Pushing Induced Fit to Its Limits

Structure 484

Notably, many cytoplasmic signaling adaptor proteins that bind activated cell surface receptors also associate with the plasma membrane through phospholipid interactions. Such a multivalent approach is advantageous in positioning the adaptor protein as well as the receptor for optimal association in specific locations at the membrane. Various mechanisms have evolved to accomplish membrane localization. Some signaling proteins contain both PH and PTB/SH2 domains in combination to achieve phospholipid and receptor binding. Additionally, some PTB domain-containing adaptor proteins such as fibroblast growth factor receptor substrate 2 (also known as SNT) contain an N-terminal myristoylation sequence, which serves to tether the proteins to the plasma membrane. In the case of Dab1, membrane association could be achieved through the built-in phosphoinositide binding activity of a single PTB/PH hybrid domain that also interacts with the cell surface receptor. Given the role of Dab1 as an adaptor for LDL receptor signaling, membrane localization is perhaps an even more essential component of its function. This pathway results in multiple membrane-associated processes, including downstream activation of the phosphatidylinositol-3 kinase pathway and receptor-mediated endocytosis [6]. The three-dimensional crystal structure of the Dab1 PTB domain/ApoER2 peptide/PI-4,5P2 ternary complex reported here provides a structural realization of a dual-mode molecular recognition by this important adaptor. It remains to be seen whether these two distinct binding modes are functionally coupled in vivo, and whether this seemingly efficient use of a single protein domain for two biological interactions is advantageous for a transient and reversible event in receptor signaling.

Kelley S. Yan and Ming-Ming Zhou Structural Biology Program Department of Physiology and Biophysics Mount Sinai School of Medicine New York University New York, New York 10029 Selected Reading 1. Forman-Kay, J.D., and Pawson, T. (1999). Curr. Opin. Struct. Biol. 9, 690–695. 2. Yan, K.S., Kuti, M., and Zhou, M.-M. (2002). FEBS Lett. 513, 67–70. 3. Kavanaugh, W.M., and Williams, L.T. (1994). Science 266, 1862– 1865. 4. Blaikie, P., Immanuel, D., Wu, J., Li, N., Yajnik, V., and Margolis, B. (1994). J. Biol. Chem. 269, 32031–32034. 5. Gustafson, T.A., He, W., Craparo, A., Schaub, C.D., and O’Meill, T.J. (1995). Mol. Cell. Biol. 15, 2500–2508. 6. Stolt, P.C., Jeon, H., Song, H.K., Herz, J., Eck, M.J., and Blacklow, S.C. (2003). Structure 11, this issue, 569–579. 7. Blomberg, N., Baraldi, E., Nilges, M., and Saraste, M. (1999). Trends Biochem. Sci. 24, 441–445. 8. Lemmon, M.A., Ferguson, K.M., and Schlessinger, J. (1996). Cell 85, 621–624. 9. Zhou, M.-M., Ravichandran, K.S., Olejniczak, E.T., Petros, A.P., Meadows, R.P., Sattler, M., Harlan, J.E., Wade, W., Burakoff, S.J., and Fesik, S.W. (1995). Nature 378, 584–592. 10. Yoon, H.S., Hajduk, P.J., Petros, A.M., Olejniczak, E.T., Meadows, R.P., and Fesik, S.W. (1994). Nature 369, 672–675. 11. Zhou, M.-M., Huang, B., Olejniczak, E.T., Meadows, R.P., Shuker, S.B., Miyazak, M., Tru¨b, T., Shoelson, S.E., and Fesik, S.W. (1996). Nat. Struct. Biol. 3, 388–393. 12. Eck, M.J., Dhe-pagnon, S., Tru¨b, T., Nolte, R., and Shoelson, S.E. (1996). Cell 85, 695–705.

Structure, Vol. 11, May, 2003, 2003 Elsevier Science Ltd. All rights reserved.

