3.2.1 Defining active and passive modes of unwinding

The mechanistic coupling of ssNA translocation to NA unwinding has been described in terms of two extreme cases – passive and active unwinding (Delagoutte & von Hippel, Reference Delagoutte and von Hippel2001; Lohman & Bjornson, Reference Lohman and Bjornson1996; Singleton & Wigley, Reference Singleton and Wigley2002). A purely passive helicase will not enter the duplex region until thermal fluctuations cause the NA at the junction to open. The passive helicase can then move forward through the duplex by binding to the newly available ssNA, preventing it from reannealing. Thus, a purely passive helicase should translocate on ssNA faster than it can unwind the duplex. In contrast, a purely ‘active’ helicase is able to directly destabilize duplex NA at the junction by lowering its free energy such that the unwinding rate for a maximally active helicase would be expected to match its translocation rate on ssNA. A mathematical description of helicase activity in terms of active and passive unwinding has been developed (Betterton & Julicher, Reference Betterton and Julicher2003, Reference Betterton and Julicher2005) in which the degree of activeness ideally follows a scale of free energy reductions per bp unwound, ΔG _d (d refers to destabilization), induced by the helicase from 0 k _BT for a purely passive helicase to 3·4 k _BT (i.e. G–C base-pairing energy) for an optimally active helicase.

To date, tweezers-based assays on three different helicases (Cheng et al. Reference Cheng, Dumont, Tinoco and Bustamante2007; Johnson et al. Reference Johnson, Bai, Smith, Patel and Wang2007; Lionnet et al. Reference Lionnet, Spiering, Benkovic, Bensimon and Croquette2007) have yielded information on the range of activeness in helicase-promoted unwinding by analyzing the dependence of unwinding kinetics on NA sequence and external force. As the larger bond energy of G–C versus A–T/U base pairs can influence both thermal fraying of ends as well as duplex destabilization, sequence-dependence of unwinding velocity is expected for both active and passive cases. Increasing force is predicted to increase the unwinding rate to a larger extent for an initially passive helicase. The experimentally measured effects of these variables on unwinding kinetics were interpreted in terms of the aforementioned theoretical model (Betterton & Julicher, Reference Betterton and Julicher2005). Results show that helicase unwinding mechanisms range from the highly active case of hepatitis NS3 helicase (Cheng et al. Reference Cheng, Dumont, Tinoco and Bustamante2007) to the partially active phage T7 gp4 helicase (Johnson et al. Reference Johnson, Bai, Smith, Patel and Wang2007) to the mostly passive phage T4 gp41 helicase (Lionnet et al. Reference Lionnet, Spiering, Benkovic, Bensimon and Croquette2007).

3.2.2 Hepatitis C virus NS3 helicase

Bustamante and co-workers (Cheng et al. Reference Cheng, Dumont, Tinoco and Bustamante2007) used optical tweezers to investigate sequence- and force-dependence of unwinding of an RNA hairpin substrate by NS3 helicase. The hairpin contained a 60 bp duplex region that was either 30 GC followed by 30 AU or vice versa, and was tethered at both ends at constant tension, such that an increase in the length of the DNA/RNA handles was measured upon helicase-promoted unwinding (set-up similar to Fig. 2h). Consistent with their previous study (Dumont et al. Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006), the unwinding velocity was independent of force; however, a strong sequence-dependence of unwinding rate was observed at multiple forces (7 and 17 pN) with pausing upon encountering a sequence barrier (i.e. GC stretch). When the measured NS3 stepping velocity was compared to that expected for a passive helicase, it was 14 000-fold faster for unwinding on a GC template (assuming 4 bp unwinding per step (Dumont et al. Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006)), which translates into ΔG _d=2·5 k _BT. These observations indicate that NS3 unwinds actively, and does not wait passively for thermal fraying to occur before stepping. These results are summarized in an active duplex destabilization model that invokes barrier sensing by NS3 followed by strand separation occurring in sub-steps via an inchworm-type mechanism. An smFRET study on NS3 (Myong et al. Reference Myong, Bruno, Pyle and Ha2007) suggested that DNA translocation in single nt steps leads to the accumulation of tension in the protein such that near-simultaneous unwinding of several bp occurs rapidly once enough elastic energy has built up. This study might help explain the observed force-independence of unwinding for NS3 (Dumont et al. Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006).

3.2.3 Phage T7 replicative helicase

Wang and co-workers (Johnson et al. Reference Johnson, Bai, Smith, Patel and Wang2007), reviewed in Ha (Reference Ha2007) used optical trapping to investigate the nature of unwinding by phage T7 replicative hexameric ring-shaped helicase in a set-up similar to Fig. 2h. Discrete measurements of ssDNA translocation as well as duplex DNA unwinding were made under constant tension, with translocation occurring faster than unwinding at low forces. As with the NS3 study (Cheng et al. Reference Cheng, Dumont, Tinoco and Bustamante2007), a periodic sequence dependence of unwinding rate was observed corresponding to the positioning of pseudo-repeat sequences in the template, indicating that high GC content presents a barrier to unwinding. However, in contrast to NS3, the T7 helicase unwinding rate increased 10-fold, approaching the ssDNA translocation rate, as force was increased from 5 to 11 pN. This was paralleled by a corresponding increase in total unwinding distance (i.e. processivity). While these results appear to suggest that T7 helicase is intrinsically passive and external force allows it to become more active, the force dependence of unwinding rate was best fit by a hybrid active–passive model with ΔG _d=1–2 k _BT per bp that assumes a translocation step size of 2–4 nt and takes into account the measured ssDNA translocation rate, ssDNA elasticity and mechanical stability of the dsDNA. The increase in unwinding rate with force parallels previous findings in which the presence of DNA polymerase also speeds up unwinding (Lee & Yang, Reference Lee and Yang2006; Stano et al. Reference Stano, Jeong, Donmez, Tummalapalli, Levin and Patel2005), suggesting that one mechanism by which polymerases improve helicase processivity is by mechanically preventing negative regulation of the helicase by the unwound 3′ strand at the fork (Ha, Reference Ha2007).

3.2.4 Phage T4 gp41 helicase

Another study on a replicative hexameric helicase, gp41 of phage T4 (Lionnet et al. (Reference Lionnet, Spiering, Benkovic, Bensimon and Croquette2007), uses a magnetic tweezers set-up similar to that shown in Fig. 2f to measure unwinding of a DNA hairpin tethered between a glass surface and a magnetic bead. The measured unwinding rate was 10-fold slower than the ssDNA translocation rate (i.e. rezipping rate in this system) at 3 pN, but it increased 10-fold as force was increased to 11 pN. When the measured data were fit to the Betterton and Julicher model (Betterton & Julicher, Reference Betterton and Julicher2005), a ΔG _d of 0·05 k _BT was obtained consistent with passive unwinding. These data support gp41 functioning via the strand exclusion mechanism proposed for ring-shaped hexameric replicative helicases (Donmez & Patel, Reference Donmez and Patel2006) and are also consistent with the low processivity of gp41 without the gp43 polymerase (Dong et al. Reference Dong, Weitzel and von Hippel1996; Schrock & Alberts, Reference Schrock and Alberts1996). The higher ΔG _d observed in the similar T7 system (Johnson et al. Reference Johnson, Bai, Smith, Patel and Wang2007) was postulated to be due to differences in DNA sequence and elasticity between the substrates used in the two studies (Lionnet et al. Reference Lionnet, Spiering, Benkovic, Bensimon and Croquette2007). A direct comparison of the unwinding velocity force dependence for the two phage replicative helicases (Fig. 3) suggests that below 7 pN, the unwinding by T4 helicase is slightly more force dependent (i.e. shows a greater slope in the unwinding rate versus force curve) relative to T7 helicase, whereas, above 7 pN, T7 helicase actually displays greater force dependence (and thus passivity) relative to T4 gp41. One possibility for the opposite interpretation presented by the aforementioned research is that this may be due to differences in the theoretical modeling of the data; thus, careful attention must be paid to this aspect when interpreting such studies.

Fig. 3.

