Cyclotides [1] are a plant-derived family of small proteins characterized by their head-to-tail cyclic backbone and a cystine knot arrangement of three conserved disulfide bonds. They were first discovered in plants from the Rubiaceae (coffee) and Violaceae (violet) families but have since been reported in a range of other plants from the Cucurbitaceae (cucurbit) and Fabaceae (legume) families and it has been predicted that they are widely distributed within the plant kingdom [2]. Cyclotides are notable for their exceptional stability and their diverse range of bioactivities, as well as for their expression in a wide range of plant tissues, including leaves, stems, flowers, dormant seeds and roots [3]. There have been a number of recent reviews on the discovery [4,5,6], structures [4,7,8], and applications [9,10,11,12] of cyclotides, to which the reader is referred for more background but here the focus is on their pesticidal and/or toxic activities.

The structure of the prototypical cyclotide, kalata B1, is shown in Figure 1 [13], highlighting that cyclotides are around 30 amino acids in size with a cyclic peptide backbone. They fold into compact three-dimensional shapes that incorporate a small β-sheet structure and a series of turns built around the cystine knot core. This compact structure and elaborate cross-bracing by disulfide bonds is a main contributing factor that makes cyclotides exceptionally stable. They are highly resistant to chemical, thermal and enzymatic treatments that would typically destroy conventional proteins [14].

Individual plants express a suite of cyclotides, sometimes comprising more than 100 different sequences. Based on the reported discovery rates for cyclotides in the Violaceae and Rubiaceae and the large number of unique cyclotides per plant species it has been predicted that there might be ~50,000 cyclotide sequences in nature [2] but this is probably an underestimate based on recent discoveries in the Fabaceae [19,20]. Table 1 shows a representative selection of cyclotides from the three major subfamilies that have so far been identified. The Möbius and bracelet subfamilies are the largest subfamilies of cyclotides and are so named because of the presence or absence of a conceptual 180° twist in the peptide backbone brought about by a cis X-Pro peptide bond in the Möbius class and the lack of this “twist” in the circular ribbon of the bracelet class [1]. A smaller subfamily of cyclotides known as the trypsin inhibitor cyclotides has quite different sequences than the other two subfamilies and is also routinely classified as “cyclic knottins” based on their sequence homology to conventional (acyclic) cystine knot proteins from the knottin family [21]. The cyclotides shown in Table 1 have been structurally characterized and their relevant PDB ID codes are given in the table, along with their sequences and some biophysical characteristics, including net charge and size (numbers of amino acids).

Cyclotides were originally discovered based on their observed bioactivities. The prototypical member of the family, kalata B1, was identified as the active uterotonic agent in a medicinal tea used by women in Africa to accelerate child birth. The plant used to make the tea is Oldenlandia affinis, a member of the Rubiaceae family of plants [41]. The first cyclotide discovered from the violet family, violapeptide 1, was fortuitously discovered while looking for haemolytic activity of saponins [42]. In one of the most significant early reports, a series of circular peptides was discovered in a screening program searching for anti-HIV molecules [29,43,44]. Similarly, cyclopsychotride A [45] was discovered during a screen for neurotensin antagonists. These initial bioassay-guided discoveries led to systematic searches for cyclotides based on their structural properties and this eventually led to the discovery of a wide range of other activities for cyclotides, including antimicrobial [46], insecticidal [22,47] and other pesticidal activities. I will describe these biological activities in more detail in Section 3, after making some brief comments on the biosynthesis and artificial synthesis of cyclotides.

Cyclotides are biosynthesized ribosomally as precursor proteins that encode one or more cyclotide domains [48,49]. Figure 2 shows the generic arrangement of a typical cyclotide precursor protein, which comprises an endoplasmic reticulum (ER) signal sequence, a pro-domain, a mature cyclotide domain and a C-terminal region. Although the excision and cyclization processes that yield cyclic mature peptides from these precursors are not fully understood, it is believed that asparaginyl endoproteinase (AEP) enzyme activity plays an important role [50,51]. This hypothesis is consistent with the presence of an absolutely conserved Asn (or Asp) residue that at the C-terminus of the cyclotide domain within the precursor proteins. (e.g., see cyclotide sequences in Table 1). It is further supported by studies involving expression of mutated cyclotides in transgenic plants that have identified residues important for processing. For example mutation of the conserved Asn to Ala abolishes production of cyclic peptides in planta [51].

