Knottins provide useful scaffolds or leads for drug design

  • ‣ Knottins are exceptional in that they are very small proteins yet with particularly well-defined scaffolds and remarkably high stability.
  • ‣ Also remarkable is the fact that knottins with very similar 3D structures have virtually no sequence identity except for cysteines. This observation has led to the conclusion that the knottin scaffold is very sequence tolerant.
  • ‣ These remarkable features suggest that knottins can provide excellent lead molecules or elementary scaffold in drug design studies [Chiche et al., 2004; Craik et al., 2006; Werle et al., 2006; Moore et al., 2012], and in biotechnological applications [Cox et al., 2016; Moore et al., 2013; Glotzbach et al., 2013; Barba et al., 2012].
  • ‣ Main efforts along this way are outlined below.
You can toggle between only one (default) or all item display
*all* is currently selected

Cell internalization

Circular permutations

Computer simulations

Homology modeling



Protein engineering

Cell internalization

The knottin scaffold is considered as a useful stabilizing framework to deliver bioactive peptide drugs. For intracellular targets, it may be necessary for the knottin-based drug to enter cells.

Internalization of MCoTI-II

It has been shown that the cyclic knottin MCoTI-II is internalized into mammalian cells by macropynocytose. [Grenwood et al., 2007]. As a consequence MCoTI-II may have the potential to transport bioactivities to intracellular targets.

Internalization of PA1b

The toxin PA1b from pea is lethal for certain insects. It has been shown that PA1b is internalized into insect cells sf9 by binding to a high-affinity site. The mecanisms leading to cell death remain to be determined. Mammalian cells are insensitive to the toxin. [Rahioui et al., 2007].

Circular permutations

Linearization of the circular peptide Kalata B1

Cyclic peptides can be linearized by breaking of a peptide bond, usually in solvent accessible loops. Structure and stability of the linear peptide give informations on the importance of the loops in the folding and the stability of the circular peptide. Such linearizations are similar in spirit to the circular permutations performed on non-cyclic peptides (see below).
Six different linear analogs of the cyclic peptide Kalata B1 have been synthesized [Daly & Craik, 2000]. Four linear peptides are able to fold correctly indicating that the broken loops are not essential for folding. It is shown that the two linear peptides that do not fold correspond to peptides in which the disulfide macrocycle typical of the Knottins has been disrupted. This highlight the importance of the knot in Knottins.

Circular permutation of EETI-II

A similar approach has been applied to the non-cyclic EETI-II Knottin [Chiche, Heitz & Strub, unpublished results]. In such cases the native termini of the peptide are linked by a regular peptide bond (usually with addition of a short peptide link), and new termini are introduced somewhere else in the peptide resulting in true circular permutations. Here the C- and N-termini are connected with addition of three glycines and the 17-18 peptide bond has been cleaved. The resulting permuted peptide has been shown to fold correctly, indicating that the 16-19 β-turn is not essential for folding.


Knottins are small but structurally well-defined proteins. They are well suited for theoretical studies, e.g. molecular dynamics simulations.

Prediction and characterization of the native state of cyclic knottins

A new computational method, NcCYP, has been proposed to predict the native state conformational ensemble of cyclic disulfide-rich miniproteins from sequence. The method does not assume any specific disulfide bond pairing. A hierarchical multi-resolution exploration is used that provides a large number of independent conformations. [Shehu et al., 2008].

Molecular dynamics simulations and MM-PBSA free energies of disulfide-rich miniproteins with unconventional disulfide connectivities

Unconventional connectivities were reported for the knottin kalata B1 and for two scorpion toxins, maurotoxin and spinoxin. Molecular dynamics simulations and MM-PBSA free energy estimations suggest that these unconventional connectivities do not correspond to the lowest free energy minimum. The unconventional maurotoxin structure might arise from a kinetically controlled folding process. [Combelles et al., 2008].

Molecular dynamics simulations

A molecular dynamics simulation protocol has been devised that is able to refold the knottin PCI (Potato Carboxypeptidase Inhibitor) from unfolded species. The method was also used to predict the structure of PCI samples with non-native disulfide bridges. [Marti-Renom et al., 1998; Marti-Renom et al., 2000].

Homology modeling

Homology modeling of two cyclotides

The three-dimensional structures of two cyclotides from Viola odorata were modeled by homology using the NMR structure of kalata B1 as template. The models displayed amphipathic structures with considerable surface hydrophobicity [Svangard et al., 2003].

Homology modeling of several squash inhibitors has been reported

Several squash inhibitors were modeled from the X-ray structure of MCTI-II either free or complexed to trypsin [Chakraborty et al., 2000]. Analysis of the structures indicate that sequence variations allow fine tune of the specific biological functions without destabilizing the structural scaffold.


Knottins provide interesting structural scaffolds for constrained libraries

The use of nonimmunoglobulin scaffolds, including knottin-based scaffolds, to engineer novel binding molecules has been reviewed by Binz et al. in 2005 and by Hosse et al. in 2006.