Degree of activeness in helicase activity probed by single-molecule force techniques. Overlay of rescaled data of the ring-shaped helicases of T4 and T7 phages. Unwinding rate versus force for T7 helicase (Johnson et al. Reference Johnson, Bai, Smith, Patel and Wang2007) was rescaled to match the force-free unwinding rate of T4 helicase (28·4 bp/s) (Lionnet et al. Reference Lionnet, Spiering, Benkovic, Bensimon and Croquette2007). The plot indicates that below 7 pN, the unwinding by T4 helicase is slightly more force dependent relative to T7 helicase, whereas, above 7 pN, T7 helicase unwinding is more sensitive to external force. See text for further details.

3.2.5 Implications for helicase mechanisms

A general implication from these studies is that how actively or processively a helicase can unwind is regulated by factors such as the DNA sequence and the presence of accessory proteins like DNA polymerase. These investigations may also provide a mechanistic distinction between structurally different helicase superfamilies. The non-ring-shaped SFI helicases UvrD, Rep and PcrA have been shown to translocate along ssDNA faster than they unwind duplex regions, indicating that the energy of base pairing is an unwinding barrier (Bertram et al. Reference Bertram, Hayes and Soultanas2002; Brendza et al. Reference Brendza, Cheng, Fischer, Chesnik, Niedziela-Majka and Lohman2005; Cheng et al. Reference Cheng, Brendza, Gauss, Korolev, Waksman and Lohman2002; Dillingham et al. Reference Dillingham, Wigley and Webb2000; Fischer et al. Reference Fischer, Maluf and Lohman2004; Maluf et al. Reference Maluf, Fischer and Lohman2003; Myong et al. Reference Myong, Rasnik, Joo, Lohman and Ha2005) similar to the aforementioned studies on phage T7 and T4 SFIV hexameric ring-shaped replicative helicases (Johnson et al. Reference Johnson, Bai, Smith, Patel and Wang2007; Lionnet et al. Reference Lionnet, Spiering, Benkovic, Bensimon and Croquette2007). The force-dependence of the unwinding rate, however, varies between superfamilies, with the T7 and T4 helicases being strongly force-dependent (Johnson et al. Reference Johnson, Bai, Smith, Patel and Wang2007; Lionnet et al. Reference Lionnet, Spiering, Benkovic, Bensimon and Croquette2007), the non-ring-shaped SFII hepatitis NS3 helicase reported as force-independent (Dumont et al. Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006), and UvrD reported both as force-independent (Dessinges et al. Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004) and impeded by force (Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008). The force-independence of unwinding indicates that, in some cases, strand separation alone is not the rate-limiting step. These differences might indicate a mechanistic distinction in how and why these helicases slow down upon encountering the junction. One possibility is that the non-ring-shaped helicases may undergo a rate-limiting dimerization step in order to unwind (Brendza et al. Reference Brendza, Cheng, Fischer, Chesnik, Niedziela-Majka and Lohman2005; Fischer et al. Reference Fischer, Maluf and Lohman2004; Ha et al. Reference Ha, Rasnik, Cheng, Babcock, Gauss, Lohman and Chu2002; Maluf et al. Reference Maluf, Fischer and Lohman2003; Myong et al. Reference Myong, Rasnik, Joo, Lohman and Ha2005).

3.3.1 Step-size definitions and measuring techniques

One of the key mechanistic questions regarding helicases is elucidation of step size or how many bp/nt are unwound/translocated per NTP hydrolyzed, also referred to as chemical step size or NTP coupling ratio. Other definitions of step size include (1) physical step size or average distance moved by the motor each time it moves on the NA lattice and (2) kinetic step size – the average number of bp unwound (or nt translocated) between two successive rate-limiting steps in the unwinding or translocation cycle (assuming it consists of n sequential steps of identical rates). Kinetic step size is not necessarily equivalent to chemical step size or the physical distance translocated. For example, a kinetic step size of 3 bp/nt might reflect a series of fast translocation steps in 1 nt increments followed by a slow rate-limiting step such as a conformational change that occurs every 3 bp. Physical translocation could also occur in larger steps comprised of multiple steps with different step sizes (Fischer & Lohman, Reference Fischer and Lohman2004; Lohman et al. Reference Lohman, Tomko and Wu2008).

Four methods have been used to determine/estimate helicase step size:

1. 1. Crystallography is used primarily to infer NTP coupling ratios by structural comparisons of helicase–DNA co-crystals in the nucleotide-bound form versus apo-form.

2. 2. Ensemble multiple-turnover kinetics can be used to obtain NTP coupling ratios by comparing the steady-state rates of unwinding or translocation to the rates of NTP hydrolysis.

3. 3. Ensemble pre-steady-state single-turnover kinetics of unwinding/translocation and NTP hydrolysis (described below) allow for more accurate determination of kinetic and physical step sizes and NTP coupling ratios under conditions where the helicase binds once and does not rebind.

4. 4. Sm methods allow for the determination of kinetic and physical step sizes.

In ensemble pre-steady-state single-turnover assays for unwinding and translocation (reviewed in Fischer & Lohman, Reference Fischer and Lohman2004; Lucius et al. Reference Lucius, Maluf, Fischer and Lohman2003), the helicase and DNA substrate are allowed to form an initial complex and then ATP is added using a stopped-flow apparatus and translocation/unwinding is measured as a function of time. Single-round kinetics is usually ensured by the addition of a protein trap such as heparin or excess DNA substrate to prevent any dissociated enzyme from rebinding the substrate. Unwinding assays measure the release of radioactively labeled single strands from the duplex or the decrease in FRET between a donor and acceptor fluorophore pair situated in close proximity on opposing strands of duplex substrates of varying lengths (Lucius et al. Reference Lucius, Maluf, Fischer and Lohman2003). Direct translocation assays monitor the arrival of a helicase at the end of a ssDNA by measuring protein-induced changes in the intensity of a fluorophore positioned at the end of ssDNAs of varying lengths (Dillingham et al. Reference Dillingham, Wigley and Webb2002). In addition, a single-turnover fluorescent-based ATPase assay performed as a function of DNA cofactor length (Dillingham et al. Reference Dillingham, Wigley and Webb2000; Tomko et al. Reference Tomko, Fischer, Niedziela-Majka and Lohman2007) allows for calculation of the ATP coupling ratio.

Bulk kinetic determination of helicase step size can vary due to differences, for instance, in helicase oligomerization state, usage efficiency of NTP hydrolyzed, enzyme pausing and solution conditions. In some cases, translocation step size is not equivalent to unwinding step size, and large steps comprised of smaller sub-steps have been measured for a given enzyme. Sm studies have helped resolve controversy in the literature regarding step sizes predicted by crystallographic and bulk kinetic studies. To explain this, we will focus on two examples: SFIA helicases PcrA/UvrD and SFII replicative RNA helicases such as HCV NS3.

3.3.2 SFI helicases UvrD and PcrA

Crystallographic studies suggest that the binding/hydrolysis of 1 ATP causes inchworm-like translocation along ssDNA in 1 nt increments for Bacillus stearothermophilus PcrA (Velankar et al. Reference Velankar, Soultanas, Dillingham, Subramanya and Wigley1999), and unwinding and translocation in 1 bp/nt increments for Escherichia coli UvrD (Lee & Yang, Reference Lee and Yang2006). In these examples, as crystal structures were obtained with monomeric species, the conclusion is that the small 1 bp stepping is achieved by helicase monomers.

In contrast, ensemble single-turnover kinetic assays performed on these helicases resulted in larger step sizes for unwinding and translocation. Pre-steady-state assays led to the determination of translocation kinetic step sizes of ~4 nt for UvrD (Fischer et al. Reference Fischer, Maluf and Lohman2004; Tomko et al. Reference Tomko, Fischer, Niedziela-Majka and Lohman2007) and PcrA (Niedziela-Majka et al. Reference Niedziela-Majka, Chesnik, Tomko and Lohman2007) and of unwinding kinetic step sizes of 4–6 bp for UvrD (Ali & Lohman, Reference Ali and Lohman1997) and 4 bp for PcrA (Yang et al. Reference Yang, Dou, Ren, Wang, Zhang, Qian, Pan and Xi2008). Although the step size for translocation and unwinding correlates for each helicase, the protein/DNA ratios and length of the ssDNA tail in substrates used in these studies along with numerous bulk kinetic studies suggest that translocation is performed by a monomeric species whereas unwinding requires dimerization (Ali et al. Reference Ali, Maluf and Lohman1999; Fischer et al. Reference Fischer, Maluf and Lohman2004; Maluf et al. Reference Maluf, Fischer and Lohman2003; Yang et al. Reference Yang, Dou, Ren, Wang, Zhang, Qian, Pan and Xi2008).

