Northeastern researchers team up with Kebotix to improve light-based cancer treatments by Laura Castañón September 9, 2019 Share Mastodon Facebook LinkedIn Twitter A model of a molecule. Photo by Matthew Modoono/Northeastern University Certain chemicals, activated by nothing more than visible light, have the ability to kill cancer cells. The treatment, known as photodynamic therapy, has been around for decades. It’s more localized and less damaging than treatments like chemotherapy. But photodynamic therapy can only be applied to small tumors in places that can be reached with a light—the skin, the esophagus, the colon. Now, Steven Lopez, an assistant professor of chemistry at Northeastern, is researching ways to expand this treatment to a wider range of cancers. To do this, he is applying quantum mechanical and artificial intelligence techniques to search for potential photodynamic therapy drugs that respond to red or near-infrared light, which can penetrate deeper into human tissue. “Visible light can only penetrate several millimeters into your tissues,” says Lopez. “We’re searching for new drugs that can absorb lower-energy light to treat larger tumors and those in currently inaccessible regions of the body.” Photodynamic therapy uses drugs that absorb visible or near infrared light. Steven Lopez, an assistant professor of chemistry, is searching for potential new drugs that could expand this treatment. Photos by Matthew Modoono/Northeastern University Lopez and his team are working with Kebotix, a Cambridge, Massachusetts, start-up company that specializes in discovering and producing new electronic materials with their artificial-intelligence-powered laboratory. Through this collaboration, the researchers expect to speed up the discovery of new drugs for photodynamic therapy to nearly eliminate side effects and move towards effectively treating brain cancer. Photodynamic therapy uses drugs that absorb visible or near infrared light. But there are far too many potential molecules to examine every possible drug candidate through lab experiments. The researchers are training machine learning algorithms to recognize molecular patterns of candidates for photodynamic therapy drugs. Photo by Matthew Modoono/Northeastern University “Imagine all of the grains of sand on planet Earth–that number has been estimated to be around 1018,” Lopez says. “Researchers estimate the number of organic molecules is orders of magnitude higher than that.” To solve this problem, the researchers are using calculations and machine learning algorithms to narrow down the possibilities. Lopez recently received a $750,000 grant from the Massachusetts Life Sciences Center to supply the computing power his team will need, and Kebotix will hire a data scientist to work on this project, splitting their time between Boston and Cambridge. “Partnering with prominent scientists who do research at the front edge of materials science enables Kebotix to solve much harder problems faster,” says Semion Saikin, the chief scientific officer of the company. “Our work with Northeastern University is aimed at discovering new molecules that can be game changers in photodynamic therapy.” To fit the bill, these molecules can’t be harmful or toxic in their base state—they need to be injected into a person without harming healthy cells. But once they’ve accumulated in cancer cells and are hit with the red or near-infrared light, the absorbed energy can be transferred to nearby oxygen molecules. That energy-transfer reaction pushes the oxygen molecules into an excited state called singlet oxygen. “Singlet oxygen is extremely toxic,” Lopez says. “It’s short-lived, reacts with almost everything in a tumor cell, and then leaves that tumor cell to die.” Because singlet oxygen is so short-lived, it doesn’t have a chance to spread and kill cells outside of the tumor. It rapidly settles back into a less excited, and less toxic, state. Lopez recently published an open-access database, VERDE materials DB, that catalogues light-absorbing molecules and their properties. The researchers will use that data set to train machine learning algorithms to recognize molecular patterns of candidates for photodynamic therapy drugs, which will significantly speed up the search. “Instead of doing these relatively time-expensive computations, and using a lot of high-performance computing, we’ll be able to make that prediction in fractions of a second,” Lopez says. Kebotix will create the best candidates in their lab, so these potential drugs can be tested. And if the researchers find the right one, they may be able to create the drugs necessary to treat hard-to-reach tumors without harming a patient’s healthy cells. “The dream is to replace chemotherapy with a non-invasive treatment, and to cure brain cancer,” Lopez says. “MLSC and Kebotix are providing an opportunity to move towards that.” For media inquiries, please contact email@example.com.