(Received for publication, March 14, 1997, and in revised form, April 8, 1997)
From the Department of Biological Science, Florida State University, Tallahassee, Florida 32306-3050
H1 histones, found in all multicellular
eukaryotes, associate with linker DNA between adjacent nucleosomes,
presumably to keep the chromatin in a compact, helical state. The
identification of multiple histone H1 subtypes in vertebrates suggests
these proteins have specialized roles in chromatin organization and thus influence the regulation of gene expression in the multicellular organism. The mechanism by which the association of H1 with nucleosomal DNA is regulated is not completely understood, but affinity for different DNA sequences may play a role. Here we report that a specific
H1 subtype in the mouse, namely H1b, selectively binds to a regulatory
element within the protein-encoding sequence of a
replication-dependent mouse H3.2 gene. We have previously
shown that this coding region element, , is the target of very
specific interactions in vitro with another nuclear factor
called the
factor. This element is required for normal gene
expression in stably transfected rodent cells. The mouse H1b protein
interacts poorly (100-fold lower affinity) with the comparable
"
" sequence of a replication-independent mouse H3.3 gene. This
H3.3 sequence differs at only 4 out of 22 nucleotide positions from the
H3.2 sequence. Our findings raise the possibility that this H1b protein plays a specific role in regulation of expression of the
replication-dependent histone gene family.
Five major classes of histones occur in the mouse and most eukaryotes, H1 and the core histones H2a, H2b, H3, and H4. These basic proteins play key roles in eukaryotic chromatin structure and organization. The core histones, H2a, H2b, H3, and H4, are responsible for nucleosome formation. H1 histones associate with "linker" DNA between adjacent nucleosomes and are thought to play a role in folding eukaryotic DNA into condensed higher order chromatin structures (1-3). It has long been known that H1 can act as a repressor of transcription; considerable evidence indicates that chromatin structure is important in the regulation of transcription because it restricts the accessibility to DNA of both general and gene-specific transcription factors (3-8). Linker histones are also known to modulate nucleosome position (9-11). Further, depletion of H1 from chromatin has been shown to activate transcription in vitro (4, 12, 13). There is evidence, however, that H1 is not completely absent from transcribed genes (14, 15), which implies that linker histones may interact differently with transcriptionally active than with inactive regions of chromatin (16).
The H1 class of histones displays the most complex pattern of subtypes among the histone gene family, including differentiation-specific and tissue-specific variant proteins. The diversity of H1 histones, their differential state of phosphorylation (17, 18), and their varied distribution with respect to the stage of growth or differentiation (19) suggest that H1 subtypes also vary in their ability to confer repression throughout the genome. Both in vitro and in vivo studies of gene expression implicate the different H1 subtypes as part of a global regulatory process that is responsible for selectivity in repression of transcription (13, 20-22).
Histone genes of all classes are highly conserved at the nucleotide
level and are among the most highly expressed mammalian protein-encoding genes (23). H1 is the least conserved class of histone
proteins. There are seven H1 protein sequence variants in the mouse. H4
is the most conserved of the histone classes; there are no sequence
variants. Histone genes can be classified as
replication-dependent or -independent (24, 25) on the basis of their differential regulation of expression in the cell cycle. The
expression of replication-dependent histone genes is
tightly coupled with DNA synthesis in the eukaryotic cell. These genes are coordinately up-regulated at the G1-S boundary of the
cell cycle, but the exact molecular mechanisms responsible remain to be
elucidated (26, 27). We previously identified a coding region
activating sequence (CRAS)1 in mouse H2a.2
and H3.2 replication-dependent histone genes that is
involved in the up-regulation of these and possibly all
replication-dependent histone genes (28, 29). Subsequently,
we identified two elements within the H3.2 CRAS, the and
elements, that interact with nuclear proteins in vitro and
are required for normal gene expression in vivo in stably
transfected Chinese hamster ovary cells (30, 31). Mutation of the 7 base pairs of either the
or the
element caused a 4-fold drop in
expression in vivo.
