activation, and the activity of specific transcription factors, the interplay of which determines the final state of gene activity15. In this way the Polycomb and Trithorax systems provide a layer of epigenetic regulation within which canonical sequence specific DNA binding transcription factors act. Additionally, it is now clear that besides the Hox genes, hundreds of other Drosophila genes, as well as thousands of mammalian genes, are under control of the Polycomb and Trithorax systems, highlighting them as important and widely used epigenetic mechanisms15.

The role of chromatin and epigenetic modification during transcription: The role of Drosophila Trithorax in gene regulation turns out to be much more widespread than just the set of PcG associated genes. Indeed, it appears that all actively transcribing genes contain H3K4m3 at their promoters15,19. The yeast (Saccharomyces cerevisiae) homolog of Trithorax is SET1, which is a key component of the COMPASS complex, and is responsible for all H3K4 methylation in yeast. SET1 is recruited to transcribed genes by interacting with the phosphorylated form of the carboxy-terminal domain of RNA polymerase II, a site at which many other factors are also binding during the process of transcription. COMPASS then appears to travel with Pol II and is responsible for H3K4 methylation that is associated with transcriptional elongation19. Interestingly, H3K4 methylation is in part dependent on a second histone modification, monoubiquitination of histone H2B. The Trithorax homolog in mammals is the MIXED LINEAGE LEUKEMIA (MLL) gene, which is frequently mutated in particular types of leukemia. Like in other organisms MLLs are associated with RNA polymerase II and co-localize to Pol II binding sites19.

Other histone modifications are also important in the process of transcription20. As just one example, H3K36 methylation is associated with the transcribed bodies of genes, and is deposited along with transcriptional elongation by RNA polymerase II. H3K36 functions to create a more repressive chromatin environment to suppress transcription from cryptic initiation sites within the body of the genes that might otherwise interfere with the primary transcript from the main Pol II promotor20. H3K36 methylation appears to function by creating a binding site for the chromodomain of Eaf3, a subunit of the RpdS histone deacetylases complex. The RpdS complex then hypoacetylates the nucleosomes within the body of genes, a condition generally associated with transcriptional suppression.

Role of nucleosome remodeling in gene regulation: Nucleosomes are not evenly spaced along DNA, and often the removal or shifting of nucleosomes from promoters is necessary for transcription factors to gain access to the DNA. A comprehensive way of studying nucleosome positioning is to use high throughput sequencing of DNAs that are associated with mononucleosomes that have been released from chromatin by digestion with micrococcal nuclease. One such recent study showed that, like in other eukaryotes, most active Drosophila genes usually contain a small region of the promoter that is generally depleted of nucleosomes21. In addition, a special variant of H2A called H2A.Z is enriched at the 5´ ends of genes. Interestingly, Drosophila genes also showed a nucleosome free region at the 3´ end of active genes, which might possibly play a role in the termination of transcription21.

Drosophila genetics again helped in our understanding of proteins that help shape these nucleosomal patterns. One class of TrxG mutations identified genes known as ATP-dependent chromatin remodeling proteins, which use the energy of ATP hydrolysis to alter nucleosome positioning. One in particular is the TrxG protein Brahma, a Drosophila homolog of the well-characterized yeast protein SWI2/SNF2. In vitro, Brahma can disrupt regular arrays of nucleosomes on DNA, and in vivo Brahma localizes with RNA polymerase II, suggesting a general role in gene activation10. In addition, Brahma containing complexes are also known in some cases to cause repression of gene expression, a function that is also likely carried out by its nucleosome remodeling capacity.

