Binding of G-quadruplexes to the N-terminal Recognition Domain of the RNA Helicase Associated with AU-rich Element (RHAU)*

Background: The helicase RHAU requires an N-terminal extension to bind quadruplex structures. Results: This extension adopts an elongated shape and interacts with the guanine tetrad face of quadruplexes. Conclusion: We provide a basis for the understanding of quadruplex binding by the N-terminal domain. Significance: The N-terminal region does not require the 2′-OH of the ribose to mediate the protein-quadruplex interaction. Polynucleotides containing consecutive tracts of guanines can adopt an intramolecular G-quadruplex structure where multiple planar tetrads of hydrogen-bound guanines stack on top of each other. Remodeling of G-quadruplexes impacts numerous aspects of nucleotide biology including transcriptional and translational control. RNA helicase associated with AU-rich element (RHAU), a member of the ATP-dependent DEX(H/D) family of RNA helicases, has been established as a major cellular quadruplex resolvase. RHAU contains a core helicase domain responsible for ATP binding/hydrolysis/helicase activity and is flanked on either side by N- and C-terminal extensions. The N-terminal extension is required for quadruplex recognition, and we have previously demonstrated complex formation between this domain and a quadruplex from human telomerase RNA. Here we used an integrated approach that includes small angle x-ray scattering, nuclear magnetic resonance spectroscopy, circular dichroism, and dynamic light scattering methods to demonstrate the recognition of G-quadruplexes by the N-terminal domain of RHAU. Based on our results, we conclude that (i) quadruplex from the human telomerase RNA and its DNA analog both adopt a disc shape in solution, (ii) RHAU53–105 adopts a defined and extended conformation in solution, and (iii) the N-terminal domain mediates an interaction with a guanine tetrad face of quadruplexes. Together, these data form the foundation for understanding the recognition of quadruplexes by the N-terminal domain of RHAU.


Polynucleotides containing consecutive tracts of guanines can adopt an intramolecular G-quadruplex structure where multiple planar tetrads of hydrogen-bound guanines stack on top of each other. Remodeling of G-quadruplexes impacts numerous aspects of nucleotide biology including transcriptional and translational control. RNA helicase associated with AU-rich element (RHAU), a member of the ATP-dependent DEX(H/D) family of RNA helicases, has been established as a major cellular quadruplex resolvase. RHAU contains a core helicase domain responsible for ATP binding/hydrolysis/helicase activity and is flanked on either side by N-and C-terminal exten-
sions. The N-terminal extension is required for quadruplex recognition, and we have previously demonstrated complex formation between this domain and a quadruplex from human telomerase RNA. Here we used an integrated approach that includes small angle x-ray scattering, nuclear magnetic resonance spectroscopy, circular dichroism, and dynamic light scattering methods to demonstrate the recognition of G-quadruplexes by the N-terminal domain of RHAU. Based on our results, we conclude that (i) quadruplex from the human telomerase RNA and its DNA analog both adopt a disc shape in solution, (ii) RHAU 53-105 adopts a defined and extended conformation in solution, and (iii) the N-terminal domain mediates an interaction with a guanine tetrad face of quadruplexes. Together, these data form the foundation for understanding the recognition of quadruplexes by the N-terminal domain of RHAU.
G-quadruplex structures spontaneously fold from RNA or DNA sequences containing consecutive runs of two or more adjacent guanines (G-tracts) and assemble into stable stacks of planar G-quartets that are stabilized by -stacking interactions (1)(2)(3)(4)(5)(6)(7). Quadruplexes are further stabilized by monovalent cations (typically K ϩ or Na ϩ ) that interact with guanine O6 oxygens at the center of the quadruplex (8). A diversity of quadruplex structures can assemble as either parallel or antiparallel configurations that vary based on the composition and length of intervening loops between guanines within a G-tract or between tracts themselves (9). Although the formation of quadruplexes in compatible sequences of DNA and RNA was originally considered to be an in vitro artifact of laboratory research, there is now compelling evidence that suggests that their formation in vivo may have great importance to a variety of biochemical processes (10,11). In silico studies suggest a widespread prevalence of RNA and DNA quadruplexes, and this trend has been confirmed experimentally by several studies that demonstrate that formation of G-quadruplexes impacts transcriptional control in promoter regions of genes and modulates translational repression or activation via quadruplex formation in the untranslated regions (UTRs) of mRNAs (12)(13)(14)(15)(16). Given that melting temperatures of quadruplexes are much higher than physiological temperature, manipulation of these unusually stable structures to provide regulatory control requires the work of specialized enzymes in living systems.
