An Intramolecular Route for Coupling ATPase Activity in AAA+ Proteins for Transcription Activation*

AAA+ proteins (ATPases associated with various cellular activities) contribute to many cellular processes and typically function as higher order oligomers permitting the coordination of nucleotide hydrolysis for functional output, which leads to substrate remodeling. The precise mechanisms that enable the relay of nucleotide hydrolysis to their specific functional outputs are largely unknown. Here we use PspF, a specialized AAA+ protein required for enhancer-dependent transcription activation in Escherichia coli, as a model system to address this question. We demonstrate that a conserved asparagine is involved in internal organization of the oligomeric ring, regulation of ATPase activity by “trans” factors, and optimizing substrate remodeling. We provide evidence that the spatial relationship between the asparagine residue and the Walker B motif is one key element in the conformational signaling pathway that leads to substrate remodeling. Such functional organization most likely applies to other AAA+ proteins, including Ltag (simian virus 40), Rep40 (Adeno-associated virus-2), and p97 (Mus musculus) in which the asparagine to Walker B motif relationship is conserved.

AAA ؉ proteins (ATPases associated with various cellular activities) contribute to many cellular processes and typically function as higher order oligomers permitting the coordination of nucleotide hydrolysis for functional output, which leads to substrate remodeling. The precise mechanisms that enable the relay of nucleotide hydrolysis to their specific functional outputs are largely unknown. Here we use PspF, a specialized AAA ؉ protein required for enhancer-dependent transcription activation in Escherichia coli, as a model system to address this question. We demonstrate that a conserved asparagine is involved in internal organization of the oligomeric ring, regulation of ATPase activity by "trans" factors, and optimizing substrate remodeling. We provide evidence that the spatial relationship between the asparagine residue and the Walker B motif is one key element in the conformational signaling pathway that leads to substrate remodeling. Such functional organization most likely applies to other AAA ؉ proteins, including Ltag (simian virus 40), Rep40 (Adeno-associated virus-2), and p97 (Mus musculus) in which the asparagine to Walker B motif relationship is conserved.
AAA ϩ proteins (ATPases associated with various cellular activities) are present in all kingdoms of life and play important roles in numerous cellular activities, including proteolysis, protein folding, membrane trafficking, cytoskeletal regulation, organelle biogenesis, DNA replication, and DNA transcription. AAA ϩ ATPases invariably contain Walker A and B motifs that define them as P-loop ATPases and a conserved sequence termed the second region of homology and function as higher order oligomers, which remodel their substrates in reactions that consume ATP (1)(2)(3)(4)(5)(6)(7). In many cases, AAA ϩ domains assemble into hexameric rings that change their conformation during the ATPase cycle. This nucleotide-dependent conformational change may, for example, apply mechanical tension to bound proteins or nucleic acids and thereby allow AAA ϩ proteins to remodel their substrate. The energy-dependent nature of their activities and their organization as ring assemblies raises important issues about how they function as molecular machines to engage with, and remodel, their targets. A common area of functionality that is poorly understood concerns how nucleotide binding and hydrolysis is relayed within the ring structure to allow formation of the functional states that accompany and drive substrate remodeling. Importantly, AAA ϩ proteins represent a large class of mechano-chemical enzymes that have evolved many ways of using a fundamentally similar conformational change in different biological settings (8). Indeed, these proteins often become specialized by the insertion of specific motifs within the minimal AAA ϩ core.
One well studied example of this specialization is represented by the family of bacterial enhancer-binding proteins (bEBPs) 3 required for 54 -dependent transcription activation (9). In contrast to 70 -dependent transcription, which is constitutively active, 54 -dependent transcription requires specific activators (the bEBPs) that couple ATP hydrolysis to isomerization of the initial transcriptionally inactive closed complex (CC), to a transcriptionally proficient open complex (OC) (10 -15). 54 -dependent transcription activation is functionally analogous to eukaryotic RNAP II, which requires energy derived from ATP hydrolysis provided by TFIIH (16,17). The bEBPs, which include the well studied activators DctD, DmpR, NifA, NtrC, NtrC1, PspF, and XylR, are characterized by an insertion: the L1 loop containing the "GAFTGA motif," which is required for specific interaction with the 54 N-terminal regulatory domain, 54 region 1 (2, 4, 9, 18 -20). These bEBPs are members of a sub-class of AAA ϩ proteins known as the presensor I ␤-hairpin super-clade and include the helicases RuvB, Ltag, and MCM as well as proteases HslU, ClpX, and Lon (21). In this study, we use the bEBP model, PspF (phage shock protein F), from Escherichia coli, which is composed of: (i) a catalytic AAA ϩ domain sufficient to activate 54 -dependent transcription in vivo and in vitro (PspF 1-275 , see Fig. 1A), and (ii) a C-terminal helix-turn-helix domain, which binds the upstream activator sequence of the pspA and pspG specific promoters. In addition, PspF activity is negatively regulated by PspA (22)(23)(24).
