Characterization of the Closed Complex Intermediate Formed during Transcription Initiation by Escherichia coli RNA Polymerase*

We have carried out detailed DNase I footprinting studies of the closed complex formed on the phage λprmup-1 Δ265 promoter under reaction conditions such that the contribution of the open complex to the footprint was negligible. Detailed quantification shows that the closed complex detected has the same binding constant as that determined in kinetic studies. The footprinting pattern of the closed complex shows major differences from that of the open complex. Not only is it about 20 base pairs shorter, there are also many fewer positions being protected around and upstream of the −35 region. We have derived potential contact regions in the closed and open complexes based on the DNase I footprinting patterns, and confirmed the contact region for the open complex by hydroxyl radical footprinting. One important finding is that most of the essential contacts with the phosphate groups in the −35 region are formed during the isomerization step, a conclusion consistent with our kinetic data showing that this step is salt dependent on this promoter. In addition, we found that the derived contact regions for the closed and open complexes are offset by about three base pairs in the −35 region, which suggests a shift of the contact during isomerization. Finally, we found that the footprinting pattern of the complex formed at 4 °C has some similarities to as well as differences from the closed complex formed under standard transcription conditions.

The frequency of transcription initiation in Escherichia coli is determined primarily by the rate of formation of the transcriptionally active RNA polymerase-promoter complex, the open complex. The process of open complex formation can be described by the following two-step model (1,2). R ϩ P L | ; RNA polymerase (R) binds reversibly to a promoter (P) to form the closed complex (RPc) with an equilibrium constant K B . Subsequently, the closed complex isomerizes into the open com-plex with a rate constant k f . The first step is in rapid equilibrium, while the second step is usually rate-limiting and leads almost irreversibly to the formation of the open complex (RPo). The structural properties of the open complex have been studied extensively. It is known that the promoter DNA in the open complex has a melted region extending approximately from the Ϫ10 region to the transcription start site (1). There is also a wealth of information about the RNA polymerase-promoter DNA interaction in the open complexes formed on different promoters from studies using both enzymatic and chemical probing techniques, such as DNase I footprinting (3,4), dimethyl sulfate modification protection (5)(6)(7)(8)(9), and UV cross-linking (8,10).
In contrast to the open complex, our understanding of the closed complex is very limited. The existence of the closed complex is based largely on kinetic analysis (1,11,12), and the observation that RNA polymerase can bind specifically to promoters at 0°C where open complex formation does not occur (13). Complexes formed on the lac UV5 and the T7-A3 promoters at 0°C have also been studied using DNase I footprinting (14). Footprinting using several chemical reagents has been studied at the T7A1 promoter as a function of temperature (15). However, it is not clear whether these complexes are the same as the closed complexes that form transiently during open complex formation under normal transcription conditions. Because detailed information regarding RNA polymerasepromoter interaction in the closed complex is important for understanding the mechanism of open complex formation, we have characterized the closed complex formed on the phage prmup-1⌬265 promoter using DNase I footprinting. Kinetic studies were performed to find suitable reaction conditions for the footprinting studies. Under the conditions chosen, we have been able to footprint the closed complex before a significant amount of open complex could form. The binding constant determined from the footprint analysis is in good agreement with the binding constant determined from kinetic assays, providing strong evidence that the footprinted complex is the closed complex proposed in the two-step model. Since only partial closed complex formation could be achieved under the experimental conditions, we have determined the fractional protection at each position by carrying out quantitative DNase I footprinting at several RNA polymerase concentrations. The analysis of these data permits the protection pattern of fullyformed closed complexes to be determined. The results show that the closed complex is distinct from that of the open complex, indicating significant changes in the RNA polymerasepromoter interaction during open complex formation. We also show that the DNase I footprint of the complex formed at 4°C has both similarities to and differences from that of the closed complex formed at higher temperatures. In addition, we have performed hydroxyl radical footprinting studies of the open complex.

