Sequence and Structure next up previous contents
Next: Biosynthesis Up: SFTI Previous: Bowman-Birk Inhibitors and the

Sequence and Structure

SFTI-1 contains 14 amino acids, two of which are Cys residues that are joined in a disulfide bond. As shown in Figure 1.11, this disulfide bond separates SFTI-1 into two loops corresponding to a reactive loop, which possesses strong sequence identity to the BBI active site, and a secondary loop which completes the cyclic backbone of the molecule. An inspection of the reactive loop sequence of SFTI-1 shows that it clearly belongs to the BBI family of trypsin inhibitors (Figure 1.11). About forty different members of the BBI family have been identified and a comparison of the trypsin reactive loops from a range of BBIs to SFTI-1 shows similarity in both inter-cystine length and sequence.

Figure 1.11: SFTI-1 and soybean BBI structures A shows a ribbon depiction of the structure of a BBI from Glycine max (pdb code 1BBI) with the chymotrypsin and trypsin reactive loops highlighted. The detailed structure of the trypsin inhibitory loop is shown below the ribbon diagram alongside the solution structure of SFTI-1 (pdb code 1JBL). In both cases the Cys residues are coloured yellow and the lysine residue in the P1 position is coloured green with the sidechain depicted in stick form. B shows an alignment of SFTI-1 with the trypsin reactive loop of a number of BBIs. The other sequences correspond to BBI's from Mung Bean (1SMF), the Adzuki bean (1TAB) and the soybean (1BBI). The Cys residues that form the disulphide bond are coloured red, the Lys residue in position P1 is coloured green and a conserved Asp/Asn residue, reminiscent of the cyclotides, is coloured blue. The active site nomenclature of Schechter and Berger [86] is displayed at the bottom of the alignment. The ribbon diagram and structural renderings were produced using MOLMOL.

With a K$ _i$ of 0.1-0.5 nM [8], SFTI-1 is the most potent trypsin inhibitor yet discovered and studies utilising minimised peptides mimicking the reactive loop of BBIs suggest that SFTI's remarkable efficiency is, at least partially, related to its reactive loop sequence. By screening a range of minimal synthetic variants of the BBI reactive loop the residues crucial for activity have been identified. Using the nomenclature of Schechter and Berger [86] in which the primary scissile bond is between P1 and P1', significant activity loss against trypsin is seen in the BBI reactive loop at positions P$ _2$, P$ _1$, P$ _2$' and P$ _3$' [87]. The optimal residue for a number of these positions has also been determined using the same methodology, and in each case SFTI-1 appears to possess this optimal residue. As the primary contact with the protease, changes in P$ _1$ are correlated with broad changes in specificity [89,90,88,85], and in this position SFTI contains a Lys which is specific for trypsin inhibition [89]. In P$ _2$ and P$ _2$' SFTI-1 contains Thr and Ile respectively. These two residues have been found to be optimal for their respective positions [92,91], although Thr in P$ _2$ was measured in chymotrypsin assays. An optimal residue for P$ _3$' has not been determined, but an alanine scan of a BBI reactive loop mimetic showed an $ \sim$2000$ \times$ reduction in activity against trypsin when this Pro was substituted [87]. SFTI-1 contains a Pro in this position and a Pro is absolutely conserved in this position across the BBI family, further highlighting its importance for activity. This remarkable convergence of optimal active site residues in SFTI-1, and its potent inhibitory effect, underscores the concept of SFTI-1 as a natural peptide mimetic of the BBI trypsin reactive loop.

Both the crystal structure [9], in complex with bovine $ \beta$-trypsin, and the solution structure [8], by $ ^1$H NMR, have been solved for SFTI-1. In both cases SFTI-1 was shown to consist of two anti-parallel $ \beta$-strands that are connected at both ends by turns -- resulting in a cyclic peptide. Remarkably, the complexed structure and the solution structure are very similar (mean RMSD for the backbone atoms of 0.25 $ \AA$) indicating that SFTI-1 must possess a highly rigid structure that does not undergo major conformational change when complexed with trypsin. Contributing to this rigidity is the disulfide bridge that separates the molecule into two loops, however, as revealed in the crystal structure, an extensive network of hydrogen bonds further stabilises the molecule [9]. As summarised in Figure 1.12, SFTI-1 possesses three intramolecular main-chain hydrogen bonds -- between Gly1 HN-Phe12 O and Phe12 HN-Arg2 O in the secondary loop and between Thr4 HN-Ile10 O in the reactive loop. The reactive loop is further stabilised by a bifurcated hydrogen bond between the hydroxyl group of Thr4 and both the main-chain amide of Ile10 and the sidechain hydroxyl of Ser6. The hydroxyl group from Ser6 is also bonded to the main-chain carbonyl group of Pro8. All of these hydrogen bonds were confirmed in the solution structure by measurement of slowly exchanging amide protons [8], however a hydrogen bond between the sidechain of Asp14 and the main-chain amide of Gly1 that was predicted by Luckett et al. [9], based on the crystal structure, could not be confirmed in the latter study. SFTI-1 contains three Pro residues and the bond between Ile7 and Pro8 has been shown to be in the cis conformation in both the crystal and solution structure [8,9]. Along with the disulfide bond and hydrogen bond network this cis-peptide bond acts to maintain the reactive site loop of SFTI in the shape of a $ \beta$-hairpin.

