From structure to the dynamic regulation of a molecular switch: A journey over 3 decades.

It is difficult to imagine where the signaling community would be today without the Protein Data Bank. This visionary resource, established in the 1970s, has been an essential partner for sharing information between academics and industry for over 3 decades. We describe here the history of our journey with the protein kinases using cAMP-dependent protein kinase as a prototype. We summarize what we have learned since the first structure, published in 1991, why our journey is still ongoing, and why it has been essential to share our structural information. For regulation of kinase activity, we focus on the cAMP-binding protein kinase regulatory subunits. By exploring full-length macromolecular complexes, we discovered not only allostery but also an essential motif originally attributed to crystal packing. Massive genomic data on disease mutations allows us to now revisit crystal packing as a treasure chest of possible protein:protein interfaces where the biological significance and disease relevance can be validated. It provides a new window into exploring dynamic intrinsically disordered regions that previously were deleted, ignored, or attributed to crystal packing. Merging of crystallography with cryo-electron microscopy, cryo-electron tomography, NMR, and millisecond molecular dynamics simulations is opening a new world for the signaling community where those structure coordinates, deposited in the Protein Data Bank, are just a starting point!

Vaccinia Virus Immunomodulator A46: A Lipid and Protein-Binding Scaffold for Sequestering Host TIR-Domain Proteins.

Vaccinia virus interferes with early events of the activation pathway of the transcriptional factor NF-kB by binding to numerous host TIR-domain containing adaptor proteins. We have previously determined the X-ray structure of the A46 C-terminal domain; however, the structure and function of the A46 N-terminal domain and its relationship to the C-terminal domain have remained unclear. Here, we biophysically characterize residues 1-83 of the N-terminal domain of A46 and present the X-ray structure at 1.55 Å. Crystallographic phases were obtained by a recently developed ab initio method entitled ARCIMBOLDO_BORGES that employs tertiary structure libraries extracted from the Protein Data Bank; data analysis revealed an all ß-sheet structure. This is the first such structure solved by this method which should be applicable to any protein composed entirely of ß-sheets. The A46(1-83) structure itself is a ß-sandwich containing a co-purified molecule of myristic acid inside a hydrophobic pocket and represents a previously unknown lipid-binding fold. Mass spectrometry analysis confirmed the presence of long-chain fatty acids in both N-terminal and full-length A46; mutation of the hydrophobic pocket reduced the lipid content. Using a combination of high resolution X-ray structures of the N- and C-terminal domains and SAXS analysis of full-length protein A46(1-240), we present here a structural model of A46 in a tetrameric assembly. Integrating affinity measurements and structural data, we propose how A46 simultaneously interferes with several TIR-domain containing proteins to inhibit NF-κB activation and postulate that A46 employs a bipartite binding arrangement to sequester the host immune adaptors TRAM and MyD88.

Combining H/D exchange mass spectroscopy and computational docking reveals extended DNA-binding surface on uracil-DNA glycosylase.

X-ray crystallography provides excellent structural data on protein-DNA interfaces, but crystallographic complexes typically contain only small fragments of large DNA molecules. We present a new approach that can use longer DNA substrates and reveal new protein-DNA interactions even in extensively studied systems. Our approach combines rigid-body computational docking with hydrogen/deuterium exchange mass spectrometry (DXMS). DXMS identifies solvent-exposed protein surfaces; docking is used to create a 3-dimensional model of the protein-DNA interaction. We investigated the enzyme uracil-DNA glycosylase (UNG), which detects and cleaves uracil from DNA. UNG was incubated with a 30 bp DNA fragment containing a single uracil, giving the complex with the abasic DNA product. Compared with free UNG, the UNG-DNA complex showed increased solvent protection at the UNG active site and at two regions outside the active site: residues 210-220 and 251-264. Computational docking also identified these two DNA-binding surfaces, but neither shows DNA contact in UNG-DNA crystallographic structures. Our results can be explained by separation of the two DNA strands on one side of the active site. These non-sequence-specific DNA-binding surfaces may aid local uracil search, contribute to binding the abasic DNA product and help present the DNA product to APE-1, the next enzyme on the DNA-repair pathway.

