Overview
Our goal is to design peptides to imitate CLEC5A docking with dengue virus. In order to ensure the peptides and the envelope protein (E protein) of dengue virus have an interaction, all the structure of peptides and proteins and their interactions were modeled using Rosetta. First, we utilized RosettaCM (Comparative modeling with Rosetta) to generate the structure of E protein from a local strain (PL046) based on the crystal structure (PDB: 1OAN). Second, we used the ab initio method to predict the peptide structures purely based on their sequences. Then, we utilized the clustering method to cluster the results and find the most probable structure of the peptide. Finally, we verified the interactions between these predicted peptides and the E protein based on the global protein-protein docking.
Protocols
RosettaCM (Comparative modeling with Rosetta)
Purpose: To generate the structure of E protein (PL046)
- Create a threaded model
- Target sequence in a fasta file
- Sequence alignment file in the Grishin format
- Template PDB file
- Run the hybridization mover
- Target sequence in a fasta file
- RosettaScripts xml file:
- The weight files
- Target sequence fragments files: 3-residue fragments and 9-residue fragments
- The PDB file of threaded template from step 1
- Final relax
- Disulfide bond
- PDB files generated from step 2
Input:
Output: A PDB file of threaded template.
Commands: partial_thread.static.linuxgccrelease @flag_threading
Flags:
-in:file:fasta Eprotein_pl046.fasta
-in:file:alignment Eprotein_pl046_B.grishin
-in:file:template_pdb Eprotein_B.pdb
-ignore_unrecognized_res
Input:
Output: PDB files after hybridization.
Commands: rosetta_scripts.static.linuxgccrelease @flag_hybridize
Flags:
-in:file:fasta Eprotein_pl046.fasta
-parser:protocol rosetta_cm.xml
-nstruct 1000
-out:suffix _001
-out:path:pdb ./result
-relax:minimize_bond_angles
-relax:minimize_bond_lengths
-relax:jump_move true
-default_max_cycles 200
-relax:min_type lbfgs_armijo_nonmonotone
-relax:jump_move true
-score:weights stage3.wts
-use_bicubic_interpolation
-hybridize:stage1_probability 1.0
-chemical:exclude_patches LowerDNA UpperDNA Cterm_amidation SpecialRotamer VirtualBB ShoveBB VirtualDNAPhosphate VirtualNTerm CTermConnect sc_orbitals pro_hydroxylated_case1 pro_hydroxylated_case2 ser_phosphorylated thr_phosphorylated tyr_phosphorylated tyr_sulfated lys_dimethylated lys_monomethylated lys_trimethylated lys_acetylated glu_carboxylated cys_acetylated tyr_diiodinated N_acetylated C_methylamidated MethylatedProteinCterm
Input:
Output: About 10,000 results.
Commands: relax.static.linuxgccrelease @flag_relax
Flags:
-in:fix_disulf Eprotein_pl046.disulfide
-in:file:l pdblist
-nstruct 1
-in:detect_disulf true
-default_max_cycles 3000
-out:suffix _relax
-out:file:fullatom
-out:path:pdb ./result
Figure 1. E protein PL046 (green) based on homology (pink).
Ab initio
Purpose: To generate the structure of the peptides
Input:
- Protein sequence in fasta format
- Fragment library: 3-residue fragments and 9-residue fragments
- Psipred-ss2 (optional)
The last two inputs were generated from Robetta Fragment
Output: About 20,000 results.
Commands: AbinitioRelax.static.linuxgccrelease @flags
Flags:
-in:file:fasta ./input_files/t000_.fasta
-in:file:frag3 ./input_files/aat000_03_05.200_v1_3
-in:file:frag9 ./input_files/aat000_09_05.200_v1_3
-abinitio:relax
-nstruct 2000
-seed_offset 10
-use_filters true
-psipred_ss2 ./input_files/t000_.psipred_ss2
-abinitio::increase_cycles 10
-abinitio::rg_reweight 0.5
-abinitio::rsd_wt_helix 0.5
-abinitio::rsd_wt_loop 0.5
-relax::fast
-out:file:silent fold_silent0001.out
-out:sf score0001.fsc
We did not find the "funnel shape" from the plot of "Rosetta energy score vs RMSD" for the best 1,000 structures, which suggests there might be no single stable conformation.
