Modelling receptor interactions is of significant interest to the scientific community, with many computational tools available. However, current tools are designed for the prediction of on-target effects and are widely used in the pharmaceutical industry, where compounds are routinely screened for binding affinity to only a single receptor of interest.
This was a poster which was presented at the 2nd Annual Drug Discovery USA Congress, 29-30 October 2015, Boston, USA by Will Krawszik, Maja Aleksic, Paul Russell and Jonathan G.L. Mullins from Moleculomics
Development and evaluation of in silico toxicity screening panels
1. Development and
evaluation of
in silico toxicity
screening panels
Will Krawszik1, Maja Aleksic2, Paul Russell2,
Jonathan G.L. Mullins3
2. Background - Modelling receptor interactions is of significant interest to the scientific community, with many
computational tools available. However, current tools are designed for the prediction of on-target effects
and are widely used in the pharmaceutical industry, where compounds are routinely screened for binding
affinity to only a single receptor of interest.
3. Approach - An off-target screening approach was
adopted, screening compounds of interest against the
structures of 44 receptors known to be associated with
toxic or adverse reactions. This work involved extensive
testing, comparison and cross-referencing of at least 3
independent docking methods to in vitro results. A
fundamental feature of this work was establishing “hit
thresholds” for both known toxins and FDA approved
compounds to provide a useful reference for compounds
of unknown efficacy or toxicity.
Stage 1: Structural modelling and validation
Stage 2: In silico screening of “blind test” compounds
Stage 3: Continued validation / refinement of the
approach
4. ToxCast compounds FDA Approved Drug Bank “experimental”
19 “test” compounds vs. Bowes et al “Panel 44 set”
Pathway Analysis
5. Targets to provide an early
assessment of the
potential hazard of a
compound or chemical
series, as recommended by
Bowes et al (2012), Nature
reviews: Drug Discovery,
11: 909-922:
G protein-coupled receptors
Adenosine receptor A2A
(ADORA2A)
α1A-adrenergic receptor
(ADRA1A)
α2A-adrenergic receptor
(ADRA2A)
β1-adrenergic receptor
(ADRB1)
β2-adrenergic receptor
(ADRB2)‡
Cannabinoid receptor CB1
(CNR1)
Cannabinoid receptor CB2
(CNR2)
Cholecystokinin A receptor
(CCKAR)
Dopamine receptor D1
(DRD1)‡
Dopamine receptor D2
(DRD2)‡
Endothelin receptor A
(EDNRA)
Histamine H1 receptor
(HRH1)‡
Histamine H2 receptor
(HRH2)
δ-type opioid receptor
(OPRD1)
κ-type opioid receptor
(OPRK1)‡
μ-type opioid receptor
(OPRM1)‡
Muscarinic acetylcholine
receptor M1 (CHRM1)
Muscarinic acetylcholine
receptor M2 (CHRM2)‡
Muscarinic acetylcholine
receptor M3 (CHRM3)
5-HT1A (HTR1A)
5-HT1B (HTR1B)
5-HT2A (HTR2A)‡
5-HT2B (HTR2B) High/
Vasopressin V1A receptor
(AVPR1A)
6. Ion channels
Acetylcholine receptor subunit α1 or α4
(CHRNA1 or CHRNA4)‡
Voltage-gated calcium channel subunit α
Cav1.2 (CACNA1C)‡
GABAA receptor α1(GABRA1)‡
Potassium voltage-gated channel subfamily H
member 2; hERG (KCNH2)
Potassium voltage gated channel KQT-like
member 1 (KCNQ1) and minimal potassium
channel MinK (KCNE1)
NMDA receptor subunit NR1 (GRIN1)‡
5-HT3 (HTR3A)‡
Voltage-gated sodium channel subunit α
Nuclear receptors
Androgen receptor (AR)
Glucocorticoid receptor (NR3C1)
Enzymes
Acetylcholinesterase (ACHE)
Cyclooxygenase 1;COX1 (PTGS1)
Cyclooxygenase 2; COX2 (PTGS2)‡
Monoamine oxidase A (MAOA)‡
Phosphodiesterase 3A (PDE3A)
Phosphodiesterase 4D (PDE4D)‡
Lymphocyte-specific protein tyrosine kinase
(LCK)
Transporters
Dopamine transporter (SLC6A3)
Noradrenaline transporter (SLC6A2)‡
Serotonin transporter (SLC6A4)‡
7. Normalisation techniques – A number of techniques were applied to normalise both
the in vitro control data and the results of a given docking in the context of other
dockings as follows;
Normalisation of the in vitro data – It was observed that the control data obtained
from the FDA Approved and ToxCast compounds naturally featured more hits than
misses. In vitro control data was normalised to ensure the resulting prediction was
not biased towards the prediction of a hit over a miss or vice versa.
