ABSTRACT
Basal and GC-box activated transcription were compared by various assays in order to
learn the basis for an 8-fold difference observed under standard conditions. The time required for
forming pre-initiation complexes and initiating and elongating RNA synthesis, and the
extent of transcription reinitiation were found to be quite similar for basal
and activated transcription, with complex formation being the slow step in both
cases. The extent of activation was found to vary widely with the amount of
template DNA used. Activated pre-initiation complexes were found to have a higher stability than basal
complexes. The data are interpreted to indicate that GC-box elements do not stimulate the rate constants for critical steps in
this system but rather increase the equilibrium constant for pre-initiation complex formation, probably by 10-30-fold.
High level transcription of RNA polymerase II genes requires activator proteins
(reviewed in
1
-
3
). In the absence of activators, transcription can occur but these basal levels
are quite low. Activated and basal transcription use a common subset of general
transcription factors with additional factors being required to mediate the
effect of activators (reviewed in
4
). Numerous studies have addressed how different activators enter into the multi-step, initiation pathway to increase transcription levels but a consensus
model has not yet emerged. Many of these studies have focused on detecting the
interactions of activator proteins with the general transcription factors
(reviewed in
2
,
4
).
The first cellular activator protein to be isolated was HeLa Sp1 which was
identified initially as a factor that activated transcription by binding to the
SV40 GC-box elements (
5
). Sp1 can function both
in vivo
and
in vitro
. Its activation mechanism has been proposed to involve at least the following
general transcription factors: TBP (
6
), IID (
7
), IIB (
8
) and IIA (
9
) as well as TAF proteins (
10
,
11
). The manner in which these interactions lead to enhanced transcription
initiation has not been settled (see
12
for a discussion).
Activators can enhance transcription by promoting one or more of a series of
broadly defined sequential steps that have been identified in simpler
transcription cycles (reviewed in
13
). In thermodynamic models basal transcription complexes form poorly and
activator increases their number by stabilizing them. In other models the basal
transcription complexes are sufficiently stable to form. In such models the
activator can increase the rate constant for assembling pre-initiation complexes (
14
). As a third alternative, activator could cause a pre-initiation complex that initiates RNA slowly to proceed rapidly (
15
). In this report we study RNA polymerase II transcription activated by Sp1 at
GC-boxes and find that the data strongly favor a thermodynamic model.
The approach involves applying a set of assays to compare basal and activated
transcription. This uses a HeLa cell nuclear extract that contains all factors
necessary for both basal and GC-box dependent activated transcription. Activated transcription is assayed
using a template containing a consensus TATA element, an Initiator (Inr
element;
16
) and the six SV40 GC-boxes located upstream to bind Sp1. This is compared with transcription
from a basal template that lacks all GC-boxes but has the same basal elements. Simple assays are established to
measure the half-time for forming pre-initiation complexes, for following the time required for
initiation, and for following how many complexes form and how stable they are.
The results lead to a model that is compatible with prior studies and provides
a means of evaluating the effects of transcription activators that complements
existing methods.
The activated template contains six SV40 GC boxes upstream of two optimally
placed basal elements: a consensus TATA element and the terminal
deoxynucleotidyl transferase (TdT) gene initiator element (plasmid VII of
17
). Basal templates were identical in the promoter region except that they lacked
upstream GC-boxes (plasmid IV of
17
). The HeLa cell nuclear extract was made as described (
18
).
The
in vitro
transcription assays contain 25 [mu]l HeLa cell nuclear extract in D buffer (
18
), 8.25 mM Mg
2+
, 5-1000 ng template (as indicated) and 1000 ng of pGEM as carrier DNA in a
final volume of 40 [mu]l. Standard reaction conditions include 0.5 mM nucleoside triphosphates
(NTPs) and incubation at 30oC for 30 min. Variations are noted in the text and legends. In single round
transcription assays, a 30 min pre-incubation was followed by a 2 min pulse with NTPs. RNA products
synthesized in each reaction were detected by extension of a labeled primer
with reverse transcriptase. The 79 nt labeled cDNA product was then separated
in a 6% urea-polyacrylamide gel and visualized by autoradiography. To quantify the RNA
product the radioactive bands corresponding to the correct product were scanned
with a gel scanner or cut from the gel and counted.