RNA helicase associated with AU-rich element (RHAU 9 ; also known as DHX36 and G4R1) has been recognized as a major source of quadruplex-resolving activity in cell lysates (17,18). RHAU belongs to the DEX(H/D) family of RNA helicases, a large group of enzymes that remodel nucleic acid structures in an ATP-dependent manner (19). Recently, specific functional roles for RHAU have emerged. Microarray studies have implicated RHAU in broad spectrum transcriptional regulation and mRNA stability (20). In support, RHAU has been shown to control gene expression through direct binding to quadruplexes within the promoter of the YY1 gene (21). RHAU can also modulate quadruplex structures in the context of non-coding RNAs. We have previously demonstrated that RHAU associates with a G-quadruplex at the 5Ј-end of human telomerase RNA (hTR) and is capable of unwinding this quadruplex in vitro (22). Interestingly, RHAU can associate with hTR in the context of the telomerase holoenzyme, an interaction that correlates with telomerase activity (23,24). G-tract substitutions in the quadruplex-forming region of hTR appear to confirm these results (23)(24)(25). Taken together, recognition and remodeling of quadruplex structures by RHAU play a central role in a number of key cellular regulatory processes.
RHAU is highly specific for G-quadruplex structures over duplex and single-stranded regions, and although RHAU has been shown to have affinity for both RNA and DNA, enhanced affinity for RNA structures has been reported (17). RHAU contains a core helicase domain (ϳ470 amino acids) responsible for ATP binding/hydrolysis/helicase activity and is flanked on either side by N-and C-terminal extensions (26). The C-terminal extension contains a predicted companion domain to the helicase domain based on sequence homology (residues 682-920), although its biological role is unknown (27). Preceded by a Gly-rich region (residues 10 -51), the N-terminal domain extends into a RHAU-specific motif (RSM; residues 54 -66), which has been demonstrated to be required for interaction with a synthetic tetramolecular RNA quadruplex (26,28). The RSM dependence is highlighted by the fact that it is the only conserved element within the N-terminal extension in the Drosophila RHAU ortholog (26). Deletion of the RSM attenuates quadruplex binding and unwinding, and a similar effect is observed upon mutagenesis of conserved small (Gly-59 and Gly-63) and hydrophobic (Leu-57 and Ile-62) residues in the RSM (26). Interestingly, mutation of basic RSM residues (Lys-58 and Arg-60) has no negative effect on quadruplex binding.
Although the importance of G-quadruplexes to a myriad of cellular processes is an emerging theme, there is scant information regarding the recognition of this structural motif by proteins. Here we present a comprehensive study detailing the interaction between the N-terminal domain of RHAU and an established binding partner, a G-quadruplex from the 5Ј-end of the human telomerase RNA. We specifically focused on an N-terminal RHAU construct containing the RSM and demonstrate that it adopts an extended conformation capable of mediating an interaction with both RNA and DNA quadruplexes. Based on our experimental approach, we propose a model in which the RSM recognizes the guanine tetrad face.
Protein Expression and Purification-RHAU 53-105 was cloned, expressed, and purified as described previously (22) with a final polishing step of size exclusion chromatography on a Superdex 200 10/300 GL column (GE Healthcare) in 10 mM HEPES, pH 7.5, 150 mM sodium chloride (I ϭ 154 mM) after removal of the hexahistidine (His 6 ) tag by thrombin. Protein identity was confirmed by mass spectrometry. RHAU 53-209 was initially cloned, expressed, and purified in an identical manner, but upon analysis by both gel electrophoresis and mass spectrometry, it was determined that a stable proteolytic product corresponding to RHAU 53-150 had been produced. Isotopically enriched 15 N-and 15 N, 13 C-labeled RHAU 53-105 (or RHAU 53-150 ) was overexpressed in M9 minimal medium as described previously (30). The concentration of RHAU 53-105 and RHAU 53-150 was measured spectrophotometrically at 280 nm using extinction coefficients of 7020 and 13,980 M Ϫ1 cm Ϫ1 , respectively, that were calculated from the sequence using the ExPASy ProtParam tool (31).
Protein⅐Quadruplex Complex Assembly-Protein and quadruplexes were prepared separately as described above; diluted to 10 M in 20 mM HEPES, pH 7.5, 100 mM KCl; mixed in an equimolar ratio; gently shaken for 15 min on a rotator; and then concentrated to a convenient volume. Afterward, the complexes were purified on a Superdex 200 10/300 GL SEC column. Complex concentration was determined by UV absorption using the extinction coefficient (⑀ 260 nm ) of the nucleic acid component.
