The Role of Bacterial Enhancer Binding Proteins as Specialized Activators of σ⁵⁴-Dependent Transcription

Residues of the GAFTGA motif.Sequence alignments of bEBP AAA⁺ domains indicate a very high level of conservation for the GAFTGA motif (Fig. 6) (234), reflecting its importance in σ⁵⁴-dependent transcription. The first glycine residue of the motif appears to be absolutely conserved, although it has not been widely studied in bEBPs. In NtrC, random mutagenesis identified the G215V mutation, which abolished transcriptional activation both in vivo and in vitro (131). In DctD, a substitution of the equivalent residue (G220D) also produced an inactive protein in vivo (214).

Likewise, a substitution of the second amino acid of the GAFTGA motif in the bEBPs NtrC, NifA, and DctD gave rise to variants that were unable to activate transcription (86, 154, 214). However, these variants showed little or no reduction in ATPase activity, suggesting that while this residue may be required for contacting σ⁵⁴, it does not communicate with the ATP hydrolysis machinery. Interestingly, in the case of the NtrC A216C variant, the defect in transcriptional activation is relieved by additional substitutions in the helix-turn-helix motif that prevent specific binding to enhancer DNA (229). Therefore, it appears that binding to the enhancers prevents the A216C variant of NtrC from contacting σ⁵⁴, suggesting a relationship between enhancer binding and AAA⁺ function.

The role of the phenylalanine in the GAFTGA motif has been extensively studied (Table 2). The mutagenesis of this residue in NtrC (F217), NifA (F307), DctD (F222), DmpR (F312), and PspF (F85) produced bEBPs that failed to activate transcription (21, 86, 131, 214, 222). The exception is the F307Y variant of NifA, which retained 20% of its activity in vivo (86). Indeed, 16 of 248 bEBPs identified in an alignment have a naturally occurring tyrosine at this position (234). This indicates that an aromatic ring at this position is essential for transcriptional activation. To further investigate the role of the phenylalanine of the GAFTGA motif in σ⁵⁴-dependent transcription, the F85 residue of PspF was systematically replaced with 10 other amino acid residues, and the functionality of the resulting variants was assessed in vitro (234). Each of the substitutions rendered the bEBP unable to activate transcription from the nifH promoter. The F85H, F85I, F85W, F85L, F85C, and F85Q variants retained the ability to hydrolyze ATP, which is explained by their ability to self-associate. However, they were unable to interact with σ⁵⁴ to form “trapped” complexes in the presence of ADP.AlF_x, indicating that this residue is critical for bEBP-σ⁵⁴ contact. Since some variants, e.g., F85Q, showed a significant decrease in ATPase activity, it was suggested that the F85 residue communicates with the ATP hydrolysis site. In contrast, the F85A, F85E, and F85R variants showed <10% of the ATPase activity of the wild-type protein, and gel filtration indicated that this was due to a defect in their ability to form higher-order oligomers. This suggests that there is a structural and functional link between the phenylalanine and the distant interface of self-association. The PspF F85Y variant was the only variant that was able to interact with σ⁵⁴ to form the ADP.AlF_x “trapped” complex. However, it too was unable to activate transcription, and this can be explained by the inability of the protein to form activator-σ⁵⁴-DNA complexes using promoter DNA probes with a mismatch at the −11/−12 position (20, 234). Importantly, the G4L substitution in region I of σ⁵⁴ (58) can rescue this σ⁵⁴-DNA interaction defect (234). Therefore, the conserved phenylalanine of the GAFTGA motif plays a role in “sensing” the conformation of DNA at the −12 promoter position, in agreement with a model based on recent cryo-EM reconstructions (23). Overall, studies using different bEBPs suggest that the conserved phenylalanine has multiple, interrelated roles during transcriptional activation. The conserved threonine has been shown to contact region I of σ⁵⁴ (see below), and the adjacent phenylalanine is also critical for this interaction. Previous studies suggested that F85 stabilizes L1-region I interactions indirectly through the positioning of T86 (21), and therefore, its major role is likely to be in contacting the promoter DNA rather than σ⁵⁴.

The role of the conserved threonine in the GAFTGA motif is well understood (Table 2). Substitutions of the T218 residue of NtrC, the T308 residue of NifA, and the T223 residue of DctD abolish the ability of the bEBP to activate transcription (86, 131, 214). The same is true for the T85A and T85V substitutions in PspF, but T86S remains partially active (49). Strong evidence for a direct interaction between the threonine residue of the GAFTGA motif and region I of σ⁵⁴ was provided by the identification of substitutions (e.g., G4L in region I) that specifically suppress the defects of the partially active T86S variant (20, 49, 58). Significantly, the G4L substitution allows the T86S variant to interact with but not activate the Eσ⁵⁴ (G4L) complex when the promoter DNA is “premelted.” This supports a role for the GAFTGA motif in the “sensing” of the promoter DNA conformation downstream of the −10 position. It has been suggested that a communication of this information to σ⁵⁴ via region I might allow Eσ⁵⁴ to establish contact with single-stranded DNA, which is required for open complex formation (58).

The role of the second glycine of the GAFTGA motif has been less well studied. The G219K variant of NtrC showed only a 50% reduction in activity but failed to initiate open complex formation. In agreement with this, the equivalent substitution in NorR (G266K) was unable to activate transcription in vivo (36). Surprisingly, The NtrC G219K variant showed improved DNA binding properties (154). In contrast, the G219C and G266C variants of NtrC and NorR, respectively, were competent to activate transcription (36, 131), and for NtrC, increased ATPase activity was demonstrated, which is most likely due to increased oligomerization (131). However, as is the case for the A216C variant of NtrC, binding to enhancer DNA prevents the G219C variant of NtrC from activating transcription, and the ability of the variant to activate transcription is restored in a form of the bEBP defective for DNA binding (229). Taken together, the data for NtrC suggest that the binding of the C-terminal DNA binding domain to enhancers influences substrate remodeling by the GAFTGA motif. A more comprehensive mutational analysis of this conserved glycine residue was conducted with NorR, where most substitutions of the equivalent residue (G266) prevented the bEBP from activating transcription in vivo. However, the replacement of G266 with either C, S, Q, or M resulted in variants that were competent to activate transcription in response to the cognate signal (nitric oxide [NO]). Interestingly, the G266D and G266N substitutions gave rise to high levels of deregulated transcription in vivo. The discovery of such variants led to the proposal of a novel mechanism for the regulation of bEBP activity (see below) (36).

In common with the other residues of the GAFTGA motif, studies of NtrC, NifA, DctD, and DmpR revealed that the second alanine residue is critical for σ⁵⁴-dependent transcription. The A220T and A220V variants of NtrC (131, 154); the A310N, A310D, and A310G variants of NifA (86); the A225T variant of DctD (214); and the A315T variant of DmpR (222) all fail to activate transcription. Only the A310S variant of NifA exhibits activity, although this is less than 20% of the activity of the wild-type protein (86). Many of the variants at this position are still able to hydrolyze ATP, and it is therefore likely that these substitutions destabilize the interaction between loop 1 and region I of σ⁵⁴ that forms at the regulatory center in the closed complex.

