COMPOUNDS.

Field of the invention

The invention is in the field of genomics and it provides an in vitro method for predicting whether a compound is genotoxic in vivo.

Background of the invention

Cancer is one of the leading causes of death accounting for 13% of all deaths worldwide in 2004 according to the World Health Organization. In 2007 and 2008, cancer was ranked the second cause of death accounting for 23% and 26% of total deaths, in the US and Europe respectively (1 , 2). Cancer is a very complicated and yet not fully understood disease, nevertheless, two causal factors for its development is appreciated. The first is the presence of specific gene mutations genetically inherited or endogenously induced, e.g. BRCA1 and BRCA2 mutations are considered responsible for breast cancer (3). The second is exposure to exogenous carcinogenic factors, such as the link between tobacco smoke and lung cancer (4). The molecular mechanism of tumor formation after carcinogenic exposure frequently comprises the induction of DNA mutations by the carcinogen or its metabolites. If mutations occur within genes responsible for cell proliferation or survival, the cells may become malignant (5). Cellular transformation to a tumor cell may also be caused through a variety of mechanisms (production of reactive oxygen species,

immunosuppression, peroxisome proliferation etc.) which do not necessarily involve DNA damage. Consequently, carcinogens are classified as genotoxic (GTX) or non- genotoxic (NGTX) (5). Since almost all GTX compounds are carcinogenic, it is important, in particular for regulatory purposes, to evaluate the genotoxic potential of chemicals to which humans are exposed, and therefore to discriminate between GTX and NGTX compounds.

The most commonly used assay, the Salmonella typhimurium test, for evaluating mutagenic properties of chemicals in vitro was developed in 1975 by Bruce N. Ames (6). Subsequently, several in vitro assays were developed aiming at assessing genotoxic properties of chemicals in mammalian cellular models and are accepted by the regulatory authorities. However, the conventional in vitro test battery consisting of a bacterial mutation assay [Ames assay], mammalian micronuclei [MN], chromosomal aberration [CA] and mouse lymphoma assays [MLA]) often fails to correctly predict in vivo genotoxic and carcinogenic potential of compounds, even reaching a 50% false positive rate in some cases (7).

Depending on the intended use of the chemicals and the purpose of the assessment, regulatory authorities may require the in vivo evaluation of genotoxic and carcinogenic properties in rodents, especially for chemicals that are genotoxic in vitro (EC 1907/2006) and/or intended for human use (8). As a consequence of the high false positive rate of these in vitro assays, a high number of unnecessary animal experiments are performed each year. Next to its limited relevance for human health, the use of experimental animals inflicts considerable costs and raises ethical issues.

In cases where animal testing is not required after positive outcomes of in vitro assays (Globally Harmonized System of Classification and Labelling of

Chemicals (GHS), 3rd revised edition, UN, 2009), false positive in vitro results cause wrong chemical classifications.

Overall, a more reliable in vitro assay for predicting in vivo genotoxicity is urgently required.

Summary of the invention

The aim of this study was to develop an in vitro transcriptomics- based prediction method for in vivo genotoxicity.

The invention provides an in vitro method for predicting whether a compound is genotoxic in vivo. In particular, the invention provides a method for predicting the in vivo genotoxicity of a compound comprising the steps of performing an Ames test for the compound and determining if the result is positive or negative, followed by a step wherein the gene expression level of at least 3 genes is determined in at least one HepG2 cell, compared to a reference value and predicting that the compound is in vivo genotoxic if the expression level of at least two genes is above the predetermined reference value.

More in particular, we found that in vivo genotoxicity could be predicted by a method for predicting the in vivo genotoxicity of a compound comprising the steps of

a. performing an Ames test on the compound and determining if the

compound is Ames positive or Ames negative,

b. providing a HepG2 cell

c. exposing the HepG2 cell for a period of time between 12 and 48 hours to said compound,

d. if the compound is Ames positive, determining the level of expression of a first gene set comprising at least genes NR0B2, PWWP2B and LOC100131914,

e. if the compound is Ames negative, determining the level of expression of a second gene set, comprising at least genes SLC40A1 , PNMA6A and C10orf65

f. Comparing the level of expression of the first gene set or the second gene set to a predetermined reference value,

wherein the compound is predicted to be in vivo genotoxic if the expression level of at least 2 genes exposed to the compound are above their predetermined reference values.

This method appeared to be superior to the conventional methods as further detailed herein.

Detailed description of the invention

In this study we aimed at developing an alternative in vitro transcriptomics-based method for predicting in vivo genotoxic properties of chemicals.

This novel approach for the prediction of in vivo genotoxicity results in an improved accuracy when compared to each of the conventional in vitro genotoxicity assays or to the combination of Ames assay with the other conventional in vitro methods.

We surprisingly found that the accuracy and sensitivity of the classical Ames test could be greatly improved when the results were combined with a gene expression assay as described herein.

In particular, the invention relates to a method for predicting the in vivo genotoxicity of a compound comprising the steps of

a. performing an Ames test on the compound and determining if the

compound is Ames positive or Ames negative,

b. providing a HepG2 cell

c. exposing the HepG2 cell for a period of time between 12 and 48 hours to said compound,

d. if the compound is Ames positive, determining the level of expression of a first gene set comprising at least genes NR0B2, PWWP2B and LOC100131914,

e. if the compound is Ames negative, determining the level of expression of a second gene set, comprising at least genes SLC40A1 , PNMA6A and C10orf65 f. Comparing the level of expression of the first gene set or the second gene set to a predetermined reference value,

wherein the compound is predicted to be in vivo genotoxic if the expression level of at least 2 genes exposed to the compound are above their predetermined reference values.