RHAU Association with RNA Quadruplex-All siRNA knockdowns, Western blotting, electrophoretic mobility shift assays, and RNA co-immunoprecipitations in HEK293T cells were performed as described previously (22) with the exception of the His 6 -RHAU 53-105 immunoprecipitations in which 1 mg of cell lysate was supplemented with 350 nM His 6 -RHAU 53-105 for 30 min followed by immunoprecipitation with 10 g of anti-His 6 antibody (Abcam) for 30 min with end-over-end mixing.
Hydrodynamic Characterization-Sedimentation velocity data of RHAU 53-105 were collected at 45,000 rpm (20°C) at 10-min intervals as described previously (32). Sedimentation coefficient values (s 20,b ) obtained at multiple concentrations were corrected to the standard solvent conditions to obtain s 20,w values, which were then plotted against concentration to obtain the s 20,w 0 . DLS data were collected as reported previously (33). All samples were subjected to filtration through a 0.1-m filter (Millipore) and equilibrated for 5 min at 20°C prior to the measurements. The sedimentation coefficient and hydrodynamic radius (R h ) information from analytical ultracentrifugation and DLS, respectively, were combined using a modified Svedberg equation and yielded a molecular mass of 6.7 kDa for RHAU 53-105 .
UV-Visible Absorption Spectroscopy-UV-visible spectra were recorded on an Evolution 260 Bio spectrophotometer (Thermo Scientific) using a temperature-controlled 1.0-mm sample cell and a 1.0-mm water-filled reference cell. Samples and buffer were measured at each temperature in triplicate. The averaged buffer spectra were subtracted from the corresponding averaged sample spectra.
Spectropolarimetry-All spectra were recorded on a calibrated (2.583 mM (1S)-(ϩ)-camphor-10-sulfonic acid monohydrate; Alfa Aesar) J-810 spectropolarimeter (Jasco Inc.). From 350 to 220 nm, a 1.0-mm cell was used with an integration time of 8 s; from 220 to 200 nm, a 1.0-mm cell and 32-s integration time were used; and from 220 to 180 nm, a 0.1-mm cell and an integration time of 32 s were used. We used sample concentrations of 30 -50 M in the 1.0-mm cell and 200 M in the 0.1-mm cell. In all cases, sample and buffer were measured in triplicate, and after averaging, the base line was subtracted from the sample spectrum. Nucleic acid spectra recorded in either 20 mM HEPES, pH 7.5, 100 mM KCl or in 20 mM sodium phosphate, pH 7.5, 100 mM KF were identical. The phosphate buffer has a much lower absorbance and was used to extend the measurable spectral range from 210 to 180 nm. Thermal melting curves following the ellipticity at 263 (DNA [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] or 264 nm (hTR [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] were recorded for each of the quadruplexes in the mentioned HEPES buffer. The spectra were normalized by the number of bases per unit volume. RHAU 53-105 spectra were identical in 10 mM HEPES, pH 7.5, I ϭ 154 mM (NaCl) or in 10 mM Tris, pH 7.5, I ϭ 154 mM (NaCl). The Tris buffer was used to extend the spectral range to 190 nm because of its more favorable absorption properties. Dialysis into phosphate buffer resulted in a reduced minimum at 200 nm and the substitution of Na ϩ with K ϩ or of Cl Ϫ with F Ϫ in the formation of aggregates. The spectra were normalized by the number of peptide bonds per unit volume. Protein⅐nucleic acid complexes were measured in the same HEPES buffer that was used for the nucleic acids, and the spectra were normalized by the number of bases per unit volume (above 250 nm, only the nucleic acid contributes to the spectrum).
Small Angle X-ray Scattering-Scattering data were collected at 20°C for RHAU 53 (0.9, 1.2, 1.5, and 1.8 mg/ml) in 10 mM HEPES, pH 7.5, I ϭ 154 mM (NaCl) for proteins and 20 mM HEPES, pH 7.5, 100 mM KCl for quadruplexes and protein⅐nucleic acid complexes (with additional 1 mM EDTA for RNA). Scattering data collection and processing were performed as described previously (34). The radius of gyration (R g ) and maximum particle dimension (D max ) were extracted using the GNOM program (35), and solution conformations were determined using DAMMIN (36) with experimental R g and D max values as constraints for protein, RNA, DNA, and their complexes. The quality of the models was verified by the goodness of fit parameter ( value) after each model calculation. Individual models for each entity were then rotated and averaged using the program DAMAVER (37). Sample quality was confirmed after SAXS experiments by DLS. The hydrodynamic properties were calculated for each model using the program HYDROPRO (38) to validate models.
