ABSTRACT
We have analyzed the chromatin structure of mouse ribosomal RNA genes (rDNA) by
partial digestion of genomic DNA with micrococcal nuclease (MNase), DNase I and
identified hypersensitive sites by indirect end-labeling. This analysis has revealed defined regions of nuclease hypersensitivity in the intergenic spacer which in turn coincide with regulatory elements. Hypersensitive sites map to the transcription initiation site, the enhancer repeats, the spacer promoter and two sequence elements which coincide with amplification-promoting sequences. Analysis of the DNA curvature by computer
modeling uncovered a striking correlation between sequence-directed structural features of regulatory regions and the position of
nuclease hypersensitive sites. Moreover, we demonstrate that nucleosomes are
specifically positioned upstream and downstream of the transcription start
site.
In vitro
studies using chromatin assembled in the presence of
Drosophila
embryo extracts show that binding of the transcription termination factor TTF-I to the upstream terminator mediates this specific nucleosome positioning
at the rDNA promoter in an ATP- dependent fashion.
Since the initial observation that actively transcribed gene sequences are
preferentially susceptible to digestion by DNase I (
1
,
2
), nuclease digestion has been widely used to study the relationship between
chromatin structure and gene expression. Numerous studies revealed that sites
hypersensitive to nuclease cleavage frequently appear in regions of chromatin
engaged in transcription (
3
), suggesting that these sequences are exposed. Hypersensitive sites represent
discontinuities, or gaps, in the nucleosome array of the chromatin fiber.
Biochemical analyses of hypersensitive regions as well as footprinting of chromatin structure in isolated nuclei support the conclusion that these regions are free of
nucleosomes (
4
). Hypersensitive sites have been mapped to specific positions of known
function, including promoters, upstream activating sequences, transcriptional
enhancers and silencers, origins of replication, recombination elements and
structural elements within or around telomeres and centromeres (for review, see ref.
5
). Thus, the mapping of nuclease sensitive sites within a gene locus offers a
means to locate potential
cis
-regulatory elements.
The importance of the structural and functional organization of chromatin is
being increasingly recognized as it becomes more experimentally accessible. The
genes encoding for ribosomal RNA provide a favorable system for studying the
interrelation between chromatin structure and transcriptional activity. The
ribosomal RNA genes of eukaryotes are organized in tandem arrays in which the transcribed regions are separated by an intergenic spacer region (for review, see ref.
6
). Most, if not all, of the sequence elements that govern transcription and
replication of rDNA are located within the intergenic spacer. Regulatory elements that direct transcription include (i) the gene promoter at the 5'-end of the pre-rRNA coding region, (ii) a transcription terminator immediately upstream of the gene promoter, (iii) enhancer elements that stimulate transcription, (iv) one or more
spacer promoters, and (v) terminator elements at the 3'-end of the pre-rRNA coding region. Moreover, sequences that play a role in
replication of rDNA include an
We are interested in understanding the role of chromatin structure in rDNA transcriptional regulation. There is no consensus in the literature
with regard to the chromatin structure of ribosomal genes. Electron microscopic
visualization of actively transcribed ribosomal genes has revealed that the DNA
comprising the pre-rRNA coding region is in a highly extended conformation suggesting that
they contain either altered nucleosomes or even no nucleosome-like structures at all (
7
,
8
). Also the intergenic spacer did not have a normal beaded structure typical of
bulk DNA (
8
,
9
). In contrast, nuclease digestion patterns of
Xenopus
oocyte ribosomal chromatin were consistent with an appreciable fraction of the
rDNA being contained in nucleosome-like structures (
10
). Accessible and protected sites, alternating with the periodicity of an array
of nucleosomes, were also found in the spacer regions between rRNA genes of
Tetrahymena
(
11
).
It was recognized early that alterations in chromatin structure accompany
transcriptional activation, and that nucleosomes occupy the promoters of
inactive genes and can repress their transcription (reviewed in ref.
12
). Regulatory sequences which are recognized by specific DNA binding proteins
are often free of nucleosomes in active and potentially active genes, and are
more sensitive to digestion by nucleases than bulk chromatin. Our long term
goal is to understand the molecular mechanisms which regulate rRNA synthesis
during cell growth and differentiation. As chromatin is the physiological
template for RNA polymerase I (Pol I), and biochemical and genetic evidence
indicates that chromatin structure plays an active role in transcription, we
have analyzed the chromatin structure of mouse rDNA. We found a number of
nuclease hypersensitive sites within the intergenic spacer which coincide with
regulatory elements involved in transcription and replication of rDNA.
