Deciphering the Interplay Between G‐Quadruplexes and Natural/Synthetic Polyamines

Abstract

The interplay between polyamines and G‐quadruplexes has been largely overlooked in the literature, even though polyamines are ubiquitous metabolites in living cells and G‐quadruplexes are transient regulatory elements, being both of them key regulators of biological processes. Herein, we compile the investigations connecting G‐quadruplexes and biogenic polyamines to understand the biological interplay between them. Moreover, we overview the main works focused on synthetic ligands containing polyamines designed to target G‐quadruplexes, aiming to unravel the structural motifs for designing potent and selective G4 ligands.

The interplay between polyamines and G‐quadruplexes has been largely overlooked in the literature, even though PAs are ubiquitous metabolites in living cells and G4s are transient regulatory elements. Herein, we compile the investigations connecting G4s and biogenic polyamines or synthetic G4 ligands containing polyamines. This work will assist to understand the biological impact of the association between G4s and natural/synthetic PAs.

1. Introduction

Polyamines (PAs) are organic molecules containing more than two amino groups, some of which are ubiquitous in living cells and contribute to the correct eukaryotic cell growth.[ ¹ , ² , ³ ] Naturally occurring polyamines are key players involved in the correct biological functions, although their mechanisms and some of their functions are still not fully understood. Similarly, G‐quadruplexes (G4s) are secondary nucleic acid structures adopted permanently or transitionally during biological processes such as genome maintenance, replication, or transcription but, again, the complete understanding of the mechanisms is still only in its early stages.[ ⁴ , ⁵ , ⁶ ]

It is well‐known that cations present in living organisms, typically K⁺ and Na⁺, are required to fold putative G4‐forming sequences (PQSs) into a G‐quadruplex structure. However, little attention has been paid to other biogenic and polycationic molecules present in living systems capable to induce the G4 formation. In particular, PAs are potential cellular G4 modulators that can strongly impact in the folding/unfolding dynamics of G4s and thus, affect the processes associated with G‐quadruplexes.[ ⁷ , ⁸ ]

Strikingly, several cancer cell genomes are enriched in PQSs, and the increase in cellular PAs concentration leads to cancer progression.[ ⁹ , ¹⁰ , ¹¹ , ¹² ] Therefore, the interplay between both PAs and G4s can assist us to understand the mechanisms of how cancer develops and progresses in order to develop new therapeutics. We firmly believe that the biological processes involving G4s must be re‐evaluated taking into consideration that PAs influence G4 dynamics and that the most commonly biophysical and biochemical approaches used to characterize the G4:ligand interactions must be updated considering the influence of PAs to mimic adequately cell and tissue conditions.

Herein, we compile the investigations connecting G4s and biogenic PAs for a better understanding of the effect of the latter on the G4 dynamics and the resulting biological processes. Moreover, we overview the polyamine‐based ligands described in the literature and their interaction with G4s and analyse the structural requirements to develop novel and highly potent G4 ligands by modifying the polyamine(s) sub‐structure.

2. Polyamines and G‐Quadruplexes

2.1. Polyamines

PAs are ubiquitous small molecules containing basic amine groups present in mammals, plants, protozoan parasites and many other microorganisms.[ ² , ¹³ , ¹⁴ , ¹⁵ ] Biogenic polyamines are structurally simple molecules which contain amine groups linked by an acyclic aliphatic chain. They can be protonated to generate ammonium groups depending on the pH, ionic strength, concentration and temperature. The most important biogenic PAs in eukaryotic cells are the triamines spermidine and spermine and the diamine putrescine which are mainly synthesised via the biosynthesis pathway (see Figure 1), although other transport mechanisms and biosynthesis pathways contribute to the regulation of their concentration in cells.[ ¹ , ² , ³ ]

Figure 1.

Polyamine metabolic pathway. ODC: Ornithine decarboxylase. APAO: N ¹‐acetylpolyamine oxidase. SSAT: Spermidine/spermine N ¹‐acetyltransferase. SMO: Spermine oxidase. PAO: Polyamine oxidase. AdoMetDC: S‐adenosylmethionine decarboxylase. 3‐AAP: 3‐acetamidopropanal. 3‐AP: 3‐aminopropanal. 4‐AB: 4‐aminobutanal. AdoMet: S‐Adenosyl methionine. dcAdoMet: S‐adenosylmethioninamine.

In the first step of the PAs biosynthesis (Figure 1), putrescine is formed from the ornithine amino acid by the ornithine decarboxylase (ODC). ^[2] Two specific aminopropyltransferases, spermidine synthetase and spermine synthetase, operate sequentially to form spermidine from putrescine, and spermine from spermidine respectively, using an aminopropyl moiety from a molecule of decarboxylated S‐adenosylmethionine. In addition, two enzymes catalyse the conversion of spermine to spermidine and spermidine to putrescine. Firstly, spermidine/spermine N ¹‐acetyltransferase (SSAT) catalyses the formation of N ¹‐acetylspermine or N ¹‐acetylspermidine whereas N ¹‐acetylpolymine oxidase (APAO) catalyses the cleavage of acetylated polyamines to yield spermidine or putrescine. An alternative conversion of spermine to spermidine by the spermine oxidase (SMO) completes the polyamine biosynthesis reactions in eukaryotic cells. Additionally, there are several PA transport systems that selectively transport polyamines in/out across the intracellular compartment.[ ³ , ¹⁶ ] These systems control both the PAs efflux and uptake but they are still poorly understood. They have been characterised in microorganisms and protozoa parasites but not in mammals. To close the PA regulation mechanisms, it has been reported that polyamines can be transported by endocytic pathways. Overall, PAs content within the cells is tightly ruled by the variations of the key enzymes involved in the polyamine biosynthesis pathway coupled with transport systems and endocytosis.

The roles of these PAs in cells are not fully understood although they are essential during cell development. The levels of intracellular PAs influence proliferation, differentiation, and apoptosis by the expression of specific genes related to the cell growth. ^[17] PAs control gene expression, regulate enzyme synthesis, activate DNA synthesis, promote DNA‐protein interactions and protect DNA molecules from damaging agents (i. e. ionizing radiation and reactive oxygen species).

Putrescine, spermidine and spermine are present at high concentrations in cells within the millimolar range, and they are fully protonated under physiological conditions. Nevertheless, they exist bound to cellular macromolecules such as RNA, DNA, proteins, ATP and phospholipids, yielding a low content of free or unbound PAs.[ ¹⁸ , ¹⁹ ]

The mechanisms of PAs gene regulation arise from modulating protein interactions via sequence‐dependent changes in DNA flexibility, facilitating the interconversion of alternative DNA structures, influencing the recognition of metabolites and through remodelling of the chromatin.[ ²⁰ , ²¹ , ²² ] For instance, biogenic PAs can promote the conversion of double stranded DNA (dsDNA, also referred as duplex) from the B‐form to the Z‐form, generating or supressing the DNA binding pockets for proteins with the concomitant inhibition/stimulation in the gene expression.[ ²³ , ²⁴ , ²⁵ ]

Biogenic PAs interact with DNA and RNA through two binding modes, either by (i) unspecific electrostatic interactions with the nucleic acid backbone, or by (ii) specific interactions within a binding site. Mainly, the interaction between nucleic acids and PAs is driven by electrostatic attractions between the positively charged ammonium groups and the negatively charged phosphate backbone, although hydrogen bonding with nucleobases and hydrophobic interactions involving the alkyl linkers of the PAs make a considerable contribution.[ ¹ , ²⁶ , ²⁷ ] Importantly, the interaction is not just relying on the chemical features of PAs, but it is also dependent on their adopted conformation under physiological conditions. Biogenic PAs possess a high conformation freedom and adopt specific conformations depending on intra‐/intermolecular interactions, dipolar effect and steric hindrance. Likely, the fully protonated PAs at physiological pH hinder the formation of intramolecular close contacts (N−H⋅⋅⋅N and C−H⋅⋅⋅N) and lead to repulsive interactions within the molecule (N−H⋅⋅⋅H−N or N−H⋅⋅⋅H−C). Therefore, all the fully protonated biogenic polyamines adopt a linear and extended conformation. The conformation adopted by the biogenic PAs relies then on the tight balance between the formation of intramolecular hydrogen bonds and the minimisation of electrostatic repulsion and steric interactions.

Cadaverine is another important biogenic PA derived from lysine and arises from a different biosynthetic pathway than putrescine, spermidine and spermine. ^[28] It causes the deathly stench in animals and plays important roles in plant growth and development, cell signalling, stress response, and insect defence. ^[29] Despite the important roles of cadaverine, its molecular mechanisms still remain elusive.

There are several pathologies associated with PAs, being cancer the most important, although heritable human diseases such as Snyder‐Robinson syndrome and Niemann‐Pick disease type C are also related.[ ³⁰ , ³¹ , ³² , ³³ ] Many types of cancer are characterised by elevated levels of these biogenic PAs in tissue or blood due to the increase of the polyamine metabolism in the initiation and progression of cancer. ^[34] PAs activate protein kinases such as tyrosine kinases and mitogen‐activated protein kinases, which in turn accelerate the expression of nuclear proto‐oncogenes.[ ³¹ , ³⁵ , ³⁶ ] The most relevant proto‐oncogene to this review is c‐Myc, which plays a central role in the regulation of the cell cycle progression.

The rest of biogenic amines are not polyamines but monoamines, such as histamine, serotonin, dopamine, phenethylamine and N‐methylphenethylamine, and participate in other biological functions, mainly in neurotransmission processes. ^[37] Furthermore, their intracellular content is very low which foresees not to strongly interact with G‐quadruplexes although this assumption must come with future investigations.

Another key function of PAs is their important role in regulating ion channels which controls the levels of Na⁺ and K⁺ by affecting their intracellular transport and concentration.[ ³⁸ , ³⁹ ] Therefore, PAs can indirectly influence the G4 dynamics through the content of cations required to form these structures.

2.2. G‐Quadruplexes

DNA is mostly recognised as the classical double helix structure identified by Crick and Watson in 1950s, although its highly dynamic nature enables it to adopt alternative non‐canonical secondary structures.[ ⁴⁰ , ⁴¹ ] Among the repertoire of non‐canonical structures, G4s have arisen most attention during the last decades because of the growing evidence of their involvement in key genome functions.

Initial bioinformatic analysis revealed over 370 000 putative G4‐forming sequences in the human genome which subsequently increased to 700 000 PQSs using next‐generation sequencing methods.[ ⁴² , ⁴³ ] These PQSs are located in telomeres, promoter regions, replication initiation sites and 5’ and 3’‐untranslated regions.[ ⁵ , ⁴⁴ ] These regions are associated with key processes and mechanisms in biology, including telomere maintenance, gene expression, replication, and genome stability.[ ⁴ , ⁵ , ⁶ ] Of utmost importance is the over‐represented PQSs in oncogenes and the overexpression of telomerase in cancer cells yielding unlimited proliferation.[ ⁴⁵ , ⁴⁶ ] Both processes, oncogene expression and telomerase immortalisation, can be inhibited using small molecules and thus, G4s have emerged as a novel therapeutic target in cancer therapies.[ ⁴⁷ , ⁴⁸ , ⁴⁹ , ⁵⁰ , ⁵¹ ]

G4s are folded in guanine‐rich sequences generating four‐stranded helical structures in which four guanines form co‐planar arrangements via Hoogsteen hydrogen‐bonds, termed as G‐quartet or G‐tetrad (Figure 2A). G4s comprise the stacking of at least two G‐quartets generating a central channel where cations, typically K⁺ and Na⁺, are coordinated to the guanines’ carbonyl groups (Figure 2B).

