In 2016, an international team of 23 researchers finally solved the 3D crystal structure of the human CB1 receptor at 2.8 angstrom resolution — 26 years after the gene was cloned — enabling scientists to see exactly how THC, endocannabinoids, and synthetic cannabinoids fit into the receptor's binding pocket.
Readers interested in how structural biology advances drug design, the technology behind understanding receptor-drug interactions, or the cutting edge of cannabinoid pharmaceutical development.
2.8 Åresolution — for the first time, scientists could see the exact 3D shape of the CB1 receptor's binding pocket, revealing how THC and other cannabinoids physically fit into the most abundant receptor in the brain
The Backstory
For twenty-six years, the CB1 cannabinoid receptor existed as a sequence of letters — 472 amino acids, spelled out in a gene database. Scientists knew the parts list. They knew the receptor was the most abundant GPCR in the brain. They knew it bound THC, anandamide, and 2-AG. They knew its pharmacology in exhaustive detail.
But nobody had ever seen it.
Knowing a gene sequence without knowing the 3D structure is like having every ingredient in a recipe without knowing what the dish looks like. You can predict some things. You can't design a drug with precision. In October 2016, twenty-three researchers across seven institutions in three countries finally took the photograph.
Why It Took 26 Years
Solving a crystal structure sounds straightforward: purify the protein, grow crystals, shoot X-rays at them, compute the 3D shape from the diffraction pattern. In practice, for membrane proteins like GPCRs, every step is a nightmare.
CB1 is embedded in a cell membrane. Remove it from the membrane and it loses its shape. It's flexible — constantly shifting between conformations. It has seven transmembrane helices that need to be just right. And to grow a crystal, you need billions of identical protein molecules to stack into a perfect lattice. A protein that won't sit still won't crystallize.
By 2016, decades of GPCR crystallography had produced solutions to each problem — but each had to be adapted specifically for CB1:
How They Did It
How They Photographed the Most Abundant Receptor in the Brain
Design a stabilizing drug
Alexandros Makriyannis at Northeastern University spent 40 years designing synthetic cannabinoid tools. For this project, his lab synthesized AM6538 — a custom antagonist designed not as a medicine but as a crystallography tool. AM6538 binds CB1 with extremely high affinity and locks it in a rigid, inactive conformation.
Without AM6538, CB1 was too flexible to crystallize. The drug was the key to the entire project.
Engineer the protein
The intracellular loop 3 region of CB1 is highly flexible — bad for crystallization. The team replaced it with T4 lysozyme, a stable protein that acts as a rigid scaffold. This fusion protein maintains CB1's binding properties while reducing conformational flexibility.
T4 lysozyme fusion is a standard technique developed for GPCR crystallography
Reconstitute in lipid
CB1 was reconstituted into lipidic cubic phase (LCP) — a gel-like matrix of lipids that mimics the cell membrane environment. This keeps the protein folded correctly while allowing crystal contacts to form.
LCP crystallization was pioneered for GPCRs by Vadim Cherezov and has enabled most GPCR structures
Grow crystals and collect data
Microcrystals were grown over weeks. X-ray diffraction data were collected at synchrotron facilities — particle accelerators that produce intense, precisely focused X-ray beams. The crystals diffracted to 2.8 angstrom resolution.
2.8 Å means individual atoms are nearly resolved — sufficient to see amino acid side chains and drug contacts
Solve and refine the structure
The 3D electron density map was computed from the diffraction pattern using molecular replacement. The atomic model was built into the density and refined iteratively. Functional studies and molecular docking validated the structural interpretation.
Hua et al. (2016), Cell 167:750-762
The scale of the effort reflects how modern structural biology works. This wasn't one lab with one technique. It was a synthetic chemistry lab (Makriyannis, Northeastern) making the drug, a structural biology institute (iHuman, ShanghaiTech) growing crystals and solving the structure, a functional pharmacology lab (Bohn, Scripps) validating the biology, and a computational group (Kufareva, UCSD) modeling drug binding — all coordinated across continents.
23
authors from 7 institutions across 3 countries — a collaboration spanning synthetic chemistry (Boston), structural biology (Shanghai), functional pharmacology (La Jolla), and computational modeling (San Diego). Modern structural biology at this level requires an army.
