WO2010031799A1

WO2010031799A1 - Method for identifying irritating and allergenic substances

Info

Publication number: WO2010031799A1
Application number: PCT/EP2009/062030
Authority: WO
Inventors: Sandra Szameit; Klemens Vierlinger; Letizia Farmer; Helga Tuschl; Elisabeth Weber; Christa NÖHAMMER
Original assignee: Ait Austrian Institute Of Technology Gmbh
Priority date: 2008-09-16
Filing date: 2009-09-16
Publication date: 2010-03-25
Also published as: EP2331712A1; AT507402A1

Abstract

The present invention relates to a method for identifying irritating or allergenic potential of a product comprising the steps of: a) contacting the product with immature dendritic cells, b) determining the amount of 65 gene products, expressed in said cells upon contact with the product, c) subtracting the amount of gene products of these genes expressed in said cells without contact with the product from the respective amounts determined under b) to obtain a set of product-caused amounts of gene products for each of said genes, d) comparing the product-caused amounts of gene products for each of said genes obtained under c) with a test set of product-caused amounts of gene products for each of said genes, wherein said test set of product-caused amounts of gene products for each of said genes was provided from at least two allergenic products and at least two irritating products and, optionally, from at least two products which are neither allergenic nor irritating; and wherein the comparing is performed by a pattern recognition method which recognises an allergenic pattern and an irritating pattern and, optionally, a non-allergenic and non-irritating pattern; and e) identifying an allergenic potential of a product when the comparison under d) results in the recognition of an allergenic pattern and identifying an irritating potential of a product when the comparison under d) results in the recognition of an irritating pattern and, optionally, identifying a non-allergenic and non-irritating product when the comparison under d) results in the recognition of a non-allergenic and non-irritating pattern.

Description

Method for identifying irritating and allergenic substances

The present invention relates to a method for identifying irritating and/or allergenic substances.

For the past few years, the development of in vitro assays as alternatives to animal tests for toxicological issues has become more and more important. One aspect is the identification of low-molecular weight chemicals able to induce allergic contact dermatitis.

Antigen-presenting cells of the skin play a major role in the induction of contact sensitization and are therefore potential candidates for the establishment of in vitro systems. Langerhans Cells were proposed to be used in such systems (Tuschl, H., and Kovac, R., 2001. Toxicol In Vitro 15, 327-331), but since these cells are difficult to isolate in sufficient numbers, Dendritic Cells (DCs) derived from peripheral blood monocytes (PBMC-DCs) or from CD34⁺-stem cells (CD34-DCs) may also be used. Early works mainly concentrated on the analysis of DC- maturation markers or of expression changes of immune-relevant genes, e.g. cytokines and chemokines.

Dietz et al . (2000. Biochem. Biophys. Res. Commun. 275, 731-738) analyzed changes in transcription during the maturation process of PBMC-derived immature DCs (iDCs) employing a cDNA mi- croarray containing probes for 4110 known genes. 291 genes were found to be up-regulated, while 78 genes were down-regulated. A similar study was performed by Chen and colleagues (2002 Biochem Biophys Res Commun 290, 66-72) with mouse DCs treated with Lipopolysaccharide (LPS) to induce maturation. A microarray containing probes for 514 genes revealed the differential expression of 72 genes. In 2001, Le Naour compared gene expression in iDCs and DCs matured by TNF (Le Naour et al . , 2001. J Biol Chem 276, 17920-17931) . Several groups used Affymetrix arrays to compare different inflammatory stimuli and LPS for their ability to induce DC maturation (Lindstedt et al . , 2002 Int. Immunol. 14, 1203-1213) . Moschella and colleagues analyzed changes in expression of 408 immune genes during DC maturation induced by TNF, CD40L, IFN-Y or IL-7 (Moschella et al . , 2001 Br J Haematol 114, 444-457) .

In a previous work an immune-specific DNA microarray was established and it could be shown that exposure of iDCs to the contact sensitizers NiSO₄ and Bandrowski's Base (BB) induced upregulation of several genes not induced after treatment of iDCs with the irritant SDS (Szameit et al . , 2008 Clin Chem 54:525-33) .

It is an object of the present invention to provide a method to identify irritating and/or allergenic substances and to discriminate between irritating/allergenic and non- irritating/non-allergenic substances .

The present invention relates to a method for identifying irritating or allergenic potential of a product comprising the steps of: a) contacting the product with immature dendritic cells, b) determining the amount of gene products of the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorf162, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2, SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941, expressed in said cells upon contact with the product, c) subtracting the amount of gene products of the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorfl62, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2, SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941 expressed in said cells without contact with the product from the respective amounts determined under b) to obtain a set of product-caused amounts of gene products for each of said genes, d) comparing the product-caused amounts of gene products for each of said genes obtained under c) with a test set of product-caused amounts of gene products for each of said genes, wherein said test set of product-caused amounts of gene products for each of said genes was provided from at least two allergenic products and at least two irritating products and, optionally, from at least two products which are neither allergenic nor irritating; and wherein the comparing is performed by a pattern recognition method which recognises an allergenic pattern and an irritating pattern and, optionally, a non-allergenic and non- irritating pattern; and e) identifying an allergenic potential of a product when the comparison under d) results in the recognition of an allergenic pattern and identifying an irritating potential of a product when the comparison under d) results in the recognition of an irritating pattern and, optionally, identifying a non- allergenic and non-irritating product when the comparison under d) results in the recognition of a non-allergenic and non- irritating pattern.

It surprisingly turned out that the set of 65 genes (ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorf162, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EP- STIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, P0LE4, PPFIBPl, RBPJ, RELB, RSAD2 , SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66) is sufficient to fully correlate the product to an allergenic or irritating potential (or identify the product as being free of allergenic or irritating potential) . In the first step of the present method, a product which is unknown with respect to its irritating/allergenic potential is contacted with immature dendritic cells. Upon contact with the product, the reaction of the dendritic cells to the product is observed in step b) with respect to the expression of the group of 65 genes mentioned above. These expression profiles are then normalised with respect to each of these genes for non- irritating and non-allergenic substances in step c) . This can be performed in a variety of ways. Preferably, if a microarray is used for determination of expression levels, and RNA is the gene product to be determined, the RNA may be fluorescently labelled (e.g. by two fluorescence colors) and subtracted from a normal (i.e. non-allergenic and non-irritating) value (e.g. a value for the solvent of the product only) . Also absolute levels of ex- pression may be determined and subtracted to measured (or known) normal values. The gene products (expression products) to be determined are preferably RNA, but can also be proteins.

In a preferred embodiment of the present invention the amount of CENTD3, MARCKS, AK026517, LGP2 and/or BU678941 can be determined. The determination of these genes allows an even more reliable identification whether a substance has an irritating or allergenic potential.

In a preferred embodiment, for example, the raw data are generated by measuring the amount of fluorescently labelled nucleic acid sequences generated on the basis of the genes expressed in the immature dendritic cells ("foreground intensity values") . This signal is then subtracted by the corresponding "background value" (e.g. the values originating from an unspe- cific signal (e.g. the unspecific signal on a microarray (being caused e.g. by unspecific binding or autofluorescence; if two fluorescent dyes are applied, subtraction from the "background value" may preferably be performed for both fluorescent dyes independently) ) . The foreground intensities of the product are then subtracted from measured or known figures for the expression patterns without contact with the product from the respective amounts determined under step b) to obtain a set of product-caused amounts of gene products for each of said genes. Preferably, these values are experimentally derived, e.g. coming from an experiment where the immature dendritic cells are contacted with the solvent of the product only (i.e. without the product) ; this embodiment has the additional advantage that effects of solvents are also visible and can be subtracted and excluded) .

