The bacterial type IV pilus (T4P) is a versatile nanomachine that functions in pathogenesis, biofilm formation, motility, and horizontal gene transfer. T4P assembly is powered by the motor ATPase PilB which is proposed to hydrolyze ATP by a symmetrical rotary mechanism. This mechanism, which is deduced from the structure of PilB, is untested. Here, we report the first kinetic studies of the PilB ATPase, supporting co-ordination among the protomers of this hexameric enzyme. Analysis of the genome sequence of Chloracidobacterium thermophilum identified a pilB gene whose protein we then heterologously expressed. This PilB formed a hexamer in solution and exhibited highly robust ATPase activity. It displays complex steady-state kinetics with an incline followed by a decline over an ATP concentration range of physiological relevance. The incline is multiphasic and the decline signifies substrate inhibition. These observations suggest that variations in intracellular ATP concentrations may regulate T4P assembly and T4P-mediated functions in vivo in accordance with the physiological state of bacteria with unanticipated complexity. We also identified a mutant pilB gene in the genomic DNA of C. thermophilum from an enrichment culture. The mutant PilB variant, which is significantly less active, exhibited similar inhibition of its ATPase activity by high concentrations of ATP. Our findings here with the PilB ATPase from C. thermophilum provide the first line of biochemical evidence for the co-ordination among PilB protomers consistent with the symmetrical rotary model of catalysis based on structural studies.
The bacterial type IV pilus (T4P) is a filamentous cell surface structure with versatile functions in many biological processes. The T4P machinery (T4PM), which is encoded by a set of pil genes widely distributed in bacteria, enables the construction and proper functioning of T4P in a multitude of organisms [1–4]. Morphologically, the T4P filament is ∼5–8 nm in diameter and extends a few microns beyond the cell body . This thin filament is assembled from membrane-embedded monomeric pilins by the T4PM which traverses the entire cell envelope of Gram-negative bacteria [6–8]. Functionally, T4P is best known for its involvement in bacterial surface motility, biofilm formation, pathogenesis, and natural transformation [1–4]. Essentially all the components of the T4PM are required for these functions, except that natural competence may only require a subset of the T4P proteins depending on the specific organisms. In bacterial pathogenesis and biofilm development, T4P mostly functions as an adhesin, whereas it is the retraction of T4P that powers cell movement in bacterial motility. In addition, the T4P has been found to function in electron transport and bacterial conjugation. With the versatility of a Swiss Army knife , the T4PM have been investigated and dissected extensively as an extraordinary biological nanomachine.
The T4P is classified into subtype a (T4aP) and subtype b (T4bP) with structural and functional implications [1–3]. Historically, the length of the signal sequence and the size and structure of the major pilin were used to differentiate T4aP and T4bP. The major prepilin of T4aP systems usually have short signal sequences and their mature pilins are also smaller than those of T4bP. As more genomes are sequenced, it has become apparent that these two subtypes can be distinguished by gene arrangements and the presence of selected T4P components. For example, a pilMNOPQ gene cluster, which has only been associated with the T4aP but not the T4bP system, provides a reliable identifying feature of a T4aP system. While both subtypes of T4P may function in attachment, biofilm formation, and competence, only T4aP has been demonstrated to mediate bacterial surface motility. It has been established that the recurrent cycle of T4aP extension and retraction is the underlying mechanism for T4P-mediated motility [9,10]. That is, the distal end of an extended T4aP can be tethered or attached to a solid surface and the ensuring retraction of the T4P results in the forward motion of a bacterium. The force generated by T4P retraction can be up to 150 pN, making T4P the strongest biological motor currently known [11,12]. PilT , which is the T4aP retraction ATPase, is present in all T4aP but only some of the T4bP systems . The type II secretion system (T2SS), which is related to the T4PM phylogenetically, lacks both the pilMNOP gene cluster and PilT [2,14,15].
The motor ATPase PilB along with other T4P proteins are responsible for the assembly and disassembly or extension and retraction of the T4aP [1–3]. The proteins from the pilMNOPQ genes together with the transmembrane protein PilC appear to form a stable T4P basal body [6,16–19]. Depending on its association with the PilT or the PilB ATPase, this stable basal structure allows either the retraction or extension of the T4aP filament. Presumably, if the retraction motor PilT is docked with the basal body, the pilus is retracted one pilin at a time as the T4P is dissembled at its base [6,13,20]. The T4P assembly motor ATPase PilB is proposed to facilitate the reverse process or T4P extension. That is, when it associates with the basal body, it assembles the monomeric pilins embedded in the cell membrane into a growing T4P filament in a stepwise fashion. Structural studies of PilB and cryo-EM observations of the T4aP machinery in situ support a rotation model of T4P assembly [6,20–22]. The N-terminal side of a PilB hexamer is proposed to directly interface with a PilC dimer on the cytoplasmic side of the cell where PilM is proposed to form a membrane-associated dodecameric ring encamping the hexameric PilB . During T4P assembly, PilB would catalyze ATP hydrolysis through a rotary mechanism to rotate the PilC dimer situated in the cytoplasmic membrane [6,20–22]. The rotating PilC would, in turn, scoop up the monomeric pilins from the membrane and assemble them into the growing end of the pilus . PilT is proposed to replace PilB and rotate PilC in the opposite direction during disassembly [6,16,17]. The structure of the ATPase core of the Thermus thermophilus PilF/B (TtPilB) and Geobacter metallireducens PilB (GmPilB) led to a symmetrical rotary model for their catalysis [20–22], consistent with the rotation model of T4aP assembly supported by cryo-EM studies . In addition, PilB has recently been implicated in signaling in biofilm formation and c-di-GMP regulation [23,24]. However, there has been no biochemical studies to support the rotary mechanisms of catalysis by PilB. The limiting factor appears to be the low ATPase activity of PilB from Myxococcus xanthus and Pseudomonas aeruginosa, the two model systems for the studies of T4P-mediate bacterial motility [16,25,26].