The term "in vivo genotoxicity" is intended to mean the ability of a chemical to cause DNA damage in vivo, as determined by a positive result in at least one in vivo genotoxicity assay, including but not limited to the MN and CA assays as described in the OECD guidelines of testing of chemicals, Test No 474 and Test No 475, respectively .

The phrase "the expression level of at least 2 genes exposed to the compound" is intended to mean "the expression level of at least 2 genes within said first or second gene set".

The expression "at least 2 genes" in the context of the testing of 3 genes is intended to mean "2" or "3".

The term "Ames test" is intended to mean the bacterial reverse mutation assay as described by the OECD guideline of testing for chemicals: Test No. 471 .

The term "Ames positive" is intended to refer to a positive mutagenic result in the Ames test.

The term "Ames negative" is intended to refer to a non-mutagenic result in the Ames test

The term "HepG2 cell" is intended to mean the cell of human hepatocellular carcinoma origin with ATCC no. HB-8065, with a karyotype as described by Wong et. al (Wong N, Lai P, Pang E, Leung TW, Lau JW, Johnson PJ. A

comprehensive karyotypic study on human hepatocellular carcinoma by spectral karyotyping. Hepatology. 2000 Nov;32 (5): 1060-8).

The term "determining the level of expression" is intended to mean the quantitative measurement of mRNA molecules expressed by a certain gene present in HepG2 cells. Such mRNA levels may be determined by several methods known in the art such as microarray platforms, Reverse-Transcriptase Polymerase Chain Reaction (RT-PCR), and deep sequencing.

The term "reference compound" is intended to mean a compound for which results are available in the Ames test and an in vivo genotoxicity assay.

The term "Ames positive in vivo genotoxic reference compound" is intended to mean a compound with mutagenic results in the Ames test and the ability to cause DNA damage in vivo, as determined by a positive result in at least one in vivo genotoxicity assay, including but not limited to the MN and CA assays as described in the OECD guidelines of testing of chemicals, Test No 474 and Test No 475, respectively.

The term "Ames positive in vivo non-genotoxic reference compound" is intended to mean compound with mutagenic results in the Ames test and lack of the ability to cause DNA damage in vivo, as determined by a negative result in all the in vivo genotoxicity assays that the compound has been tested, including but not limited to the MN and CA assays, as described in the OECD guidelines of testing of chemicals, Test No 474 and Test No 475, respectively.

The term "Ames negative in vivo genotoxic reference compound" is intended to mean compound with non-mutagenic results in the Ames test and the ability to cause DNA damage in vivo, as determined by a positive result in at least one in vivo genotoxicity assay, including but not limited to the MN and CA assays as described in the OECD guidelines of testing of chemicals, Test No 474 and Test No 475, respectively.

The term "Ames negative in vivo non-genotoxic reference compound" is intended to mean compound with non-mutagenic results in the Ames test and lack of the ability to cause DNA damage in vivo, as determined by a negative result in all the in vivo genotoxicity assays that the compound has been tested, including but not limited to the MN and CA assays, as described in the OECD guidelines of testing of chemicals, Test No 474 and Test No 475, respectively.

The term "reference value" is intended to refer to the level of mRNA expression of a certain gene in HepG2 cells not exposed to a test compound. This reference value is used as a reference to which the expression level of the gene in HepG2 cell(s) after exposure to a test compound is compared.

The term "mean expression level" is intended to mean the average of the obtained expression levels for a single gene from all conducted biological and/or technical replicates.

The term "about 24 hours" is to be interpreted as meaning 24 hours plus or minus 2 hours, preferably plus or minus 1 hour, most preferably plus or minus half an hour.

When the method according to the invention was performed using a first gene set consisting of the genes NR0B2, PWWP2B, and LOC100131914 for the Ames positive compounds, an accurate prediction was obtained in about 80% of the cases.

When the method according to the invention was performed using a second gene set consisting of genes SLC40A1 , PNMA6A and C10orf65 for the Ames negative compounds, an accurate prediction was obtained in about 90% of the cases..

The results obtained with the method according to the invention could even be improved when additional genes were included in the analysis. When the first gene set for the Ames positive compounds as mentioned above was supplemented with at least one gene selected from the group consisting of genes CEACAM1 , SLC27A1 , TTR, UBE2E2, NAT8, GMFG, RBPMS, C10orf10, PROSC, TBC1 D9, OR10H1 , APOM, C1 orf128, AVEN, ZNRF3 and SNORD8, the results improved.

The invention therefore relates to a method as described above wherein the first gene set additionally comprises at least one gene selected from the group consisting of genes CEACAM1 , SLC27A1 , TTR, UBE2E2, NAT8, GMFG, RBPMS, C10orf10, PROSC, TBC1 D9, OR10H1 , APOM, C1 orf128, AVEN, ZNRF3 and SNORD8.

The results obtained with a method according to the invention could also be improved when additional genes were added to the second set. When the second gene set for the Ames negative compounds as mentioned above was supplemented with at least one gene selected from the group consisting of genes SGK1 , SLC64A, ANXA6, BTD, FGA, NDUFA10, NFATC3, MTMR15, ANAPC5, ZNF767, SCRN2 and GSTK1 , the results improved.

The invention therefore relates to a method as described above wherein the second gene set additionally comprises at least one gene selected from the group consisting of genes SGK1 , SLC64A, ANXA6, BTD, FGA, NDUFA10, NFATC3, MTMR15, ANAPC5, ZNF767, SCRN2 and GSTK1 .

A reference value for a gene may be empirically determined by methods known in the art. The reference values may be varied depending on the desire to either improve the sensitivity of the assay or the specificity. A skilled person in the art will know the metes and bounds of choosing a reference value.

