Metabolism and regulation of canonical histone mRNAs: life
without a poly(A) tail
William F. Marzluff, Eric J. Wagner, and Robert J. Duronio
Program in Molecular Biology and Biotechnology, University of North Carolina at Chapel Hill, Chapel
Hill, North Carolina 27599, USA.
Abstract
The canonical histone proteins are encoded by replication-dependent genes and must rapidly reach
high levels of expression during S phase. In metazoans the genes that encode these proteins produce
mRNAs that, instead of being polyadenylated, contain a unique 3' end structure. By contrast, the
synthesis of the variant, replication-independent histones, which are encoded by polyadenylated
mRNAs, persists outside of S phase. Accurate positioning of both histone types in chromatin is
essential for proper transcriptional regulation, the demarcation of heterochromatic boundaries and
the epigenetic inheritance of gene expression patterns. Recent results suggest that the coordinated
synthesis of replication-dependent and variant histone mRNAs is achieved by signals that affect
formation of the 3' end of the replication-dependent histone mRNAs.
Histones are the primary protein component of chromatin. Although they were initially thought
to be mainly involved in chromosomal DNA packaging in eukaryotes, it is now recognized
that they also have a crucial role in regulating gene expression. Histones can be extensively
modified after translation and these modifications play an important part in regulating gene
expression. They are constantly being shifted, modified, evicted and re-deposited as chromatin
is continually remodelled (reviewed in REF.
1
). Thus, the cell must carefully coordinate the
replication of DNA, the synthesis of an estimated 10
8
molecules of each histone type in
mammalian cells and the rapid deposition of new and old histones to reform chromatin during
each relatively short S phase
2,3
.
In metazoans the bulk of the histone proteins, defined here as the canonical histone proteins,
are encoded by a family of replication-dependent histone genes. Their mRNAs are the only
known cellular non-polyadenylated mRNAs in eukaryotes
4
. These genes encode all four core
histones — H2A, H2B, H3 and H4 — which make up the nucleosome, and the linker H1
histones, which are found between nucleosomes. In place of a poly(A) tail, replication-
dependent histone mRNAs end in a 3′ stem–loop sequence that is crucial in their regulation
(FIG. 1a), and is formed by endonu-cleolytic cleavage of the pre-mRNA (FIG. 1b). This novel
3′ end results in the requirement for a distinct set of factors for metabolism and regulation of
these histone mRNAs. These mRNAs must be expressed rapidly at the beginning of S phase
and must persist at high levels throughout S phase to coincide with the replication of DNA.
They are destroyed at the conclusion of S phase or rapidly during S phase if DNA replication
is halted.
In addition to the canonical histones, there are several variant histones whose synthesis is not
cell-cycle-regulated and whose mRNAs are polyadenylated and expressed throughout the cell
cycle (replication-independent histone mRNAs). These histones include the H3.3 and
Correspondence to W.F.M. e-mail: E-mail: marzluff@med.unc.edu.
NIH Public Access
Author Manuscript
Nat Rev Genet. Author manuscript; available in PMC 2009 July 26.
Published in final edited form as:
Nat Rev Genet. 2008 November ; 9(11): 843–854. doi:10.1038/nrg2438.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
H2A.Z proteins (H2Av in Drosophila melanogaster), CENPA (histone H3-like centromeric
protein A, present at centromeres), macro-H2A and histone H1.0. The major variants of the
core histones are H3.3 and H2A.Z, which are found in all multicellular organisms. These
proteins mark active genes and are thought to contribute to defining chromatin boundaries.
In this Review we discuss results from recent years that have substantially advanced our
understanding of the metabolism and regulation of the canonical histone mRNAs and that have
provided the first clues into how the metabolism of the canonical histone mRNAs and variant
histones are coordinated. We also highlight the emergence of some surprising recent findings
about the evolutionary origin of histone genes.
Replication-dependent histone mRNAs
Like polyadenylated mRNAs, metazoan canonical histone mRNAs have a 7-methyl-guanosine
cap at the 5′ end. However, unlike polyadenylated mRNAs, histone mRNAs end in a stem–
loop, which consists of a 6 base stem and 4 nucleotide loop. Altogether, the 3′ end of canonical
histone mRNAs consists of a conserved 25–26 nucleotide sequence that includes the 5
nucleotides before the stem–loop, the 16 nucleotide stem–loop and the 4–5 nucleotides after
the stem–loop (FIG. 1c). The stem–loop sequence is evolutionarily conserved in metazoans
with a number of elements that are invariant (FIG. 1c), including sequences in the stem, in the
loop and before the stem–loop. A single protein, the stem–loop binding protein (SLBP), binds
to this 26 nucleotide sequence and participates in all aspects of histone mRNA metabolism.
