CYCLOTIDIAN Craik holds the peptide model. PHOTO BY JENNIFER CRAIK

But a new generation of circular and knotted peptides and proteins has more recently appeared on scientists' radar screens, demonstrating that what goes around does indeed come around. For example, circular proteins (as opposed to circular peptides) were little known a decade ago. However, in recent years, a number have been discovered in bacteria, plants, and animals--including several members of a family of well-rounded proteins called cyclotides.

CIRCULAR PROTEINS "are conventional proteins in every way, apart from the fact that their termini are seamlessly stitched together by an as-yet-unknown process in bacterial, plant, and animal cells," says professor of biomolecular structure David Craik of the University of Queensland, Australia, who specializes in cyclotides. "Tying the ends of proteins together using peptide bonds greatly enhances their stability and resistance to chemical, thermal, or enzymatic degradation," he adds. This enhanced stability suggests that cyclized proteins could prove to have a range of new applications as drugs and agricultural agents.

A key distinction between different types of cyclic peptides and proteins is that "classical cyclic peptides such as cyclosporin are biosynthesized nonribosomally"--that is, in enzyme-driven biosynthetic pathways--"while circular proteins are direct gene products," Craik says. Circular peptides and proteins can also be made synthetically in the laboratory.

Indeed, a number of groups have produced circular peptides and proteins synthetically and have used them to build more elaborate structures in the past few years. For instance, in 1993, chemistry professor M. Reza Ghadiri of Scripps Research Institute and coworkers induced cyclic peptides made of alternating d- and l-amino acid residues to self-assemble into hollow peptide nanotubes. Later, they discovered that these nanotubes are capable of inserting themselves into cell membranes and that they exhibit potent and selective antibacterial activity [Nature, 412, 452 (2001)].

Intein-based methods for cyclizing peptides and proteins have been developed by chemistry professor Stephen J. Benkovic of Pennsylvania State University; senior scientist Ming-Qun Xu of New England Biolabs, Beverly, Mass.; biological chemist Nicholas E. Dixon of Australian National University, Canberra; and their coworkers. An intein is a fragment that can excise itself, resulting in the ligation of neighboring fragments (exteins) through a peptide bond. Chemistry professor Tom W. Muir's group at Rockefeller University, New York City, has devised chemical-ligation and intein approaches for the cyclization of synthetic and recombinant peptides.

PLENTY OF KNOTTIN In a model of the knottin conotoxin gm9a, the protein backbone is shown as a smooth tube, and three disulfide bridges are shown in a ball-and-stick representation. In all knottins, a disulfide bridge (orange) penetrates a macrocycle (blue) formed by two other disulfides and interconnecting backbone segments. Although conotoxin gm9a's backbone is linear (noncyclic), the close proximity between its N- and C-termini shows how cyclic knottins can easily form by ring closure. MOLMOL [J. MOL. GRAPHICS, 14, 51 (1996)] FIGURE BY LAURENT CHICHE

Assistant professor of cell biology and chemistry Philip E. Dawson of Scripps and coworkers have developed protein-based catenanes (circular, interlocking protein rings) as building blocks for self-assembling and easily modifiable protein assemblies. And they're currently trying to thread linear peptides through cyclic peptides to create pseudorotaxanes (ring-and-axis structures) and catenanes. Dawson credits chemistry professor (now emeritus) Kurt M. Mislow of Princeton University and coworkers for early work on knotted and catenated loops in proteins.

Last year, three groups, working independently, determined the structure of the antibacterial peptide microcin J25 to be like a thread through the cyclic eye of a needle. They corrected an earlier study in which the structure was determined erroneously to be a head-to-tail cyclic peptide [J. Am. Chem. Soc., 125, 12382, 12464, and 12475 (2003); C&EN, Nov. 10, 2003, page 46]. The peptide is one of the few known natural agents that blocks bacterial RNA polymerase and is thus a potential target for the development of antibacterial agents.

And physics professor Huan-Xiang Zhou of Florida State University and coworkers have used polymer modeling and other theoretical techniques to study the folding and stability of cyclic and other topologically linked proteins. Zhou notes that "topology" has generally referred to the arrangement of protein secondary structures and that this has not posed much of a problem because most proteins are linear (noncyclic) chains. But he believes that the growing number of known cyclic and knotted proteins now makes it necessary to restrict the use of "topology" to descriptions of protein backbone configurations and that secondary structure arrangements should now be referred to as "architecture."

Unlike peptides and proteins with cyclic structures, those with knotted backbones continue to be quite scarce. "The search for knots in protein has uncovered little that would cause Alexander the Great to reach for his sword," wrote mathematical biologist William R. (Willie) Taylor [Nature, 406, 916 (2000)]. He was alluding to a legend that Alexander the Great used his sword to cut open a complex knot tied by King Gordius of Phrygia. This feat enabled Alexander to eventually fulfill a prophecy of that time: that the person who loosed the Gordian knot would one day rule all of Asia.

	TO BE OR KNOT TO BE The unique needle-and-thread structure of the antibacterial peptide microcin J25 can be seen in this ensemble of superimposed NMR-derived structures. COURTESY OF DAVID CRAIK

EXCLUDING KNOTS formed by post-translational cross-linking, the few proteins that are considered to be knotted have very simple knots, "with one end of the chain extending through a loop by only a few residues--10 in the 'best' example," Taylor explained. This type of knot "is not rigorously defined, and many weak protein knots disappear if the structure is viewed from a different angle."

