DNA G-quadruplex and its potential as anticancer drug target

Abstract

G-quadruplex secondary structures are four-stranded globular nucleic acid structures form in the specific DNA and RNA G-rich sequences with biological significance such as human telomeres, oncogene-promoter regions, replication initiation sites, and 5′ and 3′-untranslated (UTR) regions. The non-canonical G-quadruplex secondary structures can readily form under physiologically relevant ionic conditions and are considered to be new molecular target for cancer therapeutics. This review discusses the essential progress in our lab related to the structures and functions of biologically relevant DNA G-quadruplexes in human gene promoters and telomeres, and the opportunities presented for the development of G-quadruplex-targeted small- molecule drugs.

1 Introduction

DNA is widely recognized as a double-helical structure with a crucial role in genetic information storage. However, most recent results from the ENCODE project indicate that only ~3% of human genome is expressed in protein and that RNA and DNA may form non-canonical secondary structures that are functionally important [1]. One such example is the four-stranded G-quadruplex DNA secondary structures, which have been gaining considerable attention for their emerging role in biological systems. First observed in 1910 [2], the G-tetrad structure was not identified until 1962 [3]. DNA G-quadruplexes were first found to form in the single-stranded 3′ overhang of human telomeres [4–6]. More recently, DNA G-quadruplexes were found to form in the proximal promoter regions of human oncogenes to regulate gene transcription [7–10]. Most recently, DNA G-quadruplexes have been associated with replication initiation [11– 13]. In addition, RNA G-quadruplexes have been found to form in mRNA and, recently, in 5′- and 3′-UTRs [14–16].

Many proteins have been identified to interact with G-quadruplex structures; these proteins may bind and stabilize G-quadruplexes or unwind and destabilize G-quadruplexes [17–19]. Very recently, by using a G-quadruplex-specific antibody, G-quadruplex structures have been visualized in human cells at different sites on chromosomes other than telomeres and have been shown to increase in frequency in the presence of a G-quadruplex-interactive compound [20, 21]. Quarfloxin, the first G-quadruplex-interactive drug, has reached Phase II clinical trials for the treatment of cancer [22]. Thus, there is compelling evidence for the existence, biological significance, and potential drug-gability of G-quadruplexes [8, 17]. This review, which is associated with National Conference on Biophysical Chemistry (NCBPC) meeting presentations, will be focused on the progress in our lab related to the structures and functions of biologically relevant G-quadruplexes and the opportunities presented for the development of G-quadruplex-targeted small-molecule drugs.

2 Structural features of G-quadruplexes

G-quadruplex structures are formed by stacked guanine tetrads (G-tetrads) (Figure 1). Within a G-tetrad, four guanine bases are arranged in a square plane with Hoogsteen hydrogen bonding, instead of the Watson-Crick hydrogen bonding of B-DNA (Figure 1(a, b)). G-quadruplex structures form readily in solution in the presence of either Na⁺ or K⁺ [23]. These physiologically relevant monovalent cations are required to stabilize G-quadruplex structures by positioning between the G-tetrad planes in coordination with the O6 atoms of the tetrad guanines (Figure 1). G-quadruplexes can be monomeric or multimeric (e.g., dimeric or tetrameric) (Figure 1(c, d)). Within the G-tetrad, guanine residues may adopt either syn or anti-glycosidic conformation [8] (Figure 1(b)). Adjacent DNA strands in a G-quadruplex can have the same (parallel) or opposite (anti-parallel) orientation (Figure 1(d)); guanines of parallel DNA strands will adopt the same, generally anti, conformation, whereas guanines of anti-parallel strands will adopt opposing conformations [8] (Figure 1(b)).

Figure 1.

(a) Schematic illustration of a G-tetrad, four guanine bases arranged in a square plane with Hoogsteen hydrogen bonding. The H1–H1 and H1–H8 connectivity that are notable in NOESY experiments are shown in red and blue respectively; (b) a G-tetrad structure. Guanines in a G-tetrad may adopt either syn or anti glycosidic conformation; the guanines from parallel G-strands adopt the same glycosidic conformation and the guanines from antiparallel G-strands adopt opposing glycosidic conformations; (c) schematic tetrameric and dimeric G-quadruplexes with three G-tetrads; (d) examples of monomeric (intramolecular) G-quadruplexes with different folding structures and loop conformations. Monovalent cations (K⁺ or Na⁺, shown as blue spheres), are required to stabilize G-quadruplexes by coordinating with the O6 atoms of the adjacent G-tetrad planes.

