01.04.2015 Views

Pharmacophore Modeling and Database Searching

Pharmacophore Modeling and Database Searching

Pharmacophore Modeling and Database Searching

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>Pharmacophore</strong><br />

<strong>Modeling</strong> <strong>and</strong> <strong>Database</strong><br />

<strong>Searching</strong><br />

Process<br />

• Collect set of high-affinity lig<strong>and</strong>s for target<br />

(from literature)<br />

• Analyze conformations of each (start from<br />

least flexible)<br />

• Superpose pharmacophore elements to<br />

identify bioactive conformation<br />

• Examine lower affinity lig<strong>and</strong>s to identify<br />

disallowed volume<br />

Why > 1 Conformation?<br />

Conformational Analysis<br />

• Requires:<br />

• a method to calculate (relative) energies<br />

• an algorithm to explore conformational space<br />

H 3C<br />

CH 3 CH 3<br />

CH 3<br />

Energy<br />

CH 3<br />

CH 3<br />

Computing Energies<br />

Energies can be calculated at several levels of theory<br />

• Ab Initio – PC Spartan<br />

• Most theoretically rigorous<br />

• Energies calculated from electronic structure<br />

• Requires no experimental parameters<br />

• Semi-Empirical – PC Spartan<br />

• Simplifying assumptions made<br />

• Experimental parameters compensate<br />

• Molecular Mechanics – MOE <strong>and</strong> PC Spartan<br />

• Electrons essentially ignored<br />

• Many experimental parameters required<br />

Molecular Mechanics<br />

• Energy broken down into terms<br />

• Bond stretching<br />

• Angle bending<br />

• Torsional potential<br />

• Non-bonded interactions<br />

• Van derWaals, electrostatic, dipolar interactions<br />

1


Force Fields<br />

• The combination of mathematical formulae<br />

<strong>and</strong> parameters used to represent the energy<br />

of a chemical system<br />

• Different forcefields are optimized for different<br />

problems:<br />

• MMFF94: optimized for small organic compounds<br />

- wide structural variety<br />

• AMBER94: optimized for proteins - often missing<br />

parameters for other organics<br />

• PEFSAC95: optimized for carbohydrates<br />

Bond Stretching<br />

• Approximated with a harmonic potential<br />

• V = k s<br />

( r - r 0<br />

) 2<br />

• Two parameters per pair of atom types<br />

harmonic potential<br />

Energy<br />

}<br />

Interatomic distance<br />

Morse potential<br />

huge errors at relatively large<br />

interatomic distances<br />

Torsional Potential<br />

s=1<br />

Typical of sp 3 -sp 3 bond<br />

n=3<br />

Typical of sp 2 -sp 2 bond<br />

s=-1<br />

n=2<br />

0 60 120 180 240 300 360<br />

• Eω = V n<br />

(1 + s cos nω)<br />

(Two--fold term added to compensate for non-equivalent minima)<br />

Three parameters needed for each combination of atom types<br />

Van derWaals<br />

• Usually expressed as a Lennard-Jones potential<br />

(6-12 shown):<br />

12 6<br />

⎡⎛σ<br />

⎞ ⎛σ<br />

⎞ ⎤<br />

V VDW<br />

= 4ε<br />

⎢⎜<br />

⎟ − ⎜ ⎟ ⎥<br />

σ r 0<br />

⎢⎣<br />

⎝ r ⎠ ⎝ r ⎠ ⎥⎦<br />

repulsive attractive<br />

Energy<br />

Interatomic distance<br />

V VDW<br />

r 0 = 2 -1/6 s<br />

12<br />

6<br />

⎡⎛<br />

r<br />

⎤<br />

0 ⎞ ⎛ r0<br />

⎞<br />

= ε ⎢⎜<br />

⎟ − 2⎜<br />

⎟ ⎥<br />

⎢⎣<br />

⎝ r ⎠ ⎝ r ⎠ ⎥⎦<br />

Ionic Interactions<br />

• Generally approximated using partial point<br />

charges<br />

q 1<br />

r<br />

q 2<br />

q1q<br />

=<br />

Dr<br />

V<br />

2<br />

Effective dielectric constant<br />

Conformational Search<br />

Algorithms<br />

• Torsion angle driving<br />

• Monte Carlo<br />

• Artificial Intelligence<br />

• Molecular Dynamics<br />

• Simulated Annealing<br />

• Poling<br />

• Etc.<br />

2


Monte Carlo<br />

• Uses a r<strong>and</strong>om kick of coordinates followed by<br />

minimization to find new minima<br />

• More effective on highly flexible molecules<br />

• Not exhaustive -> heuristics used to define end point:<br />

• if each of the lowest energy conformations has been found<br />

~10 times, the search has probably found all the interesting<br />

ones<br />

• if duplicate conformations are found ~20 times in a row, the<br />

interesting conformations have probably all been found<br />

