MODEL
Background
Many therapeutic agents are currently restricted by their inability to reach the appropriate intracellular targets. Cell-penetrating peptides (CPPs) are short amino acid sequences that deliver cell-impermeable molecules across the cell membrane. As there are many limitations with targeting a cell with an antibody from outside the cell, such as cell receptors, we hope to more efficiently deliver the drug signals by using CPP to move antibodies within the cell. CPP, by fusioning with an antibody of interest, can bring the antibody inside the cell through the phospholipid bilayer.
In our wet lab, we had scFv(P5) modified from scFv (F8, by grafting the antigen-binding region of poorly stable anti-hen egg lysozyme (HEL) scFv(D1.3) to scFv (F8) scaffold, which is a monoclonal antibody raised against the coat protein of the plant virus AMCV. A CPP sequence from porcine circovirus type 2 was added to the N-terminus of scFv(P5) for intracellular delivery. After our team completed cloning, expression, purification, binding assay, and cell-penetrating assay to demonstrate that we have successfully designed a hyperstable antibody with CPP for intracellular targeting, we wanted to apply our project into therapeutic treatments. We have consulted with Dr. Sunghee Ahn, a former oncologist, and she advised us to target a Ras protein, which is involved in intracellular signals passed between cells that stimulate the amount of growth. A cancer-inducing mutation of Ras develops a form of the protein that is always activated. Therefore, we decided to perform a molecular dynamics modeling of our CPP-attached hyperstable single-chain variable fragment that can recognize and bind to a Ras protein, inhibiting cell proliferation.
Modeling Plan
Using the same mechanism of grafting the lysozyme targeting antigen-binding region of scFv(D1.3) onto scFv(F8), we inserted the antigen-binding region of Ras protein targeting antibody to the complementarity-determining region of scFv(F8), the scaffold. We also attached cell-penetrating peptide to the scFv for intracellular delivery. In order to determine whether our intended application reflected an accurate estimation of the probable success of the newly designed scFv, we performed homology modeling through a program called MODELLER.
Fig. 1 Grafting complementarity-determining region (CDR) of Ras-recognizing antibody to scFv(F8)
What is Homology Modeling?
-
Homology modeling is used to predict the structure of a target protein of known amino acid sequence when the target protein is related to at least one other protein of known sequence and structure
- ​If the proteins are closely related, then the known protein structures can serve as the basis for a model of the target​
- This type of modeling is used before experimental testing so that researchers can measure the probable success of the novel protein and modify their experiment before moving on to the laboratory
MODELLER
-
Homology modeling program developed by Andrej Sali and his colleagues
-
PyMOL, an open source molecular visualization system, and PyMod, a PyMOL plugin that serves as a bridge between PyMOL and other bioinformatics tools, are components of the MODELLER package
-
Our team decided to use MODELLER to perform homology modeling of our novel scFv because it is convenient to adjust optimization, alter energy minimization, arrange multiple alignments of protein sequences, and BLAST search protein data bank (PDB) in this program.
METHODOLOGY
Based on Dr. Ahn’s advice, we decided to research a Ras protooncogene and Fv complex on Protein Data Bank (PDB). We looked for a complex consisting of an intrabody without a disulfide bond because inside the cell, which is a reducing environment, disulfide bonds are easily broken. We wanted our known protein to be a complex of the antigen-binding fragment of the antibody and the Ras protein. As such, we found 2VH5, an ANTI-RAS FV (disulfide free mutant) COMPLEX on PDB, 2VH5, and used it to create a molecular dynamics simulation for hyperstable intracellular antibody targeting Ras protein.
First, we grafted the heavy and light chain regions of 2VH5 onto scFv(F8) connected by a linker and attached to a cell-penetrating peptide so that our designed scFv would recognize the Ras protein.
