1
Introduction
Nucleosomes
are efficient DNA-packaging units. The fundamental
protein unit of the nucleosome is the histone dimer, a simple α-helical
domain possessing a highly basic, curved surface that closely matches
the phosphate backbone of bent duplex DNA. Two copies each of histone
heterodimer, H3/H4 and H2A/H2B, form a histone octamer that is wrapped
with approximately 146 bp of duplex DNA in a left-handed spiral
1,2
(Figure 1). Through extensive electrostatic
and hydrogen-bonding interactions, each histone dimer coordinates
three consecutive minor grooves on the inner surface of the DNA spiral.
The bending of DNA over the protein surface brings the phosphate backbone
of the two strands closer together on the inside of the spiral, narrowing
the major and minor grooves of DNA, while widening the grooves on
the outside. This bent conformation of the DNA duplex, which would
otherwise be energetically unfavorable, is maintained through charge
neutralization from numerous arginine and lysine side chains of the
histones.
Figure 1
Overview of nucleosome architecture. (A) Illustration of H2A/H2B
and H3/H4 heterodimers and how they fit together to form the histone
octamer. (B) Face and top view of the nucleosome structure. For this
and all subsequent molecular representations of the nucleosome, the
high-resolution crystal structure (PDB code 1KX5) was used.
93
A significant consequence of the intimate DNA wrapping around
the
histone core is that it sterically occludes other DNA-binding proteins.
The inhibitory nature of this packaging is used by virtually all eukaryotic
systems to regulate access to DNA. However, nucleosomes on their own
are not static structures but dynamically fluctuate.
3,4
The most probable nucleosome state, captured in crystal structures,
is the fully wrapped structure. However, only a small fraction of
DNA–histone contacts need to be broken for the nucleosome to
partially unwrap. Using restriction enzyme digestion kinetics, Polach
and Widom
5
demonstrated that nucleosomes
partially unwrap and rewrap spontaneously, which they termed site
exposure. This behavior can be quantitatively defined as the site
exposure equilibrium constant, K
eq, which
is equal to the rate of DNA unwrapping that exposes a section of DNA
for protein binding divided by the rate of rewrapping to states where
the site is not accessible for binding. Values of site exposure K
eq range between 10–1 and
10–6 and correlate with the location of the DNA
segment on the nucleosome, with more internal DNA segments being less
accessible.
5−7
Rather than simply a binary regulator, where a DNA
segment is either accessible or completely blocked, this dynamic unwrapping/rewrapping
allows nucleosomes to regulate occupancy of DNA-binding proteins in
a tunable, analog fashion.
In addition to DNA sequence, which
influences both preferred nucleosome
positioning and unwrapping characteristics,
6,8−10
two distinct mechanisms further modulate nucleosome
stability and dynamics. One mechanism involves chemically altering
the histones themselves, which changes the energy landscape of histone–DNA
interactions and therefore greatly increases the dynamic range of
DNA accessibility. These chemical changes can be in the form of post-translational
modifications (PTMs) that can be dynamically added and removed enzymatically,
with the best-studied modifications including acetylation, methylation,
phosphorylation, ubiquitylation, and ADP-ribosylation.
11−13
These marks, plus other more recently appreciated modifications
such as crotonylation, succinylation, and malonylation,
14
have the potential to alter histone–DNA
and histone–histone interactions and thus provide a means for
transiently targeting changes in nucleosome dynamics.
15−17
The chemical nature of histones can also be changed at the
protein
sequence level, where the canonical histones used to package the majority
of the genome are substituted by histone variants.
18,19
Relative to canonical histones, histone variants can have a range
of sequence differences, from just four amino acid differences between
the H3.3 variant and canonical H3.1 to more than 50% sequence divergence
of the centromeric-specific H3 variant CENP-A.
18
Substitutions of histone variants can change multiple histone–DNA
and histone–histone contacts simultaneously and are well-known
for altering characteristics of single nucleosomes and chromatin fibers.
Nucleosomes containing the histone H2A.Z variant, for example, compact
chromatin fibers more readily than those with canonical H2A,
20
and those with the macroH2A variant wrap DNA
more stably,
21
whereas nucleosomes substituted
with another variant called H2A.Bbd (for Barr body deficient) do not
allow chromatin to readily condense
22
and
wrap DNA much more poorly.
23
Histone variants
can also receive PTMs, which can further modulate the unique effects
that variants have on nucleosome dynamics.
18,24
Complementing chemical and sequence changes to histones, a
second
mechanism for influencing nucleosome stability and dynamics is through
factors that reorganize nucleosome structure. Although many factors,
such as linker histones and DNA-binding proteins, can influence characteristics
such as DNA wrapping and fiber compaction, the greatest changes in
nucleosome structure and stability are achieved by histone chaperones
and ATP-dependent chromatin remodelers. These chromatin reorganizing
factors are instrumental in catalyzing changes in nucleosome structure
and dynamics that would otherwise be thermodynamically inaccessible
under physiological conditions. Histone chaperones are highly acidic
proteins that stabilize histones in the absence of DNA and thus play
critical roles in nucleosome assembly and disassembly.
25−27
Chromatin remodelers are essential for altering the composition
and position of nucleosomes by coupling ATP hydrolysis to exchange
of histone variants, nucleosome sliding, and octamer assembly/disassembly.
28−30
In conjunction with histone variants, PTMs not only alter intrinsic
dynamics of nucleosomes but also provide chemical signposts to help
guide cellular factors to particular locations in the genome. Through
recruitment of cellular factors that bind to PTM-marked histones,
termed the “histone code,” PTMs play an essential role
in defining and maintaining functionally distinct regions of the genome.
11,13,31,32
Histone chaperones and chromatin remodelers bind to and sense PTMs
as well, and in many cases the specificity of their activities can
be traced to PTM-dependent interactions.
27,33−38
This review focuses on PTMs that stimulate structural and
dynamic
changes in nucleosomes (Table 1). We begin
by discussing PTMs that have been shown to directly affect either
histone–DNA or histone–histone contacts in nucleosome
core particles and modulate intrinsic nucleosome dynamics. These PTMs
are located on both the tails and core of the histone octamer and
affect either the unwrapping dynamics or core stability of the nucleosome.
The second part of the review describes PTM-dependent changes in nucleosomes
that are primarily linked with actions of chromatin remodelers. Most
PTMs appear to aid in localizing particular remodeler activities in
different biological settings. However, new evidence is emerging that
some PTMs also influence the actions or specificity of remodelers.
Table 1
Sites of Post-Translational Modifications
That Stimulate Dynamic Changes in Nucleosomes
modification
intrinsic
effects
chromatin
factors influenced
H3(K4me3)
none
recruits Chd1 remodeler;
excludes NuRD and ATRX remodelers
H3(K9, K14, K18, K23ac)
entry site unwrapping
recruits SWI/SNF, RSC remodelers
H3(K36me2,3)
none
recruits ISW1b remodeler
H3(Y41ph)
entry site unwrappinga
H3(R42me2a)
entry site unwrappinga
H3(T45ph)
entry site unwrappinga
H3(K56ac)
entry site unwrapping
alters SWR1 specificity;
enhances CAF1 binding
H3(S57ph)
entry site unwrappinga
H3(K64me3, K64ac)
nucleosome destabilization
H3(K115ac)
nucleosome destabilization
H3(T118ph)
nucleosome destabilization
stimulates disassembly by
SWI/SNF
H3(K122ac)
nucleosome
destabilization
H4(K16ac)
chromatin fiber destabilization
reduces Chd1, ISWI activities
H4(S47ph)
nucleosome destabilizationa
H4(K77ac)
entry site unwrapping
H4(K79ac)
entry site unwrapping
H4(K91ac)
histone–histone destabilization
H4(R92me)
histone–histone destabilization
H2B(K123ub1) [yeast], H2B(K120ub1)
[human]
chromatin
fiber destabilization
aids FACT histone chaperone
[poly-ADP ribosylation]b
destabilizes histone-DNA
contacts
recruits
Alc1 remodeler
a
Predicted.
b
Not at a defined location.
2
Direct Impact of Histone Post-Translational
Modifications on Nucleosome Dynamics
Nucleosomes must dynamically
change so that DNA binding complexes
can access their binding sites. These dynamic changes, which include
nucleosome unwrapping, rewrapping, sliding, assembly, and disassembly,
involve the formation and/or disruption of interactions within the
interfaces between the DNA, H3/H4, and H2A/H2B components of the nucleosome
(Figure 1). During the past 10 years a significant
number of histone PTMs have been identified that are located within
DNA–histone and histone–histone interfaces.
17,39−41
These histone PTMs are poised to influence the interactions
that stabilize the nucleosome structure by shifting the free energy
difference between the fully wrapped nucleosome and altered nucleosome
structures. This results in a change in the probability that nucleosome
will spontaneously fluctuate into altered states, thereby regulating
DNA accessibility and ultimately DNA processing. The free energy,
ΔG, and its associated equilibrium constant, K
eq, are related by K
eq = e–(ΔG/RT), so a shift in free energy of 1 kcal/mol, for example, results in
a ∼5-fold change in the probability between two states.
While there are a wide range of PTM types,
11
a significant number of the PTMs within nucleosome interfaces are
acetylation and phosphorylation. Lysine acetylation removes a positive
charge, phosphorylation introduces negative charge, and both add steric
bulk. The introduction of a single PTM can reduce the free energy
of nucleosome formation by at least 2 kcal/mol.
42
This implies that a single PTM can increase the probability
of altered structural states by over a factor of 25.
The PTM
location impacts the type of structural fluctuation on
the nucleosome.
43
We focus on four structurally
and functionally distinct regions of the nucleosome: histone tails,
DNA entry/exit region, dyad symmetry axis region, and interfaces between
histone dimers (Figure 2). Histone tails, which
refer to the disordered N-termini, extend out from the nucleosome
core and help stabilize higher-order chromatin structure.
44−48
DNA entry/exit regions coordinate the outermost segments of DNA,
which are the first to detach from the histones during nucleosome
sliding or unwrapping (Figure 2A, red highlight).
In this region the DNA is not strongly bound to the histone surface.
