Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

# Structural basis for DNA unwinding at forked dsDNA by two coordinating Pif1 helicases

## Abstract

Pif1 plays multiple roles in maintaining genome stability and preferentially unwinds forked dsDNA, but the mechanism by which Pif1 unwinds forked dsDNA remains elusive. Here we report the structure of Bacteroides sp Pif1 (BaPif1) in complex with a symmetrical double forked dsDNA. Two interacting BaPif1 molecules are bound to each fork of the partially unwound dsDNA, and interact with the 5′ arm and 3′ ss/dsDNA respectively. Each of the two BaPif1 molecules is an active helicase and their interaction may regulate their helicase activities. The binding of BaPif1 to the 5′ arm causes a sharp bend in the 5′ ss/dsDNA junction, consequently breaking the first base-pair. BaPif1 bound to the 3′ ss/dsDNA junction impacts duplex unwinding by stabilizing the unpaired first base-pair and engaging the second base-pair poised for breaking. Our results provide an unprecedented insight into how two BaPif1 coordinate with each other to unwind the forked dsDNA.

## Introduction

Helicases are modular motor proteins that couple the binding and hydrolysis of ATP to unwinding of double-stranded (ds) DNA and play essential roles in nearly all aspects of nucleic acid metabolism, ranging from DNA replication to chromatin remodeling1,2. Helicases are classified into six different superfamilies based on the conserved helicase motifs, which are further divided into two subgroups, A and B according to the directionality of their translocation on single-stranded (ss) DNA1.

Saccharomyces cerevisiae Pif1 (ScPif1), the founding member of the Pif1 family helicases3, is a superfamily 1B (SF1B) helicase that has multiple functions in the maintenance of genomic homeostasis4,5. ScPif1 is a negative regulator of de novo telomere formation and elongation by strongly inhibiting telomere addition to the ends of both chromosomes and double-strand breaks (DSBs)6,7,8,9. ScPif1 contributes to DNA replication in many ways, which include processing Okazaki fragments10,11,12,13, promoting fork progression at hard-to-replicate regions such as highly transcribed tRNA genes14,15 and G-quadruplex (G4) structures16,17,18, and stimulating repair-associated DNA synthesis with Polδ during break-induced replication19,20.

Pif1 is evolutionarily conserved from bacteria to humans21. Pif1-like DNA helicases including ScPif1 and Rrm3, fission yeast Pfh1, human Pif1 (hPif1) and BaPif1 contain a degenerate sequence located between motifs II and III termed the Pif1 signature motif22. The signature motif of ScPif1 is essential in vivo for mitochondrial and nuclear functions and in vitro for ATPase activity23 while the signature motif of Pfh1 is necessary for both protein displacement and helicase unwinding activities24. It has been suggested that the Pif1 signature motif exerts its functional role indirectly through stabilizing the residues involved in ssDNA binding25 but the underlying mechanism remains elusive.

The in vitro activities of Pif1 proteins have been biochemically characterized extensively. ScPif1 requires a single-stranded DNA (ssDNA) region for binding and robustly unwinds forked dsDNA, RNA/DNA hybrids (R-loops) and some G4 structures but is very non-processive when unwinding conventional 5′-tailed dsDNA and very stable G4 structures3,17,26,27,28. Pif1 unwinds duplex DNA in a single-base-pair kinetic step, powered by the hydrolysis of one ATP molecule29.

While Pif1 is a monomer in solution, it can form a stable dimer on DNA30. Dimerization is not required for the helicase to translocate efficiently on ssDNA31 or to unwind substrates under conditions where annealing is limited32. For unwinding forked dsDNA substrates, ScPif1’s helicase activity can be stimulated by the non-translocating strand at the 3′-end3,32. These suggest the possibility that multiple Pif1 molecules bind to the forked dsDNA for efficient unwinding, but how these Pif1 molecules coordinate with each other for unwinding the forked DNA is unclear.

We and others have solved several crystal structures of Pif1 including BaPif125,33, ScPif134, and human Pif125,35. All these structures with/without ATP analog and/or ssDNA have improved our understanding of the helicase activity of Pif1. However, in the absence of dsDNA bound structures of Pif1 helicases, the effect of specific interactions with the duplex on the unwinding mechanism remains poorly understood.

To reveal the structural basis for unwinding dsDNA by Pif1, we determined the structure of BaPif1 in complex with a symmetrical dual forked dsDNA and ADP·AlF4. The structure showed that two Pif1 molecules bind to the 5′ arm and 3′ ss/dsDNA junction, respectively. The structural data combined with mutagenesis together with the kinetics and single-molecule FRET (smFRET) analysis provide key mechanistic insights into forked dsDNA unwinding by BaPif1.

