Mapping ATP-dependent Activation at a σ54 Promoter*

The σ54 promoter specificity factor is distinct from other bacterial RNA polymerase (RNAP) σ factors in that it forms a transcriptionally silent closed complex upon promoter binding. Transcriptional activation occurs through a nucleotide-dependent isomerization of σ54, mediated via its interactions with an enhancer-binding activator protein that utilizes the energy released in ATP hydrolysis to effect structural changes in σ54 and core RNA polymerase. The organization of σ54-promoter and σ54-RNAP-promoter complexes was investigated by fluorescence resonance energy transfer assays using σ54 single cysteine-mutants labeled with an acceptor fluorophore and donor fluorophore-labeled DNA sequences containing mismatches that mimic nifH early- and late-melted promoters. The results show that σ54 undergoes spatial rearrangements of functionally important domains upon closed complex formation. σ54 and σ54-RNAP promoter complexes reconstituted with the different mismatched DNA constructs were assayed by the addition of the activator phage shock protein F in the presence or absence of ATP and of non-hydrolysable analogues. Nucleotide-dependent alterations in fluorescence resonance energy transfer efficiencies identify different functional states of the activator-σ54-RNAP-promoter complex that exist throughout the mechano-chemical transduction pathway of transcriptional activation, i.e. from closed to open promoter complexes. The results suggest that open complex formation only occurs efficiently on replacement of a repressive fork junction with down-stream melted DNA.

The 54 promoter specificity factor is distinct from other bacterial RNA polymerase (RNAP) factors in that it forms a transcriptionally silent closed complex upon promoter binding. Transcriptional activation occurs through a nucleotidedependent isomerization of 54 , mediated via its interactions with an enhancer-binding activator protein that utilizes the energy released in ATP hydrolysis to effect structural changes in 54 and core RNA polymerase. The organization of 54 -promoter and 54 -RNAP-promoter complexes was investigated by fluorescence resonance energy transfer assays using 54 single cysteine-mutants labeled with an acceptor fluorophore and donor fluorophore-labeled DNA sequences containing mismatches that mimic nifH early-and late-melted promoters. The results show that 54 undergoes spatial rearrangements of functionally important domains upon closed complex formation. 54  Sequence-specific binding of the multisubunit bacterial core RNA polymerase (RNAP, 2 ␣ 2 ␤␤Ј (E)) to promoter DNA is conferred by an additional subunit, the factor. Two families of factor, 70 and 54 , bind to the core RNAP apoenzyme, and each endows the resulting holoenzyme with distinct properties (1)(2)(3). Core RNAP binding is known to induce conformational changes in the 70 subunit required for recognition of consensus promoter sequences (4 -7). In contrast to 70 , 54 can bind promoter sequences in the absence of core RNAP. Specific differences in the interactions with promoter DNA between 54 and 54 -RNAP (E 54 ) suggest that interactions with core RNAP induce conformational changes in the 54 subunit (8,9). These changes and the organization of 54 domains within the closed complexes have been probed previously using single cysteine 54 variants to map the proximity of specific 54 residues to regions within the promoter DNA in the presence of the tethered chemical nuclease reagent Fe-BABE (10 -13). These analyses revealed that 54 adopts a C-to N-terminal orientation in the 5Ј to 3Ј direction with respect to the nontemplate strand and demonstrated that the regulatory N-terminal domain of 54 (Region 1; Fig. 1A) and promoter DNA both undergo conformational changes, induced by the binding of core RNAP to an initial 54 -DNA complex leading to the formation of a conformationally stable and transcriptionally silent closed complex (11,14). The 54 Region I and Region III domains are required for recognition of the start site proximal consensus promoter recognition sequence (the Ϫ12 region) where DNA opening nucleates. The interactions between Region I and Region III with the Ϫ12 site are of regulatory significance and are required for preventing spontaneous isomerization of the closed complex to the open promoter complex in the absence of activation (10,(15)(16)(17)(18)(19)(20). Truncated versions of 54 lacking Region I ( 54 ⌬RI) can bind to core RNAP, forming E 54 ⌬RI complexes capable of activator-independent transcription from supercoiled and so called late-melted promoters that are mismatched between positions Ϫ10 and Ϫ1, thus mimicking the conformation adopted by the promoter DNA in the open complex but not on DNA constructs mismatched between Ϫ12 and Ϫ11, which mimic the conformation adopted by the promoter in the closed complex. This implies that in closed complexes formed by E 54 ⌬RI there is a loss of interactions with regions of DNA inhibiting open complex formation and an increase in interactions with single-stranded DNA (16). The C-terminal RpoN box (Fig. 1A) is a 54 signature sequence implicated in the binding to consensus DNA sequences at Ϫ24 (12,20).
