top of page

AI approach to Synthetic Biology: Fighting against COVID-19 and Emergent Infectious Diseases

The world doesn’t stop for a vaccine. For now, the best way to curb the spread of the novel coronavirus is to pursue accessible and widespread testing.

Jim J. Collins (Termeer Professor of MIT's Medical Engineering & Science; MIT's Biological Engineering; Harvard-MIT Health Sciences & Technology; Wyss Institute for Biologically Inspired Engineering at Harvard; Institute Member of the Broad Institute) and his group have been working on a relevant diagnostic and therapeutic technology, which involves the design of low-cost CRISPR and synthetic biology-based sensors. Such sensors, which are freeze-dried onto paper to improve stability, can detect viral RNA and are now being repurposed to detect SARS-CoV-2. The technology could become a complementary approach to existing coronavirus diagnostics, most of which use PCR.

Collins and his group began developing the technology in 2014, when Keith Pardee, a postdoc in Collin's lab, revealed it was possible to extract parts of a living cell and blot them onto paper. These cellular pieces - including RNA, DNA, and other components such as ribosomes - could then be preserved by freeze-drying, which allows the microscopic machinery to be stable at room temperature for long periods.

Initially, they put the technology towards developing diagnostic tools for Ebola, as an outbreak swept through West Africa. When Zika emerged two years later, the team refocused their attention on the mosquito-borne virus.

"Going back about a year ago, we extended the platform to clothing and textiles," Collins says. Luis Soenksen (MIT Ph.D. graduate and J-Clinic affiliated Venture Builder), along with Peter Q. Nguyen, Helena de Puig Guixe, and Nicolaas Angenent-Mari, drove the project forward.

Now, assisted by machine-learning, the lab focuses on designing the freeze-dried sensors to combat COVID-19. One option for this technology is creating an insert to be placed inside any standard mask to detect SARS-CoV-2. As someone breathes and talks, they would expel wet particles into the mask, reviving its cellular machinery. If SARS-CoV-2 is present, the fabric will emit a fluorescent glow. The fluorescence isn't visible to the naked eye, but detectable with an inexpensive device called a fluorometer.

Collins envisions the insert being used at home, making it easier to check for infections before exposing others. Beyond daily life, travelers could check themselves before boarding airplanes, and patients could receive a more streamlined diagnosis in hospital waiting rooms.

The group has also developed a stand-alone diagnostic - analogous to existing coronavirus tests - using the same CRISPR-based technology. A few weeks ago, Sherlock Biosciences, which is a spin-off of both the Collins Lab and the lab of Feng Zhang, received FDA emergency-use authorization for the test. Currently, the company is formalizing a deal with a manufacturer to produce tests on a large scale.

Compared to many PCR tests, a CRISPR-based test would not only be faster and cheaper - it would also use different reagents, relieving stress on the supply chain. “At the end of the day, we need as many different tests as possible,” Collins says.

Both the mask insert and the stand-alone test work by taking advantage of the gene-editing tool CRISPR. The team programmed RNA “guides,” which target specific viral genomic sequences. Those guides match with RNA that is unique to the SARS-CoV-2 genetic fingerprint and no other pathogen. “When the guide finds its target, it activates the CRISPR enzyme which serves to degrade that target,” Collins says. This leads to a cascade of reactions, which frees up neighboring RNA molecules and releases attached fluorescent molecules that can be detected using a fluorometer.

The team also works on a similar diagnostic, which uses synthetic biology sensors instead of CRISPR. These minute synthetic sensors, named toehold switches, normally remain “off,” closed in a hairpin-like shape. But when a specific piece of viral RNA binds to the sensor’s dock, the hairpin opens up to expose mRNA that can then produce a fluorescent protein.

For both SARS-CoV-2 and previous viruses, the team used deep-learning methods to design effective CRISPR-based guides and synthetic biology sensors. Machine-learning also proved helpful in identifying which segments of the viral genome to target, especially in the context of CRISPR. Luis Soenksen, Nicolaas Angenent-Mari, Jackie Valerie, and Miguel Angel Alcantar led this effort.

When it comes to combating COVID-19 with the biosciences, Collins is hopeful.

"I think scientific enterprise has really pivoted to this," Collins says. "I think we as a community have really all come together to see if we can get our collective talents properly directed, to help us get out of this as quickly as we can."

bottom of page