Sequence-specific assembly of FtsK hexamers establishes directional translocation on DNA

Edited* by Stephen C. Kowalczykowski, University of California, Davis, CA, and approved September 9, 2010 (received for review May 30, 2010)
November 3, 2010
107 (47) 20263-20268


FtsK is a homohexameric, RecA-like dsDNA translocase that plays a key role in bacterial chromosome segregation. The FtsK regulatory γ-subdomain determines directionality of translocation through its interaction with specific 8 base pair chromosomal sequences [(KOPS); FtsK Orienting / Polarizing Sequence(s)] that are cooriented with the direction of replication in the chromosome. We use millisecond-resolution ensemble translocation and ATPase assays to analyze the assembly, initiation, and translocation of FtsK. We show that KOPS are used to initiate new translocation events rather than reorient existing ones. By determining kinetic parameters, we show sigmoidal dependences of translocation and ATPase rates on ATP concentration that indicate sequential cooperative coupling of ATP hydrolysis to DNA motion. We also estimate the ATP coupling efficiency of translocation to be 1.63–2.11 bp of dsDNA translocated/ATP hydrolyzed. The data were used to derive a model for the assembly, initiation, and translocation of FtsK hexamers.
FtsK is a highly conserved dsDNA translocase involved in chromosome unlinking in bacteria, coordinating the final stages of chromosome segregation with cytokinesis, and performing chromosome dimer resolution by interacting with XerCD (1). Escherichia coli FtsK is a 1,369 amino-acid protein that comprises an N-terminal integral membrane domain that localizes FtsK to the septum early in cell division; a long linker of unknown function; and a C-terminal translocase domain (FtsKC) that forms hexameric rings on dsDNA. FtsKC has been shown in single-molecule assays to translocate on dsDNA at speeds of ∼5 kb s-1 and against a stall force of ∼60 pN (27). FtsKC comprises FtsKαβ, which constitute the hexameric motor; and FtsKγ, a winged-helix domain that both interacts with XerCD bound to dif to promote chromosome unlinking and recognizes a specific 8 bp sequence, KOPS (FtsK Orienting/Polarizing Sequence; 5′-GGG[C/A]AGGG-3′). FtsKγ binds to KOPS as a trimer, with three FtsKγ winged-helix modules interacting with consecutive GGG, MA, and GGG elements (8). KOPS were identified by bioinformatic (7) and genetic (3) screens as sequences polarized towards dif. Thus, KOPS orient FtsK towards XerCD-dif.
FtsK50C, a biochemically active form of FtsKC, has been used for most biochemical and single-molecule studies; however, the protein is subject to aggregation and large particles are visible under the light microscope (57). Initial reports suggested that FtsK50C reverses direction when KOPS are encountered in “nonpermissive” orientation during translocation (5, 7, 9). Conversely, structural and biochemical studies showed that KOPS are loading sites that establish oriented translocation (2, 8). Reversal of translocation at nonpermissive KOPS has been explained by the translocating unit being a double-hexamer, with a switch in the activity between motors (7, 10), or by assuming that a translocating hexamer encounters a second hexamer bound to, or translocating from, a nonpermissive KOPS (8). Biochemical experiments have shown that one hexamer is sufficient to activate XerCD, providing support for the latter hypothesis (8). Despite the above, it remains important to determine unambiguously if KOPS only acts as a loading site for FtsK, and that during translocation FtsK is blind to KOPS in either orientation.
Different interpretations of how ring-shaped motors move on DNA have also emerged. Three models for mechanochemical coupling have been proposed, based on the order of ATP hydrolysis of their subunits and their contacts to nucleic acid (11, 12): sequential, concerted, and stochastic. A second question therefore is: how does FtsK translocate on dsDNA and what is the role of ATP?
To address both questions, we used a millisecond time-resolution ensemble assay and kinetic modeling to extract the dsDNA translocation and ATPase activities of FtsK. Our dsDNA translocation parameters agree well with those measured using single-molecule assays. We show that FtsKC loads specifically at KOPS using FtsKγ, and find no evidence for the reading of KOPS by FtsK during translocation. Order of addition assays provide evidence for the assembly pathway of a hexameric ring of FtsK on DNA, which can be reconciled with the structure of trimeric FtsKγ domains bound to KOPS. By extracting translocation and ATPase rates, we show sigmoidal dependences on ATP concentration that are consistent with cooperative cycles of ATP hydrolysis coupled to DNA translocation. By comparing the maximum rates, we estimate that the ATP coupling efficiency of FtsK is 1.63–2.11 bp/ATP. Our results demonstrate the sensitivity of ensemble methods for measuring a fast motor and provide a basis for understanding mechanistically how FtsK and related hexameric motors pump dsDNA.


KOPS-Dependent and -Independent Initiation of Translocation by FtsK.