The ground state of hydrolysis.The ATP-bound “ground” state of the nucleotide cycle has been examined by soaking crystals of PspF_1–275 with ATP, either in the absence of Mg²⁺ to prevent hydrolysis or by using a hydrolysis-defective variant of the bEBP (22, 29, 173). The structures reveal the residues responsible for binding ATP within the nucleotide binding pocket (Fig. 8A). The Walker B glutamate (E108 in PspF) senses the γ-phosphate of ATP and forms a strong interaction with a nearby highly conserved asparagine (N64 in PspF). The adjacent aspartate (D107 in PspF) coordinates the position of a water molecule for nucleophilic attack on this γ-phosphate (173). In NtrC1, the conserved R finger from the adjacent protomer (R299 in NtrC1 and R168 in PspF) appears to engage with the γ-phosphate, stabilizing the binding of ATP (51). Significantly, the ATP-bound structures of PspF and NtrC1 indicate that L1 and L2 are in a raised conformation, consistent with low-resolution SAXS-derived structures using the ground-state analogue ADP-BeF_x in NtrC1 (50, 51, 173). Biochemical experiments confirmed that both PspF and NtrC1 can establish contact with σ⁵⁴ in the presence of this ground-state analogue (22, 50). Taken together, these data indicate that the binding and sensing of the nucleotide cause significant conformational changes in the bEBP AAA⁺ domain. The combined evidence to date suggests that GAFTGA-containing L1, assisted by L2, is released to make an initial, unstable interaction with Eσ⁵⁴ (Fig. 9).

• Open in new tab
• Download powerpoint

Fig 8

(A and B) Structure of monomeric PspF_1–275 bound to ATP (PDB accession number 2C96/2C9C) (A) and ADP (PDB accession number 2C98/2C9C) (B). Important motifs are highlighted: Walker A (brown), Walker B residue E108 (cyan), sensor I residue T148 (magenta), and “switch” residue N64 (yellow). Loop 1 (L1) and loop 2 (L2) are labeled, with the nonresolved fold of L1 indicated with a red dotted line. (C) Switching mechanism of the Walker B E108 residue. In the ATP-bound state, E108 interacts with N64 (indicated by a black dotted line). In the ADP-bound state, there is a 90° rotation in the N64 side chain so that E108 interacts with sensor I residue T148 via a water molecule (indicated by a red dotted line).

• Open in new tab
• Download powerpoint

Fig 9

Summary of the nucleotide-driven conformational changes that occur during ATP hydrolysis, as proposed for PspF (173). The ground (blue), transition (red), posthydrolysis (green), and released (purple) states are indicated. For simplicity, only the “switch” interactions are shown with the associated relocations of linker 1, helix 3 (H3), and L1/L2. (Adapted from reference 173 with permission from Elsevier.)

The transition state of hydrolysis.Studies using the transition-state analogue ADP.AlF_x have established a clear role for the GAFTGA motif of loop 1 in contacting region I of σ⁵⁴ prior to the remodeling of the closed complex (21, 39, 49). Most significantly, a stable interaction between GAFTGA-containing L1 and σ⁵⁴ is formed in the presence of the transition-state analogue, as revealed by the fitting of the PspF_1–275 crystal structure into the cryo-EM reconstruction of the activator in complex with σ⁵⁴ (172, 173). In agreement with this, low-resolution structures of NtrC and NtrC1 bound to ADP.AlF_x reveal an electron density above the plane of the oligomeric bEBP corresponding to a raised position of L1 and L2 (50, 62). Significantly, the ADP.AlF_x complex of NtrC1 was shown to be more stable than the ADP-BeF_x complex, indicating that ATP hydrolysis strengthens the unstable interaction between the bEBP and σ⁵⁴ that forms in the ground state (50). Similar results have been obtained for PspF (33). In addition, this form of the activator is able to initiate the early stages of promoter “melting” (34). Moreover, in the transition state, when bound to ADP.AlF_x, PspF is competent to induce the Eσ⁵⁴ holoenzyme to initiate RNA synthesis on single-stranded promoter DNA, identifying the prehydrolysis state of bEBPs as being functionally important (32). However, the “trapped” forms of PspF are largely defective for nucleotide hydrolysis, do not promote DNA melting, and are incapable of activating transcription on double-stranded DNA (dsDNA) (32, 33, 49), suggesting that the continuation of the cycle of nucleotide-driven conformational changes is vital for the formation of the open complex.

The postnucleotide hydrolysis state.Crystal structures of various bEBPs in their ADP-bound forms have provided extensive information regarding the nature of the “posthydrolysis” state (128, 172, 173, 181). Comparisons of the ADP-bound forms with ground- and transition-state structures have shed light on the conformational rearrangements that occur upon ATP hydrolysis (Fig. 8 and 10). Structures of PspF indicate that the release of the γ-phosphate to form ADP causes a 90° rotation of the glutamate side chain of the Walker B motif (E108 in PspF). As a result, the interaction between the glutamate and the conserved asparagine (N64 in PspF) is broken, resulting in the Walker B glutamate interacting instead with the sensor I threonine residue (T148 in PspF) via a water molecule (Fig. 8C). Communication of the altered position of Walker B by the asparagine residue leads to conformational changes in helix 3 (H3) and helix 4 (H4) that are translated to loop 1 (L1) and loop 2 (L2) via strategically placed residues in the central domain. The functional significance of these key residues has now been assessed, and it has emerged that both intra- and intersubunit interactions have a role in modulating the conformation of the σ⁵⁴ interaction surface (Fig. 10) (110). The structure of ADP-bound PspF_1–275 suggests that the Walker B-asparagine “switch” results in the disruption of an interaction between the R131 residue of L2 and the E97 residue of H3 (Fig. 10A) (173). Single substitutions at these positions abolish ATPase activity and σ⁵⁴ interactions, while residue swaps (R131E/E97R) partially restore activity, demonstrating the importance of this polar interaction (110). Following this, the E97 residue was proposed to interact with R91, while the R131 residue is thought to contact L1 residue E81. Variants of R91 in PspF were unaltered for ATPase activity, and although they showed only a slight decrease in the ability to contact σ⁵⁴, they were significantly less able to form open promoter complexes, indicating that this residue is important for substrate remodeling after the initial interaction (110). This arginine residue does not show a high level of conservation in the bEBP subfamily of AAA⁺ proteins, but in many cases, the adjacent residue may serve a similar function. Together, these new interactions result in the compaction of the loops down toward the surface of the AAA⁺ domain (Fig. 10), enabling σ⁵⁴ relocation, which is crucial to the conversion from the closed to the open complex (22, 50, 173).

• Open in new tab
• Download powerpoint

Fig 10

In cis and in trans interactions predicted to form during nucleotide hydrolysis in PspF. Interactions in cis center around the E97 (green) residue, which interacts with either R131 (blue) (ATP state) or R91 (purple) (ADP state), depending on the position of the Walker B-asparagine “switch.” Upon γ-phosphate release, E97 breaks its interaction with R131, allowing R131 to instead contact R81 (red). These new interactions result in the compaction of L1 and L2 down toward the surface of the AAA⁺ domain, enabling σ⁵⁴ relocation. Interactions in trans center around the E130 residue (orange) (subunit n), which contacts the R98 residue (magenta) (subunit n+1) in the ATP-bound state but which interacts with the R95 residue (cyan) (subunit n+1) in the ADP-bound state. (A) Region of the crystal structure of PspF_1–275 in the ATP-bound form (PDB accession number 2C96) and the ADP-bound form (PDB accession number 2C98) showing the locations of the key residues involved in inter- and intrasubunit interactions. Interactions occurring within the protomer are indicated by double-headed arrows. (B) Schematic showing the in cis and in trans interactions that occur before and after γ-phosphate release. Interactions are indicated by arrows, and residue colors correspond to those described above for panel A. (Adapted from reference 110 with permission of the publisher.)

Interestingly, it has now been shown that intersubunit (in trans) interactions confer cooperativity in the nucleotide-dependent substrate remodeling of PspF (Fig. 10B) (110). In the ATP-bound state, the E130 residue at the base of L2 was proposed to interact in trans with R98 of H3 from the adjacent subunit. Upon the release of the γ-phosphate, the E130 residue is expected to interact instead with R95 of H3, also from the adjacent protomer. Substitutions made at these positions cause the uncoupling of ATPase activity and substrate remodeling, since these variants are able to oligomerize and hydrolyze ATP but are not competent to contact σ⁵⁴. Therefore, the “switching” of this in trans interaction between protomers is thought to contribute to the coordination of L1 and L2 during nucleotide hydrolysis (110). These residues do not show strict conservation in the bEBP subfamily, but similar interactions may play a significant role in the coupling of ATP hydrolysis to substrate remodeling. For example, in NtrC1, the equivalent residues (F226, L229, and Y261) are thought to facilitate in trans hydrophobic contracts (51).

