ABSTRACT
The chromatin structure specific to the SV40 late transcription elongation complex as well as the occupancy of several sites that bind transcription factors have been examined.
These features have been determined by assessing blockage to restriction enzyme digestion. Cleavage specific to the elongation complex has been quantified
using ternary complex analysis. This method involves radioactively labeling the
complex by
in vitro
transcription followed by determining the extent of linearization by
electrophoresis in an agarose gel. It was found that not only is the origin
region devoid of nucleosomes, but there is also no stable factor occupancy at
the
Bgl
I,
Sph
I,
Kpn
I and
Msp
I restriction enzyme sites within this region. Thus these sites were cleaved to
a high degree, meaning that the binding sites for a number of transcription
factors, including OBP/TEF-1, TBP, DAP, as well as a proposed positioned nucleosome, are unoccupied
in the native viral transcription elongation complex. The absence of these
trans
-acting factors from their respective binding sites in the elongation
complex indicates that they bind only transiently, possibly cycling on and off
during the transcription cycle. This finding implies that various forms of
transcription complex are assembled and disassembled during transcription and
thus supports a `hit-and-run' model of factor function.
Considerable progress has been made recently in the identification and
characterization of
trans
-acting factors, which are protein factors that bind to specific sites in
the DNA regulatory regions of genes and modulate expression from those genes
(see
1
-
4
for reviews). However, while a substantial inventory of these factors has been
compiled, the various mechanisms by which they stimulate transcription are
still uncertain. Ultimately, a complete understanding of the process of
transcription will require a description of the structures of the transcription
complex in its various functional forms: initiation complex, elongation complex
and termination complex. A first step in these descriptions is a
characterization of the factor content and locations in the various complex
forms.
We previously initiated such a characterization of the simian virus 40 (SV40)
late viral transcription elongation complex (VTC) by assessing the occupancy of
the replication origin binding site (Site II) for T-antigen (
5
), the product of the viral oncogene. This was accomplished by determining the
extent of restriction enzyme cleavage at that site in the native complex. It
was found that T-antigen is absent from the origin site in the elongation complex, a
finding that was inconsistent with certain models for the role of T-antigen in the control of transcription. This in turn spurred efforts to
evolve alternative models for the participation of T-antigen in late transcription (
6
-
8
).
A similar study characterized the occupancy of the VTC sites for the cellular
transcription factor TEF-1 (
8
). In contrast to the T-antigen origin site, the TEF-1 site at nt 270 was found to be occupied in a fraction of the VTC
population, a result that indicates there are at least two forms of elongation
complex.
Additional studies of SV40 have led to the identification of many other cellular
trans
-acting factors and their binding sites (
1
-
4
,
9
and references therein), as well as characterization of the chromatin structure
within which these sites are located. Thus numerous studies have observed that
the SV40 region containing the sites for
trans
-acting factors, located around the replication origin (see Fig.
1
), exhibits a chromatin structure that is devoid of nucleosomes in a fraction of
the minichromosomes (
10
-
13
and references therein). This conclusion derives from the observation that both restriction and non-specific nucleases cleave preferentially in this region in a subpopulation of the viral minichromosomes; this
accessible stretch of chromatin has been termed the open region. An important
extension of these studies was the demonstration (
5
,
8
,
14
,
15
) that all of the VTC (which themselves comprise <5% of the total minichromosome population;
16
,
17
) contain an open region.
BS-C-1 cells were used for all experiments. These cells were cultured and
infected with SV40 wild-type strain 776 as described previously (
20
,
21
).
The isotonic extraction procedure previously described (
8
,
20
,
21
) was used to prepare nuclear extracts. When required, these extracts were
concentrated and desalted using a Centricon 30 microconcentrator (Amicon).
