GPR55: The Cannabinoid Receptor Pharma Can't Quite Drug

October 12, 2026

Photo by Ousa Chea via Unsplash.

Twenty-five years is a long time for a receptor to sit in limbo. GPR55 was cloned in 1999, picked up a cannabinoid-adjacent nickname by the mid-2000s, has been the subject of hundreds of papers linking it to pain, bone density, metabolic regulation, and even tumor biology -- and still, in 2024, there is not a single approved drug that targets it. Not one. Compare that to CB1, cloned in 1990, which had a selective antagonist (rimonabant) in clinical trials within a decade and on pharmacy shelves in Europe by 2006. GPR55 has had more time and arguably more scientific curiosity pointed at it, yet it hasn't produced a comparable pipeline.

The tension here isn't a lack of interest. Academic labs keep publishing on GPR55 because the biology keeps looking promising -- knockout mice show altered bone remodeling, the receptor shows up in inflammatory pain models, and it's tangled up in lipid signaling pathways that matter in oncology. Pharma's silence isn't about the disease targets being unattractive. It's about the receptor's pharmacology being genuinely strange, in ways that have made it hard to build the basic tools -- a clean selective agonist, a clean selective antagonist -- that drug discovery programs are built around.

That's the specific, nameable bottleneck this piece is about, and it's worth saying up front: it looks solvable in principle, not like a dead end. The question is how long it takes, who solves it, and what has to happen first. This is a long-horizon piece -- nothing here is an announcement of an imminent drug. It's an attempt to reason, using the history of how other orphan GPCRs got de-risked, about what a realistic 7-to-15-year path to a GPR55 therapeutic would actually look like, and where it could stall.

How a 1999 Orphan Receptor Became 'CB3'

Photo by kennethr via Pixabay.

GPR55 entered the scientific record in 1999, when Sawzdargo and colleagues cloned the human gene and mapped its expression across the brain. They found it concentrated in the hippocampus, caudate, putamen, hypothalamus, cerebellum, thalamus, and pons -- a distribution that immediately caught attention because it overlapped with, but didn't duplicate, the expression pattern of CB1. That alone was interesting. A receptor sitting in motor control and memory circuits alongside the classic endocannabinoid machinery, but not identical to it, suggested a parallel signaling system worth understanding on its own terms.

For years, though, nobody knew what activated it. GPR55 was an orphan G-protein-coupled receptor -- a sequence with no confirmed ligand -- and orphan GPCRs tend to sit quietly in databases until someone finds a reason to look harder. That's exactly what happened here: it took until the mid-2000s, once researchers started testing cannabinoid compounds against a panel of orphan receptors, for anyone to seriously propose GPR55 as a cannabinoid-responsive target. The nickname CB3 followed from that work, and it stuck in conference talks and review articles because it's a convenient shorthand.

But it's worth being precise about what that nickname is and isn't. CB3 is not an official designation from the International Union of Basic and Clinical Pharmacology (IUPHAR), the body that formally classifies receptor families. IUPHAR still lists GPR55 as a distinct, provisionally classified GPCR, not as a third member of the cannabinoid receptor family alongside CB1 and CB2. That distinction isn't pedantic -- it's a big part of why regulators and pharmacology departments treat the receptor cautiously. A name earned in the literature by convenience carries less weight in a regulatory filing than one backed by formal classification, and drug developers building a program around GPR55 can't lean on the assumption that it behaves like, or will be regulated like, a cannabinoid receptor. The overlapping-but-distinct brain expression pattern that made GPR55 interesting in 1999 is the same feature that's made it hard to pin down ever since.

Why GPR55 Doesn't Play by CB1/CB2 Rules

Photo by Esma Melike Sezer via Unsplash.

The clearest reason GPR55 resists easy classification is mechanical: it doesn't use the signaling machinery that defines CB1 and CB2. Both classic cannabinoid receptors couple to Gi/o proteins, which is why the standard pharmacology-lab assay for cannabinoid receptor activity is a cAMP inhibition assay -- activate the receptor, watch cyclic AMP levels drop. GPR55 doesn't do that. It doesn't couple to Gi, Gs, or Go at all. Instead, it signals through Gq and G12/13, an entirely different branch of the G-protein family.

