Computational drug repositioning: from data to therapeutics.

Traditionally, most drugs have been discovered using phenotypic or target-based screens. Subsequently, their indications are often expanded on the basis of clinical observations, providing additional benefit to patients. This review highlights computational techniques for systematic analysis of transcriptomics (Connectivity Map, CMap), side effects, and genetics (genome-wide association study, GWAS) data to generate new hypotheses for additional indications. We also discuss data domains such as electronic health records (EHRs) and phenotypic screening that we consider promising for novel computational repositioning methods.

Mammalian Chk2 is a downstream effector of the ATM-dependent DNA damage checkpoint pathway.

In response to DNA damage and replication blocks, cells activate pathways that arrest the cell cycle and induce the transcription of genes that facilitate repair. In mammals, ATM (ataxia telangiectasia mutated) kinase together with other checkpoint kinases are important components in this response. We have cloned the rat and human homologs of Saccharomyces cerevisiae Rad 53 and Schizosaccharomyces pombe Cds1, called checkpoint kinase 2 (chk2). Complementation studies suggest that Chk2 can partially replace the function of the defective checkpoint kinase in the Cds1 deficient yeast strain. Chk2 was phosphorylated and activated in response to DNA damage in an ATM dependent manner. Its activation in response to replication blocks by hydroxyurea (HU) treatment, however, was independent of ATM. Using mass spectrometry, we found that, similar to Chk1, Chk2 can phosphorylate serine 216 in Cdc25C, a site known to be involved in negative regulation of Cdc25C. These results suggest that Chk2 is a downstream effector of the ATM-dependent DNA damage checkpoint pathway. Activation of Chk2 might not only delay mitotic entry, but also increase the capacity of cultured cells to survive after treatment with gamma-radiation or with the topoisomerase-I inhibitor topotecan.

Selective inhibition of Abeta fibril formation.

We describe here an inhibitor of in vitro fibril formation, hexadecyl-N-methylpiperidinium (HMP) bromide, which is selective for the Alzheimer's disease peptide Abeta. At 10 microM, its IC50 for inhibiting Abeta aggregation at pH 5.8, HMP bromide does not inhibit fibril formation by other amyloidogenic polypeptides nor does it affect the folding stability of the beta-sheet-rich immunoglobulin VL domain REI. In addition, small structural modifications of HMP bromide reduce or eliminate its ability to inhibit pH 5.8 aggregation of Abeta. These indications of specificity, plus the ability of the molecule to inhibit A beta aggregation at concentrations almost an order of magnitude below its critical micelle concentration, suggest a mechanism of inhibition other than micellar solubilization of Abeta. HMP bromide is required in approximately a 1:1 stoichiometry for effective inhibition at pH 5.8. Although stoichiometric amounts of HMP bromide with respect to total Abeta inhibit Abeta fibril formation at pH 7.4, the molecule is incapable, at lower concentrations, of blocking the seeding of fibril formation by small amounts of added Abeta fibrils. The results suggest the existence of a binding surface on A beta capable of binding amphipathic molecules such as HMP bromide and which, when occupied, precludes assembly of A beta into amyloid fibrils. Molecules that bind to this site with high specificity may prove to be useful therapeutic agents for preventing or retarding the cerebral amyloid plaque formation implicated in Alzheimer's disease pathology.

Differential recognition by CD28 of its cognate counter receptors CD80 (B7.1) and B70 (B7.2): analysis by site directed mutagenesis.

CD28, which is a member of the immunoglobulin superfamily of molecules (IgSF), is a homodimer of two polypeptides containing a single V-like domain with short transmembrane and cytoplasmic regions. It serves as a co-signalling molecule for T cell activation through binding to its cognate counter-receptors CD80 and B70, expressed on antigen presenting cells. In the current study, we investigated the regions of CD28 which are involved in its interactions with CD80 and B70, using site directed mutagenesis, CD28 mAb epitope mapping, receptor based adhesion assays and direct binding of Ig-fusion proteins to cell surface receptors. Truncation or substitution of a stretch of a proline rich "hallmark" sequence, "MYPPPY", abrogates binding to CD80 or B70, while retaining CD28 mAb epitopes and cell surface expression. On an Ig-fold model of the CD28 V-domain, this fully conserved motif localizes to a CDR3-like region. Mutations introduced into other loops, including the CDRI-like and CDR2-like regions, had very little effect on CD80 or B70 binding. Mutations introduced within the predicted beta-strand regions caused loss of receptor expression. Conservative substitution of both the flanking tyrosine residues within the "MYPPPY" motif with phenylalanine, caused loss of binding to B70 but not to CD80. These results show that, although the same overall region on CD28 may be involved in the interactions with CD80 and B70, subtle but important differences distinguish recognition by the two molecules. These finding, along with previous observations on the differential pattern of expression and tissue distribution of CD80 and B70, support the contention that these molecules play distinct roles in the regulation of immune responses in vivo.