Mouse nuclear factors, the and
, have been shown to interact
very specifically with these H3.2 elements. Mutations that change the
or
element to the sequence found in a replication-independent H3.3 gene totally abolish DNA-protein interactions and reproduce the
effects on gene expression in vivo caused by mutation of all seven nucleotides of the
or
element. Although the DNA-binding protein interacting with the
element has not been purified, we have
shown by UV cross-linking that it has an apparent molecular mass of 45 kDa (31) and must be phosphorylated on a tyrosine residue to be active
in a DNA-binding assay (32).
The and
elements are present in the coding region of all four
nucleosomal (H2a, H2b, H3, and H4) classes of
replication-dependent mouse histone genes, and the
interactions of these elements with their respective nuclear factors
are very similar if not identical (30-33). To date, no common
elements, other than the CRAS
and
elements (30, 31), have been
reported to be involved in regulating the coordinate expression of
replication-dependent histone genes of all core histone
classes.
Here we report that a second nuclear factor, a specific mouse H1
subtype, H1b, also interacts with the H3.2 CRAS element in a highly
specific manner. There have been reports that total H1 shows preferred
DNA binding toward AT-rich tracts because of the intrinsic properties
associated with A-rich sequences (34, 35). Yaneva et al.
(36) showed that DNA curvature in itself is not enough to promote
strong H1 binding and that there are additional sequence requirements
for high affinity H1 binding sites. Others have reported that H1
histones prefer some non-AT-rich eukaryotic sequences over others
(37-41). All of these previous studies utilized total cellular H1. It
should be remembered that, in asynchronously growing cells, a number of
sequence variants may be present, disguising the preferential
participation or nonparticipation in such interactions by specific H1
subtypes. Herein we show in in vitro experiments that the
pure H1b protein interacts with 100-fold less affinity with the
comparable sequence from a replication-independent H3.3 gene, differing
at only 4 out of 22 nucleotides from the H3.2 sequence. The specificity
of this interaction for sequences within the
replication-dependent gene indicates that mouse H1b may
play a distinct role in regulation of coordinate expression of histone
genes in the eukaryotic cell cycle.
Mouse myeloma cells were grown in spinner cultures to a density of 5 × 105 cells/ml in Dulbecco's modified Eagle's medium, 10% horse serum, and 5% CO2, at 37 °C. Nuclear extracts were prepared by our modification of the method of Shapiro et al. (42). Briefly, cells were harvested, washed once with phosphate-buffered saline, and resuspended in hypotonic buffer (10 mM Hepes, pH 7.9, 0.75 mM spermidine, 0.75 mM spermine, 0.1 mM EDTA, 0.1 mM EGTA, 1 mM DTT, 10 mM KCl). The cells were incubated for 15 min on ice (or until swollen), and their membranes were disrupted with 13 strokes in a homogenizer. The nuclei were pelleted at 3,000 rpm and resuspended in nuclear resuspension buffer (20 mM Hepes, pH 7.9, 0.75 mM spermidine, 0.15 mM spermine, 0.2 mM EDTA, 2 mM EGTA, 2 mM DTT, 25% glycerol). KCl was added to a final concentration of 0.6 M, and the mixture was rocked for 30 min at 4 °C. Chromatin was pelleted at 40,000 rpm for 45 min in a fixed angle rotor. The supernatant was transferred to autoclaved dialysis tubing, and the nuclear extract was dialyzed against 100-fold dialysis buffer (20 mM Hepes, pH 7.9, 0.2 mM EDTA, 0.2 mM EGTA, 100 mM KCl, 2 mM DTT, 20% glycerol) for 2 h at 4 °C. Protein concentration was measured by the Bio-Rad protein assay.