3 Monomeric G-quadruplexes

Most biologically relevant G-quadruplexes are monomeric, or intramolecular, G-quadruplexes [8, 17]. Generally, a monomeric G-quadruplex-forming DNA sequence contains at least four tracts of three or more consecutive guanines, with flanking segments at the two ends and loop residues intervening between the G-tracts. Intramolecular G-quadruplexes are globular structures that form naturally under physiological conditions and exhibit great diversity in their topologies and loop/flanking conformations, which depend on the DNA sequences. Different loop conformations have been observed, such as the strand-reversal/or propeller loop connecting adjacent parallel strands, the lateral/ or edgewise loop connecting adjacent anti-parallel strands, and the diagonal loop connecting anti-parallel strands on opposite sides of a tetrad (Figure 1(d)). A single DNA sequence may adopt different G-quadruplex folding topologies, as in the case of the human telomeric DNA, or it may form multiple structures or loop isomers, as in the case of most G-quadruplex-forming gene promoter sequences. While extensive structural study has elucidated some rules governing intra-molecular G-quadruplex folding, it is not always possible to accurately predict the quadruplex folding in a specific sequence. Thus, structural characterization is a necessity in the determination of G-quadruplex folding. Most intramolecular G-quadruplex structures in the public domain have been determined by NMR spectroscopy in solution.

4 Human telomeric G-quadruplexes

Telomeres are unique DNA-protein complexes that protect the chromosome ends from degradation; the structure and stability of human telomeres are important for cancer, aging, and overall genetic stability [24–26]. Telomerase activity is activated in over 80% of cancer cells to maintain the telomere lengths [27–29]. Human telomeric DNA consists of 5–30 kb tandem repeats of d(TTAGGG)n that end in a single-stranded 3′ overhang that is 35–600 bases long [30–32]. It has been shown to form G-quadruplexes that can be targeted by small molecular compounds to inhibit telomerase activity and disrupt telomere capping in cancer cells [33–35].

Human telomeric DNA G-quadruplexes have been found to be highly polymorphic in their structures [36]. A 22-mer telomeric DNA was found to form a basket-type structure in Na⁺ solution [37] (Figure 2(A-a)), and a parallel structure in the crystalline form in the presence of K⁺ [34] (Figure 2(A-b)). More recently, human telomeric DNA was shown to form two equilibrating hybrid-form G-quadruplexes in K⁺ solution [38–42] (Figure 2(B)). While both K⁺ structures adopt hybrid folding (i.e., with three parallel strands and one antiparallel strand connected by two lateral loops and one strand-reversal loop, Figure 2(B-a)) the orders of the loops are different and therefore the molecular structures are distinct, with unique capping and loop conformations (Figure 2(B-b)). Despite the low energy difference between the two equilibrating hybrid structures, interconversion between the two appears to be kinetically slow, suggesting the presence of a high-energy intermediate [40].

Figure 2.

(A) Folding topologies of the basket-type Na⁺-solution (a) and parallel-stranded K⁺-crystal (b) intramolecular G-quadruplexes formed by human telomeric sequence wtTel22. Red box = anti guanine, Magenta box = syn guanine; (B) the folding topologies (a) and molecular structures (b) of the two equilibrating hybrid-form human telomeric intramolecular G-quadruplexes, hybrid-1 and hybrid-2, in K⁺ solution; (C) a strand-reorientation model of the interconversion between different forms of human telomeric G-quadruplexes, likely through the two-tetrad and G-triplex form intermediates; (D) the hybrid structures can effectively form packed multimers in human telomeres.

Based on our NMR structural studies, we proposed a strand-reorientation model with a partially unfolded triplex- type and two-tetrad intermediates [39, 43] (Figure 2(C)). More recently, several studies have provided insights and experimental evidence into the intermediate steps of G-quadruplex folding, including the G-hairpin [44] and the G-triplex [45]. With the 5′ and 3′ ends positioned on opposite sides of the G-quadruplex, the hybrid structures can effectively form packed multimers at the telomere end [39, 40], a conformation that has been confirmed by computer modeling study [46] (Figure 2(D)). All human telomeric G-quadruplexes appear to have small energy differences relative to each other [40, 46]. The structure polymorphism appears to be an intrinsic property of the human telomeric DNA sequence, particularly the TTA loop sequence [36]. As the human telomeric DNA sequence contains the same tandem repeats, the highly conserved telomeric sequence in higher eukaryotes may be naturally selected for its potential to form multiple G-quadruplexes so that different structures may be specifically recognized by different proteins to control biology. This recognition may present a potential opportunity for small molecule targeting.

