A closer look at the mode of binding of drug pemetrexed with CT-DNA

Abstract The interaction of antifolate drug Pemetrexed (PEM) with CT-DNA has been studied by UV-Vis, fluorescence and circular dichroism spectroscopic techniques. The results of these spectroscopic studies in combination with viscosity measurements, voltammetric and KI quenching studies suggested a less-common mode of binding of PEM with CT-DNA i.e. neither intercalation nor groove binding. Thus, metadynamic (MD) simulation is utilized to decipher the nature of binding of PEM with CT-DNA. Analysis of free energy surfaces obtained in MD simulation, reveals that PEM binds to the 3’- and 5’-ends of the DNA molecule. The thermodynamics of the interaction has been investigated by isothermal titration calorimetric experiment. The analysis shows that PEM binds with CT-DNA strongly with a binding constant of 2.6x109 M−1 and the process is found to be spontaneous (ΔG − 12.84 kcal/mol). Further, positive values of enthalpy (ΔH 6.09 cal/mol) and entropy (ΔS 43.1 cal/mol) changes indicate that the binding is an enthalpically unfavourable and, instead, entropically driven process. Communicated by Ramaswamy H. Sarma


Introduction
Deoxyribonucleic acid (DNA) plays an important role in the process of human life. It is the carrier of generic information and the basis of gene expression (Mirzaei-Kalar, 2018;Parisa & Dorraji, 2017;Chattopadhyay et al., 2007;Goldman & Zhao, 2002). The ability of ligands to interfere with the functions of DNA makes it as a major target for drug interaction studies. These studies have been performed using several spectroscopic studies and the results of such studies are being used in the field of drug designing. In the process of drug design, it is essential to understand the mode of interaction of drugs with DNA and structural specificity of their binding processes. It is presumed that, any knowledge regarding the interaction of the existing drug PEM with DNA would help in such drug design process in future and hence the present study.
PEM, chemically known as (N-[4-[2-(2-amino-3,4-dihydro-4oxo-7H-pyrrolo[2,3-d]pyrimidin-5-yl)-ehtyl]-benzoyl]-L-glutamic acid (Figure 1), is a unique multitargeted antifolate that powerfully inhibits three enzymes viz. thymidylate synthase, dihydrofolate reductase and glycinamide ribonucleotide formyltransferase. These enzymes are involved in the synthesis of pyrimidine and purine, which are the basic building blocks of DNA. Thus, inhibition of thymidylate synthase by PEM results in decreased thymidine necessary for DNA synthesis. The U.S. Food and Drug Administration (FDA) and the European Medicines Agency (EMEA) has approved PEM for the first line treatment in combination with cisplatin and as a monotherapy after prior chemotherapy in patients in locally advanced or metastatic non-squamous, non-small cell lung cancer (Ares et al., 2003;Joerger et al., 2010;Adjei, 2004).
The main objective of the present endeavour is to investigate the binding of PEM with CT-DNA by using several spectroscopic techniques such as UV-Vis, fluorescence, and circular dichroism along with viscosity measurements and voltammetric studies. The thermodynamics of the interaction was delineated using isothermal titration calorimetry. Based on the results of these analytical and spectral studies a plausible mechanism for the interaction of PEM with DNA has been proposed and discussed. Metadynamics simulation analysis has also been carried out to substantiate the experimental results of the binding event. Interestingly, PEM was found to bind with DNA in a non-usual mode of binding viz. it binds with the 3 0 -end and 5 0 -end of a DNA molecule. To the best of our knowledge this is the first report on such novel mode of binding of drug molecule with DNA.
While we are in the process of writing this manuscript, Senel et al. have reported the binding of PEM with double strain fish sperm DNA using spectroscopic, electrochemical and molecular docking and simulation studies. Based on the results of these studies they proposed that PEM binds to DNA in the minor groove through H-bonding interactions (since DH < 0 and DS > 0), which was substantiated by the molecular docking study (Senel et al., 2022). Interestingly, the results of the present study suggested an entirely different non-classical mode of binding (at 3 0 and 5 0 ends) of PEM with CT-DNA. The proposed mode of binding is based on relatively more robust experimental techniques such as circular dichroism, isothermal titration calorimetry, competitive assay using a classical groove binder Hoechst 33258 and free energy surface (FES) analysis of metadynamics simulation results. These studies would definitely provide a closer look at the mode of binding of ligands with DNA and thus the proposed binding mode would be the most probable one. The related results and discussion are spelt hereunder.

