Cell, Vol. 111, 771–778, December 13, 2002, Copyright 2002 by Cell Press
Signaling Network Model of Chromatin
Stuart L. Schreiber1,3 and Bradley E. Bernstein1,2
Department of Chemistry and Chemical Biology
and Howard Hughes Medical Institute
Harvard University
12 Oxford Street
Cambridge, Massachusetts 02138
2
Department of Pathology
Brigham and Women’s Hospital
Boston, Massachusetts 02115
1
We suggest that common principles underlie both cellular signaling networks and chromatin. To exemplify
similarities, we focus on signaling complexes that form
at membrane receptors and on nucleosomes. Multiple
signal-transducing modifications on side chain residues of receptor tyrosine kinases (RTKs) and histone
proteins are used to create docking sites that facilitate
proximal relations of enzymes and their substrates.
We argue that multiple histone modifications, like RTK
modifications, promote switch-like behavior and ensure robustness of the signal, and we compare this
interpretation with the histone code hypothesis. This
view provides insight into chromatin function and epigenetic inheritance.
Introduction
The concurrent molecular characterization of two proteins in 1996, one that catalyzes the attachment and the
other the removal of acetyl groups on lysine residues in
the tails of histone proteins (Brownell et al., 1996; Taunton et al., 1996), helped focus attention on an area of
research recognized by Allfrey three decades earlier (Allfrey et al., 1964). This new focus resulted in the discoveries of numerous regulatory enzymes that function by
catalyzing modifications of the side chains of residues
found in the histone tails and, more recently, in a region
of the histone core itself (Lacoste et al., 2002; Ng et al.,
2002; van Leeuwen et al., 2002; for review see Cheung
et al., 2000; Grunstein, 1997; Zhang and Reinberg, 2001).
The result of these studies is a new layer of our understanding of gene regulation, especially concerning the
role of chromatin as a regulatory element rather than a
passive structural scaffold.
Signal transduction research during the past decade
has largely focused on the mechanisms used by extracellular factors, their receptors, intracellular mediators,
and transcription factors to integrate information at enhancers and promoters. Chromatin provides the means
for memory and inheritance of these signals. The mechanisms that transcription factors use to mediate these
effects through chromatin are only now being illuminated.
Modern models of signal transduction have incorporated properties of networks, including bistability, ro3
Correspondence: sls@slsiris.harvard.edu
Review
bustness, and adaptability (Bhalla and Iyengar, 1999;
Ferrell, 2002; Freeman, 2000). Bistability, or switch-like
behavior, results from feedback loops in a network. For
example, positive feedback due to the processive actions of an enzyme can enable small changes in signal
intensity to have an all-or-nothing effect. Robustness
results from both feedback loops and the use of redundant mechanisms to propagate a signal. Bistability and
robustness enable a network to respond to varying signal intensities and to adapt to a fluctuating cellular or
extracellular environment. The mechanisms used to
propagate signals have implications for the behavior of
the network. A network that uses multiple redundant
intermediary signals that converge on a single endpoint
can send a robust signal whose output intensity can be
modulated. In contrast, a network that uses multiple
intermediary signals that each mediate different endpoints can send qualitatively different messages (Fambrough et al., 1999). Alternatively, a network can, in theory, use different combinations of a given number of
intermediary signals to mediate a larger number of outcomes.
As our understanding of chromatin function increases,
parallels with signaling networks are becoming evident.
To illuminate the signaling network properties of chromatin, we highlight similarities between the principles
used by receptor tyrosine kinases to relay extracellular
signals and those used by histone proteins to relay nuclear signals. These include: (1) signal-receiving docking
sites used to create high effective molarities between
enzymes and their substrates at targeted sites, (2) dense
collections of such docking sites at the membranes and
on chromatin, both apparently functioning to ensure
robustness of the signal, and (3) analogous signalprocessing topologies including feedback loops that
promote bistability, robustness, and adaptability. We
also extend these parallels to tubulin proteins that, like
histone proteins in chromatin, undergo self-association
to form higher order structures (microtubules). This analysis proposes striking parallels to signaling networks
and offers insight into the functions of and interrelationships between the multiple posttranslational histone and
tubulin modifications.
Histone Modifications
In the mid 1960s, histone proteins were shown to be
subject to multiple posttranslational modifications (Allfrey et al., 1964). 10 years later, the fundamental unit of
chromatin, the nucleosome, was discovered and found
to comprise 146 base pairs of DNA wrapped around an
octamer of histones (two each of H2A, H2B, H3, and
H4) (Kornberg and Thomas, 1974). More recently, enzymes that catalyze the acetylation (HATs), deacetylation (HDACs), methylation (HMTs), phosphorylation
(HKs), and ubiquitination of histones were characterized
as regulators of transcription, DNA repair, and replication (Cheung et al., 2000; Grunstein, 1997; Turner, 2002;
Zhang and Reinberg, 2001). Several of the modifications
catalyzed by these enzymes have now been shown to
Cell
772
recruit (Bannister et al., 2001; Lachner et al., 2001; Zeng
and Zhou, 2002) or exclude (Carmen et al., 2002; Nishioka et al., 2002; Zegerman et al., 2002) additional factors
that affect chromatin structure and function.
The acetylation of histones H3 and H4 at specific
lysine residues has been strongly associated with transcriptional activity (Grunstein, 1997). Histones associated with active genes exhibit high levels of acetylation,
while histones associated with repressed heterochromatin are hypoacetylated (Braunstein et al., 1993). Consistent with this paradigm, in general, HATs activate and
HDACs repress transcription. In contrast to acetylation,
histone methylation has alternately been associated
with transcriptional activity or repression, depending on
the specific lysine or arginine residue involved (Zhang
and Reinberg, 2001). For example, whereas methylation
of H3 Lys4 is observed within euchromatin and active
genes, methylation of H3 Lys9 is observed in heterochromatin and repressed genes (Bernstein et al., 2002;
Litt et al., 2001; Noma et al., 2001).
