Nucleosomes

Nucleosome structure. Nucleosomes are the fundamental units of chromatin. They are found in all

eukaryotes (with the exception of the dinoflagellates) and are conserved in structure. When nuclei
are lysed in a low salt solution, the chromatin decondenses and nucleosomes can be observed
resembling beads on a string.
The relationship between DNA and the nucleosome can be investigated using nucleases (Figure
3.2). Limiting digestion of decondensed chromatins eparates the nucleosomes into short fragments
consisting of single nucleosome units (mononucleosomes) and multimers thereof. The DNA associated
with the mononucleosomes is of a constant length, usually in the order of 200 bp (the exact
length is species- and cell-type-specific). Electrophoretic separation of DNA liberated from digested
chromatin thus produceas characteristicl adder of fragments witha periodicity reflecting the length
of the mononucleosomal DNA. This suggests that the nuclease cleaves at the same relative position
between each nucleosome.
In most species, further nuclease digestion reduces the length of DNA isolated from the nucleosomes
to 165 bp. This occurs in a single ssteupg,g esting that the initial cleavage is followed by rapid
trimming of the trailing DNA ends. The DNA that remainsi s protected by the proteins of the nucleosome
particle, which at this stage is termed a chromatosome.
Further nuclease digestion reduces the DNA to 146 bp inl ength, and liberates some protein from
the nucleosome. This length of DNA is conserved throughout the eukaryotes and is relatively resistant
to further nuclease attack. It represents the DNA intimately associated with the proteins which
comprise the nucleosome core. The particlet hus formed, the nucleosome core particle, consists of
a histone octamer containing two copies each of histones H2A, H2B, H3 and H4, and 146 bp of core
DNA which is wrapped around this octamer twice. The coprea rticles are joined together by linker
DNA, whose length varies between speciesa nd between different cell types. In most specie ass, the DNA
enters and leaves the core particle it is associated with a single copy of a linker histone
(histone H1) which is thought to seal the DNA in place.
The consequenceso f nuclease treatment can be summarizeda s follows: initial nucleased igestion
cleaves the linker DNA at a specific point and then trims the ends until the remaining DNA is
protected by histones. Further digestion removes the DNA associated with histone H1, whose loss
suggests that it lies outside the core, but close to it. The remaining DNA is that protected by the
nucleosome core itself. In yeast chromatin, there appears to be no linker histone. Initial cleavage
generates mononucleosome DNA of just 164 bp, which is trimmed in a single step to the 146 bp of
core DNA. The linker DNAb etween nucleosome core particles is thus 18 bp in length, the shortest
linker DNA known. However, histone H1 genes haveb een identified in the yeast genome.
Histones. Histones are highly basic proteins (rich in positively charged lysine and arginine residues)
which fold to form a compact corwe ith a protruding N-terminalt ail. The positiver esidues interact
with DNA by forming salt bridges with the negatively charged DNA backbone, while the tails are
targets for posttranslational modification and facilitate nucleosome-nucleosome interactions, and
interactions with other chromatin proteins. Several hundred histone sequences have been determined
andt hey show remarkable conservation across the eukaryote kingdom.H istones H3 and H4
are the mosts trongly conserved, whilst the linker histones show the most diversity. Altlh e histones
except H4 exist as multiple isoforms (isohistones) whose relative predominance in chromatvina ries
in a cell-type-specific manner. Thlien ker histones show the most isotypic diversity and can be divided
into several subclasses (e.g. H1, H5, HlO).
Histones possess an inherent ability to associate and form various complexes in the absence of
DNA, notably the (H3-H4)2 tetramer and the H2A-HZB dimer. Each histone contains a three-a-helix
motif called ah istone fold which facilitates thesein teractions. The tetramer can organize DNA into
nucleosome-like particlesi n vitro, and if the other histones are added, chromatinw ill assembles
pontaneously.
This occurs slowly, however, and rnolecarlur chaperones (q.v.) are required to help assemble
nucleosomes in vivo (see below). Thset ructure of the histone octamer has been investigated using Xray
crystallography and cross-linking studies, and has recently been solved at a resolution of 2.8 A.
H2A-H2B dimers fit above and below a central (H>H4)2 tetramer, and the cylindrical particle thus
formed possesses an outer surface which describes the superhelical path taken by DNA. The path is
not smooth, however, but distorted because of several bends. As DNA enters and leaves the

octamer, it is bound by extensions of the H3 histone protein.