Prolines and amyloidogenicity in fragments of the Alzheimer's peptide beta/A4.

Although it is well accepted that the structure of amyloid fibrils is dominated by some form of antiparallel beta-sheet, there are few details on the secondary structural arrangements of the constituent peptides and how these peptides pack together in the fibril. We describe here the use of scanning proline mutagenesis to map the secondary structural roles of each residue in amyloidogenic peptide fragments of the Alzheimer's amyloid peptide beta/A4. In two series of fragments related to residues 15-23 and 12-26 of beta/A4, we show that Pro replacement of any residue in the amyloidogenic sequence LVFFAED, corresponding to residues 17-23, leads to essentially complete loss of fibril formation and to excellent peptide solubility. Since peptidyl-prolyl bonds are incapable of forming standard extended chain conformations, the results suggest that residues 17-23 make up the beta-sheet core of the fibrils formed by these fragments. In contrast to the proline replacements, alanine substitutions at residues 17, 18, and 20 have no effect on fibril formation, while replacement of Phe19 reduces fibril formation to 15% of the level found for the wild type sequence. Scanning proline mutagenesis should play a useful role in mapping the secondary structural features of larger amyloidogenic peptide sequences, including longer, physiologically relevant forms of beta/A4. In addition, these results suggest explanations for some amyloidogenic effects observed in disease-related peptides and also suggest a possible role for aggregation-inhibiting insertion of prolines in protein evolution and protein design.

Protein engineering techniques for antibody humanization.

Few proteins have the therapeutic potential of antibodies, which can be developed against a wide variety of targets and genetically or chemically manipulated to further enhance their activities. A number of approaches have been taken in order to render these proteins pharmaceutically useful. Either the amino acid sequence of an antibody can be genetically altered (i.e. reshaped or resurfaced) or the antibody can be conjugated to other proteins or toxins.

A role for destabilizing amino acid replacements in light-chain amyloidosis.

Light-chain (L-chain) amyloidosis is characterized by deposition of fibrillar aggregates composed of the N-terminal L-chain variable region (VL) domain of an immunoglobulin, generally in individuals overproducing a monoclonal L chain. In addition to proteolytic fragmentation and high protein concentration, particular amino acid substitutions may also contribute to the tendency of an L chain to aggregate in L-chain amyloidosis, although evidence in support of this has been limited and difficult to interpret. In this paper we identify particular amino acid replacements at specific positions in the VL domain that are occupied at frequencies significantly higher in those L chains associated with amyloidosis. Analysis of the structural model for the VL domain of the Bence-Jones protein REI suggests that these positions play important roles in maintaining domain structure and stability. Using an Escherichia coli expression system, we prepared single-point mutants of REI VL incorporating amyloid-associated amino acid replacements that are both rare and located at structurally important positions. These mutants support ordered aggregate formation in an in vitro L-chain fibril formation model in which wild-type REI VL remains soluble. Moreover, the ability of these sequences to aggregate in vitro correlates well with the extent to which domain stability is decreased in denaturant-induced unfolding. The results are consistent with a mechanism for the disease process in which the VL domain, either before or after proteolytic cleavage from the L-chain constant region domain, unfolds by virtue of one or more destabilizing amino acid replacements to generate an aggregation-prone nonnative state.

Strong inhibition of fibrinogen binding to platelet receptor alpha IIb beta 3 by RGD sequences installed into a presentation scaffold.