Recently, we demonstrated that substitution of the highly conserved Walker B glutamate residue (Glu-108 in PspF) allowed ATP-dependent stable complex formation between PspF and 54 (or E 54 ) (25). Using a functional approach, we established roles of the Walker B Glu-108 residue in establishing nucleotide-dependent interactions between the GAFTGA motif and 54 . Our functional data, in combination with crystal structures of PspF 1-275 soaked with different nucleotides, suggest that residue Glu-108 relays ATP hydrolysis to remodel the E 54 ⅐DNA CC. Analysis of the different nucleotide-bound structures of PspF 1-275 demonstrated that a tight interaction between Walker B residues Glu-108 and Asn-64 occurs in the ATP-bound state, proposed to facilitate the exposure of the GAFTGA motif. However, Glu-108 is dispensable for ATP-dependent binding of PspF 1-275 to the CC (25). ATP hydrolysis was suggested to disrupt the E108-N64 interaction, resulting in repositioning of the GAFTGA motif (26). Despite these observations and mutagenesis studies, the precise signaling pathway relaying nucleotide hydrolysis-dependent events to OC formation remains unknown.
Sequence alignments of bEBPs show that the Walker B motif-interacting asparagine (Asn-64 in PspF) is strictly conserved, suggesting this residue may play an important role in bEBP activities. This asparagine is not present in all AAA ϩ proteins, however structural alignment of PspF with other AAA ϩ proteins demonstrates conservation of the asparagine (corresponding to Asn-64) in several proteins (Fig. 1, B and  C). Interestingly, all the AAA ϩ proteins (PspF, NtrC1, ZraR, Cdc6, Cdc6p, Ltag, RFC, RFCS, p97, Orc1, Orc2, PNK, and Rep40), which possess this conserved asparagine (Fig. 1, B and C, in red), also maintain the distance between this resi-due and the Walker B residues (Fig. 1, B and C, in green). Indeed, we note that in the case of Rep40 where the asparagine (Fig. 1, B and C, in orange) is not aligned structurally, the Walker B residues (Fig. 1, B and C, in purple) are also not aligned thereby maintaining a similar asparagine-Walker B distance as observed for other AAA ϩ proteins. These observations suggest that communication between the asparagine and Walker B residues and the distance between them could be important for protein functionality.
Understanding the communication mechanism between residues of the same or adjacent subunits of AAA ϩ proteins is important for understanding their global mechanisms of action. Determining how the positioning of the "functional motif" (GAFTGA in bEBPs), responsible for interacting with its target ( 54 ), is regulated can provide insight into how AAA ϩ proteins use ATP binding and hydrolysis, and evolve to become specialized. In this study, we investigated the contribution of residue Asn-64 to nucleotide-dependent outputs of PspF, which we used as a model system, and its role in relaying nucleotide hydrolysis to remodeling of the E 54 ⅐DNA CC.
We provide evidence for the direct contribution of residue Asn-64 in the catalytic ATPase activity and hexameric organization of PspF and demonstrate a clear role for Asn-64 in the efficient relay of ATPase activity to substrate remodeling during OC formation. In addition, we show that the negative regulation imposed by PspA on PspF ATPase activity (but not PspA binding) is dependent on Asn-64, confirming its central role in PspF functionality. Finally, we demonstrate Asn-64 variants are affected in a stage of the transcription activation process that follows 54 isomerization. We show that functionalities dependent upon Asn-64, which primarily involves interactions between PspF and 54 , are also sensitive to core RNAP enzyme and to promoter DNA conformation.
Protein Purification-PspF 1-275 proteins were purified as described (28). 54 was purified as described in a previous study (10). His-PspA was purified as described previously (29). E. coli core RNAP enzyme was purchased from Epicentre.
Filter Nucleotide Binding Assay-Nucleotide binding assays were performed in 25-l final volume containing: 20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 15 mM MgCl 2 , and 10 M PspF 1-275 variants. The mix was preincubated at 4°C for 10 min, and the reaction was started by adding 7.5 l of an ATP solution containing 0.3 Ci/l [␣-32 P]ATP (3000 Ci/mmol) or 0.3 Ci/l [␥-32 P]ATP (3000 Ci/mmol) and incubated for 10 min at 4°C. Binding reactions were then filtered through a Protan nitrocellulose 0.45-m filter (Whatman) placed on a slot blot 48-well system (Hoefler, Inc.), and a vacuum was briefly applied (10 s) to remove the liquid. After sample application, the membrane was immediately washed with 1 ml of washing buffer (20 mM Tris-HCl, pH 8.0, 50 mM NaCl, 15 mM MgCl 2 ) at 4°C. Radioactivity retained in the membrane was measured by using a phosphorimaging device (Fuji Bas-1500) and analyzed using the Aida software. All experiments were carried out at least five times, and fluctuations of binding values were up to 30% of WT values.
ATPase Activity-The ATPase activity assays were performed in a 10-l final volume, in buffer containing final concentrations of: 35 mM Tris acetate (pH 8.0), 70 mM potassium acetate, 15 mM magnesium acetate, 19 mM ammonium acetate, 0.7 mM dithiothreitol, and 5 M PspF 1-275 (or 1 M PspF 1-275 Ϯ 5.2 M His-PspA). The mix was preincubated at 37°C for 10 min, and the reaction was started by adding 3 l of an ATP solution containing 0.6 Ci/l [␣-32 P]ATP (3000 Ci/mmol) plus 0.1 mM ATP and incubated for varying times at 37°C. Reactions were quenched by addition of 5 volumes of 2 M formic acid. The [␣-32 P]ADP was separated from ATP by TLC, and radiolabeled ADP and ATP were measured by phosphorimaging and analyzed using the Aida software. Activity is expressed as a percentage of PspF 1-275 WT turnover value. All experiments were carried out in triplicate (at least), and fluctuations of turnover values were maximally 10%.