Moreover, since sequence-directed bending of DNA causes local variations in the structure of
genomes (
13
) and intrinsically bent DNA is frequently found near functionally important
sequences (
14
,
15
), we have also analyzed sequence-directed DNA curvature by computer modeling. This analysis has revealed a striking correlation between regulatory elements in the intergenic spacer, the position of nuclease hypersensitive sites, and regions of increased sequence-dependent DNA curvature.
The plasmid pMr974 contains a genomic 11.3 kb
Eco
RI fragment from mouse extending from -5635 to +5646 with respect to the transcription initiation site. The
subclone pMrWT contains mouse rDNA sequences from -170 to +155 cloned into pUC9. To map nuclease hypersensitive sites in
genomic rDNA by indirect end-labeling, Southern blots were hybridized to a probe complementary to rDNA
sequences from
-5635 to -5494 (
Eco
RI) or to a probe extending from +737 to +1291 (
Xho
I). Chromatin assembled on pMrWT was analyzed by
Nde
I digestion and hybridization to a 207 bp
Eco
RI-
Nde
I fragment derived from pUC9.
Mouse NIH3T3 cells suspended in 1.5 ml digestion buffer (15 mM Tris-HCl pH 7.5, 15 mM NaCl, 60 mM KCl, 5 mM MgCl
2
, 300 mM sucrose, 0.5 mM EGTA, 0.1% NP-40) were incubated for 8 min with varying amounts of DNase I (Worthington; 40- 200 U/ml) or micrococcal nuclease (Sigma; 20-100 U/ml). The reaction was stopped by addition of 0.5 ml
100 mM EDTA/4% SDS. After incubation for 1 h at 37oC with 0.1 mg/ml RNase A, cellular proteins were digested overnight by treatment with 0.1 mg/ml proteinase K. DNA was purified by phenol-chloroform extraction and ethanol precipitation. To map nuclease hypersensitive sites by indirect end-labeling, genomic DNA was digested with an appropriate restriction enzyme and the fragments were
separated on agarose gels. DNA was denatured, transferred to nylon filters and hybridized
to probes labeled by random oligonucleotide priming (Megaprime, Amersham).
To identify the nucleosome positions by oligonucleotide hybridization, chromatin
was digested extensively with MNase (500 U/ml) to get mainly the mononucleosome. After purification, the DNA was electrophoresed on a 1.3% agarose gel and transferred to nylon
filters. The filters were hybridized with radiolabeled oligonucleotides
complementary to defined rDNA regions both upstream (-90/-71) and downstream (+111/+130) of the transcription start site as
well as including (-7/+16) the transcription initiation site.
Preparation of assembly extracts and chromatin reconstitution was performed as
described (
16
). Reactions (40 [mu]l) containing 200 ng of either pMr974 or pMrWT and 11 [mu]l
Drosophila
embryo extract were assembled into chromatin for 6 h at 26oC. Chromatin was complemented with TTF-I after completion of chromatin assembly and incubated for further 30
min. ATP was removed by the addition of 1 U apyrase and incubation for 20 min.
To map nuclease hypersensitive sites, assembled chromatin was digested with 10
U micrococcal nuclease for 20 and 60 s in a total volume of 100 [mu]l and in the presence of 3 mM CaCl
2
. The reactions were stopped by the addition of 0.2 vol 4% SDS/0.1 M EDTA.
Proteins were digested with 10 [mu]g proteinase K for 1 h at 50oC. Isolated DNA was cleaved with the respective restriction enzyme, separated on agarose gels, blotted and hybridized with specific probes.
TTF-I was expressed by infecting 2.5 * 10
8
Sf9 cells with recombinant baculovirus derived from pBac-mTTF[Delta]N185. After 48 h, the cells were harvested, rinsed in PBS and
resuspended in 3 vol lysis buffer (50 mM HEPES-KOH pH 7.8, 300 mM KCl, 5 mM MgCl
2
, 1 mM [beta]-mercaptoethanol, 1 mM PMSF, 1 [mu]g/ml leupeptine). Cells were lysed by sonification followed by
addition of 0.5% NP-40 and centrifugation. Imidazole (1 mM) was added to the supernatant and
incubated with NTA-agarose beads (Qiagen) for 30 min at 4oC. The beads were washed with 20 column volumes of buffer 1 (50 mM HEPES-KOH pH 7.8, 300 mM KCl, 5 mM MgCl
2
, 0.5% NP-40, 1 mM imidazole, 1 mM PMSF, 1 [mu]g/ml leupeptine), 20 volumes of buffer 2 (same as buffer 1 with 1 M KCl)
and 20 volumes of buffer 3 (same as buffer 1 with 10 mM imidazole). Protein was eluted with five volumes of buffer 4 (20 mM HEPES-KOH pH 7.8, 100 mM KCl, 5 mM MgCl
2
, 200 mM imidazole, 1 mM PMSF, 1 [mu]g/ml leupeptine) and dialyzed against buffer AM-100 (20 mM Tris-HCl pH 7.9, 5 mM MgCl
2
, 0.1 mM EDTA, 20% glycerol, 2 mM DTE, 100 mM KCl, supplemented with protease
inhibitors). The DNA binding activity of TTF-I was determined by electrophoretic mobility shift assays (
17
).