The natural function of cyclotides is thought to be for the protection of plants against pests, a suggestion prompted by the discovery that cyclotides potently inhibit the growth and development of certain insect larvae [47]. It is not known why an individual plant species can express so many different cyclotides. This phenomenon might have evolved to counter the development of resistance of pests to an individual defense molecule, or it might reflect that fact that cyclotides have an array of functions in plants that have so far not been identified. For example, some host defense peptides in animals have direct antimicrobial activity but also function as signaling molecules and immune modulators. Similar multifunctional roles might be possible for cyclotides but have so far not been reported. In this article the focus is on toxic activities of cyclotides. Table 2 summarizes examples of pesticidal and/or toxic activities of cyclotides. They include hemolytic, cytotoxic, antimicrobial, insecticidal, antifouling, molluscicidal and nematocidal activities.

Although the range of biological activities of cyclotides might seem diverse, in fact they all appear to be accounted for by a common mechanism involving interactions of cyclotides with membranes and their subsequent disruption. In an early biophysical study NMR chemical shifts and diffusion measurements were used to demonstrate a specific interaction between a model membrane in the form of dodecylphosphocholine (DPC) micelles and the cyclotide kalata B1 [78]. This was followed by similar studies on B7 [79], which confirmed that a hydrophobic patch on the surface of the cyclotides (shown in Figure 1c) was the primary interaction face with membranes. The orientation of kalata B1 binding to the DPC surface is shown in Figure 3a. The precise orientation varies with different cyclotides, dependent on the location of the surface-exposed hydrophobic patch [80].

Independent studies showed that the membrane binding was functional, in that cyclotides were able to induce leakage of contents from phospholipid vesicles (Figure 3b) and form large pores in lipid bilayers [81], and that membrane binding modulated cytotoxicity [69]. Surface plasmon resonance studies [82,83] also established a specific interaction between cyclotides and phospholipid bilayers, as shown in Figure 3c [83]. These various studies have provided details on the selectivity of particular cyclotides for particular lipid subtypes. Overall, cyclotides appear to have a preference for phosphatidylethanolamine (PE) as opposed to phosphatidylcholine lipids (PC). Furthermore, for membranes of a given PE content kalata B1 has a higher affinity for membranes in a liquid disordered phase (i.e., more rigid membranes rich in cholesterol (Chol) and sphingomyelin (SM) that are raft-like domains [83]).

The ability to chemically synthesize or modify cyclotides has been important in establishing structure-activity relationships. As noted earlier, chemically modified cyclotide derivatives in which individual residues are successively replaced by alanine established that there was a patch of residues clustered on the surface of cyclotides that is important for biological activity [68]. This bioactive patch is not co-located with the hydrophobic patch implicated in membrane binding, which led to the suggestion that self-association may be involved in the mediation of cyclotide activity. Chemical modification studies have also established the importance of a key glutamic acid in regulating the activity of cyclotides. Esterification of this residue causes a marked drop in biological activity [84]. This residue is at the centre of the bioactive patch of residues on the kalata B1 cyclotide surface. The Glu residue is conserved in all but one of the currently known naturally occurring cyclotides, with the exception being kalata B12, where it is replaced with the conservative Asp residue. However, the precise role of the Glu in determining cyclotide activity is yet to be delineated.

In another example of the value of chemistry in determining structure-activity relationships of cyclotides a recent study used a chemically synthesized all-D kalata molecule to confirm the membrane binding hypothesis [83]. The insecticidal, anti-HIV and hemolytic activity of all-D and all-L mirror image forms of kalata B1 were similar, suggesting that a stereospecific receptor is not involved in the interaction. The slightly lower activity for the all-D peptide can be attributed to the slight chirality associated with membranes.

In summary, although cyclotides have a diverse range of activities, including antimicrobial, anti-HIV and hemolytic, and insecticidal, all these activities can be attributed to membrane binding interactions. In the case of insecticidal activity electron microscopy studies have shown clear disruption to the surface cells in the midgut of Lepidopteran larvae fed on a cyclotide-containing diet [75].

