By simultaneously scanning for hundreds or thousands of genes or proteins, geneticists can now detect whether a patient has a propensity toward certain forms of cancer. (See, for example, “TR10: Epigenetics.”) However, they commonly do this using DNA or protein microarrays that are costly to produce and require expensive detection equipment, limiting their application. In today’s issue of the journal Science, an MIT chemical-engineering doctoral student and his colleagues at MIT and Harvard Medical School offer an elegant cost-cutting solution.
Daniel Pregibon’s technique is a one-step process for producing capsule-shaped plastic particles, each packing a graphical ID and one or more biological probes that fluoresce when they detect a specific sequence of DNA or protein in a test sample. The result is a cheap yet sensitive system for producing diagnostics with as many as one million unique particles capable of detecting more than a million distinct biological targets. “For bedside diagnostics, you need to do a lot of tests for a reasonable price,” says Pregibon. “With the current technologies, I don’t know that the price will ever be low enough. With our system, we think it can.”
Microarrays are expensive because their manufacture requires a complex, multistep process, and many designs for producing coded particle probes analogous to Pregibon’s suffer from the same drawback. Pregibon’s Science paper demonstrates a simpler method for producing coded particles proposed last year by coauthor Patrick Doyle, an MIT chemical-engineering professor and Pregibon’s advisor. (See “Printing Press for Biosensors.”) Doyle’s printing press is a microfluidic device that produces multifunctional particles in plastic by exploiting laminar flow–the tendency of micrometer-width fluid streams to remain distinct rather than mixing.
To produce a basic particle, Doyle’s device flows two solutions containing molecular building blocks for the plastic polyethylene glycol down a 200-micrometer-wide channel etched in a silicon-polymer block, where they form parallel, 100-millimeter-wide streams. A 30-millisecond pulse of ultraviolet light projected through a stencil stimulates polyethylene-glycol precursors in both streams to solidify into a single particle, incorporating domains from each stream. Pregibon used this particle press to turn out unique biosensor particles by doping one of the precursor streams with a DNA biotag. He gave each particle a recognizable fingerprint by projecting patterns of dots on the second stream of precursors, producing particles with a unique pattern of holes that is visible with a low-magnification microscope.
UC Berkeley chemist and biosensor developer Jay Groves calls the one-step production system a “clever” step toward low-cost diagnostics. “The idea of inputting only a few liquid reagents into a device that creates indexed arrays of particles … seems powerful to me,” he says.
Pregibon’s particle probes may also be more sensitive than existing microarrays and particles because they are porous, rather than solid. With solid microarrays and particles, target molecules such as DNA from a patient’s blood bind only on the probe’s surface. In contrast, target molecules can diffuse into the porous polyethylene glycol of Pregibon’s particles, increasing the number of target molecules that bind and therefore producing a more intense fluorescent signal. “We’re able to gather a lot more signal that way,” he says.
Doyle says it would not be “unrealistic” to imagine commercializing the detection system within five years. The biggest challenge ahead, he says, is developing a more practical detection system. To date, he and his colleagues have used a jerry-rigged ultraviolet microscope–a bulky, impractical system that costs tens of thousands of dollars–to detect and identify particle probes bound to their targets. “I don’t imagine that for bedside use, people are going to want a microscope in every single room,” says Pregibon. “We need detectors that are portable and robust and can withstand repeated use.”
Doyle and Pregibon predict that if they succeed, their system could play a role in the detection of disease but also in identifying DNA or protein signals indicating which therapies will be most successful for a given patient. In 2005 the Food and Drug Administration approved a microarray that does just that: the AmpliChip test produced by Swiss-based pharmaceutical maker Roche. The test probes for two genes that influence how drugs such as certain antidepressants break down, and it can therefore predict whether someone is at risk of an adverse reaction to the drug, or may not respond to the drug. That system’s high price has limited its use.
Still, Groves warns that there’s plenty of work ahead. “It is still the Wild West when it comes to microfluidics and automation,” he says. “There are a lot of drastically different concepts out there, some of which are being developed into products, and many of which fail. It’s hard to know what will prove to be an industry standard 10 years from now.”
Groves points out that in order to be practical, the detection device will need to be able to crunch through large volumes of particles without clogging or otherwise failing. “I wonder about the fidelity of the system if it must run tens of thousands of assays continuously,” he says.
Adam Singer, an emergency-medicine specialist at Stony Brook University whose research has confirmed the value of bedside diagnostics in speeding up treatment, offers another word of warning. Singer says that cheap multitarget tests like Pregibon’s will be of little use if more science isn’t done to nail down markers that are truly instructive. Most microarray tests developed to date, Singer says, provide “general direction” in diagnosing illness, but not much more. “We’re still quite a ways away from being able to apply these technologies clinically. Sometimes the technologies are more advanced than the knowledge and the evidence.”
This new data poisoning tool lets artists fight back against generative AI
The tool, called Nightshade, messes up training data in ways that could cause serious damage to image-generating AI models.
Rogue superintelligence and merging with machines: Inside the mind of OpenAI’s chief scientist
An exclusive conversation with Ilya Sutskever on his fears for the future of AI and why they’ve made him change the focus of his life’s work.
The Biggest Questions: What is death?
New neuroscience is challenging our understanding of the dying process—bringing opportunities for the living.
How to fix the internet
If we want online discourse to improve, we need to move beyond the big platforms.
Get the latest updates from
MIT Technology Review
Discover special offers, top stories, upcoming events, and more.