**1. Introduction**

Toll-like receptors (TLRs), as featured pattern recognition receptors, are proven to operate as germline-encoded proteins recognizing conserved pathogen-associated molecular patterns [1]. Dysregulation of cellular activities by microbes or their products through TLR signaling affects both innate and adaptive immune responses [2]. Excessive activation of TLRs disrupts immune homeostasis and is triggered by persistent induction of proinflammatory cytokines and chemokines, thereby subsequently leading to the initiation of various inflammatory and autoimmune disorders such as systemic lupus erythematosus, sepsis, psoriasis, atherosclerosis and asthma [3]. TLR4, the first mammalian Toll protein characterized in humans [4], recognizes damage-associated molecular patterns in the debris released by injured tissues and necrotic cells as well as pathogen-associated molecular patterns, and is thus associated with the development of several acute and chronic disorders such as sepsis [5]. TLR4 in association with interleukin (IL) 29 plays a crucial role in synovial inflammation. IL-29 upregulates synovial-fibroblast TLR4, which enhances synovium inflammation in rheumatoid arthritis (RA). Elevated expression may be due to elevated numbers of macrophages that penetrate the synovium and promote RA [6,7]. Numerous studies have shown that TLR4 stimulates the expression of many proinflammatory cytokines that play a crucial part in myocardial inflammation, including myocarditis, myocardial infarction, ischemia-reperfusion, and heart failure [8]. Ample evidence confirms the participation of TLR4 in neuroinflammation, where activation of

**Citation:** Ahmad, B.; Batool, M.; Kim, M.-S.; Choi, S. Computational-Driven Epitope Verification and Affinity Maturation of TLR4-Targeting Antibodies. *Int. J. Mol. Sci.* **2021**, *22*, 5989. https:// doi.org/10.3390/ijms22115989

Academic Editor: Małgorzata Borówko

Received: 26 April 2021 Accepted: 29 May 2021 Published: 1 June 2021

**Publisher's Note:** MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

**Copyright:** © 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https:// creativecommons.org/licenses/by/ 4.0/).

this receptor stimulates microglial expression and activation of NF-κB, as well as the induction of inflammatory cytokines IL-1β, IL-6 and tumor necrosis factor (TNF) α [9,10]. The involvement of TLR4 in the pathogenesis and aggravation of these diseases highlights its importance as a potential drug target. In the TLR family, TLR4 remains a priority drug target and has been extensively studied for its therapeutic potential in several inflammatory disorders; many of these therapeutics are undergoing clinical trials [11].

Antibody-based medicines are currently the most widely used form of biotherapeutics, and their market expanded rapidly in recent years with increasing numbers of such modalities receiving FDA approval. Compared to small-molecule drugs, monoclonal antibodies (mAbs) are considered more selective and highly effective and have become mainstream therapies for autoimmune and other hard-to-cure diseases. Moreover, they have revolutionized the treatment of cancers, where inflammation is regarded as a crucial factor [12]. Nonetheless, small-molecule drugs that lack specificity are likely to have off-target effects. These drugs can interact unexpectedly and undesirably with tissues, cells and cellular components. The application of small-molecule drugs has a greater number of adverse effects due to their lower specificity as compared to mAbs. On the contrary, the clinical approval rate is higher for mAbs, and their development is easier and considerably faster than sorganic small-molecule compounds.

The development of high-affinity mAbs for therapeutic purposes is still a holy grail of molecular engineering [13]. Researchers are using numerous empirical procedures, including site-directed mutagenesis, complementarity-determining region (CDR)-grafting, and phage display techniques to enhance the target-binding affinity of mAbs [14]. Computational approaches are coming onto the scene and facilitating the development and affinity enhancement of mAbs. Prior sequence and structural information on both antigens and mAbs, and their epitope and paratope insights, are of grea<sup>t</sup> value for enhancing the affinity and stability of mAbs. These computational methods involve structural bioinformatic techniques such as homology modeling, protein–protein docking, protein interface analysis and molecular dynamics (MD) for rational design of mAbs. They provide detailed insights into binding and unbinding mechanisms as well as structure kinetics of protein-protein, protein-ligand and antigen-antibody that can be used to guide ligand, protein, peptide and mAb design [15]. The modelling of mAb-antigen interaction, and applying attraction or repulsion filters for masking the nonparatope residues as in Cluspro program, substantially improves the antigen-antibody docking complexes [16]. Thus, false positive of an epitope can be removed. The computational techniques alone can be used for affinity maturation of selected mAbs. Techniques such as AbDesign, RosettaAntibodyDesign, OptCDR and OptMAVEn are categorized as ab initio protocols for the design of novel paratopes [17,18]. Machine learning and deep learning methods can help to design CDRs for human IgG antibodies with target affinity that is superior to that of candidates derived from phage display panning experiments [19,20]. Thus far, a few effective TLR4-targeting mAbs have been investigated in preclinical and clinical studies [21,22]. NI-0101, previously known as Hu 15C1, has been developed in BALB/c mice and proved to efficiently block the signaling of lipopolysaccharide-triggered TLR4 both in vitro and in vivo [22,23]. NI-0101 is a humanized immunoglobulin (Ig) G1κ mAb, which is engineered to interfere with the dimerization of TLR4 and to abrogate its downstream signaling. NI-0101 has been proved effective in lipopolysaccharide-treated healthy volunteers [24]; nonetheless, recent clinical findings sugges<sup>t</sup> that to cure RA, inhibition of TLR4 alone is not sufficient [25]. A molecular understanding of antigen–antibody interaction and identification of their epitope–paratope hotspots are indispensable for affinity enhancement of known mAbs. Taking advantage of the epitope knowledge provided by Greg Elson's group through alanine mutagenesis [22], Loyau et al. utilized structural, computational, and phage display techniques to enhance TLR4-binding affinity of Hu 15C1 and suggested a C2E3 derivative with better TLR4-binding affinity [26]. Overall, these studies predicted the TLR4-mAb binding interface through site-directed mutational analyses. Nevertheless, the conformational changes that occur in the TLR4 structure owing to these mutations have been overlooked. These

mutation-driven conformational changes obstruct TLR4-mAb interaction. In this study, we used computational mutagenesis, molecular docking, MD simulations and molecular mechanics to delineate the dynamic binding interface of Hu 15C1 and its derivative C2E3 toward TLR4. The crucial data generated during this study regarding TLR4 epitopes will be helpful for designing efficient mAbs to block TLR4 signaling and to curb the associated immune complications.
