Thermoirreversible and thermoreversible promoter opening by two Escherichia coli RNA polymerase holoenzymes.

Promoter opening, in which the complementary DNA strands separate around the transcriptional start site, is generally thermoreversible. An exceptional case of thermoirreversible opening of the T4 late promoter has been analyzed by KMnO4 footprinting and transcription. T4 late promoters, which consist of an 8-base pair (bp) TATA box "-10" element, are recognized by the small, phage-encoded, highly diverged sigma-family initiation subunit gp55. The T4 late promoter only opens above 15-20 degrees C, but once it has been formed remains open and transcriptionally active for days at -0.5 degrees C. The low temperature-trapped open complex and its isothermally formed state are shown to be structurally distinctive. Two "extended -10" sigma 70 promoters, which, like the T4 late promoter, lack "-35" sites, have been subjected to a comparative analysis: the T4 middle promoter PrIIB2 opens and closes thermoreversibly under conditions of basal and MotA- and AsiA-activated transcription. The open galP1 promoter complex, whose transcription bubble is very AT-rich, also closes reversibly upon shift to -0.5 degrees C, but more slowly than does the rIIB2 promoter. Formation of a trapped-open low temperature state of the promoter complex appears to be a singular property of gp55-RNA polymerase holoenzyme.

Initiation of transcription by RNA polymerase follows a reaction sequence that generates distinguishable intermediates. 1) RNA polymerase first locates the promoter as doublestranded DNA, forming a closed (transcriptionally incompetent) complex. 2) The closed promoter complex isomerizes to a second, still closed state. 3) This second complex in turn isomerizes to open ϳ15 bp of DNA extending from bp Ϫ11 beyond the transcriptional start site; the transcribed and non-transcribed DNA strands become widely separated upon promoter opening (1)(2)(3). In general, promoter complexes open with increasing temperature and close when the temperature is lowered; in hyperthermophilic bacteria, for example, promoter opening requires elevated temperatures. 4) In the presence of nucleotides (nt), 1 formation of initial transcripts comprising the first few internucleotide linkages generally follows rapidly. Initial steps of RNA synthesis are not processive; short aborted transcripts are produced at all promot-ers and can be overwhelmingly predominant at some. 5) RNA polymerase clears the promoter by means of processive RNA chain elongation.
The production of transcripts at different promoters can be controlled at any of these steps: Activity can be limited at the first step if RNA polymerase affinity for the promoter as duplex DNA is low. The equilibrium at step 2 can be unfavorable, and this step can also be kinetically limiting.
Step 3 is generally rapid, but the equilibrium can be unfavorable. The transition to the final step of processive transcript elongation and promoter clearance is limiting at certain promoters at which initial transcripts are repetitively but abortively produced. Individual equilibrium and rate constants associated with these steps are, of course, dependent on environmental variables. Depending on temperature, different steps can present the principal barrier to transcriptional activity at a single promoter (4 -22). Activators of bacterial transcription can exert their effects by intervening at any one, or at more than one, step. Some activators (such as CRP, the cAMP receptor protein) activate transcription at different promoters by affecting different steps (e.g. Refs. 10 and 23).
The reaction sequence that generates the open, initiationready promoter complex is adequately described by three linked equilibria shown in Reaction 1, E ϩ P -| 0 where E is the RNA polymerase holoenzyme; P is the promoter; C 1 is the initially forming closed promoter complex; C 2 is an isomerized but still closed intermediate; C o is the strandseparated open complex, with associated equilibrium constants K 1 , K 2 , and K 3 , and forward and reverse rate constants k 1 , k Ϫ1 etc. (see Refs. 9 and 18 for relatively recent reviews). The following complications are encountered in inferring structures of the C 1 and C 2 intermediate complexes: (i) RNA polymerase holoenzyme may contact both the Ϫ35 and Ϫ10 promoter elements or only the Ϫ35 site in the initial closed C 1 complex at different promoters. Such differences of structure must reflect differences of relative affinity for these two sites and of DNA flexibility in the intervening spacer (8,24).
(ii) Analyzing intermediate complexes by equilibrium or steady state methods requires that they adequately accumulate. They may do so fortuitously at some promoters under particular conditions (e.g. at a particular temperature; Ref. 7), but this is not necessarily the case. For example, Escherichia coli 70 -RNA polymerase holoenzyme complexes with the phage P R and lacUV5 promoters do not accumulate the C 2 complex as the predominant equilibrium or isothermal steady state component (5,20,24). Access to analysis of intermediates is also provided by rapid-kinetic pre-steady state methods (25) or by temperature downshift to transiently accumulate an intermediate component ahead of a (kinetically) slow step (4 -6, 8).