Pushing Induced Fit to Its Limits: tRNA-Dependent Active Site Assembly in Class I Aminoacyl-tRNA Synthetases Aminoacyl-tRNA synthetases are responsible for aminoacylating their cognate tRNAs with a unique amino acid. Recent structural work shows how binding of substrates is coupled to assembly of the active site. Aminoacyl-tRNA synthetases (aaRSs) constitute a family of RNA binding proteins that play a central role in translation by forming the aminoacyl-tRNAs used in protein biosynthesis. In most organisms, there are 20 distinct aaRSs, each of them being responsible for aminoacylating its cognate tRNA(s) with a unique amino acid in a two-step catalytic reaction. The first step leads to

DOI 10.1016/S0969-2126(03)00080-7

the formation of an enzyme-bound intermediate called aminoacyl-adenylate from amino acids, ATP, and magnesium ions. At the second step, the aminoacyl moiety is transferred to one of the hydroxyl groups of the 3⬘terminal adenosine of the tRNA to form an aminoacyltRNA. Due to their fundamental importance, the aaRSs are likely to be one of the most ancient families of proteins and have therefore been analyzed extensively (for a recent review of the field, see [1]). The aaRSs are divided into two classes, which correspond to two architectures of the active site core characterized by conserved amino acid residues. Prodigious efforts spanning more than 25 years led to the determination of the crystal structures of mainly all aaRSs, either in the free state or in complexes with the other partners of the aminoacylation reaction. aaRSs now constitute the best textbook example of multidomain proteins, including insertion and terminal functional modules, appended to one of the two class-specific active site domains. In several cases, the specificity

Previews 485

of aaRSs for their cognate substrates has been analyzed in terms of chemistry and physical interactions at the atomic level, revealing a remarkable diversity in the strategies employed for amino acid and tRNA recognition. Moreover, it has been clearly demonstrated that the specificity of aaRSs for their cognate substrates relies heavily on idiosyncratic induced-fit conformational changes upon substrate binding. Induced fit upon binding has been seen in the tRNA and protein components. Three of the canonical class I aaRSs, the arginyl-, glutaminyl-, and glutamyl-tRNA synthetases (ArgRS, GlnRS, and GluRS) and the recently discovered class I type lysyl-tRNA synthetase (LysRS-I) are known to share a peculiar property in that they only catalyze the first step of the aminoacylation reaction, amino acid activation, in the presence of their cognate tRNA(s). This subset of aminoacyl-tRNAs must have a regulation mechanism to avoid aminoacyl-AMP formation in the absence of tRNA. Therefore, a scientific debate has been going on for more than 30 years concerning the detailed mechanism of the aminoacylation reaction by ArgRS, GlnRS, and GluRS. The question of why a subset of aaRSs requires tRNA to activate amino acids has been addressed by X-ray crystallography, and significant insight into this question is now in hand for ArgRS, GlnRS, and GluRS. Different crystals forms corresponding to different snapshots of the aminoacylation reaction have been obtained and solved. Comparative analyses of the states of the reaction give some precious clues to understand this complex biological process. Five different crystal forms corresponding to four different states of the arginylation reaction have been visualized for ArgRS [2–4]. The structural data have shown that the binding of tRNAArg produces conformational changes of the ATP binding cleft and builds up the productive ATP binding pocket. Several key residues of the active site play multiple roles in the catalytic pathway. Moreover, through a molecular switch based on two different conformations of a phylogenetically invariant tyrosine, it has been shown that L-Arg binding is a prerequisite that triggers the correct positioning of the CCA end of the tRNAArg in the catalytic center. Structural linkages that may communicate the anticodon binding signal to the catalytic site have been revealed. Recently, Sekine and coworkers [5] have reported four new structures of GluRS from Thermus thermophilus corresponding to different complexes, thus adding an impressive amount of data to work already published [6, 7]. They nicely show that GluRS possesses two modes for ATP binding, a “nonproductive” and a “productive” binding mode, which can be switched in a tRNA-dependent manner. In the absence of tRNAGlu, GluRS, ATP, and glutamate form a “dead-end” complex. Analyses of GluRS sequences have shown that the amino acid residues characteristic of the productive and nonproductive ATP binding in T. thermophilus GluRS are almost conserved in bacterial/organellar GluRSs, suggesting the conservation of the mechanism in this group. In this issue of Structure, Sherlin and Perona [8] report the crystal structure of ligand-free E. coli GlnRS at 2.4 A˚ resolution. The structure of E. coli GlnRS complexed