49,50
In contrast, the region around the dyad symmetry axis, which coordinates
the most internal segment of DNA, contains the strongest DNA–histone
interactions
49
(Figure 2A, blue highlight). In addition to DNA–histone interactions,
nucleosome stability also depends on protein–protein interfaces
between the histone dimers. We describe select PTMs between H3/H4
and H2A/H2B that appear positioned to weaken the histone–histone
interface. Overall, the primary level of discussion in this review
is the nucleosome core particle, as most most studies of histone PTMs
in structured regions of the nucleosome to date have been investigated
by use of mononucleosomes.
Figure 2
Schematic drawings of the nucleosome, highlighting
features that
contribute to nucleosome dynamics. (A) Illustration highlighting energetically
important contacts within the nucleosome. The DNA entry/exit region
(red) has weaker histone–DNA contacts, and the dyad region
(blue) has the most energetically important contacts.
49
Superhelical locations (SHLs) indicate the histone surfaces
where contact is made with the DNA minor groove.
92
The histone surface underneath the DNA at the dyad, where
the major groove faces the histone octamer, is considered SHL-0, and
increasing or decreasing values mark each SHL moving from the dyad
to the two entry/exit regions. (B) Map of histone residues where post-translational
modifications influence nucleosome dynamics.
Determining the direct influence of histone PTMs on nucleosome
dynamics requires the ability to prepare histones that are homogeneously
modified. This technical requirement has recently been accomplished
with multiple methodologies.
51
One approach
is to use a nonsense suppression strategy, where a tRNA synthetase
and tRNACUA are selected to incorporate nonnatural amino
acids during histone expression in bacteria. This methodology is currently
limited to the incorporation of acetyllysines into histone proteins.
52
Native chemical ligation (NCL)
53
and expressed protein ligation (EPL)
54
allow for a traceless connection between two peptides or
a peptide and a recombinant protein. The modifications that have been
incorporated into histones include acetyllysines,
55−57
phosphothreonines,
42
trimethyllysine,
58
and even ubiquitylated lysines.
59,60
One limitation
of this method is that it is restricted to the N- and C-terminal regions
of the histones. However, sequential NCL with multiple peptides allows
for the preparation of fully synthetic histones, where multiple distinct
histone PTMs can be incorporated throughout each histone.
61
Another approach is alkylation of cysteines,
which enables the introduction of methyllysine
62
and acetyllysine
63
analogues
throughout a histone. Although these analogue residues differ slightly
from the structures of native PTMs, which may affect protein recognition,
this labeling approach is significantly easier than chemical ligation
and therefore a more accessible technique. Combined, these methodologies
provide a diverse tool set for preparing homogeneously modified histones
and are necessary for identifying nucleosome characteristics influenced
by specific PTMs.
2.1
Histone Tails Influence
Nucleosome Unwrapping
and Stability
Histone acetylation had been known to influence
transcription even before the nucleosome was identified.
64,65
Long chains of nucleosomes form chromatin fibers that compact into
higher-order structures.
66,67
Removal of histone
tails with trypsin abolishes chromatin folding, implicating histone
tails in formation of higher-order chromatin compaction.
68,69
Histone tail hyperacetylation disrupts nucleosome array folding
70,70,71
and enhances RNA transcription,
71−73
suggesting a direct link between chromatin compaction and transcription.
More recent work using recombinant histones revealed that nucleosome
array compaction required residues 14–19 of the histone H4
tail.
74
In agreement with these findings,
a subsequent study showed that a single histone PTM in this region,
acetylation of H4 at lysine 16, prevents array compaction.
56
An extensive amount of work has gone into understanding
the detailed roles of histone tails and their impact on chromatin
compaction, which is discussed in a number of recent reviews.
44−48
On mononucleosomes, DNA rapidly unwraps and rewraps from the
histone core, providing transient opportunities for protein binding
within the nucleosome
5−7
(Figure 3A). For example,
nucleosomes expose about 30 base pairs of DNA for transcription factor
binding many times a second and then rewrap on the millisecond time
scale.
75,76
Depending on the free energy differences
between the fully wrapped nucleosome and partially unwrapped nucleosomes,
these rapid fluctuations in unwrapping/rewrapping provide varying
exposure and availability of binding sites.
50,77
This reduced availability effectively lowers binding affinities
by decreasing protein binding rates.
75,78
Furthermore,
site exposure by DNA unwrapping also occurs within compacted nucleosome
arrays, suggesting that nucleosomes transiently unwrap even in compacted
chromatin.
79,80
Interestingly, recent single-molecule
measurements have shown that the nucleosome also dramatically increases
the rate at which DNA-binding domains dissociate.
81
While the mechanism behind accelerated dissociation from
the nucleosome is not known, it provides an additional pathway for
regulating protein occupancy at specific DNA sites on the nucleosome.
Figure 3
Types
of nucleosome dynamics that can be affected by PTMs. (A)
DNA unwrapping transiently exposes protein-binding sites that are
buried within the fully wrapped nucleosome. (B) The DNA can slide
relative to the histone octamer and nucleosomes can be disassembled
to expose DNA-binding sites. With unmodified histones, these structural
changes require histone chaperones and chromatin remodeling complexes.
Dyad modifications can enhance both sliding and disassembly. (C) Nucleosomes
can unwrap with the H2A/H2B heterodimer attached to DNA. This type
of structural dynamics could be an initial step for nucleosome disassembly
and H2A/H2B exchange and may be accelerated by PTMs at histone–histone
interfaces.
DNA site exposure can
be influenced by histone tails.
82
A number
of studies have demonstrated that histone
tails interact with nucleosomal DNA and influence DNA unwrapping.
Cross-linking studies show that the histone tails interact with linker
DNA that extends out from the core,
83,84
and removal
or deletion of the histone tails enhances DNA accessibility to transcription
factors and restriction enzymes,
85−88
influences nucleosome positioning,
89
and reduces nucleosome stability.
90
This is likely due to electrostatic interactions
between the positive charge of the histone tails and the negatively
charged phosphate backbone of the DNA. Interestingly, a recent small-angle
X-ray scattering (SAXS) and fluorescence resonance energy transfer
(FRET) study indicated that the H3 tail suppresses nucleosome unwrapping
yet the H4 tail enhances unwrapping,
91
suggesting
nucleosome unwrapping is influenced by histone tails in distinct ways.
The positively charged histone tails, which make minimal intranucleosomal
interactions within crystal structures,
92,93
are thought
to be largely unstructured in solution and are highly dynamic.
94−97
Perhaps due to their accessibility, the histone tails are the most
post-translationally modified regions of the nucleosome,
98,99
and tail PTMs are intimately involved in many aspects of transcription,
DNA repair, and DNA replication.
100−104
Acetylation neutralizes the positive charge
of lysine, which reduces favorable interactions with DNA and likewise
can influence the impact of the histone tails on DNA site exposure.
Histone acetylation increases DNA accessibility to transcription factor
binding within nucleosomes,
85
whereas restriction
enzyme accessibility studies of unacetylated and hyperacetylated nucleosomes
have shown that acetylation increases DNA site accessibility by up
to 2-fold.
105
Single-molecule force spectroscopy
studies of nucleosome arrays have revealed that histone acetylation
decreases the force required to mechanically unwrap nucleosomes,
106
and likewise a separate single-molecule FRET
study indicated that histone acetylation by Piccolo NuA4 increases
nucleosome unwrapping.
107
To date,
however, most studies of histone tail acetylation have
relied on methodologies that prepare nucleosomes that are heterogeneously
acetylated, which makes it difficult to assign which lysines are modified.
In addition to modifications in the tails, hyperacetylated histones
can also contain acetyllysines within the folded portion of the histone
core, which could increase site exposure (see below). Interestingly,
a recent single-molecule study of nucleosomes with the histone tails
containing 12 lysine-to-glutamine mutations that mimic lysine acetylation
did not increase unwrapping.
108
However,
since different histone tails appear to have opposite impacts on nucleosome
unwrapping dynamics,
91
the extent to which
unwrapping dynamics is influenced by tail acetylation remains unclear.
Another complication is that histone tails also influence chromatin
higher-order structure both within and between fibers. Histone tail
interactions with DNA and the histone core are different in extended,
compacted, and aggregated nucleosome arrays.
47,109−111
This suggests that single-nucleosome studies
are relevant for nucleosome dynamics within extended euchromatic fibers
but that the influence of histone tails and their PTMs on unwrapping
dynamics within heterochromatin needs to be investigated within nucleosome
arrays. To elucidate how nucleosome unwrapping and DNA site exposure
are influenced by histone tail modifications, future quantitative
studies will need to be performed with nucleosomes containing well-defined
patterns of histone PTMs in chromatin fibers.
2.2
Histone
Post-Translational Modifications within
the Nucleosome DNA Entry/Exit Region Directly Influence Unwrapping
Dynamics
There are a number of histone PTMs that are near
the DNA entry/exit region of the nucleosome (Figure 2B) and positioned to potentially
interrupt the DNA–histone
interface, including H3(Y41ph), H3(R42me2a), H3(T45ph), H3(K56ac),
and H3(S57ph). These histone PTMs are involved in a range of biological
processes including transcription,
112−116
DNA repair,
117
replication,
118,119
and apoptosis.
120
Each of these PTMs are under the nucleosomal DNA, so distortion
of the nucleosome structure is required for protein binding to any
of these PTMs within the nucleosome (Figures 4 and 5). This positions them to directly
impact
nucleosome structure or dynamics. H3(K56ac) significantly increases
nucleosome unwrapping,
52
while the impact
of the other PTMs has yet to be reported. In addition, H3(K56ac) influences
nucleosome assembly by altering the interaction of H3/H4 with histone
chaperones. The studies of H3(K56ac) suggest that entry/exit PTMs
both increase site exposure to enhance DNA accessibility and alter
H3/H4 histone chaperone binding to regulate nucleosome assembly.
Figure 4
View of
the nucleosome (1KX5), with H3(K56) and H3(S57) highlighted in yellow.
Located under the DNA near the nucleosome entry/exit region, H3(K56ac)
increases site exposure by increasing the DNA unwrapping rate
52,61
and influences histone chaperone binding,
126
while H3(S57A) substitution interferes with octamer formation and
increases H2A/H2B dimer exchange.
