The DEXH Protein Product of the DHX36 Gene Is the Major Source of Tetramolecular Quadruplex G4-DNA Resolving Activity in HeLa Cell Lysates*

G4-DNA is a highly stable alternative DNA structure that can form spontaneously in guanine-rich regions of single-stranded DNA under physiological conditions. Since a number of biological processes create such single-stranded regions, G4-DNA occurrence must be regulated. To date, resolution of tetramolecular G4-DNA into single strands (G4-resolvase activity) has been observed only in recombinant RecQ DNA helicases. We previously reported that human cell lysates possess tetramolecular G4-DNA resolving activity (Harrington, C., Lan, Y., and Akman, S. (1997) J. Biol Chem. 272, 24631–24636). Here we report the first complete purification of a major non-RecQ, NTP-dependent G4-DNA resolving enzyme from human cell lysates. This enzyme is identified as the DEXH helicase product of gene DHX36 (also known as RHAU). G4-DNA resolving activity was captured from HeLa cell lysates on G4-DNA affinity beads and further purified by gel filtration chromatography. The DHX36 gene product was identified by mass spectrometric sequencing of a tryptic digest from the protein band on SDS-PAGE associated with activity. DHX36 was cloned within a His6-tagging vector, expressed, and purified from Escherichia coli. Inhibition and substrate resolution assays showed that recombinant DHX36 protein displayed robust, highly specific G4-DNA resolving activity. Immunodepletion of HeLa lysates by a monoclonal antibody to the DHX36 product removed ca. 77% of the enzyme from lysates and reduced G4-DNA resolving activity to 46.0 ± 0.4% of control, demonstrating that DHX36 protein is responsible for the majority of tetramolecular G4-DNA resolvase activity.

G4-DNA is an alternative highly stable DNA structure forming within runs of guanine bases. It has been amply described previously (1). G4-DNA structures have the potential to disrupt normal duplex DNA; therefore, it might be expected that the genome would have minimized the usage of runs of deoxyguanosine. On the contrary, a growing body of data support the hypothesis that formation of G4-DNA in vivo is a recognized structural motif of specialized utility for key biological processes. Recent studies with a fluorescent G4-DNAbinding ligand, as well as a specific G4-DNA-binding protein, support the presence of G4-DNA structures located at human telomeres in vivo (2,3). In addition to the aforementioned telomeres, other guanine-rich regions in human DNA readily form G4-DNA structures in vitro and make up specific genetic control elements, including the immunoglobin heavy chain switch region (4), guanine-rich regions of ribosomal DNA (5), the d(pCGG) repeats of the fragile X mental retardation gene (6), promoters of proliferation-associated genes, such as the c-MYC (7,8), PDGF-A (9), RET (10), and the diabetes susceptibility locus IDDM2 promoter (11). It has been shown that a unimolecular G4-DNA structure has a repressor function in the c-MYC promoter (8). Compounds that stabilize G4-DNA in vivo have generated much interest because of their antitumor activity, suggesting that G4-DNA structures might be a general control motif utilized to inhibit cell growth, possibly by stabilizing repressor elements in growth related promoters (8,10,(12)(13)(14).
In addition to possible functional roles of G4-DNA, given the concentrations of nucleic acids in the cell and the single-stranded nature of DNA involved in replication, transcription, and recombination, it is likely that aberrant G4 structures form spontaneously in vivo. In this regard, it has been recently shown that a recombinant transcription construct containing the mammalian immunoglobulin S and S␥3 switch regions formed intramolecular G4 structures in the non-transcribed DNA strand of the transcription unit upon induction of transcription (15).
Whether G4-DNA structures are formed for biological purpose or by stochastic accident, once present, G4-DNA structures are highly stable. In order for such structures to be dynamic and rapidly resolved, enzymes involved in the recognition of G4-DNA structures and their resolution back into separate nucleotide strands would be of great importance. To date, certain members of the RecQ family of DNA helicases have been observed to catalyze the resolution of G4-DNA structures back to single-stranded DNA (G4-DNA resolvase activity) (16 -18). We previously reported that human cell lysates possess traceable tetramolecular G4-DNA resolvase activity that is separable from nuclease and double-stranded helicase activities (19). In this study, we have utilized G4-DNA affinity chromatography and mass spectrometry to purify and identify the major contributing protein enabling tetramolecular G4-DNA resolvase activity in HeLa cell lysates.