5 Promoter G-quadruplexes

The potential for G-quadruplex formation in promoter regions is largely concentrated in genes associated with cell growth and proliferation [7]. Computational analyses [47– 51] showed the clustering of G-quadruplex-forming sequences in close proximity (within 1 kb) to the transcriptional start site. These oncogene promoters are typically TATA-less with G-rich regions in their proximal promoters, such as c-MYC [10, 52], VEGF (vascular endothelial growth factor) [53], BCL-2 (B-cell lymphoma 2) [54, 55], and PDGF-R-β [56], as well as HIF-1α (hypoxia-inducible factor 1α) [57], KRAS [58], c-KIT [59, 60], and RET [61]. While telomeric G-quadruplexes form in the single-stranded 3′-overhang of telomeres, promoter G-quadruplexes are formed in regions of double-stranded DNA. Using the c-MYC gene promoter, studies have demonstrated in vivo that active transcription induces negative superhelicity behind the transcriptional machinery [62–64]. This negative superhelicity leads to the intermediate single-stranded DNA, which can spontaneously fold into the more stable G-quadruplex structures [65].

Unlike telomeric DNA, the promoter G-quadruplex- forming sequences are significantly more diverse and often contain more than four tracts of guanine [7, 8, 17]. Consequently, a given sequence can form multiple G-quadruplexes through utilizing varying combinations of G-tracts or different loop isomers through utilizing varying guanines on one G-tract. Three-tetrad G-quadruplexes are most common. A notable feature in these promoter sequences is the presence of the G3NG3 motif, which readily contributes to formation of a robust parallel-stranded structure motif with a single-nucleotide strand-reversal loop (Figure 3). This G3NG3 motif is so widely prevalent in promoter G-quadruplexes that it may have been naturally selected as a basis for promoter G-quadruplex formation [8, 17].

Figure 3.

Comparison of G-quadruplex-forming sequences in selected gene promoters. The human telomeric sequence is also shown. The types of the G-quadruplexes are indicated. The tetrad-guanines are shaded, and the G₃NG₃ motifs are boxed.

The first example of the G3NG3 structural motif is the major G-quadruplex structure formed in the c-MYC promoter, (1:2:1 are the lengths of the three loops) [66] (Figure 4(A)). We showed that the single-nucleotide loop is highly favored for the robust parallel-stranded structural motif G3NG3 because of the right-handed twist of the adjacent G-strands (Figure 4(A-b)). The G-quadruplex-forming region of the c-MYC promoter is a 27-nucleotide nuclease hypersensitive element (NHE) III1 that has been shown to control about 85% of the c-MYC transcriptional activity [67, 68]. This sequence contains five consecutive runs of guanine. Mutational analysis in conjunction with DMS footprinting and luciferase reporter assays have indicated that the major G-quadruplex formed in K⁺ solution involves four consecutive 3G-runs that adopt parallel folding (Myc2345), with the major loop isomer being 1:2:1 [10, 69, 70] (Figures 3, 4(A)). While the c-MYC NHE III1 is the first and most extensively studied gene promoter for the G-quadruplex formation, we were very lucky in that it appears to be the simplest, containing only five G-runs with all the G-quadruplexes adopting parallel folding in K⁺ solution [66, 69–72]. Indeed, the G3NG3 parallel-stranded structural motif was found in all of the c-MYC G-quadruplexes and loop isomers. Structural, thermodynamic and kinetic analysis showed different stabilities associated with loop sizes, loop arrangements, and loop compositions of parallel structures [71, 72].

Figure 4.

The folding topologies (a) and molecular structures (b) of the major G-quadruplexes formed in the human c-MYC gene promoter (A) and human VEGF gene promoter (B) in K⁺ solution. The capping structures are shown in purple. Guanine = red, adenine = green, thymine = blue, cytosine = yellow.