Several observations appeared originally to represent
exceptions to the above generalizations. For example,
H4 Lys12 was found to be acetylated in heterochromatin
in Drosophila and in heterochromatin-like silent loci in
yeast (Braunstein et al., 1996; Turner et al., 1992). Separately, although Lys4 methylation is associated with
transcriptional activity, the H3 Lys4 HMT, Set1, is required for repression at the heterochromatin-like silent
loci in yeast (Briggs et al., 2001; Nislow et al., 1997).
These observations led to speculation that a particular
modification (e.g., Lys4 methylation) can have disparate
effects, depending on the overall histone modification
state (the histone code hypothesis; see below). However, two recent studies suggest alternative explanations for these apparent paradoxes. The first study used
improved reagents to show that all lysines examined in
H3 and H4, including H4 Lys12, are hypoacetylated in
yeast silent loci (Suka et al., 2001). The second found
that H3 Lys4 is relatively hypomethylated in these regions, suggesting that Set1 represses silent loci indirectly (Bernstein et al., 2002).
Multiple Histone Modifications
The previous examples raise the possibility that the consequence of a particular histone modification may be
context dependent. For example, it has been suggested
that H3 Ser10 phosporylation dictates mitotic or transcriptional function, depending on the coexistence of
acetylation on adjacent residues (Cheung et al., 2000).
However, there are caveats to this example of disparate
functions for a single modification. For example, the
mitotic histone kinase Ipl1 has essential nonhistone targets (Cheeseman et al., 2002), and H3 phosphorylation
does not appear to be necessary either for chromosome
segregation in yeast (Hsu et al., 2000) or for chromosome
condensation in cell extracts (de la Barre et al., 2001).
The issue is further complicated by evidence that the
phosphorylation of H3 Ser10 and the acetylation of adjacent residues occur downstream of distinct pathways
and only transiently exist in combination (Thomson et
al., 2001).
There is evidence that functional complexity is increased
by interplay between various histone modifications. This
is illustrated by, among others, the interdependence of
Lys9 methylation and Ser10 phosphorylation on histone
H3 (Rea et al., 2000). These modifications, which exert
opposite influences on transcription, are mutually exclusive: the Lys9 HMT is unable to methylate an H3 tail that
is phosphorylated at Ser10, and the Ser10 HK is unable
to phosphorylate an H3 tail that is methylated at Lys9.
Additional examples of interplay are reviewed by Turner
(2002).
Histone Code Hypothesis
The number, variety, and interdependence of histone
modifications led to the histone code hypothesis (Strahl
and Allis, 2000; Turner, 2000), which predicts that “multiple histone modifications, acting in a combinatorial or
sequential fashion on one or multiple histone tails, specify unique downstream functions” (Strahl and Allis,
2000). The latter prediction (that modifications act sequentially), as will be discussed, is supported by several
findings. However, the former prediction (that modifications act in a combinatorial fashion to specify unique
downstream functions) currently lacks evidence.
A Signaling Network Model of Chromatin
The histone code hypothesis has provided a framework
for the interpretation of many discoveries regarding the
mechanisms and interrelationships of histone modifications. However, many of the same aspects of chromatin
that inspired the histone code hypothesis are extensively paralleled in signaling networks. We are of the
opinion that an alternative interpretation, based on the
underlying principles of signal transduction networks,
provides a useful framework for understanding chromatin function. This signaling network hypothesis suggests
that multiple modifications combine to confer bistability,
robustness, and adaptability. The histone code model
was recently reviewed by Turner (2002). In the following
sections, we present a signaling network model of chromatin.
Commonalities between Histone and RTK Signaling
RTKs are key elements of one of the most well-studied
families of signaling pathways (for review, see Schlessinger, 2000). Here, we consider structural and mechanistic similarities between the platelet-derived growth
factor receptor (PDGFR), a prototypical RTK, and histone proteins. The similarities begin with domain organization; both proteins contain core domain(s) flanked by
unfolded regions (Figure 1). Specifically, the intracellular
portion of PDGFR contains a juxtamembrane region,
two kinase domains separated by an unfolded kinase
insert, and an unfolded C-terminal tail (Claesson-Welsh,
1994). Histones contain a single core domain flanked
by an unfolded N-terminal tail.
The propagation of extracellular signals by PDGF and
nuclear signals by, for example, a HAT involve posttranslational modifications of disordered regions. In the
former case, tyrosines in the PDGFR kinase insert are
modified by transphosphorylation upon receptor dimerization. In the latter, lysines in the N-terminal histone
tails are modified by the HAT upon recruitment by a
transcription factor or other DNA binding protein. In both
examples, the posttranslational modifications create
Review
773
Figure 1. Schematic Depicting Representative RTK and Histone Protein Domain Organizations
Both proteins comprise core domains flanked
by unfolded regions rich in posttranslationally
modified residues.
