How the Ribosome Works

Frank Joachim. American Scientist. Volume 86, Issue 5. Sep/Oct 1998.

Protein synthesis is one of the fundamental life-sustaining processes in the cell. Each protein is synthesized by stringing together its basic building blocks, called amino acids, according to a sequence specified by the genetic blueprint that resides in the cell’s DNA. The resulting polypeptide chain then folds in a unique way to form a functional protein molecule. The marvelous machine that makes it all happen is an organelle composed of RNA and protein, discovered over 40 years ago and named the ribosome. Because of its crucial importance, it is ubiquitous in cells, be they bacterial, plant or animal.

Though a mere 25 nanometers in size, the ribosome is of such complexity that it keeps its own scientific research community of some 1,000 people busy around the world. Hundreds of articles appear each year characterizing particular reactions or providing information on the structure or relative distances of some of its components, yet, until recently in all of the 40 years that the ribosome has been known, nobody had “seen” it with any clarity. Since the ribosome interacts with different molecules in some unknown “lock and key” fashion, it is clear that knowledge of the three-dimensional shape is crucial for understanding the dynamics of protein synthesis. Ultimately, one wishes to know the structure of the ribosome at atomic resolution, but this goal is still quite remote, since the ribosome of even one of the simplest organisms-the bacterium Escherichia coli-is made up of more than 50 components, mostly proteins plus three large RNA molecules. Ribosomes of higher organisms are even more complex and have more components. Scientists would like to learn the exact contribution each of these components makes to protein synthesis. Bacterial ribosomes are the targets of many antibiotics, and a better understanding of how the ribosome works will help towards making more efficient drugs.

It is now believed that all important functions of the ribosome actively involve the ribosomal RNAs as major players, whereas the ribosomal proteins act either as structural “glue” or as “helpers” that promote the specific binding reactions. This represents a complete reversal of the long-held view that the proteins perform all the important tasks and that the function of the ribosomal RNAs is merely to provide a skeleton that ensures the proper architecture of the assembly.

The past seven years or so have seen dramatic new insights brought about by the development of cryo-electron microscopy and methods for singleparticle reconstruction. Instead of trying to obtain crystalline arrangements of ribosomes, long thought to be a prerequisite for structural analysis, we extracted the information from thousands or tens of thousands of single particles and integrated these into a three-dimensional image.

The first such image, generated in my laboratory in 1991, resolved the characteristic asymmetric doughnut-shaped ribosome in which the boundaries of its two subunits can be recognized. Subsequent maps with a resolution of 20-25 angstroms were made three years ago and have provided the first detailed picture of the ribosome’s interior as well as the convoluted “landscape” of the sites onto which various components bind.

We have recently obtained a much refined three-dimensional map of the ribosome with the unprecedented resolution of 15 angstroms and are currently at work on a 12-angstrom map. At that resolution, one can begin to trace the double-stranded helices of individual RNA molecules. By imaging the ribosome with genetic templates, adapters and protein molecules attached to it, insights are now being gained rapidly by our group and others into the workings of one of nature’s most complex machines. As the new technique of three-dimensional cryo-electron microscopy of single particles unravels the intricacies of the ribosome, it is also gaining wide acceptance as an analytical tool useful for determining the structure of any large macromolecule.

Protein Synthesis

Protein synthesis begins with the genetic blueprint residing in a cell’s DNA. But DNA contains an archive of all of the cell’s genes and is of such primary importance that it never directly participates in protein synthesis. Instead, a copy is made of a segment of the DNA, corresponding to a gene, which specifies the order in which the amino acids of a single polypeptide chain should be strung together. This copy, or message, is in the form of RNA, a chemical relation of DNA. Messenger RNA (mRNA) carries the genetic message to the ribosome, where proteins are synthesized.

Both DNA and mRNA are composed of chemical units called nucleotides, to which different bases are attached. These bases are analogous to letters, of which the genetic alphabet contains only four. Each “word,” or codon, of the genetic lexicon is precisely three letters long and stands for one of the 20 amino acids found in proteins.

