Fig 1.
Scheme for Nrf2 biomarker development and characterization for screening of a mouse liver gene expression compendium.
Left, biomarker development and characterization. A number of microarray experiments were used to create the Nrf2 biomarker. Wild-type and Nrf2-null mice were treated with CDDO-Im (experiment 1; [47]). Wild-type and Keap1-null (whole body (WB) or liver-specific (LS)) mice were compared in experiments 2 [47] and 3 [49]. Differentially expressed genes (DEGs) were identified using Rosetta Resolver as indicated. Biomarker genes were identified from the DEGs after applying a number of filtering steps. Genes in the biomarker were evaluated by Ingenuity Pathway Analysis (IPA) for canonical pathway enrichment and potential transcription factor regulators. Right, biomarker testing and screening. The Nrf2 biomarker was imported into the BSCE environment in which protocols rank ordered the genes based on their fold-change. Screening was carried out by comparison of the biomarker to each bioset in the BSCE database using a pair-wise rank-based enrichment analysis (the Running Fisher algorithm). The results of the test including the direction of correlation and p-value of the test for each bioset in the compendium were exported and used to populate a master table containing bioset experimental details. A test of the accuracy of the biomarker predictions was carried out with treatments that are known positives and negatives for Nrf2 activation. Screening “hits” were characterized to determine the factors that modulate Nrf2 activation. Additionally, an external gene expression database of experiments performed with Affymetrix arrays was used in a classification analysis as well as to determine the relationship between expression of Nrf2 biomarker genes and p-values from the Running Fisher algorithm. Part of the Figure was adapted from a Figure in [44] and [40].
Fig 2.
Characterization of a Nrf2 biomarker.
A. The Nrf2 biomarker. (Top) Heat map comparing the expression of the genes in the biosets used to create the biomarker to the Nrf2 biomarker itself. Gene expression profiles from the livers of Keap1-null mice compared to corresponding control wild-type mice (1–3) and from the livers of wild-type (W) or Nrf2-null (N) mice exposed to CDDO-Im (CDDO) for 6 hours. Genes that were similarly regulated by genetic or pharmacological activation of Nrf2 were identified as detailed in the Methods. The biomarker represents the average expression of the genes across the biosets that exhibit genetic or pharmacological activation of Nrf2. The Keap1-null vs. wild-type comparisons were 1) hepatocyte-specific-null (GSE55001), 2) whole body-null (GSE55001), and 3) hepatocyte-specific null (GSE15633). B. Significant canonical pathways represented by the genes in the Nrf2 biomarker. Genes were examined by Ingenuity Pathways Analysis. C. Significant factors predicted to act as upstream regulators of the genes in the Nrf2 biomarker as determined by Ingenuity Pathways Analysis.
Fig 3.
A rank-based strategy for prediction of Nrf2 modulation.
A. Heat map showing the expression of genes in the Nrf2 biomarker across 443 biosets. Biosets were ordered based on their similarity to the Nrf2 biomarker using the p-value from the Running Fisher test. Biosets with positive correlation are on the left, and biosets with negative correlation are on the right. The black vertical lines denote a p-value of 10−4. B. The Nrf2 biomarker was compared to the biosets used to build the biomarker using the Running Fisher algorithm. Biosets with a negative correlation are shown as negative–log(p-value)s. C. Prediction of Nrf2 activation in the livers of mice exposed to oltipraz. Mice were exposed to oltipraz (4 days with 75 mg/kg/day), and the livers were examined for gene expression changes using Affymetrix mouse arrays. (Left) Heat map showing the expression of genes in wild-type and Nrf2-null mice. The genes were separated by strain-specific expression behavior and rank-ordered based on fold-change. The position of three genes in the Nrf2 biomarker are shown. (Middle) RT-PCR analysis of three biomarker genes in the livers of mice. Significantly different from corresponding control: * p < 0.05, **p < 0.01. Significantly different between controls in wild-type and nullizygous mice: # p < 0.05, ## p < 0.01. (Right) The two biosets derived from the oltipraz treated vs. control comparisons in wild-type or Nrf2-null mice were evaluated for activation of Nrf2 using the Running Fisher test. D. Summary of the prediction sensitivity and specificity of the Nrf2 biomarker. The biomarker was compared to chemicals or other biosets that were known positives or negatives for Nrf2 activation. A total of 81 biosets from 32 studies were examined.