Nuclear Magnetic Resonance (NMR) Spectroscopy-NMR samples were prepared in buffers identical to those for SAXS with the exception that a 9:1 H 2 O:D 2 O mixture was used. All protein spectra were acquired on a Varian Unity INOVA 600-MHz spectrometer, processed with NMRPipe (39), and visualized with Sparky (T. D. Goddard and D. G. Kneller, University of California, San Francisco). To assign the 1 H-15 N RHAU 53-105 HSQC spectrum, 15 N, 13 C-labeled RHAU 53-105 was subjected to a suite of multidimensional NMR experiments including CBCACONNH, HCCH total correlation spectroscopy, HNCA, HNCACB, HNCACO, HNCO, and NOESY-HSQC. Imino proton spectra were acquired using a Bruker Avance 500-MHz spectrometer and processed/visualized using iNMR.
We next purified the DNA equivalent of hTR 1-20 (DNA 1-20 ) by SEC to compare it with its RNA counterpart. The procedure separated two distinct DNA conformations, suggesting that although both species have identical molecular weight they may differ slightly in terms of their solution conformation (Fig. 1A). To achieve effective separation, a large scale preparatory column was used that resolved two separate peaks: DNA 1-20 C1 (first eluted species) and DNA 1-20 C2 (second eluted species). Both DNA species migrated similarly and stained efficiently with quadruplex-specific dye ( Fig. 2A). The imino proton regions of the NMR spectra were again consistent with quadruplex formation for both DNA conformations (Fig. 2B). As with the RNA, significant hypochromicity at 297 nm was observed upon heating of both DNA 1-20 C1 and DNA 1-20 C2 , consistent with quadruplex formation (Fig. 2C). Comparison of the far-UV CD spectra of hTR 1-20 , DNA 1-20 C1 , and DNA 1-20 C2 revealed similar overall features, characterized by a maximum ellipticity at 263 (DNA) or 264 nm (RNA) and a minimum at 242 nm. DNA 1-20 C1 and DNA 1-20 C2 have a second maximum at 207 (C1) or 209 nm (C2), whereas the RNA has a local minimum at this wavelength (Fig. 2D). Thermal melting of the quadruplex species resulted in a gradual disappearance of these spectral features (characteristic of quadruplex unstacking) with hTR 1-20 proving the most resistant to denaturation with a midpoint of the transition above 80°C followed by DNA 1-20 C1 with an estimated midpoint at 70°C and DNA 1-20 C2 at 64°C (Fig.  2E). These values were significantly higher than that expected for single-or double-stranded nucleic acid structures.
The solution conformations of both RNA and DNA quadruplexes were determined by SAXS. As an additional quality control step, we used DLS to ensure monodispersity and suitability for SAXS analysis (Fig. 1B). The R h values determined by DLS for hTR [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] , DNA 1-20 C1 , and DNA 1-20 C2 are similar, and the molecules are monodisperse at high concentration ( Fig. 1B and Table 1). SAXS data collected at multiple concentrations were merged to obtain a single scattering profile (Fig. 3A) that was used to generate the pair distance distribution function (P(r)) (a histogram of all observed distances between electron pairs in the molecule), enabling determination of the R g and the D max for each quadruplex ( Fig. 3B and Table 1). Based on the scattering data, multiple solution conformations for each quadruplex were generated, rotated, and superimposed to obtain the averaged solution conformation (Fig. 3C). Each quadruplex adopts a highly similar disc-shaped structure with a prominent concave bevel at the top and bottom. Near perfect superposition of a high resolution DNA quadruplex structure containing three guanine tetrads (Protein Data Bank code 3SC8 (40)) onto the low resolution models increased our confidence in these low resolution models and orients the guanine tetrad faces with their cavities superimposed on the concave bevels. Model quality was validated through excellent superposition of individually calculated models (normalized spatial discrepancy (NSD) parameter Յ0.51 for each nucleic acid molecule ensemble).