Our program
Curvature
implements several known algorithms for calculating DNA space curves (Schätz and Langowski, in preparation). In this study we have used three
nearest-neighbor models described before (
18
-
20
). Curvature plots were generated from the space curves by determining the
curvature of an approximating arc through DNA segments of 21 bp. One curvature unit corresponded to the mean DNA curvature in the crystallized nucleosome (1
cu = 42.8 Å). The curvature plots were smoothed using a sliding average over a
window of 250 bp. The predictions were normalized by dividing through the
average curvature of the rDNA sequences from -5500 to +10 000 and then averaged between the three models.
To determine the chromatin structure of the
murine intergenic rDNA spacer, cellular DNA was partially digested with increasing amounts of DNase I or micrococcal nuclease (MNase) in permeabilized NIH3T3
cells. The DNA was purified, digested with
Eco
RI and, following electrophoresis and blotting, labeled by hybridization with a
probe which anneals next to the upstream
Eco
RI site (from -5635 to -5494 with respect to the transcription start site). As shown on
the Southern blots in Figure
1
B, this indirect end-labeling technique labels the parental ~11 kb
Eco
RI rDNA restriction fragment as well as several distinct rDNA fragments which
are diagnostic for nuclease hypersensitive sites. There is a strong
hypersensitive site which coincides with the transcription start site (+1).
Moreover, several regions within the intergenic spacer upstream of the
initiation site exhibit a pronounced sensitivity to DNase I and MNase. Strong
cleavage was observed at nucleotides -1200, -1980, -4150, -4280 and -4440. Less pronounced hypersensitive sites
mapped to positions -2600, -2900 and -3300 (Fig.
1
B, lanes 1-4). Both the positions and relative intensities of the hypersensitive
sites identified with the upstream probe were confirmed by indirect end-labeling of genomic DNA digested with
Sal
I,
Ava
II and
Xho
I, respectively (data not shown).
An increasing number of studies have focused on the potential role of DNA
structure in directing the stereospecific assembly of nucleoprotein complexes.
It has been shown that sequence- directed bending of DNA causes local variations in regulatory gene
regions which in turn may facilitate the binding of specific proteins (
14
,
15
). Since the basic rules of DNA curvature are now well established, we have
analyzed the structure of rDNA by computer modeling. We have used three
independent models of DNA curvature (
18
-
20
), all of which have been shown to
fit to
the experimental data on electrophoretic mobility or cyclization of small DNA
fragments. Figure
2
shows the averaged computed curvature of rDNA sequences from -5500 to +5500. The magnitude of bending is expressed as the local bending
relative to the overall DNA curvature averaged over the complete rDNA sequence. The upstream region displays distinct sites of significant DNA curvature. The most prominent peaks are centered at the transcription start
site, the spacer promoter and within the APS elements. Three more peaks can be
discerned around positions -2200, -2600 and -3300. Significantly, all regions of increased curvature
coincide with hypersensitive sites identified by the indirect end-labeling analysis. Consistent with the less pronounced nuclease sensitivity of the transcribed region, the computer modeling
reveals a significantly lower average curvature of the coding region as
compared to spacer sequences.
As mentioned in the introduction, the question whether or not active rDNA genes
are in a nucleosomal configuration is not unequivocally solved. To address this
issue, we have mapped nuclease hypersensitive sites around the rDNA
transcription initiation site. In the experiment shown in Figure
3
A, cellular chromatin was partially digested with DNase I and MNase,
respectively, cut with
Xho
I at nucleotide +1291 and hybridized to a probe containing rDNA sequences from
+737 to +1291. Consistent with the results shown above, the strongest signal is
observed at the boundary between the non-transcribed and the transcribed region. The region downstream of the start
site is characterized by rather uniform distribution of regularly spaced sites
which may reflect a nucleosomal configuration of the rDNA. Interestingly, the
transcription start site is preceded by a MNase resistant region whose size (~160 bp) suggests that it may represent a positioned nucleosome.