DNA Fragment and Enzymes-A 564-base pair
HindIII-EcoRI fragment containing the bacteriophage prmup-1⌬265 promoter was isolated from 112prmup-1⌬265 phage (16) DNA and cloned into the pUC-13 plasmid. The prmup-1 mutation is a single base pair change (T 3 C) at Ϫ31 in the P RM promoter; the ⌬265 deletion removes the divergently oriented P R promoter. This DNA fragment was used both for abortive initiation assays and the footprinting experiments. When used for footprinting analysis, this DNA fragment was labeled at the EcoRI end either on the top strand with [␥-32 P]ATP (5000 Ci/mmol, Amersham Pharmacia Biotech) using polynucleotide kinase (New England Biolabs), or the bottom strand by filling in opposite the 5Ј overhanging end with [␣-32 P]dATP (3000 Ci/mmol, Amersham Corp.) using the Klenow fragment of DNA polymerase I (Life Technologies, Inc.).
E. coli RNA polymerase holoenzyme was isolated according to Burgess and Jendrisak (17) and Lowe et al. (18), and its activity (40%) was determined as described by Hawley and McClure (19). The concentrations reported here correspond to active holoenzyme.
Abortive Initiation Assays and Buffers -Abortive initiation assays were used to determine the kinetics of open complex formation as have been reported (19). The calculation of the apparent rate constant of open complex formation (k obs ) was based on the kinetics of the accumulation of abortive initiation product (the lag experiment) using a least-squares fitting program. The calculation of the RNA polymerase binding constant (K B ) and the rate constant (k f ) for isomerization from the closed complex into the open complex was based on plots of obs (1/k obs ) versus reciprocal RNA polymerase concentration according to Equation 1, as described previously (19).
The abortive initiation reaction reported here contains 0. 25 (20). This allowed footprinting studies of the closed complex to be performed at a lower and wider range of RNA polymerase concentrations. DNase I Footprinting-Transcription complexes were formed with 0.25 nM labeled prmup-1⌬265 promoter DNA and RNA polymerase at varying concentrations in 87 l of reaction buffer containing 0.1 mM CaCl 2 . The DNase I cleavage reaction was started with the addition of 5 l of 15 g/ml DNase I, and allowed to proceed for 30 s before being stopped with the addition of 12.5 l of 0.2 M EDTA, 5 l of 10% SDS, and 3 l of 3.33 mg/ml tRNA. The samples were incubated at 65°C for 20 min, then on ice for 10 min. The precipitate formed contained denatured proteins in complex with potassium-SDS, and was removed by centrifugation. The DNA sample in the supernatant was further purified by repeated ethanol precipitation, and analyzed by electrophoresis in a 9% polyacrylamide, 8 M urea sequencing gel. Autoradiography was performed with a Kodak X-Omat AR film.
Hydroxyl Radical Footprinting of the Open Complex-The hydroxyl radical footprinting was as described (21) with some modifications. Open complexes were formed with 0.25 nM labeled DNA fragment and 60 nM RNA polymerase in 87 l of standard reaction buffer following an incubation of 25 min at 25°C. The sample was shifted to 19°C, and incubated further for 30 min. Then, 1.5 l each of the following solutions was added: 1.2 mg/ml heparin, 6 mM Fe(II)EDTA (6 mM ferrous ammonium sulfate and 12 mM EDTA), and 250 mM sodium ascorbate. The hydroxyl radical footprinting reaction was started with the addition of 1.5 l of 6% H 2 O 2 , and was stopped following 30 s of incubation by the addition of 750 l of stopping solution (60 ml of 1.0 M thiourea, 70 l of 1.0 M sodium acetate, 3 l of 3.33 mg/ml tRNA, and 620 l of 95% ethanol). The samples were put into a dry ice-ethanol bath. The DNA was isolated by centrifugation, and reprecipitated with ethanol. The samples were further purified by SDS-potassium precipitation and ethanol precipitation, and were analyzed in a sequencing gel as described above.
Densitometry Analysis of Footprinting Autoradiograms-The autoradiograms were scanned in one dimension with a Zeineh 2D/1D densitometer (Biomedical Instrument Inc.); air was used as background to adjust absorbance to zero during scanning. In the scanning, we used a laser beam slightly wider than the darkest band in the autoradiogram being analyzed. Therefore, the average optical density of the area covered by the laser beam was measured. This compensated for the variations in band intensities and shapes, which have been considered to be the major causes for inaccuracy in the analysis using one-dimensional scanning (22). The densitometer tracing of a blank area of the autoradiogram was used for background subtraction of the sample lanes. To calculate the fractional protection of each feature in the footprints of the RNA polymerase-promoter complexes, the integration values were corrected for the small variations in the total amount of DNA loaded in each lane using normalization band(s), which did not change upon transcription complex formation (see Fig. 3). The fractional protection (F pro ) was calculated as shown by Equation 2.
