Improving Gleevec: Insight from the Receptor Structure

Gleevec cannot bind to the open (active) form of the Abl kinase - would collide with open conformation of the activation loop

Remove portion of molecule causing steric clash with the open (active) conformation of the activation loop

Arrived at new class of drug, PD17, predicted to still bind competitively at ATP-binding site of the Abl kinase

Gleevec PD17

6 H-bonds 2 H-bondscontacts 21 residues contacts 11 residuesIC50 = 100 nM IC50 = 5 nM

Despite fewer interactions with Abl, the drug PD17 is a better inhibitor

Inactive Abl + PD17 Active Abl + PD17

Despite making fewer contacts with the target protein, PD17 is a better inhibitor because it binds to both conformations of Abl

- Thus, losing an H-bond but removing Gleevec’s steric clash with the open conformation led to an improved drug

Improving PD17

PD17 loses the H-bond to threonine 319 that is essential for Gleevec’s activity; however, this residue remains available for H-bonding near the end of PD17

Can PD17 be improved by engineering a new H-bond to Thr319?

PD17

Hydroxyl group contributes new H-bond even better binding (IC50 = 0.4 nM)

Better binding than Gleevec(IC50 = 5 nM)

PD17

PD166326

With further improvements:Dasatinib, first 2nd-generation kinase inhibitor

Gleevec

With further improvements:Dasatinib, first 2nd-generation kinase inhibitor

- 325 times more effective than Gleevec against normal CML cancer

- effective against tumors expressing 14 out of 15 resistance mutations (all but the dreaded Thr-315 Isoleucine)

With further improvements:Dasatinib, first 2nd-generation kinase inhibitor

Tokarski et al. Cancer Res. 2006

- dasatinib binds to both active and inactive conformations

Phe-382-stackswith Gleevecpyrimidinering, locksactivationloop in inactive conformation

Gleevec occupies a

hydrophobicpocket that

is otherwisefilled by Phe-382

Tokarski et al. Cancer Res. 2006

Receptor-Based Design

Knowing that BCR/ABL fusion protein is the specific cause of CML...

(1) Identify a small molecule that selectively inhibits this kinase (Gleevec)

(2) Perform structural studies to understand mechanism of action: - discover new mode of drug action: selective binding to inactive kinase structure (varies from kinase to kinase)

(3) Use structural information to make a drug that binds either conformation (PD17)

(4) Through a second round of structural studies, add H-bonding interactions to optimize the inhibitor (PD166326)

Receptor-Based Design

Knowing that BCR/ABL fusion protein is the specific cause of CML...

(5) Create 2nd generation drug – Dasatinib

More effective than Gleevec because:

a) binds both active and inactive forms..

b) causes few distortions of protein, compared to ATP-bound form.. c) makes fewer interactions with P-loop & other parts of ABL..

Receptor-Based Examples

1. Targeting a single protein essential for disease progression

Improving Gleevec, a new anti-cancer drug

2. Taking advantage of unique features of a protein target

Prophylactic Inhibition of Cholera Toxin

Disease: Cholera (caused by bacterium Vibrio cholerae) Traveler’s diarrhea (E. coli)

- combined, kill over 1 million people per year

Target: pentameric protein toxins

The pathogenic bacteria V. cholerae and E. coli affect humans by producing a protein toxin that forms a pentamer

- toxin has 5 identical subunits that come together in a star-shape - released in lumen of the intestine - each of the 5 units binds to an oligosaccharide on epithelial cell surfaces, gaining entry into the cell

Strategy: design inhibitors to block binding of receptors to natural ligand on cell surface, thus preventing toxin from entering

Step 1: Design a small galactose mimic that binds the toxin as a single-site inhibitor, based on the receptor’s structure

Natural ligand of cholera toxin is an oligosaccharide ending in a terminal galactose sugar

Substitutions wouldn’t work at O3, O4; each acts as H-bond donor & acceptor with protein side chains

Also, no substitutions at O6, which is bonded to 2 waters

GluLys

Trp

Asn

H2O

H-bond acceptor H-bond donor

Step 1: Design a small galactose mimic that binds the toxin as a single-site inhibitor, based on the receptor’s structure

