|Home | About | Journals | Submit | Contact Us | Français|
Bacterial enhancer-binding proteins (bEBPs) oligomerize through AAA+ domains and use ATP hydrolysis-driven energy to isomerize the RNA polymerase-σ54 complex during transcriptional initiation. Here, we describe the first structure of the central AAA+ domain of the flagellar regulatory protein FlrC (FlrCC), a bEBP that controls flagellar synthesis in Vibrio cholerae. Our results showed that FlrCC forms heptamer both in nucleotide (Nt)-free and -bound states without ATP-dependent subunit remodeling. Unlike the bEBPs such as NtrC1 or PspF, a novel cis-mediated “all or none” ATP binding occurs in the heptameric FlrCC, because constriction at the ATPase site, caused by loop L3 and helix α7, restricts the proximity of the trans-protomer required for Nt binding. A unique “closed to open” movement of Walker A, assisted by trans-acting “Glu switch” Glu-286, facilitates ATP binding and hydrolysis. Fluorescence quenching and ATPase assays on FlrCC and mutants revealed that although Arg-349 of sensor II, positioned by trans-acting Glu-286 and Tyr-290, acts as a key residue to bind and hydrolyze ATP, Arg-319 of α7 anchors ribose and controls the rate of ATP hydrolysis by retarding the expulsion of ADP. Heptameric state of FlrCC is restored in solution even with the transition state mimicking ADP·AlF3. Structural results and pulldown assays indicated that L3 renders an in-built geometry to L1 and L2 causing σ54-FlrCC interaction independent of Nt binding. Collectively, our results underscore a novel mechanism of ATP binding and σ54 interaction that strives to understand the transcriptional mechanism of the bEBPs, which probably interact directly with the RNA polymerase-σ54 complex without DNA looping.
The bacterial enhancer-binding proteins (bEBPs)2 are molecular machines belonging to the AAA+ (ATPases associated with various cellular activities) superfamily (1). The conserved AAA+ domain of the bEBPs, which is made of the signature motifs like Walker A, Walker B, sensor I and sensor II (Fig. 1), controls the oligomeric states, nucleotide (Nt) binding, hydrolysis, and/or conformational changes in the loops, implicated in RNAP-σ54 binding to initiate transcription (2). Despite the conserved nature of the AAA+ domain, variation in the functional state of oligomerization and mode of Nt-binding of bEBPs emerged as matter of interest. Although Salmonella enterica NtrC and Aquifex aeolicus NtrC1 and NtrC4 are homologs made of regulatory (R), AAA+, and DNA binding domains, structural studies showed that their regulatory mechanisms are significantly different (3,–6). Upon phosphorylation, the R domain of NtrC gains interactions with neighboring AAA+ domain stabilizing the oligomeric form (3, 6). In the case of NtrC1 and NtrC4, phosphorylation or BeF3− + Mg2+ activation at the R domain (7) converts the inactive dimer of the AAA+ domain to the active hexa/heptamers (4, 5). The ATP-bound AAA+ domain of A. aeolicus NtrC1E239A, revealed a heptameric assembly with asymmetry in the central channel (8). Recently, the AAA+ domain of NtrC1 is found to form a split ring hexamer in the presence of ADP + BeF3− that argues for flexibility in packing stoichiometry and interface angles of the constituting monomers (9). ESI-MS results of A. aeolicus NtrC4 showed that, although the full-length and activated R-AAA+ proteins form hexamers, the isolated AAA+, unactivated R-AAA+, and AAA+-DNA binding domains form heptamers (10). Interestingly, despite the variation in the functional oligomeric states, the structural results of the aforesaid bEBPs underscores Nt-dependent subunit remodeling and participation of a trans-acting Arg (named as “R-finger”) in ATP binding and hydrolysis (11,–13).
In this study, we have investigated for the first time the state of oligomerization, molecular mechanism of ATP binding, and hydrolysis of a bEBP, FlrC, which is involved in flagellar synthesis of Vibrio cholerae. Transcriptional regulation of the flagellar system of V. cholerae acts as an important signaling component of the pathogenic process positively regulating the factors that assist in arrival at the colonization site (14). The transcription of the flagellar genes is organized in a hierarchy of four classes (15). The class I gene product FlrA activates σ54-dependent transcription of the class II genes flrBC, which encode FlrC and its cognate kinase FlrB (16). The expression of class III genes, which encode the basal body hook and the flagellin FlaA, is controlled by modulation of the activity of FlrC (16). Although deletion of flrC produces a nonmotile V. cholerae strain with a modest colonization defect, a strain producing hyperactive FlrC shows altered cell morphology (16,–18). Homologs of FlrC that regulate a similar class of flagellar genes are found in all the other Vibrio spp. studied (19), along with Pseudomonas aeruginosa (20) and Campylobacter jejuni (21, 22), suggesting that similar mechanisms underlie the regulation of class III genes in polar flagellates. The σ54-dependent activators typically bind to the sites located upstream of the σ54 holoenzyme-binding site and contact RNAP-σ54 complex by a DNA looping mechanism (23). In contrast, FlrC binds the elements located downstream of the σ54 binding and transcriptional start sites of the flaA and flgK promoters (17). Although the downstream location of FlrC-binding sites is unusual, similar downstream binding is also observed in FleQ of P. aeruginosa for flhA, fliE, and fliL genes. Ramphal and co-workers (24) suggested that the close proximity of FleQ-binding sites to the σ54-dependent transcriptional start sites is incompatible with a DNA looping mechanism and argues for a direct interaction between activator and RNAP without looping, which may occur in the case of FlrC as well.
