Leading Edge
e
ar
c
e
i
i
mechanisms mediate the epigenetic inheritance of gene expres-
sion states. These mechanisms can be broadly divided into
mammals, suggest that heritable changes in chromatin structure
DNA replication. Studies using pulse-chase experiments fol-
lowed by fractionation to measure chromatin-bound histones
sive evidence on the role of histone posttranslational modifica-
n
b-
n
of
d(Cavalli and Paro, 1998; Grewal and Klar, 1996; Xu et al., 2006;
Klar, 1998), thus representing systems for inheritance of informa-
tion that may be as powerful as replication of DNA. Whereas cis-
et al., 2007; Grewal and Moazed, 2003; Kaufman and Rando,
2010; Kouzarides, 2007; Rusche et al., 2003; Strahl and Allis,
2000; Suganuma and Workman, 2008) (Figure 1). With somereplication of DNA methylation patterns is well understood,
models for cis-inheritance of histone modifications that are
important differences (discussed later), this model is similar to
how the maintenance DNAmethyltransferase, Dnmt1, is thoughtplay profound roles in maintenance of the expression states of
master regulators such as the homeobox HOX genes (Beisel
and Paro, 2011; Grewal and Moazed, 2003). These studies
further indicate that changes in chromatin states are inherited
in cis through mitotic and even some meiotic cell divisions
tions in the regulation of transcription, these studies have give
rise to the proposal that histone modifications can be re-esta
lished by complexes that recognize a specific modification o
an inherited parental histone and catalyze the same type
modification on adjacent newly deposited nucleosomes (Dodtrans-acting and cis-acting. The first class relies on positive feed-
back mechanisms involving diffusible regulatory factors and
includes transcription factors, such as the phage lambda
repressor (cI) and Cro proteins, and eukaryotic cell type-specific
master regulators, such as the fungal Wor1 and mammalian
myoD proteins (Lassar et al., 1989; Ptashne, 2007; Ptashne
and Gann, 2001; Zordan et al., 2006). The second class of mech-
anisms involves the cis-maintenance of chromatin modifications
or DNA methylation (Beisel and Paro, 2011; Margueron and
Reinberg, 2010; Schaefer et al., 2007). Both types of mecha-
nisms are important for maintenance of gene expression
patterns, but genetic studies of heterochromatin in fungi and
Drosophila, and embryonic development in Drosophila and
strongly suggest that at the bulk level parental histones H3 and
H4 do not exchange with newly synthesized H3 and H4 but
remain bound to the newly replicated daughter DNA strands
(Jackson and Chalkley, 1974) (Figure 1). These studies and
electron microscope images of replicating chromatin further
suggest that during DNA replication parental histones are distrib-
uted randomly between the two daughter DNA strands (Jackson
and Chalkley, 1985; Sogo et al., 1986). More recently, genome-
wide studies in budding yeast using an epitope tag exchange
strategy that allows parental histones to be distinguished from
newly synthesized ones have defined the patterns of parental
histone inheritance, demonstrating histone retention at a gene-
specific level (Radman-Livaja et al., 2011). Together with exten-Mechanisms for the Inh
of Chromatin States
Danesh Moazed1,*
1Howard Hughes Medical Institute, Department of Cell Biology, Harv
*Correspondence: danesh@hms.harvard.edu
DOI 10.1016/j.cell.2011.07.013
Studies in eukaryotes ranging from yeast to mam
can be inherited following DNA replication via me
ment of such chromatin structures and their inh
factors and changes in histone posttranslational m
tenance of epigenetic information in which DNA s
that work cooperatively with parentally inherited h
DNA replication.
Cells with identical genomes can display distinct gene expres-
sion patterns and phenotypes that persist during numerous
cell divisions. This capacity is critical for cellular differentiation
and for development of multicellular organisms with stable
tissues, organs, and morphologies, all of which arise from
a single founder cell, the fertilized egg. The distinct gene expres-
sion and phenotypic states of genetically identical cells, which
develop without change in DNA sequence and persist in the
absence of initial inducing signals, are referred to as epigenetic
states (Gottschling, 2004; Ringrose and Paro, 2004).
Following establishment during embryogenesis, a variety of510 Cell 146, August 19, 2011 ª2011 Elsevier Inc.Perspective
ritance
d Medical School, Boston, MA 02115, USA
mals indicate that specific chromatin structures
hanisms acting in cis. Both the initial establish-
ritance require sequence-dependent specificity
odifications. Here I proposemodels for themain-
lencers or nascent RNA scaffolds act as sensors
stones to re-establish chromatin states following
consistent with the available evidence are lacking. Here I
propose models for cis-inheritance of chromatin states that
provide an explanation for the observation that in addition to
histone modifications, sequence-specific elements such as
DNA silencers and noncoding RNA, whichmediate the establish-
ment of silent chromatin domains, are also required for the main-
tenance of such chromatin structures.