Predicting protein-DNA interactions by full search computational docking.

Protein-DNA interactions are essential for many biological processes. X-ray crystallography can provide high-resolution structures, but protein-DNA complexes are difficult to crystallize and typically contain only small DNA fragments. Thus, there is a need for computational methods that can provide useful predictions to give insights into mechanisms and guide the design of new experiments. We used the program DOT, which performs an exhaustive, rigid-body search between two macromolecules, to investigate four diverse protein-DNA interactions. Here, we compare our computational results with subsequent experimental data on related systems. In all cases, the experimental data strongly supported our structural hypotheses from the docking calculations: a mechanism for weak, nonsequence-specific DNA binding by a transcription factor, a large DNA-binding footprint on the surface of the DNA-repair enzyme uracil-DNA glycosylase (UNG), viral and host DNA-binding sites on the catalytic domain of HIV integrase, and a three-DNA-contact model of the linker histone bound to the nucleosome. In the case of UNG, the experimental design was based on the DNA-binding surface found by docking, rather than the much smaller surface observed in the crystallographic structure. These comparisons demonstrate that the DOT electrostatic energy gives a good representation of the distinctive electrostatic properties of DNA and DNA-binding proteins. The large, favourably ranked clusters resulting from the dockings identify active sites, map out large DNA-binding sites, and reveal multiple DNA contacts with a protein. Thus, computational docking can not only help to identify protein-DNA interactions in the absence of a crystal structure, but also expand structural understanding beyond known crystallographic structures.

Functional and structural characterization of a thermostable acetyl esterase from Thermotoga maritima.

TM0077 from Thermotoga maritima is a member of the carbohydrate esterase family 7 and is active on a variety of acetylated compounds, including cephalosporin C. TM0077 esterase activity is confined to short-chain acyl esters (C2-C3), and is optimal around 100°C and pH 7.5. The positional specificity of TM0077 was investigated using 4-nitrophenyl-ß-D-xylopyranoside monoacetates as substrates in a ß-xylosidase-coupled assay. TM0077 hydrolyzes acetate at positions 2, 3, and 4 with equal efficiency. No activity was detected on xylan or acetylated xylan, which implies that TM0077 is an acetyl esterase and not an acetyl xylan esterase as currently annotated. Selenomethionine-substituted and native structures of TM0077 were determined at 2.1 and 2.5 Å resolution, respectively, revealing a classic α/ß-hydrolase fold. TM0077 assembles into a doughnut-shaped hexamer with small tunnels on either side leading to an inner cavity, which contains the six catalytic centers. Structures of TM0077 with covalently bound phenylmethylsulfonyl fluoride and paraoxon were determined to 2.4 and 2.1 Å, respectively, and confirmed that both inhibitors bind covalently to the catalytic serine (Ser188). Upon binding of inhibitor, the catalytic serine adopts an altered conformation, as observed in other esterase and lipases, and supports a previously proposed catalytic mechanism in which Ser hydroxyl rotation prevents reversal of the reaction and allows access of a water molecule for completion of the reaction.

A helix scaffold for the assembly of active protein kinases.

Structures of set of serine-threonine and tyrosine kinases were investigated by the recently developed bioinformatics tool Local Spatial Patterns (LSP) alignment. We report a set of conserved motifs comprised mostly of hydrophobic residues. These residues are scattered throughout the protein sequence and thus were not previously detected by traditional methods. These motifs traverse the conserved protein kinase core and play integrating and regulatory roles. They are anchored to the F-helix, which acts as an organizing "hub" providing precise positioning of the key catalytic and regulatory elements. Consideration of these discovered structures helps to explain previously inexplicable results.

Conserved spatial patterns across the protein kinase family.

