Cannabinoid Biosynthesis: The Enzymes and Cultivars Coming Next

Photo by CDC via Pexels.

Nobody in cannabinoid biotech has a demand problem. CBD is in gas-station tinctures and pet chews, CBC and CBG are showing up in formulations chasing the entourage effect, and money has been flowing into extraction and synthesis startups for the better part of a decade. The problem has always been supply-side and biochemical: getting a plant or a microbe to actually make these molecules efficiently. Cannabichromenic acid synthase (CBCAS) and cannabidiolic acid synthase (CBDAS) — the two enzymes responsible for CBCA and CBDA — are notoriously sluggish and poorly expressed compared to their cousin, THCA synthase. That gap has quietly capped what's possible in engineered production for years, no matter how clever the downstream engineering got.

That's what makes a 2026 study in Plant Biotechnology Journal worth paying attention to. A team led by Villard and colleagues didn't try to tweak the modern enzymes directly. Instead they reached backward — reconstructing enzyme sequences estimated at 25 to 27 million years old, from before Cannabis and its closest relative, hops, split into separate lineages — and found that these resurrected ancestral enzymes outperform the modern wildtype versions at producing CBDA and CBCA, sometimes by several-fold margins. That's a strange and genuinely interesting result: evolution didn't necessarily optimize these enzymes for output, and going back in time turned out to be a viable engineering strategy.

Zoom out and this is one thread in a much larger pattern: the slow migration of cannabinoid production out of the cannabis plant entirely, into yeast, fungi, and non-Cannabaceae host plants. The pitch is speed, consistency, and manufacturing that behaves like pharmaceutical fermentation rather than agriculture. But there's real tension worth sitting with here — lab-made cannabinoids promise scale, while cultivar breeding, whole-flower chemistry, and the terpene-driven entourage effect aren't going anywhere soon. This piece works through both sides.

Why CBDA and CBCA Have Been So Hard to Make at Scale

Photo by ernestoeslava via Pixabay.

To understand why this enzyme discovery matters, you need to know what CBDAS and CBCAS actually do and why they've been such stubborn bottlenecks. Cannabis synthesizes its acidic cannabinoids — THCA, CBDA, CBCA — using a family of FAD-dependent oxidocyclase enzymes that convert cannabigerolic acid (CBGA) into the specific acid form. THCA synthase does this job well: it's relatively robust, expresses reasonably in heterologous systems, and has been the workhorse behind most engineered-yeast THC projects to date. CBDAS and CBCAS are a different story. Both have comparatively low catalytic turnover and tend to express poorly when you try to put them into a non-native host, which is precisely the situation biotech companies want to exploit.

Heterologous production, in plain terms, means getting an organism other than cannabis — usually a strain of engineered yeast, like Saccharomyces cerevisiae, or occasionally a bacterium — to run the cannabinoid pathway inside its own cells. You feed it sugar, it ferments in a tank, and instead of ethanol or biomass you extract cannabinoids from the culture. Companies want this because it decouples production from a farm: no seasons, no acreage, no agricultural pest pressure, no THC-versus-hemp compliance testing tied to a living crop. In principle you get pharmaceutical-grade consistency batch after batch.

In practice, that promise has been stuck for years because of exactly this enzyme problem. You can put a beautifully engineered yeast strain through a dozen rounds of pathway optimization, but if the CBDAS or CBCAS gene inside it is inherently a weak catalyst, you hit a ceiling on titer no matter how good the fermentation conditions are. This is the unglamorous truth about a lot of synthetic biology: the chassis organism gets the headlines, but the rate-limiting enzyme is usually where the real economics live. Until someone found a better-performing version of these specific enzymes, scaling biosynthetic CBD or CBC stayed economically marginal — interesting in a paper, uncompetitive against plant extraction at commercial volume.

Resurrecting 25-Million-Year-Old Enzymes to Fix Modern Ones

Photo by Steve A Johnson via Pexels.

Villard et al.'s 2026 paper takes a genuinely different approach to fixing a weak enzyme: instead of directed evolution or rational mutagenesis on the modern protein, they used ancestral sequence reconstruction, or ASR. The method works by building a phylogenetic tree of related enzyme sequences across living species, then using statistical models of amino acid substitution to infer what the sequence at each ancestral node probably looked like tens of millions of years ago. You're not guessing — you're running a maximum-likelihood reconstruction backward through the tree, then synthesizing the resulting gene and testing whether the resurrected protein actually folds and functions.

It did more than function. One reconstructed ancestral node, labeled CBDASSBR_FAD in the paper, produced roughly 3.1 times more CBDA than modern wildtype CBDAS in their assay system, while also expressing two to three times higher — meaning more functional protein per cell, on top of each molecule working better. A separate reconstructed lineage, described as the HCa-to-CaSBR branch, showed near-perfect selectivity for CBCA production specifically, with expression three to four times higher than wildtype THCA synthase. These aren't marginal improvements; they're the kind of fold-change that changes whether a pathway is commercially interesting at all.