DNA-Cellulose Column ChromatographyNuclear extracts were
applied to DNA-cellulose resin (Pharmacia Biotech Inc., 40-ml bed
volume) equilibrated with buffer (20 mM HEPES, pH 7.9, 0.2 mM EGTA, 0.2 mM EDTA, 100 mM KCl,
20% glycerol, and 2 mM DTT). Proteins were eluted with the
same buffer containing 200 mM, 300 mM, 400 mM, and 1 M KCl. The collected fractions were analyzed for -binding activity by electrophoretic mobility shift assay (EMSA) and Southwestern analysis.
Synthetic oligonucleotides used in EMSA competition
experiments and Southwestern assays were synthesized with the mouse
H3.2 CRAS sequence (31) and that of the H3.3 "CRAS
." The
oligonucleotide sequences are
![]() |
![]() |
![]() |
![]() |
![]() |
Footprinting of the H1b-CRAS complex was performed by the DNase I footprinting procedure previously
described (31) with 40-mer duplex
oligonucleotides as probe.
![]() |
![]() |
![]() |
Southwestern Analysis
Southwestern blotting of SDS gels
containing proteins from the DNA-cellulose and reversed-phase HPLC
fractions was performed as described by Lenormand et al.
(47) except that the nitrocellulose membrane was incubated for 24 h at 37 °C in 10 mM Tris-HCl, pH 7.5, 50 mM
NaCl, 1 mM EDTA, 1 mM DTT (TNE-50) containing
2 × 106 cpm/ml end-labeled duplex oligonucleotides and the nonspecific competitor poly(dI-dC)-poly(dI-dC)
(50 µg/ml). Membranes were then washed with excess TNE-50, and
radioactive bands were detected by autoradiography.
The 1 M KCl
DNA-cellulose column fraction was subsequently further fractionated by
reversed-phase (RP)-HPLC on a Brownlee Aquapore RP-300 analytical
column (Rainin Instruments, MA) with 0.1% aqueous trifluoroacetic acid
and 0.08% trifluoroacetic acid in acetonitrile as mobile phases A and
B, respectively. The column was developed with a linear increase in
acetonitrile concentration from 25 to 45% at 0.3%/min, with a
constant flow rate of 1 ml/min. Column eluate was monitored at 214 nm,
and fractions were collected by hand, lyophilized, and stored at
20 °C. The HPLC system included Beckman System Gold software, run
on an IBM PS/2, the Beckman 126 solvent delivery system, an Altex 210A
sample injection valve, and a Waters 441 absorbance detector.
Proteins in the DNA-cellulose 1 M fraction and RP-HPLC fractions were further examined by acid-urea gel electrophoresis. The extraction of total H1 from mouse myeloma cells utilized the method of R. D. Cole (48). Acid-urea gel electrophoresis followed the method of Lennox and Cohen (49). Glass plate dimensions used were 26 × 25 × 0.5 cm. The gels were conditioned as described previously (49). After samples were loaded, the proteins were separated at 250 V for 12 h. Gels were stained with Amido Black, destained, and then silver-stained (Bio-Rad).
Proteolytic DigestsReversed-phase purified fractions were resuspended in 0.1 M Tris-HCl, pH 8.0, and the proteins were digested overnight at 37 °C with 1-2% (w/w) V8 protease. Peptides were separated by RP-HPLC on a Deltabond® Octyl analytical column (Keystone Scientific) with a linear acetonitrile gradient from 0 to 50% at 0.5%/min. Absorbance was monitored at 214 nm, and fractions were collected manually and dried for automated protein sequenation.
Amino Acid Composition AnalysisAmino acid composition data were obtained by the Pico-Tag® method (Waters Chromatography Division, MA). Briefly, dried protein samples were subjected to vapor phase hydrolysis with 6 N HCl in a MDS-200 microwave sample preparation system (CEM Corp.). Hydrolyzed samples were derivatized with phenylisothiocyanate according to standard methods and run on a Waters HPLC system controlled by a Maxima 820 Workstation for data acquisition and analysis. Amino acids were quantitated by comparison with Amino Acid Standard H (Pierce).
