Structural Insights into Gene Expression and Protein Synthesis

The following sections are included:

• Preface
• Contents

I was born in Milwaukee, Wisconsin, in 1940, and my family lived in an apartment above a paint store in the downtown area until 1949. Although my father had obtained a law degree from Marquette University in Milwaukee, he became the administrator in-charge of personnel at the Milwaukee County Hospital. My mother grew up on a farm in the Waukesha county outside of Milwaukee and graduated from Carol College, a small college in Waukesha. My mother devoted her time to the domestic chores required for raising a family, which eventually grew to five children—me and my two younger brothers and two younger sisters. My father’s parents lived about 20 blocks away and my mother’s parents and her brother’s family lived on the family farm in the Waukesha county…

This is a reminiscence of my interactions with Tom during our 50 years in the University of Cambridge and Yale University from 1967 until 2017. Tom was a colleague, friend and supporter throughout those years. It is a pleasure to recall some of the high points of our adventures together.

I met Tom in the fall of 1985 shortly after I began a temporary, part-time position at Yale University working for Professor Peter Lengyel in the Department of Molecular Biophysics and Biochemistry. One day Peter said, “Follow me—I want you to meet someone.” So I raced after him down the back stairwell in the Kline Biology Tower to the second floor, where he introduced me to Tom and then immediately returned to his office on the fifth floor. Tom’s office was small and modest and the walls were decorated with large and beautiful photos of mountain scenes. Still not knowing the purpose of this introduction, I asked him questions about the photos and began to learn of his love of the outdoors and of hiking and skiing. I learned that we shared a Midwestern heritage and style of upbringing, and a love of gardening. Our meeting was brief and very pleasant, but it left me puzzled. A day or so later, its purpose was revealed to me. Peter and Tom had devised a plan to create a new permanent, full-time administrative assistant position for me that would involve working for the both of them. By the start of the new year, in 1986, this position became a reality; I worked in the Steitz laboratory in the mornings and in the Lengyel laboratory in the afternoons. It was a pleasure getting to know Peter and Tom and the members of their research groups. Soon after, the Howard Hughes Medical Institute (HHMI) offered Tom a position as Investigator in their new research initiative at Yale. His acceptance of this new position and the additional funding it provided created a vast array of new opportunities. For example, it was now possible for him to hire a full-time assistant. He offered me the position. Since my background was in the arts, I expressed my concern that my lack of scientific knowledge might be a detriment to my work performance. Tom offered his reassurance and encouraged me to accept the position, which I did, in March 1987. Later, when I learned of the creativity involved in scientific research and the artistry of the structures that emerged from data collection, the position was clearly a perfect match. Although it was sad to say good-bye to Peter Lengyel and his laboratory, I am forever grateful to him for having introduced me to Tom…

Tom Steitz was invited to join the Molecular Biophysics and Biochemistry (MB&B) department at Yale University in 1970. Tom showed in his post-doctoral studies with David Blow at the MRC laboratory in Cambridge, England, that the enzyme hexokinase could be crystallized. Hexokinase is regulated and is the first enzyme in glycolysis. Tom’s new crystals diffracted X-rays well and proved to be suitable for structure determination. The promise, as Tom explained to me, was to learn the structural elements responsible for substrate-induced conformational changes in catalysis and control of phosphate transfer from ATP to glucose. Tom was eager to solve the structure, while I was daunted by Tom’s project. The structure determination would be challenging due to the 100,000 Dalton molecular weight of the hexokinase dimer. Furthermore the amino acid sequence for the yeast protein was not known…

The A isozyme of yeast hexokinase (ATP:D-hexose 6-phosphotransferase, EC 2.7.1.1) crystallized as a complex with glucose has a conformation that is dramatically different from the conformation of the B isozyme crystallized in the absence of glucose. Comparison of the high-resolution structures shows that one lobe of the molecule is rotated by 12° relative to the other lobe, resulting in movements of as much as 8 Å in the polypeptide backbone and closing the cleft between the lobes into which glucose is bound. The conformational change is produced by the binding of glucose (R. C. McDonald, T. A. Steitz, and D. M. Engelman, unpublished data) and is essential for catalysis [Anderson, C. M., Stenkamp, R. E., McDonald, R.C. & Steitz, T. A. (1978) J. Mol. Biol. 123, 207–219] and thus provides an example of induced fit. The surface area of the hexokinase A·glucose complex exposed to solvent is smaller than that of native hexokinase B. By using the change in exposed surface area to estimate the hydrophobic contribution to the free energy changes upon glucose binding, we find that the hydrophobic effect alone favors the active conformation of hexokinase in the presence and absence of sugar. The observed stability of the inactive conformation of the enzyme in the absence of substrates may result from a deficiency of complementary interactions within the cavity that forms when the two lobes close together.

Space-filling models of yeast hexokinase, adenylate kinase, and phosphoglycerate kinase drawn by computer clearly portray the bilobal character of these phosphoryl transfer enzymes, and the deep cleft which is formed between the lobes. A dramatic conformational change occurs in hexokinase as glucose binds to the bottom of the cleft, which causes the two lobes of hexokinase to come together. A substrate-induced closing of the active site cleft is postulated to occur in other kinases as well. This change may provide a mechanism by which some of these enzymes reduce their inherent adenosine triphosphatase activity and could be a general requirement of the kinase reaction.

Using small-angle X-ray scattering from solutions of yeast hexokinase, we have measured the radii of gyration of the monomeric B isozyme and its complexes with sugar substrates. We find that the radius of gyration decreases by 0.95 ± 0.24 Å upon binding glucose and 1.25 ± 0.28 Å upon binding glucose 6-phosphate. This observed reduction in radius of gyration in the presence of glucose is the same as that calculated from the coordinates of the high-resolution crystal structures of native hexokinase B and a glucose complex with hexokinase A. Thus, these measurements suggest that the dramatic closing of the slit between the two lobes of hexokinase observed in the crystal structures (Bennett, W. S., & Steitz, T. A. (1978) Proc. Natl. Acad. Sci. U.S.A. 75, 4848–4852) occurs in solution when either glucose or glucose 6-phosphate is bound.

The structure of a binary hexokinase: A glucose complex (HKA · G) has been determined independently at 4·5 Å resolution by the multiple isomorphous replacement method, and the co-ordinates have been refined at 3·5 Å resolution. Initially at 6 Å resolution, an electron density map based on a single isomorphous derivative was used to orient the native hexokinase B structure (Steitz et al., 1976) in the HKA · G unit cell and thus to obtain molecular replacement phases for the HKA · G complex based on the structure of the B isozyme. Both the single isomorphous replacement map and an F₀ − F_c difference electron density map calculated with molecular replacement phases at low resolution suggested a large conformational difference between the complex and the native B isozyme. An attempt to extend the phasing of the HKA · G diffraction amplitudes from 6 Å to 3·5 Å resolution by molecular replacement failed at this stage, presumably because of the considerable conformational difference between these two structures.

An improved isomorphous replacement map at 4·5 Å resolution confirmed that the polypeptide backbone of hexokinase A in the HKA · G complex is folded into two distinct lobes; within each lobe the folding is very similar to that of native hexokinase B, but the relative orientation of the two lobes is quite different in the complex and in the native enzyme. To characterize this conformational difference accurately and to extend phasing of the HKA · G data to higher resolution, an atomic model of the hexokinase B monomer was adapted to accommodate the difference in the relative orientation of the two lobes. The resulting HKA · G model was then refined by difference Fourier techniques to a final R factor of 0·26 at 3·5 Å resolution.