General materials
Commercially available analytical grade chemicals and solvents were used in the study. PEM drug was received as a gift sample from a locally available pharmaceutical company. Its purity was confirmed by its m.p. (Found: 243-244 C and Literature: 243-245 C) and 1 H NMR spectrum ( Figure S1). The calf thymus DNA (CT-DNA) was obtained from Genie and its purity was checked by monitoring the ratio of absorbance at 260/280 nm. The stock solutions of the drug, CT-DNA, ethidium bromide (EB) and Hoechst 33258 were made up with Millipore water. Freshly prepared solutions were used for all the spectral measurements. All the solutions used throughout the experiments were adjusted with HEBES buffer solution (pH 7.4).

UV-Vis spectral measurements
UV-Vis spectra were recorded in the wavelength range of 220-600 nm using a JASCO (V630) double beam spectrophotometer furnished with 1 cm quartz cuvettes at room temperature. The UV-Vis absorption spectra of PEM (24 mM) were monitored with incremental addition of CT-DNA (0-9 mM). Similarly, absorption spectra of CT-DNA (60 mM) were recorded with the increasing amount of PEM (0-3 mM) in HEBES buffer of pH 7.4.

Fluorescence spectral analysis
Fluorescence spectra of PEM (50 mM) (k ex 288 nm; k em 465 nm) with incremental addition of CT-DNA (0-3 mM) were recorded at three different temperatures viz. 298, 308 and 318 K using JASCO (FP 8500) spectrofluorimeter equipped with a Xenon lamp source and 1 cm quartz cell and the band width was 5 nm. Additionally, potassium iodide quenching studies of PEM (50 mM) in the absence and presence of CT-DNA (50 mM) with increasing amount of KI (0-25 mM) were also measured using the spectrofluorimeter in HEBES buffer solution. Fluorescence quenching constants (K sv ) were calculated using the following Stern-Volmer equation (Ma et al., 2012;Kalaivani et al., 2013, Wu et al., 2021.
Where F o and F are the fluorescence intensities of PEM in the absence and presence of CT-DNA, respectively and [Q] is the concentration of CT-DNA. Competitive displacement assays with EB and Hoechst 33258 were also carried out using the spectrofluorimeter and the excitation wavelengths used for EB and Hoechst systems were 250 and 313 nm, respectively.
The intrinsic binding constants for the interaction of PEM with CT-DNA were calculated using the following modified Stern-Volmer equation (Sun et al., 2008) where F 0 and F are the fluorescence emission intensities of PEM in the absence and presence of the quencher CT-DNA and [Q] is the concentration of CT-DNA. Using the K b values, the thermodynamic parameters such as enthalpy change (DH o ), entropy change (DS o ) and free energy change (DG o ) were calculated with the help of the following equations (Amin et al., 2021) log

Circular dichroism spectral studies
The circular dichroism spectra of CT-DNA (100 mM) were recorded using a JASCO (J810) spectrometer at a wavelength range of 220 to 310 nm in HEBES buffer (pH 7.4) at room temperature equipped with 1 cm quartz cell. Before commencement the experiment, the chamber of the spectrometer was deoxygenated with nitrogen gas and kept in the nitrogen atmosphere during all the experiments. Background spectrum of HEBES buffer solution was recorded and subtracted from CT-DNA spectra.

Electrochemical measurements
Electrochemical experiments were carried out, using a CHI-643B, Austin, TX electrochemical workstation, using a threeelectrode system viz. glassy carbon electrode (working electrode), Pt wire (counter electrode) and Ag/AgCl (reference electrode). Cyclic voltammetry was performed in 10 mM Tris-HCl buffer solution (pH 7.4) containing 50 mM of PEM at different scan rates (50-250 mV). Voltammetric experiments were performed in the Tris-HCl buffer solution containing 17 mM of PEM by adding varying CT-DNA concentrations (0 À 3 mM). The formation constant for the PEM/CT-DNA complex was determined using the following equation (Bard & Faulkner, 1980).
Where K f is the formation constant, A is the proportional constant and i o and i are the peak currents in the absence and presence of CT-DNA, respectively.

