Site-specific amino acid substitution in dodecameric peptides determines the stability and unfolding of c-MYC quadruplex promoting apoptosis in cancer cells

Abstract c-MYC proto-oncogene harbours a transcription-inhibitory quadruplex-forming scaffold (Pu27) upstream P1 promoter providing anti-neoplastic therapeutic target. Previous reports showed the binding profile of human Cathelicidin peptide (LL37) and telomeric G-quadruplex. Here, we truncated the quadruplex-binding domain of LL37 to prepare a small library of peptides through site-specific amino acid substitution. We investigated the intracellular selectivity of peptides for Pu27 over other oncogenic quadruplexes and their role in c-MYC promoter repression by dual-luciferase assays. We analysed their thermodynamics of binding reactions with c-MYC quadruplex isomers (Pu27, Myc22, Pu19) by Isothermal Titration Calorimetry. We discussed how amino acid substitutions and peptide helicity enhanced/weakened their affinities for c-MYC quadruplexes and characterized specific non-covalent inter-residual interactions determining their selectivity. Solution NMR structure indicated that KR12C, the best peptide candidate, selectively stabilized the 5′-propeller loop of c-MYC quadruplex by arginine-driven electrostatic-interactions at the sugar-phosphate backbone while KR12A peptide destabilized the quadruplex inducing a single-stranded hairpin-like conformation. Chromatin immunoprecipitations envisaged that KR12C and KR12A depleted and enriched Sp1 and NM23-H2 (Nucleoside diphosphate kinase) occupancy at Pu27 respectively supporting their regulation in stabilizing and unfolding c-MYC quadruplex in MCF-7 cells. We deciphered that selective arresting of c-MYC transcription by KR12C triggered apoptotic-signalling pathway via VEGF-A-BCL-2 axis.


INTRODUCTION
G-quadruplex nucleic acids have recently emerged as a new class of anti-cancer targets. These non-classical DNA structures are evolved over tandem guanine repeats with a stable core ofstacked G-quartets in a cyclic Hoogsteen hydrogen-bonding arrangement (1,2). Bioinformatic studies revealed around 3,70,000 putative G-quadruplexforming sequences in the telomeres, promoters, and untranslated regions of the oncogenes (3)(4)(5)(6). In vivo mapping of these consensus motifs by quadruplex-specific antibodies (e.g. BG4 antibody) reinforced their prevalent distribution into translocation hot-spots and primary tumors than normal tissues (7,8). These tetra-stranded structures play key functions in diverse cellular processes (e.g. transcription, translation, mRNA stability, etc.) and neoplastic transformation (e.g. genomic instability and genetic dysfunction) (9). Therapeutic implications of G-quadruplex structures have expanded during the past decade and gained a momentum with the finding that c-MYC transcription, which is upregulated in 80% of solid tumors, could be inversely regulated by stabilizing a G-quadruplex-motif (Pu27) in the Nuclease Hypersensitive Element (NHE) III 1 , located -142 to -115 base pairs upstream of c-MYC promoter (P 1 ) (10).
In cancer biology, c-MYC earned a formidable oncogenic reputation due to its aberrant expression through constitutive transcription (11,12), which attributes to autonomous proliferation, chromosomal translocation, relentless replication, and impaired apoptosis in a myriad of cancers (13)(14)(15)(16)(17)(18). Earlier reports provided an estimate of ∼70,000 annual deaths in the United States to be associated with high level of c-MYC transcripts (17,19). MYC knockout experiments in the animal models demonstrated that partial inactivation of the oncogene offers acute or sustained tumor regression promoting proliferation arrest, apoptosis, differentiation, and senescence in cancer cells (14,20,21).Therefore, the design of c-MYC-targeted therapeutics to restore its basal level of cellular expression holds a promising avenue in anti-cancer treatment (22). In normal proliferating cells, the 27-mer quadruplex structure (Pu27) in NHE III 1 negatively regulates c-MYC transcription and ensures a threshold level of c-MYC transcripts. In Burkitt's lymphoma, reciprocal translocation shifts c-MYC-NHE III 1 under the control of an ectopic promoter (in Ramos cells) or leads to a Pu27-deleted (8,14) c-MYC allele (in CA46 cells) resulting constitutive c-MYC transcription (23)(24)(25)(26). Simonsson and colleagues showed that the treatment of CA46 cells by synthetic 22-mer quadruplex substantially inhibits c-MYC transcription (10,26), which underscores the biological and clinical significance of Pu27 motif in cancer therapeutics. However, the major caveat in this study is the conformational heterogeneity in Pu27 showing a high likelihood of dynamic shuffling between intermediate quadruplex isomers (27)(28)(29)(30)(31). Despite having the solution NMR (Nuclear Magnetic Resonance) structures of different loop isomers of Pu27 (28)(29)(30)(31)(32)(33), c-MYC quadruplex stabilizing compounds suffer from conformational rigidity (e.g. Bisquinolinium derivatives (34)(35)(36)), lack of bioavailability (e.g. Quarfloxin) (37), and promiscuity to other oncogenic quadruplexes (e.g. thiazole peptides, Indolylmethyleneindanone derivatives, etc.) (38)(39)(40). Most of these ligands have aromatic/pseudoaromatic -delocalized system and stabilize c-MYC quadruplex via externalstacking interaction rendering them inefficient to distinguish the intracellular quadruplexes of various topologies. This is because as many as ∼3,70,000 sequences in human genome have the propensity to evolve quadruplexes and a vast majority of known quadruplex-interacting molecules are designed to make high-affinity stacking interactions with Gquadruplexes. They also fail to conform to the polymorphic and flexible skeleton of native quadruplex and remain ineffective from target recognition due to evolution of interconvertible quadruplex intermediates within Pu27 under cellular microenvironment (41)(42)(43)(44). Therefore, small compounds, designed to stabilize c-MYC quadruplex often encounter off-target effects and narrow therapeutic window limiting their anti-cancer activities (45).
In the light of these shortcomings, peptides offer an alternative and promising avenue for G-quadruplex-targeted anti-cancer therapeutics. Compared to other 'biologics' treatment (e.g. protein drugs and antibodies), peptides benefit from their smaller size, ease of synthesis and purification, cell permeability, tumor-penetrating ability, and improved biocompatibility (46,47). They exhibit minimal drug-drug interaction, less immunogenicity, faster blood clearance, and better intra-tumoral diffusion due to lower molecular weight and excellent tolerability in the patients. Therapeutic peptides are also amenable to rational design for high-affinity intracellular interaction with target quadruplexes (48). Opposed to the small molecules, peptide candidates provide exquisite specificity to their in vivo targets due to structural and functional diversity that enable them to fold into a complementary shape relative to the dynamic topology of the target quadruplex. Furthermore, they interact with the larger portion of the target quadruplex rendering higher action potencies and greater specificity with relatively few off-target side-effects (49). Unlike small molecules, peptides degrade into naturally occurring amino acids without accumulating toxic metabolites into specific organs. Indeed, peptide drugs are susceptible to cleavage by serum proteases causing shorter half-lives and lower stability accounting for their limited oral administration, difficult intracellular delivery, and the necessity of high doses. However, adding several steps to their manufacturing, e.g. blocking the C-and N-termini, use of unnatural or D-amino acids, and chemical modifications can circumvent these drawbacks improving their pharmacodynamic and pharmacokinetic properties (46,50). Neidle's group was the first to develop substituted acridines with amide bonds (3,6,9-trisubstituted acridines) as effective G-quadruplex stabilizers and telomerase inhibitors (51). Balasubramanian's group developed peptide-hemicyanine ligands (52), 3,6-bispeptideacridone-and 3,6-bispeptide-acridine conjugates showing greater binding affinity to telomeric Gquadruplexes (53,54). Linear and branched peptides with nonstandard amino acids and/or acridone, methylpyrrole, or methylimidazole derivatives are reported to significantly increase the quadruplex-to-duplex selectivity (55). Planar peptide macrocycles with an extended -delocalized system (e.g. oxazole-based peptide macrocycles) offer preferential binding to various oncogenic G-quadruplexes (56). Zagotto et al. constructed a small library of 2,6-and 2,7amino-acyl/peptidyl anthraquinones that inhibits telomerase activity by inducing G-quadruplex structure formation in human telomeres (57). Similarly, a number of Distamycin A analogues (e.g. amide-linked oligopyrroles, uncharged N-methyl amide analog, and carbamoyl analog) and peptide-nucleic acid oligomers (PNA) are designed to selectively stabilize telomeric and oncogenic quadruplexes (58)(59)(60)(61). With regard to the derivation of protein sequences, an arginine-glycine rich (RGG) peptide from fragile Xmental retardation protein (FMRP) showed higher affinity to G-quadruplex structures (62). Insulin and Insulinlike growth factors (IGF-2) are also observed to selectively interact with the antiparallel G-quadruplex in the Insulinlinked polymorphic region (ILPR) (63).
In this study, we formulated six peptides derived from human Cathelicidin LL37 protein, secreted by lungs epithelial cells (64). Earlier studies evidenced that LL37 binds to telomeric G-Quadruplex structures but shows non-specific interactions with other oncogenic quadruplexes (65). This coupled with a major shortcoming that the 37-mer protein is cell membrane-impermeable; we truncated its quadruplexbinding domain (FK13) and further reengineered its mutants (KR12A, KR12B, KR12C, KR12D and KR12E) to offer enhanced G-Quadruplex affinity and therapeutic efficiency via selective interaction with c-MYC quadruplex under cellular microenvironment. We first evaluated their intracellular selectivity for native c-MYC quadruplex (Pu27) and its conformationally restricted isomers (Myc22 and Pu19) over different oncogenic and telomeric G-quadruplexes by dual-luciferase assays and isother-mal titration calorimetry (ITC). Then, we investigated how amino acid substitution in different peptides modulates their binding affinities for c-MYC quadruplexes in vitro and under macromolecular crowding conditions. We analysed the thermodynamic partitioning of binding free energies between entropic and enthalpic components of peptidequadruplex association and determined the role of peptides to affect the thermal stability of c-MYC quadruplexes by circular dichroism (CD). We correlated differential binding profiles of the peptides and c-MYC quadruplex with detailed structural investigations at the atomic level by solution NMR (nuclear magnetic resonance) structure calculations and Molecular Dynamics (MD) simulations. In addition, we monitored the role of peptides to regulate the occupancy of numerous transcription factors (Sp1, RNA Polymerase II, NM23-H2 (Nucleoside diphosphate kinase), and Nucleolin) at NHE III 1 by siRNA knockdown and chromatin immunoprecipitation (ChIP) studies, which appended a deeper insight into peptide-driven transcription regulation in cancer cells. Based on these observations, we screened the best peptide candidate having higher intracellular selectivity to c-MYC quadruplex and determined its mode of action to drive apoptotic signalling in cancer cells by monitoring differential expression of the apoptotic markers. This study gives fundamental insights at the atomic level how site-specific substitution of the amino acids in peptides determines the folding and unfolding of the quadruplex. These ␣-helical peptides offer overlapping target-specific pharmacophore model to target the dynamic structure of Pu27 and open a new paradigm for next generation quadruplex-targeting peptides with higher therapeutic indexes and minimal off-target effects.

