ANALYTICAL METHODS AND ARRAYS FOR USE IN THE SAME

Abstract

The present invention relates to an in vitro method for identifying agents capable of inducing sensitization of human skin and arrays and diagnostic kits for use in such methods. In particular, the methods include measurement of the expression of the biomarkers listed in Table 3A and/or 3B in MUTZ-3 cells exposed to a test agent.

Claims

1-81. (canceled)

82. An in vitro method for detecting human skin sensitization inducers comprising the steps of: a) exposing a population of dendritic cells or a population of dendritic-like cells to a test agent; and b) measuring in the exposed cells of step a) the expression of nucleic acid molecules encoding each of the following biomarkers: i) squalene epoxidase (SQLE), ii) taste receptor, type 2, member 5 (TAS2R5), iii) keratinocyte growth factor-like protein 1/2/hypothetical protein FLJ20444 (KGFLP1/2/FLJ20444), iv) transmembrane anterior posterior transformation 1 (TAPT1), v) sprouty homolog 2 (SPRY2), vi) fatty acid synthase (FASN), vii) B-cell CLL/lymphoma 7A (BCL7A), viii) solute carrier family 25, member 32 (SLC25A32), ix) ferritin, heavy polypeptide pseudogene 1 (FTHP1), x) ATPase, H+ transporting, lysosomal 50/57 kDa, V1 subunit H (ATP6V1H), and xi) histone cluster 1, H1e (HIST1H1E); wherein said test agent is a human skin sensitization inducer when the expression of the biomarkers measured in step b) is different compared to the expression measured in cells exposed to a control non-sensitizing agent.

83. The method according to claim 82 further comprising: i) exposing a separate population of the dendritic cells or dendritic-like cells to a negative control agent that does not sensitize human skin and measuring in the cells the expression of the biomarkers measured in step (b), and/or ii) exposing a separate population of the dendritic cells or dendritic-like cells to a positive control agent that sensitizes human skin and measuring in the cells the expression of the biomarkers measured in step (b), wherein the test agent is a human skin sensitization inducer when the expression of the biomarkers measured in step b) is different compared to the expression measured in cells exposed to a negative control agent and/or similar to the expression measured in cells exposed to a positive control agent.

84. The method according to claim 82 wherein step (b) further comprises measuring the expression of at least one nucleic acid molecule encoding one of the following biomarkers: 4-aminobutyrate aminotransferase (ABAT), abhydrolase domain containing 5 (ABHD5), alkaline ceramidase 2 (ACER2), ATP citrate lyase (ACLY), actin-related protein 10 homolog (ACTR10), ADAM metallopeptidase domain 20 (ADAM20), aldehyde dehydrogenase 18 family, member A1 (ALDH18A1), aldehyde dehydrogenase 1 family, member B1 (ALDH1B1), alkB, alkylation repair homolog 6 (ALKBH6), anaphase promoting complex subunit 1 (ANAPC1), anaphase promoting complex subunit 5 (ANAPCS), ankyrin repeat, family A (RFXANK-like), 2 (ANKRA2), ADP-ribosylation factor GTPase activating protein 3 (ARFGAP3), Rho GTPase activating protein 9 (ARHGAP9), ankyrin repeat and SOCS box-containing 7 (ASB7), ATPase, H+ transporting, lysosomal 9 kDa, V0 subunit e1 (ATP6V0E1), bridging integrator 2 (BIN2), bleomycin hydrolase (BLMH), brix domain containing 1/ribosome production factor 2 homolog (BXDC1/RPF2), chromosome 11 open reading frame 67 (C11orf67), chromosome 12 open reading frame 57 (C12orf57), chromosome 15 open reading frame 24 (C15orf24), chromosome 19 open reading frame 54 (C19orf54), chromosome 1 open reading frame 174 (C1orf174), chromosome 1 open reading frame 183 (C1orf183), chromosome 20 open reading frame 111 (C20orf111), chromosome 20 open reading frame 24 (C20orf24), chromosome 3 open reading frame 62/ubiquitin specific peptidase 4 (proto-oncogene) (C3orf62/USP4), chromosome 9 open reading frame 89 (C9orf89), coactivator-associated arginine methyltransferase 1 (CARM1), CD33 molecule (CD33), CD86 molecule (CD86), CD93 molecule (CD93), cytochrome c oxidase subunit VIIa polypeptide 2 like (COX7A2L), corticotropin releasing hormone binding protein (CRHBP), chondroitin sulfate N-acetylgalactosaminyltransferase 2 (CSGALNACT2), Cytochrome P450 51A1 (CYP51A1), DDRGK domain containing 1 (DDRGK1), DEAD (Asp-Glu-Ala-Asp) box polypeptide 21 (DDX21), 24-dehydrocholesterol reductase (DHCR24), 7-dehydrocholesterol reductase (DHCR7), DEAH (Asp-Glu-Ala-His) box polypeptide 33 (DHX33), DnaJ (Hsp40) homolog, subfamily B, member 4 (DNAJB4), DnaJ (Hsp40) homolog, subfamily B, member 9 (DNAJB9), DnaJ (Hsp40) homolog, subfamily C, member 5 (DNAJC5), DnaJ (Hsp40) homolog, subfamily C, member 9 (DNAJC9), D-tyrosyl-tRNAdeacylase 1 homolog (DTD1), ER degradation enhancer, mannosidase alpha-like 2(EDEM2), ecotropic viral integration site 2B (EVI2B), family with sequence similarity 36, member A (FAM36A), family with sequence similarity 86, member A (FAM86A), Fas (TNF receptor superfamily, member 6) (FAS), MGC44478 (FDPSL2A), ferredoxinreductase (FDXR), forkhead box 04 (FOXO4), FTHL10-001, Transcribed processed pseudogene (FTHL10-001), fucosidase, alpha-L-2, plasma (FUCA2), growth arrest-specific 2 like 3 (GAS2L3), ganglioside induced differentiation associated protein 2 (GDAP2), growth differentiation factor 11 (GDF11), glutaredoxin (thioltransferase) (GLRX), guanine nucleotide binding protein-like 3 (GNL3L), glucosamine-phosphate N-acetyltransferase 1 (GNPNAT1), glutathione reductase (GSR), general transcription factor IIIC, polypeptide 2 beta (GTF3C2), HMG-box transcription factor 1(HBP1), histone cluster 1, H1c (HIST1H1C), histone cluster 1, H2ae (HIST1H2AE), histone cluster 1, H2be (HIST1H2BE), histone cluster 1, H3g (HIST1H3G), histone cluster 1, H3j (HIST1H3J), histone cluster 1, H4a (HIST1H4A), histone clusters 2, H2aa3/2, H2aa4 (HIST2H2AA3/4), high-mobility group box 3 (HMGB3), 3-hydroxy-3-methylglutaryl-Coenzyme A reductase (HMGCR), 3-hydroxy-3-methylglutaryl-Coenzyme A synthase 1 (HMGCS1), hemeoxygenase (decycling) 1 (HMOX1), heterogeneous nuclear ribonucleoprotein L (HNRNPL), insulin receptor substrate 2 (IRS2), iron-sulfur cluster scaffold homolog (ISCU), interferon stimulated exonuclease gene 20 kDa-like 2 (ISG20L2), potassium voltage-gated channel, Isk-related family, member 3 (KCNE3), hypothetical protein LOC100132855/ATPase, H+ transporting, lysosomal 38 kDa, V0 subunit d1 (LOC100132855/ATP6V0D1), hCG1651476 (LOC284417), lysophosphatidic acid receptor 1 (LPAR1), leucine-rich PPR-motif containing (LRPPRC), lymphocyte antigen 96 (LY96), mitogen-activated protein kinase kinase 1 (MAP2K1), mitogen-activated protein kinase 13 (MAPK13), methyltransferase like 2A (METTL2A), microsomal glutathione S-transferase 3 (MGST3), mitochondrial ribosomal protein L30 (MRPL30), mitochondrial ribosomal protein L4 (MRPL4), mitochondrial ribosomal protein S17 (MRPS17), 5-methyltetrahydrofolate-homocysteine methyltransferase (MTR), MYB binding protein (P160) 1a (MYBBP1A), neighbor of BRCA1 gene 1 (NBR1), nuclear import 7 homolog (NIP7), NLR family, pyrin domain containing 12 (NLRP12), nucleolar protein family 6 (RNA-associated) (NOL6), NAD(P)H dehydrogenase, quinone 1 (NQO1), nuclear receptor binding protein 1 (NRBP1), nucleotide binding protein-like (NUBPL), nudix (nucleoside diphosphate linked moiety X)-type motif 14 (NUDT14), nuclear fragile X mental retardation protein interacting protein 1 (NUFIP1), nucleoporin 153 kDa (NUP153), olfactory receptor, family 5, subfamily B, member 21 (OR5B21), PAS domain containing serine/threonine kinase (PASK), PRKC, apoptosis, WT1, regulator (PAWR), PDGFA associated protein 1 (PDAP1), phosphodiesterase 1B, calmodulin-dependent (PDE1B), phosphoribosylformylglycinamidine synthase (PFAS), pleckstrin homology-like domain, family A, member 3 (PHLDA3), phosphoinositide-3-kinase adaptor protein 1 (PIK3AP1), PTEN induced putative kinase 1 (PINK1), phosphomannomutase 2 (PMM2), partner of NOB1 homolog (PNO1), polymerase (RNA) II (DNA directed) polypeptide E, 25 kDa (POLR2E), polymerase (RNA) III (DNA directed) polypeptide E (80kD) (POLR3E), protein phosphatase 1D magnesium-dependent, delta isoform (PPM1D), phosphatidylinositol-3,4,5-trisphosphate-dependent Rac exchange factor 1 (PREX1), proline-serine-threonine phosphatase interacting protein 1 (PSTPIP1), RAB33B, member RAS oncogene family (RAB33B), renin binding protein (RENBP), replication factor C (activator 1) 2, 40 kDa (RFC2), ribonuclease H1 (RNASEH1), ring finger protein 146 (RNF146), ring finger protein 24 (RNF24), ring finger protein 26 (RNF26), ribosomal protein SA/small nucleolar RNA, H/ACA box 62 (RPSA/SNORA62), RNA pseudouridylate synthase domain containing 2 (RPUSD2), ribosomal RNA processing 12 homolog (RRP12), retinoid X receptor, alpha (RXRA), scavenger receptor class B, member 2 (SCARB2), SERPINE1 mRNA binding protein 1 (SERBP1), splicing factor proline/glutamine-rich (SFPQ), solute carrier family 35, member B3 (SLC35B3), solute carrier family 37, member 4 (SLC37A4), solute carrier family 5, member 6 (SLC5A6), sphingomyelinphosphodiesterase 4, neutral membrane (SMPD4), small nucleolar(sn)RNA host gene 1, non-coding/snRNA C/D box 26 (SNHG1/SNORD26), small nucleolar RNA host gene 12 (non-coding) (SNHG12), small nucleolar RNA, H/ACA box 45 (SNORA45), sorting nexin family member 27 (SNX27), sterol regulatory element binding transcription factor 2 (SREBF2), ST3 beta-galactoside alpha-2,3-sialyltransferase 6 (ST3GAL6), serine/threonine kinase 17b (STK17B), tubulin folding cofactor E-like (TBCEL), tectonic family member 2 (TCTN2), toll-like receptor 6 (TLR6), toll-like receptor 9/twinfilin homolog 2 (TLR9/TWF2), transmembrane protein 55A (TMEM55A), transmembrane protein 59 (TMEM59), transmembrane protein 77 (TMEM77), transmembrane protein 97 (TMEM97), translocase of outer mitochondrial membrane 34 (TOMM34), translocase of outer mitochondrial membrane 40 homolog (TOMM40), translocase of outer mitochondrial membrane 5 homolog/F-box protein 10 (TOMM5/FBXO10), tumor protein p53 inducible protein 3 (TP53I3), tumor protein p53 inducible nuclear protein 1 (TP53INP1), thioredoxinreductase 1 (TXNRD1), ubiquitin-fold modifier conjugating enzyme 1 (UFC1), ubiquitin specific peptidase 10 (USP10), vesicle-associated membrane protein 3 (cellubrevin) (VAMP3), valyl-tRNAsynthetase (VARS), vacuolar protein sorting 37 homolog A (VPS37A), zinc finger protein 211 (ZNF211), zinc finger protein 223 (ZNF223), zinc finger protein 561 (ZNF561), and zinc finger protein 79 (ZNF79).