We utilized DNA-cellulose chromatography as
the first step in the purification of the CRAS -binding factor(s)
from crude mouse myeloma cell nuclear extract. Fig. 1
shows a gel mobility shift assay of the DNA-cellulose fractions, using
as probe radioactively labeled
oligonucleotides that contain the
CRAS
-binding site CGAGATC (31). The
-binding activity eluted in
the 0.3 M fraction (lane 5). We have previously
shown that the binding activity eluting in the 0.3 M
fraction protects the
sequence in a DNase I footprinting assay and
that the H3.2
duplex oligonucleotides, but not the H3.3
duplex
oligonucleotides, act as specific competitors for the
-binding
activity.
We further examined the DNA-cellulose fractions by Southwestern
analysis. The results are shown in Fig. 2A.
Radioactively labeled duplex H3.2 oligonucleotides or,
alternatively, the corresponding sequence from a
replication-independent H3.3 gene that we have shown cannot act as a
binding site for the CRAS
factor (31) was used as probes. The two
oligonucleotide sequences are compared in panel C. Before
hybridization with radioactive probe, the proteins in the column
fractions were separated on SDS gels and blotted onto nitrocellulose
membranes as described under "Experimental Procedures." In
lane 1 of Fig. 2A, an intense signal is observed
as a result of an interaction between the H3.2 probe and a protein with
an apparent molecular mass of 30 kDa in the crude nuclear extract. The
0.3 M fraction showed no evidence of protein interaction
with the H3.2 oligonucleotides (Fig. 2A, lane 3),
although as shown above, this fraction contained the specific
-binding activity observed in EMSA (Fig. 1, lane 5). In
Fig. 2A, the DNA-protein interaction observed in crude
extract (lane 1) was also observed in lane 4,
showing that the interacting protein (apparent molecular mass, 30 kDa)
was contained in the 1 M DNA-cellulose fraction. This
interaction was not detected when the H3.3
oligonucleotides were
used as probe (lanes 5 and 8). Direct
quantitation of the radioactivity bound to the membranes showed that
the H3.3 oligonucleotides did interact to some extent with the protein
producing the strong signal seen in lanes 1 and 4, but the amount of bound probe was over 100-fold less than
that bound in the comparable lanes when the H3.2 oligonucleotides were used. The H3.2 and H3.3 oligonucleotide sequences differ at only four
nucleotide positions, three of which are in the
element (see Fig.
2C).
The highly specific nature of the interaction between the protein(s) in
the 1 M fraction and the H3.2 sequence was confirmed in
a separate southwestern competition experiment, shown in Fig. 2B. Unlabeled duplex H3.2
oligonucleotides competed very
efficiently with the radioactive H3.2 probe, eliminating the
interaction when in 100-fold excess (lane 3). Conversely,
the H3.3
competitor duplex did not compete with the H3.2 probe
(lanes 4 and 5) even when present in 100-fold
molar excess. Because the two oligonucleotide duplexes differ at only 4 nucleotide positions (Fig. 2C), these results show that the
DNA-protein interactions detected in the Southwestern analysis are
highly specific for the H3.2 CRAS
sequence.
Further purification of the proteins in the 1 M fraction was achieved by RP-HPLC. The RP-HPLC profile of
the 1 M fraction is presented in Fig.
3A as peaks of absorbance at 214 nm.
Subsequent SDS-polyacrylamide gel electrophoresis analysis of the
collected fractions A, B, and C is shown in Fig. 3B,
lanes 1-3. Silver staining of the SDS gel revealed that
fractions A-C contain proteins with apparent molecular masses of about
30-35 kDa, whereas lane 4 (1 M fraction)
contained multiple proteins of much higher molecular mass (data not
shown).
A repeat of the Southwestern assay in Fig. 2 using the H3.2 and H3.3
oligonucleotides as probes with the RP-HPLC fraction A is shown in
Fig. 3C. The protein(s) contained in fraction A showed the
same interaction with the H3.2
sequence as seen in Fig. 2,
indicated by the intense radioactive band observed in lane
1. As was true for the proteins in the 1 M
DNA-cellulose fraction (Fig. 2A, lane 8), the
protein(s) in RP-HPLC fraction A showed 100-fold less affinity for the
H3.3
sequence (lane 2).