Figure 2. Two of the representative structures of PTRS
Clustering
Purpose: To cluster the results and find the most probable structure
- Rescore the thousands of results
- Energy cut-off
- Extract the results
Input: The ab initio results.
Output: The clustering scores.
score_jd2.static.linuxgccrelease -in:file:s 110_7/*.pdb -out:file:silent scored_silent110_7.out -out:file:silent_struct_type binary -in:file:fullatom
Input: The clustering scores.
Output: The results passing the energy cut-off.
python make_sub_silent_file_percentile.py scored_silent110_7.out ecut110_7.out -1 0.10
Input: The results passing the energy cut-off.
Output: The structures sorted by clustering. (We chose the lowest score to simulate protein and protein docking.)
Commands: cluster.static.linuxgccrelease @flag_cluster
Flags:
-in:file:silent ecut110_7.out
-in:file:silent_struct_type binary
-in:file:fullatom
-cluster:gdtmm
-cluster:population_weight 0.0
-cluster:radius -1
-cluster:sort_groups_by_energy
-out:file:silent clustered110_7.out
-score:weights score12.wts
-score:patch
-out:file:silent_struct_type binary
-run:shuffle
Protein-protein docking (global docking)
Purpose: To find the interaction between peptides and E Protein
Input: The structures of ligand (peptide or CLEC5A) and receptor (E protein) in the same input file.
Output: About 10,000 results. (Find the most frequent sites among the results.)
Commands: docking_protocol.static.linuxgccrelease @flag_global_docking
Flags:
-in:file:s input_files/110_7_PL046noglycan.pdb
-nstruct 2000
-partners A_B
-randomize1
-randomize2
-center_at_interface
-ex1
-ex2aro
-out:path:all output_files
-out:suffix _global_dock1
Overview
The weakness of our design is that the peptides from the gold nanoparticles and the ones from the glass fiber compete for the same binding sites on the E protein. Moreover, if the E proteins on the virus particles are fully covered by the gold nanoparticles, there are no sites available to interact the peptides from the glass fiber.
Repulsion between gold nanoparticles
To assess this potential problem, we used DLVO theory to calculate the repulsion between gold nanoparticles to estimate the number of gold nanoparticles that would bind to a virus particle. The structure of dengue virus is icosahedral, and there are nine E proteins on each surface. The distances between the potential binding sites can be obtained from the structure in protein data bank (1K4R). DLVO theory can be described as Equation 1.
Equation 1. Wtotal(D) = Wa(D) + Wr(D) = -AR/12D + 2πεε0RΨδ2exp(-κD)
Wtotal(D): total energy
Wa(D): van der Waals interaction energy
Wr(D): electrostatic interaction energy
A: Hamaker constant = 2.5×10-19 J
R: radius = 1.3x10-8 m (based on the size of gold nanoparticles)
C: center to center distance
D: C-R
ε: permittivity of vacuum = 8.854x10-12 F/m
ε0: dielectric of solution = 80.4 (at 20°C)
Ψδ2: Stern layer potential (zeta potential) = 1.794 mV
κ-1: diffuse layer thickness = 7.95x10-10 m
We took several representative positions on adjoining faces of the icosahedron to calculate the interactions between the gold nanoparticles based on DLVO theory. We found the total energies are all positive (Table 1), and these total energies are also larger than the typical biological interactions (~0.5 kcal/mol or 3.49 x1021 J). The results suggest that there will always be free faces on the virus particles to interact with the peptides conjugated on the test line.
