31.01.2013 Views

The Prediction of Chromosomal Effects and Their ... - Lhasa Limited

The Prediction of Chromosomal Effects and Their ... - Lhasa Limited

The Prediction of Chromosomal Effects and Their ... - Lhasa Limited

SHOW MORE
SHOW LESS

You also want an ePaper? Increase the reach of your titles

YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.

<strong>The</strong> <strong>Prediction</strong> <strong>of</strong> <strong>Chromosomal</strong> <strong>Effects</strong><br />

<strong>and</strong> <strong>The</strong>ir Mechanism <strong>of</strong> Induction in<br />

Derek for Windows<br />

RV Williams 1 & M Hayashi 2<br />

1 <strong>Lhasa</strong> <strong>Limited</strong>, Leeds, UK<br />

2 National Institute <strong>of</strong> Health Sciences, Tokyo, Japan


Overview<br />

• In silico prediction <strong>of</strong> chromosomal effects<br />

• Development <strong>of</strong> a Derek for Windows model<br />

• Common mechanisms<br />

• Evaluation <strong>of</strong> predictive performance<br />

• Conclusions


<strong>Chromosomal</strong> effects<br />

• In vitro tests for chromosomal effects are part <strong>of</strong><br />

the established genotoxicity test battery<br />

• Such tests are <strong>of</strong>ten low-throughput <strong>and</strong><br />

relatively expensive in comparison to the Ames<br />

test<br />

• Computer systems <strong>of</strong>fer an alternative approach<br />

for the identification <strong>of</strong> chemicals that may<br />

induce chromosomal effects


N<br />

H 2<br />

13.0<br />

Approaches to in silico toxicology<br />

OH<br />

26.5<br />

8.2<br />

36.1<br />

S<br />

O<br />

O<br />

58.4<br />

Statistical model<br />

O<br />

O<br />

14.7<br />

Algorithm<br />

QSAR<br />

Multicase (MC4PC)<br />

ADMEworks<br />

Structure<br />

Reactivity<br />

Activity<br />

Mechanism<br />

Expert system rules<br />

Metabolic data<br />

Human expert<br />

SAR<br />

Derek for Windows


Derek for Windows<br />

• Derek for Windows is a knowledge-based expert system<br />

that predicts the toxicity <strong>of</strong> a compound from its<br />

chemical structure<br />

• <strong>Prediction</strong>s are qualitative <strong>and</strong> based upon structural<br />

alerts, rules <strong>and</strong> examples<br />

• A broad range <strong>of</strong> toxicological endpoints are covered<br />

� Carcinogenicity<br />

� Genotoxicity<br />

� Hepatotoxicity<br />

� HERG channel inhibition<br />

� Reproductive toxicity<br />

� Skin sensitisation


Derek for Windows


Genotoxicity Coverage<br />

• Genotoxicity coverage in Derek for Windows has<br />

historically focussed on the prediction <strong>of</strong> Ames<br />

test mutagenicity<br />

• A project was therefore initiated to improve the<br />

prediction <strong>of</strong> structural <strong>and</strong> numerical<br />

chromosomal effects (“chromosome damage”)


Mechanism<br />

related to existing<br />

mutagenicity alert<br />

Extend existing<br />

mutagenicity alert<br />

Project strategy<br />

Chemical class causing<br />

chromosome damage<br />

Mechanism not<br />

related to existing<br />

mutagenicity alert<br />

Develop new<br />

chromosome<br />

damage alert


Developing new alerts<br />

In vitro chromosomal aberration data sets<br />

Analyse<br />

Common toxicophores<br />

Prioritise<br />

Ranked common toxicophores<br />

Gather mechanistic evidence<br />

<strong>and</strong> additional examples<br />

Mechanism-based structural alerts


<strong>Chromosomal</strong> aberration data sets<br />

• S<strong>of</strong>uni data book<br />

• CGX data set (Kirkl<strong>and</strong> et al)<br />

• Ishidate data set<br />

• Vitic database<br />

• Data set <strong>of</strong> marketed pharmaceuticals (Snyder<br />

<strong>and</strong> Green, Snyder et al)


