Recognition and Unfolding of c-MYC and Telomeric G-Quadruplex DNAs by the RecQ C-Terminal Domain of Human Bloom Syndrome Helicase.

G-quadruplex (G4) is a noncanonical DNA secondary structure formed by Hoogsteen base pairing. It is recognized by various DNA helicases involved in DNA metabolism processes such as replication and transcription. Human Bloom syndrome protein (BLM), one of five human RecQ helicases, is a G4 helicase. While several studies revealed the mechanism of G4 binding and unfolding by the conserved RecQ C-terminal (RQC) domain of BLM, how RQC recognizes different G4 topologies is still unclear. Here, we investigated the interaction of Myc-22(14/23T) G4 from the c-Myc promoter and hTelo G4 from the telomeric sequence with RQC. Myc-22(14/23T) and hTelo form parallel and (3+1) hybrid topologies, respectively. Our circular dichroism (CD) spectroscopy data indicate that RQC can partially unfold the parallel G4, even with a short 3' overhang, while it can only partially unfold the (3+1) hybrid G4 with a 3' overhang of 6 nucleotides or longer. We found that the intrinsic thermal stability of G4 does not determine RQC-induced G4 unfolding by comparing T m of G4s. We also showed that both parallel and (3+1) hybrid G4s bind to the β-wing region of RQC. Thermodynamic analysis using isothermal titration calorimetry (ITC) showed that all interactions were endothermic and entropically driven. We suggest that RQC partially unfolds the parallel G4 more efficiently than the (3+1) hybrid G4 and binds to various G4 structures using its β-wing region. By this information, our research provides new insights into the influence of G4 structure on DNA metabolic processes involving BLM.

Introduction

Nucleic acids can form noncanonical structures other than conventional duplexes, including quadruplexes, triplexes, i-motifs, and pseudoknots. Wobble and Hoogsteen base pairing, in addition to Watson–Crick base pairing, are used to form noncanonical DNA structures. Among these, G-quadruplexes (G4s) are frequently formed within guanine-rich sequences.[1−3] G4s are two or more stacked G-tetrads composed of four guanines using Hoogsteen base pairing in one plane with monovalent cations such as sodium and potassium localized in the center of each plane.[4] The structure and conformation of G4s are quite heterogeneous based on the polarity of the strands and differences in the loops between G-tetrads.[5,6] Tetramolecular and bimolecular G4s can be formed by four short strands and two strands, respectively, and intramolecular G4s can be generated by one long DNA sequence.[7,8] If all contributing DNA strands have their backbones in the same orientation (5′ to 3′), the G4 is defined as parallel; conversely, if the two strands are in the same direction while two are in the opposite direction, the structure is antiparallel. The structures with three in the same direction and one in the opposite are called as “(3+1) hybrid”. The type and concentration of monovalent cations at the center of the plane also affect the stability of the G4.[9,10]

In the human genome, there are many (approximately 376 000) potential G4-forming sequences, mostly in the promoter and telomere regions.[11] The G4 structures formed by these sequences are thought to be diverse topologies in vivo. Because of G4s’ high stability and location in the genome, they must be resolved for replication or transcription to take place. In this way, G4s are involved in the critical processes of DNA metabolism.[12,13] In particular, G4s in oncogene promoters and telomere regions have attracted interest from researchers because unregulated G4 unfolding could trigger cancer progression.[12] For the proper handling of the heterogeneous G4 structures, cells can use appropriate helicases for each G4 topology in DNA metabolism. It was shown that an Saccharomyces cerevisiae DNA helicase, Pif1, preferentially unfolds the antiparallel G4 rather than a parallel G4 having similar intrinsic stability.[14]

Bloom syndrome protein (BLM) is a G4 helicase and one of the five human RecQ helicases along with RECQ1, WRN, RECQ4, and RECQ5.[15] Mutations in the BLM protein cause Bloom syndrome, a recessive genetic disorder with symptoms of premature aging and a predisposition to various types of cancer.[16] BLM unwinds DNA in the 3′ to 5′ direction using the energy of ATP hydrolysis and maintains gene integrity by regulating DNA replication, recombination, and repair.[17] In addition to canonical duplex DNA, BLM binds to several kinds of noncanonical DNA structures such as Holliday junctions, D-loops, and G4s, and it preferentially binds to G4s compared to other noncanonical DNA structures.[18,19] BLM is composed of several domains: the helicase domain, RecQ C-terminal (RQC) domain, and the helicase and RNase D C-terminal (HRDC) domain[19−21] (Figure A). Among these, the RQC domain has been established as the main DNA binding module.[20,22]

