Lead Discovery Approaches
1. Random screening - only approach before 1935; screen every compound you have; still a useful approach; streptomycin and tetracyclines identified in this way
2. Nonrandom (or Targeted or Focused) screening - only screen compounds related to active compounds
3. Drug metabolism studies - metabolites produced are screened for the same or other activities
4. Clinical observations - new activities found in clinical trials; Dramamine tested as antihistamine (allergy) - found to relieve motion sickness; Viagra tested as antihypertensive - found to treat erectile dysfunction

5. Rational approaches - identify causes for disease states:
• imbalance of chemicals in the body
• invasion of foreign organisms
• aberrant cell growth
Identify biological systems involved in disease states; use natural receptor ligand or enzyme substrate as the lead; a known drug also can be used as a lead

Chemical imbalances - antagonism or agonism of a receptor; enzyme inhibition
Foreign organism and aberrant cell growth enzyme inhibition; DNA interaction

FIGURE 2.1
Depiction of dopamine (DA) in its role as a neurotransmitter. DA is released by a neuron prior to interacting with dopamine receptors (D1
–D5) on the surface of another nearby neuron. Also shown is the dopamine transporter, which terminates the action of dopamine by transporting the released neurotransmitter from the synaptic cleft back into the presynaptic neuron. Reprinted by permission from Macmillan Publishers Ltd: Nature Reviews Drug
Discovery (Kreek, M. J.; LaForge, K. S.; Butelman, E. Pharmacotherapy of addictions. Nat.
Rev. Drug Discov. 2002, 1 , 710
–726) Copyright 2002.

Serotonin, a mediator of inflammation, was used as the lead for the anti-inflammatory drug indomethacin.
OH
NH
2
CH
3
O
N
O
CH
3
HO
O
N
H
Cl indomethacin

Steroid hormones as lead compounds

Peptide hormones as lead compounds

Transporter ligands as lead compounds

Enzyme substrates as lead compounds
Physostigmine

Kinases as targets —ATP is the lead compound
SCHEME 2.1
Reaction catalyzed by the kinase class of enzymes. Kinases catalyze the transfer of the terminal phosphate group of ATP or related molecules acceptor to the group of a substrate, in this case, an alcohol (ROH).

Other and repurposed ligands as lead compounds
Antihypertensive -> Parkinson’s
AIDS -> antihistamine
Antidepressant -> back pain

• Natural products
• Collections of compounds
• High-throughput synthesis of a library of compounds
• Solid-phase synthesis
• Solution-phase synthesis

Natural products seem to provide greater structural diversity than combinatorial chemistry.
Natural products that are biologically active in assays generally have drug-like properties.
Construction of chemical libraries based on natural product hits is a good compromise.

Solid-Phase Synthesis of a Natural
Product-like Combinatorial Library
Library based on the 2,2-dimethylbenzopyran scaffold, a privileged structure.
R 1
R 2
R 3
R 4
2.55
OH
R 5
R 6
R 7
R 8
2.57
O
SeBr
R 2
R 3
R 1
R 4
O
H
Se
R 6
R 7
R 5
H H
O
Se
O
R 8
R 1
R 2
R 3 O
Se
R 4
R 5
H
2
O
2
R 6
R 7
R 8
2.56
O substituen t elaboratio n
Se

Solid-phase synthesis is used to make libraries of compounds
SCHEME 2.2
Solid-phase synthesis of a library of 7-hydroxybenzodiazepines

SCHEME 2.3
Solid-phase synthesis of a library of 4-alkoxyproline derivatives

Combinatorial Chemistry
Synthesis or biosynthesis of chemical libraries of molecules for lead discovery or lead modification
Libraries prepared in a systematic and repetitive way by covalent assembly of building blocks to give diversity within a common scaffold.
Synthesis carried out on a solid support (polymer resin).
Isolation and purification of product of each step done by filtration and washing. Everything not attached to polymer is washed away, which allows use of excess reagents.

Alternatives: Carry out reactions in solution with excess reagents, which are scavenged with a polymer-bound scavenger. Filtration removes excess reagents bound to polymer.
Or, use polymer-bound reagents.
Main differences among various methods of combinatorial library synthesis:
1. Solid support used
2. Methods for assembling the building blocks
3. The state (immobilized or in solution) in which libraries are screened
4. Manner in which structures of active compounds are determined (encoding methods)

# synthetic steps
x
(Eq. 2.1)
# compounds in library
# building blocks in each step
If the # of building blocks varies in each step, then
N = bcd...
For example, consider a 4-step synthesis. If 25 different reactants (building blocks) are used in the first two steps, 20 in step three, and 15 in step four, there would be 25  25  20  15 = 187,500 different compounds (from only 85 different building blocks).

Peptide Libraries
Split synthesis
Also called mix and split or split and pool method
Most common lead discovery approach for large libraries (10 4 - 10 6 compounds), assayed as library mixtures.
Produces a collection of polymer beads, each containing one library member (100-500 pmol)
A combinatorial library of all pentapeptides comprised of the 20 common amino acids: 20 5 = 3.2  10 6 peptides.
For example, prepare a homogenous mixture of all tripeptides of His, Val, and Ser (3 3 = 27)

Split Synthesis of All
Tripeptides Containing
His, Val, and Ser
Merr ifi eld
resin
N- pr ot ected amino acid
H
H
Merr ifi eld
resin
V
V
Merr ifi eld
resin
S
S combine, homogeni ze and divide ( i. e., mix and spl it) into 3 equal parts, t hen deprotect
H V S H V S H V S
N- pr ot ected amino acid activated w ith a coupling r eagent
H
H H H
H V S
V
V V V
H V S
S
S S S
H V S combine, homogeni ze and divide i nt o 3 equal part s, then depr ot ect
H H H
H V S
H H H
H V S
H H H
H V S
V V V
H V S
V V V
H V S
V V V
H V S
S S S
H V S
S S S
H V S
S S S
H V S
N- pr ot ected amino acid activated w ith a coupling r eagent
H V S li br ary of
9 t ripeptides li br ary of
9 t ripeptides li br ary of
9 t ripeptides combine and homogenize, then cl eave all off of solid suppor t to gi ve a combi nat or ial library of 27 differ ent tr ipept ides

