Current Medicinal Chemistry - Volume 11, Issue 5, 2004

We live in times of overwhelmingly fast development of new genomics and proteomics methods bringing primary structures of a number of genomes.More than 20,000 protein structures are stored in the protein data bank (PDB),but it is less than one tenth of a number of already defined primary structures of proteins.Even using fastest throughput crystallization and X-ray crystallography methods scientists would not be able to solve 3D structures of all these proteins during the next decade.Medicinal chemistry is increasingly using 3D structures of proteins,DNA,and RNA for the development of drugs.There exist a number of successful stories of structure-based drug design.One can recall the development of HIV protease inhibitor made using 3D structure of HIV protease.At the same time it is obvious that scientists do not always have the solved 3D structures of biomolecules and have to use their structures predicted on the basis of their primary sequences. A number of predicted 3D structures of various proteins already are used in structure-based drug design.For example,simply identifying solvent exposed residues in some proteins saves enormous amount of work in revealing the sites critical for antibody binding or / and putative drugs binding to proteins.In many cases,like cytochrome P450 family of proteins, predictive modeling became a regular instrument for outlining the catalytic properties of various cytochromes and recommendation for possible structures of inhibitors of these enzymes. Construction of the template based pharmacophore models for further drug design on the basis of predicted protein structures gives a reasonable rate of success.At the same time prediction of RNA structures for this purpose has not developed sufficiently despite obvious opportunities to model specific shapes of RNA and consequently its binding to small compounds and docking to specific proteins. Currently protein structure prediction methods are becoming a part of broader genomic and proteomic studies.One of the promising directions here is using protein structure prediction to estimate the possible impact of single nucleotide polymorphisms (SNP)on the properties of specified proteins and subsequently on overall medical conditions of patients. Technologies for large-scale SNP genotyping are now affordable.At the same time current development of protein structure prediction technology also makes it possible to create high throughput prediction programs that would work in synchronized modes with large-scale SNP genotyping.This combination can bring enormous amount of data on possible impacts of SNPs in coding regions. Reliability of protein structure prediction methods is supported by a number of studies where authors made comprehensive point mutagenesis analysis of predicted structures and show very impressive correspondence of predicted 3D structures to mutagenesis results.Meanwhile I had to note that in so-called 'gray zones ' of low sequence identity reliability of protein structure prediction became lower and here we would need more support from experimental results.Nevertheless in many cases one does not need a prediction of the entire protein structure.Prediction of specific,functionally important regions can help in structure-based drug design even when we cannot create such a model of entire protein. Protein structure prediction programs develop very rapidly.A number of publications on this topic more than doubled during last 5 years.I think the methods of prediction would become a regular instrument in a number of high throughput methods of drug design.

The P450 family of proteins has more than 1000 representatives,which despite sometimes relatively low sequence identity have a surprisingly high level of structural similarity.This fact makes this family of proteins ideal candidate for various types of modeling based on protein structure prediction.A number of P450 proteins,including CYPs 1A1,1A2,1B1,3A4,11B2,17,and 19,play a role in the metabolism of estrogen.Inhibitors of these proteins could be very promising drugs for hormonal treatment of postmenopausal breast cancer.Population studies have yielded a significant amount of data describing the relationship between single nucleotide polymorphisms (SNP)in DNA and cancer risk related to these proteins. A combination of SNP analysis with protein structure prediction can be a very useful strategy in investigations of structure-functional relations of P450 proteins and structure-based drug design.Here we will demonstrate how protein structure prediction combined with genetic SNP analysis can be useful for potential drug design and possibly,individual treatment of breast cancer.

Over 1,500 human disease genes have been identified,of which only a small fraction have experimental structural information on the protein products.To better understand the mechanisms of these hereditary diseases,we undertook a systematic study to predict the structures of disease proteins and characterize their interactions with other proteins.This study was facilitated by two tools developed previously:(1) COBLATH,a structure-prediction method that exploits the complementarity of PSI-Blast and sequence-structure threading and PPISP,a method that predicts the residues involved in protein-protein interactions.In this initial study of human disease proteins, we were able to build structural models for 60 proteins involved in human diseases.For a number of proteins, new structural domains were identified.In the case of ABCD1,a protein responsible for adrenoleukodystrophy, the disease mutation P484R was positioned at the homodimer interface.This positioning is consistent with experimental observation that the P484R mutation impairs ABCD1 self-interaction and suggests that that the disease mechanism of this mutation lies in the impaired ABCD1 dimerization.This initial study illustrates the value of the predicted structure models and may serve as an example for expanded studies of other disease proteins.

