Team:CCU Taiwan/Model


Simulation of Cathepsin S preference to the sequence


To tackle the problem of MRSA latency, we designed a TAT-linked AMP (TAT-AMP) for intracellular bacterial infection. The linker between TAT and AMP can be recognized and cleaved by a macrophage-specific endopeptidase, Cathepsin S (CTSS), which releases AMP from TAT in MRSA latent lysosomes. We conduct molecular docking to predict the interaction between the selected tripeptides and CTSS and validate our design of CTSS cleavage sites on TAT-AMP.

▲ Figure 1: Cathepsin S cutting site designed according to its cleavage preference [1] and numbered the substrate amino acid by Schechter and Berger nomenclature.

The binding affinity of CTSS is primarily influenced by the residues at P2 position. We systematically changed the P2 residue from the P2-P1-P1’ tripeptide to evaluate its affinity to the CTSS active pocket using AutoDock 4.2.6.

Receptor (Cathepsin S)

The CTSS structure utilized as the receptor in the docking simulation is obtained from the demonstration of AlphaFold2 [2], which utilizes deep learning algorithms to predict the protein structure.

CTSS has a signal peptide, a propeptide, and a mature polypeptide [3] (shown in figure 2).

▲ Figure 2: Domain structure of Cathepsin S.

The domain of signal peptide and propeptide is the endogenous inhibitor to CTSS [4]. To simulate the docking in an activated CTSS, the residue 1-114 is removed. The residues involved in the cleavage are Cys139, His278, and Asn298, and an example of binding location can be indicated by a dipeptide nitrile inhibitor (shown in figure 3).

▲ Figure 3: The structure of CTSS predicted by AlphaFold 2 (Gold) and the crystal structure of CTSS (blue) with a dipeptide nitrile inhibitor (colored in purple) (PDB ID:1MS6).


We built the tripeptide structure and added an acetyl group on the N-terminal to generate the third peptide bond, and these structures were further optimized using ChemDraw built-in MM2 minimization. These tripeptides were candidates for the docking simulation.


The detailed parameters employed in AutoDock are listed in the supplementary information. These twenty tripeptide ligands have a different residue at the P2 position. We found that when the P2 residue is hydrophobic, the conformation of tripeptides is highly similar (shown in figure 4). Also, we confirmed that the cluster with the lowest binding energy occupies the targeted binding site. We believe the results are dominated by the hydrophobic interactions between the P2 residues and the binding pocket [5]. We confirmed that the scissile peptide bond is appropriately positioned near the Cys139-His278 catalytic site, through the visual inspection. The results of twenty tripeptides are listed in the supplementary information.

▲ Figure 4: The tripeptide docked to cathepsin S, which is colored according to surface hydrophobicity (red: most hydrophobic, blue: most hydrophilic).


We validate that our design of the CTSS cleavage site is reasonable based on the docking simulations.

We believed that we could further improve the design of CTSS-cleavage sequences with our procedure. We are establishing the Cathepsin S cleavage assay to confirm our simulations and degrading efficiency of TAT-AMP.

The simulation of thrombin activity


The enzyme activity can be monitored by the changes in the substrate or product amounts. We aimed to perform the thrombin cleavage assay and to examine the thrombin activity on the different substrate motifs using SDS-PAGE. Unfortunately, our progress was slowed down due to the COVID-19 pandemic, and hence, we were only able to obtain the rate constant of thrombin cleavage on a model protein, 6xHis-thrombin site-eGFP, with only one cleavage site (shown in Figure 5). We took the measured rate constant (k) and built a simulation model with two cleavage sites to predict the release rate of antimicrobial agents. This model would allow us to estimate the effective time of our product.

▲ Figure 5: The illustration of thrombin cleavage sites on model protein (6xHis-eGFP) and the proposed kinetic model.

In our sequence design, thrombin could cleave CBD-AMPs fusion protein (named “S”) at two different sites (shown in Figure 6), resulting in two or three polypeptides. These reactions can be described using Equations 1~4 (Figure 6).

▲ Figure 6: The illustration of thrombin cleavage sites on CBD-AMPs (named “S”) and the proposed kinetic models. A is the CBD, B is the linker region, and C is the AMP.


We simulated the two-site cleavage model based on Equation 5-10, in which S is the full-length peptide, A is the CBD, B is the linker region, and C is the AMP. The cleavage efficacy of thrombin to the specific sequence has been evaluated [6][7]. Based on the literature and our sequence, we assume that TAT-DPK-060 and DPK-060 have a fast cleavage rate, followed by CBD and D2A21. Hence, we assumed k1 is 0.1, 0.2 or 0.4 hr-1 (CBD cleavage), and k2 is 0.4 hr-1 for TAT-DPK-060 and DPK-060 or 0.04 hr-1 for D2A21. The initial concentration of S is 7 μM, and the initial concentrations of other substrates are 0 μM.

▲ Figure 7: The simulations DPK-060 and TAT-DPK-060 released by thrombin. The parameters are described in the simulation section, and the k1 are 0.4 hr-1 (A), 0.2 hr-1 (B), 0.1 hr-1 (C). S is the full-length peptide, A is the CBD, B is the linker region, and C is DPK-060 and TAT-DPK-060.

▲ Figure 8: The simulations D2A21 released by thrombin. The parameters are described in the simulation section, and k1 are 0.4 hr-1 (A), 0.2 hr-1 (B), and 0.1 hr-1 (C). S is the full-length peptide, A is the CBD, B is the linker region, and C is the D2A21.


The simulations predict the release rate of our antimicrobial agents, which could help us to estimate the antimicrobial ability during the utilization of dressing. We are still establishing the two-site cleavage assay by thrombin to validate and improve our simulations.


1. Biniossek, M. L., Nägler, D. K., Becker-Pauly, C., & Schilling, O. (2011). Proteomic identification of protease cleavage sites characterizes prime and non-prime specificity of cysteine cathepsins B, L, and S. Journal of proteome research, 10(12), 5363–5373.

2. Jumper, J et al. Highly accurate protein structure prediction with AlphaFold. Nature (2021).

3. Fuchs, N., Meta, M., Schuppan, D., Nuhn, L., & Schirmeister, T. (2020). Novel Opportunities for Cathepsin S Inhibitors in Cancer Immunotherapy by Nanocarrier-Mediated Delivery. Cells, 9(9), 2021.

4. Verma, S., Dixit, R., & Pandey, K. C. (2016). Cysteine Proteases: Modes of Activation and Future Prospects as Pharmacological Targets. Frontiers in pharmacology, 7, 107.

5. Pauly, T. A., Sulea, T., Ammirati, M., Sivaraman, J., Danley, D. E., Griffor, M. C., Kamath, A. V., Wang, I. K., Laird, E. R., Seddon, A. P., Ménard, R., Cygler, M., & Rath, V. L. (2003). Specificity determinants of human cathepsin s revealed by crystal structures of complexes. Biochemistry, 42(11), 3203–3213.

6. Petrassi, H. M., Williams, J. A., Li, J., Tumanut, C., Ek, J., Nakai, T., Masick, B., Backes, B. J., & Harris, J. L. (2005). A strategy to profile prime and non-prime proteolytic substrate specificity. Bioorganic & medicinal chemistry letters, 15(12), 3162–3166.

7. Song, J., Tan, H., Perry, A. J., Akutsu, T., Webb, G. I., Whisstock, J. C., & Pike, R. N. (2012). PROSPER: an integrated feature-based tool for predicting protease substrate cleavage sites. PloS one, 7(11), e50300.