Materials
All Fmoc-protected amino acids, benzotriazol-1-yloxytripyrrolidinophosphonium hexafluorophosphate (Py-BOP), and Rink amide MBHA resin are purchased from Novabiochem (Merck). N,N-Dimethylformamide (DMF), diethyl ether, trifluoroacetic acid (TFA), anisol, triisopropylsilane (TIS), phenol and N,N-diisopropylethylamine (DIPEA), piperidine and dimethyl sulfoxide (DMSO) are procured from Merck. Methanol, HPLC-grade water and acetonitrile are purchased from Rankem. For cell culture, Dulbecco's modified Eagle's medium (DMEM) and Gentamicin are obtained from Himedia laboratories. Trypsin-EDTA solution, Amphotericin B and DMSO for cell culture are bought from Sigma-Aldrich. Penicillinstreptomycin and fetal bovine serum (FBS) are purchased from Invitrogen. Single-stranded oligonucleotide sequences (Supplementary Table S1) for biophysical studies and the primers for real time polymerase chain reactions (qPCR) (Supplementary Table S2) are procured from Eurofins India Pvt Ltd. and Integrated DNA Technologies (IDT) respectively. For chromatin immunoprecipitations (ChIP), ChIP-grade anti-nucleolin primary antibody and NM23-H2 antibody (L15) are purchased from Abcam, and Santa Cruz laboratory respectively. Anti-Sp1 antibody and Rpb1 (RNAPII subunit b1) CTD (C-terminal domain) 4H8 antibody are obtained from Sigma and Cell signaling Table 1. Synthetic Peptides used in the study. Amino acid sequences are written in N (-NH2)-to C (-COOH)-direction Peptide sequences Dual-luciferase assays are performed to examine the intracellular selectivity of peptides for c-MYC quadruplex isomers (Pu27, Myc22 and Pu19) over other oncogenic quadruplexes (BCL-2, VEGF-A and KRAS) of different topologies. The role of peptide-quadruplex complex in oncogene promoter activation is investigated in contrast to the GQ-null promoter constructs into MCF-7 cells (67). After 24 h of peptide treatment, cells are washed and lysed into 100 l 1× Passive lysis buffer (PLB) for 15 min at room temperature. Renilla and firefly Luciferase activities in 25 l lysate are quantified by Dual-Luciferase ® reporter assay system (Promega) as per manufacturer's recommendations. Renilla luciferase enzyme activity is normalized to that of the firefly luciferase activity and luminescence of the peptide-treated samples are normalized to the untreated ones, which yield the fold induction values for Renilla activities by different quadruplex structures in presence and absence of synthetic peptides. Luminescence of each sample is detected in GloMax ® 20/20 Single-Tube Luminometer (Promega) in triplicates and averaged from three independent experiments.

Real time PCR
To inspect the transcription inhibitory roles of peptides in c-MYC, BCL-2, VEGF-A and KRAS oncogenes having putative quadruplexes in their promoters, we performed real time polymerase chain reactions. MCF-7 cells are subcultured into 6-well microtiter plates at a density of 1 × 10 6 cells per well and treated with different concentrations of peptides (FK13, KR12A, KR12B and KR12C) for 24 hours. Total RNA is isolated from both untreated and treated cells using TRIzol method (Invitrogen) as per manufacturer's instructions. 2 g of total RNA is processed for cDNA synthesis and reverse transcribed using a Super MuLV RT Kit (Biobharati Life Sciences Pvt. Ltd.). Real time PCR is performed using Maxima SYBR Green/ROX qPCR Master Mix (2×) (Thermo-Scientific) as per manufacturer's protocol. Housekeeping gene, GAPDH is used as an internal control to normalize the variability in c-MYC mRNA expression levels (Supplementary information Section 1). PCR primers are designed using Primer-BLAST, NCBI, and analyzed in OligoAnalyser 3.1-IDT (Supplementary Table S2).