90
Close-up
view (bottom) shows the two sides of the nucleosome superimposed,
with one copy of each histone in color and one copy in gray. In the
crystal structure, H3(S57) hydrogen bonds with neighboring H3(E59)
(magenta dotted line) and makes van der Waals contacts with carbonyl
oxygens of the αN-helix of histone H3 (gray spheres) but is
too far to make direct interactions with DNA. Although the neighboring
H3(K56) makes a closer approach to DNA, the lysine side chain is too
distant to directly hydrogen-bond to the phosphate backbone. The two
positions observed for the H3(K56) side chain suggest some mobility,
and the small gray spheres highlight the shortest path from the lysine
to the closest DNA phosphates.
Figure 5
View of the nucleosome (1KX5), with H3(Y41), H3(R42), and H3(T45) highlighted in
yellow. These residues are located right where the DNA enters and
exits the nucleosome. Close-up view (bottom) with the two sides of
the nucleosome superimposed shows the same rotamers for H3(Y41) and
H3(T45) and two different conformations for H3(R42). In this structure,
both H3(R42) and H3(T45) make direct hydrogen bonds to the DNA phosphate
backbone (magenta dotted lines). Phosphorylation of Y41 or T45 would
be expected to cause steric clashes and electrostatic repulsion with
the DNA. The direct impact of PTMs at these positions on nucleosome
dynamics has yet to be reported, but they are predicted to increase
DNA unwrapping on the basis of studies of H3(K56ac).
H3(K56ac) is the most studied of the entry/exit
PTMs. This PTM
has been incorporated into histone H3 by exploiting the pyrrolysyl-tRNA
synthetase/tRNACUA pairs that were evolved to incorporate
acetyllysines.
52
This study found that
H3(K56ac) does not alter chromatin compaction and only slightly impacts
(by 20%) chromatin remodeling by SWI/SNF and RSC. However, they determined
with single-molecule FRET distribution measurements that H3(K56ac)
increases by 7-fold the amount of DNA that is unwrapped at the nucleosome
entry/exit region.
52
Interestingly, this
study found that there was no measurable increase in unwrapping further
into the nucleosome at the 27th base pair. In contrast, a separate
study, which prepared H3(K56ac) by sequential NCL, found that this
PTM increased transcription factor occupancy at a site extending 27
base pairs into the nucleosome by 3-fold due to enhanced unwrapping.
61
The results of these two studies can be understood
by the observation that nucleosomes iteratively unwrap from the DNA
entry/exit region, and therefore a reduction in unwrapping free energy
at the edge of the nucleosome also enhances DNA site exposure further
into the nucleosome.
50
This property of
the nucleosome also results in cooperative binding of transcription
factors to adjacent binding sites within the nucleosome.
121,122
In addition, H3(K56Q), which is often used to mimic lysine acetylation,
produced a similar increase in transcription factor occupancy,
61
indicating that this mimic effectively captures
the impact of the acetylation on nucleosome unwrapping. Subsequent
studies, which used rapid association kinetics of transcription factor
binding to indirectly measure unwrapping rates, found that the shift
in equilibrium toward partially unwrapped nucleosomes is due to an
increase in the unwrapping rate.
78
Somewhat
counter to expectations, the X-ray crystal structure of a nucleosome
containing the H3(K56ac) mimic did not reveal distortions at the histone–DNA
interface.
123
Thus, although H3(K56ac)
does not prevent formation of a fully wrapped nucleosome structure,
this modification shifts the site exposure equilibrium toward partially
unwrapped states, enhancing nucleosomal DNA accessibility.
H3(K56ac)
is involved in nucleosome assembly during DNA replication
118
and repair
117
and
disassembly during transcription.
112−114
The location of H3(K56),
both adjacent to H2A in the nucleosome and underneath DNA at the entry/exit
site, is positioned to impact the interactions that could influence
nucleosome formation (Figure 4). An indirect
effect of H3(K56ac) on nucleosome formation is evident from altered
interactions of H3 with histone chaperones, which bind H3/H4 tetramers/dimers
and H2A/H2B dimers. A pulldown study found that H3(K56ac) increased
the affinity of the H3/H4 tetramer to the human histone chaperone
CAF-1,
118
while a study of the steps of
nucleosome assembly by the histone chaperone Nap1 found that H3(K56ac)
reduced the affinity of the tetrasome to DNA relative to tetrasome–Nap1
binding.
124
A magnetic tweezers study of
unmodified and H3(K56ac)-containing nucleosome arrays detected no
difference in DNA–histone binding following mechanical disruption
of nucleosomes.
43
NMR and isothermal titration
calorimetry (ITC) studies of H3/H4 binding to the yeast histone chaperone
RTT106 found that H3(K56ac) significantly enhanced binding to a double
pleckstrin-homology (PH) domain,
125
while
fluorescence studies found that H3(K56ac) increases H3/H4 binding
to the yeast histone chaperone CAF-1.
126
These studies suggest that H3(K56ac) influences assembly/disassembly
through interactions between H3/H4 tetramers and histone chaperones.
Following nucleosome assembly, H3(K56ac) enhances DNA site exposure
in the entry/exit region without significantly increasing the propensity
for dissociation of the H2A/H2B dimer.
61
Other histone PTMs in the DNA entry/exit region—H3(Y41ph),
H3(R42me2a), H3(T45ph), and H3(S57ph)—have not been extensively
studied. H3(Y41), H3(R42), and H3(T45) are located at the N-terminal
end of the αN-helix of H3, which extends between the two gyres
of DNA, while H3(S57) is at the C-terminal end of the same helix,
adjacent to H3(K56) (Figures 4 and 5). As for H3(K56ac), these PTMs could affect assembly/disassembly
by altering interactions with histone chaperones and have the potential
of impacting nucleosome site exposure by disrupting histone–DNA
interactions and increasing unwrapping. H3(Y41) is phosphorylated
by JAK2, implicated in transcriptional regulation
127
and anticorrelated with the occupancy of HP1α.
115
Phosphorylation of H3(T45) in human cells is
carried out by PKCδ
120
and DYRK1A,
116
is enriched in apoptotic cells,
120
and regulates HP1 binding.
116
In budding yeast, H3(T45) is phosphorylated by the S-phase
kinase Cdc7-Ddf4.
119
The level of this
modification peaks during DNA replication, while loss of H3(T45ph)
causes replicative defects. Interestingly, the double mutant H3(T45A,K56R)
caused a significantly slower growth phenotype type than either of
the single mutations.
119
Since phosphorylation
at these positions would introduce additional negative charge near
the phosphate backbone, these PTMs could significantly enhance unwrapping;
however, this remains to be directly tested.
Also on the αN-helix
of H3, H3(R42) was recently reported
by Allis and co-workers
128
to be asymmetrically
dimethylated in human cells by CARM1 and PRMT6. They used expressed
protein ligation (EPL) to prepare H3(R42me2a) and found that this
modification increased in vitro transcription from a chromatinized
template by 2.5 times. They hypothesize that this increase is due
to disruption of DNA–histone interactions. However, it is not
known whether this effect directly results from increased unwrapping
or if this PTM acts via nucleosome-interacting factors.
At the
C-terminal end of the H3 αN-helix, H3(S57) has been
shown to be phosphorylated during mitosis.
116
This phosphorylation, which appears to be carried out by the DYRK1A
kinase, antagonizes binding of some HP1 isoforms in human cells and
correlates with transcriptional activation.
116
Although this residue is too distant to directly contact nucleosomal
DNA, H3(S57) hydrogen bonds with H3(E59) and packs against carbonyl
oxygens of the H3 N-terminal helix, which would likely be destabilized
upon phosphorylation (Figure 4). An H3(S57A)
substitution was found to interfere with octamer formation and increase
exchange of H2A/H2B dimers in the nucleosome,
90
consistent with an important role of this side chain in maintaining
the canonical structure of the nucleosome.
H3(K36) is at the
boundary between the N-terminal tail and the
entry/exit region of the nucleosome (Figure 6). Since it is adjacent to the DNA as
it exits the nucleosome, a
PTM at this site or a protein bound to this site could directly influence
nucleosome unwrapping dynamics. This residue can be either methylated
129
or acetylated
130
in vivo. H3(K36me3) is located within the transcribed regions of
active genes,
100
recruits a number of histone
PTM binding domains (readers),
131
and is
involved in DNA repair, alternative splicing, and transcription
132
(see below). Acetylation of this H3 residue
is located within promoter regions of RNA polymerase II genes.
130
At present, it is not known whether H3(K36ac)
directly influences nucleosome dynamics. A recent study did investigate
nucleosomes containing the trimethyllysine analogue at H3(K36) and
showed that methylation alone does not enhance nucleosome unwrapping.
133
However, H3(K36me3) could be specifically bound
by the Tudor domain of Phf1, and in the context of the nucleosome,
this interaction enhanced transcription factor binding through increased
site exposure. These experiments suggested that tethering a protein
domain at the entry/exit site sterically interferes with nucleosome
wrapping and thus provides greater accessibility to nucleosomal DNA.
133
Given these results, it is expected that other
proteins that bind modified H3(K36) would also increase site exposure
and potentially other forms of nucleosome dynamics. Likewise, although
it is not currently known if PTMs on entry/exit residues H3(Y41),
H3(R42), or H3(T45) are specifically bound by other factors, the ability
of H3(K36me3) to increase its own accessibility through the Phf1 Tudor
domain suggests that other binding domains targeted to this region
would also have a significant impact on nucleosome unwrapping. Future
studies looking for and characterizing these types of histone readers
may have the added technical challenge of requiring the full nucleosome
for binding studies, but they will be important for identifying new
factors that take advantage of these entry/exit PTMs to regulate site
accessibility on the nucleosome.
Figure 6
View of the nucleosome (1KX5), with H3(K36) highlighted
in yellow. This residue
is located on the H3 tail just outside where the tail enters between
the two gyres of DNA at the entry/exit region. Close-up view (bottom)
with the two sides of the nucleosome superimposed shows that even
the backbone position of H3(K36) differs between the two copies in
this structure. Neither copy of H3(K36) has the lysine side chain
within direct hydrogen-bonding distance of the DNA backbone. While
modification of this residue does not influence DNA unwrapping,
43
the binding of the Phf1 Tudor domain enhances
DNA unwrapping and accessibility.