The role of small RNAs in epigenetic gene regulation: In the yeast S. pombe, small interfering RNAs (siRNAs) are critical for the establishment and maintenance of gene silencing and repressive H3K9 methylation22. This phenomenon also appears to be conserved in Drosophila because mutation of three RNAi components, PIWI, AUBERGINE or SPINDLE-E causes reduction of H3K9 methylation and delocalization of the heterochromatin factor HP122. Additional insight into this type of regulation came from studies the small RNAs that associate with the PIWI proteins, the Piwi-interacting RNAs or piRNAs23. piRNAs seem to be especially important in targeting transposons throughout the genome, and appear to act primarily by slicing up and destroying transposon RNAs. Interestingly, in mammals a similar system appears also to target transposons, and mammalian piRNAs were recently shown to play an important role in the establishment of DNA methylation during male germ cell development24. While these pathways are only beginning to bedeciphered, it is clear that small RNAs are playing vital roles in the targeting of epigenetic marks to the genome.

DNA methylation in the Honeybee: Cytosine DNA methylation is a conserved gene silencing mechanism that functions to suppress transposable elements as well as to regulate processes such as X chromosome inactivation and genomic imprinting. Surprisingly, while methylation systems are widely conserved throughout eukaryotes, evidence for the existence of methylation in Drosophila is controversial. Drosophila may have a very small amount of methylation, but has clearly lost most of the methylation machinery found in other eukaryotes25. However, this loss seems specific to certain groups of insects, because the Honeybee, Apis mellifera, was recently shown to have all of the key DNA methyltransferase enzymes needed for methylation at CpG sites, and CpG methylation has been detected at endogenous honeybee genome sequences26. While the function of methylation in honeybee is far from clear, a fascinating recent finding suggests that methylation plays a profound role in development. Adult female honeybees develop either as sterile workers, or if fed royal jelly, as fertile queens. Remarkably, silencing of one of the honeybee DNA methyltransferases, Dnmt3, causes an effect similar to feeding with royal jelly, suggesting that an inhibitor of DNA methylation in royal jelly may be responsible for its transformative effects27. In the future, genome wide profiling of DNA methylation in honeybee should shed further light this phenomenon, as well as help us understand the other roles that DNA methylation may play in insects.

Epigenetic Gene Regulation In Plants

Arabidopsis has become the premier model plant for studying a variety of processes, and has become an important system for studying epigenetic regulatory systems that are widely conserved in other eukaryotes. Like insects and other eukaryotes, it has a full complement of histone modification enzymes, nucleosome remodeling complexes, and a DNA methylation system that in many ways resembles that of mammals. The most powerful aspect of Arabidopsis is the excellent forward and reverse genetics available. Saturation mutant screens can be used to identify genes involved in any process, and loss of function mutations for nearly any gene can be obtained from community stock centers.

Polycomb regulation of gene expression: Like many other eukaryotes, plants use the conserved Polycomb system to regulate gene expression. Arabidopsis encodes the key components of the PRC2 complex, but interestingly is completely lacking in PRC1 proteins. Instead, Arabidopsis appears to have recruited its HP1 homolog, called LHP1, to bind to and silence H3K27m3 associated chromatin28,29. The most well studied example of an Arabidopsis gene regulated by the Polycomb system is FLC. FLC encodes a transcription factor that inhibits flowering. During early plant development, FLC is highly expressed and prevents flowering, thereby prolonging vegetative development. However,FLC expression can be silenced by exposure of the plants to long periods of cold, which in nature indicates the passing of winter, and this leads to flowering in favorable conditions in the spring. After the initial cold stimulus, FLCrepression is maintained for long periods of time by the Polycomb system, allowing for flowering to occur at the normal time. At the end of the plant life cycle, sometime around the time of meiosis, FLC is reactivated again, allowing for resetting and the continuation of this cycle in the next sexual generation.

