A Role for the Acidic Trimer Repeat Region of Transcription Factor aS4 in Setting the Rate and Temperature Dependence of Promoter Melting in Viuo*

u54 is an atypical u factor involved in enhancer-de- pendent transcription in Escherichia coli. In vivo assays were developed for following the kinetics and thermodynamics of dependent melting of the glnAP2 promoter start site. These assays were applied to a series of three uS4 polymerases containing zero, one, and two copies of the highly acidic trimer repeat region. The results showed that at least one copy of this acid region is required for a complete melting transition at physiological temperature, but is not required for a low temperature melting transition. The rate of melting the promoter in vivo increased with number of copies of this region. Taken together with other observations, the experiments point to a role for the acid region in triggering conformational changes that lead to specific promoter melting. The acid region is required for these changes to occur fully at physio- logical temperature and influences the rate at which they occur.

A Role for the Acidic Trimer Repeat Region of Transcription Factor aS4 in Setting the Rate and Temperature Dependence of Promoter Melting in Viuo* (Received for publication, June 16, 1992) Cai'ne WongS and Jay D. Grallag From the Department of Chemistry and Biochemistry and the Molecular Biology Institute, University of California, Los Angeles, California 90024-1569 u54 is an atypical u factor involved in enhancer-dependent transcription in Escherichia coli. In vivo assays were developed for following the kinetics and thermodynamics of dependent melting of the glnAP2 promoter start site. These assays were applied to a series of three uS4 polymerases containing zero, one, and two copies of the highly acidic trimer repeat region. The results showed that at least one copy of this acid region is required for a complete melting transition at physiological temperature, but is not required for a low temperature melting transition. The rate of melting the promoter in vivo increased with number of copies of this region. Taken together with other observations, the experiments point to a role for the acid region in triggering conformational changes that lead to specific promoter melting. The acid region is required for these changes to occur fully at physiological temperature and influences the rate at which they occur.
Escherichia coli RNA polymerase is a multisubunit enzyme consisting of an exchangeable u subunit and a core enzyme comprised of p, p', and two a subunits (Burgess et al., 1969).
The type of u factor associated with the core enzyme alters the sequence specificity of DNA binding and directs the enzyme to different classes of promoters (for review see Helmann and Chamberlin (1988) and Gross et al. (1992)). The major u protein in E. coli is u7', which produces most of the cellular mRNA. Most minor u proteins have related amino acid sequences and are thus considered to be members of the u70 family of proteins (Helmann and Chamberlin, 1988). Although the promoter sequences recognized by different members of this family are different, the overall pathway of transcription initiation appears to be similar (Gross et al., 1992).
There is one minor u factor, uW, that is not a member of the u70 family of proteins and has a distinct mechanism of transcription initiation. u54 associates with the common bacterial core enzyme and changes the sequence specificity of DNA binding in E. coli and in a number of other eubacteria (for review see Magasanik (1988); Kustu et al. (1989Kustu et al. ( , 1991, and Collado-Vides et al., (1991)). u54 was initially found as a *This research was supported in part by United States Public Health Service Grant GM35754. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
$Supported by National Institutes of Health Training Grant GM07185.
consequence of its involvement in mediating nitrogen assimilation but has since been found to mediate diverse metabolic pathways . Thus far, is the only identified bacterial u protein that appears not to be a member of the u70 protein family, as judged by amino acid sequence comparisons (Merrick et al., 1987;Gross et al., 1992). Additionally, the organization of u54 promoters differs from that of the d o promoters. The promoter elements recognized by u54 holoenzyme are centered at -12 and -24 in contrast to -10 and -35 which is common to all other u proteins. The activators of u54 transcription generally bind the DNA at some distance from the polymerase. Moreover, these activator binding sites are often enhancer-like in that their precise location and orientation is not critical (Buck et al., 1986;Reitzer and Magasanik, 1986;Collado-Vides et al., 1991;Kustu et al., 1991).
The mechanism of transcription initiation by polymerase at the glnAP2 promoter has been studied and found to differ substantially from that of promoters transcribed by d o polymerase. The promoter is bound by the u54 polymerase to form a closed complex in a process (Sasse-Dwight and Gralla, 1988;Popham et al., 1989) that does not require the DNAbinding protein, NTRC (NR1). Phosphorylation of NTRC in response to nitrogen deprivation leads to a DNA melting event within the closed complex that is required for transcription initiation . This event involves looping out of the intervening DNA between the activator and promoter sites, leading to what appears to be proteinprotein contacts, and also requires ATP hydrolysis (Su et al., 1990;Weiss et al., 1991). This is very different from the d otype mechanism of initiation and resembles eukaryotic polymerase I1 transcription in the involvement of remote activators and the requirement for ATP hydrolysis (Gralla, 1991;Wang et al., 1992).