Detailed comparison of the refined crystal structures of the hexokinase A:glucose complex (HKA · G) and native hexokinase B shows that, in addition to the 12° rotation of one lobe of the enzyme relative to the other as described previously (Bennett & Steitz, 1978) there are small systematic differences in the conformation of the polypeptide backbones of the two structures adjacent to the glucose binding site and crystal packing contacts. In the HKA · G complex, the cleft between the two lobes of the hexokinase molecule is narrowed, substantially reducing the accessibility of the active site to solvent. The HKA · G structure suggests specific contacts with a bound glucose molecule that cannot form in the more open native structure. The closed conformation of the HKA · G complex can be formed by either subunit in the heterologous dimer configuration of hexokinase B (Anderson et al. 1974); new or different interactions between subunits, or with ligands bound to the intersubunit ATP site, may be made when the upper subunit of the dimer is in the closed conformation and may contribute to the cooperative interactions observed in the crystalline dimer and in solution.

The following sections are included:

• Background
• “The Helical Hairpin Hypothesis” (Engelman and Steitz, 1981) (note again Tom’s love of wordplay—his title)
• “The GES Scale” (Engelman et al., 1986) (Tom strikes again!—even though the authors were Engelman, Goldman and Steitz, the pun was irresistible)
• References

We propose that the initial event in the secretion of proteins across membranes and their insertion into membranes is the spontaneous penetration of the hydrophobic portion of the bilayer by a helical hairpin. Energetic considerations of polypeptide structures in a nonpolar, lipid environment compared with an aqueous environment suggest that only α and 3₁₀ helices will be observed in the hydrophobic interior of membranes. Insertion of a polypeptide is accomplished by a hairpin structure composed of two helices, which will partition into membranes if the free energy arising from burying hydrophobic helical surfaces exceeds the free energy “cost” of burying potentially charged and hydrogen-bonding groups. We suggest, for example, that the hydrophobic leader peptide found in secreted proteins and in many membrane proteins forms one of these helices and is oriented in the membrane with its N terminus inside. In secreted proteins, the leader functions by pulling polar portions of a protein into the membrane as the second helix of the hairpin. The occurrence of all categories of membrane proteins can be rationalized by the hydrophobic or hydrophilic character of the two helices of the inserted hairpin and, for some integral membrane proteins, by events in which a single terminal helix is inserted. We propose that, because of the distribution of polar and nonpolar sequences in the polypeptide sequence, secretion and the insertion of membrane proteins are spontaneous processes that do not require the participation of additional specific membrane receptors or transport proteins.

The following sections are included:

• PERSPECTIVES AND OVERVIEW
• INTRODUCTION
• HELICES IN LIPID BILAYER ENVIRONMENTS
  • Theoretical Considerations
  • Experimentally Observed Membrane Protein Secondary Structures
• DETERMINATION OF HELIX LOCATIONS FROM SEQUENCE DATA
  • Appropriate Scales of Hydrophobicity for Bilayers and Protein Interiors
  • A Polarity Scale for Identifying Transmembrane Helices
  • Comparison with Other Polarity Scales
  • Overview of Polarity Scales
• USE OF SCALES TO IDENTIFY TRANSMEMBRANE HELICES
  • Energetic Consequences of Helical Conformation
  • Entropy of Immobilization
  • Choice of Window Length and Scanning Procedures
  • Comparison of Scales
  • Significance of Peaks in the Sequence Analysis
• TESTS OF PREDICTIONS USING KNOWN STRUCTURES
• CONCLUSIONS

My time as a postdoc in Tom’s laboratory from 1978 to 1984 marked an important transition in his research and in my life. Tom’s research was moving from studies of hexokinase to focus on DNA-binding proteins. I had joined the Steitz laboratory immediately after completing my doctoral dissertation with Louise Johnson in the University of Oxford. While at Yale University, I met and married my husband, Robert (Rob) Harrison, and Tom attended our wedding. Rob was a PhD student in Tom’s group and one of the last to work on hexokinase. He is now a professor in the Computer Science Department at Georgia State University, and we collaborate on several projects…

The amino acid sequence of the Escherichia coli catabolite gene activator protein has been fit into a 2.9-Å resolution electron density map. Each subunit of the dimer consists of two structurally distinct domains. The larger NH₂-terminal domain is seen to bind cyclic AMP and forms all of the contacts between the subunits. The cyclic AMP is completely buried between the interior of the “β roll” structure of the large domain and a long α helix; it makes important hydrogen-bonding interactions with residues from both subunits. The guanidinium group of a buried Arg makes an internal salt link with the phosphate of cyclic AMP. The 6-amino group of adenine interacts simultaneously with both subunits. This interaction with both subunits and the fact that cyclic GMP and cyclic IMP do not activate catabolite gene activator protein suggest that the binding of cyclic AMP may alter the relative orientation of the two subunits, which in turn would change the structure of a DNA binding site that is presumed to span the two smaller domains. The distribution and nature of side chains in the small domain do not rule out the possibility that catabolite gene activator protein binds to left-handed B-DNA.

The structure of a dimer of the Escherichia coli catabolite gene activator protein has been refined at 2·5 Å resolution to a crystallographic R-factor of 20·7 % starting with coordinates fitted to the map at 2·9 Å resolution. The two subunits are in different conformations and each contains one bound molecule of the allosteric activator, cyclic AMP. The amino-terminal domain is linked to the smaller carboxy-terminal domain by a nine-residue hinge region that exists in different conformations in the two subunits, giving rise to approximately a 30° rotation between the positions of the small domains relative to the larger domains. The amino-terminal domain contains an antiparallel β-roll structure in which the interstrand hydrogen bonding is well-determined. The β-roll can be described as a long antiparallel β-ribbon that folds into a right-handed supercoil and forms part of the cyclic AMP binding site. Each cyclic AMP molecule is in an anti conformation and has ionic and hydrogen bond interactions with both subunits.

The 3 angstrom resolution crystal structure of the Escherichia coli catabolite gene activator protein (CAP) complexed with a 30–base pair DNA sequence shows that the DNA is bent by 90°. This bend results almost entirely from two 40° kinks that occur between TG/CA base pairs at positions 5 and 6 on each side of the dyad axis of the complex. DNA sequence discrimination by CAP derives both from sequence-dependent distortion of the DNA helix and from direct hydrogen-bonding interactions between three protein side chains and the exposed edges of three base pairs in the major groove of the DNA. The structure of this transcription factor–DNA complex provides insights into possible mechanisms of transcription activation.

Tom’s untimely passing is a huge loss to all of us whose lives and work were touched by his brilliance and humor. Tom spent his career successfully deciphering the inner workings of the central dogma, and thus set the highest standard for all structural biologists. When I joined his laboratory as a postdoctoral fellow in 1992, Tom and his team had already determined crystal structures of DNA polymerase, reverse transcriptase, tRNA synthetase and several transcription regulators. I was given the project of bacterial γδ resolvase—despite my eagerness to work on the mammalian recombinase RAG1/2, and the fact that the first crystal structure of γδ resolvase had already been determined. With this assignment Tom taught me the essential lesson of choosing projects that have interesting biological questions and yet are approachable with currently available tools. He also taught me the lesson of persistence. Even though the first structure of an important biological complex attracts the lion’s share of attention, it often requires several more steps before the mechanism is truly understood. After I had determined the structure of γδ resolvase with DNA bound, which revealed several surprises, Tom walked me through on how I should give a scientific talk, for example, making one point with each slide—and to use a font large enough for the people at the back of the room to see! His lessons on choosing research projects and preparing us for talks guide me to this day…

Tom always set the highest expectations for all the projects in our group. After the crowning successes of the structures of DNA polymerases and DNA polymerase-DNA complexes, Tom turned his attention to the replisome. After solving the structures of so many individual pieces of the puzzle, it was time to put them all together. We started with the notion that the phage T4 replication system would recapitulate all the holistic features of the replisome not just from bacteria, but also of archaea and eukaryotes. The underlying foundation of the in vitro biochemistry was very strong, and the proteins themselves were typically well behaved. I had a great deal of experience in the T4 system from my work as a graduate student for Bill Konigsberg. Now the challenge was clear: put it all together as a complex for high-resolution crystallography…