Studies on the influence of chromatin structure on transcription would be greatly facilitated if the specific protein-DNA interactions could be reproduced on cloned DNA. To assemble chromatin
in vitro,
a circular plasmid containing mouse ribosomal gene sequences from -5635 to +5665 was incubated with extracts from
Drosophila
embryos (
16
). In this system plasmid DNA is known to be faithfully assembled into chromatin
(
22
). The
in vitro
assembled chromatin was partially digested with MNase, the protected DNA
fragments were separated by electrophoresis, blotted and hybridized to a labeled probe containing rDNA sequences from -170 to +155. As shown in Figure
4
A, a ladder of oligonucleosome-sized fragments was observed, indicating that the DNA was assembled into
an array of regularly spaced nucleosomes.
To analyze the chromatin structure of the ribosomal gene promoter in more
detail, pMrWT, a plasmid containing rDNA sequences from -170 to +155, was assembled into chromatin and MNase sensitive sites were
monitored. We have found that binding of the transcription termination factor
TTF-I to the promoter-proximal terminator element T
0
mediates ATP-dependent chromatin remodeling and activation of Pol I-directed transcription on nucleosomal templates (Längst
et al
., in press). Therefore chromatin reconstitution was performed on pMrWT and recombinant TTF-I was added after completion of chromatin assembly. The MNase digestion pattern of the
in vitro
reconstituted chromatin is shown in Figure
5
. The cleavage pattern of chromatin assembled in the absence of TTF-I closely resembles that of naked DNA (data not shown) indicating that the
nucleosomes are randomly distributed on the rDNA plasmid (lanes 1 and 2). If,
however, chromatin was incubated with TTF-I, a different pattern was obtained (lanes 3 and 4). Two strong
hypersensitive sites flanking the T
0
sequence (from -180 to -160) were induced and strongly preferred MNase cleavage sites were
protected in adjacent regions. The nuclease resistant region at the rDNA
promoter assembled
in vitro
covers sequences from -160 to -5, which probably corresponds to the protected region observed
in vivo.
Thus, the interaction of TTF-I with the upstream terminator led to a repositioning of nucleosomes, from
more or less random positions to defined positions flanking the TTF-I target site. Consistent with earlier results, the changes in TTF-I-mediated chromatin structure requires energy. Inclusion of apyrase which depletes the assembly reaction from ATP prevented TTF-I-induced remodeling (lanes 5 and 6) and, therefore
yielded a MNase digestion pattern that is indistinguishable from chromatin assembled in the absence of TTF-I.
rDNA is characterized by a number of interesting functional properties,
including high levels of Pol I transcription, high levels of recombination (
23
,
24
), differential replication in somatic and germ cells (
25
,
26
), and in sex chromosome pairing during meiosis (
27
). There appears to be a consensus that regulatory sequences, like promoter elements, enhancers, origins of replication, etc. which are recognized by specific DNA binding proteins, are free of
nucleosomes in active genes, and are more sensitive to digestion by nucleases
than bulk chromatin. As a prelude to a biochemical dissection of the role of
the chromatin structure on rDNA transcription, we have analyzed both the
localization of nuclease hypersensitive sites and sequence-directed curvature of a part of the intergenic spacer and the transcribed
region. We found a striking correlation between the localization of functionally important regulatory elements within the rDNA spacer, the position of nuclease
hypersensitive sites, and regions of increased local curvature. The hypersensitive sites correspond to regulatory elements
which play an important role in transcription and replication.
A previous analysis has revealed two sequence elements within the spacer which increase the frequency of amplification-dependent transformation of mouse cells. These amplification-promoting sequences, APS-1 and APS-2, have been mapped to two fragments extending from -4945 to -4576 and -4530 to -4105, respectively (
21
). MNase has been found to cleave preferentially within these two elements (
28
). The results shown in this study confirm and extend these earlier observations. The most pronounced hypersensitive site maps to the transcription initiation site of the 45S pre-rRNA gene promoter. Similarly, the start site of the spacer promoter at -1996 is preferentially cleaved by MNase. The mouse spacer promoter
has been demonstrated to exhibit a very low transcriptional activity
in vivo
, but to exert a stimulatory effect on transcription from the downstream rRNA
gene promoter (
29
). Our finding that the spacer promoter is preferentially cleaved by MNase
suggests that proteins bound to the spacer promoter may confer a particular
configuration to DNA which may favor transcription from the gene promoter. The hypersensitive sites are evidently imposed by DNA bound proteins and are not inherent in the
rDNA sequence, as shown for protein-free recombinant rDNA.