The strong affinity of mouse nuclear proteins contained in the 1 M DNA-cellulose fraction for DNA (elution with 1 M salt) and their rapid mobility in SDS-polyacrylamide gel
electrophoresis suggested that these proteins might be histones.
Although SDS-polyacrylamide gel electrophoresis theoretically separates
proteins according to their molecular weight, histones behave
anomalously in such gels because of post-translational modifications
such as phosphorylation (25). Acid-urea gels have been used
successfully to separate histones on the basis of differences in net
charge and mass (50). This method accomplishes the separation of
modified and unmodified forms of the basic histone proteins and allows
for the partial separation of the various H1 subtypes as well as their
differentially phosphorylated forms (51). In fact, only highly basic
proteins (like histones) enter these gels. In Fig. 4,
acid-urea gel analysis of column fractions is shown. Total H1 prepared
from mouse chromatin is shown in lane 1. The 1 M
DNA-cellulose fraction was loaded in lane 2. The proteins
observed in this lane migrate identically to those observed in
lane 1 (total H1). Acid gel separation of proteins in
RP-HPLC fraction A is shown in lane 3. The subset of bands
seen in this lane is also observed in lane 1 (total H1). The
results shown in Fig. 4 are consistent with the hypothesis that
protein(s) in the 1 M fraction that interact preferentially with the H3.2 sequence are H1 histones.
To identify conclusively the proteins interacting in such a highly
specific manner with the H3.2 sequence, we digested the protein
contained in RP-HPLC fraction A with Staphylococcus aureus V8 protease. The resulting peptides were fractionated by RP-HPLC and
analyzed by automated protein sequenation. Fig.
5A shows the RP-HPLC elution profile of these
peptides. Retention times are recorded above the
peaks, and the sequenced fragments are numbered at the
bases of their respective peaks. Sequence was
obtained for amino acids at positions 9-115. The amino acid sequence
of the purified protein is shown in panel B, the underlined
amino acids designating the primary sequence information obtained in this study. Comparison of the protein sequence we obtained by peptide
sequenation to published H1 sequences identified the H1 protein in
RP-HPLC fraction A as a single mouse H1 subtype, namely H1b. No
peptides were recovered from the extremely basic carboxyl-terminal 100 amino acids, despite repeated attempts, but amino acid composition analyses of RP-HPLC fraction A, performed by the Pico-Tag®
method, were consistent with the interpretation that the purified protein contained these amino acids (Table I). In
addition, mass spectrometric analysis produced results consistent with
that of full-length H1 protein (data not shown). The gene encoding this unique mouse H1 has since been cloned independently by two laboratories (see "Discussion"). Although several protein bands are observed in
the SDS gel shown in Fig. 3B, lane 1 (RP-HPLC
fraction A), these are very likely to be different post-translationally
modified forms of H1b. The protein contained in the RP-HPLC fraction A, via automated peptide sequencing, produced sequence from over 100 amino
acid positions, in some cases from two or more peptides (Fig.
5B). Every amino acid position sequenced matched the
sequence of the mouse H1b sequence (52).2
This is unambiguous evidence that a single protein is contained in fraction A.
|
Next, we examined the interactions of this H1 subtype with
the H3.2 sequence in EMSA. The results, using end-labeled duplex H3.2
oligonucleotides as probe, are shown in Fig. 6.
Purified H1 protein (RP-HPLC fraction A) was added in increasing
amounts (lanes 1-6). In lane 3 (35 ng of added
pure H1b), there was no indication of any interaction, even on much
longer exposures, but most of the probe was shifted in a diffuse smear
from the position of free probe to the top of the gel when 50 ng of
pure H1b was added (lane 4). This was also true at 55 ng
(lane 5), and when 60 ng of H1b was added (lane
6), all of the mobility-shifted probe was at the top of the gel. A
duplicate experiment performed with another pure H1 subtype (H1a)
revealed no evidence of interaction with the H3.2
element
under these conditions, even in the presence of
-binding activity
(data not shown).