The report here provides the first analysis of steady-state kinetics of PilB from the thermophilic and phototrophic bacterium Chloracidobacterium thermophilum [27–29]. Analysis of the genome sequence of this bacterium identified a full complement of genes encoding a complete T4aP machinery. C. thermophilum PilB (CtPilB) was expressed heterologously in Escherichia coli and purified by FPLC. Analysis indicated that CtPilB is a hexamer with highly robust ATPase activity at moderate temperatures in comparison with the hyperthermophilic TtPilB . Further studies show that this PilB displays a complex kinetic profile, consistent with it being a hexamer whose protomers hydrolyze ATP with co-ordination. During the course of this study, it was also discovered that there is a mutant pilB allele in the genomic DNA of C. thermophilus from an enrichment culture [28,29]. The mutant PilB variant has significantly reduced ATPase activity in comparison with the wild type. Nevertheless, both the mutant and the wild-type proteins showed similar bidirectional kinetic profiles overall, consistent with a symmetrical rotary mechanism of ATP hydrolysis by the hexameric PilB.
Growth media, enzymes, and chemicals
Lysogeny broth (LB)  was used for the growth of E. coli cells at 37°C. Ampicillin was supplemented at 100 µg/ml when necessary. Pyruvate kinase and l-lactic dehydrogenase from rabbit muscle were purchased from Sigma. Restriction endonucleases and other DNA-modifying enzymes were from New England BioLabs. ATP and NADH were from Research Product International, Phosphoenolpyruvic acid (PEP) from Alfa Aesar Chemicals and Malachite green oxalate salt from MP Biomedicals. Media components, other common chemicals, and salt were either from Fisher Scientific or Sigma.
Construction of CtPilB expression plasmids
The coding region for CtPilB was amplified by PCR using the C. thermophilum genomic DNA [28,29] as the template and the primers CtPilB_F (CGGATCCTCAGCAAAACTTGGTGAAGT) and CtPilB_R (TGACAAGCTTAGTGATGGTGATGGTGATGCAGAACCGTCTCCCGCACCA). The PCR fragment was digested with HindIII and cloned into pQE16 (Qiagen) digesting with BamHI, treated with Mung Bean Nuclease, and followed by digestion with HindIII. The resulting plasmids, pAS101 and pYY101, express the wild-type CtPilB and the R436C variant, respectively, with a 6 × His affinity tag at their C-termini. Both plasmids were sequenced at the Genomics Research Laboratory in the Biocomplexity Institute of Virginia Tech.
CtPilB expression and purification
A culture of E. coli with either pAS101 or pYY101 in 200 ml of LB was grown overnight at 37°C on a rotatory shaker at 250 rpm. Six flasks containing 1 l of LB each was inoculated with 25 ml of the overnight culture. Once the OD600 of these cultures reached ∼0.6 after incubation at 37°C, IPTG was added to a final concentration of 0.1 mM to initiate induction of protein expression. After 5 h of induction at 37°C, cells were harvested by centrifugation at 4°C. The cell pellets were weighted and resuspended at 0.2 g/ml in cold buffer A (25 mM Tris–HCl at pH 7.4, 500 mM NaCl, 25 mM imidazole, and 10% glycerol) with the addition of 0.3 mM PMSF and 5 mM β-mercaptoethanol. Cells were lysed by sonication using a Sonic Dismembrator (Model 300, Fisher) with 55% power for 10 cycles with 30 s ‘on’ and 30 s ‘off’. Cell lysis was confirmed by phase contrast microscopy. The cell lysates were centrifuged at 184 000×g for 30 min at 4°C and the supernatants were collected for the purification of CtPilB.
E. coli proteins were partially removed by subjecting the supernatants to 50°C for 20 min followed by incubation on ice for 30 min. Samples were subsequently centrifuged at 184 000×g for 20 min at 4°C. The recovered supernatant was filtered and loaded on a 30 ml Ni-NTA column equilibrated with buffer A. Weakly bound proteins were removed by washing with buffer A. CtPilB with the 6 × His tagged was eluted with a linear gradient of imidazole up to 500 mM. Peak fractions were pooled, concentrated, and filtered before loading on a HiPrep 26/60 Sephacryl TMS-300 column (GE Healthcare) for size exclusion chromatography (SEC). Gel filtration buffer (25 mM Tris–HCl, pH 7.4, 150 mM NaCl, 2 mM TCEP, and 10% glycerol) was used for the equilibration of the column and the elution of CtPilB. Fractions with CtPilB were pooled and protein concentration was determined using a Bradford assay kit (Bio-Rad). The purification process was monitored by SDS–PAGE and Coomassie Blue staining .