In a preferred embodiment, a reference value for a particular gene is obtained by determining the expression level of that particular gene in the presence and absence of a genotoxic compound. The ratio between the expression level in the presence and the absence of the genotoxic compound is termed the GTX ratio.

Thereafter, the expression level of that particular gene in the presence and absence of a non-genotoxic compound is determined. The ratio between the expression level in the presence and the absence of the non-genotoxic compound is termed the non-GTX ratio. The average value of the log 2 of the GTX ratio and the non-GTX ratio is a suitable reference value. The reliability of the reference value may be increased by determining the GTX - and non-GTX ratios in the presence and absence of multiple genotoxic and non-genotoxic compounds.

Hence, the invention also relates to a method as described above wherein the predetermined reference value for a particular gene is calculated as the mean of the log 2 of the ratios of the expression level said gene in the presence and absence of at least one genotoxic compound and at least one non-genotoxic reference compound.

A preferred criterion for predicting a compound as in vivo genotoxic is as follows.

First, the expression level of each of these 3 genes NR0B2,

PWWP2B, and LOCI 00131914 as described above is determined in a HepG2 cell in the presence and absence of the compound. The ratio between the expression levels in the presence and absence of the compound is then determined. The log2 value of this ratio is then compared with the reference values shown in table 1 .

If the log 2 value of the ratio of the expression level of at least two of the three genes in cells exposed to the compound is above the reference value, then the compound is predicted to be in vivo genotoxic. If log 2 value of the ratio of the expression level of at least two of the three genes in cell (s) exposed to the compound are below the reference value, then the compound is predicted to be in vivo non- genotoxic.

Hence, the invention also relates to a method as described above wherein the predetermined reference value for the gene is taken from table 1 .

Table 1 Genes and their reference values.

EntrezGene ID Gene Symbol Gene Name / function Reference

value

8431 NR0B2 nuclear receptor subfamily 0, -0.099

group B, member 2

170394 PWWP2B PWWP domain containing 2B -0.071

100131914 LOC100131914 hypothetical protein -0.054

LOC100131914 (custom CDF

version 1 1 ), identical with EntrezGene ID Gene Symbol Gene Name / function Reference value

LOC100505880 (custom CDF

version 14)

634 CEACAM1 Receptor ligand 0,1795

1 183 CLCN4 Voltage-gated ion-channel -0,014

2009 EML1 Generic phosphatase -0,1825

7325 UBE2E2 Generic enzyme 0,006

8975 USP13 Generic protease 0,046

9535 GMFG Generic binding protein -0,0125

1 1212 PROSC Generic protein -0,0445

7276 TTR Generic binding protein -0,2465

9027 NAT8 Generic enzyme -0,267

1 1030 RBPMS Generic binding protein -0,0495

1 1067 C10orf10 Generic protein 0,0355

23158 TBC1 D9 Generic protein -0,163

29916 SNX1 1 Generic binding protein -0,0575

54538 ROB04 Generic receptor 0,104

54880 BCOR Generic binding protein -0,1415

6092 ROB02 Generic receptor 0,081

6725 SRMS Protein kinase -0,0775

26539 OR10H1 GPCR 0,0455

27010 TPK1 Generic kinase 0

641 15 C10orf54 Generic receptor 0,0405

319103 SNORD8 RNA -0,0105

414918 FAM1 16B Generic protein 0,0655

55937 APOM Transporter -0,163

56675 NRIP3 Generic binding protein 0,0465

57095 C1 orf128 / Generic protein 0,1 155

PITHD1

57099 AVEN Generic binding protein 0,148

60677 BRUNOL6 Generic binding protein 0,086

84133 ZNRF3 Generic binding protein -0,3185

146227 BEAN Generic binding protein 0,1 19 EntrezGene ID Gene Symbol Gene Name / function Reference

value

376497 SLC27A1 Generic enzyme -0,037

Similarly, when the second gene set consisting of the three genes SLC40A1 , PNMA6A and C10orf65 is used, a preferred criterion for predicting an Ames negative compound as in vivo genotoxic is as follows.

First, the expression level of each of these 3 genes in a HepG2 cell is determined in the presence and absence of the compound. The ratio between the expression levels in the presence and absence of the compound is then determined. The log2 value of this ratio is then compared with the reference values shown in table 2.

If the log 2 value of the ratio of the expression level of at least two of the three genes in cells exposed to the compound is above the reference value, then the compound is predicted to be in vivo genotoxic. If log 2 value of the ratio of the expression level of at least two of the three genes in cell (s) exposed to the compound are below the reference value, then the compound is predicted to be in vivo non- genotoxic.

Hence, the invention relates to a method as described above wherein the predetermined reference value for the gene is taken from table 2.

Table 2 Genes and their reference values.

Entrez Gene Symbol Gene name Reference Gene ID Value

solute carrier family 40 (iron-regulated

30061 SLC40A1 transporter), member 1 0.329

84968 PNMA6A paraneoplastic antigen like 6A 0.251

chromosome 10 open reading frame 65,

HOGA1 (4-hydroxy-2-oxoglutarate

1 12817 C10orf65 aldolase 1 ) 0.146

309 ANXA6 Generic binding protein 0,1655

337 APOA4 Receptor ligand 0

686 BTD Generic enzyme 0,037

1939 LGTN Generic receptor 0,0275 Entrez Gene Symbol Gene name Reference Gene ID Value

3267 AGFG1 Generic binding protein -0,086

4705 NDUFA10 Generic enzyme 0,038

4775 NFATC3 Transcription factor 0,159

9373 PLAA Generic binding protein -0,057

22909 MTMR15 Generic binding protein 0,0755

51433 ANAPC5 Generic enzyme 0,0265

64969 MRPS5 Generic binding protein 0,0845

79970 ZNF767 Generic protein 0,0985

373156 GSTK1 Generic enzyme 0,0355

2243 FGA Generic binding protein -0,0205

6446 SGK1 Protein kinase 0,1975

6532 SLC6A4 Transporter 0,0535

90507 SCRN2 Generic protease 0,0405

200014 CC2D1 B Generic protein 0,0165

648921 / LOC648921 / -0,048 288921 LOC283693

As an illustrative example only, the following simplified model is provided for the calculation of a reference value.

