Are free energy calculations useful in practice? A comparison with rapid scoring functions for the p38 MAP kinase protein system.

Precise thermodynamic integration free energy simulations have been applied to a congeneric series of 16 inhibitors to the p38 MAP kinase protein for which the experimental binding data (IC(50)) is known. The relative free energy of binding for each compound has been determined. For comparison, the same series of compounds have also scored using the best rapid scoring functions used in database screening. From the results of these calculations, we find (1) that precise free energy simulations allow predictions that are reliable and in good agreement with experiment; (2) that predictions of lower reliability, but still in good qualitative agreement with experiment, can be obtained using the OWFEG free energy grid method, at a much lower computational cost; (3) and that other methods, not based on free energy simulations yield results in much poorer agreement with experiment. A new predictive index, which measures the reliability of a prediction method in the context of normal use, is defined and calculated for each scoring method. Predictive indices of 0.84, 0.56, 0.04, -0.05, and 0.25 are calculated for thermodynamic integration, OWFEG, ChemScore, PLPScore, and Dock Energy Score, respectively, where +1.0 is perfect correct prediction, -1.0 is perfect incorrect prediction, and 0.0 is random.

Improved scoring of ligand-protein interactions using OWFEG free energy grids.

A new approach to rapidly score protein-ligand interactions is tested on several protein-ligand systems. Results using this approach - the OWFEG free energy grid - are quite promising and are generally in better agreement with experiment (in some cases much better) than those obtained employing scoring techniques currently in wide use. The OWFEG free energy grid is generated from a one-window free energy perturbation MD simulation (Pearlman, D. A. J. Med. Chem. 1999, 42, 4313-4324). The OWFEG approach is applied to three protein systems: IMPDH, MAP kinase p38, and HIV-1 aspartyl protease. OWFEG scores are compared to experimental K(i) and IC50 data in each case. Empirical scoring functions applied to the same systems for comparison include ChemScore, Piecewise Linear Potential (PLP), and Dock energy score.

The structures of caspases-1, -3, -7 and -8 reveal the basis for substrate and inhibitor selectivity.

BACKGROUND: Peptide inhibitors of caspases have helped define the role of these cysteine proteases in biology. Structural and biochemical characterization of the caspase enzymes may contribute to the development of new drugs for the treatment of caspase-mediated inflammation and apoptosis. RESULTS: The crystal structure of the previously unpublished caspase-7 (Csp7; 2.35 A) bound to the reversible tetrapeptide aldehyde inhibitor acetyl-Asp-Glu-Val-Asp-CHO is compared with crystal structures of caspases-1 (2.3 A), -3 (2.2 A), and -8 (2.65 A) bound to the same inhibitor. Csp7 is a close homolog of caspase-3 (Csp3), and these two caspases possess some quarternary structural characteristics that support their unique role among the caspase family. However, although Csp3 and Csp7 are quite similar overall, they were found to have a significantly different substitution pattern of amino acids in and around the S4-binding site. CONCLUSIONS: These structures span all three caspase subgroups, and provide a basis for inferring substrate and inhibitor binding, as well as selectivity for the entire caspase family. This information will influence the design of selective caspase inhibitors to further elucidate the role of caspases in biology and hopefully lead to the design of therapeutic agents to treat caspase-mediated diseases, such as rheumatoid arthritis, certain neurogenerative diseases and stroke.

Consensus scoring: A method for obtaining improved hit rates from docking databases of three-dimensional structures into proteins.

We present the results of an extensive computational study in which we show that combining scoring functions in an intersection-based consensus approach results in an enhancement in the ability to discriminate between active and inactive enzyme inhibitors. This is illustrated in the context of docking collections of three-dimensional structures into three different enzymes of pharmaceutical interest: p38 MAP kinase, inosine monophosphate dehydrogenase, and HIV protease. An analysis of two different docking methods and thirteen scoring functions provides insights into which functions perform well, both singly and in combination. Our data shows that consensus scoring further provides a dramatic reduction in the number of false positives identified by individual scoring functions, thus leading to a significant enhancement in hit-rates.

