Thermodynamics for medicinal chemistry design

Thermodynamics for medicinal chemistry design

Peter W Kenny

http://fbdd-lit.blotspot.com | http://www.slideshare.net/pwkenny

Things that make drug discovery difficult

• Having to exploit targets that are weakly-linked to

human disease

• Poor understanding and predictability of toxicity

• Inability to measure free (unbound) physiological

concentrations of drug for remote targets (e.g.

intracellular or on far side of blood brain barrier)

Dans la merde, FBDD & Molecular Design blog :

Molecular Design

• Control of behavior of compounds and materials by manipulation of molecular properties

• Sampling of chemical space

– For example, does fragment-based screening allow better control of sampling resolution?

• Hypothesis-driven or prediction-driven

– There’s more to molecular design than making predictions (from Molecular Design blog): link

Montanari, Propopczyk, Sala, Sartori (2013) JCAMD 27:655-664 DOI

Kenny JCIM 2009 49:1234-1244 DOI

New year, new blog name, Molecular Design blog

TEP = log10([𝐷𝑟𝑢𝑔 𝒓,𝑡 ]𝑓𝑟𝑒𝑒

𝐾𝑑)

Target engagement potential (TEP) A basis for pharmaceutical molecular design?

Design objectives• Low Kd for target(s)• High (hopefully undetectable) Kd for antitargets• Ability to control [Drug(r,t)]free

Kenny, Leitão & Montanari JCAMD 2014 28:699-710 DOI

Property-based design as search for ‘sweet spot’

Green and red lines represent probability of achieving ‘satisfactory’ affinity and‘satisfactory’ ADMET characteristics respectively. The blue line shows the product ofthese probabilities and characterizes the ‘sweet spot’. This way of thinking about the‘sweet spot’ has similarities with molecular complexity model proposed by Hann et al.

Kenny & Montanari, JCAMD 2013 27:1-13 DOI

In tissues

Free in

plasma

Bound to

plasma

protein

Dose of drug

Eliminated drug

Simplifed view of what happens to drugs after dosing

ΔH-TΔS

Molecular interactions and drug action

Molecular design frequently focuses on structural relationships between compounds

Tanimoto coefficient (foyfi) for structures is 0.90

Ester is methylated acid Amides are ‘reversed’

Relationships between structures as framework for analysing activity and properties (G)

?

Date of Analysis N logFu SE SD %increase

2003 7 -0.64 0.09 0.23 0

2008 12 -0.60 0.06 0.20 0

Mining PPB database for carboxylate/tetrazole pairs suggested that bioisosteric replacement wouldlead to decrease in Fu . Tetrazoles were not synthesised even though their logP values are expected tobe 0.3 to 0.4 units lower than for corresponding carboxylic acids.

Birch et al (2009) BMCL19:850-853 DOI

Amide N logS SE SD %Increase

Acyclic (aliphatic amine) 109 0.59 0.07 0.71 76

Cyclic 9 0.18 0.15 0.47 44

Benzanilides 9 1.49 0.25 0.76 100

Effect of amide N-methylation on aqueous solubility is dependent on substructural context

Birch et al (2009) BMCL 19:850-853 DOI

Thermodynamics and molecular interactions

The contribution of an intermolecular contact (or group of

contacts) to affinity (or the changes in enthalpy, entropy,

volume or heat capacity associated with binding) is not an

experimental observable

Where does ITC fit into medicinal chemistry design?

• Direct, label-free method for measuring binding affinity

– Primary project assay

– Validation of higher-throughput project assays

– Input for computational affinity models

• Thermodynamic signature (H, TS) may be more

sensitive than G for detection of discontinuities in SAR

– e.g. change in binding mode within structural series

On enthalpic optimization

• How do isothermal systems like live humans sense the

benefits of an enthalpically-optimized drugs?

• Why would we expect measurement of H and S for

binding of compound to a protein to be predictive of the

behaviour of the compound in the absence of the

protein?

There’s a reason why we say standard free energy

of binding

G = H - TS = RTln(Kd/C0)

• Adoption of 1 M as standard concentration is

arbitrary

• A view of a chemical system that changes with

the choice of standard concentration is

thermodynamically invalid (and, with apologies to

Pauli, is ‘not even wrong’)

Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOIEfficient voodoo thermodynamics, FBDD & Molecular design blog

Rules, guidelines and metrics

• It’s not a rule, it’s a guideline… OK why did you call it a rule?