Figure 1.12: Hydrogen bond network in SFTI-1 The extensive hydrogen bond network that stabilises SFTI-1 is shown using the solution structure derived using $ ^1$H NMR (pdb code 1JBL). Cys bonds have been coloured yellow and the residues involved in hydrogen bonding are indicated. Actual hydrogen bonds are indicated with black lines. Oxygen atoms are coloured red and hydrogen atoms are coloured grey. The figure was produced using the program MOLMOL.

The structures of several BBI inhibitors have been solved using X-ray crystallography, both alone [94,80,93] and in complex with trypsin [96,95,97]. Additionally, the solution structure of a soyabean BBI has been solved using $ ^1$H NMR [98]. In each case the trypsin reactive loop forms a two stranded anti-parallel $ \beta$-sheet joined with a type VIb $ \beta$-hairpin. The reactive loops are stabilised by an absolutely conserved disulfide bridge and hydrogen bonds [99]. In dicot BBIs a conserved cis-Pro further stabilises the reactive loop [100]. The similarities between key elements of the structure of SFTI-1 and the structures of the reactive loops of the BBIs clearly indicate that they would share a similar fold and a superimposition of the structures of SFTI and the reactive loop of a selection of BBIs highlights this structural similarity (Figure 1.13). The conserved elements -- a disulfide bond, a cis-Pro in position P$ _3$', and intramolecular hydrogen bonding -- would therefore appear to be the three principle restraining features important for the inhibitory effect of these proteins.

Figure 1.13: Superimposition of several BBI reactive loops with SFTI-1 The structural similarity between the trypsin reactive site loops from a range of BBis and SFTI can be seen in this superimposition. SFTI-1 (pdb code 1JBL, colured blue) is superimposed with the reactive loops of BBIs from Mung Bean (pdb code 1SMF, coloured green), Adzuki Bean (pdb code 1TAB, coloured cyan), and soyabean (pdb code 1BBI, coloured red). For orientation purposes the Cys residues have been labelled on SFTI-1 and the right-hand figure is related to the left hand figure by a 180$ ^{\circ}$ rotation about the Y-axis. The figure was produced using the program MOLMOL.

As suggested by their structural similarities the interactions between SFTI and trypsin are similar to the interactions between the protease and the trypsin reactive loops of the BBIs. The crystal structure of SFTI-1 in complex with trypsin [9] shows that the interactions centre around Lys5, in position P1, which extends into the S1 pocket of the enzyme and makes both direct and water mediated contact with the specificity-determining Asp189, the hydroxyl group of Ser190 and the mainchain carbonyl of Gly219. An extensive array of hydrogen bonds and ion pairs forms between the enzyme and the inhibitor and these are summarised in Figure 1.14. These hydrogen bonds include interactions between the mainchain carbonyl of Lys5, which is within hydrogen bonding distance of Ser195 and His57 of the trypsin catalytic triad.

Figure 1.14: Summary of contacts between SFTI-1 and trypsin The hydrogen bond network formed between SFTI-1 and trypsin is summarised in A. The conformation of the three Pro residues is indicated and the reactive and secondary loop are labelled. B Shows a summary of the orientation of the contact points between the inhibitor and trypsin using the nomenclature of Schechter and Berger [86]. Figure adapted from [100].

The majority of the contact points between SFTI-1 and trypsin are made on one side of the reactive loop between the residues from Cys3 to Ile7 and the distant side of the reactive loop, between Pro8 and Cys11, appears to play an important stabilising role despite making no direct contacts with the enzyme [9]. It has been suggested that the highly stabilised reactive loop of SFTI prevents hydrolysis of the peptide by preventing the structural change that would normally accompany hydrolysis [8,9]. The importance of Thr in position P$ _2$ for inhibitory activity, at least in chymotrypsin, appears to support this conjecture, as its removal, and the subsequent loss of two hydrogen bonding interactions, could destabilise the loop leading to a loss of inhibitory activity. The secondary loop, which has no direct role in enzyme interactions, may aid binding efficiency by stabilising the entire structure through additional hydrogen bonding. This is supported by the exceptional similarity between the solution and crystal structure of SFTI-1 which suggests that a lack of binding-induced conformational change contributes to the low K$ _i$ as SFTI-1 can bind with minimal entropy costs.

next up previous contents
Next: Biosynthesis Up: SFTI Previous: Bowman-Birk Inhibitors and the
Jason Mulvenna