All histones appear to undergo some form of posttranslational modification, usually on the
N-terminal tail, and in many cases, specific patterns of modification correlate to changes in chromatin
function. The acetylation of histones H3 and H4 is a marker of accessible open chromatin,
although a number of recent experiments suggest the relationship between acetylation and genetic
activity is not a simple one. Histones H3 and Ha1r e phosphorylated and,a t least in the slime mould
Phsyarum, the level of H1 phosphorylation varies in a cell cycle-dependent manner. Various histones
are also methylated, or conjugated with ubiquitin or poly-ADP-ribose, although the significance of
these modifications is unclear. Histone tails are long and unstructured, but ordered by nucleosomenucleosome interactions. The modification of histones is therefore thought to disrupt interactions
between nucleosomes and thus perturbh igher order chromatin structure.
Nonhistone proteins. The histones represent the bulk of protein in chromatin and are relatively
homogeneous in nature. The remaining chromatin proteins, collectively described as nonhistone
proteins, represent a small but extremely heterogeneous fraction. The nonhistone proteins include
enzymes involved in DNA and histone metabolism, replication, recombinationa nd transcriptional
regulation
(q.v.). They also include the scaffold proteins which organize higher order chromatin
structure (as discussed below), and the high mobility group proteins( HMG proteins), which are
highly charged proteins with various functions in gene regulation and structural organization. Of
these, the HMG14/17 family of nucleosome-binding proteins are enriched in active chromatin and
are thought to help decondense higher order chromatin structure. The HMGI/Y family proteins,
whose precise role ins ot known, preferentiallyb ind to repetitive AT-rich DNA, and like histone HI,
they can bep hosphorylated by cyclindependent kinases (see The Cell Cycle). Thper incipal effect of
HMG proteins in packaging and transcriptional activation is to introduce sharp bends into the
DNA. In a packaging context, this may be required for DNA to adopt particular three-dimensional
configurations, whereas in a transcriptional context it mayb ring regulatory factors at different sites
into proximity (q.v. SRYfactor, enhnceosome). A further class of nonhistone proteins, termed protamines,
facilitate the packaging of DNA into the sperm head. These proteins align the major grooves
of adjacent DNA duplexes and fold the DNA into a highly compact array of parallel fibers.
HistoneDNA docking andt he /inking number paradoxD. Nase I or freer adical cleavage of nucleosome
core particles generates DNA fragments with 10-11 bp periodicity, suggesting that the DNA
wrapped round the histone octamer is in the B-conformation (see Nucleic Acid Structure). The
pattern of cleavage varies across the surface of the nucleosome, with more frequent cleavage at the
ends and less frequent cleavage in the middle, indicating that the structure of the DNA changes as
it wraps around the octamer. The DNA is wrapped twice around the core, and the path it follows
should create approximately 1.8 turns of negative supercoil. However, experiments designed to
measure the degree of supercoiling generated by the sequestration of DNA into nucleosomes show
that each nucleosome actually generates just one turn of negative supercoil. The discrepancy is
termed the linking number paradox (q.v. DNA topology) and is explained by the decreased pitch
(number of base pairs per turn) of nucleosomal DNA (10.2) compared with free DNA.
Position of nucleosomes on DNA. Nucleosomes do not form randomly on DNA, but occupy preferential
sites. This property is termed nucleosome phasing, and can be demonstrated by treating
chromatin with micrococcal nuclease and a restriction endonuclease. A discrete DNA fragment is
obtained because the restriction site in the DNA is found at the same position relative to the nucleosome
in each chromatin strand. Nucleosome phasing may occur in two ways, either by positioning
the histone octamer at a particular sequence because its structure is favorable for winding, or by
38 Advanced Molecular Biology
positioning it relative to a particular boundary wheren ucleosomes are excluded. Both types of phasing
have been observed. The tendency of particular DNA sequences to curve influences nucleosome
positioning, as does the presence of other proteins in nucleosome-free regions, e.ga.t promoters and
enhancers. In yeast chromatin, nucleosome phasing is observed in transcriptionally repressed chromatin,
whereas in active chromatin, the nucleosomes adopt random positions. This suggests that
nucleosome phasing may be ainn itial requirement for higher order chromatin structure.
3.2 Higher order chromatin organization
The 30 nm fiber. The winding of DNA into nucleosomes represents only the first level of structural
organization. When nuclei are lysed in a low salt solution, the characteristic beads on a string structure,
termed the 10 nm fiber, represents a packaging ratio of five and lacks histone HI. At a higher