Transcriptional activation in the 54 system is effected by a AAAϩ protein (ATPase associated with various cellular activities) bound to an upstream enhancer element being brought into contact through DNA looping with the closed complex (21,22). The interaction between subunits of the E 54 activator phage shock protein F (PspF) and E 54 closed complex are con-ditioned by the binding and hydrolysis of ATP, which drives PspF oligomerization and leads to an isomerization of E 54 closed complexes and subsequent formation of transcription-  (11) is shown in the middle. Regions and residues associated with binding to core RNAP, activator, or DNA are indicated along with the positions of the mutant cysteine residues and their Fe-BABE cleavage patterns on the early-melted promoter (bottom). B, SDS-PAGE on 8% polyacrylamide gels of unlabeled (lanes 1-4) and Alexa 594-labeled (lanes 5-7) 54 subunits stained with Coomassie Blue (top line) and visualized by fluorography (bottom line). Lane 1, wild-type 54 ; lanes 2 and 5, C20; lanes 3 and 6, C463; lanes 4 and 7, C474. C, gel mobility shift assay showing the effects of fluorophore labeling on the 54 variants (250 nM) binding to radiolabeled nifH (50 nM) early-melted DNA in the absence or presence of PspF 1-275 (15 M) and 2 mM ATP or 0.2 mM ADP⅐AlF x . The positions of the free DNA and 54 -DNA complexes are shown together with super-shifted (ss) isomerized and trapped species formed in the presence of PspF with ATP or ADP⅐AlF x , respectively. Lanes 1, 5, and 9, wild-type 54 ; lanes 2, 6, and 10, C20; lanes 3, 7, and 11, C463; lanes 4, 8, and 12, C474; lane 13, unbound early-melted DNA. D, gel mobility shift assay as in C but showing the effects with 54 -RNAP on late-melted DNA. E 54 was formed by the addition of core RNAP (63 nM) to the various 54 (250 nM) and radiolabeled nifH (50 nM) late-melted DNA complexes in the absence or presence of PspF 1-275 (15 M) and 2 mM ATP or 0.2 mM ADP⅐AlF x . Heparin (10 g/ml) was added after 10 min of incubation with ATP. Lanes 1, 5, and 9, wild-type 54 ; lanes 2, 6, and ally active open complexes (23,24). Substitution of ATP with analogues (ATP␥S or AMP-PNP) or the ATP hydrolysis transition state analogue, ADP⅐AlF x , suggests that nucleotides drive the formation of different functional states of PspF and alter interactions between E 54 and a fork junction DNA structure at the Ϫ12 promoter consensus region in the closed complex (23,(25)(26)(27). This is consistent with the idea that PspF acts via several discrete conformational states, each making functionally distinct interactions with E 54 -promoter DNA complexes before completing ATP hydrolysis for full transcriptional activation. The network of interactions governing these events at the point of ATP hydrolysis has recently been elucidated by combining the crystal structure of PspF with cryogenic electron microscopy images of 54 in complex with oligomeric PspF bound to ADP⅐AlF x (27,28). Using these structural data, sitedirected PspF mutants were assayed for ATP binding and hydrolysis activity and the ability to trigger open complex formation. It appears that structural changes occurring in the nucleotide binding pockets formed by the PspF hexamer are communicated via relocation of a conserved loop motif to make stable contacts with 54 (27). Assaying the ability of various 54 mutants to form transcriptionally active open complexes indicated the nucleotide-bound form of PspF targets Region I of 54 (11,29,30). However, because of the transient nature of the initial contact, the events and associated processes whereby weak interactions between PspF and E 54 closed complex in the absence of nucleotide lead to the formation of a stable complex capable of isomerization remain unclear. Spectroscopic methods provide a powerful means of analyzing complex processes involving macromolecular interactions. Fluorescence resonance energy transfer (FRET) has become established as a method of choice in the dissection of protein-protein and protein-oligonucleotide interactions due to the sensitivity of non-radiative energy transfer over nanometer scales. The ease with which fluorophores with high quantum yields can be incorporated into both DNA-binding proteins and their substrates has given added impetus to the analysis of structure-function relationships between factors and multisubunit RNA polymerases and the formation of transcriptionally active complexes (7,31).