To test whether KOPS act as assembly sites for FtsK, we used a fluorescent triplex displacement assay to report translocation. A TAMRA (tetramethyl-6-carboxyrhodamine)-labeled triplex forming oligonucleotide (TFO) was hybridized in acidic buffer conditions to its cognate triplex binding site (TBS), which was cloned into the DNA of interest. The resulting triplex does not dissociate when diluted into the reaction conditions. Collision of FtsK with the TFO results in displacement, measured by an increase in fluorescence. Rebinding of the TFO does not occur (13). If FtsK initiates from a specific location relative to the TBS, a characteristic lag in the displacement kinetics can be used to compute the translocation velocity (14). We performed a series of experiments on linear DNAs containing a TBS at fixed distances from three overlapping KOPS (“triple KOPS,” or tKOPS, GGGCAGGGCAGGGCAGGG). The tKOPS were oriented either facing towards (permissive) or away from (nonpermissive) the TBS. We analyzed residues 247–811 of FtsK from Pseudomonas aeruginosa, a monodisperse protein which was used to determine the crystal structure of the translocase domain and the FtsKγ-KOPS cocrystal structure (10).
Reaction profiles were followed on a millisecond time scale using stopped-flow fluorimetry. FtsK and DNA were preincubated for > 5 min in one syringe and rapidly mixed with ATP from the other in a 1∶1 ratio. Final DNA and ATP concentrations were 1 nM and 2 mM, respectively. Profiles obtained from a DNA with 1,008 bp between the TBS and permissive KOPS are shown in Fig. 1A, where FtsK concentration was varied from 0–800 nM monomer. On the basis of structural data (8), we expected the active motor complex to be a hexamer assembled on DNA. In the absence of FtsK (“0”), a slow background level of triplex displacement was observed (∼0.02 min-1). As the FtsK concentration was increased, we observed a distinct lag (tlag) followed by an exponential burst phase of triplex displacement. At ≤ 50 nM FtsK monomer, triplex displacement was incomplete: a fast exponential phase preceded a > 100-fold slower linear phase. These profiles are characteristic of stepwise translocation initiating from a specific point relative to the triplex (1315). At 100 nM FtsK the amplitude of the first phase increased moderately but the second phase increased in both rate and amplitude to give > 80% displacement after ∼3 min. A small additional phase (“end”) was also seen within the slow phase, which can be reconciled with initiation at DNA ends (SI Text and Fig. S1A) and was not observed on circular DNA (Fig. 1B, “circ”). A further doubling of FtsK to 200 nM caused a marked change in the displacement profile: an apparently shorter lag was followed by multiple exponential phases. Further increases in concentration (400 and 800 nM) resulted in very fast apparent lags followed by a series of complex displacement phases over at least three time decades. Thus, at low concentrations of FtsK only, the data are consistent with initiation of dsDNA translocation from a specific site. A schematic interpretation is shown in Fig. S1A.
Fig. 1.
KOPS-dependent and -independent translocation by FtsK measured by triplex displacement. DNA preincubated with FtsK at the indicated monomeric concentrations was rapidly mixed with ATP (final concentration 2 mM). Triplex displacement was monitored by fluorescence and normalized to conditions that yield 100% dissociation. Cartoons show the DNA substrate, with tKOPS as three overlaid blue triangles, TFO as a red line, and FtsK (as side-on hexamer) in gray. Black arrow (A only) shows predicted direction of translocation. (A, B, and C) Concentration-dependent translocation by FtsK on (A) linear tKOPS DNA; (B) linear and negatively supercoiled circular (“circ”) nonspecific DNA (2,732 bp); and (C) linear DNA with nonpermissive tKOPS. (D) Distance dependence of triplex displacement profiles. Profiles are shown for five linear DNAs with different spacings (dT) between tKOPS and TFO. Colored lines show fits to the model (cartoon), assuming one bp represents one . Kinetic constants are average values (± SD) from fits at each dT. DNA is shown end-on in yellow and ATP binding sites (empty or occupied) in blue. Apparent signal changes before 20 ms resulted from mixing artifacts and were discarded from fitted data.
We were concerned that at higher FtsK concentrations the profiles did not conform to models for a linear translocase initiating from a single site (14). Although none of our DNA substrates contained additional “bona fide” KOPS, we reasoned that the affinity for nonspecific DNA—perhaps sequences that resemble KOPS (SI Text)—is similar: high FtsK concentration would lead to assembly of hexamers elsewhere, resulting in distributive displacement kinetics. We tested this hypothesis using a DNA lacking KOPS (Fig. 1B). Very little displacement was observed below 50 nM FtsK. At 100 nM FtsK, a longer lag phase preceded an exponential phase, but with a markedly lower amplitude (see below). At ≥200 nM FtsK, distributive profiles were obtained equivalent to those in Fig. 1A. Similar nonspecific profiles were observed using FtsKΔγ, which lacks the domain that binds KOPS, and with a DNA bearing a single KOPS (Fig. S1). Thus, at high protein concentrations, FtsK assembles at random nonspecific sites and at DNA ends (see SI Text for further discussion). For these reasons, all subsequent assays were performed at 50 nM FtsK only.
If the lag-exponential kinetics observed in Fig. 1A were KOPS-dependent, at least for low FtsK concentrations, we predicted that the lag duration would depend on the orientation of the KOPS and the KOPS-TBS distance. We tested this prediction using a DNA with nonpermissive KOPS positioned 1,008 from the TBS (Fig. 1C). At 50 nM FtsK, triplex displacement occurred at the background rate. At higher concentration (400 nM), a profile characteristic of nonspecific initiation was observed. A similar profile was obtained when the nonpermissive KOPS was 14 bp from the triplex, where we would expect minimal nonspecific loading between KOPS and triplex (Fig. S2A). These observations suggest that FtsK motors can translocate past nonpermissive KOPS without reversal (8).
To confirm that FtsK loads specifically at KOPS, we measured triplex displacement at 50 nM FtsK, varying the permissive tKOPS-TBS distance (dT, Fig. 1D). Here, tlag increased with a linear dependence on dT, while the initial exponential phases were similar, consistent with assembly of FtsK at KOPS and translocation with uniform velocity towards the triplex. The amplitudes of the initial exponential phases, which report the extent of displacement, varied nonsystematically (0.37 ± 0.03). Below, we present a model for FtsK assembly at KOPS which can explain why displacement never exceeds 30%–40% in the first phase. To obtain kinetic information on translocation we modeled the reactions, assuming the following: FtsK is fully assembled at KOPS at the start of the reaction (see below), with no second-order binding step during initiation; nucleotide, when added, binds rapidly to FtsK followed by a single-step initiation phase (kini); and FtsK steps along the DNA at uniform translocation velocity (kstep) with one bp moved per step; dissociation was not considered (SI Text). Finally, collision with the triplex will result in displacement in a single rate-limiting step, kTFO. In this scheme values for kini and kTFO are interchangeable; we cannot assign the rate constant to either step with certainty (15). The number of FtsK hexamers that successfully initiate was not fixed in the model. We fitted the data to the model by numerical integration. The computed translocation rate (kstep = 4,790 ± 169 bp s-1 at 20 °C and 2 mM ATP) is consistent with single-molecule measurements (4–7 kb s-1) (27). We confirm the translocation parameters below using a different stopped-flow mixing regime where nonspecific binding does not contribute to displacement.

