Backbone Root Mean Square Research Articles

NMR-based Binding Screen and Structural Analysis of the Complex Formed between α-Cobratoxin and an 18-Mer Cognate Peptide Derived from the α1 Subunit of the Nicotinic Acetylcholine Receptor fromTorpedo californica

The alpha18-mer peptide, spanning residues 181-198 of the Torpedo nicotinic acetylcholine receptor alpha1 subunit, contains key binding determinants for agonists and competitive antagonists. To investigate whether the alpha18-mer can bind other alpha-neurotoxins besides alpha-bungarotoxin, we designed a two-dimensional (1)H-(15)N heteronuclear single quantum correlation experiment to screen four related neurotoxins for their binding ability to the peptide. Of the four toxins tested (erabutoxin a, erabutoxin b, LSIII, and alpha-cobratoxin), only alpha-cobratoxin binds the alpha18-mer to form a 1:1 complex. The NMR solution structure of the alpha-cobratoxin.alpha18-mer complex was determined with a backbone root mean square deviation of 1.46 A. In the structure, alpha-cobratoxin contacts the alpha18-mer at the tips of loop I and II and through C-terminal cationic residues. The contact zone derived from the intermolecular nuclear Overhauser effects is in agreement with recent biochemical data. Furthermore, the structural models support the involvement of cation-pi interactions in stabilizing the complex. In addition, the binding screen results suggest that C-terminal cationic residues of alpha-bungarotoxin and alpha-cobratoxin contribute significantly to binding of the alpha18-mer. Finally, we present a structural model for nicotinic acetylcholine receptor-alpha-cobratoxin interaction by superimposing the alpha-cobratoxin.alpha18-mer complex onto the crystal structure of the acetylcholine-binding protein (Protein Data Bank code ).

The Solution Structures of the Human β-Defensins Lead to a Better Understanding of the Potent Bactericidal Activity of HBD3 against Staphylococcus aureus

The three human beta-defensins, HBD1--3, are 33--47-residue, cationic antimicrobial proteins expressed by epithelial cells. All three proteins have broad spectrum antimicrobial activity, with HBD3 consistently being the most potent. Additionally, HBD3 has significant bactericidal activity against Gram-positive Staphylococcus aureus at physiological salt concentrations. We have compared the multimeric state of the three beta-defensins using NMR diffusion spectroscopy, dynamic and static light scattering, and analysis of the migration of the three beta-defensins on a native gel. All three techniques are in agreement, suggesting that HBD-3 is a dimer, while HBD-1 and HBD-2 are monomeric. Subsequently, the NMR solution structures of HBD1 and HBD3 were determined using standard homonuclear techniques and compared with the previously determined solution structure of HBD2. Both HBD1 and HBD3 form well defined structures with backbone root mean square deviations of 0.451 and 0.616 A, respectively. The tertiary structures of all three beta-defensins are similar, with a short helical segment preceding a three-stranded antiparallel beta-sheet. The surface charge density of each of the defensins is markedly different, with the surface of HBD3 significantly more basic. Analysis of the NMR data and structures led us to suggest that HBD3 forms a symmetrical dimer through strand beta2 of the beta-sheet. The increased anti-Staphylococcal activity of HBD3 may be explained by the capacity of the protein to form dimers in solution at low concentrations, an amphipathic dimer structure, and the increased positive surface charge compared with HBD1 and HBD2.

