MP6: Structure-based drug design

Question 1

What is serendipitous drug discovery? Give 3 examples of drugs discovered this way.

Answer 1

Serendipitous discovery is the unexpected discovery of a drug with therapeutic properties during research that was originally aimed at finding something else.

1. Aspirin
2. Penicillin
3. Viagra (in the context of erectile dysfunction)

Question 2

What did Paul Ehrlich mean by ‘the magic bullet’? Is this still relevant today? Give 3 examples.

Answer 2

It describes a theoretical drug that could specifically target and destroy a disease-causing microbe without harming the patient’s healthy cells. He envisioned a drug that would be similar to a bullet fired from a gun, targeting only the disease-causing agent and leaving healthy tissue unharmed.

Yes! Monoclonal antibodies, Gleevec, Herceptin.

Question 3

Define Eroom’s Law. Why is this happening?

Answer 3

The number of drugs approved per billion $ is halving every 9 years, despite technology improving.

(It’s the reverse spelling of Moore’s law, describing the trend of microchip capacity doubling around every 2 years).

• Drug development is more complex and expensive
• Low-hanging fruit targets have already been discovered
• Approval process is more rigorous

Question 4

What is rational drug design? Give an example of a drug that was discovered this way.

Answer 4

(Also known as structure-based drug design) is a process that uses the knowledge of the molecular structure and function of a target molecule to design drugs that can interact with the target molecule in a specific way.

e.g., Herceptin

Question 5

What are the 6 steps in the drug design process, going from R&D to launching?

Answer 5

1. Target identification and validation
2. Lead generation
3. Lead optimization
4. Pre-clinical development
5. Clinical trials
6. Review and approval

Question 6

Why is target identification and validation so important in drug design?

Answer 6

Understandings of the causes of diseases or conditions is vital in helping researchers known what processes or pathways drugs to treat the condition need to be able to target.

Once a target is identified, studies need to prove that targeting it will actually result in a positive therapeutic outcome.

Question 7

What is druggability? How can the druggability of a target be assessed?

Answer 7

Amenable to treatment with drugs or susceptible to alteration or manipulation with drugs i.e., it’s accessible and can elicit a measurable response.

It can be assessed by looking at the binding pocket properties.

Question 8

Give the steps involved in structure-based drug design.

Answer 8

1. Identify target molecule.
2. Determine the 3D structure
3. Identify potential drug candidates (molecular docking or virtual screening)
4. Optimize the drug candidates
5. Clinical trials

Question 9

Why are protein-protein interactions classed as ‘undruggable’?

Answer 9

They don’t have conventional binding pockets; they tend to be formed by flat surfaces.

Question 10

What is the ‘Trypanosome project’? What key target has it identified?

Answer 10

A research initiative to develop new treatments for diseases caused by Trypanosomes.

The bloodstream form of African trypanosomes lack oxidative phosphorylation machinery, relying only on glycolysis for energy production (…similar to cancer…?) Genome sequencing was then used to identify unique proteins involved in this that can be targeted.

Question 11

What the pitfalls of target identification? Give examples of this occurring.

Answer 11

1. Multiple causes for a condition e.g., asthma, which requires bronchodilators and anti-inflammatory compounds.
2. Misleading target e.g., the cause of nausea was believed to be dopamine D2 receptor so a drug was developed for it. Optimization of this drug led to less effective drugs…this was because the original drug was targeting the actual cause of nausea: 5HT3 receptors.

Question 12

What experimental techniques can be used in target validation? Why are pharmaceutical companies against these experiments?

Answer 12

• Gene knockout
• RNAi
• Chemical genetics
• Chemical probes

There’s a high failure rate of drugs due to target validation tests failing and issues with reproducibility of data. E.g., a target was knocked-out as part of testing a cancer treatment, yet the treatment still worked!

Question 13

What is opentargets.org?

Answer 13

A systematic identification platform that prioritizes potential therapeutic targets (open-source).

Question 14

Once a target has been identified, how can a lead compound be obtained?

Answer 14

• Substrate or co-factor analogue of the target (competitive inhibitor)
• FLIPR
• Scintillation assay
• XRC (fragment-based drug design)
• NMR
• Virtual screening
• Molecular docking

Question 15

How was salbutamol developed from adrenaline in the treatment of asthma?

Answer 15

Adrenaline is released in response to stress, but has no specificity for adrenergic receptor subtypes. This means it binds the B2 receptors in the smooth muscle cells of the airways, causing constriction in asthma patients.

