1. Introduction
Antibodies, ~150 kDa glycoproteins, consist of two light and two heavy polypeptide chains with six CDRs for specific binding to their targets [1]. However, their therapeutic use is limited due to the low tissue penetration. Recently, advances in the structural engineering of proteins led to the construction of many small-size proteins (3-20 kDa) with the ability to bind to epitopes recognized by mono antibodies [2]. Among all the engineered and recombinant proteins, single chain variable fragments (scFvs) composed of VH and VL, with the same antigen-binding specificity can retain the affinity of a whole antibody molecule [3]. Low-molecular weight, high rate of excretion, good tissue penetration, and low costs for production of these antibody mimetics converted them into a promising alternative for targeted therapies [4]. Currently, amino acid conversion on the surface area of non-immunoglobulin scaffolds resulted in the generation of various small antibody mimetics [4]. However, length, conformation, and the conserved amino acid residues of the hypervariable CDRs loop play an essential role in native binding site definition [5]. On the other hand, 30 kDa fragments of antibody mimetics developed by the antibody fragmentation technique restricted their simple chemical synthesis [4]. Using the CDRs grafting strategy, protein scaffolds with variable loops that mimic the conformation of CDRs can be extracted to generate small scaffolds with improved affinity. Nicaise et al. developed some protein scaffolds against lysozyme which harbored CDRs from a specific antibody. However, designed scaffolds showed various affinities towards lysozyme, indicating that selected scaffolds directly affect the affinity of grafted CDRs [6]. Therefore, for designing antibody mimetics, selecting proper protein scaffolds and determining appropriate CDRs peptide sequences is essential [7]. Among many introduced non-immunoglobulin scaffolds, anticalins with a smaller size of 180 amino acids consisting one polypeptide chain, a beta-sheet structure, and four flexible loops without any glycosylated structure can be produced in prokaryotic hosts [8]. Another example of protein scaffolds is the tenth domain of human fibronectin type III (Fn3, also called Adnectin). Adnectin is a 94-residue monomeric protein with a β-sandwich fold structure and six internal loops without disulfide bonds. This protein mimics the interaction of CDRs with the target through three loops on the tip of molecules [8]. The third structure selected for this study is VHH, with small molecular weight, low aggregation, and high solubility, which contains three variable loops [5]. This existing assortment represents an ideal CDR acceptor repertoire for our approach. In addition to the structure of the CDRs acceptor, the framework plays a vital role in CDR orientation and conformation. Therefore, framework residues should be considered in designing a protein scaffold with high binding affinity to the target [9]. Here, we aimed to simulate the CDRs conformation in three selected scaffolds of humanized VHH, Fn3, and lipocalin with at least three flexible loops for grafting CDRs isolated from the target antibody. Three CDRs peptides of H2, H3, and L3 that play an important role in the recognition of the target were extracted from the monoclonal antibody panitumumab that binds specifically to the EGFR DIII. These CDRs were then grafted through loop randomization in all these scaffolds. Apart from scFv, which carries all 6 CDRs in their native structure, 36 anti-EGFR DIII protein candidates were created (24 candidates from lipocalin scaffold, 6 structures of Fn3, and 6 candidates of humanized VHH). Among all these structures, three of them were selected for further analysis in molecular dynamic simulation.