Spatial Structure of Zervamicin IIB Bound to DPC Micelles: Implications for Voltage-Gating

Zervamicin IIB is a 16-amino acid peptaibol that forms voltage-dependent ion channels with multilevel conductance states in planar lipid bilayers and vesicular systems. The spatial structure of zervamicin IIB bound to dodecylphosphocholine micelles was studied by nuclear magnetic resonance spectroscopy. The set of 20 structures obtained has a bent helical conformation with a mean backbone root mean square deviation value of ∼0.2 Å and resembles the structure in isotropic solvents (Balashova et al., 2000. NMR structure of the channel-former zervamicin IIB in isotropic solvents. FEBS Lett 466:333–336). The N-terminus represents an α-helix, whereas the C-terminal part has a mixed 3 10/ α R hydrogen-bond pattern. In the anisotropic micelle environment, the bending angle on Hyp10 (23°) is smaller than that (47°) in isotropic solvents. In the NOESY (Nuclear Overhauser Effect Spectroscopy) spectra, the characteristic attenuation of the peptide signals by 5- and 16-doxylstearate relaxation probes indicates a peripheral mode of the peptaibol binding to the micelle with the N-terminus immersed slightly deeper into micelle interior. Analysis of the surface hydrophobicity reveals that the zervamicin IIB helix is amphiphilic and well suited to formation of a tetrameric transmembrane bundle, according to the barrel-stave mechanism. The results are discussed in a context of voltage-driven peptaibol insertion into membrane.

Prediction of a 12-residue loop in bovine pancreatic trypsin inhibitor: effects of buried water.

The x-ray conformations of 5-, 7-, 9-, and 12-residue loops in bovine pancreatic trypsin inhibitor (BPTI) were predicted by the use of multiple independent Monte Carlo simulating annealing (MCSA) runs starting from random conformations. Four buried water molecules interacted with a 12-residue loop that started at residue 8 and ended at residue 19, and that included the binding region. The final conformation at the end of an MCSA run was characterized. Solvation free energy based on the solvent accessible surface area was included in the energy function at low simulated annealing temperatures. Conformational states were interactively separated by a recently developed algorithm. Computed loops were characterized in terms of total energy, and backbone and side chain root mean square deviations (RMSDs) between computed native loop conformations and the x-ray conformation. The 12-residue loop was computed with and without buried water [called WL12(8-19) and L12(8-19), respectively]. The backbone was reliably and reproducibly computed to within 1.1 A in L12(8-19) and 0.9 A in WL12(8-19). L12(8-19) required significantly more MCSA runs to achieve the same level of reproducibility as WL12(8-19). Based on the size of the cluster of low energy native loop conformations, and the computational effort, WL12(8-19) had greater entropy. In calculations of 7-, 9-, and 12-residue loops without buried water, the effects of buried water became obvious in the 12-residue loop calculation, which interacted with all four buried water molecules. Nearly all conformations of the native loop conformer had a hydrogen bond between the Lys 15 side chain and the backbone of Gly 12, Pro 13, and Cys 14, which may have implications in the rate of exchange of buried water with bulk solvent and in protein folding. The present version of MCSA program was more efficient than earlier versions.

Molecular Dynamics of the KcsA K + Channel in a Bilayer Membrane

Molecular dynamics (MD) simulations of an atomic model of the KcsA K + channel embedded in an explicit dipalmitoylphosphatidylcholine (DPPC) phospholipid bilayer solvated by a 150 mM KCl aqueous salt solution are performed and analyzed. The model includes the KcsA K + channel, based on the recent crystallographic structure of Doyle et al. (1998, Science. 280:69–77), 112 DPPC, K + and Cl − ions, as well as over 6500 water molecules for a total of more than 40,000 atoms. Three K + ions are explicitly included in the pore. Two are positioned in the selectivity filter on the extracellular side and one in the large water-filled cavity. Different starting configurations of the ions and water molecules in the selectivity filter are considered, and MD trajectories are generated for more than 4 ns. The conformation of KcsA is very stable in all of the trajectories, with a global backbone root mean square (RMS) deviation of less than 1.9 Å with respect to the crystallographic structure. The RMS atomic fluctuations of the residues surrounding the selectivity filter on the extracellular side of the channel are significantly lower than those on the intracellular side. The motion of the residues with aromatic side chains surrounding the selectivity filter (Trp 67, Trp 68, Tyr 78, and Tyr 82) is anisotropic with the smallest RMS fluctuations in the direction parallel to the membrane plane. A concerted dynamic transition of the three K + ions in the pore is observed, during which the K + ion located initially in the cavity moves into the narrow part of the selectivity filter, while the other two K + ions move toward the extracellular side. A single water molecule is stabilized between each pair of ions during the transition, suggesting that each K + cation translocating through the narrow pore is accompanied by exactly one water molecule, in accord with streaming potential measurements (Alcayaga et al., 1989, Biophys. J. 55:367–371). The displacement of the ions is coupled with the structural fluctuations of Val 76 and Gly 77, in the selectivity filter, as well as the side chains of Glu 71, Asp 80, and Arg 89, near the extracellular side. Thus the mechanical response of the channel structure at distances as large as 10–20 Å from the ions in the selectivity filter appears to play an important role in the concerted transition.

