Just like protein-protein interaction, the structural information needed for drug targets usually comes with limited prediction accuracy for pepPIs. Following the absence of protein co-structures, most of the clinical studies and trials tend to rely on existing information, obtained from structural databases like the Protein Data Bank – PDB, in the identification of sequence-binding motifs for peptide designs.
PepX Database is another viable database with over 500 experimentally studied peptide interactions, which portray high-resolution structures, and which easily allow inputs of user-defined peptide template designs. In an analysis, such as silica mutation hotspots of protein-peptide interfaces, it was observed that 6 – 11 amino acid long peptides generally have between two and three residues, with the ability to form critical contacts with the protein being targeted. Though pepPI analysis portrays similarity to modeling protein-protein interactions, it can still be tough, because of the strange and diverse structural changes that may be as a result of backbones, and certain flexible side chains present in a peptide. It has also been observed that short peptides, with about 15 residues, have the ability to form simpler alpha-helix or beta-sheet structures. However, the structures of longer peptides, tend to be more challenging to predict, because of the rearrangements of their backbone. The actual nature of the complexity of a peptide structure prediction, also becomes more complicated when you begin to consider the flexibility of the target proteins. In this review, we will have a detailed look at some of the current computational models and methods that are in development, and which are aimed at overcoming these challenges, making it possible to get more reliable peptide designs that will overcome the challenges associated with PPIs.
How the selection of the initial scaffolds is done
Before going deep into the computational methods for PepPIs, it is vital to first start by understanding some of the recent developments that have taken place in the selection of initial peptide scaffolds, as well as the various roles they play in the peptide peptide development field. There are a good number of natural peptides that have been successfully selected from natural proteins, and which have shown very good results in preserving the original functions of peptides, including the ability to correctly and efficiently recognize the target molecules. For example, repeated Arg-Gly-Asp – RGD, motifs were initially obtained from cell attachment domains, belonging to fibronectin. These domains are known to bind themselves to membrane-bound receptor protein integrins, leading to the activation of cellular growth, and other biological processes, such as migration, adhesion, and differentiation, among others. Since the RGD peptides have the ability to mimic the functions of their natural proteins, they are currently being viewed as a promising strategy, which could be exploited for therapeutic PPI interferences, as well as analyzing the structure and functions of proteins. In addition to the already known chemical and phage peptide libraries, silica modeling-based design, is now being viewed as a powerful approach in peptide identification from various natural proteins. One other development that has generated a lot of interest, is the identification of microtubule-binding peptides. Microtubules refer to hollow tubular proteins that have intracellular alpha and beta-tubulin dimmers. These peptides have been shown to display nanodevice implications, which can be related to their many eukaryotic functions, including tumor progression.
The main targets for peptide-modulated nanodevice-encapsulating compounds, are usually the intracellular tubulins, with various formulations like the peptide-conjugating liposome assemblies, which usually try to exert synergistic anti-cancer effects. In a recent study, it was demonstrated that peptides obtained from microtubule-associated protein Tau, had the ability to functionalize the inner surface of the microtubule through the process of encapsulation of gold nanoparticles present inside the microtubules. Additionally, there is another exciting discovery involving tetrapeptide Ser-Leu-Arg-Pro – SLRP, obtained from a peptide library, showed that the microtubule functions of the peptide could be perturbed, leading to the death of cancer cells. It should be noted, however, that the selection of SLRP was done with the use of a computational docking method, known as Autodock Vina.
Docking Peptide –Protein Interactions
The successful docking of peptide-protein interaction – PepPI due to the extent of structural scaffolds present with the interaction complex. The astronomical increase in the number of peptide-protein structures currently available in the PDB, has highly encouraged the design and development of more potent docking and refinement methods, when it comes to the foretelling of accurate peptide-protein interactions. Peptide-protein docking strategies may fall into either global or local docking, and this is usually a factor of the extent of structural information available at the time of making the inputs.