The Walker B-asparagine switch.Analyses of active-site structures reveal that the glutamate “switch” residue of the Walker B motif (E108 in PspF) is a common feature of hydrolysis in the majority of AAA⁺ proteins (237). Whereas substitutions of the adjacent Walker B aspartate (D107) have severe effects on various aspects of PspF activity, in line with a key role for this residue in ATP hydrolysis, a substitution of the adjacent glutamate (E108 in PspF) has only moderate effects on activity. Such variants have been used to effectively study intermediate states en route to open complex formation and confirm the pivotal role of the Walker B glutamate in transmitting nucleotide-dependent conformational changes (116). Sequence alignments of bEBPs indicate that the glutamate switch-interacting asparagine (N64 in PspF, labeled as “Switch Asn” in Fig. 6) is strictly conserved, in agreement with the hypothesis that this residue plays a key role in these specialized activators (112). In PspF, variants of the N64 residue have altered oligomeric states and ATPase activities, suggesting that the conserved asparagine plays a significant role in the organization of the active site. In line with this, the positions of Mg²⁺-bound ATP and the water molecule involved in the nucleophilic attack of the γ-phosphate correspond with a catalytic role of the asparagine in ATP hydrolysis (173). Significantly, in the absence of the glutamate or asparagine side chains (i.e., in the E108A or N64A variant), PspF was shown to form a stable complex with σ⁵⁴, indicating that the interaction between the activator and σ⁵⁴ is not strictly dependent on the Walker B glutamate or switch asparagine residues (112, 116). However, the N64A variant was significantly less able to form open complexes than the wild-type activator. This defect was suppressed when premelted DNA was used, suggesting that the conserved asparagine is also required for the melting of the promoter DNA and the associated loading of the template into the RNAP active site (112). Despite the Walker B-interacting asparagine not being present in all members of the AAA⁺ superfamily, structural alignments of AAA⁺ proteins suggest that the distance between the glutamate and this residue is conserved. Consequently, it was suggested that the positioning of these two residues is important for their communication with each other and for forming a fully functional catalytic site. In the case of the specialized bEBP family, this communication helps to control the positioning of GAFTGA-containing L1. In AAA⁺ proteins that do not contain the GAFTGA insertion, it is likely that the interaction between these residues controls similar nucleotide-dependent conformational changes that regulate functionality in the oligomeric ring (112).

The arginine finger-directed “switch.”A recent report of a high-resolution crystal structure of the ATP-bound NtrC1 central domain (NtrC^C) and comparison with the ADP-bound form identified an alternative mechanism for the coupling of hydrolysis to substrate remodeling (31, 51, 128). In order to examine the configuration of an ATP-bound activator, the Walker B glutamate was replaced with alanine, allowing NtrC1 to bind but not to turn over the nucleotide. This led to the first structure of a bEBP in which the highly conserved arginine (R) finger (R299 in NtrC1) is seen to contact the γ-phosphate (Fig. 11). This is in contrast to the ATP-bound structure of monomeric PspF_1–275, where the alternative R finger residue R162 (R293 in NtrC1), rather than R168 (R299 in NtrC1), is predicted to be in close proximity to the γ-phosphate of the adjacent protomer (173). A comparison of the ATP- and ADP-bound NtrC1 structures indicates that the engagement of the γ-phosphate by the R299 R finger stimulates a rearrangement of interaction networks in the same protomer (in cis), at the R finger side of the interprotomer interface. It has been proposed that the interaction of this R finger with the γ-phosphate causes helical distortions that ultimately lead to the transition of L1 and L2 to a raised conformation. The K250 residue appears to be particularly important in this transition; the side chain exists in distinct environments in the two nucleotide-bound states (51). This model is in contrast to prior studies of PspF, which indicated that the Walker B-asparagine “switch” is responsible for modulating the conformation of the σ⁵⁴ interaction surface in the protomer that contains the Walker A and B residues of the ATP hydrolysis site (112, 116).

• Open in new tab
• Download powerpoint

Fig 11

Comparison of the ADP-bound and ATP-bound structures of the AAA⁺ domains of NtrC1 and the NtrC1 E339A variant, respectively. In each structure, the proposed R fingers are indicated (R293 and R299) in dark green. The sensor II arginine (R357) is shown in red. The Walker B “DE” residues (D238 and E239) are shown in cyan. The E242 (orange) and E256 (blue) residues may form interprotomer interactions. The asparagine “switch” proposed for PspF is shown in yellow (N195 in NtrC1). The K250 residue (magenta) exists in two distinct conformations depending on whether ATP or ADP is bound; this residue is proposed to be important for mediating the transition of loop 1 (L1) and loop 2 (L2) to a raised conformation. (Left) The ADP-bound structure of the NtrC1 AAA⁺ domain (PDB accession number 1NY6 [B-C interface] [128]). (Right) The ATP-bound structure of the NtrC1(E239A) AAA⁺ domain (PDB accession number 3MOE [51]). The magnesium ion is shown as a light pink sphere, and the active water molecule is shown as a blue sphere.

Model of nucleotide-driven conformational change.Overall, structural and biochemical studies have shown that the coupling of ATP hydrolysis to open complex formation by Eσ⁵⁴ is dependent upon conformational changes in the AAA⁺ domain (Fig. 9) (22, 51, 173). These changes center on the “sensing” of the γ-phosphate by either the asparagine “switch” residue that detects changes in the Walker B motif upon ATP hydrolysis (173) or the highly conserved R finger of the adjacent protomer (51). Depending upon the stage of the ATP hydrolysis cycle and the position of the “switch” or R finger, GAFTGA-containing L1 and L2 adopt various conformations (51, 110). Upon nucleotide binding, the loops are in an extended conformation, and the GAFTGA motif forms an unstable interaction with σ⁵⁴. ATP hydrolysis strengthens this interaction, and a remodeling of the holoenzyme can then occur to enable open complex formation. Upon phosphate release, rearrangements of both in cis and in trans interactions cause the loops to disengage the σ factor. The cycle can then complete with the exchange of ADP for ATP. The GAFTGA motif thus performs a crucial role in the “power stroke” of bEBPs in the coupling of ATP hydrolysis to conformational rearrangements of the σ⁵⁴-RNA polymerase.

Asymmetry in the bEBP oligomer.Structural studies of the activator bound to different nucleotides have helped establish the conformational changes within the AAA⁺ domain that couple nucleotide hydrolysis to σ⁵⁴ contact (173). The ADP.AlF_x-bound cryo-EM structure of PspF_1–275 indicates that not all protomers within the AAA⁺ hexamer contact σ⁵⁴ during the transition state of ATP hydrolysis, indicating asymmetry (172). In line with this, a number of heterohexameric AAA⁺ proteins exist (e.g., eukaryotic MCM2 to MCM7), comprising up to six different proteins, strongly suggesting that each subunit may have a distinct role in the activity of the hexamer (17, 77). Studies of the homohexameric bEBP PspF have subsequently confirmed that an asymmetric configuration is a key requirement for open complex formation (111). The introduction of the GAFTGA substitution T86A into single-chain forms of PspF with two or three subunits allowed Joly and Buck (111) to examine the minimal requirements for σ⁵⁴ contact and substrate remodeling. This substitution was shown to uncouple the ATPase and oligomerization activities of the bEBP from its ability to contact and remodel σ⁵⁴. The minimal configuration for a stable interaction with σ⁵⁴ was two adjacent functional subunits, revealing that more than one GAFTGA-containing L1 is likely to contact σ⁵⁴ at the point of ATP hydrolysis (111). However, a covalently linked arrangement of mutant and wild-type subunits in which wild-type protomers were opposite from each other in the ring gave rise to a hexamer that was markedly less able to interact with σ⁵⁴ or form open complexes. This result strongly suggests that asymmetry in the hexamer is important for its ability to contact and remodel σ⁵⁴, which is itself asymmetrical. Significantly, efficient open complex formation (i.e., the remodeling of σ⁵⁴ rather than just an interaction with σ⁵⁴) requires at least one more additional subunit, revealing that the minimal requirements for a stable interaction with σ⁵⁴ are different from those for substrate remodeling (111).

