Scientific Report 2006

Molecular Biology

Nuclear Magnetic Resonance of 3-Dimensional Structure and Dynamics of Proteins in Solution

P.E. Wright, H.J. Dyson, R. Burge, J. Ferreon, N. Greenman, T.-H. Huang, B.B. Koehntop, M. Kostic, B. Lee, C.W. Lee, M. Landes, M. Martinez-Yamout, T. Nishikawa, K. Sugase, J. Wojciak, M. Zeeb, E. Manlapaz, L.L. Tennant, J. Chung, D.A. Case, J. Gottesfeld, R. Evans,* M. Montminy*

* Salk Institute, La Jolla, California

We use multidimensional nuclear magnetic resonance (NMR) spectroscopy to investigate the structures, dynamics, and interactions of proteins in solution. Such studies are essential for understanding the mechanisms of action of these proteins and for elucidating structure-function relationships. The focus of our current research is protein-protein and protein–nucleic acid interactions involved in the regulation of gene expression.

Transcription Factor–Nucleic Acid Complexes

NMR methods are being used to determine the
3-dimensional structures and intramolecular dynamics of zinc finger motifs from several eukaryotic transcriptional regulatory proteins, both free and complexed with target nucleic acid. Zinc fingers are among the most abundant domains in eukaryotic genomes. They play a central role in the regulation of gene expression at both the transcriptional and the posttranscriptional level, mediated through their interactions with DNA, RNA, or protein components of the transcriptional machinery. The C₂H₂ zinc finger, first identified in transcription factor IIIA (TFIIIA), is used by numerous transcription factors to achieve sequence-specific recognition of DNA. There is growing evidence, however, that some C₂H₂ zinc finger proteins control gene expression both through their interactions with DNA regulatory elements and, at the posttranscriptional level, by binding to RNA.

The best-characterized example of a C₂H₂ zinc finger protein that binds specifically to both DNA and to RNA is TFIIIA, which contains 9 zinc fingers. We showed previously that different subsets of zinc fingers are responsible for high-affinity binding of TFIIIA to DNA (fingers 1–3) and to 5S RNA (fingers 4–6). To obtain insights into the mechanism by which the TFIIIA zinc fingers recognize both DNA and RNA, we have used NMR methods to determine the structures of the complex formed by zf1-3 (a protein containing fingers 1–3) with DNA and by zf4-6 (a protein consisting of fingers 4–6) with a fragment of 5S RNA.

Three-dimensional structures were determined previously for the complex of zf1-3 with the cognate 15-bp oligonucleotide duplex. The structures contain several novel features and reveal that prevailing models of DNA recognition, which assume that zinc fingers are independent modules that contact bases through a limited set of amino acids, are outmoded.

In addition to its role in binding to and regulating the 5S RNA gene, TFIIIA also forms a complex with the 5S RNA transcript. NMR structures of the complex formed by zinc fingers 4–6 with a truncated form of 5S RNA have been completed and give important insights into the structural basis for 5S RNA recognition. Finger 4 of the protein recognizes both the structure of the RNA backbone and the specific bases in the loop E motif of the RNA, in a classic lock-and-key interaction. Fingers 5 and 6, with a single residue between them, undergo mutual induced-fit folding with the loop A region of the RNA, which is highly flexible in the absence of the protein.

NMR studies of 2 alternate splice variants of the Wilms tumor zinc finger protein are in progress. These proteins differ only through insertion of 3 additional amino acids (the tripeptide lysine-threonine-serine) in the linker between fingers 3 and 4, yet have marked differences in their DNA-binding properties and subcellular localization. 15N relaxation measurements indicate that the insertion increases the flexibility of the linker between fingers 3 and 4 and abrogates binding of the fourth zinc finger to its cognate site in the DNA major groove, thereby modulating DNA-binding activity. The x-ray structure of the DNA complex has been determined, providing insights into the mechanism by which disease-causing mutations interfere with DNA binding. NMR studies of the RNA complex are in progress. We have also determined the structure of a novel zinc finger protein that binds to double-stranded RNA and have begun experiments to define the mechanism of RNA recognition.