35
This
suggests that other histone PTM readers that bind in the entry/exit
region may also alter nucleosome unwrapping/rewrapping dynamics.
Increased DNA unwrapping from
the nucleosome has also been observed
with acetylation of two residues, H4(K77) and K4(K79),
43
which are located approximately 50 bp from the
entry site, near superhelical location (SHL) ± 2.5 (Figure 7). These residues have
been found to be acetylated
in metazoans
39
and map to a region originally
identified with yeast genetic screens that cause a loss of rDNA silencing
(LRS), which include the point substitutions H4(R78G), H4(K79M), and
H4(T80A).
134
By use of in vivo chromatin
immunoprecipitation (ChIP), lrs mutants have been
shown to have reduced binding of the yeast silencing information regulatory
(SIR) proteins at loci normally targeted for silencing.
135−137
The Sir2–3–4 complex binds to nucleosomes via Sir3,
and the molecular interactions required for nucleosome recognition
were revealed in a crystal structure of the nucleosome bound to the
Sir3 BAH domain.
138
Interestingly, however,
some positions giving LRS phenotypes, such as H4(K79) and H3(R83),
do not directly contact the Sir3 BAH domain but instead are located
at histone–DNA interfaces. Thus, one interpretation is that
although these substitutions may not directly alter a histone surface
recognized by silencing factors, the increased unwrapping may indirectly
interfere with nucleosome binding or other aspects required for silencing.
This connection between loss of silencing and increased DNA unwrapping
was also supported by the observation that H3(K56Q) substitution leads
to loss of silencing in yeast.
139
Although
it has not been reported whether acetylation of H4(K77) and H4(K79)
occurs in yeast as in metazoans, the acetyl mimic H4(K79Q) but not
H4(K77Q) disrupted telomeric silencing in Saccharomyces
cerevisiae.
140
Recently,
a number of additional histone PTMs have been identified within the
DNA–histone interface along the first 40 base pairs of DNA,
41
and it will be interesting to investigate both
the biophysical and in vivo consequences of these modifications.
Figure 7
View of
the nucleosome (1KX5), with H4(K77) and H4(K79) highlighted in yellow.
These residues are located around SHL ± 3, where histone mutations
result in the loss of rDNA silencing (LRS).
134
Close-up view (bottom) with the two sides of the nucleosome superimposed
shows that H4(K79) can occupy two different conformations yet still
directly hydrogen-bond to the DNA phosphate backbone and that H4(K77)
is too distant to directly hydrogen-bond in this structure. Despite
their location relatively far from the edge of the nucleosome, acetylation
of these two residues increases DNA unwrapping at the entry/exit site,
43
which may underlie their connection to disrupting
transcriptional silencing.
2.3
Histone Post-Translational Modifications near
the Nucleosome Dyad Symmetry Axis Destabilize the Nucleosome
The dyad region around SHL ± 0.5 contains the strongest histone–DNA
interactions
49
(Figure 2A), and PTMs in this region significantly destabilize the
nucleosome.
40
Histone–DNA contacts
between the H3/H4 tetramer and DNA at the nucleosomal dyad are presumably
the first contacts made upon assembly and the last contacts broken
during nucleosome disassembly. There are four histone PTMs near the
dyad symmetry axis: H3(K115ac), H3(T118ph), H3(K122ac), and H4(S47ph)
39,41
(Figure 8). These PTMs do not influence nucleosome
unwrapping, suggesting that PTMs near the dyad function distinctly
from PTMs in the entry/exit region.
43
Figure 8
View of
the nucleosome (1KX5), with H3(K115), H3(T118), H3(K122), and H4(S47) highlighted
in yellow. Located around SHL ± 0.5, these residues are positioned
within the most energetically important histone–DNA contacts.
Close-up view (bottom) with the two sides of the nucleosome superimposed
shows very similar conformations for these residues. H3(T118) generates
a SIN phenotype when mutated and directly hydrogen-bonds to both the
DNA phosphate backbone and another SIN residue, H4(R45) (magenta dotted
lines). Phosphorylation of H3(T118), which would likely disrupt these
energetically important interactions, has been shown to destabilize
the nucleosome, similar to SIN mutations.
142,143,146
This region was first demonstrated to be important for nucleosome
dynamics through genetic screens in S. cerevisiae that identified five SIN (SWI/SNF-independent)
histone mutations.
141
These five separate histone point mutations
in budding yeast partially relieved the reduced transcription of the
HO gene in a SWI/SNF chromatin remodeler mutant.
141
Three of these mutations—H4(R45H), H3(R116H), and
H3(T118I)—reside in the DNA–histone interface near the
dyad symmetry axis. Amino acid substitutions at these positions significantly
increased thermal mobility of nucleosomes,
142,143
and H4(R45H) reduced higher-order chromatin structure.
144
Consistent with weakening histone–DNA
contacts, substitutions at SIN positions H4(R45) and H3(T118) were
found to reduce the nucleosomal barrier to transcription by RNA polymerase
II.
108,145
However, akin to H3(K56Q), SIN substitutions
did not significantly disrupt the wrapped organization of the nucleosome
in crystal structures.
142
Amino acid
substitutions and phosphorylation of H3(T118) significantly
impact nucleosome stability and dynamics. Interestingly, mutation
at this position had the largest impact on HO expression among the
SIN mutations.
141
The phosphorylation mimic
H3(T118E) is lethal in budding yeast, while low-level expression leads
to a loss of rDNA and telomeric silencing.
140
These results suggest that H3(T118) is an essential histone H3 residue.
While little is known about its role in vivo, the impact of this PTM
on nucleosome structure and dynamics has been investigated in vitro.
42,146
By use of nucleosomes containing H3(T118ph) prepared by EPL, these
studies found that this modification reduced the free energy of nucleosome
formation by approximately 2 kcal/mol and increased mobility at a
temperature of 53 °C by 30-fold relative to unmodified nucleosomes.
As measured by restriction enzyme and DNase I digestion, H3(T118ph)
did not increase DNA site accessibility in the entry/exit region of
the nucleosome but did increase DNA accessibility near the nucleosome
dyad. This modification was also found to induce dramatic changes
in the canonical histone–DNA organization that allowed formation
of nucleosome monomers, dimers, and alternative intermediately sized
structures called “altosomes”
146
(Figure 3B). Altosomes have previously been
observed as remodeling products of SWI/SNF
147−150
and are believed to represent disassembly intermediates
150,151
(see below). Accordingly, the disassembly activity of SWI/SNF is
dramatically increased on nucleosomes possessing H3(T118ph).
146
Although further work is needed to elucidate
when and where H3(T118) phosphorylation occurs in vivo, this PTM is
likely short-lived on nucleosomes and coupled to processes that require
nucleosome-free stretches of DNA.
The other phosphorylation
site, H4(S47), is adjacent to both H3(T118)
and the SIN residue H4(R45) (Figure 8). In
budding yeast, the mutation H4(S47E) causes a slow growth phenotype
and an increase in rDNA and telomeric silencing.
140
In human cells, PAK2 kinase phosphorylates H4(S47) and
appears to promote CAF-1-mediated nucleosome assembly.
152
There are no reported studies of the impact
of this histone PTM on nucleosome dynamics. Given its proximity to
H3(T118), however, phosphorylation of H3(S47) could function similarly
to H3(T118ph) and significantly impact nucleosome stability.
H3(K64) has been reported
to be both trimethylated
153
and acetylated.
154
This site is located near SHL ± 1.5, and
although the side
chain does not make direct contact with DNA in crystal structures,
it lies just underneath the DNA duplex and thus does not appear easily
accessible in the fully wrapped state (Figure 9). Trimethylation of H3(K64) is localized
within heterochromatic
regions and repetitive sequences and is significantly reduced during
cell differentiation.
153
Acetylation of
this residue is found in transcription start sites, with other histone
PTMs that activate transcription and with RNA polymerase II occupancy.
154
P300/CBP can acetylate H3(K64) in vitro, and
knockdowns reduced the levels of H3(K64ac) in vivo. In vitro, nucleosomes
containing H3(K64ac) were less stable and enhanced histone eviction
within a transcription assay.
154
Combined,
it appears that the acetylation of H3(K64) may function similarly
to H3(K122ac) to facilitate transcription, while methylation of H3(K64)
may either stabilize nucleosomes or simply regulate the acetylation
of this residue. However, more direct measurements of the methylation
and acetylation of this site are required to determine their impacts
on nucleosome dynamics and stability.
Figure 9
View of the nucleosome (1KX5), with H3(K64) shown
in yellow. Located at SHL ±
2, this position lies underneath the major groove of DNA and would
not be easily accessible to a histone-modifying enzyme in a fully
wrapped nucleosome. Close-up view (bottom) with the two sides of the
nucleosome superimposed shows very different positions of the lysine
side chain, neither within hydrogen-bonding distance of the DNA phosphate
backbone (gray spheres). From the crystal structure, the manner in
which modification of H3(K64) would directly impact nucleosome dynamics
is not clear.
2.4
Histone
Post-Translational Modifications at
Histone–Histone Interfaces Are Poised to Regulate Nucleosome
Stability
In addition to PTMs at the DNA–histone interface
of the nucleosome, a number of histone PTMs are located within interfaces
between H3/H4 tetramers and/or H2A/H2B dimers.
39
The most extensively studied histone PTM located within
a histone–histone interface is the acetylation of H4(K91) (Figure 10), which was first
identified by mass spectrometry
of histones from calf thymus nuclei
39
and
later in budding yeast.
155
In budding yeast,
H4(K91ac) is associated with transcriptionally active chromatin and
is reduced at telomeres. In one study, the mutation H4(K91A) caused
a reduction in telomeric silencing,
155
while
a separate study reported that H4(K91Q) induced a subtle reduction
in rDNA silencing but not in telomeric silencing.
140
This modification is associated with the histone acetyltransferase
and chaperone complex Hat1p–Hat2p–Hif1p, suggesting
that it is acetylated before deposition and involved in regulating
nucleosome assembly. Nucleosomes containing H4(K91A) have enhanced
sensitivity to micrococcal nuclease digestion and a predisposition
to disassemble at lower NaCl concentrations.
155
In addition, human HAT4 acetylates H4(K91) before H3/H4 is assembled
onto DNA.