(iii) An analysis of the lacUV5 promoter that combines rapid kinetic and temperature shift approaches brings an interesting complication to light. Lowering the temperature of the lacUV5 open promoter complex from 37 to 14°C closes the promoter and generates a C 2 complex whose inferred structure distinguishes it from the C 2 complex that forms transiently during continuous incubation at 22°C (6). Finding that a C 2 complex can be (transiently) trapped in apparently different states at temperatures that differ by only 8°C implies that a kinetically defined C 2 promoter complex does not have to be structurally homogeneous and that the multiple molecular species comprising it may be only slowly interconvertible.
A distinctive example of irreversibility in promoter interactions has also been noted recently in an analysis of the mechanism of activation of the late genes of bacteriophage T4 (26). The promoters of these genes consist of a simple TATA box (consensus: TATAAATA) serving as an extended Ϫ10 site; there is no Ϫ35 site or required specific sequence upstream of the T4 late TATA box (27,28). Recognition of T4 late promoters is conferred on the host RNA polymerase core by gp55, the initiation subunit encoded by T4 gene 55. The small (185 amino acids) gp55 has been proposed as a distantly related member of the 70 family of transcription factors that lacks the protein domain that would interact with a Ϫ35 site (29,30). The gp55 segment thought to be related to 70 homology segment 2 does interact with E. coli RNA polymerase core (31) and mutagenesis identifies a gp55 segment related to 70 homology segment 2.2 as an RNA polymerase core binding site (32).
Activation of T4 late transcription by the T4-encoded sliding clamp replication-processivity factor requires the co-activator gp33 and depends on direct interactions of gp33 and gp55 with the sliding clamp (Refs. 33  The experiments that are described below pursue the further analysis of this apparently locked-in open promoter. The low temperature-open promoter complex is shown to be transcriptionally competent and extremely long-lived over a range of temperatures. Probing its stability with the competitively binding polyanion heparin and probing its transcription bubble with KMnO 4 indicates that the trapped, low temperature-open complex and the open complex formed isothermally at 30°C are structurally distinctive. These properties of the gp55 holoenzyme promoter complex have been placed in the context of the more studied 70 holoenzyme by a parallel analysis of open complexes at two "extended Ϫ10" 70 promoters. (These are promoters that also lack a significant Ϫ35 promoter element.) The properties of these two 70 holoenzyme open promoter complexes are shown to differ from those of the open T4 late promoter complex. 1) The T4 middle promoter P rIIB2 opens and closes thermoreversibly. 2) The galP1-35 promoter (36) also closes at 0°C once it has been opened at 30°C, but this extremely AT-rich and readily melting promoter (36,37,38,39) does so more slowly. The analysis suggests that the gp55-RNA polymerase holoenzyme differs from 70 -RNA polymerase holoenzyme in allowing the generation of a trapped-open state of its cognate promoters.

MATERIALS AND METHODS
Proteins and DNA-The overproduction and purification of proteins is described elsewhere (40,26); -(1-565) was a generous gift from K. Severinov. The 568 bp PCR fragment used for KMnO 4 footprinting of bacteriophage T4 late promoter (P 23 ) complexes was generated from placO-SK110-rrnB(T1ϩT2) (26). A 653-bp PCR fragment from pRT510-Cϩ18 (41), which contains a hybrid gp55-or 70 -dependent promoter, was used for analysis of transcription. In the presence of ATP, GTP, UTP, and UpG dinucleotide, this construct allows formation of an initial 18-nt transcript; a 427-nt transcript is made in the presence of all four NTPs. The 214-bp PCR fragment used for footprinting and transcription analysis of the bacteriophage T4 middle promoter (P rIIB2 ) was generated from plasmid pSTS416 (42). This template allows formation of a 120-nt run-off transcript. The 127-bp DNA fragment used for footprinting and transcription analysis of the galP1 promoter and the 115-bp fragment from galP1-35 were generated from plasmids pAA121 and its derivative with other sequence substituted for the UP-element, respectively; these two plasmids were kindly provided by S. Minchin and S. Busby (38,39). DNA sequences of these promoters are presented in Fig. 1. Ϫ10, extended Ϫ10, and Ϫ35 promoter elements are in bold type and underlined; an A/T-rich UP-like element is boxed. Two versions of the T4 gene 23 late promoter have been used: the wild type, and a derivative 70 -gp55 hybrid promoter created by introducing a consensus 70 -35 site (sequence of the hybrid promoter diverges from P 23 upstream of bp Ϫ29) and changing sequence downstream of the transcriptional start site to eliminate C residues from the first 17 nt of the T4 late transcript. The galP1 promoter has a mutation at bp Ϫ19 that eliminates the activity of an overlapping promoter (P2; Ref. 39). An upstream T/A-rich element is eliminated in galP1-35 (36). Both forms of galP1 are notable for a very T/A-rich transcription bubble (9 AT base pairs out of 10 between bp Ϫ7 and ϩ3).