Figure 1

(A) Schematic drawing of the human BLM domain structure. (B) G-quadruplex sequences used for the current study. Myc-2345 is from the c-Myc promoter G-quadruplex, which adopts a parallel topology, and Myc-22(14/23T) is the major loop isomer of Myc-2345. hTelo is from a human telomere sequence and hTelo-3nt/6nt/15nt is its derivative. (C) Schematic representations of hTelo-3nt and Myc-22(14/23T). hTelo-3nt forms a (3+1) hybrid G4. Myc-22(14/23T) forms a propeller-type parallel G4.

Previous studies have characterized the G4 structure and its conformational changes using circular dichroism (CD) spectroscopy.[23−26] CD signals reflect the structural elements of G4s such as stacking, strand orientation, and loop arrangements.[27] By monitoring the maximum molar ellipticity change, G4 unfolding by the titration of chemical compounds (TAP, TmPyP4, etc.)[23,24] and proteins (UP1, Prion, etc.)[25,26] has been reported. Single-molecule Förster resonance energy transfer (FRET) has been used to study the unfolding dynamics of G4 in the presence of BLM. The helicase core of BLM (residues 642–1290) can effectively unfold a G4 by different mechanisms, depending on the structural environment.[28] Moreover, even without the helicase domain, the Zn-RQC–HRDC construct (858–1298) can unfold the G4 structure.[29] Also, a significant FRET population change from high FRET to low FRET was observed by the RQC–HRDC domain (1066–1298) addition to G4, which indicates that RQC–HRDC can induce partial G4 unfolding.[29] In these studies, G4s have adopted either parallel or (3+1) hybrid structures. Our previous nuclear magnetic resonance (NMR) study showed that RQC without the HRDC domain can destabilize the parallel G4 structure.[30] Previous research also revealed that G4 unfolding by BLM depends on the length of the 3′ overhang of G4 DNA. The BLM core (residues 642–1290) cannot unfold (3+1) hybrid G4s with 3′ overhang lengths shorter than 6 nucleotides (nt). For 6 nt overhangs, only ∼10% of G4 structures were unfolded, whereas for 15 nt overhangs, ∼50% unfolding was observed by FRET.[31]

Even though these details of BLM–G4 interaction and G4 unfolding by BLM were studied, it is still not known how BLM, and specifically the RQC domain, recognizes and unfolds different G4 topologies. In this study, we investigated the interaction of RQC with (3+1) hybrid G4 structures with different 3′ overhang lengths and compared it to the parallel G4–RQC interaction. Figure B,C shows the G4-forming sequences used in this study. hTelo-3nt, which forms a (3+1) hybrid G4, is a derivative of hTelo, which is based on the human telomeric sequence. We added one thymine to the 3′ overhang of the original hTelo described in a previous study[32] to ensure a 3′ overhang equal to that of Myc-22(14/23T). The structure of hTelo and the chemical shifts of its imino protons were determined by previous research.[32] Myc-22(14/23T) is a derivative of Myc-2345 from the c-Myc promoter with a highly stable parallel G4 conformation (Figure C). Previous structural studies revealed the properties of the G4s formed in the c-Myc promoter, including Myc-2345 and its derivative Myc-22(14/23T).[8,33] The imino proton chemical shifts were determined in a previous study,[33] and the RQC–Myc-22(14/23T) interaction was previously studied by NMR spectroscopy.[30]

Here, we monitored the changes in the CD spectra of each G4 during RQC titration and studied corresponding interactions with isothermal titration calorimetry (ITC) to investigate thermodynamic characteristics. Our CD data indicate that RQC more effectively unfolds the parallel G4 (Myc-22(14/23T)) than the (3+1) hybrid G4 (hTelo-3nt) with the same 3-nt 3′ overhang, even though the parallel G4 is more thermally stable. RQC only partially unfolds the hTelo G4 with 3′ overhangs of 6 nt or longer. By analyzing changes in the 1H–15N heteronuclear single quantum coherence (HSQC) spectra of RQC with increasing concentrations of G4, we determined that both parallel and (3+1) hybrid G4s bind to the β-wing region of RQC. In summary, we propose that, although the G4 binding mode of BLM RQC is the same for both the (3+1) hybrid and parallel structures, the parallel G4 is more susceptible to unfolding upon interaction with RQC. Our study provides insights into the influence of G4 structures on DNA metabolic processes involving the BLM helicase.