Downstream, that different coupling sends the signal through different machinery too. GPR55 activation runs through phospholipase C, and separately through the ROCK/RhoA pathway -- a route associated with cytoskeletal remodeling and cell migration, more commonly seen in receptors for lysophospholipids than in classic neurotransmitter or cannabinoid receptors. Mechanistically, GPR55 has more in common with something like a lysophosphatidic acid receptor than with CB1.

That's not a small technical footnote -- it's the basis for an open argument among pharmacologists about whether GPR55 belongs in the cannabinoid receptor family at all. The case for keeping it there rests mostly on shared ligands: cannabinoid compounds bind it. The case against rests on everything downstream of binding looking like a different system wearing a cannabinoid-receptor costume. Several researchers have proposed GPR55 be studied and classified on its own terms, grouped with lysophospholipid-responsive receptors rather than cannabinoid ones, precisely because its signaling logic runs on different rails.

The practical fallout shows up earliest in drug discovery, in the assay stage. Pharmaceutical screening pipelines built around CB1 and CB2 lean on cAMP inhibition readouts because that's what those receptors reliably produce. Run a candidate compound library through a cAMP assay looking for GPR55 activity, and you can get a false negative on a compound that's actually quite active at the receptor -- it's just not producing the signal that assay is built to catch. Any serious GPR55 discovery effort has to build or adopt a different readout -- calcium flux, beta-arrestin recruitment, GTPgammaS binding -- and different labs have picked different ones, which (as the next sections get into) is a big part of why the published pharmacology doesn't agree with itself from paper to paper.

LPI, the Real Endogenous Ligand Nobody Talks About

Photo by julien Tromeur via Unsplash.

If GPR55 has a legitimate endogenous ligand -- the molecule your body actually produces to activate it under normal physiology -- the strongest candidate by far is lysophosphatidylinositol, or LPI. Oka and colleagues demonstrated in 2007 that LPI robustly activates GPR55, and critically, that this effect disappeared in cells where GPR55 had been knocked down. That knockdown control is what makes the finding credible rather than coincidental: the response tracks the receptor specifically, not some off-target pathway LPI happens to touch.

LPI itself is not exotic -- it's a known lipid signaling molecule with established roles in cell proliferation and migration, independent of any cannabinoid biology whatsoever. That's part of why GPR55 research has a foothold in oncology alongside its pain and bone-metabolism angles: a receptor sitting downstream of a proliferation-and-migration lipid is mechanistically plausible as a player in tumor biology, and several in vitro cancer cell line studies have investigated it on exactly that logic.

Here's where the identity crisis actually bites, though. Phytocannabinoids and synthetic cannabinoid compounds also activate or modulate GPR55 in various assays -- which is the entire reason the receptor got pulled into cannabinoid pharmacology in the first place. But those effects show up inconsistently. One lab's assay shows a compound as a clear agonist; another lab, using a different readout, shows the same compound as inactive or even opposing. The receptor doesn't behave like a dedicated cannabinoid receptor that happens to also respond to LPI. It behaves like a lipid-signaling receptor that cannabinoid-like molecules can perturb under some conditions and not others.

That dual-ligand reality is itself a structural problem for drug design, not just a confusing footnote. Rational drug discovery usually starts from a clean chemical scaffold tied to a well-characterized endogenous ligand, then iterates toward selectivity and potency. GPR55 hands medicinal chemists two chemically unrelated ligand classes -- a glycerophospholipid on one hand, cannabinoid-type small molecules on the other -- both hitting the same binding pocket region with different rules. Designing a selective compound against a receptor with two unrelated native-ish ligand classes is a materially harder problem than designing one against a receptor with a single well-behaved endogenous ligand, and that difficulty shows up directly in how little progress has been made toward a clean tool compound in twenty-five years.

The CBD/THC Paradox: Agonist, Antagonist, or Both?