salt concentration, chromatin adopts a more compact structure which requires histone HI. This is
probably a coiled fiber of 2545 nm in diameter (it is termed the 30 nm fiber), although alternative
models suggest a zigzagging organization, or that nucleosomes do not form a regular structure, but
irregular clumps, or superbeads, which are arranged in linear fashion to form a fiber.
A number of different structures have been proposed for the coiled 30 nm fiber based on its
density (6-8 nucleosomes for every nucleosome in the 10 nm fiber - a packing ratio of 40), and On
X-ray diffraction evidence suggesting that the nucleosome discs are packed with their flat faces
parallel to the helical axis, although with a variable degree of tilt. Most controversy surrounds the
role of linker DNA and the position of histone H1, which may form polarized filaments perhaps by
cooperative binding.
Chromatin /OOPS. The gross organization of chromatin in the interphase nucleus is poorly understood,
but it is thought that the 30 nm fiber is attached at various points to the nuclear matrix to
form a series of loops containing 30-100 kbp of DNA (the folded fiber model). Thelo ops can be seen
in scanning electron micrographs of protein-depleted chromatin, with their bases attached to scaffold
proteins of the nuclear matrix (also q.v. lampbrush chromosome). The functional significance of
the loops is not understood, but they may correspond to the chromatin domains discussed below,
which have been identified by genetic and biochemical analysis.
Euchromatin andhetemchromatin.I nterphase chromatin existsin two distinct forms: diffusee
uchromatin,
which is believed to comprise loo3p0e ndm fibers minglinga nd tangling together in the nucleoplasm,
as discussed in the preceding paragraph, and highly condensed heterochromatin, which is
believed to adopt a higher order structure and tends to cluster around the nuclear periphery.
Heterochromatin is usually transcriptionally repressedan, d is assumed to have adopteda higher
order structure similar to mitotic chromatin which excludes transcription regulatory proteins. The
nature of this structure is unknown although the DNA sequence is not critical, for as well as
constitutive heterochromatin (which is condensed in all cells at all times and is often found near
the centromere), eukaryotic cells contain facultative heterochromatin, which is maintained in a
repressed form in some cells but not others (an example is the inactive X-chromosome). It is likely,
therefore, that specific nuclear proteins are involved in the control of heterochromatin structure, and
candidates have been identified in several organisms on the basis that they regulate the related
lThe nuclear matrix (also called then uclear scaffoldi)s a group of structures which remaiinn the nucleus when
.

it has been extracted with detergents, nucleases and high concentrations of salt. It comprises a fibrous
proteinaceous
network which may well subdivide the nucleus into compartments and organize individual chromosomes.
The composition of the matrixis poorly characterized, but idt oes contain topoisomerase1 1, which is
also a component of the central scaffold of the metaphase chromosome, orc hromosome core. There is
tantalizing
evidence for the involvement of chromatin-matrix interactions in the initiation of replication and
transcription,
but therei s much to be learned about the roof let he matrix in DNA function.

Chromatin 39
phenomenon of position effect variegation (see below). Heterochromatin also contains a high proportion
of specifically modified histones and, in higher eukaryotes, correlates with increased levels of
DNA methylation (qv.).
Metaphase chromosomes. Chromatin is most condensed at mitotic metaphase when each chromosome
is packaged as a tiny discrete structure to facilitate segregation. Again, little is known about
the organization of chromatin in this highly compact form (packaging ratio lo5). It is known that
highly condensed fibers are arranged in loops upon a proteinaceous scaffold which forms a central
helix fromw hich the loops of chromatin radiate. It is not known whethetrh ese loops are equivalent
to loops of chromatin in the interphase nucleus although, as discussed below, similar DNA motifs
may be involved in their attachment. The scaffolds of each sister chromatid twist in opposite directions
and appear to be joined together in the initial stages of mitosis.
A remarkable aspect of mitotic chromatin is that even in its highly condensed state it retains a
memory of which genes were transcriptionally active and which repressed in the previous interphase.
DNA is therefore not packaged uniformly in the mitotic chromosome, allowing the specific
arrangement of open and closed chromatin domains to be reinstated when the chromatin decondenses
~

at the following interphase. The differential packaging of mitotic chromatin is revealed by

disruptive chromosome banding techniques (q.v.) which generate reproducible patterns of dark
and light bands, corresponding to domains of genetic activity and inactivity determined through
biochemical analysis.
3.3 Chromatin and chromosome function
~