The structure of adrenaline was then taken and altered once, giving isoprenaline (showed some selectivity for B2), and altered again to give salbutamol. Salbutamol has the same potency of isoprenaline, acting competitively against adrenaline, but with a decrease of activity against B1 receptors by over 2000 times i.e., it’s VERY specific.

Question 16

Describe high-throughput screening used in drug design. What techniques are used for both cell-based and in vitro studies?

Answer 16

Libraries of compounds are tested for effect against a target protein, either in cells or in vitro.

In cells, you would assay the efficacy of the compounds using reporter genes or mRNA expression.

In vitro, ELISA can show whether an enzyme has bound an immobilized partner. Other techniques include scintillation assays or FLIPR.

Question 17

What do scintillation assays measure? Describe the process involved.

Why might these be regarded as a turning point in drug discovery? Why might they not?

Answer 17

Used to measure the binding of two molecules through the detection of radiation emitted by a radiolabeled molecule.

The ligand compound is labeled with a radioactive isotope, whilst the target is coated on the surface of small beads, which are composed of scintillant materials that emit light in response to the radiation.

When the ligand and bead-bound target get close, the radiation from the labeled ligand stimulates the scintillant in the bead to emit light. The intensity of the detected light is proportional to the amount of labeled molecule bound to the bead-bound molecule, allowing the measurement of the interaction between the two molecules.

• First true homogeneous HTS technology
• Allows throughput of ~30K compounds/day
• Easy to automate
• No significant amount of waste

BUT:
- radioactive
- long read times
- susceptible to artefacts
- not applicable to all targets

Question 18

What is a FLIPR assay? What does it measure? What are they used to screen?

Answer 18

(Fluorescence imaging plate reader) is a HTS method that measures changes in intracellular calcium concentrations.

Cells expressing a calcium-sensitive fluorescent dye are plated and loaded onto a FLIPR instrument that uses a laser the excite the fluorescent dye. When the cells are stimulated with a compound that activates a GPCR or other calcium-signaling pathway, the fluorescence increases.

Used to screen modulators of GCPRs, ion channels and other targets that signal through calcium pathways.

Question 19

How can novel ligands be discovered by XRC? What are the pros and cons?

Answer 19

Crystals of the target are soaked in a cocktail of potential lead compounds. The structure of the complex is then determined.

Pros:
- high precision
- large number of structures can be solved cheaply and easily

Cons:
- requires crystals of the target
- large structural rearrangements may be missed or break the crystal
- large timescale
- assumes the structures being used are correct and at the correct pH/conformation for the study

Question 20

How can novel ligands be discovered by NMR? What are the pros and cons?

Answer 20

By comparing the spectra of a protein with a protein in complex with a proposed compound, you look for changes in the HSQC spectra.

Pros:
- structures are in solution, so large structural rearrangements can theoretically occur
- rapid feedback on interactions
- information on protein dynamics

Cons:
- protein must be small and soluble at high concentration
- low precision
- expensive due to the use of 15-N

Question 21

Why is AlphaFold2 useful for drug design?

Answer 21

AlphaFold2 is a powerful deep learning algorithm that can predict the three-dimensional structure of proteins with high accuracy.

With AlphaFold2, researchers can now predict the 3D structure of a protein with much greater accuracy, allowing them to better understand how different molecules might interact with it.

This means that researchers can more quickly and accurately identify potential drug targets, as well as optimize the design of molecules to more effectively interact with these targets.

Question 22

What are the requirements for ligand-based and structure-based drug design strategies?

Answer 22

Ligand-based:
- requires a large body of data relating to structure and potency
- doesn’t require receptor structure

Structure-based:
- requires information on the receptor structure and how ligands bind to that structure

Question 23

What is QSAR? How is a QSAR model developed? What are the problems associated with it and how might these be overcome?

Answer 23

Quantitative structure-activity relationships, used in ligand-based drug design.

Method used in drug design to predict the activity, properties, and toxicity of chemical compounds based on their structure.

Models are developed by computationally analyzing chemical structures of relevant compounds and establishing patterns that can be used to assess the activity of new compounds.

Problems:
- How is this comparison achieved?
- What are the best properties to include?

> Strategies are used to identify important variables e.g., regression techniques, etc.

Question 24

What is structure-based drug design? What are the potential pitfalls to this?

Answer 24

Where the structure of the target protein is solved, in complex with a lead compound.

• Must be accurate to identify interacting residues
• Must be achieved in a relevant timescale

Question 25

What is fragment-based drug design? What is the process?

Answer 25

Fragment-based drug design is a drug discovery approach that involves identifying small, low-molecular-weight chemical fragments that bind to a target protein or enzyme and using these fragments as building blocks to design larger molecules with improved potency and selectivity.

