Search

Control of transcription allows the regulation of cell activity in response to external stimuli and research in the field has greatly benefited from efforts in structural biology. In this review, based on specific examples from the European SPINE2-COMPLEXES initiative, we illustrate the impact of structural proteomics on our understanding of the molecular basis of gene expression. While most atomic structures were obtained by X-ray crystallography, the impact of solution NMR and cryo-electron microscopy is far from being negligible. Here, we summarize some highlights and illustrate the importance of specific technologies on the structural biology of protein-protein or protein/DNA transcription complexes: structure/function analysis of components the eukaryotic basal and activated transcription machinery with focus on the TFIID and TFIIH multi-subunit complexes as well as transcription regulators such as members of the nuclear hormone receptor families. We also discuss molecular aspects of promoter recognition and epigenetic control of gene expression., (Copyright © 2011 Elsevier Inc. All rights reserved.)

In nature, some proteins partially unfold under specific environmental conditions. These unfolded states typically consist of a large ensemble of conformations; their proper description is therefore a challenging problem. NMR spectroscopy is particularly well suited for this task: information on conformational preferences can be derived, for example, from chemical shifts or residual dipolar couplings. This information, which is measured as a time- and ensemble-average, can be used to model these states by generating large ensembles of conformations. The challenge is then to select a minimum representative set of conformations out of a large ensemble to represent the unfolded state. We have developed for this purpose an algorithm called MINOES (MINimum Optimal Ensemble Selection), which is based on an iterative procedure based on a driven expansion/contraction selection process. MINOES aims at selecting an optimal and minimal ensemble of conformations that, on average, maximizes the agreement between back-calculated and experimental (NMR) data, without any a-priori assumption about the required ensemble size. This approach is demonstrated by modeling the partially unfolded state of a deletion mutant of the Photoactive Yellow Protein, Delta25-PYP, which has been previously characterized by NMR (Bernard et al., Structure 2005;13:953-962).

The three-dimensional solution structure of the alpha-subunit in the alpha, beta heterodimeric human chorionic gonadotropin (hCG), deglycosylated with endo-beta-N-acetylglucosaminidase-B (dg-alpha hCG), was determined using 2D homonuclear and 2D heteronuclear 1H, 13C NMR spectroscopy at natural abundance in conjunction with the program package XPLOR. The distance geometry/simulated annealing protocol was modified to allow for the efficient modelling of the cystine knot motif present in alpha hCG. The protein structure was modelled with 620 interproton distance restraints and the GlcNAc residue linked to Asn78 was modelled with 30 protein-carbohydrate and 3 intraresidual NOEs. The solution structure of dg-alpha hCG is represented by an ensemble of 27 structures. In comparison to the crystal structure of the dimer, the solution structure of free dg-alpha hCG exhibits: (a) an increased structural disorder (residues 33-57); (b) a different backbone conformation near Val76 and Glu77; and (c) a larger flexibility. These differences are caused by the absence of the interactions with the beta-subunit. Consequently, in free dg-alpha hCG, compared to the intact dimer, the two hairpin loops 20-23 and 70-74 are arranged differently with respect to each other. The beta-GlcNAc(78) is tightly associated with the hydrophobic protein-core in between the beta-hairpins. This conclusion is based on the NOEs from the axial H1, H3, H5 atoms and the N-acetyl protons of beta-GlcNAc(78) to the protein-core. The hydrophobic protein-core between the beta-hairpins is thereby shielded from the solvent.

The solution structure of photoactive yellow protein (PYP), a photosensory protein from Ectothiorhodospira halophila, has been determined by multidimensional NMR spectroscopy. The structure consists of an open, twisted, 6-stranded, antiparallel beta-sheet, which is flanked by four alpha-helices on both sides. The final set of 26 selected structures is well-defined for the regions spanning residues Phe6-Ala16, Asp24-Ala112, and Tyr118-Val125 and displays a root-mean-square deviation, versus the average, of 0.45 A for the backbone and 0.88 A for all heavy atoms. Comparison of the solution structure with an earlier published 1.4 A crystal structure (Borgstahl, G. E. O., Williams, D. R., and Getzoff, E. D. (1995) Biochemistry 34, 6278-6287) reveals a similarity with a root-mean-square deviation of 1.77 A for the backbone for the well-defined regions. The most distinct difference in the backbone with the crystal structure is found near the N-terminus, for residues Asp19-Leu23, which corresponds to an alpha-helix in the crystal structure and to one of the poorest defined regions in the solution structure. To characterize the dynamic behavior of PYP in solution, we undertook a 15N relaxation study and measurements of hydrogen/deuterium exchange. Determination of order parameters through the model-free Lipari-Szabo approach enabled the identification of several regions of enhanced dynamics. The comparison of atomic displacements in the backbone traces of the ensemble structures, with mobility measurements from NMR, show that the poorly defined regions feature fast internal motions in the nanosecond to picosecond time scale.