Gel Filtration through Superdex 200-PspF 1-275 WT and N64 v (at different concentrations) were incubated for 5 min at 4°C in buffer containing 20 mM Tris-HCl (pH 8.0), 50 mM NaCl, and 15 mM MgCl 2 , Ϯ 0.5 mM ATP or ADP where indicated. 50-l samples were then injected onto a Superdex 200 column (10 ϫ 300 mm, 24 ml, GE Healthcare) and equilibrated with the sample buffer with or without nucleotide. Chromatography was performed at 4°C at a flow rate of 0.5 ml/min, and columns were calibrated with globular proteins: apoferritin (443 kDa), alcohol dehydrogenase (150 kDa), bovine serum albumin (66 kDa), and carbonic anhydrase (29 kDa). All experiments were repeated at least four times, and the elution profiles obtained were similar.
␤-Galactosidase Assays-Cells were grown overnight at 37°C in LB broth containing the appropriate antibiotic and then diluted 100-fold (initial A 600 ϳ 0.025) into 5 ml of LB. Following incubation to A 600 ϳ 0.30, cultures were induced with different concentrations of arabinose for 1 h (as indicated), further grown to mid-exponential phase (A 600 ϳ 0.5-0.6) and than assayed for ␤-galactosidase activity as described before (30). Enzyme activities (in Miller units) represent the means Ϯ S.D. of the triplicate average values from at least two independent cultures.
Affinity Chromatography with Immobilized PspA-Affinity chromatography was performed at 4°C in Micro Biospin Bio-Rad columns packed with 50 l of nickel-nitrilotriacetic acidagarose (Qiagen). Solutions were passed through the columns by centrifugation at 5 ϫ g for 30 s. The columns were equilibrated with buffer A (20 mM Tris-HCl (pH 8.0), 50 mM NaCl, and 15 mM MgCl 2 ) and loaded with 500 l of 6 M His-PspA. The columns were washed with buffer A (1 ml), and purified PspF 1-275 WT or W56A or N64 v (400 l at 3.6 M) was allowed to flow through the column. Unbound proteins were removed by washing with 5 ϫ 100 l of buffer A plus 40 mM imidazole. His-tagged PspA was eluted with 2 ϫ 100 l buffer A plus 500 mM imidazole. 100-l fractions were collected, and 20 l was analyzed by 12% SDS-PAGE. Proteins were detected by Coomassie Blue staining.
Native Gel Mobility Shift Assays-Gel mobility shift assays were conducted to detect protein⅐protein or protein⅐DNA complexes. Assays were performed in a 10-l final volume containing: 10 mM Tris acetate (pH 8.0), 50 mM potassium acetate, 8 mM magnesium acetate, 0.1 mM dithiothreitol, 4 mM ADP, Ϯ NaF (5 mM) Ϯ 54 (1 M) Ϯ core RNAP (0.15 M) Ϯ 0.2 M HEX-labeled DNA probe. Where required, PspF 1-275 WT or N64 v (5 M) Ϯ AlCl 3 (0.4 mM) were added for a further 10 min at 37°C. Complexes were analyzed on a native 4.5% polyacrylamide gel. Proteins were detected by Coomassie Blue staining and fluorescent HEX-DNA was measured by phosphorimaging and analyzed using the Aida software.
In Vitro Full-length or Abortive Transcription Assays-Fulllength or abortive transcription assays were performed in a 10-l volume containing: 10 mM Tris acetate (pH 8.0), 50 mM potassium acetate, 8 mM magnesium acetate, 0.1 mM dithiothreitol, 4 mM dATP, 0.1 M core RNAP enzyme, 0.4 M 54 , and 20 nM promoter DNA. The mix was preincubated at 37°C for 5 min, and the reaction was started by addition of 5 M of PspF 1-275 WT or N64 v and incubated for varying times at 37°C. Full-length transcription (from the supercoiled Sinorhizobium meliloti nifH promoter) was initiated by adding a mix containing 100 g/ml heparin, 1 mM ATP, CTP, GTP, 0.05 mM UTP, and 3 Ci of [␣-32 P]UTP for a further 10 min. The reaction was stopped by addition of loading buffer and analyzed on 6% sequencing gels. Synthesis of the abortive transcript (UpGGG) was initiated by addition of heparin (100 g/ml), the dinucleotide UpG (0.5 mM), GTP (0.01 mM), and 4 Ci of [␣-32 P]GTP for a further 10 min. The reaction was quenched by addition of loading buffer and analyzed on a 20% denaturating gel. Radio-labeled RNA products were measured by phosphorimaging and analyzed using the Aida software.

RESULTS
To assess the contribution of the asparagine (Asn-64) to PspF activity we chose to substitute Asn-64 for: (i) alanine, to completely remove the side chain, (ii) aspartate, to add a charge but maintain the size of the side chain, (iii) glutamine, to maintain the charge but increase the size of the side chain, and (iv) serine, to reduce the size of the side chain and to alter the charge. Properties of the AAA ϩ domain of PspF (PspF 1-275 ) variants were analyzed to determine how the functionality of PspF depends on Asn-64.
