Developing microfluidic devices might be tricky, but finding antibodies that work as all-around great detection agents is even trickier. Just ask Jim Heath, a chemist at Caltech whose aim is to combine recent advances in microfluidics with a novel strategy to create highly sensitive and specific antibodies for better detection of diagnostic and prognostic disease biomarkers. A big advance out of his lab is an integrated blood barcode chip, which can take a finger-prick's worth of blood and detect proteins in the plasma within five minutes. The chip consists of a DNA array which, when the time comes to use it, is transformed into an antibody array by flowing oligo-bound antibodies over it. The oligos bind the DNA, and the antibodies then serve as capture agents for the proteins in the blood.
"We'd like to be able to monitor a patient's response to therapy by looking at how certain proteins in their blood are dynamically evolving during the few-hour window that surrounds when they first receive the therapy," Heath says.
When it comes down to it, antibodies are still the thorn in clinical detection's side. The biggest protein measurement problem is the protein capture agent, says Heath, and in his lab he's tried many alternatives, including aptamers, small molecule inhibitors, and the like. None of these could provide the high-affinity, high-stability alternative to antibodies, so Heath created his own method.
The method, called One Bead One Compound, is based on the concept of using "really big libraries of artificial and nonnatural amino-acid containing peptides and letting the protein assemble its own capture agent from those libraries," Heath says. "The protein serves as a catalyst to couple and build a covalent linkage between multiple peptides."
As an example, he says, consider mixing millions of beads in 20 different pots, each containing a unique amino acid. After mixing the beads together over multiple rounds, every bead is guaranteed to have a single, unique peptide on it. If you start with 100 million beads and 20 amino acids, says Heath, you can create a library of all possible variations of a 6-mer peptide.
The next step is to take this library, mix it with a target protein, and identify a peptide that binds to the protein but not very well. "It has lousy affinity and lousy selectivity," Heath says, "but it does bind." Then, scale that peptide up, add an azide-containing amino acid, and modify the ends of the rest of the peptide library with acetyl groups. If the protein, azide-bound peptide, and peptide library are all mixed together, the protein will search for peptides that best couple with the azide-peptide, already bound to it.
"So normally, you can get the azide and acetyl groups to react with a catalyst, but in this case the protein itself is the catalyst," Heath says. "And it only catalyzes the reaction when the two peptides are organized on its surface in just the right way — it'd be almost impossible to get it just right unless you've got a really big library."
It turns out that the bi- and tri-ligand peptides that the protein helps create are really selective. While a bi-ligand is already "pretty selective," being able to pull proteins out of blood, a tri-ligand is "not only very selective, it has antibody-like affinity," Heath says.
Heath has begun commercializing his capture agents and has set up a lab in Singapore to build out methods to make the process high-throughput. "Our overall goal is to do diagnostic and health measurements [of proteins] so they cost about a penny per protein you measure," he says, adding that right now they cost about $50.