Previously, Sasse-Dwight and Gralla (1990) attempted to identify functional domains within u54 by assaying the functions of proteins expressed from u~~ genes containing internal deletions. Two regions containing leucine-rich heptad repeats (Fig. l ) , which resemble leucine zippers, were shown to be involved in DNA binding at the -12 promoter element. Deletions within a potential helix-turn-helix motif near the carboxyl terminus eliminated all DNA binding, suggesting its involvement in -24 promoter element binding. Deletion of a 27-amino acid stretch within a very highly acidic domain ( Fig.  1) did not alter DNA binding but significantly reduced the ability of the mutant u54 holoenzyme to form open complexes at the start site of transcription. In this report we investigate the role of this acidic region in promoter melting.
Although acidic regions have not previously been implicated in the control of bacterial transcription, they have a well FIG. 1. Schematic diagram of structural and functional domains of u54 protein. This figure was adapted from Sasse-Dwight and Gralla (1990). T h e u64 protein is divided into three regions based on homology among sequenced cK4s (Merrick et al., 1987). Regions I and 111 have high homology, whereas Region I1 has low homology. T h e acidic trimer repeats region (ATR) is located within Region 11, the region of low homology but highly conserved in overall acidity. Flanking Region I1 are two leucine-rich heptad repeats which resemble leucine zippers and are involved in -12 promoter element recognition. FIG. 2. a, amino acid sequence of the acidic trimer repeats (ATR). T h e upper sequence is the 42 amino acids (in standard single letter code) encoded by the 126-base pair AlwNI-AluI fragment (see "Construction of pATuK4" under "Materials and Methods"). The brackets above and below the sequence indicate the acidic amino acids every third residue that make up the ATR. The bar above the sequence indicates amino acids deleted in the Del 9 us' protein. The arrow in the lower sequence shows where the amino acids of the AlwNII-AluI fragment are inserted into u5' to yield a mutant with two ATR regions in tandem. The bracket below indicates the original copy of the ATR region. The alanine residue deleted during the cloning procedure is indicated by the dash beneath the arrowhead. b, cartoon of the three e5's that differ primarily in ATR copy number used in this study. Del 9 uK4 is a uK4 mutant in which the ATR region was deleted. A T U~~ is a c64 mutant which has a second copy of the ATR region inserted 8 amino acids away from the existing ATR. known involvement in polymerase 11-dependent eukaryotic transcription (Ptashne and Gann, 1990). Experiments have suggested an important role of acidity per se, although acidic regions can contain other poorly understood but important elements (Cress and Triezenberg, 1991). In the E. coli, Klebsiella pneumoniae, and Salmonella typhimurium u54 proteins, amino acids 48-81 form a subregion of a high amino acid homology and negative charge density where (with one exception) every third amino acid is an acidic residue (see Fig. 2a). We have termed this motif the acidic trimer repeats (ATR)' The abbreviations used are: ATR, acidic trimer repeats; IPTG, high low high region which was deleted in the u5' mutant, AcDel 9 u5', studied earlier (Sasse-Dwight and Gralla, 1990). In this report we will study a homologous series of u54 s that differ primarily in that they contain either none, one, or two ATR regions (Fig. 2b). We expect that if the ATR region is a functional module whose role is to primarily provide an acidic domain, then there should be a progression of influence on DNA melting despite the grossly different sizes and possibly altered structures of the proteins. New in vivo assays for the rate and temperature dependence of open complex formation were developed and applied to this series. The results show that at least one ATR region is required for a physiological temperature promoter melting transition and that increasing the number of ATR regions increases the rate at which the promoter can be melted.

E. coli Strain, Plasmids, and in Vivo Test Promoter
Mutations were introduced into (rs4 through an rpoN gene cloned behind a tac promoter on a pBR322-derived plasmid, pTH7, originally supplied by Dr. Boris Magasanik (MIT). These plasmids, described in greater detail below, were used to transform YMClO9tk (thi, endA, hsr, DlacU169, rpoN::TnlO/F' pro l a c P ZU118,Tn5-102), a c5' minus E. coli host constructed by M. Hsieh and S. Sasse-Dwight in this laboratory. The dependent test promoter is the chromosomal glnAP2. In the YMClO9tk strain there are two chromosomal copies of the glnAP2 promoter. There is the original which controls transcription of the gln ALG operon and a glnAP2-lacZ fusion borne as a X lysogen (XglnlOl) (Blackman et al., 1981).