Typically 30-90 [mu]l aliquots of 1-3* concentrated and desalted nuclear extract were
combined with the appropriate 10* restriction enzyme buffer and 5 U/[mu]g DNA of the indicated restriction enzyme. Either 10* M buffer (final concentrations 50 mM NaCl, 10 mM MgCl
2
, 1 mM DTT, 10 mM Tris-HCl, pH 7.5) or 10* H buffer (as M buffer except that the final concentration of NaCl
was 100 mM and that of Tris was 50 mM; both buffers supplied by Boehringer
Mannheim Biochemicals) were used to adjust the salt concentrations of desalted
nuclear extracts. Incubations in buffer with no added NaCl resulted in
considerable nicking of VTC. Therefore, digestion with an enzyme with a low
salt optimum (e.g.
Kpn
I) was performed at 100 mM NaCl to avoid nicking. Additional enzyme was added
when necessary to compensate for any lower activity due to change in salt
concentration. Digestions were performed for 40 min at 30oC except where indicated. At the end of the incubation, the digestion was
halted by the addition of transcription mix or by the addition of EDTA to a
final concentration of 20 mM.
In vitro
transcription and subsequent preparation of ternary complexes were performed as
described previously (
5
,
17
,
22
). Samples were fractionated by electrophoresis through 1.0% agarose TBE gels,
after which the gel was stained with ethidium bromide and photographed using
4x5 Tri-X film. The gel was then dried and autoradiographed using Kodak XAR film.
Films were scanned and peaks integrated with an LKB Ultroscan densitometer (
5
,
22
,
23
).
VTC bands in the autoradiograph were completely sensitive to the presence of 1 [mu]g/ml [alpha]-amanitin in the extension mix, while >90% of the incorporated
radioactivity has been shown to be late strand specific (
20
).
Isotonic nuclear extract from SV40 strain 1x72 (
24
) was incubated with doubly end-labeled 610 bp bare DNA fragment (
Ava
II fragment nt 5118-557) from 1x72 SV40 (containing a single
Sph
I site). This mixture (10 [mu]g total DNA in 50 [mu]l) was incubated with
Sph
I (30 U) at 25oC in 1* M buffer as described above. Samples (6 [mu]l) were removed at the indicated times and subjected to ternary
complex analysis as described above.
Hypotonic nuclear extract was prepared by a modification of a method reported
earlier (
8
,
20
). A 10 cm plate of cells was washed twice with phosphate-buffered saline, after which the cells were lysed with 0.5-1.0 ml lysis buffer containing 10 mM Tris-HCl, pH 6.8, 10 mM Na
2
EDTA, 0.25% Triton X-100, 31 [mu]g/ml phenylmethylsulfonyl fluoride, 5 [mu]g/ml pepstatin A, 5 [mu]g/ml leupeptin and 10 [mu]g/ml aprotinin. The lysate was scraped into a centrifuge
tube and nuclei were pelleted at 3000
g
for 5 min. The nuclear pellet was resuspended in 60 [mu]l extraction buffer containing 10 mM Na HEPES, pH 7.8, 1 mM Na
2
EDTA and protease inhibitors as mentioned above. The nuclei were then treated
with 60 strokes in a dounce homogenizer and incubated overnight at 4oC with rotation. The suspension was centrifuged for 10 min at 11 000
g
to pellet the nuclear chromatin, leaving the supernatant which contained the
SV40 minichromosomes and VTC.
The nuclei pellet from cells infected with SV40 was prepared as described for the hypotonic or isotonic nuclear extract preparations (see above). The pellet (representing 0.16 of a 15 cm plate of cells) was suspended in 40 [mu]l of the appropriate buffer and digested with restriction enzyme for 40 min
at 30oC. The digestion was terminated by the addition of 1 [mu]l 40 mg/ml heparin, after which the sample was centrifuged for 10 min
at 11 000
g
. A sample of the supernatant, containing the SV40 ternary complexes, was then
incubated with transcription mix and processed as described above for ternary
complex analysis.
A 1 ml sample of isotonic nuclear extract was sedimented through a 10-30% linear glycerol gradient containing either TE (10 mM Tris-HCl, 0.1 mM EDTA, pH 7.9) or TE + 0.3 M (NH
4
)
2
SO
4
(
20
,
21
). The absorbance peak of the minichromosomes was then pooled, concentrated in a
Centricon 100 microconcentrator and desalted by spin chromatography over Sephadex G-50 equilibrated with TE. Aliquots of the pools were adjusted to 1* H buffer (see above) and digested with the indicated restriction
enzymes, after which they were extended
in vitro
with [[alpha]-
32
P]UTP and analyzed as ternary complexes as described above.