FlrC includes the N-terminal R domain, central AAA+/σ54 interaction domain, and C-terminal DNA binding domain. Phosphorylation occurs on the conserved Asp-54 of the R domain by the cognate kinase, FlrB (18). A V. cholerae strain containing an inactive (D54A) or constitutively active (M114I) FlrC mutant showed more severe colonization defects than strain lacking FlrC entirely, which implies that both unphosphorylated and phosphorylated forms of FlrC are required for the colonization, and locking FlrC into either an active or an inactive state would send incorrect stimuli into this stepwise colonization process (18).
Here, we describe the crystal structure of the central AAA+ domain of FlrC (FlrCC) in Nt-free and AMP-PNP (nonhydrolysable ATP analog)-bound states. Our results provide the first structural evidence for an AAA+-ATPase implicated in flagellar synthesis that forms heptamer both in Nt-free and -bound states without any Nt-dependent subunit remodeling. The major presence of the heptameric species is established in solution by the size exclusion chromatography and dynamic light scattering, in the ground and ADP + AlF3-mediated transition states of FlrCC. Strikingly, in contrast to the other bEBPs, FlrCC does not use any trans-acting residue for Nt binding, and the reason lies in the ATPase site architecture of the individual monomer. A unique “closed to open” conformational change occurs in Walker A of FlrCC to facilitate ATP binding. Structural observations coupled with fluorescence quenching and ATPase assay identified a novel trans-acting “Glu switch” that promotes the displacement of Walker A for ATP binding and hydrolysis. With a relatively wider central channel and an in-built architecture of the L1 loop, heptameric FlrCC interacts with σ54 both in Nt-free and -bound states. These intriguing observations open up a new avenue to further explore σ54-dependent transcriptional mechanism of such activators.
The genes of FlrCC (amino acids 132–381) and σ54 (amino acids 1–487) from V. cholerae were cloned in pET28a+ within NdeI and BamHI restriction sites. The recombinant proteins with N-terminal His6 tag were overexpressed in BL21(DE3) and purified by Ni-NTA affinity chromatography as per protocol described in Dey and Dasgupta (25). Mutants E286A, R319A and R349A were prepared by two-step PCR. All the mutants were verified by commercial sequencing, and purifications were done following the same protocol as WT protein.
Crystals of Nt-free and AMP-PNP-bound FlrCC were grown at 20 °C using the hanging drop vapor diffusion method. Optimal crystals of Nt-free FlrCC were obtained when 4 μl of the protein solution (12 mg/ml) in a buffer consisting of 50 mm Tris-HCl (pH 8.0), 300 mm NaCl, 5 mm MgCl2, and 10% glycerol was mixed with 2 μl of precipitant solution consisting of 10% (w/v) PEG 6000, 0.1 m MES (pH 6.0) and was incubated over a reservoir solution of same composition. Initially, Nt-free FlrCC crystals diffracted up to 2.8 Å, but after further standardization, the resolution improved to 2.3 Å.
AMP-PNP-bound FlrCC was prepared by mixing 20 μl of the protein solution (20 mg/ml) (in the same buffer as for Nt-free FlrCC) with 1 mm AMP-PNP and incubated for 30 min at room temperature. AMP-PNP-bound crystals of FlrCC were grown when 4 μl of the above mixture and 1.5 μl of the precipitant solution (consisting of 10% (w/v) PEG 6000, 0.1 m MES (pH 6.0)) were mixed and equilibrated for 4 days against 12% (w/v) PEG 6000, 0.1 m MES (pH 6.0), 0.1 m NaCl. Crystals were transferred to a cryo-protectant solution consisting of 40% (v/v) ethylene glycol, 2.5% (w/v) PEG 6000, and 50 mm Tris-HCl (pH 8.0). Crystals were then flash-frozen in liquid nitrogen, and diffraction data were collected to 2.6Å.
Diffraction data were collected at BM14 beamline of the European Synchrotron Radiation Facility (ESRF) at Grenoble, France. Data were processed and scaled using HKL2000 (26). Data collection and processing statistics are given in Table 1.
The initial phases for both Nt-free and AMP-PNP-bound FlrCC were obtained by molecular replacement using PHASER (27). Packing considerations indicated the presence of seven molecules in the asymmetric unit for both the structures. Seven molecules were organized in the form of a closed heptamer with a central channel.
Truncated coordinates of one subunit of the inactive dimeric σ54 activator NtrC4 of A. aeolicus (PDB code 3DZD) (4) having residues 133–369 (where the coordinates of the N-terminal regulatory domain were truncated) were retrieved, which produced acceptable solution in molecular replacement calculations for Nt-free FlrCC. The model of Nt-free FlrCC was then used to solve the structure of AMP-PNP-bound FlrCC. Few cycles of simulated annealing, positional refinement, individual B-factor, and TLS refinement were accomplished by PHENIX (28), and model building was done by WinCoot (29). The structures were refined well with Rcryst of 19.4% (Rfree = 25.47%) and with Rcryst of 18.63% (Rfree = 24.66%) for Nt-free and AMP-PNP-bound FlrCC, respectively. Data collection and refinement statistics are given in Table 1.
Fluorescence measurement was carried out using a spectrofluorometer, Hitachi F-7000. Changes in tryptophan (Trp-299) fluorescence were measured at an excitation wavelength of 295 nm, and the emission spectra were recorded between 300 and 400 nm with slit widths of 2.5 nm for both excitation and emission. All reactions were carried out at 25 °C. Equilibrium titrations of FlrCC, R319A, R349A, and E286A were carried out with AMP-PNP. The reactions were performed in a buffer containing 50 mm Tris-HCl (pH 8.0), 300 mm NaCl, and 5 mm MgCl2. For all proteins, the final concentrations were 5 μm and AMP-PNP concentration varied from 0 to 0.39 mm. The binding stoichiometry was determined using the protocol described in Mani et al. (30). The plot of log(F0 − F)/(F − F) against log [AMP-PNP], where F0, F, and F are the fluorescence intensities of FlrCC alone, FlrCC, in the presence of various concentrations of AMP-PNP, and FlrCC saturated with AMP-PNP, respectively, yielded a straight line whose slope was a measure of the binding stoichiometry.