Viscosity measurements
Viscosity measurements were carried out using an Ostwald viscometer in a thermostat at 25 C by keeping the CT-DNA concentration constant (30 mM) and varying the concentration of ligands up to [Ligand]/[CT-DNA] ratio of 1. The flow time was measured by using a digital stopwatch and the time was taken from the average of the individual three measurements were recorded. Total volume 20 mL was made by using Tris-HCl buffer (pH 7.4). The resulted data were plotted as (g/g o ) 1/3 versus [Ligand]/[CT-DNA] ratio, g and g o were the relative viscosity in the absence and presence of PEM.

Molecular docking/simulation studies
The computational studies were carried out as reported by us earlier (Ponkarpagam et al., 2021). The initial 2 D structure of PEM was prepared with the help of Bio-Chemdraw 3 D along with energy minimization and then saved as mol2 file. The optimized PEM structure was used for docking using Autodock 4.0 software. The structure of DNA fragment with B-form (1BNA) was acquired from PDB. Docking procedure includes grid spacing to accommodate whole DNA molecule as 0.620 Å along with dimensions of x, y, z coordinates set to be 62 Â 70 x 92 Å (Irshad et al., 2016). Among the resulted docking positions, the one possesses minimum binding energy was selected and saved as PDB, which then was used as the input for molecular dynamics studies using Desmond in Schrodinger Suite. The PEM/DNA complex was first soaked in an orthorhombic water box containing 11659 water molecules (TIP3P water model), all atom force field (OPLS-2005) and 22 sodium ions were used to neutralize the system along with NPT canonical ensembles and thermostat throughout the simulation period. For maintaining the constant volume throughout the simulation, it was ensured that the density and pressure were corrected during these simulations. The energy minimization and simulations were carried out at 300 K and 1 bar pressure. The simulation was carried out for 100 ns and the trajectory files were saved at every successive 100ps as snapshots. The two collective variables used for metadynamics simulation are: (i) distance between the centre of mass (COM) of four bases of DNA and the COM of PEM, and (ii) the angle between the fixed vector AB and vector AC, the distance between the COM of four bases of DNA and that of PEM as already reported by us. Finally, the setup was minimized and pre-equilibrated using the default relaxation tool.

Goodness of fit
The correlation coefficient, r, is the simplest measure of the goodness of fit of the regression equation to the data. The correlation coefficient is relevant to test the validity of a model. In the present study the goodness of a fit was explained using correlation coefficient, r. There is a much-used convention that values of correlation coefficients should be assessed as follows: r ¼ 0.99 to 1.00 (Excellent), 0.95 to 0.99 (Satisfactory), 0.90 to 0.95 (Fair) and < 0.90 (Poor) (Shorter, 1982).

UV-Vis spectral studies
The UV-Vis spectrum of free CT-DNA (60 mM) in HEBES buffer of pH 7.4 exhibited a band at 258 nm corresponds to p-p Ã transition of base pairs ( Figure S2) (Khajeh et al., 2018;Li et al., 2017). As seen from the figure, upon addition of incremental amounts of PEM (0-3 mM) the absorbance of the band at 258 nm increased marginally. The UV-Vis spectrum of free PEM (24 mM) in HEBES buffer of pH 7.4 showed a band at 222 nm with a shoulder at 251 nm due to p-p Ã transitions ( Figure S3). Upon addition of increasing concentrations of CT-DNA (0-9 mM), the absorbance of the band also increased. The hyperchromic effect observed in these experiments suggested that

Fluorescence spectral studies
The fluorescence spectrum of PEM (50 mM) showed an emission at 465 nm upon excitation at 250 nm in HEBES buffer (pH 7.4) (Figures 2 and S4 and S5). Addition of increasing concentrations of CT-DNA (0-3 mM) to PEM, was found to quench the fluorescence of the drug, indicating the interaction between PEM and CT-DNA. The mechanism of the quenching of the fluorescence of PEM by the added CT-DNA was analysed by the equation (1). The K SV values at three different temperatures were determined from the linear plots F o /F versus [Q] ( Figure S6) and the values are found to be 3.4 Â 10 5 , 2.7 Â 10 5 and 1.8 Â 10 5 M À1 at 298, 308 and 318 K, respectively. The observed decrease of K sv values with an increase in temperature suggested that CT-DNA quenched the fluorescence of PEM via static quenching mechanism (Lakowitcz, 1999 Figure S8). The thermodynamic parameters such as enthalpy change (DH o ), entropy change (DS o ) and free energy change (DG o ) are calculated to be 0.1 kJ mol À1 , 383 J K À1 mol À1 and À22.3 kJ mol À1 (at 298 K), respectively. Positive signs of enthalpy (DH o > 0) and entropy (DS o > 0) changes indicated that the binding process was mainly driven by hydrophobic forces. The stoichiometry of the interaction between PEM and CT-DNA was determined using Job's plot and the results depicted in Figure S9