EXPERIMENTAL PROCEDURES
G4-DNA Substrate Generation and G4-DNA Resolution Activity Assay-To construct synthetic pure G4-DNA substrate for the activity assay or G4-DNA for affinity chromatography, a 1-mol synthesis of either gel-purified 5 (and 6)-carboxytetramethylrhodamine-labeled Z33 oligomer DNA sequence: 5Ј-AAAGTGATGGTGGTGGGGGAAGGATTTCGAACCT or 3Ј-biotinlabeled Z33 oligomer was purchased from Oligos Etc. The procedures for tetramolecular G4-DNA formation and our standard G4-DNA resolution assay have been described previously (19). To test the ability of recombinant DHX36 protein to resolve other known tetramolecular G4-DNA structures, we used a 1-mol synthesis of gel-purified TP sequence of the immunoglobin heavy chain switch region (5Ј-TGGACCAGACCTAGCAGCTATGGGG-GAGCTGGGGAAGGTGGGAATGTGA) and "rD4," 3 an oligonucleotide sequence from the human 28 S ribosomal RNA gene (5Ј-TTGAAAATC-CGGGGGAGAGGGTGTAAATCTCG, GenBank TM accession number M11167). After quadruplex DNA was formed and isolated as above, it was end-labeled with [␥-32 P]dATP by T4 kinase (Promega) according to manufacturer's instructions and purified by G-25 Sephadex Microspin TM columns (Amersham Biosciences).
Generation of G4-DNA-bound Streptavidin Paramagnetic Beads (GSPB)-Paramagnetic affinity beads were prepared by standard methods (see supplemental data).
Isolation of G4-DNA Resolvase from HeLa Cells-Approximately 0.5 g of frozen HeLa cell pellet were thawed at 37°C, and immediately an equal volume of ice-cold 2ϫ lysis buffer (100 mM Tris acetate, pH 7.8, 0.4 mM EDTA, 40 mM ␤ME, 0.02% Triton X-100, 20% glycerol, 50 g/ml leupeptin, 100 g/ml pepstatin; more recently we have found that leupeptin and pepstatin can be sub-* This work was supported by Grant 2-P30 CA12197 of the Comprehensive Cancer Center of Wake Forest Medical Center. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental data. 1 S. D. C. is a primary author of this publication along with J. P. V. and S. A. A. 2  stituted with Sigma protease inhibitor mixture) was added. Lysates were sonicated on ice by a Branson Sonifier 4500. Supernatants were clarified by microcentrifugation. Lysates were preincubated with competitor DNA by combining 0.95 ml of HeLa cell lysate, 2 ml of RSB, 100 AU of randomized 33 base pair oligomers (HHN) 11 , 10 l of 0.5 M EDTA and then incubating on ice for 10 min. Fifty l of GSPB were added, and material was incubated at 37°C for 10 min (with mixing every 3 min) followed by cooling on ice and pelleting by magnet. Lysate-bound GSPB were then washed consecutively in 1-ml volumes of RSB, resolvase high salt buffer (consisting of 50 mM Tris acetate, pH 7.8, 3.0 M NaCl, 2.5 mM MgCl 2, 10% glycerol), and finally RSB again. For ATP-dependent elution from beads, 60 l of ATP-dependent elution buffer (CEB, consisting of RSB ϩ 10 mM ATP) was prewarmed at 37°C and added to washed GSPB. The material was incubated at 37°C for 1 min, followed by addition of 6 l of 5 M NaCl and further incubation at 37°C for an additional minute. Beads were pelleted and supernatant collected. This elution process was repeated once. Gel filtration chromatography of ATP-dependent elution was done by standard methods on a tandem double Sephadex G-200 column with a running buffer of RSB made 300 mM in NaCl at a flow rate of 0.25 ml/min. One-half-ml fractions were collected; 5-l aliquots were tested for G4-DNA resolvase activity. Pooled reactions were precipitated in acetone and assayed by SDS-PAGE with colloidal Coomassie G-250 stain. Bands excised from SDS gels were sent to the Keck Center (Yale University) for tryptic digest and quadrupole time-of-flight (Q-TOF) mass spectrometry analysis followed by Mascot analysis for protein matches.