Sequence randomization, phage-display and amplification-selection technics allow selection of optimal binders of protein targets. The cellulose-binding domain of the fungal enzyme cellobiohydrolase I has been used as a structural scaffold. Although this two-disulfide domain contains the CSB elementary motif also contained by knottins, it is not a knottin and does not display the typical "disulfide through disulfide knot". Seven residues, located on the surface of the domain, were varied by random mutation of the gene. The repertoire was cloned for display on filamentous bacteriophage, and selected for binding to cellulose or alkaline phosphatase. Binders of cellulose and alkaline phosphatase were isolated [Smith et al., 1998]. More recently, the two-disulfide Cystine Stablized Beta-sheet motif from a squash inhibitor (The 'Min-23' peptide) has been used succesfully to select 21 new specific binders on 7 different targets. Min-23 is one of the smallest known structural scaffold for efficient phage display and selection [Souriau et al., 2005].

The mRNA display is another approach to identify peptide sequences involved in macromolecular recognition. A constrained peptide library has been built using this technic based on the Knottin EETI-II, where the six residues in the first loop were randomized. The constrained library was screened against the natural target of wild-type EETI-II. Selected sequences revealed minimal consensus sequences of PR(I,L,V)L for the inhibitory loop of EETI-II and the wild-type sequence, PRILMR, was selected with the highest frequency [Baggio et al., 2002].

Point mutations

A large number of point mutations have been performed on squash trypsin inhibitors

The trypsin selectivity of EETI-II was changed to elastase and α-chymotrypsin by modification of the P1 residue [Favel et al., 1989; Le-Nguyen et al., 1990]. Similar results were obtained for another squash inhibitor, CMTI-III [McWherter et al., 1989].
Numerous mutants of CMTI-I and CMTI-III were reported, mainly by polish groups.
CMTI-III mutants of the Arg5 P1 residue exhibit antitrypsin, antichymotrypsin or antielastase activity [Rolka et al., 1989; Rolka et al., 1991, Rozycki 1994a].
Double mutants with Tyr27-Val mutation and Glycine in position 9, 11, 14, 17, 19 or 29 reduce association equilibrium constants by 6-7 orders of magnitude [Rozycki et al., 1993].
Several mutations outside the contact region did not change the antitrypsin activity [Rozycki et al., 1994b].
while other mutations outside the binding loop did affect binding to trypsin [Jaskiewicz et al., 1998].
Mutants of CMTI-I were recently screened for inhibition of serine protease from human blood clotting system [Grzesiak et al., 2000].
Mutation one to four of four conserved hydrophobic residues in CMTI-I were replaced by alanine. Single replacements led to 5-40 times less effective trypsin inhibitors whereas the quadruple mutant was approximately 450 times less effective [Kojima et al., 1996].
The conservative Met8-Leu mutation in CMTI-I was shown to affect both stability and folding of the protein [Zhukov et al., 2000].
The impact of mutation Trp7Leu and addition of Asn25 in MCTI-II, on inhibition of factor Xa and on factor X activation, has been studied [Kamei et al., 2000].

The Potato Carboxypeptidase Inhibitor

Residues of the C-terminal tail were screened by directed mutagenesis for their role in the interaction with carboxypeptidase [Marino-Buslje et al., 2000; Molina et al., 1994].

The cyclotide kalata B1 is tolerant to point mutations

Structural studies of [W19K/P20N/V21K] kalata B1 and [P20D/V21K] kalata B1 showed that the cyclotide framework is tolerant to residue substitutions [Clark et al., 2006].

Protein engineering

Knottin peptide-drug conjugate

Knottin peptide-drug conjugate (KDC) are a promising tools to selectively deliver drug to a target in vivo [Currier et al., 2016; Cox et al., 2016; Kim et al., 2015].

An engineered KDC to gemcitabine is able to binds to tumor cells with high affinity using integrin, penetrate cell, delivers drugs intracellularly, and is a highly potent inhibitor of various malignant cells [Cox et al., 2016; Kim et al., 2015].

Optical imaging of tumor using engineered knottin

Tumor cell can be targeted and illuminated with a fluorescent, engineered knottin, that binds with high affinity to integrin [Moore et al., 2013a; Moore et al., 2013b; Zhu et al., 2014].

Engineering and optimization of high-affinity inhibitors of protease

Yeast surface-displayed libraries of knottin were produced and screened for a matriptase-1 inhibitors activity. A subnanomolar inhibitor was discovered [Glotzbach et al., 2013].

Redox-Active Metallopeptides based on knottin scaffold

Cupricyclin-1 and -2, two novel enginnered knottin able to bind one Cu2+ ion with nanomolar affinity have been designed. These copper-based biocatalysts display redox activity and catalyze the dismutation of superoxide anions [Barba et al., 2012].

Knottins with new molecular recognition characteristics

An interesting review on using knottin for binding specific molecular targets [Moore et al., 2012].

Cyclooctapeptides as constrained analogues of the neurotoxin omega-agatoxine IVB

Constrained analogs of the binding loop of omega-agatoxine IVB were obtained via ring cyclization of octapeptides by side chain-side chain lactonization [Minta et al., 2008].