Chromatin and access to the infomation in DNA. The organization of DNA into highly ordered

structures (within which it is closely associated with proteins along its whole length) presents a
problem for the proteins which mediate DNA function. In particular, both replication and transcription
must occur in the context of nucleosomal organization, both involving large enzyme complexes
which translocate processivelya long DNA and unwind it. Both DNA polymerase and RNA
polymerase are substantially larger than a nucleosome. It is important to consider how these
proteins have access to the information in DNA when it is organized into chromatin.
Nucleosome structure during DNA replication. The unique pattern of nuclease sensitivity within
the replication forki ndicates that the separation of parental strands duringr eplication displaces histones
from the DNA. The free histones are thought to reassociate immediately with the daughter
duplexes, as histones displaced by replication in vitro rapidly assemble into nucleosomes on competitor
naked DNA. This is confirmed by scanning electron micrographs of replicating SV40 DNA,
which show nucleosome beading on the parental strand and both daughter strands immediately
adjacent to the replication fork.
The replication of chromatin is dispersive rather than conservative. Histones displaced from the
parental duplex demonstrate npore ference fore ither daughter duplex, and appeart o mix with newly
synthesized histones which accumulated uring the G1 stage of the cell cycle (q.v.). The precise assembly
mechanism in vivo is not understood, but it is thought that a molecular chaperone called N1/N2
initiates nucleosome formation bylo ading the (H3-H4)2 tetramer onto DNA, whilst another chaperone,
nucleoplasmin, facilitatesth e docking of H2A-H2B dimers. A challenging problem concerning
chromatin function is how activea nd repressed chromatin domainsa re stably propagated through
successive rounds of replication. This is likely to reflect the distribution of specifically modified
parental histones onto both daughter duplexes immediatfeollylo wing the passage oft he replication
fork, preserving preexisting chromatin structure. However, replication would allow competition
between nucleosomes and transcriptional complexes for occupation of strategic DNA sites, providing
an opportunity for the state of commitment of the cell to be changed, especially if a new
transcriptional regulatori s synthesized prior to replication. The outcomoef such competitions would
40 Advanced Molecular Biology
presumably reflect the affinity of regulatory complexes for DNA and their effective concentration in
the cell (see below).
Nucleosome s t ~ cdut ri~ng ~tra nscription. Transcription, like replication, displaces nucleosomes
from DNA, and reassembly appears to occur in the wake of the RNA polymerase. Most transcribed
genes thus retain a nucleosome structure, although the pattern of nucleosome phasing characteristic
of nontranscribed genes is lost, resulting in a smear of DNA fragments following digestion with
micrococcal nuclease and a restriction enzyme, rather than a discrete band. Experiments which
examine the progress of the polymerase complex through the nucleosome have shown that pausing
occurs about half-way through the core DNA, which may reflect the build-up of torsional strain as
the enzyme attempts to negotiate the first coil released fromth e nucleosome. The strain is released
as the enzyme moves past this point, indicating that the octamer is expelled. The octamer then reassociates
with DNA behind the enzyme, perhaps because it remains attached to the nontranscribed
strand, or perhaps because it is transiently associated with the enzyme itself.
In heavily transcribed genes sucha s the rRNA genes of Zampbrush chromosomes (q.v.), the extended
conformation of chromatin indicates that it is nucleosome-free. This probably reflects failure of
the displaced histones to reassemble on posttranscribed DNA because of a following transcriptional
elongation complex. In very active genes, there would be a convoy of RNA polymerases which
would maintain anin definite nucleosome-free regiono f chromatin.
Chromatin domains. The structural properties of transcriptionally active or potentially active chromatin
are distinct from those of inactive chromatin. Transcribed chromatin has a general increased
sensitivity to DNase I digestion, which may reflect its less condensed packaging. This is described
as open chromatin, and whilst it is true that transcription itself disrupts nucleosome structure, the
extension of DNase I sensitivity for several kilobases either side of the actual transcription unit, and