DNA methylation: Arabidopsis has become one of the best organisms for genetic studies of cytosine DNA methylation because of its facile forward and reverse genetics and its small and well annotated genome. ArabidopsisDNA methylation systems have much in common with mammalian systems. However, unlike in mammals where DNA methylation mutants are inviable, Arabidopsis can tolerate mutations that virtually eliminate methylation, allowing for detailed genetic analysis. DNA methylation in Arabidopsis is found in three different sequence contexts, CG, CHG (where H = A, T, or C) and CHH or asymmetric. Furthermore, DNA methyltransferase activities are generally classified as either maintenance or de novo. The methylation of hemimethylated symmetrical sequences (CG and CHG) following DNA replication is maintenance methylation. This results in stable patterns of methylation that are maintained throughout development or, in many cases, between generations30-32. Methylation that occurs at previously unmethylated sites is called de novo methylation. For symmetric sites, de novo methylation need only occur once, after which maintenance activity is sufficient. However, for asymmetric cytosines, which cannot be recognized as hemimethylated sites following DNA replication, methylation is “maintained” by the persistent activity of de novo methyltransferases33.

CG methylation (sometimes referred to as CpG methylation) is the most common type found in mammalian genomes and is also prevalent in Arabidopsis. It is maintained by the maintenance DNA methyltransferase called MET1, which is a homolog of mammalian Dnmt1 that performs the same function34,35. Dnmt1 acts on newly synthesized DNA at replication forks, and shows a preference for hemimethylated CG sites. Mouse Dnmt1 mutants show early embryonic lethality, but Arabidopsis met1 mutants which eliminate CG methylation are viable, yet display a number of specific developmental abnormalities. Remarkably, these abnormalities can be segregated away from met1, and they map to discrete loci such as the SUPERMAN and FWA genes important in development36,37. These DNA methylation defects, or “epialleles”, including hypermethylation of SUPERMAN and hypomethylation of FWA are heritable and can be used for forward genetic screens and other classical genetic experiments33. Recently, an important MET1/Dnmt1 accessory factor has been discovered called VIM1/ORTH in plants and UHRF1 in mammals38. These factors bind directly to hemimethylated DNA and aid in the recruitment of the DNA methyltransferase to replication foci to faithfully replicate DNA methylation.

CHG DNA methylation is mostly found in plant genomes, and is controlled by the plant specific DNA methyltransferase CHROMOMETHYLASE3 (CMT3)39. Mutant screens were used to uncover the cmt3 mutants, by looking for mutations that suppress the phenotypes associated with hypermethylation and silencing of the SUPERMAN locus40. Additionalcmt3 alleles were found from a similar but independent screen using the methylated and silent PAI loci41.These screens also uncovered a second gene required for CHG methylation called KRYPTONITE (KYP), which encodes a histone methyltransferase protein that is specific for H3K9m242,43. The key to understanding the mechanism by which CMT3 and KYP cooperate to maintain CHG methylation came from determining the function of particular domains in these proteins. CMT3 contains a chromodomain that can bind to methylated histone H3 tails44, which is the product KYP. KYP on the other hand contains a domain, the SRA, which can bind directly to DNA methylated at CHG sites45, the product of CMT3. This suggests a model in which CHG DNA methylation is maintained by a positive feed forward loop between the activities of CMT3 and KYP.

CHH DNA methylation is inherited in a manner very similar to that of de novo DNA methylation. In Arabidopsis, bothde novo DNA methylation and the maintenance of CHH methylation is mainly controlled by the DRM2 genes, which are orthologs of the Dnmt3 genes which perform the same functions in mammals46. The function of DRM2 was discovered through genetics experiments using the above-mentioned FWA gene. FWA can adopt two very stable epigenetic states, either DNA methylated and silent, or unmethylated and active. Once established, these states are remarkably stable over many plant generations; an excellent example of the ability of DNA methylation to be meiotically heritable. The methylation of the FWA gene is confined to two tandem direct repeats near the FWApromoter37 (Figure 1). When a new transgenic copy of the FWA gene is introduced into plants by Agrobacterium-mediated transformation, these repeats are reliably de novo methylated, and the new transgene is then transcriptionally silent. However, when FWA transgenes are introduced into plants with a drm2 mutation, the de novoDNA methylation of FWA is blocked, the transgene remains active, and the plants develop a late flowering phenotype due to the ectopic expression of the FWA protein46 (Figure 1). Thus, FWA transformation serves as a convenient assay to screen for mutants that block the establishment of DNA methylation.