Clearly, this code poses a linguistic problem. The genetic words are spelled out in nucleotide bases, but the final product is a string of amino acids. The translation from the nucleotides of the mRNA to the amino acids in the polypeptide chain is accomplished at the ribosome, with the help of an adapter molecule, called transfer RNA (tRNA). tRNA is a large, elbow-shaped molecule, with an anticodon “plug” on the long arm that inserts into the codon “socket” of mRNA. An amino acid plugs into the other end of the tRNA molecule, at the tip of the short arm. tRNAs exist in essentially 20 varieties, each one specifically designed to accommodate only one of the 20 amino acids, for which it carries the anticodon that recognizes only the codon on mRNA specifying that particular amino acid.

Thus the ribosome is the place in the cell that brings together the mRNA template, the amino acids and the adaptor molecules that will carry out this translation. First, the template must be properly secured at the ribosome. Then, one by one, tRNA molecules carry the correct amino acids to the ribosome and place them in the protein in the precise order demanded by the genetic code in the mRNA template. Bonds, called peptide bonds, are formed between each incoming amino acid and the one preceding it in the sequence, until the entire polypeptide chain is completed.

Many ribosomes in plant and animal cells are attached to the membrane of a convoluted cellular compartment called the endoplasmic reticulum. Some of the newly formed proteins pass into the endoplasmic reticulum through a special membrane channel. In yeast and mammalian cells, this channel is made up of a protein called Sec61.

By necessity, the ribosome is highly complex because it is designed to ensure the faithful translation of the sequence of codons of the mRNA. Errors in translation produce dysfunctional proteins and drain the cell of resources. Thus, when we image the ribosome, we expect to see that it is a high-precision machine, with each component having a specific place on the ribosome that guarantees the perfect placement of each amino acid in the protein. When we explore the ribosome’s structure, we are primarily interested in clues about the way it interacts with the most important ligands: mRNA, tRNA and certain protein factors that catalyze or promote the reactions.

Approaches to Ribosome Structure

When ribosomes are isolated from a cell extract, two different fractions are obtained, corresponding to its two subunits-one of them large, called the 50S subunit, and the other small, called the 30S subunit. In the laboratory, the subunits can be induced to come together to form an assembly that is structurally similar to a ribosome in the native state. In approaching the structure of the ribosome, one can proceed along two different tracks: Try to solve the structures of the small and large subunits separately, or study the assembled ribosome. But one must keep in mind that assembly itself necessarily forces conformational changes in the small and large subunits, so it becomes important to look at both isolated and assembled subunits under close-tophysiological conditions to fully understand the ribosome.

It has been quite difficult to grow sufficiently good crystals of the whole ribosome to use for x-ray crystallography. One of the reasons is that the ribosome is much larger than even the largest proteins solved crystallographically to date. Another reason is its inherent structural variability.

The method that has recently provided a breakthrough brings together the techniques of cryo-electron microscopy (cryo-EM) and single-particle reconstruction. This method currently gives a resolution in the range of 15 to 20 angstroms, which is still far from the desired atomic resolution. Nevertheless, it has given the first insights into the inner workings of the ribosome. Moreover, it has become clear in recent years that the study of the structure and dynamic behavior of many large biological assemblies, such as viruses and ribosomes, will eventually require the integration of intrinsically low-resolution information, coming from cryo-EM of the entire structure, with atomic-resolution information on its components. The latter is currently only provided by x-ray crystallography and nuclear-magnetic resonance spectroscopy.

Cryo-electron Microscopy

Cryo-EM provides a way to image macromolecules in a close-to-native conformation by freezing them rapidly. The technique was first developed during the 1970s in the laboratory of Robert Glaeser at the University of California, Berkeley and was perfected early in the 1980s by Jacques Dubochet and his group at the European Molecular Biology Laboratory in Heidelberg. Before that, the only method available for visualizing macromolecules was the negative-staining technique that surrounds a molecule with a “cast” of a heavy-metal salt.

Although negative staining gives high contrast, the resulting image mainly contains information about the shape of the molecule, not its interior structure. And the shape it yields is somewhat distorted, since the method dessicates, and thus deforms, the molecule under study Pictures of single ribosomes prepared this way show round particles with few distinctive features, except in certain views that show a clear division into the two subunits. Pictures of the isolated subunits are more informative. For example, the large subunit often appears in a characteristic view called the crown view, recognizable by three protrusions from an overall round shape. The shape of the small subunit in these images varies between that of a rolling pin and a dog bone.

