Article

• The EMBO Journal (1997) 16, 7457 - 7467
• doi:10.1093/emboj/16.24.7457

Multiple ATP-dependent steps in RNA polymerase II promoter melting and initiation

Ming Yan¹ and Jay D. Gralla¹

1. Department of Chemistry and Biochemistry, and Molecular Biology Institute, University of California, Los Angeles, CA 90095-1569, USA

Correspondence to:

Jay D. Gralla, E-mail: gralla@ewald.mbi.ucla.edu

Received 11 August 1997; Revised 25 September 1997

• Keywords:

  • abortive initiation,
  • ATPase,
  • permanganate footprinting,
  • transcription

Introduction

Transcription initiation by RNA polymerase II involves a complex series of ordered events (Conaway and Conaway, 1993; Orphanides et al., 1996; Roeder, 1996). The polymerase is recruited to the promoter by a subset of polypeptides within the group of general transcription factors TFIID, TFIIA, TFIIB and TFIIF. At this point, the polymerase is in a closed complex in which the transcription start site is not yet melted. Other general transcription factors are then used in a series of steps that begins with the formation of an open complex in which the start site is melted and available for the binding of the appropriate nucleoside triphosphates (NTPs). Initial RNA synthesis can then occur. Finally, in a stage often termed promoter clearance, the polymerase can escape from the promoter region and elongate the transcript. The various steps that follow the recruitment of the polymerase are not yet characterized in detail, but they require the general transcription factors TFIIH and TFIIE (Goodrich and Tjian, 1994; Ohkuma and Roeder, 1994; Holstege et al., 1995, 1996; Dvir et al., 1996b).

The steps that follow recruitment of the polymerase cannot be completed without energy from the hydrolysis of the – bond of ATP (Bunick et al., 1982; Sawadogo and Roeder, 1984). In this regard, pol II transcription differs from that of most other RNA polymerases, including polymerases I and III and the major prokaryotic enzymes. The special need for energy in the pol II case is not understood.

The energy of ATP hydrolysis is used to drive at least two steps, open complex formation and promoter clearance (Wang et al., 1992; Goodrich and Tjian, 1994; Dvir et al., 1996a,b; Holstege et al., 1996; Jiang et al., 1996). Because TFIIH is used in these steps, and contains several ATP-requiring ATPase activities, models for these steps focus upon the use of TFIIH. The TFIIH kinase appears to be used in promoter clearance, probably by phosphorylating the C-terminal domain (CTD) of the polymerase (Feaver et al., 1994; Dahmus, 1995; Serizawa et al., 1995; Shiekhattar et al., 1995; Jiang et al., 1996). TFIIH also contains helicase activities, at least one of which has been suggested to be involved in formation of open complexes (Park et al., 1992; Feaver et al., 1993; Schaeffer et al., 1993, 1994; Serizawa et al., 1993; Guzder et al., 1994), although direct evidence is still lacking.

Thus, it appears that ATP hydrolysis is used at the beginning and the end of the series of initiation steps, although studies in different experimental systems and at different promoters have not always been in full agreement (Luse and Jacob, 1987; Jiang et al., 1993; Goodrich and Tjian, 1994; Timmers, 1994; Dvir et al., 1996a,b; Holstege et al., 1996). The use of ATP for the initial opening of the DNA has been established in several experimental systems. All studies that assay opening directly, using the single strand DNA probe, potassium permanganate, are in agreement on this issue (Wang et al., 1992; Jiang et al., 1993; Jiang and Gralla, 1995; Holstege et al., 1996; Wolner and Gralla, 1988). A recent study of the adenovirus major late promoter (AdML) in a purified, unactivated system described a subsequent step in the opening reaction (Holstege et al., 1996). In a two-step opening process, ATP hydrolysis leads initially to opening of the -9 to +1 region. Subsequent addition of nucleotides leading to first bond formation triggers extension of the open region to position +8. This two-step opening model has not been tested on other promoters or in other transcription systems. One important goal of this work is to test the applicability of the two-step model to other promoters. Below, we find that the two-stage opening applies to two other promoters, both using activated transcription. In addition, we further characterize the mechanism of each opening step.

In another important goal we address the full range of the use of the energy of ATP hydrolysis in the initiation pathway. The steps of open DNA bubble extension, first bond formation and short RNA synthesis are thought not to require ATP hydrolysis, although the evidence comes either from complex analyses of highly purified basal systems or is speculative (Goodrich and Tjian, 1994; Timmers, 1994; Holstege et al., 1996; Jiang et al., 1996). In this paper we will describe the development of detailed assays to test the need for ATP hydrolysis in these various sub-steps. The unexpected results show that ATP hydrolysis is required for polymerase to progress at each point assayed, prior to its release from the activated adeno E4 promoter. This includes a need to hydrolyze ATP to make the first bond of the RNA, even when the start site is already open. This surprising finding has implications for the role of factors in initiation and for the general mechanism by which energy is used during polymerase II transcription initiation.

The G9E4T promoter, which contains nine Gal4 binding sites upstream of the adenovirus E4 promoter, is shown in Figure 1A. Transcription initiates within the T₆A sequence (underlined), which is 25 bases downstream of the E4 TATA box. Here we designate the last thymine in the T6 sequence as +1, although initiation can occur just downstream of this position as well as upstream of it. The initiation pathway at this template has been studied extensively using unfractionated nuclear extracts (Wang et al., 1992; Jiang and Gralla, 1993, 1995; Jiang et al., 1995)_. Closed pre-initiation complexes form when this template is incubated with nuclear extract and activator (Gal4-AH). Addition of ATP, or a – hydrolyzable analogue such as dATP, leads to the opening of the start site. This has been shown by the ATP-dependent hypersensitivity of the T₆ sequence in a permanganate footprinting assay. The open pre-initiation complexes are functional in that the permanganate sensitive region associated with the polymerase can be chased downstream by the addition of NTPs. The first experiments were designed to test whether initiating nucleotides can lead to opening of the DNA segment downstream from +1, as observed at the AdML promoter.

Figure 1.