A fast method to sample real protein conformational space

A fast computer program, FOLDTRAJ, to generate plausible random protein structures is reported. All-atom proteins are made directly in continuous three-dimensional space starting from primary sequence with an N to C directed build-up method. The method uses a novel pipelined residue addition approach in which the leading edge of the protein is constructed three residues at a time for optimal protein geometry, including the placement of cis proline. Build-up methods represent a classic N-body problem, expected to scale as N2. When proteins become more collapsed, build-up methods are susceptible to backtracking problems which can scale exponentially with the number of residues required to back out of a trapped walk. We have provided solutions to both these problems, using a multiway binary tree that makes the N-body problem of bump-checking scale as NlogN, and speeding up backtracking by varying the number of tries before backtracking based on available conformational space. FOLDTRAJ is independent of energy potentials, other than that implicit in the geometrical properties derived by statistical studies of known structures, and in atomic Van der Waals radii. WHAT―CHECK shows that the program generates chirally and physically valid proteins with all bond lengths, angles and dihedrals within allowable tolerances. Random structures built using sequences from PDB files 1SEM, 2HPR, and 1RTP typically have 5–15% α-helical content (according to DSSP) and on the order of 20% β-strand/extended content. Ensembles of random structures are compared with polymer theory and with experimentally determined fluorescence resonance energy transfer distances. Reasonably sized structure ensembles do sample most of the conformational space available to proteins. The method is also capable of protein reconstruction using CαCα direction vectors, and it compares favorably with methods that reconstruct protein backbones based on alpha-carbon coordinates, having an average backbone and Cβ root mean square deviation of 0.63 Å for nine different protein folds. Proteins 2000;39:112–131. © 2000 Wiley-Liss, Inc.

Nuclear Magnetic Resonance Solution Conformation of α-Conotoxin AuIB, an α3β4 Subtype-selective Neuronal Nicotinic Acetylcholine Receptor Antagonist

The neuronal nicotinic acetylcholine receptors constitute a highly diverse group, with subtypes consisting of pentameric combinations of alpha and beta subunits. alpha-Conotoxins are a homologous series of small peptides that antagonize these receptors. We present the three-dimensional solution structure of alpha-conotoxin AuIB, the first 15-residue alpha-conotoxin known to selectively block the alpha(3)beta(4) nicotinic acetylcholine receptor subtype. The pairwise backbone and heavy-atom root mean square deviation for an ensemble of 20 structures are 0.269 and 0.720 A, respectively. The overall fold of alpha-conotoxin AuIB closely resembles that of the alpha4/7 subfamily alpha-conotoxins. However, the absence of Tyr(15), normally present in other alpha4/7 members, results in tight bending of the backbone at the C terminus and effectively renders Asp(14) to assume the spatial location of Tyr(15) present in other neuronal alpha4/7 alpha-conotoxins. Structural comparison of alpha-conotoxin AuIB with the alpha(3)beta(2) subtype-specific alpha-conotoxin MII shows different electrostatic surface charge distributions, which may be important in differential receptor subtype recognition.