Figure 2.

Schematic illustration of (A) a G‐tetrad, (B) G‐tetrads stacked and (C–H) G4 structures of different molecularity and strand orientation. The PDB codes for each structure are indicated in the figure.

G4s can form intramolecular structures from a single strand or intermolecular structures by gathering two or four strands (Figure 2C–E).[ ⁷ , ⁸ ] The resulting structures can adopt a wide range of topologies depending on the relative orientation (5’ to 3’ direction) of the strands into three different topologies: parallel, antiparallel and hybrid (Figure 2C–H). In the parallel topology, all strands share the same orientation, in the antiparallel G4s two strands are oriented in one direction while the other two strands run in the opposite direction, and the hybrid topology shows three strands in the same orientation while the last strand in the opposite. Besides the G4 core, grooves and loops emerge in the structure exhibiting different lengths, dimensions and geometries depending on the sequence and topology of the G4.

The main driving factors for G4 folding and stabilisation are: (i) π‐π stacking interactions, (ii) hydrogen bonding, (iii) electrostatic cation‐dipole interactions, (iv) solvation and (v) electrostatic phosphate repulsions. There are excellent reviews discussing in detail these factors.[ ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ ] Among them, cation‐dipole interactions deserve special attention. It is assumed that cation coordination is required to template and stabilise G‐quartets and form the G4 structure, in contrast with the double helix in which cations mainly interact unspecifically with DNA to neutralise its negative phosphate backbone by electrostatic interactions. The G4 stabilisation has been mainly explained according to the ionic ratio and follows the trend: Sr²⁺ > K⁺ > Ca²⁺ > Na⁺ ~ NH₄ ⁺ ~ Rb⁺ > Mg²⁺ > Li⁺≥Cs⁺. For the biologically relevant cations, K⁺ and Na⁺, the coordination of potassium involves eight electrostatic cation‐dipole interactions M⁺‐O and is placed in the central channel between G‐quartets, while sodium comprises only four electrostatic interactions with the carbonyl groups within the same plane of the G‐quartet. As the bulkiness of the cation decreases, the possibility to establish electrostatic interactions with the lone pair of the carbonyl groups lessens and thus, the formation of the G4 structure is less favoured, being meaningless for Li⁺, which is unable to fold a G4 and is always taken as a control and inert electrolyte.

Interestingly, ammonium cation has a comparable stabilisation effect in G4s than Na⁺ and Rb⁺ although the ionic ratios, structures and solvation spheres are different.[ ⁵² , ⁵³ , ⁵⁴ , ⁵⁵ ] The structural analyses of the binding of ammonium to G4s have proved that the ammonium ions occupy a position between the G‐quartets in a similar fashion to K⁺, which is feasible according to its larger ionic ratio (1.02 Å for Na⁺, 1.37 Å for K⁺ and 1.47 Å for NH₄ ⁺). Nevertheless, ammonium cation derives from ammonia (NH₃), which is not an essential metabolite in animals although it can be formed during the urea cycle. ^[56] Among species, ammonium is one of the nitrogen resources in several plants. ^[57] Therefore, ammonium cations are unlikely present in living systems but other molecules containing the amine base/ammonium cation groups exist in high concentrations in cells, the biogenic PAs.

Only recently, it has been explored the interplay of PAs and G4s either with biogenic PAs or G4 ligands containing polyaminic moieties. We must highlight that the high concentration of polyamines in living systems has been systematically underexplored and can eventually be key to understand the G4 dynamics in cells and thus, unravelling the biological processes associated to G4s.

3. Biogenic Polyamines

3.1.Structural Analysis of Biogenic Polyamines and G‐Quadruplexes

Early attempts to obtain structural information on the binding mode between biogenic polyamines and nucleic acids using NMR and X‐ray were not very successful. The flexibility of the aliphatic chains and the ability to form non‐specific interactions of the polyamines prevented the structural characterisation of the DNA and RNA binding pockets for PAs. As a result, few high‐quality specific polyamine‐binding structures have been solved. These drawbacks are, at the same time, excellent PA features for generating crystalline structures as co‐solutes. The interaction between biogenic polyamines and anionic molecules such as DNA/RNA is predominantly electrostatic, but other factors such as the solvation degree, and the hydrophobic effect also contribute. The interaction of PA with nucleic acids is sensitive to the DNA/RNA secondary structure, which can contain specific binding sites for the biogenic PAs to afford well‐defined structural models. Early contributions have arisen from the Z‐form of double stranded DNA, differing from the B‐form duplex by having narrower grooves with the sugar‐phosphate skeletons of both strands holding closer together. Spermine and several analogues bind to Z‐form duplex DNA by forming interstrand electrostatic interactions between the ammonium groups of the polyamine and the phosphate groups of DNA.[ ²³ , ⁵⁸ ] G‐quadruplexes contain grooves of different width and size although they are shallower than the minor groove of duplexes. G4s can also generate additional binding sites in the loops and recently, left‐handed G4s have been recently reported, offering new models to get structural details of the binding between G4s and polyamines.[ ⁵⁹ , ⁶⁰ , ⁶¹ ]

Despite the challenge of working with polyamines, Keniry et al. brightly used the combination of ¹³C‐labelled biogenic polyamines and NMR spectroscopy to investigate their binding to G‐quadruplexes.[ ⁶² , ⁶³ , ⁶⁴ , ⁶⁵ , ⁶⁶ ] By using the labelled polyamine (see Figure 3 and 4), the adopted conformation of the polyamine can be evaluated through the enhancement of the intensity of each NMR peak. In the seminal works, Keniry investigated the interaction and dynamics of labelled spermine with G4s of different molecularities and duplexes.[ ⁶³ , ⁶⁴ ] They used the (i) intramolecular G4 of the thrombin aptamer (TBA, PQS _TBA : GGTTGGTGTGGTTGG) which only has two G‐tetrads, two lateral loops with T−T sequence and a central diagonal loop with T−G−T (see Figure 5), the (ii) intermolecular G4 formed by two strands of the telomeric DNA repeat of Oxytricha nova (PQS: G₄T₄G₄) which contains four G‐tetrads and generates two diagonal loops of T−T−T−T sequence and the (iii) intermolecular G4 formed by four TG₄T strands which holds four G‐tetrads but lacks of loops. All the binding models assumed that the charged ammonium groups of the spermine are allocated near the phosphate groups, yielding intermolecular contacts. The study shows the strong binding of spermine to G4s containing loops, in contrast with the weak interaction between spermine with the tetrameric G4 (PQS: TG₄T) and the duplexes. The analysis of the 1D and 2D NOESY spectra for the dimeric G4 (PQS: G₄T₄G₄) suggests that the structure of the G4 stem region remains unchanged, while thymine bases located within the loops experience a structural rearrangement. Taking into account intermolecular NOEs, the authors excluded the binding to the pocket created by the loop thymine bases and proposed the grooves as the probable location for a long‐lived spermine residence site. Although the free spermine adopts the lowest energy conformation with all bonds in trans, when bound to G4s, its conformation changes with at least one C−N bond in gauche conformation and the central diaminobutyl core becoming more static than the aminopropyl fragments.

Figure 3.

Predominant biogenic polyamines in eukaryotic and prokaryotic cells.

Figure 4.

(A) Labelling of [1’,1’’‐¹³C₂]spermine depicting the location of the α, β and γ carbons, the ¹³C label (indicated by asterisk) and the angles of rotation Ψ₁ and Ψ₂ of C−N bonds. (B) ¹H NMR spectrum of a 1 : 1 complex of [1’,1’’‐¹³C₂]spermine:G4 (PSQ: G₄T₄G₄). (C) Structure of spermine for different gauche and trans conformers. Reproduced with permission from Keniry et al. ^[64] Copyright 2003 Elsevier Science.

Figure 5.

A ‘stick’ model representation of the curved (A and C) and S‐shaped (B and D) spermines in the preferred residence sites on the electrostatic potential energy surface of the thrombin‐binding aptamer, featuring views of the narrow grooves (A and B in 1–6 groove and C and D in the 10–15 groove). Reproduced with permission from Keniry et al. ^[65] Copyright 2013 John Wiley & Sons, Ltd.

In the follow up study, Keniry et al. completed the dynamic characterisation of the interaction by means of ¹³C NMR relaxation experiments using intermolecular G4s (PQS: TG₄T and G₄T₄G₄). ^[65] The relaxation NMR data showed two main motional binding modes for spermine bound to these G4s. The lowest frequency mode of the relaxation (in the nanosecond range) was due to the overall tumbling of the G4, suggesting that a binding site was present with enough residence time to contribute effectively to the ¹³C relaxation. A relaxation time with a higher frequency (in the picosecond range) denoted the restricted internal motion of the spermine within the quadruplex. Tetramolecular G4 (PQS: TG₄T) showed fast internal motion of the methylene segments, which suggested a weak interaction. In contrast, the dimeric G4 (PQS: G₄T₄G₄) had a considerable population of binding sites in which the aminopropyl arms were effectively held rigid, indicating a strong interaction. Even if there was a tightly bound spermine to the dimeric G4 which contains only two potential binding sites, (i) the narrow groove and (ii) the loop cap, the precise location of the binding site for spermine has not been elucidated yet.

In line with previous works, spermine cannot adopt the lowest energy conformation when bound to DNA with all internal bonds at trans. The spermine bonds are reorganised to match the distance between the ammonium groups to the distance between the phosphates within DNA. The population of the trans conformers in free spermine at neutral pH was estimated to be greater than 75 %, but the data was inconsistent with the all‐trans conformation bonds of spermine when bound to G4. Spermine curves to match the shape of the narrow grooves of the TBA and is near the loops that cap the narrow grooves.

In a third work using the same spermine‐TBA system, NOE experiments confirmed the conformation changes in spermine and the immobilization of its central aminobutyl segment relative to the outer aminopropyl segments. ^[66] Spermine promoted an exchange of the solvent molecules attached to thymine imino protons either by (i) the disruption of the surrounding water network or (ii) the structural changes within the loops.

The interaction of spermine with the TBA G4 revealed that the PA adopts a more compact conformation by swapping C−C bonds to a gauche conformation, reducing the distance between the ammonium groups (Figure 4). It was particularly evident in the central aminobutyl segment, which had a larger N−N internuclear distance than the nitrogen atoms of outer aminopropyl segments. The trans to gauche change at the C6−C7 and C6’−C7’ bonds reduced the internuclear N5−N5’ distance by 0.11 nm and doubles effectively the NOE effect by bringing both methylenes closer to each other. TBA‐spermine complex favours a gauche conformation at the C6−C7 and C6’−C7’ bonds although there cannot be distinguished between a gauche+ (g+) or gauche‐ (g−) conformations. Spermine assumes the S‐shaped conformation (Figure 4) when the C6−C7 and C6’−C7’ bonds are g+g‐ or g‐g+ and a curved conformation when these bonds are g+g+ or g‐g‐.