To commercially exploit cyclotides as pesticidal agents it will be necessary to develop cost-effective approaches for delivering them to target plants. Synthetic methods to produce cyclotides have been developed over the last decade, and include solid phase peptide synthesis approaches [46,85,86,87,88] as well as chemo-enzymatic [64,65,89,90] and biological approaches involving modified inteins [91,92,93]. So far, most of the proof-of-concept work on the structure-activity relationships of cyclotides has been done using chemically synthesized cyclotides analogues. Solid phase chemistry is the preferred method here and typically the ligation of the N and the C terminus is achieved using an adaptation of “native chemical ligation” chemistry [94,95]. This requires that the peptide chain be assembled linked to the resin with a thioester and the linear precursor sequence assembled in such an order that the N-terminal residue is a cysteine. This allows a trans-thioesterification reaction to achieve the cyclization, as highlighted in Figure 4.

This technology works well for Möbius trypsin inhibitor subfamily cyclotides although in the past the folding of the bracelet cyclotides has been problematic [96]. Several approaches have recently been developed to overcome this limitation and further improvement seems likely in the future [97,98].

Another approach to making cyclotides is to assemble them as conventional linear peptides and then achieve the cyclization enzymatically. This chemo-enzymatic approach can be done by engineering, for example, a trypsin active site into the linear cyclotide precursor and using trypsin to achieve the ligation. Proof-of-concept for this approach was established early for the cyclic peptide SFTI-I [99] and then adapted to the trypsin inhibitor cyclotide MCoTI-II [64,65,89,90]. One advantage of this approach is that the enzyme can be immobilized on a column and the cyclization achieved while the substrate flows through the column, thus leading to easy purification.

Biological approaches to cyclotide synthesis have also been developed using E. coli as the host cells and modified inteins for activation of linear cyclotide precursors. This approach has been demonstrated for a range of cyclotides, including kalata B1 and MCoTI-II and has been used to produce small libraries of cyclotides and other cyclic peptides [11,91,92,93,100,101]. Although at this stage the yields are still relatively low compared to solid phase peptide synthesis approaches the biological approach has the potential to be less expensive.

In another biological approach, significant progress has been made in using plant cell culture methods for the production of native cyclotides. Specifically, Dörnenburg et al. [102,103,104,105] have developed a plant cell fermentation system and demonstrated excellent yields of native cyclotides in cultured O. affinis cells. Interestingly, the cultured cells not only produced cyclotides that had been identified in whole plant extracts of O. affinis but also produced some novel cyclotides not previously seen.

Despite the promise seen with SPPS or whole-cell-based approaches to cyclotide production, it seems likely that for large scale pesticidal applications of cyclotides a transgenic approach in which cyclotide genes are inserted directly into target plants is likely to be the most cost effective method. The main advantage of this approach is that the cyclotides are expressed only in the target organism thus potentially minimizing environmental impacts seen with traditional pesticidal applications through spraying. Proof-of-concept that cyclotides can be produced in model plants (tobacco and Arabidopsis) has been obtained in studies aimed at understanding the mechanistic basis of cyclotide processing [50,51].

The potential applications of cyclotides as pesticidal and/or pharmaceutical agents, merit safely evaluations of their effects in the environment. In a recent study [57] it was found that cyclotides were toxic to a range of test organisms, with EC₅₀ values ranging from 12 to 140 µM against algae, 9–40 µM against duckweed, 4–50 µM against lettuce and 7–26 µM against soil bacteria. It was concluded that cyclotides might adversely affect solid and aquatic environments and that this should be taken into account in future risk assessments for cropping systems producing cyclotides. However, it also needs to be noted that cyclotides appear to be widely present naturally in a wide range of plants from the Rubiaceae, Violaceae, Cucurbitaceae and Fabaceae families and there are so far no reports of toxic effects in the environments associated with these naturally occurring cyclotides.

There have only been two studies that have reported chemical breakdown products of cyclotides [31,84] and it seems that more work in understanding cyclotide turnover is warranted to assess how stable these molecules are in their natural environment.

Although cyclotides were discovered only 15 years ago, they have developed into an exciting field of research and display a wide range of biological activities. At this stage it would appear that these activities are all associated with membrane interactions and given a demonstrated preference for cyclotides for some phospholipids over others it seems likely that cyclotides are targeted to specific cell types based on membrane composition to exert their biological activity. The rate of cyclotides discovery is rapidly evolving and the prediction of at around 50,000 members of the family may turn out to be an underestimate. Significant progress has been made in the chemistry for synthesizing cyclotides as well as in their biological synthesis. Thus there are now available an array of tools for the chemical production and modification of cyclotides, allowing further refinement of structure-activity relationships. Significant progress has also been made in the production of transgenic plants expressing cyclotides and so there appears to be great potential for the use of cyclotides as pesticidal agents, and indeed for modified cyclotides as medicines.