N-(2-Benzoylphenyl)-L-tyrosine PPARgamma agonists. 2. Structure-activity relationship and optimization of the phenyl alkyl ether moiety.

We previously reported the identification of (2S)-((2-benzoylphenyl)amino)-3-¿4-[2-(5-methyl-2-phenyloxazol-4-y l)e thoxy]phenyl¿propanoic acid (2) (PPARgamma pKi = 8.94, PPARgamma pEC50 = 9.47) as a potent and selective PPARgamma agonist. We now report the expanded structure-activity relationship around the phenyl alkyl ether moiety by pursuing both a classical medicinal chemistry approach and a solid-phase chemistry approach for analogue synthesis. The solution-phase strategy focused on evaluating the effects of oxazole and phenyl ring replacements of the 2-(5-methyl-2-phenyloxazol-4-yl)ethyl side chain of 2 with several replacements providing potent and selective PPARgamma agonists with improved aqueous solubility. Specifically, replacement of the phenyl ring of the phenyloxazole moiety with a 4-pyridyl group to give 2(S)-((2-benzoylphenyl)amino)-3-¿4-[2-(5-methyl-2-pyridin-4-yloxazol+ ++- 4-yl)ethoxy]phenyl¿propionic acid (16) (PPARgamma pKi = 8.85, PPARgamma pEC50 = 8.74) or a 4-methylpiperazine to give 2(S)-((2-benzoylphenyl)amino)-3-(4-¿2-[5-methyl-2-(4-methylpiperazin+ ++- 1-yl)thiazol-4-yl]ethoxy¿phenyl)propionic acid (24) (PPARgamma pKi = 8.66, PPARgamma pEC50 = 8.89) provided two potent and selective PPARgamma agonists with increased solubility in pH 7.4 phosphate buffer and simulated gastric fluid as compared to 2. The second strategy took advantage of the speed and ease of parallel solid-phase analogue synthesis to generate a more diverse set of phenyl alkyl ethers which led to the identification of a number of novel, high-affinity PPARgamma ligands (PPARgamma pKi's 6.98-8.03). The combined structure-activity data derived from the two strategies provide valuable insight on the requirements for PPARgamma binding, functional activity, selectivity, and aqueous solubility.

N-(2-Benzoylphenyl)-L-tyrosine PPARgamma agonists. 3. Structure-activity relationship and optimization of the N-aryl substituent.

3-¿4-[2-(Benzoxazol-2-ylmethylamino)ethoxy]phenyl¿-(2S)-((2- benzoylph enyl)amino)propionic acid (1) and (2S)-((2-benzoylphenyl)amino)-3-¿4-[2-(5-methyl-2-phenyloxazol-4-y l)e thoxy]phenyl¿propionic acid (2) are peroxisome proliferator-activated receptor gamma (PPARgamma) agonists and have antidiabetic activity in rodent models of type 2 diabetes. As part of an effort to develop the SAR of the N-2-benzoylphenyl moiety of 1 and 2, a series of novel carboxylic acid analogues, 23-66, modified only in the N-2-benzoylphenyl moiety were synthesized from L-tyrosine and evaluated as PPARgamma agonists. In general, only modest changes in the N-2-benzoylphenyl moiety of 1 and 2 are tolerated. More specifically, the best changes involve bioisosteric replacement of one of the two phenyl rings of this moiety. Addition of substituents to this moiety generally produced compounds that are less active in the cell-based functional assays of PPARgamma activity although binding affinity to PPARgamma may be maintained. A particularly promising set of analogues is the anthranilic acid esters 63-66 in which the phenyl ring in the 2-benzoyl group of 1 and 2 has been replaced by an alkoxy group. In particular, (S)-2-(1-carboxy-2-¿4-[2-(5-methyl-2-phenyloxazol-4-yl)ethoxy]phen yl¿ ethylamino)benzoic acid methyl ester (63) has a pKi of 8.43 in the binding assay using human PPARgamma ligand binding domain and a pEC50 of 9.21 in the in vitro murine lipogenesis functional assay of PPARgamma activity. Finally, 63 was found to normalize glycemia when dosed at 3 mg/kg bid po in the Zucker diabetic fatty rat model of type 2 diabetes.

