Manolis Kellis develops algorithms and techniques for analyzing the entire genomes of different species, the better to understand those genomes. Kellis began his PhD work with little knowledge of biology: his undergraduate degree is in computer science. For his thesis, he compared the genomes of four yeast species to identify all the genes and regulatory sequences in one of them–a project hes glad no one told him was believed to be impossible.
Comparing the genomes of multiple closely related species has proved to be a powerful new tool for finding genes and the sequences that regulate them, and for learning about how genomes evolve (see “Finding Evolution’s Signatures”).
After validating his methods in yeast, Kellis has moved to the human genome, which he has so far compared with those of the mouse, rat, and dog. His work is providing an intimate understanding of the human genome that may give drug developers new points of entry in their attempts to combat viruses and other causes of disease.
Kellis discusses his work as a computational biologist in this video clip.
Peptide "Legos" to make new drugs.
Problem: The market for protein-based drugs, such as insulin and human growth hormone, has doubled over the last five years, to more than $50 billion. But making therapeutic proteins is difficult. Unlike small-molecule drugs, such as aspirin, which can be synthesized chemically, proteins are typically made by genetically engineering bacteria, growing them, and extracting the final protein from them.
It would be simpler if synthetic chemists could make small protein snippets called peptides from their amino-acid building blocks and then stitch them into complex proteins. But the process for making peptides is inefficient for chains more than 15 amino acids long. (Most therapeutic proteins contain anywhere from two to 30 times that many amino acids.) And existing techniques for joining peptide fragments are impractical.
Solution: Organic chemist Jeffrey Bode has come up with a more versatile way to connect peptides. Bode’s group discovered that two chemical groups not normally involved in peptide synthesis react to form amide bonds, the key link between amino acids. The researchers synthesized peptides using the standard chemical recipe; they then attached one of their chemical groups to either end of each peptide. Placed in water, the peptides join together. The only byproduct is carbon dioxide.
Using Bode’s scheme, researchers can in theory click any two peptides together. That might eventually let drugmakers synthesize any protein from scratch. The result could be a surge in new proteinbased therapeutics, many of which would contain chemical groups that are hard for bacteria to manufacture, but which would increase the drugs’ stability and reduce their toxicity in the body.
Artificially firing neurons.
Even in the hypertechie milieux of MIT and Stanford science departments, Edward Boyden stands out, bubbling with brilliance, energy, infectious enthusiasms, imaginative approaches to impossibly ambitious projects–and through it all, at his most earnest, with a vibrant sense of play. At MIT, he got a bachelor’s in physics and bachelor’s and master’s in electrical engineering and computer science, all with a perfect grade-point average–and by the age of 19. Fascination with computing led him to neurobiology. “The brain is a three-dimensional mass of densely wired tissue,” he says. “It’s kind of the ultimate computer.”
He landed at Stanford, where he got his doctorate in 2005, aged 26. There he created an ingenious new technology for analyzing and even controlling any neural circuit, including those in the cerebral cortex–important in sensation, action, thought, emotion, memory. In the science of the cortex, perhaps the toughest problem is to determine how neurons interact with their near neighbors. The cortex has some 20 billion neurons, which have many different functions. For years, neuro-biologists have known various ways to measure the outputs of individual neurons when they fire. Yet no method had been found to control inputs, to find and deliberately activate cells of one particular type.
Boyden’s technology–an elegant, tricky piece of genetic engineering–gives scientists exactly that. It begins with a curious protein called Channel-rhodopsin-2, or ChR2. This protein normally sits in the cell membrane of a green algae, and when exposed to blue light it changes the cell’s electrical state. Working with Stanford colleagues and the German researchers who isolated the protein and the gene that encodes it, Boyden linked the gene for ChR2 to the gene for a protein that fluoresces when hit by green light, creating a single, novel protein. The researchers inserted the gene for this new protein into rat neurons. Under green light, the cells that made the combination protein glowed; blue light caused the neurons to fire.
Boyden had thus invented a precise, reliable neural switching system operating at thousandths of a second–the speed at which neurons naturally interact. He has since added genetic elements that control just which type of neuron makes the new protein. Introduced into, say, a mouse’s brain, the protein could highlight individual types of neurons and allow researchers to study their functions. One possible use, Boyden says, is to analyze the neural circuits that perform particular types of “computation,” such as decision-making. Clinical applications are potentially huge: delivered to the brain and activated by implanted optical fibers, the protein could give doctors the power to activate neurons with selected functions. That could give rise to radical new medical technologies to treat brain disorders such as Parkinson’s disease or even some types of blindness, Boyden says. He believes that activating the right neurons could change mental and emotional states–perhaps curing prolonged, profound depression. More grandly, he projects enhancements of human mental capabilities, even the control of behavior.