The finding that naked DNA or chromatin assembled
in vitro
does not show this specific cleavage pattern is in apparent contrast to the
computer modeling studies which indicated that rDNA regions which are
preferentially cleaved by nucleases also exhibit a significant sequence-dependent curvature. This observation suggests that curved DNA may induce
a particular chromatin structure which in turn facilitates the interaction of sequence-specific DNA binding proteins. Consistent with this interpretation, the most
pronounced coincidence between hypersensitivity and intrinsically bent DNA
structure was observed at the sites of transcription initiation of 45S pre-rRNA and of spacer promoter transcripts. Apparently, these gene regions
exhibit unusual structural features. Indeed, analysis of sequence-directed structural features of eukaryotic ribosomal gene promoters has revealed striking
structural homologies of rDNA from different taxonomic groups (
30
,
31
). This observation, together with the fact that the sequences of ribosomal gene promoters have diverged significantly (for review, see ref.
6
) suggests that conservation of DNA structure, rather than DNA sequence, is
fundamental for rDNA promoter function.
Consistent with the notion that nuclease hypersensitive sites in chromatin are
indicative for regulatory elements which are bound by sequence-specific DNA binding proteins, enhanced cleavage is considerably less
prevalent in the region encoding 45S pre-rRNA. We observe a regular pattern of fragments which may indicate a
nucleosomal structure. Actually, there are contradictory reports on the
presence and absence of nucleosomes in the transcribed region. Electron
microscopic studies suggested that rDNA is devoid of nucleosomes because the
lengths of the transcribed and spacer region have repeatedly shown to be roughly the same as that of the B conformation of the deproteinized DNA (
7
,
32
). On the other hand, a number of biochemical studies indicate that ribosomal
chromatin is organized into a repeating, nucleosome-containing structure (
33
-
35
). These apparently controversial data are very likely due to the inherent
difficulty in biochemical experiments to distinguish between the structure of
active and inactive ribosomal genes within the cells. By using the
intercalating drug psoralen as a tool to mark accessible sites along rDNA
chromatin
in vivo
, Sogo and co-workers have shown that both in yeast and mammals, about half of the
multiple copies of ribosomal genes are transcriptionally active (
36
,
37
). Moreover, this group has demonstrated that in intact Friend cells two
distinct populations of ribosomal chromatin coexist: one that contains
nucleosomes and corresponds to inactive gene copies, and one that lacks the
repeating structure and corresponds to transcribed genes. The relative amounts
of the two types of genes are maintained independently of transcription and are stably propagated through the cell cycle (
37
).
The results presented in this study show that
in vivo
the murine rDNA promoter is in a nucleosomal configuration. There is a
hypersensitive site at the promoter proximal terminator T
0
in vivo
and a MNase resistant region downstream of T
0
whose size and properties suggest that it represents a positioned nucleosome.
We could reproduce this structure
in vitro
by assembling chromatin on rDNA-containing plasmids with extract proteins from
Drosophila
embryos. Significantly, the nucleosomal organization of the promoter region was
only revealed if the assembly was performed in the presence of TTF-I, a sequence-specific DNA binding protein which binds to the terminator element. In the absence of TTF-I, nucleosomes were formed non-specifically on DNA resulting in statistical
positioning. In the presence of TTF-I, nucleosomes were found at defined locations upstream and downstream of
the TTF-I binding site. Binding of TTF-I to its target sequence resulted in an ATP-dependent shift of a nucleosome which places a nucleosome both over the rDNA promoter and the
transcription initiation site templates. Unexpectedly, TTF-I directed nucleosome remodeling was found to be a prerequisite for
transcriptional activation on chromatin templates (Längst
et al
., in press). On the basis of this finding we suggest that ribosomal gene
transcription on chromatin templates occurs by a two-step activation mechanism which involves the primary nucleosome
arrangement by TTF-I followed by an additional remodeling step which may be coupled to
transcription complex assembly and transcription initiation. This suggestion is supported by studies in yeast which have clearly demonstrated that after replication the
two newly synthesized rDNA daughter strands are assembled into regular
nucleosomal arrays which have to be locally disrupted to allow the assembly of
productive initiation complexes (
38
). Whether or not nucleosomes are displaced, modified or `cracked' during Pol I
transcription complex assembly and transcription elongation remains to be
investigated.
We thank T. Blank and P. Becker for providing extracts from
Drosophila
embryos. This work was supported, in part, by the Deutsche
Forschungsgemeinschaft (SFB 229) and the Fonds der Chemischen Industrie.
*To whom correspondence should be addressed. Tel: +49 6221 423 423; Fax: +49
6221 423 404; Email: i.grummt@dkfz-heidelberg.de
REFERENCES
Return