Toxicogenomics; transcript profiling and potential application to chemical allergy.

Novel transcript profiling technologies allow simultaneous measurement of the changes in expression of many hundreds or many thousands of genes. The availability of these methods has brought about revolutionary changes in many areas of investigative biology, where analyses of patterns of gene expression, rather than of individual genes, are being employed. The application of these technologies to toxicology (toxicogenomics) offers new opportunities for both mechanistic toxicity research and predictive toxicology. Here we provide an overview of the basic approaches used in this field. The development of a series of toxicology-specific microarrays in our own laboratory is discussed, together with an example of one area of mechanistic research, chemical allergy, where we believe judicious application of toxicogenomics will make an important contribution.

Application of genomics to the definition of the molecular basis for toxicity.

Transcript profiling technology enables quantitative measurement of the transcriptional activity of potentially thousands of genes in biological samples. The application of such technology to toxicology, toxicogenomics, promises substantial dividends in mechanistic toxicity research and also, possibly, the ability to predict adverse toxicity for novel or untested compounds. Our laboratory has developed a custom approach to this technology, designing cDNA microarray platforms specifically for gene expression events of relevance to a large number of toxicological endpoints. Such arrays allow comprehensive coverage of genes associated with entire pathways (such as oxidative stress, signal transduction, stress response, epithelial biology) and enable simultaneous measurement of more than ten thousand gene expression events.

Obstacles to the prediction of estrogenicity from chemical structure: assay-mediated metabolic transformation and the apparent promiscuous nature of the estrogen receptor.

Information on structure-activity relationships (SAR) and pathways of metabolic activation would facilitate the preliminary screening of chemicals for estrogenic potential. Published crystallographic studies of the estrogen receptor (ER) imply an essential role of the two hydroxyl groups on estradiol (17beta-E(2)) for its binding to ER. The influence of these hydroxyl groups on ER binding and estrogenicity was evaluated by the study of 17beta-E(2) with one or both of these hydroxyl groups removed (17beta-desoxyestradiol and 3, 17beta-bisdesoxyestradiol, respectively). 6-Hydroxytetralin (17beta-E(2) with its C- and D-rings removed) and other synthetic estrogens were also studied. The estrogenicity assays comprised a yeast ER-mediated transcription assay, mammalian cell transcription assays incorporating either ER alpha or ER beta, and the immature rat uterotrophic assay. With the exception of 6-hydroxytetralin in the uterotrophic assay, all the chemicals were active in all the assays. Hydroxylation of the two desoxy compounds to estradiol was shown to occur in immature female rats, but metabolism was not implicated in the responses observed in the ER-binding and yeast systems. It is concluded that the 3-hydroxyl and 17beta-hydroxyl groups of 17beta-E(2) are not absolute requirements for estrogenicity. It would therefore be of value to the derivation of SAR for estrogenicity were the crystal structure of the bisdesoxy-E(2)/ER complex to be evaluated.

Identification of a possible association between carbon tetrachloride-induced hepatotoxicity and interleukin-8 expression.

Hepatotoxicants can elicit liver damage by various mechanisms that can result in cell necrosis and death. The changes induced by these compounds can vary from gross alterations in DNA repair mechanisms, protein synthesis, and apoptosis, to more discrete changes in oxidative damage and lipid peroxidation. However, little is known of the changes in gene expression that are fundamental to the mechanisms of hepatotoxicity. We have used DNA microarray technology to identify gene transcription associated with the toxicity caused by the hepatotoxicant carbon tetrachloride. Labeled poly A+ RNA from cultured human hepatoma cells (HepG2) exposed to carbon tetrachloride for 8 hours was hybridized to a human microarray filter. We found that 47 different genes were either upregulated or downregulated more than 2-fold by the hepatotoxicant compared with dimethyl formamide, a chemical that does not cause liver cell damage. The proinflammatory cytokine interleukin-8 (IL-8) was upregulated over 7-fold compared with control on the array, and this was subsequently confirmed at 1 hour and 8 hours by Northern blot analyses. We also found that carbon tetrachloride caused a time-dependent increase in interleukin-8 protein release in HepG2 cells, which was paralleled by a decrease in cell viability. These data demonstrate that carbon tetrachloride causes a rapid increase in IL-8 mRNA expression in HepG2 cells and that this increase correlates with a later and significant increase in the levels of interleukin-8 protein. These results illustrate the potential of microarray technology in the identification of novel gene changes associated with toxic processes.