Since the determination of the c-MYC G-quadruplex, parallel-stranded G-quadruplex structures have been found to be common in the human promoter sequences. Notably, most of these parallel structures contain 3 G-tetrads with 1-nt (nt = nucleotide) first and third loops, and a variable-length middle loop (Figure 3). For example, the major G-quadruplex formed in the VEGF promoter (1:4:1) also contains 1-nt first and third loops, but a 4-nt middle loop (Figure 4(B-a)). Interestingly, the 4-nt middle loop of the VEGF G-quadruplex stretches over the 5′ tetrad to form a unique capping structure with the flanking segment [73] (Figure 4(B-b)). In contrast, the 2-nt middle loop of the major MYC-quadruplex (1:2:1) stays in the groove with the capping structures formed solely by the two flanking segments [66] (Figure 4(A-b)). More recently, we found the major G-quadruplex (bcl2-1245G4) formed in the BCL-2 proximal promoter adopts a parallel structure (1:13:1) with a 13-nt middle loop [74] (Figure 5(A, B-a)). This is unexpected because 7 nt has been considered the maximum stable loop length and is used in all available quadruplex-predicting software. It thus appears that, by having two 1-nt loops, a stable parallel G-quadruplex can contain a rather extended middle loop. Therefore, while parallel structures are common to the promoter G-quadruplexes, our studies indicate that each G-quadruplex is likely to adopt unique capping and loop structures by its specific variable middle loop and flanking segments.

Figure 5.

Two interchangeable G-quadruplexes formed in the overlapping region of the BCL-2 gene promoter. (A) The G-quadruplex-forming BCL-2 gene promoter sequence bcl2-Pu39. The regions (black line) and G-runs (red line) for the formation of the two distinct G-quadruplexes, bcl2-MidG4 and bcl2-1245G4, respectively, are shown; (B) the folding topologies of the two interchangeable BCL-2 promoter G-quadruplexes.

Further variation of the 1-nt parallel-strand motif is exemplified by the major G-quadruplex formed in the PDGFR-β (platelet-derived growth-factor receptor β) promoter. Instead of using four runs of three or more consecutive guanines, this 3 G-tetrad quadruplex features a “broken” G-strand of guanines from two different G-runs to form a primarily parallel-stranded structure with three 1-nt strand-reversal loops and one additional lateral loop [75] (Figure 6). The formation of this novel “broken-strand” PDGFR-β G-quadruplex appears to maximize the number of 1-nt loops, again demonstrating the preference for 1-nt loops in the parallel-stranded structural motif, which may more accurately be described as GiNGj. These variations in parallel G-quadruplexes give rise to different overall structure properties, which may be specifically recognized by proteins or small-molecule ligands.

Figure 6.

The sequence and the folding topology of the major G-quadruplex formed in the PDGFR-β gene promoter, which forms a novel “broken-strand” structure. The guanines involved in the tetrad formation are underlined, whereas the four runs of three or more consecutive guanines are colored red.

In addition, certain promoter sequences may form multiple stable G-quadruplexes in equilibrium. For example, the BCL2 proximal promoter contains a 39-nucleotide G-rich region with 6 G-runs separated by one or more bases (bcl2-Pu39, Figure 5(A)). This sequence has been shown to form two stable G-quadruplexes, a hybrid-type structure with a G3NG3 motif and two lateral loops [55, 76] (bcl2-MidG4, Figure 5(B-b)) in addition to the 1:13:1 parallel structure which spans 5 G-runs (bcl2-1245G4, Figure 5(B-a)) [74]. The two structures have similar stability. Thermodynamically the parallel 1245G4 structure is slightly more stable than the MidG4; however, the hybrid-type MidG4 structure could be kinetically more favored because it forms on the four consecutive G-runs and thus has shorter loop-lengths. The presence of two distinct interchangeable G-quadruplexes in the overlapping region of the BCL-2 promoter is intriguing, and suggests a novel mechanism for gene transcriptional regulation by different proteins recognizing different G-quadruplex structures. In addition, two interchangeable structures may be recognized by different small molecules for gene modulation.