Steps in the KOPS-Dependent Assembly of a Hexameric FtsK Motor.

In the experiments above, preincubation of FtsK with tKOPS DNA appeared to saturate a fraction of the substrate; < 40% of the triplex was displaced in the first phase of the reaction (Fig. 1). Assuming the DNA is homogeneous, an explanation might be that the enzyme has low specific activity. If the inactive fraction were unable to bind DNA, we would expect an increase in FtsK concentration eventually to saturate KOPS and lead to 100% displacement. However, doubling the FtsK concentration from 50 to 100 nM produced only a small change in the amplitude of the first phase (Fig. 1A), which excludes a binding-deficient subpopulation. Alternatively, each DNA may be saturated at 50 nM FtsK but a proportion of the enzyme might not initiate immediately. A slow dissociation or rearrangement of this population would account for the slow second phase. A model to illustrate this hypothesis is shown in Fig. 2A.
Fig. 2.
A six-step model for the assembly of FtsK at KOPS. (A) Two-fork model showing assembly in the absence of ATP, when FtsK is preincubated with DNA and then mixed with ATP (upper branch), and in the presence of ATP, when FtsK is preincubated with ATP and then mixed with DNA and ATP (lower branch). Cartoon as per Fig. 1D, with each gray circle representing one FtsK monomer. Kinetic values for assembly are only considered for the lower fork. (B) Effect of preincubating FtsK with DNA in the absence of ATP. Simulations by numerical integration with 1,008 translocation steps, assuming that all DNA molecules are saturated with a hexamer at the start (Left), are compared to the data from Fig. 1A Right. (C) Effect of mixing FtsK and DNA in the presence of saturating ATP. Simulations with 1,008 translocation steps (Left) are compared to equivalent experimental data (Right). FtsK was present at the concentrations indicated and was rapidly mixed with DNA; ATP was present in both syringes at 2 mM. Constants used for both simulations: kbind1 = kbind2 = kbind3 = 1.2 × 109 M-1 s-1; kbind4 = kbind5 = kbind6 = 7.2 × 107 M-1 s-1; koff1 = koff2 = koff3 = 249 s-1; koff4 = koff5 = koff6 = 46.1 s-1; kini = 28 s-1; kstep = 4,500 s-1; koff = 0; and kTFO = 46 s-1; kbind,ATP → ∞ at 2 mM ATP.
In the model, we suggest that FtsK hexamers assemble onto KOPS by a six-step process, where each step is the association of an FtsK monomer. Models with smaller numbers of steps are discussed and can be rejected (SI Text and Figs. S3 and S4). Where FtsK is preincubated with DNA in the absence of ATP, as above (Fig. 1), the assembled hexamer undergoes a slow conformational change (ktrans) that hyper-stabilizes the ring. Given sufficient time, the DNA becomes saturated with FtsK hexamers. Upon mixing with ATP, the complex rapidly binds nucleotide and then either initiates translocation (kini), or one protomer dissociates (koff6). The relative values of kini and koff6 establish the amplitude of the initial phase. For the failed species, reestablishing the hexamer requires the rebinding of an FtsK monomer, a second-order process whose rate is dependent upon FtsK concentration. A simulation of this model is compared to empirical data (Fig. 2B). In this model the first three binding steps are fast, followed by three slower steps. Reasons for favoring this model are described (below and SI Text). Here, an increase in FtsK concentration alters only the rate of the second phase, not the amplitude of the initial phase. The complete simulation with an increasing concentration of FtsK shows a characteristic change in the displacement profile. However, because of nonspecific initiation, we could not employ concentrations > 50 nM FtsK. Nevertheless, the model shows a close correspondence with the data (also see SI Text).
If preincubation of FtsK with DNA produces a state that must undergo rearrangement, we predicted that the presence of ATP during DNA binding will assemble hexamers that can progress directly to initiation, bypassing ktrans1 [Pathway (ii); Fig. 2A]. We also predicted that any nonspecific or end-binding initiation states would assemble more slowly. Thus, triplex displacement should show a characteristic pattern as a function of the ordered binding steps only at KOPS. Simulation of pathway (ii) shows a strong dependence on FtsK concentration, due to the six second-order binding steps during assembly (Fig. 2C, Left and SI Text). At 50 nM FtsK, virtually no triplex displacement occurs, while complete displacement requires > 800 nM FtsK.
To test this alternative pathway using the stopped-flow assay, we preincubated FtsK and ATP in one syringe and rapidly mixed with DNA and ATP from the other. ATP turnover by FtsK in the absence of DNA is negligible, and DNA-independent oligomeric species do not form at the concentrations used (10). A set of profiles for the DNA with a tKOPS-TFO spacing of 1,008 bp are shown (Fig. 2C, Right). As above, model and data correspond closely. At high FtsK concentrations we note a small reduction in final amplitude, which we suggest results from FtsK binding to the free TFO. Importantly, the kstep, kini, and kTFO values used in the simulations are identical to the values determined in Fig. 1D. The kinetic profiles in Fig. 2C were also inconsistent with any significant amount of dissociation during translocation, as suggested (SI Text). Thus, our kinetic model describes assembly pathways for FtsK at KOPS in the presence and absence of ATP.