Oligonucleotide sequences
The putative wild-type and truncated G-quadruplex forming sequences ( Table 2 and Supplementary Table S1), located at the upstream of c-MYC promoter are reconstituted in 10 mM Potassium Phosphate buffer (pH 7.0) having 100 mM potassium chloride (KCl). The sequences are allowed to evolve G-quadruplex structure in solution by annealing at 95 • C followed by gradual cooling. (Supplementary information Section 1).

Circular dichroism
To understand the role of peptides (FK13, KR12A, KR12B, KR12C, KR12D and KR12E) in stabilizing na- tive and truncated isomers of c-MYC G-quadruplex, we have performed temperature-driven CD melting experiments for free and peptide-bound complexes in JASCO-J815 CD spectrometer. We have gradually titrated the peptides into 10 M of putative quadruplexes and acquired the spectra after 5 min of peptide addition to allow complex formation and equilibration. Samples are heated from 20 to 95 • C having the temperature gradient and the delay time of 2.5 • C/min and 150 s respectively. Since the quadruplexes unfold at higher temperatures and revert back to their initial conformations upon renaturation, we have assumed a 'two-state transition' model (folded and unfolded) to analyze the melting curves and estimated the melting temperatures (T m ) by fitting the data points into sigmoidal curves (Supplementary information Section 1).

Isothermal titration calorimetry (ITC)
Thermodynamic attributes of the interaction profiles between peptides and putative quadruplex sequences are analysed by ITC using iTC200 Microcalorimeter at 25 • C. Synthetic peptides (FK13, KR12A, KR12B, KR12C, KR12D and KR12E) are diluted to 20 M in 10 mM Potassium phosphate buffer (10 mM K 2 HPO 4 + 10 mM KH 2 PO 4 ) containing 0.1 M KCl (pH 7.0). Syringe is filled with 500 M quadruplex sequences present in the upstream promoter regions of c-MYC (Pu27, Pu19, and Myc22), BCL-2 (Pu30), VEGF-A (Pu22), KRAS (Pu32) oncogenes, and telomere (Tel26) dissolved into the same buffer. We also carried out the experiment for KR12C and Pu27 using MCF-7 nuclear extract as the solvent to mimic the cellular microenvironment and impart macromolecular crowding during binding events. Same concentration of oligonucleotides is injected into the identical buffer or in nuclear extract without peptides to normalize the heat of dilution of peptidequadruplex association to that of the control before curvefitting. Oligonucleotides are injected 20 times at an interval of 150 s into the calorimeter cell having the peptides to achieve binding saturation. The heat of reaction per injection (cal/s) is determined by integration of the peak areas using in-built Origin 7.0 software that provides the best-fit values of enthalpy ( H) and entropy ( S) of binding, the binding stoichiometry (n), and dissociation constant (K d ) (Supplementary Tables S5(a), S5(b) and S5(c)). Data points are further simulated with 'one-site' binding modes (Supplementary information Section 1).

Molecular docking and simulations
To analyze the stable non-covalent interactions between native c-MYC quadruplex and synthetic peptides (FK13,

Structure calculation and refinement
Three-dimensional structure of Myc22-KR12C complex is calculated following NMR restrained molecular dynamics (MD) simulation. The docked structure of Myc22-KR12C complex (Cluster 4) showing high agreement with the intermolecular NOE cross-peaks is selected to develop the pa-rameter and coordinate files using ff14SB and gaff forcefields. Tleap program of AMBER14 is employed to read the force-fields, coordinates, and topology information. Then, distance restraints are simulated from NOE cross-peak volumes and intensities into NOESY spectra. NOE intensities are qualitatively classified as strong, medium, and, weak having the upper distance limit of 3.0, 4.0 and 6.0Å and the lower limits of 1.8-3.0Å respectively. Then, NMR distances, dihedral restraints, and structure statistics are given for structure refinement (Supplementary Table S6). Restraint files are processed in AMBER14 using nmropt = 1 and then the model is minimized along with the lower and upper bound force constants of 2 kcal.mol −1 . A simulated annealing of 20 ps is further carried out in vacuum with Generalized Born Model. The model is first heated upto 400 K for 10 ps, and then gradually cooled to 300 K in next 10 ps, and finally cooled to 0 K for next 5 ps. The cycle is repeated thrice. PyMOL and Maestro softwares are used to visualize and validate the solution NMR structure (Supplementary information Section 1).

Chromatin immunoprecipitation (ChIP) assays
Chromatin immunoprecipitation studies are conducted to monitor the promoter occupancies of different transcription factors (Sp1, NM23-H2 and Nucleolin) and RNA polymerase II (RNAPII) on the region encompassing c-MYC-NHE III 1 (69). We have grown MCF-7 cells in 10cm 2 culture flask at a density of 1 ×  Table S4) specific to the quadruplex-enriched regions at c-MYC promoter and Phusion® High-Fidelity PCR Kit (NEB). Anti-rabbit IgG is employed for mock immunoprecipitation (Negative control).

Western blot
To observe the effect of KR12C in c-MYC protein expression, and to decipher the peptide-driven signaling path- Next membranes are incubated with peroxidase conjugated anti-rabbit IgG raised in Goat (Sigma, 1:8000 dil.) and peroxidase conjugated anti-mouse IgG raised in Rabbit (Sigma, 1:10 000 dil.) for 2 h at room temperature. Target proteins are visualised on membranes using Clarity TM Western ECL substrate (Bio-rad) and images are captured using ChemiDoc MP system with Image Lab™ software for PC version 6.0 (Bio-rad).

Design and synthesis of the peptides from naturally occurring human Cathelicidin protein, LL37
We constructed a small library of six peptides by pruning the quadruplex interacting domain of LL37 and further by substituting and mutating the position of the amino acids in the peptide sequence (Table 1 and Supplementary Figure S1). We considered few criteria to enable the peptides with anti-cancer properties and selective affinity for c-MYC quadruplex. (i) The amphipathic and cationic nature of the peptides are retained such that the net positive charge ranges between +2 to +9 at pH 7. Due to higher expression of phosphorylserines and O-glycosylated mucins, cancer cell membrane carries a net negative charge. Therefore, we put Arginine and lysine residues at termini to attract more phosphates and water to make stable amino acid-phosphate clusters across plasma membranes, which plays a central role in lipid membrane disruption and permeabilization (71)(72)(73). (ii) We incorporated lysines and leucines at different positions of the peptide to increase the ␣-helical stability as they are more prone to pass through the plasma membrane. (iii) Due to unbranched side chains and accessible rotamers, lysines enrichment is hypothesized to make more electrostatic interactions with quadruplex sugar-phosphate backbone in the loop regions of c-MYC quadruplex having distinct topologies. (iv) Phenylalanine and tryptophan are rationalized to make stacking interaction to improve selectivity for the quadruplex structure.