Analysis of CtPilB by analytical SEC
About 2.5 mg of CtPilB in gel filtration buffer was applied to a Superdex 200 10/300 GL analytical size exclusion column with a void volume (Vo) of 8.0 ml as determined by the elution of Blue Dextran. The column was further calibrated using protein markers with molecular weights (MWs) of 29, 69, 200, 443, and 669 kDa (Sigma). The elution of CtPilB was monitored by UV absorption, and its apparent MW was calculated based on its elution volume (Ve) and those of the MW standards.
Optimization of conditions for the CtPilB as an ATPase
An endpoint assay based on the Malachite Green (MLG) method  measuring the levels of phosphate was used to optimize the condition for the ATPase activity of CtPilB. Each reaction was set up with 5.2 µg/ml CtPilB and 1.5 mM ATP with varying buffers and salts in a PCR tube. The reactions in PCR tubes were allowed to occur at desired temperatures in an Eppendorf Mastercycler gradient thermocycler for 15 min before switching to 4°C for stoppage. Reactions with buffer instead of CtPilB or ATP were included as negative controls. An aliquot of each reaction mixture was transferred to a 96-well plate which included a concentration series of KH2PO4 as phosphate standards. MLG reagents (0.034% MLG, 10 mM ammonium molybdate, 1 N HCl, 3.4% EtOH, and 0.01% Tween 20) were mixed into each well and incubated for 10 min at room temperature before the absorbance at 620 nm was measured using a Tecan SPECTRAFluor Plus plate reader. The amount of phosphate in each sample was calculated against the standard curve constructed from the KH2PO4 concentration series. The activity of CtPilB was computed by dividing the amount of phosphate (in nanomole) by the amount of protein (in milligram) and the reaction time (in minutes) and expressed in nmol Pi/mg PilB/min. All assays were performed in triplicates for each experiment.
Studies of steady-state kinetics of the CtPilB ATPase
The buffer used for kinetic studies of CtPilB by the MLG assay contained 50 mM of TAPS (N-[Tris(hydroxymethyl)methyl]-3-aminopropanesulfonic acid) and Tris each at pH 8.7, 75 mM KCl, 50 mM NaOAc, 5 mM MgCl2, and 50 µM ZnCl2. For experiments with 12 different [ATP], aliquots from reactions with CtPilB at 5.2 µg/ml were withdrawn after 1, 2, 3, and 4 min of incubation at 54°C. For those with 16 different [ATP], samples were taken at 1, 2, and 3 min. These samples were quickly put on ice at their respective time points and trichloroacetic acid was added to a final concentration of 4% to inactivate PilB. The amount of phosphate in each sample was quantified using 96-well plates as described above. The amounts of phosphate were plotted against time for each ATP concentration, and the rate of ATP hydrolysis was calculated from the slope of the curve. Preliminary experiments with different [ATP] established that the amounts of phosphates released from ATP hydrolysis by CtPilB remained linear up to 8 min within the [ATP] range used in this study under our experimental conditions. Non-enzymatic hydrolysis of ATP was below the detection limit of the MLG assay within the timeframe of these experiments under our experimental conditions, which is consistent with a recent report .
The kinetic of CtPilB was additionally analyzed by an enzyme-coupled ATPase assay [35,36] at 37°C using the optimum buffering condition as determined above. In this assay, the ADP generated by ATP hydrolysis is enzymatically converted back to ATP with the oxidation of NADH to NAD+. The decrease in NADH can be detected by the decrease in absorbance at 340 nm. The reactions, which were optimized prior to the kinetic studies of CtPilB, contained pyruvate kinase at 20 U/ml, l-lactic dehydrogenase at 20 U/ml, 4 mM PEP, 0.5 mM NADH, and varying concentrations of ATP. The experiments were conducted at 37°C with a reaction volume of 200 µl per well in a 96-well plate with CtPilB at 52 µg/ml. The absorbance at 340 nm was monitored at 1-min intervals using a Tecan Infinite F200 Pro plate reader with temperature control. The rate of ATP hydrolysis was calculated by the slope of the absorbance over time in the linear range. Data analysis and calculations were performed using Microsoft Excel and curve fitting was conducted with KaleidaGraph Version 4.5.
Kinetic study of the CtPilB(R436C) variant as an ATPase
The kinetic analysis of the CtPilB(R436C) ATPase in steady state was conducted using the same procedure and the MLG assay as for the wild-type CtPilB, except that the amount of CtPilB(R436C) was 21.0 µg/ml.
