MIT reports new twist in microRNA biology

Computational biology group identifies new mechanism of gene regulation,
MIT scientists have found a new way that DNA can carry out its work that is about as surprising as discovering that a mold used to cast a metal tool can also serve as a tool itself, with two complementary shapes each showing distinct functional roles.

Professor Manolis Kellis and postdoctoral research fellow Alexander Stark report in the Jan. 1 issue of the journal Genes & Development that in certain DNA sequences, both strands of a DNA segment can perform useful functions, each encoding a distinct molecule that helps control cell functions.

DNA works by complementarity: paired DNA strands serve as a template for each other during DNA replication, and ordinarily only a single DNA strand serves as a template to produce RNA strands, which then go on to produce proteins. The process is similar to the way each bump or dent in a mold is paired with a corresponding dent or bump in the resulting molded object.

While many RNAs are eventually translated into proteins with specific functions, some RNA molecules instead act directly, carrying out roles inside the cell. Certain RNA genes, known as microRNAs, have been shown to play important regulatory roles in the cell, often coordinating important events during the development of the embryo. These microRNAs fold into relatively simple hairpin structures, with two stretches of near-perfect complementary sequence folding back onto each other. One of the two "arms" of a hairpin is then processed into a mature microRNA.

The surprising discovery is that for some microRNA genes, both DNA strands, instead of just one, encode RNA, and both resulting microRNAs fold into hairpins that are processed into mature microRNAs. In other words, both the tool and its mold appear to be functional. Kellis and Stark found two such microRNA pairs in the fruit fly, and eight more such pairs in the mouse.

The idea that there could be such dual-function strands, where both DNA strands encode functional RNA products, "had never even been hypothesized," Kellis says. But follow-up work confirmed that they did indeed function in this way. The work suggests that other such unexpected pairings, with both DNA strands encoding important functions, may also exist in a variety of species.

This discovery builds on a similar, earlier surprising finding about microRNA regulation. In December, Stark and Kellis reported that both arms of a single microRNA hairpin can also produce distinct, functional microRNAs, with distinct targets. Together, these two findings suggest that a single gene can encode as many as four different functions--one hairpin from each of the two DNA strands, and then one microRNA from each of the two arms of each hairpin.

These recent papers are the latest example of the power of using computational tools to investigate the genomes of multiple species, known as comparative genomics. The Kellis group has used this approach to discover protein-coding genes, RNAs, microRNAs, regulatory motifs, and targets of individual regulators in diverse organisms ranging from yeast and fruit flies to mice and humans.

"This represents a new phase in genomics-making biological discoveries sitting not at the lab bench, but at the computer terminal," Kellis says.

Kellis is the Karl Van Tassel Career Development Assistant Professor in the Department of Electrical Engineering and Computer Science and an associate member of the Broad Institute. He grew up in Greece and France, earned his B.S., M.Eng. and Ph.D. from MIT, and was appointed to the faculty in 2004. At 30, he has already earned numerous awards and accolades, including a place on the list of the 35 top innovators under 35 by Technology Review magazine in 2006.

Kellis' work is supported in part by grants from the National Institutes of Health and the National Science Foundation. Alexander Stark is supported by a Human Frontier Science Program fellowship.

MIT finds key to avian flu in humans

Transmission electron micrograph of avian influenza A H5N1 viruses (purple) released from infected human cell (blue).

MIT researchers have uncovered a critical difference between flu viruses that infect birds and humans, a discovery that could help scientists monitor the evolution of avian flu strains and aid in the development of vaccines against a deadly flu pandemic.

The researchers found that a virus's ability to infect humans depends on whether it can bind to one specific shape of receptor on the surface of human respiratory cells.

"Now that we know what to look for, this could help us not only monitor the bird flu virus, but it can aid in the development of potentially improved therapeutic interventions for both avian and seasonal flu," said Ram Sasisekharan, MIT Underwood Prescott Professor of Biological Engineering and Health Sciences and Technology, and the senior author of a paper on the work that appears in the Jan. 6 issue of Nature Biotechnology.

Flu viruses come in many strains, and not all of them can infect humans. Strains known as H1 or H3 have "jumped" from birds to humans and hence are tailored to attack cells of the human upper respiratory tract. H5 strains are usually confined to birds, but when they do infect humans they can have very high fatality rates.

In the past decade, isolated outbreaks of avian flu (H5N1) in humans have raised concerns that a deadly pandemic could arise if the avian flu evolves to a form that can easily infect humans and pass from person to person. Some scientists believe such an outbreak could rival the 1918 "Spanish flu" that killed 50 million to 100 million people worldwide.

Scientists already knew that whether an influenza virus infects humans depends on whether its hemagglutinin, a protein found on the virus surface, can bind to sugar (or glycan) receptors in the respiratory tract. Human respiratory cells have glycan receptors classified as alpha 2-6; avian respiratory cells' glycan receptors are known as alpha 2-3. This classification is based on how the sugars are linked together when they are displayed on cells.

Until now, scientists had believed that a genetic switch that allows the virus to bind to alpha 2-6 receptors instead of alpha 2-3 receptors is responsible for avian viruses' ability to jump to humans.

The MIT study shows that that view does not adequately explain how viruses evolve to infect humans. The new work reveals that, more specifically, it is the ability of a flu virus to bind to a certain shape, or topology, of specific alpha 2-6 glycan receptor that determines whether it will infect humans.

Alpha 2-6 glycan receptors come in two shapes--one that resembles an umbrella, and another that resembles a cone. The MIT team found that to infect humans, flu viruses must bind to the umbrella-shaped alpha 2-6 receptor.

Thus, Sasisekharan and his team have redefined the host receptor for influenza and the criteria for how H5N1 can jump to humans. They did so by showing that the shape of the sugars--and not the type of linkage--is the key determinant for human adaptation of these deadly viruses.

This new interpretation explains inconsistencies that plagued the previous model, according to Sasisekharan. For example, some flu strains that can bind to alpha 2-6 receptors do not infect humans very well. It turns out that those viruses bind to cone-shaped alpha 2-6 receptors, which are present in the human respiratory tract but in much smaller numbers than umbrella-shaped alpha 2-6 receptors.

This new paradigm should help researchers develop a better way to track the evolution of avian flu leading to human adaptation, Sasisekharan said. Now, they know to look for avian viruses that have evolved the ability to bind to umbrella-shaped alpha 2-6 receptors.

That knowledge could help them create vaccines tailored to combat a potential pandemic. Similarly, these findings will help in the development of more effective strategies for seasonal flu, which still is a leading cause of death.

"Subtle changes in influenza viruses over time can dramatically influence the likelihood that these viruses will be able to infect human populations, and this is a huge concern," said Jeremy Berg, director of the National Institute for General Medical Sciences, which funded the research. "This work enables researchers to look at flu viruses in an entirely new way. Dr. Sasisekharan's team achieved this through a multifaceted approach that combines laboratory experiments with the 'mining' of NIH-supported databases, leading to new insights into how the flu virus can adapt to a human host."

Other authors of the Nature Biotechnology paper are Terrence Tumpey of the Centers for Disease Control and Prevention; Aarthi Chandrasekaran, graduate student in MIT's Department of Biological Engineering (BE); Aravind Srinivasan and Karthik Viswanathan, postdoctoral associates in BE; Rahul Raman, research scientist in BE; S. Raguram, visiting scientist in BE; and Viswanathan Sasisekharan, visiting scientist in the Harvard-MIT Division of Health Sciences and Technology.