• Strength of a trend tells us how rigidly we should adhere to guidelines based on that trend

• Think carefully about physicochemical basis of guidelines and metrics

– Using logD to define compound quality metrics suggests that compounds can be made better by simply increasing the extent of ionization

Introduction to ligand efficiency metrics (LEMs)

• We use LEMs to normalize activity with respect to risk factors

such as molecular size and lipophilicity

• What do we mean by normalization?

• How predictive are risk factors of bad outcomes?

• We make assumptions about underlying relationship between

activity and risk factor(s) when we define an LEM

• LEM as measure of extent to which activity beats a trend?

Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOILigand efficiency metrics considered harmful, Molecular design blog

Scale activity/affinity by risk factor

LE = ΔG/HA

Offset activity/affinity by risk factor

LipE = pIC50 ClogP

Ligand efficiency metrics

There is no reason that normalization of activity with respect to risk factor should be restricted to either of these functional forms.

Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOI

Use trend actually observed in data for normalization

rather than some arbitrarily assumed trend

Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOI

Can we accurately claim to have normalized a data set if we have

made no attempt to analyse it?

Green: line of fitPurple: constant LEBlue: constant LipE

NHA Kd/M C/M (1/NHA) log10(Kd/C)

10 10-3 1 0.30

20 10-6 1 0.30

30 10-9 1 0.30

10 10-3 0.1 0.20

20 10-6 0.1 0.25

30 10-9 0.1 0.27

10 10-3 10 0.40

20 10-6 10 0.35

30 10-9 10 0.33

Effect on LE of changing standard concentration

Analysis from Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOINote that our article overlooked similar observations 5 years earlier by

Zhou & Gilson (2009) Chem Rev 109:4092-4107 DOI

Scaling transformation of parallel lines by dividing Y by X

(This is how ligand efficiency is calculated)

Size dependency of LE in this example is consequence of non-zero intercept

Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOI

Affinity plotted against molecular weight for minimal binding

elements against various targets in inhibitor deconstruction

study showing variation in intercept term

Data from Hajduk (2006) JMC 49:6972–6976 DOI

Each line corresponds to a different target and no attempt has been

made to indicate targets for individual data points. Is it valid to

combine results from different assays when using LE?

Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOI

Offsetting transformation of lines with different slope and

common intercept by subtracting X from Y

(This is how lipophilic efficiency is calculated)

Thankfully (hopefully?) lipophilicity-dependent lipophilic

efficiency has not yet been ‘discovered’

Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOI

Water

Octanol

pIC50

LipE

Thermodynamics and lipophilic efficiency

Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOI

There are two problems with this approach. Firstly octanol, is not ideal non-polar reference statebecause it can form hydrogen bonds with solutes (and is also wet). Secondly, logP does notmodel cost of transfer from water to octanol for ligands that bind as ionized forms

logP

Linear fit of ΔG to HA for published PKB ligands

Data from Verdonk & Rees (2008) ChemMedChem 3:1179-1180 DOI

HA

Δ

G/

kcal

mo

l-1ΔG/kcalmol-1 0.87 (0.44 HA)

R2 0.98 ; RMSE 0.43 kcalmol-1

-ΔGrigid

Ligand efficiency, group efficiency and residuals plotted for PKB binding data

Res

id|

GE

GE/kcalmol-1HA-1

Resid/kcalmol-1

LE/kcalmol-1HA-1

Residuals and group efficiency values show similar trends with pyrazole (HA = 5) appearing

as outlier (GE is calculated using ΔGrigid ). Using residuals to compare activity eliminates

need to use ΔGrigid estimate (see Murray & Verdonk 2002 JCAMD 16:741-753 DOI) which is

subject to uncertainty.

Use residuals to quantify extent to which activity beats trend

• Normalize activity using trend(s) actually observed in data (this means we have to model the data)

• All risk factors can be treated within the same data-analytic framework

• Residuals are invariant with respect to choice of concentration units

• Uncertainty in residuals is not explicitly dependent of value of risk factor (not the case for scaled LEMs)

• Residuals can be used with other functional forms (e.g. non-linear and multi-linear)

Kenny, Leitão & Montanari (2014) JCAMD 28:699-701 DOI

Some stuff to think about

• We need to make a clear distinction between what

we know and what we believe

• If we do bad data analysis then how will we be able

to convince people that drug discovery is really

difficult?