Potent dipeptide inhibitors of the pp60c-src SH2 domain.

The design, synthesis, and evaluation of dipeptide analogues as ligands for the pp60c-src SH2 domain are described. The critical binding interactions between Ac-Tyr-Glu-N(n-C5H11)2 (2) and the protein are established and form the basis for our structure-based drug design efforts. The effects of changes in both the C-terminal (11-27) and N-terminal (51-69) portions of the dipeptide are explored. Analogues with reduced overall charge (92-95) are also investigated. We demonstrate the feasibility of pairing structurally diverse subunits in a modest dipeptide framework with the goal of increasing the druglike attributes without sacrificing binding affinity.

The formation of a covalent complex between a dipeptide ligand and the src SH2 domain.

The X-ray crystal structure of the src SH2 domain revealed the presence of a thiol residue (Cys 188) located proximal to the phosphotyrosine portion of a dipeptide ligand. An aldehyde bearing ligand (1) was designed to position an electrophilic carbonyl group in the vicinity of the thiol. X-ray crystallographic and NMR examination of the complex formed between (1) and the src SH2 domain revealed a hemithioacetal formed by addition of the thiol to the aldehyde group with an additional stabilizing hydrogen bond between the acetal hydroxyl and a backbone carbonyl.

Peptide ligands of pp60(c-src) SH2 domains: a thermodynamic and structural study.

Thermodynamic measurements, structural determinations, and molecular computations were applied to a series of peptide ligands of the pp60(c-src) SH2 domain in an attempt to understand the critical binding determinants for this class of molecules. Isothermal titration calorimetry (ITC) measurements were combined with structural data derived from X-ray crystallographic studies on 12 peptide-SH2 domain complexes. The peptide ligands studied fall into two general classes: (1) dipeptides of the general framework N-acetylphosphotyrosine (or phosphotyrosine replacement)-Glu or methionine (or S-methylcysteine)-X, where X represents a hydrophobic amine, and (2) tetra- or pentapeptides of the general framework N-acetylphosphotyrosine-Glu-Glu-Ile-X, where X represents either Glu, Gln, or NH2. Dipeptide analogs which featured X as either hexanolamine or heptanolamine were able to pick up new hydrogen bonds involving their hydroxyl groups within a predominantly lipophilic surface cavity. However, due to internal strain as well as the solvent accessibility of the new hydrogen bonds formed, no net increase in binding affinity was observed. Phosphatase-resistant benzylmalonate and alpha,alpha-difluorobenzyl phosphonate analogs of phosphotyrosine retained some binding affinity for the pp60(c-src) SH2 domain but caused local structural perturbations in the phosphotyrosine-binding site. In the case where a reversible covalent thiohemiacetal was formed between a formylated phosphotyrosine analog and the thiol side chain of Cys-188, deltaS was 25.6 cal/(mol K) lower than for the nonformylated phosphotyrosine parent. Normal mode calculations show that the dramatic decrease in entropy observed for the covalent thiohemiacetal complex is due to the inability of the phosphotyrosine moiety to transform lost rotational and translational degrees of freedom into new vibrational modes.

Identification of peroxisome proliferator-activated receptor ligands from a biased chemical library.