It was possible to directionally ligate the 126-base pair AlwNI-AluI fragment into the proper pTH7 AlwNI terminal, since the two AlwNI sites are both noncompatible and nonsymmetric. The nonligated AlwNI terminal was blunt-ended with T 4 DNA polymerase I (which deleted an alanine residue between the two ATR) and subsequently blunt endligated to the A M terminal. This DNA was used to transform the u54 minus host YMClOStk. The transformants were selected by ampicillin resistance on minimal media agar plates which had arginine as the only nitrogen source (Sasse-Dwight and Gralla, 1990). In order to grow efficiently on this media, transformants had to have a functional n5'. The insertion and correct orientation of the AlwNI-AluI fragment into pTH7 was verified by restriction fragment size isopropyl-1-thio-P-D-galactopyranoside; AT, uK4 protein containing tandem ATR. and nucleotide sequence analysis. The resultant u54 mutant contains the original and a tandem copy of the ATR region separated by 8 amino acids (see Fig. 2b).
Construction of pAcDel 9 0'' was reported previously (Sasse-Dwight and Gralla, 1990). In this study we have shortened the AcDel 9 name to Del 9.
Determination of u54 Protein Expression in Vivo-YMClO9tk strain cells bearing encoded plasmids were diluted 1:lOO from saturated overnight cultures and grown to 0.3-0.4 OD,,, units in 5 ml of G/gln media (60 mM KzHP04, 33 mM KH2P04, 0.45 mM MgS04, 0.1 mM thiamine, 22.2 mM glucose, 13 mM Lglutamine, 100 pg/ml tetracycline, 100 pg/ml kanamycin, 100 pg/ml ampicillin) in a 37 "C Gyrotory Shaker water bath (New Bmnswick Instruments, Edison, NJ). Cells were pelleted at 5000 rpm for 5 min in an SS34 rotor at room temperature (Sorvall, Du Pont Instruments, Wilmington, DE). The cell pellet was then resuspended in 2 ml of fresh SO:--free G/gln media and pelleted again to wash the cells of residual free SO:-. The cells were then resuspended in 2 ml of fresh SO:--free G/gln media, to which was added 2 p1 of 0.5 M IPTG and 40 pCi of Na3'S04 (Amersham Corp.), and incubated at 37 "C for 15 min. The cells were harvested by centrifugation and washed in 1 ml of fresh SO:--free G/gln media at 4 "C. The pellet was resuspended in 0.5 ml of cold sonication buffer (0.05 M Tris, 5% glycerol, 2 mM EDTA, 0.1 mM dithiothreitol, 1 mM 0-mercaptoethanol, 0.233 M NaCl, 23 pg/ml phenylmethylsulfonyl fluoride, pH 8) and freezethawed twice in liquid nitrogen and an ice water bath. To lyse any remaining whole cells, the lysate was sonicated on ice with two 10-5 bursts from a 4-mm tip mounted on a model 300 Sonic Dismembrator set at 35% (Fisher Scientific, Tustin, CA). Cell debris was removed by microcentrifugation at 13,000 X g for 20 min at 4 "C (Eppendorf Brinkmann Instruments, Westbury, NJ). 3 pl of the cell lysates were resolved on an 8% SDS-polyacrylamide gel electrophoresis (Laemmli, 1970) minigel (Bio-Rad). The gel was stained with Coomassie Brilliant Blue R-250, dried, and an autoradiograph was made. Identification of the various u54 protein bands were made by comparison to a lysate in which no IPTG was added to induce wild-type u54.

In Vivo Dimethyl Sulfate Footprinting
The protocol used was essentially that reported by Sasse-Dwight and Gralla (1988). Rifampicin was used to pretreat these samples in order to trap the u '~ RNA polymerase at the glnAP2 promoter.

