Comparison of the X-ray structure of native rubredoxin from Pyrococcus furiosus with the NMR structure of the zinc-substituted protein.

The three-dimensional X-ray structures of the oxidized and reduced forms of rubredoxin from Pyrococcus furiosus, determined at -161 degrees C, and the NMR structure of the zinc-substituted protein, determined in solution at 45 degrees C, are compared. The NMR and X-ray structures, which were determined independently, are very similar and lead to similar conclusions regarding the interactions that confer hyperthermostability.     

Solution-state structure by NMR of zinc-substituted rubredoxin from the marine hyperthermophilic archaebacterium Pyrococcus furiosus.

The three-dimensional solution-state structure is reported for the zinc-substituted form of rubredoxin (Rd) from the marine hyperthermophilic archaebacterium Pyrococcus furiosus, an organism that grows optimally at 100 degrees C. Structures were generated with DSPACE by a hybrid distance geometry (DG)-based simulated annealing (SA) approach that employed 403 nuclear Overhauser effect (NOE)-derived interproton distance restraints, including 67 interresidue, 124 sequential (i-j = 1), 75 medium-range (i-j = 2-5), and 137 long-range (i-j > 5) restraints. All lower interproton distance bounds were set at the sum of the van Der Waals radii (1.8 A), and upper bounds of 2.7 A, 3.3 A, and 5.0 A were employed to represent qualitatively observed strong, medium, and weak NOE cross peak intensities, respectively. Twenty-three backbone-backbone, six backbone-sulfur (Cys), two backbone-side chain, and two side chain-side chain hydrogen bond restraints were include for structure refinement, yielding a total of 436 nonbonded restraints, which averages to > 16 restraints per residue. A total of 10 structures generated from random atom positions and 30 structures generated by molecular replacement using the backbone coordinates of Clostridium pasteurianum Rd converged to a common conformation, with the average penalty (= sum of the square of the distance bounds violations; +/- standard deviation) of 0.024 +/- 0.003 A2 and a maximum total penalty of 0.035 A2. Superposition of the backbone atoms (C, C alpha, N) of residues A1-L51 for all 40 structures afforded an average pairwise root mean square (rms) deviation value (+/- SD) of 0.42 +/- 0.07 A. Superposition of all heavy atoms for residues A1-L51, including those of structurally undefined external side chains, afforded an average pairwise rms deviation of 0.72 +/- 0.08 A. Qualitative comparison of back-calculated and experimental two-dimensional NOESY spectra indicate that the DG/SA structures are consistent with the experimental spectra. The global folding of P. furiosus Zn(Rd) is remarkably similar to the folding observed by X-ray crystallography for native Rd from the mesophilic organism C. pasteurianum, with the average rms deviation value for backbone atoms of residues A1-L51 of P. furiosus Zn(Rd) superposed with respect to residues K2-V52 of C. pasteurianum Rd of 0.77 +/- 0.06 A. The conformations of aromatic residues that compose the hydrophobic cores of the two proteins are also similar. However, P. furiosus Rd contains several unique structural elements, including at least four additional hydrogen bonds and three potential electrostatic interactions.(ABSTRACT TRUNCATED AT 400 WORDS)     

Solution structure of bovine heart fatty acid-binding protein (H-FABPc)

Fatty acid-binding protein (FABP) from bovine heart, a 15 kDa cytoplasmic protein has been investigated by multidimensional homonuclear and heteronuclear NMR-spectroscopy. Perdeuterated palmitic acid has been used as fatty acid ligand. The tertiary structure has been determined from distance geometry calculations with the variable target functions algorithm (DIANA) [1] utilizing 1027 interproton distance constraints, which were obtained from¹H-homo-nuclear NOESY spectra. Overlapping NOE crosspeaks were assigned by heteronuclear multidimensional NMR-experiments with a¹⁵N-labelled sample. The tertiary structure resembles a -barrel (-clam) consisting of ten anti-parallel -strands and a short helix-turn-helix motif. The -strands are arranged in two nearly orthogonal -sheets composed of 5 strands each. The solution structure is compared with the x-ray cyrstal structure of bovine heart [4] and rat intestinal FABPs.Abbreviations DOF-COSY Double Quantum Filtered Correlated Spectroscopy - TOCSY Total Correlated Spectroscopy - NOE Nuclear Overhauser Enhancement - NOESY Nuclear Overhauser Enhancement and Exchange Spectroscopy - HMQC Heteronuclear Multiple Quantum Coherence - FABP Fatty Acid-Binding Protein - FABP_c Cellular Fatty Acid-Binding Protein - H-FABP_c Cellular Heart Fatty Acid-Binding Protein - I-FABP_c Cellular Intestinal Fatty Acid-Binding Protein     