First the expression ratio of gene A is calculated. Therefore, the relative expression level of gene A is determined in the presence and absence of genotoxic compound Z. The expression level in the presence of compound Z is found to be 6 times higher than in its absence. It is then concluded that the GTX ratio of gene A is log2 of 6 = 2,58. The expression level of gene A in the presence of non-genotoxic compound Y is found to be 2 times higher than in its absence. It is then concluded that the non-GTX ratio of gene A is log2 of 2 = 1. A suitable reference value for gene A is than the average between the GTX ratio and the non-GTX ratio, in this example (2.58 + 1 ) / 2 = 1 .79.

Instead of a GTX ratio obtained with only one genotoxic compound, it may be advantageous to obtain several GTX ratios with different genotoxic compounds and calculate an average GTX ratio. The same may apply mutatis mutandis for non- GTX ratios.

When more than 3 genes are used in the method according to the invention, the reliability of the method may even be further improved when the criterion for genotoxicity is that (apart from the criterion that at least two out of three genes are above their reference value) more than half of the number of genes exposed to the compound are above their predetermined reference values.

Hence, the invention also relates to a method as described above wherein the compound is predicted to be in vivo genotoxic if the expression level of more than half of the number of genes exposed to the compound are above their predetermined reference values.

In a preferred embodiment, the step of comparing the level of expression of the first gene set or the second gene set to a predetermined reference value, is performed by a computer program.

A computer program particularly suited for this purpose is PAM (Prediction Analysis for Microarrays) or Support Vector Machines (SVM).

Representative examples of the accuracy, sensitivity and specificity of the method according to the invention are presented in Table 3.

The method according to the invention showed a clear improvement in comparison to methods of the prior art in regard to the accuracy and the specificity. A comparison of the results obtained by the method according to the invention and by conventional in vitro assays, is presented in Table 3.

When a method according to the invention was performed on a set of 62 compounds, the following results were obtained (Table 4): The raw data underlying table 4 are presented in tables 4A-4D.

Table 4: Class prediction results using the method of the invention Compound Prediction Compound Prediction

2AAF GTX+ ABP GTX

AFB1 GTX AZA GTX

APAP NGTX BZ GTX

BaP GTX Cb GTX

DES GTX cisPt GTX

DMBA GTX+ CP GTX

DMN GTX+ DEN GTX

MMC NGTX+ ENU GTX pCres GTX FU NGTX+

Ph GTX IQ GTX

TBTO GTX MOCA GTX

VitC GTX 2-CI GTX+

2CMP NGTX An is GTX

4AAF NGTX+ ASK NGTX

8HQ GTX+ BDCM NGTX ampC NGTX CAP NGTX+

An Ac NGTX CCI4 NGTX+

CsA NGTX Cou NGTX

Cur NGTX DDT NGTX

DEHP NGTX DZN NGTX

Diclo NGTX EthylB NGTX

Dman NGTX EuG NGTX+

E2 NGTX HCH NGTX

EtAc GTX NBZ NGTX+

NPD NGTX+ PCP NGTX

PhB NGTX Prog NGTX

Phen NGTX Sim NGTX

Que NGTX TCE NGTX

Res NGTX

RR GTX

Sulfi NGTX

TCDD NGTX TPA NGTX

WY NGTX

GTX: the compound is predicted genotoxic; NGTX: the compound is predicted non-genotoxic; Results indicated with bold and underlined letters indicate misclassification; Results labeled + indicate that two of the three replicates were classified in the indicated class.

Table 4A: Log2 treatment : control ratios obtained in triplicate experiments with Ames positive compounds.