The dissociation constant, Kd was determined using nonlinear curve fitting analysis as per Equations 1 and 2 (31). All experimental points for the binding isotherms were fitted by the least squares methods.
Although C0 denotes the input concentrations of the ligand AMP-PNP, Cp denotes the same for FlrCC and its mutants. ΔF is the change in fluorescence intensity at 340 nm (λex = 295 nm) for each point of titration curve, and ΔFmax is the same parameter when ligand is totally bound to the protein. A double-reciprocal plot of 1/ΔF against 1/(Cp − C0), as shown in Equation 3 was used to determine the ΔFmax.
ΔFmax was calculated from the slope of the best fit line corresponding to the above plot. All experimental data points of the binding isotherm were fitted by linear fit analysis using Microsoft EXCEL and Origin 8.0. The equilibrium titrations of FlrCC and the mutant R319A were also carried out in the presence of ADP following the same process as of AMP-PNP.
ATPase activity was determined with a procedure from the malachite green assay (32, 33) to monitor the release of inorganic phosphate (Pi). For ATPase assay, reaction mixtures containing FlrCC and the mutants E286A, R319A, and R349A (final concentration of 2.5 μm) were individually incubated with 0.1 mm ATP at 25 °C. The reaction buffer was made of 50 mm Tris-HCl (pH 8.0), 300 mm NaCl, and 5 mm MgCl2. After 25 min of incubation, the reaction mixture was assayed for Pi. FlrCC in the absence of MgCl2 served as a negative control. Colored reagent containing 10 ml of 0.44 g of Malachite green dissolved in 0.3 m H2SO4, 2.5 ml of 7.5% ammonium molybdate, and 0.2 ml of 11% Tween 20 was added to the reaction mixture after 25 min, and the absorbance was measured at 630 nm within 5 min of adding the coloring reagent. The total Pi for each reaction was compared with a Pi standard curve prepared using KH2PO4. All the experiments were minimally performed in triplicates.
Probable interactions between His6-tagged σ54 and tag-free FlrCC were monitored using Ni-NTA pulldown assay. 30 μl of Ni2+-NTA slurry (Qiagen) was washed with binding buffer containing 50 mm Tris-Cl (pH 8.0), 300 mm NaCl, 5 mm MgCl2, 10% glycerol, and the resin was then incubated in different batches with 50 μl of purified His6-tagged σ54 protein at a concentration of 0.2 mg/ml at 25 °C for 20 min with intermittent gentle shaking. The beads were then washed three times with the binding buffer before adding FlrCC, FlrCC bound to ATP, and AMP-PNP individually. For activation, FlrCC was incubated for 10 min with ATP and AMP-PNP separately (1 mm). The mixture was then added to the σ54-bound Ni2+-NTA resin in molar excess and incubated for another 10 min at 25 °C. The beads were washed with the buffer and then resuspended in 20 μl of 4× SDS-polyacrylamide gel loading buffer and were subjected to SDS-PAGE analysis and Coomassie Blue staining.
Gel filtration chromatography experiments of FlrCC ± nucleotides were performed at room temperature in the running buffer containing 50 mm Tris-HCl (pH 8.0), 300 mm NaCl, and 5 mm MgCl2 using a Superdex 200 column (16 × 700 mm, GE Healthcare). A 200-μl sample containing FlrCC in the Nt-free state or with AMP-PNP or ADP.AlFx was injected onto the column and chromatographed at a flow rate of 1 ml/min. For the AMP-PNP-bound state, 0.3 mm FlrCC (0.3 mm) was preincubated for 15 min at 4 °C with 2 mm AMP-PNP. In case of FlrCC-ADP.AlFx state, FlrCC was incubated for 15 min at 4 °C with 2 mm ADP, 4 mm AlCl3, 20 mm NaF. During gel filtration of ADP + AlF3-treated FlrCC, the buffer used for chromatography was mixed with ADP, NaF, and AlCl3 to ensure the stability of the transition state in solution. The peak fractions were collected for SDS-PAGE analysis. Average pseudo partition coefficients were calculated as Kav = (Ve − Vo)/(Vc − Vo), where Ve is the elution volume; Vo is the void volume; and Vc is the total column volume. For calculating Kav, the column was calibrated with standard molecular weight calibration kit (GE Healthcare) made of blue dextran, ferritin (440 kDa), aldolase (158 kDa), and conalbumin (75 kDa) (data not shown).
DLS measurements were done with FlrCC samples directly taken from gel filtration fractions and analyzed using DynaPro equipped with temperature control using a 12-μl microcuvette. To study the effect of AMP-PNP and ADP + AlF3 on the hydrodynamic size of the FlrCC oligomer, similar experiments were repeated with Nt-bound samples. All the experiments were done at 25 °C. The percent population was calculated and plotted against hydrodynamic radius. Dynamics Version 6 software was used to calculate the hydrodynamic radius and corresponding oligomeric state(s).
