Nucleoside Sensing

Nucleoside drugs are of paramount importance in aids or cancer therapy due to their antiviral or antitumor activity. Since it is virtually impossible for highly polar phosphorylated species to pass cell membranes, the respective drugs are administrered as neutral nucleosides, and subsequently activated by intracellular conversion into the 5′-triphosphates. A much more elegant approach is the design of Trojan horses, which are masked nucleotides, i. e., neutral lipophilic prodrugs with decreased polarity [1]. However, even for the nucleosides, it is difficult to pass the blood brain barrier [2], so that carrier molecules are needed which may act as potent transporters. The programmed switch between a neutral and a phosphorylated state poses the challenge to develop selective detection methods. Few examples are available for nucleotide-selective hosts in water, inter alia, based on cyclophane scaffolds [3]. High binding constants (micromolar K_d values) were especially achieved with receptor molecules containing a central phenanthridinium system [4]. To the best of our knowledge, however, no selective artificial receptor molecule has been presented to date which is able to recognize a neutral nucleotide in the presence of an anionic nucleotide. Over the past decades, Klärner et al. have created a large number of rigid molecular clips and tweezers, comprising an electron-rich cavity which is able to accomodate unpolar or cationic small guest molecules [5]. In a cooperation, we have recently rendered the parent molecular clip water-soluble by adorning the central benzene ring with methyl phosphonate functionalities and found efficient binding of cationic aromatic substrates such as NAD⁺ (Figure 1) [6]. Now we discovered that neutral electron-deficient guests are also included inside the electron-rich cavity. The question arose, if these molecules could indeed be utilized for nucleoside sensing.

NMR titrations indicate inclusion in the clip cavity by substantial upfield shifts of all guest protons involved. They also allow determination of binding constants and estimation of the binding mode [7]. All nucleosides and mononucleotides were tested systematically - the respective K_a values are listed in Figure 2. At first glance it becomes evident, that only the neutral nucleoside candidates are bound by the clip, whereas the negatively charged nucleotides are rejected, except for AMP. The reason is obviously electrostatic repulsion between the host and guest phosph(on)ate moieties. Nucleosides are bound with K_a values between 2000 and 7000 M^-1, most likely depending on the size of their π-face as well as their electrostatic potential surface (Figure 3). Carbonyl groups are counterproductive due to their large dipole moment - and lead to repulsion from the electron-rich inner cavity wall. Since large π-faces are hard to desolvate in aqueous solution, the nonpolar microenvironment of the clip cavity greatly facilitates this process and leads to a positive free energy balance of the inclusion process.

Upfield shifts in the nucleic bases amount to 1-2 ppm (Δδ_max), whereas those in the ribose part are small (0-0.2 ppm), indicating that the aromatic electron-poor heterocycle is preferentially bound by the clip and the much more polar ribose part remains oriented towards the solvent. The exact orientation inside the cavity has in related cases already been calculated very successfully with quantum chemical methods, and could definitely be applied to these interesting complexes as well [8]. By contrast, except for AMP with its relatively large π-face no chemical shift changes were observed after equilibration with nucleotide guests, irrespective of the number of phophate groups present (Δδ < 0.03 ppm). Thus, mono-, di- and triphosphates such as ATP are all rejected by the clip. This renders 1 highly selective for nucleosides.

Nucleoside antibiotics such as 3′-azido-2′,3′-dideoxythymidine (AZT) and 2′,3′-didehydro-2′,3′-dideoxythymidine (ddT) are highly potent inhibitors of the human immunodeficiency virus HIV [9]. These also bind to the clip, albeit with lowered affinity. Obviously, the altered ribose skeleton poses steric constraints on the inclusion of the heterocyclic fragment. Probable complex geometries, which are in agreement with NMR data, are depicted in Figure 4.

In the future, we intend to test, if the complex of a nucleoside inside the clip penetrates membranes more easily than the drug itself. If successful, the clip could be further modified to carry additional lipophilic groups and be used as a carrier for nucleoside drugs across the BBB. Precomplexation of a solvatochromic dye within the clip cavity which is subsequently released on nucleoside binding and thereby changes its color, would lead to a real sensor system with a quantifyable visible readout. Experiments in this direction are underway in our laboratories.

Synthetic details on the preparation of clip 1: see ref. 6 and 8.

General procedure for NMR titrations with clip 1 and nucleos(t)ide guests:

a)

dilution titration: host and guest were dissolved as a 1:1 mixture in D₂O and an NMR spectrum was measured. Subsequently, the solution was diluted several times with D₂O. Pure host as well as pure guest solutions of varying concentrations were also measured and compared with respect to their proton shifts. Only those proton signals which are insensitive towards dilution, can be used as probes. All NMR spectra were routinely calibrated onto the rest proton signal of the deuterated solvent (D₂O: 4.700 ppm). The corresponding binding constants and Δδ_{sat values} were obtained by standard nonlinear regression.

b)

conventional titration: ten NMR tubes were filled each with the same amount of a solution of the guest in D₂O. The host solution (D₂O) was added in increasing volume increments to tubes 2-10. All tubes were subsequently filled with pure solvent to the very same total volume to exclude potential dilution effects. Tube 1 served as the reference. The guest concentration was estimated at the reciprocal value of the expected association constant (K_d). NMR spectra were calibrated onto the rest proton signal of the deuterated solvent (D₂O: 4.700 ppm). The corresponding binding constants and Δδ_{sat values} were obtained by standard nonlinear regression.