Structural modeling of CtPilB
C. thermophilum has a full complement of genes for T4aP
The photosynthetic thermophilic acidobacterium C. thermophilum was first discovered in a microbial mat of hot springs in the Yellow Stone National Park, WY [27–29]. Our analysis of its genome sequence  indicated that it has a full complement of pil genes for the machinery of a T4aP system (Figure 1A). All the pil genes are encoded on the first of its two chromosomes and the most conserved among them are clustered at two genetic loci. One cluster, which includes pilM, pilN, pilO, pilP, and pilQ, likely forms an operon with all open reading frames (ORFs) reading on the negative/Crick strand of the chromosome. The other locus has pilB, pilT, and pilC genes, all three reading on the positive/Watson strand. The presence and the arrangement of the pilMNOPQ gene cluster indicate that C. thermophilum likely assembles a T4aP [1–3], which is consistent with the presence of pilT in the pilBTC cluster. Aside from another gene cluster for a probable T2SS, there are four other genes on this chromosome elsewhere encoding proteins with a prepilin-type signal sequence. Analysis suggests that CABTHER_RS02295 may encode the major pilin for the T4aP pilus of C. thermophilum while the other may encode three minor pilins.
T4P and T2SS components encoded on the C. thermophilum genome.
The arrangements of these pil genes in C. thermophilum resemble those in the δ-proteobacteria including M. xanthus , a model organism for studies of T4aP-mediated motility and functions [39–41]. This is in agreement with the close relationship of Chloracidobacteria with the δ-proteobacteria on the phylogenetic tree . Further sequence analysis indicates that the C. thermophilum PilB (CtPilB) is a T4aP assembly ATPase ortholog that aligns well with M. xanthus PilB (MxPilB) (Figure 1B) and PilB in other T4aP systems (Supplementary Figure S1). Among the best studied PilB orthologs, CtPilB shows the highest amino acid identity with PilB from M. xanthus and G. metallireducens, both of which are δ-proteobacteria that share a common root with acidobacteria on the phylogenetic tree . In addition, a chimera containing the N-terminus of MxPilB and the C-terminus of CtPilB supported T4P-mediated motility in M. xanthus (Supplementary Figure S2), indicating that CtPilB is likely functional in vivo and a reasonable candidate for the studies of the PilB ATPase in vitro.
Purification of CtPilB
CtPilB with a 6 × His tag fused to its C-terminus was expressed and purified using E. coli as a host (Figure 2). Briefly, the coding region of pilB was PCR-amplified from C. thermophilum genomic DNA (kindly provided by Dr Donald Bryant at The Pennsylvania State University)  and cloned into an IPTG-inducible expression vector. E. coli cells containing the recombinant expression plasmid were induced with IPTG to produce CtPilB. The soluble fraction of the whole cell lysates after a mild heat treatment was subjected to Ni-NTA affinity chromatography where CtPilB eluted ∼180 mM imidazole. CtPilB was further purified by preparative SEC where it eluted as a single peak, indicating uniformity of the oligomeric state of the purified protein.
Purification of CtPilB.
CtPilB is a hexamer
The oligomeric state of the purified CtPilB was examined by analytical SEC with a column calibrated with protein standards of known MW (Figure 3A). On this column, CtPilB eluted as a single peak centered at the elution volume of 10.25 ml (Figure 3A). This corresponds to an apparent MW of 422 kDa based on the standard curve from the elution of the MW markers (Figure 3B). This apparent MW is 6.12 times of 68.9 kDa, the theoretical MW of a CtPilB monomer with the 6 × His tag. In addition, CtPilB was analyzed by dynamic light scattering (DLS) in gel filtration buffer (Supplementary Figure S3A), which indicated a hydrodynamic diameter of 240 Å. This size is ∼100 Å larger than the diameter of the crystal form of the hexameric ATPase core of TtPilB which lacks the conserved N-terminus (Figure 1B) . It is also more than 50 Å larger than the PilM ring where PilB is proposed to fit . It appears therefore that CtPilB exists in vitro in a more extended form in solution than expected, which may explain the slight increase in its apparent molecular weight by SEC (Figure 3) over a theoretical hexamer of CtPilB. These results here indicate that the purified CtPilB is a hexamer, which is in contrast with M. xanthus PilB that purified as a monomer without a hexameric tag [16,25].
Molecular weight of CtPilB determined by analytical SEC.
CtPilB is a thermophilic ATPase requiring Mg2+ and Zn2+ for its activity
The ATPase activity of CtPilB was analyzed and optimized initially by an endpoint assay using the MLG method which measures the amount of phosphate released from ATP hydrolysis [43,44]. The optimum buffering conditions were found to require 50 mM of TAPS and Tris each, 75 mM KCl, and 50 mM NaOAc at the optimum pH of 8.7 (Figure 4A). As was expected for a member of the PilB ATPase family [21,45], CtPilB requires Mg2+ and Zn2+ for activity with the optimum concentration at 5 mM and 50 µM, respectively (Figure 4B). The optimum temperature for the ATPase activity of CtPilB was determined to be ∼54°C (Figure 4A), consistent with the temperature of the hot springs where C. thermophilus was first discovered [27–29].