Molecular replacement calculations for Nt-free FlrCC were systematically performed using different search models of AAA+ domain prepared from the coordinates of inactive dimeric A. aeolicus NtrC1 (PDB code 1NY5), ADP-bound heptameric NtrC1 of A. aeolicus (PDB code 1NY6), ATP-bound heptameric variant of NtrC1 (PDB code 3M0E), inactive dimeric NtrC4 of A. aeolicus (PDB code 3DZD), ZraR of Salmonella typhimurium (PDB code 1OJL), and PspF of E. coli (PDB code 2BJW). In each case, the coordinates of AAA+ domain were retrieved from one subunit, and mismatched residues were mutated to alanine. Although molecular replacement calculations with the models of NtrC1 and PspF produced no solution, the coordinates of NtrC4 having residues 133–369 of chain A yielded a clear-cut solution. PHASER (27) identified seven monomers (with RFZ = 5.1, TFZ = 27.9, and LLG = 1771) organized in a heptameric ring, in the space group P212121 using data between 49 and 2.3 Å resolutions. A search model of ZraR also produced a heptameric solution, but the statistics were better for the previous one. After a few cycles of refinement and model building, Nt-free FlrCC produced a final Rcryst of 19.4% (Rfree = 25.47%) (Table 1). The structure of AMP-PNP-bound FlrCC was solved using a monomer of Nt-free FlrCC structure as search model and the resulting heptameric structure refined up to the Rcryst of 18.63% (Rfree = 24.66%) (Table 1). Noncrystallographic symmetry was not used during the refinement of these two structures.
Nt-free FlrCC forms a closed heptameric ring with a central channel (Fig. 2A). The diameter and height of the ring are ~126 and ~55 Å, respectively, whereas that of the central channel is ~26 Å (Fig. 2A). Contact surface areas between the subunits (calculated by PISA) (34) are consistently similar with an average value of ~780 Å2. Adjacent protomers of Nt-free FlrCC are related roughly by a rotation angle of 50–53° indicating that the heptamer is symmetrical.
Each protomer of FlrCC is kidney-shaped consisting of a α/β subdomain that is common for P-loop NTPases and a α-helical subdomain that carries a signature of AAA+-ATPases (Fig. 2, B and C). A monomer of FlrCC in the Nt-free state superposes on that of the inactive NtrC4 (4), NtrC1 (13), and PspF (35) with root mean square deviations of 1.1 Å (for 205 Cα), 1.2 Å (for 216 Cα), and 1.7 Å (for 187 Cα), respectively. Structural superposition shows that although the central part of the α/β subdomain of FlrCC, containing the helices α1, α2, and the central β-sheet, matches well with aforesaid bEBPs, the loop regions of the α/β subdomain and a portion of the α-helical subdomain differs significantly (Fig. 2C). Interestingly, a protrusion of the L3 loop and an inclination of helix α7 constrict the ATP binding pocket of FlrCC by ~7 Å compared with the bEBPs like NtrC1, NtrC4, and Pspf (Fig. 2C).
AMP-PNP-bound FlrCC is also a heptamer (Fig. 2D) whose average contact surface area between subunits (calculated by PISA) (33) is ~750 Å2, which is comparable with that of Nt-free FlrCC. Angular rotation values are similar to the Nt-free state indicating that the symmetry of the heptamer is retained upon AMP-PNP binding (Fig. 2D). Heptamer of AMP-PNP bound FlrCC superposed on the apo-structure with an r.m.s.d. of 0.85 Å, further indicating that no major change in the oligomeric state occurs here. Inter-protomeric interactions were largely retained upon AMP-PNP binding (Table 2). However, significant local conformational changes occurred to accommodate AMP-PNP, which will be discussed subsequently.
Stoichiometry of binding between AMP-PNP and FlrCC has been determined using fluorescence spectroscopy. Because Trp-299 is within the Forster distance of the ATP binding pocket, fluorescence quenching of this Trp was monitored in the presence of AMP-PNP. The effect of AMP-PNP binding to each FlrCC monomer was determined by measuring the change in fluorescence intensity with increasing concentrations of the AMP-PNP. The plot of fluorescence intensities against ligand concentration (Fig. 2E), as per the protocol described in Mani et al. (30), yielded a slope of 1.13 ± 0.045 indicating 1:1 interaction between AMP-PNP and FlrCC monomer.
Available crystal structures of the bEBPs showed that the Nt-binding site is located in the cleft between the subdomains and between two adjacent protomers (8, 13). AMP-PNP binds canonically in a similar cleft of FlrCC with certain uniqueness (Fig. 2, F and G). Unlike other bEBPs, FlrCC binds AMP-PNP only through the cis-acting residues without any direct contribution from the trans-acting protomer (Fig. 2F, ,3,3, A–C). The adenine base fits snugly in the hydrophobic pocket made of Val-132, Val-167 of α2, Leu-312, His-315 of α7, and Val-348 of α9 (Fig. 3B). Involvement of Val-132 indicates that the linker region that connects R and AAA+ domains is stabilized upon AMP-PNP binding.
Trp-299 belonging to the connector between β5 and α7 also contributes to this hydrophobic packing to cap the adenine base (Fig. 3B). A tightly bound water molecule mediates interaction between Arg-305 and the adenine base. Arg-319 of α7 stabilizes the ribose sugar (Fig. 3, B and C). Inclination of α7 toward the ATP binding pocket is furthered a little upon AMP-PNP binding ensuring tight interactions between Arg-319 and the hydroxyl groups of the ribose (Fig. 3, C and D). Arg-319 is held in place by salt bridge interaction with nonconserved Asp-352 and a water-mediated interaction with the N terminus of FlrCC.