At the last year of my graduate study, I decided to postdoc either with Tom Steitz to deepen my crystallographic skills or with Richard Henderson to learn electron cryomicroscopy (cryoEM). Before I formally applied, I was offered a postdoc position in a molecular biology laboratory at Yale University. I told the principal investigator of my intention to work in the Steitz laboratory. He therefore arranged a meeting with Tom. Since it was not a formal interview, I was not sure what to expect. Tom, dressed in his iconic tweed jacket, greeted me warmly. However, the meeting did not go well, I thought. When Tom asked about my thesis project, I told him about my structural and functional studies of tryptophanyl tRNA synthetase and my experience in using the maximum entropy method for phase optimization in Charlie Carter’s laboratory at the University of North Carolina at Chapel Hill. Tom listened carefully. He then asked if I had solved a structure using experimental phases. I sheepishly said, “No.” On my flight back, I was disappointed with myself. However, to my surprise, an email from Tom came next day offering me a position in his laboratory. I moved to New Haven soon after…

During my tenure in his laboratory, Tom was well known for his belief that there were no impossible projects in structural biology. He always emphasized that seemingly difficult challenges could be surmounted by really understanding the crystallographic basis behind them. A prime example was fine-tuning how phase angles could be calculated as opposed to relying on the turn-key methods supplied by the available software packages of the time. More specifically, we would sit down with Jimin Wang in the core computational center for a few days and then feel ashamed at how badly we really understood crystallography. However, we were happy that we were changed for the better by the experience. Many times, this also translated into refining our crystals in terms of size and cryoprotective procedures and going on many trips to the synchrotrons in Long Island (NSLS1), Chicago (APS) and Berkeley (ALS) before it became common to send crystals via post or collect data remotely via the internet…

As one of the graduate students in Tom’s laboratory, I sometimes debated whether I should work on an important but difficult project or something easier but less exciting. Tom always said, “You need to know how to solve a real scientific problem.” His advice was always to work on the more important project, and this is why I began to study the structural mechanism of DNA recombination by serine recombinases…

The 3.3-Å resolution crystal structure of the large proteolytic fragment of Escherichia coli DNA polymerase I complexed with deoxythymidine monophosphate consists of two domains, the smaller of which binds zinc-deoxythymidine monophosphate. The most striking feature of the larger domain is a deep crevice of the appropriate size and shape for binding double-stranded B-DNA. A flexible subdomain may allow the enzyme to surround completely the DNA substrate, thereby allowing processive nucleotide polymerization without enzyme dissociation.

High-resolution crystal structures of editing complexes of both duplex and single-stranded DNA bound to Escherichia coli DNA polymerase I large fragment (Klenow fragment) show four nucleotides of single-stranded DNA bound to the 3′ – 5′ exonuclease active site and extending toward the polymerase active site. Melting of the duplex DNA by the protein is stabilized by hydophobic interactions between Phe-473, Leu-361, and His-666 and the last three bases at the 3′ terminus. Two divalent metal ions interacting with the phosphodiester to be hydrolyzed are proposed to catalyze the exonuclease reaction by a mechanism that may be related to mechanisms of other enzymes that catalyze phospho-group transfer including RNA enzymes. We suggest that the editing active site competes with the polymerase active site some 30 Å away for the newly formed 3′ terminus. Since a 3′ terminal mismatched base pair favors the melting of duplex DNA, its binding and excision at the editing exonuclease site that binds single-stranded DNA is enhanced.

The refined crystal structures of the large proteolytic fragment (Klenow fragment) of Escherichia coli DNA polymerase I and its complexes with a deoxynucleoside monophosphate product and a single-stranded DNA substrate offer a detailed picture of an editing 3′ – 5′ exonuclease active site. The structures of these complexes have been refined to R-factors of 0.18 and 0.19 at 2.6 and 3.1 Å resolution respectively. The complex with a thymidine tetranucleotide complex shows numerous hydrophobic and hydrogen-bonding interactions between the protein and an extended tetranucleotide that account for the ability of this enzyme to denature four nucleotides at the 3′ end of duplex DNA. The structures of these complexes provide details that support and extend a proposed two metal ion mechanism for the 3′ – 5′ editing exonuclease reaction that may be general for a large family of phosphoryltransfer enzymes. A nucleophilic attack on the phosphorous atom of the terminal nucleotide is postulated to be carried out by a hydroxide ion that is activated by one divalent metal, while the expected pentacoordinate transition state and the leaving oxyanion are stabilized by a second divalent metal ion that is 3.9 Å from the first. Virtually all aspects of the pretransition state substrate complex are directly seen in the structures, and only very small changes in the positions of phosphate atoms are required to form the transition state.

A 3.5 angstrom resolution electron density map of the HIV-1 reverse transcriptase heterodimer complexed with nevirapine, a drug with potential for treatment of AIDS, reveals an asymmetric dimer. The polymerase (pol) domain of the 66-kilodalton subunit has a large cleft analogous to that of the Klenow fragment of Escherichia coli DNA polymerase I. However, the 51-kilodalton subunit of identical sequence has no such cleft because the four subdomains of the pol domain occupy completely different relative positions. Two of the four pol subdomains appear to be structurally related to subdomains of the Klenow fragment, including one containing the catalytic site. The subdomain that appears likely to bind the template strand at the pol active site has a different structure in the two polymerases. Duplex A-form RNA-DNA hybrid can be model-built into the cleft that runs between the ribonuclease H and pol active sites. Nevirapine is almost completely buried in a pocket near but not overlapping with the pol active site. Residues whose mutation results in drug resistance have been approximately located.

A catalytic subdomain of HIV reverse transcriptase has the same structure as that of E. coli Klenow fragment, including three conserved carboxylates whose mutation destroys activity. The structures of Klenow fragment complexed with DNA and with divalent metal ions, taken together with mutagenic studies, are consistent with a two-metal-ion mechanism of phosphoryl transfer at both active sites. The source of processivity in DNA polymerase III is easily understood in terms of the donut shape of its β-subunit.

Klenow fragment of Escherichia coli DNA polymerase I, which was cocrystallized with duplex DNA, positioned 11 base pairs of DNA in a groove that lies at right angles to the cleft that contains the polymerase active site and is adjacent to the 3′ to 5′ exonuclease domain. When the fragment bound DNA, a region previously referred to as the “disordered domain” became more ordered and moved along with two helices toward the 3′ to 5′ exonuclease domain to form the binding groove. A single-stranded, 3′ extension of three nucleotides bound to the 3′ to 5′ exonuclease active site. Although this cocrystal structure appears to be an editing complex, it suggests that the primer strand approaches the catalytic site of the polymerase from the direction of the 3′ to 5′ exonuclease domain and that the duplex DNA product may bend to enter the cleft that contains the polymerase catalytic site.