Several novel zinc binding motifs have recently been identified that mediate gene expression at the posttranscriptional level by regulating mRNA processing and metabolism. Regulatory proteins of the TIS11 family bind specifically, through a pair of novel CCCH zinc fingers, to the adenosine-uridine–rich element in the 3′ untranslated region of short-lived cytokine, growth factor, and proto-oncogene mRNAs and control expression by promoting rapid degradation of the message. We recently determined the NMR structure of the complex formed between the tandem zinc finger domain of TIS11d and its binding site on the adenosine-uridine–rich element. This structure showed sequence-specific recognition of single-stranded RNA through formation of a network of hydrogen bonds between the polypeptide backbone and the Watson-Crick edges of the bases.

Protein-Protein Interactions in Transcriptional Regulation

Transcriptional regulation in eukaryotes relies on protein-protein interactions between DNA-bound factors and coactivators that, in turn, interact with the basal transcription machinery. The transcriptional coactivator CREB-binding protein (CBP) and its homolog p300 play an essential role in cell growth, differentiation, and development. Understanding the molecular mechanisms by which CBP and p300 recognize their various target proteins is of fundamental biomedical importance. CBP and p300 have been implicated in diseases such as leukemia, cancer, and mental retardation and are novel targets for therapeutic intervention.

We previously determined the structure of the kinase-inducible activation domain of the transcription factor CREB bound to its target domain (the KIX domain) in CBP. Ongoing work is directed toward mapping the interactions between KIX and the transcriptional activation domains of the proto-oncogene c-Myb and of the mixed-lineage leukemia protein. The solution structure of the ternary complex between KIX, c-Myb and the mixed-lineage leukemia protein has been completed and provides insights into the structural basis for the ability of the KIX domain to interact simultaneously and allosterically with 2 different effectors. Our work has also provided new understanding of the thermodynamics of the coupled folding and binding processes involved in interaction of KIX with transcriptional activation domains. We are using R₂ relaxation dispersion experiments to elucidate the mechanism by which folding of the kinase-inducible activation domain of CREB is coupled to binding to its KIX target domain. These experiments reveal formation of a transient and largely unfolded encounter complex, which then folds on the surface of the KIX domain to form the helical structure observed in the fully bound state.

Recently, we determined the structure of the complex between the hypoxia-inducible factor Hif-1α and the CH1 domain of CBP. The interaction between Hif-1α and CBP/p300 is of major therapeutic interest because of the central role Hif-1α plays in tumor progression and metastasis; disruption of this interaction leads to attenuation of tumor growth. A protein named CITED2 functions as a negative feedback regulator of the hypoxic response by competing with Hif-1α for binding to the CH1 domain of CBP. We determined the structure of the complex formed between CITED2 and the CH1 domain and were able to show that the CH1 domain is folded into a stable 3-dimensional structure even in the absence of binding partners. The intrinsically unstructured Hif-1α and CITED2 domains use partly overlapping surfaces of the CH1 motif to achieve high-affinity binding and compete effectively with each other for CBP/p300. We are continuing to map the multiplicity of interactions between CBP/p300 domains and their numerous biological targets to understand the complex interplay of interactions that mediate key biological processes in health and disease.

Publications

De Guzman, R.N., Goto, N.K., Dyson, H.J., Wright, P.E. Structural basis for cooperative transcription factor binding to the CBP coactivator. J. Mol. Biol. 355:1005, 2006.

Kostic, M., Matt, T., Martinez-Yamout, M.A., Dyson, H.J., Wright, P.E. Solution structure of the Hdm2 C2H2C4 RING, a domain critical for ubiquitination of p53. J. Mol. Biol. 363:433, 2006.