Open Complex Formation Rate Assay
The protocol for the open complex formation rate assay was an adaptation of the KMn04 footprinting procedure reported by Sasse-Dwight and Gralla (1988Gralla ( , 1990. Cells harboring mutant and wildtype encoded plasmids were grown in B shaking water bath at 37 "C to 0.3-0.4 ODao units in 100 ml G/gln media from a 1:lOO dilution of a saturated overnight culture. 100 p1 of 0.5 M IPTG was added to the log phase cells and allowed to grow for an additional 60 min. The 100-ml culture was divided into eight 10-ml aliquots and placed back into the 37 "C water bath. Except for the no rifampicin sample, 40 pl of rifampicin (Sigma) at 50 mg/ml methanol was added to each culture at time 0. 270 pl of 0.37 M KMIIO)~ was added to different aliquots at 0.0, 0.16, 0.5, 1.0, 2.0, 4.0, and 6.0 min after rifampicin addition. The KMn04 was allowed to react with the cells for 2 min. The cells were harvested by transferring to chilled glass centrifuge tubes and centrifuged 5 min at 5000 rpm in an SS34 rotor at 4 'C. The supernatant was removed and the cell pellet resuspended in 1 ml of cold SE buffer (150 mM NaC1,lOO mM EDTA, pH 8.0) and pelleted again. The resulting pellet was resuspended in 475 pl of SET buffer (150 mM NaC1, 15 mM EDTA, 60 mM Tris, pH 8.3) plus 5.0 pl of 10 mg/ml heat-inactivated RNase A (Sigma) and 20 pl of 25% SDS, pH 7.0. The lysed cells were then incubated at 37 "C for 30 min. 10 pl of 10 mg/ml proteinase K (Sigma) was added to the cell lysate and incubated overnight at 37 "C. The resulting proteinase K digest was extracted three times with 700 p1 of chloroform, each time separating the aqueous phase from the chloroform layer by microcentrifugation at 4 "C for 5-10 min. The lowered temperature aids in removing SDS. The chloroform extractions are followed by a single extraction with 700 pl of phenol, pH 7.0, and a final chloroform extraction. The chromosomal DNA was then precipitated by adding 200 pl of 7.5 M ammonium acetate and 300 pl of cold 2-propanol. The DNA pellet was resuspended in 150 p1 of water and heated at 60 "C and periodically agitated until the DNA pellet was no longer visible. The DNA was passed through a 1-ml G-50-80 Sephadex (Sigma) spin column equilibrated with water. 50% of the spin column eluent was used in 32P-end-labeled primer extension through the g l d P 2 promoter. Primer extension was performed as follows. The chromosomal DNA sample volume was brought to 85 pl with water. To the DNA was added 10 pl of 10 X Taq polymerase buffer (Promega), 4 pl of 5 mM dNTPs (Pharmacia LKB Biotechnology Inc.), 1 unit of Taq DNA polymerase (Promega), and 1 pl of 32P-5'-end-labeled primer (1-5 X lo6 cpmlpl), which hybridized to the coding strand between +54 and +38. A 100-pl mineral oil overlay was added to reduce evaporation. The primer extension was cycled 17 times with a thermal cycler (MJ Research, Watertown, MA). The samples were denatured at 94 "C for 1.5 min for the first cycle and for 1.0 min for all subsequent cycles. Hybridization of the primer was at 57 "C for 2.0 min for all cycles. Elongation was at 72 "C for 3.0 min except for the last cycle which was extended to 10.0 min. The mineral oil overlay was removed, and the primer-extended product was extracted with 100 pl of chloroform and precipitated with 30 pl of 4.0 M ammonium acetate, 20 mM EDTA, and 200 pl of cold 95% ethanol. The DNA pellet was washed with 750 pl of 70% ethanol, dried, and resuspended in 5 pl of DNA sequencing gel sample buffer. The samples were resolved on 6% DNA sequencing gels which were dried and autoradiographed. To quantify the open complexes, the radioactive bands of the dried gel, corresponding to the glnAP2 open complex, were excised with a razor blade and counted for 5 min in a Beckman LS-3133T scintillation counter (Beckman Instruments).