The DRM2 DNA methyltransferase appears to be guided by small interfering RNAs (siRNAs), because mutations in a number of RNA silencing genes were found to mimic drm2 mutants in the FWA transformation assay. These include the genes ARGONAUTE4, DICER-LIKE 3 and RNA DEPENDENT RNA POLYMERASE232. This, in many ways, resembles heterochromatin establishment in S. pombe where these same types RNA silencing factors are critical for the establishment of H3K9 histone methylation and gene silencing22. The finding of RNA silencing factors being involved in de novo DNA methylation is consistent with much earlier observations of so-called RNA directed DNA methylation (RdDM), first observed when cytoplasmically replicating RNA viroids were seen to cause de novo methylation of homologous genomic DNA sequences47. When RdDM is directed to promoter sequences, for instance as driven by inverted repeat expressing transgenes, it can cause transcriptional gene silencing48-50. These results suggest that siRNAs are at the heart of de novo DNA methylation processes, which can help explain the sequence specificity of gene silencing, likely involving the pairing of siRNAs with either DNA or nascent RNA transcripts. However, many questions remain concerning the precise mechanisms involved. Interestingly, recent studies suggest that siRNAs transfected into mammalian cells can also cause RNA directed gene silencing, and that antisense RNAs can trigger transcriptional gene silencing and DNA methylation, suggesting that certain aspects of this pathway may be conserved in mammals51,52. In addition, plant RNA directed DNA methylation resembles the phenomenon of piRNA directed DNA methylation that occurs in mammalian male germ cells24.

DNA demethylation also plays an important role in shaping DNA methylation patterns throughout the genome. For instance, the FWA gene is normally methylated throughout the plant life cycle, but FWA is an imprinted gene that becomes specifically demethylated during female gametophyte development53. DNA demethylation is required for expression of FWA and is actively carried out by the DME DNA glycosylase54. DNA glycosylases are normally involved in DNA repair, but DME encodes an enzyme that can specifically remove methylated cytosine. Another DNA glycosylase, ROS1, shows more general effects on DNA demethylation on transgenes and at various locations throughout the genome55. The molecular mechanisms for DNA demethylation in mammals have been very elusive, but it seems possible that similar mechanisms could be at play56.

Epigenomics: Arabidopsis has been one of the leading systems in the emerging area of epigenomics research, in large part due to the well-developed genomic resources in this model plant. The Arabidopsis genome is roughly 130 megabases (about the size of Drosophila), and is one of the most well annotated to date. Relative to most animal genomes the Arabidopsis genome is also relatively compact, with small introns, relatively little repetitive DNA, and an average spacing between genes of only around 5 kilobases. This has made whole genome tiling array experiments, as well as whole genome sequencing approaches, much more practical.

The first whole genome profiling of DNA methylation of any organisms came from Arabidopsis researchers using Affymetrix or Nimblegen whole genome tiling arrays to detect DNA that had been immunoprecipitated with a methyl-DNA specific antibody57,58. As expected, these studies showed that the majority of transposons and repeat elements in the genome were methylated, and that this methylation was highly correlated with endogenous siRNAs. Unexpectedly, over 1/3 of the protein coding genes were also methylated, even though it had been previously thought thatArabidopsis genes were mostly devoid of methylation. This methylation was peculiar in that it was restricted to the coding portions of genes that were generally highly expressed, was anticorrelated with siRNAs, and was strictly in a CG sequence context. The function of this methylation is presently unclear, though it has been proposed to function to suppress the initiation of cryptic transcripts within genes, similar to the known role of H3K36 histone methylation in yeast59.

