ABSTRACT
We have isolated from a subtractive cDNA library of
Xenopus laevis
a novel transcript, H2A.Zl, which belongs to the H2A.Z variant gene family. Characterization of its expression during
oogenesis and development shows significant differences from the expression of
the core histone H2A. First, H2A.Zl mRNA is mainly detected only during
oogenesis and after the midblastula transition, whereas H2A is constitutively
expressed, at much higher levels, throughout embryonic growth. Second, in
contrast with H2A, the variant H2A.Zl is polyadenylated during development. Third, expression of H2A.Zl is uncoupled from the S phase after gastrula, whereas synthesis of the
core histone H2A mRNA is tightly controlled to DNA replication. Interestingly,
H2A.Zl is less charged in the N-terminal tail which is crucial for chromatin-mediated repression. The characteristics of H2A.Zl suggest that its
incorporation into nucleosomes would lead to a chromatin structure more competent for gene expression during development.
DNA in chromatin is packaged into a fundamental nucleosome structure consisting of an octamer of core histone proteins (
1
) which are highly conserved throughout evolution. H3, H4 and H2B histone families
contain only one or several almost identical isoprotein species, whereas genes
encoding H2A histones are divided in three subfamilies: major histone H2A, and
two distinct minor histone variants: H2A.X and H2A.Z. Relatively little
information is known about the variants as compared with the major core
histones. However, available information shows that these variants exhibit
unusual features. The type Z of H2A histone is of particular interest as it
could be a member of a separate evolutionary lineage from the other H2A species
(
2
). Its primary protein structure is highly conserved in organisms as diverse as
mammals (
3
), chicken (
4
),
Drosophila
(
5
), sea urchin (
6
),
Tetrahymena
(
7
) and
Schizosaccharomyces pombe
(
8
). In contrast with major H2A gene expression which is linked to the S phase of the cell
cycle, H2A.Z gene expression is regulated in a replication-independent fashion (
9
). Deletion of the H2A.Z gene in
S.pombe
cells results in reduced mitotic stability (
8
).
The H2A.Z variant represents only 5-10% of the total H2A in all examined cell types and species (
10
,
11
). These variants are encoded by single copy genes which differ from H2A both in
their genomic structure and in the maturation of their transcripts. Indeed,
H2A.Z coding sequences contain introns, lack the 3'-terminal stem-loop structure typical of replication-dependent histone mRNA and produce polyadenylated
messenger RNAs.
One important function for the variant histone could be to substitute for S
phase linked histones in chromatin in relation to transcription. Such a
function would be favoured by their synthesis outside of S phase. Moreover,
H2A.Z is present in the transcriptionally active macronucleus of
Tetrahymena
whereas it is absent in the transcriptionally inert micronucleus (
12
,
13
). An essential role for H2A.Z has also been demonstrated in
Drosophila
early development where its deletion is lethal (
14
).
We have been interested in the identification of a family of genes specifically expressed during oogenesis and which are either absent or
poorly expressed in differentiated adult tissues (
15
). We report here the sequence of a new variant histone, H2A.Zl, resembling the
type Z of H2A. A significant divergence between H2A.Zl and other H2A.Z histones
was found in the N-terminal region of the protein which renders it constitutively less basic.
We investigated developmental expression of H2A.Zl messenger RNAs and show that
the transcripts are polyadenylated in contrast to core histone H2A messengers.
Finally, we found that H2A.Zl transcription is not temporally modulated by S
phase in post gastrula embryos.
A 180 bp fragment Mat 200 was previously selected by differential screening of a
subtracted [lambda]gt-10 library enriched for specific maternal sequences (
15
) and subcloned into a bluescript vector.
32
P-labeled complementary DNA probe of Mat 200 (10
9
c.p.m./ml) was used to isolate complete cDNAs from a nonsubtractive [lambda]gt-10 oocyte library, provided by D. A. Melton (
16
). Hybridization of ~2 * 10
5
clones on nylon filters was carried out overnight at 42oC in 50% formamide, 0.5% SDS, 5* SSPE and 5* Denhardt's solution. Filters were extensively washed in 0.2* SSPE, 0.2% SDS at 55oC. Autoradiography was performed at -70oC with an intensifying screen.