Global and local docking methods
Local docking is the most commonly used docking strategy. The reason for this, is because it goes for a potential binding pose at a specific user-defined binding site, within the structure of the target receptor. There are various methods that can be used to enhance the initial model quality at atomic resolutions of the experimental peptide conformations. Some of these methods include Rosetta FlexPepDock, DynaDock, and PepCrawler are some of the most popular methods that have been used, to consistently explore various approaches for defining peptide-binding sites. With DynaDock, the softcore potential is used in conjunction with Molecular Dynamics, to get a more refined method for conformational sampling and receptor side-chain flexibility determination. With these approaches, there was the smoothing of Coulomb and Van der Waals energy potentials, and as a result of this, there was a faster conformational sampling of the peptide-protein complex, with the soft-core potential, gradually converging to physical potential, as the simulation process went on.
Rosetta FlexPepDock, on the other hand, is a Monte-Carlo-based method that is mainly used to minimize the optimization steps to give high quality conformational sampling for properly characterized binding motifs, with hot spot residues. The validation of this protocol was done against large data sets, with rigid-body sample docking, and variable degrees of backbone modeling. Finally, PepCrawler, an algorithmic robotic motion planning, known as Rapidly-exploring Random Tree- RTT, for the optimization of peptide structural poses within the binding sites. With this method, there is the building of a conformational tree for the peptide-protein complex. This leads to the generation of models, which are then automatically clustered with the help of local shape analysis of the energy funnel.
However, it should be noted that not all peptide queries will return readily available information regarding the backbone conformation. Sampling methods that make it possible to get near-native peptide conformation are now important for any local docking processes performed. Rosetta FlexPepDock is a good example of a protocol that can utilize both ab initio peptide folding, with local docking. This is made possible by placing the query peptide into a user-defined binding site. Such a binding site can be defined through the positioning of a hot spot residue or through the standard constraints for binding sites, like those made available by the protocol. In the recent past, the HADDOCK method was used in hypothesizing the unneccessity of prior backbone information in local docking when secondary structures of canonical conformation constraint were applied to the binding site. Also, there are lots of small-molecule docking methods, such as AutoDock, Surflex, and Gold that have been proven to perform local docking for short peptides, whose structures had less than five amino acids.
Global Docking Methods
Local docking is concerned with searching for the peptide binding pose. Global docking, on the other hand, also searches for the peptide-binding site at the target protein. This is the method that is predominantly preferred when there is no information currently available about the binding site. With this method, and in cases when there is no prior information about the docking site, a spatial position-specific scoring matrix – PSSM, is typically used to develop the PepSite method, which will then be ultimately used to identify any potential binding sites with an estimated position for each residue. However, the variable degrees of peptide backbone/side-chain interactions make flexible-body docking extremely difficult, and inefficient. The standard procedures for global peptide-protein docking, therefore, still have to depend on the rigid-body docking approaches after the acquisition of input peptide conformation.
It is worth noting that there are several global docking protocols with the ability to correctly predict peptide conformation, when presented with a given query sequence. For example, ATTRACT and ClusPro protocols can utilize a pre-defined motif set of template confirmation to generate various threads for query sequences. Usually, the resulting peptide conformations are the next rigid-body that has been docked in a single simulation. Other global docking methods, such as AnchorDock, CABS-Dock, and PeptiMap are also effective, when used to provide automatic docking simulations with varying algorithms like insolvent simulation, small molecule binding simulation, and flexibility of query peptide. Also, in addition to highly accurate predictions made with the help of docking methods such as PIPER-FlexPepDock, a more recent development is the adoption of HPEPDOCK, which was used with an ensemble of peptide conformations for blind global docking. The method yielded very high success rates and low simulation time when compared to methods such as pepATTRACT.cc
- Isaacs, J. S., Xu, W. & Neckers, L. Heat shock protein 90 as a molecular target for cancer therapeutics. Cancer Cell 3, 213–217 (2003).
- Li, J., Soroka, J. & Buchner, J. The Hsp90 chaperone machinery: conformational dynamics and regulation by co-chaperones. Biochim. Biophys. Acta 1823, 624–635 (2012).
- Chen, X. et al. DCZ3112, a novel Hsp90 inhibitor, exerts potent antitumor activity against HER2-positive breast cancer through disruption of Hsp90-Cdc37 interaction. Cancer Lett. 434, 70–80 (2018).
- Porter, J. R., Fritz, C. C. & Depew, K. M. Discovery and development of Hsp90 inhibitors: a promising pathway for cancer therapy. Curr. Opin. Chem. Biol. 14, 412–420 (2010).