the maintenance of sensitivity in the absence of transcription suggests that this phenomenon
involves organization at a higher level than the nucleosome.
The extent of DNase I sensitivity defines a conceptual chromatin domain, a region of chromatin
whose activity is independent from that of other domains. Superimposed uponth e general DNase
I sensitivity are further DNase I hypersensitive sites. These are found flanking the transcription
unit, and are preferentially cleaved at low DNase I concentrations. Such sites are usually about 200
bp in length and often correspond to cis-acting regulatory elements (q.v. enhancw, locus control
region); they are believed to represent nucleosome-free regions where transcription regulatory complexes
bind to DNA. This has been confirmed in the case of the SV40 genome, where molecular
analysis of DNase I hypersensitive sites has shown correspondence to regulatory elements, and
topographical analysis by scanninge lectron microscopyh as identified nucleosome-free regions. The
molecular basis of open chromatins tructure is not fully understood, although there are some interesting
correlations. Open chromatinc ontains a generally higher proportion of N-terminal acetylated
core histones and HMG14/21 proteins than bulk chromatin, and is relatively depleted for linker
histones. Since linkehr istones are required forc hromatin to adopt the 30 nm fiber, and since histone
N-terminal tails facilitate nucleosome-nucleosome interactionsi,s ipto ssible that the transition from
repressed chromatin to open chromatin involves the decondensation of the 30 nm fiber to a simple
10 nm fiber organization, stabilized bnyo nhistone proteins. Histone acetyltransferases and deacetylases
have been shown to be recruited by some transcription factors, providing a mechanism for
chromatin remodeling as an initial step in transcriptional activation and repression. In mammals
and plants, repressed chromatin is often associated with high levels of DNA methylation, which
may also play a major role in the epigenetic regulation of gene expression (see DNA Methylation).
However, DNA methylation is absent from many lower eukaryotes, including, for example, yeast
and DrosophiZa.
Chromatin 41
The division of eukaryotic genomes into functionally discrete domains is particularly conspicuous
in mammals because the chromatin structure is reflected in the topography of the chromosomes
at the M phase (see Chromosome Structure and Function). A variety of disruptive staining
procedures (notably G-banding and R-banding, q.v.) reveal a reproducible pattern of dark and light
transverse bands which are thought to correspond to areas distinguished generally by the density
of chromatin structure. The G-bands correspondt o chromosome bands revealed by other methods,
such as transient replication banding (which can specifically label early and late replicating DNA)
and D-banding, which identifies regions of DNase I sensitivity. Molecular analysis shows that the
DNA within these regions differs with respect to sequence content and architecture, providing
evidence for a high level bkphsic organization (q.v.) of the mammalian genome (see Genomes and
Mapping).
Domains and boundary functions. Chromatin is physically divided into discrete and topologically
isolated regions. This idse monstrated byth e loop and scaffold appearance of protein-depleted chromatin,
the lateral loops of lampbrush chromosomes (4.v.) and the reproducible banding pattern of
Drosophila polytene chromosomes (q.v.). Chromatin loops appear to be tethered to the nuclear matrix
at their bases, and it is likely that this involves specific nucleoprotein complexes. The significanocf e
the chromatin loops is unclear, but it is possible that they represent the functionally independent
domains discussed above. This would allow adjacent domains to adopt different orders of chromatin
packaging, and would providea mechanism for enhancer monogamy, where thaec tivity of
a distant enhancer is confined to its target gene.
If loops are equivalent to domains, it can be predicted that specific DNA sequences would be
associated with the nuclear matrix, that these sequences should map neathr e borders of biochemically
defined chromatin domains, and that they should be able to isolate genes from the effects of
adjacent domains, i.e. they should act as insulators or boundary elements.
Putative matrix-associated regions (MARS) (also known as scaffold attachment regions
(SARs)) have beeni dentified in two complementary approachesb y their ability to bind to nuclear
matrix proteins. Fragmented DNA exposed to matrix components can trap putative MARS in the
insoluble fraction, and protein-depleted chromatin digested with nucleases should leave only
MARS protected from nuclease activity. These procedures have identified a number of AT-rich
elements with nsotr ong sequence conservation except a recognition site for topoisomerase 1 1, a
component