Results

Structure of hTelo Derivative G4s

We first confirmed whether the hTelo derivatives (3, 6, and 15 nt overhangs) used in this study have the same conformation as the originally characterized hTelo G4 using CD and NMR spectroscopies.[32]Figure A,B shows the CD and 1D 1H NMR spectra of hTelo-3nt. CD data on hTelo-3nt showed a positive band at 290 nm and a negative band at 240 nm, which is typical of (3+1) hybrid G4s.[34] Also, the imino proton NMR spectrum was identical to the previously reported one (hTelo in Figure B).[32] Because the imino protons’ chemical shifts are very sensitive to G4 topology,[35−37] it means that hTelo-3nt, which has one dT more at the 3′, maintains the same (3+1) hybrid geometry.[32] The hTelo-6nt and hTelo-15nt G4s also showed similar CD and 1H 1D NMR spectral features (Figure C–F) to hTelo-3nt. These data suggest that all three hTelo G4s mainly have (3+1) hybrid topology.

Figure 2

CD spectrum (left); imino proton region of 1D 1H NMR spectrum (middle); and schematic structure (right) of (A) hTelo-3nt, (B) hTelo-6nt, and (C) hTelo-15nt.

CD Analysis of Conformational Changes in G4s Induced by RQC Binding

A previous report showed that RQC induces partial unfolding of the Myc-22(14/23T) G4 by analyzing hydrogen–deuterium exchange data of the imino protons in the middle plane of the G4.[30] While this is an effective way to investigate the stability of the hydrogen bonds, the changes induced in the G4 stacking by the unfolding can also be directly observed with CD spectroscopy. The CD spectra of Myc-22(14/23T) showed that the main conformation is a parallel G4, which is typified by a positive band centered at 260 nm and a negative band centered at 240 nm (Figure S1).[34]Figure A,B shows the change in the CD spectrum of Myc-22(14/23T) upon the addition of RQC. The molar ellipticity at 260 nm was found to decrease in a protein-concentration-dependent manner. At [DNA]/[RQC] = 1:2, the molar ellipticity at 260 nm decreased to 64% of the original value. The decrease continued until [RQC] = 60 μM ([DNA]/[RQC] = 1:3), beyond which no further reduction was observed. The molar ellipticity at 260 nm was 55% of the original value at this concentration. Because the CD signal of RQC itself was zero from 250 to 300 nm (Figure S2), the increasing [RQC] in the sample did not contribute to the signal changes in this region. Therefore, it can be considered that expanded spectra (Figure B,D,F,H) are the same as the net change of G4 signal, and the contribution of RQC is eliminated. These results were consistent with the conclusion that RQC binding induces partial unfolding of the G4 structure.

Figure 3

(A) CD spectra of 20 μM Myc-22(14/23T) in the presence of 0–80 μM RQC. (B) Expanded view of CD spectra from (A) in the range of 250–300 nm. (C) CD spectra of 60 μM hTelo-3nt in the presence of 0–120 μM RQC. (D) Expanded view of CD spectra from (C) in the range of 250–310 nm. (E) CD spectra of 60 μM hTelo-6nt in the presence of 0–120 μM RQC. (F) Expanded view of CD spectra from (E) in the range of 250–310 nm. (G) CD spectra of 60 μM hTelo-15nt in the presence of 0–120 μM RQC. (H) Expanded view of CD spectra from (G) in the range of 260–310 nm.