Initiation nucleotides extend the open region at the activated adeno E4 promoter. (A) The promoter region (non-template strand) of the G9E4T plasmid. Transcription can start from multiple positions within the T₆A region and the +1 designation is to provide a source of reference for discussion. (B) The indicated combinations of nucleotides were added to pre-initiation complexes for 2 min and then opening was probed with permanganate. A, U and C refer to the ribonucleoside triphosphates and dA refers to the deoxy form, all present at 25 M. The dinucleotide UpA was added with the others to a concentration of 1 mM. The +1 position and direction of transcription are indicated. (C) Permanganate probing was done as in part B, lane 7. In lane 2, the CTD kinase inhibitor H8 was added to 1.2 mM, or in lane 3 -amanitin was added to 1 g/ml, in both cases before any nucleotide/dinucleotide.

View full figure (88 KB)

Open complex formation at the G9E4T promoter occurs in two distinct steps

In the first experiment different combinations of nucleotides were added to pre-initiation complexes that had been opened using dATP. We did potassium permanganate footprinting to test if the nucleotides caused the transcription bubble to extend further downstream. The starting point is the open pre-initiation complex described previously and shown in Figure 1B, lane 1. By comparison with the background reactivity in the absence of ATP (lane 5), the primary reactivity is with the five thymines from +1 to -4. There is less reactivity with thymines at -5 and at -12 and -13. Thus the pre-initiation open complex (termed POC) extends from at least -13 to at least +1.

Addition of the initiating nucleotides ATP and UTP to the POC leads to a weak permanganate signal farther downstream (Figure 1B, lane 3). The extent of initiation using this combination of nucleotides is not known because of the multiple possible initiation start sites. Additional inclusion of CTP, in principle allowing RNA synthesis to stall prior to the +7 guanine, leads to a stronger downstream permanganate signal (lane 4). In this case the open region extends from at least -13 to at least +16. This initiation open complex (termed IOC) has thus extended the open region 15 bp farther than in the POC (POC).

In order to control the initiation point with greater certainty, we used dinucleotide priming of initiation. Figure 1B, line 7 shows that the joint use of the dinucleotide UpA and CTP, which should restrict synthesis to the RNA sequence UAC, also leads to strong downstream opening. The use of UpA alone appears to lead to a very weak downstream opening with these experimental conditions (lane 6, investigated further below). Collectively, these data show that downstream opening occurs efficiently under conditions that should support formation of either a three or a six nucleotide long RNA. Because we cannot exclude the possibility of low-level NTP contamination, these lengths, and others estimated below, must be considered approximations. However, such contamination may be minimal, as clear pattern changes are induced by addition of ATP (lane 2 versus lane 3) or CTP (lane 6 versus lane 7). The extent of the opened region in the IOC seems to be about the same in these two cases (compare lanes 4 and 7).

This initiating nucleotide-dependent bubble extension reaction is blocked by the RNA synthesis inhibitor –amanitin (Figure 1C, lane 3 versus lane 1). Thus, when RNA synthesis is blocked, bubble extension is blocked. The bubble extension reaction is not blocked by the inhibitor H8 (Figure 1C, lane 2 versus lane 1; Serizawa et al., 1993a), which is known to interfere with the phosphorylation of the polymerase CTD in this experimental system (Jiang et al., 1996). Thus, bubble extension does not require the kinase activity of TFIIH. We interpret these experiments as showing that short RNA synthesis at the activated E4 promoter is accompanied by downstream expansion of the open DNA using a reaction that does not depend on CTD kinase activity.

Two successive ATP hydrolysis events trigger stable IOC formation

The POC requires ATP hydrolysis to form, both at the E4 promoter and the adeno ML (AdML) promoter (Wang et al., 1992; Holstege et al., 1996). As shown above for E4 and previously for AdML, both promoters subsequently extend the open region to form an IOC. Several critical features of these complexes have not been established, including their stability and whether ATP is needed to form the IOC from the POC.

We first measured the stability of the POC. As ATP is required for its formation, we attempted to destabilize fully formed POCs by removing the ATP. This was accomplished using the standard procedure of digesting the ATP with hexokinase and glucose (Bieker and Roeder, 1986; Arias and Dynan, 1989; Dvir et al., 1996a). When ATP is depleted in this manner, the permanganate signal 8 min later is dramatically reduced (Figure 2, lane 8). It is slightly greater than a negative control where POCs were never formed (lane 1) but much less than the starting amount of POC (lane 2). Moreover, the signal is much less than a comparison where POCs were formed with ATP and then incubated with glucose or hexokinase alone for the same 8 min (lanes 7 and 9). This latter comparison shows that ATP is required continuously during these 8 min to keep the POCs fully open; if it is depleted at the beginning POCs mostly decay during this time. We infer that it is the continuous use of ATP that keeps the DNA fully open in the POC.

Figure 2.

Continuous ATP hydrolysis is needed for full stability of the POC (POC). Pre-initiation complexes were opened with dATP, except in lane 1. After 2 min hexokinase and glucose were added to deplete the dATP as indicated. Experiments showed that dATP was depleted within 1 min. Control lanes lacked hexokinase except for lane 9 which lacked glucose. After the times indicated permanganate probing was done to determine the extent to which the start site remained open.

View full figure (70 KB)

The experiment also shows the loss of permanganate signal at shorter times (Figure 2, lanes 2–6). Based on repeated experiments we estimate that the half-life of the POC is 5 min under these conditions. Of course, under conditions where ATP is continuously present, repeated ATP hydrolysis keeps the DNA start site fully open.

It is not known whether ATP is required again for the open region within the POC to be extended into downstream sequences to form the IOC. We tested this by forming the POC, depleting ATP and then assaying for initiating nucleotide-dependent extension of the bubble. Figure 3, lane 2 shows that the extended open region does not form using this protocol, in which ATP is removed prior to adding initiating nucleotides. The ATP removal, using hexokinase and glucose, was accomplished using a 2 min treatment, which is short enough so that the start site remains largely open (lane 2 and above discussion). If the hexokinase is simply left out in control experiments, maintaining ATP levels, the initiating nucleotides UpA and CTP can trigger downstream extension of the bubble (Figure 3, lane 1). We conclude that extension of the bubble requires ATP.

Figure 3.