More recently, by coupling the technique of genomic bisulfite sequencing with ultra high throughput DNA sequencing methods using the Illumina Genome Analyzer, researchers have refined Arabidopsis methylation patterns further. Sodium bisulfite converts C to U, but methyl C is protected from conversion. Thus after PCR amplification and sequencing of bisulfite converted DNAs, Cs indicate methylation while Ts indicate unmethylation. Deep sequencing of bisulfite converted DNA was used to establish single nucleotide resolution DNA methylation maps of Arabidopsis, revealing a number of addition interesting properties60,61. For instance, precise and quantitative sequence specificity for methylation conditioned by the three different methylation systems could be readily determined. It also allowed the study of previously inaccessible repetitive elements of the genome such as telomeres and rDNA. These methods can be used in virtually any genome, and once sequencing throughput is increased, it should become practical to utilize these techniques to analyze entire mammalian genomes routinely.

High-resolution mapping techniques are also being utilized to study histone modifications. As one example, whole genome tiling array experiments showed that at least 4,400 Arabidopsis genes (17% of all genes) are associated with H3K27m362, including a wide variety of transcription factors and developmentally important loci. This suggests that the type of Polycomb mediated long-term epigenetic regulation exemplified by the FLC locus plays a major role in a wide variety of developmental and physiological processes. Clearly, high resolution profiling of other histone modifications and chromatin proteins in Arabidopsis will shed additional light on mechanisms of epigenetic inheritance.

Future of Plant and Insect Epigenetics

Epigenetics has evolved from an initially rather peculiar set of poorly understood phenomena to one of the hottest fields of biology. The explosion of epigenetics research came about in part because of the realization that the same epigenetic mechanisms that regulate phenomena such as PEV are involved in the regulation of many genes during development. The process of transcription itself involves a complex set of chromatin interactions and the main function of transcription factors is often to modify the underlying chromatin structure to permit or restrict transcription. A revolution taking place in biology is the widespread use of genomic techniques to study basic cellular processes, and the field of epigenetics is benefiting from this tremendously. The use of tiling microarrays and high throughput sequencing to discover the positions of chromatin proteins and particular histone modifications is becoming commonplace, and as these techniques are further improved they will soon become part of the basic toolkit of most laboratories. Ultimately, we need to understand the complex interplay of the various epigenetic systems and how they converge to effect proper gene control. We also know relatively little about how epigenetic regulatory complexes are targeted. For instance, it is not clear how Polycomb and Trithorax complexes find their target genes, or how DNA methylation is directed to repeats and other sequences. So while we now have a basic understanding about many epigenetic regulatory systems, there is also much to be learned. The future of exciting epigenetics research is bright indeed.

Acknowledgement

The author thanks Steve Jacobsen for assistance and critical review of the manuscript.