Here we describe a FRET assay capable of assigning relative separations of dye labels incorporated into defined sites on the 54 subunit and the ends of DNA fragments encompassing the Sinorhizobium meliloti nifH DNA promoter sequence. Labeled positions in 54 were chosen based on Fe-BABE analyses (11,12). The FRET assay has been used to probe the structural changes  Promoter binding by wild-type and dye-labeled 54 and E 54 in the presence or absence of nucleotide-dependent activator protein Gel shifts were conducted as in Fig. 1C. Values represent the percentage of input DNA bound Ϯ S.E.; n ϭ 3.

Early-melted promoter
Late-melted promoter Unactivated Activated a Activated a Unactivated Activated b a Activation of -DNA and isomerized supershifted -DNA complexes (ss) in the presence of PspF 1-275 and ATP. b Activated E represents complexes isomerized in the presence of PspF 1-275 and ATP that remain after heparin challenge.
involved in promoter complex formation on the early-melted construct (mismatched between Ϫ12 and Ϫ11, Fig. 1A (23)) and the late-melted construct (mismatched between Ϫ10 and Ϫ1; Fig. 1A (32)). Note these promoter constructs behave differently in transcription assays. The early-melted promoter can be isomerized by 54 alone (in response to activation) but not by the E 54 complex because the latter is sensitive to the restrictive DNA conformation at Ϫ12 and Ϫ11 (16). In contrast, the latemelted promoter isomerizes with E 54 (in response to activation), is competitor (heparin)-resistant, and supports transcription, i.e. it shows most of the features of the natural open complex. Structural alterations occurring upon nucleotide hydrolysis-dependent closed to open complex transition were inferred from assays utilizing truncated PspF (PspF 1-275 ) lacking a DNA binding motif. PspF 1-275 retains the ability to isom-erize 54 and E 54 promoter complexes and activate transcription efficiently (11,33). In addition, the use of PspF 1-275 facilitates experimental design since it does not rely on upstream activating sequences for transcription activation. FRET assays of the interactions of PspF 1-275 with 54 and E 54 promoter complexes (formed on the early-or late-melted fragments) in the presence or absence of hydrolysable and non-hydrolysable nucleotides provide novel insights into the structural alterations occurring during formation of the open complex.

EXPERIMENTAL PROCEDURES
Protein Expression and Purification-Klebsiella pneumoniae wildtype 54 and single-cysteine variants (C20, C463, and C474) and the catalytic domain (residues 1-275) of the 54 -dependent activator phage shock protein F (PspF 1-275 ) were purified as N-terminal His 6tagged fusion proteins. All proteins were overexpressed in BL21 Escherichia coli cells in 2YT medium in the presence of 50 g/ml kanamycin. After growth to A 600 of ϳ0.6 at 37°C, cells were induced by the addition of 0.8 mM isopropyl 1-thio-␤-D-galactopyranoside, incubated at 30°C for 3 h, then harvested by centrifugation at 11,000 ϫ g for 30 min at 4°C. All subsequent steps were performed at 4°C in the presence of protease inhibitor mixture (Complete-EDTA; Roche Diagnostics) unless stated otherwise. Cells pellets were resuspended in 50 mM phosphate buffer, pH 8.0, 150 mM NaCl, 10 mM imidazole, 5 mM ␤-mercaptoethanol, 10% (v/v) glycerol, 1% (v/v) Tween 20. Resuspended cells were then lysed with a French press, and insoluble material removed by centrifugation at 11,000 ϫ g for 30 min.