Because another nuclear protein (the factor, Fig. 1, lanes
1 and 5) was identified on the basis of its interaction
with the H3.2
sequence in EMSA, we examined the possibility that interactions of the H1b protein would be detectable in the presence of
the
factor. In Fig. 7, we show the results of such
an experiment. The source of
-binding activity in this experiment
was crude nuclear extract, added in identical amounts to all lanes.
Crude nuclear extract and probe were incubated on ice for 15 min and then increasing amounts of H1b were added over the same range of
concentrations used in Fig. 6, and the reactions were incubated an
additional 15 min (lanes 2-8). The
complex, the result
of interaction between the
factor and the H3.2
sequence, is
observed in lane 1 (compare with Fig. 1, lane 1).
There is some evidence of nuclear protein-DNA-H1b interactions at
10-20 ng of added H1b (lanes 3 and 4), but at 35 ng a "supershifting" of the
complex is observed (lane
5). This supershifting increased with the addition of higher
concentrations of H1b (lanes 6 and 7), and the
complex was no longer observed at 100 ng of added H1b (lane
8).
Two things are remarkable about these interactions. First, evidence of
H1 binding to the DNA duplex is detected at a much lower H1
concentration than that required for detection of the interaction of
the pure H1b protein with the
sequence alone (compare Fig. 7,
lanes 3-5 (10-35 ng) with Fig. 6, lane 4 (50 ng
of H1b)). Second, the amount of H1 required to shift most of the probe
to the top of the gel is also higher when the
factor is present
(compare lane 6 of Fig. 6 (60 ng of H1b) to lane
8 of Fig. 7 (100 ng)). When pure H1b was added to the binding
reaction with the
probe at a molar ratio of 1:1 (Fig. 6, lane
2), no evidence of interaction was detected. Only when the molar
ratio of H1 exceeded that of DNA probe by greater than 2:1 (Fig. 6, lane 4) were interactions observed. Conversely, when the
CRAS
-binding activity was present (Fig. 7, lane 1),
H1-CRAS
complex interactions were detected at a much lower molar
ratio (Fig. 7, lane 3, 0.5:1; lane 4, 1:1),
indicating that binding by H1b to the
DNA sequence may be enhanced
by the presence of the preincubated
factor-DNA complex. We know
that our crude nuclear extracts contain H1; these extracts are the
source of our RP-HPLC-purified H1b. The apparent enhancement of H1b
interaction with the
complex might simply be due to the presence of
H1b molecules in crude nuclear extract, but the molar amount of H1b
present in the crude extract added to the EMSA reactions is quite low.
Therefore, it is unlikely that this is the explanation for the
differences observed between Figs. 6 and 7. As the next experiment will
demonstrate (Fig. 8), it is possible that DNA structure
plays a role in the H1b-
-DNA interaction. A change in the
DNA
structure upon interaction with the
-binding activity may be
responsible for the apparent enhancement of H1b interactions observed
in Fig. 7. Until the
factor is purified, however, we cannot
distinguish between these possibilities.
Evidence of Specific Effects upon DNA Structure by H1b Participation in the
Elsewhere
(30, 31), we have shown that two elements, and
, within the
coding sequence of core histone genes interact with nuclear proteins
in vitro and are required for normal gene expression
in vivo in stably transfected Chinese hamster ovary cells.
Mutation of the seven conserved nucleotides of the histone
sequence
caused a 4-fold drop in expression of a mouse H3.2 gene in stable
transfectants and abolished the
complex observed in gel mobility
shift experiments. In Fig. 8, we compare the DNA-protein interactions
at the
element (shown in gel shift assay in Fig. 1, lane
1) by DNase I footprint analyses, performed in the absence and in
the presence of purified H1b. The labeled strand in these experiments
was the noncoding strand, and the
sequence is indicated by a
bracket. G indicates that the lane was loaded
with the Maxam-Gilbert G reaction cleavage products of the probe;
B indicates that the lane was loaded with the DNase
I-treated complex of probe-
factor after elution from a native gel
slice (see "Experimental Procedures"); and F indicates free probe
after elution as described for the B lanes.
