LIFE306 Dan\'s lectures PDF

Preview

Summary

Detailed notes from Dr. Rigden, very useful for hiss questions in final exams...

Description

LIFE 306 Daniel Rigden Lectures outline Lecture 1 Introduction     

  

Needing new drug How find new drug Properties of a good drug target Basic properties of a drug candidate (typically an inhibitor) Different class of inhibitor (competitive & non-competitive inhibitors; irreversible and suicide inhibitors) properties of a good drug lead Additional drug-like properties required How computer help find a target (genomes; systems; ligands; structures)

Lecture 3 Computer applications in drug design: targets from systems      

Lecture 4 Computer applications in drug design: ligand-based drug design      

Lecture 2 Computer applications in drug design: targets from genomes  

   

how computer help find a target : genomes blasting to look for similar proteins 1 finding a broad-spectrum target: core essential genes 2 finding a pathogenicity-related target: genome substrates 3 finding a pathogenicity-related target: genome comparsion 4. finding a narrow-spectrum target: genome comparsion 5. finding analogous enzyme: database searching the phosphoglycerate mutases pathway reconstruction scenarios: endogenous disease personal genomes and personalised medicine cancer genome guiding treatment antibiotic-resistant bacteria identify resistance mechanisms

Network topology vs essentiality Target choosingSub The flux control coefficient (C) Example: Glycolysis in T. burcei A signal transduction example: the EDF-receptor system The insulin/IGF pathway

  

Substrate mimic/analogues QSAR (Quantitative structure-activity relationship); 3D-QSAR CoMFA (Comparative Molecular Field Analysis): CONCORD, CORINA Pharmacophore Synthetic accessibility Lipinski’s rule of 5 o log P o Other properties Cross BBB Drug metabolism and toxicity Prediction: reaction/ drug-based binding/problems

Lecture 5 Computer applications in drug design: the roles of protein structures     

   

structures and druggability SBDD strategy (structural-based drug design) Example: aldose reductase AmpC β-lactamase (screening library for covalent inhibitor) Estrogen receptor o exploting subtle differences using modelling o agnosis design Thermolysin: inhibitor rigidity MMP-13: joining fragments BET bromodomins: joining across domains Structure can o explain drug resistance o help design next-generation drugs o aid drug repositioning o for structure-based metabolism prediction

Lecture 1 Introduction 1. Why need for new drugs? (1) Some diseases still have no effective treatment. (2) Few drugs are perfect, most drugs lead to side effects to some people, since people respond differently to drugs. (3) Infective agents become resistant to existing drugs, for instance, MRSA (Methicillin-resistant staphylococcus aureus) (4) New disease arises, for example, HIV. (5) Drug development is very expensive, therefore need to be as effective as possible in each step. 2. How can we find new drugs? (1)From nature ① Folk remedies: trial and error over millennia, such as aspirin from willow bark and astemisin from wormwood. ② Serendipity (be found by chance), for example, penicillin from mould. ③ Problems: drugs from natural sources ofen complex, non-selective, available in small amount, toxic etc. (2) In vitro (by lab experiment) ① HTS screening: robots can text libraries for thousands of compounds ② Problems: need to identify targets first; need to develop suitable assay to see if compound has effect; risk for false positives; the libraries only samples from all possible compounds, not precise and still need very much detection and verification. ③ Computers help in silico sequential characterisation of drug targets and small molecule effectors of these targets. The procedure contains Bioinformatics – central to drug design  Target discovery Genomics; Genetics; Proteomics; Transcriptomics Target identification Target validation Chemogenomics; Structural genomics; Homology  drug discovery modelling Lead identification Lead optimization  drug development Preclinical Biomarkers; Toxicogenomics; Pharmacogenomics; Clinical In silico modelling 3. Properties of a good drug target (NOT DRUG) [ISAD] (1) Impact: intervention should have a significant impact on the process (disease, infection etc.). That means not every protein is a target. (2) Specificity: an easy target should be different from other relevant proteins, allowing specific targeting. But it may be difficult to hit one human target in the context of many related proteins. (3) Accessible: present in an assessable compartment of the organism. (4) Druggability: there should be a structure of the target, or the possibility of modelling it. Prediction of druggability is important.