ATP hydrolysis is required for conversion of the POC to the IOC. Preinitiation open complexes were formed in the presence dATP and UpA. ATPase depletion experiments in lanes 1 and 2 used hexokinase–glucose (lane 2) or a glucose only control (lane 1). This was followed after 2 min by addition of CTP to attempt to trigger downstream opening. For lanes 3 and 4, either CTP (lane 3) or CTP/500 M ATP--S mixture (lane 4) was added to the POC. Permanganate probing was done 3 min later to establish the need for ATP hydrolysis to obtain downstream opening.

View full figure (60 KB)

In the next experiment we used the non-hydrolyzable ATP analogue, ATP--S, as a competitor for –-hydrolysis of ATP. After POC formation, either UpA/CTP or UpA/CTP/ATP--S were added. The result shows that ATP--S interferes with the conversion of POC to IOC (Figure 3, lane 4 versus lane 3). Using ATP in the same experiment does not have any inhibitory effect (data not shown). This experiment independently illustrates that the hydrolysis of ATP is required for the POC to IOC step.

As ATP appears to be used twice, first to form the POC and then again to form the IOC, we wondered if the overall process leads to a complex with greater stability. Recall that the POC is only moderately stable, with a half-life on the order of 5 min if ATP is depleted. A similar set of experiments shows that the IOC is indeed more stable. Different combinations of nucleotides were used to convert the POC to the IOC. Hexokinase and glucose were then added to deplete the ATP. Five minutes later (Figure 4A) or 8 min later (Figure 4B) permanganate assays were carried out to view the amount of downstream opening remaining (brackets). The results are shown in Figure 4A, lane 3 versus lane 4 and Figure 4B, lane 7 versus lane 8, lane 9 versus lane 10, and lane 11 versus lane 12. On average there is little loss of downstream IOC signal induced by the depletion of ATP, using hexokinase–glucose. This contrasts with the much greater loss of signal when the POC start site opening is assayed in parallel (Figure 4A, lane 1 versus lane 2 and 4B, lane 5 versus lane 6).

Figure 4.

Enhanced stability of the IOC. The indicated combinations of nucleotides/dinucleotides were added for 3 min to form either the POC (lanes 1, 2, 5 and 6) or the IOC (lanes 3, 4 and 7–12). dATP/ATP was rapidly depleted with hexokinase–glucose (Hex) as indicated. Permanganate probing was done after 5 min (A) or 8 min (B). The minor differences in pattern intensities seen in panel B result from small variations in the timing of the experiments.

View full figure (87 KB)

We conclude that the IOC and the POC differ in stability, with the IOC being substantially more stable. Thus the use of the second ATP is a stabilizing event. That is, the creation of the IOC is not simply a bubble extension reaction but also the formation of a distinct complex with greater stability. In most of the experiments, IOC formation seems to be accompanied by some loss of permanganate sensitivity around the start site (see, for example, Figure 4, lane 1 versus lane 3 and lane 5 versus lane 7). This supports the view that the POC and the IOC are distinct complexes with distinct characteristics.

ATP hydrolysis is required to begin RNA synthesis, even after the start site is opened

The above experiments show that ATP hydrolysis is needed to convert the POC to the IOC. This transition involves the use of initiating nucleotides. Thus, it is possible that ATP hydrolysis is directly required to form the first bond of the RNA. It has been established that first bond formation does not occur without a source of hydrolyzable ATP, but this was thought to be because ATP was needed to open the start site (Jiang et al., 1995; Dvir et al., 1996b). The above experiments raise a new possibility, namely that even after the DNA is opened, ATP must be used again to begin RNA synthesis. This is possible because ATP must be used again for the initiating nucleotides to convert the POC to the IOC. We tested this possibility by opening the DNA and then following first bond formation and its dependence on ATP, using an abortive initiation assay.

We established conditions for the abortive initiation assay previously on the same E4 template as that used here (Jiang et al., 1995). The assay follows formation of the trinucleotide UAC from UpA and radioactive CTP. The appearance of UAC in a reaction that is sensitive to -amanitin (a critical control) monitors appropriate first bond formation in this system. Controls on this assay have also been established previously (Jiang et al., 1995). We used this assay to explore whether POCs need ATP in order to form the first bond of the RNA.

In the initial experiments dATP is added to form these POCs, which are then passed through a CL-4B column, following established procedures (Jiang et al., 1995). This fractionation removes most of the free proteins and is necessary to obtain a clear UAC signal. The column also removes the dATP. UpA and [-³²P]CTP are added to the dATP-depleted fraction and then UAC synthesis is followed during the subsequent 3 min. These complexes fail to support any -amanitin-sensitive UAC synthesis (Figure 5B, compare lanes 1 and 2). If dATP is added back a very strong UAC signal is seen (Figure 5B, lane 3 versus lane 4).

Figure 5.

First bond formation requires ATP even when the start site is open. (A) shows the flow chart of the experiment, where treated column fractions were assayed either by a 3 min abortive initiation assay (abo.; results in B) or permanganate footprinting assay (ppm.; results in C). The dATP-depleted column fraction supports no first bond synthesis (lane 1 versus lane 2) even though 20% of the complexes are open (lane 8). Bond synthesis is fully restored by adding dATP (lane 3 versus lane 4) which also restores opening to the 100% level (lane 9). Subsequent dATP depletion (40 s treatment with 0.25/l hexokinase–12.5mM glucose) eliminates synthesis (lane 6 versus lanes 5 and 7), even though the start site remains largely open (lanes 11 and 13) throughout the abortive assay. Time elapsed after the addition of CTP in the dATP depletion experiments is indicated. 'ama' refers to addition of -amanitin.

View full figure (71 KB)

To monitor the extent of opening in these same reactions, the permanganate assay was done using these column fractions. The dATP-free complexes that failed to form any UAC signal still remained open to the 20% level following the 15 min column fractionation (Figure 5C, compare lane 8 and lane 9). When dATP is added back, open complex levels are restored to 100% (Figure 5C, lane 9). Although the dATP increases the opening from the 20% to the 100% level (Figure 5C, lanes 8 and 9) its effect on bond synthesis is far greater (Figure 5B, compare lanes 1 and 3); it essentially converts an undetectable signal to a very strong one.