I n Vivo Temperature-dependent DNA Melting-The protocol for determining temperature-dependent promoter DNA melting in vivo is also an adaptation of the KMn04 footprinting procedure reported by Sasse-Dwight and Gralla (1988Gralla ( ,1990. Cells harboring the mutants and wild-type rpoN (u")-encoded plasmids were grown separately in a shaking water bath at 37 "C to 0.4-0.5 ODm units in 50 ml of G/ gln media from a 1:lOO dilution of a saturated overnight culture. 50 pl of 0.5 M IPTG was added to the log phase cells, which were allowed to grow for an additional 30 min. The 50-ml cultures were divided into four 10-ml samples. These were then maintained at either 0, 16, 23, or 37 "C for 30 min. After the 30-min equilibration, 40 pl of rifampicin in 50 mg/ml methanol was added to each of the four samples. After treating the cells with rifampicin for 5 min, 270 pl of 0.37 M KMn04 was added to all aliquots and allowed to react with the cells for 2 min. These KMn04-treated samples were then processed using the same protocol used in the open complex formation rate assay.

Mutant Proteins Are
Expressed to Wild-type Lmek-Previous experiments used plating assays to indicate that mutant forms of 051 were expressed (Sasse-Dwight and Gralla, 1990). Now we wished to determine whether the two 051 proteins with changes in the ATR region were present at levels comparable with that of the wild-type protein. Total cellular protein was pulse-labeled with [35S]sulfate in cells harboring either the mutants or wild-type rpoN (~7~~) gene. Equal amounts of labeled soluble cell extracts were resolved by SDS-polyacrylamide gel electrophoresis. u61 could not be seen by Coomassie staining (not shown) but could be seen by autoradiography (Fig. 3). The u51 band is identified by comparison with the lysate to which no IPTG was added to induce wild-type ( l a n e 1 ) . As expected, the deletion in the gene increases the mobility of this protein (lane 4 ) and the insertion decreases the mobility (lane 3 ) . However, the mobility shifts correspond poorly to the predicted molecular weight shifts (approximately +4600 for ATu'~ and -3000 for Del 9 u51), most likely due to the highly charged nature of the ATR region. The autoradiograph in Fig. 3 shows that the two mutant C T~~ s (see arrowheads in lanes 3 and 4 ) exhibit approximately the same band intensity as wild-type (lane 2). The u51 band intensity is not influenced by the ATR region copy number, since the ATR region does not have any sulfurcontaining amino acids. Overall, the experiment shows that all three proteins are produced in approximately equal amounts.
The Mutant ATu51 Directs RNA Polymerase to Bind and Melt the glnAP2 Promoter-Previously, it was shown (Sasse- Dwight and Gralla, 1990) that AcDel 9 u54 (called Del 9 in this study), which is missing the ATR region, was capable of directing binding to the glnAP2 promoter i n vivo. We used dimethyl sulfate protection footprinting to determine if A T u~~, which has two copies of the ATR region in tandem, also directs RNA polymerase to bind to the glnAP2 promoter. Previous dimethyl sulfate footprinting studies of the glnAP2 promoter (Sasse-Dwight and Gralla, 1988Gralla, , 1990 have shown that the critical guanines at -13 and -25 are protected by RNA polymerase when a functional us4 is present in the cell. The assay involves expressing the mutant us4 in cells lacking the wild-type u54, trapping the RNA polymerase at the promoter, followed by modifying the DNA with dimethyl sulfate. The sites of attack are then determined by isolating the DNA and cutting it at the methylated sites with piperidine followed by primer extension analysis. Lanes 1-3 of Fig. 4 show the dimethyl sulfate cleavage pattern using AT, wild-type, and Del 9 u54 proteins, respectively. The patterns are essentially the same, indicating that the changes in the acid region have not altered the ability to bind the promoter. These three patterns differ from that seen when no u54 is provided ( l u n e 4 ) . The critical guanine a t position -25 is fully protected when any of the three forms of u54 are present, demonstrating full polymerase occupancy. The -13 position is also protected in all three cases, although the extent of protection is not as strong as reported previously when a less activating medium was used (Sasse-Dwight and Gralla, 1990). We interpret these patterns to mean that the acid region acts as a largely independent module in the sense that even a large insertion or a large deletion does not prevent the remainder of the protein from accomplishing its DNA binding function. The equal binding of mutants and wild-type indicates that there is no significant loss of binding affinity in the mutant transcription complexes. This binding relies on the association of core with u54 , and thus the core binding function also appears to be fully intact in these mutants.