N-terminal Domain of RHAU Mediates an Interaction with hTR-To investigate the interaction between the N-terminal domain of RHAU and hTR quadruplex, we focused on a short truncation of the domain (RHAU 53-105 ) that contains the RSM (residues 54 -66) and can bind quadruplex in vitro (22) (Fig.  4A). We have previously demonstrated that endogenous RHAU co-immunoprecipitates with the hTR from HEK293T cell lysates (22). We tested the binding specificity of RHAU 53-105 by performing RNA interaction assays in HEK293T cell lysates with RHAU 53-105 purified prior to proteolytic removal of the N-terminal His 6 tag. His 6 -tagged RHAU 53-105 was able to specifically immunoprecipitate (using anti-His 6 antibody) hTR from HEK293T cell lysates in both the presence and absence of endogenous RHAU (Fig. 4B). This enrichment was specific to a known RHAU target (hTR) but not to a negative control RNA (GAPDH). Next, we examined the interaction between RHAU and hTR via immunoprecipitations of endogenous RHAU in the presence and absence of an excess of RHAU 53-105 . RNA isolated from RHAU immunoprecipitations demonstrated a ϳ100-fold enrichment of the telomerase RNA as compared with an equal amount of total RNA. When RHAU 53-105 was introduced to the cell lysates, this enrichment was reduced ϳ5-fold, indicating that an excess of RHAU 53-105 disrupts binding of RHAU to the hTR quadruplex (Fig. 4C). Electropho-   , and DNA 1-20 C2 to a similarly sized complex under native conditions (Fig. 4D). Dissociation constants for RHAU 53-105 in complex with hTR 1-20 (202 Ϯ 17 nM), DNA 1-20 C1 (299 ϩ 18 nM), and DNA 1-20 C2 (332 Ϯ 24 nM) were estimated based on quantitation of EMSAs at multiple protein concentrations (Fig. 4E). Together, these results suggest that RHAU 53-105 is capable of mediating a specific interaction with hTR.
N-terminal Domain of RHAU Adopts an Extended Conformation-Recombinantly expressed RHAU 53-105 elutes as a single peak by SEC after affinity tag removal via thrombin diges-tion (Fig. 1A). Sedimentation velocity experiments using analytical ultracentrifugation at multiple concentrations revealed that RHAU 53-105 remains stable and monomeric with s 20,w 0 of 0.88 Ϯ 0.05 S (Fig. 5A). Interestingly, a frictional ratio (f/f o ) of ϳ1.40 was obtained from sedimentation velocity data that suggested an extended conformation in solution. DLS analysis at multiple concentrations also confirmed that RHAU 53-105 is stable and behaves as a monomer (Fig. 1B). The far-UV CD spectrum of RHAU 53-105 shows features of both random coil and ordered secondary structural elements (Fig. 5B).
The solution conformation of RHAU 53-105 was further determined by SAXS using merged data obtained from data collected at multiple concentrations to generate a single scat-  tering profile (Fig. 5C). An extended pair distance distribution function was observed that is typical for an elongated molecule ( Fig. 5D) with R g and D max values of 2.1 Ϯ 0.2 and 7.8 nm, respectively (Table 1). Based on the scattering data, multiple solution conformations for RHAU 53-105 were generated, rotated, and superimposed to obtain an averaged solution conformation (Fig. 5E). The NSD parameter of 0.58 Ϯ 0.06 obtained by comparing all models indicates an excellent agreement between independently calculated conformations. Low resolution shape analysis of RHAU 53-105 revealed an elongated conformation with two well defined subdomains, consistent with an extended and folded structure. Agreement between experimentally determined hydrodynamic parameters and those calculated based solely on the individual SAXS models served to further validate the models (Table 1). Identical DLS results were obtained pre-and post-SAXS data collection, con-firming the absence of any radiation damage to the sample after exposure to x-rays. Similarly to RHAU 53-105 , a longer RHAU 53-150 subdomain also adopts an extended conformation that enabled the assignment of the N-and C-terminal regions of the models (Fig. 5, C, D, and E, and Table 1).
