bubble collapse, RNA, position, transcription, translocation, displacement, T7 RNA polymerase, reannealing, products, DNA, complexes, RNA polymerase, results, complex, template DNA, MM, positions, elongation, template strand, promoter, promoter region, RNA polymerases, initial, the transition
JBC Papers in Press. Published on August 25, 2004 as Manuscript M409118200 Initial bubble collapse plays a key role in the transition to elongation in T7 RNA polymerase Peng Gong, Edward A. Esposito and Craig T. Martin* Department of Chemistry
, University of Massachusetts
at Amherst, 710NorthPleasantStreet LGRT 701, Amherst, Massachusetts 01003-9336 Running title: Initial bubble collapse in promoter escape To whom correspondence should be addressed. Phone: (413) 545-3299. Fax:(413)545-4490. E-mail: [email protected]
Copyright 2004 by The American Society
for Biochemistry and Molecular Biology, Inc.
Initial bubble collapse in promoter escape Summary RNA polymerases bind to specific sequences in DNA, melt open duplex DNA around the start site, and start transcription within the initially melted bubble. The initially transcribing complex is relatively unstable, releasing short abortive products. After synthesis of a minimal length of RNA (about 10-12 bases in the T7 system), RNA polymerases complete the transition to a processive (highly stable) elongation phase and lose the initial promoter contacts. The current study strongly supports a model for T7 RNA polymerase in which initial bubble collapse from position 4 to position +3 is responsible for initiating RNA displacement in the transition process. More specifically, collapse of the bubble from position 4 to position 1 indirectly and energetically facilitates the direct strand invasion offered by collapse at positions +1 to +3. Parallel work shows that promoter release, another key event occurring during this stage of transcription, begins after translocation to position +8 and is largely complete by translocation to about position +12. The timing of promoter release agrees with the timing of initial bubble collapse determined by our previous fluorescence studies, suggesting that these two events are closely related. Page 2
Initial bubble collapse in promoter escape Introduction RNA polymerases bind to specific sequences in DNA, melt open duplex DNA around the start site, and start transcription within the initially melted bubble. The initially transcribing complex is relatively unstable, releasing short abortive products. After synthesis of a minimal transcript length, RNA polymerases complete the transition to a processive (highly stable) elongation phase and release the initial promoter contacts. The single-subunit phage T7 RNA polymerase carries out de novo transcription in a highly efficient manner, with key mechanistic features very similar to those of the structurally unrelated multi-subunit RNA polymerases (1). The polymerase binds specifically to a small duplex DNA element within the promoter (from position 17 to position 5 relative to the start site), melts open the immediate downstream region to form an initial bubble of about 7-8 base pairs, and begins transcription at the start site within the initial bubble (2-5). Biochemical studies of T7 RNA polymerase have provided mechanistic information on the process of promoter escape. When the RNA transcript is about 2-8 nucleotides in length, the initially transcribing complex is relatively unstable and tends to release the transcript and restart transcription (6,7). This abortive cycle is a defining feature of the initiation phase of transcription (8,9). After synthesis of about 12 nucleotides, the RNA polymerase complex enters the elongation phase (10-15). In contrast to the initially transcribing complex, the elongation complex is highly stable, allowing transcripts to be extended over thousands of bases with little premature termination. Finally, within the elongation phase, transcription shows sequencedependent termination, completing the synthesis of the RNA transcript (16-18). Page 3
Initial bubble collapse in promoter escape Crystal structure
s of an early initially transcribing complex (with a three-base transcript) and two elongation complexes provide abundant information on the initial and final states in the transition process (19-21). Comparing all three structures, it is seen that the conformation of the enzyme, in particular in the N-terminal region (residues 1-266), changes dramatically from an early initially transcribing complex to an elongation complex . The change in structure facilitates growth of a longer RNA-DNA hybrid (about 8 nucleotides), creates an RNA exit pathway, and disrupts the initial promoter binding region. This suggests that promoter release and initial RNA displacement are important in the transition process. Biochemical studies also provide insight into the transition from initiation to elongation in T7 RNA polymerase (11,13,14,22-25). Promoter release is thought to be required for the transition to elongation and footprinting experiments have shown that initial promoter contacts are lost in translocation between positions +6 and +15 (22). Since there is some evidence that specific contacts with promoter bases around positions 9 and 5 are thought to be lost on translocation to position +7 (23), the loss of the contact with promoter bases near position 17 might be the final signature of promoter release. Exonuclease footprinting experiments have shown that on translocation to position +8, the entire DNA region from position 17 to position 5 becomes accessible for exonuclease digestion in some population (about 5-20%) of the complexes. Upon translocation to positions +9 and +10, the population increases to about 4090%, indicating that the contact around position 17 is absent or weak at this stage (26). Thus, these studies argue that promoter release occurs on translocation beyond position +7. Fluorescence experiments demonstrate that collapse of the initially melted DNA bubble occurs on translocation beyond position +8 (13). Interestingly, the size of the bubble increases from about 7-8 base pairs in initial promoter melting to about 13 base pairs on translocation to Page 4