Polytides (Polymer-Peptide hybrids) as a tool to improve bioavailability of small disulfide-rich proteins

Replacement of nonessential amino acid residues with isostere polymeric spacers in conotoxin SIIIA resulted in more potent and longer-lasting analgesics when compared to SIIIA. It is proposed that polytides provide a promising strategy for transforming disulfide-rich peptides into therapeutics [Green et al., 2007].

Grafting active peptides into EETI-II and the minimized Agouti-related protein

Thrombopoietin is the primary regulator of platelet production. Peptides shown to act as high-affinity thrombopoietin antagonists were grafted into the active loops of EETI-II (loop 'a') and of the minimized Agouti-related protein AGRP(87-132) , loop 'e'. Most interestingly, covalent dimerization of the engineered miniproteins through lysine cross-linking yielded potent bivalent agonists of the thrombopoietin receptor (c-Mpl) [Krause et al., 2007].
The RGD sequence (IPRGDYR) was also grafted into loop 'a' of EETI-II and both RGD and KGD sequences (RVAKGDWNDDT) were grafted into loop 'e' of AGRP(87-132) [Reiss et al., 2006]. The RGD-grafted EETI-II and AGRP were more potent when compared to the unsupported peptides. In contrast the KGD-grafted AGRP resulted in very poor platelet inhibition.

Structure-activity studies on the Agouti-Related Protein (AGRP)

The Agouti-Related Protein (AGRP) is an endogeneous antagonist of melanocortin receptors. It has been shown that the C-terminal knottin domain of AGRP, AGRP(87-120) known as the mini-AGRP, is sufficient to maintain biological activity [Jackson et al., 2002]. Interestingly, the shorter AGRP(91-122) containing only the two-disulfide Cystine-Stabilized β-sheet motif of the knottin fold, also displayed antagonist activity although with reduced potency [Wilczynski et al., 2005]. Moreover, chimeric peptides based on the mini-AGRP template, and obtained by grafting agonist peptides resulted in potent agonists of melanocorin receptors 1 and 3-5, however receptor selectivity was affected [Jackson et al., 2005].

Replacement of the inhibitory loop of EETI-II with a different inhibitory peptide

A seven-residue peptide, derived from the inhibitory loop of the the third domain of ovomucoid inhibitor from turkey, has been grafted onto EETI-II. The resulting peptide has been shown to inhibit porcine pancreatic elastase [Hilpert et al., 2002].

Role of the β-hairpin in omega-Atracotoxin-Hv1a

Omega-Atracotoxin-Hv1a is an insect-specific neurotoxin. Removal of the C-terminal hairpin has been shown to suppress the insecticidal activity, although the knottin fold remains essentially unaffected (compare 1AXH and 1HVW). Alanine scanning on the hairpin then indicated that spatially close residues Asn(27) and Arg(35) are essential for the activity [Tedford et al., 2001].

Replacement of the GPNG β-turn of Min-23 with the RT-loop of SH3 domains

The Nef protein of HIV binds to SH3 domains of Src protein kinases. As a first step toward the development of new anti-HIV drugs, chimeric peptides were built by grafting the RT-loop of Src protein kinases onto the Min-23 peptide [Heitz et al., 2000].

Chimeric analogs of omega-conotoxin MVIIA and MVIIC

MVIIA and MVIIC selectively block N- and P/Q-type calcium channels, respectively. Chimeric analogs of MVIIA and MVIIC in which the N- and C-terminal halves were exchanged showed that residues important for the recognition of N-type channels are located in the N-terminal half, especially position 11, whereas essential residues for P/Q-type channel recognition are spread over the whole molecule. [Sato et al., 1997; Sato et al., 2000].

Circularization of conotoxins

MVIIA conotoxin has revealed interesting therapeutic potential. Chimeric molecules combining the essential residues of the MVIIA with the stable circular scaffold of Kalata B1 have been synthesized and could improve the bioavailability of the compound for drug usage [Craik et al., 2000].

Possible use of AAI as insecticid

The Knottin AAI (α-amylase inhibitor from Amaranth) is considered as a very promising head for development of insecticids against insects like Tenebrio molitor, which are responsible for huge harvest losses [Pereira et al., 1999].

Replacement of the inhibitory loop of EETI-II with two different epitopes

Replacing the inhibitory loop of EETI-II by a 13-residue epitope of Sendai virus L-protein or by a 17-residue epitope from human bone Gla-protein, resulted in correctly folded variants. Cells expressing EETI-II variants containing an epitope were shown to be surface labeled with the respective monoclonal antibody by indirect immunofluorescence corroborating the cell surface exposure of the epitope sequences embedded in the EETI-II cystine knot scaffold [Christmann et al., 1999]

The chimeric Trypsin Carboxypeptidase Inhibitor

Early studies showed that grafting the C-terminal active site of the Knottin PCI (Potato Carboxypeptidase A Inhibitor) at the C-terminus of another Knottin EETI-II (Ecballium Elaterium Trypsin Inhibitor) does not affect the Knottin fold of EETI-II. The resulting chimeric molecule is able to inhibit carboxypeptidase A and trypsin either separately or simultaneously [Le-Nguyen et al, 1990; Chiche et al., 1993].