Previously, it was proposed (Sasse-Dwight and Gralla, 1990) that the acidic region acted as an essential module used by the bound polymerase to melt the promoter DNA. This Sigma 54 AT Wt Del9 None

FIG. 4. Dimethyl sulfate footprint of AT, wild-type ( W t ) ,
and Del 9 u6' RNA polymerase at the glnAP2 promoter. These data show that the mutant U"S allow RNA polymerase binding to the glnAP2 promoter. Critical guanines a t -13 and -25 are not protected from methylation at the glnAP2 promoter when u5' is absent (lane 4 ) . Lanes I and 3 show that ATa" and Del 9 u5' RNA polymerases bind to the glnAP2 promoter a t least as well if not better than wildtype us' (lane 2 ) , since the critical guanines are protected a t least as well if not better by the mutant proteins. was suggested by the diminished melting directed by the bound Del 9 u54 polymerase. Among the complications in this interpretation is that the deletion is a large perturbation. The AT us4 also represents a large perturbation of the same region but one that increases rather than decreases the total acidity. To test whether the AT u54 polymerase could melt the DNA, we used KMn04 footprinting i n vivo. Fig. 5a shows the assay of KMn04 hypersensitivity associated with glnAP2 start site melting. Lanes 6 of panels B and C are repetitions of a previous result showing diminished melting in Del 9 compared with wild type (Sasse-Dwight and Gralla, 1990). Lane 6 of panel A shows that start site melting is not diminished using the mutant with tandem copies of the ATR.

The Acidic Trimer Repeat Region Affects the Rate of Open
Complex Formation i n Vivo-Previous studies on many promoters have focused on the rate at which open complexes form, because this rate may limit transcription. These experiments have been restricted to in vitro conditions due to the lack of a technique to make the measurements in vivo. In order to learn whether the ATR region plays a role in setting this critical rate, we developed a method for measuring this rate in vivo.
The method is an adaptation of the i n vivo KMn04 footprinting protocol (Sasse-Dwight and Gralla, 1988). KMn04 preferentially oxidizes non-base-paired thymine residues and has been used in vivo to detect open complexes in several contexts (Morett and Buck, 1989;Kassavetis et al., 1990;Wang et al., 1992). At the glnAP2 promoter there is little KMn04 reactivity at the start site during steady-state transcription. This is because promoter melting is transient. The open complex is rapidly cleared by initiation, and the start site rapidly recloses. However, if the cells are pretreated with rifampicin, KMn04 detects substantial start site opening at the glnAP2 promoter (Sasse-Dwight and Gralla, 1988). This is because rifampicin binds the p subunit of RNA polymerase and prevents the bound polymerase from entering an elongation mode (Sippel and Hartmann, 1968;Wehrli et al., 1976;Carpousis and Gralla, 1985). Thus the open complexes can form normally, but the polymerase within them is trapped.   Gralla, 1988;this study: Fig. 5a, panel B, lane N ) . As just discussed, this is because open complexes initiate transcription rapidly and thus are removed almost as soon as they form. Immediately after rifampicin is added, KMn0, sensitivity begins to appear, since elongation of open complexes is blocked, leading to their accumulation. The KMn04 signal increases with time and apparently plateaus after approximately 1 min. The rate of appearance of the KMn0, signal is taken as a measure of the rate at which open complexes form in vivo.

Acidic Domains and Promoter D N A Melting
These data were analyzed quantitatively and the result is shown in Fig. 5b (filled square symbols). The tIl2 for open complex formation for wild-type uB4 is approximately 0.7 min under these i n vivo experimental conditions. The tl12 i n vitro has not been measured at this promoter, but other promoters exhibit t1/2 that range up to several minutes (McClure, 1985). Thus the assay yields an in vivo result that is within a comparable range.
This experiment was also performed using cells producing the mutant forms of uS4 (Fig. 5a, panels A and C). Qualitatively, the results show that in AT, when two ATR regions are present, open complexes accumulate faster than with wildtype us,, which has one ATR region (compare panels A and B ) . When the ATR region is missing (Del 9 us,), open complexes accumulate more slowly, and as noted previously, even with long rifampicin treatment, fewer open complexes accumulate (Fig. 5, panel C).
The results with the mutant proteins were also analyzed quantitatively and are compared to those from wild-type (Fig.  5b). The saturation levels for AT and w t are roughly comparable, although the ATuS4 melting might be slightly higher. In terms of the half-time for open complex formation, the mutant with two ATR regions is 2-3 times faster than wild-type, with a tlI2 of approximately 0.25 minutes. The differences are seen primarily at the earliest times which are the only ones that clearly precede the attainment of a plateau.