Protein kinases are a large family of enzymes heavily involved in signal transduction, regulation of metabolism, and control of cell growth and differentiation. These functions require precise recognition of widely diverse signals and substrates, and very detailed control of protein kinase activity. Large molecules interact primarily through recognition of surface features. Comparison of surfaces is complicated by both sequence diversity and conformational variability, including multiple possible rotameric states of side chains. We used a recently developed method of protein surface comparison to compare different serine/threonine and tyrosine kinases. As we have shown, two hydrophobic cores inside a protein kinase molecule are connected by a unique formation, called the "spine". It exists only in the active conformation of protein kinases and is dynamically disassembled during the inactivation process. Detection of such structures by any other method was not possible as the residues which comprise the spine do not form any sequence or 3D motifs in a traditional sense.

A generalized allosteric mechanism for cis-regulated cyclic nucleotide binding domains.

Cyclic nucleotides (cAMP and cGMP) regulate multiple intracellular processes and are thus of a great general interest for molecular and structural biologists. To study the allosteric mechanism of different cyclic nucleotide binding (CNB) domains, we compared cAMP-bound and cAMP-free structures (PKA, Epac, and two ionic channels) using a new bioinformatics method: local spatial pattern alignment. Our analysis highlights four major conserved structural motifs: 1) the phosphate binding cassette (PBC), which binds the cAMP ribose-phosphate, 2) the "hinge," a flexible helix, which contacts the PBC, 3) the beta(2,3) loop, which provides precise positioning of an invariant arginine from the PBC, and 4) a conserved structural element consisting of an N-terminal helix, an eight residue loop and the A-helix (N3A-motif). The PBC and the hinge were included in the previously reported allosteric model, whereas the definition of the beta(2,3) loop and the N3A-motif as conserved elements is novel. The N3A-motif is found in all cis-regulated CNB domains, and we present a model for an allosteric mechanism in these domains. Catabolite gene activator protein (CAP) represents a trans-regulated CNB domain family: it does not contain the N3A-motif, and its long range allosteric interactions are substantially different from the cis-regulated CNB domains.

Automated extraction of backbone deuteration levels from amide H/2H mass spectrometry experiments.

A Fourier deconvolution method has been developed to explicitly determine the amount of backbone amide deuterium incorporated into protein regions or segments by hydrogen/deuterium (H/D) exchange with high-resolution mass spectrometry. Determination and analysis of the level and number of backbone amide exchanging in solution provide more information about the solvent accessibility of the protein than do previous centroid methods, which only calculate the average deuterons exchanged. After exchange, a protein is digested into peptides as a way of determining the exchange within a local area of the protein. The mass of a peptide upon deuteration is a sum of the natural isotope abundance, fast exchanging side-chain hydrogens (present in MALDI-TOF H/2H data) and backbone amide exchange. Removal of the components of the isotopic distribution due to the natural isotope abundances and the fast exchanging side-chains allows for a precise quantification of the levels of backbone amide exchange, as is shown by an example from protein kinase A. The deconvoluted results are affected by overlapping peptides or inconsistent mass envelopes, and evaluation procedures for these cases are discussed. Finally, a method for determining the back exchange corrected populations is presented, and its effect on the data is discussed under various circumstances.

Identification and characterization of EX1 kinetics in H/D exchange mass spectrometry by peak width analysis.