In order to probe the structural constraints on binding of RGD sequences to the platelet receptor alpha IIb beta 3 we have used recombinant DNA techniques to install the RGD sequence into 'presentation scaffolds', small proteins of known 3-D structure chosen to present guest sequences in constrained orientations. Using Escherichia coli expression systems we made sequence variants in which loop residues of the immunoglobulin VL domain REI and of human interleukin-1 beta were replaced (without changing polypeptide length) by the RGD sequence at positions predicted, based on small molecule studies, to orient the RGD moiety into an active conformation. These variants do not compete for fibrinogen binding to alpha IIb beta 3 up to almost 1 mM concentration. Unfolded or proteolytically fragmented forms of these same proteins do compete, however, showing that the RGD sequences in the mutants must be prohibited from binding by constraints imposed by scaffold structure. To suppress the effects of such structural constraints we constructed two sequence variants in which RGD-containing sequences 42-57 or 44-55 from the snake venom platelet antagonist kistrin were inserted (this increasing the length of the loop) into the third complementarity determining loop of REI. Both of these variants compete strongly for fibrinogen binding with IC50s in the nM range. These results, plus data on kistrin-related peptides also presented here, suggest that the molecular scaffold REI is capable of providing to an installed sequence a structural context and conformation beneficial to binding. The results also suggest that in order to bind well to alpha IIb beta 3, RGD sequences in protein ligands must either project significantly from the surface of the scaffold and/or retain a degree of conformational flexibility within the scaffold. Molecular scaffolds like REI should prove useful in the elucidation of structure-function relationships and the discovery of new active sequences, and may also serve as the basis for novel therapeutic agents.

Comparison of solution structures of mutant bovine pancreatic trypsin inhibitor proteins using two-dimensional nuclear magnetic resonance.

Structural perturbations due to a series of mutations at the 30-51 disulfide bond of bovine pancreatic trypsin inhibitor have been explored using NMR. The mutants replaced cysteines at positions 30 and 51 by alanine at position 51 and alanine, threonine, or valine at position 30. Chemical shift changes occur in residues proximate to the site of mutation. NOE assignments were made using an automated procedure, NASIGN, which used information from the wild-type crystal structure. Intensity information was utilized by a distance geometry algorithm, VEMBED, to generate a series of structures for each protein. Statistical analyses of these structures indicated larger averaged structural perturbations than would be expected from crystallographic and other information. Constrained molecular dynamics refinement using AMBER at 900 K was useful in eliminating structural movements that were not a necessary consequence of the NMR data. In most cases, statistically significant movements are shown to be those greater than approximately 1 A. Such movements do not appear to occur between wild type and A30A51, a result confirmed by crystallography (Eigenbrot, C., Randal, M., & Kossiakoff, A.A., 1990, Protein Eng. 3, 591-598). Structural alterations in the T30A51 or V30A51 mutant proteins near the limits of detection occur in the beta-loop (residues 25-28) or C-terminal alpha-helix, respectively.

Confirmation of the predicted source of a slow folding reaction: proline 8 of bovine pancreatic trypsin inhibitor.

A major question in protein structural analysis concerns the applicability of results from model systems to other proteins. Theoretical approaches seem the best manner of transferring information from one system to another, but their accuracy in the model systems must first be tested with results from experiment. Since bovine pancreatic trypsin inhibitor (BPTI) is a model system for the evaluation of energy minimization and molecular dynamics routines, we can use folding and stability measurements to examine the reliability of these methods. All two-disulfide mutants of BPTI investigated thus far have two very slow folding reactions which have characteristics of proline isomerization. These reactions may occur because the non-native cis form of two of the four prolines in BPTI significantly destabilizes the folded state of the protein. Previous energy minimization studies of wild-type BPTI suggested that the cis form of Pro8 was the most destabilizing of the four prolines [Levitt,M. (1981) J. Mol. Biol., 145, 251-263]. In this paper, we show that mutation of Pro8----Gln in the two-disulfide bond mutant Val30/Ala51 results in a loss of the slowest folding reaction, consistent with Levitt's prediction.

Denaturant-dependent folding of bovine pancreatic trypsin inhibitor mutants with two intact disulfide bonds.