BACKGROUND: The peroxisome proliferator-activated receptors (PPARs) were cloned as orphan members of the nuclear receptor superfamily of transcription factors. The identification of subtype-selective ligands for PPARalpha and PPARgamma has led to the discovery of their roles in the regulation of lipid metabolism and glucose homeostasis. No subtype-selective PPARdelta ligands are available and the function of this subtype is currently unknown. RESULTS: A three-component library was designed in which one of the monomers was biased towards the PPARs and the other two monomers were chosen to add chemical diversity. Synthesis and screening of the library resulted in the identification of pools with activity on each of the PPAR subtypes. Deconvolution of the pools with the highest activity on PPARdelta led to the identification of GW 2433 as the first high-affinity PPARdelta ligand. [3H]GW 2433 is an effective radioligand for use in PPARdelta competition-binding assays. CONCLUSIONS: The synthesis of biased chemical libraries is an efficient approach to the identification of lead molecules for members of sequence-related receptor families. This approach is well suited to the discovery of small-molecule ligands for orphan receptors.

Synthesis and molecular modeling of 1-phenyl-1,2,3,4-tetrahydroisoquinolines and related 5,6,8,9-tetrahydro-13bH-dibenzo[a,h]quinolizines as D1 dopamine antagonists.

New 1-phenyl-1,2,3,4-tetrahydroisoquinolines and related 5,6,8,9-tetrahydro- 13bH-dibenzo[a,h]-quinolizines were prepared as ring-contracted analogs of the prototypical 1-phenyl-2,3,4,5-tetrahydrobenzazepines (e.g., SCH23390) as a continuation of our studies to characterize the antagonist binding pharmacophore of the D1 dopamine receptor. Receptor affinity was assessed by competition for [3H]SCH23390 binding sites in rat striatal membranes. The 6-bromo-1-phenyltetrahydroisoquinoline analog 2 of SCH23390 1 had D1 binding affinity similar to that for the previously reported 6-chloro analog 6, whereas the 6,7-dihydroxy analog 5 had significantly lower D1 affinity. Conversely, neither 6-monohydroxy- (3) nor 7-monohydroxy-1-phenyltetrahydroisoquinolines (4) had significant affinity for the D1 receptor. These results demonstrate that 6-halo and 7-hydroxy substituents influence D1 binding affinity of the 1-phenyltetrahydroisoquinolines in a fashion similar to their effects on 1-phenyltetrahydrobenzazepines. The conformationally constrained 3-chloro-2-hydroxytetrahydrodibenzoquinolizine 9 had much lower affinity relative to the corresponding, and more flexible, 6-chloro-7-hydroxy-1-phenyltetrahydroisoquinoline 6. Similarly, 2,3-dihydroxytetrahydrodibenzoquinolizine 10 had much lower D1 affinity compared to dihydrexidine 14, a structurally similar hexahydrobenzo[a]phenanthridine that is a high-affinity full D1 agonist. Together, these data not only confirm the effects of the halo and hydroxy substitutents on the parent nucleus but demonstrate the pharmacophoric importance of both the nitrogen position and the orientation of the accessory phenyl ring in modulating D1 receptor affinity and function. Molecular modeling studies and conformational analyses were conducted using the data from these new analogs in combination with the data from compounds previously synthesized. The resulting geometries were used to refine a working model of the D1 antagonist pharmacophore using conventional quantitative structure-activity relationships and three-dimensional QSAR (CoMFA).

Peptide inhibitors of src SH3-SH2-phosphoprotein interactions.