The data can then be normalised, e.g. by using first loess normalisation and then scale, but all other normalising methods can also be applied. In the method according to the present invention, the sequence of normalisation/subtraction is not critical. For example, in the case of fluorescence analysis, usually subtraction of controls (e.g. the dendritic cells treated with solvent only (especially with Iog2 (first color) - Iog2 (second color)) may preferably be performed after normalisation.

In step d) the expression patterns with respect to the group of 65 genes are compared to each other by pattern recognition methods. These methods are well available to the skilled man in the art. A preferred method for pattern recognition is employing feature selection and involving construction of a classifier. In the examples, a nearest shrunken centroid method was employed, as implemented in the pamr software available at www.bioconductor.org. This identified a range of classifiers from 70 genes down to the above mentioned 65 genes each of which achieving a 90.6% prediction accuracy. Such methods are well available in the art, for example in WO 2008/037806 A, Tibshi- rani et al . (PNAS 99 (10) (2002), 6567-6572 and Stat. Science 18 (1) (2003), 104-117) and Hastie et al . (,,The elements of Statistical Learning", Springer, 2001, especially pages 9-39, 79-111, 182-190, 225-255 and 347-433) . The method according to the present invention may include other pattern recognition techniques such as "k-nearest neighbour (KNN) ", "support vector machines (SVM)", "linear discriminant analysis (LDA)", "Artificial Neural Networks (ANN)" and others. The rule for classification of an unknown sample will usually depend on a distance measure of the unknown sample to a representation of either of the two classes, i.e. the centroid, the nearest neighbour or any other representation of the classes. Alternatively, any regression or model based approaches which calculates a decision boundary by least squares optimisation or maximum likelihood estimation will be applicable in the method according to the present invention.

With the method according to the present invention, the identification of an allergenic or irritating substance in a product (and, optionally, the lack of allergenic or irritating properties) can be identified with 100 % accuracy. This is also shown in the example section.

The test set may contain more than the at least two expression data of each group of compounds. Preferably, the test set comprises the product-caused amounts of gene products of (independently) at least 4, preferably at least 6, more preferred at least 10, especially at least 20 substances of each class (allergen, irritant, non-allergenic and non-irritating) . The larger the test set data, the more robust and less sophisticated pattern matching may be applied.

The group of 65 genes are necessary (and sufficient) for allowing a nearly 100 % accuracy of the method, however, further markers for allergens/irritants may be included in the method according to the present invention. Preferably, these markers may be chosen from the following Table A (which also includes the group of 65 genes as mentioned above, including the GeneBank accession nos. for these genes).

Table A: GenBank Accession No. in parenthesis

Immature dendritic cells undergo maturation upon contact with allergenic substances. This effect is utilized in the method of the present invention to identify irritating and allergenic substances in products tested with this system. If immature dendritic cells are contacted with a allergenic substance the cell begins i.a. to up-regulate the transcription of various genes. Although other substances which do not show any allergenic properties, but irritating properties may induce also a non-specific general stress response in dendritic cells accompanied by an up- and down-regulation of various genes these other substances do not induce the specific gene expression pattern of the genes of Table A. Therefore, the genes identified in Table A allow to unambiguously identify and discriminate between irri- tating/allergenic substances from non-irritating/non-allergenic substances .

In a preferred embodiment of the present invention the amount of gene products of at least one, of at least two, of at least three, of at least four, of at least ten, of at least 15, normalization gene(s) is also determined. The expression rate of a normalization gene in a dendritic cell is upon contact with an irritating/allergenic substance approximately identical (i.e. the amount determined varies only for about ± 10%) to the amount expressed in dendritic cells contacted with non- irritating/allergenic substances. The normalization gene(s) may be selected from the group disclosed in Table B.

Table B: Normalizing genes

As used herein, the term "gene product" refers either to RNA, in particular mRNA, or to a peptide, polypeptide or protein, resulting from expression (transcription or transcription/translation) of a gene.

Immature dendritic cells to be used in the method of the present invention can be produced by methods known in the art (see e.g. US 2004/109851, De Smedt et al . , 2002 Arch Dermatol Res 294, 109-116) .

The immature dendritic cells used in the method of the present invention are preferably immature PBMC-DCs.

According to preferred embodiment of the present invention the immature dendritic cells are derived from peripheral blood monocytes (PBMCs) or from CD34⁺-stem cells.

Methods to produce and isolate this kind of cells are known to the person skilled in the art (De Smedt et al . , 2002 Arch Dermatol Res 294, 109-116) .

According to a further preferred embodiment of the present invention the amount of at least 65 gene products in step b) is determined.

In order to increase the accuracy of the method of the present invention it is possible to determine in step b) the amount of more than the group of 65 gene products mentioned.

As described above, the gene product to be quantified can be a nucleic acid (i.e. RNA) or a proteinaceous (e.g. polypeptide) molecule. The method to be employed to determine the quantity of these molecules depends on the type.

According to a preferred embodiment of the present invention the amount of the at least 65 gene products is determined by reverse transcribing RNA, in particular mRNA, to produce cDNA and subjecting said cDNA to a real-time polymerase chain reaction (PCR) and/or to a hybridisation assay.

If the amount of RNA is determined, a reverse transcription step is needed to produce cDNA, which can be quantified by methods like real-time PCR or hybridisation assays. Of course it is also possible to combine the step of the reverse transcription with a real-time PCR.

The hybridisation assay is preferably a microarray assay or a microsphere assay.

Both techniques involve the use of target specific probes, which are immobilised on a solid support. In the case of mi- croarrays the solid support is substantially planar, whereas in the case of microspheres the solid support is substantially spherically. Solid supports to be used in such methods are known in the art. Specifically preferred surfaces to be used in the microchip according to the present invention are coated at least partially with a phenolic resin polymer with a functionality of from 6 to 15, preferably from 7 to 10, most preferably 8. Such surfaces are e.g. described in WO 03/027675 A, Preininger et al . (Anal. Biochem. 330 (2004), 29-34).

A microarray (also commonly known as gene chip, DNA chip, or biochip) is a collection of microscopic DNA spots attached to a solid surface, such as glass, plastic or silicon chip forming an array for the purpose of expression profiling, monitoring levels for a large number of amplified nucleic acids simultaneously. Microarrays can be fabricated using a variety of technologies, including printing with fine-pointed pins onto glass slides, photolithography using pre-made masks, photolithography using dynamic micromirror devices, ink-jet printing, or electrochemistry on microelectrode arrays. A microarray comprises a large number of immobilized oligonucleotide molecules provided in high density on the solid support. A microarray is a highly efficient tool in order to detect dozens, hundreds or even thousands of different amplification products according to the present invention in one single detection step. Such microarrays are often provided as slides or plates in particular microtiter plates. In the state of the art a microarray is both defined either as a miniaturized arrangement of binding sites (i.e. a material, the support) or as a support comprising miniaturized binding sites (i.e. the array).

According to a preferred embodiment of the present invention the amount of at least 65 gene products is determined by using antibodies binding to said gene products.