of both the nuclear matrix and the metaphase scaffold. It is possible that topoisomerase 11
could divide chromatin into topologically isolated domains.
Boundary elements havea lso been identified by their cytologiacnadl biochemical properties, i.e.
they map ath e boundaries of known chromatin domains acnodn tain nuclease-hypersensitive sites.
Such elements include the Drosophila special chromatin structures (scs and scs) and the chicken and
human P-globin HS5 element from the locus control region (4.). A genuine boundary element would
be able to establish an independent chromatin domain in an autonomous fashion, and this is
thought to be the role of the P-globin locus control region (LCRi)n vivo. This function can be tested
in two ways: by assaying for protection of a transgene against endogenous position effects using
flanking boundary elements, and by testing for the insulation of a gene from the effects of an adjacent
enhancer by the interposition of a boundary element. This has been achieved in the case of the
Drosophila scs-like elements, the chicken P-globin LCR 5HS site and the chicken lysozyme gene A
element. There is a degree of functional interchangability between the elements. The P-globin LCR
5HS site, for instancef,u nctions in Drosophila, but the scs-like elements do not function in transgenic
mice. The chicken lysozyme A elements suppress position effects both in transgenic mice and in
transgenic plants (also q.v. gypsy transposon).
Although the results of experiments designed to test the function of the M A R S isolated by
physical methods are inconclusive, the presence of MAR-type elements in some boundary elements
identified functionally suggests that physical division of chromatin into loops could be the basis of
the functional division of chromatin domains.
42 Advanced Molecular Biology
Heterochromatin stmcture and epigenetic gene regulation. The translocation of euchromatic DNA
into heterochromatin often results in the spread of transcriptional repression into the euchromatic
region. The extent of heterochromatinization varies from cell to cell, but is clonally propagated so
that genes near the breakpoint are expressed in a variegated manner. This phenomenon, which is
termed position effect variegation (PEV), suggests that a cis-acting silencing activity can spread
across the chromosome breakpoint. The spreading effect is linear and uninterrupted; gene loci
nearest the breakpoint are inactivated most often and the repression never skips over genes. This
suggests that heterochromatin normally spreads bpyr oteins adding on to preexisting heterochromatin
and extending it until some boundary is reached. Translocation presumably removes the
boundary, and the variable spreading which causes PEV probably reflects variable amounts of
heterochromatin-sponsoring proteins in different cells.
It is likely that heterochromatin and euchromatin proteins are found in equilibrium between
chromatin and the nucleoplasm, and that mutations encoding genes with chromatin functions
would alter the balance of these proteins and result in a shift in the extent of PEV following a
translocation event (these are termed antipodal effects). Genetically, such mutations would behave
as either suppressors or enhancers of PEV, and would identify the genes involved in higher-order
chromatin structure.
In Drosophila, a search for such PEV modifiers has identified several cellular components with a
role in heterochromatin formation. Heterochromatin itself is one of the best PEV modifiers; by
increasing the dosage of the Y-chromosome (which is mostly heterochromatin), PEV can be
suppressed, probably resulting from the depletion of heterochromatin proteins in the nucleoplasm.
Histones have also been identified as an important component heterochromatin is generally poor
in acetylated H4 but rich in a particular acetylated form. Other modifiers correspond to specific
proteins which play a direct role in heterochromatin structure. These include the general protein
HP1 and some regulators of gene expression (e.g. Polycomb; q.v. homotic genes, maintenance of
diferentiution), which mediate their effects at the level of chromatin structure. Such proteins often
possess a conserved structure, termed a chromodomain.
In yeast, heterochromatinp roteins have beeni dentified on the basis that mutants de-repress the
silent mating-type loci, which are usually sequestered in repressed chromatin near the telomeres
(q.v. mating type switching). As well as histones H3 and H4, the silent information regulators SIR3
and SIR4 and a further protein termed RAPl havbee en identified,t he spreading of heterochromatin
apparently reflecting interactions between the histones and the SIR products, facilitated by RAPl
(SIR3, for instance, interacts with the N-terminal tail of histone H4).