Asn-64 Contributes to Nucleotide Binding, Hydrolysis, and Selfassociation-We first determined whether the PspF 1-275 Asn-64 variants (N64 v ) maintained their ability to bind and hydrolyze nucleotides (Fig. 2, A and B). The Asn-64 substitutions tested resulted in either an apparent increase (N64A, N64Q, and N64S) or decrease (N64D) in ATP binding, compared with PspF 1-275 WT (WT). Having demonstrated that all the N64 v were able to bind ATP ( Fig. 2A), we next determined whether they were affected in their capacity to hydrolyze ATP (Fig. 2B). We observed that N64S ATPase activity was not affected, whereas the other N64 v tested were all deficient for ATPase activity (N64S (100%) Ͼ N64A (36%) Ͼ N64Q (12%) Ͼ N64D (Ͻ0.4%)). We conclude that Asn-64 contributes to, but is not essential for, nucleotide binding and ATPase activity. We have previously shown that the ATPase activity of PspF 1-275 is directly related to its oligomeric state (a hexamer being the most active form) and that PspF 1-275 oligomerization is strongly stimulated in the presence of nucleotides (ATP or ADP) (28). Defects in oligomer formation are therefore predicted to negatively affect hydrolysis due to a loss in cooperativity between subunits. Because Asn-64 does not appear to be absolutely required for ATPase activity, we sought to determine whether a lack of ATPase activity was due to an effect on catalytic site formation comprising "in cis" and "in trans" residues, thereby potentially a global change in hexamer organization (28,33). To investigate the effect of N64 v on PspF 1-275 oligomerization we performed gel filtration experiments and observed that in the absence of nucleotide, the elution profiles obtained with the N64 v all differ from WT (Fig. 2C). When N64 v hexamers formed, they had the same elution position as WT, suggesting no large scale changes in structure, as seen in some other PspF 1-275 variants (28). We divided the N64 v into three different classes: (i) constitutive hexamer formation (N64S), (ii) reduced hexamer formation (N64Q and N64A), and (iii) defective hexamer formation (N64D). In the presence of nucleotide (ATP or ADP), as with WT, a stimulatory effect on hexamerization was observed with N64A and N64Q (data not shown). However, N64D was still unable to form a hexamer. Overall, the results demonstrate that Asn-64 is important in ensuring the optimal oligomerization of PspF and therefore contributes to forming the active site of the protein (see "Discussion").
Formation of a Stable PspF 1-275 ⅐E 54 Complex Is Dependent on N64-We then addressed the question whether the N64 v can form biologically relevant complexes with (E) 54 . Stable in vitro interactions between PspF 1-275 and its natural target (E) 54 have been observed using non-hydrolysable nucleotide analogues ADP-AlF x , ADP-BeF x , and AMP-AlF x (34 -36). These complexes, termed "trapped complexes," are thought to capture structural and functional intermediate conformations of the ATP-ground state and -transition state en route to OC formation (14). Using the transition state nucleotide analogue ADP-AlF x , we observed stable complexes between WT, N64A, N64Q, or N64S (but not N64D) and (E) 54 (Fig. 3A). By estimating the relative quantities of complexes formed we show a defect in complex formation with the N64 v (WT (100%) Ͼ N64A ϳ N64S (ϳ50%) Ͼ N64Q (ϳ35%) Ͼ N64D (below detection)). Interestingly, using AMP-AlF x a proposed ground-state analogue of ATP, we note a more pronounced defect in trapped complex formation by N64S than for the other N64 v (supplemental Fig. S1), suggesting that serine is not compatible with certain stable nucleotide-bound states that allow binding to (E) 54 .
The Asn-64 substitutions all diminish nucleotide-dependent interactions between PspF 1-275 and (E) 54 , suggesting that Asn-64 may be involved either directly or indirectly in the nucleotide-dependent exposure of the GAFTGA containing L1 loops that bind (E) 54 .
Asn-64 Substitutions Affect OC Formation and Not Transcription Elongation-Because all N64 v , except N64D, can interact with E 54 , we then determined whether they were capable of activating transcription using an in vitro transcription assay.
Having confirmed under the chosen assay conditions that ATP hydrolysis by PspF 1-275 was required to activate E 54 transcription (data not shown) we performed in vitro transcription assays with the N64 v . To determine the rate of transcription, we incubated the E 54 ⅐DNA (CC) and WT or N64 v with dATP for varying activation times. As shown in Fig. 3B, all N64 v are negatively affected in the initial rate of transcript formation (compared with WT at 5 min activation time, N64A (66%), N64D (not detected), N64Q (14%) and N64S (34%)). Yet, after 30 min activation time N64A produces significantly more transcript than WT (N64A (135%), N64D (not detected), N64Q (55%) and N64S (38%)), and after 60 min activation time N64Q reaches WT activity levels (N64A (170%), N64D (not detected), N64Q (180%) and N64S (50%)). In contrast to N64A and N64Q, N64S is clearly affected in the efficiency of transcription and not just slowed down, because after 60-min activation time the amount of transcript obtained remains substantially lower than WT. As expected from strong defects in self-association and ATPase, N64D did not activate transcription.
Because direct binding interactions between PspF 1-275 and (E) 54 occur, we investigated whether the reduction in transcripts formed by E 54 with N64 v was a consequence of either a defect in "activation" (OC formation) or promoter "escape" (transition to the elongating complex). To address this question, we performed abortive transcription assays to monitor OC formation. A defect in promoter escape should be accompanied by an accumulation of more abortive than fulllength transcripts. To reduce the experimental error, we performed the abortive assays with the same reaction mix used for the full-length transcription assays. After activation, the sample was divided into two equal parts and supplemented with the appropriate nucleotide mix (see "Experimental Procedures"). For all the proteins tested, we observed similar amounts of abortive transcript as full-length transcript (Fig. 3B).