Cell lysates were loaded onto a 5-ml immobilized metal affinity column (HisTrap; Amersham Biosciences) and washed with 50 ml of 50 mM phosphate buffer, pH 8.0, 300 mM NaCl, 5 mM ␤-mercaptoethanol, 5% (v/v) glycerol, then 50 ml of the above wash buffer containing 20 mM imidazole. Specifically bound protein was eluted by a step elution of 15 ml of wash buffer containing 600 mM imidazole, and the eluted fraction was exchanged into 10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM DTT, 5% (v/v) glycerol by passage through a HiPrep 26/10 desalting column. Fractions containing the peak of protein were loaded onto a 1-ml Sepharose Q col- umn, washed with 5 ml of 10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM DTT, 5% (v/v) glycerol, 0.1 mM EDTA, then eluted with a linear gradient of NaCl (50 -1000 mM) in the same buffer over 20 ml. Purified protein was dialyzed overnight against 10 mM Tris, pH 7.5, 50 mM NaCl, 1 mM DTT, 50% (v/v) glycerol, 0.1 mM EDTA for storage at Ϫ80°C. Cysteine mutant 54 subunit proteins were purified in an identical fashion before fluorescent labeling except DTT was omitted from chromatography and dialysis buffers.
Fluorescent Labeling of Cysteine Mutant 54 Subunit Protein-Purified cysteine mutant 54 subunit protein (1 ml) was incubated on a Reduce-Imm column (Pierce) for 1 h at room temperature in 10 mM Tris, pH 7.5, 50 mM NaCl, 5% (v/v) glycerol, 0.1 mM EDTA. The column was eluted with the same buffer, and the fraction containing the peak of protein was incubated with a 10 -20 M excess of Alexa Fluor 594 maleimide for 2 h at room temperature after which the reaction was quenched with 10 mM ␤-mercaptoethanol. The quenched reaction mixture was adjusted to 30% saturated (NH 4 ) 2 SO 4 and loaded onto a 1-ml phenyl-Sepharose column. The column was washed with 15 ml of 50 mM Tris, pH 7.5, 1 mM DTT, 5% (v/v) glycerol, 0.1 mM EDTA, 30% saturated (NH 4 ) 2 SO 4 then eluted with a linear gradient of 30 to 0% saturated (NH 4 ) 2 SO 4 in 50 mM Tris, pH 7.5, 1 mM DTT, 5% (v/v) glycerol, 0.1 mM EDTA over 15 ml. Purified labeled fractions were dialyzed as above for long term storage at Ϫ80°C. Comparison of elution profiles of labeled and unlabeled 54 protein preparations from phenyl-Sepharose columns and SDS-PAGE analysis of eluted fractions indicates that hydrophobic interaction chromatography separates protein from free dye and resolves labeled and unlabeled protein (data not shown). Surface-enhanced laser desorption ionization (SELDI) mass spectrometry of labeled cysteine mutant 54 subunit protein samples immobilized on nickel-charged supports confirmed incorporation of single dye molecules (data not shown). Fluorophore concentrations in eluted fractions were determined by visible spectroscopy using the published extinction coefficients for Alexa Fluor 594 maleimide (96,000 M Ϫ1 cm Ϫ1 ; Molecular Probes), and protein concentrations were quantified by amino acid analysis. The efficiency of labeling of cysteine mutant 54 subunit protein preparations, as determined by the molar ratio of fluorophore to protein, was typically Ն70%.