As further confirmation of the need for ATP for the POCs to make RNA, a direct ATP depletion experiment was done. First, the column-isolated POCs were restored to the 100% open state by addition of dATP. Then a 40 s hexokinase–glucose treatment was used to deplete the dATP. This depletion of dATP abolished the capability of the POC to generate the short RNA, UAC, in the standard 3 min assay (Figure 5B, lane 6 versus lanes 5 and 7). It is very important to note that the start site remained substantially open during the entire abortive initiation assay, as followed directly using permanganate footprinting (Figure 5C, lanes 11 and 13 versus lanes 10 and 12, respectively). Thus, even though the DNA start site remained open, first bond synthesis cannot occur unless dATP is present. This series of experiments demonstrates that ATP is needed for open complexes to go on to begin RNA synthesis.

In a peripheral set of experiments we noted that the column-isolated POCs had slightly modified properties that make IOC formation more stringent. Adding dATP, UpA and CTP did not lead to efficient IOC formation but nucleotides allowing longer RNAs to form did trigger IOC formation (Figure 5C lanes 11 and 13, and experiments not shown). It is possible that this difference is caused by the loss of loosely associated protein factors during column fractionation or by trace amounts of NTPs in the extract.

Short RNA products help IOC formation

The above experiments have shown the need for ATP in two separate assays for events that occur after the start site has already been opened. That is, ATP is needed to begin RNA synthesis and also to trigger the formation of an IOC in which the bubble has been extended to downstream positions. The IOC that forms has an increased stability. One possible view of the overall initiation process is that it is a series of reactions leading to a state strongly stabilized by bound RNA. In this view, the mere presence of an appropriate RNA could have a direct influence on the properties of the various complexes. We noted that a very weak downstream signal, characteristic of IOC formation, could be obtained when a two nucleotide long RNA is added to the POCs (Figure 1B, lane 6 and Figure 6A, lane 6). We explored this effect further and found that the short RNA could trigger stronger downstream opening when experimental conditions were optimized.

Figure 6.

Optimization of the use of dinucleotide to trigger the downstream opening characteristic of the IOC. (A) Permanganate assays were done after treatment of pre-initiation complexes with combinations of UpA, CTP ('C') and the indicated concentrations of dATP ('dA'). 'Pre.' designates the preincubation of UpA with the closed complex for 4 min prior to the addition of dATP. In lanes 13 and 14 the DNA was opened with dATP and then UpA was added. Downstream opening was inhibited by depletion of dATP (B, lane 13 versus 14) or the presence of -amanitin (C, lane 16 versus 15).

View full figure (72 KB)

Two experimental variations are required to optimize the triggering of IOC formation by the dinucleotide, UpA. First, a higher concentration of dATP is needed, 125 M rather than the 25 M used previously (Figure 6A, compare lanes 3 and 6). Second, we found that the signal increased if UpA is preincubated with closed complexes for a short period of time before the addition of dATP (Figure 6A, lane 6 versus lane 12); possibly the initial opening reaction restricts access for the dinucleotide, as is known to occur for the single strand analogue, heparin. The two conditions together allow UpA to trigger fairly strong downstream opening (lanes 2 and 11). The signal can be strengthened further if CTP is added to allow a three nucleotide long RNA to form (lane 4). This effect of UpA is specific for the E4 initiation sequence. Other dinucleotides corresponding to the initiation region (UpU, CpA) have a similar effect (lanes 8 and 10), whereas the non-complementary ApA has no effect (Figure 6A, lane 9).

Because this effect is influenced by the concentration of dATP, it seems that ATP hydrolysis is used to trigger downstream opening, even in the absence of bond formation. This is supported by the experiment of Figure 6B where dATP depletion blocks UpA from triggering downstream opening. The catalytic inhibitor -amanitin also blocks this process (Figure 6C). We don't know the cause of this, but it is possible that amanitin binding influences the properties of the active site of the enzyme in a way that interferes with the action of UpA.

From this set of experiments we conclude that the incorporation of a two nucleotide long RNA complementary to the initiation sequence can trigger downstream opening. Even when downstream opening is triggered in this manner, it requires ATP. The opening can be further strengthened if the dinucleotide is used as a substrate to make a three nucleotide long RNA. As discussed above, we cannot exclude the possibility that some slightly longer RNAs are made under these conditions. Nonetheless, it appears that there are two important requirements for triggering downstream opening, ATP hydrolysis and the presence of a complementary short RNA.

Two-stage IOC formation is also observed at the DHFR promoter

In a final set of experiments we explored whether this two-stage opening process applies to the mouse dihydrofolate reductase (DHFR) promoter. The DHFR promoter is a representative of the class of TATA-less promoters (Azizkhan et al., 1993; Schilling and Farnham, 1994). Aspects of the initiation pathway have been proposed to differ at this promoter, as it has an altered dependence on the polymerase CTD (Thompson et al., 1989; Kang and Dahmus, 1993; Akoulitchev et al., 1995). It is an activated promoter with an initiator sequence downstream of several Sp1 binding sites. The initiation site is within three consecutive adenosines (Figure 7A).

Figure 7.

ATP-dependent formation of an IOC with downstream opening at the TATA-less DHFR promoter. (A) Sequences near the initiating non-template strand adenine are shown. (B) Nucleotides were added for 2 min as indicated and then the complexes were probed with permanganate. dATP and ATP were at 125 M. (C) 25 M dATP or ATP was added as indicated to trigger formation of the POC (lanes 2 and 3) or the IOC (lanes 4 and 5), opening at the DHFR promoter. dATP/ATP was then depleted in lanes 3 and 5. The samples were probed with permanganate 8 min later. The IOC opening at DHFR (upper panel) remains in lane 5 whereas the POC opening of an E4 template in the same reaction is lost (lower panel). The reactions were done with the two promoter templates, mixed and then probed with separate primers.