## Results

### Structural overview

Since ScPif1 can form a stable dimer on a forked dsDNA30, we reasoned that a symmetrical dual forked dsDNA would allow multiple Pif1 molecules to bind such a substrate, thereby facilitating crystallization. We crystallized BaPif1 in complex with ADP·AlF4 and a symmetrical dual forked dsDNA containing a 10 nt oligo-dT at both 3′ and 5′ ends (designated as BaPif1–10dT-fdsDNA; Fig. 1a) and determined its structure at a resolution of 3.3 Å. The asymmetric unit of the crystal structure contains two BaPif1 molecules plus one DNA strand, thus the model of the biologically meaningful BaPif1–10dT-fdsDNA complex is generated by the application of a crystallographic two-fold operation. The structure of BaPif1–10dT-fdsDNA showed that two BaPif1 molecules bind to each fork of 10dT-dsDNA and interact with their respective forks in the same manner (Fig. 1b). For simplicity, we only describe the BaPif1–DNA interactions at one fork, which involves molecules A and B of BaPif1. Out of the 10 base-pairs in the double-stranded region of DNA, only eight well-ordered base-pairs could be modeled (Supplementary Fig. 1a). A total of two base-pairs, one on each side of this central double-stranded region are broken. In the single-stranded regions, 7 dT could be modeled at the 5′ end while only 4 dT at the 3′ end were observed. The conformations of two BaPif1 molecules are essentially identical and are similar to those observed in our previous structures of BaPif1 in complex with ssDNA and ADP·AlF425, suggesting that the BaPif1 molecules in these structures undergo similar conformational change upon concomitant binding of ATP and DNA (Supplementary Fig. 1b). The two BaPif1 molecules bound to the junction regions of one fork are related by a rotation of 68° rather than 180°. Therefore, the two BaPif1 molecules bound to the junction regions of a fork cannot be considered as a conventional dimer.

### Interaction of BaPif1 with DNA

As shown in Fig. 2, the binding of BaPif1 to the 5′ arm causes a sharp bend between T10 and C11; consequently, the base of T10 points in the direction opposite to those of other oligo-dTs. The flipped base of T10 is sandwiched between Tyr91A and Tyr412B. It should be noted that the Tyr91A belongs to the BaPif1 bound to the 5′ arm while Tyr412B is contributed by the BaPif1 bound to the 3′ ss/dsDNA junction. A similarly flipped base was observed in the structure of BaPif1-dH but it stacks against only Tyr9125. Importantly, the bending in the 5′ ss/dsDNA junction breaks the first C11–G20 base-pair. The base of unpaired C11 is sandwiched between Ile409B from BaPif1 bound to the 3′ arm and the base of adjacent G12. In the 3′ ss/dsDNA junction; the unpaired G20 is stabilized by residues Val149B and His361B, which stack against the base and ribose ring of G20, respectively. Another notable feature is that three residues His236B, His377B and Phe379B from BaPif1 bound to the 3′ ss/dsDNA are juxtaposed at the helical junction of ss/dsDNA junction. Phe379B stacks against the base of C19 in the second G12–C19 base-pair while His236B and His377B contact the ribose rings of C19 and G12, respectively. Moreover, Lys151B and Lys152B from BaPif1 bound to the 3′arm interact with the phosphate backbone of the double-helical region of DNA, therefore anchoring BaPif1 in close proximity to the 3′ ss/dsDNA junction. Among residues involved in 3′ ss/dsDNA recognition, only V149B, Lys151B, His361B, and Phe379B are conserved across Pif1 family helicases (Supplementary Fig. 2), suggesting that they might be critical for unwinding.

### Interaction of two BaPif1 molecules at ss/dsDNA junctions

At the ss/dsDNA junctions, the interaction of two BaPif1 molecules buries a total solvent accessible surface area of 716 Å2. BaPif1 bound to the 5′ arm uses loops β11–β12, β12–β13 and β14 of the 2B domain to interact with loop α8–α9, helices α9 and α13 of the 2A domain in the BaPif1 bound to the 3′ ss/dsDNA junction predominantly through a network of salt bridge contacts and hydrogen bonds (Fig. 3). Key interactions include salt bridges between Lys321A and Glu210B, Glu323A and Lys405B, and hydrogen bonds between Asn308A and Asp209B, Leu291A and Gln205B, Glu323A and Ser208B. These interactions are further strengthened by Van der Waal’s contacts involving residues Lys292A, Val309A, Asp209B, Pro410B and Tyr412B. Interestingly, the residues involved in BaPif1A–BaPif1B interaction are not well conserved among Pif1 helicases (Supplementary Fig. 2). The weak nature of the interface may allow two BaPif1 molecules bound to the 5′ and 3′ arms translocate in the opposite directions easier, therefore coordinating with each other and regulating their activities.

### Mutational analysis of residues in the ss/dsDNA junction

The structure of BaPif1–dT10-fdsDNA showed that two BaPif1 molecules bind to the ss/dsDNA junctions. As Pif1 is a 5′–3′ helicase and has critical functions in DNA replication, we designed a substrate to determine whether the BaPif1 molecules bound to both arms of the fork are both active helicases, capable of unwinding duplex DNA. This substrate mimics a stalled DNA replication fork, which contains dual duplexes termed parent and leading duplexes (Fig. 4a). This DNA substrate enables us to examine the unwinding of dual duplexes simultaneously, hence the coordination of the activities of the two BaPif1 molecules can be discerned. Wild type BaPif1 unwinds both the forked (parent) duplex and the non-forked (leading) duplex on the 3′-arm of the fork although the amplitudes of product formation are low even at saturating enzyme concentrations (Supplementary Fig. 3). Interestingly, more products are formed from the leading duplex than the parent duplex.

The greater product formation observed with leading duplex may be due to it being only 12 bp whereas the parent duplex contained 16 bp. A substrate was tested which has 16 bp duplexes on each arm of the substrate (Fig. 4b). The quantity of product for the dual duplex with 16 bp was similar to that for the substrate containing the 12 bp leading duplex (Fig. 4c–e) which indicates that the leading duplex is preferentially unwound.