Activated pp60c-src has been implicated in a number of human malignancies including colon carcinoma and breast adenocarcinoma. Association of the src SH2 domain with tyrosine-phosphorylated proteins plays a role in src-mediated signal transduction. Inhibitors of src SH2 domain-phosphoprotein interactions are, thus, of great interest in defining the role(s) of src in signal transduction pathways. To facilitate such studies, an enzyme-linked immunosorbent assay (ELISA) was developed to detect inhibitors of src SH2-phosphoprotein interactions. This assay measures inhibition of binding of a fusion construct (glutathione S-transferase src SH3-SH2) with autophosphorylated epidermal growth factor receptor tyrosine kinase domain. Activities of phosphopeptide segments derived from potential src SH2 cognate phosphoprotein partners were determined, with the focal adhesion kinase-derived segment VSETDDY*AEIIDE yielding the highest inhibitory activity. Structure activity studies starting from acetyl (Ac)-Y*EEIE have identified Ac-Y*Y*Y*IE as the most active compound screened in the ELISA. This compound is at least 20-fold more active than the parent peptide Ac-Y*EEIE. A high resolution (2 A) crystal structure of human src SH2 complexed with Ac-Y*EEIE was obtained and provided a useful framework for understanding the structure-activity relationships. Additionally, Ac-Y*EEIE was able to block interactions between src and its cellular phosphoprotein partners in vanadate-treated cell lysates from MDA-MB-468 breast carcinoma cells. However, it is unable to abrogate proliferation of MDA-MB-468 cells in culture, presumably because of poor cell penetration and/or lability of the phosphate group on tyrosine.

Conformational analysis, pharmacophore identification, and comparative molecular field analysis of ligands for the neuromodulatory sigma 3 receptor.

Molecular modeling studies were carried out on a series of 1-phenyl-3-amino-1,2,3,4-tetrahydronaphthalenes (phenylaminotetralins, PATs), several PAT structural analogs, and various non-PAT ligands that demonstrate a range of affinities for a novel sigma 3 receptor linked to stimulation of tyrosine hydroxylase and dopamine synthesis in rodent brain. In an effort to develop a ligand-binding model for the sigma 3 receptor, a pharmacophore mapping program (DISCO) was used to identify structural features that are common to ligands that exhibit moderate to high binding affinity for sigma 3 sites. DISCO then was utilized to propose a common pharmacophoric region that included one low-energy conformation of each compound in the training set. The resulting alignment was utilized in a comparative molecular field analysis (CoMFA) study in an attempt to correlate the steric and electrostatic fields of the molecules with the respective binding affinities at the sigma 3 receptor. A suitably predictive model was obtained from the CoMFA analysis which will be employed in the development of additional PAT analogs that could potentially display high affinity and selectivity for the sigma 3 receptor. The excluded volumes which resulted from comparing molecular volumes of active and inactive compounds were visualized to examine the limits of steric tolerance imposed by the sigma 3 receptor.

Role of glutamine-61 in the hydrolysis of GTP by p21H-ras: an experimental and theoretical study.

The active GTP-bound form of p21ras is converted to the biologically inactive GDP-bound form by enzymatic hydrolysis and this function serves to regulate the wild-type ras protein. The side chain of the amino acid at position 61 may play a key role in this hydrolysis of GTP by p21. Experimental studies that define properties of the Q61E mutant of p21H-ras are presented along with supporting molecular dynamics simulations. We find that under saturating concentrations of GTP the Q61E mutant of p21H-ras has a 20-fold greater rate of intrinsic hydrolysis (kcat = 0.57 min-1) than the wild type. The affinity of the Q61E variant for GTP (Kd = 115 microM) is much lower than that of the wild type. GTPase activating protein does not activate the variant. From molecular dynamics simulations, we find that both the wild type and Q61E mutant have the residue 61 side chain in transient contact with a water molecule that is well-positioned for hydrolytic attack on the gamma phosphate. Thr-35 also is found to form a transient hydrogen bond with this critical water. These elements may define the catalytic complex for hydrolysis of the GTP [Pai et al. (1990) EMBO J. 9, 2351]. Similarly, the G12P mutant, which also has an intrinsic hydrolysis rate similar to the wild type, is found to form the same complex in simulation. In contrast, molecular dynamics analysis of the mutants G12R, G12V, and Q61L, which have much lower intrinsic rates than the wild-type p21, do not show this complex.(ABSTRACT TRUNCATED AT 250 WORDS)

Computer-assisted molecular modeling: indispensable tools for molecular pharmacology.