Formation and Analysis of Promoter Open Complexes-Standard procedures for assembly of promoter complexes have been described (26). Promoter DNA (30 and 100 fmol for footprinting and transcription, respectively) and proteins (1 pmol of RNA polymerase core, 6 pmol of gp55 or 70 ) were mixed in a 15-l reaction buffer containing 200 mM potassium acetate, 33 mM Tris acetate (pH 7.8), 10 mM magnesium acetate, 150 g/ml bovine serum albumin, 1 mM dithiothreitol, 0.05% (w/v) Brij58, and 5% (w/v) polyethylene glycol (PEG)3300. RNA polymerase core enzyme from uninfected or T4-infected E. coli was used as specified for the individual experiments described below. Promoter complexes were formed for 60 min at 30°C, then shifted to low temperature for 30 min or for the time noted (Fig. 2) before further analysis. Maintaining temperatures not exceeding 0°C was assured by performing low temperature manipulations in a 3.5°C cold room, using a constant temperature bath set to Ϫ0.5°C.
KMnO 4 Footprinting-Reaction mixtures containing promoter complexes in a 15-l volume were mixed with 2 l of KMnO 4 (to 10 mM final concentration for analysis of T4 late promoter complexes and 6 mM for T4 middle and galP1 promoter complexes). Oxidation was stopped after 30 s by adding nine volumes of stop solution containing 200 mM ␤-mercaptoethanol. Samples were extracted with phenol/chloroform/isoamyl alcohol, DNA was precipitated with ethanol, dried, reacted with 10% (v/v) piperidine/1 mM EDTA at 90°C, re-extracted with phenol, and re-precipitated, as described (26). Equal quantities of radioactivity were loaded for resolution by denaturing PAGE, visualized by phosphorimaging and quantified, as described (26).
Transcription Reactions carried out at 30°C contained 100 g/ml heparin; heparin was omitted for transcription at Ϫ0.5°C. RNA synthesis was terminated 10 min after addition of nucleotides. For long time course experiments ( Fig. 6), samples were incubated for specified times at Ϫ0.5°C, then transferred to 30°C for 1 min prior to addition of NTPs and heparin for 10 min. Samples were processed, resolved on 23% polyacrylamide gel (37.5:1 acrylamide/bis), analyzed, and quantified by phosphorimaging. For RNA chain elongation (Fig. 5), the 18-mer initial transcript was formed in the presence of 100 M UpG, 5 M each of ATP, GTP, and [␣-32 P]UTP for 10 min at 30 or Ϫ0.5°C, all four unlabeled NTPs were added to final concentrations of 800 M, and the reaction was sampled at various times thereafter. RNA was prepared and analyzed as specified above, except that 12% polyacrylamide gel was used to resolve transcripts. For the experiment shown in Fig. 6C, the reaction buffer did not contain polyethylene glycol.
Run-off transcription initiating at P rIIB2 was carried out with 1 mM ATP, 1 mM GTP, 0.1 mM UTP, 0.1 mM [␣-32 P]CTP, and 100 g/ml heparin; heparin was omitted for transcription at Ϫ0.5°C; run-off transcripts were resolved on 6% polyacrylamide gel. Run-off transcription initiating at the galP1 promoters was carried out with 1 mM ATP, 1 mM UTP, 0.1 mM GTP, 0.1 mM [␣-32 P]CTP, and 100 g/ml heparin, but heparin was omitted for transcription at Ϫ0.5°C. Run-off transcripts were resolved on 10% polyacrylamide gel. For long time course experiments with the galP1-35 promoter (Fig. 8C), samples were incubated for appropriate times at Ϫ0.5°C, then transferred to 30°C for 0.5 min prior to addition of NTPs and heparin for 6 min.