Split Synthesis of Most Active Hexapeptide for a

Opioid Receptor
A combinatorial library is prepared of pentapeptides (comprised of the 20 main amino acids) bound to the MBHA polystyrene resin (makes peptide amides when cleaved off of resin)
Separate into 20 tubes, then add a different N -acetylamino acid to each tube
1.
N -Ac-Ala
2.
N -Ac-Cys
3.
N -Ac-Arg
XXXXX
XXXXX
XXXXX
N
H
N
H
N
H
MBHA resin
MBHA resin
MBHA resin
Each tube contains a combinatorial library of 3.2 x 10 different MBHA resin-bound N -acetylhexapeptides all having the same amino acid at the N -terminus.
6
20.
N -Ac-Thr XXXXX N
H
MBHA resin
• Assay the 20 combinatorial library mixtures (either still attached to bead or after cleavage)
• Set the N-terminal amino acid of most potent library constant

If the most potent was the library with Ac-Arg at N-terminus,
Set all to Arg Vary 2nd amino acid
Next 4 amino acids are a combinatorial library
1.
2.
3.
N
N -Ac-Arg
N
-Ac-Arg
-Ac-Arg
Ala-XXXX
Cys-XXXX
Arg-XXXX
N
H
N
H
N
H
MBHA resin
MBHA resin
MBHA resin
Each tube contains a combinatorial library of MBHA resin-bound N -acetylhexapeptides all with Ac-Arg at the
N -terminus, one of the 20 amino acids at the penultimate
N -terminus, and a random mixture of all possible amino acids at positions 3-6
20.
N -Ac-Arg Thr-XXXX N
H
MBHA resin
Determine best residue at 2nd position, then repeat process for 3rd position.
Continue until you get most potent N -acetyl hexapeptide amide (in this case, AcArg-Phe-Met-Trp-Met-Thr-NH
2
).

Problems with Split Synthesis Approach
Most potent analog may be less potent than what was observed in a previous iteration
• Peptide-peptide interactions may have been responsible for previous potency (some peptides removed in next step)
• Active compound may be a complex of more than one peptide (some peptides removed in next step)
• Conformation may be different with fewer peptides present

Nonpeptide Libraries
Problems with Assaying Multiple Compounds Together
Many false negatives (active compound that does not produce a hit) and false positives (inactive compound that does produce a hit) observed.
Therefore, single entity screens are used.
Parallel synthesis - solid-phase reactions carried out to make individual compounds rapidly (and robotically)
The library (50-10 4 compounds) is made in parallel without combining tubes so each support has one compound attached.
The goal is to get as much molecular diversity with as much functionality as possible.

SCHEME 2.4
Solution-phase synthesis of a library of furanose derivatives

Libraries can be made in microtiter plates
FIGURE 2.2
(A) Schematic of a typical 96-well microtiter plate. (Reprinted with permission from Custom Biogenic
Systems ( http://www.biomedicalmarketing.com/CBS/MicrotiterCRacks.html) .) (B) Picture of a 96-well microtiter plate taken by Jeffrey M. Vinocur, 4/21/06, published on Wikipedia Commons
( http://commons.wikimedia.org/wiki/File:Microtiter_plate.JPG
)

Libraries can be made in microtiter plates
FIGURE 2.4
Image of 96-well filter plates. Reprinted with permission from Norgen Biotek Corp.

Polymer-bound scavenger reagents can be used in synthesis
FIGURE 2.3
Product of polymer-bound N -methylglucosamine with borate anion

SCHEME 2.5
Use of a solid-phase scavenger in solution-phase synthesis. In this example, a polymer-bound isocyanate is used to scavenge excess primary or secondary amine from a solution by forming the corresponding polymer-bound urea.

An antitumor drug discovered by combinatorial synthesis and highthroughput screening

Advantages and problems in high throughput synthesis
• Solid-phase synthesis can make large numbers of compounds, but purification is problematic and quantities are small.
• Solution-phase synthesis makes smaller numbers of compounds, but larger quantities and easier purification.

Properties that Influence Oral
Bioavailability
Lipinski’s Rule of 5
The “rule of 5” states that poor oral absorption and/or distribution are more likely when:
• The molecular weight is > 500
• The log P is > 5
• There are more than 5 H-bond donors (expressed as the sum of OH and NH groups)
• There are more than 10 H-bond acceptors (expressed as the sum of N and O atoms)
It is highly likely (>90%) that compounds with two or more of these characteristics will have poor absorption properties.
Antibiotics, antifungals, vitamins, and cardiac glycosides are the exception because they often have active transporters to carry them across membranes.

Alternative to Lipinski’s Rule of Five
Veber and co-workers
Poor oral bioavailability found as a result of increased molecular flexibility (independent of molecular weight), as measured by:
• greater than 10 rotatable bonds
• high polar surface area (> 140 Å 2 )
• total hydrogen bond count (> a total of 12 donors and acceptors)
Both the number of rotatable bonds and hydrogen bond count tend to increase with molecular weight, which may explain Lipinski’s first rule.
Reduced polar surface area was found to correlate better with an increased permeation rate than did lipophilicity.