View full figure (87 KB)

dATP-dependent opening is very weak at this promoter. Repeated experiments and optimization were necessary to detect even a weak permanganate signal at thymines -8 and -9 (Figure 7B, lane 2 versus lane 1). Subsequent downstream opening is stronger and can be triggered by ATP (lane 3), which allows a short RNA to form within the initiation region. The complementary dinucleotide CpA also triggers downstream opening (lane 4) and only complementary nucleotides have this effect (data not shown). This effect of the complementary dinucleotide requires that dATP be present (no signal in lane 7). Characteristic of IOC formation, this downstream opening is found to be very stable, using ATP depletion by a hexokinase–glucose protocol as in Figure 4 (Figure 7C, lane 4 versus lane 5, upper panel), in contrast to the internal G9E4T control, where less stable POCs are assayed (Figure 7C, lane 4 versus lane 5, lower panel). Thus the DHFR promoter appears to mimic the ATP-dependent, two-stage opening seen in the above experiments at the adeno E4 promoter.

The most unexpected finding of this study concerns the remarkably extensive role of ATP in the transcription initiation pathway. The experiments show that if ATP is removed at any of several stages the initiation pathway is blocked at that stage. Perhaps the most striking of these experiments shows that even after the transcription start site is opened, RNA synthesis cannot begin unless ATP is present. Similarly, ATP is needed both when the start site is opened initially and in the separate process that occurs when this opening is extended further downstream. The number of processes that need ATP appears to exceed the number of identified candidate ATPases in the initiation factors TFIIH and TFIIE (see Introduction).

These surprising considerations raise the unusual prospect that transcription initiation may not be supported by a series of discrete ATPases, but rather occurs by means of an activated state maintained by continuous ATP hydrolysis. That is, a single ATPase may support a high energy form of the transcription complex and only this form can complete all the required initiation steps. Although there are no clear precedents for such multi-step activation events, there are numerous examples, such as G-proteins, Myosin and helicases, in which activated states are triggered by nucleotide hydrolysis (Bedinger and Alberts, 1983; Lahue and Matson, 1988; Cox, 1994; Todd et al., 1994; Senior et al., 1995; Clarke, 1996; Rayment et al., 1996; Cyr, 1997). This novel speculative possibility, as well as selected important new mechanistic features of the pathway, will be discussed below.

Open complex formation proceeds in two discrete steps at three promoters, in both basal and activated contexts

Start site opening was identified initially as an ATP-dependent process at the activated adeno E4 promoter in an unfractionated transcription extract (Wang et al., 1992). The process has been confirmed using several promoters in both basal and activated contexts as well as in fractionated and unfractionated systems (Jiang et al., 1993; Holstege et al., 1996; Wolner and Gralla, 1998). Subsequently, a second stage of opening, in which the open region is extended downstream, was identified in a highly purified system using the AdML promoter (Holstege et al., 1996). The current work has extended the two-stage model to the E4 promoter and to the TATA-less mouse DHFR promoter, using an activated, unfractionated system. We show elsewhere that the model also applies to a CREB-activated AdML promoter (Wolner and Gralla, 1998). Thus, two-stage opening appears to be very common. The present data indicate that both stages of opening require ATP.

Nevertheless there are some observable differences in the promoters studied. The first stage, formation of a POC in which the start site is open, is by far the easiest to detect at the E4 promoter. We show above that the POC has a lifetime of the order of 5 min at the E4 promoter. The complex is further stabilized by transcription initiation and extension of the bubble to form the IOC. It is possible that the POC is much less stable than the IOC at the AdML and DHFR promoters, accounting for the much greater difficulty in detecting the POC, compared with the IOC, at these promoters. This lower POC signal is not related to the presence of activator, as it is retained in the CREB-activated AdML promoter (Wolner and Gralla, 1998), and the DHFR promoter is activated by endogenous SP1 (Schilling and Farnham, 1994). It may be that at many promoters, bubble extension is exceedingly important in forming a complex of sufficient stability to proceed along the initiation pathway.

The initiation pathway and potential roles for RNA and ATPases

One surprising finding in the above experiments is that complementary dinucleotides can bypass the need for RNA chain initiation and directly trigger downstream opening to form the stable IOC. Thus, short RNAs may potentially be viewed as stabilizing ligands for the IOC. It is known that the initial stages of RNA production are accompanied by drastic changes in the transcription complex, including enhanced resistance to sarkosyl, tighter binding of RNA, altered interactions with DNA and release of certain general transcription factors (Cai and Luse, 1987a,b; Luse et al., 1987; Zawel et al., 1995). There are also examples of short RNA primers modifying the transcription pathway in certain systems (Akoulitchev et al., 1995; Holstege et al., 1996). These considerations raise the possibility that the RNA itself plays a central role in stabilizing and reconfiguring the transcription complex during initiation. In an extreme view, one might consider initiation to be a process leading to formation of a ternary complex of sufficient stability to allow the polymerase to fully retain and elongate the RNA.

These considerations and the new data presented here may be used to modify existing models for the sequence of steps that occur during transcription initiation. The first two overall steps are: (i) polymerase is recruited to an appropriate location at the promoter, and (ii) the start site is opened, allowing the proper complementary NTPs to become bound. The above considerations suggest that this may be followed by step (iii) formation of short RNAs driving formation of a stable ternary complex with a longer open region. When the ternary complex becomes sufficiently stable, the polymerase, with its bound RNA, may be released from the promoter into the elongation phase.

Irrespective of the speculative details of this model, one must incorporate the critical role for ATP. The new observation that ATP removal at any of these stages halts initiation is consistent with a wide variety of prior studies (Conaway and Conaway, 1988; Arias and Dynan, 1989; Dvir et al., 1996a,b). There are two exceptions. One may be explained by the high concentration of different hydrolyzable nucleotides used to substitute for ATP (Jiang et al., 1996). The other is the lack of ATP requirement in certain systems containing heteroduplex or highly supercoiled templates (Parvin and Sharp, 1993; Pan and Greenblatt, 1994; Tantin and Carey, 1994; Holstege et al., 1996). In these cases many factor requirements are partially bypassed, as has been explained by the use of alternative mechanisms in bypass transcription in simpler systems (Wang et al., 1997).

