Fig 1.
Building and characterizing the NF-κB biomarker.
(A) Gene expression profiles of the six biosets used to construct the NF-κB biomarker. (Left) The fold-changes of the statistically significant genes from the Tian et al. (2005) study are shown after one-dimensional clustering of genes. (Center) Expression of the genes that make up the biomarker after TNFα treatment in wild-type and IκB-expressing cell lines are shown. (Right) The fold-change values averaged across treatments in wild-type cells yielded the 108 gene NF-κB biomarker, with the names of the top 10 genes shown. (B) Ingenuity Pathway Analysis of biomarker genes. (Top) Transcription factors predicted to regulate the biomarker genes using the upstream regulator analysis. (Bottom) Canonical pathways that significantly overlap with the biomarker genes. Biomarker genes were compared to the genes in the canonical pathway lists from IPA. (C) Comparison of the biomarker to the biosets used to construct the biomarker. The -log(p-value)s of the pairwise correlations are in the same order as those in A (middle).
Fig 2.
Assessment of the accuracy of the biomarker in predicting NF-κB activation.
The sensitivity of the biomarker was determined using studies from cells expected to exhibit NF-κB activation after exposure to (A) interleukin 1α/β, (B) lipopolysaccharides, and (C) tumor necrosis factor α. For each factor, the -log(p-values) for the studies were rank-ordered. The red line shows the cutoff for statistical significance (-log(p-value) = 4).
Fig 3.
The biomarker predicts activation by Toll-like receptor and interleukin receptor agonists.
Significance of the pairwise comparisons between microarray profiles and the biomarker are shown for experiments in which whole blood from normal, MyD88-defective, or IRAK4-defective patients was stimulated for 2 hours with the indicated agonists (data from GSE25742 [35].
Fig 4.
Screening a human microarray compendium for NF-κB chemical modulators.
(A) The -log(p-values) for correlations between the NF-κB biomarker and each of the biosets representing cells exposed to individual chemicals. Biosets were rank ordered by -log(p-value) of the correlation between the biomarker and the bioset. The cutoffs for statistical significance are shown with dashed lines. The biosets with -log(p-value) ≥ 4 were considered positively correlated while the biosets with -log(p-value) ≤ -4 were considered negatively correlated with the biomarker. The top five biosets predicted to be activators or inhibitors are shown with the chemical names. (B) Gene expression changes for NF-κB biomarker genes across the biosets evaluating chemical-induced changes in gene expression. The NF-κB biomarker fold-change values are shown on the left.
Fig 5.
Characterization of NF-κB activators.
(A) Time-dependent changes in NF-κB activation after exposure to particulate matter (PM)10 isolated from either indoor air from classroom settings (indoor) or outdoor air in bronchial epithelial BEAS-2B cells (from GSE34607). (B) Time-dependent changes in activation of NF-κB by 3uM sphingosine-1-phosphate in dermal fibroblasts from primary normal human dermal fibroblasts (normal) or C18 dermal fibroblasts (C18) (data from GSE56308). (C) Concentration-dependent increase in NF-κB activity after exposure to silica (left; from GSE30213 and GSE30200) and silica nanoparticles (right; from GSE63806) in A549 cells.
Table 1.
The 20 biosets with the highest correlation to the NF-κB biomarker.
Fig 6.
Characterization of NF-κB inhibitors.
(A) The -log(p-values) representing the correlation between the NF-κB biomarker and each of the 19 putative NF-κB inhibitors that fell into the five major functional categories of inhibitors (from 49 total biosets). Error bars are shown when multiple biosets assessed a single chemical. (B) Time-dependent NF-κB suppression by 10 nM mometasone in lung fibroblasts (from GSE30242) and 100 nM dexamethasone in macrophages (from GSE61880).
Table 2.
Predicted inhibitors of NF-κB identified using the biomarker screening approach.
Forty-nine chemicals predicted to inhibit NF-κB were identified. Shown are experimental conditions that led to significant correlation*.
Fig 7.
Expression of NF-κB biomarker genes in wild-type and NFKB1-1-null cells.
Wild-type and NFKB1-null HeLa cells were treated with IL1β or the indicated chemicals for 6 hrs and expression of the NF-κB-responsive gene CXCL1 or several NF-κB biomarker genes were examined by RT-qPCR. (A) Expression changes of NF-κB-responsive genes are diminished or abolished in NFKB1-null cells. *Indicates significant difference between treated and control wild-type cells; p-value < 0.05. # Indicates significant difference between treated wild-type and treated NFKB1-null cells; p-value < 0.05. (B) Changes in the expression of CXCL1 and IL6 genes after exposure to 15 Tox21 chemicals. (C) Changes in the expression of CXCL1 and IL6 genes after exposure to 17 ToxCast chemicals.
Table 3.
Activators of NF-κB identified by HTS assays examined by RT-qPCR.
There were 32 organic chemicals predicted to activate NF-κB selected for further study.
Fig 8.
Transcript profiling of chemicals in wild-type and NFKB1-null cells.
The indicated treatments were analyzed by TempO-Seq human S1500+ platform examining the expression changes in ~3000 genes. Significant expression changes were identified as described in the Methods. (A) The heat maps show the genes altered by the indicated treatment in wild-type HeLa cells and their expression after treatment in the NFKB1-null cells. Chlor, chlorhexidine diacetate; Carbo, carbocyanine. (B) Number of genes significantly altered in each treatment described in A.