Following the protocol of cryo-EM, a droplet of a buffer solution containing ribosomes is placed on an electron microscope grid (a copper or molybdenum grid about 2 millimeters in diameter). The excess liquid is blotted from the grid, which is then plunged into liquid ethane kept at a temperature of nearly -196 degrees Celsius, the temperature of liquid nitrogen. On contact with the cool liquid, a thin amorphous layer of ice (in the range of 500-800 angstroms) is formed in which the ribosomes are held in random orientations. The grid is transferred to the microscope and maintained at liquidnitrogen temperature throughout the experiment. A picture is formed in the electron microscope with a radiation dose so low that virtually no damage is done to the sample. The resulting micrograph typically shows hundreds of ribosomes lying in different orientations.

The next challenge is to construct a three-dimensional image from these more-or-less random projections. This is the objective of single-particle reconstruction, a method that we have been working on in my laboratory for the past 20 years.

Single-Particle Reconstruction A macromolecule such as the ribosome can readily be reconstructed from its projections, provided that they occur in a range of orientations and provided these orientations are known. For example, it would be easy to obtain a series of projections by tilting the specimen grid in increments ranging from -60 to +60 degrees. This reconstruction technique is analogous to CAT-scan imaging of the human body and is known as electron tomography. The disadvantage of this approach is that the electron dose accumulated during the experiment is normally so large that it destroys the specimen. As a result, a three-dimensional image obtained this way would have little resemblance to the natural object. Instead, we have sought to develop methods of reconstruction that make use of projections of molecules exposed only once, with the lowest electron dose that produces a usable image.

The most important problem to be solved is how to find the relative orientations of the molecules. We developed two different, complementary approaches in my lab. One is a bootstrapping approach that yields a low-resolution threedimensional map for a structure that is entirely unknown. The other, projectionmatching approach can be used for any subsequent work, as it is based on comparing experimental projections with an existing three-dimensional map.

The booststrapping approach seeks to establish defined geometrical relations among a subset of the particles. The idea is to consider only those particles that present the same view on the untilted grid. (They can be sorted from particles presenting other views by the computer using a method of automated classification.) The important point is that particles showing a given view may occur in many different azimuthal orientations (that is, their orientation in the plane of the grid). If the grid is tilted by a sufficiently large angle (50-60 degrees), then those same-view particles will now present a variety of different views whose appearance depends on the precise azimuthal angle (which can be found from the untiltedgrid micrograph).

For illustration, we might consider how we would go about reconstructing a hand when we had many copies of it on the specimen grid, all lying “palms down” but in random rotations. In a view of the untilted grid, we gain nothing by having more than one hand to look at, since they are all seen from the same direction. The situation changes radically if we tilt the grid; now each of the hands looks different. Each gives us a separate piece of the information that we need to recover its three-dimensional shape.

It is easy to see that the directions of the projections obtained lie on a cone, and that such a projection set is sufficient to reconstruct the particle. Since the azimuthal angles are random, the technique is known as random-conical reconstruction. Several such reconstructions, based on particles in different orientations, may be combined into a first low-resolution map.

Other techniques, based on so-called common lines, are able to determine directly the angles of projection between views of the molecule and do away with the need to tilt the specimen, but they are unable to distinguish a left hand from a right. Such methods can therefore be used after a first lowresolution structure is already in place.

The projection-matching approach uses projections that are predicted based on an existing reconstruction. These predicted projections are compared in the computer with the experimental data so that more precise angles can be assigned. Thus this approach may be used to refine the angles of the initial experimental projection set or to obtain the correct angles for projections not initially considered.