ATP-Dependence of FtsK Translocation Suggests Cooperative Cycles of ATP Hydrolysis.

If the lag-exponential kinetics were dependent on translocation speed, as well as the assembly rate and position, we predicted that the lag is also dependent on ATP concentration. Thus, we investigated the effect of changes in ATP concentration where tKOPS DNA was preincubated with 50 nM FtsK (equivalent to experiments in Fig. 1 and Fig. 3A). Indeed, as the ATP concentration was reduced from 2 mM, we observed an increase in lag-time resulting from a reduction in translocation speed. We also noted other effects (SI Text), but these did not complicate our determination of kstep because the fluorescence decay was slow relative to triplex displacement, and the lag-time is independent of second-phase effects (14, 15).
Fig. 3.
ATP coupling in dsDNA translocation by FtsK. (A) Triplex displacement as a function of ATP concentration. A linear DNA with 814 bp spacing between tKOPS and the TFO was preincubated with 50 nM FtsK monomer and rapidly mixed with ATP at the final concentrations shown. (B) Distance-dependence of the triplex displacement profiles at 600 μM ATP. Profiles are shown for four linear DNAs with different spacings (dT) between the tKOPS and TFO. Colored lines show fits by numerical integration as in Fig. 1D. (C) Relationship between kstep and ATP concentration. kstep values are averages (± SD) derived from multiple DNAs as per (B). Black line is a least squares fit to where n is the Hill coefficient. Red, cyan, and blue lines are simulations for the n values shown. Residual plots in Fig. S5. (D) (Upper) Relationship between displacement amplitude and ATP concentration. No systematic relationship with dT was observed. (Lower) Relationship between kini or kTFO and ATP concentration. Points represent values from individual fits for each DNA with no bias to the assignment of kini or kTFO (23). (E) Kinetics of phosphate release. DNA was preincubated with FtsK monomer at 50 nM and rapidly mixed with ATP at the concentrations shown. Pi release was measured from the fluorescence increase upon binding fluorescently-labeled PBP normalized to a titration of phosphate standard. Dotted lines are linear fits to the first 1 s of the traces to give kATP. (F) Relationship between kATP and ATP concentration. Black line is a least squares fit to Hill equation (C) with red, cyan, and blue lines being fits for the n values shown. Residual plots in Fig. S5.
To determine values for each translocation parameter, we repeated the ATP titrations for three other KOPS-TBS spacings (dT). Initial phases of the profiles were fitted independently using the model from Fig. 1D, returning mean values for kstep, kTFO, and kini at each ATP concentration for each spacing. Example fits are shown for 600 μM ATP (Fig. 3B), and the dependence of each of these parameters and the amplitude on the ATP concentration are shown (Fig. 3 C and D and SI Text).
The lag-times varied linearly with dT for each ATP concentration, as expected of stepwise translocation. The dependence of kstep on ATP showed a marked sigmoidal nature, giving a best fit to a Hill coefficient of 1.78 ± 0.24 (Fig. 3C, Fig. S5, and Table 1). Thus, at least two ATP are required to act cooperatively in the ATPase cycle before a step on DNA can occur.
Table 1.
Sigmoidal dependence of kstep and kATP on ATP concentration
A. Translocation (kstep)
nVmax (bp s-1)K1/2 (μM)χ2
16,470 ± 211847 ± 74.513.1
1.78 ± 0.244,790 ± 169411 ± 37.91.51
34,480 ± 55.5330 ± 7.5717.1
64,390 ± 53.5284 ± 3.3461.9
B. ATPase (kATP)
nVmax (μM s-1)K1/2 (μM)χ2
11.11 ± 0.056322 ± 43.30.0108
1.45 ± 0.140.953 ± 0.035237 ± 17.00.00325
30.816 ± 0.044190 ± 17.30.0376
60.762 ± 0.67183 ± 19.20.113

Bold values are parameters from least-square Hill fits of (A) translocation and (B) ATPase data (Fig. 3 C and F) with n = 1, 3, or 6. χ2 values (nonreduced) report relative goodness of fit.