Table 1. The representative energies of the interaction between the gold nanoparticles.
 |
 |
 |
 |
|
|---|---|---|---|---|
| C (nm) | 54.58 | 42.71 | 31.71 | 43.24 |
| Attractive energy (10-21J) | -3.3 | -4.6 | -7.2 | -4.5 |
| Repulsive energy (10-21J) | 51.0 | 51.3 | 51.6 | 51.3 |
| Total energy (10-21J) | 47.7 | 46.7 | 44.4 | 46.8 |
 |
 |
 |
 |
|
| C (nm) | 30.32 | 31.06 | 33.19 | 30.32 |
| Attractive energy (10-21J) | -7.8 | -7.5 | -6.7 | -7.8 |
| Repulsive energy (10-21J) | 51.7 | 51.7 | 51.6 | 51.7 |
| Total energy (10-21J) | 43.9 | 44.2 | 44.9 | 43.9 |
References
Combs, Steven A., Deluca, Samuel L., Deluca, Stephanie H., Lemmon, Gordon H., Nannemann, David P., Nguyen, Elizabeth D., Willis, Jordan R., Sheehan, Jonathan H. & Meiler, Jens. (2013). Small-molecule ligand docking into comparative models with Rosetta. Nature Protocols, 8(7), 1277-1298. doi: 10.1038/nprot.2013.074
Raveh, Barak., London, Nir., Zimmerman, Lior. & Schueler-Furman, Ora. (2011). Rosetta FlexPepDockab-initio: Simultaneous folding, docking and refinement of peptides onto their receptors. PLoS ONE, 6(4). doi: 10.1371/journal.pone.0018934
Alam, Nawsad., Goldstein, Oriel., Xia, Bing., Porter, Kathryn A., Kozakov, Dima. & Schueler-Furman, Ora. (2017). High-resolution global peptide-protein docking using fragments-based PIPER-FlexPepDock. PLoS Computational Biology, 13(12), 1-20. doi: 10.1371/journal.pcbi.1005905
Barrientos, Arturo. & Concha, Fernando. (1990). Phenomenological model of classification in conventional hydrocylones. Comminution, 819(1), 287-305. doi: 10.1007/978-1-61779-465-0
Li, Haiou., Lu, Liyao., Chen, Rong., Quan, Lijun., Xia, Xiaoyan. & Lü, Qiang. (2014). PaFlexPepDock: Parallel Ab-initio docking of peptides onto their receptors with full flexibility based on Rosetta. PLoS ONE, 9(5). doi: 10.1371/journal.pone.0094769
Ciemny, Maciej., Kurcinski, Mateusz., Kamel, Karol., Kolinski, Andrzej., Alam, Nawsad., Schueler-Furman, Ora. & Kmiecik, Sebastian. (2018). Protein–peptide docking: opportunities and challenges. Drug Discovery Today, 23(8), 1530-1537. doi: 10.1016/j.drudis.2018.05.006
Elena Pokidysheva, Ying Zhang, Anthony J. Battisti, Carol M. Bator-Kelly, Paul R. Chipman, Chuan Xiao, G. Glenn Gregorio, Wayne A. Hendrickson, Richard J. Kuhn, Michael G. Rossmann. Cryo-EM Reconstruction of Dengue Virus in Complex with the Carbohydrate Recognition Domain of DC-SIGN. Cell Press, 124(3). doi:10.1016
Jörg Polte. Fundamental growth principles of colloidal metal nanoparticles – a new perspective. CrystEngComm, 36. doi:10.1039
Phillip E. Mason, Adrien Lerbret, Marie-Louise Saboungi, George W. Neilson, Christopher E. Dempsey, John W. Brady. Glucose Interactions with a Model Peptide. NCBI, 79(7). doi:10.1002
Taehoon Kim, Kangtaek Lee, Myoung-seon Gong, Sang-Woo Joo. Control of Gold Nanoparticle Aggregates by Manipulation of Interparticle Interaction. ACS Publications. doi:10.1021
© iGEM Team, 2020 CCU Taiwan