Initial bubble collapse in promoter escape position +8, without any collapse at the upstream edge of the bubble. The upstream edge of the bubble begins to collapse on translocation beyond position +8, and by the time the complex has reached position +15 (or sooner), the bubble has collapsed back to about 8 base pairs (13,14,27,28). It seems logical that initial bubble collapse would help to drive initial RNA displacement, via direct strand invasion, and indeed initial RNA displacement has been shown to occur at about this point in translocation (13). Promoter release, initial bubble collapse, and initial RNA displacement are likely associated with one another, accompanied by the conformational rearrangements of the enzyme in the final stage of the transition process. But the precise translocational timing of these three events, the relationship between them, and what movement of the enzyme drives or facilitates them, remain elusive. Various results provide important insight into their relationship. Avoiding initial bubble collapse by removing or altering the relevant nontemplate DNA (by creating partially single stranded, mismatched, or nicked constructs) impairs RNA displacement, resulting in an increase in 1113mer products during normal runoff transcription (7,12). Results from similar partially single stranded DNA constructs suggested the formation of an extended RNA-DNA hybrid in complexes stalled at position +13 (29). Thus the 1113mer products seen during the above runoff experiments likely arise from improper initial RNA displacement. That improper RNA displacement could limit forward translocation is also suggested in the mismatched bubble elongation complex models. On different nucleic acid scaffolds without complementarity in the nontemplate strand within the template-RNA hybrid region, efficient elongation of the RNA primers is possible, but is limited to only 3 to 5 bases (30,31), suggesting that without the assistance of DNA annealing at the upstream edge of the bubble, the polymerase has difficulty in properly initiating the process of RNA displacement. This interpretation is Page 5
Initial bubble collapse in promoter escape strengthened by a recent study in which a scaffold is generated containing a complementary nontemplate strand (32). In this case, bubble collapse and displacement of the upstream end of the RNA should be near normal, and assembled complexes do experience no limitation in elongation. In the current study, our first goal is to address the relationship between initial bubble collapse and initial RNA displacement, and to understand their importance in the transition process. Paralleling earlier studies (7,12,29), different DNA constructs were used in a transcription assay to investigate the importance of the nontemplate region of the DNA in initial RNA displacement. The results suggest that reannealing of the DNA from position 4 to position +3 is important to proper displacement of the 5' end of the RNA. The second goal is to determine the timing of promoter release using a sink challenge assay (33). By allowing transcription to stall at every translocational position from position +6 to position +15, we show that promoter release begins on translocation beyond position +8 and is largely complete on translocation to position +12. Results The following experiments address the interrelationship between three fundamental events in transcription: initial displacement of the 5' end of the RNA, collapse of the upstream edge of the initial transcription bubble, and release of specific protein-promoter contacts. Coupling between initial bubble collapse and initial RNA displacement The transcription bubble in T7 RNA polymerase begins as a 7-8 base pair bubble, and grows to about 12-13 base pairs, as the initially transcribing complex synthesizes up to a 7-8 base RNA (13). The following explores the relationship between the initial collapse of the upstream edge of Page 6
Initial bubble collapse in promoter escape the bubble and the displacement of the 5' end of the nascent RNA transcript. By analogy to deficiencies in elongation of scaffold or mismatched bubble elongation complexes, analysis of the results below uses the premature release of 1113mer RNA products as a sign of improper initial RNA displacement. More specifically, the molar ratio
of 1113mer to 1120mer products is taken to assess the initial RNA displacement. Of the complexes that have successfully extended past position +10, this ratio represents the fraction that terminate prematurely at positions +11 to +13. Partially single stranded constructs. In order to address the relationship between initial bubble collapse (upstream DNA reannealing) and initial RNA displacement, a set of partially single stranded DNA constructs was designed to compare with the control DNA construct (fully double stranded) in a transcription assay. All DNA constructs encode the identical runoff 20mer RNA. As shown in Table I, partially single stranded DNA constructs were prepared in which the nontemplate strand extends from position 22 upstream (in all constructs) to positions ranging from 5 to +3 downstream (constructs PSS to PSS[+3]). It is expected that constructs lacking nontemplate strand bases at positions where collapse is critical to invasion-mediated displacement of the 5' end of the RNA will be impaired in RNA displacement. As the critical nontemplate strand bases are restored, RNA displacement will be restored. Page 7
Initial bubble collapse in promoter escape
5'-GCGATTAATACGACTCACTATA GGGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAAAGG-5'
PSS[-5] 5'-GCGATTAATACGACTCAC-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAAAGG-5'
PSS[-4] 5'-GCGATTAATACGACTCACT-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAAAGG-5'
PSS[-3] 5'-GCGATTAATACGACTCACTA-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAAAGG-5'
PSS[-2] 5'-GCGATTAATACGACTCACTAT-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAAAGG-5'
PSS[-1] 5'-GCGATTAATACGACTCACTATA-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAAAGG-5'
PSS[+1] 5'-GCGATTAATACGACTCACTATA G-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAAAGG-5'
PSS[+2] 5'-GCGATTAATACGACTCACTATA GG-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAAAGG-5'
PSS[+3] 5'-GCGATTAATACGACTCACTATA GGG-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAAAGG-5' Table I. Partially single stranded DNA constructs. All constructs encode
the same 20 base runoff transcript, but differ in the downstream extension of
the nontemplate strand.