A was the normalized integration value of a protected feature (i.e. band or group of bands), and A o was the integration value of the corresponding feature in the control without RNA polymerase. Calculation of K B from the Fractional Protection-The fractional protection (F pro ) in the DNase I footprint was assumed to be proportional to the fractional occupancy. Based on Reaction 1, the closed complex is in a rapid equilibrium with free RNA polymerase and promoter DNA. Therefore, the fractional protection of the closed complex is related to RNA polymerase concentration as shown in the following equation.
F max is the maximum fractional protection of a feature when closed complex formation is complete. Therefore, from a double-reciprocal plot of F pro versus [R], K B is calculated as the intercept divided by the slope; the maximum fractional protection, F max , is obtained as the inverse of the intercept.

Kinetics of Open Complex Formation-Kinetic
analysis of open complex formation served two purposes in the studies reported here. First, we used it as a tool to find suitable conditions for DNase I footprinting of the closed complex; second, the results provided a comparison between the functional assays and the footprinting results. For the first purpose, we determined the kinetics of open complex formation at several reaction temperatures, and in different buffers. The abortive initiation assay as described previously (19) was used to determine both the binding constant (K B ) of the closed complex, and the rate constant (k f )) for the isomerization of the closed complex into the open complex. The results of the kinetic experiments performed under several of the conditions studied are shown in Table I. Open complex formation on this promoter is dependent on salt concentration. The salt concentration dependence of the second order rate constant, K B k f ), measured at 25°C corresponds to the release of about 9 (8.7 Ϯ 1.2) counterions during open complex formation based on calculations using the equations of Record and co-workers (20,23). This is similar to the results reported for the P R and P R Ј promoters by Leirmo et al. (20). We found that both steps of open complex formation were salt-dependent on the prmup-1⌬265 promoter, while Leirmo et al. have shown that only the first step was salt-dependent on the P R and P R Ј promoters.
DNase I Footprinting-Based on the kinetic studies described above, we chose to perform most of our footprinting experiments under the following standard conditions: assay buffer containing 200 mM potassium glutamate and a reaction temperature of 19°C. Our footprinting studies of the closed complex involve a short DNase I digestion reaction (30 s) carried out right after RNA polymerase was mixed with promoter DNA (30 s). A calculation using the kinetic data described above predicted that the fraction of open complex formed should be less than 5% at all of the RNA polymerase concentrations used under our footprinting conditions, and is negligible comparing to the fraction of closed complex formed. Fig. 1 shows the autoradiograms obtained from DNase I footprinting on both the top and the bottom strands. Lane 2 shows the footprint obtained 30 s after RNA polymerase (20 nM) was mixed with promoter DNA; significant protection is apparent in the promoter region, particularly around the Ϫ10 region. A calculation using the K B and k f ) values shown in Table I predicted that the fraction of the closed complex formed should be about 63% while that of the open complex formed should be less than 5% under the conditions used. Therefore, this protection is likely to be caused primarily by the formation of the closed complex. In support of this, we found that such protection was completely sensitive to heparin challenge (lane 3). In addition, the footprint is distinct from that of the open complex (lane 4). In general, extensive protection seems to be restricted to the region from about Ϫ30 to ϩ10 in the closed complex, while the protection covers the whole promoter region in the open complex (see below).
Determination of K B Using Quantitative DNase I Footprinting-If the complex detected right after RNA polymerase was mixed with promoter DNA is the closed complex depicted in the two-step model, it is expected to form in a RNA polymerase concentration-dependent manner and to have a binding constant that is the same as K B derived from functional assays described above. To confirm that this is the case, we carried out DNase I footprinting analysis at varying RNA polymerase concentrations. The autoradiogram resulting from one such experiment performed with the promoter DNA labeled on the top strand is shown in Fig. 2. As expected, the extent of protection in the promoter region increased with increasing RNA polymerase concentrations.
