Development of a Novel Fluorescence Assay Based on the Use of the Thrombin-Binding Aptamer for the Detection of O6-Alkylguanine-DNA Alkyltransferase Activity

Human O6-alkylguanine-DNA alkyltransferase (hAGT) is a DNA repair protein that reverses the effects of alkylating agents by removing DNA adducts from the O6 position of guanine. Here, we developed a real-time fluorescence hAGT activity assay that is based on the detection of conformational changes of the thrombin-binding aptamer (TBA). The quadruplex structure of TBA is disrupted when a central guanine is replaced by an O6-methyl-guanine. The sequence also contains a fluorophore (fluorescein) and a quencher (dabsyl) attached to the opposite ends. In the unfolded structure, the fluorophore and the quencher are separated. When hAGT removes the methyl group from the central guanine of TBA, it folds back immediately into its quadruplex structure. Consequently, the fluorophore and the quencher come into close proximity, thereby resulting in decreased fluorescence intensity. Here, we developed a new method to quantify the hAGT without using radioactivity. This new fluorescence resonance energy transfer assay has been designed to detect the conformational change of TBA that is induced by the removal of the O6-methyl group.


Introduction
Alkylating agents are chemotherapeutic anticancer drugs that produce their cytotoxic effect by generating adducts at multiple sites in DNA [1]. A subset of alkylating agents, which includes nitrosoureas and temozolamide, have a preference for alkylating guanine at the O 6 position, which is the most relevant in terms of mutagenesis and carcinogenesis [2][3][4][5][6][7][8][9]. In particular, 1,3-bis-(2-chloroethyl)-1-nitrosourea (BCNU) attacks initially at the O 6 guanine position, causing its rearrangement in a cyclic intermediate thus giving rise to N 1 ,O 6 -ethanoguanine [10]. Finally, a cross-link with the opposite cytosine is formed, and, as a consequence, DNA replication is blocked, producing G2/M arrest [11]. In addition to the well-known side effects and limitations of chemotherapeutic agents, these substances also present problems of acquired tumor resistance. In particular, the DNA-repair human O 6 -alkylguanine DNA alkyltransferase (hAGT or MGMT) is responsible for removing alkyl adducts from the O 6 position of guanines, thereby blocking the cytotoxic effects of the alkylating agents and making a crucial contribution to the resistance mechanism [12][13][14]. It is well established that tumor cells show greater expression of this protein than healthy cells. Thus, this increased expression appears to be predictive of a poor response to chemotherapeutic drugs. This effect has been observed in a large number of cancers, ranging from colon cancer, lung tumors, breast cancer, pancreatic tumors, non-Hodgkin's lymphoma, myeloma, and glioblastoma multiforme, among others [15][16][17]. In addition methylation of the hAGT promoter and consequently the complete depletion of hAGT, it has been associated with longer survival in patients with gliomas undergoing combined radiation-chemotherapy treatment [18,19]. Therefore, pharmacological inhibition of hAGT has the potential to enhance the cytotoxicity of a diverse range of anticancer agents [20].
hAGT is a 22-kDa (207 AA) DNA-binding protein that contains a highly conserved internal cysteine, which acts as the acceptor site for alkyl groups. The S-alkyl-Cys formed is not regenerated and the protein, which behaves as a suicidal enzyme, inactivates itself in the dealkylation process [21,22]. This single turnover of hAGT renders it vulnerable to inactivation. On the basis of this observation, intense research effort has been devoted to the identification of small molecules capable of inhibiting hAGT activity and significantly enhancing the cytotoxic effect of BCNU in prostate, breast, colon, and lung tumor cells [20].
Several methods are available to characterize the mechanism of action of hAGT and its activity. Moreover, they also have the capacity to evaluate the capacity of small molecules to inhibit hAGT. Most of these methods are based on radioactivity assays while others are based on multiplestep enzymatic reactions [23][24][25][26].
G-quadruplexes are a family of four-stranded DNA structures stabilized by the stacking of guanine tetrads, in which four planar guanines form a cyclic array of hydrogen bonds [27]. These G-rich regions are connected by lateral, central, or diagonal loops of diverse sizes and composition that form base-pairing alignments, which in turn stack with the terminal G-tetrads, thus further stabilizing Gquadruplex structures [28]. Another key element in Gquadruplex formation is the presence of monovalent cations, which stabilize the negative electrostatic potential created by the guanine O 6 oxygen atoms within the quadruplex core [29,30]. However, most divalent cations have the capacity to induce the dissociation of G-quadruplex structures [31]. Finally, modifications in the base composition of the tetrads are poorly tolerated by these structures. As an example, inosine [32] and O 6 -methylguanine [33], both nonnatural bases can form a smaller number of hydrogen bonds and thus destabilize the G-quadruplex.
Given the potential relevance of hAGT as a prognostic marker of cancer and as a therapeutic target, and that its substrate O 6 -methylguanine disrupts the formation of G-quadruplex structures [33], here, we developed a new fluorescence activity assay for hAGT. For this purpose, we selected the thrombin-binding aptamer (TBA) as our Gquadruplex model. TBA is a well-known 15 mer that adopts a chair-like structure consisting of two G-tetrads connected by two lateral TT loops and a central TGT loop [34]. The modification of its sequence in positions 5 or 6 by replacing a guanine of the tetrad for an O 6 -methylguanine disrupts folding, leaving it in an extended conformation. Giving that the two conformations bring the two ends of TBA together take them further apart, our working hypothesis was that the incorporation of fluorescence probes results in a measurable change in intensity. The final aim of this fluorescence assay was to measure the DNA repair activity of hAGT and thus facilitate the search for new and more potent inhibitors which enhance chemotherapeutic drugs. Although several methods have been described to quantify hAGT activity [23][24][25][26], none of them use the conformational change of Gquadruplex for this purpose. Here, we describe the development of a straightforward, rapid, one-step fluorescence resonance energy transfer (FRET) assay.
Solvents for chromatographic analysis were prepared using triethylamine, acetic acid, and acetonitrile as mobile phase. All other chemicals were of analytical reagent grade and were used as supplied. Ultrapure water (Millipore, USA) was used in all experiments.

