**1. Introduction**

The concept of the bispecific antibody (BsAb) has been around for more than 50 years, but within the last 20 years, activity and interest in the field of study has skyrocketed [1,2]. Publications describing hundreds of BsAbs can be found in the scientific literature, and more than 100 BsAb clinical candidates are currently under development [3,4]. A handful of BsAbs have obtained health authority approval for use and are currently marketed as therapeutics in a number of disease areas (e.g., blinatumomab, emicizumab) around the world, highlighting the therapeutic potential of engaging two targets within a single molecule [4]. This is attributed to advanced biotechnologies, enhanced manufacturing knowledge of therapeutic antibody products, and strong scientific rationale for the development of biologics with the ability to engage more than one target [5,6].

BsAbs are typically designed to possess the epitope specificity and manufacturability of a conventional monoclonal antibody (mAb) but are engineered to bind two distinct targets instead of one. The actual structure of a BsAb can vary widely, and depends on a number of factors including the intended mechanism of action (MoA) of the BsAb and desired pharmacokinetic/pharmacodynamic (PK/PD) properties [7,8]. Development and commercialization of BsAbs, to engage multiple targets using only one therapeutic, has gained significant attention recently, shifting industry focus and investments on this effective therapeutic strategy.

In this review, we discuss challenges and opportunities associated with developing bioassays for BsAbs with a particular focus on recent advances in bioanalytical approaches, as supported by multiple case studies.

**Citation:** Register, A.C.; Tarighat, S.S.; Lee, H.Y. Bioassay Development for Bispecific Antibodies—Challenges and Opportunities. *Int. J. Mol. Sci.* **2021**, *22*, 5350. https://doi.org/ 10.3390/ijms22105350

Academic Editor: Yong-Seok Heo

Received: 1 April 2021 Accepted: 15 May 2021 Published: 19 May 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/).

### *1.1. Diverse Formats of BsAb*

There are more than 100 distinct BsAb formats described and reviewed in the literature, but they generally fall into two categories: IgG-like and fragment-based (see Figure 1 and Wang et al. [9]).

**Figure 1.** Examples of BsAb formats and structural diversity: (**a**–**f**) IgG-like BsAbs and (**g**–**l**) fragment-based BsAbs.

DVD-Ig: dual variable domain immunoglobulin; scFv: single-chain variable fragment; Fab: antigen-binding fragment; HSA: human serum albumin; BiTE: bispecific T-cell engager; HLE: half-life extended; DART: dual-affinity re-targeting antibody.

The IgG-like BsAbs approximate the structure of a traditional mAb and typically contain an Fc domain and two antigen binding domains. However, many designs incorporate multiple copies of one or more antigen binding domains, allowing for avidity binding of one or more targets (Figure 1a–f; [10]). For example, an IgG-like anti-human epidermal growth factor receptor 2 (aHer2)/aCD3 bispecific molecule was engineered to include two low-affinity Her2 binding domains, thereby increasing the selectivity of the BsAb for cells overexpressing Her2 and increasing selective killing of tumor cells over Her2-expressing bystander cells [11]. IgG-like BsAbs tend to have longer serum half-lives due to the presence of an Fc domain that can interact with neonatal Fc Receptor (FcRn), and they can be easily engineered to either maximize or minimize interactions with FcgammaRs, allowing for flexibility in regards to effector function activity such as antibody-dependent cellular cytolysis (ADCC), antibody-dependent cellular phagocytosis (ADCP), and complement-dependent cytotoxicity (CDC) as desired [12]. IgG-like BsAbs can be challenging to manufacture, as many platforms require in-vitro or in-vivo assembly of two distinct half antibody pairs, resulting in product-related impurities stemming from chain mispairing events that can be difficult to separate from the desired product [9]. However, a number of technologies have been developed to overcome these challenges and maximize BsAb formation including knobs-into-holes, Cross mAb, and common light chain, among others [13–16].

In contrast, fragment-based BsAbs are typically much simpler to manufacture, as they are smaller and less structurally complex. Many fragment-based BsAbs are made by combining scFv fragments of different specificities (see Figure 1g–l), and they often self assemble from a single polypeptide chain (no opportunity for chain mispairing) [17]. Their small size can lead to better tissue penetration, and it has been postulated that their small size and conformational flexibility enable a more potent receptor activation, for example when bridging two cell types, compared to their larger counterparts [7,10]. However, they tend to have very short serum half lives due to the lack of an Fc domain. For example, while efficacious, blinatumomab treatment requires continuous infusion due to its extremely short serum half life (~2h; [18]). Several fragment-based structures have been developed to increase serum half, including appending scFv fragments to Fc domains or Human Serum Albumin (HSA) [10,19]. As with IgG-like BsAbs, there is a wide range of structural variability and avidity of binding available with this class of BsAb molecules.

#### *1.2. Mechanisms of Action of BsAb*

Due largely to the high level of interest in BsAbs as potential therapeutics and because of their structural diversity in design, both the scientific literature and clinical development pipeline contain numerous examples of BsAbs whose MoAs span a wide range [3]. For the purpose of this review, we will sort the BsAbs into four general classes: cell-bridging BsAbs, receptor/ligand blockers or activators, cofactor mimetics, and "homing" BsAbs (Figure 2; [3,20]).