Privileged Structures and Drug-like Molecules
Certain scaffolds are capable of binding to multiple receptor targets
Appropriate structural modifications can change the activity
R
O
N
NHCOR'
N benzodiazepines
Y
Privileged structure
(a) Evans, B. E.; Rittle, K. E.; Bock, M. G.; DiPardo, R. M.; Freidinger, R. M.; Whitter, W. L.; Lundell, G. F.; Veber, D. F.; Anderson, P.
S.; Chang, R. S. L.; Lotti, V. J.; Cerino, D. J.; Chen, T. B.; Kling, P. J.; Kunkel, K. A.; Springer, J. P.; Hirshfield, J. J. Med. Chem. 1988 ,
31 , 2235. (b) Ariëns, E. J.; Beld, A. J.; Rodrigues de Miranda, J. F.; Simonis, A. M. in The Receptors: A Comprehensive Treatise ; O'Brien,
R. D., Ed.; Plenum Press: New York, 1979, p. 33. (b) Ariëns, E. J. Med. Res. Rev. 1987 , 7 , 367. (c) Patchett, A. A.; Nargund, R. P. Annu.
Rep. Med. Chem. 2000 , 35 , 289. (d) Hajduk, P. J.; Bures, M.; Praestgaard, J.; Fesik, S. W. J. Med. Chem. 2000 , 43 , 3443. (e) Fecik, R. A.;
Frank, K. E.; Gentry, E. J.; Menon, S. R.; Mitscher, L. A.; Telikepalli, Med. Res. Rev. 1998 , 18 , 149. (f) Horton, D. A.; Bourne, G. T.;
Smythe, M. L. J. Computer-Aided Mol. Des. 2002 , 16 , 415. (g) Klabunde, T.; Hessler, B. Chembiochem. 2002 , 3 , 928. (h) Matter, H.
Baringhaus, K.H.; Naumann, T.; Klabunde, T.; Pirard, B. Comb. Chem. High Throughput Screen. 2001 , 4 , 453. (i) Horton, D. A.; Bourne,
G. T.; Smythe, M. L. Chem. Rev. 2003 , 103 , 893.

Drug-likeness
Only 32 scaffolds describe half of all known drugs.
1
Average number of side chains per molecule is 4. If carbonyl is ignored, then 73% of the side chains in drugs are from top 20 most common side chains.
2
Drug-likeness may be an inherent property of some molecules.
3
Try these first.
1
2
Bemis, G. W.; Murcko, M. A.
J. Med. Chem.
1996 , 39 , 2887.
Bemis, G. W.; Murcko, M. A.
J. Med. Chem.
1999 , 42 , 5095.
3 Ajay; Walters, W. P.; Murcko, M. A.
J. Med. Chem.
1998 , 41 , 3314.

FIGURE 2.5
Pairs of compounds containing a privileged structure (indole, dihydropyridine, or benzimidazole) and binding to diverse target classes

Some groups should be avoided because of possible toxicity

Sucralose is an alkyl halide, but was approved as a non-nutritive sweetener

Many nondrug-like molecules show up as active in screens - false positives.
Appear to bind to numerous receptors, but not at the site of the natural ligand (called promiscuous drugs)
May be the result of aggregate formation

Virtual screening can accelerate lead identification
SCHEME 2.6
The process of virtual screening to identify compounds that conform to a hypothesis specifying properties (that are discernible from a compound’s structure) that are required for activity

Virtual screening databases are available
• Tripos LeadQuest
• Maybridge
• Accelrys ACD
• ZINC-a free database

Virtual screening process
• Identify substructure linked to desired activity
• Search database for molecules with active substructure
• 2D search just matches the structure
• 3D search can match molecular shape

Substructure search problems
FIGURE 2.6
Hypothetical example illustrating that substructure search (e.g., using pyridine as the search query) may not retrieve the most structurally similar compounds in a compound collection

How to quantify similarity?
T = N
11
/n-N
00
= 7/18-4 = 0.50

Computer-Based Methods of QSAR
3D-QSAR correlates diverse molecular structures and their biological function at a particular target.
General approach
• assay a set of molecules
• align molecules according to some predetermined orientation rules
• calculate a set of spatially dependent parameters for each molecule determined in the receptor space
• derive a function relating each molecule’s spatial parameters to their biological property
• establish self-consistency and predictability of the function

Comparative Molecular Field Analysis
( CoMFA )
• Molecule-receptor interactions are represented by steric and electrostatic fields exerted by each molecule
• A series of active compounds are identified and 3-D structural models constructed
• These structures are superimposed on one another and placed within a regular 3-D grid
• A probe atom placed at lattice points on grid where it is used to calculate steric and electrostatic potentials between itself and each superimposed structure
• A 3-D contour map is constructed to identify where lower or higher steric or electrostatic interactions increase binding

A pharmacophore based on known ligands
FIGURE 2.7
Example of a computer-generated pharmacophore model. From Laurini et al Bioorg.
Med. Chem. Lett. 2010 (Ref. 111)

Molecular Graphics
Use of computers to display and manipulate molecular structures based on Fischer lock-and-key hypothesis.
Structure-based drug design
Computer-assisted drug design
Computer-assisted molecular design synonymous names for the method
Commercial software, such as Sybyl (Tripos), Insight II (Molecular
Simulations), and Gold (Cambridge Crystallographic Data Center), are used.
Need a high-resolution (crystal or solution) structure of a receptor with a ligand bound. The ligand can be removed graphically to expose the receptor binding site.
Other molecules are docked into the binding site.

Haldol, a lead for HIV-1 protease inhibitors from docking

• Screening results in hits
• In HTS, need to identify and confirm structure of hits
• In virtual screening, need to perform assays to confirm activity
• A series of analogs may be prepared
• Ligand efficiency = ΔG/N

Fragment-based identification of leads

K i
1 μM 40 mM 19 mM 10 mM
But structural studies show that the fragments do not bind in the same site

SAR by NMR
NMR approach to identify and optimize high-affinity ligands bound to proteins
FIGURE 2.8
SAR by NMR methodology
Screen ~10 compounds at a time. Look for a 15 N-chemical shift in the protein
NMR indicating compound bound at that amino acid in the protein.
Once lead is identified, make a focused library of analogs to optimize lead.
Screen a 2nd library to find one that binds to a site near to where the first compound bound. Optimize the 2nd lead.
Link the two compounds together - binding affinity is dramatically increased.