Experiments in which the amount of DNA was varied and transcription was assayed
were modeled by considering a simple bimolecular reaction. The reactants are
considered to be a nucleoprotein complex containing DNA and a factor that react
to form a complex that inevitably leads to transcription. Plots of RNA levels
versus amount of DNA were constructed. Best fit binding curves were derived for
both activated and basal DNA templates. The relative binding constants
associated with these curves were found to differ by 30-fold, assuming that basal and activated transcription reach the same level
at infinitely high concentrations of DNA. Curve fitting the data without this
assumption reduces the 30-fold difference to 10-fold.
The relative stability of pre-initiation complexes is evaluated with the following protocol. First, pre-initiation complexes are allowed to form on both basal and activated
templates during a 1 h incubation. Next, a 10-fold excess of challenge template is added to the reactions and incubation
is continued for an additional 45 or 90 min, as indicated. NTPs are then added
to allow the remaining stable transcription complexes to produce RNA. This is
quantified using primer extension as described above and a phosphoimager. The
challenge template was G5TILuc (
19
) which contains five Gal4 binding sites upstream to the identical consensus
TATA and initiator elements used above. It was pre-bound with GAL-AH (
20
) prior to addition. Transcription from this challenge template is not included
in the signal as its RNA does not hybridize to the primer used to detect RNA
from the original templates.
The standard transcription assay uses HeLa nuclear extract, nucleoside
triphosphates (NTPs) and a mixture of promoter and non-promoter (carrier) DNA. Each reaction contains 100 ng of the appropriate
supercoiled DNA template and 1000 ng of carrier DNA. The inclusion of very
large amounts of carrier DNA is a long-established condition to reduce occlusion of the template with non-specific DNA-binding proteins (
18
). Such conditions minimize activation by anti-repression mechanisms because the activator need not clear the template to
make it available for general transcription factors (
21
). Thirty minute reactions in the presence of all the necessary components
including NTPs are done in parallel using the activated and basal templates.
This protocol compares continuous transcription from promoters that do or do
not contain GC-boxes. The RNAs from both samples are detected by primer extension.
The autoradiograph (Fig.
1
) shows that both activated (lane 1) and basal (lane 2) transcription can be
detected. The bands were excised from the gel to determine the radioactivity
associated with each RNA sample. Repeated experiments showed that the presence
of GC-boxes on the template leads to an ~8-fold activation of transcription. This activation is within the
lower portion of the range of activation by Sp1 seen in prior experiments which
varies quite considerably and can reach several hundred fold for very weak
basal promoters. The basal levels in this experiment are relatively high
because the template contains both TATA and Inr elements (
22
). These experimental conditions, associated with 8-fold activation, will be used in each of the following assays. These
assays are designed to separate sub-steps in the transcription pathway to explore whether activator enhances
any step in a way that accounts for the 8-fold activation.
First, we established an assay for the half-time of pre-initiation complex formation by adapting prior protocols (
23
). There are two purposes for this experiment. One is to learn how fast pre-initiation complexes assemble in this system. The second more important
purpose is to learn whether the 8-fold difference in basal and activated transcription levels has its source
in an 8-fold acceleration of the rate constant caused by the activator. In this
protocol the template is incubated with nuclear extract for various times. At
each of these times samples are removed to assay for the number of functional
pre-initiation complexes that have formed. This assay is accomplished by
adding NTPs and allowing each functional complex to produce RNA. In order to
minimize potential complications due to production of more than one RNA from a
pre-initiation complex the synthesis is restricted to a 2 min pulse with
nucleotides (
23
and also see below).
The experiments were done in parallel using the basal and activated template
systems. The stronger signal strength of the activated system allows for a more
accurate determination. The data show that the formation of a full complement
of activated pre-initiation complexes is nearly complete within 1 h (Fig.
2
b). The half-time for the reaction is 20-25 min, in agreement with prior studies of assembly of various pre-initiation complexes involving RNA polymerase II (
23
-
25
).
Next we measured how long it took for these pre-initiation complexes to form RNA when presented with NTPs. The experiment
begins with a 30 min incubation of DNA and transcription extract. Because this
is done in the absence of added nucleotides, pre-initiation complexes will assemble but not initiate, as just described.