2VH5 Heavy + Light Chains:
EVQLLESGGGLVQPGGSLRLSAAASGFTFSTFSMNWVRQAPGKGLEWVSYISRTSKTIYYADSVKGRFTISRDNSKNTLY LQMNSLRAEDTAVYYVARGRFFDYWGQGTLVTVS IQMTQSPSSLSASVGDRVTITVRASQSISSYLNWYQQKPGEAPKLLIYSASVLQSGVPSRFSGSGSGTDFTLTISSLQPE DFATYYAQQSVMIPMTFGQGTKVE
​
scFv(F8)_CPP_Linker:
MTYTRRRFRRRRHRPRS QVQLQESGGDLVQPGGSLKLSCAASGFTFSSYGMSWVRQTPDKRLELVATINSNGGSTFYPDSVKGRFTISRDNAKNTLYLQMSSLKSEDTAMYYCARRRNYPYYYGSRGTFDYWGQGTTVTVSS GGGGSGGGGSGGGGS DIELTQSPASLAVSLGQRATISCRASESVDSYGNSFMHWYQQKPGQPPKLLIYRALNLESGIPARFSGSGSRTDFTLTINPVEADDVATYYCQQSNEDPWTFGGGTKLEIKR
​
scFv(F8)_CPP_Linker + 2VH5 grafted:
MTYTRRRFRRRRHRPRS QVQLQESGGDLVQPGGSLKLSCAASGFTFSSYGMSWVRQTPDKRLELVATINRTGKTTYYPDSVKGRFTISRDNAKNTLYLQMSSLKSEDTAMYYCARGRFFDYWGQGTTVTVSS GGGGSGGGGSGGGGS DIELTQSPASLAVSLGQRATISCRASESVDSYMHWYQQKPGQPPKLLIYRALNLESGIPARFSGSGSRTDFTLTINPVEADDVATYYCQQSVEDPWTFGGGTKLEIKR
Because we created the target protein sequence as a text file, we converted it to a FASTA file, which is the format read in PyMod. Then, we ran a BLAST search of our target protein sequence on the Protein Data Bank to find similar proteins of known structure.
We selected the top five most similar known structures (based on the E-Value) from the BLAST search: 5B3N, 6NJL, 5GS3, 6G8R, 3UYP.
After opening them on PyMod, we aligned the amino acid sequences of the target protein and proteins of known structure through MUSCLE (MUltiple Sequence Comparison by Log-Expectation).
We fetched the PDB structure of all the sequences and decided to only use 5B3N as our known protein structure for a compact view. Then, we linked regions of insertions and deletions into the mainchain of the known protein so that a model was created for the complete mainchain of the target protein.
Finally, we refined the model by limited energy minimization, and the root-mean-square deviation (RMSD) was 0.554 Angstrom.
RESULTS
Fig. 2 2VH5 (ANTI-RAS FV COMPLEX)
Fig. 3 5B3N (Known Structure)
Fig. 4 scFv(F8) with Heavy and Light Chains of 2VH5 Grafted onto its CDR (Target Structure)
Fig. 5 Combined View of Grafted scFv(F8) and 5B3N
Fig. 6 Combined View of Grafted scFv(F8) and 2VH5
Fig. 7 Combined View of Grafted scFv(F8), 5B3N, and 2VH5
Fig. 8 PyMOL Animation of Grafted scFv(F8) and 5B3N
Fig. 9 PyMOL Animation of Grafted scFv(F8) and 2VH5
ANALYSIS
As shown in the results from MODELLER homology modeling, our team was successfully able to create molecular dynamics for CPP-attached scFv(F8) added with heavy and light chains of Ras-targeting antibody. The PyMOL animation of grafted scFv(F8) (target protein) and 5B3N (known protein) (fig. 8) indicates their structural similarity, establishing the success of our homology modeling. Furthermore, the PyMOL animation of Ras-targeting scFv(F8) and 2VH5 (ANTI-RAS FV complex) protein structure (fig. 9) demonstrates that the antigen-binding site of the single-chain variable fragment is compatible with the Ras protein of 2VH5 and structurally similar to the intracellular antibody of 2VH5. Thus, we conclude that our model of CPP-attached Ras-targeting scFv was a success.
FUTURE WORK
We believe the homology model we have created is our stepping stone for developing an intracellular antibody that targets Ras proteins. Because we have determined the feasibility of our idea through modeling, we hope to utilize our simulation to help better understand the affinity of scFv and Ras and the protein structure of CPP-attached and grafted scFv(F8) in our laboratory in the future.
REFERENCE
[1] Lesk, A. M. (2016). Introduction to protein science: architecture, function, and genomics. Oxford: Oxford University Press.
[2] Tanaka, T., & Rabbitts, T. H. (2008). Functional Intracellular Antibody Fragments Do Not Require Invariant Intra-domain Disulfide Bonds. Journal of Molecular Biology, 376(3), 749–757. doi: 10.1016/j.jmb.2007.11.085