6 Small-molecule targeting of G-quadruplexes

DNA G-quadruplexes, in particular those formed in human telomere and oncogene promoters, have become attractive cancer-specific molecular targets for anticancer therapeutics [17]. Targeting G-quadruplexes in promoters provides a potential means to modulate oncogene transcription [8, 9]. The stabilization of telomeric G-quadruplexes in cancer cells often leads to apoptosis [34, 35]. Elevated interest has been seen in the development of G-quadruplex-interactive small molecules [77]. NMR spectroscopy is a powerful tool for studying small molecule interactions with G-quadruplexes [78]. X-ray crystallography and biophysical methods such as CD, UV, and fluorescence spectroscopies are also commonly utilized in studying quadruplex-ligand interactions [79, 80]. Early dogma on the optimal G-quadruplex-interactive compounds has focused on compounds with large, planar, and symmetric cyclic rings, such as TMPyP4 (tetra- (N-methyl-4-pyridyl)porphyrin) and telomestatin [81, 82] (Figure 7(a)), because they maximize stacking interactions with the external G-tetrad. However, these drugs are found to have low specificity for intramolecular G-quadruplexes [78].

Figure 7.

(a) Structures of G-quadruplex-interactive small molecules; (b) a representative model of the NMR solution structure of the 2:1 complex of quindoline and c-MYC G-quadruplex. The quindoline molecules are shown in space-filling model in magenta. Guanine = red, adenine = green, thymine = blue; (c) axial view of the two drug-induced binding pockets at the 5′-end and the 3′-end.

Our recent NMR solution structure of the 2:1 complex of quindoline and c-MYC G-quadruplex provides some important insights: an asymmetric compound with a smaller stacking moiety and appropriate functional groups is preferred for specific binding to the intramolecular G-quadruplex [83]. Quindoline is a derivative of the natural product cryptolepine, with a crescent-shaped ring system and a diethylamino side chain (Figure 7(a), that has been shown to stabilize the c-MYC promoter G-quadruplex and lower c-MYC protein levels [84]. We have determined the NMR solution structure of a 2:1 complex of quindoline and c-MYC G-quadruplex with one quindoline bound at each end of the quadruplex, the first drug complex structure of a biologically relevant intramolecular promoter quadruplex [83] (Figure 7(b)). This NMR structure shows an unexpected drug-induced repositioning of the flanking sequences at both ends of DNA, namely, the “induced intercalated triad pocket” recognition (Figure 7(b)), in a manner somewhat analogous to a reorganized ligand-induced fit observed in riboswitches [85]. While both 3′ and 5′ complexes show similar overall features, there are identifiable differences, observable only in solution, to emphasize the importance of both stacking and electronic interactions. The 5′ complex involves more hydrophobic and stacking interactions, whereas the 3′ complex at the more hydrophilic 3′-surface involves a H-bonding interaction that is dependent on salt concentration (Figure 7(c)) [83]. The complex structures demonstrate the importance of the shape of the ligand as well as the two flanking bases in determining binding specificity. Unlike the previously recognized paradigm, our results indicate that asymmetric compounds containing a smaller stacking moiety, in particular the crescent-shaped moiety, with appropriate functional groups, are more likely to bind in a specific manner to an intramolecular G-quadruplex. Therefore, the molecular mechanism for G-quadruplex recognition by small molecules may be divided into two steps. The first step involves the induced intercalated triad pocket recognition (by the core of ligand) and second involves the groove/loop interactions (by different functional groups), which could each be exploited in selective recognition of small-molecule drugs.

7 Future directions

Our understanding of G-quadruplexes has expanded from that of a laboratory curiosity to biologically relevant and validated targets for cancer therapeutics. Intramolecular G-quadruplexes form naturally under physiological conditions and are stabilized by monovalent cations present in cells. Depending on their DNA sequence, these structures exhibit great diversity in their structures. Quarfloxin, the first-in-class G-quadruplex-interactive drug, illustrates the potential for quadruplex-targeting drugs to be well tolerated and highly efficacious [22, 86]. We are currently working on structural studies and rational design of different small molecules that bind specifically to intramolecular G-quadruplexes. In addition, proteins have been found to interact with the c-MYC promoter G-quadruplex and be involved in c-MYC gene transcription. Nucleolin specifically binds and stabilizes the c-MYC G-quadruplex and functions as a transcription repressor [87], whereas NM23-H2 binds and unfolds the c-MYC G-quadruplex and functions as a transcription activator [88, 89]. We are studying the interactions of these proteins with the c-MYC G-quadruplex to exploit their cellular functions and the potential for small molecules targeting.