Figure
1
illustrates a map of selected restriction enzyme cleavage sites in the SV40
genome. Even though VTCs comprise <5% of the SV40 minichromosome population in the nuclear extract of an infected
cell, the extent of restriction enzyme cleavage of the VTC can be quantified by
the radioactive extension labeling method (
5
,
8
,
14
,
17
). A brief description of the method follows. A sample of nuclear extract is
incubated with a restriction enzyme, after which heparin is added to convert
VTCs into ternary complexes. These are then labeled by extension
in vitro
, after which they are separated into linear, nicked and supercoiled forms by
agarose gel electrophoresis. A photograph of the ethidium bromide stained gel
reveals the relative levels of the three forms for the bulk population, while
the autoradiograph reveals the relative levels for the VTC. The extent of
exposure of the restriction enzyme site in the VTC is proportional to the
fraction of linear molecules in the digested sample. When a restriction site
lies within or near to a
trans
-acting factor binding site in the regulatory region near the SV40 origin,
the presence of that factor is detected as a blockage to digestion. Moreover,
if additional enzymes are chosen that have sites throughout the rest of the
genome, information about the chromatin organization in various regions of the
genome can be obtained. This is because DNA organized in randomly positioned
nucleosomes is characterized by 30-50% protection from restriction enzyme digestion (reviewed in
12
), so a sequence that is completely digested is likely to be nucleosome-free. Analysis of sites spanning the entire genome can therefore detect
differences between the chromatin structures of bulk minichromosomes and VTC,
which in turn can provide insight into the distinctive chromatin structural
features necessary for transcriptional activity.
Figure
2
shows the results of such an experiment. An undigested sample (Fig.
2
, UNCUT) exhibits the expected bands for supercoiled and nicked forms in both
the ethidium bromide stained gel (representing bulk minichromosomes; MCs) and
in the autoradiograph (representing VTCs). An initial examination of the
samples incubated with various restriction enzymes reveals that the bulk
minichromosome population tended to be linearized to a level of ~30-50% for all of the enzymes used (Fig.
2
, MCs). As discussed above, this level of digestion is characteristic of a
chromatin organization composed of nucleosomes positioned randomly over the
respective sites (
12
).
The high levels of VTC digestion by the origin region enzymes in Figure
2
indicate that their sites are virtually bare. To further investigate this, a
representative origin region enzyme was examined in more detail. We had
previously shown that
Sph
I digestion is in fact blocked when its site is occupied by the
trans
-acting factor TEF-1 (
8
), so a kinetic analysis of cleavage was performed with this enzyme (Fig.
3
). Since there is an ambiguity in interpretation of the
Sph
I cleavage due to the presence of two closely spaced
Sph
I sites in the duplicated 72 bp repeat of wild-type SV40, this analysis was performed on a variant of SV40 (1x72;
25
) that contains only one copy of the 72 bp repeat sequence. To provide a direct comparison to the digestion of bare DNA,
an end-labeled fragment of bare SV40 DNA containing a single
Sph
I site was added to 1x72 nuclear extract. A time course of
Sph
I digestion was performed on this mixture, followed by ternary complex analysis.
The results in Figure
3
show that both the bare DNA and the VTC were digested to completion. The plot
of the results of densitometric quantitation (Fig.
3
, bottom) reveals that the VTC was digested at the same rate as bare DNA present
in the same tube. These results indicate that the single
Sph
I site in the 1x72 VTC is entirely vacant.
The percent cleavages for bulk minichromosomes and VTC for 13 enzymes were
determined by densitometric scanning and a summary of the results is presented
in Figure
6
. This quantitative representation of the data bears out the conclusions reached
above. Figure
6
B shows that the variation of linearization with different restriction enzymes
is somewhat limited for the bulk minichromosome population, ranging between ~30 and 50%, with no particular part of the genome showing systematically
distinctive digestion. However, Figure
6
A shows that sites in the VTC origin region are digested to a markedly higher
level than are sites in the remainder of the genome. The distinction between
the two forms of complex is particularly clear in a difference plot for the VTC
versus bulk minichromosome (Fig.