85. The method according to claim 82 wherein measuring the expression of the nucleic acid molecules encoding the biomarkers in step (b) is performed using a method selected from the group consisting of Southern hybridization, Northern hybridization, polymerase chain reaction (PCR), reverse transcriptase PCR (RT-PCR), quantitative real-time PCR (qRT-PCR), nanoarray, microarray, macroarray, autoradiography and in situ hybridization.

86. The method according to claim 82 wherein measuring the expression of the nucleic acid molecules encoding the biomarkers in step (b) is performed using one or more binding moieties, each binding moiety capable of binding selectively to a nucleic acid molecule encoding one of the biomarkers.

87. The method according to claim 82 wherein step (b) is performed using an array.

88. The method according to claim 82 wherein the skin sensitization is allergic contact dermatitis (ACD).

89. The method according to claim 82 wherein the population of dendritic cells or population of dendritic-like cells is a population of dendritic-like cells.

90. The method according to claim 89 wherein the dendritic-like cells are myeloid leukaemia-derived cells selected from the group consisting of KG-1, THP-1, U 937, HL-60, Monomac-6, AML-193 and MUTZ-3.

91. The method according to claim 83 wherein the negative control agent is selected from the group consisting of 1-Butanol, 4-Aminobenzoic acid, Benzaldehyde, Chlorobenzene, Diethyl phthalate, Dimethyl formamide, Ethyl vanillin, Glycerol, Isopropanol, Lactic acid, Methyl salicylate, Octanoic acid, Propylene glycol, Phenol, p-hydroxybenzoic acid, Potassium permanganate, Salicylic acid, Sodium dodecyl sulphate, Tween 80 and Zinc sulphate.

92. The method according to claim 83 wherein the positive control agent is selected from the group consisting of 2,4-Dinitrochlorobenzene, Oxazolone, Potassium dichromate, Kathon CH (MC/MCI), Formaldehyde, 2-Aminophenol, 2-nitro-1,4-Phenylendiamine, p Phenylendiamine, Hexylcinnamic aldehyde, 2-Hydroxyethyl acrylate, 2 Mercaptobenzothiazole, Glyoxal, Cinnamaldehyde, Isoeugenol, Ethylendiamine, Resorcinol, Cinnamic alcohol, Eugenol, Penicillin G or Geraniol.

93. The method of claim 82, said method consisting of the steps of: a) exposing a population of dendritic cells or a population of dendritic-like cells to a test agent; and b) measuring in the exposed cells of step a) the expression of nucleic acid molecules encoding each of the following biomarkers: i) squalene epoxidase (SQLE), ii) taste receptor, type 2, member 5 (TAS2R5), iii) keratinocyte growth factor-like protein 1/2/hypothetical protein FLJ20444 (KGFLP1/2/FLJ20444), iv) transmembrane anterior posterior transformation 1 (TAPT1), v) sprouty homolog 2 (SPRY2), vi) fatty acid synthase (FASN), vii) B-cell CLL/lymphoma 7A (BCL7A), viii) solute carrier family 25, member 32 (SLC25A32), ix) ferritin, heavy polypeptide pseudogene 1 (FTHP1), x) ATPase, H+ transporting, lysosomal 50/57 kDa, V1 subunit H (ATP6V1H), and xi) histone cluster 1, H1e (HIST1H1E).

Description

[0141] Preferred, non-limiting examples which embody certain aspects of the invention will now be described, with reference to the following figures:

[0142] FIG. 1. Phenotype of MUTZ-3 Cells Prior to Stimulation with Sensitizing and Non-Sensitizing Chemicals

[0143] Cell surface expression levels of CD14, CD1a, CD34, CD54, CD80, CD86 and HLA-DR were assessed with flow cytometry. Gates were set to exclude debris and dead cells, and quadrants were established by comparing with relevant isotype controls. Results are shown from one representative experiment out of six.

[0144] FIGS. 2A and 2B. Changes in CD86 Expression Following Stimulation with Sensitizing and Non-Sensitizing Chemicals

[0145] Cell surface expression levels of CD86 were monitored after stimulation with chemicals for 24 h. FIG. 2A). Chemical-induced upregulation of CD86, in terms of changes in frequency of positive cells, were determined by flow cytometry, as exemplified by the comparison of 2-aminophenol-stimulated cells (right dotplot) and unstimulated controls (left dotplot). Results are shown from one representative experiment out of three. Gates were set to exclude debris and dead cells, and quadrants were established by comparing with relevant isotype controls. FIG. 2B) Compilation of frequencies of CD86-positive cells after 24 h of stimulation. Statistical analysis was performed using paired Student's t test. *p<0.05, # p<0.01. NA, not analysed.

[0146] FIGS. 3A-3D. Principal Component Analysis of Transcripts Differentially Expressed after Chemical Stimulation

[0147] mRNA levels in MUTZ-3 cells stimulated for 24 h with 20 sensitizing and 20 non-sensitizing chemicals were assessed with transcritomics using Affymetrix Human Gene 1.0 ST arrays. Structures and similarities in the gene expression dataset were investigated using principal component analysis (PCA) in the software Qlucore. FIG. 3A) PCA of genes differentially expressed in cells stimulated with sensitizing (red) as compared to non-sensitizing (green) chemicals (1010 genes identified with one-way ANOVA). FIG. 3B) PCA of genes differentially expressed in cells stimulated with sensitizing as compared to non-sensitizing chemicals (1010 genes), but now samples are coloured by the compound used for stimulation. FIG. 3C) PCA of genes differentially expressed when comparing the different stimulations with 2-way ANOVA (1137 genes). Samples are coloured according to sensitizing (red) and non-sensitizing (green) chemicals. FIG. 3D) PCA of genes differentially expressed when comparing the different stimulations with 2-way ANOVA (1137 genes), but now samples are coloured by the compound used for stimulation. P, p-value from ANOVA. Q, p-value corrected for multiple hypothesis testing.

[0148] FIGS. 4A-4C. Identification and PCA Analysis of “Prediction Signature”

[0149] FIG. 4A) The number of differentially expressed significant genes in cells stimulated with sensitizing as compared to non-sensitizing chemicals (1010 genes) was reduced using Backward Elimination. The lowest KLD is observed after elimination of 810 analytes, referred to as the Breakpoint. The remaining 200 genes is considered to be the top predictors in the data set, and is termed “Prediction Signature”. FIG. 4B) Complete separation between sensitizers (red) and non-sensitizers (green) is observed with PCA of the “Prediction Signature”. FIG. 4C) Same PCA as in B, now with samples coloured according to their potency in LLNA.

[0150] FIG. 5. Validation of Selection Procedure of “Prediction Signature”

[0151] The method by which the “Prediction Signature” was constructed was validated by repeating the process on 70% of the data set, selected at random. The remaining 30% of data was used as a test set for signature validation. A) PCA demonstrates that the “Test Gene Signature” can separate sensitizers from non-sensitizers. Only the samples of the 70% training set, displayed in bright colours, were used to build the space of the first three principal components. The test set samples, displayed in dark colours, were plotted into this space based on expression levels of the analytes in the “Test Gene Signature”. B) An SVM was trained on the 70% training set, and validated with the 30% test set. The area under the ROC curve of 1.0 proves that the group belonging of all samples in the test set was correctly predicted, demonstrating the strength of the “Test Gene Signature”, and by association also the strength of our “Prediction Signature”.

[0152] FIG. 6. The Interactome of the “Prediction Signature”

[0153] Interactome of the 200 molecules (orange) and molecules connecting theses according to evidence from IPA. Direct interactions are shown as solid lines and indirect as dotted lines.