To determine the binding constant, we quantified the frac- tional protection at each RNA polymerase concentration by densitometry scanning. In these analyses, small variations in the amount of DNA loaded to each lane were normalized based on the intensities of the normalization bands, which corresponded to positions Ϫ39 and Ϫ40 on the top strand and Ϫ35 on the bottom strand. Fig. 3 shows that DNase I cleavage at these positions was not significantly affected by the binding of RNA polymerase.
We determined the fractional protection for the entire protected regions, single bands, or small groups of bands. Subsequently, using double-reciprocal plots of fractional protection versus RNA polymerase concentration, we determined the binding constant of the closed complex. Fig. 4A shows several examples of such plots, and the binding constants derived from these as well as several other similar plots are summarized in Table II, part A. We observed good agreement between the binding constants determined for different portions of the footprint and the whole protected regions of both strands of the promoter DNA. Importantly, these binding constants are, within experimental error, the same as the K B value measured by the abortive initiation assays under the same conditions. Therefore, the complex detected in the footprinting analysis indeed is the closed complex intermediate proposed in the two-step kinetic model.
Determination of the DNase I Footprinting Pattern of the Closed Complex-The footprint observed at a single RNA polymerase concentration is not sufficient to determine the protection pattern of the fully formed closed complex. This is because the closed complex will not reach full occupancy at lower RNA polymerase concentrations (e.g. Fig. 1), while at higher RNA polymerase concentrations, where full occupancy can theoretically be achieved, nonspecific protection becomes significant and the specific protection pattern of the closed complex becomes obscured. Therefore, we have determined the protection pattern by carrying out quantitative footprinting analysis at varying RNA polymerase concentrations as described above.
The double-reciprocal plots of fractional protection versus RNA polymerase concentration, which was also used to determine the binding constant as described above, was used to determined the maximum fractional protection, i.e. the fractional protection expected at saturating RNA polymerase concentration, at each position. The quantification was carried out for each individual band whenever possible, but for the bands that were too close to each other, or too weak to be quantified alone, they were quantified as small groups. Fig. 4B shows several examples of the double-reciprocal plots from the top strand. The binding constants calculated from these plots as well as most other protected bands in the promoter region are similar to each other and to the binding constant determined for the entire regions described above. This demonstrates that the protection of these bands corresponds to the binding of a single RNA polymerase. In contrast, the intercepts of the different plots and consequently the maximum fractional protections calculated are different. The protection pattern of the closed complex derived from the maximum fractional protection at different positions in the promoter region is shown in Fig. 5 (C and D).
Comparison of the Protection Patterns of the Closed and the Open Complexes-Besides the protection pattern of the closed complex, we have also determined, using densitometry scanning, the protection pattern of the open complex, which is shown in Fig. 5 (A and B). complex, the protected region extends from Ϫ52 to ϩ20 on the top strand, and from Ϫ56 to ϩ21 on the bottom strand. In the closed complex, the protected region extends from about Ϫ45 to around ϩ10. The downstream boundaries on both strands are between ϩ10 and ϩ16; the exact position is not certain because the maximum fractional protection determined for the positions from ϩ11 to ϩ15 was less precise due to weak DNase I cleavage at these positions. The upstream boundary on the bottom strand is at Ϫ43, as judged by the fact that none of the positions beyond was protected by more than 30%, which is less than half of the average fractional protection in the major protected region. For a similar reason, the upstream boundary on the top strand is around Ϫ46. The histogram in Fig. 5D also shows that there was significant protection upstream from Ϫ51 on the top strand (dashed line in the figure). However, we found that this protection originated from binding of RNA polymerase at the overhanging end. This protection displays different RNA polymerase concentration dependence compared with those observed in the promoter region, and has a binding constant of 2.5 ϫ 10 8 M Ϫ1 , which is more than 2 times larger than that of the closed complex. Moreover, this protection was completely abolished after the overhanging end was blunted by a nucleotide filling-in reaction using the Klenow fragment.