Conserved Arg-349 of sensor II interacts canonically with the γ-phosphate of AMP-PNP (Fig. 3, A and C). Mg2+ that binds the γ-phosphate of AMP-PNP is stabilized by Asp-230, the first conserved acidic residue of Walker B (Fig. 3, C and E). Conserved Asn-187 of β2 and Thr-270 (Ala in case of NtrC1, NtrC4, and PspF) of β4 stabilize the conformation of Asp-230 (Figs. 1 and and33E). Lys-165 of α2 and Asn-272 of sensor I also participate in positioning the γ-phosphate of AMP-PNP (Fig. 3C). Superposition of all seven chains of AMP-PNP-bound FlrCC shows conserved water Wat1 that coordinates consistently with Mg2+ and the γ-phosphate (Fig. 3E). Few more water molecules are also identified in “near apical” position, as observed in case of LTag structure of SV40 (36), to act as potential nucleophile (Fig. 3E).
Comparison of the AMP-PNP-bound FlrCC with the Nt-free structure demonstrates that Walker A motif undergoes a unique conformational rearrangement to facilitate AMP-PNP binding (Fig. 3F). In the native (closed) conformation Walker A would exert steric hindrance to the incoming AMP-PNP. A displacement of ~2.5 Å of Walker A relieves this hindrance to facilitate AMP-PNP binding where the side chain of Ser-161 experiences a maximum shift of ~5.3 Å (Fig. 3, F and G). In the Nt-free state, Glu-231 of Walker B and Asn-272 of sensor I bind Ser-161 to stabilize the closed conformation of Walker A (Fig. 3G). Hydrogen bond between Ser-163 and Trp-299 also stabilizes that closed conformation (Fig. 3G). Both of these interactions are abrogated upon AMP-PNP binding, and the open conformation of Walker A is stabilized by trans-acting Glu-286 of sensor I (Fig. 3G). Gly-162, Ser-163, and Gly-164 of Walker A stabilize the β-phosphate of AMP-PNP. The primary sequence of Walker A is represented by “GXXXXGKEL” in bEBPs versus consensus “GXXXXGK(T/S)” of AAA+ superfamily (11, 37). A critical analysis of such sequences, short-listed by Bush and Dixon (2), suggests that the consensus of GXXXXGKEL should actually be “G(E/D)XGXGKE(L/V).” Interestingly, in FlrC, instead of Glu/Asp, the second residue of Walker A motif is Pro (Pro-160) (Fig. 1), and our structural results show that aforesaid conformational rearrangement of Walker A, which is unique in FlrC, starts from Pro-160 (Fig. 3F).
In the Nt-free state, cis-acting Arg-349 is positioned by an inter-protomeric salt bridge with trans-acting Glu-286 and a hydrophobic barrier made of trans-acting Tyr-290 (Fig. 3G). Upon AMP-PNP binding, Arg-349 slightly alters its conformation, breaks the salt bridge with trans-acting Glu-286, and binds γ-phosphate (Fig. 3G). trans-Acting Glu-286, on the other hand, participates in stabilizing the open conformation of Walker A. Similarly, cis-acting Asn-272 is relieved from anchoring Ser-161 and is recruited to stabilize the γ-phosphate (Fig. 3G). The movement of Walker A causes a lateral shift of α2 toward the ATP binding pocket, which in turn facilitates hydrophobic packing of Val-167 with adenine base and salt bridge interactions of Lys-165 with β- and γ-phosphates (Fig. 3D).
Based on the structural results, we investigated the potential contribution of Arg-319, Arg-349, and Glu-286 toward AMP-PNP binding through fluorescence quenching studies. Because Trp-299, which is unique in FlrC (Fig. 1), experiences conformational change upon AMP-PNP binding (Fig. 3D), fluorescence quenching of Trp-299 was monitored for FlrCC and the mutants R319A, R349A, and E286A with the addition of AMP-PNP. As expected, FlrCC showed maximum quenching by AMP-PNP with a Kd value of 11.5 ± 0.575 μm (Fig. 4 and Table 3). The minimum quenching was observed for R349A with a Kd of 309 ± 15.45 μm (Fig. 4A and Table 3). Substitution of Arg-319 by Ala showed almost ~7-fold higher Kd value compared with FlrCC, although the impact of this substitution was much less compared with that of Arg-349 (Fig. 4A). These observations imply that although Arg-319 renders significant contribution in the Nt binding through its interaction with ribose, stabilization of γ-phosphate is more important in terms of ATP binding, which is severely affected upon mutation at Arg-349. Binding of ADP with FlrCC has also been tested in a similar fashion. The result showed that the binding efficiency of ADP to FlrCC is only ~4-fold weaker than AMP-PNP, which might be attributed to the contribution of Arg-319 in stabilizing ribose sugar that may restrict the expulsion of ADP upon hydrolysis. Although Glu-286 shows no direct interaction with AMP-PNP, quenching of E286A was lesser compared with FlrCC with an ~5-fold higher Kd value (Fig. 4A) suggesting that in the absence of Glu-286 stabilization of the open conformation of Walker A would be compromised.
We further investigated the ability of FlrCC and the aforesaid mutants to hydrolyze ATP through Malachite green assay (32, 33). The reactions were carried out in a buffer containing 50 mm Tris-HCl (pH 8.0), 300 mm NaCl, and 5 mm MgCl2. Each protein was tested with Malachite green without ATP to measure the contaminant inorganic phosphate if any, and the negligible absorbance thus obtained at 630 nm, ranging between 0.001 and 0.003, was subtracted from the absorbance produced by that protein upon hydrolysis of the added ATP. The highest rate of ATP hydrolysis was observed for FlrCC (Fig. 4B). We also tested the effect of Mg2+ in ATP hydrolysis by measuring generation of inorganic phosphate from ATP by FlrCC in the absence of Mg2+. About 90% reduction in the rate of ATP hydrolysis was observed for FlrCC without Mg2+ (Fig. 4B). Interestingly, the ATP hydrolysis rate of R349A was as low as FlrCC without Mg2+, whereas R319A showed about 30% reduction compared with that of FlrCC + Mg2+ (Fig. 4B). These observations further suggest that although Arg-319 contributes to ATP binding through its interactions with ribose, Arg-349 that stabilizes γ-phosphate is more effective in terms of ATP binding and hydrolysis. Interestingly, although Glu-286 has no direct interaction with AMP-PNP, the mutant E286A shows about 20% reduction in the rate of ATP hydrolysis (Fig. 4B).