A modification of the TROSY experiment which allows for sensitivity enhancement without introduction of additional delays is presented. The method relies on the combination of two orthogonal pathways and results in a signal-to-noise gain of 2 independent of the size of the molecule. The sensitivity enhanced TROSY scheme can be combined with gradient coherence selection. The method has been applied to the 269 residue serine protease PB92., (Copyright 1998 Academic Press.)

Light irradiation of photoactive yellow protein (PYP) induces a photocycle, in which red-shifted (pR) and blue-shifted (pB) intermediates have been characterized. An NMR study of the long-lived pB intermediate now reveals that it exhibits a large degree of disorder and exists as a family of multiple conformers that exchange on a millisecond time scale. This shows that the behavior of PYP in solution is different from what has been observed in the crystalline state. Furthermore, differential refolding to ground state pG is observed, whereby the central beta-sheet and parts of the helical structure are formed first and the region around the chromophore at a later stage.

Background: Integrase mediates a crucial step in the life cycle of the human immunodeficiency virus (HIV). The enzyme cleaves the viral DNA ends in a sequence-dependent manner and couples the newly generated hydroxyl groups to phosphates in the target DNA. Three domains have been identified in HIV integrase: an amino-terminal domain, a central catalytic core and a carboxy-terminal DNA-binding domain. The amino-terminal region is the only domain with unknown structure thus far. This domain, which is known to bind zinc, contains a HHCC motif that is conserved in retroviral integrases. Although the exact function of this domain is unknown, it is required for cleavage and integration., Results: The three-dimensional structure of the amino-terminal domain of HIV-2 integrase has been determined using two-dimensional and three-dimensional nuclear magnetic resonance data. We obtained 20 final structures, calculated using 693 nuclear Overhauser effects, which display a backbone root-mean square deviation versus the average of 0.25 A for the well defined region. The structure consists of three alpha helices and a helical turn. The zinc is coordinated with His 12 via the N epsilon 2 atom, with His16 via the N delta 1 atom and with the sulfur atoms of Cys40 and Cys43. The alpha helices form a three-helix bundle that is stabilized by this zinc-binding unit. The helical arrangement is similar to that found in the DNA-binding domains of the trp repressor, the prd paired domain and Tc3A transposase., Conclusion: The amino-terminal domain of HIV-2 integrase has a remarkable hybrid structure combining features of a three-helix bundle fold with a zinc-binding HHCC motif. This structure shows no similarity with any of the known zinc-finger structures. The strictly conserved residues of the HHCC motif of retroviral integrases are involved in metal coordination, whereas many other well conserved hydrophobic residues are part of the protein core.

The three-dimensional solution structure of the immunodominant central conserved region of the attachment protein G (BRSV-G) of bovine respiratory syncytial virus has been determined by nuclear magnetic resonance (NMR) spectroscopy. In the 32-residue peptide studied, 19 residues form a small rigid core composed of two short helices, connected by a type I' turn, and linked by two disulfide bridges. This unique fold is among the smallest stable tertiary structures known and could therefore serve as an ideal building block for the design of de novo proteins and as a test case for modeling studies. A characteristic hydrophobic pocket, lined by conserved residues, lies at the surface of the peptide and may play a role in receptor binding. This work provides a structural basis for further peptide vaccine development against the severe diseases associated with the respiratory syncytial viruses in both cattle and man.

Human chorionic gonadotropin (hCG) is a heterodimeric glycoprotein hormone that is involved in the maintenance of the corpus luteum in early pregnancy. Glycosylation at Asn52 of its alpha subunit (alpha hCG) is essential for signal transduction, whereas the N-glycan at Asn78 stabilizes the structure of the protein. In this study, an almost complete 1H-NMR and a partial 13C-NMR spectral assignment for the amino acids and the N-glycans of alpha hCG and of an enzymatically deglycosylated form, which had a single GlcNAc residue at each of its two glycosylation sites, has been achieved. The secondary structure of alpha hCG is solution, which was determined based on NOE data, is partially similar to that of the alpha subunit in the crystal structure of hCG, but large structural differences are found for amino acid residues 33-58. In the crystal structure of hCG, residues 33-37 and 54-58 of the alpha subunit are part of an intersubunit seven-stranded beta-barrel and residues 41-47 constitute a 3(10)-helix. In contrast, in free alpha hCG in solution, amino acids 33-58 are part of a large disordered loop, indicating that in intact hCG interactions with the beta subunit of hCG stabilize the conformation of the alpha subunit. The NMR data of alpha hCG and its deglycosylated counterpart are very similar, indicating that removal of carbohydrate residues other than GlcNAc-1 does not notably affect the conformation of the protein part. However, numerous 1H-NOEs between the GlcNAc-1 residue at Asn78 and several amino acid residues show that this GlcNAc residue is tightly packed against the protein, being an integral part of the structure of the alpha subunit. 1H-NOEs across the glycosidic linkages of the glycan, resonance-line widths, and 1H and 13C chemical shifts of the other monosaccharides suggest that the remainder of the glycans at Asn78, and the glycans at Asn52 are largely extended in solution.