Example of SAR by NMR
Inhibition of stromelysin, a matrix metalloprotease
(degrades extracellular matrix components in tissue remodeling and in arthritis, osteoporosis, and cancer)
Site 1 Site 2 Linker
K d
= 17 mM K d
= 20  M K d
= 17 nM
Took 6 months to identify 2.45
. Prior to this, no leads had been identified.

A clinical candidate from SAR by NMR

SAR by NMR
What’s Really Involved?
To see a 15 N chemical shift in NMR spectrum:
1. Overexpress protein in microorganism grown on 15 NH
4
Cl as sole N source to incorporate 15 N in all amino acids
2. Purify protein (need > 200 mg per spectrum)
3. Determine 3D structure of protein by various
NMR techniques (mass < 40 kDa)
If structure can be determined, SAR by NMR can screen 1000 compounds a day.

FIGURE 2.9
Ellman combinatorial methodology for lead generation with an unknown or impure protein

SAR by MS
High-throughput mass spectrometry-based screen
• A set of diverse compounds screened by mass spectrometry to identify those that bind to a receptor
• Competition experiments determine compounds bound to different sites (ternary complex if bound to different sites; binary complex if at same site)
• For molecules bound to different sites, vary substituent size to determine which bind at adjacent sites
• Attach the two molecules with a linker

FIGURE 2.10
Steps of SAS for identification of protease inhibitors

FIGURE 2.11
Three approaches to linking fragments: (A) fragment evolution, (B) fragment linking, and (C) fragment self-assembly. Reprinted with permission from Macmillan Publishers Ltd: Nature
Reviews Drug Discovery (Reese, D. C.; Congreve, M.; Murray, C. W.; Carr, R. Fragment-based lead discovery. Nat. Rev. Drug Discov. 2004, 3, 660 –672) Copyright 2004.

FIGURE 2.12
Example of fragment-based lead discovery incorporating the fragment evolution approach followed by lead modification to an optimized compound. Ligand efficiencies help guide the overall process.

Fragment self-assembly —click chemistry
FIGURE 2.13
Example of fragment-based lead discovery incorporating the fragment self-assembly approach: click chemistry

FIGURE 2.14
(A) Structure of 2.51
complexed with 6-phosphogluconate dehydrogenase (6PGDH) determined by X-ray crystallography. (B) Structure of 2.53
complexed with 6PGDH predicted by computational docking. From Ruda, et al. Bioorg. Med. Chem. 2010 , 18, 5056
–5062.

Ligands for 6-phosphogluconate dehydrogenase

Fragment-based development of a urokinase-type plasminogen activator inhibitor
FIGURE 2.15
Progression from mexiletin ( 2.55
, identified by fragment-based screening) to a potent orally bioavailable uPA

Lead Optimization
Pharmacodynamics receptor interactions structure of lead is similar to that of the natural receptor ligand or enzyme substrate
Pharmacokinetics ADME - Absorption,
Distribution, Metabolism, Excretion; depends on water solubility and lipid solubility
Toxicity

Identification of the Active Part of the Lead
First consider pharmacodynamics
Pharmacophore - the relevant groups on the compound that interact with the receptor and produce activity
Auxophore - the rest of the molecule

Some of the auxophoric groups interfere with binding of pharmacophoric groups and must be excised.
Some of the auxophoric groups may neither bind to the receptor nor prevent the pharmacophoric groups from binding - these can be modified without effect on potency.
Modification of these auxophoric groups may be used to correct ADME problems.

One approach to determine which are pharmacophoric atoms, which are auxophoric atoms hindering binding, and which are auxophoric atoms that can be modified is to cut away pieces of the lead and measure the effect on potency.

Example of Pharmacophore and Auxophore
Identification
Assume the well-known addictive analgesics below are the lead compounds. The darkened part is known to be the pharmacophore (you would not know that in a reallife example).
All bind to the  opioid receptor
What happens if we excise the bridging O (shown above not in the pharmacophore)?

3-4 times more potent than morphine
Note: The structure is drawn to mimic that of morphine; it is not the lowest energy conformer.
Maybe additional conformational flexibility allows the molecule to bind more effectively.

Next, remove half of the cyclohexene ring (also not in the pharmacophore).
Less potent than morphine, but much lower addictive properties
.
about the same potency as morphine

Removal of more rings
Demerol
10-12% the potency of morphine
Potency similar to morphine

If every cut gives lower potency, either most of the molecule is in the pharmacophore or the cut causes a conformational change away from the bioactive conformation. Add groups to increase the pharmacophore.
3200 times more potent than morphine

Fentanyl and derivatives are the most potent opiates known. Is the ester group in carfentanil part of the pharmacophore?
N fentanyl
100X morphine
N
O
O
O
N
N
O carfentanil
10,000X morphine

Antibacterial agent carbutamide ( 2.64
, R = NH
2 has an antidiabetic side effect.
)
(
Replacement of NH
2.64
, R = CH
2 by CH
3 gives tolbutamide
3
), which is an antidiabetic drug with no antibacterial activity.
An obvious relationship exists between molecular structure and its activity.

