Week 2, February 2026: Conditionally Active Antibodies

Antibody therapeutics have a structural limitation: many targets such as immune checkpoints, costimulatory receptors, and broadly expressed tumor-associated antigens, cannot be engaged at pharmacologically optimal exposures without triggering dose-limiting on-target, off-tumor toxicities. In response, multiple groups are exploring antibody formats, like introducing conditionality. Recent work on checkpoint blockade, multispecific agonism, antibody-drug conjugates, and computationally designed masking elements illustrates how conditional activation is being studied.

Defining Conditional Activity in Antibody Engineering

A conditionally active antibody is engineered so that antigen binding, receptor clustering, or Fc-mediated effector recruitment is materially attenuated under baseline systemic conditions and restored only upon exposure to a defined trigger. One method for this is masking strategies, in which steric hindrance or affinity-based occlusion of the paratope is relieved through tumor-associated stimuli, most commonly protease-mediated linker cleavage. Another method relies on spatial or coincidence logic, in which productive signaling (e.g., receptor agonism) requires anchoring to a second antigen or cell surface, thereby restricting functional activation to specific microenvironments.

Tumor-Localized Checkpoint Blockade as a Model for Conditional Activation

The rationale for conditional activation is evident in the context of CTLA-4 blockade, where systemic administration is associated with immune-related adverse events. In a recent study by Arias-Badia et al. (2025), tumor-localized conditional CTLA-4 blockade was engineered using a protease-cleavable design intended to restrict activation to the tumor microenvironment. By incorporating a tumor-targeting element linked to a cleavable domain, the antibody remains functionally attenuated until exposure to tumor-associated proteolytic activity.

In mouse models of prostate cancer, the version of the antibody that was designed to be fully cleaved within the tumor slowed tumor growth to a similar extent as standard anti-CTLA-4 therapy. However, mice receiving the conditional antibody showed less treatment-associated weight loss, a commonly used indicator of systemic toxicity in preclinical studies, suggesting improved tolerability. A second design, which was only partially cleavable and remained associated with the tumor-targeting element after activation, was less effective at controlling tumor growth. This comparison highlights a common challenge in conditional antibody engineering: strategies that increase tumor localization do not always translate into optimal functional activity within the tumor microenvironment, as factors such as diffusion, receptor engagement dynamics, and local concentration can influence pharmacodynamic outcomes. Although these findings are limited to preclinical models, they support the concept that conditional checkpoint blockade is primarily intended to reduce systemic immune activation while preserving antitumor efficacy, rather than to enhance intrinsic potency beyond that of conventional antibodies.

Human-Derived Steric Masking Platforms

Steric hindrance-based masking strategies have been explored for several years, but concerns regarding immunogenicity of non-human masking domains persist. A recent study describing a calmodulin-calmodulin-binding peptide (CaM-CBP) “peptide clamp” platform provides an example of how human-derived components can be leveraged to address this limitation. In this system, calmodulin and its cognate peptide are fused to the N-termini of antibody heavy and light chains via an MMP-cleavable linker, forming a steric clamp that occludes the antigen-binding interface until protease-mediated hydrolysis occurs in the tumor microenvironment.

When this masking platform was applied to the clinically established antibodies trastuzumab and cetuximab, the modified (masked) versions showed markedly reduced ability to bind their target antigens. In parallel, activation of antibody-dependent cellular cytotoxicity (ADCC) in a reporter assay was strongly diminished, indicating that the masking strategy effectively suppressed effector function under baseline conditions. After exposure to MMP-9, a protease commonly enriched in tumor microenvironments, the masking domain was cleaved and both antigen binding and ADCC activity were restored to levels comparable to the original, unmodified antibodies. Thus, conditional antibody platforms are increasingly evaluated not only for their ability to control biological activity, but also for manufacturability, stability, and other properties essential for clinical translation.

Conditional Agonism in Bispecifics

Conditionality can also be achieved without proteolytic unmasking. In a c-MET × CD137 bispecific antibody described by Zhang et al. (2024), CD137 agonism was engineered to depend on tumor-cell anchoring via c-MET. By employing a monovalent c-MET–binding arm and configuring the construct to enable trivalent CD137 engagement only upon tumor-surface anchoring, the design seeks to confine costimulatory signaling to the tumor microenvironment.

In cell-based assays, the bispecific antibody triggered strong T-cell activation and cytokine production only when both targets were present, while activation was minimal under single-target conditions. When combined with anti-PD-1 therapy, cytokine release was further enhanced in vitro. Here, conditionality is achieved through spatial control rather than protease cleavage. Effective CD137 agonism requires receptor clustering, which occurs predominantly when the antibody is anchored to tumor cells via its c-MET–binding arm. This anchoring-dependent design aims to localize immune stimulation to the tumor environment, addressing the potency-toxicity challenges that have limited systemic CD137 agonists.

Conditional Activation in ADCs

The conditional activation concept has also extended to antibody-drug conjugates (ADCs). In a reported first-in-human phase I study, the ROR2-directed ADC ozuriftamab vedotin was described as a conditionally active construct in patients with advanced melanoma. Among five treated patients, objective responses were observed in four, including one durable complete response exceeding five years. Dose modifications were required for neuropathy and neutropenia, but treatment discontinuation for adverse drug reactions was not reported in this small cohort.

Computationally Designed Masks

Beyond specific antibody programs, recent work on de novo-designed peptide masks for miniprotein binders demonstrates how computational protein design is entering the conditional-activation landscape. In this approach, cleavable C-terminal masking helices were generated using diffusion-based backbone modeling and sequence design algorithms, achieving substantial affinity reductions in masked constructs and restoration of binding following protease cleavage. Notably, micromolar mask-binder affinity was sufficient for effective steric occlusion when tethered, highlighting how proximity effects can compensate for low intrinsic affinity. This highlights how rational and computational workflows may accelerate the generation of cleavable masking modules adaptable to diverse binder formats.