- Neckers, L. & Workman, P. Hsp90 molecular chaperone inhibitors: are we there yet? Clin. Cancer Res. 18, 64–76 (2012).
- Taipale, M. et al. Quantitative analysis of HSP90-client interactions reveals principles of substrate recognition. Cell 150, 987–1001 (2012).
- Patel, H. J., Modi, S., Chiosis, G. & Taldone, T. Advances in the discovery and development of heat-shock protein 90 inhibitors for cancer treatment. Expert Opin. Drug Discov. 6, 559–587 (2011).
- Wandinger, S. K., Richter, K. & Buchner, J. The Hsp90 chaperone machinery. J. Biol. Chem. 283, 18473–18477 (2008).
- Hainzl, O., Lapina, M. C., Buchner, J. & Richter, K. The charged linker region is an important regulator of Hsp90 function. J. Biol. Chem. 284, 22559–22567 (2009).
- Pearl, L. H. & Prodromou, C. Structure and mechanism of the Hsp90 molecular chaperone machinery. Annu. Rev. Biochem. 75, 271–294 (2006).
- Garg, G., Khandelwal, A. & Blagg, B. S. Anticancer inhibitors of Hsp90 function: beyond the usual suspects. Adv. Cancer Res. 129, 51–88 (2016).
- Solit, D. B. & Chiosis, G. Development and application of Hsp90 inhibitors. Drug Discov. Today 13, 38–43 (2008).
- Rolland T., Tasan M., Charloteaux B., Pevzner S.J., Zhong Q., Sahni N., Yi S., Lemmens I., Fontanillo C., Mosca R., et al. A proteome-scale map of the human interactome network.
- Cunningham A.D., Qvit N., Mochly-Rosen D. Peptides and peptidomimetics as regulators of protein-protein interactions. Curr. Opin. Struct. Biol. 2017;44:59–66.
- Petta I., Lievens S., Libert C., Tavernier J., de Bosscher K. Modulation of Protein-Protein Interactions for the Development of Novel Therapeutics. Mol. Ther. 2016;24:707–718.
- Warso M.A., Richards J.M., Mehta D., Christov K., Schaeffer C., Bressler L.R., Yamada T., Majumdar D., Kennedy S.A., Beattie C.W., et al. A first-in-class, first-in-human, phase I trial of p28, a non-HDM2-mediated peptide inhibitor of p53 ubiquitination in patients with advanced solid tumours. Br. J. Cancer.
- Tabernero J., Dirix L., Schoffski P., Cervantes A., Capdevila J., Baselga J., Beijsterveldt L.V., Winkler H., Kraljevic S., Zhuang S.H. Phase I pharmacokinetic (PK) and pharmacodynamic (PD) study of HDM-2 antagonist JNJ-26854165 in patients with advanced refractory solid tumors. J. Clin. Oncol.
- Wong D., Kandagatla P., Korz W., Chinni S.R. Targeting CXCR4 with CTCE-9908 inhibits prostate tumor metastasis. BMC Urol. 2014;14:12. doi: 10.1186/1471-2490-14-12
- Huang E.H., Singh B., Cristofanilli M., Gelovani J., Wei C., Vincent L., Cook K.R., Lucci A. A CXCR4 antagonist CTCE-9908 inhibits primary tumor growth and metastasis of breast cancer. J. Surg. Res. 2009;155:231–236.
- Chiquet C., Aptel F., Creuzot-Garcher C., Berrod J.P., Kodjikian L., Massin P., Deloche C., Perino J., Kirwan B.A., de Brouwer S., et al. Postoperative Ocular Inflammation: A Single Subconjunctival Injection of XG-102 Compared to Dexamethasone Drops in a Randomized Trial. Am. J. Ophthalmol.
- Lau Y.H., de Andrade P., Wu Y., Spring D.R. Peptide stapling techniques based on different macrocyclisation chemistries. Chem. Soc. Rev. 2015;44:91–102.
- Ellert-Miklaszewska A., Poleszak K., Kaminska B. Short peptides interfering with signaling pathways as new therapeutic tools for cancer treatment. Future Med. Chem. 2017;9:199–221.