Extrapolating from the co-crystal structure of rat DNA polymerase β (pol β) complexed with primer-template, dideoxycytidine triphosphate (ddCTP), and two metal ions, H. Pelletier et al. recently concluded that the orientation of the DNA primer-template in Escherichia coli DNA polymerase I Klenow fragment (KF) and the reverse transcriptase (RT) of human immunodeficiency virus-type 1 is opposite to that derived from published co-crystal structures. We disagree with this conclusion and suggest an alternative interpretation of the structural data, namely, that there is no contradiction between the orientations of the DNA inferred from these structures; rather, the apparent inconsistency is the result of an inappropriate alignment of the pol β structure with the other polymerase structures. While the crystal structures of KF, RT, and T7 RNA polymerase (RNAP) can be aligned by superposition of a homologous “palm” subdomain, pol β is not homologous to these other polymerases, and therefore should not be aligned with them by superimposing protein structures. Instead, we suggest that pol β can be oriented relative to this family only by superposition of the functionally important entities in the polymerase reaction, namely, the two catalytic divalent metal ions and the 3′ terminus of DNA primer strand. This alignment is achieved by rotating the entire pol β complex by about 180° (Figs. 1 and 2) from the structural alignment proposed by Pelletier et al. The alignment we suggest allows all four polymerases to use the identical polymerase mechanism on similarly oriented primer-template molecules without the need to re-orient the primer-templates from their previously determined positions and is therefore consistent with structural, biochemical, and molecular genetic studies of polymerase-substrate complexes. [By contrast, the proposal of Pelletier et al. that the direction of primer-template binding to KF, RT, and RNAP should be reversed contradicts the conclusions drawn from a substantial body of existing data.] A further advantage of our proposed alignment is that it reveals additional analogies between pol β and the other three polymerases in the overall structure of the polymerase domain…

The dipyridodiazepinone Nevirapine is a potent and highly specific inhibitor of the reverse transcriptase (RT) from human immunodeficiency virus type 1 (HIV-1). It is a member of an important class of nonnucleoside drugs that appear to share part or all of the same binding site on the enzyme but are susceptible to a variety of spontaneous drug-resistance mutations. The co-crystal-structure of HIV-1 RT and Nevirapine has been solved previously at 3.5-Å resolution and now is partially refined against data extending to 2.9-Å spacing. The drug is bound in a hydrophobic pocket and in contact with some 38 protein atoms from the p66 palm and thumb subdomains. Most, but not all, nonnucleoside drug-resistance mutations map to residues in close contact with Nevirapine. The major effects of these mutations are to introduce steric clashes with the drug molecule or to remove favorable protein-drug contacts. Additionally, four residues (Phe-227, Trp-229, Leu-234, and Tyr-319) in contact with Nevirapine have not been selected as sites of drug-resistance mutations, implying that there may be limitations on the number and types of resistance mutations that yield viable virus. Strategies of inhibitor design that target interactions with these conserved residues may yield drugs that are less vulnerable to escape mutations.

The reverse transcriptase from human immunodeficiency virus type 1 is a heterodimer consisting of one 66-kDa and one 51-kDa subunit. The p66 subunit contains both a polymerase and an RNase H domain; proteolytic cleavage of p66 removes the RNase H domain to yield the p51 subunit. Although the polymerase domain of p66 folds into an open, extended structure containing a large active-site cleft, that of p51 is closed and compact. The connection subdomain, which lies between the polymerase and RNase H active sites in p66, plays a central role in the formation of the reverse transcriptase heterodimer. Extensive and very different intra- and intersubunit contacts are made by the connection subdomains of each of the subunits. Together, contacts between the two connection domains constitute approximately one-third of the total contacts between subunits of the heterodimer. Conversion of an open p66 polymerase domain structure to a closed p51-like structure results in a reduction in solvent-accessible surface area by 1600 Ų and the burying of an extensive hydrophobic surface. Thus, the monomeric forms of both p66 and p51 are proposed to have the same closed structure as seen in the p51 subunit of the heterodimer. The free energy required to convert p66 from a closed p51-like structure to the observed open p66 polymerase domain structure is generated by the burying of a large, predominantly hydrophobic surface area upon formation of the heterodimer. It is likely that the only kind of dimer that can form is an asymmetric one like that seen in the heterodimer structure, since one dimer interaction surface exists only in p51 and the other only in p66. We suggest that both p51 and p66 form asymmetric homodimers that are assembled from one subunit that has assumed the open conformation and one that has the closed structure.

THE DNA polymerase from Thermus aquaticus (Taq polymerase), famous for its use in the polymerase chain reaction, is homologous to Escherichia coli DNA polymerase I (pol I) (ref. 1). Like pol I, Taq polymerase has a domain at its amino terminus (residues 1–290) that has 5′ nuclease activity and a domain at its carboxy terminus that catalyses the polymerase reaction. Unlike pol I, the intervening domain in Taq polymerase has lost the editing 3′–5′ exonuclease activity. Although the structure of the Klenow fragment of pol I has been known for ten years, that of the intact pol I has proved more elusive. The structure of Taq polymerase determined here at 2.4 Å resolution shows that the structures of the polymerase domains of the thermostable enzyme and of the Klenow fragment are nearly identical, whereas the catalytically critical carboxylate residues that bind two metal ions are missing from the remnants of the 3′–5′-exonuclease active site of Taq polymerase. The first view of the 5′ nuclease domain, responsible for excising the Okazaki RNA in lagging-strand DNA replication, shows a cluster of conserved divalent metal-ion-binding carboxylates at the bottom of a cleft. The location of this 5′-nuclease active site some 70 Å from the polymerase active site in this crystal form highlights the unanswered question of how this domain works in concert with the polymerase domain to produce a duplex DNA product that contains only a nick…

THE DNA polymerase from Thermus aquaticus (Taq polymerase) is homologous to Escherichia coli DNA polymerase I (Pol I) and likewise has domains responsible for DNA polymerase and 5′ nuclease activities. The structures of the polymerase domains of Taq polymerase and of the Klenow fragment (KF) of Pol I are almost identical, whereas the structure of a vestigial editing 3′ – 5′ exonuclease domain of Taq polymerase that lies between the other two domains is dramatically altered, resulting in the absence of this activity in the thermostable enzyme. The structures have been solved for editing complexes between KF and single-stranded DNA and for duplex DNA with a 3′ overhanging single strand, but not for a complex containing duplex DNA at the polymerase active-site. Here we present the co-crystal structure of Taq polymerase with a blunt-ended duplex DNA bound to the polymerase active-site cleft; the DNA neither bends nor goes through the large polymerase cleft, and the structural form of the bound DNA is between the B and A forms. A wide minor groove allows access to protein side chains that hydrogen-bond to the N3 of purines and the O2 of pyrimidines at the blunt-end terminus. Part of the DNA bound to the polymerase site shares a common binding site with DNA bound to the exonuclease site, but they are translated relative to each other by several ångströms along their helix axes…

Although the single-polypeptide-chain RNA polymerase from bacteriophage T7 (T7RNAP), like other RNA polymerases, uses the same mechanism of polymerization as the DNA polymerases, it can also recognize a specific promoter sequence, initiate new RNA chains from a single nucleotide, abortively cycle the synthesis of short transcripts, be regulated by a transcription inhibitor, and terminate transcription. As T7RNAP is homologous to the Pol I family of DNA polymerases, the differences between the structure of T7RNAP complexed to substrates and that of the corresponding DNA polymerase complex provides a structural basis for understanding many of these functional differences. T7RNAP initiates RNA synthesis at promoter sequences that are conserved from positions −17 to +6 relative to the start site of transcription. The crystal structure at 2.4 Å resolution of T7RNAP complexed with a 17-base-pair promoter shows that the four base pairs closest to the catalytic active site have melted to form a transcription bubble. The T7 promoter sequence is recognized by interactions in the major groove between an antiparallel β-loop and bases. The amino-terminal domain is involved in promoter recognition and DNA melting. We have also used homology modelling of the priming and incoming nucleoside triphosphates from the T7 DNA-polymerase ternary complex structure to explain the specificity of T7RNAP for ribonucleotides, its ability to initiate from a single nucleotide, and the abortive cycling at the initiation of transcription…

The structure of a T7 RNA polymerase (T7 RNAP) initiation complex captured transcribing a trinucleotide of RNA from a 17–base pair promoter DNA containing a 5-nucleotide single-strand template extension was determined at a resolution of 2.4 angstroms. Binding of the upstream duplex portion of the promoter occurs in the same manner as that in the open promoter complex, but the single-stranded template is repositioned to place the +4 base at the catalytic active site. Thus, synthesis of RNA in the initiation phase leads to accumulation or “scrunching” of the template in the enclosed active site pocket of T7 RNAP. Only three base pairs of heteroduplex are formed before the RNA peels off the template.