Sequencing was performed on both strands using the
T7
Sequencing Kit (Pharmacia). The amino acid sequence of the full-length protein was deduced. Computer similarity searches were carried out using the FASTA program as described by Pearson and Lipman (
17
).
Xenopus laevis
oocytes were collected and defolliculated as previously described (
18
) and sorted by size under a microscope according to Dumont (
19
). Eggs and embryos were obtained as previously described (
20
).
For Northern blot analyses (
21
), total RNA from embryos and cells was extracted as previously described by a urea-LiCl procedure (
20
). Poly(A)
+
and Poly(A)
-
RNA fractions were obtained from total RNA by oligo (dT)-cellulose (
21
). The
32
P-labeled DNA probes used were from the 18S RNA gene (1.8 kb DNA fragment),
Xenopus
H2A (300 bp DNA fragment) or H2A.Zl1 (825 bp DNA fragment). After
hybridization, nylon membranes were washed as described for library screening
(see above) and exposed either to autoradiography or to storage phosphor
screens (Molecular Dynamics 400A PhosphorImager).
Albino
X.laevis
embryos were processed for whole-mount
in situ
hybridization as described by Harland (
22
). H2A and H2A.Zl1 in bluescript were linearized with
Bam
HI or
Hin
dIII to generate templates for
in vitro
transcription. Sense and antisense probes were labeled by incorporation of
digoxigenin-modified UTP (Boehringer) using either T3 or T7 polymerase. Embryos were
mounted in benzyl benzoate-benzyl alcohol (2:1) for photography.
Developing embryos from the same batch of siblings were divided into a control
group and an aphidicolin (125 [mu]g/ml) group at the early gastrula stage [stage 10; according to Nieuwkoop
and Faber (
23
)]. Genomic DNA and total RNA were extracted at the time of transfer into
aphidicolin (0 h) and at three time points (2, 18 and 22 h) during the course
of experiment. DNA and RNA were extracted from the same homogenates by treatment with 250 [mu]g/ml proteinase K, phenol, phenol-chloroform and precipitation with ethanol (
24
). DNA and RNA were electrophoresed on a 0.6% agarose gel and transferred to
nylon membrane. The blot was hybridized with
32
P-labeled DNA probes for 18S RNA, H2A and H2A.Zl1 cDNAs, washed and exposed
to autoradiography as described for library screening (see above).
We have previously reported the construction of a subtractive cDNA library from
poly(A)
+
mRNA intended to isolate new genes specifically expressed as maternal RNAs in
the oocyte and preferentially active in embryogenesis (
15
). A 180 bp fragment (called Mat 200) was isolated and used as a labeled probe
to screen for full length cDNA clones (see Materials and Methods). Sequencing
on both strands of two cDNAs (H2A.Zl1, EMBL accession no. X98535 and H2A.Zl2,
EMBL accession no. X98536) revealed a single identical open reading frame
encoding a 127 amino acid polypeptide. Nucleotide sequences show 3.7%
divergence in the two coding regions which corresponds to simple changes of
codon usage for the same amino acid. The existence of these two different
sequences is likely to be a consequence of genomic tetraploidation that
occurred in the emergence of the genus
Xenopus
(
25
). The 3' untranslated region (UTR) of H2A.Zl2 contains a polyadenylation signal
(AATAAA) 14 nt upstream from a 16 nt poly(A) tail, and a T-rich sequence (TTTTTTTTAT) termed the cytoplasmic polyadenylation element (CPE) (
26
,
27
). This sequence is specifically required for cytoplasmic polyadenylation during
oocyte maturation. The CPE and its position relative to AATAAA may control the
timing and extent of polyadenylation. It has been shown that mRNAs which
contains these two distinct 3'UTR signals are translationally activated upon maturation (
28
,
29
).
