Organomercury Nucleic Acids: Past, Present and Future

Abstract Synthetic efforts towards nucleosides, nucleotides, oligonucleotides and nucleic acids covalently mercurated at one or more of their base moieties are summarized, followed by a discussion of the proposed, realized and abandoned applications of this unique class of compounds. Special emphasis is given to fields in which active research is ongoing, notably the use of HgII‐mediated base pairing to improve the hybridization properties of oligonucleotide probes. Finally, this minireview attempts to anticipate potential future applications of organomercury nucleic acids.


Introduction
Mercury is one of the metals most extensively studied for its interactions with nucleic acids. In line with the softness of Hg II , these interactions mainly take the form of coordination to the nitrogen donors of nucleobases. Mercury also forms fairly stable organometallic compounds, in many cases under conditions withstood by nucleic acids. In fact, natural nucleic acids feature two sites that are readily mercurated by simple treatment with Hg II salts, namely the C5 atoms of cytosine and uracil bases. Appropriately designed artificial nucleobases further widen the range of organomercury modifications that can introduced to oligonucleotides and nucleic acids.
Numerous applications of organomercury nucleic acids and their constituents have been proposed over the past five decades. Some, such as density labeling in pycnographic analysis have only seen limited popularity while others, such as affinity tagging, have been phased out by more modern methods after a brief period of usage. Yet others, notably the utilization of covalently mercurated nucleobases as synthetic intermediates, are still seen occasionally. Finally, organomercury oligonucleotides as probes in single nucleotide polymorphism (SNP) genotyping and mercurated nucleotides as markers in electron microscopy DNA sequencing are examples of promising future applications still in their infancy. This minireview summarizes the synthetic methods towards organomercury nucleic acids and oligonucleotides, outlines their historic and contemporary applications and attempts to shed light on future prospects.

Electrophilic aromatic substitution by Hg II
Mercuration of aromatic rings through electrophilic aromatic substitution has been known since the mid-1800s. [1,2] With unactivated aromatic hydrocarbons prohibitively harsh conditions are required but electron-rich arenes, including nucleobases, readily react under conditions tolerated by nucleic acids. [3,4] The most reactive sites are the C5 atoms of cytosine and uracil, quantitatively mercurated in 2 h with 4 mM aqueous mercuric acetate at pH 6.0 and 50°C ( Figure 1A and B). It should be noted, however, that an excess of mercuric acetate is typically required owing to competing coordinative interactions with the endocyclic nitrogen atoms, especially thymine and uracil N3 and guanine N1. [5] The purine bases are inert to electrophilic aromatic substitution by mercuric acetate under conditions feasible for the mercuration of nucleosides. 8-Methylmercurypurine nucleosides, however, can be prepared by 30 min treatment with methylmercuric nitrate at pH 7 and 50°C ( Figure 1C and D). [6] Analogously to the reaction of pyrimidine bases and mercuric acetate, covalent mercuration at C8 only takes place once all of the nitrogen donors have been saturated.
Artificial nucleoside and nucleotide analogues with sufficiently electron-rich base moieties greatly widen the scope of covalent mercuration. C7 of 7-deazaadenosine-5'-triphosphate, for example, can be mercurated under the same conditions as C5 of cytosine and uracil ( Figure 1E). [3] With 1,N 6 -ethenoadenosine, the etheno bridge introduces a new site of covalent mercuration, namely the carbon atom bonded to N1 (Figure 1F). [7] Interestingly, 3,N 4 -ethenocytidine also undergoes mercuration at the corresponding carbon atom rather than C5 ( Figure 1G). Mercuration can be promoted at a desired site by activating ortho/para directors, such as hydroxy or amino substituents, and prevented at an undesired site by alkyl substituents. Facile and selective mercuration at C2 of a 3fluoro-6-methylaniline C-nucleoside ( Figure 1H) provides an illustrative example of the power of this approach. [8] Nucleobase analogues with multiple electron-rich carbon atoms, such as phenol [9] or 6-phenyl-1H-carbazole [10] (Figure 1I and J), can be mercurated more than once although at progressively slower rates.