The mutant without an ATR region behaves differently in this analysis. The KMn04 signal reaches a maximum after approximately 1 min and then slowly declines. The results differ from those of the other us4 proteins in two ways. First, the KMn04 signal is always 50-70% less. Second, only in this case does the signal reproducibly decline after reaching a maximum. We do not know the source of this decline for the mutant lacking the ATR region, but one possibility is that the Del 9-mediated open complexes are not stable and are not easily re-formed even with lengthy rifampicin treatment. In any case, the decline at the longer times complicates the quantitative analysis, since it is difficult to determine a t l l 2 when the saturation level is not known accurately.
In order to compare the Del 9 data to that from the other us4 proteins, we compared the absolute initial rates in the three cases. This used the first four data points to estimate the initial rate of open complex accumulation. In this analysis the lack of an ATR region causes an approxiamtely 8-fold decrease in the number of open complexes that accumulate in the first minute, before the decline sets in. The presence of an extra ATR region causes an approximately 2.5-fold increase. We conclude that the number of ATR regions present determines at least in part how fast an open complex forms at the glnAP2 promoter i n vivo. At Least One A T R Region Is Required for a Physiological, but Not a Low Temperature Melting Transition in Vivo-The previous experiments have shown that the ATR region influences the rate at which the DNA is melted within the transcription complex. In addition, the lack of an ATR region leads to a lowering of the amount of open complex that can ultimately be formed. The ability of RNA polymerase to melt a promoter depends on the temperature, presumably reflecting in part the temperature-dependent melting of naked DNA. Such open complex melting curves have been obtained in vitro and have been shown to depend on the promoter DNA sequence (Grimes et al., 1991). In this study, we wished to learn if promoter melting curves could also be influenced by the amino acid sequence of the polymerase, particularly the ATR region implicated in melting. To do this, the KMn04 footprinting assay was applied to cells at different temperatures.
In this experiment, cell cultures are grown to mid-late log phase at 37 "C. u54 protein is induced with IPTG, and samples are removed and equilibrated a t several lower temperatures for 30 min. Each sample is then pretreated with rifampicin, and promoter opening is assessed with the standard KMn04 procedure.
Results obtained for wild-type uS4 a t four different temperatures are shown in Fig. 6a, lanes 5-8. As the temperature is lowered from 37 to 23 "C, there is a modest decrease in the KMn04 signal (compare lanes 7 and 8). However, when the temperature is lowered a few more degrees to 16 "C, there is a large reduction in the KMn04 signal (lane 6). A large reduction in signal is also seen as the temperature is lowered further to 0 "C (lane 5). The KMn04-sensitive bands were excised from the gel and counted. Fig. 6b shows the data plotted as a conventional DNA melting curve.
The results (Fig. 6b, filled squares) resemble known melting curves obtained with naked DNA and with polymerase-dependent open complexes in vitro (Grimes et al., 1991). The apparent melting temperature is approximately 15 "C. This value is between the values of 9 and 25 "C determined by permanganate probing of two other promoters in vitro (Grimes et al., 1991). All of these values are much lower than the melting temperature of naked DNA since these are enzyme catalyzed reactions. This experiment has not been done in vitro for the glnAP2 promoter, so it is not yet possible to compare in vitro with in vivo results at the same promoter. This experiment was also performed with the two u54 mutants, and the results are shown in Fig. 6a, lanes 1-4 (AT) and 9-12 (Del 9). The KMn04 signals were quantified and are plotted on Fig. 6b. The melting curves for AT and wildtype forms of are indistinguishable. However, for Del 9, the melting behavior is similar only at low temperatures. In fact, at 0 and 16 "C, all three proteins seem to catalyze identical extents of melting. However, above 16 "C the melting transition that is common to wild type and AT is simply absent for Del 9. This transition is nonlinear and is reminiscent of the cooperative melting behavior of DNA. We infer that needs a t least one copy of the ATR region to facilitate full promoter melting a t physiological temperatures. Promoter melting at temperatures below 16 "C is partial and occurs independently of the ATR region.

DISCUSSION
In these experiments, the acidic trimer repeats of uS4 seem to act as a required melting module. Neither a duplication nor a deletion of this region hinders the other critical functions of u54, namely binding to the -12 DNA promoter element, binding to the -24 promoter element, and by inference, the binding to core polymerase. Thus large changes in this region are tolerated with respect to those functions. On the other hand, both the deletion and duplication of the ATR region strongly influences promoter melting, although in opposite ways, as discussed further below.