A high quality nuclear magnetic resonance solution structure of peptide deformylase from Escherichia coli: Application of an automated assignment strategy using GARANT

The NMR structure of the peptide deformylase (PDF) (1–150) from Escherichia coli, which is an essential enzyme that removes the formyl group from nascent polypeptides and represents a potential target for drug discovery, was determined using 15N/13C doubly labeled protein. Nearly completely automated assignment routines were employed to assign three-dimensional triple resonance, 15N-resolved and 13C-resolved NOESY spectra using the program GARANT. This assignment strategy, demonstrated on a 17 kDa protein, is a significant advance in the automation of NMR data assignment and structure determination that will accelerate future work. A total of 2302 conformational constraints were collected as input for the distance geometry program DYANA. After restrained energy minimization with the program X-PLOR the 20 best conformers characterize a high quality structure with an average of 0.43 Å for the root-mean-square deviation calculated from the backbone atoms N, C and C, and 0.81 Å for all heavy atoms of the individual conformers relative to the mean coordinates for residues 1 to 150. The globular fold of PDF contains two -helices comprising residues 25–40, 125–138, six -strands 57–60, 70–77, 85–88, 98–101, 105–111, 117–123 and one 310 helix comprising residues 49–51. The C-terminal helix contains the HEXXH motif positioning a zinc ligand in a similar fashion to other metalloproteases, with the third ligand being cysteine and the fourth presumably a water. The three-dimensional structure of PDF affords insight into the substrate recognition and specificity for N-formylated over N-acetylated substrates and is compared to other PDF structures.     

Relaxation data in NMR structure determination: model calculations for the lysozyme-Gd3+ complex

The effect of including paramagnetic relaxation data as additional restraints in the determination of protein tertiary structures from NMR data has been explored by a systematic series of model calculations. The system used for testing the method was the 2.0 A resolution tetragonal crystal structure of hen egg white lysozyme (129 amino acid residues) and structures were generated using a version of the hybrid "distance geometry-dynamic simulated annealing" procedure. A limited set of 769 NOEs was used as restraints in all the calculations; the strengths of these were categorized into three classes on the basis of distances observed in the crystal structure. The values of 50 phi angles were also restrained on the basis of amide-alpha coupling constants calculated from the X-ray structure. Five sets of 12 structures were determined using differing sets of paramagnetic relaxation data as restraints additional to those involving the NOE and coupling constant data. The paramagnetic relaxation data were modeled on the basis of the distances of defined protons from the crystallographic binding site of Gd3+ in lysozyme. Analysis of the results showed that the relaxation data significantly improved the correspondence between the set of generated structures and the crystal structure, and that the more well defined the relaxation data, the more significant the improvement in the quality of the structures. The results suggest that the inclusion of paramagnetic relaxation restraints could be of significant value for the experimental determination of protein structures from NMR data.     

