Receptors Unite: Agonism with Antibodies
Receptors Unite: Agonism with Antibodies
Virtually all FDA-approved antibodies neutralize block cell surface receptors. Why is building an activating (agonizing) antibody interesting? Why has it been so difficult and is that changing?
Forward
Last week, I wrote a thread about antibody-mediated receptor agonism—specifically in the context of a rare, hereditary disease. This made me interested to learn more generally about receptor agonism and why there’s such a dearth of antibodies being studied as agonists—despite the wide commercial success of antagonist antibodies. This brief essay outlines some of the reasons why. I’m eager to see what peculiar fusion-protein or non-standard antibody variants the field cooks up as these seem to be outperforming IgG antibodies in pre-clinical models of receptor agonism. Generally, our ability to activate signal pathways seems like a core capability in the therapeutic toolkit, though quite challenging to effectively unlock.
Receptor Pharmacology Basics
Receptor pharmacology has matured substantially since the 20th century. We know now that receptors are membrane-bound proteins that can directly or indirectly regulate biochemical processes inside cells. The molecules that bind receptors to produce an effect are called ligands. If they occur naturally inside an organism, we call them endogenous ligands. If they’re created or modified by human beings, we call them exogenous ligands.
Generally, ligands involved in receptor signaling are small molecules (e.g., neurotransmitters) or peptides (e.g., hormones). There are a few key aspects of receptor ligands worth mentioning here—affinity, selectivity, and activity.
Affinity—All ligands must first bind to their cognate receptors to alter signaling. Affinity refers to the strength of binding. You can also think of it as the probability that a ligand is bound to a receptor at a given time. Some ligands bind irreversibly while others bind transiently.
Selectivity—Ligands may interact non-specifically with multiple receptors or sub-types. Drugs rarely are completely specific to one receptor, instead they have relative selectivity for their target receptor. You can think of selectivity as the propensity a ligand has to bind its target versus other non-targets.
Activity—Once bound, activity refers to the effect a ligand has on downstream biochemical signaling. There are several principles underlying activity such as residence time (the duration a ligand occupies its receptor), conformational dynamics (how ligand binding alters the 3D structure of the receptor), the receptor’s constitutive functionality (is it baseline active or not), but I won’t linger on those for now.
We can sub-classify ligands, whether endogenous or exogenous, according to the effects they have on receptors. It’s tempting to say that receptor agonists upregulate pathways while antagonists downregulate pathways. This is an oversimplification that conflates receptor activation and pathway activation. Here’s an example.
Let’s say that exogenous Ligand X binds Receptor Y, outcompeting the endogenous Ligand Z. Typically, when Ligand Z binds Receptor Y, it decreases a cellular function. So when Ligand X binds, preventing the binding of Ligand Z, the cellular function increases. What type of ligand is Ligand X?
It’s still an antagonist. While true that the pathway was activated, Ligand X inactivated the receptor by preventing it from binding Ligand Z. Therefore, we label it an antagonist because of its effect on the receptor—not the signaling pathway.
Let’s walk through a few more terms.
Agonists—As drugs, agonists bind to their receptors and activate them. Agonists often mimic their endogenous ligand counterparts and typically bind to similar sites on receptors (though they don’t have to). Agonists promote cellular responses by initiating downstream signaling cascades. Importantly, signal transduction isn’t a binary “on-off” switch, it exists along a continuum.
Antagonists—These molecules bind to a receptor without activating it. Antagonists prevent the cellular response normally initiated by endogenous ligand binding.
Partial/Super Agonists—As mentioned, signaling has both directionality and magnitude. Agonists that activate receptors and only partially induce the native biological response are called partial agonists while those that induce a larger-than-normal response are called super agonists.
Inverse Agonists—Some receptors are constitutively activated. That is, they’re constantly in an activated state. When inverse agonists bind, they stabilize a receptor conformation that is inactive, thereby reducing its baseline activity. While antagonists competitively block binding from other ligands, inverse agonists reduce a receptor’s signal level below that when any ligand is bound.
Agonists in the Wild
Our cells are socialites. One way our cells chat with each other is through receptor agonists. Neurons fire neurotransmitters across synapses to pass along an electrical signal from cell to cell. Our endocrine system releases hormones into our bloodstream to ensure organs are behaving appropriately. It follows logically that when cell-to-cell communication goes wrong, and it often does, that scientists would look to an agonist’s endogenous counterpart as a viable starting point to design a synthetic (exogenous) version.
Of the many FDA-approved drugs that have agonistic mechanisms of action (MOAs), many are peptides that mimic their endogenous hormones. Take type 1 diabetes (T1D) for example. T1D results from cells in the pancreas being unable to produce insulin, a hormone that helps cells absorb glucose. As shown below, both short and long-lasting synthetic insulin medicines (agonists) used to treat T1D begin with the endogenous peptide scaffold and feature slight modifications to enhance drug-like properties.