Promoter DNA Fragments-Synthetic DNA oligonucleotides corresponding to Ϫ38 to ϩ6 bp of the S. meliloti nifH promoter sequence were used in the formation of fragments to assay wildtype and fluorescently labeled cysteine mutant 54 function. Incorporation of mismatches at positions Ϫ12 and Ϫ11 or from Ϫ10 to Ϫ1 generates mismatched DNA constructs that mimic the early-and late-melted promoter conformations, respectively (2,11,16). Mismatched DNA constructs used in native gel mobility shift assays were formed after the method of Wigneshweraraj et al. (34) by annealing a radiolabeled promoter strand with a 2-fold molar excess of the unlabeled complementary strand. FRET measurements were performed using  Table 2, are shown here as a series of histograms for the various states of the complexes as indicated (error bars ϭ S.D.; n ϭ 4). These changes are relative to the initial separations of the dyes, which were as follows. A, C20 to position Ϫ38 ϭ 80 Å; C20 to position ϩ6 ϭ 80 Å. B, C463 to position Ϫ38 ϭ 60 Å; C463 to position ϩ6 ϭ 75 Å. C, C474 to position Ϫ38 ϭ 60 Å; C474 to position ϩ6 ϭ 95 Å.  3.5% (w/v) polyethylene glycol 6000 in a final volume of 20 l at 30°C using 100 nM radiolabeled promoter probe and 500 nM 54 . The ability of the preformed 54 early-melted complex to undergo isomerization was then assayed by the addition of 15 M PspF 1-275 and 2 mM ATP, and the reaction was incubated for a further 5 min. In experiments utilizing late-melted promoter, isomerized heparin-stable E 54 promoter complexes were isolated by the addition of 10 g/ml heparin after completion of the isomerization reaction. Reactions were analyzed on a 4.5% polyacrylamide gel under non-denaturing conditions in 25 mM Tris, 200 mM glycine buffer, pH 8.6, at room temperature. The dried gel was visualized, and bound DNA was quantified using phosphorimaging.
FRET Assays and Relative Separation Calculations-Fluorescence measurements were conducted using a Shimadzu RF-5301 spectrophotometer with excitation at 488 nm and excitation and emission slit band-pass settings at 5 nm. Assays were performed in a 40-l, 3-mm path length microcuvette inside a temperature-controlled sample chamber at 30°C utilizing buffer and incubation conditions identical to gel shift assays. Formation of 54 -RNAP was achieved by the addition of E. coli core RNA polymerase (Epicenter Technologies) to give a 1:4 molar ratio of core to 54 . Emission spectra were measured after 5 min and corrected for dilution. In experiments analyzing the effect of activator protein in the presence or absence of nucleotides, spectra were measured before and after the addition of PspF 1-275 and corrected for dilution and the presence of activator. Nucleotide was then added, and spectra were recorded after 10 min and corrected for dilution and the presence of nucleotide. The transition state analogue, ADP⅐AlF x , was formed in situ by the addition of 0.2 mM AlCl 3 to an assay mixture containing 0.2 mM ADP and 5.0 mM NaF.
The efficiency of fluorescence resonance energy transfer was calculated using the equation E FRET ϭ 1 Ϫ (I DA /I D ), where I DA is the integrated donor fluorescence intensity in the presence of labeled acceptor proteins, and I D is the integrated donor fluorescence intensity in the presence of unlabeled acceptor proteins. The relative change in FRET was then corrected for the percentage of each bound species estimated using the population distribution in an identical gel shift assay (see the supplemental data).
The relative separation (R) between donor and acceptor was calculated from FRET efficiencies using R ϭ R 0 ((1/E FRET ) Ϫ 1) 1/6 , where R 0 , the Förster distance, is the distance at which energy transfer is 50% of the maximum value. A value for R 0 of 54 Å for the Alexa Fluor 488/Alexa Fluor 594 dye pair calculated from ensemble FRET measurements of donor and acceptor fluorophores, separated by known distances in a DNA ladder, agreed with previously published data (35).