Cloning of DHX36 and the Isolation of Recombinant G4-DNA Resolvase-I.M.A.G.E. consortium clone 5273384 containing the DHX36 gene was purchased from Invitrogen Corp. The gene insert was corrected of errors and cloned into the PshAI-BamHI site of expression plasmid TriEx-4 (Novagen) by standard methods.
Recombinant proteins were produced by transformation of Escherichia coli strain Rosetta 2 (Novagen) with either TriEx-4 DHX36 or TriEx-4 vector alone. Recombinant proteins were initially purified by means of a His 6 tag by utilizing the TALON purification kit (Clontech). Rosetta 2 cell lysates were isolated and bound to TALON cobalt resin as recommended by the manufacturer. TALON resin (1 ml bed volume) was eluted twice with 1 ml of histidine elution buffer (0.7 M histidine, pH 6.0, 8.6 mM ␤ME, Sigma protease inhibitor mixture), followed by 1 ml of 250 mM EDTA, pH 6.0. For the second phase of purification, eluates were combined and incubated with GSPB in 0.4% ␣lactalbumin and 100 AU of 33-base randomized oligomer (HHN) 11 at 37°C for 10 min. Bound GSPB were washed two times in SSC (4ϫ) with 0.1% ␣lactalbumin. High purity recombinant DHX36 (20,000 units per g, units defined previously (19)) was obtained by ATP-dependent elution of GSPB as described for HeLa extracts, except 0.04% ␣-lactalbumin and Sigma protease inhibitor (instead of leupeptin and pepstatin) were added to the elution buffer.
G4-DNA Competition and Resolution Assays-Unlabeled G4-DNA, Y duplex DNA, 5Ј-overhang blunt duplex, and 3Ј-overhang blunt duplex were made utilizing oligomers that were previously used to study WRN helicase activity (20). G4-DNA used in competition experiments included tetramolecular Z33, TP, and ribosomal DNA. Serial dilutions of competitor structures were made from 300-fold down to an equimolar ratio of competitor to labeled G4-DNA. In standard 30-l reactions, competitor dilutions were added first to reaction tubes on ice. Next, one unit of recombinant DHX36 was added, and then the reaction was started by the addition of 0.2 pmol of 5 (and 6)-carboxytetramethylrhodamine-labeled G4-DNA. Reactions were incubated at 37°C for 30 min and stopped with 5 l of 200 mM EDTA.
Studies Involving Immunological Reagents-Monoclonal antibody generation, immunodepletion of HeLa cell lysates, and Western blotting were done by standard means (see supplemental data).

RESULTS
The Protein Product of the DHX36 Gene Is a G4-DNA Resolvase-In this study we report the isolation, purification, and identification of the major tetramolecular G4-DNA resolvase in HeLa cell lysates. Fig. 1A outlines the purification strategy used for isolating the protein responsible for G4-DNA resolvase activity. It is of note that we initially used G4-DNA affinity chromatography to bind activity without utilizing any other prior fractionation techniques, thus allowing activity to be captured by a comprehensive approach from complete lysates. Fig. 1B shows a standard G4-DNA resolvase activity assay using a previously characterized tetramolecular G4-DNA structure formed by self-annealed deoxyoligonucleotide Z33 as substrate (19). In Fig. 1B, lane 3, pure G4-DNA was exposed to HeLa cell lysates in the presence of ATP and MgCl 2 ; all the G4-DNA tetramolecular substrate of higher electrophoretic mobility was converted to monomer by enzymatic activity. In the absence of ATP or MgCl 2 , no G4-DNA conversion into monomeric DNA occurred (data not shown). Incubation of HeLa cell lysates with GSPB depleted total HeLa lysates of G4-DNA resolvase activity and in turn enriched GSPB with high levels of activity that could be assayed directly on GSPB (Fig. 1B, lane 4). G4-DNA resolving activity bound to GSPB in the presence of EDTA, and binding did not require activation of the enzyme with MgCl 2 and/or ATP. However, GSPB resolvase activity required MgCl 2 and ATP. Avid binding of G4-DNA resolvase activity to GSPB allowed us to utilize a stringent 2 M NaCl wash to strip beads of many other bound proteins while maintaining the majority of G4-DNA resolvase activity on the GSPB (data not shown).