As previously shown (31), nucleotides within the histone sequence
(GCTCTAG on the noncoding strand) show protection from cleavage by
DNase I (Fig. 8, minus H1b (
H1b) panel, B
lane). In the panel showing experiments performed in
the presence of H1b (+H1b), labeled
oligonucleotides
were preincubated with crude nuclear extract before the addition of
pure H1b exactly as described in Fig. 7. DNase I digestion followed,
and then the reaction was loaded onto a native gel for separation of
free and bound probe molecules by EMSA. For reactions shown in the
+H1b panel, the supershifted
DNA-protein complex, which
migrates slightly behind the histone
complex (Fig. 7, lanes
4 and 5), was sliced from the native gel and treated as
described above (for details, see "Experimental Procedures").
Strong protection of the first two nucleotides (GC) in the element
by protein interaction is observed in both +H1b and
H1b panels, but several DNase I-hypersensitive sites not
seen in the
H1b panel are observed in the +H1b
panel, lane B. The C nucleotide immediately outside the
element shows hypersensitivity to DNase I cleavage, as do three
more nucleotides in the nine nucleotides 5
of the
element. These
hypersensitive sites must result from the specific interactions of H1b
molecules with the preincubated histone
complex. The increased
sensitivity of these nucleotides indicates that the addition of H1b to
the
DNA-protein complex has a very specific effect on the DNA
structure immediately 5
of the
element. The structure of the
duplex
oligonucleotides in solution, whether free or bound, is such
that DNA protein interactions 3
of the seven nucleotides composing the
element cannot be detected. More detailed analyses of the
H1b-
-DNA interaction will be possible when the
-binding activity
is purified to homogeneity.
We have previously shown that the H3.2 CRAS element is
required for normal expression of the mouse H3.2 gene in
vivo (31). We have also reported that this regulatory element is
present in the coding regions of replication-dependent
histone genes of all four nucleosomal classes and that the DNA-protein
interactions are identical for these genes. In addition, we have
reported that the
factor (DNA-binding activity) is present in
G1, but not in S or M phase, nuclear extracts (32). The
fact that a replication-independent histone gene, an H3.3, has a
mutated
sequence that fails to bind or compete for binding of the
factor implicates this element in the coordinate regulation of the
replication-dependent histone genes (31, 32).
Here, we have reported the purification by DNA-cellulose chromatography
and RP-HPLC of a second nuclear factor that interacts specifically with
the element, a specific mouse H1 subtype, H1b. H1b does not
interact with high affinity with the comparable sequence from a
replication-independent H3.3 gene. In fact, the H3.2 DNA-H1
interactions are so specific that the H1b protein interacted with over
100-fold lower affinity with the H3.3
oligonucleotides (Figs. 2 and
3). Yet, the H3.2 and H3.3 oligonucleotides differ at only 4 of 22 nucleotide positions (Fig. 2C). These results, also
demonstrated in a very different set of experiments (EMSA), clearly
show the binding specificity of the H1b histone in these interactions.