Fluorescent Labeling of 54 and DNA Promoters Does Not
Affect Recognition and Isomerization-Previously single cysteine variants of 54 have been created to provide proximitybased footprinting tools. These have been used here as the points of attachment of the acceptor for FRET. Three mutants were chosen on the basis of their cysteine locations in the activator binding (C20 of Region I) or DNA binding (C463 and C474 in Region III) domains ( Ref. 11; Fig. 1). These were labeled (efficiency Ն70%) with Alexa Fluor 594 maleimide and purified away from unlabeled protein and free dye by hydrophobic interaction chromatography on phenyl-Sepharose (Fig. 1B).
The labeled proteins were then assayed for their abilities to bind radio-or fluorescently labeled (data not shown) promoter fragments. Two types of promoter fragment, early-and latemelted (see "Experimental Procedures") were used in electrophoretic mobility shift assay and isomerization assays (Fig. 1, C  and D). Gel-shift assays of 54 and E 54 complexes on either DNA fragment that have undergone nucleotide-dependent isomerization by PspF 1-275 reveal the presence of differently migrating species in the presence of ADP⅐AlF x or ATP. Complexes formed by the action of PspF  in the presence of transition state analogue are described as being "trapped" in a state conformationally distinct from that found in the fully isomerized 54 early-melted promoter construct (supershifted) species generated in response to ATP hydrolysis by PspF  . Such fully isomerized complexes on the late-melted promoter are defined by their ability to withstand a challenge with heparin. The percentages of total radiolabeled DNA bound in the non-isomerized, fully isomerized, and trapped species for both DNA fragments interacting with wild-type and dye-labeled factors are listed in Table 1. The results suggest that the labeling does not interfere with DNA binding, activation, and subsequent isomerization.
FRET Analysis of the Activation Pathway on an Early-melted Promoter-FRET assays were then carried out using donor-labeled (Alexa 488) DNA with the dye on either the non-template or template strand. Spectra were recorded for the unbound early-melted DNA, the DNA with saturating levels of 54 or core RNAP, and finally for the addition of core RNAP to the 54 early-melted DNA complex. Data were corrected by subtraction of appropriate control spectra (see details in the supplemental data). The results are shown in Fig. 2. The addition of core RNAP to early-melted DNA results in reductions in donor emission intensity compared with the protein-free promoter, suggesting that significant fluorescence quenching occurs. Adding the factor to the free DNA also causes a decrease in donor intensity and an increase in acceptor fluorescence with an emission peak at 610 nm (Fig. 2, A-D). These results confirm that FRET occurs between 54 and the early-melted DNA. Assembling E 54 with the labeled factor results in further decreases in donor emission that are larger than the effect(s) caused by core RNAP quenching, presumably due to alterations in the conformations in the E 54 complex. These FRET effects occur with differing efficiencies, as expected, for the different sigma variants and the different positions of the donor. Higher FRET efficiencies are apparent in the assays of 54 C20, 54 C463, or 54 C474 binding to the DNA with the dye on the non-template strand (Fig. 2, B, D, and F) compared with the template strand (Fig. 2, A, C, and E). This is consistent with proximity expectations based on the Fe-BABE footprinting and other structural studies of these complexes (11).
Isomerization of the 54 early-melted complex (in the absence of core RNAP) via the ATP-dependent action of PspF 1-275 results in significant conformational changes, as shown by altered migration in gel shift assays (Fig. 1C, ss). In the FRET assay, the addition of PspF 1-275 to the complex in the absence of nucleotide induces slight decreases in FRET efficiency that are most easily seen when the donor is on the template strand (Fig. 3). The addition of ATP to this mixture, which would be expected to produce the isomerization seen in Fig. 1C, FIGURE 6. Mapping changes in the relative separations of 54 domains and DNA during isomerization on the late-melted promoter. The apparent relative separations (⌬R app ; Å) of the fluorophores are shown in the legend for Fig. 4 and are relative to the initial separations, which were as follows. A, EC20 to position Ϫ38 ϭ 70 Å; EC20 to position ϩ6 ϭ 70 Å. B, EC463 to position Ϫ38 ϭ 50 Å; EC463 to position ϩ6 ϭ 65 Å. C, EC474 to position Ϫ38 ϭ 30 Å; EC474 to position ϩ6 ϭ 70 Å. NOVEMBER 3, 2006 • VOLUME 281 • NUMBER 44 results in a more significant effect. Use of the non-hydrolysable ADP⅐AlF x results in only a slight reduction in this effect compared with ATP, implying that the conformation of the complex during the transition state for nucleotide hydrolysis shares features found in the final (ATP hydrolysis-dependent) state. When the assays are repeated with a PspF 1-275 variant (T86A) that can hydrolyze the ATP but is unable to couple this to isomerization of the factor, the FRET curves are essentially identical to those with PspF 1-275 in the absence of nucleotide, establishing FRET changes as reflecting outcomes of energy coupling.