Direct mercuration of oligonucleotides and nucleic acids
Electrophilic aromatic substitution at C5 of cytosine and uracil bases, as well as electron-rich carbon atoms of artificial nucleoside analogues, by Hg II proceeds under sufficiently mild conditions to be feasible also on oligonucleotides [11] and even longer nucleic acid sequences. [4] The reaction is highly sequence dependent, homopolymers of cytidine and uridine being mercurated much faster than naturally occurring heteropolymers. In contrast, secondary structure of the nucleic acid hardly affects the rate of mercuration, probably because of the denaturing conditions of the reaction mixture. [4] Thiol-promoted demercuration, on the other hand, can be performed under conditions where the polymercurated nucleic acid retains its secondary and tertiary structure and in such cases site-dependent reactivity patterns are observed. This selectivity has been harnessed for the preparation of a monomercurated tRNA although homogeneity of the product in terms of the site of mercuration was not established. [12] With synthetic oligonucleotides, preventing off-target mercuration is straightforward as any cytosines and uracils that should remain unreacted can be replaced with the inert 5-methylcytosines and thymines, respectively.

Enzymatic polymerization of organomercury nucleotides
The utility of organomercury nucleotides as substrates of polymerases depends strongly on the ligand sphere of Hg II . With relatively weakly coordinating ligands, such as acetato or chlorido, organomercury nucleotides are potent inhibitors of both DNA and RNA polymerases, in all likelihood owing to coordination of Hg II to a critical sulfhydryl group. [3,13,14] In the presence of a thiol ligand, polymerization proceeds smoothly albeit with some concomitant demercuration. [15] Interestingly, different enzymes exhibit different requirements for the thiol ligand -although 2-mercaptoethanol is the ligand of choice in most cases, some enzymes, notably calf thymus terminal deoxynucleotidyl transferase, prefer a sterically less demanding ligand such as methane-or ethanethiol. [3] Enzymatic polymerization of 5-mercuripyrimidine nucleotides has been proven with both template-directed [16] as well as template-independent [13] polymerases. In the former case, 5mercuriuridine-5'-triphosphate was readily incorporated on a poly[d(AT)] but not on a poly(dG)·poly(dC) template while the opposite is true for 5-mercuricytidine-5'-triphosphate, thus suggesting that the fidelity of the enzymatic reaction is retained. The scope of other modifications tolerated on the organomercury nucleotide remains obscure, as the studies found in the literature are limited to derivatives of natural nucleotides as substrates.

Post-synthetic introduction of organomercury nucleobase surrogates
As discussed above, direct mercuration of oligonucleotides by electrophilic aromatic substitution is a feasible strategy when the aromatic ring to be mercurated is sufficiently electron-rich. With less reactive systems, the conditions required can be so harsh that competing reactions, such as RNA cleavage, [17] become a problem. In such cases, the organomercury moiety can first be synthesized separately using whatever conditions necessary and then introduced to the oligonucleotide by suitable conjugation chemistry. The feasibility of this strategy has been demonstrated recently by oximation of a supportbound aminooxy-functionalized oligonucleotide with 2-mercury-3-hydroxybenzaldehyde. [18] 3. Organomercury Nucleosides, Nucleotides and Nucleic Acids as Reactive Intermediates

Halodemercuration
Sequential mercuration and halodemercuration provides access to halogenated aromatic compounds. [2] In the field of nucleic acid chemistry this approach has found a niche application in the synthesis of radiolabelled nucleosides. Originally described soon after the first reports on organomercury nucleotides and nucleic acids, [47] iododemercuration of 5-chloromercuri-2'-deoxyuridine has later been refined into a facile and robust procedure for the preparation of 123 I-, 125 I-and 131 I-labelled 5iodo-2'-deoxyuridine (Scheme 2). [48,49] Remarkably, radiolabelling through iododemercuration (as well as bromodemercuration and reductive demercuration with sodium borotritiide) has been proven also on polymeric nucleic acids. [47] Combined with the site-selective mercuration of chemically synthesized oligonucleotides discussed above, halodemercuration should allow the preparation of oligonucleotides bearing a sterically conservative radiolabel at a predetermined site. Future studies will hopefully demonstrate the practical utility of such an approach.