Further support for a larger kinetic step size was obtained from an sm magnetic tweezers study of UvrD unwinding (Dessinges et al. Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004; set-up similar to Fig. 2e), in which statistical analysis of helicase-induced noise in unwinding burst kinetics yielded a kinetic step size for unwinding of ~6 bp. However, due to the large force (35 pN) applied to the ends of the nicked substrate in this work, it is possible that this measurement actually represented a translocation kinetic step size (Fischer et al. Reference Fischer, Maluf and Lohman2004). This was also suggested in a later magnetic tweezers investigation that monitored UvrD unwinding of a hairpin substrate (Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008; set-up similar to Fig. 2f), in which the unwinding rate was 4-fold lower than that reported by Dessinges et al. (Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004). In this second sm study on UvrD (Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008), a similar statistical fluctuation analysis to that of Dessinges et al. (Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004) was applied to calculate an unwinding kinetic step size of 4·2 bp. Xi and co-workers also suggest, based on the non-exponential distribution of off-times between unwinding bursts, that a UvrD dimer catalyzes unwinding in their assay (Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008).

The ATP coupling ratios for translocation calculated from single-turnover ATPase time courses in the presence of varying length oligonucleotides were ~1 nt/ATP hydrolyzed for PcrA (Dillingham et al. Reference Dillingham, Wigley and Webb2000) and also for UvrD (Tomko et al. Reference Tomko, Fischer, Niedziela-Majka and Lohman2007). Recently, an ATP coupling ratio for unwinding by PcrA in the presence of initiator protein RepD was determined to be 1 bp per ATP hydrolyzed via similar assays with linear partial duplex and plasmid substrates (Slatter et al. Reference Slatter, Thomas and Webb2009). The differences between the small ATP coupling ratio predicted by crystallographic studies and single-turnover ATPase kinetics and the larger kinetic step size calculated from ensemble single-round pre-steady-state kinetics and sm studies suggest that these helicases move 1 nt/bp per ATP hydrolyzed with a rate-limiting step in translocation or unwinding occurring every 4 ATPase cycles.

3.3.3 Replicative RNA helicases

In our second example of replicative RNA helicases, an ATP coupling ratio of 1 bp unwound per ATP hydrolyzed has also been predicted based on comparison of ATP-bound and unbound crystal structures of four SFII RNA helicases, including HCV NS3 helicase (reviewed in Hopfner & Michaelis, Reference Hopfner and Michaelis2007). However, similar to the SFIA helicase family, ensemble kinetic and sm studies have shown larger step sizes and unwinding bursts. Single-round pre-steady-state kinetic measurements of unwinding of dsRNA of different lengths by Vaccinia virus NPH-II revealed a 6 bp RNA unwinding kinetic step size (Jankowsky et al. Reference Jankowsky, Gross, Shuman and Pyle2000). A combination of ensemble steady-state and pre-steady-state kinetic measurements of NS3 unwinding of RNA of different lengths revealed both a physical and a kinetic step size for RNA unwinding of 18 bp (Serebrov & Pyle, Reference Serebrov and Pyle2004). A complementary sm approach using optical tweezers to monitor NS3 unwinding of RNA hairpins (Dumont et al. Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006) found that extension occurred with a physical step size for unwinding of ~11 bp that was independent of ATP concentration and force. However, the optical tweezers experiment also detected three smaller sub-steps of 3·6 bp within each physical step, suggesting that NS3 is not unwinding in a single motion but in discrete smaller motions coordinated and triggered by the ATPase cycle, consistent with ATP-dependence of the stepping velocity and duration of pauses between steps (Dumont et al. Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006) (Fig. 4a).

Fig. 4.

Single-molecule studies on helicase step size: (a) From Dumont et al. (Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006). (Left) Optical tweezer set-up to monitor NS3 helicase unwinding of an RNA hairpin (see Fig. 2h). (Center) Pairwise distance distribution for unwinding extension time traces reveals an 11±3 bp periodicity in pausing during extension. (Right) Unwinding substeps of 3·6±1·3 bp that appear within the 11 bp step observed at limiting ATP concentrations. Reprinted by permission from Macmillan Publishers Ltd: Nature 439, 105–108, copyright (2006). (b) From Myong et al. (Reference Myong, Bruno, Pyle and Ha2007). (Far left) Experimental smFRET set-up to measure NS3 unwinding of 18 bp partial duplex DNA. (Mid left) FRET unwinding time trace fit to the step-finding algorithm quantitates FRET values and dwell times for six steps. (Mid right) Total density plot representing two-dimensional histograms for pairs of FRET values entering and exiting each transition reveals six peaks. (Far right) Gamma distribution fitting to collected dwell times at each plateau pause reveals three irreversible steps within the 3 bp step, indicating 1 nt elemental step size within each 3 bp transition. From Myong, S. et al. (Reference Myong, Bruno, Pyle and Ha2007) Science 317, 513–516. Reprinted with permission from AAAS. See text for further details.

A separate smFRET-based assay of DNA unwinding by NS3 helicase (Myong et al. Reference Myong, Bruno, Pyle and Ha2007) revealed unwinding of a partial duplex 18 bp DNA in 3 bp steps. Dwell time analysis of pause durations at each step was consistent with three irreversible steps that need to occur before 3 bp unwinding, indicating that the elemental step size within each 3 bp is 1 nt. This led to the development of a model that translocation occurs in 1 nt increments, building up tension in a ‘spring-loaded’ NS3 such that every 3 nt, unwinding occurs in 3 bp bursts (Myong et al. Reference Myong, Bruno, Pyle and Ha2007). These sm studies thus provide a way to understand the 1 nt translocation coupling ratio predicted by crystal structures with respect to the larger 3·6 bp unwinding steps observed (Fig. 4b). However, what about the larger kinetic steps sizes of 11 and 18 bp previously observed (Dumont et al. Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006; Serebrov & Pyle, Reference Serebrov and Pyle2004)? In another series of pre-steady-state stopped flow experiments using RNA of varying lengths and sequence (Serebrov et al. Reference Serebrov, Beran and Pyle2009), the step size reduced from 18 to 11 bp upon lowering the ionic strength. The NS3 oligomeric state was correlated with the ionic strength used in the previous studies, with it operating as a monomer at low ionic strength (Dumont et al. Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006) but as a dimer at higher ionic strength (Serebrov & Pyle, Reference Serebrov and Pyle2004). As the smFRET study (Myong et al. Reference Myong, Bruno, Pyle and Ha2007) also showed that NS3 remained associated with the unwound strand, Pyle and co-workers (Serebrov et al. Reference Serebrov, Beran and Pyle2009) then proposed that the 11 or 18 bp steps corresponded to 5′ unwound tail product release in pulses after RNA unwinding has reached a certain length.

3.3.4 Implications for helicase mechanisms

The combined work on NS3 provides evidence that helicases can move in 1 nt/bp steps which are linked to large conformational physical motions that occur during unwinding such as product release that are apparent as much bigger ‘steps.’ In general for all the helicases described, the smaller kinetic step sizes of 3–6 nt/bp that are observed through sm and pre-steady-state single-turnover kinetics represent ATP-dependent rate-limiting events such as unwinding bursts that occur after moving a discrete number of bp in single-nt increments. Nonlinear global fitting of the single-cycle kinetic ensemble data is limited by two factors: (i) the assumption that helicase movement occurs without slippage or pausing, which would lead to an overestimation of step-size (Fischer et al. Reference Fischer, Maluf and Lohman2004) and (ii) the uncertainty in step-size estimation caused by analyzing multiple datasets as a group (each with a limited amount of data) as opposed to a single dataset with less variation (Lucius et al. Reference Lucius, Vindigni, Gregorian, Ali, Taylor, Smith and Lohman2002). Both of these problems can be addressed by sm techniques, which in fact have detected pausing by many helicases (Lionnet et al. Reference Lionnet, Dawid, Bigot, Barre, Saleh, Heslot, Allemand, Bensimon and Croquette2006) and derived kinetic information from datasets derived from single molecules (Myong et al. Reference Myong, Bruno, Pyle and Ha2007, Reference Myong, Rasnik, Joo, Lohman and Ha2005).