1. Fragment screening using XRC or NMR
2. Fragment linking using molecular modeling and docking
3. Optimization

Question 26

What is virtual screening and in silico docking? What is the main advantage and disadvantage of it?

Answer 26

Virtual screening is a method used to identify small molecules or compounds that may have potential as drug candidates. It involves using computer-based algorithms to screen large databases of compounds and predict their binding affinity to a particular target protein. The goal of virtual screening is to identify a set of potential drug candidates that can be further tested in laboratory experiments.

In silico docking is a computational method used to predict the binding mode and affinity of small molecules or compounds to a target protein.

+ Experiments ‘aren’t real’ so saves money
- Too many molecules that could be screened

Question 27

What are pharmacophores? Why are they useful for virtual screening?

Answer 27

3D models that represent the essential features of a molecule that are responsible for its biological activity. These features include:
- functional groups
- charges

This allows you to model certain features of a drug, but using a pharmacophore that contains multiple conformations. Virtual screening methods can use pharmacophore models to search large databases of compounds and identify molecules that match the required features of the pharmacophore. This can greatly reduce the search space and increase the likelihood of identifying potential drug candidates.

Question 28

Why are all compounds screened against hERG channels?

Answer 28

A drug, Vioxx, was sent to market but later showed fatalities amongst certain members of the public. This was due to a mutation in the sub-population of hERG channels, not identified in the clinical trials. Hence, all compounds are now screened against these channels.

Question 29

How is GRID used to generate pharmacophore data?

Answer 29

A high-quality protein and ligand structure is generated and a 3D grid map is build around the protein target. At each grid point, a probe is placed and its interaction is computed, mimicking the chemical properties of most common atom types and small moieties found in ligands.

The result is an isopotential map that shows regions favouring positive and negative probes.

Question 30

What is molecular docking? How does this compare with rigid-body and flexible docking, and which method is most commonly used?

Answer 30

Molecular docking is a computational method that predicts how a small molecule (ligand) interacts with a larger molecule (receptor), usually a protein. The goal of molecular docking is to predict the optimal orientation and conformation of the ligand within the receptor, and to estimate the strength of the resulting ligand-receptor complex (binding affinity).

Rigid-body docking: a fast molecular docking method, due to it not accounting for flexibility.

Flexible docked can get more and more complex as you increase flexibility, but this increases the price.

Instead, a combination is used: rigid protein- flexible ligand screening.

Question 31

How are poses generated by molecular docking? How are these then scored? List the types of scoring functions.

Answer 31

The angles, torsions and orientations are altered using various algorithms (e.g., MD, stimulated annealing, Monte Carlo.)

A particular orientation and conformation are found and scored according to its relatedness to the true affinity value. The poses are then ranked and re-scored to search for a commonality in the scoring (validation).

1. Forcefield-based
2. Empirical
3. Knowledge-based

Question 32

How can the best molecular docking score be found?

Answer 32

By finding the energy minimum. This usually involves Monte Carlo simulations where different poses are evaluated and the decision to progress with a pose is based on ‘tossing of a coin’ and that ‘coin’ gets weighted so that the probability of accepting the pose decreases if the new system being accepted is worse.

Question 33

What are the cons of using pharmacophores?

Answer 33

1. Limited to known ligands (no novel compounds)
2. Inherent bias to known ‘chemotypes’

Question 34

What is an ‘inverse’ pharmacophore? Why are these useful?

Answer 34

A pharmacophore built from the protein, rather than the ligand.

These allow ligands to be overlayed and check that they meet the pharmacophore requirements for protein binding.

Question 35

What is the problem with generic scoring functions?

Answer 35

They’re only good for generating ideas and ruling out compounds that aren’t worth pursuing, but not for actually finding new compounds.

This is why no drugs have been developed from this method, even with the ‘screen-saver’ approach.

Question 36

What is the ‘screen-saver’ approach for molecular docking?

Answer 36

This exploits ‘dead time’ on computers and uses this power for large-scale molecular docking experiments. This was famously applied to the FightAIDS@Home campaign and is now used in other projects.

Question 37

Flexibility is very expensive to compute, so how can this problem be overcome? What are the benefits to this approach?

Answer 37

Using fragments, rather than whole ligands –> fragment-based drug discovery.

Instead of screening (either real or in silico compounds), fragments are screened.

+ Chemical diversity space is better covered
+ Better hit rates as less steric clash due to very small sizes
+ Compounds optimized from fragments have high binding efficiencies per atom i.e., most atoms are contributing, rather than being dead-weight

Question 38

What is fragment growing and fragment linking? Give an example of where each have been used to discover a drug.