Another aspect of the present invention relates to a microarray having immobilised on its surface probes for the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorf162, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2, SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941 and wherein said microarray comprises at least 10%, preferably at least 20%, more preferably at least 30% of the total number of probes are probes for the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorfl62, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2, SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941.

It is further preferred to additionally have probes for normalising genes on the microarray, for example at least five, preferably of at least ten, more preferably of at least 15 normalizing genes selected from Table B.

According to a particular preferred embodiment of the present invention the microarray contains probes of 60%, preferably 80%, of the genes of Table A, whereby ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorfl62, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2, FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2 , SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66, are always present on the microarray.

A further aspect of the present invention relates to a kit containing probes for the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorfl62, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, IT- GAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2 , SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66. Preferably, the kit may additionally comprise at least one probe for the genes CENTD3, MARCKS, AK026517, LGP2, BU678941. It is also preferred, if the kit further comprises probes least five, preferably of at least ten, more preferably of at least 15 normalizing genes selected from Table B.

Preferred kits therefore contain probes wherein at least 20%, preferably at least 40%, more preferably at least 60% of all probes are selected from the group consisting of the genes selected from the group consisting of ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorf162, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2, FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2 , SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941 and at least five, preferably of at least ten, more preferably of at least 15 normalizing genes selected from Table B. The probes to be used according to the present invention are nucleic acid molecules derived from said genes having the same or complementary nucleic acid sequence.

Preferred probes are the following:

GeneName SystematicName Sequence

EPSTIl NM_033255 AGAAGAGAAGCATTTAGAGAGCATCAGCAATACAAAACCGCTGAGTTCTTGAGCAAACTG

ILlRN NM_173842 IAIICCIGCAIIIGIGAAAIGAIGGIGAAAGIAAGIGGIAGCIIIICCCIICIIIIICII

GPlBA NM_000173 IIICIAGIIIICIIAICAGGAIGIGAGCACICGIIGIGICIGGAIGIIACAAAIAIGGGI

PHLDB3 NM_198850 GAGGAAGAIIGCAGGGGCGGGAAGAAACIAIIIAIIGGIGCICCIGIGAIACCGAAIIAA

ARHGAP22 NM_021226 AICAGAIGCIGCICCAACCAIGCAGIICCIGGIGAGGGICAGAAGGGGACGGIACCAAGA

POLE4 NM_019896 CIGIGGAIGAAIIIGCIIIICIGGAAGGIACIIIAGAIIGAIIGCCGAGCGGGGCAGIII

AKO26517 AKO26517 AGCAAIAGAICICGGIAGIIACGIAIIGGGCAGAIACIIACIGIAIGAAIGAAAGAACAI

INFRSFl2A NM_016639 ICACICAGAIGICCIGAAAIICCACCACGGGGGICACCCIGGGGGGIIAGGGACCIAIII

SYIIl NM_152280 GCICIICIGCIAIGAAGIAGIAAAAGGCAGICIAIAAIIAACIGACAGACCIAACIGAAG

SlOOAlO NM_002966 AGAAAAGIIAAAIACCAGAIAAGCIIIIGAIIIIIGIAIIGIIIGCAICCCCIIGCCCIC

EMP2 NM_001424 IICCGGAGCIGGGIIGCIICIGCIGCAGIACAGAAICCACAIICAGAIAACCAIIIIGIA

CENID3 NM_022481 AIAIIIGAIACGIAGGGGIICCAIGAGAGAIIIIGGGIIIIAAAGGAAIGGIIIIACIGC

BIN2A2 NM_181531 GIIAIIGAGGACIIIAAAGAGCIIIIGIIIAIIIGGGIIAAIAIIIAIGACAIIIGACAI

RELB NM_006509 CAGACIIGAAGGIGGGGGGIAGGIIGGIIGIICAGAGICIICCCAAIAAAGAIGAGIIII

H2AFY2 NM_018649 AAAICCIIICAAAAIICIIAAAICIICIGIICCICCIIIIICCAAGGGAAGAGGGCAAAA

S100A4 NM_002961 AIAAGCAGCCCAGGAAGAAAIGAAAACICCICIGAIGIGGIIGGGGGGICIGCCAGCIGG

SLC9A9 NM_173653

LY75 NM 002349 IGIGGIIAICACIIIAAGIIIIGACACCIAGAIIAIAGICIIAGIAAIAGCAICCACIGG DHRS 9 NM_005771 ACTCATTTAGATCGTGCTTATTTGGATTGCAAAAGGGAGTCCCACCATCGCTGGTGGTAT