Optimization of assay conditions for the CtPilB ATPase.
CtPilB is a highly robust ATPase
Steady-state kinetics of CtPilB was first analyzed at 54°C using the MLG assay and the conditions as optimized above. The rates of ATP hydrolysis remained linear up to 8 min in our assay for all ATP concentrations and the initial rate was calculated using time points up to 4 min (Figure 5A). As shown in Figure 5B, the activity of CtPilB shows a bidirectional response to increasing concentrations of ATP. The rate of ATP hydrolysis steadily rose with increasing [ATP] initially, approaching the highest at ∼1.5 mM ATP (Figure 5B). Interestingly, a slower rate of ATP hydrolysis is observed when [ATP] increased further. The kinetic profile of CtPilB can therefore be divided into two parts, an incline followed by a decline with a peak ∼1.5 mM ATP. The highest activity previously reported for a PilB ATPase was ∼20 nmol/mg/min for the hexameric TtPilB . The activities reported for PilB from mesophiles M. xanthus and P. aeruginosa were between 3 and 10 nmol/mg/min [16,25,26,46]. In comparison, the purified CtPilB is a more robust ATPase that can hydrolyze ATP over 700 nmol/mg/min (Figure 4).
Steady-state kinetics of CtPilB ATPase.
CtPilB ATPase exhibits substrate inhibition
The results in Figure 5B indicated that the rate of ATP hydrolysis by CtPilB increased with increasing [ATP] until ∼1.5 mM. The incline turns to a decline with a further increase in [ATP]. We next analyzed CtPilB in an [ATP] range of ≥1.2 mM with increasing data density to focus on the decline of CtPilB kinetics (Figure 6A). The results confirmed the bidirectional nature of the response of CtPilB to [ATP]. The highest rate of ATP hydrolysis by CtPilB, ∼700 nmol/mg/min, was observed at 1.5 mM. Either higher or lower [ATP] resulted in a lower rate of ATP hydrolysis. The declining slope at higher [ATP] suggests that CtPilB as an ATPase is inhibited by its substrate ATP.
Substrate inhibition of CtPilB ATPase.
However, during the time course of the ATPase assay, ATP is hydrolyzed to ADP which accumulates over time. It was theoretically possible that the accumulation of ADP inhibited the ATPase activity [47–49] of CtPilB at high [ATP] even before the first sample could be taken at 1 min (Figure 5A,B). In addition, because the ATP stock used in the experiment contained trace amounts of ADP (Alfa Aesar®), the increase in [ATP] led to a concurrent increase in ADP concentrations in the reactions even before enzymatic hydrolysis of ATP occurred. To examine if CtPilB displays ATP or substrate inhibition, an enzyme-coupled ATPase assay  was used additionally to study the kinetics of CtPilB. In this assay, ADP produced from ATP hydrolysis is continuously converted back to ATP by two other enzymes with the oxidation or expenditure of NADH stoichiometrically. The decrease in NADH can be continually monitored by spectrophotometry for kinetic studies. As a consequence, ATP is maintained at a constant concentration and there would be no significant accumulation of ADP over the course of the assay. Because two other enzymes are present in this assay, it was performed at 37°C. The highest rate for the CtPilB ATPase (Figure 6B), obtained from the linear range of this assay (Figure 6B, inset), is significantly reduced in comparison with the activity at 54°C (Figure 6). Analysis using the MLG-based assay at 37°C confirmed the temperature-dependent reduction of CtPilB activity at 37°C (Supplementary Figure S4). Despite the differences in the overall kinetic profile at these two temperatures, the rate of ATP hydrolysis by CtPilB clearly decreased at the higher range of [ATP] at 37°C (Figure 6B and Supplementary Figure S4). That is, the ATPase activity of CtPilB decreased with increasing [ATP] beyond 1.5 mM. The result (Figure 6) here indicates that the CtPilB ATPase is inhibited by its substrate ATP at high concentrations.
The incline in the kinetics of CtPilB ATPase is multiphasic
The rates of ATP hydrolysis by CtPilB at ≤1.5 mM ATP showed an upward trend with increasing [ATP]. However, no significant plateau or saturation in activity was observed at the high end of [ATP] as would be expected if CtPilB had conformed to Michaelis–Menten kinetics (Figure 5B). We subsequently analyzed CtPilB ATPase activity with increased data density in the concentration range of 0–1.5 mM ATP (Figure 7A) to further our understanding of CtPilB kinetics. The data here show clearly that CtPilB kinetics presents a triphasic profile with three regions of steady increase in activity with increasing [ATP]. CtPilB ATPase activity rose at low [ATP] and reached a first plateau ∼0.5 mM ATP. A second upward shift followed with an ensuing plateau ∼1.1 mM ATP before a third upward shift is observed. CtPilB, as a hexameric ATPase, is thus not an enzyme with simple Michaelis–Menten kinetics. Instead, it employs a more complex mechanism to carry out its cycle of ATP hydrolysis.
The incline of CtPilB ATPase kinetics is multiphasic.
