What is a Drug by student19


									What is a Drug?

Realistic Design Cycle
Leads: The Good, the Bad & the Ugly
There are important differences between a given compound
from a library, selected lead compound and a marketed drug.

The ‘drugability’ depends on a number of criteria such as:
•   specificity, solubility, liphophilicity, bioavailability
•   molecular weight, flexibility, synthetic route,
•   drug target and tissue specificity, blood brain barrier
•   etc

Industry has additional ‘requirements’:
•   Inexpensive to synthesize and manufacture
•   Distinct ‘clinical endpoints’
•   Reimbursable
•   No global regulatory or patent issues
•   To market 3 years from discovery (more like 7 - 10?)
        Lipinski’s rule(s) of five

• Experimental and computational approaches for
  estimation of solubility and permeability of new
  candidate compounds.
• Identifying calculable parameters of the selected
  compound library, related to absorption and
   Main sources of drug leads
       1970’s and 1980’s

Around 1970 – large empirically based screening
From then on – knowledge base grew for rational drug
Most leads had already been in a range of physical
properties previously known to be consistent with oral
   Main sources of drug leads
          1989 and on
HTS enabled screening of hundreds of
thousands of compounds across in-vitro
Soon after – combinatorial chemistry.
Rapid progress in molecular genetics –
expression of receptors.
Drugs were dissolved in DMSO (dimethyl
    HTS - High Throughput Screening

In vivo

In vitro
Solubility of leads
  In DMSO, even very insoluble drugs could be tested.
  As a result – in vitro activity could be detected in
  compounds with very poor thermodynamic solubility
  The physico-chemical profile of leads does not depend on
  compound solubility
  A reliable method to improve in-vitro activity –
  incorporating properly positioned lipophilic groups that
  can occupy a receptor pocket
  Adding a polar group that is not required for binding can
  be tolerated if it does not add to receptor binding.
  Therefore – compounds are more easily detected in HTS
  if they are larger and more lipophilic.
Selected parameters for testing

   Molecular weight – known relationship
   between poor permeability and high
   molecular weight.

   Lipophilicity (ratio of octanol solubility to
   water solubility) – measured through LogP.

   Number of hydrogen bond donors and
   acceptors – High numbers may impair
   permeability across membrane bilayer
            Partition coefficient – LogP

 •     The ratio of the equilibrium concentrations of a dissolved
      substance in a two-phase system containing two largely immiscible
      solvents (water and n-octanol)

                 C  water 
                  C  oct .
Since the differences are usually on a
     very large scale, Log10(P) is used.
 The (First) rule of five: Lipinski

Poor absorption or permeation are
          more likely when:

  There are more than 5 H-bond donors.

  The molecular weight is over 500.

  The LogP is over 5.

  There are more than 10 H-bond acceptors.
        MLogP – Moriguchi’s correction

•   A straightforward counting of lipophilic atoms and hydrophilic
    atoms account for only 73% of the variance in the experimental
    LogP. Therefore, corrections should be applied
                                        LogP and LogD
Log D, Distribution Coefficients

Log D is the log distribution coefficient at a particular
pH. This is not constant and will vary according to the
protogenic nature of the molecule. Log D at pH 7.4 is
often quoted to give an indication of the lipophilicity of
a drug at the pH of blood plasma.

Distribution Coefficient

D=             [Unionised] (org)
      [Unionised](aq) + [Ionised](aq)

Log D = log10 (Distribution Coefficient)
               ClogP – Calculated LogP

   Method to calculate log P(ow)
     from structure by an additive-
     constitutive procedure

| SMILES: CC(=O)Oc1ccccc1C(O)=O                                               |
| Class    | Type |    Log(P) Contribution Description     | Comment | Value |
|Fragment | # 1 | Ester [aA]                               |Measured | -1.180|
|Fragment | # 2 | Carboxy (ZW-) [a]                        |Measured | -.030|
|Carbon    |      | 1 aliphatic isolating carbon           |          |   .195|
|Carbon    |      | 6 aromatic isolating carbons           |          |   .780|
|ExFragment|Hydrog| 7 hydrogens on isolating carbons       |          | 1.589|
|ExFragment|Bonds | 1 chain and 0 alicyclic (net)          |          | -.120|
|Electronic|SigRho| 2 potential interactions; 2.00 used    |WithinRing|   .220|
|Ortho     |Ring 1| 1 normal ortho interaction             |          | -.430|
|RESULT    | 4/20+| All fragments measured                 | CLOGP    | 1.023|
    The (Present) rule of five: Merck
Poor absorption or poor permeation often occur when there are:

•    More than 5 H-bond donors.
•    Mw is over 500 D.
•    The CLog P is over 5 (or MLogP is over 4.15).
•    The sum of N’s and O’s is over 10.

Substrates for transporters and natural products are exceptions.

  Computational calculations for
     new chemical entities

Applied to entities introduced between 1990-1993
Average values:
  H-bond donor sum=2.53
  Molecular weight =408
  H-bond acceptor sum=6.95
Alerts for possible poor absorption-12%
                (quantitive structure activity relationship)

A QSAR model is a multivariant mathematical relationship between a set
of physicochemical or structural properties (descriptors) and a property
of the system being studied, such as the biological activity, solubility,
lipophililcity or transport behavior.