The second major difference between the protection patterns of the closed and the open complexes is found around and upstream from the Ϫ35 region. In general, the protection in this region was much more extensive in the open complex. We refer to the positions with protection less than 30% as DNase I accessible sites. In Based on crystal structure, DNase I needs to contact up to six phosphate groups on substrate DNA for each cleavage (24). In other words, each DNase I-accessible site in the DNase I footprint actually represents a cluster of six phosphate groups that is not in contact with RNA polymerase. Based on this reason-ing, we have derived the potential RNA polymerase contact region on promoter DNA in the closed and the open complexes (Fig. 6). A similar analysis has also been performed previously (25)(26)(27). Our analysis shows that the contact region of the closed complex is about 10 base pairs shorter than that of the open complex upstream of the Ϫ35 region and that the contact in the Ϫ35 region is much narrower. It also shows that the locations of the contact regions of the two complexes are different. Specifically, the centers of the two regions are offset by about three base pairs. This finding suggests a change in the relative positioning of RNA polymerase on the promoter DNA (see "Discussion").
Changes in DNase I Footprinting Patterns during Open Complex Formation-We carried out DNase I footprinting at different times after the reaction for open complex formation was started. Fig. 7A  DNase I Footprinting at 4°C and Other Non-standard Conditions-RNA polymerase-promoter DNA complexes formed at temperatures near 0°C have been considered to resemble the closed complex and have been in some cases studied by DNase I footprinting (14). To reveal whether such complexes have the same protection pattern as the closed complex formed under standard transcription conditions, we have performed DNase I footprinting analysis on the complex formed on prmup-1⌬265 promoter at 4°C (data not shown). We found that the footprint obtained has both similarities to as well as differences from that of the closed complex described above. The comparable features are: (i) the downstream boundary is between ϩ10 and ϩ16 on the top strand, and between ϩ8 and ϩ16 on the bottom strand; (ii) the positions from Ϫ38 to Ϫ40 on the top strand, and Ϫ35 on the bottom strand were not protected. The major difference is that the protection around the Ϫ35 region is in general more extensive at 4°C. Specifically, the positions from Ϫ30 to Ϫ33 on the top strand, and position Ϫ34 on the bottom strand were well protected in the complex formed at 4°C, but not at all in the closed complex formed at 19°C. In addition, there was also generally more protection upstream from Ϫ40. For example, the positions from Ϫ42 to Ϫ44 on the top strand were protected to about the same extent as the positions in the spacer region of the promoter; on the bottom strand, the protection at positions Ϫ44 to Ϫ50 was more than 50%. The complex formed at 4°C also has a larger binding constant than that of the closed complex determined under standard conditions (Table II,

part B).
In contrast to results obtained for the complex formed at 4°C, we have found that the protection patterns of the closed complex footprinted in KCl buffer at 19°C, and in potassium glutamate buffers at 25°C were very similar to the pattern of the closed complex discussed above. The binding constant of the closed complex determined in buffer containing 100 mM KCl at 19°C is shown in Table II, part B; it is also similar to the K B value determined by abortive initiation assays shown in Table I. The K B for each region was calculated using the double reciprocal plot of fractional protection of each region versus RNA polymerase concentration as shown in Fig. 4. The errors for the K B values are shown in parentheses. (a) The footprinting was carried out as described in Fig. 2 on both the top and bottom strands. (b) The footprinting was carried out under the reaction conditions specified. KGlu, potassium glutamate.

Top strand
Bottom strand A. KGlu (200 mM, 19°C)ϩ12 to Ϫ10 7.2 (1.3) ϩ14 to Ϫ12 6.6 (1.1) Ϫ11 to Ϫ26 6.6 (1.1) Ϫ13 to Ϫ20 9.9 (1.2) Ϫ32 to Ϫ38 8.8 (0.60) Ϫ21 to Ϫ32 8.0 (1.2) ϩ12 to  Fig. 1, is plotted. For the closed complex, the maximum fractional protection (F max ) at each position on the top (C) and bottom (D) strands is plotted. F max values were determined from plots such as those shown in Fig. 4C. The errors for F max are shown only when greater than 15%. The broken line in the histogram of C upstream of Ϫ50 corresponds to protection from nonspecific complexes (see text).