The human genome and other genome sequencing projects have generated huge amounts of protein sequence information.Recently,a structural genomics project that aims to determine at least one representative three-dimensional structure from every protein family experimentally has been started.Homology modeling will play an essential role in structure based drug design such as in silico screening;because based on these representative structures the three-dimensional structures of the remaining proteins encoded in the various genomes can be predicted by homology modeling.The results of the last Critical Assessment of Techniques for Protein Structure Prediction (CASP5)demonstrated that the quality of homology modeling prediction has improved;reaching a level of reliability that biologists can now confidently use homology modeling.With improvements in modeling software and the growing number of known protein structures,homology modeling is becoming a more and more powerful and reliable tool.The present paper discusses the features and roles of homology modeling in structure based drug design, and describes the CHIMERA and FAMS modeling systems as examples.For a sample application,homology modeling of non-structural proteins of the severe acute respiratory syndrome (SARS)coronavirus is discussed. Many biological functions involve formation of protein-protein complexes;in which the protein molecules behave dynamically in the course of binding.Therefore,an understanding of protein-protein interaction will be very important for structure based drug design.To this end,normal mode analysis is useful.The present paper discusses the prediction of protein-protein interaction using normal mode analysis and examples of applications are given.

The nicotinic receptor at the motor endplate has served as a prototype for understanding structure,function and ligand recognition in the superfamily of pentameric ligand-gated ion channels.Yet despite this advanced state of knowledge, atomic-scale understanding of such elementary processes as ligand recognition has remained elusive owing to the lack of a high-resolution x-ray structure.However,the field has recently entered a state of rapid advancement following the discovery and atomic structural determination of the water-soluble acetylcholine binding protein (AChBP),a homolog of the receptor ligand binding domain.The AChBP structure provides the theoretical foundation for generating homology models of the corresponding receptor ligand binding domains within this structural family of receptors.Experimental assignment of residue equivalence between AChBP and receptor subunits subsequently yielded homology models ready for experimental testing.One such test is computational determination of ligand docking orientation in conjunction with mutagenesis of predicted contact residues and measurements of ligand binding affinity.Applied to different analogs of the competitive antagonist curare,docking computations that incorporate intrinsic protein flexibility reveal fundamentally distinct orientations of each analog bound to AChBP.The different contact residues predicted for each analog were tested and confirmed by mutagenesis of AChBP followed by measurements of ligand binding.By applying the same computational and experimental approaches to the adult human muscle AChR,we find that the two curare analogs also dock in distinctly different orientations.Thus subtle structural changes in the ligand,and by extension,structural differences in non-conserved residues among receptor subtypes and species,can dramatically alter the orientation of the bound ligand.The results have important implications for design of drugs targeting nicotinic receptors and members of the superfamily of pentameric ligand-gated ion channels.

We have studied the structure and function of Human Misshapen / NIK-related kinase (MINK-1)through a combination of computational methods and experimental approaches,including (1)fold recognition and sequence-structure alignment for each structural domain using the threading program PROSPECT,(2)gene expression and protein-protein interaction analysis of yeast homologs of human MINK-1 domains,and (3)yeast two-hybrid screening for proteins that interact with human MINK-1.Our structure prediction dissects MINK-1 into four domains:a conserved N-terminal kinase domain,followed by a coiled-coil region and a proline-rich region,and a C- terminal GCK domain.Gene expression and yeast two-hybrid analysis of yeast homologs of the MINK-1 domains suggest that MINK-1 may be involved in cell-cycle progression and cytoskeletal control.Consistent with these predicted functions,our in-house yeast two-hybrid screen for proteins that interact with human MINK-1 provides strong evidence that the coiled-coil and proline-rich domains of MINK-1 participate in the regulation of cytoskeletal organization,cell-cycle control and apoptosis.A homology model of the MINK-1 kinase domain was used to screen the NCI open compound database in DOCK,and chemical compounds with pharmaceutically acceptable properties were identified.Further medicinal chemistry compound structure optimization and kinase assays are underway.