To examine further how ATP is coupled to motion on the DNA, we measured ATPase rates using a fluorescent phosphate sensor under similar experimental conditions. This assay couples the ATPase activity of FtsK to binding of the released phosphate using a fluorescently labeled phosphate binding protein (PBP) (16). Under our conditions, the maximal steady-state ATPase rate that can be measured is > 8,000 s-1. Example phosphate release profiles are shown in Fig. 3E. Each profile was biphasic: an initial linear phase was followed by a nonlinear second phase. The second phase likely reflects the approach to a complex steady-state with initiation, translocation, dissociation, and reassembly (SI Text). However, the initial linear burst phase is on the same time scale (< 1 s) as the triplex displacement profiles. Thus, we fitted the initial linear phases to obtain the ATPase rate. The data was not corrected for ATPase activity by FtsK at nonspecific sites, which was negligible (< 1%) under these conditions (Fig. S6).
The dependence of the initial ATPase rate on ATP concentration is shown (Fig. 3F). Similarly to translocation, the data showed a sigmoidal character with a Hill coefficient of 1.45 ± 0.14 (Table 1). Although the dependence is weaker than that observed for translocation, we suggest that both values are consistent with the action of more than one ATP to drive a successful ATPase cycle (Fig. S5).
Using the fitted Vmax values (Table 1), conditions where both assays return values of similar accuracy and which also reflect the ATP concentration in vivo of ∼1–3 mM (18), we calculated a ratiometric ATP coupling value, the number of ATP consumed per base pair translocated. This ratio is independent of the oligomeric state of the motor (SI Text). To compare the two rates, we calculated the concentration of active motor complexes that contributed to the ATPase turnover number using amplitudes from the triplex assays. We assumed that a fully saturated signal in the triplex displacement assay corresponds to the activity of 1 nM motor, given 1 nM DNA concentration. Thus, at saturating ATP, the observed amplitude (0.37 ± 0.03; Fig. 1D) corresponds to 0.37 ± 0.03 nM translocating motors. The validity of this approach is discussed (SI Text). This calculation yielded an ATPase turnover number of ∼2,600 ± 250 ATP s-1 per motor assembly. Considering the experimental error (95% confidence interval), we calculate a range for the coupling ratio of 1.63 to 2.11 bp translocated per ATP consumed (ref. 15 and SI Text). This value suggests that the translocation step size is ∼2 bp/ATP, consistent with a sequential rotary mechanism. Further data consistent with this result is provided in the SI Text. We note a systematic variation that results from the apparent K1/2 for ATP hydrolysis being lower than that for translocation. This variation has been observed for other helicases (1719) and may reflect uncoupled translocation at low ATP concentrations or different sensitivities of the assays (Fig. S8D).


FtsK Assembles Specifically at KOPS and Is Blind to KOPS During Translocation.

FtsK is an extremely rapid dsDNA translocase that has an essential role in the accuracy and timing of chromosome segregation. In vivo, FtsK is recruited to the closing septum late in the cell cycle and is tethered via an integral membrane domain. Here, in addition to these constraints, we have shown in vitro that the assembly of active FtsK hexamers on DNA occurs specifically at its recognition sequence, KOPS, via the FtsKγ sequence-recognition domain. Following initiation, FtsK translocation is blind to KOPS. The data extend previous findings by determining precise kinetic parameters for assembly and translocation, which allowed us to develop a full kinetic model and estimate the energetic coupling ratio for translocation. This data also has wider implications for other ring motors where different translocation schemes have been advanced (11).
We suggest that reversal events at nonpermissive KOPS, seen previously in single-molecule tweezers assays, are an accident of experimental design. These studies measured translocation rates via changes in the position of a bead attached to the DNA. Detection of translocation requires loop extrusion, which in turn selects for artifactual events involving multiple motors. Furthermore, we observed largely nonspecific initiation at concentrations > 50 nM (see SI Text for discussion of nonspecific assembly). In previous studies, high FtsK concentrations (75–250 nM monomer; refs. 2, 4, 5, 8, 9) could have elicited nonspecific loading and complicated the analysis. Reversals without dissociation are structurally unlikely because helicases/translocases are “hard-wired” to move directionally (20, 21). A model requiring recognition of nonpermissive KOPS, motor disassembly and reassembly in the opposite orientation would also, according to the model (Fig. 2A), be kinetically limited, as well as sterically difficult. For these reasons, FtsK reversal events are more readily explained by the antagonistic action of multiple colocalized motors.
Random or nonspecific assembly of FtsK would be problematic in vivo, as the potent translocation activity (here; refs. 27) and nucleoprotein remodeling activity (22) of FtsK would interfere with essential activities such as replication or transcription. How could nonspecific loading events be averted? The local concentration of FtsK at the septum would be high despite its low copy number (∼30 per cell). However, FtsK is spatially constrained by its membrane linkage, which might disfavor nonspecific loading. Nonspecific loading would also be inhibited by the higher concentration of counterions in vivo (Fig. S2B). We suggest that nonspecific loading is an artifact of high protein and low salt concentrations in vitro. As seen here, FtsK loading in vivo may occur preferentially at contiguous or clustered KOPS sites (3).