For statistical reasons, the number of projections required is roughly proportional to the cube of resolution. The current record lies at 30,000, the number of images used to reconstruct the ribosome at a resolution of 15 angstroms. Some other hurdles-not merely of a statistical nature-must be overcome before we can obtain higher-resolution images. We often want to image the ribosome with various components required for protein synthesis attached to it. But the ribosome is a dynamic machine whose shape changes depending on what binds to it. It is therefore important to ensure that all of the ribosomes in the sample are bound in the same way and thus have assumed the same shape, since averaging over a mixed population produces a blurred three-dimensional map. A second limitation is posed by the stability and coherence of the electron beam used for imaging in the electron microscope. Higher resolution requires the high stability and coherent illumination that can be found in one of the new-generation instruments equipped with a fieldemission cathode.

The large number of images and the iterative nature of some of the image processing and reconstruction algorithms pose extraordinary computational challenges. Many of our computations are therefore performed at the National Center for Supercomputer Applications in Champaign-Urbana, Illinois.

What We See

At first glance, the cryo-EM map of the ribosome shows a particle of bewildering complexity. The two subunits are clearly resolved. It is evident, too, that they are held together by several bridges, two of which are massive, the others rather thin and pencil-like.

In addition to looking at whole ribosomes, we were able to image the individual subunits by cryo-EM. In order to make sense of these reconstructions and to understand how the individual subunits relate to the whole, we had to look for some landmarks. The large subunit has three protrusions, which we immediately recognize as the features that give rise to the “crown” appearance in the negatively stained preparations. Early work in the laboratories of James Lake at UCLA and Marina Stoffler-Meilicke at the Max-Planck Institute for Molecular Genetics in Berlin used antibodies to determine, by electron microscopy, the position on the ribosome’s surface of a number of the constituent proteins. These studies identified the globular protrusion to be the large-subunit protein that ribosome biologists knew as L1, while the knobby protrusion on the opposite side of the large subunit (best seen in Figure 6e) is formed by proteins L7 and L12. The protrusion in the middle of the large subunit (Figure 6f) is called the central protuberance, or in the jargon of electron microscopists, simply the head. The L7/L12 protrusion looks quite different under different conditions, and may be folded down against the ribosome’s surface or appear extended as a stalk. On the side facing the small subunit, the surface of the large subunit is marked by a deep groove running almost horizontally across it, termed the interface canyon, which was first observed in a reconstruction from negatively stained large-subunit specimens.

The small subunit can be divided into its anatomically named parts, called the head, neck and body. The head is triangular and curves sideways as it tapers off. In addition, below the neck the subunit sprouts a large extended mass called the platform. For later reference, we also note the cleft between the platform and the neck, believed to be the site where the tRNA anticodon contacts the codon on the mRNA. As an added curiosity, the small subunit has a thin extension at the bottom, which we have termed the spur.

Using these features on the subunits, we are able to describe the unique way they join to form the complete ribosome. Ribosomes assemble and disassemble continuously inside the living cell. Assembly takes place only after the small subunit encounters and binds to an mRNA molecule. That complex is then joined with a tRNA molecule, with the help of special proteins called initiation factors. Finally, the whole subassembly associates with a large subunit. The exact details of the assembly process remain a mystery that can only be solved when more is known about the structure of the subunits. The complete ribosome stays together as it proceeds through the message from start to finish until it reaches the final “stop” codon. At that point the ribosome dissociates into its subunits.

Without knowing all of the details of ribosomal assembly, we can infer the origin of the two massive bridges from the features of the individual subunits.

The uppermost bridge is created by the fusion of the small subunit head with the central protuberance of the large subunit. The central, most massive bridge derives from a fusion of the small subunit’s platform rim with the rim of the interface canyon of the large subunit. Both bridges are evidently required to hold the subunits in a defined configuration and to create a defined binding environment for the tRNA. As to the thin bridges, we can only speculate about their role. Are they used to stabilize the assembly? Or could they somehow facilitate coordination between the two subunits?

Looking at a side view of the ribosome, we were especially struck by the shape of the empty space between the subunits. This space looks as though it had been tailored precisely to accommodate the elbow-shaped tRNA molecule, with the anticodon end (on the long arm) sitting in the cleft of the small subunit and the other end (the short arm) in the interface canyon of the large one. The fit is so perfect, it leaves little doubt that the tRNA must pass through the ribosome in this space. It also makes absolute evolutionary sense.