Competitive binding experiments
The mode of binding of PEM with CT-DNA was resolved by carrying out competitive displacement assays with a typical intercalator EB as well as with a typical groove binder Hoechst 33258. As shown in Figure 3, EB/CT-DNA complex exhibited a strong fluorescence at 601 nm on excitation at 250 nm. Addition of incremental amounts of PEM to EB/CT-DNA complex was found to quench the fluorescence of the later, but not to a greater extent. This observation suggested that PEM may not bind to CT-DNA through a typical intercalation mode. On the other hand, addition of incremental amounts of PEM to CT-DNA/Hoechst 33258 complex, the fluorescence of the complex at 495 nm remained practically unchanged (Figure 4), suggesting absence of any groove binding mode of interaction between PEM and CT-DNA (Chao et al., 2013;Sahoo et al., 2008). The results of these competitive binding assays indicated that PEM may bind to CT-DNA in a non-regular mode of binding.

Circular dichroism spectral studies
Non-covalent interactions between ligands and DNA can affect the structural morphology of the DNA, which alters the intrinsic CD spectral behaviour of the DNA. The CD spectra of CT-DNA (200 mM) in the absence and presence of PEM (1 and 2 equivalent) are shown in Figure 5. The CD spectrum of free CT-DNA exhibited a negative peak at 245 nm due to right-handed helicity and a positive peak at 275 nm owing to p-p base stacking, which suggested that the existence of DNA in B form. It is seen from the figure that upon addition of PEM to CT-DNA, the peak at 245 nm showed hyperchromocity while the peak due to p-p base stacking exhibited hypochromocity suggesting strong interaction between PEM and CT-DNA in the PEM/CT-DNA complex (Qais et al., 2017).

Potassium iodide quenching study
It is well-known that the binding of ligands to the groove provides little protection from anionic quencher like KI and the quencher can quench the fluorescence of the ligand even in the presence of DNA. On the other hand, the ligands that have intercalated with DNA is well protected from the anionic quencher and the fluorescence of which is not quenched. In the present study, the fluorescence intensities of PEM (50 mM) upon the addition of incremental amounts of KI (0-25mM) were measured at 465 nm (k ex 288 nm) and the results are depicted in Figure S10. It is seen that the added KI quenched the fluorescence of PEM gradually and the Stern-Volmer quenching constant K sv was calculated using linear plot of F o /F versus [KI] ( Figure S11). The K SV value was determined to be 36.14 M À1 . Interestingly, the addition      Figure 11. Free energy illustration of binding PEM to CT-DNA as a function of distance and dihedral angle. From two energy minima A1 and B1 were found to be 3' and 5' end binding mode. Then A2 and B2 shows the interaction of PEM to DNA residues. of incremental amounts of KI (0-25 mM) to PEM/CT-DNA complex had no significant effect on the intensity of fluorescence of PEM. If PEM is intercalated between the base pairs of the CT-DNA molecule, then due to relatively greater protection form the quencher KI, it would have yielded a relatively lower K SV value than that of free PEM. The observed unaffected fluorescence of PEM in the CT-DNA complex may be due to the non-availability of the aromatic ring for its interaction with the iodide ion (Najbar & Mac, 1991). Thus, the KI quenching experiments suggested a non-usual mode of binding of PEM with CT-DNA.