Interestingly, the absence of strong NOE peaks from the aromatic G4 protons opposed a binding site deep within a groove. The NOE intensities of TBA‐spermine complex were predominantly determined by the distances within the spermine conformations with longer residence times at favoured sites on the TBA aptamer. TBA had the highest negative density regions at the two narrow grooves capped by TpT loops and the phosphate backbone lined the ridges that border these two narrow grooves with the protons pointing upwards and away from the groove bases (Figure 5), formed by the sequence G1‐G2‐T3‐T4‐G5‐G6 (G1–6) and G10‐G11‐T12‐T13‐G14‐G15 (G10–15). In the curved and S‐shaped spermine models, the ammonium groups were in close proximity to the aptamer phosphates and regions of high negative electrostatic potential (Figure 5). The shape of the curved spermine complemented the slight twist of both grooves. The curved spermine was inserted laterally into the G1–6, whereas it assumed a more horizontal position in the G10–15, to optimize the bonds between the ammonium and phosphate groups. The curved spermine generated less‐effective hydrogen bonds and electrostatic contacts with the backbone in the G10–15 groove than the analogous binding site in the G1–6 groove. The S‐shaped spermine cannot match the shape of the narrow grooves, hindering the insertion and disposing the spermine above the groove. It is likely that the S‐shaped conformation above the grooves is transitory before rearranging into the curved conformation.

Then, Keniry investigated the G4 structures within the promoter region of c‐Myc Pu27 (PQS _Pu27 : TGGGGAGGGTGGGGAGGGTGGGGAAGG) which contains five guanine tracks that can adopt several G4 conformations depending on the guanine tracks involved in the tetrads. ^[66] Two main structures were characterised, one of them contains the first four guanine tracks and another uses the last four guanine tracks, named Myc1234 or Myc2345 respectively. The first isomer was characterised by flanking bases stacking over an external G‐quartet, causing the chain to fold back and form a hydrophobic pocket, whereas in the second isomer, the third G‐track formed a long chain‐reversal loop, which was probably more mobile than the short chain‐reversal loops. At high molar ratios (spermine: DNA >20 : 1), the little effect on the NMR spectra and the nulled NOEs were consistent with a mobile spermine delocalised on the G4 surface without a specific binding site. The intramolecular spermine NOEs showed uniformly greater intensity in the spermine:Myc12345 complex compared to the other quadruplexes previously studied, indicating that spermine had the longest residence time at the binding site(s) on Myc12345. NOEs experiments indicated that spermine was bound to Myc12345 in a gauche conformation because, in the all‐trans conformation, the distances were too long to generate a substantial NOE. The population‐averaged immobility of spermine was enhanced in the presence of the fifth G‐tract without altering the G4 structure, which suggests a direct involvement of the fifth G‐tract in the immobilisation of spermine. The spermine laid over the region of highest negative charge density, along the groove formed by the 2‐nt chain reversal loop traversing the three G‐quartets into a region where the 5’‐terminal TGA sequence folded back under the external G‐quartet. The curved spermine followed the natural curvature of the quadruplex backbone, optimising contacts between the spermine amino groups and the quadruplex phosphates. It can be speculated that the 5’‐terminal TGGGGA sequence of Myc12345 folded back and contributed to form a negatively charged binding pocket, assisting the spermine immobilisation at the N10 terminus.

3.1. Thermodynamics on the G4‐Polyamine Interaction

In addition to the structural characterisation, thermodynamic and kinetical investigations were conducted to get the full picture of the interaction between polyamines and G‐quadruplexes. In an early study, Sugimoto and co‐workers explored the ability of the biogenic PAs and the crowding agent PEG to interact with the dimeric G4 DNA (PQS: G₄T₄G₄).[ ⁶⁷ , ⁶⁸ ] Using circular dichroism (CD) spectroscopy, the characteristic antiparallel CD bands were maintained upon addition of the putrescine, cadaverine, or spermine, in Na⁺/Ca²⁺ solutions, indicating that these PAs did not alter the antiparallel G4 topology under these conditions. The pre‐folded antiparallel G4 was investigated with putrescine in the presence and absence of the Na⁺ cation, indicating that it induced the folding of the random coil DNA sequence into an antiparallel G4 but could not alter the pre‐folded G4. The binding constants (K _a) between putrescine and G4 were 2‐fold higher in absence than in the presence of 100 mM NaCl (277 and 2.5 M⁻¹). This decrease in constants was due to the polyelectrolyte effect in which an Na⁺ excess hampered the electrostatic interactions between putrescine and the phosphate groups of DNA. Thermal experiments using CD melting spectroscopy of dimeric G4 with and without putrescine showed a destabilisation of the antiparallel G4 topology with putrescine and a decrease of the Gibbs energy (−ΔG°₂₅ from 28 to 22 kcal mol⁻¹), suggesting that the destabilisation effect was an entropy‐driven process (TΔΔS^o > ΔΔH^o). Lastly, molecular modelling for the interaction between putrescine with G4 allowed to determine the binding of the putrescine to the wider groove of the dimeric G4 without disrupting the G4 structure.

Tang‘s group studied the interaction of the biogenic PAs (putrescine, spermidine and spermine) with different G4s. ^[69] The CD bands of the telomeric G4 DNA 24TTA (hybrid structure, PQS _24TTA : TTAGGGTTAGGGTTAGGGTTAGGG) in K⁺‐rich conditions slightly increased upon spermine addition, suggesting that spermine induced G4 formation at low concentrations. In contrast, potassium could not promote the G4 formation in spermine‐containing solutions, indicating that the effect of promoting/inhibiting G4 formation induced by K⁺ is greatly dependent on the PAs concentration, which hampered any electrostatic contact between the cations and the phosphates. Thermal CD experiments showed that the plot of the T _m values versus the PAs concentration produced an asymmetric bell‐shaped curve, indicating that polyamines stabilised G4s at low concentrations whilst destabilised and even denatured them at high concentrations. The stabilisation effect reached a T _m maximum at 0.2 mM for spermine while for spermidine was higher (~1 mM), in contrast with putrescine which was unable to stabilise 24TTA G4 at any concentration. The denaturation of G4s was confirmed using CD, UV‐visible, FRET and NMR spectroscopies, which established that spermine and spermidine had a larger ability to denature G4 than putrescine and needed less amount of polyamine (5.5 mM for spermine/spermidine versus 6.6 mM for putrescine). Considering the PAs structures, the stabilisation and denaturation effects were likely associated to the chain length and the number of PA ammonium groups.

To explore the impact on the G4 topology and molecularity, spermine was studied with other PQS forming intra‐ (c‐Myc, PQS _c‐Myc : TGAGGGTGGGTAGGGTGGGTAA; c‐kit, PQS _c‐kit : AGGGAGGGCGCTGGGAGGAGGG; and Bcl2, PQS _Bcl‐2 : GGGCGCGGGAGGAATTGGGCGGG) and intermolecular G4s (H7, PQS _H7 : TTAGGGT and H12, PQS _H12 : TTAGGGTTAGG) by CD spectroscopy. ^[69] At low spermine concentration, it promoted G4 formation of intra‐ and intermolecular G4s (<0.5 mM) while denatured all G4s at higher concentration. The hardness to denature the highly stable c‐Myc G4 in comparison with the rest of G4s, indicated that the denaturing effect of PAs depended on the stability of G4s to some extent. Interestingly, to evaluate the electrostatic contribution of the interaction, ammonium‐containing buffer was used to mimic the positively charged groups of polyamines. Strikingly, the T _m value of 24TTA was unaffected under an equal concentration of ammonium in the solution than spermine, indicating that the electrostatic interactions were not the only driving force to affect the G4 structure.

Wen et al. used electrospray ionization‐quadrupole time of flight mass spectrometry (ESI−Q‐TOF‐MS) to determine stoichiometries and relative affinities of the biogenic PAs (putrescine, spermidine and spermine) in ammonium acetate buffer without K⁺. ^[70] Using the 24TTA G4, spermine formed adducts with 1 : 1, 1 : 2, 1 : 3, 1 : 4 and 1 : 5 stoichiometries (24TTA:polyamine), whilst spermidine formed adducts with 1 : 1, 1 : 2 and 1 : 3, and putrescine only 1 : 1 and 1 : 2 adducts. The mass peak intensities were used to determine the qualitatively affinities of each adduct following the trend: spermine > spermidine > putrescine.

In addition, the conformational G4 changes were studied with all PAs using CD spectroscopy and the morphological G4 changes by AFM. The CD spectra of the PQS upon the addition of biogenic polyamines (at low concentrations) in absence of K⁺, showed changes in the CD bands corresponding to a well‐defined structure, following the trend: spermine > spermidine > putrescine. In the presence of 100 mM K⁺, PAs weakened the intensity of the CD bands corresponding to the pre‐folded G4s, following the same order and indicating that the G4 topology was retained even at low PA concentrations.

AFM imaging showed a round particle‐like morphology of 24TTA G4 with and without PAs. The average dimensions increased upon PAs binding in K⁺‐containing buffer (from h x w=3.1x83 nm for 24TTA to 11.2x140.9 nm for spermine:24TTA, 7x112.3 nm for spermidine:24TTA and 4x98.4 nm for putrescine:24TTA). On the other hand, AFM experiments without K⁺ resulted in aggregates with PAs of similar or even smaller dimensions than G4 alone. Based on the images, the authors suggested that these supramolecular aggregates follow both (i) a blunt‐end stacked G4s on top of each other, and (ii) interlocked G4 systems formed by slipped G4 dimerization involving the free guanine bases at the strand ends. Finally, molecular modelling using 24TTA G4 showed the groove binding of PAs, with each polyamine adopting a different binding mode, which highlights the importance of PAs chain length and number of ammonium groups to fit into a pocket of specific dimensions.

Li and co‐workers investigated the aggregated structure, morphology and the aggregation mechanism of G4s with spermine. ^[71] Parallel G4s (c‐kit1, c‐kit2, Pu27 and RET1) showed a decrease of the UV band at 255 nm whereas a new band centred at 320 nm appeared upon increasing spermine concentration, which is associated to Tyndall scattering, indicative of the formation of nanometric aggregates. CD spectroscopy and UV melting experiments confirmed the formation of multiple adducts of c‐kit2 at low spermine concentration. Lastly, DLS experiments indicated the assembly of the c‐kit2 with spermine into a network‐like structures with an average height of 4.5 nm, and TEM images showed network‐like microaggregates larger than 1 μm size. In contrast to parallel topologies, hybrid and antiparallel G4s exhibited little effect upon interacting with spermine by UV titrations and melting experiments. DLS plots showed smaller nanoparticles (size distribution of 0.6–3.5 nm) in the presence of 0.5 mM spermine, and no large aggregates were detected with the hybrid G4. AFM confirmed that hTelo22:spermine complex formed very small nanoparticles (~2 nm) that did not assemble into microaggregates (PQS _hTelo22 :AGGGTTAGGGTTAGGGTTAGGG). Therefore, parallel G4s are able to condense into network‐like microaggregates in the presence of spermine, whereas the hybrid‐type and the antiparallel G4s cannot significantly form aggregates with spermine.