3.4.1 Pausing and reversal by RecBCD

E. coli RecBCD is a dual helicase/nuclease required in recombinational repair of dsDNA breaks. The helicase initially binds at the break site and proceeds to unwind ~30 kbp using two motors, RecD and RecB, which translocate with opposite polarity. As it unwinds DNA, the enzyme degrades both strands at differential rates. Upon encountering a χ recombination hotspot, RecBCD undergoes a conformational change resulting in decreased helicase/nuclease activity and production of a long 3′ ss tail for loading of RecA and continuation of homologous recombination. The results of several sm studies have helped elucidate the mechanism and χ-site regulation of RecBCD translocation.

Two zero/low-force sm experimental systems have provided evidence for highly processive unidirectional movement by RecBCD along dsDNA. A tethered-particle-assay (TPA) investigation (Dohoney & Gelles, Reference Dohoney and Gelles2001) monitored changes in Brownian motion of a streptavidin-coated polystyrene bead linked to RecBCD through a biotin-tagged RecD as it unwound/degraded a 1 kb tethered DNA (in an assay similar to Fig. 2g). No enzyme pausing was detected regardless of whether the DNA contained an internal χ site, with RecBCD dissociating upon reaching the tether, thus ruling out a model where the χ-site-induced conformational change involves RecD subunit ejection. Another study (Bianco et al. Reference Bianco, Brewer, Corzett, Balhorn, Yeh, Kowalczykowski and Baskin2001) used optical trapping fluorescence microscopy to measure RecBCD movement through 40-fold longer λ DNA (Fig. 2d). In this case, a small force was applied via the flow-stretch experimental set-up in order to stretch the long λ DNA. One end of the DNA was specifically attached to a microsphere and the dsDNA was visualized by YOYO fluorescence. A microfluidics set-up was used in which the DNA–RecBCD complex in one channel is captured in the optical trap and then moved to a second channel where ATP is flowed in. RecBCD unwinding/degradation was followed by monitoring changes in fluorescence while YOYO was released upon dsDNA shortening. Again, highly processive unwinding with no pausing was observed at saturating ATP over the length of the entire λ DNA.

This optical trapping fluorescence method was next applied to analyze RecBCD unwinding of 30 kb DNA containing χ sequences (Spies et al. Reference Spies, Bianco, Dillingham, Handa, Baskin and Kowalczykowski2003), and ~5 s χ-specific pauses were observed followed by unwinding at a reduced rate rate (Fig. 5a). This phenomenon was postulated to reflect either uncoupling of RecD from the holoenzyme or a conformational change in RecBCD that inactivates RecD. The detection of χ-induced pauses in this set-up as opposed to the TPA experiment was probably due to the higher spatiotemporal resolution. An extension of this experiment was performed without YOYO but instead using RecBCD tagged through its RecD subunit with a fluorescent bead (Handa et al. Reference Handa, Bianco, Baskin and Kowalczykowski2005). Again, processive unwinding was observed prior to a 6 s pause at the χ site, after which the unwinding rate decreased 2-fold. The fluorescent nanoparticle stayed with the RecBCD for the entire duration of the pause, thus agreeing with the conclusion made by Dohoney & Gelles (Reference Dohoney and Gelles2001) that RecD is not dissociating at χ. Further analysis using the same assay with a RecBCD mutant with a non-functional RecD subunit (Spies et al. Reference Spies, Amitani, Baskin and Kowalczykowski2007) led to the finding that RecD is the lead motor prior to χ recognition whereupon RecD becomes uncoupled from the holoenzyme and RecB becomes the lead motor. The unwinding rates correlated with a model in which there is an initial requirement for fast unwinding of damaged or foreign DNA, but χ recognition signals the need to slow down in order to properly assemble the RecA filament.

Fig. 5.

Single-molecule detection of pausing and reversal during helicase unwinding. (a) From Spies et al. (Reference Spies, Bianco, Dillingham, Handa, Baskin and Kowalczykowski2003). Optical trap fluorescence microscopy (see Fig. 2d) to monitor DNA unwinding by RecBCD detects χ-site-specific pausing followed by unwinding at a slower velocity by a RecD-uncoupled holoenzyme. Reprinted from Cell 114, Spies, M. et al., ‘A molecular throttle: The recombination hotspot, χ controls DNA translocation by the RecBCD helicase,’ Copyright (2003), with permission from Elsevier. (b) From Sun et al. (Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008). Magnetic tweezers study of UvrD unwinding of DNA hairpin substrate (far left) shows multiple patterns in unwinding (UW) bursts followed by rapid rehybrization (not shown) or slower rewinding (SRW). (Mid left) Pauses (P) during unwinding may reflect transient dimer dissociation. (Mid right) Transitions from a slow to fast rewinding rate (FRW) may reflect dimer dissociation during translocation on the opposing strand after strand switching. (Far right) Inverse force dependence of unwinding rate was incorporated into a strained inchworm model for unwinding in which UvrD dimers bend the 3′ ssDNA tail to form an active unwinding species. Reprinted by permission from Macmillan Publishers Ltd: EMBO J. 27, 3279–87, copyright (2008). (c) From Yodh et al. (Reference Yodh, Stevens, Kanagaraj, Janscak and Ha2009). smFRET study of unwinding of forked DNA substrate (far left) by human BLM helicase. (Mid left) Example of partial repetitive unwinding behavior by a single helicase. (Mid right) Histogram of the change in FRET values in repetitive unwinding events shows a peak at 0·3 ▵FRET efficiency. (Far right) A threshold ▵FRET of 0·2 must be achieved before BLM reverses unwinding. These data suggest that BLM ‘measures’ the amount of bp it unwinds before it reverses unwinding. Reprinted by permission from Macmillan Publishers Ltd: EMBO J. 28, 405–16, copyright (2009). See text for further details.

As a follow-up to the original TPA study (Dohoney & Gelles, Reference Dohoney and Gelles2001), a force–TPA assay was developed in which a stretching force was applied by continuous laminar flow to the DNA–bead complex (Fan & Li, Reference Fan and Li2009). Using this force–TPA set-up with low ATP and 0·2 pN stretching force, RecBCD unwound a 4·1 kb DNA at the same rate observed previously, except in this case χ-specific pausing was observed followed by a rate decrease, suggesting that the lack of pausing seen by Dohoney & Gelles (Reference Dohoney and Gelles2001) was likely due to the more limited resolution of that experimental set-up.

Block and co-workers applied a different force-based approach to investigate RecBCD activity (Perkins et al. Reference Perkins, Li, Dalal, Gelles and Block2004), in which a biotinylated RecBCD was tethered to a coverslip via a streptavidin linkage and a 7 kb DNA substrate lacking χ sites was attached at its distal end to a polystyrene bead held inside an optical trap over a varying force range as shown in Fig. 2i. The surface-bound enzyme bound to the free DNA end, and, upon ATP addition, its movement along dsDNA toward the trap, was monitored under constant force up to 8 pN. DNA unwinding correlated with decreasing length of the tether, and RecBCD was observed to move unidirectionally over ~1000 s of bp. However, in contrast to the previous zero- or low-force studies (Bianco et al. Reference Bianco, Brewer, Corzett, Balhorn, Yeh, Kowalczykowski and Baskin2001; Dohoney & Gelles, Reference Dohoney and Gelles2001), χ-independent pausing was observed, including transient force-dependent pauses as well as spontaneous pausing that was independent of both force and ATP. This high-resolution optical trapping method revealed yet another surprising behavior – sporadic uneven RecBCD-dependent backward movement during unwinding. Releasing the force converted the reversals into forward motions, indicating backsliding was caused by force. The change in tether length upon backsliding was consistent with partial ssDNA formation, indicating possibly an ssDNA loop formation after unwinding but prior to degradation. In general, the force-dependent pausing and reversal during RecBCD movement were not thought to be sequence-dependent but instead to reflect kinetically distinct persistent states of the enzyme–DNA complex.