NMR structure of the channel-former zervamicin IIB in isotropic solvents

Spatial structure of the membrane channel-forming hexadecapeptide, zervamicin IIB, was studied by NMR spectroscopy in mixed solvents of different polarity ranging from CDCl 3/CD 3OH (9:1, v/v) to CD 3OH/H 2O (1:1, v/v). The results show that in all solvents used the peptide has a very similar structure that is a bent amphiphilic helix with a mean backbone root mean square deviation (rmsd) value of ca. 0.3 Å. Side chains of Trp 1, Ile 2, Gln 3, Ile 5 and Thr 6 are mobile. The results are discussed in relation to the validity of the obtained structure to serve as a building block of zervamicin IIB ion channels.

Homology Modeling and Molecular Dynamics Simulations of PBCV-1 Glycosylase Complexed with UV-damaged DNA

The UV-light damage specific DNA glycosylase from Chlorella virus strain PBCV-1 (pyrimidine dimer glycosylase; PDG) incises DNA at sites containing UV-induced thymidine dimers by catalyzing the breakage of the N-C-1′ glycosyl bond. As the amino acid sequence of PDG is 41% identical to that of T4 endonuclease V (Endo V), and potential key active site residues are conserved, we used coordinates from a crystal structure of an Endo V complexed with DNA containing a cis-syn cyclobutane thymidine-dimer as a template to model a similar complex of PDG. Quantum mechanical calculations of the damaged base pair and the distance geometry based program DIAMOD were used to generate a PDG/DNA model whose backbone root mean square deviation (RMSD) to the Endo V/DNA structure was 0.5 A, 0.5 A, and 0.8 A for DNA, protein, and the whole complex, respectively. To better understand structural details that could account for differences in activity of the two enzymes, molecular dynamics simulations were used to follow protein-DNA interactions in an aqueous environment. The simulations of the Endo V/DNA complex indicate new roles for Arg22 and Arg26 in the active site in recognizing irregular pairing and maintaining the strand separation needed for incision of the damaged bases. The model for the PDG/DNA complex and simulations thereof indicate a similar mechanism for DNA binding by this enzyme despite significant differences in residues maintaining the flipped-out adenine and strand separation in the area of damage. According to our model, PDG′s increased affinity for substrate is probably due to a higher surface charge. Further, reduced packing density in the active site could account for PDG′s activity on trans-syn II cyclobutane dimers.

Refined solution structure of the c‐terminal DNA‐binding domain of human immunovirus‐1 integrase

The structure of the C-terminal DNA-binding domain of human immunovirus-1 integrase has been refined using nuclear magnetic resonance spectroscopy. The protein is a dimer in solution and shows a well-defined dimer interface. The folding topology of the monomer consists of a five-stranded β-barrel that resembles that of Src homology 3 domains. Compared with our previously reported structure, the structure is now defined far better. The final 42 structures display a back-bone root mean square deviation versus the average of 0.46 Å. Correlation of the structure with recent mutagenesis studies suggests two possible models for DNA binding. Proteins 1999;36:556–564. © 1999 Wiley-Liss, Inc.

Establishing a degree of order: obtaining high-resolution NMR structures from molecular alignment

I would like to thank Eva de Alba for providing structural data as well as Ad Bax and James Baber for their critical comments.

NMR Solution Structure of Complement-like Repeat CR8 from the Low Density Lipoprotein Receptor-related Protein

The low density lipoprotein receptor-related protein is a member of the low density lipoprotein receptor family and contains clusters of cysteine-rich complement-like repeats of about 42 residues that are present in all members of this family of receptors. These clusters are thought to be the principal binding sites for protein ligands. We have expressed one complement-like repeat, CR8, from the cluster in lipoprotein receptor-related protein that binds certain proteinase inhibitor-proteinase complexes and used three-dimensional NMR on the 13C/15N-labeled protein to determine the structure in solution of the calcium-bound form. The structure is very similar in overall fold to repeat 5 from the low density lipoprotein receptor (LB5), with backbone root mean square deviation of 1.5 A. The calcium-binding site also appears to be homologous, with four carboxyl and two backbone carbonyl ligands. However, differences in primary structure are such that equivalent surfaces that might represent the binding interfaces are very different from one another, indicating that different domains will have very different ligand specificities.