For context: the THC isolation paper had 2 authors. The CB1 cloning paper had 5. The anandamide paper had 10. The crystal structure needed 23.
Hua et al. (2016), Cell
The Drug That Made the Picture Possible
The unsung hero of this paper is a molecule most people will never hear of: AM6538.
Alexandros Makriyannis, the George Behrakis Chair in Pharmaceutical Biotechnology at Northeastern University, has spent over four decades designing synthetic cannabinoids. Not recreational drugs — research tools. Molecules precisely engineered to bind cannabinoid receptors in specific ways, to serve as probes for understanding how the system works.
AM6538 was his masterpiece of purpose-built pharmacology. It's a CB1 antagonist — structurally related to rimonabant — but designed with a single goal: bind CB1 so tightly and so stably that the receptor would stop moving long enough to form a crystal. The drug isn't meant to be a medicine. It's meant to be a molecular splint.
It worked. AM6538 locked CB1 into a rigid inactive conformation. Crystals formed. X-rays scattered. For the first time in history, the 3D shape of the brain's cannabinoid receptor emerged from the data.
What the Structure Reveals
Before 2016 (Gene Sequence Only)
- Knew the 472 amino acid sequence
- Knew it was a 7-transmembrane GPCR
- Could predict general binding regions by mutation studies
- Drug design was trial-and-error — synthesize, test, iterate
- Couldn't explain why certain drugs fit and others didn't
- No ability to computationally screen virtual compound libraries
- Allosteric sites were hypothetical
Parts list without floor plan
After 2016 (Atomic Structure)
- 3D shape of the complete binding pocket at 2.8 Å
- Exact amino acid contacts for antagonist binding mapped
- Computational docking predicts how THC, synthetics, endocannabinoids orient
- Virtual screening of millions of compounds before synthesis
- Allosteric sites identified — explaining CBD's mechanism
- Structural basis for designing biased agonists and peripherally-restricted drugs
- Template for understanding activation conformational changes
Atomic photograph enabling precision medicine
Hua et al. (2016), Cell; structure-based drug design literature
The structure revealed several features that couldn't be predicted from the sequence alone:
The binding pocket is deep and narrow — a long, tunnel-like cavity extending from the extracellular surface into the transmembrane core. THC slides in lengthwise, its tricyclic ring system making contacts with specific amino acids lining the tunnel walls. This geometry explains why small changes in cannabinoid structure (a pentyl chain vs a propyl chain, as in THC vs THCV) produce such different pharmacological profiles.
The antagonist-binding mode showed that AM6538 occupies the same tunnel but plugs it differently than agonists — blocking the conformational changes needed for receptor activation. This is the structural explanation for what Pertwee had characterized pharmacologically: antagonists don't just compete for the same site, they prevent the receptor from changing shape.
The structure also identified regions consistent with allosteric binding sites — locations away from the main pocket where modulators like CBD could bind and alter the receptor's response to THC without directly competing for the same spot. This structural feature directly supports the negative allosteric modulator model for CBD.
From Photograph to Drug Design
The crystal structure didn't just satisfy scientific curiosity. It launched a new era of cannabinoid drug development.
Research Timeline
The Structural Biology Era of Cannabinoid Research
CB1 gene cloned (Matsuda, Bonner)
The amino acid sequence — parts list without a floor plan
25 years of pharmacology without a structure
Drug design by trial and error — synthesize, test, iterate
CB1 crystal structure solved — antagonist-bound (Hua et al.)
2.8 Å resolution. First photograph of the receptor. Published in Cell.
Companion CB1 structure with taranabant (Shao et al.)
Independent validation — different antagonist, consistent architecture. Published in Nature.
Agonist-bound CB1 structures reported
Reveals conformational changes during receptor activation — how the shape shifts when THC or 2-AG bind
CB2 receptor crystal structure solved (Stevens group)
Both cannabinoid receptors now structurally characterized
Computational drug design using structural coordinates
Virtual screening of millions of compounds, structure-guided optimization of leads, rational design of allosteric modulators and biased agonists
Hua et al. (2016), Cell; Shao et al. (2016), Nature; Li et al. (2019)
Before the structure: Drug design meant synthesizing hundreds of compounds, testing each one in cell assays or animals, and iterating. A single drug candidate might take years of trial and error.