Instrumentation. Semipreparative reverse phase (RP)
HPLC was performed on a Waters chromatography system using Nucleosil semipreparative 120 C18 (250 × 8 mm) columns. Analytical RP-HPLC was performed using an XBridge OST C18 2.5 μm column and a Nucleosil Analytical column 120 C18 (250 × 4 mm). Oligonucleotides were detected by UV absorption at 260 nm. Mass spectra were recorded on a MALDI Voyager DE TM RP time-of-flight (TOF) spectrometer (Applied Biosystems, USA) with a nitrogen laser at 337 nm using a 3-ns pulse. Fluorometric measurements were performed on a spectrofluorometer multidetection microplate reader Biotek FL × 800 and a Jasco FP6200. Molecular absorption spectra between 220 and 550 nm were recorded with a Jasco V650 spectrophotometer. The temperature was controlled with a Jasco ETC 272T Peltier device. Hellma quartz cuvettes (0.5 and 1.0 cm path length, 500 or 1000 μL volume) were used. Table 1) were synthesized on an ABI 3400 DNA Synthesizer (Applied Biosystems, USA) using a 200nmol scale synthesis and following the standard protocols. We used dimethylformamidino-protected guanine G dmf phosphoramidite for all the syntheses. 5 -Fluorescein CE phosphoramidite (6-FAM), O 6 -methylguanine (O 6 -MeG) and G dmf phosphoramidites were from commercial sources. The isobutyryl group was used to protect the amino group of O 6 -MeG [35]. The quencher group was introduced at the 3 end using the controlled pore glass functionalized with a 3 -Dabsyl derivative CPG. O 6 -MeG-containing oligonucleotides were deblocked using concentrated aqueous ammonia overnight at room temperature following the manufacturer's instructions. The resulting products were desalted by Sephadex G-25 (NAP-10, GE Healtcare, USA) and purified by reversed-phase HPLC using Nucleosil columns. The yields and purities obtained for the products were around 85% for 5-O 6 -MeG-TBA and 6-O 6 -MeG-TBA, and >98% for the rest of the sequences. The length and homogeneity of the oligonucleotides were checked by MALDI-TOF.  The DNA-strand concentration was determined by absorbance measurements (260 nm) by calculating extinction coefficients. Oligonucleotide samples were kept at 4 • C until further use. Double-stranded O 6 -MeG-TBA was formed by annealing equimolar concentrations of complementary oligonucleotide strands at 72 • C for 5 min and then allowed to slowly cool to room temperature.

Melting Temperature Studies.
Melting curves were measured by monitoring the absorbance hyperchromicity at 295 and 495 nm in a JASCO V650 spectrophotometer equipped with a Peltier temperature-controlling unit. UV/Vis absorption spectra were recorded at 1 • C/min intervals, with a 1min equilibration time at each temperature; the sample was heated over the range 20-80 • C. The buffer solutions used were 10 mM sodium cacodylate pH 7.0 and 100 mM KCl. Sample concentration was around 4 μM. Each sample was allowed to equilibrate at the initial temperature without any external control of temperature for 5 min before the melting experiment began. The melting temperatures (Tm) are the average value of at least one pair of Tm experiments.
2.5. CD Spectra. Samples were prepared as described for the melting experiments by UV spectroscopy. Measurements were conducted in 10 mM sodium cacodylate pH 7.0 and 100 mM KCl. Sample concentration was between 1-4 μM. The CD spectra were recorded on a Jasco J-810 spectropolarimeter attached to a Julabo F/25HD circulating water bath in 1 cm path-length quartz cylindrical cells. Spectra were recorded at room temperature using a 100 nm/min scan rate, a spectral band width of 1 nm and a time constant of 4 s. All the spectra were corrected with the buffer blank and plotted using Origin software.