RESULTS
Complexes formed by E. coli 70 -RNA polymerase holoenzyme at the lacUV5 and P R promoters melt a ϳ15-bp DNA segment thermoreversibly (5,9,18,20). This feature has been utilized to trap and analyze intermediate states of closed promoter complexes (6,8,21). In contrast, it was recently found that the bacteriophage T4 late promoter open complex formed by gp55-RNA polymerase holoenzyme remains open upon shift to lower temperature (26). Further analysis of the P 23 late promoter open complex by KMnO 4 footprinting reveals a stable open conformation that forms rapidly upon temperature shift from 30 to 0°C (Fig. 2). The low temperature state of the T4 late promoter open complex is characterized by overall lower KMnO 4 reactivity in the promoter bubble, together with nearly complete loss of reactivity of TϪ12 in the transcribed (template) strand (Fig. 2). The T4 late promoter open complex withstands challenge with heparin at 25°C or 30°C, but the low temperature-open complex dissipates rapidly in the presence of 100 g/ml heparin (Fig. 2).
The T4 late promoter opens slowly in the absence of its co-activator and sliding clamp activator at 30°C, under the conditions of these experiments (26). This makes it possible to use a temperature downshift/upshift sequence to determine whether the quantitative changes of KMnO 4 reactivity at Ϫ0.5°C seen in Fig. 2 reflect a partial dissociation of promoter complexes, and whether these low temperature complexes are trapped in an inactive form (Fig. 3). Open promoter complexes were allowed to form at 30°C for 60 min (lane 1), shifted to Ϫ0.5°C for 30 min (lane 2), then shifted back to 30°C for a time interval that is too brief to allow appreciable de novo formation of open complexes at this concentration of RNA polymerase holoenzyme (1 and 2 min, lanes 3 and 4, respectively) (26) and assayed at each step by treatment with KMnO 4 . Reactivity to permanganate was very rapidly and fully restored upon shift back from Ϫ0.5 to 30°C, indicating that the low temperature and "conventional" (30°C) states of the T4 late open promoter complex are in rapidly established equilibrium. In contrast, the low temperature-open complex is not formed upon continuous incubation at Ϫ0.5°C on the time scale of the experiment (up to 1.5 h; Fig. 2A, lane 9), or even in 48 h (as we show below).
The low temperature open complex is also transcriptionally competent. Promoter escape, detected by disappearance of the promoter transcription bubble, followed upon addition of NTP substrates (Fig. 4, A and B). The open complex held at Ϫ0.5°C for 30 min also fully retained the ability to make the appropriate 19-mer transcript at the low temperature when presented with UpG, ATP, GTP, UTP, and chain-terminating 3Ј O-methyl CTP (Fig. 4C, compare lanes 8, 9, and 10 with lane 1), although it lost this capacity rapidly when challenged with heparin before nucleotide addition (compare lanes 3-7 with lanes 8 -10), due to disruption of this low temperature open complex (Fig. 2). Indeed, the 18-mer initial transcript elongated at Ϫ0.5°C upon addition of CTP to the three other rNTPs, although elongation was quite slow (ϳ0.1-0.4 nt/s) at this low temperature (Fig. 5). For this experiment, open promoter complexes were formed at 30°C for 60 min, and shifted to Ϫ0.5°C for 30 min before adding UpG, ATP, GTP, and [␣-32 P]UTP for 10 min to make the 18-mer initial transcript (Fig. 5, lane 6). All four unlabeled NTPs were added (with 800 M UTP to dilute out the label), and samples were taken at the times noted in order to analyze the resulting first cycle of RNA chain elongation (lanes 7-11). As a control, the elongation process was also followed at 30°C on an appropriately shorter time scale (lanes 1-5). Clearance of the promoter from its ϩ18 start was at least as effective at Ϫ0.5°C as at 30°C (compare lanes 6 and 11 with lanes 1 and 2,  respectively). There were no prominent sites of transient pausing during elongation at Ϫ0.5°C (compare lane 11 with lane 2, for example).
Transcription assays were used to examine the barrier to reversible promoter closing on a considerably extended time scale and at different temperatures. Persistence of the open promoter complex after shift to low temperature was measured by ability to form a 19-mer initial transcript in the presence of heparin (Fig. 4C) after a very brief shift back to 30°C allowing thermal equilibration (Fig. 3) (Fig. 6C).