Unique Insights into the Pathway of Hexamer Assembly: a Full Kinetic Model.

We derived an assembly model in which FtsK hexamers form at KOPS via ordered stepwise addition of monomers (Fig. 2). This model could describe data whether FtsK was preincubated with or without DNA. Models in which hexamers form from dimers, trimers, or by direct binding, are inconsistent with the data (SI Text). The six-step scheme is only consistent with the data if the binding rates are equal or if early binding steps are faster than later steps (SI Text). Although we cannot explicitly distinguish between these alternatives, we propose that three FtsK monomers assemble quickly and reversibly into a trimer, followed by slower reversible addition of three further monomers (Fig. 2 B and C). Our preference for this model is based primarily on structural evidence that FtsKγ-dependent trimers can form at KOPS (8). The efficiency of tKOPS for loading may be because this sequence offers a larger target for the initial binding step. Our analysis of single KOPS was limited by nonspecific binding at the higher FtsK concentrations required to observe triplex displacement (Fig. S1C). Once the DNA-trimer is formed, subsequent recruitment of FtsK monomers is likely to be independent of KOPS, leaving three FtsKγ domains unbound, and thus slower than the previous steps. FtsK and FtsKΔγ can also assemble and translocate on nonspecific DNA, but only at higher FtsK concentrations indicating a slower assembly pathway (SI Text). We also observed a minor assembly pathway at DNA ends. We speculate that this additional phase results from reduced steric interference for KOPS-independent ring formation or altered electrostatic characteristics at the DNA ends, but not from threading, as we can find no evidence for higher order oligomers in solution (10). In vivo, proteins such as RecBCD are more likely to bind dsDNA ends. Once ATP binds and the motor initiates, the γ domains disengage or are removed by the motor. We cannot exclude more complex assembly pathways (SI Text). However, the model describes the data well.
Where FtsK was preincubated with the DNA in the absence of ATP, a fit of model to data was only satisfied when an additional rearrangement step, ktrans, was included, which invokes a stable hexameric intermediate that cannot undergo initiation until ATP is added. A rationalization of ktrans is that, when preincubated with DNA, trans-acting amino acid residues in the nucleotide binding site, such as the arginine finger essential for hydrolysis, are not engaged (11, 12). In the hexameric crystal structure of FtsK, the putative arginine finger (R620) is positioned ∼10  from nucleotide, which may reflect an inactive or a posthydrolytic state (10). Supporting this hypothesis, ktrans was not necessary where FtsK and ATP were premixed.

ATP Concentration Dependence of Translocation and ATP Hydrolysis Rates.

Translocation and ATP hydrolysis were best described by a sigmoidal dependence on ATP, with Hill coefficients for kstep and kATP of 1.78 ± 0.24 and 1.45 ± 0.14, and Vmax values of 4,790 ± 169 bp s-1 and 2,600 ± 250 ATP s-1 per hexamer. This relationship suggests an allosteric use of ATP to drive translocation, with more than one ATP bound at any time during translocation. We cannot deconvolve the arrangement of ATP binding; however, we suggest that binding to adjacent subunits is most likely, as proposed for other hexameric motors (11), consistent with a sequential rotary mechanism. Our results using FtsK hexamers with random or targeted nucleotide binding and hydrolysis mutations imply that catalytic mutations in adjacent subunits have a severe effect on translocation velocity and force generation (22). The data is also consistent with uncoupling of translocation and ATP hydrolysis as the ATP concentration is reduced, conditions where adjacent monomers may not be saturated (Fig. S8D).
By comparing the maximal translocation and ATPase rates, we calculated a coupling ratio of ∼1.63 to 2.11 bp translocated per ATP hydrolysed, an error range confirmed by alternative experimental approaches (SI Text). A priori, 2 bp/ATP would be a plausible step size for a hexameric dsDNA translocase, equating to a 69° displacement of B-form DNA contacts for a 60° hand-off between monomers upon ATP hydrolysis at one active site. Although we have not yet determined a high-resolution structure of FtsK in complex with dsDNA, it is tempting to speculate that the FtsK employs a “coordinated escort” translocation mechanism (20, 23), in which binding loops within the hexameric pore form a “staircase” with nucleic acid, with the loop position in the staircase dependent on the hydrolytic state of the monomer. In the equivalent mechanisms proposed for Rho and E1 (20, 23), hydrolysis is coupled to a single-stranded polynucleotide and each ATP hydrolysis results in steps of one nucleotide. However, the principle is the same; ATP hydrolysis occurs cyclically around the ring, coupled to movement along the track. Our data are therefore consistent with a common mechanism for translocation of single-stranded and double-stranded nucleic acid by ring motors. Moreover, the analysis has revealed a theoretical model for how a hexameric enzyme is topologically linked around its substrate at a specific location, providing a framework which may be applicable to a wider range of ring motors.

