DNA usually adopts a structure in which its strands are connected through hydrogen bond alignments of G:C and A:T into a topology with two grooves. Guanine-rich nucleic acids may form nucleic acid architectures that have four bases participating in pseudo-planar hydrogen bond alignments that define topologies that have four grooves. These are known as DNA quadruplexes. These architectures are of interest due to their biological functions, as well as utility as therapeutics, and materials. They add to the chemical diversity of structures with useful functions. Solution NMR spectroscopy has played a crucial role in the initial developments in the study of quadruplex DNA. The motivating premises for the study of quadruplexes followed a train of events that were predicated on structure. Some of the initial general questions were about the hydrogen bonding alignments feasible, the bases that can participate, how are they stabilized, etc. The uses for these architectures were initially speculative; and remained so for a long period of time. As with any chemical entity, to develop understanding on their use, or their biological function, we have always had the need to understand structure. Solution NMR structural studies are ever more useful in this context. There are a number of reviews that focus on the methodologies that have been used [1], and on how specific structural characteristics can be observed in quadruplexes by solution NMR [2]. The modified nucleosides shown in Figure 1 have been used as probes for structure determination, but also in their own right as part of quadruplex structure for varied reasons. A number of reviews have been written on modified G-quadruplex nucleic acids [3,4], but none specific to structure determination. Here we review modified nucleic acids that can be used to aid in the determination of solution quadruplex structure by NMR and how they influence the folding of the topologies examined.

Quadruplexes are thus composed of four hydrogen bond aligned bases, denominated tetrads, that stack onto each other to form a well-defined nucleic acid stem. It has four grooves whose width is defined by the angular relationship defining each of the bases to their sugar pucker: they can be narrow, medium, or wide. This relationship is best defined by the glycosidic bond angle (GBA) of the bases that make up the quadruplex stem—Figure 2. We have previously defined the relationships between structural parameters of these architectures based on a geometric formalism defined by this relationship [5]. The primary requirement for formation of a quadruplex stem is thus that a minimum of two stacked tetrads with the same groove widths stack onto each other. In a reductionistic sense there can be two states for the GBA: one in which the H1' of the sugar is trans (anti) with respect to the H8 of the base, and the other in which their relationship is cys (syn). The position of specific GBAs within the quadruplex stem defines, and is defined by, the topology of the architecture. Other tetrads, triads and mismatches can stack onto a quadruplex stem. Furthermore, tetrads may include bases other than guanosines.

The process of determining solution structures of quadruplexes by NMR starts of with a solution containing a DNA or RNA molecule of interest, with known sequence of bases. The initial steps involve the identification of ¹H signals corresponding to the sequence of residues in isotopic natural abundance samples: the sequence specific assignment. Initially the distances between the H8 proton of the base and the H1' of the sugar of a nucleoside are fundamental for identifying sequence through NOE-based experiments. The identification of other proton signals within the nucleoside can allow for this sequential identification. The sequence of nucleosides can also be followed through scalar coupling NMR experiments that correlate protons of two sequential sugar puckers. However, for a number of reasons such as difficulties in identifying sequential steps involving G(syn) residues, major overlap of sugar proton signals, and others that reside outside the scope of this review, the sequential assignment of oligonucleosides may not possible. These issues can be addressed by the use of modified oligonucleosides.

Base substitutions are essential tools in the identification of NMR signals, but also in imparting biological or thermodynamic stability to particular topologies. Their use is now widespread due to the development of solid phase synthesis methods. One of the most commonly used base modifications is the substitution of the H8 of guanosine by Br-, methyl-, O-methyl-, amino-, and oxo-substituents. Its origins reside in the need to define the sequence specific assignment of residues in a DNA molecule. This substitution would result in the disappearance of NOE cross-peaks due to the H8 proton. Instinctively, it is expected that the substitution of H8 is associated with an increase of steric effects due to the size of the substituent. Thus all of these substituents result in locking, or inducing the relationship between base and sugar to a G(syn). Most of the substitution studies were performed with the bromo substituent. Various studies have shown that it is feasible to identify the G(syn) guanosines of a topology through systematic substitution of the guanosine bases in its sequence. However, the substitution of specific G(syn) bases did not result in folded topologies. Here we provide a plausible explanation.

Riboguanosine (rG) substitutions can in principle be utilized for identification of G(anti) due to the fact that the C3'-endo preferred conformation of the sugar pucker favors this GBA. Tang and Shafer have made a series of rG modifications in the TBA sequence. The ones that proved successful in folding this topology were all in G(anti) positions [28]. The structure of a dimeric propeller-type RNA G-quadruplex formed by the human telomeric RNA sequence r(UAGGGUUAGGGU) has been determined in solution by NMR [29]. In this study all guanosine bases adopt their preferred anti conformation. They utilized site-specific deoxyribose substitutions to assist resonance assignments. The substituted residues are recognized thanks to their upfield-shifted H2′/H2′′ peaks.