Hydroxyl Radical Footprinting Pattern of the Open Complex-We have used hydroxyl radical footprinting to reveal in more detail the RNA polymerase contact on the promoter DNA in the open complex. Similar studies on the closed complex have not been successful because the hydroxyl radical rapidly inactivated free RNA polymerase (Ͼ80% in 30 s). The fractional protection at each position obtained from densitometry analysis is shown in Fig. 8. The overall protection pattern is similar to what has been reported for other promoters (15, 28 -30), and includes the following major features: (i) the major protection was confined to the region from Ϫ51 to ϩ16; (ii) in the region from Ϫ20 to Ϫ50, the protected regions alternate with unprotected regions with a 10-base pair periodicity, and in contrast, the protection downstream from Ϫ20 is very extensive, and covers regions of more than 10 base pairs long. Interestingly, several positions around the transcription start site were poorly protected. The accessibility in this region may be important for nucleotide triphosphates to bind to the template during transcription initiation.
Significantly, a comparison shows that most of the protected positions (F pro Ն 0.2) in the hydroxyl radical footprint around the Ϫ35 region are located in the contact region of the open complex as derived from the DNase I-accessible positions (Fig.  6); there are only a couple of positions outside the contact region that showed significant protection in the hydroxyl radical footprint. In addition, all of the DNase I-accessible positions are located in the unprotected regions in the hydroxyl radical footprint. Thus, the RNA polymerase contact region derived from DNase I-accessible sites are in good agreement with the hydroxyl radical protection pattern. DISCUSSION That the intermediate complex we have studied is indeed the closed complex depicted in the two-step kinetic model for open complex formation is supported by the following evidence. First, this complex has the same K B as that measured using abortive initiation assays. Second, it formed rapidly after RNA polymerase was mixed with the promoter DNA, and it was heparin-sensitive. Third, its DNase I footprinting pattern is distinct from that of the open complex. The differences between the two patterns indicate changes in the RNA polymerasepromoter interaction during the conversion of the closed complex into the open complex.
Changes in RNA Polymerase-Promoter DNA Interactions during Open Complex Formation in the Ϫ35 Region-We have shown that the difference between the protection patterns of the closed and open complexes is most dramatic around and upstream of the Ϫ35 region. In general, the protection in this region was much more extensive in the open complex than in the closed complex. The potential contact region of the closed complex, as derived from the DNase I-accessible sites, is 10 base pairs shorter than, and is only a third the width of that of the open complex (Fig. 6). This suggests that new contacts must be formed during the isomerization step.
Our ethylation interference studies on the prmup-1⌬265 pro- There are mutations in the Ϫ35 region that affect both K B and k f ), indicating that RNA polymerase contacts the base pairs in the Ϫ35 region in the closed complex. Our observation that the contact region of the closed complex covers a portion of the major groove in the Ϫ35 region indicates that specific contacts with some of the bases are likely. Genetic evidence has shown that the subunit interacts with the Ϫ35 region bases (31,32). In addition, two recent studies (33,34) have shown that repressor (bound at O R 2, and near the Ϫ35 region of P RM ) interacts with the subunit when it activates this promoter. Fig. 6 also shows that RNA polymerase contacts different regions of the promoter DNA in the closed and the open com-plexes. Significantly, more than half of the positions in the contact region of the open complex are accessible to DNase I in the closed complex, but some positions outside this contact region are not accessible. Accordingly, the centers of the contact regions of the two complexes are offset by about three base pairs. This change in the relative positioning of RNA polymerase and promoter DNA during isomerization may result in part from a DNA rotational change, which would be consistent with topological unwinding of about three base pairs. This topological unwinding might facilitate DNA melting during open complex formation. In addition, the topological unwinding suggested here may correspond to the DNA structural stress measured in some other studies (35)(36)(37).
