Leu-D-Leu-) (2) were synthesized and used for a study of the interaction ofb-rings in solution. In appropriate solvents, 1 and 2 formed dimers consisting of twob-rings connected through six interannular H-bonds and having the N-Me group(s) on the solvent-exposed face. The study also afforded indications that 2 formed different dimers of this kind, differing in the relative orientation of the twos -rings. These experimental observations provide strong support to the idea that nnsubstituted o,L-alternating cyclooligopeptides, such as cyclo(hexaleucine), can selfassemble and give rise to long tubular stacks ofp -rings.
Introduction. -In 1974, De Santis and coworkers [I] predicted that peptide rings formed by alternating D-and L-residues in equivalent / 3 -like conformation @-rings) should be able to pile up and form multi-ring aggregates with tubular structure. Such aggregates, of which two types are illustrated in Fig. , are more than a fascinating example of supramolecular architecture. Indeed, they appear to be uniquely suited for a number of special applications (e.g., for mimicking biological channels, effecting separations of ions or molecules, ordering metal atoms in one-dimensional arrays), and they might even usher the development of novel optical and electronical devices. The first reports [2] [3] on synthetic D,L-alternating cyclopeptides do not provide any hint to such aggregates. In 1987, however, work from our laboratory [4] on some D,L-alternating cyclopeptides with 4, 6, or 8 valine, leucine, or phenylalanine residues showed that these cyclooligopeptides, in contrast to open-chain analogues, are remarkably insoluble*). This behavior is understandable, if these cyclopeptides have a tubular structure of the types shown in Fig. I . IR Data [4] and the observation [5] of ap-ring-like molecular conformation in crystals of cyclo(hexava1ine) and of cyclo(hexaphenyla1anine) containing cocrystallized CF,COOH provided some support for this possibility. Much stronger evidence for a tubular structure in a cyclopeptide, i.e., in cyclo[-(~-Ala-~-Glu-~-Ala-~-Gln),-], was presented recently by Ghadiri et al. . Despite these studies, however, an experimental demonstration of the stacking ofp -rings is still lacking.
Aiming at studying this interaction, we considered the use of D,L-akernating cyclopeptides having some or all L(or D)-residues N-methylated. We expected that the N-Me group(s), which would protrude from one side of theB -rings, would allow a pairing of the ') *)