We also monitored CD spectral changes of the hTelo-3nt G4 upon the addition of RQC. Figure C,D shows the change of the CD spectrum of hTelo-3nt upon RQC addition. Surprisingly, unlike Myc-22(14/23T), the positive band at 290 nm showed no change even after 2 molar equivalents of RQC were added. It implies that the hTelo-3nt maintains its stacking interaction in the presence of RQC. Previous research shows that the BLM core can unfold telomeric G4s with 3′ overhangs of 6 nt or longer.[31] Therefore, we performed a CD analysis of hTelo-6nt and hTelo-15nt. Figure E,F shows that the molar ellipticity at 290 nm of hTelo-6nt G4 decreased in a concentration-dependent manner. At [DNA]/[RQC] = 1:2, the maximum molar ellipticity value decreased to 74% of the ellipticity value in DNA-only spectrum. For hTelo-15nt, molar ellipticity at 290 nm was also decreased upon RQC addition. At [DNA]/[RQC] = 1:2, the maximum molar ellipticity value decreased to 49% of the ellipticity value in DNA-only spectrum (Figure G,H). To confirm the effect of longer 3′ overhang on the unfolding of RQC, we performed the CD experiment using Myc-15nt G4. An additional 12 dTs were added to Myc-22(14/23T) to make Myc-15nt. At [DNA]/[RQC] = 1:2, the molar ellipticity of Myc-15nt at 260 nm decreased to 57% of the original value, while Myc-22(14/23T) decreased to 64% of the original value (Figure S3). This is consistent with CD results using hTelos where a longer 3′ overhang induces more progressive unfolding of RQC. Figure S4 shows the residual molar ellipticity of each G4 structure at [DNA]/[RQC] = 1:2. These data suggest that RQC interacts with and partially unfolds parallel G4s, but not (3+1) hybrid G4s, but that addition of a long 3′ overhang facilitates an interaction between the (3+1) hybrid G4 and RQC that results in partial unfolding.

Melting Temperatures of Each G4

Our CD data on Myc-22(14/23T) and hTelo G4s with RQC raised the question of how the intrinsic stability of each G4 DNA contributes to the unfolding by RQC. We hypothesized that a G4 that is inherently more unstable is more likely to be unfolded by RQC. To investigate this possibility, the thermal stability of each G4 was analyzed by measuring its melting temperature using CD spectroscopy. The maximum molar ellipticity of each G4 over a temperature range of 25–90 °C was fitted to a Boltzmann sigmoidal curve to obtain the melting temperature (Figure ). For the hTelo G4s, as the 3′ overhang became longer, the melting temperature slightly decreased (Table ), indicating that the extra nucleotides at the 3′ end affected the G4 stability. The partial unfolding of hTelo G4(-6nt and -15nt) by RQC monitored by CD spectroscopy could be related to their intrinsic thermal stabilities. For Myc-22(14/23T), a much higher melting temperature of 78.6 °C was estimated compared to those of the hTelo G4s (melting temperature range from 53.1 to 57.3 °C). This implies that Myc-22(14/23T) is much more intrinsically stable than hTelo G4s. These results suggest that the preferred unfolding function of the RQC to parallel G4 over the (3+1) hybrid ones is not related to the intrinsic stability of the G4 DNA.

Figure 4

Boltzmann sigmoidal plot of the normalized maximum molar ellipticity of each G4 vs temperature.

Table 1

Melting Temperature of Each G4

G4	Tm (°C)
hTelo-3nt	57.2 ± 0.3
hTelo-6nt	54.5 ± 0.4
hTelo-15nt	53.1 ± 0.3
Myc-22(14/23T)	78.6 ± 4.4

Identification of Binding Surface on RQC

We monitored the changes in the backbone amide signals of RQC upon addition of the (3+1) hybrid G4 by 1H–15N HSQC spectra to investigate the binding surface of RQC toward (3+1) hybrid G4. We used hTelo-3nt to investigate the binding surface on RQC because it is the least unfolded one. hTelo-6nt and -15nt are unfolded by RQC and have higher molecular weights than hTelo-3nt. They could contribute to either disappearance of peaks or perturbation of other regions besides the initial binding area. The binding surface of RQC for Myc-22(14/23T) was revealed by a previous study.[30] Peaks were assigned based on previously published data.[22]Figure A shows 1H–15N HSQC spectra of RQC during titration of hTelo-3nt at 25 °C. Several peaks, most prominently N1162, were significantly perturbed, and several peaks disappeared upon addition of 2 equiv of G4. Figure B shows the histogram of the average chemical shift perturbations (Δδavg) vs residue number. Residues in the N-terminus, C-terminus, and N1162 in the β-wing region had Δδavg values higher than two standard deviations above the average. V1103, N1164, and residues located in the terminal region had Δδavg higher than one standard deviation above the average. Perturbations in the terminal regions are likely by the allosteric effects of G4 binding rather than direct interaction with G4 because both terminals are flexible unstructured regions.[38] Significantly perturbed residues including V1103, N1162, and N1164 were mapped on the solution structure of RQC (Figure C) (PDB ID: 2MH9).[22] N1162 and N1164 are located in the β-wing region, and V1103 is located in the α1−α2 loop. It is noteworthy that both β wing and α1−α2 loop regions were identified as the essential regions for duplex DNA and parallel G4 binding by previous research.[20,30,39] In addition, V1103, N1162, and N1164 were the residues most perturbed by the addition of Myc-22(14/23T).[30]