The equilibrium and kinetic behavior of the guanidine hydrochloride (Gdn-HCl) induced unfolding/refolding of four bovine pancreatic trypsin inhibitor (BPTI) mutants was examined by using ultraviolet difference spectroscopy. In three of the mutants, we replaced the buried 30-51 disulfide bond with alanine at position 51 and valine (Val30/Ala51), alanine (Ala30/Ala51), or threonine (Thr30/Ala51) at position 30. For the fourth mutant, the solvent-exposed 14-38 disulfide was substituted by a pair of alanines (Ala14/Ala38). All mutants retained the 5-55 disulfide. Experiments were performed under oxidizing conditions; thus, both the unfolded and folded forms retained two native disulfide bonds. Equilibrium experiments demonstrated that all four mutants were destabilized relative to wild-type BPTI. However, the stability of the 30-51 mutants increased with the hydrophobicity of the residue substituted at position 30. Kinetic experiments showed that all four mutants contained two minor slow refolding phases with characteristics of proline isomerization. The specific behavior of the phases depended on the location of the disulfide bonds. The major unfolding/refolding phase for each of the 30-51 mutants was more than an order of magnitude slower than for Ala14/Ala38 or for BPTI in which the 14-38 disulfide bond was specifically reduced and blocked with iodoacetamide [Jullien, M., & Baldwin, R. L. (1981) J. Mol. Biol. 145, 265-280]. Since this effect is independent of the stability of the protein, it is consistent with a model in which the proper docking of the interior residues of the protein is the rate-limiting step in the folding of these mutants.

Multiple replacements at position 211 in the alpha subunit of tryptophan synthase as a probe of the folding unit association reaction.

Equilibrium and kinetic studies on the folding of a series of amino acid replacements at position 211 in the alpha subunit of tryptophan synthase from Escherichia coli were performed in order to determine the role of this position in the rate-limiting step in folding. Previous studies [Beasty, A. M., Hurle, M. R., Manz, J. T., Stackhouse, T., Onuffer, J. J., & Matthews, C. R. (1986) Biochemistry 25, 2965-2974] have shown that the rate-limiting step corresponds to the association/dissociation of the amino (residues 1-188) and carboxy (residues 189-268) folding units. In terms of the secondary structure, the amino folding unit consists of the first six strands and five alpha helices of this alpha/beta barrel protein. The carboxy folding unit comprises the remaining two strands and three alpha helices; position 211 is in strand 7. Replacement of the wild-type glycine at position 211 with serine, valine, and tryptophan at most alters the rate of dissociation of the folding units; association is not changed significantly. In contrast, glutamic acid and arginine dramatically decelerate and accelerate, respectively, both association and dissociation. The difference in effects is attributed to long-range electrostatic interactions for these charged side chains; steric effects and/or hydrogen bonding play lesser roles. When considered with previous data on replacements at other positions in the alpha subunit [Hurle, M. R., Tweedy, N. B., & Matthews, C. R. (1986) Biochemistry 25, 6356-6360], it is clear that beta strands 6 (in the amino folding unit) and 7 (in the carboxy folding unit and containing position 211) dock late in the folding process.

Proline isomerization and the slow folding reactions of the alpha subunit of tryptophan synthase from Escherichia coli.

Previous studies on the refolding of the alpha subunit of tryptophan synthase from Escherichia coli assigned two slow refolding phases to rate-limiting isomerizations of two 'essential' proline residues, one in each of the two domains of the protein (Matthews, C.R., Crisanti, M.M., Manz, J.T. and Gepner, G.L. (1983) Biochemistry 22, 1445-1452). The double-jump experiment (Brandts, J.F., Halvorson, H.R. and Brennan, M. (1975) Biochemistry 14, 4953-4963) was used to further investigate this phenomenon. The reaction assigned to the carboxyl domain is consistent with the proline isomerization hypothesis. The amino domain process is more rapid than expected for proline isomerization and may reflect another type of slow folding reaction. The results permit a further refinement of the folding model for the alpha subunit and demonstrate the existence of a third unfolded species whose folding is not limited by either of these two reactions.

Characterization of a slow folding reaction for the alpha subunit of tryptophan synthase.

The equilibria and kinetics of urea-induced unfolding and refolding of the alpha subunit of tryptophan synthase of E. coli have been examined for their dependences on viscosity, pH, and temperature in order to investigate the properties of one of the rate-limiting steps, domain association. A viscosity enhancer, 0.58 M sucrose, was found to slow unfolding and accelerate refolding. This apparently anomalous result was shown to be due to the stabilizing effect of sucrose on the folding reaction. After accounting for this stabilization effect by using linear free-energy plots, the unfolding and refolding kinetics were found to have a viscosity dependence. A decrease in pH was found to stabilize the domain association reaction by increasing the refolding rate and decreasing the unfolding rate. This effect was accounted for by protonation of a single residue with a pK value of 8.8 in the native state and 7.1 in the intermediate, in which the two domains are not yet associated. The activation energy of unfolding is 4.8 kcal/mol, close to the diffusion limit. The negative activation entropy of unfolding, -47 cal/deg-mol, which controls this reaction, may result from ordering of solvent about the newly exposed domain interface of the transition state. These results may provide information on the types of noncovalent interactions involved in domain association and improve the ability to interpret the folding of mutants with single amino-acid substitutions at the interface.