In his Nature paper, Taylor reported a knot-detecting computer algorithm that did not have such viewpoint sensitivity. Using the algorithm to search a protein-structure database, he discovered a figure-eight knot in the plant protein acetohydroxy acid isomeroreductase and proposed a protein-folding mechanism by which such a knot could form.

Despite the rarity of backbone-knotted structures, one major class of peptides and proteins that are knotted through their disulfide linkages has emerged in the past couple of decades. These are called knottins (http://knottin.cbs.cnrs.fr/).

The 1982 determination of the structure of the first knottin, found in potato plants, revealed a unique grouping of knotted disulfide linkages in which one disulfide bridge penetrates a macrocycle formed by two other disulfides and interconnecting backbone segments. This unusual scaffold was later shown to be present in small proteins from other plants and from fungi, arthropods, mollusks, and vertebrates as well. The structures of more than 100 knottins have now been determined.

"Backbones of knottins are not knotted in a mathematical sense, even if we take disulfide bridges into account and even when the knottins are macrocyclic," explains senior structural biologist Laurent Chiche of the National Center for Scientific Research (CNRS) in Montpellier, France. But when Chiche and coworkers named this class of proteins in 1990, they selected "knottins" as the term that they believed best described the proteins' characteristic shape. Knottins are also sometimes referred to as "inhibitor cystine knots."

Knottins have been found to act as protease inhibitors, hormones, ion-channel toxins, antimicrobials, antitumor agents, and insecticides. Owing to their high stability, chemical accessibility, and sequence variability, knottins have recently been used as structural scaffolds for drug design studies and as templates for the synthesis of combinatorial libraries.

One knottin--the drug Prialt (ziconotide), a synthetic form of the cone snail peptide v-conotoxin MVIIA--has just passed Phase III clinical trials. Elan Corp., Dublin, is hoping to bring the drug to market next year for the treatment of severe chronic pain.

	CIRCLE GAME Structure of cycloviolacin O1, a typical cyclotide, shows the protein's circular backbone and its knotted arrangement of disulfide bonds (ball-and-stick components). Cyclotides have compact folds and behave like superstable miniproteins. COURTESY OF DAVID CRAIK

MANY KNOTTINS are linear, but a number of cyclic knottins are known, and in 1999 Craik and colleagues named them "cyclotides," short for cyclic peptides. The first cyclotide, Kalata B1, was discovered in the early 1970s as an active ingredient in an herbal medicine used by African women as a childbirth aid. Its structure was not determined until 1995, also by Craik's group. Approximately 50 cyclotides are currently known, but Craik says that number will soon increase considerably. For instance, "a new study from my group will report more than 60 new cyclotides from a single plant species," he says.

"These miniproteins are extracted from plants using methanol, can be boiled and still retain biological activity, and are completely impervious to enzymes," he notes. "All of this is due to the combination of a head-to-tail cyclized backbone and the cystine knot, in which two disulfide bonds form an embedded ring that is threaded by a third disulfide bond," he adds. Cyclotides fall into two subclasses: Möbius strips, in which the chains are twisted, and bracelet types, in which they're untwisted (bracelet-like).

In a recent study, Craik's group found that the enzyme trypsin can be used to cyclize a peptide--the reverse of its better known role as a catalyst of peptide bond cleavage. The 14-unit peptide that was cyclized, sunflower trypsin inhibitor-1, is the smallest circular-peptide natural product and one of the most potent trypsin inhibitors ever found. It doesn't meet the structural requirements of a cyclotide, but the researchers believe that similar enzymatic processes may be the way cyclotides and other circular peptides and proteins are biosynthesized in general.

Besides Craik and coworkers, important contributions to cyclotide research have been made by a number of other groups, including those of Lars Bohlin, head of the department of medicinal chemistry at Uppsala University, Sweden; and Kirk R. Gustafson, principal investigator in the Molecular Targets Development Program at the National Cancer Institute, Frederick, Md.

Cyclotides have been found to have a range of bioactivities, from insecticidal, antimicrobial, and anti-HIV functions to neurotensin binding and hemolytic activity, suggesting potential therapeutic and agricultural applications.

To pursue such applications, Craik founded two companies--Kalthera and Cyclagen, both in Brisbane, Australia--as spin-offs of the Institute for Molecular Bioscience at the University of Queensland. Kalthera uses cyclization technology to stabilize peptide-based pharmaceuticals, either by cyclizing the peptides directly or by grafting them onto cyclotide-like protein frameworks. The company "is currently seeking partnerships with pharma and biotech companies for specific therapeutic applications," ranging from cancer (antiangiogenesis) and multiple sclerosis to inflammation, Craik says.

And Cyclagen hopes to exploit the natural insecticidal activity of cyclotides to protect crop plants from insect attack. Cyclagen researchers are transferring cyclotide genes into cotton to provide built-in protection against Heliothis pests, which cause up to 15% of preharvest destruction of cotton crops. The company is seeking partnerships with agbiotech firms for a variety of other crop applications as well.

Which just goes to show that circular reasoning can certainly have its advantages. In fact, when it comes to cyclized peptides and proteins, "there are endless possibilities, really," Craik says. "Sorry--I couldn't resist."