In an alternative approach, Burrows et al. used plants instead of mammalians to investigate the link between PAs and G4s. ^[72] Plants are unique systems with a cellular environment characterised by high polyamine concentrations and contain different biogenic polyamines compared to mammalians, such as norspermine (Figure 1). ^[73] PQS have been poorly studied in plants although recent bioinformatic analysis have revealed an enrichment of PQS at introns and gene promoter regions across several plant species.[ ⁷⁴ , ⁷⁵ ] In particular, Burrow's group used the green unicellular soil algae C. reinhardtii, which presents a highly GC‐rich genome prone to contain PQS (67 % of GC‐based genome). ^[76]

Initially, they identified PSQ in relevant DNA repair and photosystem genes by bioinformatic tools to afford 19 PQS, of which 18 were validated to fold into G4s using a combination of biophysical methods. Among them, two G4s located within the promoter regions of ERCC3 and psaK1 genes (PQS _ERCC3 : AAGGGGAGAGGGGAAAAGGGAGAAGGGGTT; PQS _psaK1 : ATGGGCCTGGGCGTGGGTCTAGGGGGAG), were selected due to their mixed topologies and relatively low melting temperatures to investigate their binding to putrescine and norspermine. These PAs have the highest cellular concentration in C. reinhardtii, 120 and 20 mM, respectively. CD spectroscopy showed that none of them altered the overall fold of the ERCC3 G4, in contrast with psak1 G4, in which norspermine shifted the structural equilibrium from mixed to parallel, whilst putrescine had no such effect. The thermal experiments showed that the ERCC3 G4 became more stable upon addition of any of these PAs until no further effect was observed, likely because the G4 binding sites were saturated. In a similar fashion, norspermine stabilised the psaK1 G4, while putrescine destabilised it, highlighting that putrescine and norspermidine impact differently in the stability of G‐quadruplexes. Lastly, both polyamines destabilised these G4s at high concentrations.

Apart from their biological roles in cells, G4s have been recently reported to exhibit peroxidase activity when complexed with hemin, forming complexes known as DNAzymes.[ ⁷⁷ , ⁷⁸ , ⁷⁹ ] Shangguan et al. studied the effects of biogenic polyamines on the catalytic efficiency of these DNAzymes. ^[80] They investigated 15 G4:hemin DNAzymes with two substrates, ABTS and TMB, to catalyse the peroxidase activity (see reference 80 for further details of DNAzymes). All biogenic PAs (spermine, spermidine and putrescine) were found to have positive catalytic effects on different G4:hemin DNAzymes, in which spermine exhibited the strongest catalytic enhancement efficiency and putrescine only had a very low effect. Among the G4 topologies, intramolecular parallel G4s had the highest catalytic activity, showing the largest enhancement of the activity with the aptamers PS2.M and AS1411 (PQS _PS2.M : GTGGGTAGGGCGGGTTGG; PQS _AS1411 : GGTGGTGGTGGTTGTGGTGGTGGTGG). The authors investigated the influence of the pH in the catalysis and spermine showed the highest catalytic activity for the PS2.M:hemin DNAzyme at physiological pH range (pH 6–7). In general, the enhancement of catalytic efficiency follows the trend spermine > spermidine > putrescine.

The inactivation of G4:hemin DNAzymes was mainly attributed to the degradation of hemin by H₂O₂ and according to UV‐visible experiments, spermine protected hemin from the degradation induced by the oxidant and thus, extended the active time of the DNAzyme. The authors suggested that the DNA condensation provoked by the PAs into microaggregates protected DNA from reactive oxygen species. These microaggregates provided a more hydrophobic binding site for hemin and a better microenvironment for the formation and stabilisation of ferryl heme intermediates, and thus, it results in the enhancement of the catalytic activity of the DNAzyme.

3.2. The Biological Interplay Between G4s and Biogenic Polyamines

As it is detailed in the introduction, the polyamine biosynthesis and catabolism pathways tightly regulate the concentration of biogenic PAs but they can be additionally controlled by the folding/unfolding of PQS into G‐quadruplexes. Ornithine decarboxylase plays a central role in the polyamine biosynthesis by catalysing the formation of putrescine from ornithine and it has several transactivator genes, including c‐Myc. This gene contains a PQS within its promoter and thus, the formation of the G4 structure can potentially regulate the polyamine synthesis pathway with the concomitant alteration of the cell proliferation and apoptosis.

Hall's group investigated the presence of PQS in the untranslated regions (UTR) of the polyamine synthesis pathway using algorithm tools and found 35 PQS with a high score to fold into bitetrad G4 structures (20 at 5’‐UTRs and 15 at 3’‐UTRs). ^[81] Among these PQS, 12 sequences from the 5’‐UTRs (ARG2, AZIN1, OAZ2, ODC1, SMS), and the 3’‐UTRs (ARG2, SMS, OAZ1, OAZ3) altered the reporter expression in dual Renilla and Firefly luciferase assays, confirming the control of PA synthesis by these bitetrad G4s at several points in the PA pathway. Interestingly, only ARG2, ODC1, OAZ1 and OAZ3 regulated the expression at the mRNA level and their PQS were then further explored to fold into G4s. The single migration band with a specific rate in PAGE (Polyacrilamide Gel Electrophoresis), the higher Thioflavin T fluorescence, the reversible melting transition and the characteristic CD bands using the ARG2, OAZ2, and ODC1 sequences, indicated their folding into intramolecular G4s. The biophysical assays on the PQS in SMS and ARG2, suggested either G4s containing a long loop or intermolecular G4s. Nevertheless, the most striking conformation corresponded to the AZIN1 structure, in which all the biophysical analysis suggested an intramolecular Watson‐Crick pairing structure.

Interestingly, low micromolar concentrations of the G4 stabilising molecule PDS strongly reduced spermidine and spermine levels in cells, decreased SMS expression and lowered endogenous SMS protein levels, which was consistent with the suppression of PAs in cells driven by the G4 formation in the SMS. Because the PA synthesis is self‐regulated via the formation of G4s at the mRNAs associated to the PA Proteins Synthesis, the effects of the biogenic polyamines (putrescine, spermine and spermidine) in cells were examined expressing selected G4 reporters. The Renilla luciferase assays showed decreased activity for AZIN1 and SMS reporters, whilst the activity increased for SAT1 and OAZ2. AZIN1 and SMS drove PA synthesis, whereas SAT1 and OAZ1 suppressed it and hence, all these results indicated that G4 motifs participated in a feedback loop. The depletion of polyamines using PA synthesis inhibitors (DFMO and APCHA) increased the activity of AZIN1 and SMS, and decreased the OAZ2 activity, confirming that PAs could self‐regulate through a sub‐set of canonical and long‐looped G4‐PQSs in AZIN1, SAT1, SMS and OAZ2.

Following this, AZIN1 (PQS _AZIN1 : GGACCCAGACAUAGGCUUGGUGG) was thoroughly investigated to assess its potential as a functional on/off switch depending on the conformation. Using high‐resolution NMR spectroscopy, AZIN1 sequence showed imino protons associated to Watson‐Crick base pairs (stem‐loop conformation) and Hoogsteen‐like pairing (G4 conformation). The intensity of the later peaks increased along K⁺ concentration, whereas disappeared upon Na⁺ or Mg²⁺ addition in concordance with the folding and unfolding of the tetrameric structure. NMR experiments using ¹⁵N‐labelled sequences supported this equilibrium and the alteration of the equilibrium dynamics of AZIN1 sub‐structures with spermine, which destabilised the hairpin to form the G4 structure (Figure 6).

Figure 6.

Derived model for the equilibrium between two possible hairpin conformers and the G4 of AZIN1; K⁺ is expected to favour the G4; Na⁺/Mg²⁺ are expected to favour the hairpin conformers. Adapted from reference 81.

An early work by Maiti et al. focused on the interplay between biogenic PAs and the c‐Myc22 proto‐oncogene (PQS _c‐Myc22 :GGGGAGGGTGGGGAGGGTGGGG). ^[82] CD spectroscopy and electrophoretic mobility shift assays (EMSA) revealed that he interaction of spermidine and spermine with 22‐mer c‐Myc shifted the mixture of parallel and antiparallel G4 conformations towards the parallel one. Following this, the authors confirmed that these polyamines could drive the transition from the duplex structure, formed by the c‐Myc PQS and its complementary strand, to the G4 structure. Then, they explored the impact in the recognition of the c‐Myc G4 by a well‐known porphyrin G4 binder (TMPyp4) by evaluating the binding affinity in presence and absence of polyamines using UV‐vis titrations. The binding affinity values were one order of magnitude lower for the G4–porphyrin interaction, suggesting the critical role of polyamines in modulating the molecular recognition and the interaction with other molecules.

c‐Myc transcript levels were assessed after treating cells with increasing concentrations of spermidine and spermine and a concentration dependent increase in c‐Myc transcript levels was observed. Using a reporter gene assay, they observed an increase of the luciferase activity upon PA treatment, suggesting that they stimulated transcription although G4s are intimately associated as repressor elements. In this regard, polyamines could induce a structural transition of c‐Myc G4 into a transcriptionally active motif with distinctive molecular recognition properties for the transcription factor binding, which drove the c‐Myc expression.

A recent study by Higashi's lab described the regulation of the chondroitin synthase 1 (CHS1), ^[83] which partly controls the structural changes of glycosaminoglycans such as chondroitin sulfate (CS) by biogenic polyamines through the G4 RNA located at the 5’‐UTR. CHS1 catalyses the chain elongation of the disaccharides forming several glycosaminoglycans, which are subsequently sulfated at different positions by chondroitin sulfotransferases (C4ST and C6ST). Initial studies on the expression levels and sulfation patterns of CS in 15 cell types confirmed that intracellular polyamine levels modulate CS structure and that the CHSY1 and C4ST2 expression decreased in polyamine depleted cells. They found that CHSY1 synthesis was affected at the translational level whilst C4ST2 acted at the transcriptional level.

Then, they further investigated the CHSY1 stimulation caused by PAs using a plasmid assay. It confirmed that the high CHSY1 levels resulted from an increase of polyamines at the translational initiation step, which were mediated by a PQS located from −202 to −117 at the 5’‐UTR. This PQS folded into a bitetrad G4 RNA as confirmed by thermal UV melting and NMR and CD spectroscopies. Treatment with TMPyp4 reduced the expression of CHSY1 protein in cells, indicating that the G4 ligand stabilised this G4 RNA and inhibited the CHSY1 synthesis. Taken together, these results suggested that CHSY1 synthesis was negatively regulated by the G4 structure located at the position −145 to −135, and PAs can indirectly affect this G4 structure. The destabilisation of their structures by PA resulted in the stimulation of translation initiation of CHSY1 synthesis and affected, at least in part, the level of CS 4‐O‐sulfation and chain polymerization.