NR0B2 PWWP2B LOC100505880

2AAF 0,042 -0,045 -0,103

2AAF -0,673 -0,14 -0,643

2AAF 0,472 0,042 0,579

ABP 0,806 0,442 0,65

ABP 0,21 1 0,047 0,088

ABP 0,217 0,264 -0,072

AFB1 0,605 0,098 0,281

AFB1 1 ,482 0,275 0,774

AFB1 0,548 0,088 0,534

AZA 1 ,473 0,536 1 ,541

AZA 0,232 0,044 0,022

AZA 0,893 -0,035 1 ,33

BaP 1 ,322 0,1 19 1 ,086

BaP 1 ,8 0,439 1 ,208

BaP 0,592 0,105 0,877

BZ 1 ,254 0,013 0,217

BZ 0,556 -0,137 0,523

BZ 0,916 0,255 -0,087

Cb 1 ,254 0,399 1 ,036

Cb 0,671 -0,133 0,803

Cb 0,519 0,145 0,483

cisPt 0,367 0,095 0,35

cisPt 1 ,545 -0,147 0,602

cisPt 0,467 -0,18 0,166

CP -0,404 0,042 -0,031

CP 0,276 -0,221 -0,01

CP 0,039 0,073 0,139

DEN 0,689 0,087 0,823

DEN 0,245 0,095 0,448

DEN -0,262 0,056 -0,022

DMBA 0,064 -0,155 0,08 NR0B2 PWWP2B LOC100505880

DMBA -0,1 16 0,088 -0,059

DMBA -0,076 -0,102 -0,025

DMN -0,173 -0,01 1 0,222

DMN -1 ,832 -0,368 -0,518

DMN -0,051 -0,304 0,321

ENU 0,424 0,01 0,088

ENU 0,901 0,06 0,382

ENU 1 ,056 0,11 -0,192

FU 0,781 0,256 0,583

FU -0,197 0,175 -0,067

Fu -0,457 0,008 -0,218

IQ 0,847 0,188 3,101

IQ 0,627 -0,003 2,784

IQ -0,396 -0,052 2,082

MMC 0,071 -0,106 -0,208

MMC -0,308 -0,232 -0,256

MMC 0,38 0,022 0,595

MOCA 0,498 0,047 0,088

MOCA 0,957 0,134 0,143

MOCA 0,484 0,259 -0,424

Paracres 1 ,286 0,271 -0,41

Paracres 1 ,877 0,072 0,437

Paracres 1 ,893 0,384 0,487

2-CI 0,881 0,564 -0,222

2-CI 0, 162 0, 197 -0,041

2-CI -0,623 0,058 -0,47

2CMP -1 ,551 -0,214 -1 ,088

2CMP -1 ,683 -0,23 -1 ,225

2CMP -1 ,227 -0,031 -0,867

4AAF -0,04 -0,524 -0,217

4AAF -0,278 -0,086 -0,295

4AAF -0,088 0,002 -0,101

8HQ -0,007 0,014 -0,34

8HQ -0,753 -0,165 -0,572

8HQ 0,249 -0,069 0,558

An is 0,886 0,013 1 ,084

An is 0,751 0,076 0,697

An is -0,076 0,253 0,288

NPDhigh -0,277 0,01 1 -0,1 19

NPDhigh -0,621 -0,153 -0,365

NPDhigh 0,1 -0,238 0,008

PhB 0,352 -0,169 -0,154 NR0B2 PWWP2B LOC100505880

PhB -0,176 -0,272 -0,38

PhB -0,407 -0,154 -0,303

Que -0,635 -0,206 0,062

Que -0,69 -0,437 -0,337

Que -3,709 -0,1 13 -0,727

reference value -0.099 -0,071

Table 4B: Determination of GTX or NGTX status according to a method of the invention wherein a compound is scored as GTX when at least two out of three genes are above the reference value. Plus sign indicates a value above the reference value, minus sign indicates a value below the reference value.

At

least

2/3 Average result genes over three

Compound Standard NR0B2 PWWP2B LOC100505880 +? measurements

2AAF GTX + + - GTX GTX

2AAF GTX - - - NGTX

2AAF GTX + + + GTX

ABP GTX + + + GTX GTX

ABP GTX + + + GTX

ABP GTX + + - GTX

AFB1 GTX + + + GTX GTX

AFB1 GTX + + + GTX

AFB1 GTX + + + GTX

AZA GTX + + + GTX GTX

AZA GTX + + + GTX

AZA GTX + + + GTX

BaP GTX + + + GTX GTX

BaP GTX + + + GTX

BaP GTX + + + GTX

BZ GTX + + + GTX GTX

BZ GTX + - + GTX

BZ GTX + + - GTX

Cb GTX + + + GTX GTX

Cb GTX + - + GTX

Cb GTX + + + GTX

cisPt GTX + + + GTX GTX cisPt GTX + - + GTX

cisPt GTX + - + GTX At

least

2/3 Average result genes over three

Compound Standard NR0B2 PWWP2B LOC100505880 +? measurements

CP GTX - + + GTX GTX

CP GTX + - + GTX

CP GTX + + + GTX

DEN GTX + + + GTX GTX

DEN GTX + + + GTX

DEN GTX - + + GTX

DMBA GTX + - + GTX GTX

DMBA GTX - + - NGTX

DMBA GTX + - + GTX

DMN GTX - + + GTX GTX

DMN GTX - - - NGTX

DMN GTX + - + GTX

ENU GTX + + + GTX GTX

ENU GTX + + + GTX

ENU GTX + + - GTX

FU GTX + + + GTX NGTX

FU GTX - + - NGTX

Fu GTX - + - NGTX

IQ GTX + + + GTX GTX

IQ GTX + + + GTX

IQ GTX - + + GTX

MMC GTX + - - NGTX NGTX

MMC GTX - - - NGTX

MMC GTX + + + GTX

MOCA GTX + + + GTX GTX

MOCA GTX + + + GTX

MOCA GTX + + - GTX

Paracres GTX + + - GTX GTX

Paracres GTX + + + GTX

Paracres GTX + + + GTX

2-CI NGTX + + - GTX GTX

2-CI NGTX + + + GTX

2-CI NGTX - + - NGTX

2CMP NGTX - - - NGTX NGTX

2CMP NGTX - - - NGTX

2CMP NGTX - + - NGTX

4AAF NGTX + - - NGTX NGTX

4AAF NGTX - - - NGTX

4AAF NGTX + + - GTX At

least

2/3 Average result genes over three

Compound Standard NR0B2 PWWP2B LOC100505880 +? measurements

8HQ NGTX + + - GTX GTX

8HQ NGTX - - - NGTX

8HQ NGTX + + + GTX

Anis NGTX + + + GTX GTX

Anis NGTX + + + GTX

Anis NGTX + + + GTX

NPDhigh NGTX - + - NGTX NGTX

NPDhigh NGTX - - - NGTX

NPDhigh NGTX + - + GTX

PhB NGTX + - - NGTX NGTX

PhB NGTX - - - NGTX

PhB NGTX - - - NGTX

Que NGTX - - + NGTX NGTX

Que NGTX - - - NGTX

Que NGTX - - - NGTX

Bold and underlined means that the result of the method of the invention differs from the standard designation. Table 4C: Log2 treatment : control ratios obtained in triplicate experiments with Ames negative compounds.