Epigenetic References

  1. Lugar, K et al. Nature 389: 251-260 (1997).
  2. Strahl, BD & Allis, CD. Nature 403: 41-45 (2000).
  3. Jenuwein, T & Allis, CD. Science 293: 1074-1080 (2001).
  4. Turner, BM. Bioessays 22: 836-845 (2000).
  5. Reuter, G & Wolff, I. Mol Gen Genet 182: 516-519 (1981).
  6. Rea, S et al. Nature 406: 593-599 (2000).
  7. Nakayama, J et al. Science 292: 110-113 (2001).
  8. Peters, AH et al. Cell 107: 323-337 (2001).
  9. Khorasanizadeh, S. Cell 116: 259-272 (2004).
  10. Schulze, SR & Wallrath, LL. Annu Rev Entomol 52: 171-192 (2007).
  11. Lachner, M et al. Nature 410: 116-120 (2001).
  12. Bannister, AJ et al. Nature 410: 120-124 (2001).
  13. Jacobs, SA et al. EMBO J 20: 5232-5241 (2001).
  14. Grewal, SI & Jia, S. Nat Rev Genet 8: 35-46 (2007).
  15. Schwartz, YB & Pirrotta, V. Curr Opin Cell Biol 20: 266-273 (2008).
  16. Cao, R et al. Science 298: 1039-1043 (2002).
  17. Fischle, W et al. Genes Dev 17: 1870-1881 (2003).
  18. Min, J et al. Genes Dev 17: 1823-1828 (2003).
  19. Ruthenburg, AJ et al. Mol Cell 25: 15-30 (2007).
  20. Li, B et al. Cell 128: 707-719 (2007).
  21. Mavrich, TN et al. Nature 453: 358-362 (2008).
  22. Grewal, SI & Elgin SC. Nature 447: 399-406 (2007).
  23. Aravin, AA et al. Science 318: 761-764 (2007).
  24. Aravin, AA & Bourc’his, D. Genes Dev 22: 970-975 (2008).
  25. Kunert, N et al. Development 130: 5083-5090 (2003).
  26. Wang, Y et al. Science 314: 645-647 (2006).
  27. Kucharski, R et al. Science 319: 1827-1830 (2008).
  28. Zhang, X et al. Nat Struct Mol Biol 14: 869-871 (2007).
  29. Turck, et al. PLoS Genet 3: e86 (2007).
  30. Riggs, AD. Cytogenet Cell Genet 14: 9-25 (1975).
  31. Holliday, R & Pugh, JE. Science 187: 226-232 (1975).
  32. Henderson, IR & Jacobsen, SE. Nature 447: 418-424 (2007).
  33. Chan, SW et al. Nat Rev Genet 6: 351-360 (2005).
  34. Goll, MG & Bestor, TH. Annu Rev Biochem 74: 481-514 (2005).
  35. Finnegan, EJ et al. Proc Natl Acad Sci U S A 93: 8449-8454 (1996).
  36. Jacobsen, SE & Meyerowitz, EM. Science 277: 1100-1103 (1997).
  37. Soppe, WJ et al. Mol Cell 6: 791-802 (2000).
  38. Ooi, SK & Bestor, TH. Curr Biol 18: R174-R176 (2008).
  39. Henikoff, S & Comai, L. Genetics 149: 307-318 (1998).
  40. Lindroth, AM et al. Science 292: 2077-2080 (2001).
  41. Bartee, L et al. Genes Dev 15: 1753-1758 (2001).
  42. Malagnac, F et al. EMBO J 21: 6842-6852 (2002).
  43. Jackson, JP et al. Nature 416: 556-560 (2002).
  44. Lindroth, AM et al. EMBO J 23: 4286-4296 (2004).
  45. Johnson, LM et al. Current Biology 17: 379 (2007).
  46. Cao, X & Jacobsen, SE. Current Biology 12: 1138-1144 (2002).
  47. Wassenegger, M et al. Cell 76: 567-576 (1994).
  48. Mette, MF et al. EMBO J 19: 5194-5201 (2000).
  49. Jones, L et al. Curr Biol 11: 747-757 (2001).
  50. Sijen, T et al. Curr Biol 11: 436-440 (2001).
  51. Hawkins, PG & Morris, KV. Cell Cycle 7: 602-607 (2008).
  52. Yu, W et al. Nature 451: 202-206 (2008).
  53. Kinoshita, T et al. Science 303: 521-523 (2004).
  54. Choi, Y et al. Cell 110: 33-42 (2002).
  55. Gong, Z et al. Cell 111: 803-814 (2002).
  56. Ooi, SK & Bestor, TH. Cell 133: 1145-1148 (2008).
  57. Zhang, X et al. Cell 126: 1189-1201 (2006).
  58. Zilberman, D et al. Nat Genet 39: 61-69 (2007).
  59. Tran, RK et al. Curr Biol 15: 154-159 (2005).
  60. Cokus, SJ et al. Nature 452: 215-219 (2008).
  61. Lister, R et al. Cell 133: 523-536 (2008).
  62. Zhang, X et al. PLoS Biology 5(5): e129 (2007).