pocketz.doi.bio/2026-06-08_163204_linkedincom_what-multivalent-gpcr-conjugates-quietly-tell-us-our-blind-damko-cpaie

What multivalent GPCR conjugates quietly tell us about our blind spots | LinkedIn

URL: https://www.linkedin.com/pulse/what-multivalent-gpcr-conjugates-quietly-tell-us-our-blind-damko-cpaie/

Saved: 2026-06-08T23:32:04.468Z

Sachdev S, Cheloha RW. Multivalent antibody-based conjugates as new tools for tailored modulation of G protein-coupled receptors. Br J Pharmacol. 2026 What multivalent GPCR conjugates quietly tell us about our blind spots Ermelinda Damko Sr. Scientist @ Regeneron | Biotechnology R&D & Scientific AI | Founder, “Data‑Rich, Insight‑Poor?” | Science‑first forum on the epistemic and ethical limits of biological data and human‑centred technology June 3, 2026

Data‑Rich, Insight‑Poor? — LXIV

If any target class deserves to be called “data-rich,” it is GPCRs. Decades of dose–response curves, multiplexed signalling readouts, and an ever-growing structural atlas have densely populated the orthosteric landscape. Yet the stubborn realities of limited subtype selectivity, context-dependent bias, and narrow therapeutic windows make it clear that we remain “insight-poor” about the questions that most matter for drug design.

Sachdev and Cheloha’s review of multivalent antibody- and nanobody-based GPCR conjugates is best read against that backdrop. It does not add yet another layer of orthosteric SAR; instead, it shows how re-engineering ligand topology and valency can turn GPCRs from highly instrumented targets into systems we can actually interrogate. By forcing ligands to respect the spatial organization of receptors on the cell surface, these constructs reveal how much of our GPCR data has been generated under conditions that ignore the very features we now most want to control—assembly, allostery, and context.

Dashboard thinking: rich orthosteric data, thin mechanistic insight

Classical GPCR discovery is organized around the orthosteric site. We optimize ligands for potency and sometimes bias, run them through signalling panels, and celebrate when the “dashboard” of cAMP, Ca²⁺, β‑arrestin, and internalization looks favourable. Implicitly, we assume that a dense map of ligand behaviour in that pocket will translate into control over biology.

This is where the “data-rich, insight-poor” problem emerges. Those dashboards largely ignore:

How receptors are spatially organized: as monomers, homo- and heterodimers, or higher-order clusters. Where signalling occurs: at the plasma membrane versus endosomes or other compartments. How allosteric surfaces—extracellular loops, lipid-facing grooves, and oligomer interfaces—shape the conformational landscape and signalling trajectory.

The result is an internal-consistency trap: we get better and better at ranking ligands within our assay framework, without substantially improving our ability to predict which design choices will matter in vivo.

Multivalency as a deliberate experiment on receptor organization

Multivalent ligands—bivalent orthosteric ligands, cis- and trans-bitopic compounds, and biologic-based conjugates—break this trap by design. They are not just “more ligands per molecule”; they are spatial hypotheses rendered in chemistry.

Sachdev and Cheloha emphasize two axes of classification:

Valency: bivalent ligands (two orthosteric pharmacophores) versus bitopic ligands (orthosteric plus a secondary pharmacophore). Topology: cis-bitopic ligands, which engage orthosteric and secondary sites on the same protomer, versus trans-bitopic ligands, which span distinct protomers in a dimer or heteromer.

Classical work at muscarinic, adenosine, serotonin, dopamine, and bombesin receptors already showed that:

Linker length and rigidity determine whether dual engagement occurs or whether pharmacophores “toggle” between sites. Properly tuned geometries can create emergent subtype selectivity and biased signalling that co-administered monovalent ligands cannot replicate.

Multivalency, in other words, converts receptor organization from a nuisance variable into an explicit design dimension. Each construct asks: “What if the receptor has to be in this configuration for the ligand to work?” The answer is encoded directly in its pharmacology.

