RNA can bind and sense the shapes of other molecules by feeling them with its backbone—
and not just its bases. What gives RNA molecules this remarkable versatility?
he molecular world has always been part of my mental furniture. I grew up on the outskirts of Sandia National Laboratory in Albuquerque, New Mexico, famed for its research on energy, materials, and nuclear weapons. My dad was a physician and biomedical researcher who loved chemistry above all things, and who would interpret all of life’s vicissitudes in terms of some obscure chemical reaction or metabolic dysfunction. Learning chemistry, therefore, became a necessity for basic communication with my father. My neighbors were mostly physicists who would bring home spare bits and pieces from labs around the country. My friends and I sprayed rainbows of color on the bedroom wall with old prisms and played with a...
Although chemistry and physics were a sort of alternative reality for me growing up, those fields did not become substantively important to me until high school, when I learned about chemical synthesis. It is one thing to appreciate the microscopic world and another to realize that you can actively manipulate it with your own hands, building new molecules just as you would build a house or a sculpture. Sandia lab was full of scientists who—without many students of their own—would come teach at my public high school, offering such challenging courses as multistep organic synthesis. Under the enthusiastic guidance of our guest teacher, we synthesized and purified all kinds of compounds that smelled wonderful or bounced when you threw them against the wall.
The fun of making molecules led me to study organometallic chemistry in college and graduate school. Organometallic compounds, which contain a reactive metal ion decorated with elaborate appendages that direct its function, can be devilishly hard to make and often explode when exposed to air, making their synthesis particularly entertaining.
But during my first year of graduate school at Columbia University, my interests began to broaden into biology. I took classes with Ronald Breslow, Koji Nakanishi, and Jackie Barton, who each exemplified how chemists can play a powerful role in manipulating and understanding biological systems. Jackie Barton in particular had shown that metal complexes with unique shapes and chiralities can recognize DNA in useful ways that could potentially influence gene expression. After many discussions with Jackie, I decided to join her lab, working on new metal complexes that could hug the duplex DNA and detect tiny structural differences. The project opened my eyes to the challenging problem of nucleic acid molecular recognition: How does a seemingly uniform polymer like DNA recognize small molecules, proteins, and other nucleic acids in such strikingly specific ways? Is it all about the bases, or does the backbone or the local variations in DNA shape influence the way DNA interacts with other molecules?
he more I considered the interactions of DNA, the more I started to think about RNA. This free-wheeling relative of DNA offered a better way to learn about interactions between the shape of nucleic acid and its function, in part because RNA does not always rely on base-pairing to fold into its final form. In fact, RNA is not always helical or single-stranded and can adopt elaborate, three-dimensional shapes much like a globular protein.
The complexity of RNA structure was brought into sharp focus by the discovery of ribozymes: RNA molecules that could catalyze chemical reactions.1 Thomas Cech had discovered that a piece of RNA within the single-celled organism Tetrahymena thermophila could snip itself out of RNA segments and then ligate the flanking pieces back together in a process called RNA self-splicing. At the same time, Sidney Altman discovered that a different RNA (called ribonuclease P) trimmed transfer RNA (tRNA) molecules to their proper size, indicating that catalytic RNA molecules (ribozymes) were widespread and of general importance. Although RNA splicing had been discovered previously, it was thought to be catalyzed by a big molecular machine, called the spliceosome, which contained proteins. Cech and Altman’s discoveries (for which they would later earn a Nobel Prize) were an eye-opener for the field—they had shown that certain RNA molecules were not just passive informational tapes—they could perform chemistry! Soon thereafter, I wrote to Tom Cech at the University of Colorado and told him that I wanted to join his lab as a postdoc in order to study how the Tetrahymena ribozyme recognized its RNA substrates. Tom agreed and provided encouragement throughout the project, although he insisted that I go skiing whenever I began to take the project too seriously. As this happened frequently, I soon became a much better skier.
In addition, I learned that the Tetrahymena ribozyme was not merely binding to its RNA substrate through Watson-Crick base-pairing. Rather, the ribozyme also clamped onto the RNA’s sugar-phosphate backbone. Strong interactions involving the 2′-hydroxyl group of the ribose sugar are extremely important for molecular recognition by RNA and they provide considerable stability to RNA-RNA complexes.2
When starting my own faculty position, I decided to look at how RNA molecules fold into their complicated tertiary structures, and to elucidate the diverse types of molecular interactions that hold RNA molecules together.