To examine the role of residues contacting the ss/dsDNA junction, we made three single Ala mutants (H236A, H377A, and F379A) and one double mutant (Y91A/Y412A). Neither H236A nor F379A showed unwinding activity for either of the duplexes in a single turnover reaction while both H377A and Y91A/Y412A variants exhibited markedly reduced unwinding activities (Fig. 4c, e). We also measured the binding of H236A and F379A to ssDNA (Supplementary Fig. 4). Both variants still bind DNA so the mutations have not affected protein folding or prevented them from binding to the substrates. However, the affinities are moderately reduced. Hydrolysis of ATP on ssDNA is markedly reduced for the H236A, F379A, and Y91A/Y412A variants (Supplementary Fig. 4), which could explain the low unwinding activity. In addition to interacting with the ss/dsDNA junction, residues His236, His377 and Phe379 in the BaPif1 bound to the 5′ arm interact with 5′ ssDNA (Supplementary Fig. 5) while they stabilize the broken first base-pair and are poised to assist breaking the second base-pair in the BaPif1 bound to the 3′ ss/dsDNA junction (Fig. 2). Mutations of these three residues not only inhibit unwinding of the leading duplex but also affect the unwinding of the parental duplex.

Interestingly, BaPif1B bound on the 3ʹ-ssDNA interacts with the junction similarly to the SF1A DNA helicases PcrA and UvrD that translocate in the opposite direction on ssDNA compared to BaPif1. The stacking interaction of Phe379 against the second base-pair in BaPif1 bound to 3′ ss/dsDNA junction is reminiscent of those observed for Phe626 and Tyr621 in SF1A DNA helicases PcrA36 and UvrD37, respectively (Supplementary Fig. 6). Phe626 of PcrA and Tyr621 of UvrD are located in a Pin region (termed as a wedge for dsDNA unwinding) and stack against the first base-pair in the ss/dsDNA junctions. Structural data combined with mutational analysis suggested that these two residues directly participate in duplex unwinding driven by ATP-coupled 3′-5′ translocation. In contrast, although Phe379 in BaPif1 bound to the 3′ arm also stacks with a base-pair at the ss/dsDNA junction, it may exert its role in unwinding indirectly through assisting the unwinding activity of the BaPif1 bound to the 5′ arm.

The flipped base on the 5′ arm is sandwiched between Tyr91 of BaPif1 bound to the 5′ arm and Tyr412 of BaPif1 bound to the 3′ ss/dsDNA junction (Figs. 2, 3). Together they stabilize the bent ssDNA at 5′ ss/dsDNA junction. As the ssDNA bending caused by BaPif1 bound to the 5′ arm is a key event for breaking the first base-pair, mutation of these two Tyr residues to Ala would be expected to destabilize the bent conformation of substrate DNA, thereby impacting both the ATPase and unwinding activities of BaPif1. Indeed, the Y91A/Y412A variant exhibits reduced rates of ATP hydrolysis (Supplementary Fig. 4). The unwinding activity of the Y91A/Y412A variant is also greatly reduced (Fig. 4c, e), which is not surprising as unwinding is driven by ATP hydrolysis-dependent translocation in a 5′-3′ direction.

### Mutational analysis of the BaPif1–BaPif1 interface

Mutations of residues in the BaPif1A–BaPif1B interface would not be predicted to directly affect the enzymatic activities of BaPif1. Indeed, some of these Pif1 variants have only small changes in activity relative to the wild type enzyme (D209A and D209A/K405A) (Fig. 4d, e). However, the E323A/K405A variant unwinds more DNA in a single turnover reaction than the wild type enzyme for both the parental and the leading duplex and eliminates the preference for unwinding the leading duplex (Fig. 4d, e). The quantities of product formed are equivalent for the two duplexes for E323A/K405A. This may indicate that the two BaPif1 molecules regulate each other and in the absence of interaction between the two BaPif1 molecules, each is able to unwind and is only limited by the inherent processivity of the enzyme.

We also measured the unwinding activity of the E323K variant using the dual duplex substrate (Fig. 4d, e). The E323K variant showed increased product formation in a single turnover reaction relative to the wild type. The activity of the E323K variant is similar to that of the E323A/K405A. These data suggest that the increase in activity observed in the E323A/K405A variant is due to the mutation of E323 more so than due to the mutation of K405. This is also consistent with the reduction in activity of D209A/K405A relative to D209A, which also suggests that a K405A mutation does not increase activity. Gel filtration analysis of wtBaPif1 or E323K complexed with a single forked DNA at a molar ratio of 2:1 (protein:DNA) in the presence of ADP·AlF4 showed that less 2:1 and more 1:1 BaPif1–DNA complexes were formed for E323K than wtBaPif1 (fractions 22–25), indicating that dimerization on DNA is weakened for the E323K variant (Supplementary Fig. 7). In support of this notion, E323A forms a salt bridge with K405B and hydrogen bonds with S208B (Fig. 3). The rates of ATP hydrolysis by both the E323A/K405A and E323K variants are only slightly increased relative to the wild type enzyme (Supplementary Fig. 4), The small increase in ATP hydrolysis is not sufficient to explain the 2-fold increase in unwinding of the leading duplex and 3-fold increase in unwinding of the parental duplex by these variants.

### Two BaPif1 molecules bound to the fork are dually active

Wild type BaPif1 unwinds the parental and leading duplexes with similar rate constants (Fig. 6). Mutation of E323 and K405 increases both the unwinding rate constants and the quantity of product formed on the parental and leading duplexes (Fig. 6). Importantly, the combined amplitudes for the fraction of unwound product equals 1.2 for this BaPif1 variant. Thus, each molecule of helicase unwinds a duplex simultaneously. The increase in activity upon uncoupling the two BaPif1 subunits suggests that the interaction of two BaPif1 molecules may serve a regulatory role whereby each of the BaPif1 subunits clears the DNA on one strand of the fork but the linkage of the enzymes constrains their activity so that only a small region of the DNA is cleared. This is consistent with the known role of Pif1 in DSB repair7,38, regulation of telomerase8, and Okazaki fragment processing11,12.