EXAMPLES

[0154] Allergic contact dermatitis is an inflammatory skin disease that affects a significant proportion of the population. This is commonly caused by immunological responses towards chemical haptens leading to substantial economic burden for society. Current test of sensitizing chemicals rely on animal experimentation. New legislations on the registration and use of chemicals within e.g. pharmaceutical and cosmetic industries have stimulated significant research efforts to develop alternative human cell-based assays for the prediction of sensitization. The aim is to replace animal experiments with in vitro tests displaying a higher predictive power.

[0155] We have developed a novel cell-based assay for the prediction of sensitizing chemicals. By analyzing the transcriptome of the human cell line MUTZ-3 after 24 h stimulation with 20 different sensitizing chemicals, 20 non-sensitizing chemicals and vehicle controls, we have identified a biomarker signature of 200 genes with potent discriminatory ability. Using a Support Vector Machine for supervised classification, the prediction performance of the assay revealed an accuracy of 100%, sensitivity of 100% and specificity of 100%. In addition, categorizing the chemicals according to the LLNA assay, this gene signature could also predict potency. The identified markers are involved in biological pathways with immunological relevant functions, which can shed light on the process of human sensitization.

[0156] A gene signature predicting sensitization using a human cell line in vitro has been developed. This easy and robust cell-based assay can completely replace or drastically reduce current test systems, using experimental animals. Being based on human biology, the assay is considered to be more relevant and more accurate for predicting sensitization in humans than the traditional animal-based tests.

[0157] Results

[0158] The Cellular Rational of the In Vitro Cell Culture System

[0159] Acting as the link between innate and adaptive immunity, DCs are essential immunoregulatory cells of the immune system. Their unique property to recognize antigen for the purpose of initiating T cell responses, and their potent regulatory function in skewing immune responses, makes them targets for assay development. However, primary DCs constitute a heterogeneous and minor population of cells not suited for screening. The obvious advantages of using a cell line with characteristics compared to primary DCs for the basis of a predictive test are stability, reproducibility and unlimited supply of cells. So far, no leukemia with obvious DC-like properties has been reported, probably due to the fact that the characteristics of this cell type are determined by a complex terminal differentiation process that can occur only post-cell division [11], and thus, the generation of DC-like cell lines relies on available myeloid leukemia cell lines. MUTZ-3 is a human acute myelomonocytic leukemia cell line with a potent ability to differentiate into DCs, present antigens and induce specific T-cell proliferation. Among the available myeloid human cell lines, MUTZ-3 is by far the preferred candidate. Similar to immature primary DCs, MUTZ-3 progenitor express CD1a, HLA-DR and CD54, as well as low levels of CD80 and CD86 (FIG. 1). The MUTZ-3 population also contains three subpopulations of CD14.sup.+, CD34.sup.+ and double negative cells, previously reported to be transitional differentiation steps from a proliferative CD34.sup.+ progenitor into a non-proliferative CD14.sup.+ DC precursor [9]. Consequently, we utilized constitutively differentiating progenitor MUTZ-3 cells as the basis for the test system.

[0160] CD86 Surface Expression in Response to Sensitizer Stimulation:

[0161] CD86 is the most extensively studied biomarker for sensitization to date, in cell-systems such as monocyte derived dendritic cells (MoDCs) or dendritic cell-like human cell lines and their progenitors, such as THP-1, U-937 and KG-1. Thus, as a reference, cell surface expression of CD86 was measured with flow cytometry after 24h stimulation with 20 sensitizers and 20 non-sensitizers, as well as with vehicle controls (Table 1). CD86 was significantly up-regulated on cells stimulated with 2-Aminophenol, Kathon CG, 2-nitro-1,4-Phenylendiamine, 2,4-Dinitrochlorobenzene, 2-Hydroxyethyl acrylate, Cinnamic aldehyde, p-Phenylendiamine, Resorcinol, and 2-Mercaptobenzothiazole. Hence, an assay based on measurement of a single biomarker, such as CD86, would give a sensitivity of 47% and a specificity of 100%. Consequently, CD86 cannot classify skin sensitizers, using a cell based system such as MUTZ-3.

[0162] Analysis of the Transcriptional Profiles in Chemically Stimulated MUTZ-3 Cells:

[0163] The genomic expression arrays were used to test 20 sensitizers and 20 non-sensitizers, in triplicates, and vehicle controls such as DMSO and distilled water, the latter in twelve replicates. In total, a data set was generated based on 144 samples. RMA normalization and quality controls of the samples revealed that the Oxazolone and Cinnamic aldehyde samples were significant outliers and had to be removed, or they would have dominated the data set prohibiting biomarker identification (data not shown). In addition, one of the replicates of potassium permanganate had to be removed due to a faulty array. This left a data set consisting of 137 samples, each with data from measurements of 33,297 transcripts. In order to mine the data set for information specific for sensitizers vs. non-sensitizers, the software Qlucore Omics Explorer 2.1 was used, which enable real time principal component analysis (PCA) analysis, while sorting the input genes after desired criteria, e.g. sensitizers and non-sensitizers, based on ANOVA p-value selection. FIGS. 3A and 3B show PCAs based on 1010 transcripts with a p-value of ≦2.0×10.sup.−6 from ANOVA analysis, comparing sensitizing vs. non-sensitizing chemicals. As can be seen in FIG. 3A, a clear distinction can be made between the two groups, with non-sensitizers forming a condensed cloud in the lower part of the figure (green), while sensitizers stretch upwards in various directions (red). However, a complete separation is not achieved between the two groups at this level of significance. From FIG. 3B, now coloured according to stimulation, it is evident that one or more replicate of Glyoxal, Eugenol, Hexylcinnamic aldehyde, lsoeugenol, Resorcinol, Penicillin G and Ethylendiamine group together with the control group. In addition, one replicate or more of the non-sensitizers Tween 80, Octanoic acid and Phenol tend to group with the sensitizers. FIGS. 3C and 3D show PCA plots based on 1137 genes that all have a p-value of 7.0×10.sup.−21, when comparing the different stimulations. Identifying this large number of genes at this level of significance provides strong indications of the power of the data set. In FIG. 3D, it is clear that the replicates group together, indicating high quality data. The triplicate samples of Potassium dichromate have a discrete profile, which demonstrate a substantial impact of the cells compared to non-sensitizers. Furthermore, 2-Hydroxyethyl acrylate, 2-Aminophenol, Kathon CG, Formaldehyde, 2-nitro-1,4-Phenylendiamine, 2,4-Dinitrochlorobenzoic acid, p-Phenylendiamine, 2-Mercaptobenzothiazole, Cinnamic alcohol and Resorcinol have replicates that group together, separate from the negative group. Still, as can be seen in FIG. 3C as well as in 3A, complete separation is not achieved with neither of the gene signatures of 1010 and 1137 genes respectively.

[0164] Backward Elimination Identifies Genes with the Most Discriminatory Power:

[0165] Even though the data set contains genes with p-values down to 1×10.sup.−17, lowering the p-value cutoff did not achieve complete separation between sensitizers and non-sensitizers. Gene signatures entirely selected on p-values does not provide the best possible predictive power, since a low p-value is no guarantee that a gene provides any additional information. To further reduce the number of transcripts for a predictive biomarker signature, we employed an algorithm for backward elimination (FIG. 4A). The algorithm removes genes one by one while taking into account not only the impact of genes individually, but how they perform collectively with the entire selected gene signature. For each gene eliminated, the Kullback-Leibler divergence (KLD) value is lowered, until a breakpoint is reached, at which point 200 genes remained. Continuing eliminating genes at this point causes the KLD to rise again, indicating that information is being lost (FIG. 4A). Therefore, the 200 genes with lowest KLD value were selected for further analysis. PCA of the 200 analytes revealed that they have the ability to completely separate sensitizers vs. non-sensitizers, indicating that these transcripts can be used as predictors for sensitizing properties of unknown samples (FIG. 4B). Importantly, by coloring the samples in the PCA by their potency, according to LLNA, it is clear that potency can be predicted by these genes, as well (FIG. 4C). The 200 genes are termed the “Prediction Signature” and their identity is listed in Table 3.

[0166] Validation of the Analysis Method Used to Identify the Prediction Signature:

[0167] To validate the predictive power of our signature, we used a supervised learning method called the Support Vector Machine (SVM) [12], which maps the data from a training set in space in order to maximize the separation of gene expression induced by sensitizing and non-sensitizing chemicals. As training set, 70% of the data set was selected randomly and the entire selection process (as described above) was repeated. Starting with 29,141 transcripts, the signature was reduced to 200 transcripts, termed “Test Gene Signature”, using ANOVA filtering and backward elimination. The remaining 30% of the data set was used to test the signature obtained. The partitioning of the data set into subsets of 70% training data set and 30% test data set was done in a stratified random manner, meaning that the proportion of sensitizers and non-sensitizers in the complete data set are maintained in both the subsets, although the samples included in either of the two subsets are selected at random. Thereafter, the “Test Gene Signature” was used to train an SVM model with the training set, and the predictive power of the model was assessed with the test set. FIG. 5A shows a PCA plot based on the “Test Gene Signature” and the samples of the test set. Clearly, the separation between sensitizers (red) and non-sensitizers (green) resembles the one observed for the “Prediction Signature” in FIG. 4B. In the PCA of FIG. 5A, the samples of the sensitizing and non-sensitizing chemicals of the test set have been colored dark red and dark green respectively, indicating that they are not contributing to the principal components of the plot, but are merely plotted based on their expression values of the selected “Test Gene Signature”. As can be seen, sensitizers from the test set group with sensitizers from the training set, while non-sensitizers from the test set group with non-sensitizers from the training set. This is a very intuitive way of predicting the group belonging of the samples in the test set, using only the pattern recognition of the eye. The outcome of the SVM training and validation can be seen in FIG. 5B, where an area under the ROC curve of 1 confirms the ability of the “Test Gene Signature” to predict which group the sample belonged to in the test set. The prediction performance of the assay reveals an accuracy of 96%, sensitivity of 100% and specificity of 92%. While this experiment does not validate the prediction power of the “Prediction Signature” per se, it does indeed validate the method by which it has been selected, supporting the claim that the “Prediction Signature” is capable of accurately predict sensitizing properties of unknown samples.