Intracellular selectivity of synthetic peptides for Gquadruplex structure at c-MYC promoter
Peptide candidates suitable for the therapeutic interventions are expected to selectively interact with G-quadruplex scaffolds at the target oncogenes over duplex and other oncogenic quadruplexes under cellular environment. To evaluate the intracellular conformer selectivity of the peptides for c-MYC quadruplex isomers (Pu27, Myc22 and Pu19) over other oncogenic quadruplexes (BCL-2, VEGF-A and KRAS) of different folding topologies, we performed dualluciferase assays in MCF-7 cells transfected with the luciferase constructs containing Renilla luciferase coding sequences downstream of the G-quadruplex scaffolds of the oncogenes of interest ( Figure 1A and Supplementary Figure S2A). Earlier studies suggested that the peptides alter their secondary structures depending upon cellular microenvironment. In addition, conformational flexibility of the peptides is significantly implicated in their effective functional expression, diversification, and ability to penetrate the cell membrane (74,75). We determined the magnitude of promoter inhibition of the target oncogenes with an increasing gradient of peptides' treatment (0-60 M) considering that the stable association of peptides and quadruplexes in MCF-7 cells at the oncogene promoters would drive lower expression of downstream Renilla luciferase in contrast to the GQ-null plasmids. We observed that the constructs lacking G-quadruplex motifs (GQ null) in the oncogene promoters were constitutively activated irrespective of the peptide concentrations, while the promoters having Gquadruplex structures of various folding topologies manifested differential luciferase activities reflecting differential regulation of oncogene promoters by the peptides ( Figure  1B-E). KR12D and KR12E had no significant effects upon oncogene promoters regardless of the upstream quadruplex motif(s), which envisaged their incompetence to recognize oncogenic quadruplexes in cellulo. However, at higher concentration, KR12D showed meager inhibitory effect (∼1.6-fold) upon KRAS promoter activity ( Figure 1D and F). FK13 and KR12A abrogated c-MYC promoter activity (∼2-fold repression at 60 M) dose-dependently. Coupled with that, they were observed to significantly downregulate the promoter activation of VEGF-A and BCL-2, which indicated their failure to discriminate different oncogenic quadruplex conformers and non-specific binding with different quadruplexes in the cells ( Figure 1B, C, E and F). KR12A further repressed KRAS promoter activity (∼2.5-fold) at higher concentrations suggesting its weak interaction with KRAS quadruplexes in cellulo ( Figure 1D and F). KR12B and KR12C selectively attenuated c-MYC promoter activation in a dose-dependent manner. While KR12B was found to show inverse regulation of BCL-2 and KRAS promoters at higher doses, KR12C had no significant impact upon other oncogenic promoters, even at Relative promoter activity is determined by normalizing the Rluc/Fluc values to that of the cells transfected with P1-P2 promoter construct (GQ-null), having no quadruplex-forming motif. Error bars represent mean ± SE (N = 5). Statistical differences in the luciferase activities compared to the respective GQ-null constructs of the target oncogenes used one-way ANOVA followed by Tukey-Kramer Test ( # P < 0.05, ## P < 0.01, ### P < 0.001). Statistical differences in the luciferase activities compared to that of the peptide-untreated cells used two-tailed Student's t test (*P < 0.05, **P < 0.01, ***P < 0.001). (F) Schematic diagram of the peptides having differential selectivity to oncogenic quadruplexes with various folding topologies and differential regulation of the promoter activies of target oncogenes by the peptides. higher concentrations, conferring its conformer selectivity for c-MYC quadruplex in cellulo ( Figure 1B, C, D and F). These results prompted us to investigate the intracellular selectivity of peptides for different loop isomers (Myc22 and Pu19) of native c-MYC quadruplex (Pu27). As per earlier investigations, Pu27 enables high likelihood of conformational dynamics in the cellular microenvironment involving dynamic shuffling of multiple overlapping quadruplex conformers (10,27). We monitored if these peptides could negatively regulate c-MYC promoter in presence of the biologically relevant and stable loop isomers (Myc22 and Pu19) of Pu27 (Supplementary Figure S2B). Interestingly, we observed that KR12C led to promoter attrition (∼5-fold) in presence of Myc22 and Pu19 isomers while the inhibitory role of KR12B remained weaker in Pu19 ( Supplementary   Figures S2C and S2D). FK13 and KR12A, at higher concentrations, attenuated c-MYC promoter activation under the control of Myc22 and Pu19 (Supplementary Figures  S2C and S2D). Consistent with the earlier data, KR12D and KR12E exerted no regulatory effect in c-MYC promoter activation in presence of Myc22 and Pu19 conformers. This experiment allowed a rapid screening of synthetic peptides for their selective targeting at c-MYC quadruplex in the cells driving c-MYC promoter repression.