Materials and Methods

FtsK Purification and DNA Preparation.

C-terminally His-tagged Pseudomonas aeruginosa FtsK (residues 247–811) was purified from E. coli B834(DE3) xerD- as described (10). Linear dsDNA was derived from plasmids that carry two TBSs separated by a PmeI restriction site, with KOPS in various configurations with respect to the TBSs. Plasmids were purified using a Qiagen maxiprep kit, linearized using PmeI and the DNA purified by standard phenol-chloroform extraction and isopropanol precipitation. TAMRA-labeled TFOs were synthesized and HPLC purified by Sigma-Genosys. Triplexes were formed as previously described (14). Displacement of the TFO required a 4 nt 3′ flap (9).

Stopped-Flow Triplex Displacement Assays.

Fluorescence intensity measurements were performed using an SF61-DX2 stopped-flow fluorimeter (TgK Scientific) at 20 ± 0.1 °C in single-mix mode as described (14, 15). Reactions were performed in 25 mM Tris-OAc (pH 7.5), 2 mM Mg(OAc)2 and 1 mM DTT. Where FtsK was preincubated with DNA, syringe C contained FtsK·DNA and syringe D contained ATP. Where FtsK was rapidly mixed with DNA, syringe C contained DNA and syringe D contained FtsK; ATP was present at the same concentration in both syringes. Final DNA concentration was 1 nM; FtsK and ATP concentrations were varied. Data were acquired over a logarithmic time-base and three traces averaged. Data were fitted by numerical integration using Berkeley Madonna (24). Attempts to preincubate FtsK in the absence of ATP and DNA were prevented by loss of protein activity under these conditions.

Stopped-Flow ATP Hydrolysis Assays.

PBP labeled with 7-diethylamino-3-[(((2-maleimidyl)ethyl)amino)carbonyl)coumarin] (MDCC) was prepared as described (16). Fluorescence intensity measurements were performed at 20 ± 0.1 °C using the stopped-flow apparatus outlined above, with λex = 435 nm (2 nm bandwidth) and 455 nm band-pass filter. Pre-equilibrated samples contained 25 mM Tris-Cl (pH 7.5), 2 mM Mg(OAc)2, 1 mM DTT, 1 nM PmeI linearized duplex, and 50 nM FtsK monomer in syringe C and 25 mM Tris-Cl (pH 7.5), 2 mM Mg(OAc)2, and 1 mM DTT and ATP (2× desired cell concentration) in syringe D. 12 μM MDCC-PBP, 0.05 U/mL purine nucleoside phosphorylase and 0.2 mM 7-methyl-guanosine were present in both syringes. Data were acquired over 10 s using a linear time-base. The photomultiplier tube (PMT) response was linear with respect to Pi concentration over the ranges detected. Data were normalized to fluorescence intensity at zero time and linear fits made using Igor Pro (Wavemetrics, Inc.).


We thank Martin Webb for help with the PBP assay and Jan Löwe for the P. aeruginosa FtsKC expression plasmid. This work was supported by a Wellcome Trust PhD Studentship (077470 to J.E.G. and D.J.S.) and Wellcome Trust Programme Grants (083469 to D.J.S. and 084086 to M.D.S.).