[0168] Molecular Functions and Canonical Pathways of the “Prediction Signature”.

[0169] Using Ingenuity Pathways Analysis (IPA, Ingenuity Systems Inc.), 184 of the 200 molecules in the signature were characterized with regard to functions and known (canonical) pathways. The remaining 16 molecules could not be mapped to any IPA entries. The dominating functions identified were small molecule biochemistry (38 molecules), cell death (33), lipid metabolism (24), hematological system development (19), cellular growth and proliferation (16), molecular transport (15), cell cycle (15) and carbohydrate metabolism (15), see table 4 for details. 67 of the 184 molecules were involved in the listed functions. Of the remaining 117 molecules, 30 were known from a variety of human diseases and molecular functions, such as described biomarkers (SCARB2, RFC2, VPS37A and BCL7A) and drug targets (ABAT). Most of these molecules are metabolic markers. In the signature as a whole, there are several drug targets, such as HMGCR, HMOX1, ABAT, RXRA, CD33, MAP2K1, MAPK13 and CD86. Two are described for skin disorders: CD86 (psoriasis) and RXRA (eczema). The signature also contains skin development (DHCR24) and dendritic cell markers (MAP2K1, NLRP12 and RFC2).

[0170] Pathways possibly invoked by the molecules in the signature were also investigated using IPA. Those most highly populated were NRF2 mediated oxidative response (10), xenobiotic metabolism signaling (8), LPS/IL-1 mediated inhibition of RXR function (6), aryl hydrocarbon receptor signaling (6) and protein kinase A signaling (6). The five highest ranked of these pathways are all known to take part in reactions provoked by foreign substances, xenobiotics. Xenobiotics are natural or synthetic chemical compounds, foreign to the human body.

[0171] Conclusions

[0172] Allergic contact dermatitis (ACD) is an inflammatory skin disease caused by dysregulated adaptive immune responses to allergens [13]. Small molecular weight chemicals, so-called haptens, can bind self-proteins in the skin, which enables internalisation of the protein-bound allergenic chemical by skin dendritic cell (DC). DCs, under the influence of the local microenvironment, process the protein-hapten complex, migrate to the local lymph nodes and activate naïve T cells. The initiation and development of allergen-specific responses, mainly effector CD8+ T cells and Th1 cells, and production of immunoregulatory proteins, are hallmarks of the immune activation observed in ACD. This T-cell mediated type IV hypersensitivity reaction is characterised by symptoms such as rash, blisters and itching. ACD is the most common manifestation of immunotoxicity observed in humans [13] and hundreds of chemicals have been shown to cause sensitization in skin [14]. The driving factors and molecular mechanisms involved in sensitization are still unknown even though intense research efforts have been carried out to identify the immunological responses towards allergenic chemicals. The REACH legislation requires that all chemicals produced over 1 ton/year are tested for hazardous properties such as toxicity and allergenicity [5], which increase the demand for accurate assays for predictive power. Today, the identification of potential human sensitizers relies on animal experimentation, in particular the murine local lymph node assay (LLNA) [6]. The LLNA is based upon measurements of proliferation induced in draining lymph nodes of mice after chemical exposure [15]. Chemicals are defined as sensitizers if they provoke a three-fold increase in proliferation compared to control, and the amount of chemical required for the increase is the EC3 value. Thus, the LLNA can also be used to categorize the chemicals based on sensitisation potency. However, LLNA is in many ways not optimal. Besides the obvious ethical reasons, the assay is also time consuming and expensive. Human sensitization data often stem from human maximization tests (HMT) [16] and human patch tests (HPT). In an extensive report from the Interagency Coordinating Committee on the Validation of Alternative Methods (ICCVAM), the performance characteristics of LLNA were compared to other available animal-based methods and human sensitization data (HMT and HPT) [17]. The LLNA performance in comparison to human data (74 assessments) revealed an accuracy of 72%, a sensitivity of 72% and a specificity of 67%. Accuracy is defined as the proportion of correct outcomes of a method, sensitivity is the proportion of all positive chemicals that are correctly classified as positives, and specificity is the proportion of all negative chemicals that are correctly classified as negatives. Thus, there is a clear need to develop more reliable test methods for sensitization. Additionally, the 7th Amendment to the Cosmetics Directive (76/768/EEC) poses a complete ban on using animal experimentation for testing cosmetic ingredients by 2013 when a scientific reliable method is available. Thus, there is a great need from the industry for reliable predictive test methods that are based on human cells. Various human cell lines and primary cells involved in sensitization have been evaluated as predictive test system, such as epithelial cells, dendritic cells and T cells, however no validated test assay is currently available. Various single biomarkers have been suggested to be upregulated upon stimulation with sensitizing chemicals, such as CD40, CD80, CD54, CXCL8, IL-1β, MIP-1β, p38 MAPK as reviewed in [18], yet singlehanded, none of them have the predictive power to discriminate between sensitizing and non-sensitizing chemicals. CD86 is among the markers most extensively studied; however, determining the expression level of this marker in our assay is relevant but not sufficient as readout for sensitization (FIG. 2). Only 9 out of 19 sensitizing chemicals induced a significant upregulation of CD86. Instead, it is our firm belief that the analyses of biomarker signatures are superior to any single biomarker. We therefore utilized the power of complete-genome transcriptomics and screened the gene regulation induced by a large set of well-defined chemicals and controls. The large number of differentially expressed genes in MUTZ-3 cells stimulated with sensitizing chemicals as compared to non-sensitizing controls, as identified with ANOVA (FIG. 3) revealed that MUTZ-3 indeed has a capacity to differentiate between these two groups, thus being a suitable cell system for in vitro assays for sensitization. Efforts have been made to create assays based on full genome analysis in various cell systems, such as CD34.sup.+-progenitor cells-derived DCs [19-21]. While such assays might provide in vivo like environments, one could argue that primary cells are not well suited for an assay of this great commercial interest. Moreover, the ethical aspect needs to be considered in such a system. Furthermore, previous efforts within in vitro assay development for sensitization that rely on full genome analysis have used a very limited set of testing compounds. To date, this study is the largest study performed within this area, utilizing 20 positive and 20 negative training compounds. Efforts have been made by us to divide these training compounds into two subsets, for training and testing respectively. While these experiments have resulted in successful predictions (data not shown), it is our experience that sensitizing compounds differ greatly in their induced gene expression profile, as can be seen in FIG. 3D. In this perspective, we wished to include as many training compounds as possible when identifying our “Prediction Signature”. Thus, we did not exclude any compounds for validation.

[0173] Instead, we have validated the method by which the “Prediction Signature” was identified, by subdividing the samples into training and test sets at random, using unseen data for validation, as seen in FIG. 5. By including all sensitizing compounds, with a wide range of reactive mechanisms as well as sensitizing potencies, we argue that we have identified a prediction signature that is well suited for predicting sensitizing properties of unknown samples. Of important note, our “Prediction Signature” is able to predict the potency of sensitizing compounds, such as those defined by the LLNA, as is demonstrated in FIG. 4C. However, the potency predicted by LLNA and that of our classifier do not match for all samples. Notably, the moderate sensitizer 2-Hydroxyethyl acrylate show strong resemblance of strong and extreme sensitizers with respect to their gene expression profile. Similarly, the moderate sensitizers Ethylendiamine, Hexylcinnamic aldehyde, and Glyoxal group together with weak sensitizers. These findings provide vast implications that sensitizing potency, as defined, may need revising. We argue that our “Prediction Signature” may be used for such classifications.

[0174] In conclusion, we present an in vitro assay, based on a MUTZ-3 cell system that with an identified “Prediction Signature” consisting of 200 genes, which have the ability to correctly classify a sample as sensitizer or non-sensitizer. In addition, this assay can predict the potency of sensitizing compounds, and may be used to revise such classifications.

[0175] Materials and Methods

[0176] Chemicals

[0177] A panel of chemicals consisting of 20 sensitizers and 20 non-sensitizers were used for cell stimulations. The sensitizers were 2,4-Dinitrochlorobenzene, Cinnamaldehyde, Resorcinol, Oxazolone, Glyoxal, 2-Mercaptobenzothiazole, Eugenol, lsoeugenol, Cinnamic alcohol and p-Phenylendiamine, Formaldehyde, Ethylendiamine, 2-Hydroxyethyl acrylate, Hexylcinnamic aldehyde, Potassium Dichromate, Penicillin G, Katchon CG (MCI/Ml), 2-aminophenol, Geraniol and 2-nitro-1,4-Phenylendiamine (Table 1). The non-sensitizers were Sodium dodecyl sulphate, Salicylic acid, Phenol, Glycerol, Lactic acid, Chlorobenzene, p-Hydrobenzoic acid, Benzaldehyde, Diethyl Phtalate and Octanoic acid, Zinc sulphate, 4-Aminobenzoic acid, Methyl salicylate, Ethyl vanillin, Isopropanol, Dimethyl formamide, 1-Butanol, Potassium permanganate, Propylene glycol and Tween 80 (Table 1). All chemicals were from Sigma-Aldrich, St. Louis, Mo., USA. Compounds were dissolved in either Dimethyl sulfoxide (DMSO) or distilled water. Prior to stimulations, the cytotoxicity of all compounds were monitored, using Propidium Iodide (PI) (BD Biosciences, San Diego, Calif.) using protocol provided by the manufacturer. The relative viability of stimulated cells was calculated as

[00001]$\mathrm{Relative}.Math..Math.\mathrm{viability}=\frac{\mathrm{fraction}.Math..Math.\mathrm{of}.Math..Math.\mathrm{viable}.Math..Math.\mathrm{stimulated}.Math..Math.\mathrm{cells}}{\mathrm{fraction}.Math..Math.\mathrm{of}.Math..Math.\mathrm{viable}.Math..Math.\mathrm{unstimulated}.Math..Math.\mathrm{cells}}.Math.100$

[0178] For toxic compounds, the concentration yielding 90% relative viability (Rv90) was used. For untoxic compounds, a concentration of 500 μM was used when possible. For non-toxic compounds that were insoluble at 500 μM in medium, the highest soluble concentration was used. For compounds dissolved in DMSO, the in-well concentration was 0.1% DMSO. The vehicle and concentrations used for each compound are listed in Table 2.