SLBP has a small, 73 amino acid RNA-binding domain (RBD) that is not similar to the RBD
of any other RNA-binding protein
5,6
.
In most species, histone H1 mRNAs have the same 3′ end as the other canonical histone
mRNAs. Interestingly, however, in some organisms (for example, the nematodes
Caenorhabditis elegans and Caenorhabditis briggsae) only the core histone genes end in a
stem–loop, whereas the histone H1 genes encode polyadenylated mRNAs
7
. As the metabolism
of histone H1 differs from that of the core histones and histone H1 turns over more rapidly.
The absence of the stem–loop from histone H1 mRNA in these species does not affect the
ability to coordinately express the four nucleosomal core proteins.
Organization of canonical histone genes
The high demand for canonical histone proteins during S phase is met by the coordinated
expression of multiple copies of the histone genes in all metazoans. In mammals there are
approximately 75 distinct canonical histone mRNAs
8
. The genes that encode these transcripts
are clustered together in the genomes of all eukaryotic species studied so far, and these clusters
typically contain multiple copies of the genes that encode the five distinct histone proteins.
Thus, the canonical histone genes have remained tightly linked throughout metazoan evolution.
Two types of clusters are commonly found: tandemly repeated gene sets, with the repeat unit
containing one copy of each of the five histone genes; and ‘jumbled’ clusters, in which there
is no common gene order and the individual genes for each histone protein are not identical.
Within vertebrates, frogs have tandemly repeated clusters, whereas mammals and birds have
jumbled clusters.
A dramatic example of tandemly repeated canonical histone genes occurs in D.
melanogaster. There are about 100 copies of a ∼5 kb histone gene set, significantly more
canonical histone genes than found in mammals, which have 20-fold larger genomes. Thus,
D. melanogaster seems to have many more canonical histone genes than are required for
somatic cell cycles, with their relatively long S phases. However, this extensive duplication
might be required at the end of oogenesis to provide sufficient canonical histone proteins for
early embryonic development (see below). Alternatively, each individual gene might be
Marzluff et al. Page 2
Nat Rev Genet. Author manuscript; available in PMC 2009 July 26.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
expressed at a relatively low level. In either case, the tandem repeat organization ensures that
an equal number of each histone mRNA is produced, providing a likely explanation for why
this type of genomic organization has been maintained. The jumbled clusters found in mammals
and birds might have been maintained to allow the definition of a nuclear subdomain in which
efficient biosynthesis of histone mRNA can take place (see below).
Outside the canonical histone gene clusters in mammals there are only two canonical histone
genes encoding non-polyadenylated mRNAs. One of these encodes the histone variant
H2A.X, which is involved in recognition of DNA damage in chromatin
9
. The 3′ end of H2A.X
mRNA expressed during S phase is not polyadenylated, but in G0 and G1 phase transcription
proceeds further and the mRNA ends in a poly(A) tail
10
. It is possible that the separation of
the H2A.X gene from the major cluster plays a part in allowing the expression of both
polyadenylated and non-polyadenylated mRNAs from the same gene. As the H2A.X protein
is synthesized throughout the cell cycle
11
, the ability to make two different mRNAs ensures
that an H2A.X mRNA is constitutively expressed. In addition, a single human histone H4 gene
that is located outside any histone gene cluster expresses a non-polyadenylated mRNA in S
phase
8
. The function of this gene is not known; however, there is a syntenic copy in the mouse,
suggesting that it has been conserved in mammalian evolution.
Expression of canonical histone genes
A consequence of having a unique 3′ end is that a different set of factors is required for the
expression and function of histone mRNAs, including the U7 small nuclear RNA (snRNA)
and the Sm-like proteins LSm10, LSm11, SLBP and SlBP-interacting protein 1 (SLIP1)
(TABLE 1). Surprisingly, some factors involved in histone mRNA metabolism are not unique
to the regulation of these mRNAs, but overlap with the poly(A)
+
mRNA metabolic machinery.
Below we describe the steps in histone mRNA expression, focusing on data from experiments
in mammalian and D. melanogaster cells. An overview of histone mRNA metabolism in
mammalian cells is presented in FIG. 2.
Transcription of canonical histone genes
Mammalian genes encoding canonical histones are constitutively transcribed by RNA
polymerase II and their rate of transcription increases as cells approach S phase
12
. The nuclear
protein ataxia-telangiectasia locus (NPAT) is present constitutively throughout the cell-cycle
in Cajal bodies adjacent to histone genes (FIG. 3). NPAT is essential for entry into S phase
13
and has a domain that stimulates histone gene transcription
14,15
. However, deletion of this
domain does not prevent entry into S phase, suggesting that NPAT might have multiple roles
in histone mRNA biosynthesis
13
. At the beginning of S phase cyclin E–CDK2 (cyclin-
dependent kinase 2) phosphorylates NPAT in these bodies
14,16
, and the phosphorylated form
persists throughout S phase, resulting in increased expression of canonical histone genes.