Most promoters that are recognized by 70 family initiation subunits have the characteristic Ϫ35 and Ϫ10 DNA sites. T4 late promoters are exceptional in lacking a Ϫ35 site, and gp55 is correspondingly exceptional in lacking any counterpart of the 70 family protein domain that interacts with the Ϫ35 site (structure domain 4 ; sequence homology segment 4.2) (43,29,30,44,45). A hybrid promoter for gp55-and 70 holoenzyme has been constructed by placing the Ϫ35 70 consensus site upstream of the T4 late promoter (41) (Fig. 1). The gp55 holoenzyme keeps this gp55/ 70 hybrid promoter open after temperature downshift from 30°C to 0°C, while the 70 holoenzyme opens and closes the same promoter thermoreversibly  1 and 2) transcription was started at 30°C by adding NTPs with (ϩ) or without (Ϫ) 100 g/ml heparin. All other samples were then shifted to Ϫ0.5°C for 30 min. For control samples shown in lanes 8 -10, NTPs were added after an additional interval noted above each lane. Heparin was added (to 100 g/ml) for samples shown in lanes 3-7, and NTPs were added at the same time (lane 3) or after the interval indicated above lanes 4 -7. R.M., recovery marker. between 30 and 0°C (26). It was previously suggested that the distinction between the properties of these two holoenzymes might be due to differences of interaction with the upstream promoter site. The next experiments explore this conjecture.
T4 middle promoters are members of the "extended Ϫ10" subclass of 70 promoters (consensus: TGXTATAAT; the TG extension of the Ϫ10 site is recognized by 70 homology segment 2.5). Middle promoters lack the 70 Ϫ35 recognition site but substitute a binding site centered around bp Ϫ30 for their activator, MotA. AsiA, the co-activator of T4 middle transcription, binds to 70 at two sites located in homology segments 4.1 and 4.2 (structure domain 4 ) and also to MotA (thereby dissociating dimeric MotA into the monomer). MotA additionally interacts with the very C terminus of 70 (46 -50).
The thermoreversibility of opening the T4 middle promoter P rIIB2 with 70 -RNA polymerase holoenzyme was examined under conditions of basal and MotA/AsiA-activated transcription. When the rIIB 2 open promoter complex was shifted from 30 to Ϫ0.5°C for 30 min, KMnO 4 footprinting revealed less than 10% of residual reactivity (Fig. 7A). AsiA, alone or together with MotA, did not affect the thermoreversibility of promoter opening (Fig. 7A). A transcription experiment showed that the previously opened rIIB 2 promoter closes within 30 min at 0°C (Fig. 7, C and D). E. coli RNA polymerase core is modified after phage T4 infection by ADP-ribosylation of its ␣-subunits in the C-terminal domain that interacts sequenceselectively with AT-rich DNA upstream of the core promoter. Binding to this UP element is eliminated by ADP-ribosylation of ␣ (at Arg-265). ADP-ribosylated (T4-modified) RNA polymerase was also seen to open and close the rIIB 2 promoter thermoreversibly (Fig. 7, C and D).
The effect of deleting region 4.2 of 70 on thermoreversibility was also examined. Basal (MotA-and AsiA-independent) tran- FIG. 5. Transcript elongation by ADP-ribosylated gp55 holoenzyme at 30°C and ؊0.5°C. Open complexes at the T4 hybrid promoter were formed at 30°C, as described for Fig. 4C, and split into two parts. One sample was shifted to Ϫ0.5°C and incubated for 30 min before adding UpG, ATP, GTP, and [␣-32 P]UTP for 10 min to allow formation of the initial 18-mer transcript at Ϫ0.5°C (lane 6) prior to chasing with all four unlabeled NTPs for the times indicated above lanes 7-11. The other sample, which remained at 30°C, was allowed to form the end-labeling 18-mer initial transcript (lane 1), and elongate that transcript with unlabeled NTPs for the times noted above lanes 2-5. M, DNA size markers. r.m., recovery marker. scription at P rIIB2 was dramatically decreased when 70 was replaced with 70 -(1-565). It has been shown that AsiA becomes a strong activator of T4 middle transcription in the context of the segment 4.2-deleted 70 -(1-565) (51). Promoter opening at 30°C was correspondingly rescued by AsiA and not further affected by MotA (Fig. 7B). Thermoreversibility was not qualitatively affected by the presence of AsiA or MotA (although it is conceivable that MotA has a small quantitative effect on residual promoter opening after shift to Ϫ0.5°C) (Fig.  7, A and B).