Figure 5

(A) Overlaid 1H–15N HSQC spectra of 15N-labeled RQC at increasing molar ratios of hTelo-3nt. (B) Chemical shift perturbations (Δδavg) of 15N-labeled RQC induced by 1.0 equiv of hTelo-3nt. Residues with broadened cross-peaks upon hTelo-3nt addition are shaded in purple. The dotted lines indicate one and two standard deviations higher than the average. (C) Mapping of residues affected by G4 binding on the solution structure of RQC (PDB ID 2MH9). Residues perturbed by two standard deviations above the average are marked in red and those perturbed by one standard deviation above the average are marked in blue.

Thermodynamic Parameters of RQC Binding to G4

To further investigate the RQC–hTelos interaction’s thermodynamic properties, ITC experiments were employed. Figure A shows the binding isotherm of RQC with hTelo-3nt. Figure B,C shows the results of RQC binding to hTelo-6nt and hTelo-15nt, respectively. We measured the heat of dilutions separately, and the heat of dilution was subtracted from the integrated heat change. The resulting data were fitted to the nonlinear regression fit (Figure S5). The thermodynamic parameters, including dissociation constants (Kds), are listed in Table . Our results showed that all Kds of RQC with hTelos were in the micromolar range. The apparent Kd of hTelo-3nt with RQC is the lowest among the G4s that we tested (0.59 ± 0.06 μM). Kd values increased as the 3′ overhangs became longer: the Kd of hTelo-6nt-RQC was 1.3-fold larger and that of hTelo-15nt-RQC was 2-fold larger than the hTelo-3nt-RQC complex. Also, the enthalpy and entropy values increased as the 3′ overhangs got longer. While the interaction was endothermic throughout the experiment for hTelo-6nt and hTelo-15nt, the hTelo-3nt–RQC interaction was endothermic at earlier titration points and changed to exothermic at later points. ΔG of each interaction was calculated using ΔH and ΔS values obtained from the ITC experiment (Table ). The results of the ITC experiment using RQC domain and Myc-22(14/23T) have been published previously. The reaction was also endothermic and assumed to be an entropically driven process.[30]

Figure 6

ITC binding isotherms of (A) hTelo-3nt, (B) hTelo-6nt, and (C) hTelo-15nt titrated into RQC. Raw heat data with the subtraction of the heat of dilution (top) and integrated heat data with the nonlinear regression fits (bottom) are shown.

Table 2

Thermodynamic Parameters Obtained by ITC

	ΔH (kJ/mol)	ΔS (J/mol/K)	ΔG (kJ/mol)	Kd (μM)	n
hTelo-3nt	14.9 ± 2.0	169.3	–35.6	0.59 ± 0.06	0.54 ± 0.03
hTelo-6nt	16.8 ± 2.3	172.9	–34.7	0.80 ± 0.12	0.62 ± 0.06
hTelo-15nt	21.2 ± 2.6	184.6	–33.8	1.18 ± 0.08	0.48 ± 0.06

Discussion

Previous reports showed that the BLM RQC domain is the main binding module for duplex and G4 DNA and that the BLM core without helicase domain (858–1298) can unfold the G4 in an ATP-independent manner.[20,29,39] Moreover, RQC can partially unfold the parallel G4 structure in the absence of other domains.[30] However, whether this binding and unfolding by the RQC domain has any difference for the type of G4 structure has not yet been revealed.

In this study, using CD spectroscopy, we investigated partial disruption of (3+1) hybrid G4s by RQC. It is expressed quantitatively using the remaining molar ellipticity value in CD spectra. First, we confirmed that RQC could not unfold the (3+1) hybrid G4 with a short 3′ overhang (3 nt) and only partially unfolded the (3+1) hybrid G4s with long 3′ overhangs (≥6 nt). This is in line with previous research showing that the BLM core can unfold the telomeric G4 with 3′ overhangs of 6 nucleotides or longer.[31] Also, as the 3′ overhang got longer, the melting temperature of the (3+1) hybrid G4 was decreased, which indicates that it became more unstable. Previously, it has been shown that adding extra nucleotides to the G4 terminal makes the structure less stable.[40] Because RQC does not make additional contact with the ssDNA overhang of the DNA duplex in the crystal structure,[20] it is reasonable that the intrinsic G4 stability affects the extent of the unfolding by RQC in the (3+1) hybrid G4 case.