SAXS. As described above, scattering data were collected from each complex at multiple concentrations to ensure the absence of self-association and to increase the signal-tonoise ratio (Fig. 6A). RHAU 53-105 ⅐hTR 1-20 presents an extended pair distance distribution function with features that are roughly similar to the free protein but suggests an increased frequency of distances observed in the 4.5-6.5-nm range (Fig. 6B). Interestingly, for the RNA complex, the D max increased modestly from 7.8 (free protein) to 8.5 nm (complex), which corresponds approximately to the short edge of the quadruplex disc ( Table 1). The average solution conformation based on several models of RHAU 53-105 ⅐hTR 1-20 shares many of the features of the free protein with the exception of a significant thickening of the N-terminal subdomain upon interaction with quadruplex (Fig. 6C). Fitting of the individual average solution conformations of RHAU 53-105 and hTR 1-20 into the complex envelope orients the quadruplex at the end of the RSM-containing subdomain and appears to orient the quadruplex such that the guanine tetrad face (largest surface area; contains the concave bevel), and not the sugar-phosphate backbone edges, mediates the interaction with RHAU (Fig. 6D). Model and data quality are evidenced by the low NSD (0.52 Ϯ 0.02), indicating excellent superposition of individually calculated models, and a Lysates were supplemented with 350 nM His 6 -RHAU 53-105 for 30 min with end-over-end mixing, and RNA⅐protein complexes were purified using an anti-His 6 antibody. Co-purified RNA was isolated, and both hTR and GAPDH mRNAs were quantified by reverse transcription and PCR. -Fold enrichment was calculated using the comparative C T method relative to an equal amount of total RNA from each sample. RNAi knockdowns were performed with identical efficiency as shown previously (22). C, bar graph representing -fold enrichment of the hTR mRNA as measured by RT-PCR relative to an equal amount of total RNA. RNA co-immunoprecipitated with endogenous RHAU from untreated HEK293T cell lysates and lysates supplemented with 5 M RHAU 53-105 was purified and subjected to reverse transcription and PCR using hTR-specific primers. -Fold enrichment was calculated using the comparative C T method. D, EMSAs demonstrate complex formation between DNA 1-20 C1 , DNA 1-20 C2 , and hTR 1-20 with RHAU 53-105 . Nucleic acids were incubated at 200 nM either alone or with 1 M RHAU 53-105 for 10 min prior to separation on a 12% native Tris borate-EDTA polyacrylamide gel and stained with SYBR Gold to visualize nucleic acids. E, quantitation of EMSA results for hTR 1-20 (green), DNA 1-20 C1 (orange), and DNA 1-20 C2 (goldenrod). Nucleic acids were incubated at 150 nM with increasing concentrations of RHAU 53-105 (0 -750 nM) for 10 min prior to separation on a 12% native Tris borate-EDTA polyacrylamide gel and stained with SYBR Gold to visualize/quantify nucleic acids. Experiments were repeated in triplicate with associated S.D. indicated as error bars. Dissociation constant determination was performed according to the methodology outlined by Ryder et al. (48). IP, immunoprecipitation. DECEMBER 6, 2013 • VOLUME 288 • NUMBER 49 value of 1.1, indicating good agreement between the individual models calculated and the raw scattering data (Table 1).

Quadruplex Recognition by the RHAU-specific Motif
Conversely, complexes between RHAU 53-105 with either DNA quadruplex present a bell-shaped pair distance distribution function and are significantly more compact than the free protein alone or the RNA-containing complex (Fig. 6B). Model determination revealed two similar globular average solution conformations for the DNA-containing complexes that lack subdomain definition characteristic of both the free protein and RNA-containing complex (Fig. 6E). Rigid body modeling of individual components was therefore not feasible in the context of the DNA complexes.
The RSM Mediates an Interaction with Both RNA and DNA Quadruplex Structures-To attempt to determine the region(s) of the N-terminal domain involved in mediating the interaction, we performed NMR chemical shift perturbation studies on RHAU 53-105 in the presence and absence of G-quadruplexes. Amide crosspeaks from 15 N HSQC spectra of free [ 15 N]RHAU 53-105 were clustered heavily between 7.5 and 8.5 ppm in the proton dimension, consistent with the extended conformation observed by SAXS and lacking the dispersion characteristic of ␤-sheet structure (Fig. 7A). The spectrum was sufficiently resolved to enable spectral assignments of a large number of amide resonances based on a standard suite of triple resonance experiments. Most importantly, the vast majority of the resonances corresponding to the RSM were unambiguously assigned, including the lone tryptophan side chain amine resonance (Trp-65 sc ) significantly downfield past 10 ppm in the proton dimension.