Human Recombinant hAGT Protein.
Full-length hAGT was overexpressed and purified as previously described [36]. . Several incubation times were tested (30, 60, 90, 120, 360, and 1440 min) at 37 • C and the reaction was stopped by heating the samples at 72 • C for 5 min. The reaction products were analyzed by HPLC on a Nucleosil analytical column at 60 • C or room temperature, depending on the substrates used in the experiment (double-or single-stranded TBA). The HPLC flow rate was 1 mL/min, and a gradient of 10%-40% acetonitrile for 20 min was used.

Fluorescence
Assay for hAGT Activity. The fluorescence assay was performed in a multidetection microplate reader biotek FLx800 in optical 96-well Optical btm reaction plates

Results and Discussion
The aim of this study was to develop a real-time hAGT activity assay based on the detection of differences between extended and folded conformations of TBA. Our working hypothesis was that the quadruplex structure of TBA is disrupted when a central guanine is replaced by an O 6methylguanine. The TBA sequence also contains a fluorophore and a quencher group attached to 5 or 3 end, respectively. In the presence of O 6 -methylguanine, methylated TBA unfolds and the fluorophore and the quencher become physically separated beyond the Förster distance. When the repair protein hAGT is added to the methylated aptamer, the enzyme removes the methyl group from the mutated guanine, thus allowing TBA to fold back into its chair-like conformation. As a result, the quencher comes closer to the fluorophore and blocks its fluorescence, as illustrated in Figure 1. This loss of fluorescence is quantified as a direct measurement of hAGT activity.

Synthesis of Modified G-Quadruplex Sequences.
The Gquadruplex sequences used in this study are shown in Table 1. Oligonucleotide synthesis was performed by the solid-phase 2 -cyanoethyl-phosphoramidite method [37]. Ammonia treatment was performed at room temperature overnight to minimize decomposition of O 6 -methylguanine. For the same reason, the dimethylformamidino group was selected for the protection of 2 -deoxyguanosine [38]. The sequences were characterized by HPLC and mass spectrometry, which provided the expected molecular weights. The yields of the isolated molecular beacon oligonucleotides after HPLC purification and desalting were in the range of those obtained for unmodified oligonucleotides.

Thermal Stability of Methylguanine-Modified TBA.
In order to induce the unfolding of the quadruplex structure of TBA, we introduced an O 6 -methyl-guanine at position 5 or 6 of TBA. Melting curves of the modified sequences were performed by UV spectroscopy and compared with the unmodified sequence. These experiments were carried  T1   T1  T3   T7  T9   T4   T16  T17  T18   T19 Figure  S5). These results confirmed that methylation of guanine in either of the two positions of the TBA sequence prevents the formation of the chair structure. This observation confirms our working hypothesis. Given that the DNA repair activity exerted by hAGT is higher in double-stranded DNA than single-stranded DNA [39], we studied the stability and the quadruplex formation of elongated self-complementary TBA derivatives (see Table 1). We designed several TBA oligonucleotides elongated in their 3 end with a subset of self-complementary sequences of diverse sizes. The purpose of these elongations was to check whether the extended sequences have the capacity to form a double strand helix of different lengths and strengths with themselves without disrupting the chairlike structure. For 6-HP-TBA and 9-HP-TBA, we found that the corresponding Tm at 260 nm were 63 o C and 71.8 • C, respectively. This observation confirms our hypothesis of a double helix structure that increases in strength as the sequence length increases. However, we did not detect a melting temperature of these two sequences at 295 nm. The absence of a transition at 295 nm is consistent with the absence of an antiparallel quadruplex structure. In contrast, 3-HP-TBA, corresponding to an overhang of only 3 nucleotides, gave a melting temperature of 46 • C at 295 nm, which is slightly but similar to that obtained for natural TBA. This result indicates that 3-HP-TBA forms an antiparallel quadruplex instead of a duplex because the overhang is too short to break the chair-like structure. Circular dichroism confirms the quadruplex structure of 3-HP-TBA and the absence of quadruplex structure in 6-HP-TBA and 9-HP-TBA ( Figure S6). Models of all these structures are shown in Figure 3.