binding sites that recruit regulatory proteins. The phospho-tyrosines in the RTK insert complete binding sites
recognized by the SH2 domains of several signaling
proteins, including PI3K, Grb2, and Shc (ClaessonWelsh, 1994). Similarly, the acetyl-lysines in the histone
tails complete binding sites recognized by the bromodomains of several proteins, including TAFII250, GCN5, and
Swi2/Snf2 (Hassan et al., 2002; Hudson et al., 2000;
Jacobson et al., 2000; Owen et al., 2000; Zeng and Zhou,
2002). In addition, methylation of a lysine in the tail of
histone H3 (Lys9) completes a binding site for the chromodomain of HP1 (Bannister et al., 2001; Lachner et al.,
2001). The analogy between the recruitment of PI3K by
RTK phosphorylation and the recruitment of Swi2/Snf2
by histone acetylation is particularly intriguing from a
chemical perspective (Figure 2). Since PI3K (a lipid kinase) and Swi2/Snf2 are both enzymes, these signals
Figure 2. RTK- and Histone-Mediated Signal Transduction Mechanisms Are Analogous
Signal transduction at the cell membrane is initiated by PDGFinduced receptor dimerization, leading to transphosphorylation of
tyrosines in the kinase insert. Signal transduction in the nucleus is
initiated by a HAT that acetylates lysines in the H3 tail. In both cases,
the signal is relayed by recruitment of enzymatic activity to newly
formed docking sites. As illustrative examples, at the membrane,
phosphatidylinositol-3,4,5-triphosphate (PIP3) is generated, locally
activating Akt; in the nucleus, nucleosomes are remodeled, locally
facilitating transcription.
lead to the localization of enzymatic activity. In the former case, localized activity increases phosphatidylinositol-3,4,5-triphosphate (PIP3) concentrations at the cell
membrane, thus locally activating Akt. In the latter case,
localized chromatin remodeling facilitates local activation of transcription. In both cases, the signal, via the
docking mechanism, results in a high effective molarity
of the enzyme and its substrate.
Although most modification sites are located in unfolded tails and inserts, certain residues within the RTK
core kinase domains are also subject to modification.
Similarly, a modification site at Lys79 within the histone
H3 core has been identified (Lacoste et al., 2002; Ng et
al., 2002; van Leeuwen et al., 2002). Modifications within
the RTK core domains, rather than serving as docking
sites, regulate intrinsic kinase activity (Mohammadi et
al., 1996). Analogously, although the mechanism is unknown, the location of Lys79 within a groove that may
accommodate tails from adjacent nucleosomes suggests that it may regulate the formation of higherordered nucleosome arrays (Luger, 2002).
Redundancy in Signaling Networks and Chromatin
Insight into the functions of the multiple phospho-tyrosine docking sites in an RTK was provided by a study
by Fambrough et al. (1999). Using fibroblasts harboring
a chimeric PDGFR, these investigators dissected the
roles of several modifiable tyrosines in propagating an
RTK-mediated mitogenic extracellular signal. Specifically, mutant receptors lacking these tyrosines (and
therefore unable to recruit the downstream signaling
proteins PI3K, PLC␥, SHP2, or RasGAP) were examined
for their ability to propagate signals. Induction of immediate-early genes (IEGs) was used as a downstream
readout (Figure 3).
Fambrough and colleagues considered two extreme
alternatives by which RTKs could communicate their
signals: “either distinct ‘modules’ of IEGs are induced
by specific pathways, or each IEG depends on no single
pathway, but rather receives input from every pathway.”
Using high-density arrays, they found that a mitogenic
signal induces 66 IEGs in cells harboring a receptor with
its natural repertoire of tyrosines. Remarkably, when the
various modifiable tyrosines were replaced with nonmodifiable phenylalanines, nearly all of these IEGs could
still be induced, though with somewhat lower amplitude.
In a separate study, several kinases downstream of
PDGFR (Src, Yes, and Fyn) were found to be largely
dispensable for PDGF signaling (Klinghoffer et al., 1999).
Hence, multiple modification sites and signaling path-
Cell
774
Figure 3. Signal Transducing Modifications in PDGFR and Histone
H4 Are Redundant
As demonstrated in systematic profiling studies of IEG induction,
the multiple tyrosines in the RTK insert are redundant (Fambrough
et al., 1999). Similarly, lysine residues in the H4 tail that are subject
to acetylation are redundant, at least with respect to induction of
GAL1 (Durrin et al., 1991). In both cases, rather than specifying
unique downstream function, multiple, modifiable residues appear
to ensure transduction of a robust signal (see text).
ways, at least in the case of this RTK, converge on
the same target to modulate signal intensity and confer
robustness, rather than act on different targets to mediate multiple different endpoints.
Although no profiling studies have systematically examined the roles of the various modifiable histone residues, there is nonetheless strong evidence for a similar
redundancy in histone tails. One example involves the
lysines in the H4 tail that, as originally reported by
Grunstein and colleagues (Durrin et al., 1991), are required for activation of the GAL1 promoter in media
containing galactose (Figure 3). Replacement of a single
acetylatable lysine at residue 5, 8, 12, or 16 in H4 does
not abrogate GAL1 induction. However, replacement of
three or four of these residues with arginine results in
5- or 50-fold reduced induction, respectively. While the
structural and functional details of induction remain unclear, these data indicate an integral but redundant role
for the H4 tail lysines in propagating the galactose signal.
A second example of redundancy involves mitotic
phosphorylation of histone H3 by the aurora kinases.
Phosphorylation of histone H3 at Ser10 correlates
closely with chromosome condensation, and yeast lacking the aurora kinase Ipl1 exhibit segregation defects
(Chan and Botstein, 1993). However, yeast harboring a
mutation at H3 Ser10 do not exhibit similar defects (Hsu
et al., 2000). A likely explanation for this surprising result,
reviewed in Cheung et al. (2000), comes from the finding
that Ipl1 also phosphorylates histone H2B. Moreover, in
Tetrahymena, which lacks phosphorylatable serines in
the H2B tail, H3 Ser10 is necessary for proper segregation (Wei et al., 1999). Hence, the phosphorylatable serines in the H3 and H2B tails appear to be functionally
redundant and, in contrast to the relevant kinase, are
not individually required for mitosis and chromosome
segregation in yeast.