LNA are 2'-O-4'-C-methylene-linked ribonucleotide nucleic acid analogues [30,31] that result into a forced C3'-endo conformation which in turn favors the anti glycosidic bond angle configuration [32]. LNA is known to increase the affinity of oligonucleotides for DNA and RNA targets [33]. It is possible to influence the thermodynamically preferred structure of G-quadruplexes with the introduction of LNA modifications [34], increasing the possibility of obtaining parallel quadruplexes. In fact, NMR structures of tetramolecular quadruplexes with fully or partly LNA modified Guanine residues have resulted in all-parallel structures [35,36]. A unimolecular NMR G-quadruplex structure of the TBA with a single LNA modified guanosine at the 3'-end corresponding to a G(anti) retained its original topology [37]—Figure 4B. A systematic study with the Oxytricha nova telomeric sequence d[G₄T₄G₄]₂ in K⁺ (normally antiparallel) revealed that only one combination of four LNA modified Guanines gave rise to a well folded quadruplex where all eight guanines participate in the stem [38].

The guanosine to inosine substitution has been used in G-quadruplex structural studies for almost 20 years [39,40]. In the folded architectures the substituted residue still participates in the tetrad but causes chemical shift differences, and/or improved dispersion in the imino proton spectrum of the quadruplex. The only difference with guanosine in its chemical structure is the absence of the amino group which causes chemical shift differences mainly of its own imino proton but also of the aromatic proton of the hydrogen bonded guanosine in the tetrad. Numerous examples of single guanine-to-inosine substitutions resulting into significantly improved NMR spectra and/or a single quadruplex conformation exists [41,42,43,44,45,46,47].

Partially or fully isotopically labelled nucleosides have been used in quadruplex NMR structural studies [7,48,49,50,51,52,53,54] to assist with resonance assignments. Isotope filtering techniques allow for the discrimination of intra- and intermolecular NOEs [55]. Such techniques, have been used in the structure determination of a symmetric dimeric quadruplex, allowing for the unambiguous differentiation of intra- and inter-strand contributions for topology-defining NOEs [56]. Although this approach is dependent on knowledge of the folding topology, other methods focus on the unambiguous discrimination between intra- and intermolecular hydrogen bonds in symmetric multimeric quadruplexes [57] without prior structural knowledge requirements. More details regarding methodological aspects of isotopic enrichment have been reviewed previously.

Of particular importance, is the site-specific low isotopic enrichment (1%–2%, ¹⁵N and/or ¹³C) as an aid in the sequential assignment of G-quadruplexes [58,59]. This cost-effective approach has become part of the established methodology for structural studies [21,60] of these molecules, resulting in unambiguous assignment of imino signals (¹⁵N) and or sugar protons (¹³C). This approach may become even less expensive by utilizing only ¹⁵N-enrichment. Site-specific 2% ¹³C, ¹⁵N-labeled samples can additionally be utilized to unambiguously assign methyl and H6 protons of Thymines, using ¹³C-filtered experiments [61].

Site-specific incorporation of ¹⁵N and ¹³C-labeled nucleoside (N¹, N², N⁷-¹⁵N, C²-¹³C-labeled guanine and N¹, N⁶, N⁷-¹⁵N-labeled adenine) phosphonates have been used to resolve imino proton assignments of a guanine and adenine rich sequence forming a unique fold [19]. The structure determination was additionally aided by site-specific substitutions of 2,6-diaminoPurine and 8Br-Adenine for Adenine, 8Br-Guanine, 7-deazaGuanine and dInosine for Guanine, and dUracil and 5Br-dUracil for thymine [19], thus illustrating the full range of options in low isotopic enrichment and base modifications as a tool in quadruplex structure determination.

Currently, it is still too expensive to utilize near 100% isotopically labelled nucleic acids. In addition, in solution NMR structural studies it is customary to establish the predominance of a single fold before pursuing full structure determination. In this context the use of modified bases serves the purpose of selecting a topology from equilibria. Base modifications and substitutions are a convenient means of signal identification as well as probes for establishing pseudo-planar hydrogen bond alignments. All of these results improved spectra for which, in select cases, full structure determination has been performed.

There are a number of modifications to nucleosides that improve the thermal stability of quadruplexes, or their resistance to enzymatic degradation. There are a number of modifications that have been used to manipulate the folding of quadruplexes. Thus one can expect many more structures determined for nucleoside-modified quadruplexes as therapeutic entities, for use as sensors, catalytic systems, or materials. Solution NMR structural studies will continue to be central to these applications.