4. Inhibitor (ofen drugs are as inhibitors) (1) Different types of inhibitors Types Competitive

Binding sites Binding to the same site as substrate.

Features  Common.  Inhibitor ofen resembles substrate, so more easily to design. Also, inhibitors bind much more strongly than the real substrate.  [DREWBACK]Substrate will increase on inhibition potentially competing off the inhibitor  E.G. Ritonavir, a peptide-based inhibitor of HIV protease Non-competitive Binding to else where  No [DREWBACK]  Inhibition is independent of the amount of substrates  Need to discover second site, so more information is required.  E.G. Pentobarbital, an anaesthetic, inhibiting GABA receptors.  No [DREWBACK] Irreversible Binding covalently, usually at the catalytic  Do not be washed away or get site metabolised  E.G. some antibacterial compounds called sesquiterpene lactons, working by inactivating MurA, involved in cell wall synthesis. Enzyme here is permanently inactivated.  One of the typical examples is suicide inhibitors, exploiting the catalytic mechanism of the target. E.G. DFMO, an anti-trypanosomatid compound, irreversibly inhibits ornithine decarboxylase. Since the processing bu the catalytic site is necessary, so it is called a suicide inhibitor. (2) Basic properties of a drug candidate (typically an inhibitor) ①Specificity is a key concept. 

In ideal cases, the target is completely different to other proteins. Binding to other proteins/ targets should be avoided. Only effect is the desired one; no binding to other molecules; less potential for side effects.



In other cases, we may wish to inhibit one protein but not a similar one. To achieve this, two similar proteins should be compared and differences exploited to produce specificity.



Irreversible inhibition is possible when difference include chemically reactive groups (e.g. Lys/ Cys)

② High affinity. So that: need less (cheaper); fewer side effects; less toxicity; more easily administered ③ Soluble ④ Bioavailable and could reach circulation ⑤ Good pharmokinetic, being metabolised neither too slow nor too quick. ⑥ Cheep to synthesis and development.

5. How computer help find a target (1) Genomes  

Complete genomes necessarily contain all possible targets, Computer can compare genomes (human protein to potential pathogen targets/ the common gene among organisms/ differences between pathological and non-pathological strains in bacteria/ analogues/ drug-resistant and drug-sensitive strains).

(2) Systems  



Protein more ofen act in pathways and networks. Computer can o Mosel the emergent, non-obvious properties of networks, thereby highlighting the key nodes. o Simulate the effectives of intervention in network, for instance, synergistic pairs of drugs. o Model metabolic pathways in order to show which enzyme exert most control of flux. Sequence based production of o Subcellular localization shows how many membranes need to be crossed. o PTM o The existence and tissue-specificity of isoforms.

(3) Structures   

A structure of a target enables structure-based drug design Computer can be used to model the target and/or other related proteins. The difference between the target and the related proteins could help to seek specificity of action. Since there are many variants of a target gene in the population. Structural knowledge can explain and predict consequence of difference. It will be useful for the development of personalised medicine.

(4) Ligands 



If some ligands for the target protein are known, ①computer can analyse their shared characteristics and relate them to binding affinity. ② A description of the characteristics of a good ligand could be used to database searching for difference chemicals. ③These ligands of interest can be screened (i) For ‘drug-like properties’, (ii) For toxicity (iii) For ease of synthetic accessibility. Therefore, a better ligand could be produced.

Lecture 2 Computer Application in Drug Design: Targets from genomes   

Finding target in pathogens: BLAST/comparison (very powerful way to discover new targets) of bacterial genomes/ Analogous enzymes/ narrowly-specific targets Endogenous disease Personalised medicine and drug resistance

1. Targets  

Accessible, effective. A specific inhibitor can be designed

(1) Pathogens: ① Pathogenic organisms are assumed to have extra components conferring pathogenicity. Targets associated with pathogenicity (tissue invasion, in which the toxin production is attractive. If this process can be blocked, the hosts’ natural defences might be able to clear infection) ②Pathogen-specific pathway also attractive. (Bacterial cell walls) ③ Metabolic pathways also present in host can be exploited, but likely to cause problems. (2) Endogenous host problems  

Avoid targeting members of big families. Use knowledge of networks to consider all possibilities.