Our structure of BaPif1 in complex with 10dT-fdsDNA and unwinding assays showed that two BaPif1 molecules bind to and are active on both arms of the forked dsDNA. Since ScPif1, human Pif1 and BaPif1 are structurally similar and all are capable of unwinding the forked DNA substrates, it is envisaged that two molecules of both ScPif1 and hPif1 would bind and unwind the same substrate we used for BaPif1. Consistent with this notion, unwinding assays (Supplementary Fig. 9) showed that under single turnover conditions, ScPif1 unwinds both duplexes but with strong preference for the forked parental duplex (Supplementary Fig. 9a–c). Addition of a short tail to the displaced strand of the leading duplex allowed ScPif1 to unwind both duplexes similarly to BaPif1 (Supplementary Fig. 9d–f), consistent with ScPif1’s preference for unwinding forked duplexes3,29,32. hPif1 produced no product with either substrate in a single turnover reaction; however, multi-turnover experiments demonstrated that hPif1 unwinds both duplexes (Supplementary Fig. 9g–i). The different unwinding efficiencies and preferences for substrate duplexes for ScPif1, hPif1, and BaPif1 may be attributed to the different duplexes to be unwound by them in their respective species. Further structural evidence of ScPif1 and hPif1 in complex with forked dsDNA are required to prove this view.

### Mechanism of forked dsDNA unwinding by BaPif1

For unwinding the 5′ tailed dsDNA, previously we proposed based on the structure of BaPif1-dH that the bending at the 5′ ss/dsDNA junction caused by the simultaneous binding of ADP·AlF4 and a 5′ tailed dsDNA would break the first few base pairs and the rest of a short dsDNA region is likely to be unwound by thermal fraying25. In support of this model, our current structure shows that the binding of BaPif1 to the 5′ arm causes a similar bending and consequently breaks the first base-pair. Based on our structural observations and available biochemical data, we propose a mechanism for unwinding a forked dsDNA by BaPif1 helicase (Fig. 7). Two BaPif1 molecules synergistically bind to both arms of the fork. The BaPif1 molecule bound to the 5′ arm causes the breaking of the first base-pair through bending the tracking strand. The second BaPif1 molecule bound to the 3′ arm interacts extensively with the ss/dsDNA junction such that it not only stabilizes the first broken base-pair but also is poised to assist breaking the second base-pair. The binding of BaPif1 to the 3’ arm may also prevent re-winding the separated strands. Further unwinding of the duplex is carried out by ATPase coupled translocation of the molecule at the 5′ arm. Given that the BaPif1 bound to the 3′ ss/dsDNA junction can only translocate in a 5′–3′ direction, the model we propose herein suggests that two BaPif1 molecules bound to the fork, initiate unwinding, after which each BaPif1 monomer can function as an active helicase. However, the protein dimer ensures only short stretches of duplex are unwound, keeping with many of the reported functions of Pif1 family enzymes. Consistent with this model, biochemical studies showed that the 3′ ssDNA tail stabilizes a dimer formed on a forked dsDNA30 and stimulates the unwinding activity but hinders the annealing activity of ScPif13,32,39. Moreover, our unwinding assays showed that the two BaPif1 helicases bound to the fork are dually active (Figs. 4, 6 and Supplementary Fig. 3). Interestingly, reduction of the interactions between the monomers results in increased unwinding processivity suggesting that the protein-protein interactions may serve a regulatory mechanism whereby each BaPif1 molecule influences the activity of the other. When the protein-protein interactions are reduced, the ability of each monomer to translocate forward unhindered is increased, which could result in more processive unwinding by the E323 variants.

To date, the structures of only monomeric forms of SF1 and SF2 helicases in complex with a dsDNA have been reported36,37 although a dimeric UvrD has been proposed to be required for processive unwinding of dsDNA based on single turnover kinetic data40. In the proposed dimeric UvrD–DNA complex, one monomer binds to the 3′ ss/dsDNA junction while the second monomer is bound to the 3′ ssDNA tail, and two interacting UvrD molecules form a stable initiation complex. Moreover, UvrD can self-assemble to form dimers and tetramers in the absence of DNA whereas ScPif1 without bound DNA is in a monomeric form in solution30. However, a dimeric UvrD with or without bound DNA has not been crystallized. Therefore, our structure of BaPif1–dT10-fdsDNA represents the only one showing two interacting SF1 helicases that are bound at forked dsDNA, with one monomer bound to the 3’-arm and the other to the 5’-arm, yet both are active on their respective strands.

