Structure-based design of a heptavalent anthrax toxin inhibitor

The design of polyvalent molecules, consisting of multiple copies of a biospecific ligand attached to a suitable scaffold, represents a promising approach to inhibit pathogens and oligomeric microbial toxins. Despite the increasing interest in structure-based drug design, few polyvalent inhibitors based on this approach have shown efficacy in vivo. Here we demonstrate the structure-based design of potent biospecific heptavalent inhibitors of anthrax lethal toxin. Specifically, we illustrate the ability to design potent polyvalent ligands by matching the pattern of binding sites on the biological target. We used a combination of experimental studies based on mutagenesis and computational docking studies to identify the binding site for an inhibitory peptide on the heptameric subunit of anthrax toxin. We developed an approach based on copper-catalyzed azide-alkyne cycloaddition (click-chemistry) to facilitate the attachment of seven copies of the inhibitory peptide to a β-cyclodextrin core via a polyethylene glycol linker of an appropriate length. The resulting heptavalent inhibitors neutralized anthrax lethal toxin both in vitro and in vivo and showed appreciable stability in serum. Given the inherent biocompatibility of cyclodextrin and polyethylene glycol, these potent well-defined heptavalent inhibitors show considerable promise as anthrax antitoxins.     

Structure-based design of a pathway-specific nuclear import inhibitor

Kapbeta2 (also called transportin) recognizes PY nuclear localization signal (NLS), a new class of NLS with a R/H/Kx((2-5))PY motif. Here we show that Kapbeta2 complexes containing hydrophobic and basic PY-NLSs, as classified by the composition of an additional N-terminal motif, converge in structure only at consensus motifs, which explains ligand diversity. On the basis of these data and complementary biochemical analyses, we designed a Kapbeta2-specific nuclear import inhibitor, M9M.     

Structure-based design of model proteins

A structure-based, sequence-design procedure is proposed in which one considers a set of decoy structures that compete significantly with the target structure in being low energy conformations. The decoy structures are chosen to have strong overlaps in contacts with the putative native state. The procedure allows the design of sequences with large and small stability gaps in a random-bond heteropolymer model in both two and three dimensions by an appropriate assignment of the contact energies to both the native and nonnative contacts. The design procedure is also successfully applied to the two-dimensional HP model. Proteins 31:10–20, 1998. © 1998 Wiley-Liss, Inc.     

Structure-based design of a potent purine-based cyclin-dependent kinase inhibitor

Aberrant control of cyclin-dependent kinases (CDKs) is a central feature of the molecular pathology of cancer. Iterative structure-based design was used to optimize the ATP- competitive inhibition of CDK1 and CDK2 by O(6)-cyclohexylmethylguanines, resulting in O(6)-cyclohexylmethyl-2-(4'- sulfamoylanilino)purine. The new inhibitor is 1,000-fold more potent than the parent compound (K(i) values for CDK1 = 9 nM and CDK2 = 6 nM versus 5,000 nM and 12,000 nM, respectively, for O(6)-cyclohexylmethylguanine). The increased potency arises primarily from the formation of two additional hydrogen bonds between the inhibitor and Asp 86 of CDK2, which facilitate optimum hydrophobic packing of the anilino group with the specificity surface of CDK2. Cellular studies with O(6)-cyclohexylmethyl-2-(4'- sulfamoylanilino) purine demonstrated inhibition of MCF-7 cell growth and target protein phosphorylation, consistent with CDK1 and CDK2 inhibition. The work represents the first successful iterative synthesis of a potent CDK inhibitor based on the structure of fully activated CDK2-cyclin A. Furthermore, the potency of O(6)-cyclohexylmethyl-2-(4'- sulfamoylanilino)purine was both predicted and fully rationalized on the basis of protein-ligand interactions.     