The modern pharmacopeia is filled with drug modalities beyond just small molecules and peptides. Each method has advantages and disadvantages to weigh against the target and disease biology at hand. Conceptually, when designing a receptor agonist, it makes sense to me to narrow the chemical or peptide universe by beginning from molecular scaffolds that resemble the natural human ligand. But what if you didn’t? What drug modality is clinically-validated, has excellent drug-like properties, can be selective for cell-surface receptors, and stands to benefit from decades of developability and manufacturing infrastructure? Antibodies!
Why entertain such an idea? In the context of receptor agonism, small molecules may not be sufficiently large to stabilize a specific receptor conformation or outcompete an endogenous high-affinity, peptide ligand. They may also be too promiscuous. The fact that small molecules are cell-permeable isn’t directly advantageous here. Meanwhile, peptide mimetics suffer from poor stability relative to larger proteins like antibodies. There’s fascinating research being done to ameliorate all of these challenges, so I’m not saying we should ditch them in favor of antibodies. What I’m saying is that antibodies are an interesting (and perhaps non-intuitive) platform to develop receptor agonists.
Antibody Agonists—What Have You Done for Me Lately?
First commercialized in 1986, monoclonal antibodies (mAbs) are an established and still-surging therapeutic modality. Virtually all approved antibodies act as antagonists. Some of the most established antibody therapeutics are in the field of cancer immunotherapy. For example, pembrolizumab (Keytruda) topped the global sales charts in 2023—generating >$25 billion in revenue.
Keytruda is a programmed cell death protein 1 (PD-1) antagonist. As shown below, Keytruda inactivates the PD-1 receptor displayed on the surface of a T-cell, preventing the receptor from binding PD-L1/PD-L2. By doing so, Keytruda prevents the normal signaling pathway activated by PD-1—PD-L1/PD-L2 binding. If that pathway were to fire, it would result in the T cell’s death—preventing it from attacking a cancer cell.
Antibody-mediated receptor agonism has lagged behind significantly. While there is an orphan drug approval for a CD40 agonist antibody (sotigalimab), I couldn’t find any other FDA-approved agonist antibodies. In fact, there seem to be very few (~3) in late-stage clinical development.
Of the agonist antibodies in early development, most also focus on immunotherapy. Though a few years old, a 2018 study catalogued nearly fifty Phase I/II trials evaluating agonist antibodies in immunotherapy. Specifically, these assets go after receptors in the tumor necrosis factor superfamily (TNFSF). Walking through TNFSF, I think it becomes apparent why receptor agonism is such a tall order for drug development.
Superfamily Reunions
We’ve established that receptor agonism exhibits complex pharmacology and used this to explain the relative dearth in approved agonist antibodies. In this section, we’ll go more in depth, taking a moment to highlight the underlying complexity.
Many receptor classes don’t act alone. Instead, once bound to their cognate ligands, some receptors will come together through a process called dimerization. Once linked, the protein-ligand ensemble is stabilized into a membrane conformation that enables internal signaling.
When I drafted my last thread on Twitter (X), I wrote about Diagonal Therapeutics going after HHT with a bispecific antibody agonist. HHT causes dysregulation of a multi-member receptor complex (a superfamily), disabling it from properly signaling in the context its endogenous ligand (TGF-beta). This example showcases how some signaling pathways involve bringing together more than two receptors, as shown below.
The TNFSF represents another many-member receptor family involved in a key biological process—one that dozens of therapeutics companies are prosecuting with agonist antibodies.
With 19 ligands and 29 distinct receptors, the TNFSF is a large and diverse signaling apparatus. TNFSF members often for trimers between three different receptors. Expressed on the surface of T cells, the TNFSF has a key role when the body is marshaling an immune response against a tumor. Typically, optimal effector function from a T cell involves a TCR interaction with a peptide-MHC and a co-stimulatory signal from TNFSF receptors. Under normal circumstances, endogenous ligands can bind these receptors and fully activate T cells.
The tumor microenvironment (TME) is anything but normal. The TME can be highly immunosuppressive. In the context of TNFSF, my understanding is that tumors can secrete immunosuppressive cytokines or other inhibitor molecules to shut down native signaling. Ostensibly, introducing a competitive agonist to replace lost signaling can restore immune functioning in tumor-proximal T cells.
TNFSF is complex and I encourage those interested to read a full rundown of the target rationale. For our current sake, the important takeaway is that agonist pharmacology is complicated, requiring multi-specific or multi-valent interactions and conformational tuning.
Agonism—A Tall Design Task
Beyond the complex MOA, I suspect an unfortunate catch-22 is the reason agonist antibodies have lagged behind in the clinic. The inherent complexity could make investors balk at the stacked biological risks with antibody agonism, especially when there’s still so much wood to chop with antibody antagonism. Because of a relative lack of funding or interest, there’s a concomitant sluggishness with methods development. Agonist-specific discovery and development methods are critical because the host of antagonist-optimized tools isn’t sufficient to chip away at the complex pharmacology. Thus, the cycle continues.
Think about it this way. Developing an antagonistic antibody is already challenging. One must narrow down the enormous design space to arrive at a sequence that has high affinity for its target while simultaneously having desirable drug-like properties, low immunogenicity, and low affinity for closely related anti-targets. Recall that target engagement is a necessary, but not sufficient characteristic for an agonist. Agonists must often bind multiple receptors to induce a conformation that activates the receptor. In that sense, agonists are subsets within subsets.