156
These findings are consistent
with the hypothesis that H4(K91ac) regulates nucleosome assembly,
but more work is needed to clarify the extent that these effects arise
from chaperone-mediated contacts versus direct histone–histone
destabilization.
Figure 10
View of the nucleosome (1KX5), with H4(K91) and H4(R92) highlighted
in yellow.
These residues are located near the center of the histone octamer,
at the interface between H2A/H2B dimers and H3/H4 tetramer, and are
not readily accessible from the exterior of the nucleosome in the
crystal structure. In the far view (top), the backbone of H3/H4 on
the right side is semitransparent so that H4(R92) can be seen. Close-up
view (bottom) with the two sides of the nucleosome superimposed shows
that both copies are in very similar conformations, with each making
direct hydrogen bonds to residues on H2B (magenta dotted lines). Modification
of either H4(K91) or H4(R92) would be expected to interfere with these
hydrogen bonds. H4(K91ac) is involved in nucleosome assembly and may
increase fluctuations or dissociation of H2A/H2B on the nucleosome.
Additional sites of histone PTMs
within histone–histone
interfaces that have been investigated include H4(R92), which can
be methylated, and H3(C110), which can be glutathionylated. In budding
yeast, mutations H4(R92A) and H4(R92K) eliminate telomeric silencing,
suggesting that this modification is important in regulating transcription.
140
Glutathionylation of H3(C110) was recently
reported to occur within proliferating cells and destabilize nucleosomes.
157
Furthermore, there are over 10 additional histone
PTMs that reside near or within histone–histone interfaces
that have been detected by mass spectrometry but remain to be studied.
On the basis of studies of H4(K91ac), histone PTMs within histone–histone
interfaces appear to regulate nucleosome assembly and disassembly.
Assembly/disassembly could be regulated by two nonexclusive mechanisms
where PTMs (i) directly disrupt nucleosome stability and/or (ii) alter
histone chaperone binding. The report that lower concentrations of
NaCl are required to disassemble nucleosomes with H4(K91ac) suggests
that this PTM directly disrupts nucleosome stability. Interestingly,
two of the five SIN mutations identified in genetic screens, H3(E105
K) and H4(V43I), are also located near histone–histone interfaces.
141
In vitro, H4(V43I) only slightly altered nucleosome
thermal mobility,
142,143
suggesting that changes at histone–histone
interfaces do not always directly impact nucleosome stability. In
addition, unmodified nucleosomes have been reported to transiently
breathe where the H2A/H2B dimer transiently dissociates from the H3/H4
tetramer.
158
Although such breathing may
change the extent to which nucleosomal DNA is wrapped, this mode of
structural fluctuation is distinct from site exposure. Given this
dynamic nature of histones in nucleosomes, the degree of structural
fluctuations may be impacted by PTMs at histone–histone interfaces
(Figure 3C). With multiple PTMs at histone–histone
interfaces,
41
several PTMs may work together
or in parallel to influence chromatin remodeling and/or nucleosome
assembly/disassembly. Future work in this area will be essential to
understand the impact of these histone PTMs on nucleosome stability,
dynamics, and assembly/disassembly.
2.5
Nucleosome
Destabilization by Poly(ADP-ribosylation)
A PTM that has
long been associated with chromatin and changes
in chromatin structure is poly(ADP-ribosylation).
159−162
Carried out by poly(ADP-ribose) polymerases (PARPs), ADP-ribosylation
shares some commonalities with, but also has important differences
from, other PTMs. As for other PTMs, single or multiple ADP-ribose
units, polymerized into so-called PAR chains, are recognized specifically
by reader domains, called macrodomains, and are removed by eraser
enzymes called poly(ADP-ribose) glycohydrolases (PARGs).
163,164
However, what distinguishes PARylation from other modifications
is its dominant electrostatic character. With two phosphates for every
ADP-ribose unit, and often many dozens of such units strung together
in branched and unbranched chains, PARylation produces strong anionic
polyelectrolytes that can compete with DNA and RNA binding.
159−162
Although the predominant substrate for PARP-1 is itself,
165
it also modifies histone H1 and the core histones,
particularly H2B, in vivo.
166
In vitro,
both direct PARylation of histones and noncovalent associations with
PAR chains disrupt histone–DNA interactions,
167−169
correlating with early observations of PARP-dependent decondensation
of purified chromatin fibers.
170
In vivo,
PARP activity in Drosophila embryos
was responsible for normal chromatin decondensation, visualized as
chromatin puffs, and robust gene expression upon heat shock.
171
Similarly, in Drosophila S2 cells, the rapid spread of PARP-1 and PAR chains in the
activated
HSP70 locus was shown to be necessary for histone eviction, preceding
transcription by RNA polymerase II.
172,173
The
mechanisms of nucleosome destabilization by PARylation have been difficult
to discern due to technical challenges and complexity of the system.
PARP-1 is a large, multifunctional enzyme regulated through DNA and
protein interactions and post-translational modifications.
174,175
PARylation activity is greatly stimulated by binding to double-stranded
DNA breaks, histones, and polynucleosomes.
175−180
The mechanisms of PARP-1 activation are not currently clear, but
transcription-coupled activity has been shown to require C-terminal
phosphorylation of the histone variant H2A.Z
181
and N-terminal acetylation of histone H2A.
173
Subsequent to PARP activation, the recruitment and stimulation of
PARG enzymes add further complexity to the system, as the lengths
and distributions of PAR chains change dynamically.
182
At sites of DNA damage, PARylation is required for recruiting
the macrodomain-containing chromatin remodeler Alc1 (amplified in
liver cancer 1), which likely plays an active role in destabilizing
and evicting nucleosomes
183−185
(see below). For PARylation,
an intriguing idea is that PAR chains may be important for stabilizing
histones that have been displaced from DNA.
169,171−173
In this model, a dense mesh of PAR chains,
synthesized in situations where rapid and large-scale nucleosome eviction
is occurring (e.g., activated heat shock loci or sites of DNA damage),
could provide a scaffolding to locally maintain a reservoir of displaced
histones. The subsequent breakdown of the PAR chains by PARGs would
then make histones available for redeposition, a process referred
to as histone shuttling.
169
To further
complicate matters, however, PARylation is not only involved in stripping
histones from DNA but also has been shown to be needed to maintain
chromatin structure.
180,186
PARylation has been found to
be important for maintenance of heterochromatin and rDNA silencing,
186,187
and in mammals, the inactive X chromosome is enriched in a macrodomain-containing
H2A variant called macroH2A.
188
Thus, although
great strides have been made, more work will be needed to elucidate
the nature and mechanisms by which PARylation alters chromatin structure.
As this system will likely continue to be challenging to interrogate
in vitro, progress will be greatly aided by application of high-resolution
and real-time cell imaging.
189−191
3
Epigenetic
Guidance of Chromatin Remodelers
as a Means of Determining Nucleosome Stability and Dynamics
In addition to influencing nucleosome dynamics directly, PTMs also
play important roles in reshaping the chromatin landscape by directing
actions of chromatin remodelers. Chromatin remodelers appear to be
involved in most processes that reorganize nucleosomal architecture,
such as nucleosome assembly, disassembly, histone variant exchange,
and histone octamer repositioning.
28−30
These distinct nucleosomal
rearrangements are all driven by a conserved helicase-like ATPase
motor, named Snf2 for the founding SWI/SNF remodeler, that can translocate
along DNA.
192−194
For SWI/SNF and ISWI remodelers, the Snf2
motor has been shown to engage with nucleosomal DNA at an internal
location, approximately 20 bp from the nucleosomal dyad.
195−199
At a simple level, shifting DNA past the histone octamer is presumably
achieved by translocation of the Snf2 motor on nucleosomal DNA while
maintaining contact with some portion of the histone octamer. However,
recent single-molecule studies with ISW2 unexpectedly revealed that
DNA appears to exit the nucleosome before more DNA is pulled onto
the histone core.
200
Although the details
remain to be worked out, these results suggest that the canonical
structure of the nucleosome is likely distorted during the remodeling
process.
Whereas the ATPase motor plays a central role in powering
the structural
reorganization of nucleosomes, the outcome of the remodeling reaction
appears to be guided by auxiliary domains outside the Snf2 motor.
194
The number and types of auxiliary domains and
subunits vary extensively among different remodeler families.
201−203
Many domains are recognizable for a direct connection to reading
and modifying the histone code. Associated reader domains that can
recognize unmodified, acetylated, and methylated lysines include bromodomains,
chromodomains, PHD (plant homeodomain) fingers, BAH (bromodomain-associated
homology) domains, ADD (ATRX–DNMT3–DNMT3L) domains,
and PWWP (Pro-Trp-Trp-Pro) domains.
32,131,204,205
One or more of these
lysine-recognition domains are commonly found in chromatin remodeling
complexes, either covalently attached to the Snf2 motor or as part
of non-covalently-associated subunits.
30,203
Another type
of reader domain is the macrodomain, which recognizes ADP-ribosylation
and appears unique to the Alc1 remodeler.
183,184
These reader domains can help localize remodeling activities
to
chromatin containing particular PTMs. The SWR1 remodeler, for example,
is believed to be preferentially recruited to nucleosomes with acetylated
H2A and H4 tails through its bromodomain-containing subunit Bdf1.
206
While individual reader–PTM interactions
are generally weak (micromolar K
D), many
remodelers possess multiple reader domains, suggestive of multivalent
interactions that would increase both affinity and specificity.
207,208
One example of a multivalent reader is the BPTF subunit of NURF,
an ISWI-type remodeler, which preferentially binds nucleosomes possessing
both H3(K4me3) and H4(K16ac) marks through a PHD–bromodomain
module.
209
Reader domains can also increase
specificity through preferences for unmodified residues at specific
positions, alone or in combinations with other PTMs. Methylation of
H3(K4) interferes with recruitment of the multisubunit NuRD remodeler,
210
and similarly, the DPF3b subunit of the BAF
remodeler
211
and ADD domain of the ATRX
remodeler,
212
which recognize H3(K14ac)
and H3(K9me3), respectively, require unmodified H3(K4) for binding
to the H3 tail.