FABP4 NM_001442 CTCACCTTGAAGAATAATCCTAGAAAACTCACAAAATGTGTGATGCTTTTGTAGGTACCT

SPOCDl AKO 97227 TGGAGGTGGCTGATAAGAGTAGAGTTTCAAAATCTCTTTAAACCTTCCTAAAGCAATGAT

MS4A3 NM_006138 TAATATCCAGTCATTAAGGAGTTGTCACTCTTCATCAGAGTCACCGGACCTATGCAATTA

BAIl NM_001702 CGGCCGGCCTGGCACCGTTTTTTAAACACCCCCATCCCTCGGGAAGCAGCCAGCTCCCCA

LOC541471 AK001796 GCAAGACGTATGCAGTTTTCATTGACATCTTTTGGAGAAACTGACAAACTGGACTTGACT

FABP4 NM_001442 AAGGGACGTTGACCTGGACTGAAGTTCGCATTGAACTCTACAACATTCTGTGGGGATATA

ABCG2 NM_004827 CTGGGGCTTGTGGAAGAATCACGTGGCCTTGGCTTGTATGATTGTTATTTTCCTCACAAT

LGP2 NM_024119 CATACTGTACTCAGAATCACGACATTCCTTCCCTACCAAGGCCACTTCTATTTTTTGAGG

AOAH NM_001637 TTTACAAACTTCAATCTTTTCTACATGGATTTTGCCTTCCATGAAATCATACAGGAGTGG

PPFIBPl NM_003622 GGAGIAAIGIGCCGAIICIGAAGIIGCCACAAAAAAIAAGACACIGGIGAAIGAGAGIAI

ILlRN NM_173843 AAGAIIIIAIIGIAAAACAGAGCIGAAGICACAGGAAGIAGGGAACIIIGCACCCAACAI

MARCKS NM_002356 GGIAAGGICAIGAACCACIAIIIIIGAICCAIIAIICCAAIIAAGAAIGCGIGICAAAAC

FBN2 NM_001999 GIICCCIIGAAAGGGAACACCIGGCAIICIGIGGIGIIICGIGCIGICIIAAAIAAIGGI

LOC124220 NM_145252 CCACCAGIIAAICICACAIACICAGCAAACICACCCGIGGGICGCIAGGGIGGGGIAIGG

SPOCDl NM_144569

PDLIM4 NM_003687 IGCICCCACGCCIGCIICIIAAGGICCCIGCICGGCCGGIGIAAAIAIGIIICACCCIGI

IIGAD NM_005353 GCIAAGCACCIICICGGAGAGAIAGAGAIIGIAAIGIIIIIACAIAICIGICCAICIIII

CYBRDl NM_024843

MBOAI2 NM_138799 IIGIICCIAAAIGGIAIIIICAAGIGIAAIAIIGIGAGAACGCIACIGCAGIAGIIGAIG

POLE4 NM_019896 ICAGAGGAGAGACIIGGAIAAIGCAAIAGAAGCIGIGGAIGAAIIIGCIIIICIGGAAGG

SYIIl NM_152280 GCIGACIIIGGCIIICACAIIIGIICIIICCAGAGCIAACIGAIAAGAGIGGAGGAGGAA

PLA2G4C NM_003706 ICCAGAIGGCCAGAAIGAAIGIGAIAGIICAGACCAAIGCCIICCACIGCICCIIIAIGA

BU678941 BU678941 AGCGACCAICCAGICAIIIAIIICCCICCAIICCCAAIGAIGIACACACGACIAAGAAGG

EHF NM_012153 lAICIGIIIACIGICICAICIGAACIGAICCCAGGIGAACGGIIIAIIGCCIAGAIIIGI

SLAMFl NM_003037 ICIGIGCCICAGIIICICICICAGGAIAAAGAGIGAAIAGAGGCCGAAGGGIGAAIIICI

CD38 NM_001775 IGAAAAAICCIGAGGAIICAICIIGCACAICIGAGAICIGAGCCAGICGCIGIGGIIGII

IL4I1 NM_172374 CCAGIIAICICICCAAAACACGACCCACACGAGGACCICGCAIIAAAGIAIIIICGGAAA

BIN2A2 NM_181531 CCIGGICGAGCAGGGCAGIACIGGACCAGGICIACGICAGCAIICAGGIICAAIGGGGAC

SLAMFl NM_003037 AGGCGCAGAACAGAGCGIIACIIGAIAACAGCGIICCAICIIIGIGIIGIAGCAGAIGAA

Clorf162 NM_174896 ICAGIGCACAGAGIAIICACCCAGCAICAIGAAICAACIIGGGAGGAGICAACCAAAIGA

CDYL2 NM_152342 IIAGAICGGGIGICAGAGGACGACCAAICIIAGGGAAIIICCAGACCAAIGAGCAAAAIG

DNAJCl2 NM_021800 AGIGGGAAGCIIIGAAIGACICAGIGAAGACGICAAIGCACIGGGIIGICAGAGGIAAAA NM_201262

G0S2 NM_015714 CCAACACIGIGIGAAIIAICIAAAIGCGICIACCAIIIIGCACIAGGGAGGAAGGAIAAA

IFI13 NM_001031683 AIICGAAIAAAGCICIIGAGAAGGGACIGAAICCICIGAAIGCAIACICCGAICICGCIG NM^O 01549

CCDC26 NM_145050 IIIAGGGCIAAGAICIIGIIGIGIICAGCCACAAAIGAIGGCCAGCAGIGAGGAAICIGI

CLCFl NM_013246 IGGCAAIICIACACAAAAAGAGAIGAGAIIAACAGIGCAGGGIIGGGGICIGCAIIGGAG

FAS NM_000043 AIGICIAICCACAGGCIAACCCCACICIAIGAAICAAIAGAAGAAGCIAIGACCIIIIGC NM_152871 NM 152872 NM 152873

NM 152874

NM 152875

NM 152876

NM 152877

GEM NM 005261 GGACCTTGCTGGAAACAAAGGCTTAGCAAACAATTTTTGTTCAATGCCCACCAAGACATA

NM 181702

KCNE3 NM 005472 TCATATACATTAAGTTGAGCCATATGTAATCACTGTGTTTGTAGGTTAGAAACAGCTGAG

AOAH NM 001637 TTTACAAACTTCAATCTTTTCTACATGGATTTTGCCTTCCATGAAATCATACAGGAGTGG

ARID5B NNMM 003322119999 AAGACTGGCAAACAGATGGCAAGGGATGCCCCTCTTTTTCATAAAACTCTCCAAGGTTCA

CD274 NM 014143 GAGTTTTTCCTATTTATTTTGAGTCTGTGAGGTCTTCTTGTCATGTGAGTGTGGTTGTGA

CTPS NM 001905 TTTCTTCCAGGATGGTGTTACTGCAGTTGAAGGGCAATATGAAGTTACTTTCTTAATGTG

FZDl NM 003505 AAATGCCGGTACTTAGGACCTAAATTTATCTATGTCTGTCATACGCTAAAATGATATTGG

HSD3B1 NM 000862 AAGTCCAAGACTCAGTGATCGAAGGATGACAGAGATGTGCAGTGGGTATTGTTAGGAGAT

KLHL24 NM 017644 CTTGTTCTGATAAGGTTCAATCTTATGATCCAGAAACCAATTCTTGGCTACTTCGTGCAG

LPCAT2 NM 017839 TGTAACTCTTGTTTCTAGGTAATCGTTCTCTCTCAACAAACTTCTCAAGCGTCTGTGTAA

NEDD4L NM 001144964 TGACTTACCTCCATATGAAACCTTTGAAGATTTACGAGAGAAACTTCTCATGGCCGTGGA

NM 001144965

NM 001144966

NM 001144967

NM 001144968

NM 001144969

NM 001144970

NM 001144971

NM 015277

RSAD2 NM 080657 ACTCTGAGTCAGTTGAAATAGGGTACCATCTAGGTCAGTTTAAGAAGAGTCAGCTCAGAG

TMC5 NM 001105248 TGGGCATTCCATGCTATTTTTAATACCTGGATTGCTGATTTTTCAAGACAAAATACTTGG

NM 001105249

NM 024780

WDR66 NM 144668 CCAGGATGGTCAGTGAAGTTACCAGGAATGTTTAAAGCACAAAGGACTTTGGGTGTGTGT

MET NM 000245 GTGTCTGGACAGATTGTGGGAGTAAGTGATTCTTCTAAGAATTAGATACTTGTCACTGCC

NM 001127500)

PKIB NM 032471 ATGAAGGCTCATAATCTATCAAGAGTGCTGAATTTCTGCATGTTGAAAGACTTAGTGGTT

NM 181794

NM 181795

TCP11L2 NM 152772 AAATATACTTCGAAAGCTGCTCTTCAATGAGGAAGCCATGGGGAAGGTAGATGCTTCACC

TPMl NM _000366 TTTTCAGTCCTTCATGTTAAAGATTTAGACACCACATACAACTGGTAAAGGACGTTTTCT

NM^~001018004 NM_001018005 NM_001018006 NM_001018007 NM_001018008 NM 001018020

N4BP3 NM 015111 GCCCCAAGGTCCCGAAGGGCAGGTCAGAGGGAGAGAGGCTGGAGACCTGGGCTGGGGCCT

RBPJ NM 005349 CAGTGGGGAGCCTTTTTTATTCATCTCTTGGATGATGATGAATCAGAAGGAGAAGAATTC

NM ^"015874

NM ^"203283

NM ^"203284

TM4SF1 NM 014220 CATGAAAACTGTGGCAAACGATGTGCGATGCTTTCTTCTGTATTGGCTGCTCTCATTGGA

TRIM15 NM 033229 AGCCAGCGGGCTTTCGGGGAAAAAAAAGAAAAAGACATCTAAAATAAAATGTTTAAACTG

The kit of the present invention contains probes which can be used to detect the presence of those gene products of immature dendritic cells which are expressed when said cells are contacted with irritating and/or allergenic substances. The present invention is further illustrated by the following figures and examples, however, without being restricted thereto .