We have solved the crystal structures of the bacteriophage RB69 sliding clamp, its complex with a peptide essential for DNA polymerase interactions, and the DNA polymerase complexed with primer-template DNA. The editing complex structure shows a partially melted duplex DNA exiting from the exonuclease domain at an unexpected angle and significant changes in the protein structure. The clamp complex shows the C-terminal 11 residues of polymerase bound in a hydrophobic pocket, and it allows docking of the editing and clamp structures together. The peptide binds to the sliding clamp at a position identical to that of a replication inhibitor peptide bound to PCNA, suggesting that the replication inhibitor protein p21^CIP1 functions by competing with eukaryotic polymerases for the same binding pocket on the clamp.

Possibly the earliest enzymatic activity to appear in evolution was that of the polynucleotide polymerases, the ability to replicate the genome accurately being a prerequisite for evolution itself. Thus, one might anticipate that the mechanism by which all polymerases work would be both simple and universal. Further, these enzymatic scribes must faithfully copy the sequences of the genome into daughter nucleic acid or the information contained within would be lost; thus some mechanism of assuring fidelity is required. Finally, all classes of polynucleotide polymerases must be able to translocate along the template being copied as synthesis proceeds. The crystal structures of numerous DNA polymerases from different families suggest that they all utilize an identical two-metal-ioncatalyzed polymerase mechanism but differ extensively in many of their structural features…

We describe the 2.6 Å resolution crystal structure of RB69 DNA polymerase with primer-template DNA and dTTP, capturing the step just before primer extension. This ternary complex structure in the human DNA polymerase α family shows a 60° rotation of the fingers domain relative to the apo-protein structure, similar to the fingers movement in pol I family polymerases. Minor groove interactions near the primer 3′ terminus suggest a common fidelity mechanism for pol I and pol α family polymerases. The duplex product DNA orientation differs by 40° between the polymerizing mode and editing mode structures. The role of the thumb in this DNA motion provides a model for editing in the pol α family.

To make messenger RNA transcripts, bacteriophage T7 RNA polymerase (T7 RNAP) undergoes a transition from an initiation phase, which only makes short RNA fragments, to a stable elongation phase. We have determined at 2.1 angstrom resolution the crystal structure of a T7 RNAP elongation complex with 30 base pairs of duplex DNA containing a “transcription bubble” interacting with a 17-nucleotide RNA transcript. The transition from an initiation to an elongation complex is accompanied by a major refolding of the amino-terminal 300 residues. This results in loss of the promoter binding site, facilitating promoter clearance, and creates a tunnel that surrounds the RNA transcript after it peels off a seven–base pair heteroduplex. Formation of the exit tunnel explains the enhanced processivity of the elongation complex. Downstream duplex DNA binds to the fingers domain, and its orientation relative to upstream DNA in the initiation complex implies an unwinding that could facilitate formation of the open promoter complex.

RNA polymerase functions like a molecular motor that can convert chemical energy into the work of strand separation and translocation along the DNA during transcription. The structures of phage T7 RNA polymerase in an elongation phase substrate complex that includes the incoming nucleoside triphosphate and a pretranslocation product complex that includes the product pyrophosphate (PP_i) are described here. These structures and the previously determined posttranslocation elongation complex demonstrate that two enzyme conformations exist during a cycle of single nucleotide addition. One orientation of a five-helix subdomain is stabilized by the phosphates of either the incoming NTP or by the product PP_i. A second orientation of this subdomain is stable in their absence and is associated with translocation of the heteroduplex product as well as strand separation of the downstream DNA. We propose that the dissociation of the product PP_i after nucleotide addition produces the protein conformational change resulting in translocation and strand separation.

The crystal structure of the catalytic α–subunit of the DNA polymerase III (PolIIIα) holoenzyme bound to primer–template DNA and an incoming deoxy-nucleoside 5′-triphosphate has been determined at 4.6-Å resolution. The polymerase interacts with the sugar–phosphate backbone of the DNA across its minor groove, which is made possible by significant movements of the thumb, finger, and β-binding domains relative to their orientations in the unliganded polymerase structure. Additionally, the DNA and incoming nucleotide are bound to the active site of PolIIIα nearly identically as they are in their complex with DNA polymerase β, thereby proving that the eubacterial replicating polymerase, but not the eukaryotic replicating polymerase, is homologous to DNA polymerase β. Finally, superimposing a recent structure of the clamp bound to DNA on this PolIIIα complex with DNA places a loop of the β-binding domain into the appropriate clamp cleft and supports a mechanism of polymerase switching.

DNA polymerases can only synthesize nascent DNA from single-stranded DNA (ssDNA) templates. In bacteria, the unwinding of parental duplex DNA is carried out by the replicative DNA helicase (DnaB) that couples NTP hydrolysis to 5′ to 3′ translocation. The crystal structure of the DnaB hexamer in complex with GDP-AIF₄ and ssDNA reported here reveals that DnaB adopts a closed spiral staircase quaternary structure around an A-form ssDNA with each C-terminal domain coordinating two nucleotides of ssDNA. The structure not only provides structural insights into the translocation mechanism of superfamily IV helicases but also suggests that members of this superfamily employ a translocation mechanism that is distinct from other helicase super-families. We propose a hand-over-hand mechanism in which sequential hydrolysis of NTP causes a sequential 5′ to 3′ movement of the subunits along the helical axis of the staircase, resulting in the unwinding of two nucleotides per subunit.

Tom arrived at Yale University in late 1970, without a beard and with a firm vision of the future (Fig. 1). Although he and I shared an administrative office with Frederic Richards, our chair, Tom and I had hardly any scientific interactions for the next 15 years—Tom did sophisticated structural work on topics unrelated to protein translation, while I studied the many roles of transfer RNA. However, there were many conferences of the genetic code and translation field that we both attended, where Joan and Tom were regulars (Fig. 2). By the mid-1980s, we had in a decade-long effort developed Escherichia coli glutaminyl-tRNA synthetase into a versatile genetic and biochemical system to examine the mechanism of tRNA synthetase-catalyzed aminoacyl-tRNA formation. Clearly, structural information was now required to advance our knowledge. In the meantime Tom’s 20-year-old research interest in aminoacyl-tRNA synthetases had come to the fore. This led to a collaboration of two graduate students (Mark Rould from Tom’s group and John Perona from my lab) that resulted in the high resolution structure of the GlnRS-tRNA^Gln complex; this provided remarkable insight into the mechanism of tRNA recognition by a corresponding tRNA synthetase. This study also explained how tRNAs can be misrecognized by non-cognate tRNA synthetases leading to tRNA mischarging in vivo (Perona et al., 1988; 1989; Round et al., 1989)…

Even before I ever heard about Tom Steitz, I had read his papers. Some of these most deeply intriguing papers I had read as an undergraduate convinced me to apply to Yale University for graduate school in 1991. They were those that described the structurefunction of nucleic acid binding proteins—the structure of catabolite activator protein (CAP) demonstrating its ability to bend DNA by 90° (Rould et al., 1989) and the Structure of Glutaminyl tRNA synthetase bound to tRNA^Gln demonstrating the beauty of its specific RNA nucleotide recognition (Schultz et al., 1991). Both were seminal papers at the time and contributed to my initial sense that Yale was heavily investing in the field of nucleic acids. And although my undergraduate mentor had originally been persuaded to believe that NMR spectroscopy was the technique of choice, I decided to give Tom Steitz’s X-ray crystallography laboratory a try for my last rotation…

The crystal structure of Escherichia coli glutaminyl-tRNA synthetase (GlnRS) complexed with its cognate glutaminyl transfer RNA (tRNA^Gln) and adenosine triphosphate (ATP) has been derived from a 2.8 angstrom resolution electron density map and the known protein and tRNA sequences. The 63.4-kilodalton monomeric enzyme consists of four domains arranged to give an elongated molecule with an axial ratio greater than 3 to 1. Its interactions with the tRNA extend from the anticodon to the acceptor stem along the entire inside of the L of the tRNA. The complexed tRNA retains the overall conformation of the yeast phenylalanine tRNA (tRNA^Phe) with two major differences: the 3′ acceptor strand of tRNA^Gln makes a hairpin turn toward the inside of the L, with the disruption of the final base pair of the acceptor stem, and the anticodon loop adopts a conformation not seen in any of the previously determined tRNA structures. Specific recognition elements identified so far include (i) enzyme contacts with the 2-amino groups of guanine via the tRNA minor groove in the acceptor stem at G2 and G3; (ii) interactions between the enzyme and the anticodon nucleotides; and (iii) the ability of the nucleotides G73 and U1 · A72 of the cognate tRNA to assume a conformation stabilized by the protein at a lower free energy cost than noncognate sequences. The central domain of this synthetase binds ATP, glutamine, and the acceptor end of the tRNA as well as making specific interactions with the acceptor stem. It is structurally similar to the dinucleotide binding motifs of the tyrosyl- and methionyl-tRNA synthetases, suggesting that all synthetases may have evolved from a common domain that can recognize the acceptor stem of the cognate tRNA.