Formation of the 3′ end of histone mRNA
The genes encoding metazoan canonical histones lack introns, and thus an endonucleoytic
cleavage that releases the nascent pre-mRNA from the DNA template is the only processing
event necessary to form mature histone mRNA. In fact, insertion of an intron into a histone
gene interferes with histone 3′ end formation, resulting in formation of polyadenylated histone
mRNA
17
.
There are many similarities between the endonucleoytic cleavage reaction that forms
polyadenylated mRNAs and the reaction that forms histone mRNAs. For polyadenylated pre-
mRNAs, cleavage occurs between two cis elements: the hexanucleotide sequence consensus
AAUAAA and a Gu-rich element located about 40 nucleotides downstream in poly(A)
Marzluff et al. Page 3
Nat Rev Genet. Author manuscript; available in PMC 2009 July 26.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
mRNAs
18
. For histone pre-mRNAs, cleavage occurs between the stem–loop and the histone
downstream element (HDE), a purine-rich sequence located about 15 nucleotides after the
cleavage site (FIG. 1b). Both cleavage reactions preferentially occur after the dinucleotide CA,
giving rise to a 3′-hydroxyl (3′-OH), and both reactions occur in vitro in the presence of EDTA,
a property that helped identify the canonical histone pre-mRNA cleavage factor
19–21
.
In the first step of histone mRNA 3′ end formation
22
, SlBP binds the stem–loop and then the
HDE binds with a novel snRNA, U7 snRNA (a component of the U7 small nuclear
ribonucleoprotein, snRNP (FIG. 1b)). U7 snRNA is similar to the spliceosomal snRNAs (U1,
U2, U4 and U5) as it has a 2,2,7-trimethylguanylate cap, precipitates with the anti-Sm antisera
and ends in a stem–loop. It is the 5′ end of U7 snRNA that base-pairs with the HDE
23,24
. In
spliceosomal snRNPs, the seven Sm proteins (SMB, SMB′, SMD1, SMD2, SME, SMF and
SMG) form a heptameric ring around the Sm-binding site. U7 snRNA contains a slightly
different Sm site, which deviates from the consensus Sm-binding sites in spliceosomal snRNAs
at two of the ten positions. Two novel Sm-like proteins, LSM10 and LSM11, replace SMD1
and SMD2 (Refs 25
,26
) in U7 snRNP, and their inclusion is required to bind the non-consensus
Sm-binding site in the U7 snRNA
27
. Changing the novel Sm site in U7 snRNA to the consensus
Sm site results in binding of the spliceosomal Sm core structure and inactivation of U7
snRNP
27
.
Together, SLBP and U7 snRNP recruit a cleavage factor that contains at least CPSF73
(cleavage and polyade-nylation specificty factor 73), CPSF100 and symplekin
28
, and probably
also FIP1 (Refs 21
,28,29
), all of which are also part of the complex that processes the 3′ end of
polyadenylated mRNAs. The components of the cleavage factor have been identified by both
in vivo and in vitro approaches. CPSF73 can be crosslinked to the cleavage site
21
, and
symplekin can restore processing to a heat-treated extract
28
. CPSF73 contains tightly bound
zinc, which accounts for the ability of the cleavage reaction to occur in EDTA. A recent study
showed that a fusion protein containing CPSF73, CPSF100 and symplekin can restore
processing to a heat-treated extract, and that this activity requires crucial residues for zinc
binding in CPSF73 and CPSF100 (REF.
30
). This provides direct evidence for activity of these
proteins in cleavage, and indicates that CPSF73 is almost certainly the endonuclease that
cleaves histone pre-mRNA and polyadenylated pre-mRNAs
21,31
. Interestingly, the 3′ end of
snRNAs is probably formed through cleavage by an integrator complex containing integrator
9 and 11 subunits that are homologous to CPSF100 and CPSF73, respectively, and that are
members of the β-lactamase family
32
. This suggests that the mechanism for formation of the
3′ end of all RNA polymerase II transcripts has been conserved.
Histone locus bodies: a factory for histone mRNA synthesis?
One possible reason for the conserved physical linkage of histone genes is that it allows them
to be brought within a domain of the nucleus that is specifically enriched in factors required
for the expression of replication-dependent histone genes, for example, the Cajal body. In
mammalian cells and frog oocytes, Cajal bodies contain the U7 snRNP
33,34
and, like ‘nuclear
speckles’ that contain splicing factors, they either act as stockpiles of processing factors or are
sites of processing. Vertebrate Cajal bodies are also sites of snRNA maturation and snRNP
assembly
35,36
. Cajal bodies are found at several sites in the nucleus, including near the
mammalian
37
and frog
38
replication-dependent histone genes and, in mammals, Cajal bodies
located near these genes differ from others as they contain NPAT
13,14
(FIG. 3).