We conclude that the defect in the initial transcription rates observed with the N64 v is not due to a deficiency in promoter escape but due to a fault in using ATP hydrolysis to drive OC formation. Notably, N64A and N64Q appear to be slower in OC formation (transcript levels similar to WT after longer activation times following OC accumulation), whereas N64S is slower and less efficient (because after 60-min activation time, the  MAY 16, 2008 • VOLUME 283 • NUMBER 20 amount of transcript formed is substantially lower than WT). Although N64S forms more stable complexes with (E) 54 than N64Q, it is more defective in OC formation, suggesting a problem in using target binding for remodeling E 54 .

Functional Pathway in AAA ؉ Proteins
N64 v Are Less Active Than PspF  WT in Vivo-The in vivo activities of the N64 v were then assayed to validate the results of the in vitro transcription experiments. In vivo assays were conducted using a strain lacking pspF and pspA (the negative effector of PspF, see below), in which we measured the amounts of ␤-galactosidase made by a single chromosomal lacZ gene copy under the control of pspAp using pBAD18C plasmids harboring pspF 1-275 WT or pspF 1-275 N64 v genes. At maximal PspF 1-275 induction levels transcription activities, compared with WT, were N64S (40%) Ͼ N64A (34%) with N64Q and N64D not detected, yet all N64 v had similar levels of protein production (protein accumulation, see supplemental PspA Interacts with the N64 v but Does Not Inhibit Their ATPase Activity-Because PspF ATPase activity is negatively regulated by PspA and Asn-64 contributes to PspF ATPase, we measured the sensitivity of N64 v to negative regulation by PspA using an in vitro PspF ATPase assay in the presence of purified PspA. In the presence of PspA the ATPase activity of WT is ϳ70% inhibited; however, with the N64 v no significant decrease of ATPase was observed (Fig. 4A). Importantly, direct binding interactions between PspA and WT or N64 v were observed, but not with the negative control PspF 1-275 W56A, which is specifically defective in binding to PspA (Fig. 4B). Taken together these results suggest that the repressive regulatory interaction between PspA and PspF (via residue Trp-56) occurs through Asn-64 acting to reduce the ATPase activity of PspF (see "Discussion").
The Asn-64 Side Chain Affects Productive Communication between PspF and the 54 ⅐DNA Complex-We hypothesized that the lower rate of transcription observed, in vivo (supplemental Fig. S2) and in vitro, for N64S could be due to a major defect in relaying of ATP hydrolysis to remodeling of the CC. To distinguish between interactions with 54 from E 54 , we used a 54 "supershift" assay in the absence of core RNAP. Cannon et al. (10) showed that 54 forms a stable complex with a linear promoter DNA probe harboring a mismatch at positions Ϫ12 and Ϫ11 on the non-template strand (Ϫ12Ϫ11/wt). In the presence of a hydrolysable nucleoside triphosphate (dATP), PspF 1-275 WT can convert this binary 54 ⅐DNA complex to an isomerized ss 54 ⅐DNA complex (super shifted 54 ⅐DNA com-plex), which does not contain PspF 1-275 and migrates differently on a native gel (10,25). In addition, this DNA (Ϫ12Ϫ11/ wt) is active, albeit at a significantly lower level than that observed with homoduplex DNA (0/wt), for OC formation (data not shown). As shown in Fig. 5A, the N64 v , with the exception of N64D, were all able to form the ss 54 ⅐DNA complex, although some differences were observed. N64A formed a similar amount of ss 54 ⅐DNA complex as WT, but N64S and N64Q formed clearly less ss 54 ⅐DNA complex. Interestingly, when using N64Q an additional band (C A ) was also observed. Characterization of the C A complex by UV cross-linking demonstrated the presence of N64Q, suggesting that C A is a putative intermediate in the pathway to form the ss 54 ⅐DNA complex (Fig. 6A, lanes 5 and 11). A similar complex was observed with E108D, suggesting an overlapping phenotype between N64 v and E108 v (25). Using another DNA probe (mismatched between Ϫ12 and Ϫ1 on the non-template strand; Ϫ12-1/wt) that supports ss 54 ⅐DNA complex formation by E108 v , but not by N64 v (25), N64Q forms the C A complex, suggesting C A is related to protein isomerization rather than DNA structure changes (supplemental Fig. S3).
In conclusion, with the exception of N64D, all the N64 v tested supported formation of the ss 54 ⅐DNA complex in an ATP hydrolysis-dependent manner on the Ϫ12Ϫ11/wt DNA, but not on pre-opened DNA (Ϫ12Ϫ1/wt). Interestingly, in the presence of N64Q a stable PspF 1-275 ⅐ 54 ⅐DNA complex (C A ) similar to that formed by E108D was observed, suggesting overlapping phenotypes.
PspF Activity Is Sensitive to the DNA Opening Step-We further explored the basis for the differences observed in the abilities of the N64 v to use ATP hydrolysis to remodel the CC. For N64A, the amount of ss 54 ⅐DNA complex formed is comparable to WT, although this variant is clearly affected in the rate of OC formation. For N64A the rate-limiting step in OC formation may not be 54 ⅐DNA isomerization, but a core RNAP-dependent stage, potentially involving the conformation of the promoter DNA region melted within the OC. We tested this idea using abortive transcription assays with linear promoter DNA probes reflecting the closed DNA conformation (0/wt) or open DNA conformation (Ϫ10Ϫ1/wt).