2. How does computer find a target – Genomes (1)Since complete genomes necessarily contain all possible targets, so computer can be used for searching in and comparing genomes, to find new targets in pathogens and check if the same protein are present in the host. Steps: 1. Finding targets: which genes are present in many pathogens? The difference between pathogenic and non-pathogenic strains of bacteria? 2. Checking specificity: The important reactions catalysed by unrelated, analogous enzyme in parasite and host? Are there similar human proteins to potential pathogen targets? (2) For the pathogen-host scenarios: Favourable 1 Favourable 2

Target is in a pathway only present in the pathogen Target is in a pathway present in both pathogen and host. However, the enzymes are analogous, which means unrelated proteins can catalysed similar reaction, and therefore structurally dissimilar. So, specificity is easy to be achieved. Less favourable Target is in a pathway present in both pathogen and host. Also, the enzymes are homologous and therefore structurally similar. So, specificity is much harder to be achieved. 3. BLASTing to look for similar proteins (1) BLAST:   

The main tool to look in database, the human genome, for proteins similar to a query, i.e. your interested target. Similarity expressed as an e-value: smaller means closer relationship, greater than 0.01 presumed insignificant. having simple application in drug design o Is there a human/host protein similar to a pathogen’s potential drug target? If it is, then this drug cannot be designed, since it may attack human normal proteins.

o

Does pathogen B have a protein corresponding to a drug target in pathogen A? If it does, this drug may be useful in attack both pathogens.

(2) BLAST steps (Searches can be designed for both broad- and narrow-spectrum tragets) ① Finding a broad-spectrum target: core essential genes This gene is essential for the growth of the pathologic bacterial growth, and this will be present in every genome. Comparison between several bacterial, will suggest several genes are essential. But within these gene, only a part will be useful, since either related or unrelated proteins can carry out missing activity sufficiently to allow growth; also, not all genes may be essential under all conditions. ② Finding a pathogenicity-related target: genome subtraction (减法，差集 减法，差集) Comparing a relative benign bacteria with a meningitis causing bacteria, finding the identical gene between them. Then, filter out the genes that ONLY in one pathogen, in which may include the gene responsible for the pathogenicity. Actually, around half of the genes that ONLY in the pathogen are found exclusively in pathogenic bacteria; also, these genes have virulence-related function, such as surface protein and proteins that synthesise toxins. These are candidate drug targets although the target would still need to be validated. For example: is a mutant lacking the gene still virulent? ③ Finding a pathogenicity-related target: genome comparison Searching among several pathogens for novel virulence-associated genes. Took the proteins which are described as hypothetical/unknown function. Next, made mutants, to see if some will be attenuated. If yes, then these genes may be novel virulence factors. However, most of these mutants do not affect growth of the bacterium, only affecting the ability to infect the human host. That is normal, since immune system would clear less virulent forms. Bioinformatics can predict subcellular location: surface or exported proteins more likely to interact with host. ④ Finding a narrow-spectrum target: genome comparison Sometimes, the ideal drug target might only be in single species. For instance, Helicobacter pylori, which is associated with stomach cancer. （We want to kill the specific bacterial rather than all the bacteria in the gut.） [The process about finding and analysing Helicobacter pylori: 1. All gene of Helicobacter pylori are searching against other microbial genomes. Only proteins distantly related to any of these genomes are kept. 2. In the kept proteins, remove membrane proteins (hard biochemistry), remove hypothetical/unknown function (no clue as to biochemistry), remove those encoding inessential proteins. 3. Testing these in lab by deleting genes. [[即挑选基因要和其他的有害，有利，以及人体基因相比对，不能伤害人体，要和有害基因有一定的相似度，如果是这 即挑选基因要和其他的有害，有利，以及人体基因相比对，不能伤害人体，要和有害基因有一定的相似度，如果是这 个病原体特别独有的基因，则很有可能作为这个病原体的特殊致病相关基因。 个病原体特别独有的基因，则很有可能作为这个病原体的特殊致病相关基因。]]

4. Database of essential genes. Since limited coverage, mainly model organisms, merely human, yeast, mouse, fly, zebra fish; lack of data on essentiality from particular conditions, unsure if they will be relevant for medical conditions. 5. Analogous enzyme. In various cases, nature has envolved the same catalytic activity twice, which means for the same catalytic activity, there are two enzymes, but they share no similarity expect for the ability to catalyse the same reaction. Thus, they should be significant structural differences. If one kind is drug target in bacteria, then specificity over the human counterpart should be easily achieved.]