Proteins that undergo cooperative unfolding events display EX1 kinetic signatures in hydrogen exchange mass spectra. The hallmark bimodal isotope pattern observed for EX1 kinetics is distinct from the binomial isotope pattern for uncorrelated exchange (EX2), the normal exchange regime for folded proteins. Detection and characterization of EX1 kinetics is simple when the cooperative unit is large enough that the isotopic envelopes in the bimodal pattern are resolved in the m/z scale but become complicated in cases where the unit is small or there is a mixture of EX1 and EX2 kinetics. Here we describe a data interpretation method involving peak width analysis that makes characterization of EX1 kinetics simple and rapid. The theoretical basis for EX1 and EX2 isotopic signatures and the effects each have on peak width are described. Modeling of EX2 widening and analysis of empirical data for proteins and peptides containing purely EX2 kinetics showed that the amount of widening attributable to stochastic forward- and back exchange in a typical experiment is small and can be quantified. Proteins and peptides with both obvious and less obvious EX1 kinetics were analyzed with the peak width method. Such analyses provide the half-life for the cooperative unfolding event and the relative number of residues involved. Automated analysis of peak width was performed with custom Excel macros and the DEX software package. Peak width analysis is robust, capable of automation, and provides quick interpretation of the key information contained in EX1 kinetic events.

Crystal structure of the E230Q mutant of cAMP-dependent protein kinase reveals an unexpected apoenzyme conformation and an extended N-terminal A helix.

Glu230, one of the acidic residues that cluster around the active site of the catalytic subunit of cAMP-dependent protein kinase, plays an important role in substrate recognition. Specifically, its side chain forms a direct salt-bridge interaction with the substrate's P-2 Arg. Previous studies showed that mutation of Glu230 to Gln (E230Q) caused significant decreases not only in substrate binding but also in the rate of phosphoryl transfer. To better understand the importance of Glu230 for structure and function, we solved the crystal structure of the E230Q mutant at 2.8 A resolution. Surprisingly, the mutant preferred an open conformation with no bound ligands observed, even though the crystals were grown in the presence of MgATP and the inhibitor peptide, IP20. This is in contrast to the wild-type protein that, under the same conditions, prefers the closed conformation of a ternary complex. The structure highlights the importance of the electrostatic surface not only for substrate binding and catalysis, but also for the mechanism for closing the active site cleft. This surface mutation clearly disrupts the recognition and binding of substrate peptide so that the enzyme prefers an open conformation that cannot trap ATP. This is consistent with the reinforcing concepts of conformational dynamics and the synergistic binding of ATP and substrate peptide. Another unusual feature of the structure is the observation of the entire N terminus (Gly1-Thr32) assumes an extended alpha-helix conformation. Finally, based on temperature factors, this mutant structure is more stable than the wild-type C-subunit in the apo state.

Crystal structure of a cAMP-dependent protein kinase mutant at 1.26A: new insights into the catalytic mechanism.

The catalytic subunit of cAMP-dependent protein kinase has served as a paradigm for the entire kinase family. In the course of studying the structure-function relationship of the P+1 loop (Leu198-Leu205) of the kinase, we have solved the crystal structure of the Tyr204 to Ala mutant in complexes with Mg.ATP and an inhibitory peptide at 1.26A, with overall structure very similar to that of the wild-type protein. However, at the nucleotide binding site, ATP was found largely hydrolyzed, with the products ADP-PO(4) retained in the structure. High-resolution refinement suggests that 26% of the molecules contain the intact ATP, whereas 74% have the hydrolyzed products. The observation of the substrate and product states in the same structure adds significant information to our understanding of the phosphoryl transfer process. Structural examination of the mutation site substantiates and extends the emerging concept that the hydrophobic core in the large lobe of the kinase might serve as a stable platform for anchoring key segments involved in catalysis. We propose that Tyr204 is critical for anchoring the P+1 loop to the core. Further analysis has highlighted two major connections between the P+1 loop and the catalytic loop (Arg165-Asn171). One emphasizes the hydrophobic packing of Tyr204 and Leu167 mediated through residues from the alphaF-helix, recently recognized as a signal integration motif, which together with the alphaE-helix forms the center of the hydrophobic core network. The other connection is mediated by the hydrogen bond interaction between Thr201 and Asp166, in a substrate-dependent manner. We speculate that the latter interaction may be important for the kinase to sense the presence of substrate and prepare itself for the catalytic reaction. Thus, the P+1 loop is not merely involved in substrate binding; it mediates the communication between substrate and catalytic residues.