The current data do not allow one to distinguish how many ATPases are used during initiation. As discussed above, the data support a minimum number of two, the CTD kinase needed for promoter clearance and one or more ATPases required during the steps that precede clearance. Even if the intriguing single activating ATPase model presented above proves true, it does not necessarily imply that multiple ATP molecules need to be hydrolyzed during the successive initiation steps. During normal initiation, the ATP activated state could possibly exist long enough for the complex to proceed through the entire pathway. If this were not universally true then the need for multiple uses of ATP at some promoters could conceivably be used as a checkpoint mechanism for their regulation by providing additional opportunities for altering transcription initiation.

General materials

Ultrapure NTPs and dNTPs were from Pharmacia. Ribo-dinucleotides (UpA, CpA, ApA, UpU), -amanitin, hexokinase, Sepharose CL-4B and Sephadex G50–80 resin were from Sigma. Hexokinase was dialyzed into buffer D (20 mM HEPES pH 7.9, 100 mM KCl, 0.1 mM EDTA, 20% glycerol, 1 mM dithiothreitol) before use. [-³²P]CTP was from New England Nuclear. Nuclear extract was prepared as described (Dignam et al., 1983). The DNA template G9E4T contains nine GAL4 binding sites upstream of a truncated adenovirus E4 promoter (Carey et al., 1990a,b). The plasmid pSS625, generously provided by Dr P.Farnham, contains the Sau3A(+275)–SmaI(-356) fragment of mouse dhfr 5' end inserted into Bam/SmaI sites of Puc9 (Farnham and Schimke, 1986). All DNA templates were prepared with Maxiprep (Qiagen).

Permanganate footprinting

Potassium permanganate footprinting was carried out as described, with minor modifications (Jiang and Gralla, 1995). First, the transcription pre-initiation complex was assembled in a 40 l final volume for 30 min at 25°C, with 12.5 l HeLa nuclear extract, 12.5 l D buffer, 6 mM magnesium chloride, 200 ng of carrier plasmid DNA, and 10 ng supercoiled DNA template. Gal4-AH (Giniger and Ptashne, 1987) was present at this stage for the G9E4T template. dNTPs, NTPs, dinucleotides and hexokinase–glucose were added as designated. Unless indicated, the final concentrations used are: 25 M for any dNTPs or NTPs, 1 mM for dinucleotide, 0.125 unit/l for hexokinase and 2.5 mM for glucose. The POC or IOC formation process was allowed for 2 or 3 min. Then potassium permanganate to 6 mM was added for 2 min, to probe the single-stranded region of DNA. 3 l 2-Mercaptoethanol were used to quench the potassium permanganate, and 100 l transcription stop buffer (10 mM EDTA, 0.3 M sodium acetate pH 5.5, 0.2% SDS, 50 g/ml yeast tRNA) and 20 mg proteinase K were added. After a 1 h incubation at 37°C, hot phenol/chloroform and subsequently chloroform were used to extract proteins. DNA was precipitated with ethanol, redissolved in water, passed through a Sephadex G50–80 spin column and subjected to 35 cycles of primer extension using an end-labeled primer. After chloroform extraction, ethanol precipitation and wash, the sample was loaded onto a 6% sequencing gel. The potassium permanganate hypersensitive signals, indicative of DNA melting, were revealed by PhosphorImager analysis (Molecular Dynamics).

Abortive initiation

Abortive initiation analysis was as described with minor modifications (Jiang et al., 1995). Briefly, a 400 l pre-initiation mixture was prepared as above. dATP was added to 125 M for 2 min to allow open complexes to form. Then the mixture was loaded onto a 6 ml Sepharose CL-4B column, equilibrated with 0.6 buffer D with 6 mM magnesium chloride. The column was eluted with the same solution at room temperature, taking 15 min. The excluded fractions were pooled and primer UpA (1 mM) was added. 20 l aliquots were treated with variations described in the text and UpApC synthesis was started with the addition of [-³²P]CTP (25 M, 32 Ci/mmol). The reactions continued for 3 min and were stopped with 20 l of formamide–/8 M urea and loaded directly onto a 20% polyacrylamide–8 M urea sequencing gel. The gel was run at 1000 V for 10 h and UAC was quantified using a PhosphorImager. In parallel, 40 l samples of the excluded fraction were treated identically except that they were probed with potassium permanganate to monitor the status of template opening.

We thank Dr Peggy Farnham for providing pSS625 plasmid and general advice about the DHFR promoter. We also thank Dean Tantin, Yijian Bai and members of the Gralla group for reading the manuscript. This research was supported by grant GM49048 from the PHS.

Akoulitchev S, Makela TP, Weinberg RA and Reinberg D (1995) Requirement for TFIIH kinase activity in transcription by RNA polymerase II. Nature, 377, 557–560. | Article | PubMed | ISI | ChemPort |

Arias JA and Dynan WS (1989) Promoter-dependent transcription by RNA polymerase II using immobilized enzyme complexes. J Biol Chem, 264, 3223–3229. | PubMed | ISI | ChemPort |

Azizkhan JC, Jensen DE, Pierce AJ and Wade M (1993) Transcription from TATA-less promoters: dihydrofolate reductase as a model. Crit Rev Eukaryot Gene Expr, 3, 229–254. | PubMed | ChemPort |

Bedinger P and Alberts BM (1983) The 3'–5' proofreading exonuclease of bacteriophage T4 DNA polymerase is stimulated by other T4 DNA replication proteins. J Biol Chem, 258, 9649–9656. | PubMed | ChemPort |

Bieker JJ and Roeder RG (1986) Characterization of the nucleotide requirement for elimination of the rate-limiting step in 5S RNA gene transcription. J Biol Chem, 261, 9732–9738. | PubMed | ChemPort |

Bunick D, Zandomeni R, Ackerman S and Weinmann R (1982) Mechanism of RNA polymerase II–specific initiation of transcription in vitro: ATP requirement and uncapped runoff transcripts. Cell, 29, 877–886. | Article | PubMed | ISI | ChemPort |

Cai H and Luse DS (1987a) Transcription initiation by RNA polymerase II in vitro. Properties of pre-initiation, initiation, and elongation complexes. J Biol Chem, 262, 298–304. | PubMed | ChemPort |

Cai H and Luse DS (1987b) Variations in template protection by the RNA polymerase II transcription complex during the initiation process. Mol Cell Biol, 7, 3371–3379. | ChemPort |