After the structure: Computational chemists can dock millions of virtual molecules into the 3D binding pocket, score how well each fits, and select only the best candidates for actual synthesis and testing. This doesn't replace experiments — but it compresses years of screening into weeks of computing.
The practical applications include:
- Peripherally-restricted CB1 antagonists — using the structure to design molecules that block CB1 in the gut and liver (for metabolic disease) but can't cross the blood-brain barrier (avoiding rimonabant's psychiatric side effects)
- Biased agonists — molecules designed from the structure to activate therapeutic signaling pathways (pain relief) without activating pathways that produce psychoactivity or tolerance
- Allosteric modulators — compounds targeting sites identified in the structure that fine-tune receptor activity rather than fully activating or blocking it
The Arc From Molecule to Atom
This study completes a 52-year arc in cannabinoid science:
1964: Mechoulam identifies the THC molecule — the key.
1988-1990: Howlett proves a receptor exists, then Matsuda clones the gene — the lock.
1992-1997: Anandamide and 2-AG are found — the body's own keys.
2006-2008: Pacher maps the therapeutic landscape, Pertwee characterizes the pharmacology — understanding how the lock and keys interact.
2016: This paper — the first photograph of the lock, at atomic resolution.
Each step required the one before it. You can't photograph a receptor you haven't cloned. You can't clone a receptor you haven't proven exists. You can't prove a receptor exists without the molecule it responds to. Fifty-two years from organic chemistry to structural biology, each generation handing the next the tools it needed.
Related Research
From Gene to Photograph — The CB1 Story
The crystal structure sits at the end of a 26-year journey from cloning the CB1 gene to seeing its 3D shape.
Structure of a cannabinoid receptor and functional expression of the cloned cDNA
Matsuda et al. (1990)
The gene — 472 amino acids spelled out, but no 3D shape. 26 years later, the photograph.
The diverse CB1 and CB2 receptor pharmacology of three plant cannabinoids
Pertwee (2008)
The pharmacology the structure now explains at atomic resolution — partial agonism, antagonism, allosteric modulation
Molecular characterization of a peripheral receptor for cannabinoids
Munro et al. (1993)
CB2 — the sibling receptor whose crystal structure followed in 2019, also from Stevens' group
Isolation, Structure, and Partial Synthesis of an Active Constituent of Hashish
Gaoni & Mechoulam (1964)
THC — the molecule whose binding to CB1 can now be visualized atom by atom
Why did it take 26 years to see the CB1 receptor's 3D shape?
Membrane proteins like GPCRs are among the hardest structures in biology to solve. CB1 is embedded in a cell membrane, it's flexible (constantly shifting between conformations), and it tends to denature when removed from its native environment. The breakthrough required three key innovations: a custom-designed drug (AM6538) that locked the receptor in a stable shape, a protein engineering trick (T4 lysozyme fusion) to reduce flexibility, and a specialized crystallization technique (lipidic cubic phase) that mimics the membrane environment. All of these technologies had to be developed, tested, and optimized over many years.
What can scientists do with the crystal structure that they couldn't do before?
Before the structure, drug design for CB1 was trial and error — synthesize a compound, test it, iterate. With the 3D structure, computational chemists can dock millions of virtual molecules into the receptor's binding pocket and predict which ones will fit well before synthesizing anything. This enables rational drug design for next-generation cannabinoid therapeutics: peripherally-restricted antagonists for metabolic disease, biased agonists for pain without psychoactivity, and allosteric modulators that fine-tune rather than fully block the receptor.
What the researchers found
The 2.8 angstrom crystal structure of human CB1 was solved in complex with AM6538, a stabilizing antagonist specifically designed for this purpose. The structure revealed the receptor's seven-transmembrane architecture, the precise geometry of the orthosteric binding pocket, and critical amino acid contacts for antagonist binding. Combined with functional studies and molecular modeling, the structure provided insight into how THC and other natural cannabinoids orient within the binding pocket, why synthetic cannabinoids bind differently, and where allosteric modulation sites are located. The structure enables rational drug design — the ability to computationally predict how new compounds will interact with CB1 before they are synthesized.