4. G‐Quadruplex Ligands Containing Polyamines

In the last decade, G4s have emerged as novel therapeutic targets because of their roles in telomere maintenance and inhibition of oncogene expression, which can cause inhibition of the tumour growth and induce tumour regression in vivo. In this line, a large number of ligands have been described to strongly bind G4s in vitro and exhibit high selectivity for G4s over the major conformation of duplex DNA.[ ⁸⁴ , ⁸⁵ , ⁸⁶ , ⁸⁷ ]

Some of the validated G4 ligands have reached antitumoral clinical trials such as fluoroquinolone derivatives CX‐3543 and CX‐5461 and other ligands are currently under the clinical scope. CX‐3543 inhibits RNA polymerase through the interaction with ribosomal G4s, while CX‐5461 interacts with oncogenic (c‐Myc and c‐Kit) and telomeric G4s, resulting in the blockage of the replication forks, induction of DNA damage and the inhibition of rRNA biogenesis.[ ⁸⁸ , ⁸⁹ ] Taking into account the structural features of G4s, several hallmarks of potent and selective G4 ligands can be ruled. The large planar surface of the G‐tetrads provides the main binding contact for ligands through π‐stacking interactions and thus, large π ‐deficient cyclic moieties tightly bind to the G‐tetrads. A complementary benefit of this scaffold arises from the incapacity of large cycles to intercalate between base‐pairs in dsDNA due to the steric restrictions of the phosphate backbone. A second G4 structural feature arises from the distinctive loops and grooves, which provides binding sites for pendant ligand motifs. Therefore, a variety of side chains or pendant arms have been incorporated into G4 ligands such as pyrrolidine, piperidine, morpholine, 1‐ethylpiperazine, N,N‐diethylethylenediamine, and guanidine.

G4 ligands bearing polyamine chains lead to enhancements in the binding affinities to G4s by electrostatic interactions and/or formation of hydrogen bonds with the phosphate groups and bases in addition to improve their aqueous solubility. Therefore, it has been often designed G4 ligands with multiple amine groups to make multiple contacts, enhancing the interaction with G4s. To mention that polyamine‐based G4 ligands can exploit the active Polyamines Transport System (PTS) to enter in the cells and enhance the pharmacological profile. Finally, the incorporation of PAs within the G4 ligand scaffolding can protect against double‐strand breaks and irradiation and regulate the cellular growth and differentiation using the biogenic PAs pathways.

In this regard, we have taken together the most representative G4 ligands containing polyamine‐based chains, either similar to biogenic ones or synthetically prepared. It is noteworthy that G4 ligands including monoamine groups are out of the scope of this review and we will only focus on the G4 ligands containing polyamine moieties.

4.1. Acyclic Polyamine G4 Binders

4.1.1. Acyclic Monobranched G4 Binders

Fei, Liu and co‐workers pioneered the investigations on the interaction of synthetic polyamines with G4s using the triethylene tetramine (trien, Figure 7). ^[90] CD spectroscopy studies showed the preference of trien for the antiparallel G4 topology using intra‐ and intermolecular G4s (telomeric sequences (T₂AG₃)₄ and TG₅T). Thermal UV melting experiments proved the capacity of trien to stabilise pre‐folded telomeric G4 DNA and induce the G4 formation of the unfolded putative G4‐forming sequence. Growth inhibition in HeLa cells after trien treatment was determined, having an IC₅₀ value of 74.8 μM, which was lower than in normal cells (HLF), suggesting that telomerase sensitive cells were more sensitive than normal cells to trien. Then, the telomerase effect of trien was evaluated by using a TRAP (telomerase repeat amplification protocol) assay in HeLa cells, affording a high inhibition of telomerase activity with a of IC₅₀ value of 7.8 μM. Lastly, they concluded that trien inhibited telomerase by using a Tap polymerase assay. ^[91]

Figure 7.

Structures of trien, dien, L1 and L2.

The same authors investigated the effect of trien on c‐Myc G4. ^[92] This ligand induced the folding of the PQS _c‐Myc into a parallel G4 topology at low K⁺ concentration but had little effect in K⁺‐rich solutions. Then, a PCR‐stop assay using the PQS _c‐Myc and its complementary strand confirmed the G4‐dependent polymerase inhibition caused by trien (IC₅₀=28 μM). To conclude the work, Western blot and RT‐qPCR analysis demonstrated a decrease in c‐Myc expression at both the RNA and protein levels in HeLa cells treated with trien. ^[93]

Rodriguez, Balasubramanian et al. reported the interaction of three acyclic polyamine‐based ligands with telomeric G4s (see Figure 7). ^[94] L1 caused the folding of the telomeric (T₃AG₃)₄ sequence into a parallel G4 topology, in contrast to the ligand lacking of the anthracyl unit (diethylene triamine or dien) and the amide analogue (L2). It highlighted the importance of the anthracene moiety and the amine group attached to the aromatic core, either by reducing the net charge or the geometrical freedom of the molecule. The presence of potassium cations prevented the interaction and folding‐induction of L1, probably because of the saturation of the binding sites by the cations. However, in Na⁺‐containing solution, L1 addition switched the antiparallel G4 conformation to a parallel one, according to the changes in the CD bands. Additionally, ¹H NMR spectroscopy confirmed the interaction between L1 and 24TTA G4, causing a downfield shift of the ligand NMR peaks. They proposed a unique binding mode for these G4 ligands, in which the polyamine chain replaced the potassium ions within the G4 channel. The precise binding mode of acyclic polyamine‐based molecules to G4s remains unclarified and only can be addressed using specifically designed structural studies.

Carla Cruz's team published several works of acyclic polyamine ligands containing different aromatic units, either a naphthalene, indole, quinoline or acridine moiety (Figure 8).[ ⁹⁵ , ⁹⁶ , ⁹⁷ ] In a first study, they assessed the interaction of a series of naphthalene derivatives bearing polyamines of different lengths and number of amine groups (L3–L5) and a quinoline derivative (L6) with PQS formed in the immunoglobulin switch regions Sμ3 (named 58Sγ3) and Sμ (3Sμ), c‐Myc22 and hTelo22 (also termed as 22AG). Both FRET and CD melting experiments showed the high stabilisation effect induced by the longer L5 ligand on the c‐Myc22 and 58Sγ3 G4s while L4 yielded the highest stabilisation effect for hTelo22 G4, confirming that the polyamine moiety played a key role in the G4 stabilisation. Then, CD spectroscopy of the 58Sγ3 G4 confirmed the binding of these ligands without further disruption of the G4 conformation. Interestingly, L3 induced the parallel G4 conformation of hTelo22 G4 according to CD experiments. The stabilisation effect of the ligands towards 58Sγ3 G4 was assessed in a more biologically relevant condition by using a polymerase stop assay in which 16 equivalents of the ligands halted the prime extension due to the G4‐formation. Interestingly, the low displacement (<15 %) of thiazole orange (TO) in Fluorescence Indicator Displacement (FID) assays pointed out that the ligands bound to the G4 grooves since the ligand could not impair the G4‐TO interaction.

Figure 8.

Structures of L3–L9.

In the subsequent study, Carvalho et al. investigated a series of indole derivatives (L7–L9, Figure 8). ^[96] CD and FRET melting assays with hTelo22 and c‐Myc22 G4s showed the stabilisation effect following the trend L9 > L8 > L7, in line with the increase of the length and number of amines of the polyamine moiety. The required presence of the aromatic moiety in the ligand structure for the G4 stabilisation was confirmed because of the low ΔT _m of trien (~2 °C), in contrast to the higher ΔT _m of L9 (8.4 °C, for hTelo22). The lack of any stabilisation upon addition of the ligands to duplex DNA suggested a certain degree of selectivity towards G4 over dsDNA. In the same way that previous acyclic ligands, CD spectroscopy indicated the interaction without disrupting the G4. The quenching of the fluorescence emission of the indole moiety once titrated with c‐Myc22 and hTelo22 was used to afford Stern‐Volmer binding constants (K_SV), with values of one order of magnitude higher for G4 than for duplex (1.95×10⁵ M⁻¹ for c‐Myc22, 1.0×10⁵ M⁻¹ for hTelo22 and 9.9×10³ M⁻¹ for duplex).

A tentative binding mode of L9 and c‐Myc G4 was proposed using fluorescence and NMR spectroscopies. The Job's plot analysis using the fluorescence of L9 indicated a 1 : 2 stoichiometry (DNA:L) for both c‐Myc and hTelo G4s. The largest NMR signal shifts (Δδ), assigned to the guanines of the terminal G‐tetrads for the c‐Myc G4:L9 system, suggested the stacking of the indole unit on the G‐tetrads and the binding of two L9 molecules per G4. Molecular docking studies afforded several binding models in which the indole moiety interacted with the G‐tetrad, while the polyamine moiety was accommodated in the loops and grooves.

2 A comprehensive work on acyclic monobranched G4 ligands were reported by Xue's group. ^[98] The authors conjugated a TO moiety to different amine‐based chains categorised into five classes (see Figure 9): (i) alkylamines (L10–L12), (ii) polyamines (L13–L15), (iii) oligo(ethylene glycol) amines (L16–L17), (iv) pyrrolidine‐containing chain (L18–L19), and (v) piperidine‐containing chain (L20–L21). An initial screening using FRET melting assays showed a strong and linear stabilisation effect as a function of the ligand concentration for L13, L14, L15 and L18 for the hybrid G4 hTelo22. The stabilisation degree in ΔT _m values followed the trend L13 < L14 < L15 in agreement with the increase of amine groups and length between them, whilst the pyrrolidine side arm of L18 was well‐known moiety to bind G4s. Among all the TO‐conjugates, L15 and L18 exhibited the strongest G4 stabilisation effect which was then confirmed by CD melting experiments using the same PQS _hTelo22 in both K⁺ and Na⁺ conditions forming hybrid and antiparallel conformations, respectively.

Figure 9.

Structures of L10–L21.

Using the fluorescent enhancement of the TO moiety upon DNA interaction, fluorimetric titrations were performed to determine the binding affinities of L13, L15, L18, L19, L20 and L21 with hTelo22. The dissociation constants values (K _d) were found in the low micromolar range for L19, L20 and L21 (1.65, 1.13 and 2.18 μM, respectively), which showed a lower affinity for hTelo22 than the TO control (0.87 μM), whereas L13, L15 and L18 (0.14, 0.07 and 0.3 μM, respectively) exhibited a higher affinity for the antiparallel telomeric G4. In addition, L13, L15 and L18 titration curves fitted better to a two‐site model, suggesting the stoichiometry 1 : 2 (DNA:L), in contrast with the rest of ligands, that fitted perfectly to a 1 : 1 stoichiometry model. Using a duplex DNA, the dissociation constants (K _d) were calculated to be within the sub‐micromolar range for all the ligands and the TO control. A selectivity index was calculated from the dissociation constants (K _d(G4)/K _d(duplex)) to highlight that the addition of amine groups to the side chain increased the binding. It also reduced the selectivity for G4 vs. duplex DNA because of the formation of a higher number of non‐specific contacts between the positively charged ammonium groups and the negatively charged phosphates of the DNA backbone.