Methylation of histones at a variety of lysine and arginine residues is a complex and poorly understood process. It will be important to determine whether there are
pairs or groups of methyl marks that, like analogous
acetyl and phosphoryl marks, act in a functionally redundant manner. In support of this possibility are biochemical studies that suggest that the transcriptional activator
CARM1 methylates histone H3 at Arg17 and Arg26, and
that the Polycomb-group repressor E(z) methylates histone H3 at Lys9 and Lys27 (Cheeseman et al., 2002;
Kuzmichev et al., 2002; Schurter et al., 2001). Overall,
there is good evidence for functional redundancy among
posttranslational modifications to histone proteins.
Functional redundancy in RTK signaling and other cellular networks is thought to enable the transduced signal
to be modulated quantitatively and to confer robustness
(i.e., the response is not dependent on any single pathway and can be elicited under a variety of conditions)
(Fambrough et al., 1999). We suspect that functional
redundancy among histone modifications serves an
analogous role in chromatin.
Feedback Loops in Signaling Networks
and Chromatin
In the frog oocyte MAP kinase cascade, positive feedback loops lead to bistability and ensure an all-or-nothing cell fate decision (Ferrell, 2002). In RTK signaling, a
positive feedback (processivity) feature of Src kinases
confers bistability (Stover et al., 1995; Zhou and Cantley,
1995). Src enzymes contain, in addition to their kinase
domain, an SH2 domain that binds selectively to Srcphosphorylated products. After it phosphorylates tyrosines within the RTK, Src binds to the RTK. This processivity increases effective Src concentrations, leads
to phosphorylation of nearby signaling proteins and results in effective propagation of signal (Figure 4A).
We point out two examples of positive feedback in
chromatin that closely parallel those in RTK signaling
(Figures 4B and 4C). The first involves SUV39H1, the
H3 Lys9 HMT, and HP1, a chromodomain protein that
specifically binds H3 tail methylated at Lys9 (Bannister
et al., 2001; Lachner et al., 2001). SUV39H1 and HP1
colocalize and interact biochemically. Hence, like the
Src kinase, the SUV39H1-HP1 complex both modifies a
substrate (Lys9) and binds the product (methylated
Lys9). Concomitant increase in local enzyme concentration should lead to methylation of H3 Lys9 in nearby
nucleosomes, and perhaps of other tail lysines. Since
Lys9 methylation is a repressive event, this processivity
should cause the chromatin to adopt an off state. A
second example involves a class of enzymes that activate transcription, the bromodomain-containing HATs.
Since bromodomains bind acetylated lysines, their presence within HAT proteins such as TAFII250 and GCN5
should promote processive acetylation and, therefore,
effect a chromatin on switch (Hudson et al., 2000; Jacobson et al., 2000; Kouzarides, 2000; Owen et al., 2000).
Several recent reports provide direct evidence for the
processive functions of chromatin modifying complexes
(Cao et al., 2002; Czermin et al., 2002; Hassan et al.,
2002; Muller et al., 2002; reviewed in Turner, 2002).
Double-negative feedback, another network motif
that can promote bistability (Ferrell, 2002), also appears
Review
775
modifications on the same histone tail, modifications on
one tail can affect those on adjacent tails (Dover et al.,
2002; Sun and Allis, 2002; reviewed in Henry and Berger,
2002; Turner, 2002). Although it is unclear how these
trans-tail effects are mediated, they most likely also result in feedback loops and processivity. The complex
interactions observed among modifications on the same
and adjacent histone tails likely form part of an extensive
network that dictates the structure and function of chromatin.
Figure 4. Feedback Mechanisms in Chromatin
A classic example of feedback in signal transduction is the processivity of Src kinases. The Src SH2 domain has affinity for the phosphorylated product of its kinase domain. As depicted in (A), this results
in positive feedback, which promotes switch-like behavior in RTK
signaling. Positive feedback in chromatin signaling is mediated by
enzymes that contain chromodomains (B) and bromodomains (C).
Double-negative feedback, which also promotes switch-like behavior and robustness in a network, is observed in the interplay between
Lys9 methylation and Ser10 phosphorylation of histone H3 (D). For
discussion of the relevance of Figures 4B and 4C to epigenetic
inheritance, see text. (E) TDAC is unique among deacetylases in
that it has two deacetylase domains. One of these domains may be
used for binding, resulting in either negative or positive feedback.
to act in chromatin. A prime example emerges from
biochemical studies that demonstrate mutual exclusivity
of H3 modifications at Lys9 and Ser10 (Rea et al., 2000).
The Lys9 HMT (SUV39H1) is unable to methylate an H3
tail that is phosphorylated at Ser10. Conversely, the
Ser10 HK (Ipl1) is unable to phosphorylate H3 tail that
is methylated at Lys9. Phosphorylation of Ser10, in addition to having a mitotic function, is associated with transcriptional activity, while methylation of Lys9 is repressive. Hence, in the context of a signaling model, the
interrelationship between these modifications should invoke switch-like behavior in chromatin activation and
repression.
Another interrelationship that may produce bistability
involves modifications to H3 tail residues Lys4 and Lys9.
Methylation of Lys4 is an activating event and hence
influences chromatin in a manner opposite that of Lys9
methylation. Methylated Lys4 appears to prevent deacetylation of Lys9 by excluding the NuRD complex
(Nishioka et al., 2002; Zegerman et al., 2002). Since an
acetylated lysine is not a substrate for methylation (Rea
et al., 2000), this function of methylated Lys4 has the
end effect of preventing Lys9 methylation. Conversely,
there is evidence that methylated Lys9 prevents Lys4
methylation (by Set7), thereby completing the basis for
another double-negative feedback loop in chromatin
(Wang et al., 2001).