The results presented in Figure 1 show that in a construct containing the nontemplate strand downstream only to position 5, 1113mer products are substantially increased relative to transcription from the fully double stranded control, consistent with previous results (12). The data in lanes 1-5 of Figure 1 show that as the nontemplate strand is extended from position 5 to position 1, the accumulation of 1113mer products remains at high levels (more than four-fold above native levels). Collapse in this region is not expected to compete directly with RNA in the heteroduplex. As the nontemplate strand is lengthened downstream to positions +2 and +3, the levels of 1113mer products drop to near those of the double stranded control (Figure 1, lanes 7
Initial bubble collapse in promoter escape and 8), demonstrating that reannealing of the DNA bases from position 5 to position +3 is sufficient for proper initial RNA displacement. Extension of the nontemplate strand farther downstream does not alter the levels of 1113mer products (Gong and Martin, unpublished results). More importantly, the molar ratio of 1113mer to 1120mer products mirrors these trends, confirming that premature termination at positions +11 to +13, a property of complexes lacking the nontemplate strand, is fully rescued by extension of the nontemplate strand to only position +2. Figure 1. Coupling between bubble collapse and initial RNA displacement. Transcription from partially single stranded constructs in which the nontemplate strand extends downstream from position 5 to +3 (lanes 1-8) and from the double stranded control (lane 9) are compared (transcript lengths are indicated above the gel). Relative molar amounts of 1113mer transcripts are indicated in the bar chart
. Of the complexes that have successfully extended past position +10, the ratio of 1113mer to 1120mer products represents the fraction that terminate prematurely at positions +1113. The nomenclature is that of Table I. Mismatched bubble constructs. In order to confirm the generality of the results from the partially single stranded constructs, a set of DNA constructs was designed, each containing a Page 9
Initial bubble collapse in promoter escape window of 4 mismatched bases, with the window centered at various positions from 4 to +20 (Table II, constructs MM[4,1] to MM[+16,+20]).
5'-GCGATTAATACGACTCACgcgc GGGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGatat CCCTCTGGTGTTGCCAAAGG-5'
MM[+1,+4] 5'-GCGATTAATACGACTCACTATA tttcGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATAT ccctCTGGTGTTGCCAAAGG-5'
MM[+5,+8] 5'-GCGATTAATACGACTCACTATA GGGAtcaaACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTctggTGTTGCCAAAGG-5'
MM[+9,+12] 5'-GCGATTAATACGACTCACTATA GGGAGACCcaccCGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGtgttGCCAAAGG-5'
MM[+17,+20] 5'-GCGATTAATACGACTCACTATA GGGAGACCACAACGGTggaa-3' 3'-CGCTAATTATGCTGAGTGATAT CCCTCTGGTGTTGCCAaagg-5' Table II. Localized mismatch-containing DNA constructs. All constructs
contain the same template strand as the constructs in Table I (and so encode
the same 20 base transcript). Mismatched bases are shown in lower case.
Constructs with the mismatched bases in the initially melted region, MM[4,1] and MM[+1,+4], both give high ratios (about three fold above the native level) of 1113mer to 1120mer products (Figure 2, lanes 2 and 3), indicating an impairment in RNA displacement. The results for the MM[+1,+4] construct are consistent with a model in which bubble collapse from position +1 to position +3 is essential. That a construct with a mismatch bubble from position 4 to position 1 also yields high levels likely reflects that reannealing of bases in the 4 to 1 region cooperatively facilitates reannealing of the DNA from position +1 to position +3. As the mismatched region is moved downstream, seen in constructs MM[+5,+8] and MM[+9,+12], the ratio of 1113mer to 1120mer products is less than observed for mismatches around the start site, and close to that of fully duplex DNA (Figure 2, lanes 4 and 5), indicating that on these
Initial bubble collapse in promoter escape constructs RNA displacement proceeds with a minimal disturbance. As expected, mismatching of region farthest downstream as in construct MM[+17,+20], brings the ratio of 1113mer to 1120mer products to native level (Figure 2, lanes 6). Together with the results from the partially single stranded constructs, we conclude that reannealing of the DNA in the initially melted region (from position 4 to position +3) is the primary determinant for proper initial RNA displacement. Figure 2. Localized mismatched bases impair bubble collapse. Transcription from constructs with windows of four mismatched bases (lanes 2-6) are compared to that from a control construct (lane 1). The nomenclature is that of Table II. Transcripts are indicated as described in the legend to Figure 1. Energetic cooperativity in bubble collapse. The data presented in Figure 1 suggest that reannealing of the DNA bases at position +2 has the largest impact on the production of 1113mer products. In particular, the ratio of 1113mer to 1120mer products drops the most for the extension of the nontemplate strand from position +1 to position +2. Based on this observation, we expect that reannealing of the DNA at this position and the positions immediately upstream (positions +1 and 1) is most critical to proper RNA displacement. To test this assumption, a set of DNA constructs was designed containing single or double base pair mismatches, as summarized in Table III. Introducing different adjacent two-base mismatches in Page 11
Initial bubble collapse in promoter escape this critical region results in the highest (about three fold above native level) ratios of 1113mer to 1120mer products (Figure 3, lanes 2 and 3). Furthermore, any single-base mismatch in this region also leads to a high ratio (more than two fold above native level) of 1113mer to 1120mer products (Figure 3, lanes 4-6). These results demonstrate that even a small weakening of bubble collapse in this region is sufficient to impair proper RNA displacement.
DNA MM[-1,+1] MM[+1,+2] MM[-1] MM[+1] MM[+2] MM[-4,-3] MM[-4,-3]GC[-2,-1] MM[-2,-1] GC[-2,-1]
5'-GCGATTAATACGACTCACTATc tGGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATAt cCCTCTGGTGTTGCCAAAGG-5'
5'-GCGATTAATACGACTCACTATA ttGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATAT ccCTCTGGTGTTGCCAAAGG-5'
5'-GCGATTAATACGACTCACTATc GGGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATAt CCCTCTGGTGTTGCCAAAGG-5'
5'-GCGATTAATACGACTCACTATA tGGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATAT cCCTCTGGTGTTGCCAAAGG-5'
5'-GCGATTAATACGACTCACTATA GtGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATAT CcCTCTGGTGTTGCCAAAGG-5'
5'-GCGATTAATACGACTCACgcTA GGGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGatAT CCCTCTGGTGTTGCCAAAGG-5'
5'-GCGATTAATACGACTCACgcGC GGGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGatCG CCCTCTGGTGTTGCCAAAGG-5'
5'-GCGATTAATACGACTCACTAgc GGGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATat CCCTCTGGTGTTGCCAAAGG-5'
5'-GCGATTAATACGACTCACTAGC GGGAGACCACAACGGTTTCC-3' 3'-CGCTAATTATGCTGAGTGATCG CCCTCTGGTGTTGCCAAAGG-5'
Table III. Mismatches very near the start site. DNA constructs with
mismatched bases in the region from position 4 to position +2 and/or designed
GC pairs in positions 2 and 1. Mismatched bases are shown in lower case.