The overall contact region for RNA polymerase in the closed complex is about 50 -55 base pairs long. The longest dimension of RNA polymerase is about 160 Å (38), about the length of 47 base pairs of B-form DNA. This suggests that RNA polymerase could bind to promoter DNA and form the closed complex without major conformational changes. However, since the contact region in the open complex is significantly longer, about 70 base pairs, significant conformational changes in RNA polymerase and/or the promoter DNA during open complex formation must occur during isomerization. An elongation of RNA polymerase has been suggested by neutron small angle scattering studies that showed a decrease in the cross section of RNA polymerase upon open complex formation on the T7-A1 promoter (339). A major conformational change of RNA polymerase has also been suggested based on thermodynamic studies (40). Although DNA conformational changes, such as extensive bending or wrapping around RNA polymerase (30), may also contribute to the extended protection, no DNA conformational change other than DNA melting (and topological unwinding) has been shown for open complex formation.
Comparison of Closed Complexes Formed on Other Promoters-To reveal RNA polymerase-promoter contact in the closed complex, several groups have performed footprinting studies on RNA polymerase-promoter complexes formed at low temperatures with 70 holoenzyme (14,15) or other holoenzymes containing minor sigma factors (41)(42)(43). Footprinting studies on the closed complexes formed with 54 holoenzyme under transcription conditions have also been reported, taking advantage of the fact that the conversion of such closed complex into the open complex requires both the nitrogen regulatory protein, NtrC, and ATP (44).
At low temperatures, open complex formation does not occur, and thus the low temperature footprints have been considered to be the footprints of the closed complexes. Our studies show that the footprint obtained for complexes formed at 4°C on the prmup-1⌬265 promoter indeed resembles the footprint of the closed complex detected under standard transcription conditions. At the same time, there are clear differences as well. Therefore, we believe it is important to perform footprinting studies on the closed complex under standard transcription conditions to allow detailed comparison between the closed and the open complexes.
Despite the fact that the protection pattern of the complexes formed at lower temperatures may differ in certain aspect from that of the closed complex formed under standard conditions, we have compared the protection pattern of the closed complex formed on the prmup-1⌬265 promoter and two previously reported DNase I footprints for 70 holoenzyme containing complexes formed on the lacUV5 promoter and the T7-A3 promoter at 0°C (14). We have found that these three patterns share some similarities compared with those of the open complexes: (i) the overall protected regions are shorter; (ii) except for the footprint for T7-A3 promoter, there are more DNase I-accessible positions; (iii) interestingly, certain positions seemed to be better protected in these complexes than in the open complexes.
Besides the shared similarities, the footprints for the putative closed complexes show major differences as well. First, the protected regions are different: the overall protected regions are from Ϫ5 to Ϫ51, from about ϩ1 to Ϫ55, and from about ϩ10 to Ϫ45, on T7-A3, UV5, and prmup-1 ⌬265 promoters, respectively. Second, the extent of the protection is different. In the case of T7-A3, the whole promoter region is almost completely protected with very few interruptions, and was comparable to that in the open complex. For the UV5 promoter, the protected and unprotected regions alternated throughout the whole region displaying a 10-base pair periodicity. Third, the locations of the DNase I-accessible positions are different. For example, the unprotected positions (relative to the Ϫ35 region of prmup-1⌬265 promoter) around the Ϫ35 regions on the top strand are: Ϫ32 to Ϫ34 for the UV5 promoter, and Ϫ30 to Ϫ33 for the prmup-1⌬265 promoter; the unprotected positions on the bottom strand are Ϫ28 for the T7-A3 promoter, Ϫ36 to Ϫ39 for the UV5 promoter, Ϫ34 and Ϫ35 for the prmup-1⌬265 promoter. These differences are very intriguing because it has been shown that the RNA polymerase actually makes very similar contacts in the open complexes formed on different promoters (1,3,6). We speculate that the differences in RNA polymerasepromoter contacts in the closed complexes as implicated by these studies may be partly responsible for the unique kinetic and thermodynamic behaviors of open complex formation on different promoters. As an example, our results have shown that both steps in the open complex formation are salt concentration-dependent on the prmup-1⌬265 promoter. Based on our footprinting results, the salt dependence of the isomerization step can be at least partly attributed to the formation of new phosphate contacts in the Ϫ35 region. In contrast to prmup-1⌬265 promoter, it has been shown that on the P R and P R Ј promoters, the first but not the second step is salt-dependent. We speculate that on such promoters, RNA polymerase may form more extensive contact with the promoter DNA in the closed complex.