Recently, Gall and co-workers reported the first example of a Cajal body outside vertebrates
by showing that D. melanogaster has a single Cajal body in each nucleus
39
. Strikingly, D.
melanogaster nuclei also contain a separate body, the histone locus body (HLB), that is
associated with the replication-dependent histone gene cluster (FIG. 3). The HLB contains
factors involved in histone mRNA expression, but does not contain U85 (REF.
39
), a small
Marzluff et al. Page 4
Nat Rev Genet. Author manuscript; available in PMC 2009 July 26.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
RNA required for modification of the spliceosomal snRNAs
40
that is present in the Cajal body.
The Cajal body and the HlB are adjacent to the nucleolus and are occasionally close to each
other but are non-overlapping. Thus, in D. melanogaster there are distinct nuclear organelles
for snRNA maturation and replication-dependent histone pre-mRNA processing. The HLB can
also be visualized with the MPN2 monoclonal antibody, which recognizes a cyclin E–CDK2-
dependent phospho-epitope in the HLB of D. melanogaster cells
41
, suggesting that flies and
mammals have similar cell cycle regulatory inputs into replication-dependent histone gene
expression.
Translation of replication-dependent histone mRNA
Replication-dependent histone mRNAs, like most other mRNAs, are exported from the nucleus
by antigen peptide transporter
42–44
, and once exported they are efficiently translated.
Translation of polyadenylated mRNA requires that the 3′ poly(A) tail be brought into the
proximity of the 5′ cap. This is mediated through protein–protein interactions between the poly
(A) binding protein and eukaryotic translation initiation factor 4-γ (EIF4G), which also
interacts with the cap-binding protein
45
. A distinct mechanism has evolved for the
circularization of replication-dependent histone mRNAs. Like the poly(A) tail on other
mRNAs, the 3′ end of these histone mRNAs is essential for translation in vivo
46
. SLBP is bound
to the cytoplasmic histone mRNA
47
and is required for its translation
48,49
. The translation
activation region has been mapped to a conserved region of SLBP
50
. SLBP does not interact
directly with EIF4G, but the domain required for translation of histone mRNA interacts with
the recently identified protein SLIP1, which also interacts with EIF4G
50
. Thus, these proteins
circularize histone mRNA and help mediate the efficient translation of histone mRNA by a
mechanism similar to the translation of polyadenylated mRNAs.
Regulated degradation of replication-dependent his-tone mRNAs
As cells approach the end of S phase, the need for new canonical histone proteins diminishes
and the levels of replication-dependent histone mRNAs decrease. This attenuation of mRNA
levels is achieved at the level of transcription in the yeast Saccharomyces cerevisiae, in which
the half-life of the mRNAs is intrinsically short
51
. By contrast, in mammals the decrease results
from a rapid reduction in mRNA half-life when DNA replication is blocked or completed. The
stem–loop at the 3′ end of replication-dependent histone mRNAs is the cis-element that
mediates this degradation
52
. Degradation requires that the stem–loop be close to the stop
codon
53,54
, which accounts for the observation that in all metazoan histone mRNAs the stem–
loop begins 25–60 nucleotides from the termination codon. Degradation of histone mRNA
requires SLBP, which is involved in recruiting the proteins necessary to add a short oligo(U)
tail to the histone mRNA that is being translated
55
. This oligo(U) tail is an optimal binding site
for the LSM1–7 heptamer and leads to the degradation of the histone mRNA by pathways that
are similar to those involved in the degradation of poly(A) mRNA after deadenylation
56,57
.
Thus, both the 3′ ends of replication-dependent histone mRNAs and SLBP are crucial for three
steps in histone mRNA metabolism — processing, translation and degradation (FIG. 2).
Cell-cycle regulation of histone mRNAs
Multiple modes of coordination with the cell cycle
In mammalian cells, there are four distinct regulatory mechanisms that contribute to the proper
rate of canonical histone protein accumulation during the cell cycle: transcription of the histone
genes, efficiency of pre-mRNA processing, change in mRNA half-life and degradation of
excess histone protein. Replication-dependent histone mRNAs, for all five classes of canonical
histones, are among the most highly cell-cycle-regulated mRNAs in mammalian cells.
Marzluff et al. Page 5
Nat Rev Genet. Author manuscript; available in PMC 2009 July 26.
NIH-PA Author Manuscript NIH-PA Author Manuscript NIH-PA Author Manuscript