Electrochemical studies
Typical cyclic voltammograms of 50 mM PEM at varying scan rates (0.05 À 0.25 V/s) in 0.05 M Tris-HCl buffer (pH 7.3) are shown in Figure 6. As seen from the voltammogram, in forward sweep potentials with a scan rate of 0.05 V/s a cathodic peak at potential of À0.6 V was obtained, which is due to the reduction of the amide moieties of PEM (Ponkarpagam et al., 2020). The dependence of the cathodic peak current on the scan rate was also depicted in the Figure 6. Further, a plot of cathodic peak current (i p ) versus scan rate was found to be linear ( Figure S12; r 0.989), which suggested that the electrode process is an adsorption controlled process (Zia et al., 2020). The formation of a complex between PEM and CT-DNA was also investigated using more sensitive differential pulse voltammetry (DPV) techniques in the absence and presence of increasing amounts of CT-DNA in Tris-HCl buffer solution (pH 7.3) and the results are depicted in Figure 7. In the absence of CT-DNA, PEM (50 mM) showed a reduction peak at À0.53 V (Ag/AgCl), which upon addition of increasing concentration of CT-DNA (0-3 mM) shifted towards negative potential with a concomitant decrease in peak current suggesting the formation of PEM/CT-DNA complex. The formation constant for the complex was determined using the equation (2) (Bard & Faulkner, 1980;Wu et al., 2022). The value of K f was calculated from the linear plot of (i o /i o -i) versus 1/[CT-DNA] ( Figure S13; r 0.997) and was calculated to be 7.5 Â 10 5 M À1 , suggesting strong interaction between the partners in the complex (pH 7.2 À 7.4).

Viscosity measurements
Viscosity experiment is considered as one of the unambiguous methods to elucidate the mode of binding of small molecule with DNA. A classical intercalator upon binding to DNA would yield an increase in separation of base pairs and consequently causes a significant increase in viscosity of the DNA solution. In contrast, a ligand that binds exclusively in the DNA grooves or electrostatic interaction causes minimal or no appreciable change in viscosity of the DNA solution (Gaber et al., 2020;Ling et al., 2008). In the present study, the effect of addition of EB (a classical intercalator) and Hoechst 33258 (a classical groove binder), along with PEM, on the change in the viscosity of the CT-DNA solution were also recorded under identical conditions for comparison purpose. The viscosity plots of (g/g o ) versus [ligand]/[CT-DNA] were obtained for the three ligands viz. PEM, EB and Hoechst 33258 ( Figure 8). As seen from the figure, the changes in viscosity of the DNA solution upon addition of increasing concentration of PEM didn't matched with either that of EB or that of Hoechst 33258. The viscosity curve of PEM/CT-DNA complex was found to position in between that of EB/CT-DNA and Hoechst 33258/CT-DNA complexes, suggesting a non-usual mode of binding of PEM with CT-DNA.

Isothermal titration calorimetric analysis
It is well known that ITC is a Gold standard tool which enables direct measurement of comprehensive thermodynamic profiles such as free energy change (DG), enthalpy change (DH) and entropy change (DS) and also detailed insights of drug-CT-DNA complex in terms of binding constant (K a ) and number of binding sites. Isothermal titrations of PEM (15 mL aliquots, 1.25 mM, at 2 mints intervals) into a solution of CT-DNA (200 ml, 0.5 mM) were performed in Tris-HCl buffer of pH 7.4. The raw ITC data of titration of PEM with CT-DNA is depicted in Figure 9 (upper panel) and the heat burst of the reaction plotted against molar ratio of the PEM to CT-DNA is shown in the lower panel of the figure. The analysis showed that the binding of PEM with CT-DNA is characterized by a binding constant of 2.6 Â 10 9 M À1 . The magnitude of the binding constant indicated a strong interaction between PEM and CT-DNA, which is in line with the observation made in the CD spectral studies. The observed negative free energy change (DG À 12.84 kcal/mol) revealed that the binding process is spontaneous. The positive heat of formation (DH 6.09 cal/mol) of PEM/CT-DNA complex indicated that this binding is an enthalpically unfavorable and, instead, entropically driven process (DS 43.1 cal/mol) (Barcelo et al., 2002;Mårtensson & Lincoln, 2018). The results of the ITC study corroborated that obtained from fluorescence studies.