## Discussion

In eukaryotes, the progression and integrity of DNA replication fork (RF) are tightly regulated to maintain genomic stability and for coupling DNA synthesis with other processes. Many physical impediments such as protein–DNA complexes, replication fork barriers (RFB) and DNA damage can hinder the DNA replication fork progression41. ScPif1 and Rrm3 in Saccharomyces cerevisiae have been shown to function in DNA replication15,42,43. Most recently, Deegan et al.44 showed that both ScPif1 and Rrm3 stimulate fork convergence by helping to unwind the final stretch of parental DNA during DNA replication termination44. If BaPif1 functions similarly to ScPif1, then two helicase molecules at the fork should help clear both strands for replication. Consistent with the findings that ScPif1 and Rrm3 have overlapping functions at replication forks, the C-terminal helicase domains of ScPif1 and Rrm3 are well conserved. It has been proposed that Rrm3 might move with the replication fork, enhancing the activity of the replicative helicase to enable it to bypass protein–DNA complexes45 while ScPif1 and Rrm3 may help to overcome geometrical strain at converging forks to unwind the final stretch of parental dsDNA44. Our structure showed that one BaPif1 binds to the 3′ arm, corresponding to the leading strand template of a fork and the other BaPif1 binds to the 5′ arm, corresponding to the lagging strand template of a fork. The BaPif1 bound to the leading strand template, on one hand, would displace the protein–DNA complexes by its translocase activity; on the other hand, it would enhance parent duplex unwinding by assisting the BaPif1 bound to the lagging strand template. It is envisaged that the coordinating actions of two BaPif1 molecules bound to the replication fork would enable efficient fork progression and convergence. Although ScPif1, Rrm3, and hPif1share low sequence similarity with BaPif1, the structural and biochemical studies on ScPif1 and BaPif1 indicated that they are functionally similar. Therefore, the functional insights we gained from BaPif1 in DNA replication are likely applicable to other Pif1 family helicases.

Many reported activities for Pif1 do not require unwinding of long stretches of duplex DNA. In addition to helping initiate duplex unwinding, the protein interactions of two Pif1 molecules at a DNA fork could serve a regulatory role by influencing processivity. The removal of secondary structures and bound proteins by Pif1 could be limited to the local region of the replication fork. Pif1 dependent removal of telomerase6,7,8 and generation of long flaps during Okazaki fragment maturation11,12 do not require unwinding long stretches of dsDNA. Disruption of the protein-protein interactions observed here between Pif1 molecules (as seen in Fig. 5) increases processivity, which is consistent with the idea that each monomer regulates the activity of the other.

In conclusion, our structure of BaPif1 bound to a partially unwound forked dsDNA shows that two BaPif1 molecules bind to and unwind the forked dsDNA. The structure revealed several important features that distinguish BaPif1 from any other known SF1 and SF2 helicases. First, the BaPif1 molecule bound to the 5′ arm initiates the forked DNA unwinding by breaking the first base-pair through geometrical strain while the BaPif1 molecule bound to the 3′ arm plays an accessory role in unwinding through stabilizing the first broken base-pair and engaging the second base-pair in a pre-breaking state. Second, the BaPif1 molecule bound to the 3′ arm is positioned to prevent re-winding of the displaced strand with the tracking strand. Third, the two BaPif1 molecules contact each other through electrostatic interactions, and the nature of such interactions may allow two BaPif1 molecules regulate their unwinding activities. These observations together with kinetics and smFRET data enabled us to propose a mechanism by which BaPif1 unwinds a forked dsDNA. These results not only explain the previous biochemical observations but also provided a framework for further elucidating the mechanism by which Pif1 helicases unwind DNA/RNA hybrids and G4 DNAs.

## Methods

### Protein purification and crystallization

Pif1 from Bacterioides fragilis (BaPif1) and the mutants of BaPif1: D209A, H236A, F379A, H377A, E323K, Y91A/Y412A, E323A/K405A, and D209A/K405A were prepared, expressed and purified using the protocol described in Zhou et al.25. A self-complementary oligonucleotide (sequence: 5′-TTTTTTTTTTCGCGCGCGCGTTTTTTTTTT-3′) was annealed to generate a symmetrical dual forked dsDNA (referred to as dT10-fdsDNA). Purified wild type BaPif1 was incubated with the dT10-fdsDNA at a molar ratio of 4:1 in the presence of ATP transition-state analogue ADP·AlF4. The resulting BaPif1–dT10-fdsDNA complex was purified to homogeneity using size exclusion chromatography and then concentrated to approximately 12 mg ml−1. Crystallization screening for the BaPif1–dT10-fdsDNA was performed at 16 °C using the sitting drop vapor diffusion method. Crystals of BaPif1–dT10-fdsDNA were obtained in a condition containing 80 mM NaCl, 40 mM sodium cacodylate pH 6.0, 35% MPD, and 12 mM spermine.

### Structure determination of BaPif1–dT10-fdsDNA

X-ray diffraction data were collected at Shanghai Synchrotron Radiation Facility. The crystal structure of BaPif1–dT10-fdsDNA was determined with successive rounds of molecular replacement with PHASER46 using 1A–2A and 2B domains of BaPif1 (PDB id: 2FHD). The final model was built to acceptable stereo-chemical values with iterative model building and refinement cycles using CCP447 and Phenix48. The difference (Fo−Fc) map was utilized to identify and build DNA, and ADP·AlF4 and Mg2+ were modeled based on PDB id: 5FHD. Data collection and refinement statistics are given in Table 1. A representative portion of electron density map is shown in Supplementary Fig. 10.

### Oligonucleotides for assays

Oligonucleotides listed in Supplementary Table 1 were purchased from Integrated DNA Technologies and gel purified by denaturing 20% PAGE, visualized by UV shadowing to excise the band, and electroeluted from the acrylamide using an Elutrap (Whatman). Oligonucleotides were desalted by loading onto on a C18 SepPak (Waters) which had been pre-equilibrated by washing with acetonitrile, H2O, and 10 mM NH4OAc. SepPaks were washed with H2O, and oligonucleotides were eluted with 60% methanol and dried in a Speed Vac. After re-suspending samples in 10 mM Hepes pH 7.5, 1 mM EDTA, the concentrations were determined using extinction coefficients at 260 nm. Cy5-labeled oligonucleotides which were purchased HPLC and were not gel purified. Duplexes were formed by mixing 5 µM of each strand in 10 mM Hepes pH 7.5, 1 mM EDTA, heating to 95 °C for 5 minutes, and slowly cooling to room temperature.