Answer 38

In fragment-based drug discovery, fragments can either be grown within the target binding site, or multiple fragments can be linked together and optimized for high binding affinities.

e.g., CDK1 and 2 inhibitor AT7519 to treat solid tumors, and ABT-263 (Bcl-X inhibitor) to treat lung cancer.

Question 39

How are leads optimized?

Answer 39

1. Comparative structure-activity relationships (SAR and QSAR)
2. Structure-directed multiple parallel synthesis
3. Optimizing interactions between ligand and target based upon intermolecular forces, either by eye or via automated methods.

Question 40

What is the structure-activity relationship, and its related paradox?

Answer 40

Similar molecules have similar activities. The SAR paradox refers to the fact that it’s not the case that ALL similar molecules have similar activities.

Question 41

What is structure-directed multiple parallel synthesis?

Answer 41

Once you have your lead molecule, this can be used to optimize it when the lead molecule structure suggests the possibility of modifications at more than one site. Compounds are then synthesized with each possible combination of potential modifications to try to optimize the binding kinetics.

NB: sometimes the modification sites are called vectors.

Question 42

What strategies can be used to improve binding affinities of ligands to their targets in drug design?

Answer 42

1. Bury more hydrophobic surface between the ligand and the protein for an entropic gain when water is released.
2. Pre-form the bound conformation of the inhibitor so it doesn’t pay this cost when binding the protein.
3. Cater for any unpaired hydrogen bonding potential.
4. Form cooperative hydrogen bonds with water molecules.

Question 43

What is the free energy perturbation method in molecular dynamics? What can it be used for in drug design? What can you do when you don’t have a starting molecule?

Give an example where this was vital in drug discovery.

Answer 43

It can calculate the free energy difference between two states of a molecular system, giving information about the affinities for different lead compounds.

The basic idea of the FEP method is to use a series of simulation runs to gradually transform one state of the system into the other, while simultaneously measuring the free energy change associated with each step of the transformation. The free energy difference between the two states can then be calculated as the sum of the free energy changes measured in each step.

• If we don’t know any of the values, we can grow the presence of the ligand in solution and then solve this thermodynamic cycle, using absolute free energies.

E.g., BRD4 inhibitors that were predicted correctly, but scored very poorly. FEP calculations drastically improved this ranking.

Question 44

What is the con to free-energy perturbation calculations, and what trick can be used to overcome this?

Answer 44

Calculations are expensive, so if compounds look similar, you can calculate the difference between the two compounds instead.

Question 45

Why is it important to know where waters are within a protein-ligand interaction? Give an example of where this was important.

Answer 45

One of the strategies to improve the binding affinities is to displace the water to achieve an entropic gain, so you need to know where these waters are.

E.g., HIV protease contains a water right in the middle of its binding site, so inhibitors were designed to eliminate this, resulting in a large gain of affinity.

Question 46

Why aren’t all inhibitors also drugs?

Answer 46

Whilst inhibitors may work in theory, there’s no point in using them is the compound can’t actually get to the target site in vivo…ADME.

Question 47

What is Lipinski’s rule of 5?

Answer 47

1. A molecular weight of <500Da
2. No more than 5 hydrogen bond donors
3. No more than 10 hydrogen bond acceptors
4. A logP value <5

Question 48

Using a specific example, how are Trojan horse molecules being used to improve drug availability?

Answer 48

PepT2 is a proton-coupled peptide transporter within the gut. Acyclovir has poor bioavailability, but by adding a valine, the molecule can trick pepT2 into translocating it and thus improving uptake.

Could be used for blood-brain barrier drug delivery.

Question 49

How are drugs metabolised in the liver? Why is cytochrome P450 a problem for predicting drug metabolism?

Answer 49

Within the liver, lipophilic drugs and converted into hydrophilic metabolites that can then be excreted in urine or bile. This involves two main groups of enzymes, one of which is cytochrome P450.

There are many types of cytochrome p450 within the population, meaning there could be different metabolites which makes it hard to predict side effects.

It also has a large binding pocket, so more than one drug could bind and have drug-drug interactions.

As these substrates can be toxic, we must be able to predict them, either through prior knowledge of similar compounds, or via quantum mechanics.

Question 50

What is glaucoma, and how has structure-based drug design aided in treating it?

Answer 50

Glaucoma is a chronic eye condition, caused by increased pressure in the eye, that can cause blindness.