There's a deeper evolutionary story tucked into this too. The ancestral enzyme the researchers call HCa appears to predate the roughly 25-to-27-million-year-old divergence between Cannabis and Humulus, its sister genus — hops. The gene for this cannabinoid-oxidocyclase-like function seems to have been lost somewhere in the lineage leading to modern hops, which don't make cannabinoids. That raises a real open question in plant evolutionary biology: did cannabinoid biosynthesis actually originate in Cannabis, or did it exist in a shared ancestor of both genera and simply survive in one lineage while disappearing from the other? The paper doesn't settle this definitively, but it reframes cannabinoid production as possibly older and more evolutionarily contingent than the standard story of Cannabis-specific innovation.

The caveat here matters and shouldn't get lost in the excitement: this is one peer-reviewed study demonstrating improved expression and activity in a lab construct. That's a meaningfully different milestone from a validated, scaled, regulatorily cleared manufacturing process. Plenty of enzymes look great in a plate assay and hit unexpected problems — protein stability, toxicity to the host cell, downstream purification headaches — once someone tries to run them at fermenter scale.

Beyond Cannabis: Engineering Other Organisms to Make Cannabinoids

Photo by Nennieinszweidrei via Pixabay.

Cannabis itself isn't the only organism being pressed into cannabinoid service, and the search for a better production chassis extends well past enzyme swaps. Xinteza, working with Professor Asaph Aharoni's lab at the Weizmann Institute of Science, has reportedly been developing a non-Cannabaceae host plant engineered to express the cannabinoid biosynthetic pathway throughout its tissue mass rather than confined to specialized structures. The logic is straightforward: cannabis only produces cannabinoids in glandular trichomes concentrated on female flowers, meaning most of the plant's biomass — stalk, leaf, root — is metabolically dead weight from an extraction standpoint. A host plant that makes cannabinoids in every cell would, in theory, offer a dramatically higher ratio of active ingredient to total harvested biomass.

Worth flagging directly: this particular story traces back to reporting that appears to have circulated earlier and then resurfaced in coverage dated March 2026, so readers should treat the exact project timeline and current commercial status as unconfirmed rather than fresh news. It's a real research direction worth watching, not a settled product.

Fungi are the other major front. Late-2025 review literature has pointed to engineered strains of Aspergillus niger, Aspergillus oryzae, and Trichoderma reesei showing enhanced cannabinoid titers through pathway optimization and precursor feeding — essentially supplying the fungal culture with cannabinoid precursor molecules and tuning its native metabolism to convert them more efficiently. Fungi are attractive hosts partly because industrial fermentation using Aspergillus species is already a mature discipline; the food and enzyme industries have run these organisms at massive scale for decades making everything from citric acid to industrial enzymes, so the fermentation infrastructure and regulatory familiarity already exist.

Industry market analyses have also cited CRISPR-Cas9 strain engineering achieving up to a 40% increase in cannabinoid production in microbial hosts. That figure comes from industry-side reporting rather than a single controlled peer-reviewed trial, so it's worth treating as a directional claim rather than a universal constant across strains and conditions. Still, the appeal of fungal and other non-plant hosts to manufacturers is easy to understand: a fermentation run can finish in days rather than the three-to-four-month flowering cycle cannabis requires, and a fungal bioreactor doesn't carry the same regulatory classification, security requirements, or seed-to-sale tracking baggage that comes with cultivating an actual controlled cannabis crop.

The Efficiency Case: Water, Land, and Speed

The global cannabinoid biosynthesis market is projected to grow nearly fourfold, from $2.57 billion in 2025 to $9.67 billion by 2035, reflecting rapid expansion in enzyme-driven cannabinoid production.

The efficiency argument for biosynthesis rests on a few big numbers that get repeated across industry decks and trade press: that eliminating large-scale cannabis farming in favor of fermentation could cut water and land use by roughly 60% compared to conventional cultivation, and that biosynthetic production can run up to 10 times faster than growing and extracting from plants. Both numbers are plausible on their face — a fermentation tank genuinely doesn't need irrigated acreage or months of photoperiod-dependent flowering — but it's important to be precise about where they come from. These are industry-sourced efficiency claims, not figures independently audited by a third party or verified against a standardized lifecycle analysis. Read them as directional signals of where the economics are pointing, not settled accounting.

There's a real counter-case that tends to get buried under those headline percentages. Fermentation-based production isn't free of capital intensity — it just shifts where the capital goes. Instead of land, irrigation systems, and greenhouse infrastructure, you need precision bioreactors, sterile fermentation environments, and downstream purification trains capable of isolating a specific cannabinoid from a complex fermentation broth to pharmaceutical or food-grade purity. That purification step is often the expensive, unglamorous part that market projections gloss over. On top of that, a cannabinoid produced in yeast or fungus for human consumption typically has to clear novel-food or novel-ingredient regulatory review in most jurisdictions — a process field-grown, extracted cannabinoids from an already-legal plant don't face in the same way, since regulators already have decades of familiarity with cannabis extraction as a category.