Docking experiments of representative ligands (L10, L15 and L18) with hybrid‐type G4 indicated a general binding mode in which the TO moiety stacks with the G‐quartet, while the side chains lie into one of the grooves, forming electrostatic contacts. Finally, a TRAP assay was used to investigate the inhibition of human telomerase activity, showing that L15 had the strongest inhibition effect in both telomerase and polymerase activities.

4.1.2. Acyclic Dibranched Ligands

It is noteworthy that most of the G4 ligands reported so far contain two side chains (dibranched) probably because of the greater accessibility to synthesise symmetric ligands and the increase of their aqueous solubility. In this regard, Ulven's team prepared a series of dibranched ligands based on PhenDC, a strong G4 binder, as scaffold and studied the interaction with G4s. ^[99]

These novel ligands maintained the PhenDC core in whichwhere the phenanthroline scaffold was further substituted at the 4 and 7 positions with amino side chains categorised by (i) aminoalkyl chains (L22–L26, see Figure 10):), (ii) polyamines (L27–L32) and (iii) guaninidium chains (L33–L35). Firstly, FRET melting studies were carried out using the G4s from the telomeric G‐overhang (hTelo21; PQS _hTelo21 : GGGTTAGGGTTAGGGTTAGGG), c‐kit2, c‐Myc22 and a hairpin duplex DNA (see reference 99 for sequences). Polyamine and guanindinium containing ligands showed high stabilisation ΔT _m values, being the quaternary diamine derivatives, L31 and L32, the strongest G4 stabilisers. In addition, all the ligands did not show any stabilisation effect of the duplex hairpin DNA, indicating that they displayed good selectivity of G4 over duplex. It is worth to mention that aminoalkyl derivatives (L22–L26) exhibited a lower stabilisation effect than PhenDC, which could be attributed to the increase of the electronic density of the phenanthroline moiety causing a less efficient π‐π stacking on the G‐quartet. The ΔT _m values were close along the diamine (L27–L29) and guaninidium series (L33–L35), suggesting that the increase in chain length had little effect on the interaction. Interestingly, quaternary diamines derivatives, L31 and L32, were the most potent G4 ligands, even though their end groups were incapable of forming hydrogen‐bond interactions with the G4 loops. The introduction of an extra‐amino group did not increase the G4 stabilisation effect neither (L30), probably due to the increase in the entropic penalty of binding the highly flexible side chains.

Figure 10.

Structures of L22–L35.

Fluorescent displacement assays and ESI‐MS experiments confirmed L31 and L32 as strong and selective G4 ligands with affinity constants of two orders of magnitude higher for G4s than duplexes and a stoichiometry of 1 : 2 (DNA:L). Structural studies using CD spectroscopy pointed out that these ligands could induce the G4 folding of the random coil sequence into an antiparallel G4 topology and could shift the equilibrium from hybrid to antiparallel G4 topologies.

A series of pyrelene diimides were used by Savino and collaborators as G4 ligands. ^[100] In the initial work, they conjugated the pyrelene moiety to two polyamine side chains (Figure 11) to explore the influence of the number and length of alkyl chains between amine groups on the ability to bind G‐quadruplex and exert biological activity. EMSA experiments using the hybrid‐type G4 24TTA showed the inability of L36 and L37 to induce the intramolecular G4 folding, probably because of their low aqueous solubility, which promoted their self‐aggregation. The rest of the ligands could induce G4 folding following the order L38 ≈ L40 > L39 ≈ L41 > L43 > L42 according to the apparition of a new mobility band assigned to the folded G4 in the gel. The ligands showed a lesser capacity to induce the G4 folding of intermolecular G4s. In the same work, a TRAP assay indicated the same trend observed by EMSA and confirmed L38 as the most promising G‐quadruplex ligand. Overall, this study suggested that the ligands with the stronger G4 interaction were the ethylenic triamine derivatives followed by the dipropylenic diamine derivatives as well as further methylation could not improve the ligand interaction with G4s.

Figure 11.

Structures of L36–L43.

The same authors evaluated the interaction of this series of ligands with G4 and duplex DNAs by FRET melting experiments.[ ¹⁰¹ , ¹⁰² ] ΔT _m values were higher for ligands containing side chains with three amine groups in comparison to side chains containing only two amines. The stabilisation of duplex DNA was slightly affected by the ligands and unaffected by the acetylation at the side arms. Interestingly, L43 exhibited the greatest selectivity and stabilisation effect for G4, highlighting L43 and L38 as the most potent G4 ligands.

CD spectroscopy studies of these ligands suggested the G4 interaction through the monomer species of the ligands and the binding between the pyrelene moiety and the G‐quartet. Finally, molecular modelling studies located two molecules of L38 by G4 structure in which the pyrelene cores were stacked on each terminal G‐tetrad of the G4. The side arms interacted with the DNA grooves and phosphate groups whereas the hydrophobic ends of the side arms appeared to govern the interaction with G4 DNA grooves. The selectivity observed for L43 could be ascribed to the fewer contact interactions with the solvent molecules of the edges.

In a following study, they investigated the hTERT promoter for putative G4‐forming sequences and after confirming six PQS within this region (see reference 103 for the sequences), the interaction of L38, L39 and L43 towards these G4s was assessed. ^[103] Using the DNA polymerase stop assay, all these ligands halted the polymerase by folding a G4 within the promoter of hTERT, specially L38. CD experiments showed that the ligands slightly favoured the antiparallel G4 conformation, and the apparition of an induced circular dichroism (ICD) band indicated the formation of monomer species of the ligands that interact with the G4s by π‐π stacking at the terminal G‐quartet.

Milelli, Minarini et al. developed a series of asymmetric dibranched ligands containing a naphthalene diimide (NDI) core, a methoxybenzyl chain and different polyamine chains (see Figure 12). ^[104] A screening on the thermal stabilisation effect of the ligands towards hTelo21 G4 and duplex was conducted. Ligand L46 was a stronger G4 binder than L47, indicating that the 2,3,4‐methoxybenzyl moiety had an unfavourable effect on G4 stabilisation (ΔT _m=21.4 °C for L46 and 13.1 °C for L47 respectively) as well as L47 being a better duplex binder, being less selective than L46. This effect was opposite to L44/L45, in which the trimethoxy derivative was a better G4 ligand, suggesting a different binding mode towards G4 of these spermidine derivatives. The shortening of the linker length from four (L44) to three (L52) carbon atoms decreased the G4‐stabilisation effect while the further shortening to a 3–2‐3 polyamine‐based chain (L51) conserved the thermal effect on G4s. Among the linkers with only two amines (L48–L50), L49 showed the highest stabilisation effect which was larger than the exhibited by L46, indicating that the number of amine groups and therefore, the net charge was not the main factor that governs the interaction with G4s. Methylation of the terminal amines enhanced the ΔT _m values in comparison with the unmethylated analogues but causing an increase of duplex stabilisation and thus, decreasing the selectivity. Finally, the substitution of the amine for ether groups induced a lower stabilisation for G4 which highlighted the key roles of nitrogen atoms to establish contacts. From the overall FRET melting analysis, L46 and L49 emerged as the most potent and selective G4 ligands. Both L46 and L49 could inhibit the enzymatic processing of topoisomerase and polymerase but could not inhibit telomerase in a selective fashion. Molecular modelling indicated the binding of L49 towards hybrid‐type G4s (both type 1 and type 2) via the grooves establishing many electrostatic contacts.

Figure 12.

Structures of L44–L57.

The same team prepared a series of asymmetric dibranched ligands to target simultaneously enzymes and DNA, containing an hydroxamic acid and a naphthalene diamide core to interact with histones and G4 DNA, respectively. ^[105] The molecular design included different polyamine side chains (see Figure 13) as structural modification to enhance pharmacological profile and increase the G4 DNA interaction. They assessed the thermal stability induced by the ligands of G4 (hTelo22) and duplex models by means of FRET melting assays. All the ligands, excluding L62, showed a lower G4 stabilisation effect than the reference L46 in addition to interacting much more efficiently with the duplex DNA. The higher stabilisation towards dsDNA in comparison with the trimethoxybenzyl derivative indicated the less selectivity for G4s of the new series of ligands and that substituting the methoxybenzyl moiety for a hydroxymethyl attenuated the binding towards G4s. The ΔT _m values decreased following the order L62 > L60 > L61 > L59 > L58, suggesting a less efficient interaction when decreasing the length of the linkers and number of amine groups. Of utmost importance, L62 could inhibit histone deacetylases and highlighted that it is a promising ligand targeting both DNA and HDACs.

Figure 13.

Structures of L58–L62.

4.1.3. Acyclic Tribranched Ligands

A new series of G4 ligands based on the triphenylamine (TPA) core was developed by Pont and colleagues (see Figure 14), from which one, two, or three side polyamine chains emerged.[ ¹⁰⁶ , ¹⁰⁷ , ¹⁰⁸ ] The side chains were macrocyclic (L63–L65), N,N‐dimethylethylendiamine (L66–L68) or N,N‐dimethylpropylentriamine (L69–L71) moieties. To assess the interaction with G4s, FRET melting experiments were conducted using a panel of G4s of different topologies and a duplex DNA. Along the series, the variation in the melting temperature was larger as the number of side arms and thus, amine groups, increased, following the trend L65 > L64 > L63, L68 > L67 > L66 or L71 > L70 > L69. This effect denoted the impact of the number of amine groups interacting with DNA and a general rule of “the more, the better”. The best G4 ligands were the tribranched ligands L65 and L71, which showed a high selectivity according to the low stabilisation effect on the duplex DNA model and the meaningless decrease in the melting values by FRET melting competition experiments. The fluorescence emission of the triphenylamine moiety was used to determine affinity constant values (K _a), which ranged between 10⁵ and 10⁶ M⁻¹ for G4s models with L65 and L71. Strikingly, affinity values of 10⁴ M⁻¹ were obtained for L65 with dsDNA in contrast to the lower values obtained for L71 (K _a=10² M⁻¹), suggesting a very high selectivity of the ligand bearing propylenic chains in comparison with the ligand bearing three macrocyclic moieties. CD spectroscopy confirmed a stronger interaction of the ligands containing three side chains followed by ligands containing two while the less effective ligands were those with only one side chain. Molecular modelling of L65 and L71 with G4/dsDNA models gave light into the binding mode. Both ligands interacted by the stacking of the TPA moiety on the G‐quartet of the G4 while the side chains interacted with the grooves and phosphates by electrostatic and hydrogen bonding interactions. Nevertheless, the duplex minimum energy conformers showed different binding patterns, with L65 located into the minor groove with one of the side macrocyclic chains interacting simultaneously with both DNA strands and bending the structure. On the other hand, L71 was placed externally in the minor groove with only a few phosphate groups contacts and thus, exhibiting a less effective binding.

