When RNA rules

Image of Arabidopsis thaliana seedlings

Whitehead postdoctoral fellow Allison Mallory achieved dramatic results in Arabidopsis thaliana seedlings by overexpressing a single microRNA gene. The mutant plant (right) has fused seed leaves that form cup-shaped structures.

Image by Allison Mallory

May 5, 2006

Tags: Bartel LabRNA

What do newly discovered molecules called microRNAs and the Internet have in common? Both reshaped entire fields in the past decade, says Whitehead postdoctoral fellow Andrew Grimson.

“That’s a fairly grandiose claim for microRNAs,” acknowledges Grimson, who studies them. “But the discovery of the widespread role of these molecules changed the landscape of biology very quickly.”

“Labs across the world, working on a variety of biological questions, are now integrating microRNAs into their research,” says David Bartel, Whitehead Member and Howard Hughes Medical Institute investigator.

Bartel and his colleagues have helped to fuel the frenzy by identifying hundreds of the small RNA molecules and providing compelling evidence that they regulate the production of thousands of proteins in plants and animals.

Until the early 1990s, no one had a clue about microRNAs, which flew under the radar because of their tiny size. Each one contains only 21 to 24 nucleotides, or letters of the genetic alphabet, so scientists simply missed them. Victor Ambros’s group found the first microRNA—lin-4—in 1993 at Harvard Medical School while studying a mutation in the worm Caenorhabditis elegans.

Another Harvard researcher detected a second microRNA in 2000. One year later, the floodgates opened with the discovery of nearly a hundred in worms, insects and humans. At this point researchers began calling these tiny regulatory molecules “microRNAs.”

The discoveries changed conceptions of RNA. Scientists have known for decades that RNA molecules serve as messengers and translators, building proteins from DNA sequences. But microRNAs determine which DNA sequences get translated in a given cell, a responsibility once considered the purview of proteins known as transcription factors. MicroRNAs essentially choreograph biological ballets, helping to determine where and when proteins can appear to perform. Thus RNA can add “regulator” to the roles listed on its résumé.

MicroRNAs bind to messenger RNAs that code for proteins involved in activities ranging from development to cancer, and disrupt the production of these proteins. In humans, microRNAs regulate roughly one-third of protein-coding genes, and that’s a conservative estimate.

Going through the genome

“This is the first discovery of a broad biological mechanism that’s been made since genomics,” says Nobel laureate Phillip Sharp, who is investigating how microRNAs work at MIT, where he is an Institute Professor.

Scientists determined the scope of microRNA activity in a matter of years by mining recently published DNA sequences. Bartel, an RNA biochemist, and computational biologist Christopher Burge of MIT played a leading role. They collaborated to develop computer programs that scanned genomes to identify microRNAs and their messenger RNA targets. Their work helped to ignite interest in microRNAs as biologists in labs around the world realized the tiny molecules regulate a large portion of the protein-coding genes in plant and animal cells.

“Computational work has produced a very big picture of what microRNAs are likely to be doing in a very short time,” says Sharp. “ “It feels like the field is moving at warp speed,” agrees Burge, a Whitehead Career Development Associate Professor of Biology. “Genomic approaches have provided a number of important insights, and there has been nice synergy with molecular and biochemical studies.”

Finding the first microRNAs

Rosalind Lee and Rhonda Feinbaum, researchers in the Ambros lab, were conducting painstaking experiments on C. elegans when they bumped into the first microRNA.

They knew that early development of worm larvae required proper levels of the novel protein lin-14. They also knew that something was regulating those levels and assumed it was another protein, so they set out to isolate the gene for that protein. The result amazed them.

The gene fell on a stretch of DNA once termed “junk” by some, a stretch outside the protein-coding region of the chromosome. It appeared to code for a small RNA molecule— lin-4—that somehow regulated lin-14 levels.