The sections below give examples where PTMs
help direct actions
of chromatin remodelers. Each example describes the known activities
of particular remodelers and how these activities fit into the biological
contexts in which the PTMs are found. As an efficient packaging medium,
chromatin is relatively stable on its own and therefore requires active
intervention by chromatin remodelers to rapidly alter nucleosome position,
occupancy, and composition. In many cases, PTMs stimulate changes
in chromatin through binding and recruitment of remodelers, with the
specificity of the remodeling reaction dictated by the type of remodeler
that is recruited. In addition to recruitment, PTMs can influence
remodeler specificity by changing the chemical nature of histones,
which can both affect intrinsic nucleosome dynamics and how remodelers
engage with their nucleosome substrates.
3.1
Nucleosome
Disruption through Recruitment
of SWI/SNF Remodelers
One of the earliest connections between
PTMs and chromatin remodeling was made with the discoveries that the
SWI/SNF remodeler and acetylation accompany transcriptional activation.
From work in budding yeast focused on genes required for mating-type
switching (SWI) and growth on sucrose (SNF, for sucrose nonfermenting),
subunits of the SWI/SNF remodeler were first identified as transcriptional
activators
213−216
that participate in altering chromatin structure.
217
The discovery that SWI/SNF-related remodelers stimulated
transcription by disrupting chromatin structure
218−222
spawned several exciting areas of exploration that continue to this
day: what are the mechanisms underlying chromatin remodeling, in what
cellular processes are remodelers involved, where do they act, and
how are remodelers specifically localized? Just on the heels of finding
SWI/SNF to be a chromatin remodeler was the discovery that the GCN5
transcriptional activator is an acetyltransferase that modifies nucleosomes
to regulate transcription,
223,224
which provided the
foundation for modern epigenetics.
Histone acetylation has since
been shown to stimulate chromatin remodeling activities by SWI/SNF
remodelers.
225−228
This stimulation stems from acetylation-dependent targeting to nucleosome
substrates via bromodomains, which primarily read out the acetylation
state of histone H3.
227−231
Although directly recognizing epigenetic modifications, SWI/SNF
remodelers have also long been known to be recruited to chromatin
through sequence-specific DNA-binding factors.
232−236
These distinct mechanisms for remodeler recruitment not only provide
an environmentally sensitive means for focusing SWI/SNF action but
also likely underlie the dynamic cycles of SWI/SNF-stimulated nucleosome
disruption followed by re-establishment of the chromatin barrier.
237−239
SWI/SNF remodelers are perhaps best known for their roles
in disrupting
chromatin structure, an important step for transcriptional activation.
240−242
By reorganizing histone–DNA contacts, SWI/SNF remodelers
promote binding of sequence-specific factors that would otherwise
be occluded by nucleosomes.
218,219,222,243
This increased access to DNA
arises from the ability to reposition or slide nucleosomes along DNA.
One defining characteristic of SWI/SNF remodelers is that they are
insensitive to whether DNA flanking the nucleosome is available to
accept new positions of the histone core.
244,245
On mononucleosome substrates, this insensitivity allows movement
of nucleosomes up to 50 bp “off the ends” of DNA, which
corresponds to translocation of DNA ends up to the internal location
of the Snf2 motor on the nucleosome, approximately 20 bp from the
dyad.
196,197,244,245
This insensitivity has two consequences. First, nucleosomes
can be shifted greater distances, thus exposing larger stretches of
DNA that were previously occluded by histones (Figure 11A). Second, this insensitivity
allows nucleosomes to be shifted
into each other. Nucleosomes become unstable when positioned too closely
together, and the active collision of neighboring nucleosomes has
been proposed to underlie the disruptive ability of SWI/SNF.
150,151
Figure 11
Changes in nucleosome organization carried out by chromatin remodelers.
(A) Most chromatin remodelers can reposition or “slide”
nucleosomes along DNA. Depending on the direction of sliding, this
repositioning can bury or expose DNA binding sites. Remodelers like
Chd1 and many ISWI-type remodelers are sensitive to DNA flanking the
nucleosome and generate evenly spaced nucleosome arrays (top). In
contrast, other remodelers such as SWI/SNF and RSC can shift nucleosomes
into their neighbors, generating dimeric or altosome structures, which
are believed to be intermediates for nucleosome disassembly (bottom).
The altosome organization depicted here was adapted from a model of
Ulyanova and Schnitzler.
357
(B) Some remodelers
specialize in histone variant exchange. Exchange of canonical and
variant H2A/H2B dimers, highlighted here, is carried out by SWR1 and
INO80.
Consistent with the ability to
shift nucleosomes into each other,
SWI/SNF has been shown to generate altosomes, a noncanonical organization
that resembles a dinucleosome on a short stretch of DNA
147−149
(Figure 11A). Although difficult to study
in vitro, these alternative structures are less stable than canonical
nucleosomes and therefore provide an attractive starting point for
nucleosome disassembly.
148
Despite their
active involvement in increasing nucleosome dynamics, however, SWI/SNF
remodelers likely do not evict nucleosomes single-handedly. Histone
chaperones are essential for maintaining soluble pools of histone
dimers in the absence of DNA and take part whenever histones are deposited
or removed from DNA.
25−27
Remodelers such as SWI/SNF and histone chaperones
are therefore ideally suited to work together: remodelers can disrupt
the histone architecture, either directly or through collisions with
neighboring nucleosomes, and chaperones may both increase the ease
of ATP-dependent disruption and provide efficient acceptors for displaced
histones. Indeed, although the RSC remodeler can displace histone
dimers and weakly transfer octamers its own in vitro,
246,247
nucleosome eviction is greatly aided in the presence of the NAP1
histone chaperone.
248
How exactly these
two systems work together, though, and how the remodeling reactions
are resolved, is far from clear. On its own, RSC promotes loss of
the H2A/H2B dimer during transcriptional elongation by RNA polymerase
II; however, inclusion of NAP1 under identical conditions actually
prevented dimer loss and effectively stabilized the nucleosome.
249
Thus, whereas SWI/SNF remodelers possess an
intrinsic ability to destabilize nucleosomes and increase nucleosome
dynamics through repositioning, the outcome of the remodeling reaction
is dictated by the presence of other chromatin-associated factors,
250
which ultimately rely on DNA sequence and epigenetic
signatures.
3.2
Poly(ADP-ribosylation)
Recruits and Activates
the Alc1 Remodeler
In metazoans, PARP-1 is activated in response
to certain stresses, such as DNA damage and heat shock, and rapidly
poly(ADP-ribosyl)ates (PARylates) itself, histones, and other factors
as an early step in chromatin reorganization and signaling.
159−162
In addition to intrinsically destabilizing nucleosomes by generating
PAR chains (see above), PARP-1 further promotes chromatin reorganization
by recruiting and activating the Alc1 remodeler. Alc1, related to
Chd-type remodelers and also known as Chd1L for Chd1-like, is a monomeric
and relatively small remodeler, with a C-terminal macrodomain being
the only identifiable domain outside of the Snf2 motor.
183,184
DNA- and nucleosome-dependent ATPase activities and nucleosome sliding
ability of Alc1 are stimulated in the presence of PARP-1 and NAD+, the substrate for
generating PAR, but not PARP-1 alone.
This PARylation-dependent stimulation requires the macrodomain, suggesting
that the macrodomain plays a crucial role in localization by binding
to PAR chains.
183,184
Accordingly, Alc1 rapidly colocalizes
with PARylation at sites of DNA damage in cells, and this localization
relies on the PAR-binding ability of the macrodomain.
183,184
It has not yet been demonstrated what remodeling products
are preferentially generated by Alc1, though nucleosome eviction,
observed to accompany PARylation, would be consistent with promoting
disassembly. The related Chd1 remodeler assembles but does not remove
nucleosomes from DNA.
251
However, yeast
Chd1 gained the potential to disrupt nucleosomes when the native DNA-binding
domain was replaced with monomeric streptavidin.
252
This Chd1–streptavidin fusion remodeler shifted
nucleosomes off the ends of DNA fragments and generated SWI/SNF-like
nucleosome products when the histones were biotinylated. In contrast,
when DNA was biotinylated, the remodeler was not disruptive and instead
repositioned nucleosomes on top of the biotinylated DNA sites, which
reduced nucleosome binding affinity. These results suggested that
the Chd1 remodeler could be transformed into a more disruptive remodeler
simply by utilizing recruitment sites that are unaffected by nucleosome
sliding.
252
For Alc1, recruitment via PAR
chains, attached to either histones or PARP-1, may similarly enable
the remodeler to disrupt nucleosomes in a SWI/SNF-like fashion.
3.3
Recruitment of Chd1 to the +1 Nucleosome Helps
Early Elongation of RNA Polymerase II
Promoters of activated
genes typically maintain nucleosome-free or nucleosome-depleted regions
(NFRs/NDRs) flanked by well-positioned nucleosomes, with a nucleosome
that overlaps or is adjacent to the transcription start site (TSS),
called the +1 nucleosome.
253
Two important
epigenetic marks at the promoter/gene body boundary of active genes
are the H2A.Z variant histone and H3(K4me3).
12,13,254−256
In vitro, the H3(K4me3)
mark increases the association of mammalian Chd1 with nucleosomes
and appears to enhance chromatin remodeling toward modified nucleosomes.
258,259
Interestingly, although the H3(K4me3) mark is present in budding
yeast, the yeast Chd1 chromodomains do not maintain the aromatic cage
necessary for binding to this modification,
260,261
and accordingly deletion of Chd1 has little effect on +1 nucleosomes
in yeast.
262,263
As with recognition of most
PTMs, the interaction of the mammalian Chd1 chromodomains with H3(K4me3)
is relatively weak.
260,264
Though the presence of H3(K4me3)
did stimulate transcriptional elongation in a Chd1- and ATP-dependent
manner, the H3(K4me3) mark alone was not sufficient for robust recruitment
of Chd1 in vitro.
259
Instead, recognition
of this mark is one of several interactions that localizes Chd1 to
the transcriptional machinery; protein–protein interactions
have been found between Chd1 and the mediator subunits med1 (S. cerevisiae)
259
or med15
(Schizosaccharomyces pombe),
265
the splicing factor SF3a subcomplex of the
spliceosome,
266
and components of the Paf1,
FACT (Spt16–Pob3), and DSIF (Spt4–Spt5) elongation factors.