Prediction of the tertiary structure of the alpha-subunit of tryptophan synthase.

The tertiary structure of the alpha-subunit of tryptophan synthase was proposed using a combination of experimental data and computational methods. The vacuum-ultraviolet circular dichroism spectrum was used to assign the protein to the alpha/beta-class of supersecondary structures. The two-domain structure of the alpha-subunit (Miles et al.: Biochemistry 21:2586, 1982; Beasty and Matthews: Biochemistry 24:3547, 1985) eliminated consideration of a barrel structure and focused attention on a beta-sheet structure. An algorithm (Cohen et al.: Biochemistry 22:4894, 1983) was used to generate a secondary structure prediction that was consistent with the sequence data of the alpha-subunit from five species. Three potential secondary structures were then packed into tertiary structures using other algorithms. The assumption of nearest neighbors from second-site revertant data eliminated 97% of the possible tertiary structures; consideration of conserved hydrophobic packing regions on the beta-sheet eliminated all but one structure. The native structure is predicted to have a parallel beta-sheet flanked on both sides by alpha-helices, and is consistent with the available data on chemical cross-linking, chemical modification, and limited proteolysis. In addition, an active site region containing appropriate residues could be identified as well as an interface for beta 2-subunit association. The ability of experimental data to facilitate the prediction of protein structure is discussed.

Synergism in folding of a double mutant of the alpha subunit of tryptophan synthase.

The urea-induced unfolding of the inactive single mutants Tyr-175----Cys and Gly-211----Glu and the active double mutant Cys-175/Glu-211 of the alpha subunit of tryptophan synthase from Escherichia coli was examined by using ultraviolet difference spectroscopy. Equilibrium techniques were used to determine the equilibrium free energies of unfolding for the mutant proteins to permit comparison with the wild-type protein. The sum of the changes in stability for the single mutants is not equal to the change seen in the double mutant. This inequality is evidence for a structural interaction between these two residues. Kinetic studies show that this synergism, which destabilizes the native form by 1.5-2.0 kcal/mol at pH 7.8, 25 degrees C, occurs only after the final rate-limiting step of domain association.

Effects of the phenylalanine-22----leucine, glutamic acid-49----methionine, glycine-234----aspartic acid, and glycine-234----lysine mutations on the folding and stability of the alpha subunit of tryptophan synthase from Escherichia coli.

The effects of four single amino acid replacements on the stability and folding of the alpha subunit of tryptophan synthase from Escherichia coli have been investigated by ultraviolet differences spectroscopy. In previous studies [Miles, E. W., Yutani, K., & Ogasahara, K. (1982) Biochemistry 21, 2586], it had been shown that the urea-induced unfolding at pH 7.8, 25 degrees C, proceeds by the initial unfolding of the less stable carboxyl domain (residues 189-268) followed by the unfolding of the more stable amino domain (residues 1-188). The effects of the Phe-22----Leu, Glu-49----Met, Gly-234----Asp, and Gly-234----Lys mutants on the equilibrium unfolding process can all be understood in terms of the domain unfolding model. With the exception of the Glu-49----Met replacement, the effects on stability are small. In contrast, the effects of three of the four mutations on the kinetics of interconversion of the native form and one of the stable partially folded intermediates are dramatic. The results for the Phe-22----Leu and Gly-234----Asp mutations indicate that these residues play a key role in the rate-limiting step. The Glu-49----Met mutation increases the stability of the native form with respect to that of the intermediate but does not affect the rate-limiting step. The Gly-234----Lys mutation does not affect either the stability or the kinetics of folding for the transition between native and intermediate forms. The changes in stability calculated from the unfolding and refolding rate constants agree quantitatively with those obtained from the equilibrium data. When considered with the results from a previous study on the Gly-211----Glu replacement [Matthews, C. R., Crisanti, M. M., Manz, J. T., & Gepner G. L. (1983) Biochemistry 22, 1445], it can be concluded that the rate-limiting step in the conversion of the intermediate to the native conformation involves either domain association or some other type of molecule-wide phenomenon.