Figure 14.

Structures of L63–L71.

A detailed analysis of the FRET melting results and the net charge of the ligands at the pH of the DNA binding experiments showed very interesting findings (Figure 15). A good correlation was obtained from the plot of the ΔT _m values versus net charge at pH 7.4 (Figure 15), which highlighted that the net charge of these polyamine ligands governed the overall interaction with G4s.

Figure 15.

Plot of the ΔT _m (°C) values for hTelo (0.2 μM) and ligands L63–L71 (ligand‐to‐G4 ratio of 5) versus the positive net charge at pH=7.4. Adapted from reference [108].

Once the trisubstituted ligands were identified as the most potent G4 ligands, L72 and L73 were developed containing three either linear or macrocyclic pendant substituents but replacing the triphenylamine central core for a 1,3,5‐triphenylbenzene moiety (see Figure 16). Both ligands possessed a larger aromatic surface able to interact with a G‐tetrad while lacking of the central amine atom, which can facilitate a more efficient planar arrangement. Both ligands showed a strong G4 stabilisation effect for the hTelo21 G4 (ΔT _m=24.6 and 30 °C for L72 and L73, respectively), confirming a strong interaction with G4s. Moreover, they exhibited a very low stabilisation effect on the duplex DNA, suggesting a high selectivity for G4 over duplex structures. Interestingly, L72 and L73 had high cytotoxicities in cancer cells (MCF‐7, HeLa and LN229) in comparison with the triphenylamine analogues L65 and L71. It highlighted that these ligands could efficiently stabilize G4s because of the planar arrangement of the central core moiety while improving their cytotoxic effect in cancer cells since their pharmacological properties were different. ^[108]

Figure 16.

Structures of L72–L73.

Tribranched G4 ligands were developed by Shchekotikhin and collaborators using an anthraquinone core as scaffold. ^[109] They prepared a series of ligands composed by an anthra[2,3‐b]thiophene‐5,10‐dione or anthra[2,3‐b]furan‐5,10‐dione scaffolds bearing three side chains with polyamine terminal guanidine groups (Figure 17). Molecular modelling of the dibranched ligand and a hybrid‐type 24TTA G4 showed the stacking of the aromatic core on the G‐quartet and multiple van der Waals and hydrophobic interactions of the side chains. The analysis of the minimal conformers, binding affinities and thermal stabilisation indicated that the attachment of a third chain at the 2‐position increased the binding towards G4s. Despite the outstanding binding to telomeric G4, they showed poor activity in vitro and in vivo due to low cell permeability or bioaccessibility.

Figure 17.

Structures of L74–L89.

4.1.4. Acyclic Tetrabranched Ligands

Some tetrabranched polyamine ligands containing a porphyrin core were reported as G4 ligands.[ ¹¹⁰ , ¹¹¹ ] Commonly, the large aromatic surface of the porphyrin moiety drives the interaction with G4 via end‐stacking although additional binding modes by intercalation between tetrads, within the grooves and with the loops, have been described. Yatsunyk et al. reported a meso‐tetrakis‐(4‐carboxysperminephenyl) porphyrin (L90) and its Zn(II)‐derivative (L91) (Figure 18) bearing four spermine side‐chains and their interaction with hTelo22. Both ligands tightly bound to G4 and the binding isotherms showed 2–3 molecules of L90 per hTelo22 G4 whilst a higher stoichiometry of 9 molecules per G4 was achieved for L91, indicating multiple binding modes besides end‐stacking. Biophysical experiments indicated strong electronic communication between porphyrin moieties, suggesting that both porphyrins were in close proximity to the G‐quartets. They excluded the formation of large and non‐stoichiometric ligand‐G4 aggregates because ligands could not aggregate in the absence of DNA, confirming the formation of discrete porphyrin ligand – hTelo22 G4 adducts. The spermine arms acted with the four tentacles reaching into grooves and stabilizing the G4. The mild selectivity for G4 over dsDNA was likely due to strong electrostatic interactions between the polycationic ligand and negatively charged DNA backbone but at the same time, the presence of the four spermine arms in L90 circumvented the insertion of the porphyrin core within the DNA bases because of their steric hindrance.

Figure 18.

Structures of L90–L91.

4.2. Cyclic Polyamine G4 Binders

Attaching bearing side chains to a planar large aromatic core has been the main strategy to develop G4 ligands and continues affording new ones. Nevertheless, different approaches can be followed to afford potent G4 ligands such as the cyclisation of two aromatic G‐tetrad stacking units described by the seminal work of Teulade‐Fichou et al.[ ¹¹² , ¹¹³ ] The supramolecular concepts for nucleotide recognition prompted them to design cyclic ligands containing two intercalative moieties bridged together by polyamine chains. The intercalative moieties were dibenzo[b,j]‐1.10‐phenanthroline units, which hold a large and crescent‐shaped structure able to stack with the G‐quartets. They brightly envisaged that the cyclic conformation adopted by the ligand would hamper intercalation in the duplex DNA structure and thus, potentiate a higher selectivity for G4s (Figure 19).

Figure 19.

Structures of L92–L96.

In this pioneering work, FRET melting experiments using hTelo21 G4 showed the high thermal stabilisation effect of L92 in comparison with the monomer ligands used as controls (L93–L96) to evaluate the cyclisation (ΔT _m =28 °C for L90 and between 0–12 for L93–L96). ^[114] The binding affinity constants and the stoichiometries were determined using both Surface Plasmon Resonance and fluorescence experiments showing close outcomes. The affinity for this hTelo21 G4 structure is 1.2x10⁷ M⁻¹ for L92 with two molecules binding per G4 in contrast to the monomeric ligands which showed one order of magnitude lower (K _a=10⁶ M⁻¹). In addition, the binding affinities for two duplex DNA structures containing different stem sequences were within the range 10⁶ M⁻¹ for the cyclic ligand, denoting the selectivity for G4 over duplex of L92.

Electrophoretic mobility shift assays showed the ability of promoting the intermolecular G4 formation of all the ligands, being L92 the most efficient G4 inducer. Competition dialysis experiments confirmed the preference of the cyclic ligand for G4 structures with a slight preference for parallel conformations. The telomerase activity was evaluated and indicated the potent activity in the sub‐micromolar range of L92. In the proposed binding model, L92 adopted a semiclosed structure in which each dibenzo[b,j]‐1,10‐phenanthroline stack on the top of a G‐quartet although a second potential binding mode could be envisaged from the interaction within the G4 grooves of the semiclosed ligand conformation too.

In a follow up screening of G4 ligands of the same group, the interaction of a series of 26 polyamine‐based ligands and their selectivity for G4s was evaluated by means of FRET melting experiments.[ ¹¹⁴ , ¹¹⁵ ] In those experiments, they used the hTelo21 G4 in K⁺ and Na⁺ conditions to study the hybrid and antiparallel conformations, respectively. The need of two intercalative units was mandatory for the G4 interaction as confirmed by the poor stabilisation of the ferrocene derivative (L113). The size of the aromatic cores was important although the lower ΔT _m values of the acridine‐based ligands in comparison with the naphthalene analogues highlighted the much more important role of the structural arrangement of the semiclosed ligand conformation. Finally, the aromaticity rigidity appeared to be key as well, since the unfused rings of L97–L116 displayed lower variation of the melting (Figures 20 and 21).

Figure 20.

Structures of L97–L116.

Figure 21.

Structures of L117–L121.

The substitution pattern was determinant for the interaction because the different connection between the aromatic cores within the series afforded different ΔT _m values. It could be explained in terms of the freedom degree of rotation of the bonds, which affected the final conformation. Importantly, the linkers were polyamine chains which able a stronger stabilisation to G4s as denoted by the decrease of ΔT _m values upon swapping any nitrogen atom for oxygen or sulphur atoms. It is due to the combination of the larger steric hindrance of sulphur in comparison to nitrogen and oxygen and the potential involvement of the ammonium or amine group in establishing strong electrostatic interactions with the G4. The non‐dependence of the G4 interaction with the ligand net charge denoted that the G4 recognition of the ligands was not governed by the polyamine net charge. With regards to the selectivity, the authors highlighted that the semiclosed ligand conformation of naphthalene derivatives, positioned them as hits and that the attachment of one but no two bearing side chains increased the selectivity because of the formation of specific contacts.

An alternative approach was followed by Sissi and collaborators using a series of ligands in which only one large aromatic acridine moiety was cyclised at the 2,7‐position with polyamines of different lengths (see Figure 22). ^[116] The acyclic ligand used as a control showed a high stabilisation effect and an intercalative binding mode in duplex DNA whereas the larger macrocycle did not exhibit these binding features and alternatively bound to the grooves of duplex. Interestingly, the macrocycle having the shorter linker exhibited an intermediate behaviour promoting duplex stabilisation. The presence of the shorter ethylene spacer rather than propylene spacer featured in L124 appeared to be essential for obtaining a correct match between the binding sites of L123 and the nucleobases.

Figure 22.

Structures of L122–L124.

The efficiency of these ligands was modulated by the nature of the side chains. In particular, L123 was the most active in stabilizing G4s, followed by L124 and L122. Both L123 and L124 gave rise to stronger π‐π interactions with the 3’‐end G‐tetrad and the bent conformations of L123 and L124 did not prevent G‐quadruplex recognition.

Milelli, Sissi, Neidle et al. designed a series of macrocycles containing one aromatic moiety (NDI), which could effectively stack on a G‐quartet, and a small benzene unit bridged by polyamine side chains (Figure 23). ^[117] Some ligands were characterised by spermine‐like side chains differing in the length of the spacer between inner nitrogen atoms (L125–L129). They also studied analogue ligands (i) gathering two side chains to investigate the cyclisation and (ii) replacing the inner nitrogen atoms with oxygen atoms to explore the possibility of stablishing hydrogen bonds but unable to form ionic interactions. Using FRET melting experiments on hTelo21 and c‐kit2 G4s, ligand macrocyclization led to an increase in G4 stabilisation except for the oxygen‐swapped ligand. The most effective G4 ligand was characterized by the spermine‐like side chains (L128). The shortening in the length of the bridged chains connecting the inner nitrogen atoms along the series of ligands L126, L127 and L128 induced a decrease on the stabilisation. The elimination of one of the amine groups from L128 to L125 also led to a decrease in the G4s stabilisation effect, underlying the requirement of nitrogen atoms for effective interactions. The replacement of nitrogen with oxygen atoms of the ligands abolished the ability to interact with G4s.

Figure 23.

Structures of L125–L131.

L129 emerged as the most interesting compound considering its higher G4 stabilisation effect and selectivity. It is worth to highlight that the spermine‐like side chains prevented intercalation within the guanine‐cytosine region of duplex because of the steric hindrance and the polyamine linker was mostly involved in the electrostatic interactions with the negative counterpart of the phosphate backbone. The hit ligand L129 bound into the grooves of either G4s or duplexes exerted by the electrostatic effect provided by the ammonium groups. Interestingly, the four protonated aliphatic secondary amine groups of the spermine‐like side chains (as found for compounds L128, L129 and L130) were the most significant part of the designed macrocyclic NDI for G4 recognition. In addition, growth‐inhibitory activity and telomerase inhibition in cells showed an opposite activity to DNA‐binding, which highlighted the unfavourable pharmacological properties of highly charged polyamine ligands.