Three-dimensional 1H NMR structure of the nucleocapsid protein NCp10 of Moloney murine leukemia virus

Summary The nucleocapsid protein of Moloney murine leukemia virus (NCp10) is a 56-amino acid protein which contains one zinc finger of the CysX₂CysX₄HisX₄Cys form, a highly conserved motif present in most retroviruses and retroelements. At pH5, NCp10 binds one zinc atom and the complexation induces a folding of the CysX₂CysX₄HisX₄Cys box, similar to that observed for the zinc-binding domains of HIV-1 NC protein. The three-dimensional structure of NCp10 has been determined in aqueous solution by 600 MHz ¹H NMR spectroscopy. The proton resonances could be almost completely assigned by means of phase-sensitive double-quantum-filtered COSY, TOCSY and NOESY techniques. NOESY spectra yielded 597 relevant structural constraints, which were used as input for distance geometry calculations with DIANA. Further refinement was performed by minimization with the program AMBER, which was modified by introducing a zinc force field. The solution structure is characterized by a well-defined central zinc finger (rmsd of 0.747±0.209 Å for backbone atoms and 1.709±0.187 Å when all atoms are considered), surrounded by flexible N- and C-terminal domains. The Tyr²⁸, Trp³⁵, Lys³⁷, Lys⁴¹ and Lys⁴² residues, which are essential for activity, lie on the same face of the zinc finger, forming a bulge structure probably involved in viral RNA binding. The significance of these structural characteristics for the various biological functions of the protein is discussed, taking into account the results obtained with various mutants.     

The NMR side-chain assignments and solution structure of enzyme IIBcellobiose of the phosphoenolpyruvate-dependent phosphotransferase system of Escherichia coli.

The assignment of the side-chain NMR resonances and the determination of the three-dimensional solution structure of the C10S mutant of enzyme IIBcellobiose (IIBcel) of the phosphoenolpyruvate-dependent phosphotransferase system of Escherichia coli are presented. The side-chain resonances were assigned nearly completely using a variety of mostly heteronuclear NMR experiments, including HCCH-TOCSY, HCCH-COSY, and COCCH-TOCSY experiments as well as CBCACOHA, CBCA(CO)NH, and HBHA(CBCA)(CO)NH experiments. In order to obtain the three-dimensional structure, NOE data were collected from 15N-NOESY-HSQC, 13C-HSQC-NOESY, and 2D NOE experiments. The distance restraints derived from these NOE data were used in distance geometry calculations followed by molecular dynamics and simulated annealing protocols. In an iterative procedure, additional NOE assignments were derived from the calculated structures and new structures were calculated. The final set of structures, calculated with approximately 2000 unambiguous and ambiguous distance restraints, has an rms deviation of 1.1 A on C alpha atoms. IIBcel consists of a four stranded parallel beta-sheet, in the order 2134. The sheet is flanked with two and three alpha-helices on either side. Residue 10, a cysteine in the wild-type enzyme, which is phosphorylated during the catalytic cycle, is located at the end of the first beta-strand. A loop that is proposed to be involved in the binding of the phosphoryl-group follows the cysteine. The loop appears to be disordered in the unphosphorylated state.     

The three-dimensional structure of a helix-less variant of intestinal fatty acid-binding protein.

Intestinal fatty acid-binding protein (I-FABP) is a cytosolic 15.1-kDa protein that appears to function in the intracellular transport and metabolic trafficking of fatty acids. It binds a single molecule of long-chain fatty acid in an enclosed cavity surrounded by two five-stranded antiparallel beta-sheets and a helix-turn-helix domain. To investigate the role of the helical domain, we engineered a variant of I-FABP by deleting 17 contiguous residues and inserting a Ser-Gly linker (Kim K et al., 1996, Biochemistry 35:7553-7558). This variant, termed delta17-SG, was remarkably stable, exhibited a high beta-sheet content and was able to bind fatty acids with some features characteristic of the wild-type protein. In the present study, we determined the structure of the delta17-SG/palmitate complex at atomic resolution using triple-resonance 3D NMR methods. Sequence-specific 1H, 13C, and 15N resonance assignments were established at pH 7.2 and 25 degrees C and used to define the consensus 1H/13C chemical shift-derived secondary structure. Subsequently, an iterative protocol was used to identify 2,544 NOE-derived interproton distance restraints and to calculate its tertiary structure using a unique distance geometry/simulated annealing algorithm. In spite of the sizable deletion, the delta17-SG structure exhibits a backbone conformation that is nearly superimposable with the beta-sheet domain of the wild-type protein. The selective deletion of the alpha-helical domain creates a very large opening that connects the interior ligand-binding cavity with exterior solvent. Unlike wild-type I-FABP, fatty acid dissociation from delta17-SG is structurally and kinetically unimpeded, and a protein conformational transition is not required. The delta17-SG variant of I-FABP is the only wild-type or engineered member of the intracellular lipid-binding protein family whose structure lacks alpha-helices. Thus, delta17-SG I-FABP constitutes a unique model system for investigating the role of the helical domain in ligand-protein recognition, protein stability and folding, lipid transfer mechanisms, and cellular function.     