Structural biology: from static pockets to constrained ensembles

The GPCR structural revolution has given us exquisite snapshots of orthosteric and allosteric engagement, often at atomic resolution. Yet most of those structures capture a single receptor protomer and focus on endpoint active or inactive states.

Multivalent designs implicitly assume something richer:

That receptors exist as ensembles of conformations and assemblies. That altering distances and orientations between binding events changes which routes through that landscape are accessible.

When a nanobody docks on an extracellular epitope and a tethered ligand reaches into the orthosteric pocket or an allosteric vestibule, it does more than bind two places at once. It restricts which receptor states and oligomeric configurations can be productively engaged, and it changes the kinetics of movement between them.

The functional readouts—potency, bias, trafficking, receptor-pair selectivity—are therefore structural statements. They tell us which topologies are compatible with signalling in a way that monovalent ligand screens seldom can.

Allosteric pharmacology: maps in hand, experiments missing

We have made real progress in mapping allosteric pockets: identifying conserved “hot spots,” cryptic sites that appear upon activation or dimerization, and intracellular allosteric centres. But our experiments still largely treat allostery as “orthosteric plus a second pocket,” usually within a single protomer.

The constructs in Sachdev and Cheloha’s review expand this view. They show that:

Allosteric control can be exerted by the way we localize a ligand to the receptor surface (for example, via a nanobody) as much as by adding a second small-molecule warhead. Oligomer interfaces and co-expression patterns can become allosteric selectors: a trans-bitopic ligand has high avidity only when the right receptor pair is present and geometrically compatible. Combining orthosteric chemistry, allosteric pockets, and assembly-specific docking into one design problem yields effects—like receptor-pair-specific activation and logic gating—that monovalent strategies almost never achieve.

We are not short of allosteric maps; we are short of perturbations that truly test them.

Biologics as topology encoders

Antibodies and nanobodies are often discussed in terms of affinity and specificity, but in this context their most important property is geometric.

IgGs bring bivalency, size, and long half-life. When conjugated to peptides or small molecules (as in AMG 133), they become macromolecular scaffolds that define where in the body a ligand accumulates, which cells see it, and how long the receptor–ligand complex persists. Nanobodies introduce compact, convex paratopes capable of reaching concave or recessed epitopes, including vestibules and deep pockets. They can be arranged in series, in parallel, or fused to ligands to generate biparatopic, bitopic, or multivalent architectures with nanometre-scale control of geometry.

From the “data-rich, insight-poor” vantage point, biologics are valuable not because they simply “bind better,” but because they let us decide which spatial hypotheses our ligands will test.

Bioconjugation: the chemistry that determines what we can learn

Our ability to draw mechanistic insight from multivalent constructs depends critically on how we build them.

Stochastic Lys or Cys modification is chemically convenient but produces heterogeneous mixtures with poorly defined geometry. Such conjugates can still be active but offer muddy structure–function relationships. Site-specific strategies—engineered cysteines, noncanonical amino acids, enzymatic tags like sortase A—allow us to fix copy number, attachment site, and linker identity. Combined with bio-orthogonal reactions, they support systematic variation of geometry at constant scaffold.

For multivalent GPCR ligands, this precision is the difference between “more data” and “more insight.” If attachment sites, valency, and linker chemistries are all floating, interpreting changes in bias or receptor selectivity becomes guesswork. When they are controlled, each change can be linked to a specific geometric alteration.

Case study 1: PTH1R – when “neutral binders” become engines of bias

The PTH1R system is a particularly clear example of how multivalency can convert data into insight. Short PTH fragments such as PTH(1–11) and PTH(1–14) are modest, relatively non-selective agonists at PTH1R and PTH2R. Neutral nanobodies that bind the PTH1R ECD without strong signalling effects look uninformative in standard dashboards.

Once tethered together, however:

The conjugates become dramatically more potent at PTH1R than the free peptide. Selectivity over PTH2R increases substantially. Signalling becomes highly biased: sustained Gαs–cAMP with minimal β‑arrestin recruitment, limited Gq engagement, and altered internalization, making these among the most biased PTH1R agonists reported.