Barring an extraterrestrial origin of life, the ribosome and tRNA must have coevolved during the first billion and a half years of life’s tenure on earth. Neither the ribosome nor tRNA alone can accomplish translation, and the correct interaction between the two is absolutely vital for the survival of an organism-and by extension, a species. As life forms grew in complexity from the initial single-celled organisms to more sophisticated multicellular forms, their ribosomes also acquired some new bells and whistles in order to become more efficient. However, the tRNAs of plants and animals are virtually interchangeable with those in the bacterium E. coli. We were not surprised, therefore, when cryo-EM revealed the same tRNA-shaped space in ribosomes from yeast, wheat germ and rat liver that we had previously found in the bacterial ribosome.

Another interesting clue we found to the way the ribosome might function is the presence of a channel through the neck of the small subunit and a tunnel that begins at the bottom of the large subunit’s interface canyon and exits at its back. Since the small-subunit channel leads into the cleft region where we expect the anticodon to contact the codon, the channel might provide a conduit for the incoming mRNA. This arrangement makes sense in terms of traffic control, since it keeps the L7/L12 end of the interface space open for incoming bulky molecules. The tunnel entry lies precisely where we expect the polypeptide chain to be strung together, so the tunnel itself (a feature first observed some ten years ago by the groups of Ada Yonath at the Weizmann Institute in Israel and Nigel Unwin at Stanford in low-resolution reconstructions of thin ribosome crystals) is likely to be the conduit through which the newly synthesized polypeptide chain exits the ribosome. Again, channel and tunnel occupy the same respective places in the ribosomes of both bacteria and yeast, reinforcing the view that they are essential parts of the translation apparatus, even for organisms widely separated in evolution.

The hypothesis that the tunnel through the large subunit serves as a conduit for the polypeptide chain has received additional experimental support for eukaryotes. Many ribosomes sit on the membrane of a structure called the endoplasmic reticulum (ER), and it is known that newly synthesized proteins enter the ER through a membrane channel formed by the Sec61 protein. In collaboration with Gunther Blobel and colleagues at the Rockefeller University, we looked at the relation between the yeast ribosome and the Sec61 protein channel. We found that the ribosomal tunnel aligns perfectly with the Sec61 channel in the membrane of the ER (Figure 9). From our work, we conclude that in the ER-associated ribosomes, the newly manufactured proteins pass through the large-subunit tunnel, through the Sec61 channel and into the ER.

The Elongation Cycle

Cryo-EM has given us the first threedimensional map of the ribosome, and with it the first opportunity to integrate structural information with information obtained by a variety of biochemical and biophysical experiments. Moreover, cryo-EM allows us to visualize tRNA and protein factors attached to the ribosome in functionally meaningful complexes. With this information, we can now test various models of how tRNA moves through the ribosome during protein synthesis. In essence, we have produced a series of snapshots of the ribosome as protein synthesis proceeds.

The process, as already noted, is initiated when mRNA binds to the small ribosomal subunit. Then the tRNA anticodon binds the appropriate codon of mRNA. Finally, the large ribosomal subunit attaches, with the mRNA molecule sandwiched between large and small subunits.

Recall that the purpose of the ribosome is to add new amino acids to a polypeptide chain. Therefore, it has long been assumed that the ribosome contains at least two sites, one to accommodate the tRNA holding the polypeptide chain, called the peptidyl site (P site), and the other to accommodate the tRNA holding an incoming amino acid, called the aminoacyl site (A site). More recently, a third site has been proposed, the site from which tRNAs exit the ribosome when they are no longer needed. This site is called the E site.

Starting protein synthesis poses a problem. Each amino acid must add to the polypeptide chain in the P site, but when synthesis begins, no chain exists. For this first step alone, a special amino acid called formylmethionyl or fMet is used to initiate the polypeptide chain. fMet is carried to the ribosome by its specialized tRNA, called fMet-tRNAeMetf and is placed in the P site right away

Three-dimensional visualization of the fMet-tRNAMetf attached to the ribosome has allowed us to locate the P-site position of tRNA with an accuracy to within 1 angstrom, in spite of the fact that the reconstruction is limited in resolution to 15 angstroms. The high accuracy is a consequence of the fact that the grooves of the tRNA helices are visible when the display threshold is raised sufficiently, leaving little freedom in the fitting of the atomic structure. As we had anticipated from the shape of the intersubunit space, the tRNA fits into this space precisely in lock-and-key fashion.