Histone Modification-Based Chromatin Inheritance
Current models of chromatin inheritance are based on experi-
mental evidence on the fate of nucleosomal histones following

to re-establish DNA methylation patterns by preferentially asso-
ciating with and methylating hemimethylated DNA (Holliday,
1987; Schaefer et al., 2007). The model requires that histone
modifications provide sufficient specificity to directly or indirectly
recruit cognate-modifying enzymes and that the kinetics of their
erasure is slower than the kinetics of postreplication re-estab-
lishment. Although in principle this mechanism based entirely
on histones could account for the epigenetic inheritance of chro-
Figure 1. Re-establishment of Epigenetic States from Parental
Histone Modifications
During chromatin replication, parental histones and their posttranslational
modifications are retained and randomly associate with the newly synthesized
daughter DNA strands. Themodifications of parental histones are proposed to
be copied onto newly deposited histones by chromatin modification
complexes that contain a subunit that recognizes the modification on the
parental histone and another subunit that is an enzyme that catalyzes the same
modification on an adjacent nucleosome. Note that distribution of histones to
daughter DNA strands is random. For simplicity, equally spaced nucleosomes
are depicted.matin states, experiments in yeast and flies, discussed below,
suggest that histone modifications alone are not sufficient for
epigenetic inheritance.
Establishment and Maintenance of Silent Chromatin
Domains
Silent or heterochromatic DNA domains in eukaryotic organisms
ranging from yeast to human share a number of central
properties, including their mode of epigenetic inheritance (Beisel
and Paro, 2011; Grewal and Moazed, 2003). Here I briefly
review our current knowledge of how yeast silent chromatin
domains are established and maintained and what these studies
tell us about epigenetic inheritance. In the budding yeast
Saccharomyces cerevisiae, heterochromatin-like silent chro-
matin domains occur at the silent mating-type loci (called HM
loci) and telomeres (Moazed, 2001; Rusche et al., 2003). The
formation of silent chromatin requires input from three different
classes of molecular players (Figure 2A). The first class is spec-
ificity elements. DNA regions, called silencers, direct the
assembly of silent chromatin at the HM loci. Silencers are
composed of binding sites for two general transcription factors,
Rap1 and Abf1, and the origin recognition complex (ORC)
(Bell et al., 1993; Brand et al., 1985; Foss et al., 1993; McNally
and Rine, 1991; Shore and Nasmyth, 1987). At telomeres,
silencing is initiated by tracks of Rap1-binding sites and the
chromosome end, which is bound by the Ku70 and Ku80
proteins (Gasser and Cockell, 2001). Silencer- or telomere-
binding proteins act combinatorially to recruit a second class
of regulators, the Sir1, Sir2, Sir3, and Sir4 proteins, which spread
along the chromatin fiber away from the nucleation site and
create modified chromatin domains that are refractory to
productive transcription (Moazed, 2001; Rusche et al., 2003).
The Sir2 and Sir4 proteins assemble together into a heterodimer
that associates with Sir3 to form the SIR complex (Hoppe et al.,
2002; Moazed et al., 1997; Moretti et al., 1994; Rudner et al.,
2005; Strahl-Bolsinger et al., 1997). The Sir1 protein forms
a bridge between silencer-bound ORC and the Sir3 and Sir4
subunits of the SIR complex, which is important for efficient
recruitment (Gardner et al., 1999; Triolo and Sternglanz, 1996)
(Figure 2A). Histones are the third class of regulators. In
particular, the conserved N terminus of histone H4 and lysine
16 within this region are critical for silencing (Johnson et al.,
1992; Kayne et al., 1988). Any model for the mechanism of inher-
itance must take into account the fact that all three classes of
regulators are required for establishment as well as inheritance
of the silent state.
The SIR complex has three different activities, histone deace-
tylation, histone binding, and self-association, which play critical
roles in establishment and maintenance of silent chromatin. The
Sir2 protein is an NAD-dependent deacetylase with preference
for histone H4 lysine 16 (H4K16), the H4 residue that is required
for silencing (Imai et al., 2000; Johnson et al., 1992; Landry et al.,
2000; Tanny and Moazed, 2001). The Sir3 protein binds prefer-
entially to histone peptides (Hecht et al., 1995) and nucleosomes
that contain deacetylated H4K16 (Liou et al., 2005; Onishi et al.,
2007). In addition to interactions with H4, Sir3 binds to the
globular domain of histone H3, around lysine 79, on the surface
of the nucleosome, and methylation of histone H3 lysine 79
Cell 146, August 19, 2011 ª2011 Elsevier Inc. 511