Knottins and Cyclic trypsin inhibitors from Cucurbitaceae next up previous contents
Next: Cyclic Bacteriocins Up: Other Cyclic Peptides Previous: Other Cyclic Peptides

Knottins and Cyclic trypsin inhibitors from Cucurbitaceae

As discussed above small linear, cystine-rich proteins that possess the typical cystine knot topology of I-IV, II-V and III-VI have been categorised as knottins. This structural class is presently thought to contain at least 12 different protein families, with virtually no sequence identity, including conotoxins from cone snails, spider toxins, potato inhibitors and the antifungal peptide from Phytolacca americana (PAFP-S). The widespread occurrence of the knottin fold within nature suggests that there was either an ancestral knottin or that the fold has evolved independently in many different lineages. The squash trypsin inhibitors (TI) are a class of knottin that were discovered in the late 1970's in the seeds of the winter squash [114]. Before the discovery of SFTI-1 they were the smallest known inhibitors of serine proteinases and also one of the most potent with association constants as high as $ 10^{-12}$M$ ^{-1}$ [115]. Almost 30 homologs have since been identified, exclusively from plants of the Cucurbitaceae family, and all these peptides contain six Cys residues arranged into three disulfides connected in the typical cystine knot arrangement [114,117,115,116].

Until recently the cyclotides were thought to be the only example of the knottin fold that included macrocyclisation of the peptide backbone, however this changed with the discovery of the 34 residue trypsin inhibitors, MCoTI-I and II, from the dormant seeds of Momordica cochinchinensis [30]. These TIs combine the typical knottin fold and disulphide connectivity with backbone cyclisation. Although sequence homology with the cyclotides is low and the loop lengths differ, the cyclic proteins from Momordica cochinchinensis have been classed as cyclotides on the basis of the conserved CCK motif [7]. Intriguingly, unlike the cyclotides, the seeds of Momordica cochinchinensis contain not only the macrocyclic MCoTI-I and II but also a range of other species including a linear version of the TI [30].

Figure 1.16: Alignment of select knottins. An alignment of various squash TIs and kalata B1. Cys residues are coloured red, and the conserved C-terminal Gly, thought to be the C-terminal processing point of the squash TIs, is coloured green, the disulphide connectivity is indicated at the bottom of the alignment. Sequence belonging to the precursor protein in TGT-II has been italicised. Also shown is the PAFP-S peptide that highlights the difference in Cys residue positioning between the cyclotides and the squash TIs on the one hand and the rest of the known knottins. The proximity of CysIV to CysV in the cyclotides and squash TIs is thought to allow the closer positioning of the N and C termini prior to cyclisation.

It can be seen in Figure 1.16 that the sequence of both MCoTI-I and II are similar to other linear squash TI's apart from the addition of a short linker, consisting predominantly of Ser and Gly residues, that completes the cyclic backbone of the peptides. The linear precursor of these peptides is unknown, however the presence of a conserved Gly after the final Cys residue, common to other squash TIs, would seem to indicate that this may be the C-terminal processing point. Additionally, the MCoTI-II linker region (SGSDGGV) shares strong sequence identity with the pro-sequence of the linear precursor of the squash TI TGT-II (SGRHGGI) [118], suggesting that the cyclic variant is produced via a typical squash TI precursor protein. It is interesting to note that the two residues at either end of this putative linear precursor of MCoTI-II are the same residues that have been shown to be critical for the cyclising reaction that produces the cyclic pilin.