[0179] Chemical Exposure of the Cells

[0180] The human myeloid leukaemia-derived cell line MUTZ-3 (DSMZ, Braunschweig, Germany) was maintained in a-MEM (Thermo Scientific Hyclone, Logan, Utah) supplemented with 20% (volume/volume) fetal calf serum (Invitrogen, Carlsbad, Calif.) and 40 ng/ml rhGM-CSF (Bayer HealthCare Pharmaceuticals, Seattle, Wash.), as described [10]. Prior to each experiment, the cells were immunophenotyped using flow cytometry as a quality control. Cells were seeded in 6-well plates at 200.000 cells/ml. Stock solutions of each compound was prepared in either DMSO or distilled water, and were subsequently diluted so the in-well concentrations corresponded to the Rv90 value, and in-well concentrations of DMSO were 0.1%. Cells were incubated for 24h at 37° C. and 5% CO.sub.2. Thereafter, cells were harvested and analysed with flow cytometry. In parallel, harvested cells were lysed in TRIzol reagent (Invitrogen) and stored at −20° C. until RNA extraction. Stimulations with chemicals were performed in three individual experiments, so that triplicates samples were obtained.

[0181] Phenotypic Analysis with Flow Cytometry

[0182] All cell surface staining and washing steps was performed in PBS containing 1% BSA (w/v). Cells were incubated with specific mouse mAbs for 15 min at 4° C. The following mAbs were used for flow cytometry: FITC-conjugated CD1a (DakoCytomation, Glostrup, Denmark), CD34, CD86, and HLA-DR (BD Biosciences), PE-conjugated CD14 (DakoCytomation), CD54 and CD80 (BD Biosciences). Mouse IgG1, conjugated to FITC or PE were used as isotype controls and PI was used to assess cell viability (BD Biosciences). FACSDiva software was used for data acquisition with FACSCanto II instrument (BD Bioscience). 10,000 events were acquired and gates were set based on light scatter properties to exclude debris and nonviable cells. Further data analysis was performed using FCS Express V3 (De Novo Software, Los Angeles, Calif.).

[0183] Preparation of cRNA and Gene Chip Hybridization

[0184] RNA isolation and gene chip hybridization was performed as described [22]. Briefly, RNA from unstimulated and chemical-stimulated MUTZ-3 cells, from triplicate experiments, were extracted and analysed. The preparation of labeled sense DNA was performed according to Affymetrix GeneChip Whole Transcript (WT) Sense Target Labeling Assay (100 ng Total RNA Labeling Protocol) using the recommended kits and controls (Affymetrix, Santa Clara, Calif.). Hybridization, washing and scanning of the Human Gene 1.0 ST Arrays were performed according to the manufacturer's protocol (Affymetrix).

[0185] Microarray Data Analysis and Statistical Methods

[0186] The microarray data were normalised and quality checked with the RMA algorithm, using Affymetrix Expression Console (Affymetrix). Genes that were significantly regulated when comparing sensitizers with non-sensitizers were identified using one-way ANOVA, with false discovery rate (FDR) as a correction for multiple hypothesis testing. In order to reduce the large number of identified significant gene, we applied an algorithm developed in-house for Backward Elimination of analytes (Carlsson et al, unpublished). Wth this method, we train and test a Support Vector Machine (SVM) model [12] with leave-one out cross-validation, with one analyte left out. This process is iterated until each analyte has been left out once. For each iteration, a Kullback-Leibler divergence (KLD) is recorded, yielding N KLDs, where N is the number of analytes. The analyte that was left out when the smallest KLD was observed is considered to provide the least information in the data set. Thus, this analyte is eliminated and the iterations proceed, this time with N−1 analytes. In this manner, the analytes are eliminated one by one until a panel of markers remain that have been selected based on each analyte's ability to discriminate between sensitizers and non-sensitizers. The script for Backwards Eliminations was programmed for R [23], with the additional package e1071 [24]. ANOVA analyses and visualisation of results were performed in Qlucore Omics Explorer 2.1 (Qlucore, Lund, Sweden). The selected biomarker profile of 200 transcripts were designated the “Prediction Signature”.

[0187] Validation of the Method for Identification of the “Prediction Signature”

[0188] In the absence of an external test data set, the data set was divided into a training set of 70% and a test set of 30% of the samples. The division was performed randomly, while maintaining the proportions of sensitizers and non-sensitizers in each subset at the same ratio as in the complete data set. A biomarker signature was identified in the training set using ANOVA filtering and Backward Elimination, as described above. This test signature was used to train an SVM, using the training set, which was thereafter applied to predict the samples of the test set. The distribution of the area under the Receiver Operating Characteristic (ROC) curve [25] was used as a measurement of the performance of the model.

[0189] Assessment of Biological Functions of “Prediction Signature” Using Pathway Analysis

[0190] In order to investigate the biological functions the gene profile was analyzed using the Ingenuity Pathway Analysis software, IPA, (Ingenuity Systems, Inc. Mountain View, USA). The 200 top genes resulting from Backward Elimination were analyzed using the ‘build’ and ‘Path Explorer’ functions to build an interactome of the core genes from the “Prediction Signature” and connecting molecules suggested by IPA. The 200 molecules in the “Prediction Signature” were connected using the shortest known paths. In this process only human evidence from primary cells, cell lines and epidermal tissue was used. All molecules except for endogenous and chemical drugs were allowed in the network and all kinds of connections were allowed. Known ‘Functions’ and ‘Canonical Pathways’ from IPA were mapped to the interactome using the ‘Overlay’ function.

Abbreviations

[0191] ACD, atopic contact dermatitis;
AML, acute myeloid leukemia cell;

APC, Antigen Presenting Cell;

DC, Dendritic Cell;

[0192] GM-CSF, Granulocyte macrophage colony-stimulating factor;
GPMT, Guinea pig maximization test;

HMT, Human Maximation Test;

H PTA, Human Patch Test Allergen;

IL, Interleukin;

LLNA, Local Lymph Node Assay;

PCA, Principal Component Analysis

REFERENCES

[0193] 1. Akhavan, A. and S. R. Cohen, The relationship between atopic dermatitis and contact dermatitis. Clin Dermatol, 2003. 21(2): p. 158-62. [0194] 2. Mortz, C. G., et al., Prevalence of atopic dermatitis, asthma, allergic rhinitis, and hand and contact dermatitis in adolescents. The Odense Adolescence Cohort Study on Atopic Diseases and Dermatitis. Br J Dermatol, 2001. 144(3): p. 523-32. [0195] 3. Nielsen, N. H., et al., Allergic contact sensitization in an adult Danish population: two cross-sectional surveys eight years apart (the Copenhagen Allergy Study). Acta Derm Venereol, 2001. 81(1): p. 31-4. [0196] 4. Fonacier, L. S., S. C. Dreskin, and D. Y. Leung, Allergic skin diseases. J Allergy Clin Immunol. 125(2 Suppl 2): p. S138-49. [0197] 5. EC 1907/2006, in Regulation (EC) No 1907/2006 of the European Parliament and of the Council of 18 Dec. 2006 concerning the Registration, Evaluation, Authorisation and Restriction of Chemicals (REACH), establishing a European Chemicals Agency, amending Directive 1999/45/EC and repealing Council Regulation (EEC) No 793/93 and Commission Regulation (EC) No 1488/94 as well as Council Directive 76/769/EEC and Commission Directives 91/155/EEC, 93/67/EEC, 93/105/EC and 2000/21/EC. [0198] 6. Basketter, D. A., et al., Local lymph node assay—validation, conduct and use in practice. Food Chem Toxicol, 2002. 40(5): p. 593-8. [0199] 7. Magnusson, B. and A. M. Kligman, The identification of contact allergens by animal assay. The guinea pig maximization test. J Invest Dermatol, 1969. 52(3): p. 268-76. [0200] 8. Masterson, A. J., et al., MUTZ-3, a human cell line model for the cytokine-induced differentiation of dendritic cells from CD34+ precursors. Blood, 2002. 100(2): p. 701-3. [0201] 9. Santegoets, S. J., et al., A CD34(+) human cell line model of myeloid dendritic cell differentiation: evidence for a CD14(+)CD11b(+) Langerhans cell precursor. J Leukoc Biol, 2006. 80(6): p. 1337-44. [0202] 10. Larsson, K., M. Lindstedt, and C. A. Borrebaeck, Functional and transcriptional profiling of MUTZ-3, a myeloid cell line acting as a model for dendritic cells. Immunology, 2006. 117(2): p. 156-66. [0203] 11. Rasaiyaah, J., et al., Dendritic cells and myeloid leukaemias: plasticity and commitment in cell differentiation. Br J Haematol, 2007. 138(3): p. 281-90. [0204] 12. Cortes, C. and V. Vapnik, Support-Vector Networks. Machine Learning, 1995. 20(3): p. 273-297. [0205] 13. Kimber, I., et al., Allergic contact dermatitis. Int Immunopharmacol, 2002. 2(2-3): p. 201-11. [0206] 14. Gerberick, G. F., et al., Compilation of historical local lymph node data for evaluation of skin sensitization alternative methods. Dermatitis, 2005. 16(4): p. 157-202. [0207] 15. Kimber, I., et al., Classification of contact allergens according to potency: proposals. Food Chem Toxicol, 2003. 41(12): p. 1799-809. [0208] 16. Kligman, A. M., The identification of contact allergens by human assay. 3. The maximization test: a procedure for screening and rating contact sensitizers. J Invest Dermatol, 1966. 47(5): p. 393-409. [0209] 17. Haneke, K. E., et al., ICCVAM evaluation of the murine local lymph node assay. Data analyses completed by the National Toxicology Program Interagency Center for the Evaluation of Alternative Toxicological Methods. Regul Toxicol Pharmacol, 2001. 34(3): p. 274-86. [0210] 18. dos Santos, G. G., et al., Progress on the development of human in vitro dendritic cell based assays for assessment of the sensitizing potential of a compound. Toxicol Appl Pharmacol, 2009. 236(3): p. 372-82. [0211] 19. Hooyberghs, J., et al., A cell-based in vitro alternative to identify skin sensitizers by gene expression. Toxicol Appl Pharmacol, 2008. 231(1): p. 103-11. [0212] 20. Lambrechts, N., et al., Gene markers in dendritic cells unravel pieces of the skin sensitization puzzle. Toxicol Lett, 2010. 196(2): p. 95-103. [0213] 21. Lambrechts, N., et al., Assessment of chemical skin-sensitizing potency by an in vitro assay based on human dendritic cells. Toxicol Sci, 2010. 116(1): p. 122-9. [0214] 22. Lindstedt, M., et al., Genomic and functional delineation of dendritic cells and memory T cells derived from grass pollen-allergic patients and healthy individuals. Int Immunol, 2005. 17(4): p. 401-9. [0215] 23. R Development Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing. 2008; Available from: http://www.R-project.org [0216] 24. R package e1071. Available from: [0217] http://cran.r-project.org/web/packages/e1071/index.html. [0218] 25. Lasko, T. A., et al., The use of receiver operating characteristic curves in biomedical informatics. J Biomed Inform, 2005. 38(5): p. 404-15. [0219] 26. Warbrick, E. V., et al., Influence of application vehicle on skin sensitization to methylchloroisothiazolinone/methylisothiazolinone: an analysis using the local lymph node assay. Contact Dermatitis, 1999. 41(6): p. 325-9. [0220] 27. Roberts, D. W., et al., Mechanistic applicability domain classification of a local lymph node assay dataset for skin sensitization. Chem Res Toxicol, 2007. 20(7): p. 1019-30. [0221] 28. Gerberick, G. F., et al., Contact allergenic potency: correlation of human and local lymph node assay data. Am J Contact Dermat, 2001. 12(3): p. 156-61. [0222] 29. Basketter, D. A., D. Sanders, and I. R. Jowsey, The skin sensitization potential of resorcinol: experience with the local lymph node assay. Contact Dermatitis, 2007. 56(4): p. 196-200. [0223] 30. Ryan, C. A., et al., Activity of human contact allergens in the murine local lymph node assay. Contact Dermatitis, 2000. 43(2): p. 95-102. [0224] 31. APPENDIX C; Comparative LLNA: BrdU-FC, Traditional LLNA, Guinea Pig Skin Sensitization, and Human Data. 2008 Available from: http://iccvam.niehs.nih.gov/methods/immunotox/fcLLNA/Appx/AppendixC_LL NA_FC07Jan08FD.pdf. [0225] 32. Patlewicz, G., et al., Skin-sensitization structure-activity relationships for aldehydes. Contact Dermatitis, 2001. 44(6): p. 331-6. [0226] 33. Basketter, D. A., G. F. Gerberick, and I. Kimber, Strategies for identifying false positive responses in predictive skin sensitization tests. Food Chem Toxicol, 1998. 36(4): p. 327-33. [0227] 34. Ashby, J., et al., Structure activity relationships in skin sensitization using the murine local lymph node assay. Toxicology, 1995. 103(3): p. 177-94. [0228] 35. Sakaguchi, H., et al., Development of an in vitro skin sensitization test using human cell lines; human Cell Line Activation Test (h-CLAT). II. An inter-laboratory study of the h-CLAT. Toxicol In Vitro, 2006. 20(5): p. 774-84. [0229] 36. The Murine Local Lymph Node Assay: A Test Method for Assessing the Allergic Contact Dermatitis Potential of Chemicals/Compounds. 1999; Available from: http://iccvam.niehs.nih.gov/docs/immunotox_docs/IIna/IInarep.pdf.