Lee, B.M., Xu, J., Clarkson, B.K., Martinez-Yamout, M.A., Dyson, H.J., Case, D.A., Gottesfeld, J.M., Wright, P.E. Induced fit and “lock and key” recognition of 5S RNA by zinc fingers of transcription factor IIIA. J. Mol. Biol. 357:275, 2006.

Folding of Proteins and Protein Fragments

P.E. Wright, H.J. Dyson, C. Nishimura, D. Felitsky, Y. Yao, J. Chung, L.L. Tennant, V. Bychkova,* T. Uzawa,** S. Takahashi**

* Institute of Protein Research, Puschino, Russia
** Kyoto University, Kyoto, Japan

The molecular mechanism by which proteins fold into their 3-dimensional structures remains one of the most important unsolved problems in structural biology. Nuclear magnetic resonance (NMR) spectroscopy is uniquely suited to provide information on the structure of transient intermediates formed during protein folding. Previously, we used NMR methods to show that many peptide fragments of proteins have a tendency to adopt folded conformations in water solution. The presence of transiently populated folded structures, including reverse turns, helices, nascent helices, and hydrophobic clusters, in water solutions of short peptides has important implications for initiation of protein folding. Formation of elements of secondary structure probably plays an important role in the initiation of protein folding by reducing the number of conformations that must be explored by the polypeptide chain and by directing subsequent folding pathways.

Apomyoglobin Folding Pathway

A major program in our laboratory is directed toward a structural and mechanistic description of the apomyoglobin folding pathway. Previously, we used quenched-flow pulse-labeling methods in conjunction with 2-dimensional NMR spectroscopy to map the kinetic folding pathway of the wild-type protein. With these methods, we showed that an intermediate in which the A, G, and H helices and part of the B helix adopt hydrogen-bonded secondary structure is formed within 6 milliseconds of the initiation of refolding. Folding then proceeds by stabilization of additional structure in the B helix and in the C and E helices. We are using carefully selected myoglobin mutants and both optical stopped-flow spectroscopy and NMR methods to further probe the kinetic folding pathway. For some of the mutants studied, the changes in amino acid sequence resulted in changes in the folding pathway of the protein. These experiments are providing novel insights into both the local and the long-range interactions that stabilize the kinetic folding intermediate. Of particular importance, long-range interactions have been observed that indicate nativelike packing of some of the helices in the kinetic molten globule intermediate.

Apomyoglobin provides a unique opportunity for detailed characterization of the structure and dynamics of a protein-folding intermediate. Conditions were previously identified under which the apomyoglobin molten globule intermediate is sufficiently stable for acquisition of multidimensional heteronuclear NMR spectra. Analysis of ¹³C and other chemical shifts and measurements of polypeptide dynamics provided unprecedented insights into the structure of this state.

The A, G, and H helices and part of the B helix are folded and form the core of the molten globule. This core is stabilized by relatively nonspecific hydrophobic interactions that restrict the motions of the polypeptide chain. Fluctuating helical structure is formed in regions outside the core, although the population of helix is low and the chain retains considerable flexibility. The F helix acts as a gate for heme binding and only adopts stable structure in the fully folded holoprotein.

The acid-denatured (unfolded) state of apomyoglobin is an excellent model for the fluctuating local interactions that lead to the transient formation of unstable elements of secondary structure and local hydrophobic clusters during the earliest stages of folding. NMR data indicated substantial formation of helical secondary structure in the acid-denatured state in regions that form the A and H helices in the folded protein and also revealed nonnative structure in the D and E helix regions.

Because the A and H regions adopt stabilized helical structure in the earliest detectable folding intermediate, these results lend strong support to folding models in which spontaneous formation of local elements of secondary structure plays a role in initiating formation of the A-[B]-G-H molten globule folding intermediate. In addition to formation of transient helical structure, formation of local hydrophobic clusters has been detected by using ¹⁵N relaxation measurements. Significantly, these clusters are formed in regions where the average surface area buried upon folding is large. In contrast to acid-denatured unfolded apomyoglobin, the urea-denatured state is largely devoid of structure, although residual hydrophobic interactions have been detected by using relaxation measurements.