6
C).
Numerous studies observe increased enzyme accessibility (
32
-
35
) in the regulatory region of the SV40 bulk minichromosome population and this
region of accessibility has been termed the open region of SV40. Electron
microscopy indicates that this accessibility reflects an absence of nucleosomes
in this region (
32
,
34
). While the data in Figure
6
B show no significant indication of an open origin region in the bulk
minichromosomes of our preparation, the apparent coincidence between the extent
of the open region of the VTC (Fig.
6
A) and that reported by others (see references above) for the bulk
minichromosome implies that nucleosomes may be responsible for the VTC open
region as well.
To investigate this, samples of nuclear extract were analyzed following exposure
to an ionic strength [300 mM (NH
4
)
2
SO
4
] at which nucleosome sliding is known to occur (
12
,
36
,
37
). Thus nuclear extract samples were sedimented through glycerol gradients containing different ionic strengths, after which accessibilities to various restriction enzymes were determined.
The results are shown in Figure
7
. VTCs sedimented through a gradient containing low ionic strength showed higher
cleavage at the
Bgl
I site, a representative origin region enzyme, than at the other sites in the
genome (Fig.
7
A, TE). This result is similar to those from less processed samples (Figs
2
,
4
and
5
). However, exposure to 300 mM (NH
4
)
2
SO
4
in the glycerol gradient decreased the relative level of digestion of the
Bgl
I site to where it approximated the levels of the other sites [Fig.
7
A, 300 mM (NH
4
)
2
SO
4
]. Figure
7
B shows the results of quantitation of the data for both the VTC and the bulk
minichromosomes. While exposure to elevated ionic strength caused little change
in the pattern of exposure of sites in the bulk minichromosomes, the pattern of
the VTC sites changed significantly, becoming more like the bulk pattern.
Specifically, the
Bgl
I site has changed from being highly accessible in the low salt-treated sample to being only partially accessible and similar to the other
sites in the high salt-treated sample. Thus the results are consistent with elevated ionic
strength having caused nucleosomes to invade the open region VTC by sliding,
partially blocking the previously exposed
Bgl
I site. This suggests that nucleosomes are indeed responsible for the open
region in the VTC and further supports the conclusion that the VTC contains
features which are both similar to and different from those of the bulk
population (
22
).
SV40 has proven to be a useful model in the study of both viral and cellular
gene expression. In an effort to gain insight into mechanisms by which
trans
-acting factors regulate transcription, we have examined the structure of SV40 late transcription complex. Two related features of VTC structure have been characterized: the open
region in the chromatin of the VTC and the occupancy of
trans
-acting factor binding sites in that open region.
We have found that the extent of the open region in the VTC corresponds to that
reported for the bulk minichromosome population. Thus Varshavsky
et al
. (
35
) and Scott
et al
. (
19
) observed in bulk minichromosome studies that the open region extends from nt ~1 to 400. The high levels of cleavage observed here for
Bgl
I,
Sph
I,
Kpn
I and
Msp
I in the origin region define an open region from nt 1 to 346. In contrast,
Ppu
MI (nt 587) and
Bst
XI (nt 4759) cleave to a low level and this low cleavage, along with the
similarly low cleavages at the
Bcl
I,
Bam
HI,
Eco
RI,
Hae
II and
Eco
RV sites, define limits to the extent of the open region. It should be noted
that while the open region is quite apparent in the data for the VTC (Fig.
6
A), the data here for bulk minichromosomes (Fig.
6
B) do not show a strong indication of an open region. This variation in the
observation of the open region in the bulk population has been reported by other groups (see for example
38
,
39
) and may reflect a variation in the composition of the minichromosome population in different nuclear extract preparations.
It was demonstrated that hypertonic treatment of nuclear extract caused
digestion of the VTC origin region to change from a high level of cleavage to
one characteristic of nucleosome organization (Fig.