The DNA-binding protein HU from Bacillus stearothermophilus (HUBst) is a dimer with a molecular weight of 195 kDa that is capable of bending DNA. An x-ray structure has been determined previously [Tanaka et al. 1984) Nature, vol. 310, pp. 376-381], but no structure could be established for a large part of the supposed DNA-binding beta-arms. Distance geometry and restrained molecular dynamics using nmr restraints were used to generate a set of 25 structures. These structures display a backbone rms deviation (RMSD) of 0.36 A for the well-defined region (residues 2-54 and 75-90). The structure of the core is very similar to that observed in the x-ray structure, with a pairwise RMSD of 1.06 A. The structure of the beta-hairpin arm contains a double flip-over at the prolines in the two strands of the beta-arm. Heteronuclear 15N relaxation measurements indicate that the beta-arm and the tip of the beta-arm is flexible. This explains the disorder observed in the solution and x-ray structures of the beta-arm with respect to the core of the protein. Overlayed onto itself the beta-arm is better defined, with a backbone RMSD of 1.0 A calculated for residues 54-59 and 69-74. The tip of the arm adopts a well-defined 4 : 6 beta-hairpin conformation. Changes in amide 15N and 1H chemical shifts upon titrating DNA are most pronounced for the residues in the beta-hairpin arm and for the residues in the second half of the third alpha-helix. Heteronuclear 15N relaxation data for free and complexed HUBst show that that the arms become structured upon DNA binding. Together with chemically induced nuclear polarization measurements on a mutant HUBst (M69Y; V76Y) this shows that the beta-hairpin arm is involved in direct DNA interaction.

The lipolytic enzyme phospholipase A2 (PLA2) is involved in the degradation of high-molecular weight phospholipid aggregates in vivo. The enzyme has very high catalytic activities on aggregated substrates compared with monomeric substrates, a phenomenon called interfacial activation. Crystal structures of PLA2s in the absence and presence of inhibitors are identical, from which it has been concluded that enzymatic conformational changes do not play a role in the mechanism of interfacial activation. The high-resolution NMR structure of porcine pancreatic PLA2 free in solution was determined with heteronuclear multidimensional NMR methodology using doubly labeled 13C, 15N-labeled protein. The solution structure of PLA2 shows important deviations from the crystal structure. In the NMR structure the Ala1 alpha-amino group is disordered and the hydrogen bonding network involving the N-terminus and the active site is incomplete. The disorder observed for the N-terminal region of PLA2 in the solution structure could be related to the low activity of the enzyme towards monomeric substrates. The NMR structure of PLA2 suggests, in contrast to the crystallographic work, that conformational changes do play a role in the interfacial activation of this enzyme.

The Arc repressor of Salmonella bacteriophage P22 is a dimeric sequence-specific DNA-binding protein. The solution structure of Arc has been determined from 2D NMR data using an "ensemble" iterative relaxation matrix approach (IRMA) followed by direct NOE refinement with DINOSAUR. A set of 51 structures was generated with distance geometry and further refined with a combination of restrained energy minimization and restrained molecular dynamics in a parallel refinement protocol. Distance constraints were obtained from an extensive set of NOE build-ups in H2O and 2H2O via relaxation matrix calculations from the ensemble of structures. Methyl group rotation, aromatic ring flaps and internal mobility effects (via order parameters obtained from a free molecular dynamics run in water) were included in these calculations. The best structures were finally refined with direct NOE constraints following a slow-cooling simulated annealing protocol. In this final refinement stage, theoretical NOE intensities were directly compared with the experimental data and forces were derived using a simple two-spin approximation for the gradient of the NOE function. Dynamic assignment was applied to the peaks involving unassigned diastereotopic groups. The structure is determined to a precision (r.m.s.d. from the average excluding the ill defined C and N-terminal region) of 0.55 and 1.10 A for backbone and all atoms, respectively. The final structures, with R factor values around 0.35, have good stereochemical qualities, contain an extensive network of hydrogen bonds consistent with the secondary structure elements and structural features in concordance with genetic data. The overall folding of the solution and crystal structures is the same.