In addition to interrelationships that occur between
Role of Multiple Modifications in Signaling
Networks and Chromatin
Although studies suggest that redundancy is an underlying feature of RTK signaling networks, there are also
examples where specific downstream pathways (e.g.,
PI3K and Src signaling downstream of Met) are required
for appropriate cellular response to an RTK-mediated
signal (Madhani, 2001; Maina et al., 2001). A recent study
by Agalioti and colleagues may have identified an analogous situation in chromatin (Agalioti et al., 2002). These
investigators followed acetylation patterns within the
IFN-␤ promoter during IFN-␤ gene induction. Although
several lysines in the tails of histones H3 and H4 were
modified during gene induction, specific residues (e.g.,
H4 Lys8, H3 Lys9, and H3 Lys14) appear to be required
for recruitment of the bromodomain-containing proteins, Brg1 and TAFII250, and for induction of IFN-␤ to
wild-type levels.
The principles of redundancy and specificity are compatible with a signaling network model, and each is likely
to govern certain aspects of chromatin function. The
role of combinatorics, where multiple modifications are
read out in combination, thereby enabling a single modification to exert disparate effects depending on its context, has yet to be determined. Although this combinatorial transmission of information has the potential to
mediate an extraordinarily large number of endpoints,
it alone will not impart robustness or other features of
networks. Histone modifications are extremely complex
and poorly understood at this time. Just as in studies
of RTKs, it will be important to determine the relative
contributions of redundancy, specificity, and combinatorics to chromatin function.
Despite the many parallels between chromatin and
other cellular networks, histones appear to be subject to
a considerably wider variety of chemical modifications
than other signaling proteins. We speculate that this is
a remnant of a more primitive biological system from
which the highly conserved elements of chromatin have
evolved. For example, although prokaryotes lack histone proteins, they do contain HDAC orthologs (Finnin
et al., 1999; Frye, 2000). Additional support for this view
can be argued from similarities between chromatin and
microtubules. Both structures are vital for eukaryotic
cell division and may have coevolved (Kaczanowski and
Jerzmanowski, 2001). Like histones, the building block
of microtubules, tubulin, is subject to a wide array of
posttranslational modifications including phosphorylation, tyrosination, detyrosination, acetylation, and deacetylation (the latter via a tubulin deacetylase, TDAC
[Hubbert et al., 2002]). Indeed, parallels between histones and tubulin are extensive (Strahl and Allis, 2000).
Cell
776
Both proteins constitute building blocks of higher-ordered
structures: histones form nucleosomes that associate
into polymeric structures; tubulins form the polymeric
microtubules. These higher-ordered polymers become
physically linked during mitosis at the centromere. It is
especially intriguing that the recently discovered TDAC
(Hubbert et al., 2002), formerly known as HDAC6 (Grozinger et al., 1999), is unique among the family of deacetylases in that it has two deacetylase domains. Only one
of the two TDAC domains is catalytically active in the
deacetylation of tubulin (S.J.H., K.M.K., and S.L.S., unpublished data). In the polymerized microtubule, multiple TDAC substrates (Lys40 of ␣ tubulin) reside within
reasonable proximity. We speculate that the catalytically
inactive domain may function as a binding domain that
recruits the catalytically active domain to tubulin,
thereby leading to either negative or positive feedback
(Figure 4E). In either case, microtubule structure-function is another area where signal transduction paradigms will likely provide valuable insight.
Nonetheless, the implications of the diverse nature
and large number of chemical modifications to which
histones are subjected should be considered. One set
of modifications in RTKs is responsible for regulating
numerous processes all at once: transcription, translation, and secretion. Distinct (but likely overlapping) sets
of modifications in histone proteins are responsible for
regulating numerous processes but at different times:
transcription during interphase, DNA replication during
S phase, and chromosome segregation during mitosis.
In all cases, it is likely that the features of networks
leading to redundancy, robustness, feedback, and
switch-like behavior are essential. The acquisition of
these features requires multiple modifications–and at
multiple times–for histone proteins.
Epigenetic Inheritance in a Chromatin
Signaling Network
Despite its mechanistic and functional similarities to
other signaling networks, chromatin retains the unique
ability to encode epigenetic information via the maintenance of histone modification patterns. The feedback
principles presented in our earlier discussions and in
Figure 4 play important roles in maintaining these modification patterns. For example, in nondividing cells,
chromodomain-containing HMT complexes and bromodomain-containing HATs bind to and propagate heterochromatin- and euchromatin-associated modifications,
respectively (Cosma et al., 1999; Grewal and Elgin, 2002;
Hassan et al., 2002). The importance of these processive
functions is highlighted by recent observations that
newly synthesized histones are continuously deposited
at active loci and must therefore be posttranslationally
marked in an appropriate fashion (Ahmad and Henikoff,
2002). Epigenetic memory can also persist through cell
division (reviewed in Lyko and Paro, 1999; Turner, 2000;
Wolffe and Matzke, 1999). For example, DNA methylation patterns are maintained during DNA replication
through the actions of a DNA methylase that recognizes
hemi-methylated duplex DNA (reviewed in Wolffe and
Matzke, 1999). During DNA replication, histone octamers segregate randomly onto the two emergent duplexes (reviewed in Wolffe and Matzke, 1999). Thus, a
given duplex should contain histone octamers with an
established modification pattern and an equal number
of just-deposited histone octamers lacking this pattern.
The docking capabilities of the original histone octamers
likely direct the enzymes required to replicate their
modification patterns onto adjacent histone octamers
(Grewal and Elgin, 2002; Hassan et al., 2002). Again, a
key property of these enzymes and complexes is their
domain structure–they have a protein binding domain
that recognizes the product of their enzymatic domain
(Figure 4). This feature leads to processive enzymatic
activity and, in this case, epigenetic memory (in nondividing cells) and inheritence (in dividing cells).