Fig 4.
Activation of Nrf2 in females compared to males.
A. Activation of Nrf2 in the livers of female mice compared to male mice. (Upper) Female vs. male comparisons were examined for effects on Nrf2. Biosets were divided into the indicated four groups based on underlying factors in common between the mice. Control biosets are from wild-type mice which in some studies include treatment with control vehicle. Only sexually mature mice were used in the analysis. (Lower) Each of the biosets were also compared to the biomarker for STAT5b [33]. The results show the expected suppression of the male-specific STAT5b gene expression pattern in most of the female mice compared to male mice. B. Expression of Nrf2 and STAT5b biomarker genes in female to male comparisons. The biosets of female vs. male comparisons from A. were rank-ordered based on the significance of the overlap with the Nrf2 biomarker. C. Relationship between Nrf2 activation and STAT5b suppression. Biosets from (A) were plotted based on the significance of the Nrf2 and STAT5b biomarker comparisons. The plot shows the trend toward more significant Nrf2 activation with more significant suppression of STAT5b. The points in the box are those female vs. male comparisons with significant activation of Nrf2 and significant suppression of STAT5b.
Fig 5.
Nrf2 activation is hormonally-regulated.
A. (left) Nrf2 activation is suppressed by testosterone and activated by estradiol. Effects on the Nrf2 biomarker were examined after treatment with estrogen, testosterone, or dihydrotestosterone in the indicated studies. Biosets were derived from GSE13265 and GSE13388. (Right) The same biosets were examined for effects on STAT5b. B. Nrf2 is activated by ethinyl estradiol in male mice. Mice were treated with ethinyl estradiol for 7 or 28 days at the indicated doses and then evaluated for effects on Nrf2 (left) or STAT5b (right). Biosets were derived from GSE84590.
Fig 6.
Parallel activation of Nrf2 and suppression of STAT5b in genetic models of disrupted growth hormone signaling.
A. Expected Nrf2 biomarker behavior in biosets from Keap1-null or Nrf2-null mice. Biosets from nullizygous vs. wild-type comparisons from the indicated studies were assessed for Nrf2 activation. B. Nrf2 is activated in dwarf mice. Significance of the correlation of dwarf mouse vs. wild-type comparisons from the indicated mice to the Nrf2 or STAT5b biomarkers are shown. C. Relationship between activation of Nrf2 and suppression of STAT5b in the dwarf mouse comparisons from B. Points in the shaded box are those biosets in which there is significant Nrf2 activation and suppression of STAT5b.
Fig 7.
Feminization is associated with Nrf2 activation after chemical exposure in male mice.
A. Coincidence of Nrf2 activation and STAT5b suppression after chemical exposure in male mice. Shaded boxes divide those biosets of significant Nrf2 activation into those with and without significant suppression of STAT5b. Specific chemicals mentioned in the text are shown. PFNA, perfluorononanoic acid; GC-1, a thyroid hormone receptor beta-specific agonist. B. Nrf2 activation and STAT5b effects in female mice. C. Uncoupling of Nrf2 activation and STAT5b suppression in chemically-treated primary hepatocytes. Biosets from chemically treated primary hepatocytes or hepatocyte-derived cell lines almost always exhibit Nrf2 activation in the absence of feminization. Two biosets showed activation of Nrf2 and suppression of STAT5b.
Fig 8.
Greater numbers of Nrf2 biomarker genes are activated in females than males after chemical exposure.
Male and female mice were exposed to the indicated chemicals as described in the Methods and Table 1 and assessed for Nrf2 activation. From top to bottom panels: expression of Nrf2 biomarker genes, the number of positively regulated Nrf2 biomarker genes altered, the -log(p-value)s of the correlation of the bioset to the Nrf2 biomarker, total number of genes altered by the chemical treatment, and -log(p-value)s of the correlation to the indicated biomarkers for xenobiotic receptors.
Table 1.
Exposure conditions of 11 chemicals evaluated for sex differences in Nrf2 activation.