A Monte Carlo method is described for automated docking of proteins on DNA. The simulation program MONTY keeps the entire DNA and the protein backbone and core fixed while protein surface side-chains are allowed to rotate freely. The entire protein is rotated and translated by small random steps in order to find the best fit with the DNA. New configurations are accepted on basis of their Boltzmann probability. Protein-DNA interaction is represented by square well potentials for hydrogen bond and van der Waals interactions. The structure with the largest interaction energy encountered during the simulation is saved. The method is tested on complexes of the 434 Cro protein and its operator DNA where the protein is shifted up or down one or two base-pairs and is subsequently allowed to find back its native binding site. This protocol is performed for shifted complexes derived from the crystal structure, shifted complexes where the crystal structure DNA is replaced by standard B-DNA and shifted complexes where in addition the protein is replaced by protein from the uncomplexed crystal structure. In all three cases the six lowest energy structures correspond to complexes close to the native complex. The quality of sequence specific recognition diminishes, however, when the molecular surface complementarity between protein and DNA decreases.

The recent development of three-dimensional NMR spectroscopy has alleviated the problem of overlap of resonances. However, also for the 3D experiments resonance assignment strategies have usually relied upon knowledge about spin systems, combined with information about short (sequential) distances. For doubly (15N/13C)-labelled molecules, a novel assignment strategy has been developed. In this paper we address the possibilities of an assignment strategy for proteins, based solely upon the use of NOE data. For this, the 3D NOE-NOE experiment seems most suitable. Therefore, we have made a theoretical evaluation of double NOE transfer pathways in 28 protein crystal structures. We identify 95 connectivities which are most likely to be observed as cross peaks in a 3D NOE-NOE spectrum of a protein. Given the occurrence of one of these 95 connectivities, we evaluate the chances of occurrence for the others. Analysis of these conditional probabilities allowed the construction of five patterns of related, highly correlated cross peaks which resemble the conventional idea of spin systems to some extent and may provide a basis for assignment and secondary structure analysis from 3D NOE-NOE data alone.

A set of computer programs called DINOSAUR has been developed, which allows the refinement of biomolecular structures directly from 2D NOE intensities. The NOE restraining potential implemented emphasises the weak intensities corresponding to larger distances which are more likely to determine the three-dimensional structure. An approximation based on a two-spin approximation is proposed for the gradient of the NOE intensities instead of the exact solution which is extremely time-consuming. The DINOSAUR routines have been implemented in various refinement programs (Distance bound Driven Dynamics, Molecular Dynamics and Energy Minimisation) and tested on an eight-residue model peptide.

A suite of computer programs (CLAIRE) is described which can be of assistance in the process of assigning 2D 1H NMR spectra of proteins. The programs embody a software implementation of the sequential assignment approach first developed by Wüthrich and co-workers (K. Wüthrich, G. Wider, G. Wagner and W. Braun (1982) J. Mol. Biol. 155, 311). After data-abstraction (peakpicking), the software can be used to detect patterns (spin systems), to find cross peaks between patterns in 2D NOE data sets and to generate assignments that are consistent with all available data and which satisfy a number of constraints imposed by the user. An interactive graphics program called CONPAT is used to control the entire assignment process as well as to provide the essential feedback from the experimental NMR spectra. The algorithms are described in detail and the approach is demonstrated on a set of spectra from the mistletoe protein phoratoxin B, a homolog of crambin. The results obtained compare well with those reported earlier based entirely on a manual assignment process.

The increase in dimensionality of three-dimensional (3D) NMR greatly enhances the spectral resolution in comparison to 2D NMR. It alleviates the problem of resonance overlap and may extend the range of molecules amenable to structure determination by high-resolution NRM spectroscopy. Here, we present strategies for the assignment of protein resonances from homonuclear nonselective 3D NOE-HOHAHA spectra. A notation for connectivities between protons, corresponding to cross peaks in 3D spectra, is introduced. We show how spin systems can be identified by tracing cross-peak patterns in cross sections perpendicular to the three frequency axes. The observable 3D sequential connectivities in proteins are tabulated, and estimates for the relative intensities of the corresponding cross peaks are given for alpha-helical and beta-sheet conformations. Intensities of the cross peaks in the 3D spectrum of pike III parvalbumin follow the predictions. The sequential-assignment procedure is illustrated for loop regions, extended and alpha-helical conformations for the residues Ala 54-Leu 63 of parvalbumin. NOEs that were not previously identified in 2D spectra of parvalbumin due to overlap are found.