The refined crystal structure of Escherichia coli glutaminyl transfer RNA synthetase complexed with transfer RNA^Gln and ATP reveals that the structure of the anticodon loop of the enzyme-bound tRNA^Gln differs extensively from that of the known crystal structures of uncomplexed tRNA molecules. The anticodon stem is extended by two non-Watson–Crick base pairs, leaving the three anticodon bases unpaired and splayed out to bind snugly into three separate complementary pockets in the protein. These interactions suggest that the entire anticodon loop provides essential sites for glutaminyl tRNA synthetase discrimination among tRNA molecules.

The structure of Escherichia coli glutaminyl-tRNA synthetase complexed with tRNA₂^Gln and ATP refined at 2.5-Å resolution reveals structural details of the catalytic center and allows description of the specific roles of individual amino acid residues in substrate binding and catalysis. The reactive moieties of the ATP and tRNA substrates are positioned within hydrogen-bonding distance of each other. Model-building has been used to position the glutamine substrate in an adjacent cavity with its reactive carboxylate adjacent to the α-phosphate of ATP; the interactions of the carboxyamide side chain suggest a structural rationale for the way in which the enzyme discriminates against glutamate. The binding site for a manganese ion has also been identified bridging the β- and γ-phosphates of the ATP. The well-known HIGH and KMSKS sequence motifs interact directly with each other as well as with the ATP, providing a structural rationale for their simultaneous conservation in all class I synthetases. The KMSKS loop adopts a wellordered and catalytically productive conformation as a consequence of interactions made with the proximal β-barrel domain. While there are no protein side chains near the reaction site that might function in acid-base catalysis, the side chains of two residues, His43 and Lys270, are positioned to assist in stabilizing the expected pentacovalent intermediate at the α-phosphate. Transfer of glutamine to the 3′-terminal tRNA ribose may well proceed by intramolecular catalysis involving proton abstraction by a phosphate oxygen atom of glutaminyl adenylate. Catalytic competence of the crystalline enzyme is directly shown by its ability to hydrolyze ATP and release pyrophosphate when crystals of the ternary complex are soaked in mother liquor containing glutamine.

Isoleucyl–transfer RNA (tRNA) synthetase (IleRS) joins Ile to tRNA^Ile at its synthetic active site and hydrolyzes incorrectly acylated amino acids at its editing active site. The 2.2 angstrom resolution crystal structure of Staphylococcus aureus IleRS complexed with tRNA^Ile and Mupirocin shows the acceptor strand of the tRNA^Ile in the continuously stacked, A-form conformation with the 3′ terminal nucleotide in the editing active site. To position the 3′ terminus in the synthetic active site, the acceptor strand must adopt the hairpinned conformation seen in tRNA^Gln complexed with its synthetase. The amino acid editing activity of the IleRS may result from the incorrect products shuttling between the synthetic and editing active sites, which is reminiscent of the editing mechanism of DNA polymerases.

THE recA protein catalyses the ATP-driven homologous pairing and strand exchange of DNA molecules. It is an allosteric enzyme: the ATPase activity is DNA-dependent, and ATP-bound recA protein has a high affinity for DNA, whereas the ADP-bound form has a low affinity. In the absence of ATP hydrolysis, recA protein can still promote homologous pairing, apparently through the formation of a triple-stranded intermediate. The exact role of ATP hydrolysis is not clear, but it presumably drives the triplex intermediate towards products. Here we determine the position of bound ADP diffused into the recA crystal. We show that only the phosphates are bound in the same way as in other NTPases containing the G/AXXXXGKT/S motif. We propose that recA protein may change its conformation upon ATP hydrolysis in a manner analogous to one such protein, the p21 protein from the ras oncogene. A model is presented to account for the allosteric stimulation of DNA binding by ATP. The mechanism by which nucleoside triphosphate hydrolysis is coupled to the binding of another ligand in recA protein and p21 may be typical of the large class of NTPases containing this conserved motif.

The crystal structure of the recA protein from Escherichia coli at 2.3-Å resolution reveals a major domain that binds ADP and probably single- and double-stranded DNA. Two smaller subdomains at the N and C termini protrude from the protein and respectively stabilize a 6₁ helical polymer of protein subunits and interpolymer bundles. This polymer structure closely resembles that of recA/DNA filaments determined by electron microscopy. Mutations in recA protein that enhance coprotease, DNA-binding and/or strand-exchange activity can be explained if the interpolymer interactions in the crystal reflect a regulatory mechanism in vivo.

The structure of γδ resolvase complexed with a 34 bp substrate DNA has been determined at 3.0 Å resolution. The DNA is sharply bent by 60° toward the major groove and away from the resolvase catalytic domains at the recombination crossover point. The C-terminal one third of resolvase, which was disordered in the absence of DNA, forms an arm and a 3-hellx DNA-binding domain on the opposite side of the DNA from the N-terminal domain. The arms wrap around the minor groove of the central 16 bp, and the DNA-binding domains interact with the major grooves near the outer boundaries of the binding site. The resolvase dimer is asymmetric, particularly in the arm region, implying a conformational adaptability that may be important for resolvase binding to different DNA sites in the synaptosome. It also raises the possibility of a sequential single-strand cleavage mechanism.

The structure of a synaptic intermediate of the site-specific recombinase γδ resolvase covalently linked through Ser¹⁰ to two cleaved duplex DNAs has been determined at 3.4 angstrom resolution. This resolvase, activated for recombination by mutations, forms a tetramer whose structure is substantially changed from that of a presynaptic complex between dimeric resolvase and the cleavage site DNA. Because the two cleaved DNA duplexes that are to be recombined lie on opposite sides of the core tetramer, large movements of both protein and DNA are required to achieve strand exchange. The two dimers linked to the DNAs that are to be recombined are held together by a flat interface. This may allow a 180° rotation of one dimer relative to the other in order to reposition the DNA duplexes for strand exchange.