Designed GC pairs are shown in bold.
Initial bubble collapse in promoter escape Figure 3. Critical regions for proper RNA displacement. Transcription from constructs with mismatched bases in the proposed critical DNA region (lanes 2-6) are compared to that of control construct (lane 1). Transcription from constructs with mismatched bases and/or designed GC pairs (lanes 7-10), is compared to that from the control construct (lane 1). Construct nomenclature is that of Table III (a star notes the position of the GC dinucleotide step). Transcripts are indicated as described in the legend of Figure 1. The strong effect of the single-base mismatch at position 1 supports our earlier proposal that reannealing of DNA immediately upstream of the initial RNA displacement site (+1) facilitates the strand invasion immediately downstream. To confirm that collapse of bases from position 4 to position 1 helps the downstream collapse, another set of DNA constructs was designed. Mismatches at either positions 4/3 or positions 2/1 induce a high ratio of 1113mer to 1120mer products (Figure 3, lanes 7 and 9), implying that collapse of the bubble one to four bases upstream of the RNA facilitates its release. This appears to be a simple matter of energetics, however, in that the energetics can be restored by replacement of the adjacent AT pairs by GC pairs. Specifically, the construct MM[4,3]GC[2,1] shows near native ratios of 1113mer to 1120mer RNA products (Figure 3, lane 8). Thus, collapse of the upstream edge of the initial bubble from position 4 to position 1 contributes to the overall driving force to Page 13
Initial bubble collapse in promoter escape facilitate collapse at positions downstream. Finally, replacement of the TA step at positions 2 and 1 by a GC step in an otherwise wild type
construct leads to ratios of 1113mer to 1120mer products lower than seen in the fully duplex wild type construct (Figure 3, lane 10). Translocational timing of promoter release Promoter release, initial bubble collapse, and initial RNA displacement are expected to be closely related to one another in translocational timing. The results presented here suggest that initial bubble collapse and presumably the consequent displacement of the 5' end of the RNA from the initial hybrid are most critical to allowing the proper transition. But how is promoter release involved in this process? Is promoter release coupled to initial bubble collapse or initial RNA displacement or both? Determination of the translocational timing of promoter release would help to address these questions. The occurrence of promoter release was previously localized between translocational positions +7 and +15 by footprinting and UV-photo crosslinking studies (22,23). Preventing promoter release by the introduction of a covalent crosslink between the polymerase and the 17 template base of the promoter DNA results in substantial accumulation of 1213mer products (Esposito & Martin, unpublished results), suggesting that promoter release likely happens or has to happen at some point before the complex steps to position +12 or +13. Page 14
Initial bubble collapse in promoter escape Figure 4. The sink challenge assay. Initially transcribing complexes can release abortive RNAs without dissociating from the promoter. In this abortive cycle, the complex is not accessible to a sink promoter. In contrast, complexes that have escaped the promoter, are accessible to the sink on the path to reinitiation. Hence, susceptibility to sink measures promoter release. In an attempt to precisely determine the timing of promoter release, a set of DNA constructs derived from the T7 10 promoter was designed and is shown in Table IV. These constructs allow transcription to stall at every position from +6 to +15 in the presence of only GTP, ATP and CTP. In translocation of the complex to positions +6 and +7, the promoter has not been released (22,23). As illustrated in Figure 4, if RNA dissociates but polymerase and DNA remain associated, the polymerase is then able to resume the starting conformation and start a second round of transcription without rebinding to the promoter. In this mode, the complex is resistant to the challenge of a competitive promoter sink (33). On translocation to position +15, however, all promoter contacts are lost. To start a second round of transcription, the polymerase now has to rebind to a promoter and therefore is susceptible to challenge by the sink. Page 15
Initial bubble collapse in promoter escape
5'-TAATACGACTCACTATA GGGAGATACAAACGACTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTATGTTTGCTGATGC-5'
5'-TAATACGACTCACTATA GGGAGAATCAAACGACTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTTAGTTTGCTGATGC-5'
5'-TAATACGACTCACTATA GGGAGAACTAAACGACTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTTGATTTGCTGATGC-5'
5'-TAATACGACTCACTATA GGGAGAACATAACGACTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTTGTATTGCTGATGC-5'
5'-TAATACGACTCACTATA GGGAGAACAATACGACTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTTGTTATGCTGATGC-5'
5'-TAATACGACTCACTATA GGGAGAACAAATCGACTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTTGTTTAGCTGATGC-5'
5'-TAATACGACTCACTATA GGGAGAACAAACTGACTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTTGTTTGACTGATGC-5'
5'-TAATACGACTCACTATA GGGAGAACAAACGTACTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTTGTTTGCATGATGC-5'
5'-TAATACGACTCACTATA GGGAGAACAAACGATCTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTTGTTTGCTAGATGC-5'
5'-TAATACGACTCACTATA GGGAGAACAAACGACTTACG-3' 3'-ATTATGCTGAGTGATAT CCCTCTTGTTTGCTGAATGC-5'
A GCGAATTATGCTGAGTGATAT CCC-5'
Table IV. Constructs used in the sink challenge assay. Constructs
DS-6 through DS-15 contain both the template and nontemplate
strand from 17 to +20 (including the consensus 10 sequence from
17 to +6). The first TA pairs introduced in the encoding region are
shown in bold (which allows transcription to stall at different position
with only GTP, ATP and CTP).