Homology modeling has become an essential tool for studying proteins that are targets for medical drug design.This paper describes the approach we developed that combines sequence decomposition techniques with distance geometry algorithms for homology modeling to determine functionally important regions of proteins.We show here the application of these techniques to targets of medical interest chosen from those included in the CASP5 (Critical Assessment of Techniques for Protein Structure Prediction)competition,including the dihydroneopterin aldolase from Mycobacterium tuberculosis ,RNase III of Thermobacteria maritima ,and the NO-transporter nitrophorin from saliva of the bedbug Cimex lectularius .Physical chemical property (PCP)motifs,identified in aligned sequences with our MASIA program,can be used to select among different alignments returned by fold recognition servers.They can also be used to suggest functions for hypothetical proteins,as we illustrate for target T188.Once a suitable alignment has been made with the template,our modeling suite MPACK generates a series of possible models. The models can then be selected according to their match in areas known to be conserved in protein families. Alignments based on motifs can improve the structural matching of residues in the active site.The quality of the local structure of our 3D models near active sites or epitopes makes them useful aids for drug and vaccine design.Further,the PCP motif approach,when combined with a structural filter,can be a potent way to detect areas involved in activity and to suggest function for novel genome sequences.

Rational design in protein engineering leads to significant progresses in medicinal chemistry research.It alleviates the difficulty of exploring unreasonable biological functions.Combining with analysis of biophysical-chemical properties,a three-dimensional (3D)structure provides fruitful information for rational design by revealing functionally important residues.Comparative (homology)modeling,one of the 3D structural prediction techniques,takes advantage of that homologous proteins share similarity in their 3D structures despite the lack of sequence similarity.Of the most value,3D models provide functional clues even though the function may have been modified during evolution. We illustrate here two applications to medicinal chemistry research where comparative models made a significant improvement on the understanding of relevant biological functions of two proteins.These multiple collaborative projects involve the identification of solvent-exposed residues in a membrane anchoring domain of human coagulation factor V,and revealing critical residues in the interfaces of an antibody and a polynuclear aromatic hydrocarbon ligand.Since the protocol of comparative modeling technique we employed is essential to proposing useful hypotheses for experimental testing,we also present our methodology underlying our modeling programs.Our results show that inaccuracies in comparative models do not hamper functional evaluation as long as an in depth analysis of 3D structures is performed.

The Catharanthus (or Vinca )alkaloids comprise a group of about 130 terpenoid indole alkaloids. Vinblastine is now marketed for more than 40 years as an anticancer drug and became a true lead compound for drug development.Due to the pharmaceutical importance and the low content in the plant of vinblastine and the related alkaloid vincristine,Catharanthus roseus became one of the best-studied medicinal plants. Consequently it developed as a model system for biotechnological studies on plant secondary metabolism. The aim of this review is to acquaint a broader audience with the recent progress in this research and with its exciting perspectives.The pharmacognostical aspects of the Catharanthus alkaloids cover botanical (including some historical),phytochemical and analytical data.An up-to-date view on the biosynthesis of the alkaloids is given.The pharmacological aspects of these alkaloids and their semi-synthetic derivatives are only discussed briefly. The biotechnological part focuses on alternative production systems for these alkaloids,for example by in vitro culture of C.roseus cells.Subsequently it will be discussed to what extent the alkaloid biosynthetic pathway can be manipulated genetically (“metabolic engineering ”),aiming at higher production levels of the alkaloids.Another approach is to produce the alkaloids (or their precursors)in other organisms such as yeast. Despite the availability of only a limited number of biosynthetic genes,the research on C.roseus has already led to a broad scientific spin-off.It is clear that many interesting results can be expected when more genes become available.

Serotonin (5-HT)is involved in various physiological and pathological processes by interaction with 14 distinct receptor subtypes,grouped in seven classes of receptors (5-HT 1-7 )on the basis of amino acid sequence,pharmacology,and signal transduction pathways.The 5-HT 7 R is a G-protein coupled receptor with seven transmembrane domains.It was found by the application of molecular cloning and has been identified in rat,mouse,human,pig,and guinea pig.Although the biological functions of the 5-HT 7 Rs are poorly understood,preliminary evidence suggests that it may be involved in depression,control of circadian rhythms,and relaxation of vascular smooth muscles.For these reasons,the 5-HT 7 R has become a target for the development of novel drugs.This review will give a brief introduction of the pharmacology of 5-HT 7 R (molecular structure,distribution of 5- HT 7 R mRNA,localization of the 5-HT 7 R protein,functional correlates,and therapeutic potential)and a detailed survey of the 5-HT 7 R ligands,which have appeared in the literature in both papers and patents.Structure- activity relationships (SAR)of these ligands will also be described.