ATP-dependent G4-DNA resolvase activity was selectively eluted from GSPB into solution by adding 10 mM ATP to beads and titrating the NaCl level to 300 mM (beads had previously seen up to 2 M NaCl). The asterisk over the arrow in Fig. 1A marks the point that bumped soluble GSPB eluate was analyzed by SDS-PAGE (Fig. 1C, lane 1). Although a number of prominent protein species were detected, an arrow in Fig. 1C marks an ϳ120-kDa protein band found to be associated with activity in subsequent gel filtration experiments. Once activity was substantiated, this actual protein band was used for mass spectrometry analysis. Fig. 1D shows a standard G4-DNA resolvase activity assay of ATP-dependent GSPB eluate fractionated by gel filtration chromatography on Sephadex G-200. G4-DNA resolvase assays of total fractionated eluate from gel filtration chromatography demonstrated only a single peak of resolvase activity, suggesting one major species of G4-DNA resolvase. G4-DNA resolvase activity was highest in fractions 13 and 14, and elution time correlated with a mass of ϳ120 kDa (determined using molecular weight standards). Gel filtration fractions were pooled in pairs (depicted in boxes labeled by Roman numerals in Fig. 1D) for analysis by SDS-PAGE. A polypeptide band at molecular mass ϳ120 kDa of peak staining intensity (Fig. 1E) correlated with the peak of G4-DNA resolvase activity observed in pool III. Q-TOF mass spectrometry measurements of tryptic digests of this protein species identified five peptide matches (see supplemental data) with the puta- tive DEXH box helicase DHX36 (21). The DHX36 protein has a predicted molecular mass of 115 kDa that correlates well with the observed molecular mass on SDS-PAGE of ϳ120 kDa. Examination of the DHX36 sequence revealed the protein to be a DEXH box helicase with putative NTPase activity and Mg 2ϩ dependence, both characteristics we had previously reported for G4-DNA resolvase activity in crude extracts (19).
In Fig. 2, the identification of DHX36 as a G4-DNA resolvase is confirmed by showing that expression and purification of recombinant DHX36 in E. coli is associated with robust G4-DNA resolving activity. The majority of material remaining after two rounds of purification was a protein of molecular mass 120 kDa ( Fig. 2A, lane 6). Q-TOF mass spectrometry of tryptic digests of the 120 kDa protein determined it to be recombinant human DHX36 (see supplemental data). No significant protein of 120 kDa was isolated by application of the purification scheme to lysates of cells transformed by TriEx-4 vector without insert ( Fig. 2A, lane 5). Purified recombinant DHX36 protein isolated from TriEx-4 DHX36-transformed E. coli had robust G4-DNA resolvase activity (Fig. 2B, lane 6); however, material purified from TriEx-4 (vector only) transformed E. coli had no detectable G4-DNA resolvase activity (Fig. 2B, lane 5). These data conclusively assign G4-DNA resolvase activity to DHX36.
DHX36 Protein Prefers Tetramolecular G4-DNA Structures to Watson-Crick Duplex DNA-An inhibition experiment (Fig. 3, A and B) demonstrates the substrate specificity of recombinant DHX36 product for G4-DNA compared with three duplex DNA helicase substrates. While equimolar unlabeled G4-DNA significantly diminished resolution of labeled G4-DNA, a 300-fold molar excess of a Y form duplex was required to have a similar inhibitory effect. A 300-fold molar excess of 5Ј-and 3Ј-overhangs had less of an inhibitory effect than equimolar, unlabeled G4-DNA. In other experiments, direct resolution of Y form duplex DNA required at least 27-fold more recombinant enzyme than was required for resolution of equimolar amounts of Z33 G4-DNA (data not shown). These data demonstrate that DHX36 has high specificity for the G4-DNA substrate in comparison to other duplex DNA substrates.