Comparative analyses of the interaction profiles between peptides and c-MYC quadruplex isomers
Luciferase assays provided a real time read out of c-MYC promoter regulation and quadruplex selectivity by synthetic peptides underpinning their different behaviour in c-MYC Nucleic Acids Research, 2018, Vol. 46, No. 19 9939 quadruplex interaction under cellular environment. This observation led us investigate the thermodynamic parameters of in vitro binding reactions between the peptides and different oncogenic quadruplexes (c-MYC (Pu27, Myc22, and Pu19), BCL-2, VEGF-A, KRAS) and telomeric quadruplex by Isothermal titration calorimetry (ITC). We slowly titrated different quadruplex sequences into synthetic peptides dissolved into Potassium phosphate buffer (pH 7.0) and analysed the integrated heat release of the titrations to explain the thermodynamic partitioning of binding free energy ( G) between enthalpic and entropic components. Indeed, we witnessed a series of enthalpy-dominated exothermic binding reactions between peptides and the quadruplexes since the binding enthalpy ( H) overcompensated for the unfavorable loss in conformational entropy ( S) (Figure 2A and Supplementary Figures S3-S5). The net entropy changes ( S) were negative values, which might attribute to the restriction of the conformational freedom at Pu27 interacting sites and release of the constrained water molecules upon complex formation (Table 3 and Supplementary Tables S5(a)-(c)). The magnitude of H were relatively more negative than S, hence contributed substantially to the binding free energies during the interactions (Table 3 and Supplementary Tables S5(a)-(c)). Consistent with the previous results of dual-luciferase assays, sitespecific substitution of the amino acids in peptides offered differential magnitude of enthalpy-entropy compensation suggesting varying mode of interactions during the binding reactions between peptides and different quadruplexes. Replacing Q5 (in KR12A) with K5 in KR12B, KR12C, and KR12D significantly lowered their binding dissociation constants (k d ) for Pu27 and its isomers as well as telomeric quadruplex (Table 3 and Supplementary Tables  S5(a) and S5(b)). The amide side chain of Glutamine does not ionize at physiological pH while ε-amino group of lysine (pK a > 10) remains positively charged at pH 7.0. This enabled the hydrogen donor ammonium (NH 3 + ) in Lysine to favour the electrostatic interactions with the negatively charged sugar-phosphate backbone of quadruplex DNA resulting the binding H relatively more negative. In contrast, FK13, which also has a glutamine at fifth position, showed enhanced affinity for Pu27 compared to KR12A, KR12B, KR12D and KR12E despite having lower binding affinity with the isomers of Pu27 and the telomeric quadruplex. Earlier studies showed that phenylalanines possess additional stability due to -electron cloud across their aromatic rings, which enable them to make aromaticaromatic interactions (76) and stacking interactions with nucleic acids. Phenylalanines also play important role in stabilizing ␣-helix and phe-phe pair in peptides are enthalpically favoured to impart additional stability in peptide-DNA complexes (77,78). FK13 has two phenylalanines at 1st and 11th positions, which might impart additional stability to the helicity of the peptide upon binding with Pu27 rendering improved affinity towards Pu27. On the other hand, single-residue replacement (R12 > L12) in KR12E at C-terminus severely compromised the binding affinity of KR12E for Pu27, Myc22, Pu19 and telomeric quadruplexes (Table 3 and Supplementary Tables S5(a) and S5(b)). Since the guanidium moiety of R12 has a greater intrinsic pK a (<∼12) allowing its side chains to be invariably proto-nated at neutral pH (79), the peptides with C-terminal Arginine preferably made electrostatic interactions with quadruplex backbone, which concomitantly enhanced their binding affinity (except KR12E) for c-MYC quadruplexes. Furthermore, the conformational flexibility of Arginine-side chain at C-termini enabled the guanidinium group to search for polar atoms to satisfy its propensity to donate up to five hydrogen bonds to make more stabilized H-bonded network with the phosphate backbone of the quadruplex. This corroborated with the inefficiency of KR12E to recognize c-MYC quadruplex into NHE III 1 in the cellular context. KR12B showed moderate binding affinity towards c-MYC quadruplexes and telomere. It was also found to inversely regulate the promoter activity of other oncogenes. Indeed, KR12B exhibited enhanced affinity for the quadruplexes compared to FK13 and KR12A due to having a greater net positive charge (+7). However, it failed to discriminate different quadruplexes like that of FK13 and KR12A under cellular milieu. This accounted for their non-specific interactions with different quadruplex topologies rendering FK13 and KR12B to be equally effective in reducing c-MYC promoter activity compared to KR12C. On the contrary, KR12C demonstrated ∼2-fold higher binding H for Pu27 association and ∼5 times higher affinity for Pu27 (K d = 8.01 M) and its conformationally restricted biologically significant conformers (Myc22 and Pu19) compared to KR12B and KR12D (Table 3 and Supplementary Table S5(a)). This might account for the spatial position of K5. K5 in KR12C was buried in a non-polar environment in KR12C while was followed by a positive charge (R6) in KR12B. The pK a of lysine side chain (εamino group) is subject to fluctuate depending upon the neighbouring residues (80,81), which might have altered its localized charge distribution and accessibility to make salt bridge interactions. In addition, KR12C showed very poor or no binding affinity with other oncogenic quadruplexes (BCL-2, KRAS and VEGF-A). It binds to BCL-2 and telomeric quadruplex having ∼3and ∼4-fold decreased affinity compared to c-MYC quadruplexes (Supplementary Figure S4 and S5 and Supplementary Table S5(b-c)). This was also in agreement with intracellular conformer selectivity of KR12C for c-MYC quadruplexes. However, peptide conformation is subject to change under cellular microenvironment. Therefore, peptide affinities for c-MYC quadruplex would be radically different in the cells compared to the potassium phosphate buffer. To address this concern, we reiterated the ITC experiment between KR12C and c-MYC quadruplex (Pu27) using MCF-7 nuclear extract as the solvent to mimic the cellular microenvironment and impart macromolecular crowding (82). We observed as significant as ∼8-fold increase in the binding affinity (K d = 901 nM), which envisages a sharp enhancement in c-MYC quadruplex selectivity due to the crowding effect of macromolecules in MCF-7 nuclear extract ( Figure 3C).
Since Pu27 exhibits conformational dynamics between different isomers, its conformational entropy is correlated to a broadened energy landscape which enables free Pu27 to sample more conformational states than its peptide-bound counterparts (83). To evaluate if these peptides could alter the secondary structure of the quadruplexes and increase its thermal stability by decreasing the entropy of quadruplex   unfolding, we titrated the peptides into G-quadruplexes and quantitated the changes in the melting temperature ( T m ) of Pu27 and its conformers (Myc22 and Pu19) upon peptide binding ( Figure 2B and Supplementary Table S5(a), Supplmentary Figures S6(A) and S6(B)). These peptides did not significantly affect the secondary conformation of the quadruplexes. However, striking find of the study was that KR12A and KR12E, which exhibited lower binding affinity as well as non-specific and negligible intracellular selectivity for c-MYC quadruplex, decreased the T m of Pu27 by ∼30 • C rendering conformational destabilization. In contrast, KR12C, which offered maximal affinity and intracellular specificity for Pu27 and its conformers (Myc22 and Pu19), promoted a sharp rise in the T m by 9 • C (Pu27), 6.91 • C (Myc22), and 10.87 • C (Pu19). This signified that KR12C could adopt a flexible ␣-helical conformation that allowed the peptide to conform to the flexible skeleton of Pu27 and stabilize the biologically significant quadruplex isomers of native c-MYC quadruplex ( Figure 3A and B). This might attribute to the lysine in KR12C, which is known to produce a much greater population of accessible rotamers compared to arginine and glutamine causing enhanced stability of Pu27 by KR12C and a sharp decrease in the conformational entropy of Pu27 ( S) during unfolding (84). KR12B and KR12D showing moderate affinities for Pu27 did not significantly induce thermal stability. These observations gave a fair explanation to the varying intracellular specificity of peptides for c-MYC quadruplexes and their roles in differential promoter regulation.

Differential affinity of peptides for Pu27 depends on their tethering point at the quadruplex and non-covalent interactions
The thermodynamic profiling of peptide-Pu27 association and their differential selectivity for Pu27 under cellular microenvironment impinge on the assessment of intermolecular interactions. Unfortunately, the structural ambiguity in Pu27 limits the mapping of the binding interface using solution NMR studies due to the broad envelope of overlapping 9-guaninyl-N1(H) resonances in the imino proton region (␦, 10-12 ppm) of Pu27. To resolve the problem, we performed constrained molecular docking between Pu27 and the peptides. We established docking benchmarks by clustering similar binding poses. The clusters with root mean square deviation (RMSD) of 1.0-2.0Å were considered as the near-native conformations and taken up for the molecular dynamics (MD) simulation studies for the validation and evaluation of the intermolecular interactions (Supplementary Table S3). KR12C was involved in a multitude of electrostatic and weaker CH 3 -interactions to impart selectivity and stability to Pu27 structure ( Supplementary Figure S7A), confirmed by the average RMSD of the binding interface during simulation (Supplementary Figure S7B). The guanidino group of R12 attached to the sugar phosphate backbone of A6 by electrostatic interaction (Supplementary Figure S7C). Similarly, the penultimate R2 at Nterminus was clung to the sugar-phosphate backbone of G14 and G16 (Supplementary Figure S7D). These two in-teractions were highly stable and observed in more than 80% of the complexes in last 10 ns of the MD trajectories. They protected the G-quartet proximal at the 5 -propeller loop by gripping the backbone of the guanines delineating the quartet. Furthermore, L11 and I7 came into spatial proximity of the terminal quartet and stabilized the reverse Watson-Crick hydrogen bonding between A6 and A15 via CH 3 -interactions (Supplementary Figure S7E). These were weaker interactions, wherein the delocalizedelectrons of Adenines served as the acceptors and the CHvector directly pointed at the ring centre and was collinear with the ring normal, which maintained the integrity of the G-quartet (Supplementary Figures S7E and S7F) FK13 involves R3 to make bisected electrostatic interactions with G3 and G4 bases and R7 to clip G14 backbone. In addition, the side chain of F11 is spatially inclined between 130 • to 140 • with the sugar backbone of Pu27 near 5 -terminus (Supplementary Figure S8A). Earlier studies showed that this type of interactions mediated by phenylalanines induce non-specific interactions at DNA grooves (85). Similar interactions were reflected in FK13-Pu27 complex, which agrees with the non-specific binding between FK13 and quadruplexes. In KR12A, K5 was replaced with Q5. This altered the helical property of the peptide during binding reaction such that L11 and I7 were far apart giving no protection to the A6-A15 pair. This destabilized the quartet, which corroborated with the decline of the thermal stability as observed in the CD melting experiment. However, KR12A was still able to tether to the quadruplex at C-termini wherein R12 and R6 played the key role to make H-bonds with sugar backbone of G6 and G20 respectively (Supplementary Figure S8B). In KR12B, R12 and K9 made an electrostatic interaction with G3 and G14 backbones respectively while R2 gripped G11 backbone of Pu27 via strong electrostatic interactions. L6>R6 replacement perturbed the helicity of KR12B that could no longer induce further stability to the quartet. KR12B formed a network of electrostatic interactions in quadruplex loop via terminal residues, which enhanced its affinity for quadruplex DNA at the cost of compromised selectivity for c-MYC quadruplex (Supplementary Figure S8C). In KR12D, inclusion of V6 incapacitated the helical stability, which hindered quadruplex stability and compromised its binding affinity for quadruplex. R2 and R12 developed transient hydrogen bond interactions with G11, G13, G4 and G5 backbones respectively, which accounted for its poor affinity to Pu27 (Supplementary Figure S8D). In KR12E, C-terminal R12 was replaced with L12 rendering it unable to grip the quadruplex backbone. Only the N-terminal K1 made transient H-bond with G3 backbone. R2 established electrostatic interactions with G4 backbone and K5 made transient polar contact with G11 (Supplementary Figure S8E). These interactions were less prevalent (observed in <60% in MD trajectory) due to solvent exposure that allowed water molecules to sequester the H-bond offering weaker interaction between KR12E and Pu27. This was also consistent with our previous results of KR12E-induced structural destability and poor binding affinity.