Use of cDNA microarrays to probe and understand the toxicological consequences of altered gene expression.

Genomic sciences offer the ability to measure quantitative modulation of transcription in cells and tissues under a wide variety of conditions. We have developed a series of custom cDNA microarrays specifically to investigate toxicity processes. Around 600 marker genes for toxicity were selected and representative cDNA clones were obtained, amplified and purified by polymerase chain reaction (PCR), before being immobilised on nylon membranes. A detailed database on biochemical function, role in disease and allelic variation has been assembled for each gene. Applications in our laboratory include mechanistic investigation of a number of toxic endpoints such as hepatotoxicity and endocrine disruption.

Oestradiol is a potent mitogen and modulator of GnRH signalling in alphaT3-1 cells: are these effects causally related?

GnRH acts via phospholipase C (PLC) activating G-protein coupled receptors to stimulate secretion of gonadotrophins from gonadotrophs. These cells are also regulated by gonadal steroids, which act centrally to influence GnRH secretion, and peripherally to modulate GnRH action. We have shown that oestradiol can stimulate proliferation and modulate GnRH-stimulated [(3)H]inositol phosphate ([(3)H]IP(x)) accumulation (used as a measure of PLC activity) in a gonadotroph-derived cell line (alphaT3-1). Here we show that when alphaT3-1 cells were incubated in medium with 2% foetal calf serum (FCS), [(3)H]thymidine incorporation was not stimulated by oestradiol but was reduced to <2% of control by the oestrogen antagonist, raloxifene. The inhibitory effect of 10 or 1000 nM raloxifene was reversed competitively by oestradiol. A similar pattern of effects was seen when effects of oestradiol and raloxifene on the proportion of cells in the S-phase of the cell cycle (as measured by flow cytometry of propidium iodide-labelled cells) and on oestrogen receptor activity (as measured by trans-activation of the oestrogen-response elements in the vitellogenin promoter) were quantified. In addition, RT-PCR revealed expression of alpha and beta (but not beta2) subtypes of oestrogen receptors. Thus, oestrogen is an essential mitogen for alphaT3-1 cells, its mitogenic effect is oestrogen receptor mediated and is associated with a marked alteration of cell cycle distribution, and the full extent of these effects are best revealed in the presence of raloxifene. Using this strategy, we found that cells cultured for 4 days with 10 nM raloxifene expressed GnRH receptors (K(d) for (125)I-buserelin 4.33 nM) and that their activation by GnRH caused a concentration-dependent increase in [(3)H]IP(x) (in cells labelled with [(3)H]inositol) and inositol 1,4,5 trisphophate (in unlabelled cells). Addition of 10 nM oestradiol (to overcome receptor blockade by raloxifene) reduced GnRH receptor number by 31% but increased maximal effects on [(3)H]IP(x) and Ins(1,4,5)P(3) approximately 4-fold. The effects of oestradiol on GnRH receptor number and signalling were not, however, mimicked by culture for 2 days in medium with 10% FCS and the S-phase blocker, thymidine (15 mM). This treatment increased the proportion of cells in the S-phase 2- to 3-fold but did not alter GnRH receptor number or signalling. Other treatments which altered cell cycle transition (hydroxyurea, colcemid, methotrexate) also failed to alter GnRH receptor number or signalling and no correlation was seen between GnRH receptor number or GnRH-stimulated [(3)H]IP(x) accumulation and the proportion of cells in the S-phase or G2/M-phases of the cell cycle. Thus, oestradiol has pronounced effects on GnRH signalling, proliferation and cell cycle distribution in alphaT3-1 cells, but these trophic effects do not underlie the modulation of GnRH signalling.

Specific methylation events contribute to the transcriptional repression of the mouse tissue inhibitor of metalloproteinases-3 gene in neoplastic cells.

The tissue inhibitor of metalloproteinases-3 (TIMP-3) gene is specifically down-regulated in neoplastic cells of the mouse JB6 progression model, suggesting a role for TIMP-3 inactivation in neoplastic progression. On the basis of 5-azacytidine reversal, the mechanism for this down-regulation appears to involve changes in the methylation state of the TIMP-3 promoter. Although total genomic methylation levels are comparable, specific differences in the methylation of the TIMP-3 promoter were observed between preneoplastic and neoplastic JB6 cells at three Hpall sites, with preneoplastic cells being less methylated. Expression of antisense methyltransferase in a neoplastic JB6 variant known to be hypermethylated in TIMP-3 resulted in reactivation of the endogenous TIMP-3 gene and restoration of hypomethylated status to the three implicated Hpall sites. Thus, hypermethylation at specific sequences in the TIMP-3 promoter appears to contribute to the silencing of the gene in neoplastic cells.