Other species of the squash TIs have been isolated from Momordica cochinchinesis including species of MCoTI-II possessing a $ \beta$-Asp-Gly bond in the linker region and a succinimide cyclic intermediate of the same bond formed during the conversion of the $ \alpha$-Asp-Gly to the $ \beta$ form. Similarly a species of MCoTI-I has been identified that would appear to also possess a $ \beta$-Asp-Gly in the same position [30,7]. MCoTI-III was also isolated and it appears to be a regular linear member of the squash TI family, with the exception of a N-terminal pyroglutamate residue [30]. The presence of a linear species is of great interest. It has been pointed out that processing of the squash TIs appears to be variable [117]. It can be seen in Figure 1.16 that examples from the same species often display almost complete sequence identity apart from the addition of a N-terminal segment that, significantly, possesses multiple Glu residues. Whether the existence of two species of TI is a result of two distinct genes or the result of post-translational modifications is not known, although in the peptides from Cucurbita maxima and Momordica charantis there are enough substitutions, apart from the N-terminal, to suggest separate genes. It is interesting, therefore, to note that MCoTI-III could represent a version of the expanded peptide in Momordica cochinchinensis and the apparent presence of precursor sequence in the cyclic versions is a consequence of mutations in the N-terminal region of the precursor protein for the shorter version of the TI -- mutations resulting in cyclisation. Although this is an interesting parallel, resolution of these questions relies on the determination of a range of squash TI precursor sequences, particularly the linear precursors of the cyclic peptides from Momordica cochinchinensis, and these data are eagerly awaited.

Figure 1.17: Structure of some squash trypsin inhibitors Structures of A. EETI-II (PDB code 2ETI) from Ecballium elaterium, B. CMTI-I (PDB code 1CTI) from Cucurbita maxima, C. cyclic MCoTI-II (PDB code 1HA9) from Momordica cochinchinensis and for comparison a ribbon diagram of kalata B1 (PDB code 1NB1). In each depiction the Cys residues have been denoted.

The 3D structure of MCoTI-II has been determined by NMR spectroscopy [7,119] and it displays a similar fold to the linear versions of the squash TIs, EETI-II [121,120] and CMTI-I [123,122] (see Figure 1.17). The structure of MCoTI-II is characterised by the presence of a $ \beta$-hairpin and the molecule contains several turns. Compared to the cystine knot of the cyclotides the squash TIs exhibit a looser ring, with eleven backbone residues making up the ring through which the third disulphide threads. Nonetheless the cyclotides and MCoTI-II both contain the identical structural elements of a cystine knot and a triple stranded $ \beta$-sheet. Although the overall structure of MCoTI-II is well defined the linker region, unlike the cyclotides, shows considerable disorder. One possible advantage of cyclisation is reduced conformational flexibility leading to increased binding efficiency, however this does not appear to be the case in MCoTI-II as the linker region does not seem to impart conformational rigidity. It has been suggested that cyclisation has evolved in Momordica cochinchinensis to impart resistance to proteolytic activity or to increase stability [7]. Although the latter is feasible, it should be noted that linear squash inhibitors possess a very stable framework, with the melting temperature of EETI-I reported as 140$ ^\circ$C [72] and consequently the most likely advantage that would be conferred would be resistance to exo-peptidases. Given the existence of a linear homologue in Momordica cochinchinensis and if other Cucurbitaceae species do not contain cyclic TIs then it is also quite possible that cyclisation in Momordica is a relatively recent event, possibly brought about via serendipitous mutations in the precursor sequence.

Although the possibility exists that cyclic squash TIs are present in other Cucurbitaceae species but have been overlooked due to sequencing failure brought about by backbone cyclisation, the existence of cyclic peptides in one isolated species follows the trend discussed elsewhere within this chapter. If the Momordica peptides are the result of a fortuitous mutation than it may be illustrating a more general trend in the evolution of cyclic proteins. It is possible that cyclisation relies only on the close proximity of the N and C-termini and the presence of key residues in the precursor sequence -- cyclisation then proceeds using regular proteases or other processing enzymes typical to the biosynthetic pathway of the particular species. Interestingly it has been pointed out that the cysteine topology of the cyclotides and the squash TIs is conducive to the positioning of the N and C-termini in close proximity, as these two classes are the only knottins to have CysIV closer to CysV than to CysIII (see Figure 1.16), hence increasing the probability of cyclisation.

next up previous contents
Next: Cyclic Bacteriocins Up: Other Cyclic Peptides Previous: Other Cyclic Peptides
Jason Mulvenna