[0230] Tables

TABLE-US-00001 TABLE 1 List of sensitizing and non-sensitizing chemicals, based on murine LLNA classification, tested in the cell-based assay. Compound Abbreviation Potency LLNA HMT.sup.1 HPTA.sup.1 Sensitizers 2,4-Dinitrochlorobenzene DNCB Extreme [15] +[15] Oxazolone OXA Extreme [15] +[15] Potassium dichromate PD Extreme [14] +[14] + + Kathon CH (MC/MCI) KCG Extreme [14, 26] +[14, 26] Formaldehyde FA Strong [15] +[15] + + 2-Aminophenol 2AP Strong [27] +[27] 2-nitro-1,4-Phenylendiamine NPDA Strong [27] +[27] p-Phenylendiamine PPD Strong [28] +[28] + + Hexylcinnamic aldehyde HCA Moderate [15] +[15] 2-Hydroxyethyl acrylate 2HA Moderate [27] +[27] + 2-Mercaptobenzothiazole MBT Moderate [27] +[27] + + Glyoxal GO Moderate [27] +[27] + Cinnamaldehyde CALD Moderate [28] +[28] + + Isoeugenol IEU Moderate [28] +[28] + Ethylendiamine EDA Moderate [14] +[14] Resorcinol RC Moderate [29] +[29] − + Cinnamic alcohol CALC Weak [27] +[28] Eugenol EU Weak [28] +[28] + Penicillin G PEN G Weak [28] +[28] + Geraniol GER Weak [14] +[14] − + Non-sensitizers 1-Butanol BUT −[30] 4-Aminobenzoic acid PABA −[31] − + Benzaldehyde BA −[32] Chlorobenzene CB −[14] Diethyl phthalate DP −[28] Dimethyl formamide DF −[26] Ethyl vanillin EV −[32] Glycerol GLY −[28] Isopropanol IP −[28] Lactic acid LA −[14] Methyl salicylate MS −[14] − Octanoic acid OA −[33] Propylene glycol PG −[31] Phenol PHE −[33] − p-Hydroxybenzoic acid HBA −[34] Potassium permanganate PP − Salicylic acid SA −[14] − Sodium dodecyl sulphate SDS +.sup.2[14, 33] − Tween 80 T80 −[35] + Zinc sulphate ZS +.sup.2[36] .sup.1HMT, Human Maximation Test; HPTA, Human Patch Test Allergen. Information is derived from [17]. .sup.2False positives in LLNA.

TABLE-US-00002 TABLE 2 Vehicle and concentrations used for testing. Max Concentration solubility Rv90 In culture Compound Abbreviation Vehicle (μM) (μM) (μM) Sensitizers 2,4-Dinitrochlorobenzene DNCB DMSO —  4 4 Oxazolone OXA DMSO 250 — 250 Potassium dichromate PD Water    51.02    1.5 1.5 Kathon CG (MC/MCI)* KCG Water — 0.0035% 0.0035% Formaldehyde FA Water —  80 80 2-Aminophenol 2AP DMSO — 100 100 2-nitro-1,4-Phenylendiamine NPDA DMSO — 300 300 p-Phenylendiamine PPD DMSO 566  75 75 Hexylcinnamic aldehyde HCA DMSO    32.34 — 32.24 2-Hydroxyethyl acrylate 2HA Water — 100 100 2-Mercaptobenzothiazole MBT DMSO 250 — 250 Glyoxal GO Water — 300 300 Cinnamaldehyde CALD Water — 120 120 Isoeugenol IEU DMSO 641 300 300 Ethylendiamine EDA Water — — 500 Resorcinol RC Water — — 500 Cinnamic alcohol CALC DMSO 500 — 500 Eugenol EU DMSO 649 300 300 Penicillin G PEN G Water — — 500 Geraniol GER DMSO — — 500 Non-sensitizers 1-Butanol BUT DMSO — — 500 4-Aminobenzoic acid PABA DMSO — — 500 Benzaldehyde BA DMSO 250 — 250 Chlorobenzene CB DMSO  98 — 98 Diethyl phthalate DP DMSO  50 — 50 Dimethyl formamide DF Water — — 500 Ethyl vanillin EV DMSO — — 500 Glycerol GLY Water — — 500 Isopropanol IP Water — — 500 Lactic acid LA Water — — 500 Methyl salicylate MS DMSO — — 500 Octanoic acid OA DMSO 504 — 500 Peopylene glycol PG Water — — 500 Phenol PHE Water — — 500 p-Hydroxybenzoic acid HBA DMSO 250 — 250 Potassium permanganate PP Water  38 — 38 Salicylic acid SA DMSO — — 500 Sodium dodecyl sulphate SDS Water — 200 200 Tween 80 T80 DMSO — — 500 Zinc sulphate ZS Water 126 — 126 .sup.1 Kathon CG is a mixture of the compounds MC and MCI. The concentration of this mixture is given in %.

