Kanvas Bio speeds the search for microbiome therapies through industrial-scale imaging of gut and host cells

Why is Kanvas Biosciences, a spatial biology company we’ve been backing since 2021, succeeding in its efforts to develop new microbiome-based ther­a­peu­tics, when so many drugs companies failed to harness the microbiome in the past? 

We think it’s because of Kanvas’s technology — specif­i­cally, its ability to use multiplexed spectral imaging to (literally) illuminate and measure inter­ac­tions between the microor­gan­isms in our GI tracts and the human cells that surround them. This method of generating novel, proprietary data about host-microbiome inter­ac­tions fuels AI models that are already helping the company come up with advances in cancer treatment. Kanvas’s platform is a striking example of a core conviction at DCVC: that it is new tools that enable new science, rather than the reverse.

Now Kanvas is gathering data about human cells and microbes in a new way. The company has built a new kind of instrument — a high-resolution, multi­spec­tral lightsheet microscope — that works at unprece­dented speed, enabling Kanvas to accelerate its data collection, further populate its training database, and more deeply understand the microbiome as a kind of organ system. 

Researchers have long suspected that some gut microbes can activate cancer-fighting immune cells in their human hosts. In our recent writings about Kanvas, we’ve highlighted the company’s progress on KAN-001, a mix of microor­gan­isms intended to boost the effects of cancer drugs called immune checkpoint inhibitors, or ICIs. KAN-001 used its first-generation imaging technology — which tags RNA molecules with a variety of fluo­rophores then uses different wavelengths of light to show their locations — to identify the bacterial strains most likely to turn ICI non-responders into responders. The company is moving toward its first clinical trials of the drug this year.

As it pushes forward with KAN-001 and other candidate ther­a­peu­tics, the company has continued to lean into its core strength, engineering innovation. Recently the company decided to bet on an idea long nurtured by co-founder and chief technology officer Hao Shi. He wanted to replace the commercial confocal microscopes Kanvas had been using for its multiplexed imaging with its own lightsheet microscope.

The basic problem was this: to show exactly where RNA and proteins are expressed and how gut bacteria and surrounding host cells interact, Kanvas needed to scan large slices of tissue. But confocal microscopes shine only a single point of light at a time. Scanning a whole sample means illu­mi­nating one point, moving the focus, observing another point, and so on, until a final 2D image is assembled. Shi says scanning a square of tissue just 135 microns on a side can take five minutes or more in a spectral mode, and imaging a large tissue sample (5−15 mm²) at a single depth can take 24 to 48 hours.

Shi knew that lightsheet microscopes can light up an entire slice of tissue with a planar sheet of laser light, gleaning much more data at once. This approach would also allow Kanvas’s researchers to assemble 3D volumes, simply by stacking multiple planar scans. There were only a couple of problems. The best commercial lightsheet microscopes, such as the Zeiss Lattice Lightsheet 7, can shine only a few colors of light, which meant Kanvas’s multiplexed RNA detection system wouldn’t work on those machines. On top of that, they lack the spatial resolution needed to show bacteria and their RNA transcripts clearly.

And so Kanvas hired Kevin Keomanee-Dizon, a physics fellow who had developed several novel microscopy tech­nolo­gies at Princeton University, to work with Shi to build a new instrument with both the resolution and the spectral capa­bil­i­ties Kanvas needed. ​“Kevin has been building microscopes for almost a decade now,” Shi says. ​“So we talked with him, sketched out the technical specs, did some quick calcu­la­tions, and concluded that it was possible in principle to build such an instrument.” By November 2024, Shi and Keomanee-Dizon had assembled a tabletop prototype of its Kanvas Spectral Lightsheet (KSL) microscope. And it worked: Shi says it’s capable of scanning the same 135x135-micron tissue slice in just a few seconds. 

In other words, it promises to help Kanvas expand its data collection efforts to industrial scale — making it a classic example of the kind of paradigm-breaking deep tech innovation we at DCVC seek to encourage in our portfolio companies.

“We’re gaining so much in terms of speed that we are now able to process much larger pieces of tissue,” says Kanvas co-founder and CEO Matt Cheng. That gives the company confidence that the microbe-microbe and host-microbiome inter­ac­tions it’s documenting under the microscope are truly repre­sen­ta­tive of what’s going on in the human gut. And that, in turn, is central to Kanvas’s argument that they can char­ac­terize the microbiome as a distinct organ system that can be manipulated using new therapeutics.

“We think that the data coming out of this lightsheet microscope is going to allow us to chart out the first true host-microbiome atlas,” Cheng says. ​“We developed KAN-001 by reverse-engineering the microbiome of a superdonor, using our platform to optimize it. But the next generation of ther­a­peu­tics is going to be built out de novo. And to get there, you need mountains of data.”

Microscope makers will likely be interested in licensing Kanvas’s new lightsheet technology, Shi says. But for now, he’s focused on proving that the instrument can collect the type of data Kanvas needs — data that, we’re confident, will unlock ideas that build on the company’s lead in the new realm of microbiome therapeutics.