SLC40A1 PNMA6A C10orf65/HOGA1

APAP 0,057 -0,186 0,057

APAP 0,056 0,414 0,049

APAP -0,052 -0,062 -0,002

DES 0,723 0,135 0,206

DES 1 ,504 0,286 0,146

DES 0,717 0,203 0,516

Phenol 0,41 1 1 ,052 0,796

Phenol 0,65 0,262 0,1 13

Phenol 0,921 0,831 0,209

TBTO 0,604 0,909 0,426

TBTO 1 ,649 0,663 0,098

TBTO 0,208 0,456 0,858

VitC 0,972 1 ,027 0,333

VitC 0,225 0,378 0,348 SLC40A1 PNMA6A C10orf65/HOGA1

TCDD 0,169 -0,041 -0,107

TCDD -0,21 0,26 0,056

TCDD 0,104 0,072 0,151

TCE 0,195 -0,244 -0,36

TCE -0,121 -0,041 -0,274

TCE -0,304 0,062 -0,003

TPA -0,327 -0,493 0,108

TPA 1 ,338 -0,137 -0,423

TPA 0,199 -0,26 0,14

WY -0,312 0,059 -0,061

WY -0,393 -0,515 -0,158

WY -0,643 1 ,157 -0,053

Reference

Value 0,329 0,251 0,146

Table 4D: Determination of GTX or NGTX status according to a method of the invention wherein a compound is scored as GTX when at least two out of three are above the reference value.

Bold and underlined means that the result of the method of the invention differs from the standard designation.

An important increase of the specificity, and therewith a reduction of the false positive results, of up to 32% is achieved when the method according to the invention is compared to the outcome of the conventional in vitro assays.

The false positive rate of the conventional in vitro assays exceeds 50%, with the exception of Ames (23%) (7), whereas the false-positive rate of the method according to the invention is approximately 16%.

The false positive rate of our assay results from the misclassification of 5 NGTX compounds, namely RR, 2-CI, PhB, Anis and Sim. All of these compounds, with the exception of Sim, have delivered positive results in the conventional in vitro genotoxicity assays (see Table 5).

Due to its high accuracy, and especially due to its high specificity, the method according to the invention may be used in several applications in order to avoid unnecessary experiments on animals. For instance, it may facilitate the hazard identification of existing industrial chemicals to serve the purposes of the EU chemical policy program REACH, for which it has been estimated that some 400,000 rodents may be used for testing genotoxicity in vivo (14); specifically, chemical prioritization by grouping chemicals for further testing for genotoxicity in vivo may be supported.

The method according to the invention may also be applied for assessing genotoxic properties of novel cosmetics, since in the EU, for cosmetic ingredients, animal testing is generally prohibited since 2009 (EC Regulation

1223/2009). Furthermore, our approach may be effective in drug development, by significantly avoiding false positive results of the standard in vitro genotoxicity test battery, implying that promising lead compounds will no longer be eliminated due to wrong assumptions on their genotoxic properties and that rodents would not be unnecessarily sacrificed in costly experimentation.

Examples

Example 1 : Chemicals

Table 5 shows the doses for the 62 compounds used in this study and provides information on the stratification of the compounds based on the Ames assay, and on in vivo genotoxicity data.

Table 5: Chemicals used in this study, selected doses and information on in vitro and in vivo genotoxicity data. In In