Additional experiments demonstrate (Fig. 3C) that substrate specificity of recombinant DHX36 protein product includes other tetramolecular G4-DNA structures. Both mammalian sequences of TP G4-DNA, generated from the immunoglobulin switch region, and human ribosomal G4-DNA effectively inhibit resolution of Z33 G4-DNA, whereas single-stranded deoxyoligonucleotide Z33 oligomer did not. Activity assays comparing the efficiency of resolution of these three substrates (Fig. 3D) demonstrated that ribosomal G4-DNA is readily resolved. In contrast, TP G4-DNA structure is only very slowly resolved. The fact that TP G4-DNA is an excellent inhibitory structure, yet is poorly resolved, suggests that DHX36 recognizes and binds TP G4-DNA well but resolves slowly.

DHX36 Protein Is an Abundant Protein in HeLa Cell
Lysates That Is Responsible for the Majority of G4-DNA Resolving Activity-To quantify levels of DHX36 protein in HeLa lysates and to determine the extent to which DHX36 gene product contributes to the total G4-DNA resolvase activity by specific immunodepletion of HeLa cell lysates, we developed a mAb to DHX36. Quantitative immunoblotting indicated that DHX36 protein accounts for 0.08 Ϯ 0.006% (mean Ϯ S.D., n ϭ 3) of protein mass present in total HeLa cell lysates (see supplemental data). Fig. 4A shows a Western blot analysis of HeLa cell lysate immunodepleted by anti-DHX36 protein product mAb compared with lysate immunodepleted by a IgG control mAb. Titrations of IgG-depleted lysate allow a direct comparison of DHX36 protein product levels between the mAb immunodepletions. When loading was normalized against ␤-actin, and DHX36 protein levels were compared with a linear plot of the IgG titrations, ϳ77% of DHX36 protein product was estimated to be removed from the lysates compared with the IgG control. In standard G4-DNA resolvase activity assays (Fig. 4B), the  1, 3, and 5) or TriEx-4 DHX36 (lanes 2, 4, and 6) transformed E. coli strain Rosetta 2. Shown are: total bacterial cell lysates (lanes 1 and 2), proteins bound by TALON cobalt affinity resin (lanes 3 and 4), followed by GSPB binding and ATP-dependent elution (lanes 5 and 6). The predominant 120-kDa protein observed in lane 6 was determined to be recombinant human DHX36 by band excision and Q-TOF mass spectrometry analysis of tryptic digests. B, G4-DNA resolvase assay demonstrates that protein purified from TriEx-4 DHX36 transformed Rosetta 2 has robust activity (lane 6), while TriEx-4 (vector only) proteins purified identically have no G4-DNA resolvase activity (lane 5). G4, quadruplex form Z33. Z, Z33 monomer. FIGURE 3. Inhibition assays and activity assays show that recombinant DHX36 has a strong preference for quadruplex structures. A, an illustrative non-denaturing 10% polyacrylamide gel demonstrating G4-DNA resolvase assay with increasing amounts of unlabeled competing DNA structures inhibiting resolving activity. Lanes 1-7 correspond to 0, equimolar, 3-fold, 10-fold, 30-fold, 100-fold, or 300-fold molar excess of unlabeled competitor, respectively. B, graphic representation of change in G4-DNA resolvase activity caused by unlabeled competitors (mean Ϯ S.D. of percent change in activity compared with no competitor, reactions set up in triplicate; SD Ͻ1% in all cases). Percent of activity is measured by the amount of monomer produced with competitor normalized to amount of monomer formed without competitor. C, graphic representation of a second inhibition assay (gels not shown) with other unlabeled G4-DNA structures TP and rD4 as well as single-stranded Z33 as competitors. D, non-denaturing 10% polyacrylamide gel demonstrates G4-DNA resolvase activity on other quadruplex structures with recombinant DXH36 protein.  lysate specifically immunodepleted for the DHX36 protein had 46.0 Ϯ 0.4% (mean Ϯ S.D., n ϭ 3) of the G4-DNA resolvase activity as did lysates immunodepleted with a control IgG. Thus, DHX36 protein product is the major, but not the sole, G4-DNA resolvase present in HeLa cell lysates. Extrapolation of these data suggests that 100% immunodepletion of DHX36 protein product removes ϳ70% of the G4-DNA resolvase activity from HeLa lysates. Therefore, about 30% of G4-DNA resolvase activity of HeLa cell lysates remains unidentified.