Site-specific substitution of amino acids is the deterministic factor of quadruplex unfolding and stability
Stable association of KR12C at the 5 -quartet of Myc22 by solution NMR structure determination. So far, our investigations suggested opposing role of KR12A and KR12C in c-MYC quadruplex stability and binding affinities. MD simulations indicated that terminal positively charged residues of the peptides drove major intermolecular cationic contacts between KR12A-Pu27 and KR12C-Pu27. However, KR12A and KR12C differed by amino acid substitutions at fifth and sixth positions albeit having identical net charge (+6). To gain a detailed insight into the binding interface, intermolecular interactions, and the plasticity of the quadruplex structure upon peptide binding, we determined the solution NMR structure of Myc22-KR12C complex ( Figure 4A). In this study, Myc22 was regarded as the biologically relevant quadruplex structure of Pu27 because it yields unambiguous assignment of the imino proton resonances in proton NMR spectra and retained the potential to downregulate c-MYC promoter activation. We performed 1 H-1 H NOESY (Nuclear Overhauser Effect Spectroscopy) experiment of unbound Myc22 and Myc22complexed with KR12C at 6:1 molar ratio. NMR samples were prepared into 100 mM KCl and 10 mM potassium phosphate at pH 7.0 . G-tetrad alignment was determined by the intra-residual connectivities of H1 and H8 protons in exchangeable proton NOESY spectra since base H8 of one guanine comes into closer vicinity to imino H1 of adjacent guanine upon Hoogsteen H-bond formation in the G-tetrad (Supplementary Figure S9). The H8/H1 intensities of G4/G8, G8/G13, G18/G13 and G10/G6 in the complex were significantly increased, which corroborated with KR12C driven enhanced stability of the quadruplex structure (Supplementary Figure S9). The glycosidic angles remained in anti-conformation, as confirmed by the H8-H1 NOE (nuclear overhauser effect) intensities. We observed the broadening of the imino signals between 10 to 12 ppm region upon KR12C titration, which indicated intermediate to slow exchange between the free and bound states or preequilibrium conformational change during binding reactions or the structural rearrangement and allosteric changes of the complex. The intraresidual H1 /H2 and H4 /H1 connectivites and interresidual NOE signals between H1 of guanines and NH of amino acids supported that KR12C made electrostatic interactions with the sugar-phosphate backbone of quadruplex to alleviate the stability (Supplementary Figure S10). K9 and K5 strongly interacted with the sugar phosphate backbone of G17 and G13, which resulted moderate NOE signals between Hb of K9, K5 and H1 of G17, G13 respectively ( Figure 4B and C). Another electrostatic interaction between R12 and G8 backbone resulted strong Hb/H1 crosspeak intensity, which in turn enabled the G8 base into spatial proximity of R12 reflecting weak NOE between N1(H1) of G8 and R12(NH) (Figure 4D and E). These compelling evidences confirmed that KR12C was associated at the 5 -quartet plane of Myc22 ( Figure 4A) as to which weak to moderate intermolecular NOE signals were found between N1(H1) of G4, G17, G13 and NH of R12, W10 and K9 respectively ( Figure 4F, G and H).

KR12A destabilizes c-MYC quadruplex by unfolding the quartets to single-stranded hairpin loop-like conformation.
Opposed to KR12C-Myc22 association, we observed the evolution of 1D proton signals between 12.5 and 14 ppm in KR12A-Myc22 complex, which reflected conformational change in the quadruplex upon KR12A titration ( Figure 5A). Upon further titration at 12:1 molar ratio (KR12A:Myc22) the imino signals between 10 to 12 ppm completely disappeared whereas the signals at 12.5-14 ppm sustained and were significantly broadened ( Figure  5B). This observation validated our earlier finding, which showed significant depletion (∼30 • C) of the melting temperature of Pu27 and Myc22 upon KR12A binding. Based on these experimental observations, we conducted constrained docking of Myc22 and KR12A and further simulated the complex for 100 ns, which clearly detected the per- turbation of the quartets upon KR12A interaction ( Figure  5C). We mapped the binding interface at atomic level, which evidenced strong electrostatic interactions between R6 and G20, K8 and G18 ( Figure 5D). In contrast to KR12C, KR12A preferred to target the 3 -termini of Myc22 quadruplex, which was further substantiated by significant broadening and downfield shift of the proton resonances near the binding site ( Figure 5B-D). This denotes greater exposure of the imino protons to the solvent, which sequestered the stable Hoogsteen bonding and resulted in conformational fluctuation. We also observed formation of reverse Watson-Crick interaction between G4-G17, G5-G18 and G14-G18. Furthermore, G18 was involved to maintain an equilibrium between G5 and G14 via Hoogsteen and Reverse Watson-Crick interactions respectively. We observed Guanine-guanine (G-G) pairing via Hoogsteen interaction (G13-G17, G15-G19, G4-G13 and G14-G18) and Reverse Watson-Crick interactions, which contributed to the signals between 12.5 and 13 ppm ( Figure 5E). To our surprise, we found an unusual G-G interaction involving N7imino(N1(H1)) and amino(N2(H2))-O6, which occurred in 80% of the MD trajectories ( Figure 5F). These interactions might have further caused downfield shift of the 1D signal at 13.5 ppm due to enhanced solvent accessibility of the base pair.

