Designing tomorrow’s drugs

30th January 2010

Adrian MulhollandCentre for Computational Chemistry

School of Chemistry, University of Bristol

Why do we need new drugs?• For emerging diseases (e.g. to combat

new strains of flu; to circumvent bacterial antibiotic resistance)

• To avoid side-effects of current drugs• More specific drugs tailored to individual

patients (e.g. based on genetic differences)

• For diseases without current effective treatments

Drug discovery and development is a slow process

• Currently it takes approximately 15 years to go from an idea to a marketable drug

• Investment of £100s millions needed for each drug

• New drug approvals are decreasing alarmingly

A crisis in drug discovery

B. Hughes, Nature Reviews Drug Discovery 8, 93-96 (February 2009).

1

• Fewer new drugs are being approved

We need new, better ways to design and develop new drugs

• Computer-aided design and molecular modelling can help

• New methods based on quantum mechanics

• More accurate

What are drugs?• E.g. ibuprofen – an ‘over-the-counter’ painkiller

Ibuprofen – a common painkiller

• Like most drugs, ibuprofen is a small molecule

How does ibuprofen work?• Like most drugs, ibuprofen is a small molecule, and binds to a large protein molecule in the body

• The protein is an enzyme – a biological catalyst; its job is to make molecules by a chemical reaction

• Ibuprofen stops the enzyme working• It is an enzyme inhibitor

A biochemical pain signal

• This enzyme adds oxygen to make a hormone molecule

O2

How does ibuprofen work?

Transmits pain signals to the brain, causes inflammation

Ibuprofen inhibits (blocks) this enzyme, stopping the pain signal from being made

Enzymes are biological catalysts

• Almost all chemical reactions in a cell are catalysed by enzymes

• All aspects of biochemistry depend on enzyme catalysis

• Catalysts make reactions happen faster but are not changed by the reaction

Energy

Reactants

Products

Activation energy: energy barrier to

reaction

Transition state

HH

H

Cl

HH

H

Cl

HH

H

ClCl

Transition state

+ Cl-Cl

-+

-

Enzymes make reactions faster by lowering the energy barrier

Energy

Reactants

Products

Reaction without enzyme

Reaction with enzyme: lower energy barrier

They can do this by stabilizing the transition state, i.e. binding to it strongly

Enzymes as drug targets• Many drugs work by inhibiting

enzymes• But how do they work – at the

molecular level?• What interactions are involved?• Knowing how enzymes catalyse

reactions can help in the design of new drugs

The active site•A small part in the enzyme where the chemical reaction happens

Enzyme inhibitors as drugs

New drugs from understanding how enzymes work

• Enzymes bind transition states tightly

• Design molecules that resemble the transition state

• Should bind strongly to enzyme active site

Enzyme

Active site

Transition state

Transition state

analogue drug

Transition state analogues as drugs

To design a drug, we need to know:

• The structure of the protein (e.g. enzyme) target

• Knowing the structure of the transition state for the reaction in the enzyme, and how it interacts with the enzyme, should also help a lot

How can we find out what proteins look like?

• Can determine protein structure by X-ray crystallography

• To do this, you need a crystal of the protein…

• …and X-rays!

Protein crystals

Dr. Toshiya Senda, Dept. of BioEngineering, Nagaoka University of Technology, Japan http://bio.nagaokaut.ac.jp/~senda/welcome.html

‘Diamond’ synchrotron, Harwell

X-rays from synchrotrons

Protein structure from X-ray diffraction

Different ways of representing protein structure

Many drugs are enzyme inhibitors• The drug binds at the active site and stops

the enzyme from working

Ibuprofen bound to its enzyme target

Ibuprofen

Why do we need computer modelling?

Can do things that experiments can’t: • Model how chemical bonds break and

form, i.e. model reactions in enzymes• Model transition state structures• Model how proteins move and flex• Modelling can study proteins ‘in action’• Predict how tightly new drugs will bind

University of Bristol supercomputer: ‘BlueCrystal’

• Among the top 100 most powerful in the world

Modelling antibiotic breakdown in a bacterial enzyme

Modelling antibiotic breakdown in a bacterial enzyme

Modelling antibiotic breakdown• Understand molecular mechanisms of antibiotic resistance

• Identify which groups in the enzyme are responsible for catalysing the reaction

• Model the transition state• Design modified antibiotics to overcome bacterial resistance

The bliss molecule• Anandamide (ananda is Sanskrit for bliss)• Released naturally in the body in

response to pain• An ‘endocannabinoid’

Natural pain relief• Stopping the breakdown of

anandamide relieves pain • Anandamide is broken down by fatty

acid amide hydrolase • Inhibitors of fatty acid amide hydrolase

are potentially useful drugs• Clinically useful aspects of marijuana

without side-effects

Fatty acid amide hydrolase• The enzyme that breaks down

the bliss molecule• Modelling shows how

anandamide is broken down• Shows how inhibitors bind to

the enzyme• Helping in the design of new,

better medicines

Modelling shows how the drug binds to the enzyme

• Model of URB597 as it reacts and binds to the enzyme

Natural pain relief• URB597: an inhibitor with pain relief

properties, ready to enter clinical trials

Tamiflu (Oseltamivir)

Tamiflu

Influenza neuraminidase

•Flu enzyme, drug target

•Large, complex, difficult to model

Tamiflu molecule binds at enzyme active site

It binds because it is a transition state analogue

Neuraminidase inhibitors as flu drugs

Calculate which molecule binds more tightly to the protein

• i.e. which is the better potential drug?

Modelling drug metabolism

Modelling drug metabolism• Drugs are broken down by enzymes• Aim to predict how drugs interact

with each other, or other substances, in the body

• E.g. grapefruit juice contains enzyme inhibitors that slow drug breakdown!

Modelling drug metabolismUse molecular modelling of reactions of

drugs in enzymes to help to predict: • Toxicity• Side effects• Genetic effects (in future: tailor drug and

dose to the patient)• Adverse drug reactions

Modelling ibuprofen metabolism

Reaction coordinate [Å]

Ener

gy [k

cal/m

ol]

0.5 1.0 1.5 2.0 2.5 3.0 3.50

10

20

30

40

50

pathway 4 - site 1pathway 4 - site 2pathway 5

Computer-aided drug design produces drugs that save lives

• Nelfinavir: HIV protease inhibitor

Thanks to:University of Bristol• Dr. Christine Bathelt• Dr. Johannes Hermann• Dr. Richard Lonsdale• Dr. Kara Ranaghan• Dr. Christopher Woods• Dr. Jolanta Zurek• Katie Shaw

Prof. Jeremy HarveyDr. Fred Manby

Thanks to: University of Parma• Dr. Alessio Lodola • Prof. Marco MorUniversity of California-Irvine• Prof. Daniele PiomelliHeinrich Heine Univers. Düsseldorf• Prof. H.-D. Höltje

Funding: EPSRC (Engineering and Physical Sciences Research Council)

Thanks to:

Further information• Biomolecular simulation (Journal of the

Royal Society Interface special issue, freely available online:Introduction; Status, progress and prospects

• A model solution to depression? (RSC Chemical Biology highlight on FAAH)

• Rewriting the biochemistry textbooks• QM/MM calculations on enzymes• Designing the drugs of tomorrow