The researchers wondered if lin-4 was an esoteric molecule or a harbinger of a new class of RNAs. “We had no basis for saying that lin-4 was part of something much broader,” says Ambros, who now works at Dartmouth Medical School.

His lab had no luck searching for additional RNAs in the next few years. He was thrilled when researchers in the lab of Harvard Medical School’s Gary Ruvkun discovered another gene in C. elegans that coded for a small RNA called let-7 in 2000. In addition to cloning let-7, Ruvkun’s group examined the genomes of a number of other animals and found the gene for let-7 in most of them. The study foreshadowed the role of genomics in later research.

In 2001, Rockefeller University associate professor Thomas Tuschl (formerly a postdoctoral fellow in the Bartel lab), Ambros and Bartel independently found dozens of additional small RNA genes in worms, flies and humans and decided to call them microRNAs.

Leveraging genomics

Bartel realized he needed to look outside the toolbox of classical biology. In 2001, he approached Burge—who had previously developed algorithms to identify protein-coding genes in the human genome—and Lee Lim, who had just completed his PhD training with Burge. The researchers jumped at the chance to explore a new class of genes. Lim worked jointly with the two labs to write a computer program that could scan DNA sequences and predict microRNA genes.

He started by examining known microRNAs. Each microRNA is generated from a piece of RNA that folds back on itself to form a structure that resembles a hairpin. Lim scanned the genome of C. elegans for DNA sequences that would give rise to hairpins after being transcribed into RNA. He then looked for ways to further refine the search.

The double-stranded RNA of a hairpin is chopped and processed into a single-stranded microRNA by proteins called Drosha and Dicer. But apparently these proteins don’t recognize every hairpin. Lim whittled down the list of potential microRNAs by eliminating DNA templates for hairpins that lacked Dicer-friendly characteristics.

Lim then screened the remaining microRNA candidates by comparing the genomic sequence of C. elegans with that of the related worm C. briggsae. He reasoned that most of the genuine microRNAs, those performing critical biological functions, would be conserved across species.

Eventually, the team showed that the human genome contains more than 200 microRNA genes. “We were excited to find new microRNAs,” says Burge. “But then the big question was—what do they do?”

This question had been largely answered in plants. Matthew Jones-Rhoades, a graduate student in the Bartel lab, had discovered that plant microRNAs have extensive and highly conserved pairing to plant messenger RNAs, so he could easily identify many targets of the plant microRNAs.

“At a time when we had about 50 plant targets, we were still in the dark regarding which genes were targeted in animals,” says Bartel.

Benjamin Lewis, a graduate student in both the Bartel and Burge labs, developed a second computer program to bridge this gap. He took the sequences of known micro-RNAs, scanned animal genomes for corresponding messenger RNA targets and, like Lim, used conservation across species to screen the results. The goal was to find many more conserved microRNA-mRNA pairings than would result by chance. But the initial program failed to deliver.

The researchers then tried another twist. Previous work showed that some microRNAs pair only partially with their mRNA targets, so the team hypothesized that one part of each microRNA sequence might be particularly important. They were right. Lewis hit the jackpot when he required perfect pairing near one end of the microRNAs. He found tiny sequences, matching short stretches of microRNAs, conserved much more frequently than chance would dictate in the mRNAs of mice, rats and humans.

Lewis named the critical stretch that matches targeted mRNAs the “seed” of the microRNA. The discovery of the seed gave scientists working on the biochemical interaction between microRNAs and mRNAs a big boost. It also allowed Bartel, Burge and Lewis to move forward with predicting targets.

They showed that many animal microRNAs have hundreds of conserved targets involved in a variety of processes, and in January 2005, they conservatively estimated that micro-RNAs regulate one-third of protein-coding genes in humans. This was a shock, as each plant microRNA appears to have just a few targets linked to development.

By the end of 2005, Kyle Kai-How Farh, another graduate student in Bartel’s lab, together with Andrew Grimson, showed there is also a large potential for species-specific targeting, and that in many cases protein-coding genes are evolving to avoid pairing with microRNAs. Thus micro-RNAs are affecting the majority of human protein-coding genes, at either a functional level or an evolutionary level.

