Large and stable transmembrane pores from guanosine-bile acid conjugates.

A variety of natural and synthetic compounds are known to selfassemble to give transmembrane ion channels.1 Hydrogen-bonded macrocycles that can π-stack are a new type of channel motif.2 Thus, folate quartets stack to give ion channels in lipid bilayers.3 This folate assembly had a single-channel conductance of 10-20 picosiemens (pS), values consistent with the quartet’s 3 Å diameter. We found that a noncovalent assembly of 16 guanosine monomers could be cross-linked to give a “unimolecular” G-quadruplex that can transport Na+ across lipid membranes.4 We now report that the ditopic guanosine-lithocholate 1 forms discrete channels in phospholipid membranes (Figure 1). These pores are large (nS conductance) and stable, with “open” times of seconds, distinguishing them from most synthetic channels, which typically conduct in the pS range with millisecond lifetimes.1,5 Lehn and Barboiu have independently shown that ditopic monomers with guanine end groups form supramolecular polymers in cation-templated processes.6 Possible supramolecular structures built from these ion-templated G4-quartets are depicted in Figure 2. In addition to the G4-quartet channel, such structures might well stack to form pores for transmembrane transport. Of relevance was Barboiu’s demonstration that Na+ and K+ could be transported across films made from G4-quartet polymers.6b The nucleoside-sterol conjugate 1 has two guanosine groups connected by a bis-lithocholate linker. This spacer was inspired by Kobuke’s studies that showed that bis-cholic acid derivatives formed cation-selective channels with pS conductance.7,8 We envisioned that membrane insertion of 1, followed by formation of G4-quartets, might well provide functional pores (Figure 2). Compound 1 was made by coupling 2′,3′-tBDMSi-5′-amino G,9 with a bis-lithocholic acid. Compound 2, with -NMe amide end groups, was a control. The 1H NMR spectrum of 1 gave sharp peaks in DMSO-d6, a polar solvent that inhibits self-assembly mediated by hydrogen bonding. In contrast, the NMR spectrum of 1 in CDCl3 gave much broader signals, consistent with self-association in this “poor” solvent. We used CD spectroscopy to gain evidence that 1 forms stacked G4-quartets in a nonpolar solvent.10 Figure 3 shows CD spectra for samples of 1 in CHCl3. The CD spectrum of 1 (blue) was taken after isolation from a silica gel. This sample showed a weak Cotton band in the 200-280 nm region, suggesting some stacked G4quartets.10 We added [2.2.2]-cryptand to ensure that any adventitious cations bound by 1 were sequestered. Indeed, the resulting spectrum (green trace) was inactive. We next stirred the mixture of 1 and [2.2.2]-cryptand in the presence of excess K+ 2,6-dinitrophenolate (DNP). The CD spectrum of ditopic 1 was much different after extraction of K+DNP(red trace). This sample showed a CD signature diagnostic for stacked G-quartets, with a positive band at λ ) 266 nm and a negative peak at λ ) 240 nm.10 The complex formed by 1 and K+ also showed a strong Cotton band at λ ) 295 nm, a signal that has been previously observed for G-quadruplexes formed from both nucleosides and DNA strands.11,12 The control compound, NMe amide 2, showed no CD activity under identical conditions. These data indicate that the K+ ion can template formation of stacked G4-quartets by 1 in a nonpolar environment. We next used voltage-clamp experiments to show that 1 forms single channels in planar membranes; some of these channels had remarkably long lifetimes and large conductance values for pores made from a synthetic compound.13 Figure 4 shows representative records of conductance, as mediated by 1, in a planar bilayer at an applied voltage of -10 mV in 1 M KCl (trans)/KCl (cis) solution † Department of Chemistry and Biochemistry. ‡ Department of Biology. Figure 1. Ditopic G-sterol 1 and control bis-lithocholamide 2. Typical traces of conductance vs time, after addition of 1 or 2, are depicted. The guanosine end-groups are needed to form a transmembrane channel.