Molecular docking and dynamics studies
After performing molecular docking, the optimal geometry of the PEM/DNA complex with the lowest free energy of binding was selected and depicted in Figure 10A. As shown in the figure, PEM binds to DNA at the end of the helix. The interaction diagram clearly indicated that one of the carboxyl group of PEM interacts with DNA through hydrophobic interactions. The C ¼ O moiety of the other carboxyl group binds to DG24 residue via H-bonding ( Figure 10B). These results matched well with the proposed binding model based on experimental studies. Further, nature of the interaction between PEM and DNA in the complex is also harmonized with that suggested by thermodynamics of the interaction. With an aim to predict the structure of PEM-DNA complex accurately and reliably, metadynamics simulation has been carried out by adopting the same procedure and collective variables (CV) as reported by us earlier (Ponkarpagam et al., 2021). The real time simulation for 100 ns was carried out and for every 1000 ps snapshots of the PEM-DNA complex structures were obtained. The conformational changes of PEM-DNA complex were monitored by the free energy surface plotted as a function of distance (d) (CV1 in Å) vs dihedral angle (CV2 in degrees). The results in Figure 11 showed two free energy minima, that are labelled as A and B. The insets A1 and B1, in Figure 11, illustrates representative structures of the PEM-DNA complex in the minima A and B, respectively. At minima A, PEM molecule binds to the 3 0 -end of the DNA molecule and this pose is stabilized by number of favourable interactions such as p-p stacking between the phenyl ring of PEM with free energy À10.44 kcal/mol and DC11 and interaction with DA18 and DC11 nucleotide bases (inset A2) ( Figure S13). Minima B having free energy À10.13 kcal/mol represents the lowest energy pose where PEM molecule binds to the 5 0 -end of DNA (inset B1). Here, this pose is stabilized by the interaction between PEM and DG22 residue of DNA (inset B2) ( Figure S14). The phenyl ring in this pose is nearly stacked with the DG24, but the distance between them may be slightly longer than the critical distance to visualize in the interaction diagram B2. Thus, MD simulation results clearly indicated that PEM binds to DNA neither through intercalation nor through groove binding but through a non-usual mode of binding i.e. at the 3 0 -and 5 0 -ends of the DNA helix. The foregoing results of analytical, spectral and voltammetric studies strongly supported this observation. This is a novel mode of binding of ligand to the DNA molecule and so far, only very few reports have been documented in literature. Di Leva et al. have disclosed a hopping binding mechanism wherein 3-(benzo[d]thiazol-2-yl)-2Hchrom-en-2-one binds both to the groove and 3 0 -end of the G-quadruplex (Leva et al., 2014). Liu et al. have shown that the ligand BMVC first binds the 5 0 -end of G-quadruplex with higher affinity to form a 1:1 complex and at higher ratio, the ligand also binds to the 3 0 -end to form a second complex (Liu et al., 2019).

Conclusions
A detailed experimental and theoretical investigation on the mode of binding of the antifolate drug PEM with CT-DNA was carried out. Absorption, fluorescence and CD spectrophotometry in combination with voltammetric studies suggested the formation of a strong complex between PEM and CT-DNA and CT-DNA quenched the fluorescence of PEM via static quenching mechanism. The results of competitive displacement assays with EB (a classical intercalator) and Hoechst 33258 dye (a classical groove binder) along with that of viscosity measurements and potassium iodide quenching studies suggested that the mode of binding of PEM with CT-DNA is neither intercalation nor groove mode of binding. Hence, the less-common mode of binding has been decrypted by metadynamics simulation analysis and the results of which showed that PEM binds to the 3 0 -and 5 0 -ends of the DNA molecule, which is a novel result in the study of binding of small molecules with DNA. The thermodynamic aspects of the novel binding process have been investigated using isothermal calorimetric titration experiment. The binding constant of the PEM/CT-DNA complex was determined to be 2.6 Â 10 9 M À1 indicating a strong binding between the partners, which is in good agreement with the results of the spectrophotometric studies. The negative free energy change (DG À 12.84 kcal/mol) and positive enthalpy (DH 6.09 cal/mol) and entropy (DS 43.1 cal/mol) changes demonstrated that the binding process is spontaneous, enthalpically unfavourable and an entropically driven process. The results of the experimental and computational studies indicated that the proposed mode of binding of PEM with CT-DNA would be the most probable one when compared to that suggested by Senel and co-workers. Un-doubtfully, the suggested novel mode of binding of PEM drug with DNA would shed some light on the mechanism of action of the drug and guide us future drug development attempts.