Data mining exercises using information on >2Mio compounds
                                            Virtual Screening
   Database preparation
 ligand structures, QSAR,
  ionization, error checks

    Database prefilters                                    Receptor preparation
physchem properties, QSAR,                                 Energy minimization, 
   reactive substructures                                     grid calculation

                                 pro­leads docking,

                             scoring, graphical browsing
                                  other descriptors


             Docking / Arrimage
         de Ligands dans les Protéines
- molécule qui interagit avec la protéine
- DNA, substrats et ses analogues, drogues ou ses
précurseurs, etc.
Centre(s) actif(s) de Protéine(s)
- interaction de nature alostérique
- interaction de nature competitive
Fonction de l'interaction
- naturelle ou artificielle
    C'est quoi le Docking de
   Ligands dans les Protéines?
     Prevision par des méthodes computationnelles les structures
   de complexes entre ligands et protéines en partant d'un
   ensemble varié de conformations et orientations. L'orientation
   du ligand qui maximise l'interaction aura potentiellement une
   correspondence avec la structure réelle du complexe

Importance des complexes
   Etablir des relations structure -> fonction
    Différents Aspects du Docking

Structure et Centre Actif de la Proteine
- connaissance a priori (PDBs, etc.)
- Base de données PROCAT: matrices 3D de centres actifs

Structure du Ligand
- substrat et/ou son analogue
- Précurseur de substance active (fragment de base = Lead)
- groupes chimiques connus

Docking Rigide ou Flexible
- solution ou dans le vide
- structure
 Approches Algorithmiques au Docking

 Complement de forme et ajustement

 Calcul d'énergie
 Determination du minime globale d'énergie
 Estimation de l'énergie libre

 Complémentarité géométrique et énergétique
 Approche à 2 phases : docking approximatif et docking fin
    Composantes Algorithmiques Communes
    - l'espace des solutions est typiquement très vaste

    - resolution basse vs. grande

    - la flexibilité oblige une recherche dans un espace de dimension N

    Evaluation (Scoring)
    - evaluation des résultats obtenus (configurations) en termes d'énergie
    totale et/ou écart quadratique moyen (RMSD - root mean square

    - minimisation de l'énergie potentielle
    - peut tenir em compte la flexibilité, solvent, entropie, hydrophobicité,
    interactions de van der Waals, interactions électrostatiques, etc
              Preparation du Docking
Utilization d'une Representation Interne
    conversion des coordonnées 3D des deux molécules (ligand et protéine)
- ex:
        representations de la surface accessible (Connolly ou autre)
        AutoGrid / Grid / ChemGrid...
        D'autres grilles avec des valeurs discrètes (docking rigide)
 GRID          Goodford 1985

Identification de Sites d'Interaction avec les Protéines
La fonction d'énergie est:  E =Eij + Eel + Ehb
         Eij - Lennard Jones   Eel   Électrostatique   Ehb Liaison d'hydrogène

Recherche avec des sondes (probes)
    H2O, NH3+, “O- Carboxylique”

La sonde se déplace sur une grille
L'énergie d'interaction est calculée à chaque point de la
L'énergie est décrite sous la forme d'un contour 3D
Les energies negatives correspondent aux sites
d'interaction favorables
  AutoDock         Olson 1990             • Angle de liaison de H

                                          • Description des Torsions

• 1) Approche Monte-Carlo (version 2)

• 2) Algorithmes Génétiques (version 3)

            2   3
                      4       OH

H       1           NH2

            2   3
                      4       OH

H       1           NH2
DOCK 1, 2, 3, 4 et 5      Kuntz 1992-présent

Design de Drogues Basé sur des Descripteurs
de Forme / Matrice 3D

Descripteurs sphériques du site d'interaction
(binding site)

Ligand rigide pendant le screening

Permet le docking de leads, auxquels d'autres
fragments extra pourront s'ajouter – Design
de Ligands / Drogues
Prediction of Ligand Binding Energies for
  Virtual Screening in Ligand Discovery

                           Good fit
                           Bad fit

            X-ray crystallography
            and Modelling
                                      Potential leads, selected
                                      for biological screening
  Lead optimisation,
  Med. Chem., libraries
          Methods for Estimation of Binding
Quality                                    Throughput
           • Molecular descriptors /

           • Docking: Ludi, FlexX, Gold,
             Dock etc.

           • Force Field Energies

           • QM/MM

           • Free Energy Perturbation
           The Virtual Screening Approach

                              Good fit
                              Bad fit

High Rank
                                                          Top ranks selected
                                                          for biological screening

Low Rank
             Equal distribution          Virtual Screen
Docking and Scoring for Virtual Screening

 • Data set of 360 thrombin compounds:
   37 compounds < 40 nM
   50 compounds < 100 nM
   80 compounds < 500 nM
   280 compounds > 500 nM - mM, “inactives”

 • Fully automated docking (GOLD) and scoring,
   medium throughput to maintain good quality

 • By Chance 3 - 4 actives in top 10% ranks
              Enrichment of Actives towards Top

      Equal    GOLD   GOLDmin MW   DOCK   FlexX   BLEEP FF vdW FF (E)
        Virtual Screening: Discussion Points
$ Docking and scoring are amongst the most complex problems in Molecular

$ Because of the complexity of binding interactions, a good dock and score does
   not imply a good binding energy

$ Not very successful at pin-pointing single active compounds from a set of
   compounds, problem of false positives

 Inactive templates with good fit in the binding site can be used to generate
   binding hypothesis and lead to targeted libraries

 Docking and scoring does achieve significant enrichment

 Docking programs are of good quality, scoring is the bottleneck

 A good dock indicates a better than chance probability of binding
? Do scoring functions mainly score size and shape-fit?

To top