For all of their promise, many drugs today have a hard time with specificity—they have difficulties in discerning the bad cells, those that need treatment, from similarly looking good cells, those that shouldn’t be treated. Because of this many disease treatments basically treat all similarly looking cells, damaging good, healthy cells as they target diseased cells.
That may change soon as a team of researchers at Columbia University Medical Center, New York City, working with collaborators at the Hospital for Special Surgery, New York City, have created a fleet of molecular “nano-robots” (or automata) that can home in on specific cells and mark them for drug therapy or destruction.
The nano-robots—collections of DNA molecules, some attached to antibodies—were designed to seek out a specific set of human blood cells and attach a fluorescent tag to the cell surfaces. Details of this advance were published recently in Nature Nanotechnology.
“This opens up the possibility of using such molecules to target, treat, or kill specific cells without affecting similar healthy cells,” said Milan Stojanovic, lead author of the Nature Nanotechnology article and an associate professor of medicine and of biomedical engineering at Columbia University Medical Center. “In our experiment, we tagged the cells with a fluorescent marker, but we could replace that with a drug or a toxin to kill the cell.”
Stojanovic said the power of the system is its ability to discern the features of the cells it encounters. Though other DNA nanorobots have been designed to deliver drugs to cells, the advantage of Columbia’s fleet is its ability to distinguish cell populations that do not share a single distinctive feature.
“We have a system that takes into account, autonomously and without human input, multiple cell features before it labels or targets a cell,” he explained. “It’s Boolean algebra analysis of cells.”
Cells, including cancer cells, rarely possess a single, exclusive feature that sets them apart from all other cells. This makes it difficult to design drugs without side effects. Drugs can be designed to target cancer cells with a specific receptor, but healthy cells with the same receptor also will be targeted.
For example, Stojanovic explained, rituximab (an anti-CD20 monoclonal antibody used in the treatment of non-Hodgkin type lymphomas) “kills pretty much all B-cells it identifies. There are subtypes that one does not need and want to eliminate.”
The only way to target cells more precisely is to identify cells on a collection of features.
“If we look for the presence of two, three, going up to six or more proteins on the cell surface, we can be more selective,” said Stojanovic, who first became interested in this field through an interest in molecular computing and a push from NASA into autonomous therapeutics.
Large cell sorting machines have the ability to identify cells based on multiple proteins, but until now, molecular therapeutics have not had that capability.
Instead of building a single, complex molecule to identify multiple features on the cell surface, Stojanovic and his colleagues used a different, potentially easier approach based on multiple simple molecules, which together form the robots, or automata, as Stojanovic prefers to call them.
To identify cells possessing three specific surface proteins, the team first constructed three different components for their molecular automaton. Each component consists of a piece of double stranded DNA attached to an antibody specific to one of the surface proteins. When these components are added to a collection of cells, the antibody portions of the automaton bind to their respective proteins and work in concert.
On cells where all three components are attached, the automaton is functional and a fourth component initiates a chain reaction among the DNA strands. Each component swaps a strand of DNA with another, until the end of the swap when the last antibody obtains a strand of DNA that is fluorescently labeled.
At the end of the chain reaction, which takes less than 15 minutes in a sample of human blood, only cells with the three proteins are labeled with the fluorescent marker.
“Individual components by themselves are not robots, it's their combination,” he said. “Theoretically, there should be no limit to number of components we could add, but in practice we would expect those based on two, three, or maybe four, antibody conjugates to find the most use.”
The team demonstrated the concept with blood cells “because their surface proteins are well known, but in principle our molecules could be deployed anywhere in the body,” Stojanovic said.
“Accuracy is the same as multi-color flow cytometry, in other words, for two or three antibody components all cells that you could label in flow cytometry, would be labeled in our system as well,” he added.
The next step is tests on living organisms and Stojanovic is gearing up for studies with mice.
“It’s always easier to demonstrate things in vitro than in a living organism,” he added, “and we have to do that to see whether there is anything we missed that can preclude eventual practical, therapeutical applications.”