Photo by felixioncool via Pixabay.

Read enough of the GPR55 literature back to back and a pattern emerges that would be funny if it weren't such a practical problem: compounds don't consistently do what they're supposed to do. LPI and the synthetic compound O-1602 generally behave as agonists across studies, which is reasonably solid ground. But rimonabant and AM251 -- both developed and validated as CB1 antagonists -- show up in several GPR55 assays behaving like agonists, not antagonists. A compound engineered to block one cannabinoid-adjacent receptor turns around and activates a different one entirely. That's the kind of result that, if it only happened once, you'd suspect an assay artifact. It's been reported enough times that it has to be taken as real receptor pharmacology.

CBD complicates things in the other direction. Several functional assays report CBD acting as a GPR55 antagonist, and that finding has become one of the more interesting proposed mechanisms for why CBD shows anti-inflammatory and anti-seizure effects that can't be fully explained by CB1 or CB2 engagement, since CBD's affinity for those two receptors is famously weak. If CBD is quietly doing meaningful work at GPR55, that would help explain part of its clinical profile independent of the classic endocannabinoid story -- but it's reported in some assay systems and not others, so it remains a proposed mechanism, not an established one.

THC's story is murkier still. Some assays report THC as a weak partial agonist at GPR55; others report no activity at all. The disagreement traces back to the same root problem flagged in the signaling section -- different labs are using different functional readouts. A beta-arrestin recruitment assay, a calcium flux assay, and a GTPgammaS binding assay are all legitimate ways to measure GPCR activation, but they don't always agree with each other, especially for a receptor like GPR55 that signals through unconventional pathways in the first place. A compound can look active on one readout and silent on another, and there's no current consensus on which assay is the correct standard.

This is the actual crux of pharma's hesitation, more than any single confusing result. A drug discovery program needs a reproducible, validated assay to run thousands of compounds through and trust the output. Without agreement on which assay tells the truth about GPR55 pharmacology, a company can't commit a discovery budget with any confidence that the hits it finds will replicate elsewhere -- and no pharmacology department wants to spend years chasing a target that keeps changing its behavior depending on how you ask the question.

What a Selective Agonist Would Actually Unlock

Photo by Roberto Sorin via Unsplash.

Set the assay confusion aside for a moment and look at what a clean, selective GPR55 agonist (or antagonist) would actually be useful for, because the preclinical case is not thin. Rodent and in vitro work has tied GPR55 to chronic pain signaling, to inflammatory bowel conditions, to bone metabolism -- GPR55 knockout mice show altered osteoclast function and bone density changes, which is a fairly specific and reproducible phenotype -- and to proliferation in some cancer cell lines, consistent with its activation by LPI. That's four distinct therapeutic areas with some degree of published support, which is a wider spread of potential indications than most newly characterized GPCRs get.

The reason none of that has translated into a program comes down to a pattern that's held across GPCR drug discovery for decades: a selective tool compound has to exist before a disease hypothesis becomes testable in a way pharma trusts. Look at how rimonabant came to exist in the first place -- SR141716 only became viable as an obesity drug candidate once CB1 pharmacology had been cleanly separated out from everything else cannabinoid compounds were doing, giving Sanofi a receptor-selective lever to pull. That selectivity is what let the obesity-drug push happen through the 1990s and into the 2000s, right up until rimonabant's 2008 market withdrawal in Europe over psychiatric side effects -- a cautionary note worth carrying into the next section, but a separate problem from the selectivity question itself.

Reasoning from that same historical pattern, a fair forecast is that GPR55 tool compound development keeps happening mostly in academic medicinal chemistry labs over the next 5 to 10 years, rather than inside a pharma pipeline. That's the track record for other orphan GPCRs that eventually mattered clinically -- GPR119 and GPR120, both metabolic-disease targets, took well over a decade between first pharmacological characterization and the first clinical candidates reaching trials, with academic and small-biotech chemistry doing the unglamorous early selectivity work that de-risked the target enough for anyone bigger to pay attention.