7
), suggesting that the high salt caused the nucleosomes to slide into the origin
region. An additional feature of VTC nucleosome organization is revealed by
Msp
I digestion (Figs
4
-
6
and Table
1
). The high level of cleavage by
Msp
I indicates that the nucleosome, which has been proposed to be positioned over
that site in the bulk population (
40
,
41
), is not present in the VTC. If such a positioned nucleosome is important for
transcription, as the authors propose (
40
,
41
), then it is possible that the nucleosome is present in the preinitiation state
and then displaced once transcription ensues. Alternatively, rather than being a component of a transcriptionally active complex, perhaps a more likely role for such a nucleosome would be to repress late transcription, since the data here indicate that it is absent from the VTC.
A nucleosome has also been proposed to be specifically positioned over the
Bam
HI site and has been assigned a role in transcription termination (
42
,
43
). This would predict that the
Bam
HI site in the VTC should be quantitatively blocked. The data in Figures
2
,
6
and
7
demonstrate that such is not the case, a finding which makes the proposal
unlikely.
The open region contains binding sites for numerous
trans
-acting factors that stimulate transcription from the late promoter (
6
,
7
,
28
-
30
,
44
-
49
). In order to gain insight into mechanisms by which
trans
-acting factors function, we determined the level of occupancy of several
trans
-acting factor binding sites in the VTC origin region. This was
accomplished by analysis of cleavage sites for
Bgl
I,
Kpn
I,
Msp
I and
Sph
I in the VTC.
The
Bgl
I site of the VTC was cleaved to a high degree in both extract as well as nuclei
(Figs
3
,
4
,
6
and
7
and Table
1
), a result which confirms and extends our earlier findings (
5
). The
Bgl
I cleavage site lies within T-antigen Site II in the origin of replication, a region that has been
identified to be important for T-antigen-dependent stimulation of late transcription activity (
29
). Our findings indicate that neither T-antigen nor any other factor is bound to this site in the VTC.
The sequence surrounding the
Kpn
I site (nt 294) has been assigned significance as a
trans
-acting factor binding site (
30
,
31
) and a second study reports that the DNA double helix is melted in this region
in the bulk minichromosome population (
41
). However, the data here indicate that this site is unoccupied and in the
normal DNA B-form in the VTC.
The sequence surrounding the
Msp
I site (nt 345) is the binding site for IBP-s, a repressor protein that is proposed to be bound to the early
transcription complex in order to maintain late transcription repression (
50
). The shift from early to late transcription has been proposed to involve
removal of this IBP-s. Our finding that the
Msp
I site is vacant therefore supports this proposal. A study by Ayer and Dynan (
28
) proposes that the
Msp
I site is the binding site for downstream activator protein (DAP). The authors
suggest that DAP may stimulate late transcription by facilitating assembly of
the preinitiation complex. Again, our data indicate that this site is vacant in
the VTC, so if DAP participates in the preinitiation complex, it would need to
dissociate from the complex once transcription elongation begins. This will be
discussed in more detail below.
This finding that the binding sites for numerous
trans
-acting factors are unoccupied in the VTC raises an important question: are
these sites in fact vacant in the cell or is the apparent lack of occupancy due
to the procedure used to analyze them? We feel that the site vacancy observed
is not a procedural artifact, for the following reasons.
First, the same high level of cleavage was observed whether the experiment was performed on nuclear extract or in nuclei (Figs
4
and
5
and Table
1
), indicating that the factors were not lost during the preparation of extract
from nuclei. The nuclei sample was prepared with minimal manipulation and is
similar to what has been termed an `
in vivo
' state of chromatin used in another analysis (
51
).
Second, high cleavage was observed when hypotonic conditions were used for preparation of nuclei and extract (Figs
4
and
5
and Table
1
), indicating that salt-induced removal of factors had not occurred. Indeed, as described above,
reduced cleavage occurred with hypertonic treatment of extract, but this can be
attributed to salt-induced nucleosome sliding over the open region.
Third, the high level of digestion could be accomplished within minutes for
Sph
I (Fig.