The GAFTGA motif of loop L1 is implicated in σ54 binding for several bEBPs (2). In FlrC, L1 and its neighboring loops L2 and L3 do not experience any dramatic conformational rearrangement upon AMP-PNP binding because extensive intra- and inter-subunit interactions provide an in-built architecture to this region (Fig. 5, A–D). Although the residues of α3 and α4 pack hydrophobically with those of β2 and β3 at the base of L1, salt bridge interactions among Arg-254, Glu-204, and Glu-220 constitute an intra-subunit electrostatic equilibrium (Fig. 5C). Protruded conformation of L3 is supported by the hydrophobic packing at the interior of that loop. A unique mode of intra-subunit hydrophobic packing is observed between L1 and L3 with the participation of Met-195 of L3 and Tyr-203, Phe-208 of L1 (Figs. 1 and and55C). Furthermore, inter-subunit interactions of L3 with trans-acting α4 and sensor II provide added compactness to this region.
B-factor plot of Nt-free and -bound structures, however, shows that upon AMP-PNP binding L1 and L3 loops of FlrCC acquire high atomic vibration (Fig. 5, E and F). Because similar enhancement of vibration occurs in L1 and L3 loops of all the protomers (Fig. 5, E and F), inter- and intra-protomeric interactions are not compromised (Fig. 5, C and D). However, slight asymmetry is observed in the central channel of AMP-PNP-bound FlrCC. The average diameter of the central channel is 26 Å for the Nt-free FlrCC, but it varies between 23 and 26 Å for the AMP-PNP-bound form. This asymmetry is attributed to a small displacement of L1 upon AMP-PNP binding, which is maximum (2.2 Å) for D chain and minimum (1 Å) for F chain. The movement of α2 toward the ATP binding pocket causes a shift of β2 and β3 (~1.5 Å) away from the pocket, and eventually the shift is propagated to L1 through L3 culminating to a cartwheel kind of movement of L1, L2, and L3 upon AMP-PNP binding. Interestingly, similar high thermal vibration was also observed in L1 loops of ATP bound NtrC1C structure (Fig. 5G).
In-built architecture of L1, L2, and L3 loops prompted us to qualitatively assess the binding ability of FlrCC with σ54 through in vitro pulldown assays. σ54 having N-terminal His6 tag was immobilized on Ni2+-NTA resin. The resin was then washed thoroughly and incubated with Nt-free FlrCC as well as FlrCC treated with ATP and AMP-PNP individually. Our results consistently showed interaction between σ54 and Nt-free, AMP-PNP and ATP-bound FlrCC (Fig. 5H). Only FlrCC, FlrCC + ATP, and FlrCC + AMP-PNP without σ54 were used as controls to see the basal level of adherence of FlrCC (if any) with Ni-NTA in free and Nt-bound states. Only σ54 nullifies the possibility of any contaminating band. Variation in ATP or AMP-PNP concentration did not show any measurable variation in binding (data not shown).
We have investigated the oligomeric states of Nt-free as well as AMP-PNP and ADP + AlF3 (that mimics transition state)-treated FlrCC in solution through gel filtration experiments in Superdex-200 column. All three profiles consistently show a major presence of the heptamers with a minor trailing at the lower molecular weight region (Fig. 6A). The observations indicate that the state of oligomerization of FlrCC remains intact upon Nt binding.
To further investigate, we performed DLS experiments with the eluted fractions corresponding to the peak region and trailing region of gel filtration. The peak regions of Nt-free FlrCC, AMP-PNP-treated FlrCC, and ADP + AlF3-treated FlrCC showed a monomodal population with a hydrodynamic radius RH of 6.6 ± 0.3 nm (Fig. 6B) that corresponds to the molecular mass of 277 ± 30 kDa. An oligomeric assembly with a central pore is expected to have a larger hydrodynamic radius than the compact globular proteins of similar molecular weight. The trailing region identified species with RH of 4.5 ± 0.1 and 5.1 ± 0.1 nm that correspond to molecular masses 115 ± 10 and 150 ± 10 kDa, which probably appear due to gradual dilution. Nonetheless, DLS results further established that the major oligomeric structure of FlrCC in solution is not influenced by Nt binding (Fig. 6B).
Two models were proposed to explain how hydrolysis is coordinated in the AAA+ family of proteins (38). Homogeneous Nt occupancy was observed for a number of AAA+ protein crystal structures, such as SV40 helicase LTag (2, 39,–41), where Nt binds simultaneously to all the pockets of the oligomer with full occupancy, supporting a model of concerted/synchronized hydrolysis. Other AAA+ structures showed mixed occupancy with ATP/ADP, which supports a model of sequential hydrolysis (42, 43). The 1:1 binding stoichiometry between FlrCC monomer and AMP-PNP indicates that all seven pockets of the heptamer are able to act simultaneously as potential ATP-binding sites (Fig. 2E). Our structural results also show that AMP-PNP binds to all seven pockets of FlrCC heptamer with full occupancy (Fig. 2, D and F) which resembles the “all or none” Nt-binding model proposed in support of synchronized hydrolysis. Unlike PspF or NtrC1, where inter-subunit interactions confer cooperativity in Nt binding (8, 44), FlrCC, in its heptameric state, renders an intriguing cis-mediated all or none ATP binding without any Nt-dependent subunit remodeling.
