Natural killer cells could form the next generation of cancer immunotherapies, but in order to fulfill this promise, they’ll need to overcome some of the challenges of the first generation of cell therapies. Those FDA-approved treatments can address blood cancers, but solid tumors have proven to be a more difficult target. Biotech companies are also working on ways to produce cell therapies at scale. Startup Modulus Therapeutics aims to address those challenges and more, and artificial intelligence is a key part of its approach.

The prevailing way of developing a cell therapy is high throughput screening, said Max Darnell, Modulus’s chief scientist. Companies screen to find a single gene that can be knocked out to improve a cell’s behavior as a therapy. By conducting experiments, scientists can test and see what happens—one gene at a time. But the number of things to try out is astronomical, Darnell said. He likens the startup’s technology to a flashlight that helps scientists figure out where to look.

“Where in this really dark room should we shine the flashlight to find what we are looking for,” he said. “To do that we need to illuminate enough of the room to see what’s there, and follow the light.”

Seattle-based Modulus has been incubating within the Allen Institute of Artificial Intelligence. The startup recently emerged from stealth, revealing $3.5 million in seed financing. The company’s approach, which it has dubbed “Convergent Design,” combines several different technologies, said Chief Data Scientist Bryce Daines. Gene editing and screening capabilities enable Modulus to create thousands of custom-designed cells. Those cells are then tested in animal model experiments. The output of those tests undergoes computational analysis. Machine learning interprets and predicts the effects of any unobserved or unseen modifications, which in turn helps the company determine the engineering changes needed for the next layer of screening.

As an example of the kinds of changes that Modulus would make to a cell, Darnell pointed to the goal of getting cell therapies to target solid tumors. Scientists know how previous cell therapies have failed in this endeavor. Therefore, Modulus will make changes that improve a cell’s metabolism, its ability to home in on a tumor, and the capability to resist immunosuppressive signals from a tumor. The results of one experiment become the input for the next one, Daines said. The overall process is a feedback loop. The goal is to search for and identify the genetic modifications to cells that will make them more effective therapies, giving them the capability to overcome barriers presented by the tumor microenvironment, Daines said.

“We’re building a platform and repeatable process,” he said.

Modulus is joining a growing field of companies chasing natural killer, or NK, cell therapies. Over the summer, Wugen raised $172 million in a Series B round of funding as it continues clinical development of its NK therapies. That company, which splits its operations between St. Louis and San Diego, is developing “off-the-shelf” therapies with NK cells sourced from healthy donors. Artiva Biotherapeutics, which launched last year with $78 million in financing, is also pursuing off-the-shelf therapies. That San Diego-based biotech generates its NK cells from the umbilical cord blood of healthy donors.

Other companies are developing their treatments by working with NK cells derived from induced pluripotent stem cells (iPSCs). Caribou Biosciences of Berkeley, California, raised $304 million from its recent IPO, some of which is earmarked for preclinical discovery research of iPSC-derived NK cells. The company is trying to figure out if such cells can be edited to become cell therapies for solid tumors. In some cases, NK research is leading to partnerships. In June, BeiGene paid Shoreline Biosciences $45 million up front to begin an alliance focused on developing off-the-shelf therapies NK cell therapies using Cambridge, Massachusetts-based Shoreline’s technology for designing cell therapies from iPSCs.

Darnell declined to disclose the source of Modulus’s NK cells, though he added that the startup’s technology is compatible with cells from either healthy donors or iPSCs. That capability may lend itself to partnerships with larger companies. NK cells are the initial focus of Modulus. But Daines added that the startup’s technology can be applied to various types of cells. Breast cancer is the first target, which could validate the technology and enable Modulus to expand the approach to additional cancers.

Modulus could develop one or two cell therapies, then take them all the way to the market, Daines added. But the startup could also outlicense assets or strike up a partnership with a larger company as a way of developing those assets. Alternatively, partnership efforts could sign emerging cell therapy developers or larger pharma companies looking to use the Modulus technology as a way to develop their internal drug pipelines. Daines said Modulus is building relationships with drug companies that could become such partners.

Modulus’s seed round of investment was led by Madrona Venture Group. Also participating in the financing were KdT Ventures and the Allen Institute for AI. Darnell said that new financing will be used to expand the team and hire staff for wet lab and computational work. The cash will also be used for experiments with the technology, research that he hopes will validate the technology and yield a therapeutic candidate.

Image by Modulus Therapeutics