Differential activation by xenoestrogens of ER alpha and ER beta when linked to different response elements.

The discovery of a second estrogen receptor (ER beta) has significant implications for our understanding of the molecular basis for the diverse actions of estrogen. Here we report the differential activation by natural and xenobiotic estrogens of ER alpha and ER beta when linked to different response elements. Receptor mediated activation of reporter constructs containing either the estrogen response element (ERE) from the vitellogenin (Vit) gene or from the luteinizing hormone beta (LH) gene were examined in transiently transfected Cos-1 cells. ER beta preferentially activated the consensus Vit ERE whereas ER alpha showed greater activation at the divergent LH ERE. This differential activation was observed for a number of ligands including estradiol, estrone, bisphenol A, octylphenol and diethystilbestrol. These findings show that the nature of the ERE, as well as the ratio of ER subtypes in a particular cell/tissue, will influence whether particular estrogen responsive genes are activated in the presence of natural or xenobiotic estrogens.

The position and length of the steroid-dependent hypersensitive region in the mouse mammary tumor virus long terminal repeat are invariant despite multiple nucleosome B frames.

Stimulation of the mouse mammary tumor virus with steroids results in the generation of a DNase I-hypersensitive region (HSR) spanning the hormone responsive element (HRE) in the long terminal repeat. Restriction enzymes were used to characterize the accessibility of various sites within the HSR of mouse mammary tumor virus long terminal repeat-reporter constructions in four different cell lines. The glucocorticoid-dependent HSR was found to span minimally 187 bases, a stretch of DNA longer than that associated with histones in the core particle. Although the 5'-most receptor binding site within the HRE is downstream of -190, hypersensitive sites were found further upstream to at least -295. The relationship in the accessibility between pairs of sites in the vicinity of the HSR was further examined in one cell line by a two-enzyme restriction access assay. In the uninduced state, the accessibilities at these sites were found to be independent of each other. In contrast, when stimulated with hormone, the accessibilities at these sites were observed to become linked. That is, once a distinct promoter was activated, all of the sites within the HSR of that molecule became accessible. The HSR formed along an invariant stretch of DNA sequence despite the multiplicity of nucleosome frames in the nucleosome B region, where the HRE is located. The results indicate that the macroscopic length of the HSR does not arise from core length-remodeling events in molecules containing Nuc-B in alternative positions.

Expression of dominant negative Erk2 inhibits AP-1 transactivation and neoplastic transformation.

The mitogen activated protein (MAP) kinases or extracellular signal-regulated kinases (Erks) are activated in response to Ras expression or exposure to tumor promoters or to growth factors, and have been implicated in AP-1 transactivation in some models. We have shown that tumor promoter induced activation of the transcription factor AP-1 is required for induced neoplastic transformation in the Balb/C JB6 cell model. Jun and Fos family protein levels have been found not to be limiting for AP-1 response. The present study asks whether activation of Erks1 and 2 is required for AP-1 transactivation and transformation of JB6 cells and whether Erks might be targeted for cancer prevention. Expression of either of two different dominant negative kinase inactive Erk2 mutants in transformation sensitive (P+) JB6 cells substantially inhibited the tumor promoter induced activation of Erks1 and 2 and of AP-1 measured by a collagenase-luciferase reporter. Multiple mutant Erk2 expressing clonal lines were also rendered non-responsive to induced neoplastic transformation. These observations, together with our recent finding attributing AP-1 non-responsiveness to Erk deficiency in a clonal line of transformation resistant (P-) cells, argue for a requirement for Erks1 and/or 2 activation in AP-1 transactivation in the mouse JB6 neoplastic progression model, and suggest the utility of Erks as a prevention target.

Differential functional significance of AP-1 binding sites in the promoter of the gene encoding mouse tissue inhibitor of metalloproteinases-3.