Fig. 1 shows dose-response experiments after 24h-exposure of iDCs to increasing concentrations of chemicals. CD86 expression compared to the solvent control was determined after the method of Overton (1988) (solid line). PI staining was performed in order to determine cytotoxicity and viability is expressed as 100% - cytotoxicity [%] (dotted line) . Means and SD of at least 3 independent experiments are shown. Dashed vertical lines mark the concentrations chosen for all further exposures. Analysis of iDCs exposed to NiSO₄, BB and SDS have already been previously presented (Szameit et al . , 2008).

Fig. 2: The line in fig. 2 shows the misclassification error from an 8-fold cross-validation as a function of the value of threshold (the number of genes used for the pattern recognition) . A range of classifiers from 4500 genes down to 6 genes was identified, each of which achieving a 100% prediction accuracy. Three random sets of genes were taken from the dataset: a 6-gene set, a 48-gene set and a 4500-gene set. The discriminative power of these genes was then assessed again in an 8-fold cross-validation approach. This procedure (random selection of gene sets and 8-fold cross-validation) was repeated 100 times and the respective cross-validation error rated are shown in the boxplots. The high error rates show that randomly selected gene sets are largely useless in the discrimination between allergen and irritants.

Fig. 3 shows the fold changes (M-values) of 6 genes necessary for the classifier (PHLDB3, FABP4, ABCG2, FBN2, SPOCDl and SLAMFl) as measured in the single microarray experiments (chemical treated vs. solvent-treated DCs). Fold changes derived from experiments employing an allergenic chemical are shown in red, expression values derived from experiments employing an irritating chemical are shown in green. Differences in gene expression between experiments employing allergenic chemicals and experiments employing irritating chemicals are clearly visible. Fig. 3 shows

EXAMPLE :

Materials and Methods:

Culture of iDCs and chemical exposure iDCs were generated as previously described (De Smedt et al., 2002. Arch. Dermatol. Res. 294, 109-116). Briefly, 400 U/ml IL4 and 1000 U/ml GMCSF (Strathman Biotech) were added to the incubation medium. Medium was refreshed every second day.

On day six, iDCs were treated with different concentrations of the following chemicals and the corresponding solvent controls for 24 hours:

Sensitizers (Allergens): NiSO₄, Bandrowski's Base (BB), Di- nitrobenzenesulfonic acid (DNBS) , Cinnamaldehyde (CA) , Dinitro- chlorobenzene (DNCB) , Eugenol (Eug) , TMTD (Tetramethylthiuram disulfide) .

Non-Allergens: Sodium dodecyl sulphate (SDS), TritonX-100 (Tri), Propylene glycol (PropG) , Ethanol (EtOH), Tryptamin- hydrochlorid (TryH) , Phenol (Phe) , Glycerol (GIy), Dimethylsul- foxide (DMSO) , para-aminobenzoic acid (pABA) , Methyl salicylate (MS) .

CA, DNCB, Eug, TMTD, MS and BB were dissolved in 0.1% DMSO, PropG, EtOH, TryH, Phe, GIy, DMSO, pABA, NiSO₄, DNBS, SDS and Tri were dissolved in water. Controls were performed with the solvents only (i.e. without the allergen/non-allergen) to evaluate solvent effects.

For NiSO₄, BB, DNBS, SDS and Tri, DCs derived from 4 human donors were used for each chemical. For CA, DNCB, Eug, TMTD, PropG, EtOH, TryH, Phe, GIy, DMSO, pABA and MS, DCs derived from 1 human donor were used for each chemical.

Flow cytometric analysis

Flow cytometric analysis was performed as previously described (Szameit et al . , 2008). The following antibodies were used: FITC anti-human CDIa, PE anti-human CD14, PE-anti-human CD86 (BD Pharmingen) . Measurements were performed on a Coulter Epics XL-MCL (Beckman Coulter Inc.) with EXPO32™ ADC XL 3 Color v1.1C - Expo32 vl.2 Analysis vl .1C (Beckman Coulter Inc.). DCs were defined by light scatter, dead cells were gated out and fluorescence histograms were evaluated after the method of Over- ton (Overton, 1988. Cytometry 9, 619-626).

Propidium iodide (PI) staining was performed to determine cytotoxicity (Sigma-Aldrich) . Cells were harvested and 500 μl of the culture were incubated with 10 μl PI (0.2 mg/ml in PBS) for 5 min at 4°C. The percentage of Pi-positive cells was measured with flow cytometry and viability was expressed as "100 % - cy- totoxicity [%]".

RNA extraction

DCs were harvested by centrifugation (10 min, 300 g, 4°C) and cells were resuspended in Trizol (Invitrogen) . Total RNA was isolated using Trizol as described in the manufacturer's protocol. RNA pellets were resuspended in 100 μl RNase-free water, and 350 μl buffer RLT (Qiagen) and 250 μl EtOH were added. Samples were applied to RNeasy micro columns (Qiagen) and RNA cleanup was performed according to the manufacturer's protocol. On-column DNase treatment was performed employing the supplied RNase-free DNase. RNA was eluted with 2 x 14 μl of RNase-free water .

RNA amplification, labelling and hybridization

Whole genome array

RNA samples derived from DCs exposed to CA, DNCB, Eug, TMTD, PropG, EtOH, TryH, Phe, GIy, DMSO, pABA, MS, NiSO₄, BB, DNBS, SDS, Tri were analyzed with 44k human whole genome oligo mi- croarrays (Agilent Technologies) .

RNA expression levels from chemical-treated DCs and from DCs exposed to solvent control were analyzed on one microarray using the Two-Color Low RNA Input Linear Amplification Kit PLUS (Agilent Technologies) according to the manufacturer's protocol. For each amplification, 200 ng of total RNA were employed and amplified samples were prepared for hybridization using the Gene Expression Hybridization Kit (Agilent Technologies) . Hybridization was performed over night at 65°C in a rotating hybridization oven (Agilent Technologies) .

Data Analysis

Hybridizations performed on whole genome arrays were scanned using an Agilent DNA microarray scanner and raw image data were extracted using the Agilent Feature Extraction Software. The software package Limma for the R computing environment was used for data analysis of both systems (Smyth and Speed, 2003. Methods 31, 265-273) . Limma is part of the Bioconductor project at www.bioconductor.org (Gentleman et al . , 2004. Genome Biol. 5, R80). For every hybridization, background intensities were subtracted from foreground intensities for each spot. For whole genome experiments, global loess normalization was performed.

For probe selection, classification and cross-validation of the whole genome data a nearest shrunken centroid method (imple- merited in the pamr software available at www . bioconductor.org) was chosen (Tibshirani et al . , 2002 Proc. Nathl . Acad. Sci. USA 99, 6567-6572). It was selected for several reasons: it has been extensively used in literature, it allows multiclass classification and it runs feature selection, classification and cross- validation in one go.

Briefly, it calculates several different possible classifiers using different shrinkage thresholds (i.e. different number of genes) and finds the best threshold from leave-one-out- crossvalidation (loocv) . Here we picked the classifier with the smallest number of genes (largest threshold), if more than one threshold yielded the same crossvalidation results. Probe selection, classification and cross-validation was performed for 2 subsets of data. The first subset includes data derived from experiments with NiSO₄, BB, DNBS, SDS, Tri (subset 1) . The second subset includes data derived from experiments with CA, DNCB, Eug, TMTD, PropG, EtOH, TryH, Phe, GIy, DMSO, pABA, MS (subset 2) .