Structure of the Insect Cytokine Peptide Plasmatocyte-spreading Peptide 1 from Pseudoplusia includens

The structure of the recently identified plasmatocyte spreading peptide from the moth Pseudoplusia includens (PSP1) has been determined by NMR spectroscopy. This novel insect cytokine consists of 23 amino acid residues and a single disulfide bond. Torsion angle dynamics calculations utilizing a total of 337 distance constraints yielded an ensemble of 30 structures with an average backbone root mean square deviation for residues 7-22 of 0.18 A from the mean structure. The structure consists of a disordered N-terminal region and a well defined core that is stabilized by numerous hydrophobic interactions and a short beta-hairpin. Structural comparisons confirm that PSP1 adopts an epidermal growth factor (EGF)-like fold with close similarity to the C-terminal subdomain of EGF-like module 5 of human thrombomodulin. The combination of the three-dimensional structure of PSP1 and the extensive literature on EGF-receptor interactions should accelerate the process of identifying the specific residues responsible for receptor binding activity of this family of immunoregulatory peptides.

Structural Model of the HIV-1 Tat(46–58)-TAR Complex

The trans-activator protein (Tat) of human immunodeficiency virus type 1 (HIV-1>) binds to an uridine-rich bulge of an RNA target (TAR; trans-activation responsive element) predominantly via its basic sequence domain. The structure of the Tat(46–58)-TAR complex has been determined by a novel modeling approach relying on structural information about one crucial arginine residue and crosslink data. The strategy described here solely uses this experimental data without additional “modeling” assumptions about the structure of the complex in order to avoid human bias. Model building was performed in a fashion similar to structure calculations from nuclear magnetic resonance (NMR)-spectroscopic data using restrained molecular dynamics. The resulting set of structures of Tat(46–58) in its complex with TAR reveals that all models have converged to a common fold, showing a backbone root mean square deviation (RMSD) of 1.36Å. Analysis of the calculated structures suggests that HIV-1 Tat forms a hairpin loop in its complex with TAR that shares striking similarity to the hairpin formed by the structure of the bovine immunodeficiency virus Tat protein after TAR binding as determined by NMR studies. The outlined approach is not limited to the Tat-TAR complex modeling, but is also applicable to all molecular complexes with sufficient biochemical and biophysical data available.

Molecular dynamics simulations of epidermal growth factor and transforming growth factor‐α structures in water

AMBER v. 4.1 force field in 1.5 ns NPT molecular dynamics simulations of murine epidermal growth factor (mEGF), human epidermal growth factor (hEGF), and human transforming growth factor-α (hTGF-α) structures with explicit TIP3P solvation were used to investigate differences in backbone stability, changes in secondary structure, interdomain flexibility, and weakly polar interactions. Backbone root mean square deviations of sections of each peptide show that the most stable regions in mEGF and hEGF are the A-, B-, and C-loops, whereas the most stable regions in hTGF-α are the A- and B-loops. The secondary structure in the B-loops of mEGF and hEGF differ significantly from the nuclear magnetic resonance (NMR) structures of mEGF and hEGF. The position and type of turns in the B-loop of mEGF and hEGF increase the interstrand distance of the antiparallel β-sheets thereby disrupting their structure. The interdomain flexibility of simulated hTGF-α structure is greater than in either mEGF or hEGF. The ϕ, ψ dihedrals of hTGF-α occupy two distinct populations of phase space corresponding to either a C or an α-helical conformation. This change in dihedral angle is stabilized by Phe15 with Arg42 and Phe17 with Arg42 N-π weakly polar interactions that are present only in hTGF-α but not in mEGF or hEGF. Proteins 33:396–407, 1998. © 1998 Wiley-Liss, Inc.