267−270
It is presently unclear how localization of Chd1 to +1 nucleosomes
may stimulate transcription in mammalian cells. Decreasing the levels
of active Chd1 in mammalian cells correlated with a small but significant
reduction in histone occupancy and turnover at the promoter, and also
an increased fraction of RNA polymerases that were paused at promoter-proximal
nucleosomes.
271
Nucleosomes are natural
barriers for RNA polymerase II, with distinct pause sites corresponding
to stable histone–DNA interactions on the nucleosome.
49,272−274
Whether the increased pausing from lack
of Chd1 activity is due to a direct or indirect effect is not presently
known. Chd1 can assemble and reposition nucleosomes,
251,275−277
and therefore it could potentially assist
RNA polymerase II directly by disrupting histone–DNA interactions.
It has been demonstrated that several different chromatin remodelers
introduce torsional strain into DNA during the remodeling reaction.
278
Interestingly, nucleosomes were recently found
to be sensitive to torsional strain in single-molecule experiments.
279
By use of a specialized optical trap that allowed
nucleosomal DNA to be both stretched and twisted, the H2A/H2B dimer
was found to be selectively lost at high torque (∼38 pN·nm).
In addition to chromatin remodelers, DNA torque accompanies all enzymes
that translocate along duplex DNA, and significant torque is generated
from transcribing RNA polymerases.
279,280
Although
nucleosomes pose a significant barrier for RNA polymerases, increasing
ionic strength to facilitate transcription through the nucleosome
results in loss of H2A/H2B dimers
281
or
entire octamers.
274
A challenge for factors
assisting RNA polymerase is to therefore coordinate with the mechanics
of transcription to reduce the nucleosomal barrier while maintaining
or reestablishing histones and their marks for subsequent polymerases.
3.4
Post-Translational Modifications in Gene Bodies
Reduce Histone Turnover and Coordinate Chromatin Dynamics with Passage
of RNA Polymerase II
Although RNA polymerase II can transcribe
through nucleosomes and leave the histone octamer intact, it often
stimulates loss of H2A/H2B and sometimes the entire octamer from DNA.
274,281−286
RNA polymerase II is accompanied by several factors, many of which
are elongation factors, that play important roles in maintaining the
chromatin barrier during transcription. Failure to effectively maintain
the chromatin barrier can allow improper transcriptional initiation
at cryptic sites within coding regions, resulting in truncated gene
products and antisense transcripts.
287−289
Also associated with
the elongating polymerase are PTMs, which appear to help establish
and maintain the transcriptionally refractory environment of chromatin
yet also aid polymerase transiently as it transcribes through nucleosomes.
Two marks that play important roles in these processes are monoubiquitylation
of the H2B C-terminus (K123 in S. cerevisiae and K120 in humans) and methylation of
H3(K36).
Di- and trimethylation
of H3(K36) is established in gene bodies by Set2, which interacts
with the phosphorylated C-terminus of elongating RNA polymerase II.
290−293
This methyl mark appears to serve as a basic recruitment signal
for several factors. The Rpd3 histone deacetylase complex recognizes
methylated H3(K36) through a chromodomain-containing subunit, Eaf3,
and its localization to gene bodies reduces acetyl marks to maintain
chromatin in a less transcriptionally accessible state.
294−296
Additional factors associated with the H3(K36me3) mark were identified
by MudPIT (multidimensional protein identification technology) mass
spectrometry, which, in addition to Rpd3 subunits, included several
chromatin remodelers (ISW1a, ISW1b, ISW2, and Chd1) and histone chaperones
(Spt16, Pob3, Spt6, Rtt106, Hir1, Hir2, and Hir3).
297
The ISW1b remodeler, consisting of the Isw1 ATPase subunit
and two auxiliary subunits, Ioc2 and Ioc4,
298
appeared to be the most likely candidate for a direct interaction
with H3(K36me3) due to PHD and PWWP domains on its Ioc2 and Ioc4 subunits,
respectively. Although these subunits failed to directly bind methylated
H3(K36)-containing peptides in vitro and the PWWP domain only modestly
improved interactions with methylated nucleosomes, the PWWP domain
was shown to be required for localization of Ioc4 to gene bodies in
a Set2-dependent manner.
297
Also consistent
with H3(K36me3) recognition, deletion of the ISW1 gene produced a
similar phenotype as set2Δ, with increased
cryptic transcription and histone turnover.
297,299
Both cryptic transcription and increased histone turnover were more
pronounced in an isw1Δchd1Δ background, consistent with colocalization to H3(K36me3)-containing
chromatin and functional redundancy for these two remodelers.
263,297
Although Chd1 failed to specifically recognize methylated H3(K36)
peptides or nucleosomes,
297
the interactions
between Chd1 and elongation factors (FACT,
267−269
DSIF,
269
and Paf1 complex
268−270
) may account for its colocalization to the H3(K36me3) mark.
In the coding region, Chd1 and ISWI-type remodelers may help maintain
the chromatin barrier by several mechanisms. One common characteristic
shared by Chd1 and many but not all ISWI-type remodelers is the ability
to generate evenly spaced arrays both in vitro
251,300−302
and in vivo
263,303
(Figure 11A). Close packing of nucleosomes is expected to
be refractory to transcription initiation, as it would limit the availability
of DNA. Unlike SWI/SNF remodelers, which can shift nucleosomes into
their neighbors to create altosomes and nucleosome-free regions,
147,150,151
Chd1 and ISWI-type remodelers
maintain nucleosomes at a minimum distance from each other. This distinct
remodeling characteristic arises from a preference to shift nucleosomes
toward longer stretches of flanking DNA.
275,276,304,305
Another potential mechanism for reducing cryptic transcription
and reestablishing the chromatin barrier is through nucleosome assembly.
In vitro, both Chd1 and Iswi-type remodelers have been shown to catalyze
the formation of nucleosomes with histones that have been deposited
on DNA but not properly wrapped into nucleosomes.
251,277,300−302
For highly expressed genes where polymerase density is high, most
nucleosomes are evicted, and reassembly by remodelers and histone
chaperones is essential for resetting the chromatin barrier once transcription
levels are reduced.
287,306−308
Another connection between Chd1 and maintenance of the chromatin
barrier in gene bodies is the ubiquitylation of H2B(K123) [yeast numbering,
corresponding to H2B(K120) in humans]. Although no direct link has
yet been made, Chd1 was found to be necessary for high levels of the
H2B monoubiquitin mark.
309
Ubiquitylation
of H2B is a dynamic mark associated with elongating RNA polymerase,
which both regulates histone methylation pathways and participates
in reorganization of chromatin structure.
310
In S. cerevisiae, and similarly in
metazoans, ubiquitin is added to H2B(K123) by Rad6 (a ubiquitin-conjugating
E2 enzyme) and Bre1 (an E3 ubiquitin ligase), a reaction that depends
on the Paf1 complex and transcriptional elongation.
311−316
The monoubiquitylation mark on H2B is required for subsequent di-
and trimethylation of H3(K4) and H3(K79) by Set1/COMPASS and Dot1
methyltransferases
317,318
and also reduces methylation
of H3(K36) by Set2.
319
The presence of
ubiquitin on the H2B C-terminus appears to directly influence nucleosome–nucleosome
packing, as this mark interferes with chromatin fiber compaction.
320
The ubiquitin mark is only present transiently,
however, as deubiquitylation enzymes, such as Ubp8 in S. cerevisiae, also travel along
with the elongating
polymerase complex and dynamically remove ubiquitin from H2B.
321
Preventing either removal of ubiquitin (ubp8Δ) or its deposition [H2B(K123A) substitution]
reduces gene expression, highlighting the importance of the dynamic
ubiquitylation/deubiquitylation cycle for proper transcriptional elongation
and gene expression.
321−323
Intriguingly, beyond just influencing
other PTMs, the ubiquitin
mark on H2B(K123) has been found to affect chromatin dynamics and
passage of RNA polymerase. With a reconstituted chromatin transcription
system, robust transcription was found to require both ubiquitylation
of H2B and the presence of the FACT elongation factor.
316
Although ubiquitylated H2B does not alter transcription
through chromatin on its own, the increased stimulation of transcriptional
elongation in the presence of FACT suggests that the ubiquitin modification
either intrinsically destabilizes the nucleosome or assists destabilization
through action of FACT.
316
In vivo, simultaneous
disruption of FACT and prevention of H2B ubiquitylation (K123A substitution)
produced a transcription-dependent lowering of histone occupancy.
323
Blocking ubiquitylation of H2B with K123A also
disrupted chromatin structure and increased cryptic transcription.
323
Since disruption of Chd1 and ISWI-type
remodelers, FACT, and H2B-ubiquitin
and H3(K36me3) marks increase histone turnover and decrease histone
occupancy, a generally accepted view is that these factors are important
for histone octamer reassembly after passage of RNA polymerase II.
However, the increased histone turnover and decreased occupancy may
instead result indirectly from increased polymerase pausing. After
forward motion of polymerase unwraps some DNA from the nucleosome,
the rewrapping of DNA can trap polymerases in a backtracked state
that prevents nucleotides from being added to the 3′ end of
the growing RNA strand.
274,285,324
On moderately transcribed genes, increased pausing would allow some
polymerases to catch up to others. When traveling in pairs, polymerases
can assist each other in transcribing through nucleosomes, as the
trailing polymerase limits the backtracking of the leading polymerase,
and the leading polymerase helps maintain an unwrapped state for the
trailing polymerase.
324
In contrast to
single polymerases transcribing through nucleosomes, which can leave
the nucleosome intact, multiple polymerases are much more destructive,
presumably because they diminish opportunities for nucleosomes to
re-form histone–DNA contacts.
284,325
A major remaining
challenge is to elucidate the mechanisms by which transcription of
RNA polymerase through nucleosomes can be assisted by chromatin remodelers
and histone chaperones, which help to leave the nucleosome barrier
intact.
3.5
H3(K56) Acetylation Modulates the Specificity
of Histone H2A.Z Variant Exchange by the SWR1 Remodeler
Histone
modifications not only provide platforms for recruitment but also
influence how chromatin remodelers act on their nucleosome substrates.
Recently, the H3(K56ac) mark was shown to dramatically affect histone
exchange activity of the SWR1 remodeler.