3.4.2 Force-dependent unwinding/rewinding bursts by UvrD and replicative helicases

Magnetic tweezers studies of UvrD helicase have revealed pausing and reversal behavior (Dessinges et al. Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004; Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008). Croquette and co-workers (Dessinges et al. Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004) monitored the unwinding of an 11 kb singly nicked duplex DNA that was anchored to a magnetic bead at one end of the duplex and to a glass surface at the other end (as in Fig. 2e). At UvrD <1 nm, the extension versus time curves obtained at high force >30 pN exhibited isolated unwinding bursts. Each burst showed a regular increase in extension corresponding to unwinding. The slope and height of the burst were used to calculate the rate of unwinding (U) and extent of unwinding, respectively. Unwinding proceeded processively without pausing but occasionally the extension signal decreased back to baseline in one of two fashions – UH bursts showed an instantaneous decrease corresponding to enzyme dissociation and rapid rehybridization (H), whereas UZ bursts showed a much slower decrease reflecting a slower rezipping (Z) event. In UZ bursts, the slow rezipping rate was equivalent to the unwinding rate, and was ATP-dependent indicating that ssDNA translocation might be involved. These findings led the authors to propose a model in which UvrD is capable of reversing unwinding by switching strands and translocating 3′–5′ on the opposing single strand, thus inducing reannealing behind the helicase. The high force required to achieve unwinding at low UvrD concentrations in this study most likely helps prevent reannealing during unwinding and was proposed to be the reason why unwinding is achieved at sub-nm concentrations unlike in bulk studies (Maluf et al. Reference Maluf, Fischer and Lohman2003).

A second study (Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008) analyzed force-dependent UvrD unwinding of a λ DNA hairpin situated between a λ dsDNA tethered to the surface and a 3′ ssDNA oligonucleotide tethered to a magnetic bead (similar to Fig. 2f). At saturating ATP, UvrD <10 nm and forces >4 pN, UvrD produced the UH and UZ unwinding burst patterns observed previously (Dessinges et al. Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004); however, additional patterns were also observed. First, pausing was observed in 25% of unwinding events during both U and Z phases as well as between them. The pausing frequency was inversely proportional to UvrD concentration, thus relating pausing to the enzyme oligomeric state. Second, slow and fast rezipping (Z) rates were observed, indicating there may be two kinds of translocases operating. The U and slow Z rates in the hairpin study (Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008) were 10-fold lower than those measured previously on nicked DNA (Dessinges et al. Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004), which were actually closer to the monomeric UvrD ssDNA translocation rate measured in ensemble studies (Fischer et al. Reference Fischer, Maluf and Lohman2004). The fast rezipping rate observed in the hairpin study was also closer to the ssDNA translocation rate. The distributions of off-times between unwinding bursts followed a non-exponential distribution that was inconsistent with single enzyme binding kinetics (Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008). In light of these findings, Xi and co-workers proposed that their unwinding bursts reflected unwinding by a dimeric helicase with the pausing representing transient or permanent dimer dissociation (Fig. 5b), which is reminiscent of the unwinding initiation and reinitiation model proposed for Rep helicase based on sm analysis (Ha et al. Reference Ha, Rasnik, Cheng, Babcock, Gauss, Lohman and Chu2002). The switch between slow and fast rezipping may reflect dimer dissociation/monomer–monomer reassociation, with both dimers and monomers being able to switch strands and translocate on ssDNA (Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008).

Perhaps the most intriguing result of the second magnetic tweezers study (Sun et al. Reference Sun, Wei, Zhang, Zhang, Dou, Li and Xi2008) was that the UvrD unwinding rate decreased with increasing force, indicating that force kinetically impeded unwinding (Fig. 5b). From this the authors proposed that UvrD uses a strained inchworm mechanism for unwinding involving a UvrD dimer that must bend the 3′ ssDNA tail in order to achieve a modeled interaction of both 2B domains required for activation of unwinding. The experimental force on the ssDNA tail in this system may hinder the bending of the DNA at the junction and thus unwinding.

Croquette and co-workers also observed primarily UH/UZ-type events when monitoring unwinding of a hairpin substrate by T4 hexameric gp41 helicase via magnetic tweezers as shown in Fig. 2f (Lionnet et al. Reference Lionnet, Spiering, Benkovic, Bensimon and Croquette2007) at saturating ATP, and forces between 6–10 pN. In UH events, the helicase unwound the hairpin fully and then dissociated, causing the hairpin to reanneal instantaneously. In UZ events, the slow rezipping phase was ATP-dependent, signifying helicase dependence; however, this was not caused by strand switching as proposed for UvrD (Dessinges et al. Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004) because the original substrate was a hairpin. Instead, the gp41 was proposed to continue translocating along the newly unwound DNA with the hairpin reannealing in its wake. Similar reverse motions characterized by unwinding/rewinding bursts have also been observed albeit infrequently for the T7 hexameric helicase using an optical trap-based unzipping assay at forces between 7 and 12 pN (Johnson et al. Reference Johnson, Bai, Smith, Patel and Wang2007). Finally, rare stepwise rewinding events have also been detected during NS3 unwinding of an RNA hairpin measured at 17 pN using an optical tweezers-based assay (Dumont et al. Reference Dumont, Cheng, Serebrov, Beran, Tinoco, Pyle and Bustamante2006).

3.4.3 Reverse motions during helicase activity detected by smFRET

smFRET studies on helicases have also revealed reverse motions and stalling during helicase activity at zero force. Initial studies on E. coli Rep helicase (Ha et al. Reference Ha, Rasnik, Cheng, Babcock, Gauss, Lohman and Chu2002) using an unwinding substrate – a partial duplex DNA with both donor and acceptor dyes at the ssDNA–dsDNA junction (Fig. 2a) – revealed that Rep could efficiently unwind an 18 bp duplex but not a 40 bp duplex. Seventy percent of the time, unwinding of the 40-mer was punctuated by stalls that usually led to dissociation of Rep and reannealing of the substrate. The stalls were proposed to be due to dissociation of dimeric rep, the functional oligomeric state for unwinding. In some cases, reassociation of monomers led to unwinding reinitiation on the same substrate.

Ha and co-workers also discovered repeating unwinding patterns in smFRET time traces for two SFII helicases – HCV NS3 (Myong et al. Reference Myong, Bruno, Pyle and Ha2007) and the human Bloom Syndrome helicase (BLM) helicase (Yodh et al. Reference Yodh, Stevens, Kanagaraj, Janscak and Ha2009). Using partial duplex DNA templates with a surface-tethered 18 bp duplex and acceptor and donor dyes situated at the junction on adjacent strands, the addition of ATP caused NS3 to unwind the DNA (with a corresponding FRET decrease); however, upon reaching the surface, the DNA reannealed instantaneously. This pattern was repeated in succession on the same molecule and persisted even after washing out free protein, indicating it was catalyzed by a single NS3 species. Incomplete, repetitive unwinding was also observed using duplexes longer than 18 bp, indicating that a steric blockade is not necessary for unwinding reversal. The authors proposed a model that NS3 remains in contact with the displaced strand during unwinding and, upon reaching a steric or length-based block, returns to the junction with concomitant strand reannealing and begins the unwinding cycle again (Myong et al. Reference Myong, Bruno, Pyle and Ha2007).