Antihypertensive with diuretic activity
Antihypertensive without diuretic activity

Structure-Activity Relationships (SARs)
1868 - Crum-Brown and Fraser
Examined neuromuscular blocking effects of a variety of simple quaternary ammonium salts to determine if the quaternary amine in curare was the cause for its muscle paralytic properties.
Conclusion: the physiological action is a function of chemical constitution

FIGURE 2.16
Structure of D -tubocurarine, a constituent of curare

Structurally specific drugs (most drugs):
Act at specific sites (receptor or enzyme)
Activity/potency susceptible to small changes in structure
Structurally nonspecific drugs:
No specific site of action
Similar activities with varied structures (various gaseous anesthetics, sedatives, antiseptics)

Example of SAR
Lead: sulfanilamide (R = H)
Thousands of analogs synthesized
From clinical trials, various analogs shown to possess three different activities:
• Antimicrobial
• Diuretic
• Antidiabetic

SAR
General Structure of Antimicrobial Agents
R = SO
2
NHR  , SO
3
H
• Groups must be para
• Must be NH
2
(or converted to NH
2 in vivo)
• Replacement of benzene ring or added substituents decreases or abolishes activity
• R can be
(but potency is reduced)
• R = SO
2
NR 
2 gives inactive compounds

Antidiabetic Agents
X = O, S, or N

SAR
Diuretics (2 types) hydrochlorothiazides
R 2 is an electrophilic group high ceiling type
(high level of diuresis)

SAR for Paclitaxel
Paclitaxel ( 2.72
, Taxol) - anticancer drug

SAR Conclusions
FIGURE 2.17
Use of a molecular activity map structural drawing of a lead annotated to show where structural changes affect activity or potency

Structural Modifications
Increase potency, therapeutic index and ADME
Increase therapeutic index - measure of the ratio of the concentration of a drug that gives undesirable effects to that which gives desirable effects e.g., LD
50
ED
50
(lethal dose for 50% of the test animals)
(effective dose to give maximum effect in
50% of test animals)
Therefore, want LD
50 to be large and ED
50 to be small .
The larger the therapeutic index, the greater the margin of safety.
The more life threatening the disease, the lower is an acceptable therapeutic index.

Therapeutic index = 23

Types of Structural Modifications
Homologation - increasing compounds by a constant unit (e.g.,
CH
2
)
Effect of carbon chain length on drug potency
Figure 2.18
Pharmacokinetic explanation : Increasing chain length increases lipophilicity and ability to cross membranes; if too high lipophilicity, it remains in the membrane
Pharmacodynamic explanation : Hydrophobic pocket increases binding with increasing length; too large and does not fit into hydrophobic pocket

in throat lozenges
Branched chain groups are less lipophilic than straight chain groups.

Chain Branching
Often lowers potency and/or changes activity; interferes with receptor binding
More potent than the tertiary homolog

10-Aminoalkylphenothiazines (X = H)
Promethazine-antispasmodic/antihistamine activities predominate
All bind to different receptors
Promazine-greatly reduced antispasmodic/antihistamine
Activities, greatly enhanced sedative/tranquilizing activities
Trimepazine-reduced tranquilizing activity enhanced antipruritic (anti-itch) activity

Bioisosterism
Bioisosteres - substituents or groups with chemical or physical similarities that produce similar biological properties. Can attenuate toxicity, modify activity of lead, and/or alter pharmacokinetics of lead.

Classical Isosteres

Non-Classical Isosteres
Do not have the same number of atoms and do not fit steric and electronic rules of classical isosteres, but have similar biological activity.

Examples of Bioisosteric Analogues

Changes in Activity by Bioisosterism
If the S in phenothiazine neuroleptic drugs ( 2.75
) is replaced by -CH=CH- or -CH
2
-CH
2
- bioisosteres, then dibenzazepine antidepressant drugs ( 2.76
) result.

Changes resulting from bioisosteric replacements:
Size, shape, electronic distribution, lipid solubility, water solubility, p K a
, chemical reactivity, hydrogen bonding
Effects of bioisosteric replacement:
1. Structural (size, shape, H-bonding are important)
2. Receptor interactions (all but lipid/H
2 important)
O solubility are
3. Pharmacokinetics (lipophilicity, hydrophilicity, p K a
H-bonding are important)
,
4. Metabolism (chemical reactivity is important)
Bioisosteric replacements allow you to tinker with whichever parameters are necessary to increase potency or reduce toxicity.

Bioisosterism allows modification of physicochemical parameters
Multiple alterations may be necessary:
If a bioisosteric modification for receptor binding decreases lipophilicity, you may have to modify a different part of the molecule with a lipophilic group.
Where on the molecule do you go to make the modification?
The auxophoric groups that do not interfere with binding.

FIGURE 2.19
Conformationally rigid analogs to determine bioactive conformation

Ring-Chain Transformations
Transformation of alkyl substituents into cyclic analogs, which generally does not affect potency.
Ring-chain transformation can have pharmacokinetic effects, such as increased lipophilicity or decreased metabolism.
2.80
and 2.81
have equivalent tranquilizing effects
2.82. and 2.83 have similar antipruritic activity
2.84 has better antiemetic activity than 2.80

Ring-chain transformations can affect toxicity toxic nontoxic

Peptidomimetics
Peptides are important endogenous molecules neurotransmitters, hormones, neuromodulators
Peptide drugs - analgesics, antihypertensives, antitumor agents
Peptides generally do not make good drug candidates
• rapidly proteolyzed in GI tract and serum
• poorly bioavailable
• rapidly excreted
• bind to multiple receptors
Peptidomimetic - a compound that mimics or blocks the biological effect of a peptide, but without undesirable structural characteristics

Use the peptide as a lead - modify to minimize undesirable pharmacokinetic properties
Try to mimic structure of the peptide when it is bound to the target receptor
Replace as much of the peptide backbone as possible with nonpeptide fragments - leave the pharmacophoric groups
Initially, retain conformational flexibility, but then refine to more conformationally-rigid analogs to hold groups in a bioactive conformation.