To determine the region(s) perturbed in quadruplex binding, we superimposed the HSQC spectra of [ 15 N]RHAU 53-105 , free FIGURE 5. RHAU 53-105 adopts an extended conformation. A, sedimentation coefficient profile obtained by c(s) analysis for RHAU 53-105 at 0.7 mg/ml indicating its purity. The concentration dependence of the sedimentation coefficient is shown in the inset. B, far-UV CD spectra of RHAU 53-105 (red) in comparison with reference spectra (49) of secondary structure elements random coil (rc), ␣-helix (␣), and ␤-strand (␤). C, merged SAXS data for RHAU 53-105 (red) and RHAU 53-150 (gray) obtained from multiple concentrations for each protein. The fit of the calculated scattering profile of the obtained models in E to the raw scattering data is overlaid. D, corresponding pair distance distribution functions. E, averaged low resolution models for RHAU 53-105 (left) and RHAU 53-150 (middle) suggesting an elongated shape of both proteins. Superimposition of the two protein models (right) allowed the assignment of the N and C termini. The longitudinal extension (corresponding to D max ) is indicated below the models.

Quadruplex Recognition by the RHAU-specific Motif
and in complex with hTR 1-20 (Fig. 7A). Although the majority of amide resonances were unchanged, significant chemical shift perturbations were observed in the RSM and the adjacent ␣1 helix (Lys-58, Arg-60, Glu-61, Ile-62, Gly-63, Met-64, Trp-65, Tyr-66, Ala-67, Lys-68, Lys-69, Asn-74, Lys-75, Ala-77, and Glu-78). We repeated NMR experiments using the longer [ 15 N]RHAU 53-150 to determine the impact of additional C-terminal residues. Although spectral overlap precluded assignment of the amide resonances, comparison of the HSQC spectrum of the protein, free and in complex with hTR 1-20 , demonstrated that no additional resonances were shifted (data not shown). Notably, a second tryptophan in the longer construct is unperturbed, whereas that in the RSM shifts upon quadruplex binding.
Chemical shift mapping experiments on DNA-containing complexes demonstrated an identical pattern of perturbations in the RSM and adjacent ␣1 helix as in the complex containing hTR [1][2][3][4][5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20] (Fig. 7B). Therefore, both RNA and DNA quadruplexes interact in a similar manner with the RSM region of RHAU. In contrast to [ 15 N]RHAU 53-105 ⅐hTR 1-20 , numerous new resonances appear upon formation of the DNA complexes in various spectral regions. It is an important distinction that these resonances were not originally observable in the spectrum of the free protein, indicating that an unobservable region of the protein has become more ordered upon quadruplex binding. Although we did not attempt assignment of resonances in the DNA complexes, the most straightforward interpretation consistent with the SAXS results is that a second region in addition to the RSM is mediating the interaction with the DNA.

DISCUSSION
RNA and DNA G-quadruplex structures have emerged as important cis-acting regulatory elements and have recently been quantitatively visualized in human cells in a number of contexts (11). Although the structural and biophysical characterization of G-quadruplexes has progressed rapidly, the mechanistic basis for their recognition and remodeling is poorly understood. RHAU has emerged as a primary quadruplex resolvase in eukaryotic cells, and the functional consequences of quadruplex remodeling has been established in a small number of important processes (17,18,21,41,42). Therefore, our goal was to make biophysical inroads to the process by examining the recognition of G-quadruplexes by the N-terminal region of RHAU that contains the RSM, a required region for quadruplex recognition.
Our primary concern with the experimental approach was the biological relevance of using the RHAU 53-105 truncation to characterize quadruplex recognition. Although the full-length helicase has significantly higher affinity for quadruplex than the N-terminal subdomain (26), we have previously established that RHAU 53-105 can interact with hTR 1-20 by EMSA in vitro (22) and extended this work to demonstrate that RHAU 53-105 is capable of mediating an interaction in the context of a cell lysate (Fig. 4). We have demonstrated by a number of biophysical methods that RHAU 53-105 adopts an extended conformation in which the RSM is likely localized to the largest N-terminal subdomain. It is important to stress that, although extended, the data suggest a well defined, as opposed to a highly dynamic, structure. The solution conformation determined by SAXS in particular reinforces this point as individual models superimpose well on each other (NSD ϭ 0.58 Ϯ 0.06). The observed solution conformation is not unexpected in the context of the full-length protein as it is preceded by a glycine-rich region and followed by a region that we have observed is susceptible to proteolysis (Fig. 4A). NMR data on RHAU 53-105 demonstrate dispersion that is consistent with secondary structure predictions (Fig. 7A) and are supported by CD data indicating a mixture of defined secondary structure elements and random coil (Fig. 5B).