Crucially, this transformation is achieved not by changing the peptide’s core structure, but by anchoring it to a receptor-specific nanobody at a defined epitope. The data tell us that geometry and assembly—not just orthosteric chemistry—are primary determinants of bias here. Without multivalent constructs, that insight would remain latent.

Case study 2: AMG 133 – from combination therapy to embodied topology

AMG 133 (maridebart cafraglutide) is rightly celebrated as a clever combination of GIPR antagonism and GLP‑1R agonism in a single long-acting antibody–peptide conjugate. But from a “data-rich, insight-poor” perspective, its deeper value is as a topological intervention.

The anti-GIPR antibody defines where the conjugate goes and which GIPR-expressing cells see high local concentrations. The tethered GLP‑1 analogues experience a spatial and kinetic environment defined by GIPR internalization and trafficking, not by GLP‑1R alone. In co-expressing cells, the construct effectively enforces a defined relationship between GLP‑1R and GIPR signalling, including co-internalization and endosomal cAMP, that free peptides do not.

The clinical weight loss and metabolic improvements observed in early trials are therefore readouts of a specific geometric and pharmacological hypothesis, not just of “more GLP‑1R agonism plus GIPR blockade.” That is mechanistic information we did not have before.

Case study 3: Nanobody bitopics and GPCR pairs – logic as data

Recent work on nanobody-based bitopic conjugates targeting small-molecule-binding GPCRs pushes the idea of “designed perturbations” even further. By linking a small-molecule agonist to a nanobody that recognizes an engineered N-terminal epitope on the same or a partner receptor, these constructs:

Activate only tagged receptors, or specific receptor pairs, with high potency. Can be wired so that signalling is strong only when both the nanobody target and the small-molecule target are present in the same cell, effectively implementing an AND gate. Show prolonged signalling compared with free small molecules, due to altered binding kinetics and trafficking.

From a data standpoint, each construct is a topology-aware perturbation: it tells us not just that a ligand can bias signalling, but that bias and potency can be made conditional on receptor co-expression and spatial arrangement. That is the kind of context-dependence our monovalent dashboards hint at but rarely resolve.

GPCR-targeting therapeutics: bitopic nanobody conjugates vs AMG 133–like bispecifics

Thinking in pipeline terms, bitopic nanobody–ligand conjugates and AMG 133–style bispecific antibody–peptide conjugates represent two complementary ways of transforming data into insight and therapy.

Bitopic nanobody conjugates are high-resolution probes. They let us ask finely grained questions about how specific geometries and receptor pairs shape signalling, bias, and trafficking. They are ideal for mapping receptor organization–phenotype relationships and for de-risking designs before committing to clinically heavy formats. AMG 133–like constructs are translational testbeds. They show how a particular combination of targets, geometry, and PK behaves in real patients. Their success or failure feeds back into which topological hypotheses are worth pursuing at scale.

In a genuinely insight-driven pipeline, these would not be separate silos. Nanobody bitopics would be used systematically to explore design space; a subset of those designs would then be “promoted” into AMG 133–like formats where long-acting PK, manufacturability, and regulatory familiarity matter.

Biased agonism and off-target toxicity: from curves to constraints

Biased agonism has generated vast amounts of data and sophisticated analysis frameworks, but the core problem persists: bias measured in one expression system often shifts in another. The missing piece is usually context—expression level, coupling partners, trafficking.

Nanobody tethering and multivalency offer a way to build that context into the ligand itself:

By anchoring ligands to defined epitopes, we constrain which receptor states and assemblies can be engaged. By altering internalization and recycling, we determine where and for how long key pathways are activated. By targeting specific receptors or receptor combinations on specific cells, we limit where biased signalling can occur at all.

For targets where efficacy and toxicity map to different pathways or compartments (opioids, some chemokines, cardiovascular GPCRs), this suggests a path to fewer off-target effects that relies less on the intrinsic “cleanness” of the chemotype and more on the geometric and cellular constraints imposed by the scaffold.