Structure-based design of a fluorimetric redox active peptide probe

Structure-based iterative design was used to prepare a disulfide-containing nonapeptide as a fluorimetric probe for chemical and biochemical disulfide forming and breaking reactions. The peptide is composed entirely of natural amino acids and exhibits a marked (42%) change in fluorescence between its oxidized and its reduced states. The probe is easily synthesized and highly water soluble and exhibits well-behaved kinetics on reduction with the reductant tris-carboxyethylphosphine. The reduced peptide is an excellent substrate of the enzyme quiescin-sulfhydryl oxidase and may find utility in the characterization of other disulfide oxidoreductases.     

Structure-based design of PDK1 inhibitors

A macrocyclic 2-anilino-4-phenyl-pyrimidine CDK/Flt3/JAK2 inhibitor was found to have moderate PDK1 activity. After docking into a PDK1 X-ray structure it was suggested that the pyrimidine ring could be substituted for a purine thereby increasing the number of hydrophobic contacts with the protein and forming an additional hydrogen bond to the kinase hinge. Deletion of the macrocyclic linker allowed a more rapid optimisation of the aromatic substituents as well as the introduction of an amino-amide solubility tag. This improved both binding to the enzyme and physiochemical properties without compromising ligand efficiency.     

Structure-based design of potent histatin analogues

Conformational studies of human salivary peptide, histatin 3 (Hst3), were performed by nuclear magnetic resonance (NMR) and circular dichroism (CD) spectroscopy in a membrane-mimicking environment. The structural information that was obtained was used in the design of peptide analogues with improved antifungal activity. In the presence of increasing concentrations of L-alpha-dimyristoylphosphatidylcholine (L-alpha-DMPC) lipid vesicles, a dramatic increase in a minimum at 198 nm is observed in the CD spectra of Hst3. The NMR data of Hst3 in the presence of L-alpha-DMPC lipid vesicles reveal the proximity of residues Y(10) and S(20), indicating the existence of a more compact structure. Peptide analogues were designed on the basis of this observation, which incorporated a disulfide bond to stabilize an extended loop in this region of the sequence. One of these, peptide 4, was 100 times more potent than Hst5 against Saccharomyces cerevisiae cells. Conformational analysis of peptide 4 revealed a looped structure with charged residues protruding on the outside surface, while a combination of aromatic residues and histidines are packed into an internal core.     

Structure-based drug design of pyrazinone antithrombotics as selective inhibitors of the tissue factor VIIa complex

Structure-based drug design coupled with polymer-assisted solution-phase library synthesis was utilized to develop a series of pyrazinone inhibitors of the tissue factor/Factor VIIa complex. The crystal structure of a tri-peptide ketothiazole complexed with TF/VIIa was utilized in a docking experiment that identified a benzyl-substituted pyrazinone as a P(2) surrogate for the tri-peptide. A 5-step PASP library synthesis of these aryl-substituted pyrazinones was developed. The sequence allows for attachment of a variety of P(1) and P(3) moieties, which led to synthesis pyrazinone 23. Compound 23 exhibited 16 nM IC(50) against TF/VIIa with >6250x selectivity versus Factor Xa and thrombin. This potent and highly selective inhibitor of TF/VIIa was chosen for pre-clinical intravenous proof-of-concept studies to demonstrate the separation between antithrombotic efficacy and bleeding side effects in a primate model of thrombosis.     

Structure-based protein design with deep learning

Since the first revelation of proteins functioning as macromolecular machines through their three dimensional structures, researchers have been intrigued by the marvelous ways the biochemical processes are carried out by proteins. The aspiration to understand protein structures has fueled extensive efforts across different scientific disciplines. In recent years, it has been demonstrated that proteins with new functionality or shapes can be designed via structure-based modeling methods, and the design strategies have combined all available information — but largely piece-by-piece — from sequence derived statistics to the detailed atomic-level modeling of chemical interactions. Despite the significant progress, incorporating data-derived approaches through the use of deep learning methods can be a game changer. In this review, we summarize current progress, compare the arc of developing the deep learning approaches with the conventional methods, and describe the motivation and concepts behind current strategies that may lead to potential future opportunities.     