with its cognate tRNAGln and ATP was the first to be solved 14 years ago [9], followed several years later by a ternary complex involving GlnRS, tRNAGln, and a stable glutaminyl-adenylate analog [10]. Comparison of this new structure with the tRNA-bound complexes reveals that tRNAGln binding generates subtle but significant conformational changes in several regions of the protein, particularly in and adjacent to the active site cleft. Their analyses highlight the fact that binding of substrates is coupled to active site assembly and that neither the ATP nor the glutamine binding sites are fully formed in the unliganded GlnRS. A careful comparison of the mechanisms used by ArgRS, GlnRS, and GluRS still remains to be done in order to establish some general guidelines. Indeed, several features of the conformational changes observed upon substrate binding are likely to be idiosyncratic to each system. However, as pointed out by Sherlin and Perona, induced conformational transitions arising from proteins contact with different regions of the tRNA converge to modify the structure and interactions in the catalytic site. The observed structural changes stress the importance of the induced fit and rationalize the role of the tRNA as an obligate macromolecular cofactor in the first step of the aminoacylation reaction by ArgRS, GlnRS, and GluRS. Despite the elucidation of more than 15 crystal structures, additional kinetic and crystallographic analyses are required for a full description of such an induced-fit pathway. The pictures of several states are still missing; for example, the ATP binding manner of GlnRS or ArgRS in the absence of tRNA remains unknown. Moreover, for ArgRS, GlnRS, and GluRS, it has not been possible to freeze together all the partners of the aminoacylation reaction in a given state of the reaction. X-ray crystallography is challenged to produce a highly desired image of the transition state quaternary complex. Unfortunately, the problem of obtaining high-quality crystals of complexes containing a given aaRS and its cognate tRNA remains a bottleneck. Jean Cavarelli De´partement de Biologie et Ge´nomique Structurales UMR7104, Universite´ Louis Pasteur, Strasbourg IGBMC 1 rue Laurent Fries BP 10142 67404 Illkirch Cedex France Selected Reading 1. Francklyn, C., Perona, J.J., Puetz, J., and Hou, Y.M. (2002). RNA 8, 1363–1372. 2. Cavarelli, J., Delagoutte, B., Eriani, G., Gangloff, J., and Moras, D. (1998). EMBO J. 17, 5438–5448. 3. Delagoutte, B., Moras, D., and Cavarelli, J. (2000). EMBO J. 19, 5599–5610. 4. Shimada, A., Nureki, O., Goto, M., Takahashi, S., and Yokoyama, S. (2001). Proc. Natl. Acad. Sci. USA 98, 13537–13542. 5. Sekine, S., Nureki, O., Dubois, D.Y., Bernier, S., Chenevert, R., Lapointe, J., Vassylyev, D.G., and Yokoyama, S. (2003). EMBO J. 22, 676–688. 6. Nureki, O., Vassylyev, D.G., Katayanagi, K., Shimizu, T., Sekine, S., Kigawa, T., Miyazawa, T., Yokoyama, S., and Morikawa, K. (1995). Science 267, 1958–1965.

Structure 486

7. Sekine, S., Nureki, O., Shimada, A., Vassylyev, D.G., and Yokoyama, S. (2001). Nat. Struct. Biol. 8, 203–206. 8. Sherlin, L.D., and Perona, J.J. (2003). Structure 11, this issue, 591–603.

9. Rould, M.A., Perona, J.J., Soll, D., and Steitz, T.A. (1989). Science 246, 1135–1142. 10. Rath, V.L., Silvian, L.F., Beijer, B., Sproat, B.S., and Steitz, T.A. (1998). Structure 6, 439–449.