As shown through CD spectroscopy, RQC–G4 bindings that we examined in this study are associated with the conformational change of the G4. It implies that at least two events such as binding and unfolding occur during the titration of G4 into RQC. However, our ITC data only show a combined binding isotherm of all events at each titration, which was well-fitted to one event binding model. It is likely to contain the sum of all events occurring during each titration.[41] The previous study showed that reverse titration separated heat isotherms of each event.[42] Unfortunately, we could not obtain the separated isotherms of the binding and unfolding event by the titration of RQC into hTelo G4s (data not shown). Hence, we consider that the thermodynamic parameters obtained in this study reflect a combination of all molecular events (binding and partial unfolding). We suggest a thermodynamic model for the RQC interaction with hTelo-6nt or -15nt to include two events: (1) the initial binding of RQC with the G4 and (2) the coupling of the RQC binding energy to the RQC-induced unfolding of the G4 (Figure A). In this respect, the overall ΔG estimated from our ITC data could be considered as the sum of ΔG of the G4–RQC binding (ΔGb) and ΔG of the partial unfolding of G4 (ΔGu) (Table and Figure ). Because hTelo-3nt seems to be rarely unfolded by RQC, the ITC-measured ΔG value can be considered as ΔGb. If we assume that the ΔGbs for hTelo-6nt or -15nt is similar to that of hTelo-3nt because the 3′ overhang does not affect the G4 core structure, then ΔGu of hTelo-6nt and hTelo-15nt can be calculated as +0.83 and +1.74 kJ/mol, respectively. As we mentioned above, the required energy for the partial G4 unfolding could be supplied by the coupling of the RQC binding energy. It is noteworthy that BLM hydrolyzes ATP to support complete and processive G4 unfolding.[17] Our study presented that the RQC domain can provide binding energy for the partial unfolding of G4 in specific cases.

Figure 7

(A) Schematic drawing of Gibbs free energy state of each reaction coordinate. “Q” means folded G4 (quadruplex), “P” means RQC (protein), and “I” means the partially unfolded G4 (intermediate). ΔGb is the ΔG of RQC–G4 binding and ΔGu is the ΔG of G4 partial unfolding. (B) Comparison of ΔG values of each G4 interaction with RQC.

We observed that the interaction between RQC and hTelo-3nt was endothermic at earlier titration, then changed to exothermic at later titration in ITC experiments. Similar ITC isotherms were reported in many previous studies and considered as a result of nonspecific protein–ligand (DNA or other protein) interactions.[43−46] In the aspect that RQC interacts with various DNA structures such as duplex, D-loop, Holliday junction, and G4s,[18−20] it is reasonable that RQC does not have perfectly complemented interface for specific DNA. Consistently, RQC could interact with both the parallel and (3+1) hybrid G4 with the same binding surfaces as revealed by our NMR data.[30] In the case of hTelo-6nt and -15nt with RQC, positive heat changes throughout the titrations were observed. The unfolding process after the initial binding could contribute to the positive enthalpy and entropy change by the cation release when the hydrogen bond breaks in the guanine plane.[30,43]

Contrary to hTelo-3nt, Myc-22(14/23T) could be partially disrupted by RQC. The decreased molar ellipticity in the CD spectrum (Figure A,B) is in line with the accelerated deuterium exchange of the guanines in the middle plane with RQC.[30] These imply that RQC induces the disruption of stacking interactions and hydrogen bonds. It is remarkable because Myc-22(14/23T) G4 is intrinsically more stable than hTelo-3nt, as revealed by CD melting experiments.

Even though more investigation is necessary to answer why Myc-22(14/23T) shows preferential unfolding by RQC, the preferred unfolding of the parallel G4 by RQC could be physiologically meaningful. It has been revealed that many of the G4s observed in promoter regions, including those of c-myc, KRAS, VEGF, HIF-1α, and PDGF-A, have parallel topology.[8,33,47−52] The G4 structures at transcription start sites and the first introns were shown to be the genomic target for BLM and BLM-dependent transcription regulation.[53] Based on this, we would predict that RQC’s preferential unfolding of parallel G4s would lead to the preferred activity of BLM during transcription.