Structure-based antigen design: a strategy for next generation vaccines

Vaccine design is progressing from empiricism towards the increasingly rational presentation of the targets of protective immunity. Nevertheless, most current vaccine antigens are essentially the native macromolecules of pathogens. These molecules are adapted to evade, not induce, immunity. High resolution structures reveal the electrostatic surfaces recognized by neutralizing antibodies and the architectures underlying these surfaces, thereby identifying which substructures must be left intact and which can be changed to optimize biochemical and immunologic performance. Armed with detailed structural information, we can engineer optimized antigens that are more stable, homogeneous, and efficiently produced, making immunization more practical and affordable. Understanding the structural basis for immunogenicity and immunodominance will allow us to improve vaccine efficacy and broaden the range of vaccine-preventable diseases.     

Structure-based design of caspase-1 inhibitor containing a diphenyl ether sulfonamide

A series of compounds was designed and prepared as inhibitors of interleukin-1beta converting enzyme (ICE), also known as caspase-1. These inhibitors, which employ a diphenyl ether sulfonamide, were designed to improve potency by forming favorable interactions between the diphenyl ether rings and the prime side hydrophobic region. An X-ray crystal structure of a representative member of the diphenyl ether sulfonamide series bound to the active site of caspase-1 was obtained.     

Structure-based design of Tet repressor to optimize a new inducer specificity

We constructed a mutant of the tetracycline-inducible repressor protein TetR with specificity for the tc analogue 4-de(dimethylamino)anhydrotetracycline (4-ddma-atc), which is neither an antibiotic nor an inducer for the wild-type protein. The previously published relaxed specificity mutant TetR H64K S135L displays reduced induction by tc but full induction by doxycycline (dox), anhydrotetracycline (atc), and 4-de(dimethylamino)-6-demethyl-6-deoxytetracycline (cmt3). To create induction specificity for tc derivatives lacking the 4-dimethylamino grouping such as cmt3 and 4-ddma-atc, the residues at positions 82 and 138, which are located close to that moiety in the crystal structure of the TetR-[tc-Mg](+)(2) complex, were randomized. We anticipated that a residue with increased size may lead to sterical hindrance, and screening for 4-ddma-atc-specific induction indeed revealed the mutant TetR H64K S135L S138I. Out of 24 exchanges only the addition of S138I to TetR H64K S135L yielded a mutant with a pronounced reduction of affinity for atc and dox, while the one for 4-ddma-atc is not affected. The ratio of binding constants revealed a 200-fold specificity increase for 4-ddma-atc over atc. The contributions of each single mutant to specificity indicate that the tc variants bind slightly different positions in the TetR tc binding pocket.     

Structure-based design of a thiazolidinedione which targets the mitochondrial protein mitoNEET

Several PPAR-gamma agonists containing a thiazolidinedione moiety (referred to as glitazones) have been proposed to be neuroprotective and appear to alter mitochondrial function. Recently, a search for mitochondrial proteins that bind pioglitazone identified a novel protein, mitoNEET, which was later shown to regulate the oxidative capacity of the mitochondria. This identified an alternative target for the glitazones suggesting a possible new drug target for the treatment of neurodegenerative diseases. Molecular docking studies employing the reported crystal structure revealed five possible binding pockets on mitoNEET. We focused on two sites based on their physical characteristics. Using binding information gained from the analysis of two glitazones docked in these pockets, we designed and synthesized a ligand (NL-1) that would preferentially bind to site 1. Based on [³H]-binding data of the glitazones and comparisons to computer generated K_is, we were able to predict that site 1 was likely the target of the glitazones. NL-1 uncoupled isolated mitochondrial complex I respiration with an IC₅₀ of 2.4 μM and inhibited state III respiration up to 45%. To investigate the ability of NL-1 to block rotenone initiated free radicals from complex I, we found it was able to protect the human neuronal cell line SH-SY5Y against rotenone induced cell death. These data demonstrate that mitoNEET is a viable target for the design and synthesis of novel therapeutic agents aimed at altering mitochondrial function.     