The genes from both resulting classifiers were combined and accuracy of the classifier was again determined using loocv.

RNA amplification , labelling and hybridization

Immune toxicity chip

Template-switching PCR (TS-PCR) amplification of 50 ng of total RNA was performed as previously described (Petalidis et al., 2003 Nucleic Acids Res. 31, el42; Lauss et al . , 2007 Vir- chows Arch. 451, 1019-1029) using Power Script RT Kit and Advantage 2 PCR Kit (Clontech) . All resulting first strand cDNA (10 μl) was used as template for the second strand amplification reaction. PCR was performed with the following settings: 95°C for 1 min, 17 cycles of 95°C for 10 s, 65°C for 10 s, 68°C for 6 min. The cycle number for the PCR was adjusted in order to obtain cDNA derived from the exponential phase of the PCR. PCR products were purified employing the QIAquick PCR purification kit with Buffer PB (Qiagen) following the manufacturer's protocol and labelled with Klenow Polymerase exo-minus Epicentre IOOOU (Biozym) . 3 μg of PCR-product were mixed with 3 μl (N) i₀ primer (OD=125) to a final reaction volume of 94.5 μl . The reaction was incubated at 98°C for 5 min and chilled on ice for 5 min. Next, the following reagents were added: 12 μl 1Ox Klenow buffer, 10 μl of a dNTP mix containing 0.23 inM dA/dT/dGTP and 0.07 mM dCTP, 2 μl Cy3- or Cy5 dCTP (GE Healthcare), and 1.5 μl Klenow Polymerase. The reaction was incubated at 37°C for 2 hours and terminated at 95°C for 3 min. After labelling, cDNAs derived from chemical-treated DCs and solvent-treated DCs were combined and a microcon cleanup was performed employing Microcon YM30 columns (Millipore Corporation) .

After blocking the chip surface as previously described (Szameit et al . , 2008), cDNAs were co-hybridized at 42°C over night on one microarray using 20 μl DigEasy hybridisation buffer (Roche Diagnostics) .

Setup of the immune toxicity chip

66 immune-relevant genes, 7 housekeeping genes, 8 negative controls and external and internal normalization controls were spotted onto ARChip Epoxy glass slides (Austrian Research Centers GmbH - ARC, Seibersdorf, Austria) using an Omnigrid Arrayer (GeneMachines) . Probe design and external normalization controls (Lucidea Universal Score Card System, GE Healthcare) have been described in detail previously (Szameit et al . , 2008). In addition to external normalization controls, probes for internal normalization were included in the chip design, enabling normalization without the need to add extra RNA. Therefore, the risk of inaccurate normalization due to pipetting errors is reduced. Similar to the system described by Yang et al . (Yang et al., 2002 Nucleic Acids Res. 30, el5) , a series of 10 non- differentially expressed control spots was created from a pool of human RNA (Universal Human Reference RNA, Stratagene) . The RNA was reverse transcribed and amplified using the TS-PCR amplification as described earlier in this publication. The resulting cDNA was spotted on the immune toxicity chip in serial dilutions (1:3) starting with 96 ng/μl . Since the expression of most genes is not affected by exposure of DCs to chemicals, signal intensities of the fluorescent dyes Cy3 and Cy5 should be equal on these probes in a two-colour microarray experiment.

Data Analysis

Hybridizations performed on the immune toxicity chip were scanned using a GenePix 4000A scanner (Molecular Devices) with adjusted settings (equal for both wavelengths) avoiding saturated signals. Raw image data were extracted using GenePix 3.0 software (Molecular Devices) .

Data derived from the immune toxicity chip were normalized employing the internal normalization controls using loess normalization. For whole genome experiments, global loess normalization was performed.

To identify the most promising candidate genes for the identification of sensitizers, experiments performed with RNA derived from allergen-exposed DCs were combined, a linear model was fitted and the empirical Bayes method (Smyth, 2004 Stat .Appl .Genet .MoI .Biol . 3, Article3 ) was applied. The significance of the differentially expressed genes was ranked by the B-statistic (Smyth, 2004). In addition, gene expression changes in DCs treated with each chemical were analyzed separately. For the immune toxicity chip, within-array replicates were integrated in the analysis as described previously (Smyth et al., 2005 Bioinformatics . 21, 2067-2075 ) . If a gene was represented by more than one probe or several equal probes on the whole genome array, only the fold-induction and B-value of the probe with the highest B-value were taken. Technical replicates (dye swaps) and biological replicates (experiments employing DCs from different blood donors) were specified for each chemical. For whole genome arrays, only biological replicates were performed.

To visualize differences in gene expression between DCs exposed to diverse chemicals, hierarchical clustering of genes significantly expressed (B > 1) after treatment of iDCs with either chemical was performed.

Real-time PCR

Quantitative real-time PCR was performed as described previously (Szameit et al., 2008) using TaqMan Gene Expression Assays (Applied Biosystems) with RNA isolated from 4 DC-cultures treated with SDS or DNBS and the corresponding solvent controls for 24 h (Table 1) .

Table 1: Genes analyzed with quantitative real-time PCR. Genes were randomly selected from genes identified as differentially expressed with the whole genome array after exposure of iDCs to sensitizers. TaqMan gene expression assays were used (assay ID) .

gene symbol accession number assay ID

ATF3 NM 001030287.2; NM 001040619.1; NM 001674.2; Hs00231069 ml gene symbol accession number assay ID

NM_004024.4

B2M NM_004048.2 Hs99999907_ml

CLCFl NM_013246.2 Hs00757942_ml

CXCL2 NM_002089.3 Hs00236966_ml

FREQ NM_014286.2 Hs00179522_ml

IL8 NM_000584.2 Hs00174103_ml

RIPK2 NM_003821.5 Hs00169419_ml

RPLPO NM_053275.3; NM_001002.3 Hs99999902_ml

SOD2 NM_001024465.1; NM_001024466.1 ; NM_000636.2 Hs00167309_ml

STK17A NM_004760.2 Hs00270504_ml

The same RNA that was used for the microarray experiments was employed. Of each sample, triplicates were performed, nor^¬ malized to 2 housekeeping genes (RPLPO, B2M) and biological rep^¬ licates were analyzed using the 2^{~Δ ΔCT} method (Livak and Schmitt- gen, 2001 Methods 25, 402-408) . After Iog2-transf ormation of the resulting fold-inductions, a one sample t-test was performed to analyze whether fold-inductions were significantly different from 0 (p < 0.05) .

Results

Flow cytometric analysis and dose-response experiments

Before exposure, cells were analyzed for expression of the monocyte/macrophage marker CD14 and the DC-marker CDIa employing flow cytometry. All cultures were constantly CD14-negative and approximately 90% of the cells were highly CDla-positive (data not shown) . Effects of increasing concentrations of HQ, CA, DNBS, DNCB, Eug, MS and Tri on expression of the maturation marker CD86 and PI incorporation were analyzed. Concentrations resulting in 10% - 20% cytotoxicity were chosen for all chemicals (240 μg/ml DNBS, 2 μg/ml DNCB, 212 μg/ml CA, 320 μg/ml Eug, 25 μg/ml HQ, 100 μg/ml MS and 7.5 μg/ml Tri) (Fig. 1). Cytotoxic concentrations (> 20% cytotoxicity) might result in unspecific up-regulation of CD86 expression (Szameit et al . , 2008) and were therefore not used. In previous studies, dose-response curves for iDCs treated with different concentrations of NiSO₄, BB and SDS was already presented (Szameit et al . , 2008). In these studies, gene expression in iDCs exposed to concentrations resulting in ≤ 10% cytotoxicity was analyzed. To obtain 10% to 20% cytotoxicity in the current work, the doses of BB and SDS were in- creased to 10 μg/ml and 90 μg/ml, respectively.