Captopril-inhibits angiotensin converting enzyme

Phenylalanine Peptidomimetics
FIGURE 2.20
Conformationally restricted phenylalanine analogs
Increased lipophilicity and conformational rigidity - better absorption and poor recognition by proteases

Conformationally-Restricted Peptides
FIGURE 2.21
Conformationally restricted dipeptide analogs

Secondary Structure Mimetics
FIGURE 2.22
Conformationally restricted secondary structure peptidomimetics
 -turns -helix
 -strand
 -loop

Scaffold Peptidomimetics
Arg-Gly-Asp (RGD) common  -turn motif that binds to receptors
FIGURE 2.23
RGD scaffold peptidomimetics

somatostatin agonist scaffold peptidomimetic

Thyrotropin-releasing hormone (TRH)
A scaffold peptidomimetic for TRH

Peptide Backbone Isosteres
Peptide amide bond replaced with alternative groups
(statine)

FIGURE 2.24
Hydrolysis of a peptide bond showing the tetrahedral intermediate arising from nucleophilic attack of water on the carbonyl group of the peptide bond. The hydroxyethylene isostere analog is designed to mimic the tetrahedral intermediate.

Structure Modifications to Increase Oral
Bioavailability
• ~ 75% of drug candidates do not go to clinical trials because of pharmacokinetic problems
• Less than 10% of drug candidates in clinical trials go to market
• ~ 40% of molecules that fail in clinical trials do so because of pharmacokinetic problems
Therefore, examine pharmacokinetic problems, such as oral bioavailability and plasma half life, as early as possible in the drug discovery process.

Low water solubility (high lipophilicity) can be an important limiting factor for oral bioavailability .
Highly lipophilic compounds are easily metabolized or bind to plasma proteins.
Low lipophilicity leads to poor absorption (cannot cross membranes).
Therefore, need to know lipophilicities of molecules and of substituents.
Determination of lipophilicities of substituents by
Corwin Hansch and co-workers is based on the
Hammett equation for electronic effects.

Electronic Effects
The Hammett Equation
Hammett’s postulate:
Electronic effects (both inductive and resonance) of a set of substituents should be similar for different organic reactions.
Therefore, assign values for the electronic effect of different substituents in a standard organic reaction, then use these values to estimate rates in a new reaction.

Hammett Equation
Standard system: benzoic acid derivatives
Equilibrium Constants
Scheme 2.7
As X becomes more e -withdrawing, the reaction to the right should be favored (increased K a
).

Rate Constants
Scheme 2.8
As X becomes more e -withdrawing, rate should increase because of transition state stabilization
(lower activation energy).

Linear Free-Energy Relationship
Rates of hydrolysis of ethyl benzoates
Dissociation constants of benzoic acids not for ortho substituentssteric/polar effects

log k k o
=  log
K
K o slope of line k o and K o are rate and equilibrium constants, respectively, for the parent compound (X = H)
If log then
K
K o log is defined as  , k k o
= 
Reaction constant (depends on reaction conditions) carbocation intermediate, - slope carbanion intermediate, + slope
Hammett equation
Electronic parameter (depends on electronic properties of substituent) the more e -withdrawing, more + the more e -donating, more -
H = 0.0

Lipophilicity Effects: Hansch Equation
Hansch thought there should be a linear free-energy relationship between lipophilicity and biological activity.
Action of drug depends on 2 processes:
1. drug has to get to site of action
(pharmacokinetics)
2. drug has to interact with site of action
(pharmacodynamics)

Fluid Mosaic Model of Membrane
Drug must pass through various membranes to reach the site of action
Figure 2.26
lipid bilayer
Correlations noted earlier between lipid solubility and biological activity hydrophilic heads (OH,
NH
3
+ , sugars) lipophilic tails (steroid and hydrocarbon chains)

Measured Lipophilicities
Model for transport of drug to site of action
Ability of compound to partition between
1-octanol (simulates membrane) and water (simulates cytoplasm)
Measure of lipophilicity: partition coefficient
(P) between 1-octanol and water:
[compound]
1-octanol
P =
[compound] water
(1 )
(Eq. 2.7) degree of dissociation in H
2 from ionization constants
O

concentration of drug that produces the biological effect
C
2 + k  (log P) + k 
For a compound more soluble in H
2 octanol P < 1; log P is -
O than in 1-
For a compound more soluble in 1-octanol than in H
2
O
P > 1; log P is +
The larger the P, the more interaction of the drug with membranes (more lipophilic)

Parabolic Relationship Between
1
C optimum partition coefficient for biological activity log P
0 should be  2 to penentrate
FIGURE 2.27
Effect of log  P on biological response. P is the partition coefficient, and C is the the CNS oncentration of the compound required to produce a standard biological effect. Log P o optimal log P for biological activity.
is the

Ionization of compounds leads to greater water solubility than predicted from the neutral structure.
For compounds that can ionize (COO , NH
3
+ , etc.) use log D - log of distribution coefficient.
Ionization is a function of p K a of a compound and pH of the medium; therefore, must specify pH.
log D is the log P of an ionizable compound at a particular pH.
log D
7.4
is the log P at pH 7.4

Effect of pH on log D fully protonated amine decreased concentration of protonated amine

p
Derived the same as electronic substituent constants  p = log P
X
- log P log P for a compound with substituent X
H
= log
P
X
P
H log P for a compound with no substituent

p (like  ) is additive and constitutive
Multiple substituents exert lipophilicity equal to the sum of individual substituents
Effect depends on the molecule to which it is attached
Alkyl groups are least constitutive p
CH
3
≈ 0.50
By definition, p
H
= 0
Therefore, p
CH
2
= p
CH
3

How does folding affect π values?

Log P = 2 π
CH3
+ 2 π
CH2
+ π
CH=CH
+ 2logP
PhOH
= 2(0.50)+2(0.50)+0.69+2(1.46)-0.40
-0.40
=5.21 (experimental, 5.07)

Example of Additivity of p
Constants
Branching lowers log P or p by
0.2/branch branching log P
2.83
= 2 p
Ph
+ p
CH
+ p
OCH
2
+ p p
OCH
2
= log P
CH
3
CH
2
OCH
2
CH
3
2 p
CH
3
CH
2
p
+ p
NMe
2
CH
2
0.2
= 0.77 - 2(0.5) - 0.5 = -0.73
branching p
NMe
2
= log P
Ph(CH
2
)
3
NMe
2
p
Ph
3 p
CH
2
= 2.68 - 2.13 - 3(0.5) = -0.95
log P
2.83
= 2(2.13) + 0.5 + (-0.73) + 0.5 + (-0.95) - 0.2
= 3.38
Experimental log P = 3.27

Computerization of Log P Values
Several software packages are available to calculate log P.
However, log P values can vary by 2 or more units among software packages.
p values are constitutive - cannot account for all substituents on every scaffold.
Measure log P for one member of a library, then see which software package comes closest to the experimental value.