Ball-and-stick mechanical models, typically associated with chemists, have been helpful in understanding structural problems and the relationship between structure and biologic activity. With progress in computer speed, graphics performance, and software innovation, molecules of biological interest can be subjected to rigorous calculations. Computational chemistry and biology are rooted in the belief that theoretical physics can be used to calculate accurate molecular structures. Although in its infancy, computer-assisted molecular modeling is gaining attention and acceptability as an increasing number of researchers turn their attention toward rational molecular design. The trend to use theoretical methods can be traced to the greater availability of computer graphics work-stations, decreasing computer costs, faster central processing units, more robust algorithms, and "user-friendly" software codes. Every major pharmaceutical company has invested in these resources to reduce the time it takes to design and develop pharmaceutical agents. Because of the vast financial and manpower investments needed to introduce a single drug, medicinal chemists and pharmacologists are interested in understanding and predicting drug action at the molecular level. Although drug action is still poorly understood, molecular modeling should reduce some of the labor in the development of pharmaceutical agents.

Molecular dynamics simulation of bovine prothrombin fragment 1 in the presence of calcium ions.

Early solvation-induced structural reorganization of calcium prothrombin fragment 1 is simulated with molecular dynamics. Initial coordinates are those of the 2.2-A resolution crystal structure [Soriano-Garcia, M., Padmanabhan, K., de Vos, A. M., & Tulinsky, A. (1992) Biochemistry 31, 2554-2556]. The molecular dynamics code AMBER, appropriately modified to include long-range (less than or equal to 22.0 A) ionic forces, was employed. The solution structure appears to equilibrate within 100 ps. Although minor changes are seen in various structural domains, the early solution structure basically maintains an intricate network of nine gamma-carboxyglutamic acid (Gla) residues encapsulating seven calcium ions. However, the Gla domain moves with respect to the kringle domain. This motion is mainly due to the movement of Ser34-Leu35 that appears to be a flexible hinge between the domains. The N-terminus of Ala 1 is in a tightly bound complex with three Gla residues that remains stable in the solution structure when the long-range electrostatic cutoff is employed and the near planar alignment of the seven calcium ions is only slightly distorted. The simulation structure is discussed in terms of experiments that studied calcium ion-induced quenching of the intrinsic fluorescence, protection of the N-terminal amino group from acetylation by calcium ions, chemical modification of the N-terminus to a trinitrophenyl derivative, and the possibility of a calcium-binding site(s) in the kringle domain.

Simulation of the solution structure of the H-ras p21-GTP complex.

An unconstrained simulation of the GTP-bound form of the H-ras protein p21 is performed in an aqueous environment with charge-neutralizing counterions. The simulation is compared to the 1.35-A structure of Pai et al. [(1990) EMBO J. 9, 2351] and a proposed alternate structure, in which the loop at residues 60-65 is modeled into a form which may activate a water molecule for the GTP hydrolysis. The simulation suggests that some protein intermolecular H-bond contacts which are present in the crystal structure are lost in the solvation process and this loss may lead to localized refolding of the molecule. For instance, we find that the gamma-phosphate of the GTP has somewhat weaker contact with the protein in the simulation structure. The antiparallel beta-sheet (residues 38-57) partially melts. The 60-65 loop, which is hypervariable in the X-ray study, is initially relatively distant from the gamma-phosphate region. However, this loop moves so as to sample the space around the gamma-phosphate. For a significant fraction of the simulation time, forms similar to the alternate structure are observed, and a water molecule is localized near the hydrolytic site. The molecular dynamics simulations of p21-GTP in solution support a postulated hydrolysis mechanism for the biological inactivation of the nucleotide complex based on crystallographic data.

Expression and characterization of human factor IX. Factor IXthr-397 and factor IXval-397.