Structure-based design of aliskiren,a novel orally effective renin inhibitor

Hypertension is a major risk factor for cardiovascular diseases such as stroke, myocardial infarction, and heart failure, the leading causes of death in the Western world. Inhibitors of the renin-angiotensin system (RAS) have proven to be successful treatments for hypertension. As renin specifically catalyses the rate-limiting step of the RAS, it represents the optimal target for RAS inhibition. Several peptide-like renin inhibitors have been synthesized previously, but poor pharmacokinetic properties meant that these compounds were not clinically useful. We employed a combination of molecular modelling and crystallographic structure analysis to design renin inhibitors lacking the extended peptide-like backbone of earlier inhibitors, for improved pharmacokinetic properties. This led to the discovery of aliskiren, a highly potent and selective inhibitor of human renin in vitro, and in vivo; once-daily oral doses of aliskiren inhibit renin and lower blood pressure in sodium-depleted marmosets and hypertensive human patients. Aliskiren represents the first in a novel class of renin inhibitors with the potential for treatment of hypertension and related cardiovascular diseases.     

Structure-based drug design with geometric deep learning

Structure-based drug design uses three-dimensional geometric information of macromolecules, such as proteins or nucleic acids, to identify suitable ligands. Geometric deep learning, an emerging concept of neural-network-based machine learning, has been applied to macromolecular structures. This review provides an overview of the recent applications of geometric deep learning in bioorganic and medicinal chemistry, highlighting its potential for structure-based drug discovery and design. Emphasis is placed on molecular property prediction, ligand binding site and pose prediction, and structure-based de novo molecular design. The current challenges and opportunities are highlighted, and a forecast of the future of geometric deep learning for drug discovery is presented.     

Structure-based design of 7-carbamate analogs of geldanamycin

The 7-carbamate groups of geldanamycin and its 17-(2-dimethylaminoethyl)amino-17-demethoxy derivative (17-DMAG) bind the N-terminal domain of Hsp90 by establishing a network of hydrogen bonds which involve four buried water molecules. In this study, a structure-based approach was used to investigate the effects of displacing some of these waters by modification of the 7-carbamate. A general loss of binding to human Hsp90 was observed, except for replacement of the carbamate with a hydroxamate group which gave an analog with weak activity. Modeling of Hsp90-ligand interactions suggested that the hydroxamate was not able to displace the buried water molecules, while bulkier substituents able to do so proved inactive.     

Structure-based design of cyclooxygenase-2 selectivity into ketoprofen

We have recently described how to achieve COX-2 selectivity from the non-selective inhibitor indomethacin (1) using a combination of a pharmacophore and computer 3-D models based on the known X-ray crystal structures of cyclooxygenases. In the present study we have focused on the design of COX-2 selective analogues of the NSAID ketoprofen (2). The design is similarly based on the combined use of the previous pharmacophore together with traditional medicinal chemistry techniques motivated by the comparative modeling of the 3-D structures of 2 docked into the COX active sites. The analysis includes use of the program GRID to detect isoenzyme differences near the active site region and is aimed at suggesting modifications of the basic benzophenone frame of the lead compound 2. The resulting series of compounds bearing this central framework is exemplified by the potent and selective COX-2 inhibitor 17 (LM-1669).     

Structure-based protein-protein interaction networks and drug design

Proteins carry out their functions by interacting with other proteins and small molecules, forming a complex interaction network. In this review, we briefly introduce classical graph theory based protein-protein interaction networks. We also describe the commonly used experimental methods to construct these networks, and the insights that can be gained from these networks. We then discuss the recent transition from graph theory based networks to structure based protein-protein interaction networks and the advantages of the latter over the former, using two networks as examples. We further discuss the usefulness of structure based protein-protein interaction networks for drug discovery, with a special emphasis on drug repositioning.