One of the main targets for glaucoma drug development is the enzyme carbonic anhydrase, which is involved in the production of aqueous humor, the fluid that fills the front part of the eye. By inhibiting carbonic anhydrase, the production of aqueous humor can be reduced, which can lower the pressure in the eye and help prevent optic nerve damage.

By studying the three-dimensional structure of carbonic anhydrase and its active site, researchers have identified specific structural features that can be targeted to develop highly specific inhibitors. For example, the development of dorzolamide, a widely used carbonic anhydrase inhibitor for glaucoma treatment, was aided by structure-based drug design.

Question 51

How has structure-based drug design aided the fight against HIV? What was the main issue with the first HIV drug to be produced? What alternative strategy was then used instead?

Give specific examples of the therapies now being used.

Answer 51

HIV protease is a key drug target, and structures show a unique architecture that’s distinct from mammalian proteases.

As the drug industry already had lots of details on other protease inhibitors, a pharmacophore could be generated. This was used to search chemical databases and generated various compounds. It was also noticed that there was a water group within the active site that could be replaced with methoxy groups for an entropic gain.

Whilst a iterations of SBDD did produce a drug, the severe side effects made it unusable as HIV patients are long-term.

Instead, a weak dipeptide inhibitor that was known to inhibit these proteases was modified. Several drugs have been made using this method, allowing for combination treatments to be used that overcomes clinical resistance:
1. Protease inhibitor + RTase inhibitor (e.g., AZT)
2. Protease inhibitor + non-nucleoside RTase inhibitor

Question 52

How has structure-based drug design aided in designing drugs to treat flu? What are the problems with the drugs that have so far been identified with this method?

Answer 52

One class of drugs developed using structure-based drug design to treat flu is the neuraminidase inhibitors (NAIs). NAIs work by inhibiting the activity of the neuraminidase enzyme, which is essential for the release of new influenza virus particles from infected cells.

Relenza was discovered by crystallizing it with neuraminidase, and then GRID was used to modify the compound, increasing the Ki. However, resistance against relenza was quick to occur, and had poor bioavailability.

Tamiflu was created by removing the oxygen from the Relenza ring structure, increasing lipophilicity for better bioavailability…however, it’s only thought to reduce symptoms by one day…so not really worth it.

Question 53

How has structure-based drug design aided in treatments against arthritis? How are these drugs delivered and why does this aid the treatment?

Answer 53

Arthritis is a result from excessive inflammation in the joints.

The inflammatory cascade involves MAPK and so has been a key target for many years. Most compounds compete with the ATP binding site, and there have been many successful drugs achieved using SBDD.

Structure work has also revealed a loop that covers a lipophilic pocket which, when targeted, has a far better selectivity.

MAPK inhibitors are mostly inhaled and work by having a slow off-rate, meaning it remains bound for far longer.

Question 54

How has structure-based drug design aided in treatments against hypertension?

Answer 54

Captopril is a medication used for the treatment of hypertension (high blood pressure) and congestive heart failure. It belongs to a class of drugs known as angiotensin-converting enzyme (ACE) inhibitors.

The researchers used a technique called “enzyme substrate analog design” to develop captopril. This involved creating a molecule that was similar in structure to the substrate that ACE normally acts upon (angiotensin I), but which had chemical modifications that would allow it to bind tightly to the active site of the enzyme and inhibit its activity.

Question 55

How was fragment-based drug discovery involved in developing venetoclax, a drug for treating leukemia? Describe the process.

Answer 55

Venetoclax targets a protein called B-cell lymphoma 2 (BCL-2), which is overexpressed in CLL and other cancers and helps cancer cells evade apoptosis (cell death).

1. Fragment screens against BCL-2 used NMR.
2. Leads were optimized
3. Improvements of pharmacodynamic and -kinetic properties
4. Clinical development

Question 56

How is personalized medicine being used? Explain how this is used for those being put onto warfarin.

Answer 56

The genotyping of drug targets and enzymes should allow for better clinical trial design and enable highly personalized drug optimization.

We know that in some patients with particular SNPs on CytP450 causes an adverse reaction. A trial tried to link the SNPs and the response to warfarin, so now you must be genotyped before you go onto warfarin.

Question 57

What are the overall pros and cons of structure-based drug design?

Answer 57

Pros:
- SBDD is more efficiency
- Allows atom by atom optimization
- Has good success with difficult targets
- It can be coupled with biophysics for identification of new binding sites and kinetics

Cons:
- expensive and requires lots of resources
- optimization can be achieved with other techniques

Question 58

Why isn’t AI going to solve all drug design problems?

Answer 58

AI needs the right questions, and won’t be able to actually reduce the clinical failure rate currently.