Mapping ATP-dependent Activation at a 54 Promoter
In FRET assays utilizing PspF 1-275 in the absence of nucleotides, the slight difference observed for the C20 mutant on the early-melted template strand was no longer apparent. Clearly, upon binding of hydrolyzable or non-hydrolyzable nucleotide, PspF 1-275 interacts with Region I of 54 bound to distort DNA, thereby altering FRET efficiencies without full isomerization. The importance of Region I in formation of the regulatory center at the Ϫ12 fork junction DNA was established in experiments using labeled homoduplexes (data not shown). The absence of FRET between C20 and donor label on the homoduplex confirms the importance of DNA distortion at this position for interaction with Region I, forming a nucleo-protein interface that is the target for interaction with the activator. These observations confirm that FRET can be used to identify and discriminate between the various forms of the activation complex and also that the mutant T86A PspF 1-275 is unable to influence the conformation of the promoter bound factor.
The precise extent of conformational change occurring during the assays shown in Fig. 3 is not readily apparent from the spectra because only a fraction of the free DNA binds 54 and only a fraction of the bound 54 complex undergoes PspF 1-275mediated isomerization. The gel shift assays shown in Fig. 1C were carried out under essentially identical conditions to the FRET assays. Thus, it is possible to estimate the amount of each of these complexes present in the FRET assays by densitometry of the bands in Fig. 1C (Table 1). Very little affinity or functional difference between the wild-type and variant factors was evident.
Correcting the FRET curves for both the effects of quenching and the amounts of each species present allows us to estimate the relative dye separations when isomerization occurs ( Table  2; Fig. 4; see the supplemental data). ATP hydrolysis-dependent isomerization or formation of the trapped transition state results in increases in separation of ϳ15-40 Å between the 5Ј end of the early-melted non-template strand and the factor for all three single cysteine 54 mutants. This suggests something like a rigid body movement between the protein and the promoter in the direction of the transcript start (see Fig. 7A). However, there is also an apparent increase in separation (ϳ10 -25 Å) from the 5Ј end of the template strand that is not nucleotide-dependent. This distance would be expected to decrease if the proteins moved toward the transcript start. Therefore, under these conditions either the factor is not properly engaged with the DNA or there is a "bending/scrunching" of the DNA and protein complex, probably due to the conformation of the early-melted promoter (see Fig. 7A).
FRET Analysis of the Activation Pathway on a Late-melted Promoter-To test the idea that the conformational changes described above were the result of the choice of promoter fragment, we then carried out equivalent assays with the latemelted promoter DNA. Acceptor fluorophore-labeled E 54 binds to this DNA leading to decreases in donor fluorescence and increases in acceptor emission. The relative FRET efficiencies are different from those recorded for E 54 on the earlymelted promoter, suggesting that the holoenzyme is in a slightly different conformation on the two DNAs (Fig. 5, cf. Fig.  2). Similarly, in contrast to the results with the early-melted

Effect of wild-type and mutant PspF 1-275 on FRET between donor-labeled nifH promoters and acceptor-labeled 54 and E 54
Energy transfer efficiencies were corrected for presence of non-isomerized species (see Table 1). ND, not done; Ϫ, no change.