We measured residual dipolar couplings for unfolded states of apomyoglobin by using partial alignment in strained polyacrylamide gels. These data provide novel insights into the structure and dynamics of the unfolded polypeptide chain. We have shown that the residual dipolar couplings arise from the well-known statistical properties of flexible polypeptide chains. Residual dipolar couplings provide valuable insights into the dynamic and conformational propensities of unfolded and partly folded states of proteins and hold great promise for charting the upper reaches of protein-folding landscapes.

To probe long-range interactions in unfolded and partially folded states of apomyoglobin, we introduced spin-label probes at several sites throughout the polypeptide chain. These experiments led to the surprising discovery that structures with nativelike topology exist within the ensemble of conformations formed by the acid-denatured state of apomyoglobin. They also indicated that the packing of helices in the molten globule state is similar to that in the native folded protein.

The view of protein folding that results from our work on apomyoglobin is one in which collapse of the polypeptide chain to form increasingly compact states leads to progressive accumulation of secondary structure and increasing restriction of fluctuations in the polypeptide backbone. Chain flexibility is greatest at the earliest stages of folding, in which transient elements of secondary structure and local hydrophobic clusters are formed. As the folding protein becomes increasingly compact, backbone motions become more restricted, the hydrophobic core is formed and extended, and nascent elements of secondary structure are progressively stabilized. The ordered tertiary structure characteristic of the native protein, with well-packed side chains and relatively low-amplitude local dynamics, appears to form rather late in folding.

We recently introduced a variation on the classic quench-flow technique, which makes use of the capabilities of modern NMR spectrometers and heteronuclear NMR experiments, to study the proteins labeled along the folding pathway in an unfolded state in an aprotic organic solvent. This method allows detection of many more amide proton probes than in the classic method, which required formation of the fully folded protein and the measurement of its NMR spectrum in water solutions. This method is particularly useful in documenting changes in the folding pathway that result in the destabilization of parts of the protein in the molten globule intermediate. We recently showed that self-compensating mutations designed to change the amino acid sequence such that the average area buried upon folding is significantly changed while the 3-dimensional structure of the final folded state remains the same. These studies showed that the average area buried upon folding is an accurate predictor of those parts of the apomyoglobin molecule that will fold first and participate in the molten globule intermediate. Quench-flow hydrogen exchange experiments performed on a series of hydrophobic core mutants indicated that the overall helix-packing topology of the kinetic folding intermediate is like that of the native protein, despite local nonnative interactions in packing of the G and H helices. Finally, using a rapid mixing device, we have reduced the dead time of the kinetic refolding experiments and have shown that a compact helical intermediate is formed within 400 microseconds after initiation of apomyoglobin refolding.

Folding-Unfolding Transitions in Cellular Metabolism

Many species of bacteria sense and respond to their own population density by an intricate autoregulatory mechanism known as quorum sensing; the bacteria release extracellular signal molecules, called autoinducers, for cell-cell communication within and between bacterial species. A number of bacteria appear to use quorum sensing for regulation of gene expression in response to fluctuations in cell population density. Processes regulated in this way include symbiosis, virulence, competence, conjugation, production of antibiotics, motility, sporulation, and formation of biofilms.

We determined the 3-dimensional solution structure of a complex composed of the N-terminal 171 residues of the quorum-sensing protein SdiA of Escherichia coli and an autoinducer molecule, N-octanoyl-1-homoserine lactone (HSL) (Fig. 1). The SdiA-HSL system shows the “folding switch” behavior associated with quorum-sensing factors produced by other bacterial species. In the presence of HSL, the SdiA protein is stable and folded and can be produced in good yields from an E coli expression system. In the absence of the autoinducer, the protein is expressed into inclusion bodies. Samples of the SdiA-HSL complex can be denatured but cannot be refolded in aqueous buffers. The solution structure of the complex provides a likely explanation for this behavior. The autoinducer molecule is tightly bound in a deep pocket in the hydrophobic core and is bounded by specific hydrogen bonds to the side chains of conserved residues. The autoinducer thus forms an integral part of the hydrophobic core of the folded SdiA.