Toxicophore identification<br />

• Toxicophores were identified primarily by<br />

analysis <strong>of</strong> the S<strong>of</strong>uni data book<br />

• Two techniques were used during this process<br />

� Visual analysis<br />

� Computer analysis using ChemTK lite


Example case for toxicophore<br />

identification<br />

5-Substituted uracils


Toxicophore identification<br />

• Visual analysis highlighted that the data sets<br />

contained four 5-substituted uracils<br />

• All four compounds gave positive results in the<br />

in vitro chromosomal aberration test<br />

• <strong>The</strong>se were developed into a toxicophore,<br />

comprised <strong>of</strong> the common structural features<br />

allowing for some generalisation


Toxicophore identification<br />

F<br />

O N<br />

HN<br />

O<br />

O<br />

HN<br />

O<br />

O<br />

O<br />

N<br />

O<br />

N<br />

F<br />

1,3-Bis(2-tetrahydr<strong>of</strong>uranyl)<br />

-5-fluorouracil<br />

1-(2-Tetrahydr<strong>of</strong>uranyl)<br />

-5-fluorouracil<br />

HO<br />

O<br />

O<br />

HN<br />

O<br />

O<br />

O<br />

N<br />

N<br />

H<br />

F<br />

F<br />

OH<br />

5-Fluorodeoxyuridine<br />

5-Fluorouracil


Known active<br />

O<br />

HN<br />

O<br />

N<br />

H<br />

F<br />

Toxicophore identification<br />

O<br />

N<br />

O<br />

N<br />

R<br />

Where R =<br />

any heteroatom<br />

Unknown activity<br />

O<br />

HN<br />

O<br />

N<br />

H<br />

Cl


Toxicophore identification - results<br />

• 103 toxicophores have been identified using this<br />

approach


Toxicophore development<br />

• <strong>The</strong> next step was to prioritise the toxicophores,<br />

using a weight <strong>of</strong> evidence approach<br />

• Prioritised toxicophores were then further<br />

developed using information in the published<br />

literature including<br />

� Mechanistic evidence<br />

� Additional example compounds


Example case for toxicophore<br />

development<br />

5-Substituted uracils


Initial<br />

toxicophore:<br />

O<br />

N<br />

O<br />

N<br />

Toxicophore development<br />

R<br />

Where R =<br />

any heteroatom<br />

Additional<br />

examples included:<br />

O<br />

N<br />

O<br />

HN<br />

N<br />

OH<br />

O<br />

O<br />

F<br />

OH<br />

Capecitabine, a clastogenic<br />

5-fluorocytidine<br />

Underlying<br />

mechanism:<br />

5-Fluorouracils <strong>and</strong> 5fluorocytidines<br />

inhibit<br />

thymidylate synthetase


O<br />

Alert features<br />

Mechanism-based<br />

structural alert for<br />

5-fluoropyrimidines<br />

N<br />

NH 2<br />

N<br />

F<br />

Based on the mechanism <strong>of</strong> action, the alert is restricted to 5-fluoropyrimidines<br />

O<br />

N<br />

O<br />

N<br />

F


O<br />

N<br />

O<br />

HN<br />

N<br />

OH<br />

O<br />

O<br />

F<br />

OH<br />

O<br />

Alert features<br />

Mechanism-based<br />

structural alert for<br />

5-fluoropyrimidines<br />

N<br />

NH 2<br />

N<br />

F<br />

<strong>The</strong> alert is also restricted to 5-fluoropyrimidines that can form or mimic nucleotides<br />