Nevertheless, a preliminary analysis suggested that the rate of ATP hydrolysis by CtPilB at [ATP] ≤0.5 mM (Figures 5B and 7A) might follow the Michaelis–Menten steady-state kinetics. We therefore analyzed the rate of ATP hydrolysis by CtPilB in this concentration range with increased data density (Figure 7B). Analysis indicates that the data in this [ATP] range obey Michaelis–Menten kinetics (inset in Figure 7B). Curve fitting produced a theoretical Vmax of 541 nmol/mg/min and a Km of 0.164 mM with an R2 value of 0.996. These results indicate that CtPilB follows the steady-state kinetics of a simple enzyme at sub-millimolar concentrations of ATP despite its more complex kinetics at higher [ATP].
We next focused on the kinetics of CtPilB in the [ATP] range ∼1.0 mM with increasing data density (Figure 7C). This new data set recapitulated the bidirectional nature of the kinetic profile of CtPilB (Figures 5B and 6), showing peak activity at 1.5 mM ATP and lower hydrolysis rate on both sides of the curve (Figure 7C). More importantly, these results confirmed the second upward shift and the plateau ∼1.0 mM ATP in CtPilB activity (Figure 7A,C). We have been unable to achieve any reasonable fit for the data in Figure 7A,C to established enzyme kinetic equations, indicating complicated effects of ATP on the ATPase activity of CtPilB.
A conserved arginine is important for CtPilB activity
During the cloning and sequencing of the constructs for CtPilB expression in this study, it became apparent that there were two versions or alleles of pilB present in the C. thermophilum genomic DNA, which was isolated from cells that were fractionated from a mixed culture by differential centrifugation [28,29]. One version, which encodes the amino acid sequence presented in Figure 1B, is designated as the wild type in this study. Its protein was used for all the results presented thus far (Figures 2–7). The other version, which is in the published genome sequence of this organism , has a single-nucleotide polymorphism changing a CGT codon to TGT. This results in the substitution of Arg-436 with a Cys (R436C) in CtPilB. This Arg, which resides adjacent and C-terminal to the Walker B box (Figure 1B), is strictly conserved in PilB and related ATPases . Structural modeling based on the crystal structure of TtPilB  indicated that this Arg residue contributes to the positively charged surface of a channel between two adjacent protomers leading to a bound ATP molecule  (Figure 8A). It is noted that Arg-436 is distinct from the Arg finger residues (Arg-293 and Arg-310) of CtPilB residing N-terminal to the Walker A box (Figure 1B) .
ATPase activity of CtPilB(R436C) is severely reduced.
To examine the effect of the R436C substitution on CtPilB activity, the CtPilB(R436C) variant was expressed and purified similarly as the wild-type CtPilB for studies in vitro. DLS analysis of CtPilB(R436C) indicated a similar particle size (data not shown) as the wild-type CtPilB (Supplementary Figure S3). As shown in Figure 8B, the maximum activity of the CtPilB(R436C) variant as an ATPase was reduced to <40 nmol/mg/min at 54°C, a 18× fold change in comparison with its wild-type counterpart under the same assay conditions. This result indicates that a R436C mutation likely affects T4P assembly significantly at the cellular level in vivo. Nevertheless, the overall kinetic profile of this CtPilB mutant variant largely resembled that of the wild type (Figure 5B). That is, its ATPase activity increased with increasing [ATP] up to 1.5 mM and higher [ATP] inhibited its activity as was observed with the wild type (Figure 5A). These results, while confirming that CtPilB ATPase is inhibited by its substrate ATP at high concentrations, indicate that the highly conserved Arg-436 neighboring the Walker B box is crucial for the ATPase activity of PilB.
We demonstrate here that CtPilB is a natural hexamer (Figure 2 and Supplementary Figure S3) with robust ATPase activity in vitro (Figures 5–7 and Supplementary Figure S5). At the cellular level, the activity of PilB is expected to correlate with the rate of T4P assembly observed in vivo. P. aeruginosa T4aP extends ∼0.5 µm/s . Since the change in length is 8–10 Å per pilin in the T4P filament , this rate of extension translates to ∼500 pilins processed by one T4PM per second or 30 000 per minute. In the current model, one T4PM is associated with one PilB hexamer . If we consider an expenditure of 2 ATPs per assembled pilin [6,20,21,50], 60 000 ATPs would be hydrolyzed per minute per PilB hexamer during active T4P assembly. TtPilB, an atypical hyperthermophilic PilB with a long N-terminal extension, hydrolyzed 13 ATPs per minute per hexamer in vitro . PilB from the mesophiles M. xanthus and P. aeruginosa hydrolyzed 1–3 ATPs in vitro per minute per hexamer equivalent [16,25,26]. In comparison, the highest rate of ATP hydrolysis by CtPilB here translates to ∼290 ATPs per minute per hexamer. This improvement in activity allowed the first kinetic studies of PilB, providing biochemical insights into the catalytic mechanism of PilB for the first time. On the other hand, the activity of CtPilB reported here is still more than two orders of magnitude below the T4P assembly rate in vivo, suggesting that other T4P proteins in or exposed to the cytoplasm play significant roles in stimulating the activity of PilB during T4P assembly. For perspectives, a bacterial cell with a volume of 1.0 µm3  would contain ∼600 000 ATP molecules at a given moment at 1 mM intracellular [ATP]. The ATP pool is replenished in part by the bacterial F0F1 ATP synthase which has a turnover rate up to ∼12 000 molecules per minute per enzyme in vitro . This enzyme can also function in the reverse as a rotary motor ATPase capable of hydrolyzing ∼300 000 ATP molecules per minute in vitro at an optimum temperature . The rate of CtPilB as an ATPase is ∼1000-fold lower in comparison, despite its higher activity than its counterparts currently in the literature [16,25,26].