### DNA unwinding

BaPif1 (2 μM) was pre-incubated with 20 nM fluorescently labeled dual duplex DNA in 10 mM Tris, pH 7.5, 50 mM NaCl, 0.1 mM EDTA, 2 mM DTT, 0.1 mg ml−1 BSA, 5% glycerol for 5 min. Reactions were initiated by addition of ATP to 5 mM, MgCl2 to 10 mM, and 400 nM of annealing trap complementary to the displaced strands at 25 °C. For single turnover reactions, dT50 (100 μM) was added with the ATP to prevent re-association of the enzyme with the substrate after dissociation. Reactions were quenched by mixing with an equal quantity of 100 mM EDTA, 0.3% SDS, 1 μM dT50, 5% glycerol, 0.05% orange G and separated by 20% native PAGE. Some reactions were performed using a rapid chemical quench flow instrument. For these experiments, reactions were quenched with 400 mM EDTA and loading buffer was added to 5% glycerol, 0.05% orange G. Gels were visualized with a Typhoon Trio using a 488 nm laser for excitation and a 520 BP 40 filter for emission of FAM and a 633 nm laser for excitation and a 670 BP 30 filter for emission of Cy5. The blue and red channels of the gels were quantified separately using ImageQuant Software and the fraction product for each duplex was determined using Equation 1.

$${\mathrm{Fraction}}\,{\mathrm{product}} = \frac{{\frac{{P_t}}{{S_t + I_t + P_t}} - \frac{{P_0}}{{S_0 + I_0 + P_0}}}}{{1 - \frac{{P_0}}{{S_0 + I_0 + P_0}}}}$$
(1)

Pt is the fraction product at a time point, P0 is the fraction product at the initiation of the reaction. St is the fraction substrate at a time point, S0 is the fraction substrate at the initiation of the reaction. It is the fraction intermediate with the other duplex unwound at a time point, I0 is the fraction intermediate with the other duplex unwound at the initiation of the reaction. Data were fit to a 5-step sequential mechanism using KinTek Explorer software.

### smFRET experiments

Oligonucleotides

(5′-/5Biotin/ CCAGGCGACATCAGCGTTTTTTTGCAGTGACCAGACAGG-3′;

5′-CCTGTCTGGTCACTGC/3Cy3/-3′; and

5′-/5Cy5/TTTTTTTTTTCGCTGATGTCGCCTGG-3′) were purchased from Integrated DNA Technologies and annealed by mixing 10 µM of the biotin strand and 11 µM of the Cy3 and Cy5 strands in 10 mM Tris-HCl pH 7.5, 50 mM NaCl, and 150 mM KCl. The mixtures were incubated at 95 °C for 2 min then slowly cooled to 37 °C at a rate of 2 °C per minute, and then cooled to room temperature at a rate of 5 °C per minute. Single-molecule FRET flow experiments were performed on a prism-type total internal reflection fluorescence (TIRF) microscope at room temperature (23 ± 1 °C). The microscope, flow chamber preparation, and DNA substrate immobilization were done as described49. The unwinding reaction buffer consisted of 20 mM Tris-HCl pH 7.5, 50 mM NaCl, 5 mM MgCl2, 2 mM DTT, and 2% glycerol. For photobleaching and blinking reduction during imaging, buffers also contained 0.8% (w/v) dextrose, 165 U/mL glucose oxidase, 2170 U/mL catalase, and 2–3 mM Trolox. 100 nM Pif1 and 250 µM ATP, or 25 nM Pif1 and 6.25 µM ATP, were flowed into the sample chamber simultaneously and imaged at an exposure time of 200 ms. 532 nm and 633 nm laser excitations were alternating for every frame during the course of a 100 s movie. Single-molecule trajectories were generated by IDL particle tracker and analyzed by MATLAB (R2015b) scripts to determine the unwinding time for parent and leading duplexes.

### Reporting summary

Further information on research design is available in the Nature Research Reporting Summary linked to this article.

## Data availability

The atomic coordinates and structural factors for the BaPif1–10dT-fdsDNA complex have been deposited with the Protein Data Bank under accession code 6L3G. The source data underlying Figs. 4c–e, 6a–d and Supplementary Figs. 3b–d, 4a, b, 7b, c, 9b, c, f, h, i are provided as a Source Data file. All data supporting the findings of this study are available from the corresponding authors on reasonable request.

## References

1. Singleton, M. R., Dillingham, M. S. & Wigley, D. B. Structure and mechanism of helicases and nucleic acid translocases. Annu Rev. Biochem 76, 23–50 (2007).

2. Lohman, T. M., Tomko, E. J. & Wu, C. G. Non-hexameric DNA helicases and translocases: mechanisms and regulation. Nat. Rev. Mol. Cell Biol. 9, 391–401 (2008).

3. Lahaye, A., Leterme, S. & Foury, F. PIF1 DNA helicase from Saccharomyces cerevisiae. Biochemical characterization of the enzyme. J. Biol. Chem. 268, 26155–26161 (1993).

4. Chung, W. H. To peep into Pif1 helicase: multifaceted all the way from genome stability to repair-associated DNA synthesis. J. Microbiol 52, 89–98 (2014).