Toward the solution of the protein structure prediction problem

Since Anfinsen demonstrated that the information encoded in a protein’s amino acid sequence determines its structure in 1973, solving the protein structure prediction problem has been the Holy Grail of structural biology. The goal of protein structure prediction approaches is to utilize computational modeling to determine the spatial location of every atom in a protein molecule starting from only its amino acid sequence. Depending on whether homologous structures can be found in the Protein Data Bank (PDB), structure prediction methods have been historically categorized as template-based modeling (TBM) or template-free modeling (FM) approaches. Until recently, TBM has been the most reliable approach to predicting protein structures, and in the absence of reliable templates, the modeling accuracy sharply declines. Nevertheless, the results of the most recent community-wide assessment of protein structure prediction experiment (CASP14) have demonstrated that the protein structure prediction problem can be largely solved through the use of end-to-end deep machine learning techniques, where correct folds could be built for nearly all single-domain proteins without using the PDB templates. Critically, the model quality exhibited little correlation with the quality of available template structures, as well as the number of sequence homologs detected for a given target protein. Thus, the implementation of deep-learning techniques has essentially broken through the 50-year-old modeling border between TBM and FM approaches and has made the success of high-resolution structure prediction significantly less dependent on template availability in the PDB library.     

TOUCHSTONEX: protein structure prediction with sparse NMR data

TOUCHSTONEX, a new method for folding proteins that uses a small number of long-range contact restraints derived from NMR experimental NOE (nuclear Overhauser enhancement) data, is described. The method employs a new lattice-based, reduced model of proteins that explicitly represents C(alpha), C(beta), and the sidechain centers of mass. The force field consists of knowledge-based terms to produce protein-like behavior, including various short-range interactions, hydrogen bonding, and one-body, pairwise, and multibody long-range interactions. Contact restraints were incorporated into the force field as an NOE-specific pairwise potential. We evaluated the algorithm using a set of 125 proteins of various secondary structure types and lengths up to 174 residues. Using N/8 simulated, long-range sidechain contact restraints, where N is the number of residues, 108 proteins were folded to a C(alpha)-root-mean-square deviation (RMSD) from native below 6.5 A. The average RMSD of the lowest RMSD structures for all 125 proteins (folded and unfolded) was 4.4 A. The algorithm was also applied to limited experimental NOE data generated for three proteins. Using very few experimental sidechain contact restraints, and a small number of sidechain-main chain and main chain-main chain contact restraints, we folded all three proteins to low-to-medium resolution structures. The algorithm can be applied to the NMR structure determination process or other experimental methods that can provide tertiary restraint information, especially in the early stage of structure determination, when only limited data are available.     