• (actual numbers to use depend on the flexibility of the<br />

system <strong>and</strong> your interest in a nearly exhaustive search!)<br />

• In MOE there are two methods that include this<br />

• Compute -> Conformations -> Stochastic Search<br />

• Compute -> Conformations -> Hybrid Monte Carlo<br />

Exercise<br />

• Build a structure for morphine <strong>and</strong> perform a<br />

stochastic conformational search<br />

• Use Window->Potential Control to activate the MMFF94<br />

force field<br />

• Use default conditions (your choice on output database<br />

name) to run the search<br />

• Analyze your results<br />

• How many conformations?<br />

• What energy range?<br />

• How different are they (might try Edit->Interactive<br />

Superpose)<br />

Process<br />

• Collect set of high-affinity lig<strong>and</strong>s for target<br />

(from literature)<br />

• Analyze conformations of each (start from<br />

least flexible)<br />

• Superpose pharmacophore elements to<br />

identify bioactive conformation<br />

• Examine lower affinity lig<strong>and</strong>s to identify<br />

disallowed volume<br />

<strong>Pharmacophore</strong> Element<br />

Representations<br />

• While a pharmacophore is defined based on lig<strong>and</strong><br />

structure, activity occurs due to an interaction with a<br />

receptor<br />

• <strong>Pharmacophore</strong> elements can be represented as:<br />

• Lig<strong>and</strong> Points<br />

• Site Points<br />

Morphine <strong>Pharmacophore</strong><br />

Elements<br />

• Choose a conformation of morphine from<br />

your search as the ‘bioactive’ conformation<br />

• What are the distances between potential<br />

pharmacophore elements?<br />

HO<br />

O<br />

N<br />

Refining a <strong>Pharmacophore</strong><br />

• Most pharmacophore modeling cases do not<br />

include a molecule as inflexible as morphine<br />

• These cases require consideration of several<br />

structures to identify a most likely bioactive<br />

conformation for each<br />

• Compute->Flexible Alignment can be used to<br />

find superimpositions of several structures<br />

based on types of pharmacophore element<br />

HO<br />

Morphine<br />

3


Exercise – Flexible Alignment<br />

• Use flexible alignment to superimpose one of the<br />

flexible structures below on your selected<br />

conformation of morphine (fix morphine atoms using<br />

Edit->Fix)<br />

• Compute->Flexible Alignment can be used to find<br />

superimpositions of several structures based on<br />

types of pharmacophore element<br />

H 2N CH C<br />

CH 2<br />

O<br />

H<br />

N<br />

CH<br />

H<br />

O<br />

C<br />

H<br />

N<br />

CH<br />

H<br />

X=methionine or leucine<br />

O<br />

C<br />

H<br />

N<br />

CH<br />

CH 2<br />

O<br />

C<br />

X<br />

O<br />

N<br />

N<br />

OCH 3<br />

Sufentanil<br />

S<br />

Using a <strong>Pharmacophore</strong><br />

• Distances between pharmacophore elements<br />

can be used as input to 3D database<br />

searches<br />

• Public 3D database searching available<br />

through the National Cancer Institute (link on<br />

course home page)<br />

• <strong>Pharmacophore</strong> elements can be drawn<br />

• Distance ranges can be specified<br />

• Additional constraints (MW range, properties) can<br />

be used to reduce number of potential hits<br />

OH<br />

Why Distance Ranges?<br />

Accessible Conformations<br />

1. 3D databases contain a limited set of<br />

representative conformations, not every<br />

accessible conformation<br />

2. Queries defined by lig<strong>and</strong> points are overly<br />

restrictive<br />

OH<br />

OH<br />

N<br />

OH<br />

N<br />

N<br />

OH<br />

Dopamine D2 agonist<br />

that defines the 3D<br />

pharmacophore<br />

N<br />

Lig<strong>and</strong> Point Restriction<br />

Exercise<br />

• Go to the NCI database<br />

• Define a query using the morphine<br />

pharmacophore (<strong>and</strong> any other limitations<br />

you like)<br />

• What do you find?<br />

4


Related Reading<br />

• The Organic Chemistry of Drug Design <strong>and</strong><br />

Drug Action<br />

• Chapter 2.2A<br />

• Problems 2.4: 3<br />

• Textbook of Drug Design <strong>and</strong> Discovery<br />

• Chapter 4<br />

5

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!