⑤Finding analogous enzymes: (i) Database searching OR (ii) Pathway instruction (i) Database searching: the ‘EC ENZYME’ database groups enzymes according to the reaction catalysed, regardless whether they are homology or analogy. This database could be used to find analogous by looking for more than one sequence cluster in a single EC number. E.G. Phosphoglycerate mutase is an enzyme of glycolysis/gluconeogenesis. Many pathogens such as bacteria can use alternatives to glycolysis for energy production. But glycolysis is essential in the human forms of trypanosomatid parasites such as Trypanosoma brucei that causes sleeping sickness and Leishmania.

We now search for Phosphoglycerate mutase in as EC 5.4.2.1. The results shows that they are not all homologous. The relationship could be visualised with BLAST. Since the analogous enzymes have little effect on human enzyme, they should be good target for drugs.

(ii) Pathway reconstruction Figure out one pathway A-B-C-D-E, which appears in both known genes and genes in new genome. Since there are fore enzymes to align to achieve the final product, we could BLAST the four groups of the same-function enzymes separately. If two enzymes in one group are found not similar to each other, we could say they are analogues. And also, the one in the new genome could be designed as a good drug target.

(3) For the endogenous disease scenarios Favourable

The target is not related to other proteins in the patient cell. Easier to design an inhibitor of the target that will not bind to any other protein. Less favourable The target is related to several other proteins in the patient cell, this will be similar structures to the target. Harder to design an inhibitor of the target without bind to any other proteins. The above conditions could be detected by BLAST. DNA topoisomerase exists in few isoforms in humans, but is not part of a big family, so specificity readily achieved. Aurora kinase is related to dozens of other protein kinase, so specificity will be hard. 4. Personal genomes and personalised medicine  

People differ in response to drugs because of many factors (variants of drug target, different absorbance or metabolism). Genome will help select appropriate medicine.

(1) Cancer is a disease of multiple origins. E.G. Tongue cancer are rare and unresponsive. The genome of this cancer is sequenced and show RET oncogene response. Thus, the tongue cancer is treated with RET inhibitors. Later, it may recurrence later, this is due to the resistance mechanisms. 

Resistant mechanisms o Identified by analysing the genomes of antibiotic-resistant bacteria. o There are variety of resistance mechanisms  Impaired flux  Efflux  Target mutation/modification  Overproduction of the target mimic  Factor associated protection  Drug modification/degradation

Lecture 3 Computer Application in Drug Design: Targets from systems 

Essential protein as hubs and bottlenecks in (metabolic) networks o How to model them o Target example: T. brucei glycosis (Signalling networks); EGF receptor (Drug repositioning)

1. Network topology and essentiality    

The four main types of proteins in network are hub, bottleneck, hub and bottleneck and others. Hub proteins with many interactions, tend to be essential. Bottleneck protein, connections between sub-networks, tend to be essential, particularly in signalling where information flows in a direction. However, the information about these proteins are rare.

2. Detection procedure: (1) Choosing targets from pathways and networks  



The target pathway could be biosynthesis of cell components, for instance, penicillin interferes with the way that bacteria produce their cell walls. Networks are particularly common in signalling phenomena, in which signals from different ligands interact, e.g. cancer. o Feedback is common in networks. So, for non-trivial systems it becomes very hard to predict what will be the result of an invention, there are too many feedback, either positive or negative. o Networks may consist of small molecules (substrates, products or consecutive/continuous reactions of a pathway) and proteins (that interact with and modify each other, e.g. signalling pathways of kinases and phosphatases). o The key idea is to move away qualitative description and using mathematical description, which is necessary to understand the behaviour of the network which is not predictable considering components in isolation. Therefore, simple intuitive approach is insufficient. Not all proteins in a network are equally susceptible to inhibition and meanwhile, inhibition a protein will be very complex, since a single protein could be included in parallel routes and feedback. So the inhibition results may not what is expected.

(2) Metabolic pathway modelling In a linear metabolic pathway, some are more sensitive to inhibition than others, because there may be more capacity of one protein than another. Enzyme with little spare capa...