In vitro
translation from the cDNA clone in a reticulocyte lysate indicates that H2A.Zl1
encodes for a protein of apparent molecular weight 15 kDa (data not shown), in
close agreement with the calculated mass (14.1 kDa).
The predicted
X.laevis
protein sequence shares a significant identity with the
X.laevis
H2A protein (57%), and contain the H2A box AGLQFPVGR (Fig.
1
). However, a closer identity was detected with the type Z family of H2A variant
proteins, found in various organisms (70-90%). But, the
Xenopus
sequence diverges from H2A.Z of other species in two aspects. First, sea urchin,
Drosophila
and chicken H2A.Z have a 97-99% identity with human H2A.Z in its most conserved region (3-115), whereas the identity between
Xenopus
H2A.Z and human H2A.Z is 88%. Second, two amino acid changes (indicated by
arrows in Fig.
1
) affect two basic lysine and arginine residues localized in the N-terminal part of
Xenopus
H2A.Z and render it constitutively less charged. The N-terminal tails of histones are known to protrude from the nucleosome and
play an important role both in chromatin structure and interaction with
regulatory factors (
30
). The same changes were found in the two independent nucleotide sequences of
Xenopus
H2A.Z clones isolated. Because we cannot exclude the presence of other H2A.Z
sequences in the
Xenopus
genome, we have named this variant histone `H2A.Z-like', or H2A.Zl.
In order to characterize H2A expression, Northern blot analyses were carried out
using RNA extracts from oocytes and developmental stage embryos. RNAs were
electrophoresed on formaldehyde-agarose gel, transferred to nylon membranes and hybridized under stringent conditions. Hybridization with an 18S cDNA probe was used as a control. In order to get a better understanding of the
H2A.Zl expression, we compared its level of transcription with the non variant
H2A, using two specific probes (1 * 10
6
c.p.m./ml for each probe) concomitantly in Northern blot experiments (Materials
and Methods and Fig.
2
). H2A.Zl mRNA migrates as a 900 bp band whereas H2A is detected at 520 bp.
H2A.Zl mRNA expression was mainly detected during oogenesis, and after the midblastula transition (MBT). It is difficult to analyse pre-MBT stages as during oogenesis a number of proteins involved in
chromosomal replication have been stored. These include the core histones (
32
,
36
,
37
) which will be necessary to assemble the newly synthesized DNA during the 12
divisions without transcription which occur prior to MBT. However, from the
gastrula stage, when the maternal pool is exhausted, we could expect that
histone synthesis would become coupled to DNA synthesis. It was therefore
interesting to analyze both the variant and non variant H2A levels when DNA
replication is blocked after the gastrula stage. It was previously observed
that embryos incubated in the presence of aphidicolin, a DNA synthesis inhibitor, continue to differentiate at least to some extent (
24
). A batch of developing embryos which have reached the early gastrula stage
(stage 10) was divided into two populations, one left in the same medium, the
other with the addition of aphidicolin. As previously observed (
24
), embryos in both the control and aphidicolin group continued to differentiate
and have reached the early tailbud stage (stage 24/25) at the end of the
experiment. For each time point, total genomic DNA and total RNA were analyzed
on the same gel (Fig.
4
A). Figure
4
A shows that aphidicolin indeed arrests DNA synthesis, whereas no change was
observed in the level of 28S and 18S rRNAs during development, consistent with
the use of the store of maternal ribosomes during this developmental period. To
analyze both H2A.Zl and H2A expression on the same blot, hybridization was
performed first with H2A. Hybridization with H2A.Zl was then performed 10 days
after in order to obtain a similar signal level for both mRNAs. In aphidicolin-treated embryos, the amount of the core histone H2A mRNA dropped abruptly
within 2 h of aphidicolin treatment (Fig.
4
B). In contrast, the inhibitor had no effect on the level of H2A.Zl mRNA. These
data show that the H2A.Zl variant histone level is not down regulated in response to a DNA synthesis inhibitor, in contrast with its normal
counterpart H2A. These observations were reproduced in five independent experiments.