O<br />

N<br />

O<br />

N<br />

F<br />

O<br />

O<br />

N<br />

O<br />

O<br />

N<br />

F


<strong>Prediction</strong> for 5-fluorouracil


Results <strong>of</strong> toxicophore development<br />

• To date, 50 <strong>of</strong> the 103 toxicophores have been<br />

investigated<br />

• Currently the Derek for Windows knowledge<br />

base contains 63 alerts for chromosome damage<br />

• From this work, several common mechanisms<br />

for the induction <strong>of</strong> chromosomal effects that do<br />

not involve direct damage to DNA have become<br />

apparent


Why is mechanism important?<br />

• Contributes to the definition <strong>of</strong> alert scope<br />

• Aids determination <strong>of</strong> in vivo significance<br />

• Provides transparency to predictions<br />

• Is included as one <strong>of</strong> the OECD Principles for<br />

(Q)SAR Validation designed to facilitate<br />

regulatory acceptance <strong>of</strong> in silico predictions


Common mechanisms leading to<br />

chromosomal effects


<strong>Chromosomal</strong> effect mechanisms<br />

Disruption or inhibition <strong>of</strong> DNA synthesis or repair<br />

Spindle function disruption<br />

Generation <strong>of</strong> reactive oxygen species<br />

Energy depletion<br />

Thiol reactivity<br />

Intercalation


<strong>Chromosomal</strong> effect mechanisms<br />

Disruption or inhibition <strong>of</strong> DNA synthesis or repair<br />

Spindle function disruption<br />

Generation <strong>of</strong> reactive oxygen species<br />

Energy depletion<br />

Thiol reactivity<br />

Intercalation<br />

Xanthines:<br />

Inhibition <strong>of</strong> cell cycle checkpoint function<br />

O<br />

N<br />

O<br />

N<br />

N<br />

N<br />

(Caffeine)<br />

4-Hydroxystilbenes:<br />

Inhibition <strong>of</strong> ribonucleotide reductase<br />

HO<br />

OH<br />

OH<br />

(Resveratrol)


<strong>Chromosomal</strong> effect mechanisms<br />

Disruption or inhibition <strong>of</strong> DNA synthesis or repair<br />

Spindle function disruption<br />

Generation <strong>of</strong> reactive oxygen species<br />

Energy depletion<br />

Thiol reactivity<br />

Intercalation<br />

N<br />

H 2<br />

Thymine <strong>and</strong> derivatives:<br />

Imbalance <strong>of</strong> the nucleotide pool<br />

HN<br />

O<br />

N<br />

HO<br />

HO<br />

N<br />

N<br />

O<br />

O<br />

N<br />

H<br />

NH<br />

O<br />

O<br />

O<br />

P<br />

O<br />

O<br />

Na +<br />

Na +<br />

(Thymine)<br />

Guanine <strong>and</strong> derivatives:<br />

Imbalance <strong>of</strong> the nucleotide pool<br />

(Sodium 5’-guanylate)


<strong>Chromosomal</strong> effect mechanisms<br />

Disruption or inhibition <strong>of</strong> DNA synthesis or repair<br />

Spindle function disruption<br />

Generation <strong>of</strong> reactive oxygen species<br />

Energy depletion<br />

Thiol reactivity<br />

Intercalation<br />

HO<br />

Vinca alkaloids:<br />

Numerical chromosomal aberrations<br />

N<br />

H<br />

O<br />

O<br />

O<br />

N<br />

N<br />

OH<br />

H<br />

N<br />

OH<br />

H<br />

O O<br />

O<br />

(Vinblastine)<br />

Di- or tri-phenylethylenes:<br />

Numerical chromosomal aberrations<br />

OH<br />

O<br />

(Diethylstilbestrol)


<strong>Chromosomal</strong> effect mechanisms<br />

Disruption or inhibition <strong>of</strong> DNA synthesis or repair<br />

Spindle function disruption<br />

Generation <strong>of</strong> reactive oxygen species<br />

Energy depletion<br />

Thiol reactivity<br />

Intercalation<br />

N +<br />

N +<br />

e-<br />

O 2<br />

Bipyridinium compounds:<br />

N +<br />

-<br />

O 2<br />

N +<br />

Cl Cl<br />

(Paraquat)<br />

N N +


<strong>Chromosomal</strong> effect mechanisms<br />

Disruption or inhibition <strong>of</strong> DNA synthesis or repair<br />

Spindle function disruption<br />

Generation <strong>of</strong> reactive oxygen species<br />

Energy depletion<br />

Thiol reactivity<br />

Intercalation<br />

Polynitrophenols:<br />

Uncoupling <strong>of</strong> oxidative phosphorylation<br />

O 2 N<br />

OH<br />

NO 2<br />

(2,4-Dinitrophenol)


<strong>Chromosomal</strong> effect mechanisms<br />

Disruption or inhibition <strong>of</strong> DNA synthesis or repair<br />

Spindle function disruption<br />

Generation <strong>of</strong> reactive oxygen species<br />

Energy depletion<br />

Thiol reactivity<br />

Intercalation<br />

Isothiazolinones:<br />

Covalent interaction with proteins<br />

O<br />

NH<br />

S<br />

(Benzisothiazolinone)<br />

N-Polyhaloalkylthio compounds:<br />

Covalent interaction with proteins<br />

O<br />

O<br />

N S<br />

Cl<br />

Cl<br />

Cl<br />

(Captan)