Supporting Information

Supporting Information (PDF)
Supporting Information


S Bigot, et al., FtsK, a literate chromosome segregation machine. Mol Microbiol 64, 1434–1441 (2007).
S Bigot, et al., Oriented loading of FtsK on KOPS. Nat Struct Mol Biol 13, 1026–1028 (2006).
S Bigot, et al., KOPS: DNA motifs that control E. coli chromosome segregation by orienting the FtsK translocase. EMBO J 24, 3770–3780 (2005).
OA Saleh, C Pérals, F-X Barre, J-F Allemand, Fast, DNA-sequence independent translocation by FtsK in a single-molecule experiment. EMBO J 23, 2430–2439 (2004).
JL Ptacin, M Nöllmann, C Bustamante, NR Cozzarelli, Identification of the FtsK sequence-recognition domain. Nat Struct Mol Biol 13, 1023–1025 (2006).
PJ Pease, et al., Sequence-directed DNA translocation by purified FtsK. Science 307, 586–590 (2005).
O Levy, et al., Identification of oligonucleotide sequences that direct the movement of the Escherichia coli FtsK translocase. Proc Natl Acad Sci USA 102, 17618–17623 (2005).
J Löwe, et al., Molecular mechanism of sequence-directed DNA loading and translocation by FtsK. Mol Cell 31, 498–509 (2008).
V Sivanathan, et al., The FtsK gamma domain directs oriented DNA translocation by interacting with KOPS. Nat Struct Mol Biol 13, 965–972 (2006).
TH Massey, et al., Double-stranded DNA translocation: structure and mechanism of hexameric FtsK. Mol Cell 23, 457–469 (2006).
MR Singleton, MS Dillingham, DB Wigley, Structure and mechanism of helicases and nucleic acid translocases. Annu Rev Biochem 76, 23–50 (2007).
E Enemark, L Joshua-Tor, On helicases and other motor proteins. Curr Opin Struct Biol 18, 243–257 (2008).
K Firman, MD Szczelkun, Measuring motion on DNA by the type I restriction endonuclease EcoR124I using triplex displacement. EMBO J 19, 2094–2102 (2000).
SE McClelland, DTF Dryden, MD Szczelkun, Continuous assays for DNA translocation using fluorescent triplex dissociation: application to type I restriction endonucleases. J Mol Biol 348, 895–915 (2005).
R Seidel, J Bloom, C Dekker, M Szczelkun, Motor step size and ATP coupling efficiency of the dsDNA translocase EcoR124I. EMBO J 27, 1388–1398 (2008).
MR Webb, Development of fluorescent biosensors for probing the function of motor proteins. Mol Biosyst 3, 249–256 (2007).
RM Smith, J Josephsen, MD Szczelkun, The single polypeptide restriction-modification enzyme LlaGI is a self-contained molecular motor that translocates DNA loops. Nucleic Acids Res 37, 7219–7230 (2009).
J Neuhard, P Nygaard Escherichia coli and Salmonella typhimurium: Cellular and Molecular Biology, eds F Neidhardt, et al. (ASM Press, Washington, DC, 1987).
LJ Roman, SC Kowalczykowski, Characterization of the adenosinetriphosphatase activity of the Escherichia coli RecBCD enzyme: relationship of ATP hydrolysis to the unwinding of duplex DNA. Biochemistry 28, 2873–2881 (1989).
ND Thomsen, JM Berger, Running in reverse: the structural basis for translocation polarity in hexameric helicases. Cell 139, 523–534 (2009).
K Saikrishnan, et al., Mechanistic basis of 5′-3′ translocation in SF1B helicases. Cell 137, 849–859 (2009).
E Crozat, et al., Separating speed and ability to displace roadblocks during DNA translocation by FtsK. EMBO J 29, 1423–1433 (2010).
EJ Enemark, L Joshua-Tor, Mechanism of DNA translocation in a replicative hexameric helicase. Nature 442, 270–275 (2006).
R Seidel, et al., Dynamics of initiation, termination, and reinitiation of DNA translocation by the motor protein EcoR124I. EMBO J 24, 4188–4197 (2005).

Information & Authors


Published in

Go to Proceedings of the National Academy of Sciences
Go to Proceedings of the National Academy of Sciences
Proceedings of the National Academy of Sciences
Vol. 107 | No. 47
November 23, 2010
PubMed: 21048089


Submission history

Published online: November 3, 2010
Published in issue: November 23, 2010


  1. ASCE ATPase
  2. DNA translocation
  3. FtsK translocase
  4. stopped-flow
  5. triplex displacement


We thank Martin Webb for help with the PBP assay and Jan Löwe for the P. aeruginosa FtsKC expression plasmid. This work was supported by a Wellcome Trust PhD Studentship (077470 to J.E.G. and D.J.S.) and Wellcome Trust Programme Grants (083469 to D.J.S. and 084086 to M.D.S.).


*This Direct Submission article had a prearranged editor.



James E. Graham1 [email protected]
Department of Biochemistry, University of Oxford, South Parks Road, Oxford, OX1 3QU, United Kingdom; and
Present address: Department of Microbiology, University of California, One Shields Avenue, Davis, CA 95616-8665.
David J. Sherratt
Department of Biochemistry, University of Oxford, South Parks Road, Oxford, OX1 3QU, United Kingdom; and
Mark D. Szczelkun1 [email protected]
DNA-Protein Interactions Unit, School of Biochemistry, Medical Sciences Building, University of Bristol, Bristol, BS8 1TD, United Kingdom


To whom correspondence may be addressed. E-mail: [email protected] or [email protected].
Author contributions: J.E.G., D.J.S., and M.D.S. designed research; J.E.G. and M.D.S. performed research; J.E.G. and M.D.S. analyzed data; and J.E.G., D.J.S., and M.D.S. wrote the paper.

Competing Interests

Conflict of interest statement: J.E.G. declares that the preappointed editor, S.C.K., is his present employer, but that there is no scientific or financial conflict of interest.

Metrics & Citations


Note: The article usage is presented with a three- to four-day delay and will update daily once available. Due to ths delay, usage data will not appear immediately following publication. Citation information is sourced from Crossref Cited-by service.

Citation statements



If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.

Cited by


    View Options

    View options

    PDF format

    Download this article as a PDF file


    Get Access

    Login options

    Check if you have access through your login credentials or your institution to get full access on this article.

    Personal login Institutional Login

    Recommend to a librarian

    Recommend PNAS to a Librarian

    Purchase options

    Purchase this article to get full access to it.

    Single Article Purchase

    Sequence-specific assembly of FtsK hexamers establishes directional translocation on DNA
    Proceedings of the National Academy of Sciences
    • Vol. 107
    • No. 47
    • pp. 20145-20592







    Share article link

    Share on social media