Early in the summer of 2000, Tom and Peter Moore sent Jeff Hansen, then a postdoc in Tom’s lab, to present at an international conference. This was during a particularly exciting time for the laboratory. Nenad Ban, Poul Nissen and Jeff had recently collected diffraction data to a resolution of 2.4 Å, allowing them to calculate high resolution maps of the large ribosomal subunit for the first time. In his talk, Jeff presented preliminary insight into ribosomal structure and function to a captivated audience. When it came time for questions, Jamie Williamson, a professor at the Scripps Research Institute, asked whether the structure provided insight into the phenotype resulting from a mutation in his favorite ribosomal protein. Jeff answered, “Actually, I’m not sure. We postdocs have been modeling the rRNA. The protein structures are being built by rotation students.” This drew gasps and laughter from the audience. It led Jamie to reply, “How do I apply to become a rotation student in the Steitz laboratory?”…

The determination of the structure of the large ribosomal subunit which Tom, Peter, Nenad, Poul and Jeff accomplished was a monumental undertaking. It required technical advances, ambitious and creative approaches, and years of hard work. Its fruition was amazing and outstandingly impactful. As an illustration, for the previous 15 years, groups around the world had been taking a divide-and-conquer approach by solving structures of isolated ribosomal proteins. These structures were sometimes published in the highest profile journals. With the determination of the 50S structure and of the 30S structures from Venki’s and Ada’s groups, practically all the prokaryotic ribosomal protein structures were solved in their native context in one fell swoop. Moreover, the 50S structure bound to the “Yarus inhibitor” visualized the peptidyl transferase active site and showed it was composed of RNA, suggesting that the mother of all proteins was indeed an RNA enzyme, as Francis Crick had reasoned decades before. So it is not surprising that soon after such a major achievement, the postdocs Tom had assembled to solve the 50S structure left the laboratory to start their own research groups. Nenad moved to ETH Zurich and Poul to Aarhus, where they still run their own very successful groups to this day. But there was so much more to be learned about the structure and function of the 50S subunit, and the H. marismortui crystal form gave us a powerful tool to do so. So Tom gave a new group of young scientists golden opportunities to contribute to this important work and join the ribosome subgroup…

With the 50S structure the RNA structure database was suddenly five-doubled or so. It was obvious that we might discover important principles and motifs of the RNA tertiary structure. Tom would often point out that conserved residues are either important for direct function or tertiary structure (Nissen et al., 2001). So, we marked all the conserved residues to guide us. While it certainly highlighted core structures of ribosomal RNA, it still overwhelmed the picture and was not very helpful. It made however a great panel for the Science cover illustration with four Warhol inspired representations of 50S. An automated method for motif recognition would be a great help, but how would one look for something that had not been identified yet? Not exactly algorithm-friendly of the time, so the challenge was up for visual inspections, structural analysis and creative thinking. Perhaps machine learning could be helpful today—seems like a good challenge…

Over the course of five years the Steitz–Moore group transformed the textbook blob of the H. marismortui 50S subunit into an atomic model of the 50S subunit determined from experimental maps at 2.4 Å resolution…

The 50S subunit of the ribosome catalyzes the peptidyl-transferase reaction of protein synthesis. We have generated X-ray crystallographic electron density maps of the large ribosomal subunit from Haloarcula marismortui at various resolutions up to 9 Å using data from crystals that diffract to 3 Å. Positioning a 20 Å resolution EM image of these particles in the crystal lattice produced phases accurate enough to locate the bound heavy atoms in three derivatives using difference Fourier maps, thus demonstrating the correctness of the EM model and its placement in the unit cell. At 20 Å resolution, the X-ray map is similar to the EM map; however, at 9 Å it reveals long, continuous, but branched features whose shape, diameter, and right-handed twist are consistent with segments of double-helical RNA that crisscross the subunit.

We have calculated at 5.0 Å resolution an electron-density map of the large 50S ribosomal subunit from the bacterium Haloarcula marismortui by using phases derived from four heavy-atom derivatives, intercrystal density averaging and density-modification procedures. More than 300 base pairs of A-form RNA duplex have been fitted into this map, as have regions of non-A-form duplex, single-stranded segments and tetraloops. The long rods of RNA crisscrossing the subunit arise from the stacking of short, separate double helices, not all of which are A-form, and in many places proteins crosslink two or more of these rods. The polypeptide exit channel was marked by tungsten cluster compounds bound in one heavy-atom-derivatized crystal. We have determined the structure of the translation-factor-binding centre by fitting the crystal structures of the ribosomal proteins L6, L11 and L14, the sarcin–ricin loop RNA, and the RNA sequence that binds L11 into the electron density. We can position either elongation factor G or elongation factor Tu complexed with an aminoacylated transfer RNA and GTP onto the factor-binding centre in a manner that is consistent with results from biochemical and electron microscopy studies.

The large ribosomal subunit catalyzes peptide bond formation and binds initiation, termination, and elongation factors. We have determined the crystal structure of the large ribosomal subunit from Haloarcula marismortui at 2.4 angstrom resolution, and it includes 2833 of the subunit’s 3045 nucleotides and 27 of its 31 proteins. The domains of its RNAs all have irregular shapes and fit together in the ribosome like the pieces of a three-dimensional jigsaw puzzle to form a large, monolithic structure. Proteins are abundant everywhere on its surface except in the active site where peptide bond formation occurs and where it contacts the small subunit. Most of the proteins stabilize the structure by interacting with several RNA domains, often using idiosyncratically folded extensions that reach into the subunit’s interior.

Using the atomic structures of the large ribosomal subunit from Haloarcula marismortui and its complexes with two substrate analogs, we establish that the ribosome is a ribozyme and address the catalytic properties of its all-RNA active site. Both substrate analogs are contacted exclusively by conserved ribosomal RNA (rRNA) residues from domain V of 23S rRNA; there are no protein side-chain atoms closer than about 18 angstroms to the peptide bond being synthesized. The mechanism of peptide bond synthesis appears to resemble the reverse of the acylation step in serine proteases, with the base of A2486 (A2451 in Escherichia coli) playing the same general base role as histidine-57 in chymotrypsin. The unusual pK_a (where K_a is the acid dissociation constant) required for A2486 to perform this function may derive in part from its hydrogen bonding to G2482 (G2447 in E. coli), which also interacts with a buried phosphate that could stabilize unusual tautomers of these two bases. The polypeptide exit tunnel is largely formed by RNA but has significant contributions from proteins L4, L22, and L39e, and its exit is encircled by proteins L19, L22, L23, L24, L29, and L31e.

Analysis of the 2.4-Å resolution crystal structure of the large ribosomal subunit from Haloarcula marismortui reveals the existence of an abundant and ubiquitous structural motif that stabilizes RNA tertiary and quaternary structures. This motif is termed the A-minor motif, because it involves the insertion of the smooth, minor groove edges of adenines into the minor groove of neighboring helices, preferentially at C-G base pairs, where they form hydrogen bonds with one or both of the 2′ OHs of those pairs. A-minor motifs stabilize contacts between RNA helices, interactions between loops and helices, and the conformations of junctions and tight turns. The interactions between the 3′ terminal adenine of tRNAs bound in either the A site or the P site with 23S rRNA are examples of functionally significant A-minor interactions. The A-minor motif is by far the most abundant tertiary structure interaction in the large ribosomal subunit; 186 adenines in 23S and 5S rRNA participate, 68 of which are conserved. It may prove to be the universally most important long-range interaction in large RNA structures.

Analysis of the Haloarcula marismortui large ribosomal subunit has revealed a common RNA structure that we call the kink-turn, or K-turn. The six K-turns in H.marismortui 23S rRNA superimpose with an r.m.s.d. of 1.7 Å. There are two K-turns in the structure of Thermus thermophilus 16S rRNA, and the structures of U4 snRNA and L30e mRNA fragments form K-turns. The structure has a kink in the phosphodiester backbone that causes a sharp turn in the RNA helix. Its asymmetric internal loop is flanked by C–G base pairs on one side and sheared G–A base pairs on the other, with an A-minor interaction between these two helical stems. A derived consensus secondary structure for the K-turn includes 10 consensus nucleotides out of 15, and predicts its presence in the 5′-UTR of L10 mRNA, helix 78 in Escherichia coli 23S rRNA and human RNase MRP. Five K-turns in 23S rRNA interact with nine proteins. While the observed K-turns interact with proteins of unrelated structures in different ways, they interact with L7Ae and two homologous proteins in the same way.