After this accumulation, NTPs are added to initiate RNA synthesis. Samples are
removed at various subsequent times and the RNAs are detected as just
described. The assay for RNA is a primer extension assay that requires
hybridization to an RNA sequence 79 bases downstream from the start site. Thus
the assay detects only those transcription complexes that have initiated and
transcribed as far as position +79. The time required for pre-initiation complexes to initiate and reach this downstream position is
determined by quantifying the amount of RNA produced at the various times after
initiation was begun upon the addition of NTPs.
This experiment was done in parallel for basal and activated transcription and
the results are shown in Figure
3
. There are two aspects of this experiment worth noting. First, transcription is
very rapid once NTPs are added to pre-initiation complexes. Under these conditions the 79 nt RNA is initiated
from the pre-formed pre-initiation complex and completed in <1 min. This burst is followed by a slower gradual increase in RNA
production from re-initiation events (unpublished data). In contrast, the results of Figure
2
showed that it takes ~100 times longer to complete the formation of pre-initiation complexes. Thus under these conditions the slow step in
the pathway is clearly forming pre-initiation complexes; once formed they initiate and elongate the RNA very
rapidly.
These experiments have focused on the formation and properties of pre-initiation complexes derived from the interaction of free DNA and factors
in the nuclear extract. However, the RNA produced in a continuous transcription
reaction, such as that shown in Figure
1
, can include an important contribution from events occurring after these steps.
This is because some of the RNA can be produced by the process of re-initiation wherein templates are used a second time. The rate of re-initiation can be quite fast in some circumstances (
26
) and thus, in principle, it would be possible to activate transcription by
causing multiple re-initiation events selectively on the activated template.
We tested this possibility by comparing RNA levels at long times to the RNA
levels associated with primarily the first round of transcription. Pre-initiation complexes were assembled during a pre-incubation and then RNA synthesis was begun by the addition of
nucleotides. As shown in Figure
3
, there is a burst of initial RNA synthesis, in which the first round is
completed within 1 min. Samples were removed at 2 min, which measures the
amount of RNA produced in this first round and a small fraction of second round
RNA (
26
; unpublished data). Samples were also removed after multiple rounds had
occurred at 30 min. The results (Fig.
4
) show that roughly three times as much RNA is produced at 30 min compared with
2 min (Fig.
4
, lanes 2 versus 1 and 4 versus 3). However, the important point is that this
ratio does not differ dramatically for activated (lanes 3 and 4) and basal
templates (lanes 1 and 2). Thus the 8-fold higher level of transcription for activated templates is not a
consequence of an 8-fold increase in re-initiation events.
These experiments have shown that activation does not involve reduction in the
time required for pre-initiation complexes to form or to initiate and re-initiate RNA synthesis. An alternative possibility is that the
number of templates engaged in forming pre-initiation complexes is increased by the activator. This could occur by
promoting any of the succession of bimolecular reactions between DNA and
transcription factors that lead to formation of functional pre-initiation complexes. In this way activator drives more components from
the free state into the bound state and thus more transcription complexes form.
Components of bimolecular reactions can also be driven together by mass action
by simply increasing their concentration. Thus, if aspects of this
thermodynamic model are applicable then increasing the concentrations of
factors or DNA may lessen the need for activator to drive them together. The
hypothesis may be tested in this system by varying the concentration of DNA and
assaying transcription; technically it is not possible to vary the
concentration of factors over a wide range because of lack of control over the
unknown factors in the transcription extract.
In preliminary experiments (not shown) we found that increasing the amount of
DNA from 100 to 400 ng led to an increase in transcription, but increasing the
concentration of DNA further led to slight declines in activated transcription.
Declines were not observed over this range for basal transcription but did set
in when additional promoter or non-promoter DNA was added (a presumptive artifact also reported in ref.
27
). We also found that lowering the amount of DNA led to progressive decreases
for both activated and basal transcription, although the extent of the decrease
was not comparable in the two cases. We decided to assay transcription
systematically over the lower part of this range of concentrations.