Tissue inhibitor of metalloproteinases-3 (TIMP-3) is an extracellular-matrix-associated protein that suppresses tumorigenicity or invasion in several model systems. We have identified, by in vitro footprinting, six AP-1 (activator protein-1) or AP-1-like binding sites in the mouse TIMP-3 promoter that bind purified c-Jun homodimers. Electrophoretic mobility shift assays revealed that the non-consensus fifth AP-1 binding site (AP-720; nt -720 to -714) had the strongest binding activity for recombinant c-Jun protein, and that the fourth binding site (AP-763; nt -763 to -754) and AP-720 showed strong binding activity for cellular nuclear proteins. Antibody supershift and blocking experiments suggest that AP-720, but not AP-763, binds authentic AP-1 components. Transient transfection reporter assays of deletion constructs showed that the region spanning AP-720 has the highest transcriptional activity, and that sequences 5' to this region (nt -2846 to -747) may contain negative regulatory elements. The deletion construct containing about 500 nt 5' to the transcriptional start, but no AP-1 sites, showed lower but significant activity, suggesting both AP-1-dependent and -independent regulation of the mouse TIMP-3 promoter. Mutational inactivation of AP-720 abolished the activity increment that distinguished the reporter construct containing both AP-720 and sixth AP-1 binding site (AP-617; nt -617 to -611) from that containing only AP-617. In summary, we report here that both AP-1 and non-AP-1 elements contribute to activity, with the non-consensus AP-1 site at -720 showing the greatest functional significance among the AP-1 sites.

Nucleoprotein structure influences the response of the mouse mammary tumor virus promoter to activation of the cyclic AMP signalling pathway.

Recent studies have provided evidence of crosstalk between steroid receptors and cyclic AMP (cAMP) signalling pathways in the regulation of gene expression. A synergism between intracellular phosphorylation inducers and either glucocorticoids or progestins has been shown to occur during activation of the mouse mammary tumor virus (MMTV) promoter. We have investigated the effect of 8-Br-cAMP and okadaic acid, modulators of cellular kinases and phosphatases, on the hormone-induced activation of the MMTV promoter in two forms: a transiently transfected template with a disorganized, accessible nucleoprotein structure and a stably replicating template with an ordered, inaccessible nucleoprotein structure. Both okadaic acid and 8-Br-cAMP synergize significantly with either glucocorticoids or progestins in activating the transiently transfected MMTV template. In contrast, 8-Br-cAMP, but not okadaic acid, is antagonistic to hormone-induced activation of the stably replicating MMTV template. Nuclear run-on experiments demonstrate that this inhibition is a transcriptional effect on both hormone-induced transcription and basal transcription. Surprisingly, 8-Br-cAMP does not inhibit glucocorticoid-induced changes in restriction enzyme access and nuclear factor 1 binding. However, association of a complex with the TATA box region is inhibited in the presence of 8-Br-cAMP. Thus, cAMP treatment interferes with the initiation process but does not inhibit interaction of the receptor with the template. Since the replicated, ordered MMTV templates and the transfected, disorganized templates show opposite responses to 8-Br-cAMP treatment, we conclude that chromatin structure can influence the response of a promoter to activation of the cAMP signalling pathway.

The 5' enhancer of the mouse mammary tumor virus long terminal repeat contains a functional AP-2 element.

The mouse mammary tumor virus (MMTV) retrovirus causes mammary adenocarcinomas in mice by proviral insertion near members of the wnt family of proto-oncogenes, leading to their deregulation and cellular transformation. The 5' end of the MMTV long terminal repeat (LTR) has been implicated in tissue-specific activation of these genes. In this study, we characterize an enhancer element (Ban2; -1075 to -978) at the 5' end of the MMTV LTR. We show that this enhancer is 5-fold more active in a murine mammary carcinoma cell line (34i) than in a fibroblast cell line (NIH3T3), and is inactive in the liver carcinoma cell line HepG2. Mutagenesis of the enhancer reveals four cis-acting elements that are required for maximal activity. DNA-binding proteins that interact with each of the four elements have been identified. One of these factors, designated mp5, is either identical to, or closely related to, the transcription factor AP-2. The mp5/AP-2 DNA binding activity co-migrates with recombinant AP-2 and is supershifted by anti-AP-2 antibodies. We also show that the lack of enhancer activity in HepG2 cells results from the absence of AP-2 protein in these cells. Co-transfection of an AP-2 expression vector restores the activity of this enhancer in HepG2 cells, and requires an intact mp5-binding site.

Analysis of the transforming functions of bovine papillomavirus type 4.