We see that the anticodon of the fMet-tRNAMetf meets the mRNA codon in the cleft of the small subunit, right next to the point where the channel penetrates the subunit neck. The tip of the short arm of the elbow-shaped tRNA molecule, where the amino acid attaches, is thought to contact the ribosome at a site called the peptidyltransferase center. This is the place on the large subunit where the peptide bond is formed between the incoming amino acid and the polypeptide chain. It is remarkable to see from our data that the short arm of the tRNA in the P site points toward the mouth of the tunnel in the large subunit-another indication that the tunnel is indeed the conduit through which the nascent polypeptide exits the ribosome.

In the next elongation step, the second amino acid is added to the chain. This amino acid is carried to the ribosome by the appropriate tRNA molecule, and its anticodon must match the codon of the mRNA template. To prevent the wrong kind of tRNA from entering the A site, its short arm and the attached amino acid are covered by a protein called Elongation factor-Tu (EF-Tu), which leaves only the anticodon on the long arm free to bind. Removing EF-Tu and inserting the tRNA into the A site requires some energy, and that is provided by the socalled energy molecule GTP, which is bound to EF-Tu. We have to imagine, now, that the complex of tRNA, EF-Tu and GTP is in some way “springloaded,” ready to release tRNA only if the match between codon and anticodon is perfect. Only in that case can the incoming tRNA move into the A site. Cryo-EM of ribosomes to which both A- and P-site tRNAs or the ternary complex is bound, done by our group and the group of Marin van Heel, now at Imperial College, has firmly established the position of A-site tRNA and the peculiar octopus-like hold that the ternary complex has on the two subunits (Figure lid).

Once the new tRNA is properly bound in the A site, the amino acid it carries is in the correct position to form a peptide bond with the peptide chain (or in the case of the second amino acid, with the fMet amino acid) attached to tRNA sitting in the ribosome’s P site. The transfer of the bond probably results in an intermediate state where the A-site tRNA-bound amino acid is already attached to the polypeptide chain while the P-site tRNA has lost its amino acid. With one tRNA in the P site and another in the A site, the ribosome is now in what is called the pretranslocational state.

In the third step, called translocation, the whole assembly (that is, the tRNAs and attached mRNA) moves by a distance corresponding to one codon along the ribosome, freeing up the A site for the next tRNA. How this movement is accomplished remains a mystery.

What is known is that translocation requires the help of another protein, called Elongation factor G (EF-G). This is a large protein whose structure was recently solved by x-ray crystallography To everyone’s surprise, its shape was found to closely mimic that of the ternary complex that is formed by tRNA, EF-Tu and GTP. One portion of the EF-G molecule is carefully “crafted” to exhibit the same shape and charge distribution as the exposed anticodon arm of the tRNA molecule. Apparently, EF-G is designed to bind to the ribosome at the same sites and in the same way as the ternary complex, lacking only the ability to bind specifically to any one codon. As in the case of the ternary complex, the binding of EF-G is accompanied by the expenditure of energy in the form of GTP that is bound to it.

We recently imaged the ribosome with EF-G bound to it and found it in a position similar to that of the ternary complex. In addition, our visualization contains an important clue about translocation. It appears that EF-G changes its shape upon binding to the ribosome, in such a way that the part of the molecule that mimics the anticodon arm of tRNA actually takes the place of the A-site tRNA.

Translocation also moves the P-site tRNA, now devoid of its peptide, one place over, into the exit site (E site). This movement is perhaps part of a domino effect brought about when the A-site tRNA moves into the P site, since the A- and P-site tRNAs are within touching range of each other. Alternatively, the removal of tRNA from the E site might be the first step in a cascade of reactions in the course of which the P-site tRNA moves into the E site, and the A-site tRNA moves into the P site. Data from one set of images suggest that the E site lies in the vicinity of the L1 protein of the large subunit, at the far end of the intersubunit space. It is intriguing to speculate that L1 binds to the “empty,” amino-acid-depleted tRNA molecule to facilitate its orderly removal from the ribosome. This may also control the pace of protein synthesis, since it has been found that, just as in an overfilled bus, the incoming tRNA molecule cannot enter until the outgoing tRNA makes its exit.