Analysis of Pi-staining in DCs exposed to increasing concentrations of the chemicals revealed a decrease in viability for NiSO₄, CA, DNBS, DNCB, Eug, HQ and SDS. No concentration dependency was detected for the analyzed concentration ranges of BB, MS and Tri . While higher concentrations of Tri were analyzed in preliminary experiments and resulted in a rapid decrease in viability, higher concentrations of BB and MS could not be analyzed due to limits of solubility. Except for DNCB and HQ, treatment with rising chemical concentrations resulted in an increased CD86 expression. For DNCB and HQ, the amount of CD86++ cells was reduced after exposure of DCs to concentrations > 2 μg/ml (viability 87 +/- 2% or 25 μg/ml (viability 81 +/- 7%, respectively (Fig. 1).

Immune toxicity chip

All experiments employing RNA derived from iDCs exposed to allergens (NiSO4, BB, DNBS, DNCB, HQ, CA or Eug) were combined and a linear model was fitted. Log fold-inductions and B-values of the genes significantly up- or down-regulated were calculated and ranked by the B-statistic. 36 probes were found which represent a panel of markers that might be suitable for the discrimination of sensitizers and irritants. The probe for TNFRSFlA is unspecific (Szameit et al . , 2008). Therefore, this gene was excluded from the analysis.

To analyze the applicability of the marker panel, the number of differentially expressed genes after exposure of iDCs to single chemicals was calculated as percentage of the number of genes of the marker panel (Table 2) . Up to 26.7% of the markers were found to be up-regulated even after exposure of iDCs to irritants, and ≤ 14.3% were down-regulated. The weak or moderate allergens (HQ, CA, Eug) were positive for > 33.3% of the up- regulated markers and ≥ 19.1% of the down-regulated markers. The stronger allergens (NiSO₄, BB, DNBS, DNCB) clearly separate from the irritants with > 46.7% up-regulated and ≥ 57.1% down- regulated genes.

Table 2: Number of differentially expressed genes after exposure of iDCs to single chemicals as percentage of the number of genes of the marker panel. all sen- NiSO₄ BB DNBS DNCB HQ CA Eug SDS MS Tri sitizers up- regu- lated 100 93.3 60 46.7 66.7 40 33.3 60 26.7 6.7 20 down- regulated 100 76.2 71.4 66.7 57.1 33.3 38.1 19.1 14.3 0 14.3

Whole genome array

In addition to analysis with the immune toxicity chip, changes in gene expression in allergen (NiSθ₄, BB, DNBS, CA, DNCB, Eug, TMTD) or irritant/non-allergen (SDS, Tri, PropG, EtOH, TryH, Phe, GIy, DMSO, pABA, MS) treated cells were analyzed compared to solvent-treated cells to find genes not represented on the inventive chip that might be suitable as marker genes .

Data evaluation of subset 1 showed that out of > 44000 probes on the whole genome array, 1741 genes were found to be up-regulated after exposure of iDCs to allergens, but not to irritants. Additionally, 2007 genes were identified as down- regulated.

To assess the predictive value of these genes for determining allergenic potential, feature selection, classification and cross-validation were performed, both for subset 1 and for subset 2. The combination of the classifiers resulted in a list of 65 genes that are necessary to predict allergenic potential with 90% accuracy in crossvalidation (Table 4).

For subset 1, a similar classifier was constructed to predict the different chemicals used. 64 genes were necessary to predict the 5 different chemicals with 100% accuracy in cross- validation (Table 3) .

Table 3: Genes necessary to predict different chemicals as determined by cross-validation. gene name accession number

genes necessary to predict different chemicals HS3ST1 NM_005114

EDNRB NM 003991 gene name accession number

EPSTIl NM_033255

KITLG NM_000899

CDK5RAP2 NM_018249

G0S2 NM_015714

TPMl NM_000366

TALI NM_003189

DPYSL3 NM_001387

CG018 NM_052818

THC2269852 THC2269852

GBPl NM_002053

C17orf60 (ENST00000332935) NM_001085423

RGS2 NM_002923

ENCl NM_003633

DPYSL3 NM_001387

TPMl NM_001018004

MCEMPl NM_174918

SPR NM_003124

DKFZP686A01247 NM_014988

S0D2 NM_000636

BDKRB2 NM_000623

F2RL2 NM_004101

SNX5 BC002724

RABl5 NM_198686

RCNl NM_002901

LIPH NM_139248

GBPl NM_002053

IL15 NM_172174

SNN NM_003498

CESl NM_001266

KIAA1274 NM_014431

CR601322 CR601322

SPINKl NM_003122

STAG3 NM_012447

CRH NM_000756

TPMl NM_001018004

GREMl NM_013372

IFIT3 NM_001549

AF116678 AF116678

CD36 S67044 gene name accession number

EBI3 NM_005755

LEPRELl NM_018192

RGSl NM_002922

PKIB NM_181795

NDRG2 NM_201535

ABCC2 NM_000392

SLAMF7 NM_021181

TPMl NM_001018004

PROC NM_000312

CD36 NM_001001547

METTL7A NM_014033

KLFlO NM_005655

LOC222171 NM_175887

AF131834 AF131834

PMP22 NM_000304

GDF15 NM_004864

SNN NM_003498

SIGLEClO NM_033130

HSD3B1 NM_000862

DUSPl NM_004417

PDK4 NM_002612

ENCl NM_003633

ENST00000367932 ENST00000367932

Table 4 : Genes neces sary to predict the al lergenic potential of chemical s as determined by cros s -val idation

Feature selection, classification and cross-validation were performed employing a nearest shrunken centroid method (Tibshi- rani et al . , 2002) . The classifier with the smallest number of genes (largest threshold) was chosen, if more than one threshold yielded the same crossvalidation results. The method was applied to predict the allergenic potential of the chemicals in contrast to the irritating potential and to predict the 5 different chemicals used in the assay.

Hierarchical clustering of the genes significantly up- or down-regulated after exposure of iDCs to either of the chemicals showed the two major groups. The irritants (SDS, Tri) clustered in one group, while the sensitizers (NiSθ4, BB, DNBS) clustered separately. Moreover, the 2 major clusters are further separated into 5 smaller subclusters, representing the 5 different chemicals .

Real-time PCR

The expression of 8 genes found up-regulated in DNBS- treated DCs with the whole genome array was additionally analyzed with real-time PCR, employing RNA from DNBS- and SDS- treated DCs. Up-regulation of expression after DNBS-treatment could be confirmed for all genes (p ≤ 0.05) . Changes in expression of IL8 and CXCL2 were additionally measured with the immune toxicity chip, revealing a lower but non-ambiguous up-regulation compared to the whole genome and PCR results. Real-time PCR of RNA derived from SDS-treated DCs revealed a slight but significant up-regulation of ATF3, CLCFl, FREQ and STK17A. No significant induction of gene expression was found for CXCL2, IL8, RIPK2 and SOD2, neither with microarrays, nor with PCR.