HPLC Analysis of DNA Repair Activity of hAGT.
In order to observe the efficiency of hAGT to remove alkyl groups from single-stranded oligonucleotides, we performed several assays to determine the optimal conditions of the reaction.
For this purpose, the full-length hAGT was first overexpressed and purified as described previously [36]. Increasing concentrations of the protein were incubated with double-stranded 5-O 6 -MeG and 6-O 6 -MeG TBA, annealed with its complementary sequence. Figure 4(a) shows the HPLC profile of the final product of the reaction with double-stranded 5-O 6 -MeG-TBA. In order to separate the two strands of the TBA substrate (Figure 4(a) top right panel), HPLC analyses were done at 60 • C. After incubation with hAGT, we detected a peak with a shorter retention time, which corresponds to the restored TBA sequence caused by the removal of the methyl group. We obtained the same results when we used double-stranded 6-O 6 -MeG-TBA as a substrate (data not shown). As expected, hAGT activity was not affected by the position of the alkyl group within the sequence. We then tested hAGT activity over single-stranded methylated TBA in the previously optimized conditions and obtained similar results to those shown in Figure 4(b). The top right panel shows the HPLC chromatogram in the absence of hAGT. In this case, the HPLC analyses were performed at room temperature because the substrate was single-stranded TBA. Our results confirmed that hAGT has the capacity to remove methyl groups from single-or doublestranded TBA with the same efficacy. Therefore, we selected single-stranded TBA as substrate for the development of our fluorescence assay.

Fluorescence hAGT Activity Assay.
On the basis of the results obtained in the melting temperature and the HPLC experiments, we tested the effectiveness of our proposed model to the DNA repair activity of hAGT by means of fluorescence. First of all, we determined the minimum amount of TBA required to achieve a detectable and reliable Journal of Nucleic Acids 7 difference in intensity compared to the background. As expected, the negative control natural TBA quadruplex gave low background fluorescence, because of the proximity of the fluorophore and the quencher groups in the chair-like structure ( Figure 5(a)). Although the fluorescence of MB-TBA was low, this fluorescence was subtracted from the fluorescence value in each experiment. The concentration of 5 nM of fluorescently labelled MB-O 6 -MeG-TBA was considered the optimal concentration as the fluorescence signal was intense and only a small amount of hAGT protein was required to achieve a substantial decrease in fluorescence in 20-40 min. In the optimal conditions, the presence of hAGT produced a remarkable decrease in fluorescence intensity, caused by the demethylation of the O 6 position of guanines, thereby allowing the TBA to form its typical quadruplex structure, which brings together the fluorophore and the quencher groups, as occurred in the negative control. The rate of decrease in fluorescence intensity correlated directly with the amount of hAGT in the reaction mixture ( Figure 5(b)). Moreover, the inactive mutant hAGT-C145S did not exhibit any decrease in fluorescence ( Figure 5(a)). This result was expected because of the inability of this mutated protein to repair the modified TBA. Figure 5(b) shows the fluorescence intensities for the real time hAGT assay. All these observations are consistent with the hypothesis and design of our FRET method.

Conclusion
Although radioactivity has been widely used in the search of potential inhibitors of hAGT [23,24], this technique does not allow real-time data acquisition and, in addition, requires radioactive materials. Here, we developed a sensitive fluorescence method that allows the quantification of hAGT activity in a single step and in a straightforward manner. Although another fluorescence method has already been developed for this purpose [25], it requires the addition of a restriction enzyme, followed by the addition of an exonuclease after the hAGT activity reaction. Consequently, although it is a real-time assay, a three-step reaction is required before observing an increase in fluorescence. In contrast, in our assay, a change in fluorescence is detected in a single step (homogeneous), and this method does not depend on the activity of two additional enzymes.
Our assay is based on the detection of conformational changes of TBA in the presence or absence of O 6methylguanines in its structure. The quadruplex structure of TBA is disrupted when a central guanine is replaced by an O 6 -methylguanine. Fluorophore groups can be added to the modified sequence in order to detect the conformational changes by fluorescence. The variation in fluorescence can be quantified as a direct measurement of hAGT activity. In addition, this technique requires lower amounts of substrate and does not call for the use of radioactive materials. Furthermore, this method can be easily transferred to a high throughput experiment for the evaluation of small molecules as potential hAGT inhibitors [36]. Research in this direction is currently underway in the laboratory. Thrombin-binding aptamer cTBA: Complementary TBA TEAA: Triethylammonium acetate TFA: Trifluoroacetic acid Tris: Tris(hydroxymethyl)aminomethane Tm: Melting temperature UV: Ultraviolet.