Normalisation of the ligand results - Results were normalised due to the observation
of the tendency of the in vitro studies to identify a higher level of larger molecule
"hits" and smaller molecule "misses". The developed algorithm normalises the
predicted energy of interaction independent of the number of atoms.
Normalisation of the docking results – This was undertaken to account for scoring
functions that take into account how tightly a drug binds relative to other ligands that
are predicted to bind at the same site and; and for the purpose of comparison of the
respective binding affinities at different sites within the same protein.
8. Heavy atoms in the
receptor protein,
indicated in Silver
Dockings extracted
from training data sets,
indicated by blue
circles
Dockings of “low
interactivity” and
“interactivity” ligands
from, indicated in
yellow and finally;
Dockings of “high
interactivity” ligands,
indicated by red ‘x’s.
Figures below indicating high throughput docking locations of validation data sets and
specific docking orientations of “blind test” compounds.
9. In summary, the project developed a workflow
capable of reliable prediction, at >90% accuracy, of
protein-ligand hits when compared with recorded
in vitro interactions, detailed in DrugBank and
ToxCast. This is the first successful implementation
of an in silico panel for pharmacological profiling
and is a highly promising development.
10. Predicted interaction scores below are in the range [0, 1]. Results have been
normalised such that >0.5 is the criterion for a predicted hit with a true positive rate of
90% and a true negative rate of 30-40% (estimated). To aid analysis of these results a
colour coding system has been applied whereby blue denotes a score <0.5, and
therefore a miss. Hits are indicated by a score >0.5 shaded by an increase in intensity
of red towards a score of 1.0 (full confidence hit). This colour coding enables the user
to readily identify trends within the results.
11. Conclusions - The project benefited from an extensive validation exercise,
involving large databases of in vitro results. This enabled analysis of the
accuracy of prediction with reference to in vitro results, for which the
prediction of hits was pleasingly high, although there remains work to improve
upon the delineation of misses. In addition, comparisons with the in vitro data
for the blind test compounds were promising. The resulting technology system
provides extremely valuable molecular knowledge which provides the basis to
a novel screening tool as it enables a paradigm shift from reliance on observing
effects at a system level, including a reduced reliance upon animal testing, to
predicting effects based on understanding at the molecular level, whilst also
reducing drug development costs through the ability to screen for toxic or
adverse reactions earlier within the drug development cycle. Such molecular
knowledge is a valuable commodity to a range of industries including;
agrochemical, biotech, synthetic biology and medical/health research.
12. 1 – Moleculomics In Silico Discovery Inc (Canada),
500 Boulevard Cartier Ouest, Bureau 115, Laval,
Quebec, H7V 5B7, Canada
2 – Safety and Environmental Assurance Centre,
Unilever, Colworth Science Park, Sharnbrook,
Bedfordshire, MK44 1LQ, UK
3 – Moleculomics Ltd (UK), Institute of Life Sciences,
Swansea University Medical School, Singleton,
Swansea, SA2 8PP, UK
Contact email: will@moleculomics.com Contact
telephone: 514-701-2771
http://moleculomics.com