Discussion

To identify a panel of marker genes with the immune toxicity chip, a linear model was fitted for the combined experiments employing RNA derived from allergen-treated DCs. 36 probes showed significant up- or downregulation and might therefore be valuable markers for the identification of sensitizers. Then it was looked at the outputs of linear models fitted for each chemical, to find out how many genes from the marker panel are still differentially expressed if weak sensitizers are analyzed, compared to irritants. The results (Table 2) indicate that identification of strong allergens (NiSO₄, BB, DNBS, DNCB) is possible using the present assay. For these chemicals, > 46.7% of the genes were up-regulated, and > 57.1% down-regulated, compared to ≤ 26.7% and ≤ 19.1%, respectively, for the irritants. However, for weak or moderate allergens (HQ, CA, Eug) , the lowest percentage of up-regulated genes was 33.3%, and 19.1% for the down- regulated genes. Therefore, a discrimination of irritants and weak sensitizers is difficult with the current marker panel, but a bigger set of marker genes might facilitate the analysis.

It has been shown in this example that the expression of many immune-relevant genes changes during DC maturation and mi- gration and that the expression of these genes is also changed by exposure of iDCs to allergenic chemicals. Furthermore, other genes until today not related to the maturation process of DCs might also be valuable marker genes. Therefore, gene expression in DCs treated with selected sensitizers (NiSO₄, BB, DNBS) and irritants (SDS, Tri) employing a whole genome array was analyzed. The results (1741 genes up-regulated, 2007 genes down- regulated) indicate that additional relevant marker genes can be found, potentially enabling the identification of weak sensitizers. To visualize the relevance of these genes as markers, a hierarchical cluster analysis of all genes significantly differentially expressed after exposure of iDCs to one of the chemicals was performed. Furthermore, using these genes, smaller clusters can be found for each chemical, again indicating their relevance .

Finally, besides merely identifying differentially expressed genes, it was also assessed whether any combination of these genes can be used to predict allergenic potential of chemicals. 17 chemicals were tested with our classifier and the classification accuracy was 90%, showing the high reliability of our test system. This accuracy was reached with 65 genes for predicting allergen vs non-allergen.

While quantitative real-time PCR verified the up-regulation of selected genes after DNBS-treatment of iDCs, expression analysis of SDS-treated DCs revealed a significant up-regulation for ATF3, CLCFl, FREQ and STK17A. For ATF3, FREQ and STK17A, log fold-inductions of the microarray results also indicate changes in gene expression, however, these results are not significant (B < 0) . Since real-time PCR has been shown to be more sensitive than microarray analysis, a slight induction of expression of these genes by SDS is likely. However, compared to PCR results with RNA derived from DNBS-treated cells, the induction is only weak, still allowing a differentiation of sensitizers and irritants .

Log fold-inductions of IL8 and CXCL2 detected with the immune toxicity chip were lower than those detected with the whole genome array or with PCR. This indicates that further optimization of the hybridization protocol might even improve gene expression detection with the immune toxicity chip for some probes. The up-regulation of expression of other genes found differentially expressed in this study employing the immune toxicity chip (CCR7, CXCR4, IL12B p40, IL15) was already confirmed by real-time PCR with RNA derived from NiSθ4-exposed cells in an earlier work (Szameit et al . , 2008).

Claims

Claims :

1. Method for identifying irritating or allergenic potential of a product comprising the steps of: a) contacting the product with immature dendritic cells, b) determining the amount of gene products of the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorf162, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MB0AT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2, SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941 expressed in said cells upon contact with the product, c) subtracting the amount of gene products of the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorfl62, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2, SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941 expressed in said cells without contact with the product from the respective amounts determined under b) to obtain a set of product-caused amounts of gene products for each of said genes, d) comparing the product-caused amounts of gene products for each of said genes obtained under c) with a test set of product-caused amounts of gene products for each of said genes, wherein said test set of product-caused amounts of gene products for each of said genes was provided from at least two allergenic products and at least two irritating products and, optionally, from at least two products which are neither allergenic nor ir^¬ ritating; and wherein the comparing is performed by a pattern recognition method which recognises an allergenic pattern and an irritating pattern and, optionally, a non-allergenic and non- irritating pattern; and e) identifying an allergenic potential of a product when the comparison under d) results in the recognition of an allergenic pattern and identifying an irritating potential of a product when the comparison under d) results in the recognition of an irritating pattern and, optionally, identifying a non- allergenic and non-irritating product when the comparison under d) results in the recognition of a non-allergenic and non- irritating pattern.

2. Method according to claim 1, characterised in that said immature dendritic cells are immature Langerhans cells.

3. Method according to claim 1, characterised in that said immature dendritic cells are derived from peripheral blood monocytes (PBMCs) or from CD34⁺-stem cells.

4. Method according to any one of claims 1 to 3, characterised in that the amount of gene products is determined by reverse transcribing mRNA to produce cDNA and subjecting said cDNA to a real-time polymerase chain reaction and/or to a hybridisation assay.

5. Method according to claim 4, characterised in that said hybridisation assay is a microarray assay or a microsphere assay.

6. Method according to any one of claims 1 to 3, characterised in that the amount of gene products is determined by using antibodies binding to said gene products.

7. Microarray having immobilised on its surface probes for the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorf162, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MB0AT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, P0LE4, PPFIBPl, RBPJ, RELB, RSAD2, SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941 and wherein said microarray comprises at least 10%, preferably at least 20%, more preferably at least 30% of the total number of probes are probes for the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorfl62, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2 , FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, IT- GAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2 , SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941.

8. Microarray according to claim 7, characterised in that it additionally comprises probes for at least five, preferably of at least ten, more preferably of at least 15 normalizing genes selected from Table B.

9. Microarray according to claim 7 or 8, characterised in that it has a surface which is coated at least partially with a phenolic resin polymer with a functionality of from 6 to 15, preferably from 7 to 10, most preferably 8.

10. Kit, containing probes for the genes ABCG2, AOAH, ARHGAP22, ARID5B, BAIl, BTN2A2, Clorfl62, CCDC26, CD274, CD38, CDYL2, CLCFl, CTPS, CYBRDl, DHRS9, DNAJC12, EHF, EMP2, EPSTIl, FABP4, FAS, FBN2, FZDl, G0S2, GEM, GPlBA, H2AFY2, HSD3B1, IFIT3, ILlRN, IL4I1, ITGAD, KCNE3, KLHL24, LOC124220, LOC541471, LPCAT2, LY75, MBOAT2, MET, MS4A3, N4BP3, NEDD4L, PDLIM4, PHLDB3, PKIB, PLA2G4C, POLE4, PPFIBPl, RBPJ, RELB, RSAD2 , SlOOAlO, S100A4, SLAMFl, SLC9A9, SPOCDl, SYTIl, TCP11L2, TM4SF1, TMC5, TNFRSF12A, TPMl, TRIM15 and WDR66 and optionally CENTD3, MARCKS, AK026517, LGP2 and/or BU678941.

11. Kit according to claim 10, characterised in that it further comprises probes least five, preferably of at least ten, more preferably of at least 15 normalizing genes selected from Table