Heterogeneous nucleotide occupancy.The finding that only a subset of the GAFTGA-containing L1 loops is required to mediate σ⁵⁴ interactions and open complex formation in turn reveals that only a subset of ATP hydrolysis sites is required for bEBP activity (111). In support of this, work carried out to determine how ATP hydrolysis between protomers is coordinated suggested that ATPase activity in PspF is partially sequential (117). In the AAA⁺ family of proteins, two models of nucleotide occupancy may explain how hydrolysis is coordinated (Fig. 12) (2). Homogeneous nucleotide occupancy has been observed for a number of AAA⁺ protein crystal structures (79, 129, 236), supporting a model of concerted/synchronized hydrolysis (Fig. 12D), in which subunits of the AAA⁺ ring simultaneously hydrolyze ATP. Other AAA⁺ structures show mixed nucleotide occupancy with ATP, ADP, or no nucleotide bound at the catalytic sites between subunits (18, 211). This supports a model of sequential or rotational hydrolysis (Fig. 12B and C), in which heterogeneous nucleotide occupancy is coordinated between protomers. Studies using the bEBP PspF revealed that either ATP or ADP stimulates the oligomerization of the activator and that physiological ADP concentrations stimulate the ability of the protein to hydrolyze ATP. This suggests that in PspF, ADP binding promotes the formation of the stable hexamer and causes structural changes leading to increased ATPase activity in adjacent protomers. Furthermore, where a nonoptimal binding of nucleotides occurs, there are negative homotropic effects (184). High ATP concentrations at which every catalytic site in the hexamer is likely to be in the ATP-bound form inhibit ATP hydrolysis and the activation of transcription in vitro. Taken together, these data strongly suggest that heterogeneous nucleotide occupancy, coordinated between protomers in the hexameric ring, plays a crucial role in the activation of σ⁵⁴-dependent transcription by bEBPs. Therefore, the concerted/synchronized model of hydrolysis can be discounted (Fig. 12D). Because of the “ATP inhibition” and “ADP stimulation” of PspF ATPase activity, nucleotide hydrolysis in bEBPs is unlikely to be a stochastic process in which each catalytic site is independent (Fig. 12A). Rather, the arrangement of the hydrolysis site and in vitro data suggest that cooperativity exists between protomers of the hexamer (117). Indeed, R-finger residues have been shown to function in trans, coordinating the bound nucleotide in the adjacent subunit of the hexamer (27, 51, 87, 235). Furthermore, recent mutagenesis studies of PspF have identified non-R-finger residues involved in interprotomer interactions that help to coordinate the positions of L1 and L2 during ATP hydrolysis (110). Subsequently, a model of sequential or rotational hydrolysis is favored (Fig. 12B and C). These mechanisms would create an asymmetry in the exposure of GAFTGA motifs in the hexamer, which has been shown to be important for the mechanical action of the activator (111, 117). However, the exact number of nucleotides bound and the conformation of individual protomers at discrete steps of hydrolysis still remain unclear. Determinations of high-resolution crystal structures of mixed nucleotide-bound hexamers would therefore be beneficial.

• Open in new tab
• Download powerpoint

Fig 12

Models for the coordination of nucleotide hydrolysis between protomers in the AAA⁺ hexamer. In solution, the majority of bEBPs exist in equilibrium between dimeric and hexameric forms. In the case of PspF, binding of ATP or ADP promotes hexamerization and ATP hydrolysis. (A) The stochastic model assumes that each catalytic site is independent. (D) The synchronized/concerted model is based on homogeneous nucleotide occupancy and assumes that each site simultaneously hydrolyzes ATP. (B and C) Data for PspF suggest that ATP hydrolysis occurs via either the rotational (B) or sequential (C) mechanisms that utilize heterogeneous nucleotide occupancy. Both models are based on cooperativity between protomers in the hexamer. (Adapted from reference 117 with permission of the publisher.)