Dynamic features of cAMP-dependent protein kinase revealed by apoenzyme crystal structure.

To better understand the mechanism of ligand binding and ligand-induced conformational change, the crystal structure of apoenzyme catalytic (C) subunit of adenosine-3',5'-cyclic monophosphate (cAMP)-dependent protein kinase (PKA) was solved. The apoenzyme structure (Apo) provides a snapshot of the enzyme in the first step of the catalytic cycle, and in this unliganded form the PKA C subunit adopts an open conformation. A hydrophobic junction is formed by residues from the small and large lobes that come into close contact. This "greasy" patch may lubricate the shearing motion associated with domain rotation, and the opening and closing of the active-site cleft. Although Apo appears to be quite dynamic, many important residues for MgATP binding and phosphoryl transfer in the active site are preformed. Residues around the adenine ring of ATP and residues involved in phosphoryl transfer from the large lobe are mostly preformed, whereas residues involved in ribose binding and in the Gly-rich loop are not. Prior to ligand binding, Lys72 and the C-terminal tail, two important ATP-binding elements are also disordered. The surface created in the active site is contoured to bind ATP, but not GTP, and appears to be held in place by a stable hydrophobic core, which includes helices C, E, and F, and beta strand 6. This core seems to provide a network for communicating from the active site, where nucleotide binds, to the peripheral peptide-binding F-to-G helix loop, exemplified by Phe239. Two potential lines of communication are the D helix and the F helix. The conserved Trp222-Phe238 network, which lies adjacent to the F-to-G helix loop, suggests that this network would exist in other protein kinases and may be a conserved means of communicating ATP binding from the active site to the distal peptide-binding ledge.

Finding needles in haystacks: Reranking DOT results by using shape complementarity, cluster analysis, and biological information.

We present an evaluation of our results for the first Critical Assessment of PRedicted Interaction (CAPRI). The methods used include the molecular docking program DOT, shape analysis tool FADE, cluster analysis and filtering based on biological data. Good results were obtained for most of the seven CAPRI targets, and for two systems, submissions having the highest number of correctly predicted contacts were produced.

Protein structural alignment for detection of maximally conserved regions.

An algorithm for comparison of homologous protein structures and for study of conformational changes in proteins, has been developed. The method is based on identification of pieces of the two molecules that have similar shapes, as determined by the local conformation of the polypeptide chain. Pieces that superpose within a specified tolerance are assembled into domains based on similar transformations for superposition. The result is sets of pieces that represent conserved structural elements and conserved spatial relationships between structural elements within the proteins being compared. A similarity criterion based on maximum distance rather than on root mean square deviation reduces bias by outliers. The utility of the method is demonstrated by using examples from the protein kinase family.

Computational reconstruction of multidomain proteins using atomic force microscopy data.

Classical structural biology techniques face a great challenge to determine the structure at the atomic level of large and flexible macromolecules. We present a novel methodology that combines high-resolution AFM topographic images with atomic coordinates of proteins to assemble very large macromolecules or particles. Our method uses a two-step protocol: atomic coordinates of individual domains are docked beneath the molecular surface of the large macromolecule, and then each domain is assembled using a combinatorial search. The protocol was validated on three test cases: a simulated system of antibody structures; and two experimentally based test cases: Tobacco mosaic virus, a rod-shaped virus; and Aquaporin Z, a bacterial membrane protein. We have shown that AFM-intermediate resolution topography and partial surface data are useful constraints for building macromolecular assemblies. The protocol is applicable to multicomponent structures connected in the polypeptide chain or as disjoint molecules. The approach effectively increases the resolution of AFM beyond topographical information down to atomic-detail structures.

Comparative surface geometry of the protein kinase family.