NMR structure of the human doppel protein

The NMR structure of the recombinant human doppel protein, hDpl(24-152), contains a flexibly disordered "tail" comprising residues 24-51, and a globular domain extending from residues 52 to 149 for which a detailed structure was obtained. The globular domain contains four alpha-helices comprising residues 72-80 (alpha1), 101-115 (alpha2(a)), 117-121 (alpha2(b)), and 127-141 (alpha3), and a short two-stranded anti-parallel beta-sheet comprising residues 58-60 (beta1) and 88-90 (beta2). The fold of the hDpl globular domain thus coincides nearly identically with the structure of the murine Dpl protein. There are close similarities with the human prion protein (hPrP) but, similar to the situation with the corresponding murine proteins, hDpl shows marked local differences when compared to hPrP: the beta-sheet is flipped by 180 degrees with respect to the molecular scaffold formed by the four helices, and the beta1-strand is shifted by two residues toward the C terminus. A large solvent-accessible hydrophobic cleft is formed on the protein surface between beta2 and alpha3, which has no counterpart in hPrP. The helix alpha2 of hPrP is replaced by two shorter helices, alpha2(a) and alpha2(b). The helix alpha3 is shortened by more than two turns when compared with alpha3 of hPrP, which is enforced by the positioning of the second disulfide bond in hDpl. The C-terminal peptide segment 144-149 folds back onto the loop connecting beta2 and alpha2. All but four of the 20 conserved residues in the globular domains of hPrP and hDpl appear to have a structural role in maintaining a PrP-type fold. The conservation of R76, E96, N110 and R134 in hDpl, corresponding to R148, E168, N183 and R208 in hPrP suggests that these amino acid residues might have essential roles in the so far unknown functions of PrP and Dpl in healthy organisms.     

The structure of Escherichia coli heat-stable enterotoxin b by nuclear magnetic resonance and circular dichroism.

The heat-stable enterotoxin b (STb) is secreted by enterotoxigenic Escherichia coli that cause secretory diarrhea in animals and humans. It is a 48-amino acid peptide containing two disulfide bridges, between residues 10 and 48 and 21 and 36, which are crucial for its biological activity. Here, we report the solution structure of STb determined by two- and three-dimensional NMR methods. Approximate interproton distances derived from NOE data were used to construct structures of STb using distance-geometry and simulated annealing procedures. The NMR-derived structure shows that STb is helical between residues 10 and 22 and residues 38 and 44. The helical structure in the region 10-22 is amphipathic and exposes several polar residues to the solvent, some of which have been shown to be important in determining the toxicity of STb. The hydrophobic residues on the opposite face of this helix make contacts with the hydrophobic residues of the C-terminal helix. The loop region between residues 21 and 36 has another cluster of hydrophobic residues and exposes Arg 29 and Asp 30, which have been shown to be important for intestinal secretory activity. CD studies show that reduction of disulfide bridges results in a dramatic loss of structure, which correlates with loss of function. Reduced STb adopts a predominantly random-coil conformation. Chromatographic measurements of concentrations of native, fully reduced, and single-disulfide species in equilibrium mixtures of STb in redox buffers indicate that the formation of the two disulfide bonds in STb is only moderately cooperative. Similar measurements in the presence of 8 M urea suggest that the native secondary structure significantly stabilizes the disulfide bonds.     

Nuclear magnetic resonance assignment and secondary structure of an ankyrin-like repeat-bearing protein: myotrophin.

Multidimensional heteronuclear NMR has been applied to the structural analysis of myotrophin, a novel protein identified from spontaneously hypertensive rat hearts and hypertrophic human hearts. Myotrophin has been shown to stimulate protein synthesis in myocytes and likely plays an important role in the initiation of cardiac hypertrophy, a major cause of mortality in humans. Recent cDNA cloning revealed that myotrophin has 11B amino acids containing 2.5 contiguous ANK repeats, a motif known to be involved in a wide range of macromolecular recognition. A series of two- and three-dimensional heteronuclear bond correlation NMR experiments have been performed on uniformly 15N-labeled or uniformly 15N/13C-labeled protein to obtain the 1H, 15N, and 13C chemical shift assignments. The secondary structure of myotrophin has been determined by a combination of NOEs, NH exchange data, 3JHN alpha coupling constants, and chemical shifts of 1H alpha, 13C alpha, and 13 C beta. The protein has been found to consist of seven helices, all connected by turns or loops. Six of the seven helices (all but the C-terminal helix) form three separate helix-turn-helix motifs. The two full ANK repeats in myotrophin are characteristic of multiple turns followed by a helix-turn-helix motif. A hairpin-like turn involving L32-R36 in ANK repeat #1 exhibits slow conformational averaging on the NMR time scale and appears dynamically different from the corresponding region (D65-169) of ANK repeat #2.     