5. Byrd, A. K. & Raney, K. D. Structure and function of Pif1 helicase. Biochem Soc. Trans. 45, 1159–1171 (2017).

6. Boule, J. B., Vega, L. R. & Zakian, V. A. The yeast Pif1p helicase removes telomerase from telomeric DNA. Nature 438, 57–61 (2005).

7. Schulz, V. P. & Zakian, V. A. The saccharomyces PIF1 DNA helicase inhibits telomere elongation and de novo telomere formation. Cell 76, 145–155 (1994).

8. Zhou, J., Monson, E. K., Teng, S. C., Schulz, V. P. & Zakian, V. A. Pif1p helicase, a catalytic inhibitor of telomerase in yeast. Science 289, 771–774 (2000).

9. Myung, K., Chen, C. & Kolodner, R. D. Multiple pathways cooperate in the suppression of genome instability in Saccharomyces cerevisiae. Nature 411, 1073–1076 (2001).

10. Pike, J. E., Henry, R. A., Burgers, P. M., Campbell, J. L. & Bambara, R. A. An alternative pathway for Okazaki fragment processing: resolution of fold-back flaps by Pif1 helicase. J. Biol. Chem. 285, 41712–41723 (2010).

11. Pike, J. E., Burgers, P. M., Campbell, J. L. & Bambara, R. A. Pif1 helicase lengthens some Okazaki fragment flaps necessitating Dna2 nuclease/helicase action in the two-nuclease processing pathway. J. Biol. Chem. 284, 25170–25180 (2009).

12. Budd, M. E., Reis, C. C., Smith, S., Myung, K. & Campbell, J. L. Evidence suggesting that Pif1 helicase functions in DNA replication with the Dna2 helicase/nuclease and DNA polymerase delta. Mol. Cell Biol. 26, 2490–2500 (2006).

13. Rossi, M. L. et al. Pif1 helicase directs eukaryotic Okazaki fragments toward the two-nuclease cleavage pathway for primer removal. J. Biol. Chem. 283, 27483–27493 (2008).

14. Tran, P. L. T. et al. PIF1 family DNA helicases suppress R-loop mediated genome instability at tRNA genes. Nat. Commun. 8, 15025 (2017).

15. Osmundson, J. S., Kumar, J., Yeung, R. & Smith, D. J. Pif1-family helicases cooperatively suppress widespread replication-fork arrest at tRNA genes. Nat. Struct. Mol. Biol. 24, 162–170 (2017).

16. Paeschke, K., Capra, J. A. & Zakian, V. A. DNA replication through G-quadruplex motifs is promoted by the Saccharomyces cerevisiae Pif1 DNA helicase. Cell 145, 678–691 (2011).

17. Paeschke, K. et al. Pif1 family helicases suppress genome instability at G-quadruplex motifs. Nature 497, 458–462 (2013).

18. Lopes, J. et al. G-quadruplex-induced instability during leading-strand replication. EMBO J. 30, 4033–4046 (2011).

19. Wilson, M. A. et al. Pif1 helicase and Poldelta promote recombination-coupled DNA synthesis via bubble migration. Nature 502, 393–396 (2013).

20. Saini, N. et al. Migrating bubble during break-induced replication drives conservative DNA synthesis. Nature 502, 389–392 (2013).

21. Bochman, M. L., Sabouri, N. & Zakian, V. A. Unwinding the functions of the Pif1 family helicases. DNA Repair (Amst.) 9, 237–249 (2010).

22. Bochman, M. L., Judge, C. P. & Zakian, V. A. The Pif1 family in prokaryotes: what are our helicases doing in your bacteria? Mol. Biol. Cell 22, 1955–1959 (2011).

23. Geronimo, C. L., Singh, S. P., Galletto, R. & Zakian, V. A. The signature motif of the Saccharomyces cerevisiae Pif1 DNA helicase is essential in vivo for mitochondrial and nuclear functions and in vitro for ATPase activity. Nucleic Acids Res. 46, 8357–8370 (2018).

24. Mohammad, J. B., Wallgren, M. & Sabouri, N. The Pif1 signature motif of Pfh1 is necessary for both protein displacement and helicase unwinding activities, but is dispensable for strand-annealing activity. Nucleic Acids Res 46, 8516–8531 (2018).

25. Zhou, X. et al. Structural and functional insights into the unwinding mechanism of bacteroides sp Pif1. Cell Rep. 14, 2030–2039 (2016).

26. Zhou, R., Zhang, J., Bochman, M. L., Zakian, V. A. & Ha, T. Periodic DNA patrolling underlies diverse functions of Pif1 on R-loops and G-rich DNA. Elife 3, e02190 (2014).

27. Boule, J. B. & Zakian, V. A. The yeast Pif1p DNA helicase preferentially unwinds RNA DNA substrates. Nucleic Acids Res 35, 5809–5818 (2007).

28. Chib, S., Byrd, A. K. & Raney, K. D. Yeast helicase Pif1 unwinds RNA:DNA hybrids with higher processivity than DNA:DNA duplexes. J. Biol. Chem. 291, 5889–5901 (2016).

29. Ramanagoudr-Bhojappa, R. et al. Yeast Pif1 helicase exhibits a one-base-pair stepping mechanism for unwinding duplex DNA. J. Biol. Chem. 288, 16185–16195 (2013).

30. Barranco-Medina, S. & Galletto, R. DNA binding induces dimerization of Saccharomyces cerevisiae Pif1. Biochemistry 49, 8445–8454 (2010).

31. Galletto, R. & Tomko, E. J. Translocation of Saccharomyces cerevisiae Pif1 helicase monomers on single-stranded DNA. Nucleic Acids Res 41, 4613–4627 (2013).