Hexamers and heptamers.Recently, the exact functional oligomeric state of bEBPs has become a matter of debate. The first structure determined for an activator of σ⁵⁴-dependent transcription was that of the isolated ADP-bound ATPase domain of the NtrC1 protein from the extreme thermophile Aquifex aeolicus (128). The 3.1-Å structure revealed a heptameric ring with a height and a diameter of 40 Å and 124 Å, respectively. The recently reported crystal structure of the ATP-bound form of NtrC1 confirms a heptameric arrangement, and furthermore, negative-stain EM of the Walker B mutant derivative (that can bind but not hydrolyze the nucleotide) shows that it is competent to form a complex with σ⁵⁴ (31, 51). Additionally, low-resolution WAXS/SAXS NtrC1 structures reveal a seven-membered ring (50, 51), indicating that the heptamer is not simply an artifact of crystallization. However, other bEBP structures have revealed hexameric arrangements (Table 3). When the crystal structure of the isolated ATPase domain of PspF (PspF_1–275) is fitted into the cryo-EM structure of the activator in complex with σ⁵⁴, the electron density can accommodate six monomers (172). Indeed, electrospray ionization-mass spectrometry (ES-MS) shows that six monomers of PspF_1–275 form a complex with monomeric σ⁵⁴, consistent with bEBPs functioning as hexamers (133, 155). The 3-Å X-ray structhe zinc-responsive protein ZraR from Salmonella enterica serovar Typhimurium lacking the N-terminal regulatory domain also reveals a hexameric arrangement (181). These more recent studies have called into question whether the heptameric configuration of NtrC1 (128) represents a physiologically relevant form of the bEBP. In the absence of the C-terminal (DNA binding) and N-terminal (regulatory) domains, the stoichiometry of the full-length NtrC1 protein is unknown. In addition, the odd number of subunits does not match up with the dimeric arrangement of the receiver domains in both active and inactive forms (66, 128). To resolve the effect that different domains have on the oligomerization state of σ⁵⁴-dependent activators, a more thorough analysis was conducted by using different domain combinations of the NtrC4 protein from Aquifex aeolicus (12). As is the case for NtrC1, the ATPase activity of NtrC4 is subject to negative regulation. The assembly of the active oligomer is repressed by the receiver domain, and phosphorylation is likely to remove this repression (11). Unlike NtrC1, NtrC4 has a partially disrupted receiver-AAA⁺ domain interface and can assemble into active oligomers at high protein concentrations, independent of phosphorylation (11). ES-MS experiments showed that full-length NtrC4 (NtrC4^RCD) and activated NtrC4 lacking the DNA binding domain (NtrC4^RC) form hexamers. In contrast, the isolated ATPase domain (NtrC4^C), nonactivated NtrC4 lacking the DNA binding domain (NtrC4^RC), and NtrC4 lacking the regulatory domain (NtrC4^CD) all form heptamers. A heptameric arrangement for the central ATPase domain in isolation is consistent with the heptamer observed when this domain of NtrC1 is crystallized (128). Therefore, it seems that for the extreme thermophile Aquifex aeolicus, a heptamer is the most stable arrangement for the AAA⁺ domain in the absence of regulatory and DNA binding domains. This is in contrast to the central domain of PspF, which forms hexamers when in isolation (172), and therefore, despite the high level of conservation of AAA⁺ domains, the PspF and NtrC1/NtrC4 central domains must have some differences. Interestingly, when the regulatory domain of NtrC4 is absent, a heptamer is formed, but when it is present and activated by phosphorylation, hexamerization occurs. Since the activated receiver domain stabilizes the hexameric form of NtrC4, it appears that an intermediate mechanism of regulation exists, somewhere between the negative mechanism of NtrC1/DctD and the positive mechanism of NtrC (11, 12). Overall, studies examining the oligomeric state of NtrC4 have conclusively shown that truncated or nonactivated proteins may have a propensity to exhibit altered stoichiometries. Although structural studies of ZraR suggested the possibility that the AAA⁺ domain is held in a hexameric configuration by the DNA binding domains, which are dimeric in nature (181), a truncated form of NtrC4 containing both the central domain and the C-terminal domain (CTD) is heptameric. Due to the difficulty in activating full-length NtrC1, the oligomeric state of this construct cannot be assessed, but the similarity of NtrC4 to NtrC1 suggests that the activated form of NtrC1 is likely to be hexameric. In addition to bEBPs, heptameric arrangements have also been observed frequently for other members of the AAA⁺ family (Table 3), such as MCM (56, 57, 232), RuvB (142), ClpB (3, 126), magnesium chelatase (174), HslU (178), Lon (198), and the C-terminal domain of p97 (61). Significantly, hexamers have also been observed for each of these proteins (56, 57, 142, 159, 223, 232, 236). Various explanations exist for the existence of two different isoforms of the same AAA⁺ protein. In the case of RuvB from Thermus thermophilus and the MCM protein from Methanothermobacter thermautotrophicus, heptamers form in the absence of DNA, but hexamerization occurs when DNA is present (56, 57, 142, 232). It is possible that the heptamer facilitates the loading of DNA into the central channel of the protein ring before the loss of a single subunit results in the hexamer (232). The bacterial protein-disaggregating chaperone ClpB forms heptamers in the absence of a nucleotide but undergoes rearrangements to form hexamers when ATP or ADP binds (3, 126). This implies that during the ATP hydrolysis cycle (as ATP binds and is hydrolyzed and ADP is released), there is “switching” between hexameric and heptameric states. This partial ring dissociation has been suggested to facilitate the “prying apart” of aggregated substrates (3). In the case of the ATPase HslU, rings of 7-fold and 6-fold symmetry have apparently been observed under the same conditions (178). Here, it is unclear as to whether the heptameric form of the protein is competent to associate with the partner protease, HslV. However, it was suggested that the symmetry mismatch between a heptameric HslU and a hexameric HslV may facilitate the loading of the substrate into the proteolytic chamber, as was suggested for ClpAP (125). Unlike the ATP-dependent proteases HslUV and ClpAP, the Lon protease combines both proteolytic and ATPase functions within a single subunit. The Lon protease from Saccharomyces cerevisiae has been reported to consist of seven flexible subunits (198), possibly reflecting the requirement for a “mismatch symmetry” of the HslUV and ClpAP systems. However, electron microscopy of the Lon protease from Escherichia coli indicates a hexameric arrangement (159). Likewise, the magnesium chelatase subunit BchI from the proteobacterium Rhodobacter capsulatus has 6-fold symmetry (223), but the equivalent subunit in the cyanobacterium Synechocystis has 7-fold symmetry (174). This suggests that it may also be possible for the same AAA⁺ protein to have different oligomeric states in different organisms or evolutionary groups. Finally, as described above for the bEBP NtrC4 (12), proteins may be heptameric when expressed as isolated AAA⁺ domains but hexameric in their intact forms. For example, when crystallized, the isolated C-terminal (D2) domain of p97 reveals a 7-fold symmetry (61), but cryo-EM studies of the full-length form indicate that, in fact, the protein has a hexameric arrangement (236).

Phosphorylation.Many bEBPs are part of two-component systems (TCSs) that couple an external stimulus to an internal response (199). Such systems are commonly composed of a histidine protein kinase (HK) with a conserved kinase core domain and a response regulator (RR) protein with a conserved regulatory domain. Extracellular stimuli are sensed by the HK to modulate its activity in phosphotransfer. The HK transfers a phosphoryl group to a conserved aspartate in the RR (a reaction catalyzed by the RR), and the phosphorylated RR is able to activate a downstream effector domain that elicits a specific response in the bacterial cell. For example, the bEBPs NtrC, NtrC1, NtrC4, DctD, ZraR, and FlgR all have RR domains (Fig. 5) that are phosphorylated by specific sensor kinases. The best-studied RR in this group is the bEBP NtrC, which is phosphorylated at the conserved D54 residue by the sensor kinase NtrB in response to the nitrogen status of the cell (175). Briefly, the phosphorylation cascade is controlled by the uridylyltransferase (GlnD), which transmits the nitrogen status to the NtrB protein via the PII protein GlnB. Under nitrogen-limiting conditions, NtrB phosphorylates NtrC, activating it as a bEBP. Under nitrogen-excess conditions, the phosphatase activity of NtrB prevents NtrC activation (reviewed in reference 65). The NtrC1 and NtrC4 bEBPs from the thermophile A. aeolicus, which have been studied extensively at the structural level, are both classed as NtrC family members based on their high amino acid similarity (59%), but their cognate HKs and the signals controlling these two-component systems have not yet been identified (63, 128). The σ⁵⁴-dependent DctD RRs in Sinorhizobium meliloti and Rhizobium leguminosarum (140, 233) respond to phosphorylation by the HK DctB and activate the expression of DctA, a transport protein that allows bacteria to use C₄-dicarboxylic acids for growth as free-living cells or to power nitrogen fixation within symbiotic bacteroids (158). The ZraR RR is another example of a group I bEBP, which is phosphorylated by the HK ZraS in response to high Zn²⁺ concentrations. The activation of the ZraR effector domain results in the expression of a periplasmic Zn²⁺ binding protein, ZraP (130). In contrast, the FlgR RR from Helicobacter pylori is a group V bEBP (Fig. 5), which lacks a DNA binding domain (discussed below). This activator is phosphorylated by the sensor HK FlgS, leading to the transcription of genes required for flagellar biosynthesis (26, 197).

Ligand binding.Other σ⁵⁴ activators have a regulatory domain that binds small effector molecules (Fig. 5). The direct binding of aromatic compounds to a vinyl 4 reductase (V4R) domain activates the group II bEBPs DmpR and XylR (161, 189, 194). In response to small-ligand binding, DmpR activates the expression of the dmp operon, which encodes enzymes involved in the catabolism of phenol and methylphenols (191–193). XylR binds to toluene, m-xylene, and p-xylene to activate transcription at the Pu promoter of the TOL plasmid, allowing Pseudomonas putida to grow on toluene and related hydrocarbons (64). N-terminal-domain swaps between the XylR and DmpR proteins confirm that the specificity of the response is conferred by the regulatory domain (193). Another regulatory domain frequently found in bEBPs is the GAF (cyclic GMP [cGMP]-specific and -stimulated phosphodiesterases, Anabaena adenylate cyclases, and E. coli FhlA) domain (Fig. 5), a member of a large and diverse domain family that is found in all kingdoms of life (5). FhlA contains two GAF domains that bind formate to activate the transcription of the formate hydrogen lyase system (100). NorR contains a single GAF domain, which binds NO to activate the transcription of the norV and norW genes, enabling NO detoxification (60, 106). The Per, ARNT, and Sim (PAS) domain (168) often detects signals via a bound cofactor such as heme or flavin (204) and, like the GAF domain, is present in many bEBPs. Although the PAS and GAF domains share little sequence similarity, they have similar structures and may share an ancestor (97, 168). Finally, the aspartokinase, chorismate mutase, and TyrA (ACT) domain is common in metabolic enzymes that are regulated by amino acid concentrations. The bEBP-like protein TyrR contains both a PAS domain and an ACT domain (Fig. 5) and facilitates the activation or repression of the transcription of genes involved in aromatic amino acid biosynthesis and transport, although it is not an activator of σ⁵⁴-dependent transcription (164, 165, 209). The ACT domain is most likely the binding site for the aromatic amino acid tyrosine, phenylalanine, or tryptophan, whereas the PAS domain has been suggested to have a role in contacting the αCTD of RNAP (165).