In summary, promoter opening at the "extended Ϫ10" T4 middle promoter P rIIB2 , which has no Ϫ35 promoter element, by E. coli 70 -RNA polymerase holoenzyme is thermoreversible. Eliminating residual nonspecific DNA interaction at the upstream side of the promoter complex by removing the 70 homology segment 4.2 or by ADP-ribosylation of the C-terminal DNA-binding domain of the RNA polymerase ␣-subunit does not make promoter opening irreversible. MotA and AsiA, the activator and co-activator of T4 middle transcription, do not qualitatively change the reversibility of promoter opening.
The analysis of extended Ϫ10 promoters also included the extensively analyzed galP1 promoter, which has a relatively strong UP element (37,36,38,52,39), and its derivative galP1-35, which removes that element by substituting other sequence upstream of bp Ϫ35 (38). The galP1 promoter opens at a relatively low temperature (39), and replacing the UP element further lowers the characteristic promoter opening temperature (38). Under the conditions of our experiments, the galP1 and galP1- 35 promoters did not open in 90 min at Ϫ0.5°C (Fig. 8A), but clearly remained open after 30 min at that low temperature once they had been opened for 60 min at 30°C. The distribution of permanganate reactivities at Tϩ2, Ϫ2, Ϫ3, Ϫ4, and Ϫ6 was seen to be approximately the same at 30°C and after shift to Ϫ0.5°C, thus giving no indication of partial promoter closure at the lower temperature (Fig. 8A), as is the case for the T4 late promoter, for which permanganate reactivity at TϪ12 was lost upon temperature downshift (Fig.  2). Indeed, the galP1 and galP1-35 open complexes retain a high proportion of activity in a single cycle of transcription after 30 min at Ϫ0.5°C (Fig. 8, B and D). A difference in the lengths of the ϳ45-nt run-off transcripts made at the two temperatures can be noted in Fig. 8B, and probably corresponds to a progressive shift previously noted for transcription at temperatures below 37°C (Fig. 5 of Ref. 39). Substituting the ADP-ribosylated T4-modified RNA polymerase core, which should generally decrease specific (UP element) as well as nonspecific DNA interactions of the two ␣-subunit C-terminal  Fig. 6). Open complexes were formed as specified above and assessed for retention of transcriptional competence (lane 1) after shift down to Ϫ0.5°C for the times shown (lanes 2-11). Control samples were tested for retention of polymerase activity during long term incubation at Ϫ0.5°C by shifting back to 30°C for 1 h before analyzing single-cycle transcription (lanes [12][13][14][15]. D, quantification of data in panel B. E, quantification of data in panel C. domains, had no effect on the retention of transcription activity after shift to Ϫ0.5°C for 30 min (Fig. 8, B and D).
The retention of strong transcription activity of the low temperature-open promoter complex is (somewhat) at variance with a prior observation that the galP1 promoter is transcriptionally inactive at low temperatures at which partial but substantial promoter opening is detected by KMnO 4 footprinting (39). A mobile equilibrium between transcriptionally competent open and inactive closed states of a promoter should be driven toward complete opening upon addition of nucleotide substrates, consistent with the results shown in Fig. 8, B and D. The contrary prior observations came from an experiment in which transcription initiated in the presence of heparin and in the presence of a relatively low concentration of UTP: UMP is the second nucleotide of the galP1 transcript and its effective K m may be temperature-dependent; the stability of open galP1 promoter complexes at low temperature to challenge by heparin was also not reported as having been directly examined in the prior work (but compare with Fig. 2). Because the relative transcript yield at Ϫ0.5°C was seen to diminish at low NTP concentrations (data not shown), high nucleotide concentrations were used for the experiment presented in Fig. 8B (in particular, 1 mM each of ATP and UTP, which are incorporated at the first 2 steps of RNA chain elongation).