At each concentration of DNA, parallel reactions were done using basal and
activated templates. The amounts of RNA produced were measured and are plotted
in Figure
5
A. One simple observation is probably the most important aspect of this
experiment. This is emphasized in Figure
5
B which plots the ratio of activated to basal transcription as calculated from
the two curves of Figure
5
A. The analysis shows that the quantitative effect of activator varies
dramatically with the concentration of DNA used. Activation is very high at low
concentrations of DNA (25-fold at 10 ng) and very low at high concentrations of DNA (3-fold at 400 ng). Because the addition of multi-microgram amounts of DNA causes a general inhibition of
transcription, we cannot reliably determine if this ratio will continue to
decrease with higher concentrations of DNA. This larger effect of activator at
low DNA concentrations is consistent with a variety of observations, although
the phenomenon has not been studied directly (
25
,
28
).
This lesser reliance on activator is not caused by the increasing amounts of DNA
acting as a sink for histones or other non-specific DNA binding proteins that might act to repress transcription (
29
). All the reactions occur in the presence of 1000 ng of non-template carrier DNA. Thus when the activation ratio falls from 25 to 10,
by increasing the amount of template from 10 to 50 ng, the total amount of DNA
serving as a sink for histones and other non-specific DNA-binding proteins increases only from 1010 to 1050 ng. Thus the
presence of a large excess of carrier minimizes the potential contribution of
such anti-repression mechanisms.
This reaction can be modeled using a simple bimolecular scheme where two
components, one containing DNA and one containing an essential factor, react to
form an intermediate complex that inevitably leads to a pre-initiation complex. The solid lines in Figure
5
A show the shapes of two best-fit binding curves expected from a simple bimolecular reaction in which
DNA is one of the reactants. The experimental data points fit such a simple
theoretical model. No doubt the actual reaction pathway is much more
complicated but this simple model can be used to account for the drastic
changes in activation level seen as different amounts of template are used.
These results allow evaluation of the contributions to transcription activation
via GC-boxes in an
in vitro
transcription system. Kinetic experiments showed that forming a pre-initiation complex is the rate-limiting step in this system for both activated and basal
transcription. Although the activator does work at this step to enhance
transcription it does not do so by decreasing the half-time for pre-initiation complex formation. Instead it increases the number of
such complexes that can form, presumably by increasing their thermodynamic
stability; each complex, however, forms and proceeds through the pathway at a
rate that is similar to that observed for basal transcription. Thus more
transcripts are produced because the activator has caused more templates to
become engaged in the pathway, but each step in the pathway proceeds with an
unchanged rate constant.
In an attempt to identify a step that was facilitated by activator we assayed
the time required to complete various steps in the transcription pathway. Using
a standard condition that involved an 8-fold activation by GC-boxes we found that the half-times for the following steps were not affected by GC-boxes: (i) forming a pre-initiation complex; (ii) initiation and elongation
of RNA by the pre-initiation complex; (iii) re-initiation. The formation of the pre-initiation complex was found to be the slow step with a half-time on the order of 20-30 min. In contrast, when starting with a pre-initiation complex it was possible to
initiate and elongate a 79 nt transcript in <1 min. We did not directly measure the rate of re-initiation but simply showed that the activator did not lead to a large
increase in the number of rounds of transcription occurring in 30 min. These
experiments led us to conclude that the 8-fold stimulation by activator was not due to its enhancing the rate
constant for any of these steps.
These conclusions are in accord with a variety of prior studies using HeLa cell
extracts and fractionated systems, although the rates of all of these steps
have apparently not previously been compared directly with each other. The half-time of pre-initiation complex formation has been studied in a variety of
mammalian systems and estimates are in the range of 10-30 min (
23
-
25
,
33
). Pre-formed mammalian pre-initiation complexes can produce RNA rapidly (
23
). In a
Drosophila
transcription extract both steps occur more rapidly but formation of the pre-initiation complex is still rate-limiting (
34
). These studies have used a variety of promoters and activators which suggests
that it will be quite common that formation of a pre-initiation complex is rate-limiting. In addition, the data indicate that the rate constant for this step
is not strongly affected by activator, although it is too early to assume that
this important conclusion will be true generally.