32. Singh, S. P., Koc, K. N., Stodola, J. L. & Galletto, R. A monomer of Pif1 unwinds double-stranded DNA and It Is Regulated by the Nature of the Non-Translocating Strand at the 3’-End. J. Mol. Biol. 428, 1053–1067 (2016).

33. Liu, N. N. et al. The Bacteroides sp. 3_1_23 Pif1 protein is a multifunctional helicase. Nucleic Acids Res. 43, 8942–8954 (2015).

34. Lu, K. Y. et al. Insights into the structural and mechanistic basis of multifunctional S. cerevisiae Pif1p helicase. Nucleic Acids Res. 46, 1486–1500 (2018).

35. Dehghani-Tafti, S., Levdikov, V., Antson, A. A., Bax, B. & Sanders, C. M. Structural and functional analysis of the nucleotide and DNA binding activities of the human PIF1 helicase. Nucleic Acids Res. 47, 3208–3222 (2019).

36. Velankar, S. S., Soultanas, P., Dillingham, M. S., Subramanya, H. S. & Wigley, D. B. Crystal structures of complexes of PcrA DNA helicase with a DNA substrate indicate an inchworm mechanism. Cell 97, 75–84 (1999).

37. Lee, J. Y. & Yang, W. UvrD helicase unwinds DNA one base pair at a time by a two-part power stroke. Cell 127, 1349–1360 (2006).

38. Jimeno, S. et al. The helicase PIF1 facilitates resection over sequences prone to forming G4 structures. Cell Rep. 24, 3262–3273 e4 (2018).

39. Ramanagoudr-Bhojappa, R., Byrd, A. K., Dahl, C. & Raney, K. D. Yeast Pif1 accelerates annealing of complementary DNA strands. Biochemistry 53, 7659–7669 (2014).

40. Maluf, N. K., Fischer, C. J. & Lohman, T. M. A Dimer of Escherichia coli UvrD is the active form of the helicase in vitro. J. Mol. Biol. 325, 913–935 (2003).

41. Branzei, D. & Foiani, M. Maintaining genome stability at the replication fork. Nat. Rev. Mol. Cell Biol. 11, 208–219 (2010).

42. Ivessa, A. S., Zhou, J. Q. & Zakian, V. A. The Saccharomyces Pif1p DNA helicase and the highly related Rrm3p have opposite effects on replication fork progression in ribosomal DNA. Cell 100, 479–489 (2000).

43. Calzada, A., Hodgson, B., Kanemaki, M., Bueno, A. & Labib, K. Molecular anatomy and regulation of a stable replisome at a paused eukaryotic DNA replication fork. Genes Dev. 19, 1905–1919 (2005).

44. Deegan, T. D., Baxter, J., Ortiz Bazan, M. A., Yeeles, J. T. P. & Labib, K. P. M. Pif1-family helicases support Fork convergence during DNA replication termination in Eukaryotes. Mol. Cell 74, 231–244 e9 (2019).

45. Ivessa, A. S. et al. The Saccharomyces cerevisiae helicase Rrm3p facilitates replication past nonhistone protein-DNA complexes. Mol. Cell 12, 1525–1536 (2003).

46. McCoy, A. J. et al. Phaser crystallographic software. J. Appl Crystallogr 40, 658–674 (2007).

47. Winn, M. D. et al. Overview of the CCP4 suite and current developments. Acta Crystallogr. D. Biol. Crystallogr. 67, 235–242 (2011).

48. Adams, P. D. et al. PHENIX: a comprehensive Python-based system for macromolecular structure solution. Acta Crystallogr. D. Biol. Crystallogr. 66, 213–221 (2010).

49. Yang, O. & Ha, T. Single-molecule studies of ssDNA-binding proteins exchange. Methods Enzymol. 600, 463–477 (2018).

## Acknowledgements

We would like to thank the beamline scientists at BL17U of Shanghai Synchrotron Radiation Facility in China for assistance and access to synchrotron radiation facilities. This work was supported by Natural Science Foundation of China (Grant No: 31670820) and the Agency for Science, Technology and Research in Singapore and the National Institutes of Health (R35GM122601).

## Author information

Authors

### Contributions

H.S., K.D.R., and T.H. conceived and coordinated the study. N.S., A.K.B., S.R.B., O.Y., Y.J., and X.T. performed the experiments. N.S., A.K.B., S.R.B., O.Y., T.H., K.D.R., and H.S. analyzed the data and wrote the paper. S.R.B. and O.Y. contributed equally to this work.

### Corresponding authors

Correspondence to Taekjip Ha, Kevin D. Raney or Haiwei Song.

## Ethics declarations

### Competing interests

The authors declare no competing interests.

Peer review information Nature Communications thanks the anonymous reviewer(s) for their contribution to the peer review of this work. Peer reviewer reports are available.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## Rights and permissions

Reprints and Permissions

Su, N., Byrd, A.K., Bharath, S.R. et al. Structural basis for DNA unwinding at forked dsDNA by two coordinating Pif1 helicases. Nat Commun 10, 5375 (2019). https://doi.org/10.1038/s41467-019-13414-9

• Accepted:

• Published:

• DOI: https://doi.org/10.1038/s41467-019-13414-9

• ### G4-quadruplex-binding proteins: review and insights into selectivity

• Vanessa Meier-Stephenson

Biophysical Reviews (2022)

• ### Dynamic regulation of Pif1 acetylation is crucial to the maintenance of genome stability

• Onyekachi E. Ononye
• Christopher W. Sausen
• Lata Balakrishnan

Current Genetics (2021)