Peptides regulate the occupancy of Sp1, NM23-H2 and Nucleolin at c-MYC quadruplex motif to downregulate c-MYC transcription
We have earlier witnessed that FK13, KR12A, KR12B and KR12C inhibited c-MYC promoter activation in MCF-7 cells albeit showing differential affinity or selectivity to Pu27 and other quadruplex topologies. To substantiate these results, we determined the differences in the occupancy of RNA Polymerase II (RNAPII) across the promoter upon peptide treatment and estimated the level of c-MYC transcripts in MCF-7 cells upon increasing peptide concentrations ( Figure 6A and B). Consistent with the earlier results, KR12B and KR12C significantly depleted RNAPII recruitment in the promoter and repressed c-MYC transcription by ∼5-fold. Interestingly, FK13 did not significantly altered RNAPII occupancy across c-MYC promoter and did not downregulate c-MYC transcription despite having an inverse role in c-MYC promoter repression in MCF-7 cells. KR12A downregulated c-MYC transcription in a dose-dependent manner and also depleted RNAPII occupancy in the promoter albeit having poor affinity on the quadruplex ( Figure 6A and B). However, the magnitude of transcription repression by KR12A is highly reduced compared to KR12C and KR12B. This might be due to their non-specific binding to other oncogenic quadruplexes as shown in the luciferase assays and their associated off-target effects. FK13 and KR12A also abrogated the transcriptions of other oncogenes (BCL-2, VEGF-A, KRAS), which are known to regulate c-MYC transcription through an intricate network (Supplementary Figure S11). Therefore, the inverse effects of FK13 and KR12A in c-MYC promoter inhibition might be integrated with their interferences in the upstream signalling pathways, which compromised the stable complexation with c-MYC quadruplex in the cells.
To gain a deeper insight into the role of peptides in c-MYC transcription, we monitored the occupancy of Sp1, NM23-H2 (Nucleoside diphosphate kinase), and Nucleolin in NHE III 1 . Earlier reports claim that Sp1-induced negative super helicity declines nucleolin occupancy at NHE III 1 driving an increased c-MYC transcription (86). There are reports that NM23-H2 disentangles c-MYC quadruplex to enhance the rate of transcription while Nucleolin, which attenuates the promoter activation by stabilizing Pu27, antagonizes this effect (87,88). In this study, we conducted Chromatin immunoprecipitation studies on a region encompassing c-MYC-NHE III 1 and monitored the changes in promoter occupancy of Sp1, NM23-H2 and Nucleolin upon peptide treatment ( Figure 6C). We observed that KR12C depleted the enrichment of Sp1, NM23-H2 and nucleolin at c-MYC-NHE III 1 while FK13 exhibited no role in the recruitment or depletion of these transcription factors at Pu27 (Figure 6D-G). Since Sp1 and NM23-H2 are the positive regulators of c-MYC transcription, a significant de-pletion in their recruitment across NHE III 1 by KR12C accounted for its selective binding with c-MYC quadruplex under cellular microenvironment and its inverse effect in c-MYC transcription regulation. KR12B had no role in Sp1 occupancy but depleted the abundance of NM23-H2 and nucleolin ( Figure 6D-G). Therefore, the inhibitory effects of KR12B in c-MYC transcription and promoter activation were associated with its non-specific interactions with other oncogenic quadruplexes and off-target effects because c-MYC transcription is stringently regulated by cellular crosstalk involving the congregation of different transcription factors across the P 1 /P 2 promoter (Supplementary Figure S11). Smits et al. showed that the decreased expression of miR-125b by VEGF results in an upregulation of MAZ (Myc-associated zinc finger protein) expression, which specifically binds to a GA box sequence (GGGAGGG) in c-MYC-P 2 promoter to regulate c-MYC transcription (88,89). Since FK13 and KR12B strongly inhibited VEGF-A promoter activity and its transcription, the negative effects of these peptides on c-MYC promoter might be due to their non-specific targeting at VEGF-A quadruplexes. KR12A did not significantly affect Nucleolin recruitment but slightly enriched NM23-H2 and Sp1 occupancy ( Figure 6D-G). Since KR12A destabilized the quartets facilitating the passage of G-rich motif into hairpinloop like structure, it favoured the recruitment of NM23-H2 at Pu27. However, KR12A did not unfold c-MYC quadruplex in a transcriptionally active form. Instead, it destabilized the tetrad core allowing the formation of reverse Watson-Crick G-G pairs and Hoogsteen base pairs; i.e., significantly different from the quartet core observed in the canonical G-quadruplexes. This type of secondary structures were highly flexible and less stable opposed to the quadruplex structures as observed in the CD melting experiments. However, such conformation can potentially act as a roadblock for RNAPII as evidenced by our ChIP results. This is a plausible explanation why KR12A depleted the abundance of RNAPII across NHE III 1 which resulted in a downregulation of c-MYC transcription in MCF-7 cells. However, we observed that KR12A nonspecifically interacted with other oncogenic quadruplexes and repressed their transcription like FK13 and KR12B. Therefore, KR12A-mediated off-target effects and its involvement in the upstream signalling cascade could also affect c-MYC transcription. Wierstra I et al. demonstrated that FOXM1c (Forkhead Box M1) transactivates P 1 /P 2 promoter synergistically with Sp1 (86,87). KR12A was also found to show poor affinity to c-MYC quadruplex and enhance Sp1 recruitment in NHE III 1 , which further corroborated why it could not strongly abrogate c-MYC transcription compared to KR12B and KR12C.

KR12C drives apoptosis signalling cascade via E2F-1-VEGF-A-BCL-2 axis in MCF-7 cells
So far, we observed that KR12C is the best peptide candidate that selectively repressed c-MYC promoter activation in cancer cells. Earlier reports showed its ability to trigger apoptosis in cancer MCF-7 cells (89). We further validated these results through monitoring the expression profiles of different oncogenes and apoptosis markers in order to deduce the mechanism how G-quadruplex stabilization at c-MYC promoter triggered selective apoptosis in cancer cells. We observed significant upregulation of p53, cleaved PARP (poly (ADP-ribose) polymerase), caspase 8 (cysteineaspartic proteases), and APAF-1 (apoptotic protease activating factor-1), which had been well studied to positively regulate the apoptotic cascade ( Figure 7A and C). To our surprise, we observed that KR12C suppressed BCL-2 expression at both transcription and translation level despite having no regulation upon its promoter under the control of BCL-2 quadruplex ( Figure 7A and B). This result envisaged that quadruplex stabilization at c-MYC promoter by KR12C turned on a specific downstream cascade as to which the expression of BCL-2 was depleted to such an extent to promote apoptosis in MCF-7 cells. To unravel this downstream pathway, we checked the expression of E2F-1 and VEGF-A at transcription and protein level, which were the upstream oncogenes of BCL-2 and significantly regulate its transcription. Compelling evidences are available about the crosstalk between c-MYC and VEGF-A and their positive role and overexpression in breast cancer progression (90). VEGF-A is also found to stimulate the growth of breast cancer cells in a c-MYC dependent manner in xenograft models (91). c-MYC binds to an E-box motif at VEGF-A promoter which positively regulates its transcription in cancer cells. Another study revealed that c-MYC activates miR-9, which in turn elevates VEGF-A expression via ␤-catenin signalling pathway (92). c-MYC also upregulates E2F-1 transcription (93), which is shown to positively regulate VEGF-A expression (94,95). We observed that VEGF-A and E2F-1 were significantly suppressed due to repressed transcription and lower abundance of c-MYC ( Figure 7C and D). Furthermore, VEGF-A promotes survival in both tumor and endothelial cells and prevents apoptosis via inducing BCL-2. This observation was strengthened by the treatment with recombinant VEGF (rVEGF) which partially rescued the inhibitory effects of KR12C, highlighting the role of VEGF-A in this context (Supplementary Figures S12A-C). Our data complied with the available reports and showed that KR12C is arresting c-MYC transcription leading to decreased survival and apoptosis via perturbing the VEGF-BCL-2 axis (Figure 8).