High concentrations of DNA were shown here to largely overcome the need for
activation via GC-boxes. Thus GC-boxes stimulate transcription 25-fold in reactions containing 10 ng of DNA but only 3-fold in reactions containing 400 ng of DNA (always
buffered by the presence of 1000 ng of carrier DNA to minimize anti-repression). Prior studies of stimulation by GC-boxes have led to estimates ranging from 3- to 500-fold (
11
,
17
,
35
). We suspect that some of this variation is due to variations in the amount of
DNA used. It is likely that some is also due to the use of different promoters
and anti-repression effects in cases where high amounts of non-promoter DNA was lacking. The result re-emphasizes the need to standardize reaction conditions when
evaluating the strength of activators. It also emphasizes the desirability of
using very low concentrations of DNA when attempting to assay for transcription
activators. Similar phenomena have been observed in other systems although
explanations have not been discussed (
25
,
28
).
Two experiments suggest that activator increases the equilibrium constant of pre-initiation complex formation to form a complex with enhanced stability.
First, activated complexes were found to be significantly more resistant to
destabilization in a template challenge protocol. Secondly, as just discussed,
the effect of activator becomes minimal at high amounts of DNA, which suggests
that mass action has partially substituted for activator in driving complex
formation. We attempted to put these observations into semi-quantitative terms by modeling the complex reaction in terms of a simple
bimolecular scheme. The rationale is that one effect of activator may be to
facilitate the binding of a factor to a stable nucleoprotein complex containing
DNA template and possibly other factors. The reaction scheme was simplified to
that between two components: an unknown factor, whose concentration is fixed,
and DNA, the concentration of which may be varied
in vitro
. Because the concentrations of factors cannot be determined, the absolute value
of the binding constant cannot be estimated from these data. Modeling indicated
that the activator increases the equilibrium constant by 30-fold, although this could be an overestimate by up to a factor of 3, as
discussed above (see Materials and Methods). This change presumably reflects an
increase in the affinity of some critical factor for the nucleoprotein complex
containing the template DNA. The analysis allows this contribution of activator
to be interpreted in this manner, but by no means excludes additional effects
of activator in more complex pathways.
These conclusions are compatible with a wide variety of studies using partially
purified transcription factors. Many studies have led to the suggestion that
activators can contribute to transcription by `recruiting' certain general
transcription factors into the assembling pre-initiation complex (reviewed in
2
). Evidence points toward activators changing the ability to bind TFIIB although
this is not sufficient to account for the full level of activation (
8
,
36
). This evidence may be reconciled with the current data by proposing that basal
templates are poorly transcribed in large part due to a low affinity for TFIIB
and that Sp1 increases the affinity of TFIIB for promoter by up to the factor
of 30 calculated from the new binding data.
The current data do not show a change in the half-time for pre-initiation complex assembly (consistent with
24
,
25
,
37
). This implies that the critical factor is not recruited more rapidly by
activator. Instead once the factor is bound it associates with more stability;
the longer lifetime of the complex formed gives it a greater opportunity to
bind the subsequent factors and complete the pathway to functional pre-initiation complex. This is consistent with proposals that activation can
involve promoting a form of the activator-TFIID-TFIIA complex that can bind TFIIB in a stable and functional
manner (
31
,
38
). One problem with this simple model is that adding a large excess of purified
TFIIB causes only small increases in transcription rather than the large
increases that might be expected if transcription is limited by its low
affinity for the nucleoprotein template (
8
,
25
; also see
12
for discussion). However, TFIIB apparently is not recruited as an isolated
protein but rather as part of a large multiprotein transcription complex (
39
,
40
). Thus adding excess isolated TFIIB may only lead to the modest effects
observed.
These data also provide a rationale for the large differences in extents of
activation that can be observed
in vitro
. The up to 30-fold increase in equilibrium constant seen here will be predicted to have
its greatest influence when low concentrations of DNA are present or when
promoters with weak, low affinity, basal elements are used. Additionally,
experiments that omit large amounts of carrier DNA may include an additional
activation by anti-repression. These considerations should be very useful in comparing
different experiments involving activated transcription and in providing a
continuing basis for rationalizing the mechanism of activation in terms of
traditional enzymology.
This research was supported by grants from the NSF (MCB-9203293) and the USHHS (GM49048). DY is a trainee under NIH 5-T32-GM07185. We thank members of the group for their advice.
REFERENCES
Return