Bovine papillomavirus type 4 (BPV-4) morphologically transforms primary bovine cells in vitro only in the presence of an activated ras gene. The transformed cells are capable of anchorage-independent growth, but are not immortal and are incapable of inducing tumors in nude mice. BPV-4 does not possess an E6 ORF and failure to induce full transformation may be due to the lack of this gene. Here we report the contribution of individual BPV-4 genes to cell transformation and the effect of adding the E6 ORF of HPV-16 to the system. We show that BPV-4 E7 ORF is responsible for morphological transformation, the E8 ORF is responsible for anchorage-independent growth, and addition of HPV-16 E6 ORF rescues cells from senescence. By immunocytostaining, E7 and E8 have been visualized in transiently transfected cultured cells. E8 is localized in the membranes while E7 is found both in the cytoplasm and in the nucleus.

BPV-4 induces amplification and activation of 'silent' BPV-1 in a sub-line of C127 cells.

BPV-4-induced malignant transformation of C127 mouse fibroblasts in vitro is the result of a 'hit and run' mechanism. Viral DNA is lost on continued subculture of transformed cell lines without loss of any malignant characteristics. The DNA from these cells harbours specific amplified sequences. Two such amplified fragments of approximately 10 kb and 12 kb were molecularly cloned and designated HL-10 and HL-12 respectively. HL-10 transformed C127 cells efficiently and therefore encodes a transforming function, whereas the 12 kb clone did not. Heteroduplexes showed that HL-12 was homologous to HL-10 except for two additional tandem copies of an approximately 1.7 kb sequence. Sequence analysis of HL-10 revealed that the clone contained a 5.2 kb region from BPV-1 including the transforming ORFs. Transformation studies have shown differences between HL-10 and BPV-1, indicating that the host flanking sequences may contribute to the transforming potential of the BPV-1 ORFs. The BPV-1 DNA was associated with sequences homologous to murine autonomously replicating sequences (ARS) implicated in the establishment of multiple tandem DNA repeats. As the parental cells contain the set of sequences amplified in transformed cells in single copy and show none of the characteristics of transformed cells, we conclude that BPV-4 has activated these sequences by amplification and rearrangement. These phenomena may be mediated through an interaction between BPV-4 proteins and the BPV-1 origin of DNA replication or via the ARS region.

Synergism between bovine papillomavirus type 4 and the flavonoid quercetin in cell transformation in vitro.

Bovine papillomavirus type 4 (BPV-4) morphologically transforms primary bovine cells in vitro only in the presence of an activated ras gene. The transformed cells are capable of anchorage-independent growth, but are not immortal and are incapable of inducing tumors in nude mice, suggesting that other events are needed to convert the cells to the fully transformed phenotype. We show here that treatment of the cells with a single dose of the flavonoid quercetin leads to full oncogenic transformation of cells transfected with BPV-4 and ras. Quercetin is one of the most potent mutagens found in bracken fern, the environmental cofactor in BPV-4-associated carcinogenesis of the upper alimentary canal of cattle. Our results point to quercetin as the probable in vivo cocarcinogen synergizing with BPV-4 in malignant progression.

The B subgroup bovine papillomaviruses lack an identifiable E6 open reading frame.

Analysis of the corrected DNA sequence for the bovine papillomavirus type 4 (BPV4) genome revealed that there is no open reading frame (ORF) that might encode an E6 protein. The other two B subgroup bovine papillomaviruses, BPV3 and BPV6, were found to have the same arrangement of ORFs in this region as BPV4. Thus, we conclude that E6 functions are either not required by these viruses or are performed by another viral (or host) protein. Furthermore, the position that might be expected to be occupied by E6, between the long control region and the E7 ORF, contains the E8 ORF, which has the potential to encode a 42-residue polypeptide with considerable similarity to the E5 transforming protein of BPV1. Therefore, it appears that during the evolution of the B subgroup of BPVs, genomic rearrangements may have occurred resulting in the present layout of the early ORFs.

Cooperation between bovine papillomavirus type 4 and ras in the morphological transformation of primary bovine fibroblasts.

Primary bovine fibroblasts derived from foetal palate can be transformed by bovine papillomavirus type 4 DNA only in the presence of an activated ras gene, indicating that the virus does not encode all the information required for morphological transformation of non-established cells. A subgenomic fragment containing the complete E8 and E7 open reading frames (ORFs) induces transformation in cooperation with activated ras but transformation is abolished when the E7 ORF is deleted at the 3' end, showing that this ORF encodes a necessary transforming function. Transformation is more aggressive when the E8 and E7 ORFs are placed under the transcriptional control of the long terminal repeat of the mouse Moloney leukaemia virus, suggesting that the degree of transformation is dependent on the level of expression of these genes.