Molecular dynamics simulations of epidermal growth factor and transforming growth factor-α structures in water

AMBER v. 4.1 force field in 1.5 ns NPT molecular dynamics simulations of murine epidermal growth factor (mEGF), human epidermal growth factor (hEGF), and human transforming growth factor-alpha (hTGF-alpha) structures with explicit TIP3P solvation were used to investigate differences in backbone stability, changes in secondary structure, interdomain flexibility, and weakly polar interactions. Backbone root mean square deviations of sections of each peptide show that the most stable regions in mEGF and hEGF are the A-, B-, and C-loops, whereas the most stable regions in hTGF-alpha are the A- and B-loops. The secondary structure in the B-loops of mEGF and hEGF differ significantly from the nuclear magnetic resonance (NMR) structures of mEGF and hEGF. The position and type of turns in the B-loop of mEGF and hEGF increase the interstrand distance of the antiparallel beta-sheets thereby disrupting their structure. The interdomain flexibility of simulated hTGF-alpha structure is greater than in either mEGF or hEGF. The phi, psi dihedrals of hTGF-alpha occupy two distinct populations of phase space corresponding to either a Ceq7 or an alpha-helical conformation. This change in dihedral angle is stabilized by Phe15 with Arg42 and Phe17 with Arg42 N-pi weakly polar interactions that are present only in hTGF-alpha but not in mEGF or hEGF.

The three-dimensional structure of Ca(2+)-bound calcyclin: implications for Ca(2+)-signal transduction by S100 proteins.

Calcyclin is a member of the S100 subfamily of EF-hand Ca(2+)-binding proteins. This protein has implied roles in the regulation of cell growth and division, exhibits deregulated expression in association with cell transformation, and is found in high abundance in certain breast cancer cell lines. The novel homodimeric structural motif first identified for apo calcyclin raised the possibility that S100 proteins recognize their targets in a manner that is distinctly different from that of the prototypical EF-hand Ca2+ sensor, calmodulin. The NMR solution structure of Ca(2+)-bound calcyclin has been determined in order to identify Ca(2+)-induced structural changes and to obtain insights into the mechanism of Ca(2+)-triggered target protein recognition. The three-dimensional structure of Ca(2+)-bound calcyclin was calculated with 1372 experimental constraints, and is represented by an ensemble of 20 structures that have a backbone root mean square deviation of 1.9 A for the eight helices. Ca(2+)-bound calcyclin has the same symmetric homodimeric fold as observed for the apo protein. The helical packing within the globular domains and the subunit interface also change little upon Ca2+ binding. A distinct homology was found between the Ca(2+)-bound states of the calcyclin subunit and the monomeric S100 protein calbindin D9k. Only very modest Ca(2+)-induced changes are observed in the structure of calcyclin, in sharp contrast to the domain-opening that occurs in calmodulin and related Ca(2+)-sensor proteins. Thus, calcyclin, and by inference other members of the S100 family, must have a different mode for transducing Ca2+ signals and recognizing target proteins. This proposal raises significant questions concerning the purported roles of S100 proteins as Ca2+ sensors.

Role of gamma-carboxyglutamic acid in the calcium-induced structural transition of conantokin G, a conotoxin from the marine snail Conus geographus.