Structure, Vol. 11, April, 2003, 2003 Elsevier Science Ltd. All rights reserved.

DOI 10.1016/S0969-2126(03)00078-9

Folding by Consensus Recent design efforts have produced folded linear repeat proteins from duplicated consensus sequences with stabilities exceeding those of naturally occurring repeat proteins. Although the study of structural biology is nearly a half century old, our understanding of the relationship between sequence, structure, and stability is far from complete. One approach to test and advance our understanding of this important subject is to try to design proteins that adopt predetermined structures using first principles (chemical intuition) and “knowledge-based” information such as sequence homology. Some exercises in design have been limited to parts of folded proteins, as exemplified by the successful design of novel metal binding sites within the framework of naturally occurring protein domains [1]. Others have subjected the entire sequences of small, folded proteins to redesign, as exemplified in the work of Mayo and coworkers [2]. Another principle that can guide design is symmetry. By designing a structure with internal symmetry, a large protein can be constructed from a relatively small designed protomer, or module. DeGrado and coworkers made use of rotational symmetry to design helix bundles of varying stoichiometry and size [3]. Another way to take advantage of symmetry in protein design is to target a linearly repeating unit that possesses translational (or screw) symmetry. Unlike rotational symmetry, which is “closed,” translational symmetry can in principle be continued indefinitely, resulting in very large structures. Nature is full of proteins that contain repeated units of secondary structure that are arranged in linear, tandem arrays. Examples of such repeats include ␤-prism repeats, which consist of three ␤ strands, leucine-rich repeats, which contain a short ␤ strand and either a 310 or an ␣ helix, and a variety of helical repeats such as ankyrin, tetratricopeptide (TPR), and armadillo repeats. In this issue of Structure, Main et al. [4] present the successful design of a stable tetratricopeptide repeat (TPR) domain. TPR domains contain repeated motifs of 34 amino acids that form two antiparallel ␣ helices. Each repeat is packed against its neighbors with a geometry that results in a linear, superhelical structure. Main et al. used a collection of approximately 2000 TPR sequences to determine a consensus sequence based exclusively on the identity of the most probable amino acid at each

Figure 1. The Structure of a Consensus-Designed Ankyrin Repeat Protein Containing Four Identical Sequence Repeats Individual repeats in this 1.5 A˚ structure, determined by Mosavi et al. [5], are colored from red (N terminus) to blue (C terminus). Individual repeats superpose with an average C␣ RMSD of 0.33 A˚.

position [4]. Their designed repeat proteins match this consensus sequence with one exception: a cysteine residue is replaced with an alanine to prevent disulfide bond formation. In their design, the authors have also included a polar helix to terminate the C-terminal repeat, and three N-terminal residues to cap the N-terminal ␣ helix. Constructs that contain one to three copies of their consensus repeat are folded, show sharp thermal unfolding transitions, and have high melting temperatures (Table 1). Both of the larger constructs are amenable to structural analysis, giving high-quality NMR spectra and well-ordered crystals. The authors present the crystal structures of the two- and three-repeat TPR domains at 1.55 and 1.6 A˚, respectively, which demonstrate the success of their design strategy: both the two- and three-repeat constructs closely match the structure of known, naturally occurring TPR proteins. Recent efforts to produce consensus ankyrin repeat proteins have been similarly successful. Ankyrin repeats are 33-residue repeats containing two ␣ helices (slightly shorter than those of TPR repeats) followed by an extended loop-␤-turn structure and a different interrepeat packing geometry. Based on both consensus and chemical intuition, Mosavi et al. produced ankyrin polypeptides containing from one to four identical repeats [5] (Figure 1). As with the designed TPR protein, the threeand four-repeat versions of these proteins also showed high thermostability (Table 1), and the high-resolution crystal structures of these consensus ankyrin repeat proteins confirm the success of the design strategy. Using a slightly different approach, Kohl et al. combined consensus information with random surface substitutions (approximately 20% of the total sequence) to create a regular scaffold displaying a range of potential