Abbreviat CAS

Compound Dose Solvent Ames vitro vivo ion no

GTX GTX

2-acetyl

2AAF 53-96-3 50 μΜ DMSO + + + aminofluorene

1162-

Aflatoxin B1 AFB1 1μΜ DMSO + + +

65-8

Benzo[a]pyere BaP 50-32-8 2 μΜ DMSO + + +

7,12-Dimethyl

DMBA 57-97-6 5 μΜ DMSO + + + benzantracene

Dimethyl

DMN 62-75-9 2 mM DMSO + + + nitrosamine

200

Mitomycine C MMC 50-07-7 DMSO + + + nM

120-71-

Para-cresidine pCres 2 mM EtOH + + +

8

2-(chloromethyl) 6959- 300

2CMP DMSO + + - pyridine.HCI 47-3 μΜ

4-acetyl 28322- 100

4AAF DMSO + + - aminofluorene 02-3 nM

4-Nitro-o-

NPD 99-56-9 2 mM DMSO + + - phenylenediamine

148-24-

8-quinolinol 8HQ 15 μΜ DMSO + + - 3

117-39-

Quercetin Que 50 μΜ DMSO + + - 5

Phenobarbital PhB 50-06-6 1 mM DMSO + + -

103-90- 100

Acetaminophen APAP PBS - + +

2 μΜ

Diethylstilbestrol DES 56-53-1 5 μΜ EtOH - + +

108-95-

Phenol Ph 2 mM DMSO - + +

2

0.02

Tributylinoxide TBTO 56-35-9 EtOH - + + nM

458-37-

Curcumin Cur 1 μΜ DMSO - + - 7

118-92- o-anthranilic acid AnAc 2 mM DMSO - + - 3 In In

Abbreviat CAS

Compound Dose Solvent Ames vitro vivo ion no

GTX GTX

108-46-

Resorcinol RR 2 mM EtOH - + - 3

127-69-

Sulfisoxazole Sulfi 5 μΜ DMSO - + - 5

17beta-estradiol E2 50-28-2 30 μΜ DMSO - + -

140-88-

Ethylacrylate EtAc 1 mM EtOH - + - 5

Phenacetin Phen 62-44-2 1 mM EtOH - + -

L-ascorbic acid VitC 50-81-7 2 mM PBS - - +

7177- 250

Ampicillin trihydrate AmpC DMSO - - - 48-2 μΜ

15307- 100

Diclofenac Diclo PBS - - - 86-5 μΜ

250

D-mannitol Dman 69-65-8 PBS - - - μΜ

59865-

Cyclosporine A CsA 3 μΜ DMSO - - - 13-3

di(2-ethylhexyl) 117-81-

DEHP 10 mM DMSO - - - phthalate 7

12.5

Reserpine Res 50-55-5 DMSO - - - μΜ

2,3,7,8-tetrachloro 1746-

TCDD 10 ηΜ DMSO - - - dibenzo-p-dioxin 01-6

Tetradecanoyl 16561-

TPA 500ηΜ DMSO - - - phorbol acetate 29-8

50892- 200

Wy 14643 Wy DMSO - - - 23-4 μΜ

4-aminobi phenyl ABP 92-67-1 80 μΜ DMSO + + +

446-86- 250

Azathioprine AZA DMSO + + +

6 μΜ

Benzidine BZ 92-87-5 1 mM DMSO + + +

305-03-

Chlorambucil Cb 20 μΜ DMSO + + +

3 In In

Abbreviat CAS

Compound Dose Solvent Ames vitro vivo ion no

GTX GTX

15663-

Cisplatin cisPt 20 μΜ PBS + + +

27-1

6055-

Cyclophospham ide CP 2 mM PBS + + +

19-2

500

Diethylnitrosamine DEN 55-18-5 DMSO + + + μΜ

1-ethyl-1- 759-73-

ENU 1 mM DMSO + + + nitrosourea 9

110-00-

Furan Fu 2 mM DMSO + + +

9

2-amino-3-

76180- 800

m ethyim idazo[4, 5-f ] IQ DMSO + + +

96-6 μΜ

quinoline

4,4'-

101-14- methylenebis(2- MOCA 60 μΜ DMSO + + +

4

chloroaniline)

107-07-

2-chloroethanol 2-CI 2 mM DMSO + + - 3

104-94- p-anisidine Anis 60 μΜ DMSO + + - 9

Bromodichloro

BDCM 75-27-4 2 mM DMSO - + - methane

Carbon

CCI4 56-23-5 2 mM DMSO - + - tetrachloride

100-41- 800

Ethyl benzene EthylB DMSO - + - 4 μΜ

500

Eugenol EuG 97-53-0 DMSO - + - μΜ

Nitrobenzene NBZ 98-95-3 2 mM DMSO - - -

1 ,1 ,1-trichloro-2,2- di-(4-chlorophenyl) DDT 50-29-3 80 μΜ DMSO - - - ethane

Pentachlorophenol PCP 87-86-5 10 μΜ EtOH - - -

Progesterone Prog 57-83-0 6 μΜ EtOH - - - In In

Abbreviat CAS

Compound Dose Solvent Ames vitro vivo ion no

GTX GTX

127-18-

Tetrachloroethylene TCE 2 mM EtOH - - - 4

Lindane γ-HCH 58-89-9 2 mM DMSO - - -

55589-

Acesulfame-K ASK 2 mM DMSO - - - 62-3

105-60-

Caprolactam CAP 2 mM DMSO - - - 2

250

Coumaphos COU 56-72-4 DMSO - - - μΜ

333-41- 250

Diazinon DZN DMSO - - - 5 μΜ

122-34-

Simazine Sim 50 μΜ DMSO - - - 9

^* Ames results based on NTP data

† in vitro genotoxicity is considered positive when at least one in vitro genotoxicity assay (Ames, MN, CA, MLA) showed positive results,

$ in vivo genotoxicity is considered positive when at least one in vivo genotoxicity assays (MN, CA) showed positive results. Equivocal in vivo data are considered positive.

Example 2: Cell culture and treatment

HepG2 cells were cultured in 6-well plates as previously described (15). When the cells were 80% confluent, medium was replaced with fresh medium containing the corresponding dose of each compound or with the corresponding control treatment (DMSO, EtOH, or PBS 0,5%).

All doses were selected based on a MTT assay resulting to 80% viability at 72h incubation, or a maximum dose of 2mM was used when no cytotoxicity was observed, or the maximum soluble dose was used, whichever is the lowest (15). Cells were exposed for 24h. These exposure periods were selected based on the time that GTX need to be metabolized (15) and the cell cycle duration of HepG2 cells (approximately 20h) (16). Thereafter the culture medium was replaced by TRIZOL (Gibco/BRL) for RNA isolation. Three independent biological replicates were conducted.

Example 3: Total RNA isolation and microarray experiments

Total RNA was extracted using 0.5 ml TRIZOL according to the manufacturer's instructions and purified using RNeasy® Mini Kits (Qiagen). Sample preparation, hybridization, washing, staining and scanning of the Affymetrix Human Genome U133 Plus 2.0 GeneChip arrays were conducted according to the

manufacturer's protocol as previously described (17). Quality controls were within acceptable limits. Hybridization controls were called present on all arrays and yielded the expected increases in intensities.

Example 4: Annotation and normalization of microarray data

The obtained data sets were re-annotated to the MBNI Custom CDF- files versions 1 1 and 14 .

(http://brainarray.mbni.med.umich.edu/Brainarray/Database /CustomCDF/genomic_cur ated_CDF.asp) (18) and RMA normalized (19) using the NuGOExpressionFileCreator in GenePattern (20). Log2 ratios were calculated for each replicate to the

corresponding control treatment. Example 5: Selection of classifiers for genotoxicity.

The 34 chemicals were stratified into two groups based on the results of the Ames mutagenicity assay (Table 5) and consequently assigned to Ames-positive and Ames-negative. Within each group both in vivo GTX and in vivo NGTX chemicals are present. For the Ames-positive group, 13 t-tests were performed to select classifiers for discriminating in vivo GTX compounds from in vivo NGTX compounds. Genes significant in all t-tests were then selected. Within this geneset, sub-sets were investigated with regards to their predictive power. The best prediction was obtained for the geneset with three genes, namely NR0B2, PWWP2B, and LOC100131914.