The steady-state kinetics of CtPilB exhibits two striking features (Figures 5–7 and Supplementary Figure S5) that distinguish it from Michaelis–Menten kinetics. One is substrate inhibition as signified by the downward shift or the decline in the kinetics of CtPilB at high [ATP]. The other is the multiphasic nature of the kinetic incline before the decline starts at [ATP] >1.5 mM. The size of CtPilB as determined by DLS (Supplementary Figure S3B) was unchanged in the presence and absence of a non-hydrolyzable ATP analog in the enzyme activity buffer. The reduced ATPase activity of CtPilB at high concentrations of ATP is therefore not due to the disintegration of the CtPilB hexamer into monomers or lower orders of oligomers. We suggest that these complex kinetics behaviors reflect co-ordination among protomers and the highly dynamic nature of CtPilB as a hexameric ATPase in vitro. These observations are consistent with the proposed rotary models of catalysis by PilB and of the T4P assembly by T4PM [6,20–22] as discussed next.
The substrate inhibition of CtPilB may be explained by increased rigidity when ATP occupies the binding sites of all six protomers of a PilB hexameric ring. Simulations based on the crystal structure of PilB and the rotary mechanism suggest that the conformational changes in the three pairs of PilB protomers are highly co-ordinated . The proposed mechanism of force generation by PilB as a motor ATPase requires the dynamic and fluid movement of all protomer pairs in sync as hinged blocks . This means that the co-ordinated movement among the different protomers requires the overall fluidity of the hexameric ring during catalysis. As such, increased rigidity across the hexameric ring is expected to interfere with the requirement of such dynamic movements during the ATP hydrolytic cycle. It is possible that the two-fold symmetry of PilB, which is considered to be critical for its ATPase activity in the current model [20–22], may be broken at high [ATP]. The Archaeoglobus fulgidus hexameric ATPase GspE, which crystalized as an asymmetric hexamer, was found to assume a six-fold symmetry at high concentrations of an ATP analog, for example . This lends support to the proposed mechanism of substrate inhibition described above. This mechanism is also reminiscent of the inhibition of the hexameric ATP synthase by ADP [55,56]. In this case, the co-operativity among the protomers is blocked by excess ADP as a substrate which entraps this rotary enzyme in an inhibited state. There are additional models for substrate inhibition  that may also apply to CtPilB. In this context, there are at least two other hexameric ATPases, RuvB and 26S proteasomes, that are known to be similarly inhibited by ATP at high concentrations [58–60].
The rise and saturation of CtPilB activity prior to substrate inhibition do not obey classical Michaelis–Menten kinetics either (Figures 5, 7 and Supplementary Figure S5). Instead, there are perhaps three phases in the incline prior to the decline at >1.5 mM ATP. The first phase covers [ATP] up to 0.5 mM, the second from 0.6 to 1.1 mM and the third from 1.2 to 1.5 mM. The first phase fits the Michaelis–Menten equation quite well (Figure 7C). That is, in the sub-millimolar [ATP] range, CtPilB behaves as a simple ATPase with a Km ∼0.16 mM and its activity plateaus ∼0.5–0.6 mM ATP (Figure 7B). As [ATP] increases further, the activity of CtPilB rises again until it reaches a second plateau ∼1.0 mM ATP. In the third phase, the activity of CtPilB further elevates until it peaks at 1.5 mM ATP. The increases in CtPilB activity in the latter two phases can be explained by the engagement of additional active sites or protomers for catalysis in the hexamer at high [ATP] . That is, it is reasonable to assume that at a given moment, the protomers within an asymmetric hexamer may have different binding affinities for the substrate [20,21,61], resulting in multiple Km values in kinetics. Alternatively, these increases could be due to allosteric regulation by ATP binding to additional protomers in the hexamer. The observations with CtPilB are reminiscent of the non-Michaelis–Menten kinetics of the Na+/K+ ATPase  and the relief of attenuation of the RuvB ATPase by additional ATP [58,59]. Co-ordination and allosteric influence among promoters of these hexameric ATPases are provided as possible explanations of these observations [59,62]. If a PilB hexamer has one pair of protomers hydrolyzing ATP simultaneously as proposed in the symmetrical rotary model [21,22], the interaction of nucleotides with the ATP-binding pockets of a third or additional protomers may allosterically stimulate the overall activity of the hexamer [57–59]. ATP binding to neighboring subunits and intersubunit allosteric coupling is recently proposed to underlie the function of proteasomal ATPases . It is also possible that ATP may interact with unknown binding sites on the enzyme to exert an allosteric regulation [21,23,57,59].