<strong>Chromosomal</strong> effect mechanisms<br />

Disruption or inhibition <strong>of</strong> DNA synthesis or repair<br />

Spindle function disruption<br />

Generation <strong>of</strong> reactive oxygen species<br />

Energy depletion<br />

Thiol reactivity<br />

Intercalation<br />

Psoralens:<br />

Disruption <strong>of</strong> DNA helix<br />

O<br />

O O O<br />

Hydroxyanthraquinones:<br />

Disruption <strong>of</strong> DNA helix<br />

OH O OH<br />

O<br />

(5-Methoxypsoralen)<br />

(Danthron)


Evaluation <strong>of</strong> predictive performance


How computer systems are evaluated<br />

• Computer systems <strong>of</strong>fer an alternative approach<br />

for the prediction <strong>of</strong> chromosomal effects<br />

• <strong>The</strong> performance <strong>of</strong> computer systems can be<br />

evaluated using data sets <strong>of</strong> chemicals with<br />

known activities<br />

• Key measurements include sensitivity <strong>and</strong><br />

specificity


Evaluation <strong>of</strong> three computer systems<br />

• Three computer systems have been developed<br />

<strong>and</strong> evaluated by the NIHS<br />

� Derek for Windows, ADMEworks <strong>and</strong> Multicase<br />

• <strong>Chromosomal</strong> aberration data from the<br />

Japanese database <strong>of</strong> existing chemicals was<br />

used for the evaluation (GLP tests)<br />

Positive 98<br />

Negative 121<br />

Total 219


Evaluation results<br />

• <strong>The</strong> best results were obtained when the<br />

computer systems were combined<br />

• In this approach a majority verdict was used to<br />

give an overall result<br />

Derek for Windows<br />

Multicase<br />

Admeworks<br />

Overall result<br />

+<br />

+<br />

+<br />

-<br />

+<br />

+<br />

Positive<br />

- -<br />

- +<br />

- -<br />

Negative


Combined systems - detailed results<br />

Chrom ab. test<br />

Chrom ab. test<br />

+<br />

-<br />

In silico<br />

+ -<br />

63 26<br />

18 81<br />

Applicability: 85.8% (188/219)


Combined systems - detailed results<br />

Chrom ab. test<br />

Chrom ab. test<br />

+<br />

-<br />

In silico<br />

+ -<br />

63 26<br />

18 81<br />

<strong>The</strong> activity <strong>of</strong> 63/89 positive<br />

compounds was correctly predicted


Combined systems - detailed results<br />

Chrom ab. test<br />

Chrom ab. test<br />

+<br />

-<br />

In silico<br />

+ -<br />

63 26<br />

18 81<br />

<strong>The</strong> inactivity <strong>of</strong> 81/99 negative<br />

compounds was correctly predicted


Combined systems - detailed results<br />

Chrom ab. test<br />

Chrom ab. test<br />

+<br />

-<br />

In silico<br />

+ -<br />

63 26<br />

18 81<br />

Sensitivity<br />

70.8 %<br />

Specificity<br />

81.8 %<br />

Concordance<br />

76.6 %<br />

Applicability: 85.8% (188/219)


Conclusion<br />

• This work has demonstrated that it is feasible to define<br />

alerts for the prediction <strong>of</strong> chromosomal effects<br />

• Derek for Windows now contains 63 alerts for the<br />

chromosome damage endpoint <strong>and</strong> work is on-going<br />

• <strong>The</strong> approach allows mechanistic insight <strong>and</strong><br />

demonstrates promising predictive performance<br />

• When computer systems are used in combination,<br />

predictive performance can be improved


• <strong>Lhasa</strong> <strong>Limited</strong><br />

� Kate Langton<br />

� Carol Marchant<br />

� Russ Naven<br />

Acknowledgements<br />

• National Institute <strong>of</strong> Health Sciences<br />

� Akihiko Hirose<br />

� Eiichi Kamata

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!