“We can maybe apply some simple versions of the technology to solving urgent questions of today,” Boyden says. “There’s a lot of synergy between that and the technology that we use to do very, very hard things, such as addressing psychiatric disorders, confronting questions of consciousness, making brain-machine interfaces–and on the way to climbing those mountains, we can have a lot of little picnics!”
Disposable AIDS diagnosis.
Utkan Demirci wants doctors to take one look at his invention and “trash it.” That’s no knock on the device, a fast, easy-to-use–and disposable–test that measures the concentration of CD4 cells in the blood; doctors use that number to monitor HIV infections.
The size of a business card, this microfluidic instrument provides an accurate cell count in less than three minutes. At less than a buck apiece, the tests, which could reach the market in a few years, could literally be a lifesaver in HIV-ridden poor countries.
Demirci developed his expertise in microfluidics as a PhD student in electrical engineering. As part of his thesis, he devised a way to print semiconductor polymers by shooting sound waves through a small reservoir of fluid, squirting out millions of droplets per second. Now an assistant professor in the Harvard-MIT Division of Health Sciences and Technology, he’s adapting his printing technique to give doctors a way to grow new organs for transplant patients. Many tissue engineers build organ-shaped scaffolds, then coat them with cells, using conventional ink-jet printing techniques. Alas, the process “cooks” most of the cells. By replacing the polymers in Demirci’s printer with cells, engineers can deposit one unharmed cell with each droplet. Demirci hopes to use the method to begin building his first organ next year. He’s starting with one of the most challenging: the human heart.
Find out about Demirci’s thoughts on interdisciplinary research in this video.
Decoding brain signals.
Today, researchers can record and interpret brain signals with such sophistication that “mind reading” is close to becoming a reality. One of the young leaders in the field is computational neuroscientist Liam Paninski, who uses statistics to decipher electrical signals from the brain.
Because neurons fire in complex patterns, it’s tricky to identify which neurons encode which actions and how stimuli provoke them. Paninski creates mathematical models to make sense of those patterns. As an undergraduate at Brown University, he developed an algorithm that decodes arm-movement commands from the brain. Equipped with this neural code, Brown neuroscientist John Donoghue developed an implant that lets paralyzed people use their minds to control a robotic arm, manipulate a cursor, or play video games (see “Implanting Hope”).
Now a professor at Columbia, Paninski is using his statistical methods to decode vision. In the future, he hopes, implanted “video cards” may restore sight to the blind by translating digital images into neural patterns. He’s also exploring ways to treat epilepsy; as researchers decode neural signals more precisely, Paninski hopes to one day create a complete map of normal brain activity. Using the map, researchers could detect deviations such as epileptic events. Paninski envisions a warning device that will recognize abnormal events early, so that patients can take drugs to stave off a seizure–or at least get to a safe place before it begins.
The $1,000 genome.
Biotechnologist and medical student Jay Shendure is revolutionizing genetics with a new way to sequence DNA.
In 2005, he used off-the-shelf parts to determine the order of all the DNA bases in a bacterial genome, at 20 times the speed and one-ninth the cost of traditional DNA sequencing. Shendure is now working to make the process even more efficient; by 2015, he says, it may enable biologists to sequence a person’s genome for just $1,000.
The technique builds on polony sequencing, a method developed in George Church’s lab at Harvard. Shendure spreads millions of tiny beads on a glass slide, each attached to a small DNA fragment. He then adds fluorescently labeled DNA bases. The bases bind to short, complementary DNA sequences, and a standard fluorescence microscope records which base is at each position on a fragment.
Shendure next plans to use the technique to sequence the genome of a lung tumor in order to identify the genetic mutations that caused it.
A vision in bacteria.
Synthetic biologist Christopher Voigt has created an unusual image: the Virgin Mary on a lawn of E. coli. In turning microbes into a “photographic” medium, Voigt and his team have illustrated his approach to synthetic biology: creating genetic parts that can be used interchangeably to achieve different results. They hooked a light receptor from blue-green algae to a protein that normally controls E. coli genes’ response to the cell’s surroundings.
They then swapped one of these genes for one that turns a certain chemical black. The finished bacterium turned black in the dark; by holding a stencil between a Petri dish and a light source, the researchers could impose an image onto the bacteria.
Though Voigt says the bacteria are just a “toy technology,” he is using the same precepts to develop parts for more complex organisms, such as bacteria that spin spider silk or target and kill cancer cells.