Organomercury Nucleotides as Density Labels in Pycnographic Analysis
One of the earliest proposed applications of organomercury nucleotides relied on the sheer mass of the heavy mercury atom, as well as the applicability of 5-mercuricytidine-5'triphosphate as a substrate in enzymatic polymerization. The buoyant density of extensively mercurated DNA is considerably higher than that of native DNA and this difference can be exploited in CsCl density gradient centrifugation. Such DNA has been prepared through replication in permeabilized bacterial cells and found to band at a higher density than its unmodified counterpart. [14] The use of organomercury nucleotides as pycnographic probes has not, however, gained widespread popularity.

The Potential of Organomercury Nucleotides as Markers in Electron Microscopy DNA Sequencing
The possibility of DNA sequencing by electron microscopy was first explored before the advent of Sanger sequencing and more modern methods. [50] As the idea is again attracting attention, it is interesting to note that mercury was proposed as a heavy atom marker to facilitate the interpretation of DNA electron micrographs as early as 1974. [7] The suggested procedure involved sequential treatment of the DNA with chloroacetaldehyde and mercuric acetate, resulting in nearquantitative conversion of adenine and cytosine bases to covalently mercurated 1,N 6 -ethenoadenine and 3,N 4 -ethenocytosine bases ( Figure 1F and G), respectively. To distinguish between mercurated adenine and cytosine bases, the DNA was first subjected to acid-promoted depurination, after which the treatment described above afforded a sample having only the cytosine bases covalently mercurated. Finally, the positions of guanine and thymine bases would be inferred based on the complementarity rules of Watson-Crick base pairing. The technological hurdles were too high to overcome at the time but more recently a very similar approach has proven successful. [51] Instead of covalently mercurated 1,N 6 -ethenoadenine and 3,N 4 -ethenocytosine, methylmercury complex of 5mercaptouracil was used as the heavy atom marker, allowing undisrupted base pairing with adenine. The use of 5-mercuricytosine would nicely complement this method as the labelled DNA could be prepared directly from native DNA by simple incubation with mercuric acetate.

Organomercury Nucleotides in X-ray Crystallographic Structure Determination
Multiwavelength anomalous diffraction (MAD) is a phasing method used in the interpretation of X-ray crystallographic data from biomacromolecules. [52] The method requires incorporation of an anomalous scatterer, that is, a heavy atom, in the structure to be studied, ideally in such a way that the native threedimensional structure is not disturbed. In the case of nucleic acids, the easy introduction to C5 of pyrimidine bases makes mercury a particularly attractive candidate. [53] The relatively wide major groove of DNA can accommodate a 5-mercuri substituent with little disturbance of the double helix. In RNA the major groove is significantly narrower and mercuration should, hence, be confined near the ends of double-helical regions.

Organomercury Nucleotides as Affinity Tags
The very high stability of Hg II -thiol complexes can be harnessed for affinity chromatographic purification of covalently mercurated nucleic acids. Quantitative retention on thiol-functional-ized stationary phase, such as agarose, controlled-pore glass or cellulose, is achieved through mercuration of as few as one base out of 200. [16] Addition of a competing thiol, such as 2mercaptoethanol, in the mobile phase releases the mercurated nucleic acid from the stationary phase, allowing complete recovery. The method saw use in the early 1980s in the isolation of nascent nucleic acids from biological sources. [54][55][56][57][58] Mercuration was accomplished randomly throughout the sequence by incubating either permeabilized cells or isolated nuclei with 5mercuripyrimidine nucleoside triphosphates.
Covalent mercuration allows cellular nucleic acids to be not only isolated but also visualized within the cell. [59][60][61] Accordingly, nucleic acid probes of an appropriate sequence were mercurated by treatment with mercuric acetate and then allowed to hybridize sequence-specifically with metaphase chromosomes or interphase nuclei. Finally, the cells were soaked with a thiol-functionalized hapten ligand, either trinitrophenyl, biotinyl or fluorescyl. Visualization by fluorescence microscopy, either directly (in the case of fluorescyl) or after immunochemical amplification (in the case of trinitrophenyl and biotinyl) revealed localization of the ligands at the expected target sites.