Effects of Ionization on Lipophilicity and Oral
Bioavailability
Amino acid residues in receptors are ionized
Asp/Glu -COOH p K a
4 - 4.5
Cys -SH 8.5 - 9
Tyr
OH
9.5 - 10
Ser/Thr
His
-OH
HN
NH
13.5 - 14
6 - 6.5
Lys -NH
3
+
NH
2
+
10 - 10.5
Arg
NH
2
12 - 13
N
H
At physiological pH (pH 7.4) each is partially ionized and partially neutral.
pH = pKa –log [HA]/[A-]

Ionization of a drug depends on p K a of groups and pH .
Ionization has profound effect on lipophilicity
(pharmacokinetics) and interaction with a receptor
(pharmacodynamics).
What if ionization of a drug is important for its binding to a receptor, but ionization may block its ability to cross membranes?
Need to vary p K a to adjust equilibrium of ionized
(bind to receptor) and neutral (cross membranes) form.
Neutral form crosses membranes, then re-establishes equilibrium with ionized form on other side.

Ionized molecules that did not cross the membrane re-establish equilibrium with the neutral form, which can cross the membrane.
Drug metabolism and excretion prevent continual re-establishment of equilibrium.
The p K a of most alkaloids that act as neuroleptics, local anesthetics, and barbituates have values between 6 and 8.
At neutral pH there is a mixture of neutral and cationic forms.

Aminacrine is more active at pH > 7

The uricosuric drug phenylbutazone has a p K a is active as an anion.
of 4.5 and
SCHEME 2.10
Ionization equilibrium for phenylbutazone
The pH of urine is ~ 4.8.
Sulfinpyrazone (R=CH
2
CH
2
SOCH
3
)has a p K a of 2.8
Therefore all in anionic form.
20 times more potent than phenylbutazone

Effect of ionization can be rationalized from pharmacokinetic or pharmacodynamic perspective .
Pharmacokinetic - neutral form increases crossing of membranes
Pharmacodynamic - neutral form binds in a hydrophobic pocket in receptor
These can be differentiated using an isolated receptor (no membranes to cross so only pharmacodynamics) and whole cells or tissue
(membranes to cross so pharmacokinetics and pharmacodynamics are important).

Pyrimethamine is absorbed as the neutral molecule, but active as the protonated ion

In a cell-free system (no membranes), the antibacterial activity of sulfamethoxazole is directly proportional to the degree of ionization
(pharmacodynamics).
SCHEME 2.11
Ionization equilibrium for sulfamethoxazole
In intact cells, where a drug must cross a membrane to get to the site of action, the antibacterial activity is proportional to its lipophilicity (neutral).

p K a
Actual p K a
Values values depend on the microenvironment
Molecular dynamics simulations indicate that the interior of proteins have dielectric constants about
2-3 (like benzene or dioxane); water has a high dielectric constant (78.5).
COOH in a nonpolar environment - p K a because ionized form is destabilized.
higher
Asp-99 in 3-oxo 5 -steroid isomerase is 9.5!
Represents a change in equilibrium of 10 5 in favor of neutral form.

COOH that forms a salt bridge - p K a lower
Multiple COOH groups adjacent - p K a avoid adjacent anionic groups higher to
NH
2 in a nonpolar environment - p avoid cationic form
K a lower to
NH
2 in a salt bridge is stabilized - p K a higher
Multiple basic groups adjacent - p K a adjacent cationic groups lower to avoid
Lys amino group in acetoacetate decarboxylase has p K a
5.9 because adjacent residue is another Lys.
Again about 10 5 change in equilibrium.

Such large changes in p K a in microenvironment indicate that lead modification also should involve large p K a variances.
Determine in vitro and in vivo effects with p K a changes.

CNS drugs must cross the blood-brain barrier
Inactive Active Active

Quantitative Structure-Activity
Relationships
1868 - Crum-Brown and Fraser predict a mathematical relationship would be found between structure and activity
1962 - Corwin Hansch attempts to quantify effects of substituent modifications

QSAR
Basis for quantitative drug design:
Biological properties are a function of the physicochemical parameters, e.g., solubility, lipophilicity, electronic effects, ionization, stereochemistry, etc.
 and p constants should be important considerations in lead modification.
Steric factors also should be important for receptor binding.

Steric Effects: Taft Equation
Taft defined a steric parameter E s
Reference reaction:
XCH
2
CO
2
Me H
3
O + XCH
2
CO
2
H + MeOH
E s
(X = H) = 0
E s
= log k
XCH2Me
CH
3
- log k
CH3CO2Me
= log k
0
X
Hancock: E c s
= E s
=0.306(n-3) corrects for hyperconjugation

Correlation of Physicochemical Parameters
(descriptors) with Biological Activity
Hansch analysis
Because there are at least two considerations for drug activity, lipophilicity (to get drug to the site of action) and electronic factors (at the site of action), Hansch derived: log
1
C
= -k p 2 + k p +  + k  potency regression coefficients
Hansch Equation

Hansch equation generalized: log
C potency
= -a p 2 + b p +  + cE s
+ dS + e lipophilicity constant electronic constant steric constant other physicochem. parameters
Linear multiple regression analysis
The best least-squares fit of the dependent variable
(biological activity) to a linear combination of independent variables (descriptors) is determined.