In order to identify new products preferentially expressed during early development, we screened a differential library (
15
) enriched in cDNA fragments from maternal origin. We isolated a cDNA, (named H2A.Zl), the
deduced primary protein sequence of which shows highest identity with the
variant type Z of H2A identified in several species, although several amino
acid changes are peculiar to
Xenopus
H2A.Zl. H2A.Z is a core histone variant which has been conserved in several
species, from
Tetrahymena
to man.
Xenopus
H2A.Zl is transcribed at a relatively high rate during oogenesis. Major core
histones are also synthesized during this period but their corresponding RNAs
are deadenylated after maturation (
31
). In contrast, the
Xenopus
H2A.Zl variant remains adenylated and an increase in length of its mRNA was observed at maturation and at the gastrula stage, and might be due to the additional polyadenylation. Translational control is a major mechanism controlling gene activity during early development and correlates with changes in polyadenylation (
28
,
29
). Two categories of mRNAs are found in the oocyte, one containing a short poly(A) tail (<100 nt) which is translationally masked and one containing a long poly(A) tail
(>100 nt) which is translated during oogenesis. Interestingly, sequence
analysis indicates that H2A.Zl belongs to the first class of maternal mRNAs,
and also contains the two 3'UTR-specific signals shown to be required for translational activation
at maturation: the polyadenylation site and the U-rich or CPE element (
28
,
29
). This observation suggests that polyadenylation of H2A.Zl mRNA subsequent to oogenesis might regulate its activity. The
polyadenylated nature of H2A.Zl mRNA, as well as its increase in length at
maturation and gastrulation, could be a consequence of its recruitment by the
translation machinery. There is no available anti-variant histone antibody and therefore we could not assay for the specific
detection of H2A.Zl protein. Analysis of
Xenopus
oocyte proteins by 2-dimensional gel electrophoresis have identified H2A.X, but it is not yet
clear if H2A.Z is present (
38
-
40
).
It has been proposed that the H2A.Z variant retains normal histone-histone interaction but will decrease the histone-DNA interaction when it substitutes for the normal H2A (
1
). Among the H2A.Z proteins from different organisms,
Xenopus
H2A.Zl is unique in having two basic residues localized in the conserved N-terminal region, replaced by two uncharged amino acids. N-terminal tails of histones are known to be subjected to important
postranslational modifications (including acetylation) which affect chromatin.
Recent observations with H3 and H4 histones have shown direct interactions with
regulatory factors in the transcription process through the N-termini (
41
,
42
). Moreover, a recent detailed analysis of transcriptional repression by histones H2A and
H2B demonstrates that an N-terminal domain extending to amino-acid 20 of H2A is crucial for chromatin mediated repression (
43
), leading to the suggestion that the N-terminal tail of H2A could be considered as a silencing domain (
30
). The modification of the N-terminus of H2A.Zl may therefore be highly significant and suggest that
incorporation of H2A.Zl into chromatin would lead to a nucleosome structure
less stable and more permissive to transcription factors. Such incorporation
might be facilitated in view of the observation that H2A and H2B exchange out
of the nucleosome relatively easily
in vivo
(
44
).
Chromatin assembly during the first 12 divisions is mainly ensured by the core
histone protein pool previously synthesized in the oocytes (
32
). Thus one single egg contains enough core histone to built up 20 000 nuclei (
32
), e.g. the amount of nuclei at the blastula stage. In post blastula embryos,
the expression of H2A.Zl is uncoupled from DNA synthesis, in contrast with H2A.
Both the structural nature of H2A.Zl and its expression indicate that it might
be a histone variant more adapted to chromatin remodeling for transcription
during development.
We are grateful to D.A. Melton for the oocyte cDNA library, M. Perry for the
core H2A plasmid construct, and E. M. De Robertis for the Chordin plasmid
construct. We also thank A. Hair for critical reading of the manuscript. This
work was supported by grants from the Association pour la Recherche sur le
Cancer, the European Community (EBBSC1*CT000677), the GREG, the GEFLUC, and the
Ligue Nationale Contre le Cancer.
REFERENCES
Return