Protein-protein interactions.A further group of bEBPs regulates the activity of the AAA⁺ domain through protein-protein interactions with another protein called an antiactivator. In the nitrogen-fixing organism Azotobacter vinelandii, the bEBP NifA is bound by the antiactivator NifL to prevent the transcription of nif genes under conditions that are inappropriate for nitrogen fixation (136, 137, 145). In addition, the NifA protein itself contains a regulatory GAF domain. The binding of 2-oxoglutarate (2-OG) to the GAF domain antagonizes the influence of adenosine nucleotides on the NifL-NifA interaction to ensure that the bEBP is not inhibited by NifL under nitrogen-fixing conditions (132, 135).

Some bEBPs, such as PspF, lack an N-terminal regulatory domain altogether (200). PspF is instead negatively regulated by PspA in trans (70, 73, 114). Initially, PspA was suggested to be an escaped regulatory domain of PspF, but phylogenetic analyses have placed PspF into a distinct clade of response regulator bEBPs (200). The pspABCDE operon encodes several proteins that help maintain membrane integrity. It has been suggested that upon proton motive force (PMF) dissipation, PspB and PspC act as positive regulators of transcription by binding PspA and relieving the inhibition of PspF. This enables PspF to activate σ⁵⁴-dependent transcription from the psp promoter (1, 59, 114, 143). Like PspF, the related HrpR and HrpS activators are group IV bEBPs that lack an N-terminal regulatory domain (Fig. 5). The HrpR/HrpS system regulates transcription from the hrp (hypersensitive response and pathogenicity) gene cluster (105, 188), which encodes plant pathogenicity genes, including components and effectors of a type III secretion pathway in Pseudomonas syringae (4, 54). The extracytoplasmic function (ECF) σ factor (141) HrpL is the primary transcription factor that controls the expression of the hrp gene cluster. HrpR and HrpS have been shown to interact, forming a stable heteromeric complex that activates the σ⁵⁴-dependent transcription of hrpL (105). In a manner analogous to that of the regulation of PspF, HrpS (but not HrpR) is specifically bound by another protein, HrpV, to repress the activity of the heterohexamer (122). However, there is no apparent homology between HrpV and PspA (170, 200).

Global regulatory signals.In addition to the regulation of transcription via the signal-sensing functions of bEBP regulatory domains, the output from σ⁵⁴-dependent promoters can additionally be controlled in response to global regulatory signals (190). This regulation could be mediated by factors that counteract the binding either of the holoenzyme at the promoter or of the activator (exclusion). In Pseudomonas sp. strain ADP, the bEBP NtrC has been shown to activate the σ⁵⁴-dependent transcription of the atzR gene from solution (169). The AtzR protein acts as a regulator of cyanuric acid metabolism but also autoregulates its own expression by binding to a site that overlaps the atzR promoter to inhibit the formation of the closed complex by the σ⁵⁴-RNA polymerase (169). In Klebsiella aerogenes, NtrC activates transcription from the σ⁵⁴-dependent nac promoter under nitrogen-limiting conditions. In this case, Nac negatively autoregulates its own expression by a mechanism of antiactivation; it binds within the intergenic region (between the upstream NtrC enhancer sites and the promoter) to prevent productive interactions between the activator and the σ⁵⁴-RNA polymerase holoenzyme (75). Both exclusion and antiactivation mechanisms seem to be utilized in the regulation of the activation of transcription by DctD at the σ⁵⁴-dependent dctA promoter. Cyclic AMP (cAMP)-bound CRP (cAMP receptor protein) is able to bind to two sites that overlap the binding sites of the bEBP, but it is also able to interact directly with the holoenzyme to inhibit transcription (215). In vitro data suggest that the cAMP-CRP interaction with the holoenzyme results in an alternative closed complex that slowly converts into one that is subject to bEBP activation.

Negative regulation as a dominant mechanism of control.As discussed above, the N-terminal regulatory (R) domain allows the bEBP to regulate transcription at σ⁵⁴-dependent promoters in response to environmental cues. However, bEBPs have developed different methods for the transduction of this signal from the R domain (the site of detection) to the catalytic C domain. Generally, this transduction can be subject to (i) positive control or (ii) negative control (189). Assaying the activity of the bEBP (i.e., ATP hydrolysis, oligomerization, or activation of transcription) in a truncated form that lacks the N-terminal R domain has become the standard method for determining the mechanism of control. The first indication that the R domains of bEBPs may function in the repression of AAA⁺ domain activity came from the identification of semiconstitutive variants of XylR and DmpR that had substitutions in either the R domain, the C domain, or the interdomain linker (often referred to as either the B linker, Q linker, or L₁ linker) (64, 76, 194). Furthermore, N-terminally truncated forms of these proteins that lacked the R domain exhibited constitutively active phenotypes in vivo, indicating that the C domain is subject to repression by the R domain. For DmpR, constitutive ATPase activity in vitro was demonstrated in the absence of this negative control (76, 194). These observations led to a model of interdomain repression in which the R domain represses the ATPase activity of the C domain in the absence of a small molecule. Ligand binding to the R domain is then expected to cause derepression, allowing the bEBP to hydrolyze ATP and activate transcription (Fig. 13A). More recently, this mechanism of negative control has been identified in the response regulator bEBPs DctD and NtrC1 (128). Here, phosphorylation and not effector binding relieves interdomain repression (66, 128). The removal of the R domain and the L₁ linker in DctD produced an active protein without the need for phosphorylation (88). Similar results were obtained with NtrC1, where activity was repressed both in vivo and in vitro in the presence of the R domain and the L₁ linker but was derepressed in their absence (128). The founder member of the σ⁵⁴-dependent class of transcription factors, NtrC, in contrast, is positively regulated (Fig. 13B). The deletion of the R domain to give a form of the activator that can no longer be phosphorylated by NtrB results in a constitutively inactive form of the protein, indicating that the AAA⁺ domain is subject to positive regulation (69, 216). Here, the phosphorylation of the R domain has a genuine stimulatory, rather than a derepressive, function. Therefore, despite sharing ∼60% sequence similarity, the NtrC and NtrC1 bEBPs have evolved entirely different mechanisms of regulation. In the absence of a transcriptional assay, it is not known whether ZraR is subject to positive or negative regulation, but on the basis of structural similarities, it is likely to belong to the negatively regulated NtrC1/DctD group (181). Overall, the relative advantages of protein-protein interactions, phosphorylation, or effector binding as a control mechanism are not understood, but it seems that whatever the mechanism of sensing, negative regulation is the dominant mechanism of control (189).

• Open in new tab
• Download powerpoint

Fig 13

Negative (A) and positive (B) control of AAA⁺ domain activity. In the more common mechanism of negative control, ligand binding (or phosphorylation) relieves the repression of the regulatory (R) domain on the central (C) domain, which is intrinsically competent to hydrolyze ATP. The AAA⁺ domain is then able to carry out ATP hydrolysis. Accordingly, when the R domain is removed, the bEBP is active irrespective of the presence or absence of a signaling molecule (shown as a purple triangle) or an available kinase. In positive control, ligand binding or phosphorylation has a genuine stimulatory function. The phosphorylated or ligand-bound form of the R domain activates the C domain, which is not intrinsically competent to hydrolyze ATP. The AAA⁺ domain is then able to carry out ATP hydrolysis. Accordingly, when the R domain is removed, the bEBP is inactive irrespective of the presence or absence of a signaling molecule or an available kinase.