The apparent trapping of the 70 holoenzyme complex with the galP1-35 promoter in a low temperature-open state was analyzed further at longer time scale and at different temperatures. Under the conditions of these experiments (see "Materials and Methods"), galP1 opened sufficiently slowly at 30°C to allow use of the experimental design shown in Fig. 6: promoter opening at Ϫ0.5, 6, and 10°C was assayed after shift to 30°C for 0.5 min (allowing temperature equilibration but barely any formation of new open complexes; data not shown); retention of polymerase activity during long incubations at these lower temperatures was assayed by shifting back to 30°C for 60 min (Fig. 8, C and E). The galP1- 35 promoter failed to open in 23 h at Ϫ0.5°C, but slowly and only partially opened at 6°C (in 23 h), and eventually opened almost completely at 10°C (in 23 h). In contrast to the pre-opened T4 late promoter, which remained open after temperature downshift and long term incubation (Fig. 6B), the pre-opened galP1-35 promoter closed, albeit slowly, after long term incubation at Ϫ0.5°C ( Heparin sensitivity can be a simple reflection of a dynamic equilibrium, in which this polyanion, which is at vast excess, traps RNA polymerase as it dissociates from the promoter complex (53). This cannot be the case here, since the low temperature-open state is stable for many hours, while heparin destroys the open complex at Ϫ0.5°C within minutes (Fig. 2B). Evidently, heparin at 100 g/ml (Fig. 2) and even at 50 or 25 g/ml (data not shown) actively displaces gp55 holoenzyme from the open promoter complex at this low temperature. 2) Its sensitivity to DNA oxidation by KMnO 4 is qualitatively as well as quantitatively distinctive, with reactivity of transcribed strand TϪ12 much lower at Ϫ0.5°C than at 30°C, both absolutely as well as relative to TϪ10 to Tϩ2. The complementary strands of the open transcription bubble are widely separated (1-3); loss of reactivity at TϪ12 could reflect closing of the transcription bubble at its upstream end at low temperature, or occlusion of the 5-6 double bond of TϪ12 by interaction with protein.
Thus, it is possible that the open promoter complex undergoes some structural change upon shift from 30°C to Ϫ0.5°C. However, if distinct states of the open complex exist at the higher and lower temperature, they are rapidly interconverted upon temperature downshift ( Fig. 2A) and upshift (Fig. 3A) under conditions in which the de novo formation of open complexes is quite slow. The low temperature-open complex is transcriptionally fully active at Ϫ0.5°C (Fig. 4C); elongation of RNA chains through a ϳ400-bp transcription unit is slow but relatively steady without prominent sites of pausing (Fig. 5).
How does this irreversibility of promoter opening arise, and where does the block to establishing equilibrium reside? In terms of the standard kinetic model stated in the introduction (Reaction 1): a) Is it in the forward direction of promoter opening, either in converting C 1 to the C 2 closed complex or in the C 2 3 C o promoter opening step? A kinetic block solely in the forward direction implies that the equilibrium state of the basal promoter complex at Ϫ0.5°C is open. However, the late promoter does not open in 48 h at Ϫ0.5°C or even in 23 h at 6 or 9°C, and only barely opens in 7 h at 15°C. It is overwhelmingly likely that the equilibrium state of the basal promoter complex at Ϫ0.5°C is closed. b) Does the kinetic barrier to establishment of equilibrium operate on the reverse direction? A block here might be generated by a very sharp temperaturedependence of conversion of the open complex to the C 2 closed complex. However, the late promoter also does not close reversibly at 6, 9, or 12°C. That makes it exceedingly unlikely that this is the simple explanation of irreversible retention of promoter opening at Ϫ0.5°C.
With evidence lacking for either of the preceding explanations, one is left with less simple alternatives. It appears that the open promoter complex may exist in multiple forms: we propose that the C 2 closed complex opens reversibly into an open complex C o (1) , which converts into a trapped-open form C o (2) . Our limited explorations have failed to uncover an "escape" pathway from the trapped-open C o (2) state by varying temperature. Removing the macromolecular crowding agent, PEG, which might stabilize otherwise mobile structural states of the promoter complex (54), from the reaction medium also does not eliminate trapping.
The T4 late promoter consists solely of an 8 bp-long "extended Ϫ10" element. Whether the lack of a Ϫ35 element is sufficient to generate thermoirreversible promoter opening has been explored by examining two "extended Ϫ10" 70 promoters that also lack Ϫ35 elements: the T4 middle promoter P rIIB2 and the galP1 promoter. Opening of the T4 rIIB 2 middle promoter is fully reversible between 30 and Ϫ0.5°C, regardless of whether the promoter functions in its basal mode or in conjunction with its MotA activator and AsiA co-activator (Fig. 7A). One might reason that 70 homology segment 4.2, which recognizes the Ϫ35 (promoter) site, must also have some nonspecific affinity for upstream DNA and could mediate reversible promoter closing, but removing segment 4.2 does not prevent promoter closing when the temperature is shifted down to Ϫ0.5°C (Fig. 7B). Similarly, blocking DNA binding of the ␣-CTD by ADP-ribosylation, a T4 phage infection-associated modification, does not eliminate reversibility of opening at the rIIB 2 promoter (Fig. 7, C and D).