Repetitive unwinding was also catalyzed by BLM helicase, a human RecQ homologue, on forked DNA of varying duplex lengths with the dyes situated in close proximity to the fork junction (Yodh et al. Reference Yodh, Stevens, Kanagaraj, Janscak and Ha2009) (Fig. 5c). This behavior was observed for wild-type BLM and a mutant comprising the RecQ catalytic core that lacks the oligomerization domain (Janscak et al. Reference Janscak, Garcia, Hamburger, Makuta, Shiraishi, Imai, Ikeda and Bickle2003). This fact as well as the protein concentration-independence of unwinding times led to the conclusion that successive unwinding events on the same DNA molecule are carried out by a single BLM monomer. Each unwinding event was characterized by three phases: (I) a phase during which FRET decreases by a certain amount, ΔFRET, due to unwinding, followed by (II) a time period during which FRET remains at a constant low value, followed by (III) a reannealing phase characterized by a rapid switch back to high FRET. The ATP-dependence of reannealing time was reminiscent of the ATP-dependence of the slow-rezipping events detected for UvrD previously (Dessinges et al. Reference Dessinges, Lionnet, Xi, Bensimon and Croquette2004). It was proposed that BLM partially dissociated after unwinding a critical length of DNA, and then switched its helicase motors to the opposing strand and translocated 3′–5′ on this strand with the strands reannealing in its wake. However, unlike UvrD and T7, which terminated unwinding stochastically after a random number of bp were unwound, BLM appeared to terminate after a short but defined length of DNA was unwound – based on a defined peak observed in ΔFRET distributions. Thus, BLM appears to be able to ‘measure’ the number of bp it has unwound, and once it has unwound a critical length, it rapidly reverses unwinding via strand switching (Fig. 5c). The ability to switch strands and measure DNA unwound may facilitate BLM processing of stalled replication forks and illegitimately formed recombination intermediates.

In these investigations on NS3 and BLM, smFRET was able to detect partially unwound intermediates in real time, leading to the discovery of repetitive unwinding. Repetitive behaviors during helicase translocation by Rep and RIG-I proteins (Myong et al. Reference Myong, Cui, Cornish, Kirchhofer, Gack, Jung, Hopfner and Ha2009; Myong et al. Reference Myong, Rasnik, Joo, Lohman and Ha2005) were also discovered using sm fluorescence (described in section 3.5).

3.5.1 Repetitive shuttling during translocation

smFRET was used to measure E. coli Rep translocation on ssDNA using a partial duplex substrate with acceptor dye positioned at the junction end of the 3′ ssDNA tail while the enzyme was donor-labeled (Myong et al. Reference Myong, Rasnik, Joo, Lohman and Ha2005). These experiments were performed at sub-nm concentrations of Rep that do not allow unwinding which requires more than a monomer of Rep. In the presence of ATP, ssDNA translocation by Rep was expected to initiate with the enzyme binding to the available 3′ end of the ssDNA tail producing a low FRET signal and then moving toward the junction with a corresponding FRET increase related to the distance translocated. However, the enzyme did not dissociate upon reaching the junction but instead appeared to snap back to the 3′ end of the ssDNA (observed as an abrupt FRET decrease), and showed repeated cycles of translocation and snap-back. The same phenomenon was observed with dual-labeled Rep and an unlabeled translocation substrate, indicating that Rep undergoes a conformational change during translocation from an open to closed form, possibly exposing a secondary binding site that then rebinds the free 3′ tail, transferring Rep from the junction back to the 3′ end of the ssDNA tail whence it begins another cycle (Fig. 6a). This translocation behavior was termed ‘repetitive shuttling’ and was proposed to play a role in the removal of proteins such as RecA filaments in order to prevent unwanted recombination. Note that Rep shuttling behavior was also observed using sm protein-induced fluorescent enhancement (PIFE) with unlabeled Rep on a Cy3-labeled DNA (Myong et al. Reference Myong, Cui, Cornish, Kirchhofer, Gack, Jung, Hopfner and Ha2009).

Fig. 6.

Single-molecule detection of helicase conformational dynamics. (a) From Myong et al. (Reference Myong, Rasnik, Joo, Lohman and Ha2005). (Left) The smFRET assay demonstrates that dual-labeled Rep helicase catalyzes repetitive translocation along unlabeled ssDNA, indicating that Rep itself undergoes a conformational change during the shuttling cycle. (Right) Model for the translocation cycle in which Rep transitions from open to closed forms, exposing a secondary binding site for the free 3′ ss tail, allowing Rep to transfer from the junction to the 3′ end and reinitiate the cycle. Reprinted by permission from Macmillan Publishers Ltd: Nature 437, 1321–1325, copyright (2005). (b) From Theissen et al. (Reference Theissen, Karow, Kohler, Gubaev and Klostermeier2008). (Top left) Dual fluorescent labeling of YxiN, DEAD-box helicase with donor and acceptor on different sides of the interdomain cleft separating the two RecA-like helicase domains. (Right) FRET histograms±ligands for fluorescently labeled YxiN* S108C/S229C. FRET in the absence of any ligands or in the presence of any individual ligand is 0·5. Cooperative binding of ADPNP and RNA promotes interdomain cleft closure, increasing FRET to 0·8. (Bottom left) FRET histograms±ligands for fluorescently labeled YxiN* A115C/S229C. FRET in the absence of any ligands or in the presence of any individual ligand is 0·42. Cooperative binding of ADPNP and RNA promotes interdomain cleft closure, increasing FRET to 0·72. Adapted from Theissen, B. et al. (Reference Theissen, Karow, Kohler, Gubaev and Klostermeier2008). Copyright (2008). National Academy of Sciences, USA. See text for further details.

The PIFE technique was also applied to show that the dsRNA-dependent SFII DExH ATPase, RIG-I, a cytosolic viral sensor, can translocate on dsRNA in an ATP-dependent manner (Myong et al. Reference Myong, Cui, Cornish, Kirchhofer, Gack, Jung, Hopfner and Ha2009). Similar to Rep translocation along ssDNA, a single RIG-I catalyzed translocation in successive repetitions on the same RNA molecule. Although the physical mechanism of RIG-I dsRNA translocation is still unclear, the finding that this activity could be trans-activated by 5′ triphosphate caps that are typically present on viral RNA post-infection is consistent with a role for this activity in viral sensing by RIG-I.

3.5.2 Replicative helicase–DNA tail interaction dynamics

smFRET has also been applied to investigate interaction dynamics between replicative helicases and ssDNA tails at the fork (Rothenberg et al. Reference Rothenberg, Trakselis, Bell and Ha2007). The archaeal minichromosome maintenance protein (MCM) from Sulfolobus solfataricus (SsoMCM), a homologue of eukaryotic hexameric MCM replicative helicases, was previously characterized to preferentially unwind Y-shaped substrates with accessible 3′ tails. Using acceptor-labeled SsoMCM and donor-labeled substrates, smFRET was applied (Rothenberg et al. Reference Rothenberg, Trakselis, Bell and Ha2007) to determine the orientation of MCM along the 3′ tail, and confirm the helicase's preference for fork substrates. Binding of unlabeled SsoMCM to forked substrates with acceptor dye placed at the end of the 5′ tail and donor dye placed at various locations on the 3′ tail increased FRET from low to high, consistent with the idea that MCM bound the 3′ tail through its central channel while interacting with the 5′ tail through an external site on the protein surface. While MCM bound very stably without dissociating for up to an hour, temporal fluctuations in FRET were also observed with N-terminally acceptor-labeled MCM and the donor-labeled 5′ tail but not with the donor-labeled 3′ tail, indicating a specific dynamic interaction between the helicase and the 5′ tail. This work provided evidence that a replicative helicase may also interact with the non-tracking strand during unwinding, possibly playing a role in substrate specificity, helicase stability at the fork, and unwinding (Rothenberg et al. Reference Rothenberg, Trakselis, Bell and Ha2007), and suggesting that ‘exclusion’ of this strand may not be the main mechanistic role.