Other experiments done under different conditions suggest that there might actually be two E sites that become occupied successively: One is in the immediate vicinity of the P-site tRNA (which is still presumably in contact with the mRNA via codon-anticodon binding) and represents an intermediate exit point. The second E site is the one already mentioned, close to the L1 protein, which could be the point of final exit.

Models of Translocation

As I noted previously, it is still unknown how the tRNAs are moved from one to another site or how they move together with the mRNA along the ribosome by the exact distance of a codon at each step. Prompted by their experimental observations, Harry Noller’s group at the University of California at Santa Cruz proposed some time ago that the two ends of tRNA advance alternately through the ribosome. For example, as the tRNA moves from the A site to the P site, it passes through a hybrid state, wherein its long arm is attached to the small subunit at the A site while its short arm is attached at the P site of the large subunit. According to this model, in the course of the complete cycle, the tRNA “walks through” a few of these hybrid states, A/P and P/E, before it finally exits.

Knud Nierhaus’s group at the Max Planck Institute of Molecular Genetics in Berlin developed an alternative model, which they call the alpha/epsilon model. Citing evidence that the contacts between the tRNA and the ribosome remain virtually unchanged through translocation, they propose the existence of a movable domain in the ribosome on which the tRNAs “ride” in a fixed arrangement from one position to another as the ribosome moves. Indeed, at the resolution we have obtained, we see that the tRNAs maintain the same relative arrangement in the pre- as in the post-translational state, just as would be predicted by the Nierhaus model.

Both models imply highly novel interactions between RNA molecules. RNA molecules usually interact in what is called a Watson-Crick pairing, where the nucleotide bases of one RNA molecule bind to the nucleotide bases of another. But the Noller and Nierhaus models of translocation imply that tRNA binds to some crevices formed by the RNA molecules that make up the ribosome, in a manner more analogous to the way proteins bind their substrates than to anything observed before between RNA molecules. Furthermore, both models assume considerable conformational flexibility of RNA regions involved in those contacts. This is also highly unusual and describes behavior more typical of proteins than of RNA molecules. If nothing else, study of the ribosome is indeed revealing previously unimagined new abilities of RNA molecules.

What Next?

Our primary goal is to continue to improve the resolution of our techniques. Here the stability and coherence of the imaging instrument, as well as the conformational purity of the ribosomes in the sample, are determining factors. With some luck, the 10-angstrom hurdle might be reached within a year. A resolution of 9 angstroms has just been achieved for the large subunit of a saltloving bacterium, Haloarcula marismortui, using a combination of low-resolution cryo-EM images generated in my lab and x-ray crystallography performed at Yale University in the laboratories of Tom Steitz and Peter Moore, from data collected at Brookhaven National Laboratories. At that resolution, it becomes possible to fit proteins or RNA segments whose atomic structures are known from x-ray studies into the overall map and obtain their positions with great accuracy.

Ultimately, we would like to study the time course of translation events by freezing the ribosome at various stages in the elongation cycle as it moves from one codon to the next. Time-resolved images can be obtained by starting the reaction on the microscope grid at room temperature and then waiting anywhere from 5 to 200 milliseconds before plunging the grid into the coolant. A reconstruction could then be obtained, say, every 5 milliseconds, showing the progression of tRNA through the ribosome. Once the technical difficulties are solved, we might be able to watch protein synthesis as a movie in three dimensions, slowed down about 1,000 times. Among other issues, such studies will resolve questions about the existence of hybrid states and movable domains.


The author would like to thank Rajendra Agrawal for helpful comments and Amy Heagle for assistance with the design of illustrations featuring experimental maps. Some of the ligand binding experiments and interpretations were done in collaboration with the group of Knud Nierhaus, Max Planck Institute for Molecular Genetics in Berlin. Funding for this work was provided, in part, by the National Institutes of Health and the National Science Foundation.