DISCUSSION
Selective targeting of the oncogenic quadruplexes by therapeutic peptides raised great expectations for future pharmacological and clinical applications due to their variability and plasticity of the three-dimensional structures and potential to mimic protein-quadruplex interactions, which manifest into enhanced intracellular quadruplex affinity and selective modulation of the oncogenic functions with minimal off-target effects in cancer cells. Since native quadruplex structures upstream c-MYC promoter exhibit structural heterogeneity allowing optimal kinetic advantage for the shuffling of the transcription factors, small molecules often fail to conform to the flexible conformation of the target quadruplex resulting non-specific binding to other oncogenic quadruplexes in the cells. Therefore, the clinical pharmacology of quadruplex-based treatment is still in its dawn albeit offering a promising avenue for anticancer therapies.
In this study, we designed six peptides by pruning the quadruplex-binding domain of a naturally occurring human cathelicidin peptide (LL37) and substituted the amino acids in the peptide sequences with the rationale of selective anti-tumor properties and enhanced quadruplex affinity. First, we conducted dual-luciferase assays for rapid screening of the synthetic peptides to identify the candidates having preferential targeting at c-MYC quadruplex over other intracellular quadruplexes of different folding topologies. We observed that KR12A, KR12B and FK13 exhibit nonspecific targeting at oncogenic quadruplexes while KR12C specifically suppressed c-MYC promoter activation via stabilized interaction with the wild-type quadruplex upstream its promoter region. We analysed differential selectivity of the peptides for c-MYC quadruplexes by thermodynamic profiling of their binding reactions in ITC experiments, which again supported that KR12C had maximum binding affinity for c-MYC quadruplex and its isomers (Myc22 and Pu19) while rest of the peptides showed poor and moderate binding to the quadruplex element. The binding affinity of KR12C for c-MYC quadruplex is enhanced by ∼8 fold under macromolecular crowding of MCF-7 nuclear extract, which suggests that the volume exclusion by the increased chemical activity of the macromolecules in nuclear extract increases binding affinity by favouring the association of KR12C and c-MYC quadruplex. This underscores intracellular selectivity of KR12C for c-MYC quadruplexes. We correlated the differences in the magnitude of binding  affinities with specific non-covalent interresidual interactions by in silico modelling of the peptide-quadruplex complexes and their simulation for 100 ns in explicit solvent. We observed that the electrostatic interactions by positively charged amino acids (Arginine and Lysine) prevailed in the sugar-phosphate backbone of the quadruplexes. However, the binding affinity and selectivity is driven by differential helical properties of the peptides, which alleviate CH 3interaction with the tetrad plane and Hoogsteen hydrogen bonding between the adenine residues (A6 and A15). These interactions are pronounced in KR12C and Pu27, which accounts for its selective interaction with c-MYC quadruplex.
In most cases, ligands with aromatic/pseudoaromaticdelocalized system are developed to bind c-MYC quadruplex via externalstacking. These compounds generally suffer from non-specific target recognition resulting toxic off-target effects (e.g. porphyrine derivative TmPyP4(41), perylene derivative PIPER (42,43), quindoline derivative SYUIQ-05(44)) since around 3,70,000 nucleic acid sequences in human genome have the propensity to fold into quadruplex structures and maximum of the lead compounds offer G-quadruplexe-to-duplex selectivity through stacking interaction. Therefore, the loop characteristics of c-MYC quadruplex should definitely be considered (i.e., the length and folding pattern of the loops; (e.g. diagonal or propeller-loop), the groove width, torsion angles of glycosidic bonds, strand polarity) for designing a specific drug candidate. For example, c-MYC and c-KIT quadruplexes allow ligand accessibility at 5 and 3 terminal quartets, whereas telomeric and BCL-2 quadruplexes (1:13:1 isomer (Pu39WT) and 1:1:11:1 isomers (P1G4)) restrict this access of due to steric clashes at the terminal quartet regions. The groove width of the BCL-2 quadruplex is also significantly different from others. (e.g. groove widths of BCL-2 quadruplex (PDB: 2F8U) are 10.3, 6.4, 17.5 and 7.7Å, in c-MYC (PDB: 1XAV) the groove widths are 9.7, 14.9, 12.7, 11.9Å). Considering these unique properties of quadruplex loops in c-MYC quadruplex, we put more emphasis on selective interaction in quadruplex loops by the peptides. In Pu27, the planar conformation of A6 and A15, stacked over the first quartet generates a specific binding pocket for KR12C at the 5 -terminus groove of c-MYC quadruplex. The peptide, KR12C makes highly stable Coulombic interactions with the sugar-phosphate backbone at the 5propeller loop to strengthen A5-A16 Reverse Watson interaction. This interaction between A15-A6 is further stabilized by L11 and I7 via CH 3 -interactions, which gives a bona fide advantage to maintain the integrity of the quartet core of c-MYC quadruplex without external stacking interaction, opposed to other known quadruplex-targeting small compounds. Solution NMR structure calculations were also consistent with our premonitions as KR12C is preferentially targeted at the vicinity of the 5 -tetrad plane of Myc22 (the restricted conformer of Pu27), which in turn stabilized the 5 -propeller loop by a network of electrostatic interactions supported by strong, moderate, and weak interresidual crosspeak intensities at the proximity of the quartet and sugar backbone. These interactions not only confer quadruplex-to-duplex selectivity but also specifically target c-MYC quadruplex under cellular microenvironment. This observation is further supported by a sharp rise (∼8-fold) in the binding affinity (K d = 901 nM) of KR12C for Pu27 under macromolecular crowding. In contrast, we witnessed the unfolding of c-MYC quadruplex by KR12A that resulted formation of reverse Watson-Crick interactions and guanine-guanine Hoogsteen pairing at the G-rich motif. This accounts for the depletion of quadruplex thermostability and enhancement of the entropy of unfolding.
Our investigations further perpetuate with the role of peptides in regulating the promoter occupancy of transcription factors (NM23-H2 and Nucleolin) at c-MYC quadruplex. We observed that KR12B and KR12C inhibit the recruitment of NM23-H2, a protein known to upregulate the transcription by unfolding the quadruplex into singlestranded motif. However, KR12A increases NM23-H2 enrichment in the quadruplex as similar as the control and provides minimal repression in the transcription. Therefore, the flexible skeleton of KR12C not only adapts an optimized posture upon Pu27 binding, but its selective contacts at its loop region induces allosteric changes in the receptor such that NM23-H2 no longer binds to the native quadruplex. This led us further investigate how selective stabilization of Pu27 by KR12C propagates downstream signalling cascade to promote apoptosis in cancer cells. We deduced that selective arresting of c-MYC promoter via stabilizing G-quadruplex activates a unique crosstalk with E2F-1 and VEGF-A, that concomitantly reduces BCL-2 expression triggering apoptosis in cancer cells. Our study improves the insight of increasing the selectivity of the peptides for c-MYC quadruplex at the atomic level and opens a new paradigm for next generation quadruplex-targeting peptides with higher therapeutic index and minimal offtarget effects in the cells.