Topliss Operational Schemes - Lead
Optimization
Nonmathematical, nonstatistical, noncomputerized use of Hansch principles.
Consider benzenesulfonamide as lead (R = H)
Vary p and  and determine effect on potency.
This method requires an unfused benzene ring.

Topliss Decision Tree
Because potency often increases with lipophilicity, first try a substituent with a + p value (e.g., Cl; p
4-Cl
= 0.71,

4-Cl
= 0.23).
Three possible outcomes - more potent (M), equipotent
(E), or less potent (L) than parent.
If more potent, can be because of + p , +  , or both.
Hold one constant and vary other: p
4-PhS
= 2.32, 
4-PhS
= 0.18 would test lipophilicity p
4-CF3
= 0.88, 
4- CF3
= 0.54 would test electron withdrawal
If 4-PhS compound is more potent than 4-Cl, increase lipophilicity more.

Figure 2.16
Topliss Decision Tree
4-H
L E M
4-Cl
L E M
4-PhS
L
L
4-CF
3
E
E M
M

If 4-Cl analog is equipotent with parent, may be a counterbalance of favorable + p and unfavorable +  or vice versa.
Try 4-Me p
4-Me
= 0.56, 
4-Me
= 0.17 or
4-NO
2 p
4- NO2
= -0.28, 
4- NO2
= 0.78
If 4-Cl analog is less potent than parent, there may be a steric problem at C-4 or need p and/or  .
Try 3-Cl p
3-Cl
= 0.71, 
3-Cl
= 0.37

Craig plot of σ and π values
FIGURE 2.29
Craig plot of σ constants versus π values for aromatic substituents. This material is reproduced with the permission of John Wiley & Sons, Inc. From Craig, P. N. (1980) In
"Burger’s
Medicinal Chemistry," (M. E. Wolff, ed.), 4th ed., Part I, p. 343. Wiley, New York.

Batch Selection Methods
With Topliss operational scheme you need results from one compound before you know what to make next.
Batchwise methods involve making a small group of preselected analogs and testing.
Rank order the potencies, then compare with the rankings in
Table 2.17.
This table tells you which parameter is most important.
E.g., if rank is 3,4-Cl
2
> 4-Cl > H > 4-Me > 4-OMe, then  is most important.

Then consult Table 2.18 for other substituents with that parameter.
Cannot extend this method without computation.

Cluster Analysis
Select one member from each cluster
See which dominates

Donepezil was studied by QSAR

BA = a i
X i
a i
X i
+ μ
= contribution of a to BA
=1 or 0
μ = activity of parent skeleton
R
1
= R
If R
5
2
= R
3
= R
4 is Cl or CH
3 and Cl > CH
3
Then assume Cl always better, regardless of other substituents

FIGURE 2.30
Example of scaffold hopping to identify new cholesterol-lowering statins from mevastatin

FIGURE 2.31
A variation of scaffold hopping that involves disconnection of bonds, rotation or flipping of the core, and reconnection of the pharmacophoric groups

Discovery of the influenza drug zanamivir (Relenza TM )
A protein (hemagglutinin) at the surface of a virus binds to sialic acid residues on receptors at the host cell surface.
Sialic acid Lead
The virus has another protein (neuraminidase), which is an enzyme that cleaves sialic acid residues from the cell surface
The virus enters the cell and replicates in the nucleus.

Progeny virus particles escape the cell, which no longer has sialic acid residues on its surface to trap the virus.
The virus particles spread to other cells.
Inhibition of neuraminidase would prevent the release and spread of virus particles.

Random screen identified 2.129
(R =
OH) as a weak inhibitor.
Crystal structure of influenza A neuraminidase with inhibitor bound was obtained.
Computation (using GRID program) suggested 4-OH should be replaced by 4-
NH
2
, which when protonated would interact with Glu-119 (Fig 2.32).
From the crystal structure, extension of
4-ammonium with 4-guanidinium ( 2.93
, zanamivir) would extend binding to Glu-
227 as well (Fig. 2.32).
NH
3
+ group
NH
2
NH NH
2 group
FIGURE 2.32
Crystal structure of neuraminidase active site with inhibitors bound. (A) Interaction of the protonated amino group of 2.129
(R = NH
3
+ ) with Glu-119. (B) Interaction of the protonated guanidinium group of 2.127
with Glu-119 and (red arrow) Glu-227. Adapted with permission from Macmillan
Publishers Ltd: Nature (von Itzstein, M., et al. Rational design of potent sialidase-based inhibitors of influence virus replication.
Nature 1993, 363 , 418 –423) Copyright 1993.

Generally, drug discovery is not this easy - need a cyclic approach.
• Identify receptor binder
• Obtain crystal structure of receptor with molecule bound
• Do calculations and docking
• Synthesize new compounds
• Assay
• Obtain another crystal structure with more potent binder
• Repeat cycle

Computer modeling can be used to identify the pharmacophore.
By overlaying the structure of the antitumor drug paclitaxel ( 2.72
, Taxol) with those of 4 other natural products that also stabilize microtubules in competition with paclitaxel,
(see next slide)

FIGURE 2.33
Five natural products found to promote stabilization of microtubules. The boxed sections were used to identify a common pharmacophore. With permission for I. Ojima (1999).
Reprinted with permission from Proc. Natl. Acad. Sci. USA 1999, 96 , 4256
–4261. Copyright 1999
National Academy of Sciences, U.S.A.
a common pharmacophore (Fig. 2.34, next slide) was identified.

FIGURE 2.34
Common pharmacophore based on the composite of boxed sections in
Figure 2.33. With permission for I.
Ojima (1999). Reprinted with permission from Proc. Natl. Acad.
Sci. USA 1999, 96 , 4256
–4261.
Copyright 1999 National Academy of
Sciences, U.S.A.
A hybrid structure was then constructed ( 2.130
) .

The design of a bryostatin analogue