3
) as well as
Msp
I and
Kpn
I (data not shown). This digestion time is typical of methods used to detect
factor occupancy that employ non-specific footprinting agents such as DNase I and dimethyl sulfate (
51
-
53
). Complete digestion with such short digestion times makes the possibility of
the putative bound factor cycling off its binding site during digestion
unlikely.
Fourth, different enzymes that cleave at the same site each gave equally high
levels of digestion, indicating that the cleavage level at a site reflects VTC
structure and not the particular preferences of the enzyme used (see
54
for a discussion of this point). Enzymes analyzed were
Acc
65I and
Kpn
I for the site at nt 294 (Figs
4
and
5
and Table
1
),
Hpa
II and
Msp
I for the site at nt 346 (Figs
2
,
4
and
5
) and
Sph
I and
Nsi
I for the site at nt 128 (data not shown).
Fifth, protein complexes that form at each of the sites are typically very
stable, as indicated by band shift and footprint data for proteins which bind
at the
Sph
I site [TEF-1 (
26
), OBP (
55
), `band A' factors (
56
)], the
Bgl
I site [T-antigen (
5
,
57
)] and the
Msp
I site [DAP (
28
), positioned nucleosome (
41
)]. Thus, if these proteins were bound to their respective sites in the VTC
in vivo
, the demonstrated
in vitro
stability of each indicates that they would remain bound under the
comparatively gentle conditions of isolation and analysis used here.
Sixth, while the VTC origin sites are quantitatively cleaved, these sites are
partially blocked in the bulk population, even after extended digestion times
(Figs
2
and
4
-
7
). This provides evidence for the presence of another protein structure, the
nucleosome, on an alternative functional form of the same DNA sequence and
constitutes a general positive control that bound protein will block digestion
at each of the sites.
Seventh, this approach demonstrated factor occupancy at nt 270 under comparable
conditions (
8
).
The procedure described here determines the occupancy level of several
trans
-acting factor binding sites in a complex that is in a specific stage of
transcription, elongation. As such it provides insight into the mechanism of
factor function that goes beyond that which is readily available from current
methods. Thus such transcription stage-specific information is not generally available from mutation-based methods; while this genetic approach has proven to be
extremely powerful in the functional identification of
trans
-acting factors, it is less useful for elucidating the mechanism of action
of factors. Details of mechanisms are better addressed by an
in vitro
strategy involving reconstitution, but a significant concern in this approach
is the uncertainty of whether proper chromatin substrates have been reconstructed. Consequently, the method here that characterizes factor content in an
in vivo
-assembled transcription elongation complex serves as an important
complement to the genetic and
in vitro
reconstitution approaches for factor analysis.
A major conclusion of this study is that factors which bind to the
Sph
I,
Bgl
I,
Msp
I and
Kpn
I sites are not directly involved in maintaining the transcription elongation
state, because they are not present on the native elongation complex. This
indicates that the transcription activation roles identified by the mutational
analysis of their respective binding sites must occur at some other stage of
the transcription process, such as initiation complex formation. Thus,
construction of a functional transcription initiation complex may require the
presence of the large number of factors implicated by the genetic studies.
However, entry into the elongation state may then require disassembly of this
complex, resulting in opening of the sites to restriction enzyme digestion as
observed. Following termination, these sites would then be reoccupied in the
construction of a new initiation complex. This is reminiscent of the `hit-and-run' mechanism of transcription activation proposed for the
glucocorticoid receptor (
51
). However, such transient occupancy by factors contrasts with the situation in
the 5S rRNA gene, where factors remain bound to sites through multiple rounds
of transcription (reviewed in
58
). This different mechanism of factor action may in turn represent that of the
factor bound to the
Pvu
II site at nt 270 (
8
). Thus both types of occupancy characteristics appear to be present, as
revealed by our study on the elongating transcription complex, and the system
described here can serve to classify factors according to mechanism of action.
This work was supported by grant VM-22 from the American Cancer Society. We thank Dr Winship Herr for
providing the 1x72 strain of SV40.
REFERENCES
Return