Future
We hope that the signaling network model will provide
an additional framework for interpreting results concerning chromatin function. Additional experimentation will
be required to determine which elements of the histone
code and signaling network hypotheses are most illuminating, whether they are compatible with each other, and
whether additional hypotheses will be required before
a true understanding of the underlying principles is at
hand.
Acknowledgments
We would like to thank members of the Schreiber laboratory for
helpful discussions. We are grateful for the many insightful comments provided by the referees during the review process and for
support of our chromatin research by the National Institute for General Medical Sciences. B.E.B. is supported by a K08 Development
Award from the National Cancer Institute. S.L.S. is an Investigator
at the Howard Hughes Medical Institute.
References
Agalioti, T., Chen, G., and Thanos, D. (2002). Deciphering the transcriptional histone acetylation code for a human gene. Cell 111,
381–392.
Ahmad, K., and Henikoff, S. (2002). Epigenetic consequences of
nucleosome dynamics. Cell 111, 281–284.
Allfrey, V.G., Faulkner, R., and Mirsky, A.E. (1964). Acetylation and
methylation of histones and their possible role in regulation of RNA
synthesis. Proc. Natl. Acad. Sci. USA 51, 786–794.
Bannister, A.J., Zegerman, P., Partridge, J.F., Miska, E.A., Thomas,
J.O., Allshire, R.C., and Kouzarides, T. (2001). Selective recognition
of methylated lysine 9 on histone H3 by the HP1 chromo domain.
Nature 410, 120–124.
Bernstein, B.E., Humphrey, E.L., Erlich, R.L., Schneider, R., Bouman,
P., Liu, J.S., Kouzarides, T., and Schreiber, S.L. (2002). Methylation
of histone H3 Lys 4 in coding regions of active genes. Proc. Natl.
Acad. Sci. USA 99, 8695–8700.
Bhalla, U.S., and Iyengar, R. (1999). Emergent properties of networks
of biological signaling pathways. Science 283, 381–387.
Braunstein, M., Rose, A.B., Holmes, S.G., Allis, C.D., and Broach,
J.R. (1993). Transcriptional silencing in yeast is associated with
reduced nucleosome acetylation. Genes Dev. 7, 592–604.
Braunstein, M., Sobel, R.E., Allis, C.D., Turner, B.M., and Broach,
J.R. (1996). Efficient transcriptional silencing in Saccharomyces cerevisiae requires a heterochromatin histone acetylation pattern. Mol.
Cell. Biol. 16, 4349–4356.
Briggs, S.D., Bryk, M., Strahl, B.D., Cheung, W.L., Davie, J.K., Dent,
S.Y., Winston, F., and Allis, C.D. (2001). Histone H3 lysine 4 methylation is mediated by Set1 and required for cell growth and rDNA
silencing in Saccharomyces cerevisiae. Genes Dev. 15, 3286–3295.
Brownell, J.E., Zhou, J., Ranalli, T., Kobayashi, R., Edmondson, D.G.,
Review
777
Roth, S.Y., and Allis, C.D. (1996). Tetrahymena histone acetyltransferase A: a homolog to yeast Gcn5p linking histone acetylation to
gene activation. Cell 84, 843–851.
Cao, R., Wang, L., Wang, H., Xia, L., Erdjument-Bromage, H.,
Tempst, P., Jones, R.S., and Zhang, Y. (2002). Role of histone H3
lysine 27 methylation in Polycomb-group silencing. Science 298,
1039–1043.
Carmen, A.A., Milne, L., and Grunstein, M. (2002). Acetylation of the
yeast histone H4 N terminus regulates its binding to heterochromatin
protein SIR3. J. Biol. Chem. 277, 4778–4781.
Chan, C.S., and Botstein, D. (1993). Isolation and characterization
of chromosome-gain and increase-in-ploidy mutants in yeast. Genetics 135, 677–691.
Cheeseman, I.M., Anderson, S., Jwa, M., Green, E.M., Kang, J.,
Yates, J.R., Chan, C.S., Drubin, D.G., and Barnes, G. (2002). Phospho-regulation of kinetochore-microtubule attachments by the aurora kinase ipl1p. Cell 111, 163–172.
Cheung, P., Allis, C.D., and Sassone-Corsi, P. (2000). Signaling to
chromatin through histone modifications. Cell 103, 263–271.
Claesson-Welsh, L. (1994). Platelet-derived growth factor receptor
signals. J. Biol. Chem. 269, 32023–32026.
Cosma, M.P., Tanaka, T., and Nasmyth, K. (1999). Ordered recruitment of transcription and chromatin remodeling factors to a cell
cycle- and developmentally regulated promoter. Cell 97, 299–311.
Hsu, J.Y., Sun, Z.W., Li, X., Reuben, M., Tatchell, K., Bishop, D.K.,
Grushcow, J.M., Brame, C.J., Caldwell, J.A., Hunt, D.F., et al. (2000).
Mitotic phosphorylation of histone H3 is governed by Ipl1/aurora
kinase and Glc7/PP1 phosphatase in budding yeast and nematodes.
Cell 102, 279–291.
Hubbert, C., Guardiola, A., Shao, R., Kawaguchi, Y., Ito, A., Nixon,
A., Yoshida, M., Wang, X.F., and Yao, T.P. (2002). HDAC6 is a microtubule-associated deacetylase. Nature 417, 455–458.
Hudson, B.P., Martinez-Yamout, M.A., Dyson, H.J., and Wright, P.E.
(2000). Solution structure and acetyl-lysine binding activity of the
GCN5 bromodomain. J. Mol. Biol. 304, 355–370.
Jacobson, R.H., Ladurner, A.G., King, D.S., and Tjian, R. (2000).