Initial bubble collapse in promoter escape Figure 5. Detection of promoter release by the sink challenge assay. The data represent the number of transcription cycles observed in the first minute following the addition of sink. Enzyme and DNA were incubated in the presence of (cold) GTP, ATP, and CTP only, providing for stalling at the indicated positions. After 1.0 min, [32P]ATP and a twenty-fold excess of sink DNA were added and transcription continued for an additional 1.0 min. Each column thus represents resistance to the trap promoter in complexes stalled at positions from +6 to +15. A sharp decrease in transcription cycles is observed as the complex steps from position +8 to +9. Complexes stalled beyond position +12 are fully accessible to the trap promoter, such that almost no transcription is observed. Overlaid onto this graph in hatched bars are fluorescence data indicating collapse of the initially melted bubble, as probed by changes in the fluorescence from 2aminopurine placed at position 2 (from Liu & Martin, 2002). The "sink" used in this work is a hairpin DNA construct with a consensus duplex promoter region from position 17 to position 5, but containing only the template strand from position 4 to position +3 (Table IV and Figure 4). Constructs with similar structure have been shown to Page 17
Initial bubble collapse in promoter escape bind T7 RNA polymerase at least five times more tightly than does the fully duplex promoter (34) and so can efficiently inhibit transcription from normal DNA constructs. The results presented in Figure 5 show that after initiating transcription in the absence of sink and walking to positions +6, +7 or +8, addition of sink in 20fold excess over the normal DNA construct allows each polymerase, on average, to carry out about 5 cycles of transcription in 1 min. This indicates that the polymerase has not released its promoter contacts, and shows full retention through translocation to position +8, consistent with previous studies (22,23,33) In contrast, walking to position +9 decreases significantly (about 50%) the number of cycles observed in the presence of sink. The number of cycles decreases still further as the polymerase is stalled at positions from +10 to +12. Stalling by position +13 and beyond allows only trace amount of transcription in the presence of sink (less than 0.1 cycle is observed). This not only demonstrates that promoter release starts to happen as transcription steps beyond position +8, but also shows that promoter release is complete at translocational position +12 or +13. This timing agrees well with previous fluorescent studies measuring the occurrence of bubble collapse (13). Some of these fluorescence data are overlaid onto the functional data in Figure 5, illustrating clearly that promoter release and bubble collapse are closely related. Exonuclease footprinting studies suggest that promoter release begins on translocation to position +8 and is complete on translocation to position +9 or +10 (26), however as noted in the study, the processive action of the exonuclease might drive release to occur sooner than under normal conditions. Page 18
Initial bubble collapse in promoter escape Discussion Following normal promoter-dependent initiation of RNA synthesis, RNA polymerases enter a relatively unstable "initially transcribing" phase, characterized by the release of abortive transcripts (6,8,9,35). After translocation to about position +10 to +14, complexes then enter the more stable elongation phase (10,36). Similarly, a variety of studies show
that RNA polymerase retains its initial promoter-specific DNA contacts early in transcription (22,37,38). This process of converting from an unstable initially transcribing complex to a stable elongation complex contains within it a number of events. Promoter release, upstream bubble collapse, and initial RNA displacement. Previous results and the work shown here indicate that a critical event in this process, collapse of the initial DNA bubble, starts as the complex progresses from position +8 to position +9 (13). Another critical event, promoter release, has been shown by both exonuclease footprinting studies (26) and by the functional studies here, to occur at about the same stage of transcription. This correlation suggests that initial bubble collapse and promoter release are mechanistically associated. A simple model would have the 5' end of the RNA competitively displaced from the template strand by the reannealing of the nontemplate strand DNA in that same region. Indeed, the nontemplate DNA strand has previously been proposed to play an important role in RNA displacement (11,29,33). The production of 1113mer products from templates lacking the nontemplate strand downstream of the start site has been proposed to arise from formation of a persistent hybrid, as a result of improper RNA displacement (29). One must be careful to note, however, that in multiple turnover experiments, constructs lacking the nontemplate strand in the transcribed region (partially single stranded constructs) quickly convert to nicked or gapped Page 19
Initial bubble collapse in promoter escape constructs containing an RNA:DNA hybrid within the transcribed regions, regardless of whether the hybrid forms during the first round (persistent) or whether released RNA subsequently anneals to the partially single stranded DNA construct. Results presented here demonstrate that both the complementarity and the integrity of the DNA region from position 5 to +3 is essential to achieve proper initial RNA displacement. RNA products of length 11-13 bases are also observed at low levels in native transcription from fully duplex DNA, suggesting that even with intact complementary DNA strands
, complexes passing through this transition have an inherent difficulty in displacing the 5' end of the RNA. The studies presented here show that initial bubble collapse is indeed important for initial RNA displacement. More specifically, any disturbance that limits (directly or indirectly) reannealing of the DNA at positions 1 through position +2 leads to a significant increase in 1113mer products. In a very recent study, a similar approach was used to investigate the role of the nontemplate strand in the transition from initiation to elongation (W. T. McAllister, Personal communication
). In that study, the nontemplate strand is extended to positions 5, 1, +6, +10, +15 and +22. Although mostly consistent with the data presented here, that study concludes that a complementary nontemplate strand is required as far downstream as 6 to 10 base pairs, In contrast, our results show clearly that extension downstream to only position +3 is necessary to achieve double stranded levels of 1113mer products. Previous studies suggest that promoter release begins as early as translocation to position +8, but may also occur at positions one or two bases farther on (22,23,26). Similarly, the results presented in Figure 5 demonstrate that the sink begins to show an effect on cycling with translocation to position +9, but does not exert its full challenge efficiency until translocation Page 20