Functions of the C domain targeted by the R domain.In order for the output of the bEBP to be regulated, the sensory domain must respond to the detection of an environmental or metabolic signal by controlling the activity of the AAA⁺ domain that is indispensable and often sufficient for σ⁵⁴-dependent transcription (15, 121, 222, 226). Irrespective of whether the R domain regulates the C domain positively or negatively, it has been shown to target three different aspects of AAA⁺ activity: (i) the oligomerization of the AAA⁺ domain, (ii) the ATPase activity of the AAA⁺ domain, and (iii) the interaction with σ⁵⁴ (Fig. 14). These targets will now be considered.

• Open in new tab
• Download powerpoint

Fig 14

Possible targets of regulatory domain-mediated regulation. (A) Most commonly, the regulatory domain represses oligomerization, e.g., NtrC1 and DctD (66, 128), or promotes self-association in response to the signal, e.g., NtrC (62). (B) In PspF, negative regulation directly targets the nucleotide hydrolysis machinery (115). While the binding of ATP releases L1 and L2 to establish a weak interaction with σ⁵⁴, hydrolysis is required to produce a strong interaction that results in the remodeling of the holoenzyme (173). (C) NorR may represent a newly identified mechanism of control in which the interaction with σ⁵⁴ is the target of the regulatory domain. Where oligomerization is not the target, the preassembly of a hexamer prior to activation may have a physiological advantage, e.g., a rapid response to stress. The mechanisms of regulation that target ATP hydrolysis, oligomerization, and σ⁵⁴ interactions are likely to be highly interconnected, with the enzymatic activity of the AAA⁺ domain being the ultimate target of regulation.

(i) Controlling AAA⁺ oligomerization.As noted above, a self-association of the AAA⁺ domains of the bEBP must occur in order to form the functional activator (171, 235). Therefore, the oligomeric determinants of the C domain represent an ideal target for the N-terminal regulatory domain for either a positive or negative mechanism of control (Fig. 15). Structural studies of full-length and truncated forms of NtrC1 and DctD from Aquifex aeolicus and Sinorhizobium meliloti, respectively, indicated that the N-terminal R domain targets the oligomeric determinants of the AAA⁺ C domain in the mechanism of negative control (Fig. 15B) (52, 66, 128, 140, 158). The crystal structure of a form of NtrC1 comprising the regulatory domain joined to the central AAA⁺ domain by linker L₁ (R-L₁-C) reveals a dimeric structure in which the arrangement of the subunits is incompatible with AAA⁺ ring assembly (128). Here, the unphosphorylated receiver domains form a homodimer that holds the AAA⁺ protomers in an inactive front-to-front configuration via interactions involving the coiled-coil structure of linker L₁. Structures of the activated regulatory domain indicate that phosphorylation results in an alternative homodimer configuration that disrupts the repressive interaction between the R and C domains, allowing the reorientation of the AAA⁺ protomers into a front-to-back configuration (66). This phosphorylation-dependent rearrangement allows self-association to take place to form an oligomer competent to hydrolyze ATP. In line with this, the crystal structure of the isolated AAA⁺ domain of NtrC1 reveals a heptameric arrangement in the absence of the R domain (128). It appears that the coiled coil of the L₁ linker is critical for holding the central domain in an inhibitory configuration; its presence has become indicative of this type of regulation (66). Indeed, for the ligand binding XylR and DmpR proteins, mutational analysis has shown that the integrity of the linker between the regulatory and central domains is crucial for the repression of activity (83, 156). The mechanism of regulation in NtrC also targets the oligomeric determinants of the AAA⁺ domain but is in stark contrast to those in NtrC1 and DctD (Fig. 15A) (52, 62, 66). A truncated form of the protein that lacks the R domain is constitutively inactive, indicating a genuine stimulatory rather than a derepressive role for phosphorylation. The activation of oligomerization occurs upon phosphorylation, which exposes a hydrophobic patch on the R domain allowing it to bind to the N-terminal region of the central AAA⁺ domain. Recent X-ray solution scattering (SAXS/WAXS) and electron microscopy studies indicated that the R domain interacts with the C domain of an adjacent protomer on the outside edge of the AAA⁺ ring, promoting self-association and contributing to the stability of the resulting hexamer (62). Put simply, in the positive regulation of NtrC, the phosphorylation of the regulatory domain creates a new interaction that leads to oligomerization, whereas in negative regulation, typified by NtrC1 and DctD, phosphorylation releases an interaction that leads to the formation of the functional oligomer (Fig. 15). Interestingly, sequence analysis reveals a correlation between the mechanism of negative control in NtrC1 and DctD and the presence of a structured linker between the R and C domains (linker L₁) and an unstructured linker between the C and D domains (linker L₂). Conversely, the positively regulated NtrC contains an unstructured L₁ linker and a structured L₂ linker. In both classes, the structured linker seems to play a significant role in the stabilization of the inactive dimer, and examinations of the L₁ and L₂ sequences of other bEBPs may help identify whether self-association is subject to positive or negative control (66). For example, a mutation of the linker between the R and C domains in NtrC does not affect its activity (225), in agreement with a function for the regulatory domain in positive rather than negative control. The bEBP NtrC4 from Aquifex aeolicus has a partially disrupted receiver-AAA⁺ domain interface and can assemble into active oligomers at high protein concentrations independent of phosphorylation, a process that does not occur with NtrC1 (11). The activated receiver domain has been shown to stabilize the hexameric form of NtrC4, thus functioning as an intermediate between the negative mechanism of NtrC1/DctD and the positive mechanism of NtrC (11, 12).

• Open in new tab
• Download powerpoint

Fig 15

Models of bEBP activation by phosphorylation through the promotion of oligomerization by stimulatory (A) and derepressing (B) functions of the response regulator (RR) domain. In activated NtrC, the DNA binding domain is hidden underneath the hexamer ring. For DctD and NtrC1, no information is available to define the positions of DNA binding domains. R, regulatory domain; L1, linker 1; C, central domain; L2, linker 2; D, DNA binding domain. Models were built by using published structures of NtrC fragments R (off state, PDB accession number 1KRW; on state, accession number 1KRX) and L2-D (PDB accession number 1NTC) and NtrC1 fragments R (PDB accession number 1ZY2), R-L1-C (PDB accession number 1NY5), and L1-C (PDB accession number 1NY6). (Adapted from reference 66 with permission from Elsevier.)

(ii) Controlling ATPase activity.Ultimately, the target of R-domain-mediated regulation is the enzymatic activity of the bEBP. Where the R domain targets the oligomeric determinants, the effect is to promote or prevent the formation of an oligomer that is capable of hydrolyzing ATP. However, the regulatory domains of some bEBPs may specifically target the nucleotide hydrolysis machinery without influencing the oligomerization state (Fig. 14B). This has been shown for PspF, which is regulated in trans through direct interactions between the activator and the negative regulator PspA (Fig. 16) (70, 72, 73, 114). Here, oligomerization is driven by the binding of ADP and ATP to the individual protomers (117), whereby the “DE” residues of the Walker motif prevent nucleotide-independent hexamer formation (116). PspA has been shown to negatively regulate the ATPase activity of PspF through the formation of an interaction that is dependent on a surface-exposed tryptophan residue (W56 of PspF) (72, 73, 114). Recently, it was shown that the PspA-mediated inhibition of PspF ATPase activity is likely to involve the repositioning of the conserved asparagine (N64 in PspF) required for the sensing of the γ-phosphate during nucleotide hydrolysis. Substitutions of this asparagine in PspF do not prevent PspA binding, but ATPase activity is not significantly decreased by the presence of PspA as in the wild-type activator (112). Consequently, a model has been proposed to link the binding of PspA to the inhibition of ATP hydrolysis in PspF. The binding of PspA is detected via the W56 residue, which relays this information to N64 via β-sheet 2. This leads to the repositioning of the N64 side chain, altering the distances between ATP, the conserved asparagine, and the Walker B glutamate (E108 in PspF) (112). These distances are thought to be critical for ATP hydrolysis and the coordination of the resulting conformational changes in the AAA⁺ domain. Significantly, it has been demonstrated that the inactive regulatory complex consists of approximately six PspA subunits and six PspF subunits (113). Therefore, in contrast to the bEBPs NtrC1 and DctD, the negative regulation of PspF activity is unlikely to target the oligomeric determinants (66, 128). In addition, PspA does not inhibit the interaction between PspF and σ⁵⁴, suggesting that negative regulation does not target the σ⁵⁴ interaction surface of PspF. The PspA-PspF regulatory complex is instead expected to have an altered arrangement in the key ATP hydrolysis determinants that form the catalytic site at the interprotomer interfaces of the PspF hexamer. Potentially, the inhibition of a preassembled PspF hexamer by PspA allows the cell to rapidly respond to membrane damage (113, 114).