To identify the different tricks that RNA uses to touch other molecules, I needed an RNA system that was highly dependent on backbone interactions for its shape and function. Several years earlier, Philip Perlman reported a new kind of ribozyme that could catalyze its own splicing from flanking RNA molecules.3 Classified by Francois Michel as “group II introns,” they differed from Cech’s self-splicing intron because they were much larger (usually ~1000 nucleotides) and they released themselves as “lariat molecules,” which are branched RNAs that look like a noose. What’s more, these RNAs were extremely interesting structurally and evolutionarily. Perlman had shown that the most genetically conserved and catalytically important region of these introns, known as domain V (DV), could be cut away from the intron and added back as a separate molecule, thereby restoring splicing activity. In other words, you can take the heart out of this intron, float it back in, and the intron is no worse for wear.4 Since DV is completely base-paired to itself, it cannot use conventional Watson-Crick interactions to find its way into the ribozyme frame. Instead, it must feel its way to the right location through unusual interactions involving the sugar-phosphate backbone.
Perhaps because of these sequence-independent backbone interactions, it had been largely impossible to predict the overall shape of group II introns from base sequence. Only a chemical approach that probed the role of each individual nucleotide atom would unlock the architectural features of this RNA. Thus, DV represented an ideal focus for a young chemist who wanted to study RNA tertiary structure. It was small enough to make “by hand” on my oven-sized benchtop DNA-RNA synthesizer, and its association with the rest of the intron RNA could be monitored through classical enzyme kinetic analysis. My students and I therefore proceeded to make many synthetic DV variants, systematically probing the structural basis for DV function one atom at a time.
After many experiments we showed that DV had two faces, a “binding face,” which allowed it to interact with the rest of the intron, and a “ chemistry face,” where DV atoms formed an active site that cleaved RNA and DNA.4 Work by Joseph Piccirilli and Alain Jacquier had shown that many backbone atoms on the chemical face bound metal ions that were important for catalysis, suggesting that DV contains an enzyme-active site where metals play a starring role.
Although we had made progress in understanding DV, this tiny domain did not function in isolation—the rest of the intron was important as well. However, the remaining 800 nucleotides could not possibly be studied, atom by atom, by stitching them together on my desktop synthesizer. To examine the role of atoms throughout the intron RNA, we needed to develop combinatorial methods (approaches where thousands of atoms are mutated at once) to introduce single-atom mutations within the RNA. We then sifted through this pool of mutant molecules, pulling out the ones that retained ribozyme function and linking the sites of mutation with specific functions. These methods allowed us to visualize the overall organization of the intron, but they would never give us a high-resolution view of the active site. It was like peering at a much-loved relative through a piece of filmy glass.
To really see this intron—clearly and in focus—we had to solve its crystal structure, which would be a feat for an RNA molecule of this size. However, over the years we had screened a menagerie of different group II introns to identify one that was likely to crystallize: one that was compact, highly stable, and highly reactive.
We finally found a winner when Navtej Toor, a postdoc in the lab, crystallized a small new subtype called a group IIC intron (~400 nucleotides in size).5 After solving and refining the structure in collaboration with Kevin Keating, Sean Taylor, and Raj Rajashankar, Nav showed that the intron is assembled from an elaborate network of unusual tertiary interactions that involve extensive base-stacking and backbone hydrogen bonds. The bulk of the intron domains form a stable shell that enfolds catalytic DV, specifically orienting its functional groups to form a highly reactive site of chemical catalysis, where RNA and DNA bonds can be made and broken.4,6
The 3D structure explained and confirmed decades of biochemical and genetic studies of the intron, and provided new insights into the basic mechanisms of RNA assembly. But perhaps the most interesting insight was the glimpse it offered into our molecular history.
We know that group II introns are mobile genetic elements that jump around, copying and inserting themselves into new genomic locations and new hosts. Through this process, they bring new genes with them, or they chop up long genes into multiple pieces that can be used in various combinations, potentially leading to great diversity of expressed protein types (alternative splicing). Through eons of jumping around, group II introns may have extensively shaped the construction of early terrestrial genomes, leading to the complex genomic organization that is typical of many organisms today.4 The crystal structure was obtained on a particularly ancient lineage of group II introns, providing a clear glimpse of the molecular trickster that put its stamp on our genetic history.
