Quercus Biosolutions, an ag biotech startup using generative AI to create a new category of crop protection products based on designer mini proteins, has emerged from stealth with pre-seed funding from undisclosed investors.
The firm, founded last year by ag industry veterans Dr. Jon Lightner and Matt Crisp, has a strategic partnership with drug discovery startup Ordaos Bio to apply the latter’s generative AI platform—originally designed for human medicine—to agriculture.
The approach could potentially deliver a step change in crop protection, Crisp told AgFunderNews.
“We’ve been saying for 20 years that ‘In the next five years, biologicals are going to overtake chemistry-based solutions,’ and we’re all looking at each other right now and saying, ‘Well, we’ve gotten a few points of market share after 20 years and many billions of dollars of investment.’”
With Quercus’ platform, he claimed, “I feel like we’re going to achieve more in crop protection using this technology in the next two years than we have in the last 20.”
At its core, said Lightner, “We expect this platform can finally deliver the efficacy that growers and farmers associate with small molecules [and chemical crop inputs] to the biologicals segment.”
Designer proteins
Several firms in ag biologicals now use mini proteins, also known as peptides or micropeptides (small chains of amino acids) rather than living microorganisms for crop protection purposes, citing increased stability, precision, and consistency.
However most players typically screen libraries of naturally occurring proteins to identify those with potential to target pests or perform other crop protection functions, said Lightner, who is working with Solis Agrosciences for testing, data generation, optimization, and validation.
Quercus, by contrast, is using AI to design mini proteins from scratch, he said. “We look at all possible proteins, model all possible structures, and then test and build subsets of those in a days-long process for computational testing. We can then produce them via cell-free expression for testing in vitro the next day.”
Ordaos Bio CEO Dave Longo added: “A lot of people claim they are doing truly de novo protein design, but most AI companies claiming this are taking an existing protein and simply replacing the sequence that makes up that protein to get the same structure.
“For instance, if you have something that binds to HER2 [Human Epidermal Growth Factor Receptor 2, a common target for breast cancer drugs] they would take the original protein and ask the AI, can you give us a different sequence so that we can patent it, with the same structure as this existing protein?
“It’s hard to call that design. What we’re doing is giving the system all it needs to understand what your mini protein has to do such as crossing the plant cuticle or crossing the cell membrane, and then designing a protein truly from scratch.”
‘Head-to-head comparisons with the best chemical solutions’
According to Crisp: “I actually think the same problems persist for many of the biologicals out there, whether it’s peptides, RNAi, small proteins, large proteins, or whole microorganisms. I think a similar set of challenges continues to plague this entire space.”
Developing designer proteins from scratch can solve many of these problems, claimed Lightner.
“We’re building our assays around head-to-head comparisons with the best current chemical solutions. We think we’ll create an equally large crop protection category that’s based entirely on these novel proteins, with all the favored regulatory and environmental advantages of biologicals, but also a rock solid, consistent efficacy of the kind that people have achieved with chemistry in agriculture.”
Designing for multiple factors at once
Typically, said Crisp, firms developing crop protection products look at the problem they are trying to solve in a linear fashion “starting from a mode of action perspective.”
After that, they then tackle all the other challenges that come along with developing a commercial product such as formulation, stability, supply chain constraints, whether it will mix in a tank with other crop protection inputs, and whether it can be cost-effectively manufactured via precision fermentation on a commercial scale, he said.
Quercus can design products that factor in all of these things from the outset, he claimed.
“That’s what gets me so jazzed. We can build all those requirements right into our very early screens and then feed those results back into our AI.”
Go where the money is: drug discovery
To place Quercus in context, Crisp said, “I’d say ag is 5-15 years behind all of the technological enabling platform work that’s done in the most well-funded arena of innovation, which is human medicine.
“We looked across the AI space in agriculture,” said Lightner, “but Matt’s view was, if we really want to see the cutting edge of AI, we need to look at the pharmaceutical sector.”
Crisp explained: “I left [seed giant] Benson Hill a couple years ago and since then I’ve been looking at a number of different innovations and opportunities including newer generation biologicals. But I didn’t see the kind of Gen AI that was beginning to be deployed in the arena of cancer therapeutics, for example.
“So my view is, go where the money is. Companies in human medicine have been receiving millions of dollars to develop some of the most innovative platform technologies to do bespoke biological design, including for peptides and proteins. So could we repurpose one of those platforms for agriculture?
“So Jon and I did some diligence on companies in human healthcare and were really impressed with Ordaos Bio’s platform, and pretty soon came to the conclusion that it could accelerate a quantum of value creation in crop protection.”
The attractive thing about some of the bespoke mini proteins Quercus is developing is that they could be as effective as synthetic chemicals, but will be regulated as biopesticides and will not persist in the environment for decades, claimed Lightner.
“Bt technologies [deployed in GM maize and cotton for example] are beginning to fail and so we need innovation desperately. And these protein crop protectants can operate under the biopesticide framework with the EPA that’s typically a million dollars to an approval, and less than two years. We can create solutions that we can bring to market on a really quick timeframe.”
Manufacturing the mini proteins at scale
So can Quercus’s mini proteins be chemically synthesized at scale or would they be expressed in microbial cells via precision fermentation?
According to Lightner: “Many of the proteins we’re working with in the R&D phase we’ve had built synthetically, but for large-scale production we would be looking at recombinant protein expression in microbial systems.”
The proteins would then be extracted, he said, but this would not involve super-costly downstream processing steps. “We believe at scale we can compete head-to-head with small molecule chemistry on a cost basis, with attractive margins.”
‘The models get better and better’
According to Lightner: “What we’re doing in the next six to 12 months is focusing on known sites of action for particular crop protection problems. So that’s a place where we might have a small molecule [chemical] inhibitor today or we might have a naturally-derived molecule like a Bt protein, and we know the receptor or the enzyme that those actives inhibit to get the crop protective effect.”
Quercus knows the proteins the receptors interact with and usually already has crystal structures “as those proteins have been around for decades,” he said. “We then predict structures for our [designer] proteins to fit the active site or inhibit particular catalytic residues, for example.
“We do that computationally first and then test in vitro for those actual interactions. One of the unique things about Ordaos Bio’s approach is that they’ve built a connection between the computational prediction models and the in vitro testing, so the test results inform the model, and the models get better and better.”
He added: “One thing we’ve learned across statistical and genomic selection models over the last 20 years in ag is that the best model is the one we built today from all the available data… and that model will be obsolete tomorrow.”
Field level validation ‘in as little as six months’
In the next three to six months, said Crisp, Quercus aims to start conducting “greenhouse and growth chamber level studies with the potential to move to field level validation in as little as six months, given the speed of the platform.”
He added: “We see huge opportunity for IP in the space, so we’re looking to establish a strong IP portfolio founded on a unique data set around the application of AI protein design to this crop protection space.”
Defining mini proteins
As for what constitutes a “mini” protein, there’s no firm definition, said Lightner.
“The important thing is we’re looking at proteins with tertiary structure. In other words, not just a linear sequence of amino acids that’s wiggling around, but proteins large enough to have structural attributes. I haven’t seen any hard and fast definitions, but we’re looking at a minimum of 15-20 amino acids and probably far more likely to be 25-50 amino acids.”
The post Exclusive: Quercus Biosolutions emerges from stealth to disrupt crop protection with designer mini proteins appeared first on AgFunderNews.