Why it matters
Knowing a receptor's gene sequence tells you the amino acid order but not the 3D shape. For drug design, shape is everything — you need to know the exact geometry of the binding pocket to design molecules that fit precisely. This structure transformed cannabinoid drug discovery from trial-and-error pharmacology to structure-based rational design. It revealed why THC fits CB1, why synthetic cannabinoids bind differently, where allosteric sites are (relevant to CBD's mechanism), and provided a template for computationally screening thousands of potential drug candidates before synthesizing any.
How the study worked
The human CB1 receptor was engineered with a T4 lysozyme fusion to stabilize the intracellular loop region and expressed in insect cells (Sf9). The receptor was bound to AM6538, a custom-designed high-affinity antagonist synthesized by the Makriyannis laboratory at Northeastern University, which locked the receptor in a stable inactive conformation. Crystals were grown using lipidic cubic phase (LCP) crystallization — a technique where the protein is reconstituted into a lipid bilayer that mimics its native membrane environment. X-ray diffraction data were collected at synchrotron facilities and processed to 2.8 angstrom resolution. The structure was solved by molecular replacement and refined using standard crystallographic methods. Complementary functional studies (signaling assays) and computational molecular docking validated the structural findings.
What this study cannot tell us
The structure captures CB1 in one conformation — antagonist-bound, inactive state. The active (agonist-bound) conformation was solved subsequently. Crystal structures are static snapshots of dynamic proteins and may not capture the full range of conformational states. The T4 lysozyme fusion and crystallization conditions introduce non-physiological constraints. The resolution (2.8 Å) is good but not atomic-level — some side-chain positions are approximate. The structure was solved in detergent/lipid conditions, not in a native cell membrane.
How to read the evidence
Rated strong because X-ray crystallography provides direct experimental evidence of protein structure at near-atomic resolution. The structure has been independently validated by a companion structure from another group (Shao et al., Nature 2016) and by subsequent agonist-bound structures. Published in Cell, one of the most prestigious journals in biomedical science.
When this study was published
Published in 2016, this 10-year-old structure remains the foundational reference for CB1 receptor architecture. Subsequent structures (agonist-bound, cryo-EM) have refined and extended the picture but build on this initial framework. The structural coordinates (PDB) are actively used in computational drug design worldwide.
The bigger picture
The CB1 crystal structure completed a 52-year arc from molecule to atomic-resolution receptor: THC characterized (1964) → receptor binding proven (1988) → gene cloned (1990) → 3D structure solved (2016). With the structure in hand, the field entered the computational era — molecular docking, virtual screening, structure-activity relationships at atomic resolution. The CB2 structure followed in 2019 (also from Stevens' group). Agonist-bound CB1 structures have since been solved, revealing the conformational changes that occur during receptor activation. The practical impact is in drug design: next-generation CB1 modulators — peripherally-restricted antagonists, biased agonists, allosteric modulators — are being designed directly from these structural coordinates.
Questions still open
- How does CB1's shape change when activated by THC or endocannabinoids versus when blocked by an antagonist?
- Can the allosteric sites revealed by the structure explain how CBD modulates CB1?
- Will structure-based drug design produce CB1-targeting drugs that avoid the psychiatric side effects of rimonabant?
Common questions
Why did it take 26 years to see the CB1 receptor's 3D shape?
What can you do with a crystal structure that you can't do with just a gene sequence?
Read the original research
Crystal Structure of the Human Cannabinoid Receptor CB1
Cell, 167(3), 750-762.e14
Citation
Hua, T; Vemuri, K; Pu, M; Qu, L; Han, G W; Wu, Y; Zhao, S; Shui, W; Li, S; Korde, A; Laprairie, R B; Stahl, E L; Ho, J H; Zvonok, N; Zhou, H; Kufareva, I; Wu, B; Zhao, Q; Hanson, M A; Bohn, L M; Makriyannis, A; Stevens, R C; Liu, Z J. (2016). Crystal Structure of the Human Cannabinoid Receptor CB1. Cell, 167(3), 750-762.e14. https://doi.org/10.1016/j.cell.2016.10.004