Distance geometry generates native-like folds for small helical proteins using the consensus distances of predicted protein structures.

For successful ab initio protein structure prediction, a method is needed to identify native-like structures from a set containing both native and non-native protein-like conformations. In this regard, the use of distance geometry has shown promise when accurate inter-residue distances are available. We describe a method by which distance geometry restraints are culled from sets of 500 protein-like conformations for four small helical proteins generated by the method of Simons et al. (1997). A consensus-based approach was applied in which every inter-Calpha distance was measured, and the most frequently occurring distances were used as input restraints for distance geometry. For each protein, a structure with lower coordinate root-mean-square (RMS) error than the mean of the original set was constructed; in three cases the topology of the fold resembled that of the native protein. When the fold sets were filtered for the best scoring conformations with respect to an all-atom knowledge-based scoring function, the remaining subset of 50 structures yielded restraints of higher accuracy. A second round of distance geometry using these restraints resulted in an average coordinate RMS error of 4.38 A.     

Solution structure of the Arabidopsis thaliana telomeric repeat-binding protein DNA binding domain: a new fold with an additional C-terminal helix

The double-stranded telomeric repeat-binding protein (TRP) AtTRP1 is isolated from Arabidopsis thaliana. Using gel retardation assays, we defined the C-terminal 97 amino acid residues, Gln464 to Val560 (AtTRP1(464-560)), as the minimal structured telomeric repeat-binding domain. This region contains a typical Myb DNA-binding motif and a C-terminal extension of 40 amino acid residues. The monomeric AtTRP1(464-560) binds to a 13-mer DNA duplex containing a single repeat of an A.thaliana telomeric DNA sequence (GGTTTAG) in a 1:1 complex, with a K(D) approximately 10(-6)-10(-7) M. Nuclear magnetic resonance (NMR) examination revealed that the solution structure of AtTRP1(464-560) is a novel four-helix tetrahedron rather than the three-helix bundle structure found in typical Myb motifs and other TRPs. Binding of the 13-mer DNA duplex to AtTRP1(464-560) induced significant chemical shift perturbations of protein amide resonances, which suggests that helix 3 (H3) and the flexible loop connecting H3 and H4 are essential for telomeric DNA sequence recognition. Furthermore, similar to that in hTRF1, the N-terminal arm likely contributes to or stabilizes DNA binding. Sequence comparisons suggested that the four-helix structure and the involvement of the loop residues in DNA binding may be features unique to plant TRPs.     

A simple topological representation of protein structure: implications for new, fast, and robust structural classification

A topological representation of proteins is developed that makes use of two metrics: the Euclidean metric for identifying natural nearest neighboring residues via the Delaunay tessellation in Cartesian space and the distance between residues in sequence space. Using this representation, we introduce a quantitative and computationally inexpensive method for the comparison of protein structural topology. The method ultimately results in a numerical score quantifying the distance between proteins in a heuristically defined topological space. The properties of this scoring scheme are investigated and correlated with the standard Calpha distance root-mean-square deviation measure of protein similarity calculated by rigid body structural alignment. The topological comparison method is shown to have a characteristic dependence on protein conformational differences and secondary structure. This distinctive behavior is also observed in the comparison of proteins within families of structural relatives. The ability of the comparison method to successfully classify proteins into classes, superfamilies, folds, and families that are consistent with standard classification methods, both automated and human-driven, is demonstrated. Furthermore, it is shown that the scoring method allows for a fine-grained classification on the family, protein, and species level that agrees very well with currently established phylogenetic hierarchies. This fine classification is achieved without requiring visual inspection of proteins, sequence analysis, or the use of structural superimposition methods. Implications of the method for a fast, automated, topological hierarchical classification of proteins are discussed.     