This influence on promoter melting was assayed by developing and applying new methods for measuring the rate and temperature dependence of open complex formation in vivo. These are adaptations of the standard in vivo KMn04 footprinting assay. This allows for the first time an assessment of the kinetics and thermodynamics of open complex assembly in vivo. The rate of open complex formation at the glnAP2 promoter was found to be slightly less than 1 min in activating media. Although this measurement has not been made at the glnAP2 promoter in vitro, it has been made at many other bacterial promoters in vitro, and those measurements generally gave tIl2 of up to a few minutes. The in vitro rates can vary with solution conditions such as the concentration and nature of ions and the concentrations of macromolecules (McClure, 1985). It will be informative to study the same promoters in vivo and in vitro to assess the relevance of the variety of solution conditions used for in vitro transcription. Similarly, it will be useful to apply the simple melting curve procedure used here to compare with the melting behavior in solution. Although rifampicin is postulated not to affect steps prior to chain initiation (McClure and Cech, 1978), minor effects on the rate of open complex formation cannot be excluded. If precise rate differences are important to determine, one should compare the in vivo rate to the in vitro rates, both in the presence and absence of rifampicin.
The application of these methods to the ATR region of confirmed its importance in melting the promoter during transcription. The melting curves showed that at least one ATR region is required for a physiological temperature melting transition in the promoter. A lower temperature transition was not affected by the deletion of the ATR region. Although the full ATR region has a net negative charge of -12, its partial deletion leaves intact an adjacent region with an approximate net negative charge of -25. Perhaps it is this residual acidity that allows the low temperature melting to It could also perturb the conformation of the ATa64 holoenzyme to more closely resemble the enzyme in the open complex state.
The ATR region seems to act as a module, presumably electrostatic. However, although the primary chemical characteristic of the ATR region is acidity, it is not presently known which feature of this region is critical for melting the DNA. The hypothesis that acidity is the critical feature for DNA melting is supported by the lack of homology in this region in eight sequenced uS4s (GenBank, University of Wisconsin). The sequence of this region is not generally conserved, and the size can be up to 21 amino acids larger than the E..coli region. Perhaps this helps explain why our insertion and deletion constructs are so well tolerated. Despite the lack of length and sequence conservation, the acidic character of the region is conserved, suggesting a functional role for the acidity.
How does the acidic region assist in promoter melting? Acidic regions are known to be important for activation of certain genes by eukaryotic RNA polymerase 11. Experiments in those systems have led to a number of proposals, including suggestions that acidic regions bind basal transcription factors and "recruit" them to the promoter. In one case, this may occur by binding to a basic patch on a basal transcription factor (Lin et al., 1991). Several observations suggest that factor recruitment is unlikely to be the role of the acidic domain. There are only two other protein components of the transcription complex: NTRC-phosphate and core RNA polymerase. As discussed by Sasse-Dwight and Gralla (1990), NTRC does not appear to have a basic patch that might interact with the acidic patch of u64. Moreover, we have found that when the Del 9 mutant (where the ATR region is deleted) is recombined with a second mutant form of d 4 which does not require NTRC-phosphate for activation, the double mutant retains the defect in melting.2 This defect occurs in cells both expressing and lacking NTRC. This would suggest that the ATR region is not involved in the recruitment of this protein. The ATR region is also unlikely to be involved in core polymerase binding. This is because u54 is required along with the core RNA polymerase for glnAP2 promoter recognition , and promoter recognition is still * M. Hsieh and J. D. Gralla, unpublished observation. intact in the mutant with no ATR region. Thus the acidic trimer repeat region of u54 is unlikely to function by recruiting either of the other protein components of the glnAP2 transcription apparatus.
An alternative role of the acidic region would be to promote conformational changes leading to melting within the assembled closed transcription complex. Sasse-Dwight and Gralla (1990) suggested that this could occur by the negatively charged acidic region of u54 being driven by NTRC-phosphate and ATP hydrolysis toward the negatively charged DNA. This could create an energetically unfavorable electrostatic environment which would be relieved by the DNA duplex melting. The separated strands would be delivered to singlestrand binding sites, possibly in the core enzyme where the template strand is read during transcription. Alternatively, the acidic region could approach another region within the transcription complex and trigger conformational changes, possibly again driven by electrostatic repulsion. These models are consistent with the results reported here. The acidity of the ATR region would induce the observed physiological temperature melting transition, and as observed, an extra copy of the ATR region would provide the additional negative charge to cause this melting to occur faster.