We first confirmed that the levels of abortive transcription from the linear (0/wt) and supercoiled nifH DNA were similar; demonstrating that the abortive initiation assays faithfully reflect the full-length transcription experiments (compare Figs. 3B and 5B). We then compared (Fig. 5, B and C) the activity of N64 v to WT for the initial rates of OC formation (5-min activation time) on closed (0/wt; WT Ͼ N64A Ͼ N64S ϳ N64Q Ͼ Ͼ N64D) or pre-opened DNA (Ϫ10Ϫ1/wt; N64A Ͼ WT ϳ N64S Ͼ N64Q Ͼ Ͼ N64D). In the presence of pre-opened DNA, we note a global increase in the amount of OC formed. Indeed, for N64A we observed ϳ2-fold more OC than WT, suggesting that the asparagine side chain may negatively influence a step during OC formation that is dependent on DNA conformation. In addition, the amount of OC observed with N64S on preopened DNA is similar to WT (yet ϳ3-fold lower on 0/wt DNA). Because we observed lower amounts of ss 54 ⅐DNA complex with N64S, it appears that the defect in transcription observed for N64S is most likely due to a deficiency in the isomerization of 54 (see "Discussion"). Interestingly, N64Q showed a linear increase in OC formation with time, independent of the DNA conformation used, whereas all the other N64 v and WT reached a plateau or showed reduced levels of OC formation at later time points. N64Q appears to have traded fast initial rates of OC formation for a prolonged period of activation competency.
Overall in OC formation assays the N64 v tested, compared with WT, exhibited very different sensitivities to pre-opened DNA, in initial rates and over prolonged time courses. The abilities among N64 v to maintain OC were very different at later time points. Competency in 54 isomerization was not always accompanied by an equivalent competency in OC formation, and pre-opening DNA recovered some N64 v , especially N64A, suggesting Asn-64 functions to effectively link 54 isomerization to DNA opening within the CC. 54 ⅐DNA Interactions Are Modified by N64 v -Because substituting Asn-64 alters the transcription activation efficiency of PspF 1-275 in a DNA template-dependent manner, it would seem likely that protein-DNA interactions made during remodeling of (E) 54 would be different among the N64 v . To investigate the nature of these protein-DNA interactions we employed a UV cross-linking experiment on the DNA probes (Ϫ12Ϫ11/wt and Ϫ12Ϫ1/wt) used in the 54 isomerization assays. The photoreactive DNA probes were constructed by conjugating a single, strategically placed phosphorothioate with APAB, between positions Ϫ7/Ϫ6 (Ϫ7), within the melted region in the E 54 OC (single-stranded between positions Ϫ12 and Ϫ1) and between positions Ϫ1/ϩ1 (Ϫ1), the downstream edge of the transcription bubble and the transcription start site (15,31).
Using the Ϫ12Ϫ11/wt DNA (Fig.  6A), the 54 ⅐DNA proximities at Ϫ7 or Ϫ1 are similar in the presence of WT or N64 v (except N64D), although the intensity of the cross-linked 54 species at position Ϫ7 is clearly stronger than those at Ϫ1. Cross-linking at position Ϫ7 (Fig. 6A) appears to reflect binding of 54 in initial and isomerized complexes and cross-linking at position Ϫ1 (Fig.  6A) reflecting 54 isomerization (Fig. 5A, lanes 1, 2, 4, and 5), suggesting that a range of Asn-64-dependent 54 ⅐DNA interactions are detected in these assays. Clearly a different set of interactions between 54 , N64 v , and DNA exists at, or close to, these positions (Fig. 6A, compare lanes 9 -12 and 3-6). In addition, we note in the presence of N64Q a weak PspF 1-275 ⅐DNA band at position Ϫ1, similar to that observed with E108D, further suggesting an overlapping phenotype between Glu-108 and Asn-64 (supplemental Fig. S4).
To determine whether DNA conformation could affect interactions between 54 and DNA in the presence of N64 v a pre-opened DNA probe (Ϫ12Ϫ1/wt) was used. The cross-linking pattern of 54 at position Ϫ7 is clearly changed in the presence of the N64 v compared with WT (Fig. 6B, lanes 14 -18). Significantly, cross-linked 54 at position Ϫ1 (Fig. 6B, lane 20) was comparable to WT in all the N64 v tested, except N64D (Fig.  6B, lanes 20 -24). We conclude that the interactions 54 makes with the single-stranded promoter DNA at position Ϫ7, but not at position Ϫ1, are altered in the presence of the N64 v .
Asn-64 Mutations Affect the Core RNAP-DNA Interactions-We next determined whether N64 v could alter the set of interactions that E 54 makes with DNA. When the Ϫ12Ϫ11/wt DNA is conjugated at Ϫ7, the cross-linked 54 species observed with WT (Fig. 6C, lane 26) is significantly increased in the presence of all N64 v , including N64D (Fig. 6C, lanes 26 -30). Clearly N64D can modify 54 ⅐DNA interactions, although it cannot fully remodel the CC. Notably, a very weak cross-linked core RNAP band (corresponding to the ␤/␤Ј subunits) was also observed with N64Q (identical to that of E108D, supplemental Fig. S4), suggesting that the organization of the N64Q⅐E 54 ⅐ DNA and E108D⅐E 54 ⅐DNA complexes are similar. A similar cross-linking profile for WT and N64 v was obtained when using the pre-opened Ϫ12Ϫ1/wt DNA (data not shown).