Factor IXLong Beach has a single amino acid substitution at 397 (Ile to Thr) in the catalytic domain which results in severe hemophilia B. Recent investigations have shown that the substitution of threonine for isoleucine at 397 may affect a part of the macromolecular substrate binding site. Because threonine has a hydroxyl group in its side chain, it is possible that this hydroxyl group makes new hydrogen bonds and disturbs the substrate binding site. We used three techniques: molecular biology, which includes site-directed mutagenesis and recombinant protein expression in tissue culture; computer-aided kinetic data analysis; and molecular modeling to study this mutation site. We have produced two mutant factor IX molecules that have isoleucine 397 replaced by valine or threonine. Factor IXwild type and the two mutants (factor IXVal and factor IXThr) were expressed in human kidney cells and purified using a conformation-specific monoclonal antibody column. After the activation by factor XIa, these three molecules were able to bind p-aminobenzamidine and increase its fluorescence intensity in a similar manner. Factor IXVal and factor IXwild type had indistinguishable activities in an activated partial thromboplastin time (aPTT) assay and similar kinetic parameters with factor X as a substrate. Factor IXThr had only 5% clotting activity compared with normal factor IX, a slightly lower Km and significantly reduced kcat, using factor X as a substrate. We developed energy-refined (AMBER v.3.1) computer models of the three factor IX molecules based on previous work. Three factor IXa models (Ile, Val, or Thr at 397) with a fragment of the factor X activation site were used to predict the effect of the mutation at 397 and evaluate the significance of the new hydrogen bond thought to form between the side chain hydroxyl group of threonine 397 and the carbonyl oxygen of tryptophan 385. This new hydrogen bond would affect the position of an amide proton of adjacent glycine 386 which has been proposed to make a hydrogen bond with a backbone carbonyl oxygen of the P3 residue of factor X. In addition to the new hydrogen bond, there is significant movement in the side chain of tryptophan 385 between the factor IXawild type-factor X model and the factor IXaThr-factor X model that could interfere with substrate binding. This movement could be caused by the change in the molecular volume, the orientation of the side chain at 397, and the new hydrogen bond.

Solution conformations of the gamma-carboxyglutamic acid domain of bovine prothrombin fragment 1, residues 1-65.

Molecular dynamics simulations have been performed (AMBER version 3.1) on solvated residues 1-65 of bovine prothrombin fragment 1 (BF1) by using the 2.8-A resolution crystallographic coordinates as the starting conformation for understanding calcium ion-induced conformational changes that precede experimentally observable phospholipid binding. Simulations were performed on the non-metal-bound crystal structure, the form resulting from addition of eight calcium ions to the 1-65 region of the crystal structure, the form resulting from removal of calcium ions after 107 ps and continuing the simulation, and an isolated hexapeptide loop (residues 18-23). In all cases, the 100-ps time scale seemed adequate to sample an ensemble of solution conformers within a particular region of conformation space. The non-metal-containing BF1 did not unfold appreciably during a 106-ps simulation starting from the crystallographic geometry. The calcium ion-containing structure (Ca-BF1) underwent an interesting conformational reorganization during its evolution from the crystal structure: during the time course of a 107-ps simulation, Ca-BF1 experienced a trans----cis isomerization of the gamma-carboxyglutamic acid-21 (Gla-21)-Pro-22 peptide bond. Removal of the calcium ions from this structure followed by 114 ps of additional molecular dynamics showed significant unfolding relative to the final 20-ps average structure of the 107-ps simulation; however, the Gla-21-Pro-22 peptide bond remained cis. A 265-ps simulation on the termini-protected hexapeptide loop (Cys-18 to Cys-23) containing two calcium ions also did not undergo a trans----cis isomerization. It is believed that the necessary activation energy for the transitional event observed in the Ca-BF1 simulation was largely supplied by global conformational events with a possible assist from relief of intermolecular crystal packing forces. The presence of a Gla preceding Pro-22, the inclusion of Pro-22 in a highly strained loop structure, and the formation of two long-lived salt bridges prior to isomerization may all contribute to this finding.