TABLE-US-00003 TABLE 3 Differentially expressed genes in MUTZ-3 cells stimulated with sensitizing chemicals as compared to non-sensitizing agents and controls. NCBI reference Gene Title Gene Symbol sequence Table 3A fatty acid synthase FASN NM_004104 squalene epoxidase SQLE NM_003129 taste receptor, type 2, member 5 TAS2R5 NM_018980 keratinocyte growth factor-like protein 1/2/hypothetical protein KGFLP1/2/FLJ20444 AF523265 FLJ20444 transmembrane anterior posterior transformation 1 TAPT1 NM_153365 Sprouty homolog 2 SPRY2 NM_005842 B-cell CLL/lymphoma 7A BCL7A NM_020993 solute carrier family 25, member 32 SLC25A32 NM_030780 ferritin, heavy polypeptide pseudogene 1 FTHP1 GENSCAN00000008165 ATPase, H+ transporting, lysosomal 50/57 kDa, V1 subunit H ATP6V1H NM_015941 Histone cluster 1, H1e HIST1H1E NM_005321 Table 3B 4-aminobutyrate aminotransferase ABAT NM_020686 abhydrolase domain containing 5 ABHD5 NM_016006 alkaline ceramidase 2 ACER2 NM_001010887 ATP citrate lyase ACLY NM_001096 actin-related protein 10 homolog ACTR10 NM_018477 ADAM metallopeptidase domain 20 ADAM20 NM_003814 Retrotransposed pseudogene AL391261.2-201 AL391261.2-201 GENSCAN00000063078 aldehyde dehydrogenase 18 family, member A1 ALDH18A1 NM_002860 aldehyde dehydrogenase 1 family, member B1 ALDH1B1 NM_000692 alkB, alkylation repair homolog 6 (E. coli) ALKBH6 NM_032878 anaphase promoting complex subunit 1 ANAPC1 NM_022662 anaphase promoting complex subunit 5 ANAPC5 NM_016237 ankyrin repeat, family A (RFXANK-like), 2 ANKRA2 NM_023039 ADP-ribosylation factor GTPase activating protein 3 ARFGAP3 NM_014570 Rho GTPase activating protein 9 ARHGAP9 NM_032496 ankyrin repeat and SOCS box-containing 7 ASB7 NM_198243 ATPase, H+ transporting, lysosomal 9 kDa, V0 subunit e1 ATP6V0E1 NM_003945 bridging integrator 2 BIN2 NM_016293 bleomycin hydrolase BLMH NM_000386 brix domain containing 1/ribosome production factor 2 BXDC1/RPF2 ENST00000368864 homolog chromosome 11 open reading frame 61 C11orf61 NM_024631 chromosome 11 open reading frame 67 C11orf67 NM_024684 chromosome 12 open reading frame 57 C12orf57 NM_138425 chromosome 13 open reading frame 18 C13orf18 NM_025113 chromosome 15 open reading frame 24 C15orf24 NM_020154 chromosome 19 open reading frame 54 C19orf54 NM_198476 chromosome 1 open reading frame 174 C1orf174 NM_207356 chromosome 1 open reading frame 183 C1orf183 NM_019099 chromosome 20 open reading frame 111 C20orf111 NM_016470 chromosome 20 open reading frame 24 C20orf24 BC004446 chromosome 3 open reading frame 62/ubiquitin specific C3orf62/USP4 BC023586 peptidase 4 (proto-oncogene) chromosome 9 open reading frame 89 C9orf89 BC038856 coactivator-associated arginine methyltransferase 1 CARM1 NM_199141 CD33 molecule CD33 NM_001772 CD86 molecule CD86 NM_175862 CD93 molecule CD93 NM_012072 cytochrome c oxidase subunit VIIa polypeptide 2 like COX7A2L NM_004718 corticotropin releasing hormone binding protein CRHBP NM_001882 chondroitin sulfate N-acetylgalactosaminyltransferase 2 CSGALNACT2 NM_018590 Cytochrome P450 51A1 CYP51A1 NM_000786.2 DDRGK domain containing 1 DDRGK1 NM_023935 DEAD (Asp-Glu-Ala-Asp) box polypeptide 21 DDX21 NM_004728 24-dehydrocholesterol reductase DHCR24 NM_014762 7-dehydrocholesterol reductase DHCR7 NM_001360 DEAH (Asp-Glu-Ala-His) box polypeptide 33 DHX33 NM_020162 DnaJ (Hsp40) homolog, subfamily B, member 4 DNAJB4 NM_007034 DnaJ (Hsp40) homolog, subfamily B, member 9 DNAJB9 NM_012328 DnaJ (Hsp40) homolog, subfamily C, member 5 DNAJC5 NM_025219 DnaJ (Hsp40) homolog, subfamily C, member 9 DNAJC9 NM_015190 D-tyrosyl-tRNA deacylase 1 homolog DTD1 NM_080820 ER degradation enhancer, mannosidase alpha-like 2 EDEM2 NM_018217 ecotropic viral integration site 2B EVI2B NM_006495 family with sequence similarity 36, member A FAM36A NM_198076 family with sequence similarity 86, member A FAM86A NM_201400 Fas (TNF receptor superfamily, member 6) FAS NM_000043 MGC44478 FDPSL2A NR_003262 ferredoxin reductase FDXR NM_024417 forkhead box O4 FOXO4 NM_005938 FTHL10-001, Transcribed processed pseudogene FTHL10-001 NR_002200 fucosidase, alpha-L- 2, plasma FUCA2 NM_032020 growth arrest-specific 2 like 3 GAS2L3 NM_174942 ganglioside induced differentiation associated protein 2 GDAP2 NM_017686 growth differentiation factor 11 GDF11 NM_005811 glutaredoxin (thioltransferase) GLRX NM_002064 guanine nucleotide binding protein-like 3 GNL3L NM_019067 glucosamine-phosphate N-acetyltransferase 1 GNPNAT1 NM_198066 glutathione reductase GSR NM_000637 GTF2I repeat domain containing 2B/2/2 pseudogene GTF2IRD2B/2/2P BC067859 general transcription factor IIIC, polypeptide 2 beta GTF3C2 NM_001521 HMG-box transcription factor 1 HBP1 NM_012257 histone cluster 1, H1c HIST1H1C NM_005319 histone cluster 1, H2ae HIST1H2AE NM_021052 histone cluster 1, H2be HIST1H2BE NM_003523 histone clusters 1, H2bm/2, H3, pseudogene 2/2, H2b/a HIST1H2BM/HIST2H3PS2/BF/A NM_001024599 histone cluster 1, H3g HIST1H3G NM_003534 histone cluster 1, H3j HIST1H3J NM_003535 histone cluster 1, H4a HIST1H4A NM_003538 histone clusters 2, H2aa3/2, H2aa4 HIST2H2AA3/4 NM_003516 high-mobility group box 3 HMGB3 NM_005342 3-hydroxy-3-methylglutaryl-Coenzyme A reductase HMGCR NM_000859 3-hydroxy-3-methylglutaryl-Coenzyme A synthase 1 HMGCS1 NM_001098272 heme oxygenase (decycling) 1 HMOX1 NM_002133 heterogeneous nuclear ribonucleoprotein L HNRNPL NM_001533 insulin receptor substrate 2 IRS2 NM_003749 iron-sulfur cluster scaffold homolog ISCU NM_014301 interferon stimulated exonuclease gene 20 kDa-like 2 ISG20L2 NM_030980 potassium voltage-gated channel, Isk-related family, member 3 KCNE3 NM_005472 hypothetical protein LOC100132855/ATPase, H+ transporting, LOC100132855/ATP6V0D1 NM_004691 lysosomal 38 kDa, V0 subunit d1 hCG1651476 LOC284417 NM_001085488 golgi autoantigen, golgin subfamily a, 6 pseudogene/ LOC729668/MTPAP NM_018109 mitochondrial poly(A) polymerase lysophosphatidic acid receptor 1 LPAR1 NM_057159 leucine-rich PPR-motif containing LRPPRC NM_133259 lymphocyte antigen 96 LY96 NM_015364 mitogen-activated protein kinase kinase 1 MAP2K1 NM_002755 mitogen-activated protein kinase 13 MAPK13 NM_002754 methyltransferase like 2A METTL2A NM_181725 Brain cDNA clone: similar to human METTL2 METTL2B NM_018396.1 Methyltransferase like 2B METTL2B NM_018396.2 microsomal glutathione S-transferase 3 MGST3 NM_004528 mitochondrial ribosomal protein L30 MRPL30 NM_145212 mitochondrial ribosomal protein L4 MRPL4 NM_146388 mitochondrial ribosomal protein S17 MRPS17 NM_015969 5-methyltetrahydrofolate-homocysteine methyltransferase MTR NM_000254 MYB binding protein (P160) 1a MYBBP1A NM_014520 neighbor of BRCA1 gene 1 NBR1 NM_031858 nuclear import 7 homolog NIP7 NM_016101 NLR family, pyrin domain containing 12 NLRP12 NM_144687 nucleolar protein family 6 (RNA-associated) NOL6 NM_022917 NAD(P)H dehydrogenase, quinone 1 NQO1 NM_000903 nuclear receptor binding protein 1 NRBP1 NM_013392 nucleotide binding protein-like NUBPL NM_025152 nudix (nucleoside diphosphate linked moiety X)-type motif 14 NUDT14 NM_177533 nuclear fragile X mental retardation protein interacting protein 1 NUFIP1 NM_012345 nucleoporin 153 kDa NUP153 NM_005124 olfactory receptor, family 5, subfamily B, member 21 OR5B21 NM_001005218 PAS domain containing serine/threonine kinase PASK NM_015148 PRKC, apoptosis, WT1, regulator PAWR NM_002583 PDGFA associated protein 1 PDAP1 NM_014891 phosphodiesterase 1B, calmodulin-dependent PDE1B NM_000924 phosphoribosylformylglycinamidine synthase PFAS NM_012393 pleckstrin homology-like domain, family A, member 3 PHLDA3 NM_012396 phosphoinositide-3-kinase adaptor protein 1 PIK3AP1 NM_152309 PTEN induced putative kinase 1 PINK1 NM_032409 phosphomannomutase 2 PMM2 NM_000303 partner of NOB1 homolog PNO1 NM_020143 polymerase (RNA) II (DNA directed) polypeptide E, 25 kDa POLR2E NM_002695 polymerase (RNA) III (DNA directed) polypeptide E (80 kD) POLR3E NM_018119 protein phosphatase 1D magnesium-dependent, delta isoform PPM1D BC042418 phosphatidylinositol-3,4,5-trisphosphate-dependent Rac PREX1 NM_020820 exchange factor 1 proline-serine-threonine phosphatase interacting protein 1 PSTPIP1 NM_003978 prothymosin, alpha PTMA NM_016184/024809 RAB33B, member RAS oncogene family RAB33B NM_031296 renin binding protein RENBP NM_002910 replication factor C (activator 1) 2, 40 kDa RFC2 NM_181471 ribonuclease H1 RNASEH1 NM_002936 ring finger protein 146 RNF146 NM_030963 ring finger protein 24 RNF24 NM_007219 ring finger protein 26 RNF26 NM_032015 Havana pseudogene RP1-274L14.2-001 RP1-274L14.2-001 NM_032020 ribosomal protein SA/small nucleolar RNA, H/ACA box 62 RPSA/SNORA62 NM_014570 RNA pseudouridylate synthase domain containing 2 RPUSD2 NM_152260 ribosomal RNA processing 12 homolog RRP12 NM_015179 retinoid X receptor, alpha RXRA NM_002957 scavenger receptor class B, member 2 SCARB2 NM_005506 SERPINE1 mRNA binding protein 1 SERBP1 NM_001018067 splicing factor proline/glutamine-rich SFPQ NM_005066 solute carrier family 35, member B3 SLC35B3 BX538271 solute carrier family 37, member 4 SLC37A4 NM_001467 solute carrier family 5, member 6 SLC5A6 NM_021095 sphingomyelin phosphodiesterase 4, neutral membrane SMPD4 NM_017751 small nucleolar(sn)RNA host gene 1, non-coding/snRNA C/D SNHG1/SNORD26 NM_002032 box 26 small nucleolar RNA host gene 12 (non-coding) SNHG12 NM_207356 small nucleolar RNA, H/ACA box 45 SNORA45 NR_002977 SnRNA SnRNA Affymetrix: 7966223 sorting nexin family member 27 SNX27 NM_030918 sterol regulatory element binding transcription factor 2 SREBF2 NM_004599 RRNA SSU_rRNA_5 SSU_rRNA_5 ENST00000386723 ST3 beta-galactoside alpha-2,3-sialyltransferase 6 ST3GAL6 NM_006100 serine/threonine kinase 17b STK17B NM_004226 tubulin folding cofactor E-like TBCEL NM_152715 tectonic family member 2 TCTN2 NM_024809 toll-like receptor 6 TLR6 NM_006068 toll-like receptor 9/twinfilin homolog 2 TLR9/TWF2 NM_007284 transmembrane protein 55A TMEM55A NM_018710 transmembrane protein 59 TMEM59 NM_004872 transmembrane protein 77 TMEM77 BC091509 transmembrane protein 97 TMEM97 NM_014573 tumor necrosis factor receptor superfamily, member 10c TNFRSF10C NM_003841 translocase of outer mitochondrial membrane 34 TOMM34 NM_006809 translocase of outer mitochondrial membrane 40 homolog TOMM40 BC001779 translocase of outer mitochondrial membrane 5 homolog/F- TOMM5/FBXO10 NM_012166 box protein 10 tumor protein p53 inducible protein 3 TP53I3 NM_004881 tumor protein p53 inducible nuclear protein 1 TP53INP1 NM_033285 thioredoxin reductase 1 TXNRD1 NM_003330 ubiquitin-fold modifier conjugating enzyme 1 UFC1 NM_016406 ubiquitin specific peptidase 10 USP10 NM_005153 vesicle-associated membrane protein 3 (cellubrevin) VAMP3 NM_004781 valyl-tRNA synthetase VARS NM_006295 vacuolar protein sorting 37 homolog A VPS37A NM_152415 zinc finger protein 211 ZNF211 NM_006385 zinc finger protein 223 ZNF223 NM_013361 zinc finger protein 561 ZNF561 NM_152289 zinc finger protein 79 ZNF79 NM_007135

