Short-time gene expression response to valproic acid and valproic acid analogs in mouse embryonic stem cells.

Prediction of developmental toxicity in vitro could be based on short-time toxicogenomic endpoints in embryo-derived cell lines. Microarray studies in P19 mouse embryocarcinoma cells and mouse embryos have indicated that valproic acid (VPA), an inducer of neural tube defects, deregulates the expression of many genes, including those critically involved in neural tube development. In this study, we exposed undifferentiated R1 mouse embryonic stem cells to VPA and VPA analogs for 6 h and used CodeLink whole-genome expression microarrays to define VPA-responsive genes correlating with teratogenicity. Compared with the nonteratogenic analog 2-ethyl-4-methylpentanoic acid, VPA and the teratogenic VPA analog (S)-2-pentyl-4-pentynoic acid deregulated a much larger number of genes. Five genes (of ∼2500 array probes correlating with the separation) were sufficient to effectively separate teratogens from nonteratogens. A large fraction of the target genes correlating with teratogenicity are functionally related to embryonic development and morphogenesis, including neural tube formation and closure. Similar responses in R1 were found for most genes previously identified as VPA responsive in P19 and embryos. A subset of target genes was evaluated as candidate markers predictive of potential teratogenicity against a range of known teratogens using TaqMan expression arrays. These marker genes showed a positive predictive value for the teratogens butyrate and trichostatin A, which like VPA and (S)-2-pentyl-4-pentynoic acid are known histone deacetylase (HDAC) inhibitors but not for compounds that are likely to act by other mechanisms. This indicates that HDAC inhibition may be a major mechanism by which VPA induces gene deregulation and possibly teratogenicity.

Early transcriptional responses in mouse embryos as a basis for selection of molecular markers predictive of valproic acid teratogenicity.

Cell-based in vitro assays would potentially reduce animal testing in preclinical drug development. Mouse embryos exposed to the teratogenic drug valproic acid (VPA) in utero for 1.5, 3 or 6h on gestational day 8 were analyzed using microarrays. Significant effects on gene expression were observed already at 1.5h, and 85 probes were deregulated across all time points. To find transcriptional markers of VPA-induced developmental toxicity, the in vivo data were compared to previous in vitro data on embryonal carcinoma P19 cells exposed to VPA for 1.5, 6 or 24h. Maximal concordance between embryos and cells was at the 6-h time points, with 163 genes showing similar deregulation. Developmentally important Gene Ontology terms, such as "organ morphogenesis" and "tube development" were overrepresented among putative VPA target genes. The genes Gja1, Hap1, Sall2, H1f0,Cyp26a1, Fgf15, Otx2, and Lin7b emerged as candidate in vitro markers of potential VPA-induced teratogenicity.

Valproic acid-induced deregulation in vitro of genes associated in vivo with neural tube defects.

The utility of an in vitro system to search for molecular targets and markers of developmental toxicity was explored, using microarrays to detect genes susceptible to deregulation by the teratogen valproic acid (VPA) in the pluripotent mouse embryonal carcinoma cell line P19. Total RNA extracted from P19 cells cultured in the absence or presence of 1, 2.5, or 10mM VPA for 1.5, 6, or 24 h was subjected to replicated microarray analysis, using CodeLink UniSet I Mouse 20K Expression Bioarrays. A moderated F-test revealed a significant VPA response for 2972 (p < 10(-3)) array probes (19.4% of the filtered gene list), 421 of which were significant across all time points. In a core subset of VPA target genes whose expression was downregulated (68 genes) or upregulated (125 genes) with high probability (p < 10(-7)) after both 1.5 and 6 h of VPA exposure, there was a significant enrichment of the biological process Gene Ontology term transcriptional regulation among downregulated genes, and apoptosis among upregulated, and two of the downregulated genes (Folr1 and Gtf2i) have a knockout phenotype comprising exencephaly, the major malformation induced by VPA in mice. The VPA-induced gene expression response in P19 cells indicated that approximately 30% of the approximately 200 genes known from genetic mouse models to be associated with neural tube defects may be potential VPA targets, suggestive of a combined deregulation of multiple genes as a possible mechanism of VPA teratogenicity. Gene expression responses related to other known effects of VPA (histone deacetylase inhibition, G(1)-phase cell cycle arrest, induction of apoptosis) were also identified. This study indicates that toxicogenomic responses to a teratogenic compound in vitro may correlate with known in vitro and in vivo effects, and that short-time (< or =6 h) exposures in such an in vitro system could provide a useful component in mechanistic studies and screening tests in developmental toxicology.

Molecular targets and early response biomarkers for the prediction of developmental toxicity in vitro.

There is an urgent need for new in vitro methods to predict the potential developmental toxicity of candidate drugs in the early lead identification and optimisation process. This would lead to a reduction in the total number of animals required in full-scale developmental toxicology studies, and would improve the efficiency of drug development. However, suitable in vitro systems permitting robust high-throughput screening for this purpose, for the most part, remain to be designed. An understanding of the mechanisms involved in developmental toxicity may be essential for the validation of in vitro tests. Early response biomarkers - even a single one - could contribute to reducing assay time and facilitating automation. The use of toxicogenomics approaches to study in vitro and in vivo models in parallel may be a powerful tool in defining such mechanisms of action and the molecular targets of toxicity, and also for use in finding possible biomarkers of early response. Using valproic acid as a model substance, the use of DNA microarrays to identify teratogen-responsive genes in cell models is discussed. It is concluded that gene expression in P19 mouse embryocarcinoma cells represents a potentially suitable assay system, which could be readily used in a tiered testing system for developmental toxicity testing.