For the Ames-negative group 21 t-tests were performed to select classifiers for discriminating in vivo GTX from in vivo NGTX chemicals. Genes significant in all t-tests were then selected. Within this geneset, sub-sets were investigated with regards to their predictive power. The best prediction was obtained for the geneset with three genes, namely SLC40A1 , PNMA6A and C10orf65. Example 6: Class prediction of the training and validation sets of reference compounds

Prediction analysis according to our method was conducted for each of the selected genesets. The gene expression data of the three replicates was compared to the respective reference values. A compound was predicted to be in vivo GTX or in vivo non-GTX when at least two out of the three replicates were assigned to one class.

The accuracy was calculated as the percentage of the correctly classified chemicals to the total number of tested chemicals; the sensitivity as the percentage of the correctly classified GTX to the total number of tested GTX

compounds and the specificity as the percentage of the correctly classified NGTX to the total number of tested NGTX compounds.

References

1 . Jemal A, Siegel R, Xu J, Ward E. Cancer statistics, 2010. CA Cancer J Clin.

2010 Sep-Oct;60(5):277-300.

2. OECD. Mortality from Cancer, in OECD, Health at a Glance: Europe 2010

OECD Publishing. 2010 34-5.

3. Petrucelli N, Daly MB, Feldman GL. Hereditary breast and ovarian cancer due to mutations in BRCA1 and BRCA2. Genet Med. 2010 May;12(5):245-

59.

4. Clapp RW, Jacobs MM , Loechler EL. Environmental and occupational causes of cancer: new evidence 2005-2007. Rev Environ Health. 2008 Jan- Mar;23(1 ):1 -37.

5. Oliveira PA, Colaco A, Chaves R, Guedes-Pinto H, De-La-Cruz PL, Lopes C.

Chemical carcinogenesis. An Acad Bras Cienc. 2007 Dec;79(4):593-616. 6. Ames BN, Lee FD, Durston WE. An improved bacterial test system for the detection and classification of mutagens and carcinogens. Proc Natl Acad Sci U S A. 1973 Mar;70(3):782-6.

7. Kirkland D, Aardema M, Henderson L, Mijller L. Evaluation of the ability of a battery of three in vitro genotoxicity tests to discriminate rodent carcinogens and non-carcinogens I. Sensitivity, specificity and relative predictivity. Mutat

Res. 2005 Jul 4;584(1 -2):1 -256.

8. I C H . G u ida nce on gen otoxicity testi ng an d data i nterpretation for pharmaceuticals intended for human use S2(R1 ). 2008.

9. IARC. IARC Monographs on the Evaluation of Carcinogenic Risks to

Humans. Supplement 7: Overall Evaluations of Carcinogenicity: An Updating of IARC Monographs Volumes 1 to 42. 1987. IARC. IARC Monographs on the evaluation of carcinogenic risks to humans. Volume 73: Some Chemicals that Cause Tumours of the Kidney or Urinary Bladder in Rodents and Some Other Substances. 1999.

Voogd CE. Azathioprine, a genotoxic agent to be considered non-genotoxic in man. Mutat Res. 1989 Sep;221 (2):133-52.

Bergman K, Muller L, Teigen SW. Series: current issues in mutagenesis and carcinogenesis, No. 65. The genotoxicity and carcinogenicity of paracetamol: a regulatory (re)view. Mutat Res. 1996 Feb 1 ;349(2):263-88.

Nagafuchi K, Miyazaki K. Modulation of genotoxicity of azathioprine by intracellular glutathione in hepatocytes. J Ca n ce r Res C l i n O n co l . 1991 ;1 17(4):321 -5.

van derJagt K, Munn S, T0rsl0v J , Bruijn Jd . Alternative Approaches can reduce the use of test animals under REACH: Addendum to the report "Assessment of additional testing needs under REACH. Effects of (Q)SARS, risk based testing and voluntary industry initiatives". EUROPEAN CO M M I SS IO N , DI RECTO RATE G E N ERAL J RC, JO I NT RESEARC H CENTRE, Institute for Health and Consumer Protection. 2004 November. Jennen DG, Magkoufopoulou C, Ketelslegers H B , van Herwijnen M H , Kleinjans JC, van Delft JH. Comparison of HepG2 and HepaRG by whole genome gene expression analysis for the purpose of chemical hazard identification. Toxicol Sci. 2010 Jan 27.

Knasmijller S, Parzefall W, Sanyal R, Ecker S, Schwab C, Uhl M, et al. Use of metabolically competent human hepatoma cells for the detection of mutagens and antimutagens. Mutat Res 1998 Jun18;402(1 -2):185-202. Jennen DG, Magkoufopoulou C, Ketelslegers HB, van Herwijnen MH, Kleinjans JC, van Delft JH. Comparison of HepG2 and HepaRG by whole genome gene expression analysis for the purpose of chemical hazard identification. Toxicol Sci. Jan 27.

Dai M, Wang P, Boyd AD, Kostov G, Athey B, Jones EG, et al. Evolving gene/transcript definitions significantly alter the interpretation of GeneChip data. Nucleic Acids Res. 2005;33(20):e175.

Irizarry RA, Hobbs B, Collin F, Beazer-Barclay YD, Antonellis KJ, Scherf U, et al. Exploration, normalization, and summa ri es of h i g h d e n s ity oligonucleotide array probe level data. Biostatistics. 2003 Apr;4(2):249-64. De Groot PJ, Reiff C, Mayer C, Muller M. NuGO contributions to GenePattern. Genes Nutr. 2008 Dec;3(3-4):143-6.