Carey M, Leatherwood J and Ptashne M (1990a) A potent GAL4 derivative activates transcription at a distance in vitro. Science, 247, 710–712. | Article | PubMed | ISI | ChemPort |

Carey M, Lin YS, Green MR and Ptashne M (1990b) A mechanism for synergistic activation of a mammalian gene by GAL4 derivatives. Nature, 345, 361–364. | Article | PubMed | ISI | ChemPort |

Clarke AR (1996) Molecular chaperones in protein folding and translocation. Curr Opin Struct Biol, 6, 43–50. | Article | PubMed | ChemPort |

Conaway RC and Conaway JW (1988) ATP activates transcription initiation from promoters by RNA polymerase II in a reversible step prior to RNA synthesis. J Biol Chem, 263, 2962–2968. | PubMed | ChemPort |

Conaway RC and Conaway JW (1993) General initiation factors for RNA polymerase II. Annu Rev Biochem, 62, 161–190. | Article | PubMed | ISI | ChemPort |

Cox MM (1994) Why does RecA protein hydrolyse ATP? Trends Biochem Sci, 19, 217–222. | Article | PubMed | ISI | ChemPort |

Cyr DM (1997) Coupling chemical energy by the hsp70/tim44 complex to drive protein translocation into mitochondria. J Bioenerg Biomembr, 29, 29–34. | Article | PubMed | ISI | ChemPort |

Dahmus ME (1995) Phosphorylation of the C-terminal domain of RNA polymerase II. Biochim Biophys Acta, 1261, 171–182. | Article | PubMed | ISI | ChemPort |

Dvir A, Conaway RC and Conaway JW (1996a) Promoter escape by RNA polymerase II. A role for an ATP cofactor in suppression of arrest by polymerase at promoter-proximal sites. J Biol Chem, 271, 23352–23356. | Article | PubMed | ISI | ChemPort |

Dvir A, Garrett KP, Chalut C, Egly JM, Conaway JW and Conaway RC (1996b) A role for ATP and TFIIH in activation of the RNA polymerase II pre-initiation complex prior to transcription initiation. J Biol Chem, 271, 7245–7248. | Article | PubMed | ISI | ChemPort |

Farnham PJ and Schimke RT (1986) In vitro transcription and delimitation of promoter elements of the murine dihydrofolate reductase gene. Mol Cell Biol, 6, 2392–2401. | PubMed | ChemPort |

Feaver WJ, Svejstrup JQ, Bardwell L, Bardwell AJ, Buratowski S, Gulyas KD, Donahue TF, Friedberg EC and Kornberg RD (1993) Dual roles of a multiprotein complex from S. cerevisiae in transcription and DNA repair. Cell, 75, 1379–1387. | Article | PubMed | ISI | ChemPort |

Feaver WJ, Svejstrup JQ, Henry NL and Kornberg RD (1994) Relationship of CDK-activating kinase and RNA polymerase II CTD kinase TFIIH/TFIIK. Cell, 79, 1103–1109. | Article | PubMed | ISI | ChemPort |

Giniger E and Ptashne M (1987) Transcription in yeast activated by a putative amphipathic alpha helix linked to a DNA binding unit. Nature, 330, 670–672. | Article | PubMed | ISI | ChemPort |

Goodrich JA and Tjian R (1994) Transcription factors IIE and IIH and ATP hydrolysis direct promoter clearance by RNA polymerase II. Cell, 77, 145–156. | Article | PubMed | ISI | ChemPort |

Guzder SN, Sung P, Bailly V, Prakash L and Prakash S (1994) RAD25 is a DNA helicase required for DNA repair and RNA polymerase II transcription. Nature, 369, 578–581. | Article | PubMed | ISI | ChemPort |

Holstege FC, Tantin D, Carey M, van der Vliet PC and Timmers HT (1995) The requirement for the basal transcription factor IIE is determined by the helical stability of promoter DNA. EMBO J, 14, 810–819. | PubMed | ISI | ChemPort |

Holstege FC, van der Vliet PC and Timmers HT (1996) Opening of an RNA polymerase II promoter occurs in two distinct steps and requires the basal transcription factors IIE and IIH. EMBO J, 15, 1666–1677. | PubMed | ISI | ChemPort |

Jiang Y and Gralla JD (1993) Uncoupling of initiation and reinitiation rates during HeLa RNA polymerase II transcription in vitro. Mol Cell Biol, 13, 4572–4577. | PubMed | ISI | ChemPort |

Jiang Y and Gralla JD (1995) Nucleotide requirements for activated RNA polymerase II open complex formation in vitro. J Biol Chem, 270, 1277–1281. | Article | PubMed | ISI | ChemPort |

Jiang Y, Smale ST and Gralla JD (1993) A common ATP requirement for open complex formation and transcription at promoters containing initiator or TATA elements. J Biol Chem, 268, 6535–6540. | PubMed | ChemPort |

Jiang Y, Yan M and Gralla JD (1995) Abortive initiation and first bond formation at an activated adenovirus E4 promoter. J Biol Chem, 270, 27332–27338. | Article | PubMed | ChemPort |

Jiang Y, Yan M and Gralla JD (1996) A three-step pathway of transcription initiation leading to promoter clearance at an activation RNA polymerase II promoter. Mol Cell Biol, 16, 1614–1621. | PubMed | ChemPort |

Kang ME and Dahmus ME (1993) RNA polymerases IIA and IIO have distinct roles during transcription from the TATA-less murine dihydrofolate reductase promoter. J Biol Chem, 268, 25033–25040. | PubMed | ISI | ChemPort |

Lahue EE and Matson SW (1988) Escherichia coli DNA helicase I catalyzes a unidirectional and highly processive unwinding reaction. J Biol Chem, 263, 3208–3215. | PubMed | ChemPort |

Luse DS and Jacob GA (1987) Abortive initiation by RNA polymerase II in vitro at the adenovirus 2 major late promoter. J Biol Chem, 262, 14990–14997. | PubMed | ISI | ChemPort |

Luse DS, Kochel T, Kuempel ED, Coppola JA and Cai H (1987) Transcription initiation by RNA polymerase II in vitro. At least two nucleotides must be added to form a stable ternary complex. J Biol Chem, 262, 289–297. | PubMed | ChemPort |