In summary, our study provided the data on the interaction between BLM RQC and (3+1) hybrid G4s. CD spectroscopy showed that RQC could unfold a parallel G4 partially and (3+1) hybrid G4s with an extended 3′ overhang, while RQC did not unfold a (3+1) hybrid G4 with a short 3′ overhang. NMR data showed that the β wing of RQC is commonly used for interacting with both the parallel and (3+1) hybrid G4 structures. These data will expand our understanding of the initial recognition and interaction between BLM and G4 DNA, which is an essential step in the regulation of DNA metabolism, especially transcriptional control within promoter regions.

Materials and Methods

Sample Preparation

The unlabeled BLM RQC (residues 1067–1210) and 15N-labeled BLM RQC were expressed and purified as previously described.[30] All of the DNA sequences were purchased from IDT Inc. and dissolved in 20 mM Tris, 100 mM KCl, and pH 7.0 buffer to prepare the G4 solution at 1 mM. The dissolved G4s were heated to 95 °C for 10 min and cooled to room temperature for 1 h. Samples were stored at 4 °C after cooling.

CD Spectroscopy

Unless stated otherwise, all of the spectra were collected from 200 to 320 nm at a scanning speed of 100 nm/min and with a spectral bandwidth of 2 nm for each sample, and the spectra were observed after overnight incubation of samples. The average of three scans was recorded. For the Myc-22(14/23T) sample, spectra were collected from 200 to 300 nm.

For investigating the melting temperature of G4, a JASCO J-1500 CD spectrometer (KBSI, Ochang) was used. The concentration of four G4s was adjusted to have a maximum molar ellipticity value of 20 mdeg. The sample was heated from 25 to 90 °C at a rate of 1 °C/min. The maximum molar ellipticity values at 260 nm for Myc-22(14/23T) and at 290 nm for hTelo were normalized and fitted to the Boltzmann sigmoidal curve (eq ) using Origin 2019 software where dx implies the slope of the curve and describes the steepness of the curve. Y is the molar ellipticity value, x is the temperature, and V50 is the melting temperature. Error bar shows the standard error of the 95% confidence intervalTo examine G4 unfolding by RQC, the prefolded Myc-22(14/23T) concentration was fixed at 20 μM, and BLM RQC was added to yield molar ratios of 0.05, 0.1, 0.2, 0.4, 0.5, 1, 1.5, 2, 2.5, 3, 3.5, and 4 in the 20 mM Tris, 100 mM KCl, and pH 7.0 buffer. CD experiments were performed using a JASCO J-815 CD spectrometer (GIST, Gwangju) at 20 °C. For hTelo (3, 6, and 15 nt overhang), the prefolded G4 concentration was fixed at 60 μM, and BLM RQC was added to yield molar ratios of 0, 0.5, 1, 1.5, and 2 in the same buffer.

NMR Spectroscopy

NMR experiments were performed using a Bruker Avance II 900 MHz spectrometer equipped with a cryogenic probe (Korea Basic Science Institute, Ochang). 1D proton spectra of 0.3 mM hTelo-3nt G4 were obtained at 25 °C. 1H–15N HSQC spectra of 0.3 mM 15N-labeled RQC in the absence or presence of G4s were also obtained at 25 °C. All NMR data were processed with Topspin (Bruker) and analyzed with SPARKY software. The following equation was used to calculate the average chemical shift perturbation values (Δδavg). Δδavg values higher than one standard deviation above the average are selected as the significantly perturbed residues

Isothermal Titration Calorimetry

A Nano-ITC SV instrument (GIST, Gwangju) was used for the ITC experiments. Aliquots of a highly concentrated G4 solution (400 μM) were titrated into a diluted 40 μM solution of RQC. Protein and DNA samples were dialyzed against 4 L of 20 mM Tris, 100 mM KCl, and pH 7.0 buffer before the experiment. One microliter of DNA stock was added to the protein sample for the first injection. Subsequent titration points were done with 5 μL injections into the cell for a total of 21 titration points. We applied other experimental conditions, which are as follows: interval, 300 s; stirring speed, 300 rpm; and cell temperature, 25 °C. The heat of dilutions was measured by titrating G4s into the buffer with the same experimental conditions. The measured dilution heat was subtracted from the data with the average area mode of area correction function in NanoAnalyze software (TA Instrument). The subtracted heat isotherm was obtained by point-to-point subtraction of the heat of dilution from the integrated heat change.