When the Ϫ12Ϫ11/wt DNA is conjugated at Ϫ1, the cross-linking profile observed with WT was clearly different to that obtained when using 54 alone (Fig. 6A, lanes 7-12).
In the presence of core RNAP and WT, the intensity of the crosslinked 54 clearly increases and an additional band, corresponding to cross-linked core RNAP is also apparent. In all the N64 v tested except N64D, which does not support OC formation, the cross-linking profile obtained was similar to that obtained with WT (Fig. 6A,  compare lanes 1 and 4). Similar cross-linking patterns were observed using the pre-opened DNA (data not shown).
Interestingly, when we used the cross-linking assay to examine the stable complexes formed between E 54 and PspF 1-275 with the nonhydrolysable ATP transition-state analogue ADP-AlF x , we observed that in the presence of either N64Q or N64D the cross-linked PspF 1-275 species was absent (Fig. 6D, lanes  44 -45). This is not surprising for N64D, because this variant was unable to form a stable complex with E 54 (Fig. 3A). However, the differences observed with N64Q likely reflects an altered organization within the N64Q⅐E 54 ⅐DNA complex when ADP-AlF x was used, because N64Q was weakly crosslinked in the 54 isomerization reactions (Fig. 6A, lane 11). The results from the cross-linking assays suggest that N64A and N64S more closely resemble WT than N64D and N64Q, with N64Q having similar properties to E108D. Overlapping properties of Glu-108 and N64 v suggest that they could each form part of the same nucleotidedependent signaling pathways. A common basis for overlapping properties may reside in interactions Asn-64 makes with Walker B motif residues.

DISCUSSION
Determining the internal communication route operating between residues of AAA ϩ proteins is key to understanding how nucleotide-dependent outputs of AAA ϩ proteins are achieved. Structural alignments of AAA ϩ proteins with a range of cellular activities demonstrate the presence of an asparagine (Asn-64 in PspF) proximal to Walker B motif residues (Fig. 1, B and C), suggesting a common importance for this arrangement. In specialized AAA ϩ proteins (bEBPs) the asparagine is highly conserved suggesting the significance of this residue in the functionality of bEBPs. In this study, using the AAA ϩ domain of the model bEBP PspF, we show that Asn-64 contributes to the internal hexameric organization and regulation of the ATPase activity. Using different substitutions, we established the importance of this residue in substrate remodeling. Several steps from the initial interaction between PspF and the CC until OC formation are affected by N64 v . Asn-64 has a critical role in regulating nucleotide-dependent contacts between PspF and its specific target, the CC, establishing the importance of this asparagine (Asn-64) in the optimal coupling of ATPase activity to OC formation.
Conserved Asparagine Affects ATPase Activity and Self-association of an AAA ϩ Protein-In this study, we showed that substituting Asn-64 causes changes in the self-association of PspF  . The gel filtration profiles suggest Asn-64 functions in the internal organization of the PspF 1-275 hexamer (Fig. 2C). We identified three different phenotypes associated with the specific Asn-64 substitutions: (i) an increase in hexamer formation as a function of PspF 1-275 concentration (N64S), (ii) a decrease in hexamer formation as a function of PspF 1-275 concentration (N64A and N64Q), and (iii) an absence of hexamerization (N64D). If we compare the different ATPase activities of the N64 v and their elution profiles (Fig. 2, B and C), we note in the absence of oligomerization (N64D), ATP hydrolysis did not occur. Yet, an increase in hexamer formation (N64S) was not accompanied by an increase in ATPase activity (100% of WT level). In addition, despite similar elution profiles for N64A and N64Q, their ATPase activities are different. These results suggest a role for Asn-64 in the catalytic activity of PspF at the level of the detailed organization of the active site. This view is supported by structural data in which the position of an activating water molecule, used for the nucleophilic attack of the ␤-␥ bond of ATP, and the Mg-ATP clearly suggests a possible involvement of Asn-64 in the ATPase activity of PspF (26).
The Asparagine Side Chain Plays a Crucial Role in the Communication between the AAA ϩ Protein and Its Target-We have identified distinct biochemical properties associated with N64 v thereby inferring the importance of residue Asn-64 in the functionality of PspF. Deletion of the Asn-64 side chain (N64A) negatively affects concentrationdependent hexamer formation and ATPase activity but not the overall binding interactions between PspF 1-275 and (E) 54 . Despite similar levels of 54 isomerization as WT, N64A showed a defect in transcription activation (Figs. 3B and 5B, after 5-min activation). This defect in transcription activation (ϳ50 -70% of WT, after 5-min activation) can be directly correlated with its low ATPase activity (Fig. 2B, ϳ36% of WT). When using preopened DNA (Ϫ10Ϫ1/wt), we observed similar amounts of OC formation as WT, suggesting that Asn-64 is involved in DNA melting and the associated loading of DNA into the RNAP during OC formation.