Initial bubble collapse in promoter escape beyond position +12, suggesting that promoter release does not happen at a single translocational step. The timing of promoter release shown here is similar to the occurrence of initial bubble collapse observed in our earlier fluorescence studies (13). As measured by a decrease in the fluorescence signal from a fluorescent base analog at position 2 within the template DNA, the initially melted bubble starts to collapse on translocation to position +9. The fluorescence signal further decreases on translocation to position +10, and only on translocation to position +11 are the DNA bases at position 2 completely annealed. Taken together, this indicates that promoter release and bubble collapse are closely coupled and that the occurrences of these two events are most likely not homogenous (i.e., do not happen at a single translocational step). We suggest that a delay of either event in a complex can lead to some probability of improper RNA displacement, with a resulting higher production of 1113mer products. Recent results provide evidence to support this argument. Forced retention of promoter contacts, on a DNA construct that is otherwise normal, leads to a similar large increase in 1213mer products (Esposito & Martin, unpublished results). This suggests that the delay of promoter release possibly leads directly to a decrease in DNA reannealing, resulting in improper RNA displacement. In our previous work, the translocational timing of initial RNA displacement was determined by incorporating a fluorescent base analogue into the template DNA at position +1 or +2, while having either matched or mismatched nontemplate bases opposite fluorophore position (13). In complexes stalled at positions +10 through +12, results from the correctly paired constructs demonstrate that the template bases at positions +1 and +2 remain duplexed, in either an RNA:DNA hybrid or in a DNA:DNA duplex. Fluorescence from complexes with a single base mismatched opposite the fluorescent base analog shows that dissociation of the 5' end of Page 21
Initial bubble collapse in promoter escape the RNA occurs on translocation beyond position +10. Complexes stalled at position +11 show the beginning of RNA release at position +1, while complexes stalled at position +12 show initial peeling off of the RNA at position +2. The fluorescence indicates that RNA displacement is similarly a stepwise, non-homogeneous process. In the same study, we noted that the peeling away of the 5' end of the RNA, if driven by reannealing of the DNA duplex, might occur later in the constructs with mismatched bases at +1 or +2 in the nontemplate strand. Our results from constructs with mismatched bases at those same positions strongly support this argument (constructs MM[+1] and MM[+2]). With only one mismatched base in the critical region, much more of the complexes encounter difficulty in properly displacing the 5' end of the RNA, yielding more than twice the ratio of 1113mer to 1120mer products compared with the fully duplex construct. We conclude therefore that in normal transcription on fully duplex DNA, displacement of the 5' end of the RNA likely occurs earlier than was suggested in the fluorescence study. Cooperative collapse of the bubble. The initially melted bubble in the DNA extends from position 4 to about position +3. The results presented in the current study are consistent with a model in which the upstream edge of the bubble remains melted through translocation to position +8, extending the bubble downstream to about position +9. Subsequent translocation leads to a dramatic collapse of the upstream edge of the bubble. Bases at positions 4 to +1 or +2 could collapse all at once, in a cooperative fashion. Thus changes in base pairing strength at positions 4 through 1 are expected to have an effect on the strength of the collapse at positions +1 and +2. Indeed, mismatched bases at positions 4 and 3 or at positions 2 and 1 lead to a weakening of strand displacement and a resulting increase in 1113mer RNA products. Similarly, strengthening of the base pairs at positions 1 and 2 is expected to facilitate collapse- Page 22
Initial bubble collapse in promoter escape mediated strand displacement and leads to a decrease in 1113mer products. This is seen both in an otherwise native context (GC[2,1]) and in the context of mismatched bases at positions 4 and 3 (MM[4,3]GC[2,1]). Interestingly, although overall transcription from the GC[2,1] is decreased, the ratio of abortive products to full length transcript is lower than for the native promoter, suggesting an approach for transcribing RNAs from constructs that otherwise produce high levels of abortive transcripts. Why release of 11-13mer RNA transcripts? The model discussed above proposes that limitation of DNA bubble collapse leads to a larger portion of complexes poised at about positions +9 through +10 having improperly displaced RNA. The crystal structures of elongation complex models show the formation of an RNA exit channel in the protein in the elongation complex. We propose that complexes poised at translocational positions +9 to +10 are optimal for directing the displaced 5' end of the RNA into the channel. Complexes that do not displace the 5' end of the RNA at these normal translocational positions would be expected to have difficulty threading the 5' end into the channel, as the 5' end will have both spatially translated and rotated relative to that optimal positioning. Such complexes will then be limited sterically in further elongation as the hybrid encounters a protein block (residues 50-70) (19,20). The results presented here suggest that complexes that fail to displace can add at most 2-4 more bases (yielding 1113mer RNA products). This is directly analogous to the observation that elongation scaffolds impaired in collapse can extend about 3-5 bases beyond the normal 7-8 base pairs dictated by the hybrid (yielding a maximal, and presumably strained, 13 base pair hybrid) (30,31). Obviously, forward translocation of the polymerase is competitively limited by premature release of the transcript. When the RNA polymerase is stalled at positions +11 to +13 on a construct similar to PSS, turnover is substantially more rapid than for a similar stall on a Page 23
