Next: Evolution and Distribution Up: SFTI Previous: Sequence and Structure
Unlike other cyclic proteins SFTI is a reduced version of a class of much larger protein. For this reason the biosynthetic pathway that produces mature cyclic SFTI is of great interest, as it potentially involves post-translational modifications to a much larger protein that is functional in other plant species. However nothing is known about a possible SFTI precursor or its biosynthetic pathway. In a recent review Korsincsky et al.  speculated that there are at least three possible routes to mature SFTI-1 -- non-ribosomal synthesis, processing of a larger precursor protein, or the expression of a short linear peptide that is subsequently cyclised. Non-ribosomal synthesis, analogous to that utilised by bacteria in the synthesis of gramicidin  or cyclosporin , would involve the production of SFTI on large multi-enzyme complexes. Such complexes have been described in bacteria and fungi [103,104] but have not been reported in sunflower. Indeed, such complexes have not been reported in any eukaryotic organisms and, accordingly, the evolution of such complex biosynthetic machinery in one genera of plant seems remote.
As discussed above, the possibility exists that SFTI is derived from a larger gene-expressed, BBI-like precursor protein. Such a biosynthetic pathway would presumably involve the excision of the trypsin reactive loop via a peptide bond cleavage followed by a transpeptidation reaction or cleavage of two peptide bonds and subsequent ligation of the resulting N and C-termini. Excision of the reactive loop would presumably proceed from a point distal to the disulphide bridge, in which case, as illustrated in Figure 1.15, any of the peptide bonds in the secondary loop may be the bond that is ligated to form the cyclic peptide. It is interesting to note, however, the presence of a conserved Asp/Asn in the BBIs (Figure 1.11). An Asn/Asp residue is conserved at the ligation point of the cyclotide precursor, and as discussed above legumains, which cleave after Asn or Asp, are suspected to be one of the processing enzymes in the biosynthesis of the cyclotides . These enzymes are involved in the processing of many proteins, especially storage proteins that are destined for the seed. It is quite possible that they may perform a similar role in excising SFTI-1 from a larger BBI-like protein at Asp14. In this case Gly1 would then become the N-terminal residue of the linear form -- a situation that also occurs in the cyclotides. The report  of an iso-Asp in a synthetic, linear isomer of SFTI-1, in which linearisation occurred at the scissile bond between Lys5 and Ser6 (denoted as SFTI-1[6,5]), adds weight to the possibility of cyclisation between Asp14 and Gly1 by highlighting the potential for increased reactivity in this site.
It is also possible that SFTI-1 is derived from a larger precursor that shares little similarity with other BBI proteins. One mechanism whereby this could occur is by a domain swap placing the reactive loop of a BBI into a novel precursor protein. Such a process has been postulated in the evolution of animal ICKs in which a precursor peptide exon was shuffled to an ancestral non-secreting cystine knot, resulting in a secreting cystine knot peptide . Such events, however, are not common  and the extent to which exon shuffling plays a role in evolution is highly debated [107,108]. Nonetheless, the repetition of reactive domains in monocot and dicot BBI's show that gene duplication is an active force in BBI evolution [109,78] and displacement of the reactive loop may have occurred through serendipity or by an, as yet, unidentified mechanism.
Although all cyclic peptides so far characterised are derived from a larger precursor  there is a further possibility that SFTI is expressed as a short linear peptide that is subsequently cyclised. In this case cyclisation could occur at any of the bonds around the molecule. However, as pointed out by Korsincsky et al.  two positions appear more likely than the others -- the scissile bond between Lys5 and Ser6 and, once again, the bond between Gly1 and Asp14. In the same study discussed above, the linear isomer SFTI-1[6,5] was found to undergo cyclisation after incubation with trypsin . Consequently, this isomer inhibited trypsin with the same K as cyclic SFTI-1, although complex formation with trypsin was slower, consistent with the reformation of the scissile bond. Other studies have shown the reversibility of protease catalysed reactions and the reformation of scissile bonds [111,110] and this represents a possible mechanism whereby expression of a short linear peptide could lead to the cyclic form of SFTI-1 through a plant protease with similar activity to trypsin. This result also explained the observations that approximately 10% of the SFTI-1 in complex with trypsin may have been cleaved in the crystal structure  and that a small percentage of SFTI-1 appears to be hydrolysed in seeds and that one of these hydrolysed species is SFTI-1[6,5] [112,81].
Given the general reversibility of protease reactions it is also possible that any bond within the sequence of SFTI-1 that is capable of being hydrolysed by an enzyme is the cyclisation point. The rigid structure of SFTI-1 is believed to be the major cause of its resistance to hydrolysis by trypsin [8,9] and it is possible that this same effect could play a role in allowing the reversal of a proteolytic enzyme reaction at other positions in the sequence. Once again a legumain could be involved in cyclising a short linear peptide at Gly1 and Asp14. As discussed above, these enzymes are known to catalyse the formation of a peptide bond in concavalin A  and a similar mechanism could be at work here. The presence of an iso-Asp in SFTI-1[6,5]  could also indicate a self-catalysed cyclising mechanism may occur at Gly1 and Asp14, possibly via a succinimide or cyclic anhydride, to yield a cyclic peptide, thus obviating the need for a processing enzyme.
Next: Evolution and Distribution Up: SFTI Previous: Sequence and Structure Jason Mulvenna