A unique closed to open type conformational change of Walker A facilitates ATP binding and subsequent hydrolysis in FlrC (Figs. 3 and and4).4). Interestingly, the conformation of Walker A in the “native” (ADP-bound) NtrC1 (PDB code 1NY6) and ATP-bound NtrC1C (PDB code 3M0E) (8, 13) closely resemble the open conformation of Walker A of FlrCC implying that no such rearrangement of Walker A is required in NtrC1 (Fig. 2G). In FlrC, the closed conformation of Walker A exerts steric hindrance to the β- and γ-phosphates of the incoming AMP-PNP (Fig. 3F). A closed to open movement, assisted by the trans-acting sensor I residue Glu-286, is thus required for efficient binding (Fig. 3, F and G). Upon AMP-PNP binding, Lys-165, along with the helix α2, moves toward the ATP binding pocket and stabilizes β- and γ-phosphates. Consequent repositioning of the neighboring residues like Val-167 of α2, Leu-312 and His-315 of α7, and N-terminal Val-132 constitutes a hydrophobic pocket capped by Trp-299 to house the adenine base (Fig. 3D).
Further stabilization to AMP-PNP binding is incurred by the novel cis-acting Arg-319 of α7 that binds the ribose of AMP-PNP (Fig. 3C). The corresponding residue Lys-327 of NtrC1 resides about 6 Å away from the ribose (8). Likewise, Lys-230 of PspF, instead of interacting with ribose, forms a salt bridge with neighboring Glu-234 (45). In FlrC, the inclination of α7 toward the ATP binding pocket makes the interaction of Arg-319 with ribose feasible, which was not the case for NtrC1 or PspF.
Asn-272 of sensor I and Arg-349 of sensor II contribute significantly in positioning the γ-phosphate. Comparison of the Kd values and release of Pi showed that substitution of Arg-349 by Ala drastically reduces ATP binding and hydrolysis, although the effect of R319A is not so damaging (Table 3 and Fig. 5). ATP binding and hydrolysis are thus influenced more by the stabilization of γ-phosphate than that of the ribose. However, the contribution of Arg-319 cannot be under-rated. Only a 4-fold increase in Kd upon ADP binding (Table 3 and Fig. 4A) suggests that Arg-319, because of its interaction with ribose, acts as a deciding factor in the rate of ATP hydrolysis by retarding the release of ADP.
Walker B and sensor I also play regulatory roles in ATP binding and hydrolysis. Asp-230 of Walker B is positioned by the neighboring Asn-187 and Thr-270 for efficient Mg2+ binding. Conversely, the next residue Glu-231 stabilizes the closed conformation of Walker A in the Nt-free state and that of Asn-272 in the AMP-PNP-bound state, features which point toward a regulatory role of Glu-231. Additionally, Asn-272 of sensor I has emerged as a new “cis-acting Asn switch” in FlrC because of its dual role in stabilizing the closed conformation of Walker A and in sensing γ-phosphate upon AMP-PNP binding (Fig. 3G).
A trans-acting Arg, defined as R-finger, is observed in quite a few AAA+-ATPases that participate in ATP binding and hydrolysis, although the mechanistic conclusions on the role(s) of such an R-finger have yet to be drawn (2). In NtrC1 or PspF, trans-acting arginine(s) belonging to the “RXDXXXR” motif of sensor I stabilize the γ-phosphate of ATP. Strikingly, despite the conservation of 285REDXXXR291 in FlrC (Fig. 1), neither of these two arginines participates in AMP-PNP binding. Rather, trans-acting Arg-285 stabilizes the cis-acting L3 loop beneath the γ-phosphate (Fig. 5D), and the Arg-291 side chain stays about 10 Å away from the phosphates of AMP-PNP (Fig. 3C). Despite that, the role of the 285REDXXXR291 motif in ATP binding and hydrolysis of FlrC cannot be ignored. Stabilization of γ-phosphate is the most essential step in ATP hydrolysis, and in FlrC, the cis-acting Arg-349 of sensor II is one of the major contributors in this direction (Figs. 3G and and4).4). Even in the Nt-free state, Arg-349 remains oriented toward the ATP binding pocket through the polar interaction with trans-acting Glu-286 and by the hydrophobic packing with trans-acting Tyr-290, both of which belong to the 285REDXXXR291 motif. Additionally, trans-acting Glu-286 stabilizes the open conformation of Walker A and thus serves as a novel trans-acting “Glu switch” that facilitates ATP binding and hydrolysis.
The precise organization of the interface between two adjacent subunits is the key element for oligomerization of bEBPs. Starting from very similar monomeric structures, when PspF(1–275) and NtrC1C organize into hexamers and heptamers, respectively, their inter-protomeric interfaces adopt different configurations (11). Interestingly, the unique architecture of the FlrCC monomer leads to a heptamer with a much wider central channel (diameter ~26 Å) compared with NtrC1C (diameter ~17 Å) (Figs. 2C, and and6,6, C and D). The cis-acting mode of AMP-PNP binding in FlrCC is also guided by the characteristic architecture of the individual monomer. The inclination of α7, coupled with the protrusion of L3 loop, constricts the ATP binding pocket of FlrCC protomer by about 7 Å compared with that of NtrC1C (Figs. 2C, and and6,6, C and D). Because of this constriction, the adjacent protomers in FlrCC stay relatively away at the ATP-binding site (Fig. 6, C and D), eventually occluding the trans-acting residues from ATP binding. However, the binding and hydrolysis of ATP are not compromised in FlrC. Packing of the adenine base in a hydrophobic pocket, stabilization of the ribose, and most importantly the locking of the γ-phosphate make an efficient cis-mediated productive ATP binding without any direct contribution from the trans-acting residues.