Conantokin G is a gamma-carboxyglutamic acid- (Gla-) containing conotoxin isolated from the venom of the marine cone snail Conus geographus. This 17-residue polypeptide, which contains five gamma-carboxyglutamic acid residues, is a N-methyl-d-aspartate- (NMDA-) type glutamate receptor antagonist. To investigate the role of gamma-carboxyglutamic acid in the calcium-induced structural transition of conantokin G, we determined the three-dimensional structure of the conantokin G/Ca2+ complex by two-dimensional 1H NMR spectroscopy and compared it to the high-resolution structure of conantokin G in the absence of metal ions [Rigby et al. (1997) Biochemistry 36, 6906]. Complete resonance assignments were made by two dimensional 1H NMR spectroscopy at pH 5.6 in the presence of saturating amounts of Ca2+. Distance geometry and simulated annealing methods were used to derive 23 convergent structures from a set of 302 interproton distance restraints and two torsion angle measurements. A high-resolution structure, with the backbone root mean square deviation to the geometric average of the 23 structures of 0.6 +/- 0.1 A, contains a linear alpha-helix from Gla 3 to Lys 15. Gla residues 3, 7, 10, and 14 are aligned in a linear array on one face of the helix. A genetic algorithm was applied to determine the calcium positions in conantokin G, and the conantokin G/Ca2+ complex refined by molecular simulation. Upon binding of Ca2+ to gamma-carboxyglutamic acid, conantokin G undergoes a conformational transition from a distorted curvilinear 310 helix to a linear alpha-helix. Occupancy of the metal binding sites, defined by gamma-carboxyglutamic acids, results in formation of a calcium-carboxylate network that linearizes the helix and exposes the hydrophobic amino acids on the opposite face of the helix.

Nuclear magnetic resonance studies of the C-terminal human growth hormone fragment I179-C182-[SS]-C189-P191 and the related trisulfide peptide I179-C182-[SSS]-C189-P191.

The synthetic C-terminal hGH fragment I179-C182-[SS]-C189-P191 and the related trisulfide peptide I179-C182-[SSS]-C189-P191 have been studied using homonuclear 1H-NMR methods and distance geometry calculations. The 1H-NMR spectra of both the disulfide (diS) and the trisulfide (triS) were completely assigned. Amide proton exchange rates, NOEs and the temperature dependence of the NH chemical shifts indicate a hydrogen bond in triS between Val185 and Ser188 stabilizing a turn in this region. 3JH,H coupling constants and NOEs were measured and used as input for distance geometry calculations. For triS two families of structures with averaged pairwise backbone root mean square deviations for Cys182-Cys189 of 1.3-1.5 A were found, only one of which is compatible with experimental data. For diS only one family of structures was found, but with such a low structural definition (back bone rmsd > 2 A) that no interpretation into a consensus structure is useful. The generated structures were compared to the crystal structure of the terminal loop in hGH, complexed to its binding proteins. The resemblance was low between the solution structures of the tridecapeptides and the terminal hGH loop.

A method for predicting protein structure from sequence.

The ability to predict the native conformation of a globular protein from its amino-acid sequence is an important unsolved problem of molecular biology. We have previously reported a method in which reduced representations of proteins are folded on a lattice by Monte Carlo simulation, using statistically-derived potentials. When applied to sequences designed to fold into four-helix bundles, this method generated predicted conformations closely resembling the real ones. We now report a hierarchical approach to protein-structure prediction, in which two cycles of the above-mentioned lattice method (the second on a finer lattice) are followed by a full-atom molecular dynamics simulation. The end product of the simulations is thus a full-atom representation of the predicted structure. The application of this procedure to the 60 residue, B domain of staphylococcal protein A predicts a three-helix bundle with a backbone root mean square (rms) deviation of 2.25-3 A from the experimentally determined structure. Further application to a designed, 120 residue monomeric protein, mROP, based on the dimeric ROP protein of Escherichia coli, predicts a left turning, four-helix bundle native state. Although the ultimate assessment of the quality of this prediction awaits the experimental determination of the mROP structure, a comparison of this structure with the set of equivalent residues in the ROP dime- crystal structure indicates that they have a rms deviation of approximately 3.6-4.2 A. Thus, for a set of helical proteins that have simple native topologies, the native folds of the proteins can be predicted with reasonable accuracy from their sequences alone. Our approach suggest a direction for future work addressing the protein-folding problem.