Initial bubble collapse in promoter escape fully duplex construct (unpublished results). This instability would presumably be similarly manifest during passage through this position during attempted runoff transcription. In any case, complexes that achieve proper RNA displacement at positions +9 to +10 then pass through position +13 normally and enter the processive elongation phase. Once past this transition, the complexes produce runoff transcripts with high efficiency. This is true both of the majority of complexes in normal transcription and of the smaller fraction of complexes in the bubble collapse-impaired constructs studied here. In a very recent study characterizing T7 RNA polymerase elongation complexes stalled at DNA lesions, an 8 nucleotide RNA primer was successfully extended to a 133 nucleotide runoff product using a promoter-independent scaffold approach (32). However, in this case, a fully complementary nontemplate DNA strand was added after assembly of a complex including the polymerase, template DNA and an 8mer RNA primer. As a result, upstream bubble collapse can occur, allowing proper displacement of the 5' end of the RNA primers and the generation of a bona fide elongation complex. Correlation with changes in enzyme structure. In T7 RNA polymerase, the transition from an initiation to an elongation complex involves a conformational change of the enzyme and a significant reorganization of the Nucleic Acids
within the protein as promoter contacts are lost and an RNA:DNA hybrid is formed (13,14,19-21,25). The crystal structure of the early initially transcribing complex suggests that the transition from initiation to elongation must start to happen in the earliest stages of transcription (21). In particular, steric clash between the growing hybrid and the RNA polymerase must occur on translocation beyond position +3. Mutagenesis studies show that the core subdomain (residues 72-151 and 206-257) and the thumb domain (residues 324-411) are probably involved in the conformational change in subsequent Page 24
Initial bubble collapse in promoter escape translocational steps (39,40). UV-photo crosslinking results suggest that contacts between the protein and the DNA at positions 5 and 9 are likely disturbed as the complex progresses to position +6 and +7 (23). These data suggest that the complex is accumulating conformational change stepwise toward the final elongation configuration. In a proposed model for the late initiation complex stalled at position +8, the conformation of the enzyme is that of the elongation complex (promoter binding region having both rotated and translated relative to its starting position), but the promoter has not yet been released and the initially melted bubble region remains held open by the enzyme (19). This model would then represent the moment just before initial bubble collapse and is mostly consistent with the results from biochemical studies. A more recent model suggests an initial spatial translation of the promoter binding region of the protein, without substantial rotation (Theis, et al., unpublished results). This model also allows for translocation to position +8 prior to promoter release and bubble collapse, and is also consistent with the results presented here. Summary. The results presented here demonstrate that bubble collapse and promoter release commence as the complex progresses beyond position +8, defining the beginning of the final stage of the transition process. DNA reannealing from the upstream edge of the bubble to position +3 helps to displace the 5' end of the nascent RNA. Those complexes that successfully complete this displacement proceed on to fully competent elongation complexes. For those complexes that do not successfully displace the 5' end of the RNA, translocation is limited to only a few nucleotides, leading to accumulation of 1113mer RNAs. DNA modifications that limit initial bubble collapse lead to fewer elongation competent complexes, while modifications that enhance initial bubble collapse lead to an increase in elongation competent complexes. This Page 25
Initial bubble collapse in promoter escape suggests that under normal transcription an intrinsic energetic barrier exists at this key transition in transcription. Materials and Methods protein expression
and purification. His-tagged wild type T7 RNA polymerase was prepared from E. coli strain BL21, carrying the plasmid pBH161 (kindly supplied by William T. McAllister), and purified using a Qiagen Ni-NTA column (41). Protein purity was determined by SDS-PAGE gel analysis. The purified protein was concentrated and dialyzed against 20mM KH2PO4-K2HPO4, 50% glycerol, 100 mM NaCl, 1mM Na2EDTA, pH 7.8, concentration was calculated from the measured absorbance at 280 nm using the molar extinction coefficient of 1.4105 M-1 cm-1, DTT was then added to a final concentration of 1.0 mM, and the sample was then stored at -20°C. Activity was determined by a transcription assay described below. Oligonucleotide synthesis and purification. DNA oligonucleotides were synthesized trityloff using an Applied Biosystems Expedite 8909 DNA synthesizer, purified by denaturing polyacrylamide gel electrophoresis, excised from the gel, and eluted using an Elu-Trap device (Schleicher and Schuell Inc., Keene, NH). Concentrations of single stranded DNAs were calculated using the weight sums of the three different measured molar extinction coefficients for each base at 253, 259, and 267nm (42). Single stranded DNAs were stored in TE buffer (10 mM Tris, pH 7.8, 1 mM EDTA) at -20°C. Partially single stranded or double stranded DNA construction. Nontemplate DNA and the relevant template DNA strands were combined in equal molar concentrations, heated to 75°C, and then cooled slowly (about 2 hrs) to room temperature
for annealing. Annealed DNA constructs were stored in TE buffer at -20°C. Page 26
Initial bubble collapse in promoter escape Transcription assays. Transcription reactions were carried out in a total volume of 16µL at 37°C for 5 minutes before being quenched with an equal volume of formamide stop solution (95% formamide, 40mM EDTA, 0.02% bromophenol blue). Equal molar concentrations of DNA construct and enzyme were used at final concentrations of 0.125µM in a reaction buffer containing 30mM HEPES (pH 7.8), 15mM magnesium acetate, 25mM potassium glutamate, 0.25mM EDTA, and 0.05% Tween 20. Reactions were initiated by addition of NTPs to a final concentration of 400 µM each, and labeled with [-32P]GTP. RNA transcripts were resolved by 20% polyacrylamide/7M urea gel electrophoresis, visualized and quantified on a Storm Phosphorimager (Molecular Dynamics). Sink challenge assays. In these studies, reactions were carried out as above for 1 minute before the addition of sink. Buffer conditions and concentrations of DNA and enzyme before the addition of sink were the same as described above. Reactions were initiated by the addition of GTP, ATP and CTP to a final concentrations of 400µM each. After allowing transcription (in the absence of radiolabel) for 1 min, 2µL TE buffer, containing 20µM sink DNA and [-32P]ATP was added. The reactions were quenched by addition of an equal volume of formamide stop solution 1 minute after the addition of sink. RNA transcripts were resolved, visualized, and quantified as described above. Page 27
Initial bubble collapse in promoter escape References 1. McAllister, W. (1993) Cell Mol Biol Res 39, 385-391 2. Strothkamp, R. E., Oakley, J. L., and Coleman, J. E. (1980) Biochemistry 19, 1074-1080 3. Muller, D. K., Martin, C.