Donor
Acceptor Changes in apparent dye separation DNA, the addition of PspF 1-275 in the absence of nucleotide and heparin results in increased FRET efficiencies when the donor is present on the template strand ( Figs. 5 and 6, B, D, and F; Table 3). However, no significant changes to FRET occur with the donor on the non-template strand. Adding ATP to these complexes results in decreased FRET efficiencies when the donor is on the non-template strand, similar to the results with the early-melted DNA. However, they increase when the donor is on the template strand, consistent with movement of 54 away from the upstream sequences and toward the transcript start (Fig. 7B). Heparin was used to challenge the complexes being formed. Only ATP hydrolysis-dependent isomerized species are stable to heparin challenge, whereas intermediates are not (25,31). The use of the transition state analogue, ADP⅐AlF x , results in changes that are as great and in the same direction as those seen with ATP, consistent with the idea that this conformation is closer to the fully isomerized state than the starting state in the absence of nucleotides. As with the early-melted promoter experiments, use of the T86A PspF 1-275 variant mostly ablates these changes, con-firming that they reflect specific PspF-mediated conformational rearrangements in the activation complex.
Correcting the FRET curves as described above, including following heparin challenge where appropriate, yields the FRET efficiency values listed in Table 3. Fig. 6 shows the inferred alterations in relative separation of the dyes in each case. A number of structural studies have been carried out on this system, allowing us to estimate the relative separations of protein domains from the ends of the promoter. Inspection of such structures suggests that the distance estimates for the nonisomerized E 54 complexes are consistent with current cryoelectron microscopy models (36), confirming that the FRET corrections are sensible. Increases in separation from the upstream dye are closely matched by decreases in separation toward the dye located downstream. These results are consistent with an isomerization event in which the factor alters its spatial arrangement on the DNA and moves toward the transcript start (Fig. 7B). This is entirely consistent with other biochemical assays of open complex formation (7,31), although it cannot exclude a conformational rearrangement entirely within the DNA. Interestingly, the increases and decreases are different for the different mutant proteins and for the different functional states within each mutant and are not consistent with a simple rigid movement of the 54 protein domains with respect to the DNA, implying that core RNAP mediates some of the changes, which is not easily reconciled with an entirely DNA-mediated change.
Transcriptional activation is a complex, multistep process in which multiple protein complexes and promoter DNA undergo a series of defined conformational transitions. Biochemical and structural studies can only characterize and probe a limited number of these states, making it difficult to extract a temporal sequence of events defining the transcriptional activation pathway. In principle, single molecule fluorescence studies can be used to extract such information. The results presented here show that suitably labeled substrates will undergo the expected conformational rearrangements that are the prerequisite to transcription initiation. The fluorescent properties of the species generated have unique signals allowing their presence to be detected, making practical such experiments, and these are in hand. FIGURE 7. Schematic showing the effects of nucleotide-dependent activator protein on 54 early-melted and E 54 late-melted complexes. A, region I of 54 interacts with the fork junction DNA structure at Ϫ12/Ϫ11 in early-melted promoter DNA to form a closed complex. PspF 1-275 interacts transiently with 54 at Region I. The addition of ADP⅐AlF x enables PspF 1-275 to engage further with 54 -DNA, trapping the complex at the point where ATP hydrolysis is mechano-chemically transduced into alteration of 54 structure and promoter DNA bending. In contrast, the addition of ATP enables PspF 1-275 to complete the nucleotide hydrolysis pathway, leading to open complex formation, resulting in melting of DNA to Ϫ5, isomerization of 54 , and further DNA bending. B, late-melted promoter mismatched at Ϫ10 to Ϫ1 is bound by E 54 to form a closed complex with which PspF 1-275 can interact transiently. The addition of ADP⅐AlF x causes E 54 to translocate away from downstream promoter DNA and toward the transcription start site, trapping the promoter complex in a conformation similar to that seen in the isomerized open complex formed after hydrolysis of ATP by PspF 1-275 . Translocation of E 54 toward the promoter start site is indicated by arrows.