• Open in new tab
• Download powerpoint

Fig 16

Negative regulation of PspF AAA⁺ activity by PspA targets the nucleotide hydrolysis machinery via the W56 residue. (A) Crystal structure of PspF_1–275 (PDB accession number 2C96) in the ATP-bound state showing the key residues involved. (B) Model of the signaling pathway coupling negative regulation to substrate remodeling. PspA interacts directly with PspF, an interaction that is detected via the surface-exposed W56 (purple) residue of PspF. W56 relays this information to the conserved asparagine (N64) (red) via β-sheet 2 (blue). This causes the repositioning of the Walker B glutamate (E108) (green) to prevent ATP hydrolysis. Upon the dissipation of the PMF, PspA inhibition is prevented (possibly facilitated by PspB and PspC), and ATP hydrolysis can occur, strengthening the σ⁵⁴ interaction and leading to substrate remodeling. The removal of the γ-phosphate leads to a 90° rotation of the E108 side chain, breaking the interaction with N64. This change is translated to GAFTGA-containing L1 (orange) via helix 3 (H3) (yellow), and the loops compact back downwards. (Adapted from reference 112 with permission of the publisher.)

(iii) Controlling σ⁵⁴ interactions.For a bEBP to activate transcription, it must ultimately make contact with the Eσ⁵⁴ holoenzyme bound at the promoter. As described above, this interaction is facilitated by the highly conserved GAFTGA motif within the central AAA⁺ domain of the hexamer, in response to the binding and hydrolysis of ATP (21). Therefore, it is feasible that in some activators, the targets of the regulatory domain could be the determinants of σ⁵⁴ interactions (Fig. 14C). Upon the receipt of a signal, the R domain could either promote the formation of new contacts to form the interaction surface (positive control) or remove inhibitory contacts that enable σ⁵⁴ contact (negative control). Previous work identified the nitric oxide (NO)-responsive transcription factor NorR as a member of the bEBP subfamily of AAA⁺ proteins, which are subject to negative regulation. When the N-terminal regulatory (GAF) domain was removed, the truncated form of NorR was shown to be constitutively active (81, 167), suggesting a role for the R domain in interdomain repression. Electron paramagnetic resonance (EPR) spectroscopy has since shown that NO binds to the nonheme iron center of the N-terminal GAF domain of NorR to form a mononitrosyl {Fe(NO)}⁷ (S = 3/2) species, triggering conformational changes that relieve the repression exerted by the regulatory domain upon the AAA⁺ domain (60). NorR is then able to hydrolyze ATP, leading to open complex formation and the expression of the norVW genes, the products of which enable E. coli to reduce NO to the less toxic N₂O (82, 85, 106). More recently, a number of AAA⁺ variants of NorR that escape GAF-mediated repression have been identified and are located in a key region of the central domain that undergoes significant conformational changes as ATP is hydrolyzed (36). Significantly, two bypass mutations were identified in the GAFTGA motif, which is absolutely required for σ⁵⁴-dependent transcription. The G266D and G266N variants (GAFTGA) enabled complete escape from the repression exerted by the regulatory domain. This invokes a model whereby the GAF domain negatively regulates the AAA⁺ domain by preventing the access of the surface-exposed loops to σ⁵⁴ (36, 37). Genetic and biochemical studies using the GAFTGA motif escape variant (G266D) suggested that the GAF domain does not regulate AAA⁺ activity through the control of oligomerization or by directly targeting the ATP hydrolysis machinery. The role of the σ⁵⁴ interaction surface in negative regulation by the GAF domain is further supported by genetic suppression data. The R81 residue in the GAF domain appears to play a crucial role in the repression mechanism, since an alanine substitution at this position led to significant constitutive NorR activity, while the R81D charge change gave rise to a fully constitutive phenotype. Hydrophobic substitutions, particularly leucine, restored repression only when combined with specific bypass mutations in the AAA⁺ domain, including those in the GAFTGA loop (36). Previous structural modeling of the GAF domain suggested that the R81 residue is surface exposed (206) and therefore well placed to make contact with the AAA⁺ domain. In this model, it is located at the opposite end of an α-helix to the R75 residue, which is postulated to be a ligand for the hexacoordinated iron and is the most suitable candidate to be displaced upon NO binding (206). Therefore, it is possible that the formation of the mononitrosyl iron complex would displace the R75 ligand, causing a conformational change in the helix that repositions R81. Interactions between the R81 residue and the residue(s) in the AAA⁺ domain may thus facilitate the switch from the “off” state to the “on” state. Overall, NorR appears to represent another mechanism of negative regulation in which the N-terminal regulatory domain targets the σ⁵⁴-interacting region of the AAA⁺ domain, which includes the key GAFTGA motif (Fig. 17). This mode of regulation of NorR, as in the case of PspF, may reflect the requirement for a rapid stress response, facilitated by the preformation of the bEBP hexamer. In the case of those “rare” bEBPs that contain a naturally occurring aspartate or asparagine residue at the second glycine of this motif, it is anticipated that the AAA⁺ domain of such proteins is still subject to regulation by the N-terminal domain. Indeed, the FlgR protein of H. pylori contains a “GAFTDA” motif but still requires phosphorylation by FlgS in order to activate transcription (26). Therefore, these bEBPs are not likely to be regulated by targeting the σ⁵⁴ interaction surface as in the case of NorR; instead, control may be at the point of oligomerization or ATP hydrolysis.

• Open in new tab
• Download powerpoint

Fig 17

Model of NorR-dependent activation of norVW (adapted from reference 37). (A) The binding of NorR to the norR-norVW intergenic region that contains the three NorR binding sites (sites 1, 2, and 3) (highlighted in red) is thought to facilitate the formation of a higher-order oligomer that is most likely to be a hexamer (207). (B) Although bound to DNA, in the absence of NO, the N-terminal GAF domains (blue rectangles) negatively regulate the activity of the AAA⁺ domains (green circles) by preventing the access of the surface-exposed loops to σ⁵⁴. (C) In the “on” state, NO binds to the iron center in the GAF domain, forming a mononitrosyl iron species. The repression of the AAA⁺ domain is relieved, enabling ATP hydrolysis by NorR coupled to conformational changes in the AAA⁺ domain. (D) In the presence of ATP, the surface-exposed loops (orange) that include the GAFTGA motifs move into an extended conformation to establish an initial interaction with σ⁵⁴ that is strengthened upon hydrolysis, resulting in the remodeling of the closed complex. Upon phosphate release, L1 and L2 compact downwards, enabling the relocation of the sigma factor (173). For simplicity, DNA is not illustrated in panel B, C, or D.