3.5.3 DEAD-Box RNA helicase conformational switching

Cofactor-induced conformational changes in DEAD-box RNA helicases have also been probed using smFRET-based techniques (Aregger & Klostermeier, Reference Aregger and Klostermeier2009; Karow & Klostermeier, Reference Karow and Klostermeier2009; Linden et al. Reference Linden, Hartmann and Klostermeier2008; Theissen et al. Reference Theissen, Karow, Kohler, Gubaev and Klostermeier2008). Crystal structures of members of this helicase family reveal a common core structure consisting of two RecA-like helicase domains separated by an interdomain cleft, and suggest that simultaneous binding of ssRNA and nucleotide induces a conformational change from an ‘open’ state in which the helicase domains are separated to a ‘closed’ state in which the helicase domains make multiple contacts with each other. As there is no DEAD-box RNA helicase for which both open and closed structures are available, Klostermeier and co-workers utilized a confocal smFRET method (shown in Fig. 2b) to analyze interdomain cleft movement in Bacillus subtillus YxiN, a helicase involved in ribosome biogenesis (Theissen et al. Reference Theissen, Karow, Kohler, Gubaev and Klostermeier2008). YxiN double mutants were constructed for attachment of a donor and an acceptor in each helicase domain. In the absence of ligands, double-labeled YxiN followed a unimodal mid-FRET distribution consistent with a single open conformation. In the presence of nucleotide or RNA alone, the FRET distributions remained unaltered; however, simultaneous addition of ADPNP/ATP plus RNA resulted in a switch to a high FRET species for all the YxiN double mutants. Thus RNA and ATP substrates act cooperatively to induce closure of the interdomain cleft, bringing the helicase domains in close contact (Fig. 6b). Thus, smFRET captured both open and closed conformations of YxiN and correlated them with RNA/ADP-bound and RNA/ATP-bound states of the helicase, respectively (Theissen et al. Reference Theissen, Karow, Kohler, Gubaev and Klostermeier2008). In a follow-up study (Aregger & Klostermeier, Reference Aregger and Klostermeier2009), it was shown that YxiN remained in the closed state in the presence of ATP analogues plus RNA, suggesting that helicase reopening post-ATP hydrolysis is most likely coupled to P_i release. Furthermore, YxiN mutants deficient in unwinding were still able to undergo interdomain cleft closure, indicating that cleft closure is important but not sufficient for unwinding (Karow & Klostermeier, Reference Karow and Klostermeier2009). This assay was also applied to investigate the substrate specificity of a related DEAD-box RNA helicase, Hera from Thermus thermophilus (Linden et al. Reference Linden, Hartmann and Klostermeier2008). Two RNA substrates that were identified by their ability to induce interdomain cleft closure were RNase P RNA and 23S rRNA hairpin 92, suggesting a role for Hera in ribosome biogenesis. These studies illustrate the success of smFRET applications to determining not only conformational changes undergone by these DEAD-box RNA helicases but also information on their cellular functions (Linden et al. Reference Linden, Hartmann and Klostermeier2008).

3.6.1 Phage T4 primosome assembly

Sm techniques have been used to probe the assembly of the bacteriophage T4 primosome, comprised of the hexameric helicase (gp41), primase (gp61), ssDNA binding protein (gp32) and accessory helicase loader protein (gp59). This complex works together with DNA polymerase (gp43), which is aided by sliding clamp (gp45) to carry out DNA synthesis as a replisome. While gp41 can bind to forked templates by itself, its binding is stimulated by gp59, which recruits and loads gp41 helicase onto gp32-coated DNA during recombinational replication. Hammes and co-workers determined the order of assembly of primosome components using a combination of ensemble and smFRET techniques (Zhang et al. Reference Zhang, Spiering, Trakselis, Ishmael, Xi, Benkovic and Hammes2005). In this work, all four primosome proteins were single-labeled with donor or acceptor fluorophores and analyzed for interactions between the various proteins in the presence of ssDNA (ensemble FRET) to obtain binding stoichiometries, or with a biotinylated forked DNA containing both leading and lagging strand junctions (smFRET) to prove direct interactions between proteins and the order in which they were required for assembly. The results led to a model in which a 1:1 complex of gp32 and gp59 first forms on forked DNA followed by recruitment of gp41 helicase and its hexamerization. Subsequent ATP hydrolysis by gp41 leads to dissociation of gp32 and gp59. Primase (gp61) then binds to the gp41–DNA complex to complete primosome assembly. Recently, sm photobleaching experiments showed that the hexameric form of gp59 is required for helicase loading (Arumugam et al. Reference Arumugam, Lee and Benkovic2009). The same group has also used this assay to further probe the role of helicase loader (gp59) in initiation of lagging strand replication (Xi et al. Reference Xi, Zhang, Zhuang, Yang, Spiering, Hammes and Benkovic2005a, Reference Xi, Zhuang, Zhang, Selzer, Spiering, Hammes and Benkovicb).

3.6.2 Phage T7, T4 and E. coli replisome dynamics

van Oijen and co-workers have used sm flow-stretch techniques to investigate replisome activity in the bacteriophage T7 and E. coli systems (Hamdan et al. Reference Hamdan, Johnson, Tanner, Lee, Qimron, Tabor, van Oijen and Richardson2007, Reference Hamdan, Loparo, Takahashi, Richardson and van Oijen2009; Lee et al. Reference Lee, Hite, Hamdan, Xie, Richardson and van Oijen2006; Tanner et al. Reference Tanner, Hamdan, Jergic, Loscha, Schaeffer, Dixon and van Oijen2008, Reference Tanner, Loparo, Hamdan, Jergic, Dixon and van Oijen2009). The experimental set-up is as shown in Fig. 2c, in which a forked λ DNA template is tethered to the surface via a ssDNA region at the 5′ end of the lagging strand template; its 3′ end is attached to a tracking bead; and the entire template is subject to laminar flow at ~3 pN. A DNA primer is annealed to the other strand, creating a primer–template junction for leading strand synthesis. Leading strand synthesis by T7 polymerase was measured in this system (Lee et al. Reference Lee, Hite, Hamdan, Xie, Richardson and van Oijen2006) by assembling a minimal replisome (helicase–primase (gp4) and thioredoxin–polymerase (trx–gp5)) and monitoring DNA shortening of the ss lagging strand template during fork unwinding/DNA synthesis. The rate of DNA synthesis matched that measured in bulk assays; however, the processivity of DNA synthesis/unwinding in this concerted system was 1–2 orders of magnitude higher relative to that measured with polymerase or helicase alone. When rNTPs were added to this system to enable primase, short pauses during leading strand synthesis trajectories were observed. These were interpreted to reflect pausing of the polymerase while primers were being formed on the lagging strand template. When excess polymerase was additionally added to this system to enable lagging strand synthesis, the pauses were followed by continued shortening of the DNA and then abrupt lengthening. This pattern was interpreted to represent pausing during primer formation followed by the creation and release of a DNA loop during lagging strand synthesis (i.e. trombone model). These real-time trajectories of DNA length changes created during action of the minimal T7 replisome were incorporated into a primase-brake model to explain how lagging and leading strand synthesis are synchronized (Lee et al. Reference Lee, Hite, Hamdan, Xie, Richardson and van Oijen2006) (Fig. 7a).

Fig. 7.

Single-molecule studies of helicases within replisome complexes. (a) From Lee et al. (Reference Lee, Hite, Hamdan, Xie, Richardson and van Oijen2006). Flow-stretch assay (see Fig. 2c) of phage T7 replisome activity monitoring DNA shortening of the lagging strand template generated during leading strand synthesis. Pauses (I) were interpreted to reflect primer synthesis after which lagging strand synthesis continues with the formation of a lagging strand replication loop (II). Loop release (III) is observed as lengthening of the DNA. Reprinted by permission from Macmillan Publishers Ltd: Nature 439: 621–4, copyright (2006). (b) From Manosas et al. (Reference Manosas, Spiering, Zhuang, Benkovic and Croquette2009). Magnetic tweezers assay (see Fig. 2f) of phage T4 primosome (gp41 helicase/gp61 primase) activity using a hairpin substrate. Trajectories show an increase in DNA extension representing hairpin unwinding followed by fast or slow (shown) rehybridization (decrease in DNA extension) as helicase continues moving around the hairpin. (Left) T4 primosome does not pause while priming (i.e. the unwinding velocity remains constant) and rezipping traces are interrupted by primer-specific periods of constant extension interpreted to represent blocks in reannealing caused by a dissociated primase–primer complex. (Right) Detection of priming loop formation in DNA extension time traces represented by periods of slower unwinding velocity followed by jumps back to the original rate reflecting loop release. The priming loop is created by intact primosomes such that the unwound lagging strand template forms a loop that is released upon primer completion and transfer to the lagging strand polymerase. Reprinted by permission from Macmillan Publishers Ltd: Nature Chemical Biology 5, 904–912, copyright (2009). See text for further details.