Translation and regulation: turning multivalency into a tractable story

The flip side of multivalency’s richness is translational complexity. Nanobody-based constructs bring half-life, manufacturability, and immunogenicity challenges; IgG-based bispecifics and antibody–peptide conjugates must satisfy evolving regulatory expectations for mechanism, potency assays, and product quality.

From a “data-rich, insight-poor” standpoint, the risk is that we drown in CMC and safety data without preserving the link back to our original mechanistic questions. The way to avoid this is to treat topology and mechanism as part of the product definition:

Pre-define which aspects of geometry and assembly are critical to mechanism. Design conjugation and scaffold choices to make those features assayable and manufacturable. Build potency and functional assays that reflect the logic gates and biased profiles we care about, not just generic readouts.

Done well, the same design features that make these constructs mechanistically interesting become the ones that make them scientifically and regulatorily legible.

From dashboards to designed perturbations

The deeper message of Sachdev and Cheloha’s review is that our GPCR problem is not a data deficit; it is a perturbation deficit. We have spent years generating large, homogeneous datasets under a narrow set of perturbations (monovalent orthosteric or simple allosteric ligands), then wondered why our models of receptor organization and tissue-specific pharmacology remain weak.

Multivalent GPCR conjugates—nanobody–peptide, antibody–peptide, nanobody–nanobody, nanobody–small-molecule—offer a different path. Each is a designed perturbation of receptor topology, assembly, and context. When read that way, the data they produce are not just “more curves”; they are direct tests of how GPCRs work in cells and organisms.

If the goal of “data-rich, insight-poor?” is to argue for better experiments rather than more of the same, then this is what better looks like in GPCR drug discovery: fewer, more carefully designed molecules that encode geometry, allostery, and context into their very architecture.

References

Sachdev S, Cheloha RW. Multivalent antibody-based conjugates as new tools for tailored modulation of G protein–coupled receptors. Br J Pharmacol. 2026;1–20. doi:10.1111/bph.70519. Sachdev S, Creemer BA, Gardella TJ, Cheloha RW. Highly biased agonism for GPCR ligands via nanobody tethering. Nat Commun. 2024;15:4687. Sachdev S, Roy S, Cheloha RW. Nanobody-based conjugates targeting small molecule-binding GPCRs and exhibiting logic-gated signaling. PLoS Biol. 2025;23(11):e3003424. Huang X, Roth BL. Nanobody–drug conjugates for targeting specific GPCR pairs. PLoS Biol. 2025;23(11):e3003483. Laffitte A, et al. A GIPR antagonist conjugated to GLP-1 analogues promotes weight loss with improved metabolic parameters in preclinical and phase 1 settings. Nat Metab. 2024;6(2):290–303. Maurel D, Ferré S, et al. Unmasking allosteric-binding sites: novel targets for GPCR drug discovery. Br J Pharmacol. 2022;179(18):4048–4064. Rodríguez D, Tóth G, et al. Conservation of allosteric ligand binding sites in G protein–coupled receptors. J Med Chem. 2022;65(22):15092–15108. Cao C, et al. Covalent allosteric modulation: an emerging strategy for GPCRs. Pharmacol Ther. 2020;211:107537. Skiniotis G, et al. Structural findings reveal how distinct GPCR ligands create different levels of activation. Nature. 2025;623:112–120. Wacker D, Stevens RC, Roth BL. G protein–coupled receptors (GPCRs): advances in structures and pharmacology. Signal Transduct Target Ther. 2024;9:112. Maridebart cafraglutide (AMG 133) ligand page. IUPHAR/BPS Guide to PHARMACOLOGY. Accessed 2026. Amgen Inc. The Story of Maridebart Cafraglutide (MariTide). Amgen.com. 2026. Ma X, et al. Therapeutic antibodies targeting G protein-coupled receptors. SSRN. 2026; preprint. U.S. Food and Drug Administration. Bispecific Antibody Development Programs: Guidance for Industry. FDA; 2021.

Data-Rich, Insight-Poor? is a forum on the gap between what we measure and what we understand. Comments, challenges, and contradictory data are welcome.