[0231] Table 3 Legend.

[0232] The table shows the profile genes found by t-test and Backward Elimination. Genes were annotated, using the NetAffx database from Affymetrix (www.affymetrix.com, Santa Clara USA). When found, the Unigene (www.ncbi.nlm.nih.gov/UniGene/) ID was chosen as the gene identifier. In the twelve cases where no Unigene ID was reported the best alternative ID was given. Gene names and IDs were checked against the IPA database where 189 of the 200 could be matched. In one instance only an Affymetrix ID was reported. 6 duplicate genes were removed.

TABLE-US-00004 TABLE 4 Dominating functions in the “Prediction signature”. 184 of the 200 molecules were investigated functionally using IPA. Only functions populated by 15 or more molecules were reported. Number of molecules from Most prominent Function signature Molecule names sub functions small molecule 38 ABHD5, ACLY, ALDH18A1, BLMH, CD86, Metabolism (23), biochemistry CSGALNACT2, CYP51A1, DHCR24, DHCR7, DNAJC5, biosynthesis (14), FAS, FASN, FDXR, GLRX, GNPNAT1, HMGCR, modification (12), HMOX1, IRS2, LPAR1, LY96, MGST3, MTR, NQO1, synthesis (10) PASK, PDE1B, PINK1, PMM2, RENBP, RXRA, SLC25A32, SLC37A4, SLC5A6, SMPD4, SQLE, SREBF2, ST3GAL6, TLR6, TMEM55A cell death 33 CD33, DDX19A, DHCR24, DNAJB9, DNAJC5, FAS, Apoptosis (30), FASN, FDXR, FOXO4, GLRX, GNPNAT1, GSR, cell death (13) HIST1H1C, HMGB3, HMOX1, IRS2, LPAR1, MAP2K1, MAPK13, NQO1, PAWR, PDE1B, PHLDA3, PINK1, PPM1D, RXRA, SERBP1, SPRY2, STK17B, TLR6, TNFRSF10C, TP53INP1, TXNRD1 lipid 24 ABHD5, ACLY, CYP51A1, DHCR24, DHCR7, FAS, Metabolism (17), metabolism FASN, FDXR, HMGCR, HMOX1, IRS2, LPAR1, LY96, modification (11), MGST3, PASK, RENBP, RXRA, SLC37A4, SMPD4, synthesis (10) SQLE, SREBF2, ST3GAL6, TLR6, TMEM55A hematological 19 CARM1, CD33, CD86, FAS, FOXO4, HIST1H1C, Proliferation (10), system HMGB3, HMGCR, HMOX1, IRS2, LY96, NBR1, NQO1, apoptosis (5) development PAWR, PIK3AP1, PPM1D, STK17B, TP53INP1, VAMP3 cellular growth 16 CARM1, CD33, CD86, FAS, FOXO4, HIST1H1C, Proliferation (16), and HMGB3, HMGCR, HMOX1, IRS2, LY96, NBR1, NQO1, growth (4) proliferation PAWR, PIK3AP1, PPM1D, STK17B, TP53INP1, VAMP3 molecular 15 CARM1, CD33, CD86, FAS, FOXO4, HIST1H1C, Accumulation (8), transport HMGB3, HMGCR, HMOX1, IRS2, LY96, NBR1, NQO1, quantity (5) PAWR, PIK3AP1, PPM1D, STK17B, TP53INP1, VAMP3 cell cycle 15 DNAJB4, DTD1, FAS, FASN, FOXO4, GDF11, HBP1, Cell cycle HMOX1, IRS2, MAP2K1, PAWR, PPM1D, SFPQ, progression (13), SPRY2, TP53INP1 cell division (5) carbohydrate 15 DNAJB4, DTD1, FAS, FASN, FOXO4, GDF11, HBP1, Metabolism (9), metabolism HMOX1, IRS2, MAP2K1, PAWR, PPM1D, SFPQ, biosynthesis (5) SPRY2, TP53INP1

TABLE-US-00005 TABLE 5 Support Vector Machine (SVM) algorithm filnamnTraining<−“training set.txt” filnamnTest<−“test set.txt” lista <− read.delim(“biomarker signature.txt”,header=FALSE) # EN FIL MED DE ANALYTER lista <− as.character(lista[[1]]) listaBoolean <− is.element(ProteinNames, lista) group1<− “pos” group2<− “neg” rawfile <− read.delim(filnamnTraining) samplenames <− as.character(rawfile[,1]) groupsTraining <− rawfile[,2] dataTraining <− t(rawfile[,-c(1,2)]) ProteinNames <− read.delim(filnamnTraining,header=FALSE) ProteinNames <− as.character(as.matrix(ProteinNames)[1,]) ProteinNames <− ProteinNames[-(1:2)] listaBoolean <− is.element(ProteinNames, lista) rownames(dataTraining) <− ProteinNames colnames(dataTraining) <− samplenames logdataTraining <− dataTraining logdataTraining <− logdataTraining[listaBoolean,] rawfile <− read.delim(filnamnTest) samplenames <− as.character(rawfile[,1]) groupsTest <− rawfile[,2] dataTest <− t(rawfile[,-c(1,2)]) ProteinNames <− read.delim(filnamnTest,header=FALSE) ProteinNames <− as.character(as.matrix(ProteinNames)[1,]) ProteinNames <− ProteinNames[-(1:2)] rownames(dataTest) <− ProteinNames colnames(dataTest) <− samplenames logdataTest<−dataTest logdataTest <− logdataTest[listaBoolean,] svmfacTraining<− factor(rep(‘rest’,ncol(logdataTraining)),levels=c(group1, group2, ‘rest’)) subset1Training<− is.element(groupsTraining , strsplit(group1,“,”)[[1]]) subset2Training<− is.element(groupsTraining , strsplit(group2,“,”)[[1]]) svmfacTraining[subset1Training] <− group1 svmfacTraining[subset2Training] <− group2 facTraining <−factor(as.character(svmfacTraining [subset1Training|subset2Training]),levels=c(group1,group2)) svmfacTest<− factor(rep(‘rest’,ncol(logdataTest)),levels=c(group1, group2, ‘rest’)) subset1Test<− is.element(groupsTest , strsplit(group1,“,”)[[1]]) subset2Test<− is.element(groupsTest , strsplit(group2,“,”)[[1]]) svmfacTest[subset1Test] <− group1 svmfacTest[subset2Test] <− group2 facTest <−factor(as.character(svmfacTest [subset1Test|subset2Test]),levels=c(group1,group2))  n1 <− sum(facTest ==levels(facTest )[1])  n2 <− sum(facTest ==levels(facTest )[2])  nsamples <− n1+n2 SampleInformation <− paste(levels(facTest )[1],“ “,n1,” , “,levels(facTest )[2],” “,n2,sep=””) svmtrain <− svm(t(logdataTraining) , facTraining , kernel=“linear” ) pred<−predict(svmtrain , t(logdataTest) , decision.values=TRUE) res<−attr(pred, “decision.values”) names <− colnames(logdataTest, do.NULL=FALSE) orden <− order(res , decreasing=TRUE) Samples <− data.frame(names[orden],res[orden],facTest[orden]) ROCdata <− myROC(res,facTest) SenSpe <− SensitivitySpecificity(res,facTest) ROCplot(list(SampleInformation=SampleInformation,ROCarea=ROCdata[1],p .value=ROCdata[2],SenSpe <− SenSpe,samples=Samples), sensspecnumber=4) #rows in blue are needed only for ROC evaluation. #to assess an unknown sample, print res. #(vector with prediction values)