**Figure 2.** Mechanisms of actions of BsAb: (**a**) Schematic diagram of cell-bridging BsAb MoA (e.g., TDB or NK-recruiting BsAb); (**b**) Schematic diagram of receptor activating/inhibiting MoA (e.g., receptor dimerization inhibitor or activator); (**c**) Schematic diagram of cofactor mimicking MoA (e.g., emicizumab); and (**d**) Schematic diagram of "homing" BsAb MoA (e.g., blood brain barrier crosser).

#### 1.2.1. MoA Type 1—Cell-bridging BsAbs

Cell-bridging BsAbs bind two distinct cell surface receptors—one on the surface of an effector cell and one on the surface of a target/tumor cell—resulting in activation of downstream signaling networks and killing of the target cell. One of the most prevalent examples of this MoA currently under clinical development is the T-cell dependent BsAbs (TDBs; [1]). These molecules most often target CD3e within the T-cell receptor (TCR) of cytolytic T cells and a tumor-specific antigen on the surface of target cells [8,21–34]. However, there are examples of BsAbs that activate T cells by engaging other epitopes, such as CD5 or co-stimulatory receptors such as CD28 [35,36]. Bridging of the target cell and the T cell by the BsAb leads to the formation of an immunological synapse, inducing T-cell activation and resulting in the release of perforin and granzymes that lyse the target cell [37]. Thus, TDBs harness a patient's own immune system to kill tumor cells

independent of TCR epitope specificity by circumventing activation through the major histocompatability complex [2]. TDB immunotherapy is similar in concept to CAR-T therapy, in which a patient's T cells are extracted and engineered with a chimeric antigen receptor (CAR) designed to recognize and kill tumor cells [38]. However, while TDBs are often more complex and difficult to produce than a standard mAb biologic, they are currently cheaper and less logistically challenging to manufacture than CAR-T therapies, which must be prepared individually for each patient [39]. Additionally, TDBs can have more favorable safety profiles compared to CAR-T therapies, with fewer and less severe adverse events such as systemic cytokine release syndrome—the most common adverse event associated with immune-modulating therapies [27,40,41]. In addition to TDBs, there are several examples of BsAbs that recruit and activate NK cells by simultaneously binding CD16 (FcgammaRIII) and a tumor-specific receptor [42–45], as well as a BsAb that recruits and activates macrophages by targeting CD89 [46].

#### 1.2.2. MoA Type 2—Receptor/Ligand-Blocking or -Activating BsAbs

By virtue of their ability to target more than one receptor, BsAbs can be developed to target and activate a receptor in a specific cellular context (e.g., a therapeutically-relevant complex). This allows for a level of selectivity that cannot be achieved with conventional mAbs alone or in combination. For example, the anti-Fibroblast Growth Factor Receptor (aFGFRI)/anti-β-Klotho (aKLB) BsAb activates the FGFRI/KLB receptor complex, leading to weight loss and a reduction in obesity-linked disorders in preclinical models [47]. By selectively targeting FGFRI/KLB, the molecule activates FGFRI when complexed with KLB, thereby avoiding widespread FGFRI activation—FGFRI receptor is expressed in a wide range of tissues—and reducing the unintended side effects associated with mAb FGFRI agonists.

In addition to acting as receptor agonists, BsAbs can also be effective receptor antagonists. Resistance to various Her2-targeting mAbs (e.g., trastuzumab) has led to the development of novel therapeutics for blocking Her2-associated signaling, including several BsAbs [48]. While many of these molecules are TDBs (MoA discussed above), there are also examples of BsAbs that bind to Her2 and Her3, preventing ligand-activated Her3 from heterodimerizing with Her2, and dampening PI3K signaling in Her2-overexpressing cancers [48]. There is also an example of an antibody-drug conjugate (ADC) BsAb that targets two distinct, non-overlapping epitopes on Her2, leading to more efficient internalization, lysosomal degradation, and release of cytotoxic payload [49]. Beyond treatments for Her2-overexpressing cancers, there are many examples of BsAbs that target combinations of receptors and/or cognate ligands, as well as cytokines [50–59]. These BsAbs sometimes serve the same purpose as that of a combination treatment of mAb therapeutics, but there are instances in which a BsAb provides a particular advantage. For example, an aCTLA4/PD1 BsAb was developed to preferentially inhibit CTLA-4 on PD1+ cells, leading to fewer adverse events associated with immune activation than have been observed when treating patients with combinations of the conventional mAb aCTLA-4 and aPD1/L checkpoint inhibitors [60]. Monovalent targeting of CTLA-4 significantly reduces the ability of the BsAb to inhibit CTLA-4, but monovalent binding has a much lower impact on the ability of the molecule to inhibit PD1 compared to a conventional bivalent aPD1 mAb. As a result, the BsAb is able to saturate CTLA-4 receptors on PD1+ cells, without widespread inhibition of CTLA-4 leading to fewer adverse events. Bispecific targeting of CTLA-4 and PD1 with this BsAb also leads to internalization and degradation of PD1—an effect that is not observed with combinations of aCTLA-4 and aPD1 mAbs.

#### 1.2.3. MoA Type 3—Cofactor Mimicking BsAbs

Emicizumab (marketed name Hemlibra®®) is a BsAb that was developed to treat hemophilia A. The BsAb binds to coagulation Factors X and IX and is therefore able to play the role of Factor XIII—the coagulation factor missing in many hemophilia A patients [61,62].