This flow-stretch assay also confirmed that an electrostatic interaction between the C-terminal tail of the helicase and a basic region in the trx-binding region of the polymerase was crucial for T7 replisome processivity via retention of trx-gp5 in the event that it dissociates during leading strand synthesis (Hamdan et al. Reference Hamdan, Johnson, Tanner, Lee, Qimron, Tabor, van Oijen and Richardson2007). The kinetics of T7 trombone loop formation and dissolution were further analyzed (Hamdan et al. Reference Hamdan, Loparo, Takahashi, Richardson and van Oijen2009), suggesting that the trombone loop is released in response to either new primer formation or Okazaki fragment completion. Loop size also decreased with increasing stretching force, consistent with the idea that the loop is held together by predicted interactions between helicase at the fork and leading and lagging strand polymerases. Imaging of a dsDNA intercalator, YOYO, confirmed that DNA shortening represents both leading and lagging strand formation.

This experimental system was also used to measure primer extension and leading strand synthesis by the E. coli PolIII holoenzyme (Tanner et al. Reference Tanner, Hamdan, Jergic, Loscha, Schaeffer, Dixon and van Oijen2008). Whereas primer extension on an ssDNA template occurred in repeated bursts of ~1 kb separated by pauses, leading strand synthesis on the dsDNA template by the replisome, which also included DnaB, helicase loader DnaC and a minimal clamp loading complex, occurred in ~10 kb increments. The addition of primase (DnaG) decreased the processivity of DNA synthesis 3-fold through its interaction with DnaB, but no pausing was observed suggesting an alternate mechanism for slowing down leading strand synthesis during slow primer synthesis on the lagging strand.

A magnetic tweezers set-up similar to Fig. 2f was recently used to monitor phage T4 primosome (i.e. gp41–gp61 helicase–primase) activity on a DNA hairpin substrate (Manosas et al. Reference Manosas, Spiering, Zhuang, Benkovic and Croquette2009). In this system, unwinding activity is reported as an increase in DNA extension, followed by either rapid rezipping or slow reannealing of the hairpin after helicase dissociation. No pausing was observed during the unwinding phase even when primers are created. Two classes of behavior were observed during primer synthesis. In the first, primase dissociated from the helicase as suggested by periods of no extension during reannealing, which were shown to be caused by transient blocking of hairpin rezipping by the primase–primer complex (Fig. 7b). In the second, unwinding traces showed periods of slower velocity unwinding followed by a switch back to the original unwinding rate. This pattern was interpreted to represent the formation of a priming loop by an intact primosome in which the unwound DNA forms a loop between the helicase and primase that is released once the primer is synthesized. The slower unwinding velocity is due to the fact that only one of the two unwound strands contributes to the extension (Fig. 7b). A fused primosome (i.e. a single gp41–gp61 polypeptide) worked entirely via the priming loop mechanism, and addition of T4 replisome accessory proteins increased the probability of priming loop formation. This work has thus provided evidence for two types of primosome behavior during primer synthesis involving either primosome disassembly or a priming loop.

A related study on the phage T7 system investigated the coordination of leading/lagging strand synthesis via ensemble and smFRET kinetic assays (Pandey et al. Reference Pandey, Syed, Donmez, Patel, Ha and Patel2009). Using smFRET unwinding substrates such as in Fig. 2a, and similar radiolabeled substrates for ensemble DNA elongation assays, a primase activity-dependent slowing/pausing was detected during unwinding by gp4 helicase–primase but not during unwinding/DNA synthesis by the T7 replisome, in contrast to previous reports (Lee et al. Reference Lee, Hite, Hamdan, Xie, Richardson and van Oijen2006). In a separate experiment, an smFRET substrate with donor placed near the priming sequence and acceptor 40 bp away was utilized to provide evidence for primer-sequence dependent creation of a priming loop during unwinding by both gp4 and the T4 replisome. Finally, ensemble DNA elongation assays showed that lagging strand synthesis is 30% faster than leading strand synthesis by the T4 replisome. Combined, these results suggest three ways of coordinating dual strand synthesis: (1) DNA synthesis occurs concomitantly and is not interrupted by primer synthesis, (2) primase remains engaged with the priming site during unwinding/DNA synthesis, leading to loop formation in the nascent lagging strand such that the nascent primers are in close proximity for hand-off to the polymerase, and (3) the lagging strand polymerase synthesizes DNA faster, allowing more time for recycling steps and to keep up with leading strand synthesis. In light of this study, prior results that relate pausing during T7 DNA synthesis to primase action (Lee et al. Reference Lee, Hite, Hamdan, Xie, Richardson and van Oijen2006) may actually reflect uncoupled lagging and leading strand synthesis.

3.6.3 Effect of human Replication Protein A (hRPA) on BLM unwinding dynamics

The human BLM helicase has been shown to interact with numerous proteins both in vitro and in vivo. One important partner is the human ssDNA binding protein, replication factor A (hRPA), which facilitates unwinding of long duplex regions by BLM (Brosh et al. Reference Brosh, Li, Kenny, Karow, Cooper, Kureekattil, Hickson and Bohr2000). This processivity effect was postulated to be due to a direct specific interaction between BLM and hRPA. As described earlier, an smFRET study (Yodh et al. Reference Yodh, Stevens, Kanagaraj, Janscak and Ha2009) revealed a repetitive unwinding behavior by BLM on forked DNA substrates in which BLM partially unwound a critical length of DNA followed by reversal of unwinding by strand switching, and translocation on the opposing strand with concomitant reannealing of the strands in its wake. After a certain point (i.e. wait time), BLM switched back to the tracking strand and repeated the unwinding cycle. Surprisingly, the repetitive unwinding pattern persisted in the presence of hRPA with only a limited increase in unwinding processivity per cycle. The same effect was observed using the Saccharomyces cerevisiae homologue, scRPA, which interacts with BLM through the same residues as hRPA but not with the structurally non-homologous SSB from E. coli. This indicated that the hRPA effect is mediated through a direct physical interaction with BLM as opposed to solely hRPA prevention of strand reannealing. The presence of hRPA had no effect on reannealing time, indicating it is most likely dissociating during this phase. And while hRPA did not change the wait time between repetitive unwinding events for wild-type BLM, it did significantly increase the wait time for a mutant BLM that comprises the RecQ helicase core (Janscak et al. Reference Janscak, Garcia, Hamburger, Makuta, Shiraishi, Imai, Ikeda and Bickle2003) but lacks the high-affinity hRPA binding site. The latter result indicates that hRPA facilitates BLM transfer back to the tracking strand and reinitiation of unwinding via direct physical interaction between the two proteins.

3.6.4 Translocation on hRPA-coated DNA by archaeal XPD helicase

Archaeal Rad3 from Ferroplasma acidarmanus, a homologue of human XPD helicase required in nucleotide excision repair, possesses a unique Fe–S cluster in helicase domain 1. Spies and co-workers took advantage of the fluorescence quenching property of the Fe–S cluster to examine translocation by this helicase on RPA-coated DNA (Honda et al. Reference Honda, Park, Pugh, Ha and Spies2009). F. acidarmanus RPA1 and RPA2 have been shown to enhance unwinding by XPD, but whether this is mediated through direct interactions is unknown (Pugh et al. Reference Pugh, Lin, Eller, Leesley, Cann and Spies2008). Using short linear ssDNA biotinylated at the 5′ end with Cy3 tethered at the 3′ end, XPD ATP-dependent ssDNA translocation was measured using sm fluorescence microscopy as distance-dependent quenching of Cy3 as a function of time. In the presence of Cy5-labeled RPA1 and RPA2, XPD continued translocation by one of two mechanisms – either with concurrent dissociation of RPA1 or while bypassing RPA2. Both these patterns were confirmed in experiments monitoring Cy5-RPA quenching induced by translocating XPD on unlabeled DNA. These sm experiments thus provide information on the general mechanisms used by DNA repair helicases when encountering substrates coated with ssDNA binding proteins.