We present a comprehensive series of SAXS, NMR chemical shift perturbation, CD, DLS, SEC, and analytical ultracentrifugation data to build a strong case for understanding quadruplex recognition by the N-terminal domain of RHAU. SAXS reconstruction suggests that the tetrad face, as opposed to the sugarphosphate backbone, is the recognition surface for the RSM (Fig. 6D). We have deconvoluted the RNA and protein components of the model using rigid body approaches, which are feasible in this situation based on CD and imino proton data indicating minimal perturbation of quadruplex structure (Fig. 8, A  and B). Mutations of charged amino acids in the RSM, which would typically mediate the electrostatic recognition of the sugar-phosphate backbone, have minimal impact on RHAU function (26). Conversely, quadruplex binding and helicase activity are attenuated with small and hydrophobic amino acid mutations in the RSM (26). Qualitatively, the staining intensity of G-quadruplexes bound by RHAU 53-105 with N-methyl and 209 nm (C2), respectively, whereas the RNA has a local minimum at this wavelength. Conformational differences between the quadruplexes manifest in the 280 -315-nm region. Above 250 nm, only the nucleic acids contribute to the spectra. Below 250 nm, the CD signal is modulated by the protein signal. B, NMR chemical shifts of imino protons of DNA 1-20 C1 (230 M) and DNA 1-20 C2 (224 M), either free (black) or in equimolar complex with RHAU 53-105 (red). DECEMBER 6, 2013 • VOLUME 288 • NUMBER 49 mesoporphyrin IX, a quadruplex-specific dye that stacks on the G-tetrad faces (43), is significantly reduced relative to free quadruplex. This mode of recognition is novel relative to the few existing high resolution structures of a peptide bound to an RNA duplex-quadruplex junction (44) and a telomeric protein⅐DNA complex (45). NMR chemical shift perturbation experiments indicate that both RNA and DNA quadruplexes perturb identical backbone amide resonances in the RSM and ␣1 helix of RHAU 53-105 , strongly suggesting that the 2Ј-OH of the ribose sugar is not required to mediate the interaction (Fig.  7B). Although at this point we do not have the resolution to be able to address the importance of different solution conformations observed between the RNA-and DNA-containing complexes, it appears likely that the lack of 2Ј-OH enables an additional interaction with a region distal to the RSM. Although we await high resolution structures to confirm these observations, the results obtained are consistent with the RSM mediating the recognition but not unwinding of G-quadruplex structure.

Quadruplex Recognition by the RHAU-specific Motif
Although an increasing number of helicases, including RHAU, have been associated with quadruplex recognition and unwinding, the mechanism of quadruplex unwinding remains elusive. The high resolution structure of Prp43p, an RNA helicase that lacks the N-terminal extension of RHAU but shares significant sequence identity in the core helicase domain and C-terminal extension, revealed homology to existing DNA helicases and the presence of an ATP/Mg 2ϩ binding pocket (27,46). Furthermore, the structure mechanistically suggests that ATP binding/hydrolysis induces a conformational change to expose the RNA binding site required for unwinding/remodeling. We propose a working framework for the recognition of G-quadruplexes by RHAU. The RSM-containing N-terminal region is unique to RHAU and adopts an extended conformation capable of G-quadruplex binding via a G-tetrad face, leaving the sugar-phosphate backbone edges exposed. We do not expect the addition of the glycine-rich region to impede quadruplex binding as a construct retaining an N-terminal His 6 tag and linker does not impede the interaction with hTR in cell lysates (Fig. 4). The specific role of the Gly-rich motif is unclear at this point, although these motifs have been observed proximal to single-stranded RNA recognition motifs to function in RNA binding (47). Connection of the RSM to the core helicase domain via a flexible linker (consistent with our inability to produce a stable RHAU truncation containing residues 151-209) would enable RSM-bound quadruplex to directly access the helicase active site. Alternatively, RSM binding to a quadruplex would enable positioning of the helicase domain on a single-stranded RNA region close by, and translocation through the quadruplex could occur. Whether quadruplex unwinding would ultimately require dissociation from the N-terminal extension of RHAU once the helicase domain is positioned and begins translocation remains an open question. The proposed model is consistent with the observation that RHAU can bind and unwind both DNA and RNA quadruplexes but does not directly provide insight into the substrate specificity of RHAU in terms of either orientation (parallel versus antiparallel) or molecularity (uni-, bi-, or tetramolecular) (17). Model confirmation and further mechanistic insights into quadruplex unwinding will await high resolution structure determination of protein comprising both the N-terminal and functional helicase domains.