Hg II -Mediated Base Pairing of Organomercury Nucleobases
The concept of metal-mediated base pairing [62][63][64][65][66][67][68][69][70] was first introduced in 1963 with mercury as the bridging metal ion [71] and mercury still remains the most extensively studied metal in this context. Most of the research efforts have been directed at coordinative Hg II -mediated base pairs, in particular the T-Hg II -T homo base pair. [72][73][74][75] Such base pairs typically feature a dicoordinate bridging Hg II and a linear coordination geometry although with artificial nucleobase surrogates higher coordination numbers have been reported as well. [76] Various applications for coordinative Hg II -mediated base pairing are under active development, [63] ranging from sensors for Hg II [77][78][79][80][81] to molecular wires. [82] High affinity, rapid association and dissociation [83] and responsiveness of nucleic acid secondary structure to subtle changes in the binding mode [84] make Hg IImediated base pairs attractive components for the construction of DNA nanostructures.
The earliest mention of Hg II -mediated base pairing of an organometallic nucleobase, albeit not termed as such, is the report from 1984 on formation of N3-Hg II -C5-linked polymers of 5-mercuriuridine. [85] The idea was revisited in 1996 in a systematic 1 H and 199 Hg NMR study employing (1,3-dimethyluracil-5-yl) mercury(II) as a model compound. [86] N3-methylation ruled out formation of the kind of coordination polymers observed previously with 5-mercuriuridine, allowing Hg II -mediated hetero base pairs with canonical nucleobases to be investigated. In line with previous studies on methylmercury, [5,87] coordination at guanine N1 and thymine and uracil N3 with concomitant deprotonation of the donor atom was preferred.

ChemBioChem
Minireviews doi.org/10.1002/cbic.202000821 More recently, Hg II -mediated base pairing of organomercury nucleobases has been studied as a means to increase the hybridization affinity of oligonucleotide probes in biological media. [11] These studies have involved recording concentrationdependent 1 H NMR spectra at monomer level and UV melting profiles at the oligomer level and, when applicable, the results have generally been in good agreement. In other words, high stability of an individual Hg II -mediated base pair in solution translates into a high melting temperature of a double-helical oligonucleotide incorporating the same base pair. Owing to the very rapid ligand exchange of Hg II , the NMR spectra typically only show one set of signals representing an average of all species in equilibrium. For the same reason, the UV denaturation and renaturation curves are superimposable even with heating and cooling rates typical of UV melting studies on unmodified oligonucleotide duplexes (e. g., 0.5°C min À 1 ).

Mononuclear Hg II -mediated base pairs
Flipping of a pyrimidine base from anti to syn conformation places its C5 in the position normally occupied by N3. Therefore, the C5-Hg II -N3-linked base pairs of 5-mercuricytosine [88] and 5-mercuriuracil [89] with thymine (Figure 2A and B) are in all likelihood isosteric with the well-documented N3-Hg II -N3-linked base pair between two thymines ( Figure 2D). [72][73][74][75] A similar geometry seems likely also for the 3-fluoro-2-mercuri-6-meth-ylaniline-thymine base pair ( Figure 2C). [8] Within oligonucleotides, all of these base pairs have a similar effect on the thermodynamic parameters of hybridization, namely increased (less negative) enthalpy and entropy owing to dehydration of the bridging Hg II ion. [73,75] Especially in the case of 3-fluoro-2mercuri-6-methylaniline, the latter effect more than compensates for the former, resulting in considerable duplex stabilization. [8]