Identifying conserved pockets on the surfaces of a family of proteins can provide insight into conserved geometric features and sites of protein-protein interaction. Here we describe mapping and comparison of the surfaces of aligned crystallographic structures, using the protein kinase family as a model. Pockets are rapidly computed using two computer programs, FADE and Crevasse. FADE uses gradients of atomic density to locate grooves and pockets on the molecular surface. Crevasse, a new piece of software, splits the FADE output into distinct pockets. The computation was run on 10 kinase catalytic cores aligned on the alphaF-helix, and the resulting pockets spatially clustered. The active site cleft appears as a large, contiguous site that can be subdivided into nucleotide and substrate docking sites. Substrate specificity determinants in the active site cleft between serine/threonine and tyrosine kinases are visible and distinct. The active site clefts cluster tightly, showing a conserved spatial relationship between the active site and alphaF-helix in the C-lobe. When the alphaC-helix is examined, there are multiple mechanisms for anchoring the helix using spatially conserved docking sites. A novel site at the top of the N-lobe is present in all the kinases, and there is a large conserved pocket over the hinge and the alphaC-beta4 loop. Other pockets on the kinase core are strongly conserved but have not yet been mapped to a protein-protein interaction. Sites identified by this algorithm have revealed structural and spatially conserved features of the kinase family and potential conserved intermolecular and intramolecular binding sites.

R-subunit isoform specificity in protein kinase A: distinct features of protein interfaces in PKA types I and II by amide H/2H exchange mass spectrometry.

The two isoforms (RI and RII) of the regulatory (R) subunit of cAMP-dependent protein kinase or protein kinase A (PKA) are similar in sequence yet have different biochemical properties and physiological functions. To further understand the molecular basis for R-isoform-specificity, the interactions of the RIIbeta isoform with the PKA catalytic (C) subunit were analyzed by amide H/(2)H exchange mass spectrometry to compare solvent accessibility of RIIbeta and the C subunit in their free and complexed states. Direct mapping of the RIIbeta-C interface revealed important differences between the intersubunit interfaces in the type I and type II holoenzyme complexes. These differences are seen in both the R-subunits as well as the C-subunit. Unlike the type I isoform, the type II isoform complexes require both cAMP-binding domains, and ATP is not obligatory for high affinity interactions with the C-subunit. Surprisingly, the C-subunit mediates distinct, overlapping surfaces of interaction with the two R-isoforms despite a strong homology in sequence and similarity in domain organization. Identification of a remote allosteric site on the C-subunit that is essential for interactions with RII, but not RI subunits, further highlights the considerable diversity in interfaces found in higher order protein complexes mediated by the C-subunit of PKA.

Rapid Estimation of Solvation Energy for Simulations of Protein-Protein Association.

We have formulated the Energy by Linear Superposition of Corrections Approximation (ELSCA) for estimating the electrostatic and apolar solvation energy of bringing two proteins into close proximity or into contact as defined by the linearized Poisson-Boltzmann model and a linear function of the solvent-accessible surface area. ELSCA utilizes potentials of mean force between atom types found in the AMBER ff99 force field, a uniform distance-dependent dielectric, and a potential that mimics the change in solvent accessible surface area for bringing two solvated spheres into contact. ELSCA was trained by a linear least-squares fit on more than 39 000 putative complexes, each formed from pairs of nonhomologous proteins with a range of shapes, sizes, and charges. The training set was also designed to capture various stages of complex formation. ELSCA was tested against over 8000 non-native complexes of 45 enzyme/inhibitor, antibody/antigen, and other systems that are known to form complexes and gives an overall correlation of 0.962 with PBSA-derived energies for these complexes. The predictions have a slope of 0.89 on the actual values with a bias of 11.1 kcal/mol. When applied to native complexes of these 45 protein systems, ELSCA reproduces PBSA results with a correlation of 0.787, a slope of 1.13, and a bias of 13.0 kcal/mol. We report parameters for ELSCA in the context of the AMBER ff99 parameter set. Our model is most useful in macromolecular docking and protein association simulations, where large portions of each molecule may be considered rigid.