Ohkuma Y and Roeder RG (1994) Regulation of TFIIH ATPase and kinase activities by TFIIE during active initiation complex formation. Nature, 368, 160–163. | Article | PubMed | ISI | ChemPort |

Orphanides G, Lagrange T and Reinberg D (1996) The general transcription factors of RNA polymerase II. Genes Dev, 10, 2657–2683. | Article | PubMed | ISI | ChemPort |

Pan G and Greenblatt J (1994) Initiation of transcription by RNA polymerase II is limited by melting of the promoter DNA in the region immediately upstream of the initiation site. J Biol Chem, 269, 30101–30104. | PubMed | ISI | ChemPort |

Park E, Guzder SN, Koken MH, Jaspers-Dekker I, Weeda G, Hoeijmakers JH, Prakash S and Prakash L (1992) RAD25 (SSL2), the yeast homolog of the human xeroderma pigmentosum group B DNA repair gene, is essential for viability. Proc Natl Acad Sci USA, 89, 11416–11420. | PubMed | ChemPort |

Parvin JD and Sharp PA (1993) DNA topology and a minimal set of basal factors for transcription by RNA polymerase II. Cell, 73, 533–540. | Article | PubMed | ISI | ChemPort |

Rayment I, Smith C and Yount RG (1996) The active site of myosin. Annu Rev Physiol, 58, 671–702. | Article | PubMed | ChemPort |

Roeder RG (1996) The role of general initiation factors in transcription by RNA polymerase II. Trends Biochem Sci, 21, 327–335. | Article | PubMed | ISI | ChemPort |

Sawadogo M and Roeder RG (1984) Energy requirement for specific transcription initiation by the human RNA polymerase II system. J Biol Chem, 259, 5321–5326. | PubMed | ISI | ChemPort |

Schaeffer L, Roy R, Humbert S, Moncollin V, Vermeulen W, Hoeijmakers JH, Chambon P and Egly JM (1993) DNA repair helicase: a component of BTF2 (TFIIH) basic transcription factor. Science, 260, 58–63. | Article | PubMed | ISI | ChemPort |

Schaeffer L, Moncollin V, Roy R, Staub A, Mezzina M, Sarasin A, Weeda G, Hoeijmakers JH and Egly JM (1994) The ERCC2/DNA repair protein is associated with the class II BTF2/TFIIH transcription factor. EMBO J, 13, 2388–2392. | PubMed | ISI | ChemPort |

Schilling LJ and Farnham PJ (1994) Transcriptional regulation of the dihydrofolate reductase/rep–3 locus. Crit Rev Eukaryot Gene Expr, 4, 19–53. | PubMed | ChemPort |

Senior AE, al-Shawi MK and Urbatsch IL (1995) The catalytic cycle of P-glycoprotein. FEBS Lett, 377, 285–289. | Article | PubMed | ISI | ChemPort |

Serizawa H, Conaway RC and Conaway JW (1993) Multifunctional RNA polymerase II initiation factor delta from rat liver. Relationship between carboxyl-terminal domain kinase, ATPase, and DNA helicase activities. J Biol Chem, 268, 17300–17308. | PubMed | ISI | ChemPort |

Serizawa H, Makela TP, Conaway JW, Conaway RC, Weinberg RA and Young RA (1995) Association of Cdk-activating kinase subunits with transcription factor TFIIH. Nature, 374, 280–282. | Article | PubMed | ISI | ChemPort |

Shiekhattar R, Mermelstein F, Fisher RP, Drapkin R, Dynlacht B, Wessling HC, Morgan DO and Reinberg D (1995) Cdk-activating kinase complex is a component of human transcription factor TFIIH. Nature, 374, 283–287. | Article | PubMed | ISI | ChemPort |

Tantin D and Carey M (1994) A heteroduplex template circumvents the energetic requirement for ATP during activated transcription by RNA polymerase II. J Biol Chem, 269, 17397–17400. | PubMed | ISI | ChemPort |

Thompson NE, Steinberg TH, Aronson DB and Burgess RR (1989) Inhibition of in vivo and in vitro transcription by monoclonal antibodies prepared against wheat germ RNA polymerase II that react with the heptapeptide repeat of eukaryotic RNA polymerase II. J Biol Chem, 264, 11511–11520. | PubMed | ISI | ChemPort |

Timmers HT (1994) Transcription initiation by RNA polymerase II does not require hydrolysis of the phosphoanhydride bond of ATP. EMBO J, 13, 391–399. | PubMed | ISI | ChemPort |

Todd MJ, Viitanen PV and Lorimer GH (1994) Dynamics of the chaperonin ATPase cycle: implications for facilitated protein folding. Science, 265, 659–666. | Article | PubMed | ISI | ChemPort |

Wang W, Carey M and Gralla JD (1992) Polymerase II promoter activation: closed complex formation and ATP-driven start site opening. Science, 255, 450–453. | Article | PubMed | ISI | ChemPort |

Wang JT, Syed A and Gralla JD (1997) Multiple pathways to bypass the enhancer requirement of sigma 54 RNA polymerase: Roles for DNA and protein determinants. Proc Natl Acad Sci USA, 94, 9538–9543. | Article | PubMed | ChemPort |

Wolner B and Gralla JD (1998) Promoter activation via a cyclic AMP response element in vitro. J Biol Chem, in press.

Zawel L, Kumar KP and Reinberg D (1995) Recycling of the general transcription factors during RNA polymerase II transcription. Genes Dev, 9, 1479–1490. | Article | PubMed | ISI | ChemPort |

Main navigation

• Search EMBO Journal

  Search

• Advanced search

• The EMBO Journal

  • Home
  • Aims and scope
  • Current issue
  • Advance online publication
  • Web focuses
  • Archive:
  • Browse by issue
  • Browse by subject
  • Browse by category
  • Free online sample issue
  • Press releases

• Authors and Referees

  • Editorial process
  • Guide for authors
  • Submit an article
  • Guide for referees
  • Editors and Editorial Board
  • Contact editorial office

• Customer Service

  • Subscribe
  • Order sample copy
  • Purchase articles
  • Reprints and permissions
  • Contact NPG
  • Advertising