There's also a chemistry argument that doesn't show up in an efficiency spreadsheet at all. Whole-plant cultivation delivers something biosynthesis in a single-pathway microbial host can't cheaply replicate: the full terpene and minor-cannabinoid entourage that comes bundled together in an intact flower. A microbial strain engineered to pump out CBDA is, by design, optimized to make one molecule efficiently — it isn't simultaneously producing the dozens of terpenes, flavonoids, and minor cannabinoids that give a specific cannabis cultivar its characteristic aroma, flavor, and reported effect profile. For products marketed on strain identity — and a meaningful share of the legal cannabis market is exactly that kind of product — synthetic single-molecule output isn't really competing in the same category. It's a different ingredient serving a different formulation need.

What This Means for Cultivars and Breeding Programs

Photo by Tima Miroshnichenko via Pexels.

None of this makes cannabis breeders obsolete, and it's worth being direct about why. Novel cultivars still matter for reasons biosynthesis doesn't touch: the entourage effect that depends on a full terpene-and-cannabinoid profile working together, disease and pest resistance bred into a living plant that has to survive an actual growing season, yield-per-square-foot economics that matter enormously to commercial cultivators, and terroir-driven branding — the idea that a specific strain grown in a specific region carries recognizable characteristics — that consumers increasingly pay a premium for. A fermentation tank doesn't have a growing region or a genetic lineage a brand can market.

What's more interesting is the possibility that ancestral enzyme insights loop back into plant breeding itself rather than staying confined to bioreactors. If a sequence like CBDASSBR_FAD genuinely produces more CBDA with better expression, there's no fundamental reason that improved sequence has to live only in engineered yeast. In principle, it could be gene-edited or bred directly into cannabis cultivars, raising CBDA yield in an actual field-grown plant rather than requiring a switch away from agriculture entirely. That would let breeders keep everything cultivation already offers — the entourage chemistry, the existing regulatory pathway for a legal crop, the established supply chains — while fixing the yield problem at its biochemical root rather than routing around the plant.

History gives a useful, humbling precedent for how this tends to actually play out. Synthetic vanillin, produced cheaply from lignin or guaiacol, has been commercially available for well over a century and dominates the flavor market by volume — and yet vanilla farming in Madagascar and elsewhere didn't disappear, because a meaningful market segment specifically wants and will pay for real vanilla bean extract. Recombinant insulin manufacturing didn't eliminate biological diversity in drug production either; it added a scalable option alongside other manufacturing methods rather than erasing the field. The pattern across biotech disruption stories is segmentation by price point and authenticity claims, not one method fully replacing another.

Timeline expectations should be set accordingly. Enzyme discoveries like Villard et al.'s typically take five to ten-plus years to move from a published study to a commercial strain or shelf product — that's roughly the lag CRISPR-based crop traits and earlier plant synthetic biology projects took to reach real markets, and there's no obvious reason cannabinoid biosynthesis skips that maturation curve. Meanwhile the underlying market is genuinely growing: the cannabinoid biosynthesis space has been valued at $2.57 billion in 2025, with projections putting it at $9.67 billion by 2035 at a 14.3% compound annual growth rate. That growth trajectory assumes continued regulatory tolerance and no major cost or safety setback along the way — an assumption worth watching rather than banking on.

Strip away the more speculative framing and the near-term story isn't lab-grown cannabinoids replacing the plant — it's ancestral enzyme insights and non-plant hosts giving formulators cheaper access to compounds like CBC and CBDA that are currently too costly to pull out of flower at real scale. Minor cannabinoids have been the awkward middle child of the industry: everyone agrees they're interesting, few products contain meaningful concentrations of them, because extracting them economically from a plant that makes them in small quantities has never penciled out. A better enzyme, wherever it ends up being expressed, changes that math.

The path worth watching most closely over the next several years is the one that keeps the plant in the picture: gene-edited cultivars that borrow ancestral enzyme sequences like CBDASSBR_FAD directly, bred or edited into cannabis itself rather than shipped off into a yeast strain. That route preserves the agricultural and regulatory infrastructure the cannabis industry has already spent a decade building, while fixing the yield problem at its biochemical source instead of routing around it entirely. It's a less dramatic story than lab-grown cannabinoids conquering the market, but it's the one with more historical precedent behind it.

In the meantime, treat every efficiency figure circulating in this space — the 60% water and land savings, the 10x speed claims over plant extraction, the 40% CRISPR yield gains in microbial hosts — as industry-reported numbers worth tracking across the next few product cycles, not settled facts. The enzyme science here is real and peer-reviewed. The commercial claims built on top of it are still mostly promises, and promises in biotech have a well-documented habit of taking longer to cash than the press releases suggest.