Substitutions would work at O1, O2 - only lose 1 H-bond, to a displaceable H2O

35 galactose analogues purchased + tested to see if they could inhibit binding of natural ligand to the toxin protein

7 had lower IC50’s than galactose itself

GluLys

Trp

Asn

H2O

H-bond acceptor H-bond donor

Step 1: Design a small galactose mimic that binds the toxin as a single-site inhibitor, based on the receptor’s structure

Most potent inhibitor was m-nitrophenyl--D-galactoside (MNPG)

Step 1: Design a small galactose mimic that binds the toxin as a single-site inhibitor, based on the receptor’s structure

Most potent inhibitor was m-nitrophenyl--D-galactoside (MNPG)

- retains favorable binding interactions of the natural ligand

- nitrophenyl group displaces a water molecule

- structure-based design came up with an inhibitor Kd of 10 M, a 100-fold improvement over galactose alone...

- however, still much lower than the affinity for the natural ligand

Options for designing high affinity protein inhibitors:

(1) Make a drug that binds tightly to the binding site - 5 molecules must bind per toxin pentamer, independently

(2) Make a penta-valent inhibitor, that is, one molecule with 5 inhibitory “fingers” linked to a central core

- 1 molecule binds per toxin pentamer, but fingers bind semi- cooperatively

In multivalent binding, binding of 1 finger aligns other fingers with their receptor sites

- this increases the overall binding affinity, by decreasing entropic costs associated with multiple ligands binding independently

- linkers can also make favorable contacts with the protein surface, further promoting binding

allows you to make a potent inhibitor even if the fingers on their own aren’t such good binders

each low affinity high affinity strong binding

vs.

Step 2: Determine whether making a pentavalent ligand improves binding

Multi-valent drug design is a strategy to get higher binding affinity by exploiting the presence of multiple, identical binding sites on a target protein

- for instance, many proteins are multimeric, meaning composed of several identical subunits

- design a single, large molecule which presents multiple copies of an inhibitor, arranged to jam all binding sites on the target

Step 2: Determine whether making a pentavalent ligand improves binding to cholera toxin

Attach galactose to a scaffold, using flexible linkers to space out 5 sugar residues joined to a central core

galactose

scaffold

flexible linker arm (R1)

each one of these arms is the same as the one shown above

Step 2: Determine whether making a pentavalent ligand improves binding

Attach galactose to a scaffold, using flexible linkers to space out 5 sugar residues joined to a central core

IC50 (M)

Galactose-based finger, alone 5,000

Galactose-based pentavalent ligand 16

Step 3: Combine the 2 ways to improve binding: make a pentavalent ligand using the improved galactose derivative

Attach m-nitrophenyl--D-galactoside (MNPG) to a scaffold, with linkers to position the fingers over the 5 binding sites of the pentamer

Step 3: Combine the 2 ways to improve binding: make a pentavalent ligand using the improved galactose derivative

Attach m-nitrophenyl--D-galactoside (MNPG) to a scaffold, with linkers to position the fingers over the 5 binding sites of the pentamer

IC50 (M)

Galactose-based finger 5,000

Galactose-based 16 pentavalent ligand

MNPG finger alone 195

MNPG pentavalent ligand 1

pentavalent ligand shows ~200-fold improvement over the best single-site derivative

Yellow = MNPG ligand

Green = 1 arm of pentavalent ligand

Red = a water molecule that forms hydrogen bonds w/ natural galactose + protein amide;

- displaced by an oxygen of the inhibitor’s nitrophenyl ring

Pentavalent ligand fills the toxin pocket in similar manner as the free MNPG inhibitor, but with the higher binding affinity that comes with multivalency

Step 4: Continue to improve binding affinity: change scaffold

(1) Improve fit of linkers

- make more rigid: less conformations, binding is more entropically favored

- enhance interactions with protein surface

- present linker makes van der Waals contacts w/ side chains glu, tyr, his, lys, arg

(2) Increase valency: go from penta-valent (5 ligands) to deca-valent (10 ligands)

Now design a drug that will bind to 2 toxin pentamers simultaneously

green = natural ligand (oligosaccharide w/ terminal galactose)

blue = 1 arm of pentavalent (5-armed) ligand

brown = 1 arm of decavalent (10-armed) ligand