Most high-throughput screening (HTS) technologies (e.g., phage display) as well as data-driven (read: ML) approaches for antibody design are focused on optimizing affinity. This isn’t enough for agonists where the MOA involves receptor clustering and activation. Binding assays alone aren’t sufficient for finding agonist antibody hits. We need robust functional and/or activity-based HTS tools. Indeed, this seems to be what Diagonal Therapeutics is doing—basing its antibody selection based on an activity measurement, as suggested below.
There is a set of antibody properties known to be highly relevant for designing effective agonists. These constitute the ‘knobs and levers’ drug developers can tune as they seek to navigate up the manifold of agonist effectiveness.
Valency—This refers to the number of antigen binding sites on an antibody. Typical IgG antibodies are bivalent. Several pre-clinical studies have shown that increasing valency enables greater receptor super-clustering and activation.
Specificity—Drug developers can imbue an antibody with affinity towards two different receptors. These so-called bispecific antibodies (bsAbs) are incredibly useful for agonizing heterodimers—situations where two different receptors must combine to induce signaling.
Isotype—There are several natural antibody isotypes (IgG 1-4) that differ by the number of cysteine bonds within hinge regions, leading to various degrees of flexibility between Fab regions.
Strangeness Abounds—Unconventional Antibody Formats
It’s clear that there exist higher-level optimization problems to be solved for transforming standard IgG antibodies into effective, drug-like receptor agonists. I was curious to see what out-of-the-box solutions researchers are dreaming up to rise to the occasion. Though certainly not exhaustive, here are two examples that serve to highlight how conformational tuning of standard IgG antibodies can induce agonism across multiple receptor types.
A portmanteau of “contortion” and “antibody”, the Contorsbody was published on in 2020. The authors hypothesized that significantly altering the 3D conformation of an established antibody—trastuzumab in this case—could convert its antagonist functionality into agonist functionality. As shown below, they introduce linkers into the antibody that cause the Fab arms to bend backwards on themselves, creating a highly compact, multivalent construct.
Unlike trastuzumab, which the authors describe as keeping Her2 receptors apart to prevent signal activation and subsequent cell proliferation, the Contorsbody version seems to aid in receptor dimerization. They note that the increased rigidity compared to standard IgG antibodies “has an increased potential to trigger cell surface antigen or receptor ‘cis’-dimerization without ‘trans’-bridging of cells or mere receptor blockade”. Indeed, treating a Her2+ breast cancer cell line with Contorsbodies showed a proliferative effect above baseline as compared to standard trastuzumab.
Continuing on this theme, Genentech published a paper on a constrained antibody format purpose-built for receptor agonism: i-shaped antibodies or “iAbs”. As shown below, iAbs have additional hinge linkages between Fab arms that bring them together in a parallel conformation. The authors cite rare cases of iAbs existing in the wild—specifically, they highlight broadly neutralizing HIV antibodies discovered in infected humans and rhesus macaques that share the linear, i-shaped format.
In the study, the authors created introduce linking mutations into standard IgG antibodies, locking them into three structurally-similar iAbs. They tested these constructs for agonistic potential in in vitro models across both OX40 (a key member of the TNFSF) and the IL-2 pathway given its relevance in cancer and autoimmunity. Generally, this work demonstrates that iAbs can act as potent agonists across multiple receptor types.
The authors used cell based assays to study receptor agonism. Wells contained a Jurkat cell line engineered to express OX40 and a luciferase reporter to indicate signal transduction. They tested wild-type (WT), contorsbodies, and different variations of iAbs. As shown below, it’s interesting to not only see concentration-dependent agonism with iAb variants, but also iAb agonism that nears ~75% of that induced by the native OX40 ligand (OX40L). That agonism differed materially depending on whether the iAb contained an Fc region is also curious.
In another example, they test iAbs on the cytokine-mediated IL-2 pathway, which typically undergoes heterodimerization. Similarly, the authors engineer a Jurkat-reporter cell line, but also test whether their designs could induce cell proliferation in NK and CD8+ T cells in vitro. Finally, they assess the gene expression profiles of T cells following exposure to various designs, highlighting the similarity between native IL-2 and constrained agonist formats like contorsbodies and iAbs.
Wrapping Up
The ability to precisely activate cellular pathways through selective receptor agonism seems like a foundational therapeutic capability. Some of the most broadly adopted therapeutics do just that. Small molecules and peptides have dominated this field and they’re not showing any signs of stagnation. At the same time, antibody therapeutics have come online, establishing themselves as critical antagonists in fields like immunotherapy. Indeed, this rush on antibodies has fueled a wave of high-throughput experimental platform development and pioneering ML methods in kind. There’s reason to think design principles from antibodies could generalize to other proteins, potentially serving as a boon for antibody as well as peptidomimetic agonists. Regardless of the format, agonism is an exciting frontier with seemingly renewed fervor.