Crystal structures of the Haloarcula marismortui large ribosomal subunit complexed with the 16-membered macrolide antibiotics carbomycin A, spiramycin, and tylosin and a 15-membered macrolide, azithromycin, show that they bind in the polypeptide exit tunnel adjacent to the peptidyl transferase center. Their location suggests that they inhibit protein synthesis by blocking the egress of nascent polypeptides. The saccharide branch attached to C5 of the lactone rings extends toward the peptidyl transferase center, and the isobutyrate extension of the carbomycin A disaccharide overlaps the A-site. Unexpectedly, a reversible covalent bond forms between the ethylaldehyde substituent at the C6 position of the 16-membered macrolides and the N6 of A2103 (A2062, E. coli). Mutations in 23S rRNA that result in clinical resistance render the binding site less complementary to macrolides.

Structures of anisomycin, chloramphenicol, sparsomycin, blasticidin S, and virginiamycin M bound to the large ribosomal subunit of Haloarcula marismortui have been determined at 3.0 Å resolution. Most of these antibiotics bind to sites that overlap those of either peptidyl-tRNA or aminoacyl- tRNA, consistent with their functioning as competitive inhibitors of peptide bond formation. Two hydrophobic crevices, one at the peptidyl transferase center and the other at the entrance to the peptide exit tunnel play roles in binding these antibiotics. Midway between these crevices, nucleotide A2103 of H. marismortui (2062 Escherichia coli) varies in its conformation and thereby contacts antibiotics bound at either crevice. The aromatic ring of anisomycin binds to the active-site hydrophobic crevice, as does the aromatic ring of puromycin, while the aromatic ring of chloramphenicol binds to the exit tunnel hydrophobic crevice. Sparsomycin contacts primarily a P-site bound substrate, but also extends into the active-site hydrophobic crevice. Virginiamycin M occupies portions of both the A and P-site, and induces a conformational change in the ribosome. Blasticidin S base-pairs with the P-loop and thereby mimics C74 and C75 of a P-site bound tRNA.

The large ribosomal subunit catalyses the reaction between the α-amino group of the aminoacyl-tRNA bound to the A site and the ester carbon of the peptidyl-tRNA bound to the P site1, while preventing the nucleophilic attack of water on the ester, which would lead to unprogrammed deacylation of the peptidyl-tRNA. Here we describe three new structures of the large ribosomal subunit of Haloarcula marismortui (Hma) complexed with peptidyl transferase substrate analogues that reveal an induced-fit mechanism in which substrates and active-site residues reposition to allow the peptidyl transferase reaction. Proper binding of an aminoacyl-tRNA analogue to the A site induces specific movements of 23S rRNA nucleotides 2618–2620 (Escherichia coli numbering 2583–2585) and 2541(2506), thereby reorienting the ester group of the peptidyl-tRNA and making it accessible for attack. In the absence of the appropriate A-site substrate, the peptidyl transferase centre positions the ester link of the peptidyltRNA in a conformation that precludes the catalysed nucleophilic attack by water. Protein release factors may also function, in part, by inducing an active-site rearrangement similar to that produced by the A-site aminoacyl-tRNA, allowing the carbonyl group and water to be positioned for hydrolysis.

Peptide bond formation is catalyzed at the peptidyl transferase center (PTC) of the large ribosomal subunit. Crystal structures of the large ribosomal subunit of Haloarcula marismortui (Hma) complexed with several analogs that represent either the substrates or the transition state intermediate of the peptidyl transferase reaction show that this reaction proceeds through a tetrahedral intermediate with S chirality. The oxyanion of the tetrahedral intermediate interacts with a water molecule that is positioned by nucleotides A2637 (E. coli numbering, 2602) and ^methylU2619(2584). There are no Mg²⁺ ions or monovalent metal ions observed in the PTC that could directly promote catalysis. The A76 2′ hydroxyl of the peptidyl-tRNA is hydrogen bonded to the α-amino group and could facilitate peptide bond formation by substrate positioning and by acting as a proton shuttle between the α-amino group and the A76 3′ hydroxyl of the peptidyl-tRNA.

Crystal structures of H. marismortui large ribosomal subunits containing the mutation G2099A (A2058 in E. coli) with erythromycin, azithromycin, clindamycin, virginiamycin S, and telithromycin bound explain why eubacterial ribosomes containing the mutation A2058G are resistant to them. Azithromycin binds almost identically to both G2099A and wild-type subunits, but the erythromycin affinity increases by more than 10₄-fold, implying that desolvation of the N2 of G2099 accounts for the low wild-type affinity for macrolides. All macrolides bind similarly to the H. marismortui subunit, but their binding differs significantly from what has been reported in the D. radioidurans subunit. The synergy in the binding of streptogramins A and B appears to result from a reorientation of the base of A2103 (A2062, E. coli) that stacks between them. The structure of large subunit containing a three residue deletion mutant of L22 shows a change in the L22 structure and exit tunnel shape that illuminates its macrolide resistance phenotype.

Viomycin and capreomycin belong to the tuberactinomycin family of antibiotics, which are among the most effective antibiotics against multidrug-resistant tuberculosis. Here we present two crystal structures of the 70S ribosome in complex with three tRNAs and bound to either viomycin or capreomycin at 3.3- and 3.5-Å resolution, respectively. Both antibiotics bind to the same site on the ribosome, which lies at the interface between helix 44 of the small ribosomal subunit and helix 69 of the large ribosomal subunit. The structures of these complexes suggest that the tuberactinomycins inhibit translocation by stabilizing the tRNA in the A site in the pretranslocation state. In addition, these structures show that the tuberactinomycins bind adjacent to the binding sites for the paromomycin and hygromycin B antibiotics, which may enable the development of new derivatives of tuberactinomycins that are effective against drug-resistant strains.

The increasing prevalence of antibiotic-resistant pathogens reinforces the need for structures of antibiotic-ribosome complexes that are accurate enough to enable the rational design of novel ribosome-targeting therapeutics. Structures of many antibiotics in complex with both archaeal and eubacterial ribosomes have been determined, yet discrepancies between several of these models have raised the question of whether these differences arise from species-specific variations or from experimental problems. Our structure of chloramphenicol in complex with the 70S ribosome from Thermus thermophilus suggests a model for chloramphenicol bound to the large subunit of the bacterial ribosome that is radically different from the prevailing model. Further, our structures of the macrolide antibiotics erythromycin and azithromycin in complex with a bacterial ribosome are indistinguishable from those determined of complexes with the 50S subunit of Haloarcula marismortui, but differ significantly from the models that have been published for 50S subunit complexes of the eubacterium Deinococcus radiodurans. Our structure of the antibiotic telithromycin bound to the T. thermophilus ribosome reveals a lactone ring with a conformation similar to that observed in the H. marismortui and D. radiodurans complexes. However, the alkyl-aryl moiety is oriented differently in all three organisms, and the contacts observed with the T. thermophilus ribosome are consistent with biochemical studies performed on the Escherichia coli ribosome. Thus, our results support a mode of macrolide binding that is largely conserved across species, suggesting that the quality and interpretation of electron density, rather than species specificity, may be responsible for many of the discrepancies between the models.

A mechanism is proposed for the RNA-catalyzed reactions involved in RNA splicing and RNase P hydrolysis of precursor tRNA. The mechanism postulates that chemical catalysis is facilitated by two divalent metal ions 3.9 Å apart, as In phosphoryl transfer reactions catalyzed by protein enzymes, such as the 3′, 5′-exonuclease or Escherichia coli DNA polymerase I. One metal ion activates the attacking water or sugar hydroxyl, while the other coordinates and stabilizes the oxyanion leaving group. Both ions act as Lewis acids and stabilize the expected pentacovalent transition state. The symmetry of a two-metal-ion catalytic site fits well with the known reaction pathway of group I self-splicing introns and can also be reconciled with emerging data on group II self-splicing introns, the spliceosome, and RNase P. Tbe role or the RNA Is to position the two catalytic metal ions and properly orient the substrates via three specific binding sites.

The following sections are included:

• Additional Photographs of Tom
• Obituaries
• Corpus Acknowledgments