The galP1 promoter was selected for comparative analysis because of its ability to open at low temperature in linear DNA (37,36,38,39): partial promoter opening has been noted at temperatures as low as 6°C, in some experiments (39). The transcription bubble of the galP1 promoter open complex is especially AT-rich, with 9 out of 10 A:T base pairs between Ϫ7 and ϩ3. Opening galP1 at the low temperature does not require the T/A-rich UP element-like cluster that is replaced in galP1-35 (Fig. 1). Conforming the Ϫ10 site to its consensus (TATGGT 3 TATAAT) or introducing a consensus Ϫ35 site (TTGACA) further lowers the promoter opening temperature (36,38), but removing the TG extension of the Ϫ10 site (TG 3 TT) and introducing a Ϫ35 site (to create the promoter designated galP con ) markedly increases the temperature at which the promoter opens (39). Two additional observations with this collection of promoter constructs motivate a direct comparison with the T4 late promoter. (a) The galP1 promoter was reported to be transcriptionally inactive or seriously reduced in activity at low temperatures (6 and 14°C, respectively) at which it was respectively partly or almost completely open. Promoter opening at very low temperature without concomitant transcriptional activity has also been reported to result from deleting a large internal segment of the RNA polymerase ␤-subunit (55). (b) An observation indicating apparently irreversible promoter opening was reported for the galP con promoter (a conventional 70 promoter with Ϫ35 and Ϫ10 sites and no TGx extension of the Ϫ10 site): after it has been opened at 37°C, this promoter retains transcriptional activity (for 30 min) after shift to a lower temperature at which it does not open (in 30 min; assayed with KMnO 4 ) (39).
In the reaction medium that allows a direct comparison with the T4 late promoter, galP1 and galP1- 35 do not open at Ϫ0.5°C, but remain open (for 30 min) after having first been opened at 30°C (Fig. 8A), and initially retain transcriptional activity at the low temperature (Fig. 8, B and D). However, qualitative as well as quantitative differences between the low temperature-open complexes at the T4 late and galP1 promoters are seen when the analysis is extended to longer time scales (Fig. 8, C and E). Given sufficient time, the galP1-35 promoter closes (Fig. 8E), just as the T4 rIIB 2 promoter does. GalP1-35 opens completely but slowly at 10°C, so it is not surprising to find the promoter remaining open at 10°C after downshift from 30°C; at 6°C galP1-35 opens and closes slowly (and is probably only partly open at equilibrium). These properties of the galP1 promoter are adequately explained in simple terms (Reaction 1, see the Introduction): temperature dependence of rates of opening and closing, and temperature dependence of the equilibrium state of this low temperature-melting promoter. Thus, in our view, the rIIB 2 -and galP1-extended Ϫ10 promoters do not differ mechanistically in regard to promoter opening and closing, although the T4 rIIB 2 promoter clearly closes much more rapidly than the gal promoter does at Ϫ0.5°C (possibly as a result of sequence and nucleotide composition differences between these two promoters in the DNA segment that opens into the transcription bubble).
It is only the open complex of the gp55 holoenzyme with the T4 late promoter that has been found to be capable of converting to a trapped-open state, but we suspect that this cannot be a property solely of gp55. The transcription bubble traverses the core ␤and ␤Ј-subunits, and strand separation at the downstream end of the transcription bubble is maintained by interaction with ␤and ␤Ј-subunits (2, 3). Three 70 promoters, one with a Ϫ35 site (26) and two without one (this work) open and close thermoreversibly. It is conceivable that 70 constrains polymerase core or DNA indirectly to prevent the structure change that traps a low temperature-open state. Screening of sigma and core mutants (in conjunction with a promoter that does not open at low temperature) may well yield further insights in this regard. It would also not surprise us if open promoter complexes with duplex DNA, even those that close reversibly at 0°C, could be generally trapped by rapid quenching to sufficiently low temperatures. This could prove useful for structure studies.