The group II intron structure may also help explain how RNA is spliced in the nucleus of our cells. Given the similarities in mechanism between the spliceosome and the group II introns, it has been hypothesized that group II introns are evolutionary predecessors of our spliceosome. If true, group II introns could be valuable model systems for understanding spliceosomal function and active site architecture. Indeed, the crystal structure of the group II intron does suggest a specific network of tertiary interactions that one can test genetically within the spliceosome.7 Time will tell whether these systems are related and whether group II introns have more stories to tell us about RNA structure and evolution.
s we studied forces that help bind RNA strands together, we became curious about the mechanisms that blow them apart. Once RNA structures have formed, they are often very stable. And as important as RNA helices and tertiary structures actually are, there is a time in the life of every cell when even the most important RNA has to be refolded, disassembled or recycled so that something new can happen. Not surprisingly, there is a huge family of motor proteins that exclusively perform this job, and they are used in almost every aspect of cellular metabolism. Classified originally by Eugene Koonin in 1993, the superfamily 2 helicases (SF2) are now known to have a diverse array of functions in RNA folding and remodeling. We set out to study SF2 remodeling proteins and to define their basic capabilities as molecular motors.8
In the early 1990s, Stewart Shuman showed that the NPH-II helicase, an SF2 protein from pox viruses, can unwind long RNA duplexes. With his help, using NPH-II as a model system, we showed that NPH-II unwound RNA duplexes in a stepwise manner through a mechanism that resembles a molecular wire stripper: while the helicase speeds along one of the two strands, the second strand of the duplex is peeled off. We then put NPH-II to an important test: If NPH-II can shove an RNA strand out of its path, might it also displace tightly bound proteins? If true, this meant that SF2 proteins were likely to have unanticipated functions in the cell. For example, they could be specialized devices for dismantling RNA-protein complexes.
A postdoc in my lab, Eckhard Jankowsky, tested this theory by building a model system in which he placed a protein roadblock in the path of an oncoming NPH-II molecule. Much to our surprise, protein displacement by NPH-II was fast and efficient, giving NPH-II a new property, termed RNPase activity, or enzymatic protein displacement from RNA.9 Now that the biological roles of many SF2 proteins are finally being elucidated, it is becoming clear that protein displacement is among the more common roles of SF2 proteins in the cell. Indeed, we and others have now shown that “helicase” or unwinding activity by SF2 proteins is only one of their many functions and that different family members serve as translocases (moving stepwise along either single- or double-stranded RNA), ATP-dependent RNA binding proteins, or ATP-dependent annealing factors that can stabilize folding intermediates along an RNA assembly pathway. Using the same molecular building blocks (the ATPase core domain of all SF2 proteins looks almost identical), nature has created a diverse family of molecular motors for regulating behavior of nucleic acids in the cell.
hroughout these studies, I sought a system that would help us understand the function of SF2 proteins in a biologically relevant context, because most of them function as small cogs in bigger machines. To this end, I began working on the NS3 helicase from hepatitis C virus (HCV). The NS3 protein is but one component in a large replication complex that copies the virus. Replication is mediated by a complicated assembly of viral and host proteins, together with RNA from the viral genome. It is disappointing that, with approximately 2% of the US population infected with HCV and given that it is the leading cause of liver cancer, there is little known about the structure or function of the HCV replicative machinery. Insights into its function might go a long way toward helping find appropriate drugs for treatment of HCV.
To better mimic the natural context for NS3 function, we set out to study the mechanism by which full-length NS3 unwinds RNA, which is its natural target.8 Using bulk assays and single-molecule experiments conducted in collaboration with the labs of Carlos Bustamante and Taekjip Ha, we observed that NS3 unwinds nucleic acids through multiple dynamic processes that involve ATP-driven translocation, bursts of duplex unwinding, and slow stages of strand release. But we also showed that NS3 is an accessorized motor, which by itself is a relatively weak machine. To enhance its function, the conserved “helicase” core is appended to extra domains and partner proteins that enhance its activity.10 For example, binding of NS3 to RNA requires the presence of its appended protease domain, which was not presumed to play a role in helicase function. Similarly, the associated proteins and enzymes in the replicative complex all play a synergistic role in regulating one another’s activities. Together with Yale virologist Brett Lindenbach, our current work focuses on reconstructing the intact replicative machine, and understanding the role of the helicase and the polymerase enzymes in its function. We hope our work might lead to improved screens for drugs against HCV and new knowledge about the function of macromolecular machines.
We now know from the human genome project and from studies of the human “transcriptome” that the vast majority of our DNA does not encode proteins after all; rather, it encodes RNA. RNA is far more important in biology than any of us imagined even 5 years ago. Now more than ever, we must understand how RNA folds, how it serves as a scaffold and enzyme, and how it is taken apart by the engines of the cell. Studies of RNAs like group II introns and remodeling proteins like NS3 have shed a little light on this problem, but there is much more to learn. We are looking forward to the revolution in experimental biology that will illuminate the dynamic world of RNA gymnastics.
Anna Marie Pyle is the William Edward Gilbert Professor of Molecular Cellular and Developmental Biology and Professor of Chemistry at Yale University. She is an investigator of the Howard Hughes Medical Institute.