The honest counter-case deserves equal space, though. GPR55's dual Gq/G12-13 coupling and its overlap with lysophospholipid receptor biology might make it a harder selectivity problem than GPR119 or GPR120 ever were -- those targets didn't carry GPR55's two-unrelated-ligand-class complication. It's entirely possible the field stays tool-compound-limited well past the 15-year mark here. That's not pessimism for its own sake; it's a real, live possibility given how long the problem has already resisted solution, and anyone forecasting a GPR55 drug should hold that scenario alongside the more optimistic one.

The Business Case Pharma Is Waiting On

Photo by Maciej Drążkiewicz via Unsplash.

Pharmaceutical companies don't commit discovery budgets on scientific interest alone -- they wait for de-risking signals: a validated biomarker, a reproducible assay that multiple independent labs agree on, and ideally some early human proof-of-concept, even from a small academic or biotech trial. GPR55 currently has none of the three. That's the business-side mirror of everything the science sections above describe, and it's worth stating plainly rather than diplomatically: there is nothing yet that tells a pharma risk committee this bet is more likely to pay off than any other orphan receptor program competing for the same budget.

If a de-risking event does happen, the more plausible early movers are cannabis-adjacent biotechs, not traditional big pharma. Companies already working with CBD-derived compounds have built-in advantages here: regulatory familiarity with cannabinoid-interacting molecules, existing manufacturing and handling infrastructure for plant-derived and synthetic cannabinoid-like compounds, and scientific teams already fluent in the confusing literature described above. That familiarity lowers the activation energy for taking a GPR55 program seriously in a way a generalist pharma company doesn't have by default.

The realistic pathway, if a selective ligand does emerge from an academic lab, is licensing into a mid-size biotech rather than big pharma building an in-house discovery effort from scratch. That's the pattern GPR119 agonists for type 2 diabetes followed -- academic and small-biotech chemistry produced the selective compounds, and those assets moved through licensing deals to companies positioned to run the clinical development, well before any reached approval. There's no reason to expect GPR55 to break that pattern; if anything, the assay inconsistency described earlier makes it more likely that whoever solves the selectivity problem first will license out rather than build a full clinical program solo.

One more shadow is worth naming directly: rimonabant's 2008 withdrawal from the European market, driven by psychiatric adverse events including depression and suicidal ideation, made the entire pharmaceutical industry measurably more cautious about CNS-active, cannabinoid-receptor-adjacent targets. That caution wasn't specific to CB1 -- it rippled outward to anything sitting near the same signaling neighborhood, GPR55 included. It's a reasonable, evidence-based read that this institutional memory still shadows GPR55 programs today, even though GPR55's signaling mechanism and brain distribution differ meaningfully from CB1's. Risk committees don't always draw fine mechanistic distinctions when a prior drug class produced headline-generating harm; they remember the drug class, and are slower to fund anything that sounds adjacent to it.

Strip away the confusion and GPR55's story isn't really a story about failure. It's a field that has done a quarter-century of legitimate, often careful work -- cloning, expression mapping, knockout phenotyping, ligand characterization -- and arrived at a single missing reagent class standing between what's known and what's actionable: a selective agonist or antagonist clean enough to reclassify years of inconsistent, assay-dependent data into something a drug company can actually act on. That's a narrower problem than it looks from the outside, even if it's proven stubborn.

If the history of CB1 and other orphan GPCRs like GPR119 is any guide, the breakthrough -- if it comes -- is more likely to surface from an academic medicinal chemistry lab or a small biotech than from a major pharma discovery pipeline, and it will probably take most of the next decade to show up in a form anyone outside the field notices. That's not a hedge; it's the pattern these programs have followed before, repeatedly, and there's no strong reason to expect GPR55 to skip the line.

The single clearest signal worth watching for isn't a press release or a funding round -- it's duller and more useful than that. When multiple independent labs converge on the same validated GPR55 assay and start publishing results that agree with each other, that's the moment the pharmacology stops being a moving target. That's the bellwether. Commercial interest tends to follow reproducibility, not the other way around, and right now reproducibility is still the thing GPR55 research is missing.