326
SWR1, together with INO80, constitute a unique remodeler family
with the defining characteristics of a large (>250 residue) insertion
in the Snf2 motor and tight association with AAA+ RuvB helicase-like
ATPases.
327−331
These remodelers are involved in many aspects of DNA processing
and maintenance, including DNA damage signaling and repair, DNA replication,
telomere maintenance, stability of centromeres, and transcriptional
regulation.
330,331
The unique and defining characteristic
of these remodelers is an ability to exchange H2A variants into and
out of nucleosomes. Multiple variants of histone H2A have appeared
throughout evolution, and SWR1 and INO80 target the only two variants
common to all eukaryotes, H2A.X and H2A.Z
18
(Figure 11B).
Deposition of the histone
H2A.Z variant is tightly integrated into transcriptional activation.
For both transcriptionally active genes and those poised for activation,
H2A.Z histones are specifically deposited into nucleosomes flanking
the promoter NFR.
256,332−335
This deposition relies on the ATPase remodeling action of the SWR1
complex.
328,329
The SWR1 complex is specifically
stimulated by nucleosomes possessing the canonical H2A/H2B dimers
and, with the help of the NAP1 or Chz1 chaperones, replaces these
canonical dimers with H2A.Z/H2B in a unidirectional reaction.
328,336,337
The reverse reaction, where
nucleosomes containing H2A.Z variants are replaced with canonical
H2A/H2B dimers, has been shown to be specifically catalyzed by INO80.
338
Thus, these two remodelers complement one another,
with removal by INO80 important for sharpening the distribution of
H2A.Z deposited by SWR1.
338
The +1
nucleosomes of active promoters are rapidly turned over,
which means that they possess the H3(K56ac) mark indicative of a newly
deposited H3/H4 tetramer.
306−308,339
The high correlation between the H3(K56ac) mark and the presence
of H2A.Z on +1 nucleosomes led Peterson and co-workers
326
to investigate the influence of this mark on
SWR1 activity. Remarkably, it was found that the H3(K56Q) acetylation
mimic disrupts the ability of the remodeler to distinguish H2A from
H2A.Z nucleosomes. Unlike unmodified substrates, where SWR1 did not
exchange H2A.Z/H2B dimers into nucleosomes already containing two
copies of H2A.Z,
337
the presence of the
H3(K56Q) acetylation mimic allowed SWR1 to replace existing H2A.Z/H2B
on the nucleosome with H2A.Z/H2B dimers in solution. Therefore, some
element in SWR1 senses H3(K56) and, when unmodified, effectively blocks
ATPase stimulation and dimer exchange when the nucleosome already
possesses the H2A.Z variant.
326
Two different
H2A.Z-interacting elements in SWR1 have been proposed to link recognition
H2A.Z on the nucleosome with the acetylation status of H3(K56). One
is the Swc2 subunit,
326
which is required
for interacting with H2A.Z/H2B dimers.
340
Loss of Swc2 reduced overall ATPase activity of SWR1 but, importantly,
also prevented preferential ATPase stimulation by H2A-containing nucleosomes
over those with H2A.Z.
326
However, SWR1
complexes lacking Swc2 are severely impaired for H2A.Z exchange
340
and also show some instability of other SWR1
subunits,
340,341
complicating interpretation
of these experiments. Another candidate sensor element is the Swr1-Z
domain, which is N-terminal to the Snf2 motor on the Swr1 subunit.
342,343
This element forms an extended chain that wraps over one end of
the H2A.Z/H2B dimer.
343
Although not yet
tested biochemically, the Swr1-Z domain contacts the H2A.Z/H2B dimer
in a location that would be in close proximity to H3(K56) in the context
of the nucleosome, making it attractive as a potential sensor of H3(K56)
acetylation status.
343
What might
be the purpose in coupling H3(K56ac) with loss of the
ability of SWR1 to discriminate H2A.Z- from H2A-containing nucleosomes?
Like other histones, H2A.Z can acquire PTMs after deposition into
nucleosomes, and the exchange of H2A.Z in nucleosomes with soluble
pools of H2A.Z may be one mechanism of removing those marks. In S. cerevisiae, H2A.Z
can be acetylated by NuA4 and
SAGA at positions K5, K8, K10, and K14,
344−346
and in mammals it can be monoubiquitylated at its C-terminus.
347
These PTMs alter the cellular response to H2A.Z
in chromatin. Acetylation of the H2A.Z tail is important for DNA damaging
sensing. In S. cerevisiae, widespread
H2A.Z deposition, which occurs in the absence of functional INO80,
renders cells highly sensitive to DNA replication inhibitors, DNA
damaging agents, and double-stranded DNA breaks.
327,338,348,349
These sensitivities can be suppressed when the four lysines that
become acetylated are simultaneously mutated to glutamine, serving
as acetyl mimics.
338
Therefore, the DNA
damaging checkpoints that fail to resolve upon mislocalization and
overincorporation of H2A.Z in an ino80Δ background
arise from recognition of nonacetylated H2A.Z.
338
Another mark that would be expected to be disrupted by
exchanging nucleosome-associated H2A.Z with the soluble pool is monoubiquitylation,
which has been observed in mammalian cells.
347
Monoubiquitylated H2A.Z is highly enriched on the inactive X chromosome
and, similarly to H2A, deubiquitination has been found to be required
for gene activation.
347,350,351
The ability for SWR1 to replace these modified forms of H2A.Z therefore
has the potential to have a dramatic impact on H2A.Z-sensitive signaling
pathways. Importantly, however, the H3(K56ac) mark is required for
this behavioral change in SWR1, which limits the SWR1-dependent refreshing
of H2A.Z nucleosomes to those where H3/H4 has been newly deposited,
perhaps as a mechanism for keeping H2A.Z in a more naïve
state.
4
Concluding Remarks
Nucleosome dynamics plays an important role in relieving the intrinsically
inhibitory nature of chromatin and granting access to DNA. Post-translational
modifications (PTMs) can modulate dynamics by directly altering the
energy landscape of the nucleosome and by influencing binding of histone
chaperones and chromatin remodelers. PTMs near histone–DNA
interfaces can promote DNA unwrapping, which increases access for
DNA-binding proteins such as transcription factors (Figure 3A), or can destabilize
the entire nucleosome by
perturbing energetically important contacts near the dyad (Figure 3B). Likewise, PTMs
at histone–histone interfaces
can increase fluctuations of histone dimers in the octameric core
to aid assembly/disassembly (Figure 3C). PTMs
that enhance the specificity of chromatin remodelers may not only
promote nucleosome sliding, disassembly, and histone exchange (Figure 11) but also
direct remodelers to maintain the integrity
of the chromatin barrier in the presence of disruptive machinery such
as RNA polymerase II.
An area in which very little is currently
known is how chromatin
remodelers and core PTMs that affect nucleosome dynamics may work
together or in opposition to alter histone–histone and histone–DNA
interactions. One reported example of functional synergy is between
phosphorylation of the SIN residue H3(T118) and the SWI/SNF remodeler:
each destabilizes the nucleosome on its own, but together they are
much more disruptive.
146
Another PTM that
potentially acts synergistically with remodelers is H3(K56ac), which
increases unwrapping at the entry/exit region
52,61
but also alters specificity of H2A.Z exchange for the SWR1 remodeler.
326
Although this effect on SWR1 specificity may
be due to direct recognition of the acetylation status of H3(K56),
the increased DNA unwrapping due to this or other PTMs may bias exchange
of H2A/H2B dimers. In addition to acting synergistically, increased
DNA unwrapping dynamics due to PTMs could potentially have negative
consequences if, for example, unwrapping were to interfere with actions
of Chd1 and ISWI-type remodelers that are sensitive to DNA around
the entry/exit region.
Compared to modifications on the flexible
histone tails, PTMs on
the histone core are generally less accessible. Core histone PTMs
can be introduced onto histones before assembly into nucleosomes,
such as with H3(K56ac) and H4(K91ac). In the context of the nucleosome,
however, enzymes to add or remove such marks likely require assistance
to access target residues. Some access may be gained through the increased
nucleosome dynamics provided by the core PTMs themselves, which would
be expected to also increase the likelihood that modifications are
added or removed at neighboring residues. The dramatic structural
changes catalyzed by chromatin remodelers would also offer greater
opportunities for adding or removing core PTMs. Indeed, the NuRD remodeler
is named for having a nucleosome remodeling Snf2 motor associated
with histone deacetylase subunits HDAC1 and HDAC2
202
and has also been reported to associate with the lysine
demethylase LSD1.
352
Other types of modifying
enzymes have been shown to associate with remodelers, such as the
kinase-containing WSTF protein that joins ISWI-type remodelers
353,354
and the TIP60 acetytransferase complex that associates with the
SWR1 remodeler.
355
A relatively uncharted
area for future discovery will be to identify what structural and
dynamic changes in the nucleosome are required to write and erase
core PTMs, and how remodelers and other factors directly contribute
to these processes.
In addition to perturbing nucleosome dynamics,
PTMs on the histone
core may also be utilized to recruit other factors. The binding of
reader domains to core histone PTMs would likely stabilize nucleosomes
in altered conformations, further enhancing nucleosome dynamics. While
the face of the histone octamer interacts with protein domains, such
as RCC1
356
and Sir3 BAH domain,
138
there are currently no known factors that bind
histone PTMs in the DNA–histone or histone–histone interfaces.
The apparent lack of such readers could reflect the reduced accessibility
of core PTMs but may also stem from poor reagents to detect such readers.
A continuing technical challenge in studying PTMs on the histone core
is raising good antibodies for chromatin immunoprecipitation. Although
this has been achieved for some modifications, such as H3(K56ac) and
H3(K122ac), the distinct structure and environment of PTMs in the
context of the histone core is not maintained when these PTMs are
presented on peptides. The loss of structural context for core PTMs,
as well as potential steric interference that would reduce binding,
make it difficult to isolate antibodies that robustly recognize these
marks in the context of native nucleosomes.
Since their identification
of histone PTMs within the nucleosome
core in the early 2000s, we have begun to learn how these modifications
can influence nucleosome dynamics. Given the large and growing number
of histone PTMs that have since been identified and the limited number
of core histone PTMs that have been studied, what we have learned
so far may be only the tip of the iceberg.