Conformational sampling by NMR solution structures calculated with the program DIANA evaluated by comparison with long-time molecular dynamics calculations in explicit water

The NMR solution structure of bovine pancreatic trypsin inhibitor (BPTI) obtained by distance geometry calculations with the program DIANA is compared with groups of conformers generated by molecular dynamics (MD) simulations in explicit water at ambient temperature and pressure. The MD simulations started from a single conformer and were free or restrained either by the experimental NOE distance restraints or by time-averaged restraints; the groups of conformers were collected either in 10 ps intervals during 200 ps periods of simulation, or in 50 ps intervals during a 1 ns period of simulation. Overall, these comparisons show that the standard protein structure determination protocol with the program DIANA provides a picture of the protein structure that is in agreement with MD simulations using “realistic” potential functions over a nanosecond timescale. For well-constrained molecular regions there is a trend in the free MD simulation of duration 1 ns that the sampling of the conformation space is slightly increased relative to the DIANA calculations. In contrast, for surface-exposed side-chains that are less extensively constrained by the NMR data, the DIANA conformers tend to sample larger regions of conformational space than conformers selected from any of the MD trajectories. Additional insights into the behavior of surface side-chains come from comparison of the MD runs of 200 ps or 1 ns duration. In this time range the sampling of conformation space by the protein surface depends strongly on the length of the simulation, which indicates that significant side-chain transitions occur on the nanosecond timescale and that much longer simulations will be needed to obtain statistically significant data on side-chain dynamics.     

Structure prediction of protein complexes by an NMR-based protein docking algorithm

Protein docking algorithms can be used to study the driving forces and reaction mechanisms of docking processes. They are also able to speed up the lengthy process of experimental structure elucidation of protein complexes by proposing potential structures. In this paper, we are discussing a variant of the protein-protein docking problem, where the input consists of the tertiary structures of proteins A and B plus an unassigned one-dimensional ¹H-NMR spectrum of the complex AB. We present a new scoring function for evaluating and ranking potential complex structures produced by a docking algorithm. The scoring function computes a `theoretical' ¹H-NMR spectrum for each tentative complex structure and subtracts the calculated spectrum from the experimental one. The absolute areas of the difference spectra are then used to rank the potential complex structures. In contrast to formerly published approaches (e.g. [Morelli et al. (2000) Biochemistry, 39, 2530–2537]) we do not use distance constraints (intermolecular NOE constraints). We have tested the approach with four protein complexes whose three-dimensional structures are stored in the PDB data bank [Bernstein et al. (1977)] and whose ¹H-NMR shift assignments are available from the BMRB database. The best result was obtained for an example, where all standard scoring functions failed completely. Here, our new scoring function achieved an almost perfect separation between good approximations of the true complex structure and false positives.     

Improving protein tertiary structure prediction by deep learning and distance prediction in CASP14

Substantial progresses in protein structure prediction have been made by utilizing deep-learning and residue-residue distance prediction since CASP13. Inspired by the advances, we improve our CASP14 MULTICOM protein structure prediction system by incorporating three new components: (a) a new deep learning-based protein inter-residue distance predictor to improve template-free (ab initio) tertiary structure prediction, (b) an enhanced template-based tertiary structure prediction method, and (c) distance-based model quality assessment methods empowered by deep learning. In the 2020 CASP14 experiment, MULTICOM predictor was ranked seventh out of 146 predictors in tertiary structure prediction and ranked third out of 136 predictors in inter-domain structure prediction. The results demonstrate that the template-free modeling based on deep learning and residue-residue distance prediction can predict the correct topology for almost all template-based modeling targets and a majority of hard targets (template-free targets or targets whose templates cannot be recognized), which is a significant improvement over the CASP13 MULTICOM predictor. Moreover, the template-free modeling performs better than the template-based modeling on not only hard targets but also the targets that have homologous templates. The performance of the template-free modeling largely depends on the accuracy of distance prediction closely related to the quality of multiple sequence alignments. The structural model quality assessment works well on targets for which enough good models can be predicted, but it may perform poorly when only a few good models are predicted for a hard target and the distribution of model quality scores is highly skewed. MULTICOM is available at https://github.com/jianlin-cheng/MULTICOM_Human_CASP14/tree/CASP14_DeepRank3 and https://github.com/multicom-toolbox/multicom/tree/multicom_v2.0 .