T., and Coleman, J. E. (1989) Biochemistry 28, 3306-3313 4. Ъjvбri, A., and Martin, C. T. (1996) Biochemistry 35, 14574-14582 5. Cheetham, G. M., Jeruzalmi, D., and Steitz, T. A. (1999) Nature 399, 80-83 6. Milligan, J. F., Groebe, D. R., Witherell, G. W., and Uhlenbeck, O. C. (1987) Nucleic Acids Res
15, 8783-8798 7. Martin, C. T., Muller, D. K., and Coleman, J. E. (1988) Biochemistry 27, 3966-3974 8. Gralla, J. D., Carpousis, A. J., and Stefano, J. E. (1980) Biochemistry 19, 5864-5869 9. Carpousis, A. J., and Gralla, J. D. (1980) Biochemistry 19, 3245-3253 10. Tyagarajan, K., Monforte, J. A., and Hearst, J. E. (1991) Biochemistry 30, 10920-10924 11. Mentesana, P. E., Chin-Bow, S. T., Sousa, R., and McAllister, W. T. (2000) J Mol Biol 302, 1049-1062 12. Jiang, M., Rong, M., Martin, C., and McAllister, W. T. (2001) J Mol Biol 310, 509-522. 13. Liu, C., and Martin, C. T. (2002) J Biol Chem 277, 2725-2731 14. Ma, K., Temiakov, D., Jiang, M., Anikin, M., and McAllister, W. T. (2002) J Biol Chem 277, 43206-43215 15. Jia, Y., and Patel, S. S. (1997) Biochemistry 36, 4223-4232 Page 28
Initial bubble collapse in promoter escape 16. Dunn, J. J., and Studier, F. W. (1983) J Mol Biol 166, 477-535 17. Macdonald, L. E., Zhou, Y., and McAllister, W. T. (1993) J Mol Biol 232, 1030-1047 18. Macdonald, L., Durbin, R., Dunn, J., and McAllister, W. (1994) J Mol Biol 238, 145-158 19. Tahirov, T. H., Temiakov, D., Anikin, M., Patlan, V., McAllister, W. T., Vassylyev, D. G., and Yokoyama, S. (2002) Nature 420, 43-50 20. Yin, Y. W., and Steitz, T. A. (2002) Science 298, 1387-1395 21. Cheetham, G. M., and Steitz, T. A. (1999) Science 286, 2305-2309 22. Ikeda, R. A., and Richardson, C. C. (1986) Proc Natl Acad Sci U S A 83, 3614-3618 23. Place, C., Oddos, J., Buc, H., McAllister, W. T., and Buckle, M. (1999) Biochemistry 38, 4948-4957 24. Brieba, L. G., and Sousa, R. (2001) Biochemistry 40, 3882-3890. 25. Mukherjee, S., Brieba, L. G., and Sousa, R. (2002) Cell 110, 81-91 26. Brieba, L. G., and Sousa, R. (2001) Embo J 20, 6826-6835 27. Liu, C., and Martin, C. T. (2001) J Mol Biol 308, 465-475. 28. Huang, J., and Sousa, R. (2000) J Mol Biol 303, 347-358 29. Gopal, V., Brieba, L. G., Guajardo, R., McAllister, W. T., and Sousa, R. (1999) J Mol Biol 290, 411-431 30. Daube, S. S., and von Hippel, P. H. (1992) Science 258, 1320-1324 31. Temiakov, D., Anikin, M., and McAllister, W. T. (2002) J Biol Chem 277, 47035-47043 32. Jung, Y., and Lippard, S. J. (2003) J Biol Chem 278, 52084-52092 Page 29
Initial bubble collapse in promoter escape 33. Diaz, G. A., Rong, M., McAllister, W. T., and Durbin, R. K. (1996) Biochemistry 35, 10837-10843 34. Ъjvбri, A., and Martin, C. T. (1997) J Mol Biol 273, 775-781 35. Furuichi, Y. (1981) J Biol Chem 256, 483-493 36. Grachev, M. A., and Zaychikov, E. F. (1980) FEBS Lett 115, 23-26 37. Straney, D. C., and Crothers, D. M. (1987) J Mol Biol 193, 267-278 38. Cai, H., and Luse, D. S. (1987) Mol Cell Biol 7, 3371-3379. 39. He, B., Rong, M., Durbin, R. K., and McAllister, W. T. (1997) J Mol Biol 265, 275-288 40. Brieba, L. G., Gopal, V., and Sousa, R. (2001) J Biol Chem 276, 10306-10313. 41. He, B., Rong, M., Lyakhov, D., Gartenstein, H., Diaz, G., Castagna, R., McAllister, W. T., and Durbin, R. K. (1997) Protein Expr Purif 9, 142-151 42. Schick, C., and Martin, C. T. (1993) Biochemistry 32, 4275-4280 Page 30
P Gong, EA Esposito, CT Martin