In 1993, Brown and Sittman developed an HPLC protocol to separate mouse H1 protein sequence variants (54). On the basis of relative mobilities of purified HPLC peak fractions in SDS and acid gel electrophoresis, they determined which variant was represented in each HPLC peak. Their HPLC peaks had the overall mobility characteristics on gels of the major H1 variants previously demonstrated by others (25, 55). Our RP-HPLC purification (Fig. 3A) and acid-urea gel electrophoresis (Fig. 4) reproduced these results. Peptide sequenation yielded information on over 100 amino acid positions of RP-HPLC fraction A, identifying the protein as H1b. This H1 is highly related to rat H1b, rabbit H1.3, and human H1-3 variants (56). Comparison of the protein sequence to published H1 sequences identified the H1 subtype in RP-HPLC fraction A as a unique mouse H1 subtype (H1b), the cloning of which was recently reported by the laboratory of D. Doenecke (52) and which has also been independently cloned by the laboratory of W. F. Marzluff.2 The Doenecke group designated the H1b subtype H1var.4, but we have chosen to adhere to the original nomenclature introduced by Lennox and Cohen (25, 57). The mouse H1b subtype identification was confirmed by mass spectrometry of fraction A (data not shown), in the laboratory of A. I. Skoultchi.
The identification of numerous histone H1 variants in vertebrates suggests that these proteins accomplish specialized functions during development and that their relative abundances are perhaps dependent upon the differentiated cell type and whether cells are proliferating or growth-arrested. Also, H1 subtypes differ in their relative rates of synthesis and degradation in dividing and in nondividing cells (56). For example, H1b is unstable after cells cease to proliferate and is not synthesized in nonlymphoid, nondividing cells (25). In the newborn mouse, H1b declines after 4 weeks of age and cannot be detected by Coomassie Blue after 16 weeks (58). H1b is very low in adult mouse tissues, representing for example only 3% of total H1 in liver chromatin (59).
In addition, there are differences in the amount and pattern of phosphorylation among H1 subtypes in humans and mice (18, 25, 53, 56). H1b was observed as having the largest number of phosphorylated forms among the H1 subtypes (25). It has been known for many years that there are variations in the amount of phosphorylation of specific H1 subtypes during the cell division cycle (for a review, see Ref. 25), and a large amount of data shows that H1 dephosphorylation correlates with chromatin condensation (13). Consistent with these observations, phosphorylated H1 was not detected in nonproliferating adult tissues (58). It is completely consistent with these data to hypothesize that H1b histones play a key role in the regulation of genes expressed only in proliferating cells, i.e. the replication-dependent histone gene family.
Our results are in agreement with other reports that H1 histones prefer
some non-A/T eukaryotic sequences over others (37-41). However, this
may be the first set of experiments to examine site-specific DNA
interactions of a single H1 subtype. Our data support the hypothesis
that specific H1 subtypes have specific roles to play in the regulation
of specific genes or gene families. We have shown that a specific H1
subtype, H1b, exhibits a highly specific affinity for a
cis-acting DNA regulatory sequence, the CRAS element, in
the Southwestern assay. We have also shown by EMSA that H1b is capable
of binding to the
element in a fashion noncompetitive with the CRAS
factor (Fig. 7). The fact that a specific H1 subtype is responsible
for this high affinity interaction, observed in vitro,
provides evidence that, although different H1 subtypes may be present
in the same transformed cell line in asynchronous logarithmic growth
(mouse myeloma cells), these H1 subtypes may vary in their abilities to
interact with specific DNA sequences.
Our observations raise a number of interesting possibilities. First, in
the interaction between the element and H1b, is the phosphorylated
or nonphosphorylated form of H1b required? Second, is H1b involved in
nucleosome positioning on replication-dependent histone
genes by way of the
sequence? Finally, does the
factor play a
role in this interaction? A very attractive hypothesis is that the
mouse H1b protein is involved in repression of the nucleosomal
replication-dependent histone genes. The involvement of a
specific H1 subtype in regulation of all
replication-dependent histone gene expression could provide
at least part of a common mechanism for up-regulation of this large
gene family at the G1-S boundary of the eukaryotic cell
cycle. We will examine this possibility in future studies.
We thank Dr. Richard Cooke at the Core Protein Sequenation Facility at Baylor College of Medicine for expertise and Drs. Arthur I. Skoultchi and William F. Marzluff for making available unpublished results. Finally, we thank Dr. Allen M. Sirotkin in the laboratory of Dr. Arthur Skoultchi for assisting in the mass spectrometry of our RP-HPLC fractions.