Considering that short, mainly heterochiral, polypeptides with a high glycine content are expected to have played a prominent role in evolution at the earliest stage of life before nucleic acids were available, we review recent knowledge about polypeptide three-dimensional structure to predict the types of conformations they would have adopted. The possible existence of such structures at this time leads to a consideration of their functional significance, and the consequences for the course of evolution.

A majority of the standard texts dealing with proteins portray the peptide link as a mixture of two resonance forms, in one of which the nitrogen atom has a positive charge. As a consequence, it is often believed that the nitrogen atom has a net positive charge. This is in apparent contradiction with the partial negative charge on the nitrogen that is used in force fields for molecular modeling. However, charges on resonance forms are best regarded as formal rather than actual charges and current evidence clearly favors a net negative charge for the nitrogen atom. In the course of the discussion, new ideas about the electronic structure of amides and the peptide bond are presented.

Considering how biological macromolecules first evolved, probably within a marine environment, it seems likely the very earliest peptides were not encoded by nucleic acids, or at least not via the genetic code as we know it. An objective of the present work is to demonstrate that sequence-independent peptides, or peptides with variable and unreliable lengths and sequences, have the potential to perform a variety of chemically useful functions such as anion and cation binding and membrane and channel formation as well as simple types of catalysis. These functions tend to be performed with the assistance of the main chain CONH atoms rather than the more variable or limited side chain atoms of the peptides presumed to exist then.

A novel tripeptide motif called a nest has recently been described in proteins with the function of binding anionic, or partially anionic, atoms such as carbonyl O atoms. In the present work, a search for nests in small polypeptides stored in the Cambridge Structural Database is reported. 37 unique examples were found: over half form part of hydrogen-bond arrangements resembling those in proteins, such as Schellman/paperclip loop motifs, various types of beta-turn and Asx-turns or Ser/Thr-turns, while a third are in novel situations, some involving binding to anionic groups from other molecules within the crystal complex. An example is the antibiotic vancomycin, which incorporates a prominent nest forming part of a peptide-binding site. This nest binds the carboxylate of the C-terminal D-alanine of the bacterial cell-wall precursor peptide, thereby inhibiting the final step of bacterial cell-wall synthesis. As in proteins, a number of nests occur in short peptides with an alternating glycine/L-amino-acid sequence but, uniquely to non-ribosomally synthesized short peptides, several nests within them are constructed from alternating D- and L-amino acids, and such sequences seem to specially favour nests.