Structure and function of a human TAFII250 double bromodomain
module. Science 288, 1422–1425.
Kaczanowski, S., and Jerzmanowski, A. (2001). Evolutionary correlation between linker histones and microtubular structures. J. Mol.
Evol. 53, 19–30.
Klinghoffer, R.A., Sachsenmaier, C., Cooper, J.A., and Soriano, P.
(1999). Src family kinases are required for integrin but not PDGFR
signal transduction. EMBO J. 18, 2459–2471.
Kornberg, R.D., and Thomas, J.O. (1974). Chromatin structure; oligomers of the histones. Science 184, 865–868.
Kouzarides, T. (2000). Acetylation: a regulatory modification to rival
phosphorylation? EMBO J. 19, 1176–1179.
Czermin, B., Melfi, R., McCabe, D., Seitz, V., Imhof, A., and Pirrotta,
V. (2002). Drosophila Enhancer of Zeste/ESC complexes have a
histone H3 methyltransferase activity that marks chromosomal Polycomb sites. Cell 111, 185–196.
Kuzmichev, A., Nishioka, K., Erdjument-Bromage, H., Tempst, P.,
and Reinberg, D. (2002). Histone methyltransferase activity associated with a human multiprotein complex containing the Enhancer
of Zeste protein. Genes Dev. 16, 2893–2905.
de la Barre, A.E., Angelov, D., Molla, A., and Dimitrov, S. (2001).
The N-terminus of histone H2B, but not that of histone H3 or its
phosphorylation, is essential for chromosome condensation. EMBO
J. 20, 6383–6393.
Lachner, M., O’Carroll, D., Rea, S., Mechtler, K., and Jenuwein, T.
(2001). Methylation of histone H3 lysine 9 creates a binding site for
HP1 proteins. Nature 410, 116–120.
Dover, J., Schneider, J., Tawiah-Boateng, M.A., Wood, A., Dean, K.,
Johnston, M., and Shilatifard, A. (2002). Methylation of histone H3
by COMPASS requires ubiquitination of histone H2B by Rad6. J.
Biol. Chem. 277, 28368–28371.
Durrin, L.K., Mann, R.K., Kayne, P.S., and Grunstein, M. (1991). Yeast
histone H4 N-terminal sequence is required for promoter activation
in vivo. Cell 65, 1023–1031.
Fambrough, D., McClure, K., Kazlauskas, A., and Lander, E.S. (1999).
Diverse signaling pathways activated by growth factor receptors
induce broadly overlapping, rather than independent, sets of genes.
Cell 97, 727–741.
Ferrell, J.E., Jr. (2002). Self-perpetuating states in signal transduction: positive feedback, double-negative feedback and bistability.
Curr. Opin. Cell Biol. 14, 140–148.
Finnin, M.S., Donigian, J.R., Cohen, A., Richon, V.M., Rifkind, R.A.,
Marks, P.A., Breslow, R., and Pavletich, N.P. (1999). Structures of
a histone deacetylase homologue bound to the TSA and SAHA
inhibitors. Nature 401, 188–193.
Freeman, M. (2000). Feedback control of intercellular signalling in
development. Nature 408, 313–319.
Frye, R.A. (2000). Phylogenetic classification of prokaryotic and
eukaryotic Sir2-like proteins. Biochem. Biophys. Res. Commun. 273,
793–798.
Grewal, S.I., and Elgin, S.C. (2002). Heterochromatin: new possibilities for the inheritance of structure. Curr. Opin. Genet. Dev. 12,
178–187.
Grozinger, C.M., Hassig, C.A., and Schreiber, S.L. (1999). Three proteins define a class of human histone deacetylases related to yeast
Hda1p. Proc. Natl. Acad. Sci. USA 96, 4868–4873.
Grunstein, M. (1997). Histone acetylation in chromatin structure and
transcription. Nature 389, 349–352.
Hassan, A.H., Prochasson, P., Neely, K.E., Galasinski, S.C., Chandy,
M., Carrozza, M.J., and Workman, J.L. (2002). Function and selectivity of bromodomains in anchoring chromatin-modifying complexes
to promoter nucleosomes. Cell 111, 369–379.
Henry, K.W., and Berger, S.L. (2002). Trans-tail histone modifications: wedge or bridge? Nat. Struct. Biol. 9, 565–566.
Lacoste, N., Utley, R.T., Hunter, J., Poirier, G.G., and Cote, J. (2002).
Disruptor of telomeric silencing-1 is a chromatin-specific histone
H3 methyltransferase. J. Biol. Chem. 277, 30421–30424.
Litt, M.D., Simpson, M., Gaszner, M., Allis, C.D., and Felsenfeld, G.
(2001). Correlation between histone lysine methylation and developmental changes at the chicken beta-globin locus. Science 293,
2453–2455.
Luger, K. (2002). The tail does not always wag the dog. Nat. Genet.
32, 221–222.
Lyko, F., and Paro, R. (1999). Chromosomal elements conferring
epigenetic inheritance. Bioessays 21, 824–832.
Madhani, H.D. (2001). Accounting for specificity in receptor tyrosine
kinase signaling. Cell 106, 9–11.
Maina, F., Pante, G., Helmbacher, F., Andres, R., Porthin, A., Davies,
A.M., Ponzetto, C., and Klein, R. (2001). Coupling Met to specific
pathways results in distinct developmental outcomes. Mol. Cell 7,
1293–1306.
Mohammadi, M., Dikic, I., Sorokin, A., Burgess, W.H., Jaye, M., and
Schlessinger, J. (1996). Identification of six novel autophosphorylation sites on fibroblast growth factor receptor 1 and elucidation of
their importance in receptor activation and signal transduction. Mol.
Cell. Biol. 16, 977–989.