Fig. 1. Folding of protein and protein fragments. Ribbon diagram showing the lowest energy structure of the complex between HSL and E coli SdiA.

Chaperone–Cochaperone–Client Protein Interactions

Understanding the role of unfolded states in cellular processes will require an understanding of the structural basis of their interactions, but unfolded proteins are impossible to characterize structurally by x-ray crystallography, and spectroscopic methods of all kinds are limited. It is necessary to explore unfolded proteins under conditions that approximate their physiologic milieu: in solution, at physiologic pHs and salt concentrations, and in the presence of specific cofactors. Structural insights will be obtained not only from the delineation of 3-dimensional structures but also from the description of conformational ensembles and of the motions of polypeptide chains under various conditions.

To gain new insights into the structural basis for the ability of unfolded and partly folded proteins to function in living systems, we are studying the interactions of “client” proteins and cochaperones with a well-known eukaryotic chaperone, Hsp90. Some of the protein components are much larger than have traditionally been studied by using solution NMR. However, we have designed a set of experiments that will allow us to draw valid conclusions about the extent and role of disorder in Hsp90 interactions. In particular, we will apply techniques recently developed in our laboratory for the analysis of hydrogen-deuterium exchange from unstable partially folded proteins by trapping the 2H-labeled species in the aprotic solvent dimethyl sulfoxide. This powerful new technique will be used to probe the structure, stability, and interactions of client proteins and cochaperones with Hsp90.

Publications

Dyson, H.J., Wright, P.E. According to current textbooks, a well-defined three-dimensional structure is a prerequisite for the function of the protein: is this correct? IUBMB Life 58:107, 2006.

Dyson, H.J., Wright, P.E., Scheraga, H.A. The role of hydrophobic interactions in initiation and propagation of protein folding. Proc. Natl. Acad. Sci. U. S. A. 103:13057, 2006.

Kamikubo, Y., Kroon, G., Curriden, S.A., Dyson, H.J., Loskutoff, D.J. The reduced, denatured somatomedin B domain of vitronectin refolds into a stable, biologically active form. Biochemistry 45:3297, 2006.

Martinez-Yamout, M.A., Venkitakrishnan, R.P., Preece, N.E., Kroon, G., Wright, P.E., Dyson, H.J. Localization of sites of interaction between p23 and Hsp90 in solution. J. Biol. Chem. 281:14457, 2006.

Nishimura, C., Dyson, H.J., Wright, P.E. Identification of native and non-native structure in kinetic folding intermediates of apomyoglobin. J. Mol. Biol. 355:139, 2006.

Papadopoulos, E., Oglecka, K., Mäler, L., Jarvet, J., Wright, P.E., Dyson, H.J., Gräslund, A. NMR solution structure of the peptide fragment 1-30, derived from mouse Doppel protein, in DHPC micelles. Biochemistry 45:159, 2006.

Yao, Y., Martinez-Yamout, M.A., Dickerson, T.J., Brogan, A.P., Wright, P.E., Dyson, H.J. Structure of the Escherichia coli quorum sensing protein SdiA: activation of the folding switch by acyl homoserine lactones. J. Mol. Biol. 355:262, 2006.

Yao, Y., Martinez-Yamout, M.A., Dyson, H.J. Backbone and side chain ¹H, ¹³C and ¹⁵N assignments for Escherichia coli SdiA1-171, the autoinducer-binding domain of a quorum sensing protein [letter]. J. Biomol. NMR 31:373, 2005.