The different phase of the kinetics of CtPilB (Figures 5–7 and Supplementary Figure S5), including substrate inhibition, all occur within physiologically relevant concentrations in bacteria. The intracellular [ATP] in bacteria can vary from below 1 mM to more than 3 mM [64–69]. The fluctuation in cellular [ATP] may be attributed to bacterial growth phase  and media compositions . For photosynthetic cyanobacteria, intracellular concentrations of ATP may vary in response to light and circadian rhythm [67,71]. Since C. thermophilum is a photosynthetic bacterium [27,28], its intracellular ATP concentration is likely responsive to light conditions and periodicity. In addition, it is known that the cyanobacterium Synechocystis regulates its T4P-mediated motility to perform phototaxis [72,73]. The cellular localization of PilB tends to correlate with the direction of T4P-mediated motility in M. xanthus and Synechocystis [17,74]. As such, the regulation of PilB activity by ATP may influence tactic and motility behavior in vivo. Heterogeneity in T4P-mediated motility has been observed in M. xanthus , P. aeruginosa, and Synechocystis . That is, only a subpopulation of bacterial cells are actively motile under the same growth conditions. While the reasons for this heterogeneity are not understood, variations in intracellular [ATP]  and their effects on PilB activity may partially contribute to this heterogeneity at the population level. In addition, nutrient levels are known to influence M. xanthus motility , and differences in ATP levels under varying conditions may modulate T4P-mediated bacterial surface motility. It is noted that the substrate inhibition of RuvB and 26S proteasomes also occurs at physiologically relevant concentrations of ATP [58–60].
We designated the CtPilB with Arg-436 as the wild type (Figures 1B and 8A). For one, this Arg is strictly conserved in the sequences of related ATPases . For another, the draft genome of another C. thermophilum became available recently , and it has an Arg instead of a Cys in the same position of CtPilB. This conserved Arg is no doubt critical for the activity of CtPilB as an ATPase (Figure 8), and the CtPilB(R436C) variant likely results in phenotypic heterogeneities in piliation if it exists in nature. Based on the crystal structure of TtPilB, this conserved Arg lines the surface of a positively charged channel that is proposed to escort the exit of the negatively charged phosphate resulting from ATP hydrolysis . This channel is formed by two adjacent protomers (Figure 8A). In CtPilB, it involves Arg-436 as well as Arg-293 and Arg-325 in one protomer and Arg-420, Arg-424, His-459 and Arg-469 in the other (Figures 1B and 8A). The highly reduced ATPase activity of the CtPilB(R436C) variant (Figure 8B) is consistent with Arg-436 being part of the channel with critical functions in the ATP hydrolysis cycle.
In the course of this study, it was observed that the steady-state kinetics of CtPilB at 37°C is distinct from that at 54°C (Figures 5–7 and Supplementary Figure S4), the latter being the optimum temperature for this enzyme (Figure 4). Supplementary Figure S4 represents the kinetic profile of CtPilB at 37°C analyzed by the MLG-based and the enzyme coupled assays, respectively, in the same [ATP] range as in Figure 7A. The incline no longer appeared multiphasic at 37°C. The other noticeable difference is the ATP concentration at which the rate of the enzyme approaches half Vmax, which we will refer to as K0.5 for convenience. At 37°C (Supplementary Figure S4), the K0.5 value is near 0.05 mM ATP, which was the lowest ATP concentration in these experiments. In comparison, the value of K0.5 is ∼10 times higher at 54°C (Figures 5B and 7A). These observations are consistent with the well-documented decrease of Km of Michaelis–Menten enzymes at lower temperature . As far as we are aware, the effect of low temperature on the kinetics of non-Michaelis–Menten enzymes is yet to be investigated experimentally. It is nevertheless reasonable to assume a profound effect of temperature on the properties of such an enzyme [79,80]. These observations here suggest caution in the analysis of the enzymatic properties of a protein at a non-optimum temperature.
C. thermophilum PilB
dynamic light scattering
M. xanthus PilB
open reading frames
size exclusion chromatography
type II secretion system
T4P subtype a
T4P subtype b
type IV pilus
Thermus thermophilus PilF/B
A.S. performed experiments. A.S. and Z.Y. designed studies, analyzed data, and prepared the manuscript.
This work was partially supported by the National Science Foundation [MCB-1417726 to Z.Y. and Florian Schubot]. A.S. is the recipient of a Fulbright fellowship.
We are grateful to Dr Donald Bryant at The Pennsylvania State University for kindly providing the C. thermophilus genomic DNA and Dr David Ward at Montana State University for permitting its use. We thank Dr Tim Long at Virginia Tech for access to and assistance in the use of DLS instruments. We appreciate Michael Klemba, Jordan Mancl, FS for their technical help and advice. We thank Yue Yang for constructing pYY101 and Keane Dye for critical reading of the manuscript.
The Authors declare that there are no competing interests associated with the manuscript.