Springboard for new studies

While the human genome is clearly full of potential microRNA targets, scientists in the lab have confirmed only a handful of interactions between mammalian microRNAs and mRNAs in living cells. Investigators are just beginning to use classical tools to probe the functions of the interactions identified computationally by Bartel and Burge, who are refining their computer programs and designing experiments to test past predictions.

“We’re improving the prediction programs to make them more inclusive and more accurate, and we’re sequencing millions of small RNAs in plants and animals to get a clearer picture of what’s really in the cell environment,” says Bartel.

Graduate student Graham Ruby, for example, is overhauling Lim’s microRNA prediction program. The original application missed many real microRNAs, and Ruby hopes to catch some of the molecules that fell through the cracks. Lim narrowed the list of microRNA precursors by scoring each hairpin according to its microRNA-like characteristics. Ruby adds a new twist. His program includes more than one round of scoring, like the American Idol show. After each round, he eliminates the lowest-scoring hairpins from the pool of candidates. He examines the rejects and uses their characteristics to fine-tune the scoring criteria for the next round, which should make predictions more accurate.

Other researchers in Bartel’s lab are working to determine the mechanism by which microRNAs lower protein levels, as much of it remains a mystery. The picture is clearer in plants, where microRNAs pair fully with and direct the cleavage of messenger RNAs.

But most of the new studies on microRNAs deal with their specific functions. Cancer researchers are particularly interested in the tiny RNAs, as many of them appear to regulate cell proliferation. Several papers last year confirmed this link. Gregory Hannon of Cold Spring Harbor and Scott Hammond of the University of North Carolina, for example, showed that overabundance of a specific group of micro-RNAs probably contributes to human B cell lymphomas.

“Studies are beginning to show the relevance of micro-RNAs to human disease,” says postdoctoral fellow Michael Lam, who is working with mice in Bartel’s lab to probe some of the other microRNAs connected to cancer.

“It’s exciting to watch the parallel currents in microRNA research,” says Ambros. “As a classical geneticist, I find it interesting to know how particular micro-RNAs work in particular situations. But I’m also intrigued by the work of people such as Dave Bartel, who are taking a more genomic view and discerning general patterns of microRNA function and evolution.”

“This will occupy thousands of people for years,” Sharp says. “It will take decades to work out the specifics of many different microRNA-regulated processes and integrate those into whole-organism biology.”


Running interference

RNA interference (RNAi) advanced genomic exploration a few years before the abundance of microRNAs was recognized. Scientists found they could knock down the output of genes by introducing double-stranded RNA into an animal. They soon found that this double-stranded RNA was processed into short strands of RNA, and that these small interfering RNAs (siRNAs) pair to and direct the cleavage of messenger RNA molecules that control protein production.

Sound familiar?

It turns out that microRNAs and siRNAs use similar cellular machinery to achieve their goals. Both rely on a protein called Dicer for processing, and a “silencing complex” facilitates their interaction with mRNAs. Within this silencing complex, the two types of RNA molecules both can direct messenger RNA cleavage when they have extensive pairing to the messengers, or dampen protein output by other means when pairing is not as tight.

However, a microRNA comes from a single strand of RNA coded in the genome that folds back on itself to form a structure resembling a hairpin, while an siRNA comes from a long piece of double-stranded RNA.

Written by Alyssa Kneller.

This article first appeared in the Spring 2006 issue of Paradigm magazine.

Whitehead Institute is a world-renowned non-profit research institution dedicated to improving human health through basic biomedical research.
Wholly independent in its governance, finances, and research programs, Whitehead shares a close affiliation with Massachusetts Institute of Technology
through its faculty, who hold joint MIT appointments.

© Whitehead Institute for Biomedical Research              455 Main Street          Cambridge, MA 02142