Polyamine macrocycles following a similar structural design than Teulade‐Fichou's team were reported by Carvalho et al. to bind G4s. ^[118] They described ligands containing one or two phenanthroline moieties which are connected by polyamine chains differing in the number of nitrogen atoms and length of the linker (L132–L135, see Figure 24). FRET melting screening showed that all the ligands stabilised both hTelo21 and c‐Myc G4s, being L132 the stronger binder. Since L135 stabilises efficiently G4, it suggested that one phenanthroline moiety was sufficient for G4 binding. The higher stabilisation along the ligands indicated that the number of methylene groups in the linker separating the aromatic moieties has a profound effect and that the hydrophobic interactions of the ligand conformation play a crucial role for L132. The low ΔT _m values for the duplex models indicated the selectivity for G4s over duplex DNA. CD and fluorescence spectroscopies confirmed the ligands preference for the parallel G4 topology and their moderate affinity for G4s (K _a=10⁶–10⁵ M⁻¹). All the ligands interfered the G4 unwinding and L132 and L135 halted the polymerase elongation. Finally, the ligands did not significantly affect the growth of HeLa cancer cells.

Figure 24.

Structures of L132–L135.

Kataev and collaborators recently reported polyamine and polyamine‐polyamide macrocycles (L136–L139) containing naphthalamide substituents as “turn on” sensors of G4s (Figure 25). ^[119] Their strategy was based on the recognition of phosphate DNA backbone by the macrocyclic core while the naphthalamide moieties formed π‐π stacking interactions with the G‐quartets. L137 and L139 demonstrated a strong fluorescence enhancement with G4s. The amide‐amide ligand L139 bearing the dimethylamine groups was the most selective fluorescence G4 sensor for parallel G4 topologies with affinities in the micromolar range (K=1.09 μM for c‐Myc).

Figure 25.

Structures of L136–L139.

Small polyamine macrocycles were also used as structural motifs of G4 ligands. In this regard, cyclen (1,4,7,10‐tetraazacyclododecane) and cyclam were used as clamps to hold together other moieties that selectively bind G4s.[ ¹²⁰ , ¹²¹ , ¹²² ] In a different approach, Vilar's team reported a series of bimetallic G4 ligands containing a cyclen unit, which was designed for metal chelation and explored the metal coordination to the bases of G4s (Figure 26). ^[123] Interestingly, the ammonium salts of the mononuclear metal complex and the binuclear metal complex provided high affinity for G4 similar to. These results indicated that the aromatic core could stack on the G‐quartet whereas the cyclen‐salt free and the cyclen‐metal reinforced the G4 interaction in a similar extension through electrostatic and covalent interactions, respectively.

Figure 26.

Structures of L140–L141.

4.3. Biscyclic Polyamine G4 Binders

Another strategy for developing G4 ligands containing polyamines relies in the reinforcement of their structural framework to afford tris‐macrocycles or cryptands (Figure 27). The incorporation of a third connecting unit led to an increase of the thermal effect (L143) in comparison with the macrocycle analogue (L108). ^[114] Therefore, the higher stabilisation could derive from the rigidity exerted by the three connecting units which was confirmed by the strong conformation restrain of the energetic minima obtained by molecular simulations. Interestingly, L143 exhibits an exceptional selectivity, although more detailed studies are needed to fully understand the origin of it.

Figure 27.

Structures of L142–L145.

Following this, Monchaud, Granzhan and colleagues reported the strong interaction of ligands L142 and L145 with hTelo21 G4 and three‐way junctions (3WJ), ^[124] showing high selectivity for the latter. In the work, they applied competition FRET melting, PAGE and fluorimetric titration experiments to confirm the high selectivity for three‐way junctions. Regarding the biological activity, all the ligands showed the characteristic DNA damage attributed to the conventional G4 ligands, although the authors proposed an additional cellular mechanism of action apart from targeting G4s. Recently, the same authors evaluated the selectivity of a series of biscyclic polyamine ligands for 3WJs but using G4s and duplex as competitor structures. By using a tandem of FRET melting, fluorescence, polymerase stop, NMR and molecular dynamics, they concluded several structure‐activity relationships in connection with G4s such as the design of ligands with large aromatic units (i. e. anthracene) led to a high 3WJ interaction but also a considerable G4 and duplex binding. Moreover, ligands containing smaller units yielded lower G4 interaction but increased the 3WJ preference. ^[125]

4.4. Polymeric Amine G4 Binders

There are several synthetic polymers containing amine groups such as polyethylenimine (PEI). PEI is widely used as gene transfecting agents because it can compact and pack DNA, which is shielded against degradation. Tang et al. reported the interaction of PEI towards G4s of different topologies. PEI largely stabilised hTelo22 and induced the formation of intramolecular G4s from different PQS. This study showed the effects of PEI as a molecular crowding agent, forming molecular crowding conditions at low PEI concentration as other agents (i. e. PEG).The authors excluded the molecular crowding effect as the driving force for the formation of the complexes with G4s. ^[126] In addition, they investigated the G4 condensation with the short polyamines spermine and diethylene triamine, which could not induce G‐quadruplex formation and concluded that condensation was a special effect of PEI with G4s. Interestingly, they found a more effective stabilising effect on intramolecular G4 topology over intermolecular as well as a relatively weak selectivity for G4s versus duplexes.

Recently, polyamido‐amine PAMAM dendrimers were combined with the thrombin (TBA) and AS1411 aptamers to form complexes for drug delivery.[ ¹²⁷ , ¹²⁸ ] The positively charged polymer strongly bound to TBA G4, likely by electrostatic interactions. Thesize of the non‐covalent complexes depended on the PAMAM:TBA ratio, solvent and temperature. The hydrodynamic size of the complexes increased with the ionic strength and the ratio, being the highest for the 1 : 1 molar ratio. Moreover, the zeta potential of the complexes changed from negative to positive when using [PAMAM]:[TBA] ratios lower than 1 : 1.5. With these results, the authors suggested that at low molar ratios the G4s are bound to the outer surface of PAMAM while at higher ratios, the aptamers could be enclosed into the dendrimer. ^[127]

5. Conclusion and Future Directions

Biogenic polyamines have been continuously omitted in the nucleic acid field, even though they are cellular metabolites at very high concentrations in cells capable to strongly bind to both RNA and DNA. To complete the picture of PA, there is still a lack of knowledge about their exact roles in biological processes such as cell growth. It is known that polyamines can strongly interact with classical Watson‐Crick duplex, but non‐canonical nucleic acid structures such as G‐quadruplexes have been likely understudied with PA to date. The G4 structure comprises several guanines from one or several strands forming G‐tetrads which are wrapped around a cation ion, likely potassium and sodium but also ammonium cations can be located into this ionic channel and fold a G‐quadruplex. Biogenic polyamines exhibit a dual effect of stabilising/destabilising depending on the concentration, and the nature and the number of amine groups of the PAs. Therefore, we strongly believe that there shall be updated the conditions used in the nucleic acids binding assays by including PAs to correctly mimic the biological context since these metabolites can strongly influence the G4 folding/unfolding. Herein, we have compiled the bibliography of the works connecting biogenic PAs and G4s and have highlighted the cellular effects of the interplay between PAs and G4s.

In the second part of the review, we have focused on the use of polyamines as structural motifs of ligands developed to target G4s. All reports to date have focused on the aromatic core able to bind to the aromatic G‐tetrads but no attention has been paved to the “attached complements”, which can be the key to achieve the desired selectivity for these non‐canonical structures. We have covered the most important works in the field using polyamines as ligand scaffolding and the main molecular features to yield potent and selective G4 ligands. This work pretends to be the essential guide for designing and developing novel G4 ligands with polyamines considering the benefits and drawbacks of using PAs.

Conflict of Interests

The authors declare no conflict of interest.

Biographical Information

Ariadna Gil Martínez holds a PhD in Chemistry and a master in Sustainable Chemistry from the University of Valencia. Her expertise includes the synthesis and characterization of polyaminic ligands and studies on DNA interactions. She is currently a junior postdoctoral researcher at the Universidad Miguel Hernández de Elche (Spain), specialised in organic chemistry. Ariadna combines her academic background with a commitment to education, research innovation, and a passion for designing and illustrating for science communication.

Biographical Information

Cristina Galiana holds both degrees in Pharmacy from CEU Cardenal Herrera University (UCHCEU) and Biotechnology from the Polytechnic University of Valencia and a master's in Bioinformatics from the University of Valencia (UV). She obtained a PhD in Pharmacy from UCHCEU focused on Medicinal Chemistry. Upon her postdoctoral period at UV, she worked with non‐canonical DNAs as therapeutic targets. From 2024, she is researcher in the Computational Biomedicine Research Group at the Principe Felipe Research Center. Her interests cover the identification of biomarkers for the diagnostic and progression of neurodegenerative diseases through the analysis and integration of omics data.

Biographical Information

Andrea Lázaro‐Gómez received her BSc in Chemistry from the University of València in 2021, completing an undergraduate stay in Vilar's group at Imperial College London in 2020. She pursued her MRes in Molecular Nanoscience and Nanotechnology of the Institute of Molecular Science (ICMol) at the University of Valencia (Spain) under the supervision of Dr. Jorge González García, winning the Royal Spanish Society of Chemistry′s Best Master's Thesis Award 2023. She is currently conducting her PhD research on the synthesis of fluorescent probes to visualise G‐quadruplexes in live cells and studying their properties through photophysical and computational methods.

Biographical Information

Laura Mulet Rivero is a PhD student at the Inorganic Chemistry Department of the University of Valencia (Spain) under the supervision of Dr. Jorge González García and Prof. Enrique García‐España. She received a bachelor's degree in Chemistry from the University of Valencia and a master's degree in Molecular Nanoscience and Nanotechnology from the same University. Her research interests focus on the interaction of metal complexes with non‐canonical DNA G‐quadruplexes.

Biographical Information

Jorge holds a Chemistry Degree and PhD in Supramolecular and Bioinorganic Chemistry from the University of Valencia. He conducted several postdoctoral stays at the University of Bordeaux, Imperial College London, Institute Curie and University of Kansas, specialising in G‐quadruplex DNAs. Jorge returned to his alma mater as Distinguished Researcher in 2019 and promoted to Assistant Professor in 2024. He coordinates the Group of Chemical Biology of the University of Valencia, focusing on the development of supramolecular and bioinorganic tools to investigate non‐canonical nucleic acid structures.

Gil-Martínez A., Galiana-Roselló C., Lázaro-Gómez A., Mulet-Rivero L., González-García J., ChemBioChem 2025, 26, e202400873. 10.1002/cbic.202400873

Data Availability Statement

Data sharing is not applicable to this article as no new data were created or analyzed in this study.