Muller, J., Hart, C.M., Francis, N.J., Vargas, M.L., Sengupta, A., Wild,
B., Miller, E.L., O’Connor, M.B., Kingston, R.E., and Simon, J.A.
(2002). Histone methyltransferase activity of a Drosophila polycomb
group repressor complex. Cell 111, 197–208.
Ng, H.H., Feng, Q., Wang, H., Erdjument-Bromage, H., Tempst, P.,
Zhang, Y., and Struhl, K. (2002). Lysine methylation within the globular domain of histone H3 by Dot1 is important for telomeric silencing
and Sir protein association. Genes Dev. 16, 1518–1527.
Nishioka, K., Chuikov, S., Sarma, K., Erdjument-Bromage, H., Allis,
C.D., Tempst, P., and Reinberg, D. (2002). Set9, a novel histone H3
methyltransferase that facilitates transcription by precluding histone
tail modifications required for heterochromatin formation. Genes
Dev. 16, 479–489.
Nislow, C., Ray, E., and Pillus, L. (1997). SET1, a yeast member of
the trithorax family, functions in transcriptional silencing and diverse
cellular processes. Mol. Biol. Cell 8, 2421–2436.
Cell
778
Noma, K., Allis, C.D., and Grewal, S.I. (2001). Transitions in distinct
histone H3 methylation patterns at the heterochromatin domain
boundaries. Science 293, 1150–1155.
Owen, D.J., Ornaghi, P., Yang, J.C., Lowe, N., Evans, P.R., Ballario,
P., Neuhaus, D., Filetici, P., and Travers, A.A. (2000). The structural
basis for the recognition of acetylated histone H4 by the bromodomain of histone acetyltransferase gcn5p. EMBO J. 19, 6141–6149.
Rea, S., Eisenhaber, F., O’Carroll, D., Strahl, B.D., Sun, Z.W., Schmid,
M., Opravil, S., Mechtler, K., Ponting, C.P., Allis, C.D., and Jenuwein,
T. (2000). Regulation of chromatin structure by site-specific histone
H3 methyltransferases. Nature 406, 593–599.
Schlessinger, J. (2000). Cell signaling by receptor tyrosine kinases.
Cell 103, 211–225.
Schurter, B.T., Koh, S.S., Chen, D., Bunick, G.J., Harp, J.M., Hanson,
B.L., Henschen-Edman, A., Mackay, D.R., Stallcup, M.R., and Aswad, D.W. (2001). Methylation of histone H3 by coactivator-associated arginine methyltransferase 1. Biochemistry 40, 5747–5756.
Stover, D.R., Becker, M., Liebetanz, J., and Lydon, N.B. (1995). Src
phosphorylation of the epidermal growth factor receptor at novel
sites mediates receptor interaction with Src and P85␣. J. Biol. Chem.
270, 15591–15597.
Strahl, B.D., and Allis, C.D. (2000). The language of covalent histone
modifications. Nature 403, 41–45.
Suka, N., Suka, Y., Carmen, A.A., Wu, J., and Grunstein, M. (2001).
Highly specific antibodies determine histone acetylation site usage
in yeast heterochromatin and euchromatin. Mol. Cell 8, 473–479.
Sun, Z.W., and Allis, C.D. (2002). Ubiquitination of histone H2B regulates H3 methylation and gene silencing in yeast. Nature 418,
104–108.
Taunton, J., Hassig, C.A., and Schreiber, S.L. (1996). A mammalian
histone deacetylase related to the yeast transcriptional regulator
Rpd3p. Science 272, 408–411.
Thomson, S., Clayton, A.L., and Mahadevan, L.C. (2001). Independent dynamic regulation of histone phosphorylation and acetylation
during immediate-early gene induction. Mol. Cell 8, 1231–1241.
Turner, B.M. (2000). Histone acetylation and an epigenetic code.
Bioessays 22, 836–845.
Turner, B.M. (2002). Cellular memory and the histone code. Cell 111,
285–291.
Turner, B.M., Birley, A.J., and Lavender, J. (1992). Histone H4 isoforms acetylated at specific lysine residues define individual chromosomes and chromatin domains in Drosophila polytene nuclei.
Cell 69, 375–384.
van Leeuwen, F., Gafken, P.R., and Gottschling, D.E. (2002). Dot1p
modulates silencing in yeast by methylation of the nucleosome core.
Cell 109, 745–756.
Wang, H., Cao, R., Xia, L., Erdjument-Bromage, H., Borchers, C.,
Tempst, P., and Zhang, Y. (2001). Purification and functional characterization of a histone H3-lysine 4-specific methyltransferase. Mol.
Cell 8, 1207–1217.
Wei, Y., Yu, L., Bowen, J., Gorovsky, M.A., and Allis, C.D. (1999).
Phosphorylation of histone H3 is required for proper chromosome
condensation and segregation. Cell 97, 99–109.
Wolffe, A.P., and Matzke, M.A. (1999). Epigenetics: regulation
through repression. Science 286, 481–486.
Zegerman, P., Canas, B., Pappin, D., and Kouzarides, T. (2002).
Histone H3 lysine 4 methylation disrupts binding of nucleosome
remodeling and deacetylase (NuRD) repressor complex. J. Biol.
Chem. 277, 11621–11624.
Zeng, L., and Zhou, M.M. (2002). Bromodomain: an acetyl-lysine
binding domain. FEBS Lett. 513, 124–128.
Zhang, Y., and Reinberg, D. (2001). Transcription regulation by histone methylation: interplay between different covalent modifications
of the core histone tails. Genes Dev. 15, 2343–2360.
Zhou, S., and Cantley, L.C. (1995). Recognition and specificity in
protein tyrosine kinase-mediated signalling. Trends Biochem. Sci.
20, 470–475.