The necessary amount of ATP hydrolyzed by an AAA+-ATPase might be different for different functions (11). An all or none binding of AMP-PNP + Mg2+ complex (with waters molecules in near apical positions) in FlrCC points toward a synchronized mechanism of ATP hydrolysis. Local conformational changes occur here in a subtle manner to accommodate ATP without destroying the inter-protomeric interactions (Table 2). ATP binding and hydrolysis are, however, regulated by the conformational restriction of Walker A and the retarded expulsion of ADP. Upon requirement of ATP hydrolysis, Walker A moves to the open conformation, stabilized by the trans-acting Glu switch as a result of which the hydrophobic pocket for the adenine base is created. Hydrolysis of ATP and release of Pi then allow Walker A to return to its closed conformation that eventually destroys the hydrophobic pocket for adenine base causing expulsion of ADP.
The bEBPs typically bind to the enhancer elements far upstream of the σ54-binding site, and upon DNA looping interact with RNAP-σ54 at the promoter (2). FlrC, together with the flagellar regulators FlrA of V. cholerae and FleQ of P. aeruginosa, forms a new set of bEBPs that bind to the enhancer elements located downstream of the σ54-binding and transcriptional start sites (17, 24, 46). Interestingly, a sequence comparison of these flagellar regulators with NtrC1, NtrC4, and PspF showed that they consistently possess a 40–50-residue longer linker region between ATPase and the DNA binding domain. Although the process of DNA looping seems to be incompatible with such downstream enhancer binding (17, 24, 46), their mechanism of RNAP-σ54 binding at the promoter to initiate DNA melting has yet to be investigated.
Cross-linking and EM reconstruction studies have recently shown that only one oligomeric assembly of bEBP is sufficient enough to simultaneously bind RNAP-σ54 and the upstream promoter region using varying numbers of participating L1 loops (2, 9, 47, 48). Notably, σ54 binds bEBP and RNAP through its highly conserved regions I and III. Considering the very high degree of sequence conservation at these two regions of σ54 and the conserved nature of the GAFTGA loop of bEBPs, a similar binding stoichiometry may be expected between FlrC heptamer and RNAP-σ54.
Extensive studies on PspF or NtrC1, however, showed that participation of L1 loops in binding σ54 at the promoter is actually guided by Nt binding, hydrolysis-induced subunit remodeling, and a spatio-chemical environment offered by the asymmetric arrangement of the L1 loops (8, 44, 45). Although our results on FlrCC exclude the possibility of any such Nt-dependent subunit remodeling, even in the presence of ADP.AlFx, the probability of local structural changes are not ruled out. In FlrC, loop L1 and its neighboring regions have in-built architecture that is qualitatively supported by in vitro pulldown assay with σ54 (Fig. 5, C, D, and H). It seems that FlrC has a potential to recognize RNAP-σ54 even without Nt binding, although the exact scenario in the presence of promoter has yet to be investigated. This is not a very rare observation because Lee and Huber (49) reported that Rhizobium meliloti C4-dicarboxylic acid transport protein D cross-links with σ54 even without Nt binding. In FlrC, the small asymmetry created in the central pore and their elevated thermal vibration upon AMP-PNP binding probably point toward the generation of local asymmetry upon ATP binding, which may enhance during ATP hydrolysis, as observed previously for the other bEBPs (8, 9), although the extent of asymmetry may differ in this case. Dimension of the central pore and disposition of the L1 loops around the central pore would seemingly play a crucial role in causing asymmetry (8, 9). The central pore of FlrCC heptamer is strikingly wider having a diameter of ~26 Å compared with the other bEBP structures determined so far (~17 Å for NtrC1C heptamer) (Fig. 6, C and D). In spiral or split ring hexamers, asymmetric movement of the L1 loops makes them separate enough to simultaneously interact with RNAP-σ54 and the upstream promoter region (9, 48). However, the extent of asymmetry to make a productive complex with σ54 at promoter may not be that dramatic with an ATPase ring having much wider central pore where adjacent GAFTGA loops already stay substantially away (~12 Å in FlrCC) from each other.
Collectively, the structure of FlrC with a wide central pore in the ATPase ring guided by unique ATPase site architecture, cis-mediated synchronized ATP binding, and hydrolysis without subunit remodeling and a long linker that connects the ATPase ring with the DNA binding domain are indicative of a novel transcriptional initiation mechanism for this bEBP, involved in downstream enhancer binding. In the future, additional structures in the presence of ADP.AlFx combined with systematic biochemical and structural analysis of σ54 binding in the presence of cognate promoters should facilitate progress in defining the mechanochemical action of such unusual bEBPs.
We thank the EMBL staff Dr. Hassan Belrhali and Dr. Babu A. Manjasetty for providing support in data collection on the beamline BM14 at EMBL-DBT, ESRF, France. We also thank to Prof. B. Gopal of IISc, Bangalore, India, for support at various stages of this work. We thank the staff associated with DBT-CU IPLS (BUILDER) program at University of Calcutta for their help in different experiments. We also thank Seema Nath of SINP, Kolkata, India, for generous help in analyzing DLS data. We are grateful to Rev. Dr. J. Felix Raj, SJ, Principal, St. Xavier's College, Kolkata, India, for constant support and encouragement.
2The abbreviations used are: