5′ to 3′ Unfolding Directionality of DNA Secondary Structures by Replication Protein A

The replication protein A (RPA) is a single-stranded DNA-binding protein that plays an essential role in DNA metabolism. RPA is able to unfold G-quadruplex (G4) structures formed by telomeric DNA sequences, a function important for telomere maintenance. To elucidate the mechanism through which RPA unfolds telomeric G4s, we studied its interaction with oligonucleotides that adopt a G4 structure extended with a single-stranded tail on either side of the G4. Binding and unfolding was characterized using several biochemical and biophysical approaches and in the presence of specific G4 ligands, such as telomestatin and 360A. Our data show that RPA can bind on each side of the G4 but it unwinds the G4 only from 5′ toward 3′. We explain the 5′ to 3′ unfolding directionality in terms of the 5′ to 3′ oriented laying out of hRPA subunits along single-stranded DNA. Furthermore, we demonstrate by kinetics experiments that RPA proceeds with the same directionality for duplex unfolding.

The replication protein A (RPA) is a single-stranded DNAbinding protein that plays an essential role in DNA metabolism. RPA is able to unfold G-quadruplex (G4) structures formed by telomeric DNA sequences, a function important for telomere maintenance. To elucidate the mechanism through which RPA unfolds telomeric G4s, we studied its interaction with oligonucleotides that adopt a G4 structure extended with a singlestranded tail on either side of the G4. Binding and unfolding was characterized using several biochemical and biophysical approaches and in the presence of specific G4 ligands, such as telomestatin and 360A. Our data show that RPA can bind on each side of the G4 but it unwinds the G4 only from 5 toward 3. We explain the 5 to 3 unfolding directionality in terms of the 5 to 3 oriented laying out of hRPA subunits along single-stranded DNA. Furthermore, we demonstrate by kinetics experiments that RPA proceeds with the same directionality for duplex unfolding.
Human telomeres are composed of tandem repeats of the sequence 5Ј-TTAGGG-3Ј and bear a 3Ј single-stranded extension (also known as 3Ј G-overhang). This G-rich telomeric single-stranded extension can adopt non-canonical DNA conformations known as G-quadruplexes (G4s) 5 (1). These fourstranded structures are based on guanine quartets stabilized by Hoogsteen hydrogen bonds; these structures are stabilized by cations in the central cavity. The human telomeric sequence can form intramolecular G4 of different conformations (2). There is evidence that G4 forms in cells at telomeres during lagging strand DNA replication and at the 3Ј G-overhang (3)(4)(5)(6). Several proteins interact with telomeric G4s to regulate telomerase activity and maintain telomere integrity (7). Among them, the replication protein A (RPA) has been shown to have a key role in telomere maintenance and telomerase action (8 -14).
RPA is a highly conserved protein in eukaryotes (15,16); it is involved in essential processes such as replication, recombination, and DNA repair (17). RPA is a heterotrimeric protein composed of RPA1, RPA2, and RPA3 subunits. RPA carries six DNA-binding domains (DBD), four of them are located in RPA1 (DBD-A, DBD-B, DBD-C and DBD-F), one is located in RPA2 (DBD-D), and one in RPA3 (DBD-E). RPA binds to unstructured, single-stranded DNA (ssDNA) through three different binding modes involving five of the six DBD domains: DBDs A, B, C, D, and E (17)(18)(19). It is now accepted that RPA binds to ssDNA in a sequential pathway with a defined polarity (20,21). First, the high-affinity DBD-A and DBD-B domains of RPA1 bind to 8 -10 nucleotides (nt) (22); this initial binding is designated as the "compact" conformation or 8 -10-nt binding mode. This step is followed by weaker interactions of DBD-C of RPA1 with the 3Ј side of the ssDNA, leading to an intermediate named "elongated contracted" conformation or 13-22-nt binding mode (18,23,24). Finally, binding of DBD-D of RPA2 and DBD-E of RPA3 to the 3Ј side of the ssDNA leads to a stable "elongated extended" complex in which RPA covers 30 nt (30-nt binding mode); this is the most stable RPA-ssDNA complex (18). RPA binds ssDNA in a non-sequence-specific manner (15); however, it has a higher affinity for pyrimidine sequences than for other sequence compositions. Recently, Prakash et al. (25) showed, using a SELEX experiment, that DBD-A and DBD-B preferentially bind pyrimidine-rich sequences, DBD-C has no sequence preference, and DBD-D and DBD-E bind G-rich sequences.
In 2006, RPA was shown to unfold telomeric G4 structures in vitro (26). Since then, many studies have focused on the interaction between RPA and G4 to determine its unfolding mechanism. Electrophoretic mobility shift assays (EMSA) and fluorescence resonance energy transfer (FRET) experiments indicate that the initial binding of the human RPA (hRPA) requires a ssDNA region (27), probably generated by G4 structural breathing that transiently exposes ssDNA (28 -30), and that this initial step is rate-limiting (31,32). Second, an inverse relationship between telomeric G4 structural stability and hRPA binding has been established (27,33). Third, the hRPA binding mode (1:1 complexes where one RPA is bound on the DNA in an elongated mode, or 2:1 complexes where two RPA are bound on the DNA in a compact mode) depends on hRPA concentration (18,27). Fourth, cross-linking experiments showed that the RPA1 and RPA2 subunits of the hRPA directly contact the G4 sequence (27). These data were recently confirmed by the characterization of the RPA1-D228Y mutation that impaired both the binding and the opening of the telomeric G4 (34). In addition, it has been shown that hRPA binds G4 DNA in a 5Ј to 3Ј spatial manner (27) in agreement with its polarity of binding to unstructured ssDNA (18,21,35) and to a non-telomeric G4 (25). Based on these data, we have suggested a sequential hRPA-mediated G4 unfolding mechanism where hRPA unfolds G4 with a 5Ј to 3Ј directionality, however, experimental evidence of such a directionality is still missing (27). Here, we have performed a set of experiments combining the use of a telomeric G4 sequence extended by a ss-tail on either side of the G4 sequence and the use of G4-ligands, to provide experimental evidence that hRPA unfolds G4 according to a 5Ј to 3Ј directionality and to unravel the reasons of this directionality. Tail Attached to the Telomeric G4 Enhances hRPA Binding-We investigated, by EMSA, how a single-stranded tail attached to either end of the htelo G4 sequence impacted the efficiency of binding of the G4 structure by hRPA. We designed four modified versions of the htelo G4 (Table 1): 1) 5t-htelo with a single-stranded tail of five thymidines on the 5Ј side of the G4; 2) htelo-5t with a single-stranded tail of five thymidines on the 3Ј end; 3) 10t-htelo with a single-stranded tail of 10 thymidines on the 5Ј end; and 4) htelo-10t with a singlestranded tail of 10 thymidines on the 3Ј side. The CD spectra of these extended htelo sequences did not strongly differ from the spectrum of the original htelo (data not shown), suggesting that the single-stranded tails did not affect the G4 conformation. The presence of a single-stranded tail decreased the melting temperature (T m ) of the G4 structure (Table 1). Compared with the htelo oligonucleotide, a single-stranded tail at the 5Ј end decreased the T m by about 9°C and a single-stranded tail at the 3Ј end decreased the T m by about 7°C; the extent of destabilization was nearly independent of the length of the tail.

EMSA Experiments Show That the Presence of a Singlestranded
hRPA binding to these extended htelo sequences was studied by EMSA on a native agarose gel (Fig. 1A). Slow migrating species corresponding to hRPA-DNA complexes appeared for htelo and extended htelo DNAs as the protein concentration increased. Quantification of the EMSA clearly revealed that the presence of a single-stranded tail increased the affinity of hRPA for the G4 oligonucleotide (Fig. 1, B and C) and that hRPA affinity was proportional to the length of the tail. Indeed, the concentration of hRPA required to complex 50% of the DNA was 60 nM for htelo, 25 nM for 5t-htelo and htelo-5t, and 18 nM for 10t-htelo and htelo-10t. This indicates that a singlestranded region of 5 or 10 nt on either side of the G4 formed by htelo enhances binding of hRPA.
FRET Experiments Show That hRPA Efficiently Unfolds the Telomeric G4 Structures with Single-stranded Tails-To determine whether binding of hRPA to the extended 10t-htelo and htelo-10t oligonucleotides resulted in unfolding of the G4 structure, we carried out FRET experiments with these extended oligonucleotides labeled with a 6-carboxyfluorescein (FAM) and a tetramethylrhodamine (TAMRA) at the 5Ј and 3Ј end of the G4 region, respectively (Table 1). Melting experiments showed that both extended labeled G4s have the same stability (curves not shown, T m reported in Table 1). Fluorimetric titrations of 10t-FhteloT and FhteloT-10t with hRPA were performed, and data were quantified to determine the energy transfer efficiency (P, defined under "Experimental Procedures") as a function of hRPA/DNA ratio (r). With both oligonucleotides, the p value increased as the ratio r increased (Fig.  1D) indicating that FRET was suppressed by the addition of protein to DNA. The maximal p value was nearly the same as the one previously obtained for the non-extended FhteloT (27). These results demonstrate that hRPA completely unfolds the G4 structure of the extended htelo oligonucleotides, irrespective of whether the single-stranded region is at the 5Ј or 3Ј side of the G4.
Cross-linking Experiments Show That hRPA Binds the Extended G4 Structures via Its RPA1 and RPA2 Subunits-To determine the positioning of hRPA along the extended telomeric G4 sequence, a photocross-linking strategy based on thionucleobases was applied (18, 36 -38). Eight different extended oligonucleotide analogues containing a single photoactivable 4-thiothymine residue (s 4 T), at the extremity of the single-stranded extension, at the 5Ј side of the G4, in the middle of the G4, and at the 3Ј side of the G4, were used (Table 1). Upon UV irradiation of the hRPA-DNA complex, the s 4 T residue can form a covalent bond linking the DNA with the protein.  Photocross-linked complexes obtained in the presence of 20 nM hRPA and 2 nM htelo-S DNA were resolved by SDS-PAGE. At these concentrations, the major stoichiometry of the hRPAhtelo complex was 1:1 (26,27). No cross-links were formed without UV irradiation (data not shown). Upon UV irradiation, in addition to the free radioactively labeled DNA band, two bands of radioactively labeled products with low electrophoretic mobility appeared (Fig. 2). The apparent molecular masses of the species present in these bands were 35 and 90 kDa, consistent with DNA cross-linking to the RPA2 and RPA1 sub-units, respectively (27). Despite the length, the extended htelo is sufficient to allow RPA3 interacting with the DNA (18), we did not detect any cross-link between this subunit and the extended htelo. We conclude from these cross-linking experiments that hRPA interacts with the extended htelo sequences via its RPA1 and RPA2 subunits.
To establish the relative position of each subunit on DNA, we calculated the relative percentage of the cross-linked species. We observed that RPA1 was primarily cross-linked on the 5Ј side of the DNA, whereas RPA2 was located on the 3Ј side ( Fig.   FIGURE 1. EMSA and FRET of htelo and extended htelo with hRPA. A, EMSA of 32 P-htelo and 32 P-extended-htelo. Oligonucleotides (2 nM) were incubated with hRPA (from 5 to 100 nM) for 20 min at 20°C; free DNA and hRPA-DNA complexes are indicated. The 5t-htelo figure is composed of two gels separated by a line. B and C, quantification of EMSA: percentage of DNA in hRPA-DNA complexes as a function of hRPA concentration for htelo and oligonucleotides with 5-nt tails (B) and for htelo and oligonucleotides with 10-nt tails (C). D, FRET efficiency (P) as a function of hRPA/FhteloT-10t (circles) and hRPA/10t-FhteloT (triangles) ratio (r); the oligonucleotide concentration was 100 nM and the protein concentration was from 100 to 1000 nM; error bars were Ͻ10%.
2). Thus, these results are in agreement with a 1:1 stoichiometry and show that hRPA binds extended htelo G4 with a 5Ј to 3Ј polarity (i.e. RPA1 and RPA2 are bound at the 5Ј and 3Ј sides of the oligonucleotides, respectively). This result is in accordance with results we previously obtained with the 21-nt telomeric G4 (htelo) (27) and with a 31-nt oligo(dT) (26). In addition, crosslinking experiments also showed that in the 1:1 complexes, hRPA covered the entire length of the unfolded 10t-htelo and htelo-10t oligonucleotides, indicative of the elongated extended complex, which is a 30-nt binding mode.
The Presence of G4 Ligands Reveals a Decreased Binding Affinity of hRPA for 5Ј Extended htelo and Suggests a 5Ј to 3Ј Directionality of the G4 Unfolding-EMSA and FRET did not reveal any hRPA unfolding directionality. We decided to investigate the effect of G4 stabilizing ligands (39) on binding and unfolding by hRPA. Indeed, a stabilizing G4 ligand strongly diminishes the unfolded or partially folded intermediate states, constraining RPA to mainly interact with the single-stranded tails located at the 5Ј and 3Ј side of the G4 formed by extended htelo. We used telomestatin and the pyridine dicarboxamide derivatives 360A and 360A-Br (structures shown in Ref. 27), which are remarkably selective for G4 relative to duplex DNA (40 -42).
Melting experiments followed by FRET indicated that each of the three ligands stabilized to a similar extent both the extended G4 structures 10t-FhteloT and FhteloT-10t, independently of the orientation of the extension (⌬T m ϭ 20°C for 1 mM 360A and 360A-Br; ⌬T m ϭ 20°C for 1 mM telomestatin, in KCl 10 mM plus LiCl 90 mM). Interestingly, in EMSA experiments, the presence of telomestatin decreased the affinity of hRPA for all four extended htelo sequences, but to different extents (Fig. 3, A and B). Binding of hRPA to htelo bearing 5-nt tails was strongly impaired (around 60% of inhibition) whether the tails were on the 5Ј or 3Ј side of the G4 (Fig. 3A). This result suggests that when the G4 structure is strongly stabilized by a G4 ligand, a single-stranded tail of 5 nt is not sufficient to allow binding of hRPA. The presence of 10-nt single-stranded tails allowed hRPA to bind to the oligonucleotides in the presence of telomestatin (Fig. 3B). Importantly, the affinity of hRPA depended on the position of the tail: hRPA bound 10t-htelo less efficiently than htelo-10t. This behavior was observed not only with telomestatin but also with 360A and a 360A-Br (Fig. 3C), showing that the difference in the binding between htelo-10t and 10t-htelo was not due to the nature of the ligand. These results show that hRPA has a lower binding efficiency to singlestranded regions at the 5Ј side of the G4 structure than to single-stranded tails at the 3Ј side as previously observed on hairpin and duplex structures (43,44).
To investigate whether binding of hRPA to 10t-htelo and htelo-10t in the presence of G4 ligand results in G4 unfold- ing, we carried out FRET experiments in the presence of 360A with both 10t-FhteloT and FhteloT-10t DNAs (Fig.  3D). Results show that hRPA opens the G4 when the singlestranded extension is at its 5Ј side (the p value increased from 0.3 to 0.8), whereas the G4 bearing the single-stranded tail at its 3Ј side is poorly opened (the p value increased from 0.35 to 0.5). We point out that, in the absence of hRPA (r ϭ 0), the p value in the presence of ligand was higher than in the absence of ligand (0.35 versus 0.15). This is due to the fact that 360A quenches TAMRA emission to a larger extent than FAM emission, as verified with htelo labeled with a single FAM at the 5Ј terminus or with a single TAMRA at 3Ј end showed that 360A (data not shown). Quenching of FRET processes by G4 ligands is a well known phenomenon (45). Hence, on one hand EMSA experiments in the presence of 360A show that hRPA binds to both 10t-htelo and htelo-10t, with a higher affinity for structures bearing a protruding 3Ј tail; on the other hand FRET experiments show that hRPA unwinds the G4 structure only when the single-stranded tail is on the 5Ј side of the G4, suggesting a 5Ј to 3Ј directionality in G4 unfolding by hRPA.
To ascertain that the observed 5Ј to 3Ј directionality in G4 unfolding was not due to the nature of the two fluorophores (i.e. to a different effect of FAM and TAMRA on the interaction of hRPA with the extended oligonucleotides), we initially designed the inverted systems bearing a TAMRA at the 5Ј side and a FAM at the 3Ј side of the G4. Unluckily 10t-ThteloF was less stable than ThteloF-10t, as revealed by FRET, making the inverted systems not suitable for such a control experiment (data not shown). Kinetic Experiments Reveal a 5Ј to 3Ј of Unfolding Directionality by hRPA-To further investigate the directionality of DNA unfolding by hRPA, we conceived a kinetic experiment based on a recently published helicase assay (46). We used four different DNA constructs (sequences are listed in Table 2). The first construct (11ss-htelo-dx) is the one used by Mendoza et al. (46); it is composed of a 11-nt single-stranded tail at the 5Ј side of the htelo sequence and a 15-bp duplex region (at the 3Ј side of the G4) labeled with a FAM and a Dabcyl dye allowing detecting the unwinding of the construct by fluorescence emission (Fig.  4A, inset). Two nucleotides separate the G4 from the duplex structure; this short spacer is not expected to provide a binding site for hRPA (23). Here we designed a symmetrical construct (dx-htelo-11ss) bearing the single-stranded tail at the 3Ј end (Fig. 4B, inset). The two symmetrical DNA constructs had similar melting profiles (not shown), making them suitable for the study of unfolding by hRPA. Finally we tested two additional constructs where the htelo sequence was replaced by a non-G4 sequence (named "mut") not able to fold into a G4 (Fig. 4, C and D, insets). Results showed that hRPA opened the constructs with the protruding 5Ј single strand, but not the constructs with the 3Ј protruding single strand (Fig. 4). A better efficiency to open duplex with a 5Ј single-stranded tail has already been reported (43,47). The opening curves of 11ss-htelo-dx were identical to the opening curves of 11ss-mut-dx, indicating that the opening of the duplex was the rate-limiting step of the unfolding of the construct.
Although this assay does not allow directly monitoring G4 unfolding but the final disruption of the duplex, it shows that unwinding by hRPA of the construct containing a G4 proceeds in the 5Ј to 3Ј direction, in agreement with our FRET experiments. In addition, we confirm that hRPA is able to open duplex structures (43,(47)(48)(49), and soundly demonstrate, for the first time, by kinetics experiments, that duplex unwinding proceeds in the 5Ј to 3Ј direction.

Discussion
In this work, we investigated the mechanism of G4s unfolding by hRPA. To this purpose, we studied the interaction of hRPA with oligonucleotides that adopt a G4 structure extended with a single-stranded tail on either side of the G4, in the absence or presence of a G4 stabilizing ligand.
We first showed that a single-stranded region of 5 or 10 nt on either side of the G4 enhanced binding of hRPA. This can be explained by two factors. On one hand, the G4s formed by the extended htelo have a lower stability than the G4 formed by htelo, and, as we previously showed, a decrease in G4 stability increases the binding affinity of hRPA (26,27). On the other hand, the presence of single-stranded tails increases the number of potential hRPA binding sites (15,16). Interestingly, Pif1 helicase was also reported to bind more efficiently to a singlestranded tailed G4 than a tailless G4 (50).
On one side, EMSA and FRET showed that, in the absence of G4 stabilizing ligands, hRPA bound and unfolded the G4 structure formed by the extended htelo oligonucleotides, regardless of whether the single-stranded extension was at the 5Ј or the 3Ј side. On the other side, EMSA and FRET in the presence of G4 stabilizing ligands revealed a surprising behavior: hRPA bound to both extended htelo oligonucleotides, however, it efficiently unwound only the G4 bearing a single-stranded extension at its 5Ј side, suggesting that hRPA unfolds G4s from 5Ј toward 3Ј. Moreover, a kinetic assay allowed unambiguously demonstrating a 5Ј to 3Ј unfolding directionality on duplex structures. Hence, overall, our results support that hRPA unfolds G4s and duplexes according to a 5Ј to 3Ј directionality.
Interestingly, hRPA 5Ј to 3Ј unfolding directionality is in agreement with results we obtained in a recent study of a mutation in RPA1 (D228Y), which alters the ability of hRPA to bind DNA and in particular G4 sequences and causes dramatic telomeric replication defects in fission yeast (Rpa1-D223Y strain) (34). We showed that overexpression of the 5Ј to 3Ј Pif1 helicase family members (51,52) rescues the telomere defects of the Rpa1-D223Y strain, whereas overexpression of the 3Ј to 5Ј helicase Rqh1 (53) or Pot1 that unfolds G4 from 3Ј to 5Ј (54) did not.
This 5Ј to 3Ј unfolding directionality can be explained in terms of the 5Ј to 3Ј oriented laying out of hRPA subunits along single-stranded DNA (Fig. 5A). The spatial polarity, previously observed for ssDNA and for the htelo sequence, has been here confirmed for extended G4 by cross-linking experiments. When hRPA binds next to the 5Ј side of a G4, its laying out toward 3Ј destabilizes and unfolds the G4 structure "in front" of it (Fig. 5B). Conversely, when it binds next to the 3Ј side of a G4, its laying out toward 3Ј does not affect the G4 structure (Fig. 5C) "behind" it.
In the absence of ligands, hRPA can interact with the singlestranded tails located at the 5Ј and 3Ј side of the G4, but also with transiently accessible single-stranded regions of the G4 core sequence (28 -30). Whereas, in the presence of stabilizing ligands, hRPA is constrained to mainly interact with the singlestranded tails. For this reason, the unfolding directionality was revealed only by experiments with extended htelo sequences in the presence of G4 stabilizing ligands.
EMSA in the presence of a ligand also revealed a decrease in hRPA binding efficiency to the htelo with a 5Ј extension. This might be explained by the 5Ј to 3Ј binding directionality of hRPA with respect to the single-stranded DNA. When hRPA binds to a few nucleotides next to the 5Ј side of a highly stable G4 in the 8 -10-nt binding mode, the DBD-C of RPA1 and DBD-D of RPA2 are oriented toward 3Ј (21), where the G4 is located; the G4 may hence physically hinder the interaction of hRPA with the DNA. A lower binding efficiency of hRPA to a 5Ј protruding single-stranded tail has already been reported for a hairpin and duplex structures with 5Ј or 3Ј single-stranded tails of 10 -19 nt (43,44).
In conclusion, our work demonstrates that hRPA unfolds G4s and duplexes from 5Ј to 3Ј. Interestingly our results (present and previous) show that hRPA opens more easily the htelo G4 than the duplex. Indeed 20 eq of hRPA were needed to obtain a fraction of unfolded duplex (ss-mut-dx) of about 0.7 and equilibrium was attained in about 30 min (Fig. 4); whereas only 5 eq of hRPA were needed to obtain the same unfolded fraction of htelo G4 and equilibrium was attained in less than 30 s (see Fig. 7 in Ref. 26). Likely, the unwinding mechanisms of G4s and duplexes are different. On one hand, the relative fast kinetics and the stoichiometric conditions of G4 unfolding supports that the unfolding directionality of G4s results from the sequential binding of the subunits of a single hRPA to DNA from 5Ј toward 3Ј. On the other hand, the relative slow kinetics and the high amount of hRPA needed to unfold the duplex suggests that duplex unfolding might result from a cooperative binding of hRPA proteins in the compact 8 -10-nt binding mode (23).
Thus, even if hRPA is not a helicase, it is able to unfold G4s, in addition to duplexes, with the same defined 5Ј to 3Ј unwinding polarity. In contrast to helicases that require ATP hydrolysis to unwind nucleic acid duplexes in a polar manner (55-57), hRPA does it in an energy-independent manner. It is possible that the polar unwinding activity of hRPA stems from its sequential binding mode involving intermediates with increasing binding site sizes. In addition, the polar unwinding activity of hRPA that we report in this study may facilitate the oriented access of specific proteins on single-stranded DNA. In particular, the ability of hRPA to unfold G4s from 5Ј to 3Ј is consistent with its important role in unfolding G4 structures on the lagging strand during telomere replication (Fig. 5D) (34).

Experimental Procedures
Materials-BSA was from Roche, [␥-32 P]ATP was from PerkinElmer Life Sciences, and T4 polynucleotide kinase was from New England BioLabs. Oligonucleotides were purchased from Eurogentec (Belgium). The sequences of the oligonucleotides used are listed in Tables 1 and 2. DNA concentrations were determined by UV spectroscopy using the extinction coefficients provided by the manufacturer. Recombinant hRPA was expressed in the Escherichia coli BL21(DE3) strain transformed with the plasmid pET 11a -hRPA that permits the co-expression of RPA1, RPA2, and RPA3. hRPA was purified using Affi-Gel Blue, hydroxyapatite (Bio-Rad), and Q-Sepharose chromatography columns as previously described (58). Telomestatin was a generous gift from Kazuo Shin-ya (University of Tokyo, Japan). The pyridine dicarboxamide derivatives 360A and 360A-Br were synthesized by Patrick Mailliet in our laboratory and Marie-Paule Teulade-Fichou (Institut Curie, Orsay, France), respectively. Telomestatin, 360A, and 360A-Br were dissolved at 10 mM in dimethyl sulfoxide immediately before use.
Electrophoretic Mobility Shift Assay-Oligonucleotides were labeled with [␥-32 P]ATP using T4 polynucleotide kinase. Nonincorporated [␥-32 P]ATP was removed using a Biospin 6 column (Bio-Rad) equilibrated in TE buffer. For all EMSA experiments, hRPA was diluted and preincubated (10 min at 4°C) in buffer containing 50 mM Tris-HCl (pH 7.5), 100 mM KCl, 1 mM DTT, 10% glycerol, 0.2 mg/ml of BSA, and 0.1 mM EDTA. In a standard reaction, radiolabeled oligonucleotide (2 nM) was incubated with various amounts of protein in 10 l of reaction buffer (50 mM HEPES (pH 7.9), 0.1 mg/ml of BSA, 100 mM KCl, and 2% glycerol) for 15 min at 20°C. Longer incubation times (up to 1 h) did not affect the patterns or intensities of the bands, indicating that the system had reached thermodynamic equilibrium within 15 min. Samples were then loaded on a native 1% agarose gels in 0.5ϫ TBE buffer. Electrophoresis was performed for 90 min at 5 V/cm at room temperature. After electrophoresis, the gel was dried and exposed to a PhosphorImager screen. After being exposed for at least 10 h, the screen was scanned with the Phosphorimager TYPHOON instrument (Molecular Dynamics). The band intensities were quantified using ImageQuant version 5.1. For each hRPA concentration, the fraction of radiolabeled oligonucleotide bound to hRPA was calculated as follow: I DNA bound to hRPA /(I free DNA ϩ I DNA bound to hRPA ). Each experiment was reproduced at least twice. When indicated, error bars correspond to the standard deviation calculated from at least two independent experiments. When G4 ligands were used, radiolabeled oligonucleotide (2 nM) was preincubated with 0.5 M G4 ligand for 10 min at room temperature in the standard reaction buffer before adding hRPA.
Thermal Melting Studies-UV-melting profiles were acquired on an Uvikon XL spectrometer (Secoman). The oligonucleotide were dissolved in 50 mM HEPES (pH 7.9) and 100 mM KCl, at 3 M strand concentration. Samples were heated at 90°C for a few minutes, cooled to 5°C, kept at 5°C for 30 min, heated to 90°C, and cooled to 5°C; heating and cooling were done at 0.2°C/min. Absorbance of oligonucleotides was recorded at 245, 260, 273, 295, and 400 nm as a function of temperature. Melting temperatures (T m ) were defined as the intercept between the melting curve at 295 nm and the median line between low-temperature and high-temperature absorbance linear baselines. At the end of melting experiences, CD spectra were collected, at 5°C, on a J-810 spectropolarimeter FIGURE 5. Model of G4 unfolding by RPA. A, according to literature, hRPA binds to ssDNA in a sequential pathway with a defined spatial polarity: first it binds ssDNA in the 8 -10-nt binding mode involving DBD-A and DBD-B of RPA1, then it lays out toward 3Ј in the 30-nt binding mode involving DBD-C of RPA1 and DBD-D of RPA2. B, when RPA binds to a single-stranded tail next to the 5Ј side of a G4, its sequential and spatialoriented laying out from 5Ј toward 3Ј destabilizes and unfolds the G4 in front of it. C, conversely, when RPA binds to a single-stranded tail next to the 3Ј side of a G4, laying out toward the 3Ј does not affect the G4 behind it. D, putative role of 5Ј to 3Ј unfolding directionality of RPA during telomere replication. RPA is recruited at telomeres with the replication machinery. ssDNA generated on the telomeric lagging strand template (G-rich) may fold into G4s. RPA contributes to the progression of the replication fork (which proceeds from 5Ј to 3Ј relative to the lagging strand template) by unfolding G4s from 5Ј to 3Ј.
