Scindo raises $5.4m seed round for AI-powered enzyme discovery platform
[Disclosure: AgFunderNews’ parent company AgFunder is an investor in Scindo]
- Scindo—a UK-based startup building an AI-powered enzyme discovery and design platform—has raised a £4 million ($5.4 million) seed round.
- The round was co-led by Kadmos Capital and Clay Capital, with participation from PINC, the venture arm of food and beverage company Paulig, and existing investors Synbioven, AgFunder, SOSV, Farvatn Venture and Savantus Ventures.
- Scindo develops enzymes—nature’s tiny biological catalysts—that can transform a wide range of feedstocks into ingredients that have historically been sourced from petrochemicals.
Founded in 2020 by Dr. Gustaf Hemberg, Dr. Ben Davis, and Juliet Sword, Scindo combines AI models with proprietary data to accelerate enzyme discovery and optimization.
The firm, which has established partnerships with leading specialty chemical manufacturers, develops enzymes for several industry verticals including food and flavorings, cosmetics, and specialty chemicals.
With the new funding, it will expand its platform, scale wet-lab capabilities and strengthen its team.
“The specialty chemicals industry has long sought to move away from petrochemical-derived ingredients, but existing approaches have struggled with complex natural feedstocks,” said Ali Morrow, partner at lead investor Clay Capital.
“Scindo’s approach creates molecular craftsmen: enzymes designed for specific industrial jobs that offer cost-competitive natural alternatives and unlock previously inaccessible feedstocks, creating significant opportunities globally to end the industry’s reliance on crude oil.”
Designer enzymes
Scindo CEO Gustaf Hemberg told AgFunderNews: “When we started, we focused on mapping enzymes with functionalities that are difficult to achieve selectively with traditional chemistry and relatively uncommon in nature—particularly C–H activations and C–C bond cleavages [thereby opening up route for degrading stubborn molecules such as plastics].”
“By discovering and characterizing these underexplored enzymes, we built datasets that could be fed into our machine learning models—not only to identify new enzymes with previously unknown functionalities, but also to enable generative design of novel enzymes with entirely new capabilities.
“That’s really at the core of what we do: by gathering new examples and generating proprietary datasets of enzymes with novel and defined functionalities and selectivities, we enable our machine learning models to learn which parts of the sequence or structure drive the performance or selectivity that we’re targeting.
“And that’s the real challenge with public datasets—they’re concentrated on a few well-studied enzyme families with narrow, specific functionalities. They’re often incomplete and sometimes even contain mischaracterized examples, which makes them limiting when training models for prediction or generative design of enzymes for novel transformations and the specific applications that we are targeting.”
Once Scindo has identified suitable candidates, it engineers these enzymes further—first to optimize selectivity and transformation efficiency, and then to improve physical traits such as thermostability and expressibility, said Hemberg. Scindo also collects and enriches its predictive models with rich metadata to identify candidates most likely to scale in industrial settings and express at high yields in microbial systems. This ensures maximum viability for rapid scale-up and collaboration with manufacturing partners, he explained.
“We have quite a big chemistry screening platform, so we are able to test the enzymes in the lab, characterize them and then feed that data back into the machine learning. Closing the loop between real life results and machine learning has been really critical for us.”
Once it has tested some candidates, it can do further work to rank them based on viability for scaling up in a microbial expression system and then work with an enzyme manufacturer to scale up production, said Hemberg.
Cell-free biomanufacturing
Scindo’s first two products are enzymes that can create key building blocks of flavors & fragrances, and enzymes that can enable cost-effective petrochemical-free production of a high-value cosmetic ingredient via cell-free biomanufacturing.
In the case of flavor and fragrance ingredients, he said, “We can use a wide range of agricultural fatty acid feedstocks and have designed enzyme systems that selectively convert them into flavor molecules. We’re now advancing into pilot-scale production through a partnership we haven’t yet announced.
“Some of those flavor ingredients could be produced with precision fermentation [by engineering microbes to express them in costly steel fermentation tanks], but that is much more expensive [than using a cell-free approach just using enzymes], the titers are quite low, and you generate a lot of waste metabolites.”
By using a cell-free approach that utilizes the internal machinery of microbial cells (such as enzymes) to convert feedstocks into the target flavor molecules, Scindo can significantly reduce production costs, he claimed.
Operating outside the constraints of a cell—and without relying on costly cellular cofactors—allows Scindo to run faster reactions across a wider range of conditions, such as pH and temperature, and in ways that significantly reduce energy consumption as there is less heating and cooling required, he explained. It also generates cleaner products that require less costly downstream processing.
“We’re hoping to target a market launch in the next 12 months or so for our first two products.”
Proprietary data sets
Stepping back, he said, the world’s largest enzyme companies tend to concentrate on a few specific enzyme families for traditional applications, particularly in food, laundry detergents, and some pharma applications.
“We are instead focusing on novel applications that have historically been much more difficult to target with traditional chemistry and enzymes alike.
“Our key differentiator is the proprietary data we’ve built around novel enzymes—their functionalities, specificities, and characteristics—with broad applicability to carbon-chain transformations. That’s really what separates us: we’re working with data that isn’t publicly available.”