DISCUSSION
Since the Sen and Gilbert's initial observation of tetramolecular G4-DNA (4), its remarkable stability has raised a conundrum concerning how it might be dynamically resolved when formed in a cell. Candidate proteins to fulfill that role include RecQ helicases, a G4-specific nuclease for both DNA and RNA quadruplex that has been found in yeast and identified as KEM1/SEP1 (22) and an analogous DNA quadruplex nuclease activity has been found in human Raji cells (23). The protein responsible for the latter activity has yet to be identified. Yeast and mammalian proteins have been found to bind G4-DNA by affinity chromatography approaches (24,25), but to our knowledge the DHX36 protein product is the first NTP driven G4-DNA resolving enzyme to be isolated directly from human cell lysates. Thus this protein may hold a key role for catalytically destabilizing quanine quadruplex structures. Since G4-DNA resolving activity may be a primary role for this protein, we propose to add the name G4 Resolvase-1 (G4R1) for the DHX36 gene product, which is also known as RHAU and MLEL1. Mammalian RecQ helicases are ATP-dependent enzymes capable of resolving certain G4-DNA structures in vitro and have been identified by genetic linkage studies from different cancer and aging syndromes that carry their name. It is noteworthy that RecQ family proteins were not identified by our G4-DNA affinity enrichment approach; this may be due to loss of RecQ proteins during the high salt wash of G4-DNA affinity beads where a detectable but small contribution of activity is lost. However, immunodepletion experiments show that G4R1 makes a greater contribution to G4-DNA resolving activity in HeLa lysates than the RecQ protein members.
The identity of DHX36 as being responsible for a major G4-DNA resolvase activity may shed light on its biological function. The motifs of the gene have been discussed previously (26) and are typical of DEAD/DEAH box helicases. These helicases can be primarily active on DNA or RNA or both nucleic acids. The closest homologous human protein to DHX36 by BLAST comparison is the DEXH protein nuclear DNA helicase (NDH II), which has 35% identity with DHX36 (27). This protein has roles in facilitating enhanced transcription, binding both DNA and RNA, in part through association with RNA pol II (28 -30). Interestingly, another DEXH helicase termed dog-1 from the nematode Caenorhabditis elegans appears important for maintenance of certain long guanine-rich DNA tracts and has been suggested to resolve G4-DNA arising on the parental lagging strand of DNA replication, although biochemical specificity of dog-1 activity awaits characterization (31). One mouse homologue of dog-1 termed Rtel (Regulator of telomere length) was found to induce loss of telomeres and a high frequency of chromosomal fusions when knocked out in mouse ES cells (32). If both Rtel and DHX36 protein helicases are indeed recognizing G4-DNA structures, certain DEXH proteins may represent a new family of DNA caretaker enzymes specializing in preventing quadruplex G4-DNA induced mutations.
Only one prior investigation in the literature explores the function of the DHX36 protein (26). The DHX36 protein product was isolated from HeLa cell lysates as part of a protein complex that bound proteins to the AU-rich element of urokinase plasminogen activator mRNA (and termed RHAU for RNA helicase associated with AU-rich element) including the exosome complex responsible for 3Ј-5Ј degradation of AU-rich RNAs (26). Overexpression of RHAU was associated with accelerated degradation of certain AU containing mRNAs (26). Immunohistochemical staining indicated that the majority of HeLa cell DHX36 is nuclear, although a splice variant of RHAU had primarily cytoplasmic staining (26). This work does suggest that the cytoplasmic variant of DHX36 protein may have a key function in processing AU mRNA metabolism. DHX36 mRNA expression in human organs is highest in testis (33), which is consistent with the protein being utilized in cells with high rates of proliferation. High levels of DHX36 may be needed by rapidly growing cells, including certain tumor types. Regions of DNA that become single-stranded during replication and transcription are vulnerable to forming G4-DNA structures. With the knowledge that the DHX36 protein product (G4R1) is a major G4-DNA resolvase, we are conducting experiments to monitor sequences that are likely to form G-quadruplex structures when G4R1 activity is conditionally knocked out. These experiments will give us a better understanding of the biological function of G4R1 and a better understanding of the roles and extent of G-quadruplex formation in the cell.