(Jasco). Fluorescence melting profiles of oligonucleotides labeled with FAM and TAMRA were recorded on a MX3000P (Stratagene) at a strand concentration of 0.2 M in 10 mM cacodylic acid (pH 7.2) supplemented with 10 mM KCl and 90 mM LiCl, or with 100 mM KCl. The temperature was increased from 25 to 95°C at 1°C/min and the emission was recorded as a function of temperature (excitation wavelength 492 nm and emission wavelength 516 nm). Fluorescence melting profiles were normalized between the maximum and the minimum of fluorescence and T m was defined as the temperature at which the normalized fluorescence intensity was equal to 0.5.
Fluorescence Titrations Assay-Fluorescence spectra of 100 nM 10t-FhteloT or FhteloT-10t oligonucleotides were recorded at 20°C in a buffer containing 50 mM KCl, 2 mM MgCl 2 , and 5 mM lithium cacodylate (pH 7.2). 10t-FhteloT or FhteloT-10t were preincubated (10 min at 20°C) with or without 0.5 M G4 ligand. The protein (from 100 to 1000 nM) was directly added to the solutions containing the oligonucleotides alone or in the presence of the G4 ligand. The spectra (490 to 660 nm) were collected after 2 min of incubation while exciting at 470 nm. The fluorescence intensity of fluorophore was measured at 518 nm (I D ) for the donor FAM and at 586 nm (I A ) for the acceptor TAMRA. The ratio P was calculated as P ϭ I D /(I D ϩ I A ), where I D and I A are the emission intensities of the donor FAM and the acceptor TAMRA, respectively.
Fluorescence Kinetic Assay-All oligonucleotides used in this assay are listed in Table 2. DNA constructs (11sshtelo-dx and dx-htelo-11ss, 11ss-mut-dx and dx-mut-11ss) were prepared by annealing a mixture of Dabcyl-labeled oligonucleotide (1 M) and FAM-labeled oligonucleotide (0.85 M), in 20 mM Tris-HCl buffer (pH 7.3) containing 5 mM MgCl 2 , 1 mM KCl, and 99 mM NaCl. Mixtures were heated 5 min at 90°C and then slowly cooled to room temperature. DNA samples were then stored at Ϫ20°C. Unfolding reactions were carried out at least in triplicate in 96-well plates at 25°C and fluorescence was monitored in a microplate reader Infinite M1000 PRO (Tecan). Every replicate, containing 50 l of solution of 10 nM DNA construct (previously prepared at 1 M) and 100 nM of an oligonucleotide complementary to the FAM-labeled strand (named Trap1 in Table 2), was incubated with the indicated amount of hRPA (0, 10, 20, 50, 100, 150, or 200 nM). The 96-well plate was stirred for 10 s and the fluorescence emission recorded every 10 s. The excitation wavelength was set at 492 nm and the emission wavelength at 520 nm. Once the steady-state was attained (after about 30 -45 min), 100 nM of an oligonucleotide complementary to the Dabcyl-labeled sequence (named Trap2 in Table 2) was added to reaction wells to completely unfold the DNA construct. Well plates were then stirred for 10 s, and emission was monitored every 10 s until reaching again the steady-state of fluorescence emission. For each DNA construct, the fluorescence emission as a function of time was then normalized between the minimal and maximal values. To check the stability of 11ss-htelo-dx and dx-htelo-11ss DNA constructs, UV melting experiments were carried out with an UVmc2 spectrophotometer (SAFAS). Samples were heated at 96.5°C for 5 min, then cooled from 96.5 to 0.5°C, kept at 0.5°C for 10 min, and then heated from 0.5 to 96.5°C with a rate of 0.2°C/min. The absorbance as a function of temperature was monitored at 335, 295, 273, 260, and 240 nm.
Photocross-linking Experiments-Radiolabeled oligonucleotides (2 nM) were mixed with the indicated amount of protein using the protocol described under the EMSA section. Each sample (20 l) was irradiated for 30 min at 10°C at 360 nm. Laemmli gels (10%) were used to separate cross-linked species.
Author Contributions-C. S. and J. F. R. designed the study; C. S., J. F. R., and P. A. wrote the manuscript; L. S. performed EMSA, fluorescence titration, and cross-linking assays; L. S., F. T., I. P., and E. D. prepared hRPA; P. A. carried out UV and FRET melting experiments and collected CD spectra; N. G. performed fluorescence kinetic assays; N. G., A. B., and O. M. analyzed kinetic assays; O. L. contributed to the discussion of data. All authors revised and approved the final version of the manuscript.