Dinuclear Hg II -mediated base pairs and triples
Covalent attachment of more than one Hg II ion allows novel binding modes not resembling that of the T-Hg II -T base pair. Multinuclear organomercury nucleobases can be categorised as mono-or bifacial depending on the positioning of the Hg II ions relative to each other. In monofacial bases, such as 1,8dimercuri-6-phenyl-1H-carbazole, [10] the Hg II bridges converge to bind to a single nucleobase, typically on its Watson-Crick face. NMR spectrometric or X-ray crystallographic data on such base pairs is not available but high-level DFT calculations have predicted remarkably similar structures for two dinuclear Hg IImediated base pairs with thymine, one coordinative (Figure 3A) [90,91] and one organometallic ( Figure 3B). [10] In both of these base pairs, the two Hg II ions coordinate to O2 and O4 of the thymine base. With the coordinative 1,N 6 -ethenoadenine-Hg II 2 -thymine base pair, more recent calculations suggest additional coordination to thymine N3 when this base pair is embedded within the base stack of a double helix. [91] While very high stabilities have been reported for a number of multinuclear Ag I -mediated base pairs, [92][93][94][95][96][97][98] corresponding results on these dinuclear Hg II -mediated base pairs (organometallic or otherwise) were less impressive, [10,90] perhaps owing to electrostatic repulsion between the Hg II ions.
Bifacial multinuclear organomercury nucleobases are characterized by diverging Hg II bridges binding to two other nucleobases. 2,6-Dimercuriphenol, for example, forms stable dinuclear Hg II -mediated base triples with adenine, cytosine and thymine. [9] The latter, in particular, proved highly stabilizing in the middle of a homothymine*homoadenine·homothymine triple helix. As C2 and C6 of the 2,6-dimercuriphenol nucleobase analogue are equivalent to C5 and N3 of pyrimidine nucleobases, base pairing at both Watson-Crick and Hoogsteen faces most likely exhibits similar geometry as the mononuclear Hg IImediated base pairs discussed above ( Figure 3C).

Hg II -mediated base pairing in SNP genotyping
Various methods for the detection of single nucleotide polymorphisms (SNPs) rely on differences in the hybridization affinities of oligonucleotide probes for the target sequence. [99][100][101][102] Unfortunately, canonical Watson-Crick base pairing is less than ideal for this approach. Although the matched base pair is usually much more stable than any of the mispairs, stabilities of the latter do not differ sufficiently from each other to allow reliable identification of the polymorphic nucleobase ( Figure 4A). Metal-mediated base pairing offers a way to overcome this limitation. Organomercury nucleobases exhibit very different base pairing preferences from their natural counterparts and 3-fluoro-2-mercuri-6-methylaniline, in particular, stands out favourably for SNP genotyping. UV melting temperatures of short double-helical oligonucleotides pairing this organomercury nucleobase with either adenine, cytosine, guanine or thymine all differed by at least 7°C, enough for reliable identification of the variable base ( Figure 4A). [8] The identity of the base pairing partner was also clearly reflected in the 19 F chemical shift of 3-fluoro-2-mercuri-6-methylaniline. The feasibility of SNP genotyping based on metal-mediated base pairing was recently demonstrated with a molecular beacon -type probe incorporating the 3-fluoro-2-mercuri-6-methylaniline in the middle of the recognition loop. [103] At appropriate temperature, stabilities of the different Hg II -mediated base pairs translated into different hairpin-heteroduplex equilibria and further into different fluorescence emission intensities ( Figure 4B). Organomercury oligonucleotides, hence, hold great future potential as hybridization probes in SNP genotyping.

Summary and Outlook
Covalent mercuration of nucleic acids predates the efficient chemical synthesis of oligonucleotides. Organomercury nucleic acids were, hence, first employed in applications that do not rely on site-specific mercuration, such as density labelling for pycnographic analysis or affinity tagging. These applications largely fell out of favour before oligonucleotide synthesis became mainstream, leaving their full potential unattained. Recently, synthetic oligonucleotides site-specifically mercurated at predetermined natural or artificial hot spots have brought about a renaissance of organomercury nucleic acid chemistry. Potential new applications include the use of organomercury oligonucleotides as hybridization probes, notably in SNP genotyping. The scope of some established applications, such as radiolabelling through halodemercuration, could potentially be expanded to oligonucleotides. Finally, while coordinative Hg II -mediated base pairing has already been harnessed in DNA nanotechnology, the use of organomercury oligonucleotides in this field remains an unexplored but interesting possibility.