An increase in Asn-64 side chain length (N64Q) negatively affects concentration-dependent hexamer formation and greatly reduces the ATPase activity, binding interactions with (E) 54 and isomerization of the 54 ⅐DNA complex. In particular, N64Q affects the nucleotide-dependent contact with 54 . Indeed, a new C A complex was observed in the isomerization experiments, clearly demonstrating that N64Q affects the process of 54 ⅐DNA isomerization, thereby generating a new stable state similar to that observed with E108D (25). These results suggest that Glu-108 and Asn-64 may have interconnected functionalities. In agreement with this, the DNA cross-linking results (Fig. 6) show a similar set of interactions between E 54 and DNA in the presence of either N64Q or E108D (supplemental Fig. S4).
Reducing the Asn-64 side chain length and altering the charge (N64S) favors PspF hexamerization, had no detectable effect on ATPase activity, but reduced the nucleotide-dependent interaction with (E) 54 thereby reducing the amount of ss 54 ⅐DNA and OC formed. Because the amount of OC formed in the presence of N64S was low and never reached WT levels from the closed DNA template, we infer that N64S is negatively affected in the coupling of ATPase activity to OC formation, at least at the level of changing the 54 organization for DNA opening. The latter view is supported by the recovery of initial rates of OC formation by N64S (near WT levels) on pre-opened DNA.
Asparagine Couples PspA Binding to PspF ATPase-negative Regulation-PspA, via PspF residue Trp-56, negatively regulates PspF ATPase activity through an as yet unidentified mechanism that does not involve reduced binding of ATP (29). 4 Here we show that, although N64 v can bind PspA, their ATPase 4 N. Joly, personal communication. activities are not significantly affected, suggesting that the negative regulation imposed by PspA on PspF ATPase activity may occur via Asn-64. We propose a functional pathway that links PspA binding (to PspF) with its negative effect on PspF ATPase activity (Fig. 7). Here Trp-56 senses an interaction with PspA and relays this binding event via ␤-sheet 2, to Asn-64, altering the position of the Asn-64 side chain, ultimately effecting the distance (and potentially the coordination of the water molecule) between N64-ATP and N64-E108. In N64S, the -OH side chain could change the chemical state and/or the coordination of the water molecule (responsible for the ␤-␥ bond cleavage). Absence of the asparagine side chain may explain why N64S (and the other N64 v ) is insensitive to PspA-negative regulation. It may also explain why, although N64S can bind ATP ϳ4-fold better than WT, they have similar ATPase activity. These results further indicate the importance of the relative position of asparagine (Asn-64) to ATP and Walker B residues in the nucleotide binding pocket.
Asparagine-Walker B Distance and Optimal Coupling of ATPase Activity to Substrate Remodeling in an AAA ϩ Protein-Previous researchers (26) have proposed a model based on structural studies of PspF 1-275 crystals soaked with different nucleotides, in which they suggest that at the point of ATP hydrolysis a tight interaction between the side chains of residues Glu-108 and Asn-64 would stabilize exposure of the GAFTGA-containing L1 loop, thereby reinforcing the interaction between PspF and the E 54 ⅐DNA CC. Upon P i release, they proposed that the Glu-108 side chain rotates 90°, disrupting the Asn-64-Glu-108 interaction, causing rotation of helix 3 leading to a significant relocation of the GAFTGA-containing L1 loop into an unproductive, buried conformation. From functional data obtained on residues Glu-108 (25) and Asn-64 (this study), we revisited this model (Fig. 7). Significantly, we have established that in the absence of these residues' side chains (E108A and N64A) PspF 1-275 can still form a stable complex with 54 in the presence of ATP or ADP-AlF x , demonstrating that a stable interaction between PspF and 54 is clearly not strictly dependent on Glu-108-Asn-64 side chain interactions. Phenotypes of Asn-64 and Glu-108 substitutions, including alanine substitutions, suggest these residues are involved at different levels of a pathway coupling ATP hydrolysis to OC formation (Fig. 7). The pathway couples ATP hydrolysis to substrate remodeling by controlling the productive interaction between PspF and 54 , changes in 54 allowing DNA loading into the RNAP during transcription activation (15). Determinants of the pathway emanating from the ATPase active site may well include Asp-107 and residues controlling the positioning of the central ␤-sheet of the AAA ϩ domain (37).
Because structural alignments of AAA ϩ proteins also point to conservation of the distance between the asparagine and Walker B residue, we suggest that such organization is important for other AAA ϩ proteins. The positioning of these two residues (Asn-64-Walker B) may be required for their communication with each other and for creating a fully functional active site. The negative regulator of PspF (PspA) used this common property to control PspF activity. Indeed, PspA, via ␤-sheet 2, probably alters the position of Asn-64 affecting the optimal distance between Asn-64, ATP, and the Walker B res-idues. In consequence, the ATPase activity of PspF is reduced and transcription activation is repressed. In the case of bEBPs, the communication between the asparagine and the Walker B motif residues contributes to controlling the positioning of the L1 loop (inserted in helix 3). In AAA ϩ proteins that possess this asparagine but not the L1 loop insertion, we propose that the functional interactions between these two motifs could allow more global conformational changes in the oligomeric ring thereby regulating the functionality of the AAA ϩ protein. Indeed, changes in the positions of residues proximal to the nucleotide pocket are relatively small when different nucleotide bound states are compared. The substrate remodeling must then depend upon amplification by the hexameric assembly of the small local nucleotide-dependent changes at the subunit level. The distance between Walker B residues and the asparagine studied here seems a good candidate for enabling such nucleotide dependent changes.