Article Figures & Data
Figures
- Fig. 1.
Gene expression profiles clustered UM-SCC and HKC cells using PCA and unsupervised cluster analysis. A, three-dimensional presentation of PCA of three major clustered sample groups. Each dot represents one sample calculated based on values of all 9,273 genes, whereas the percentage of variances represented were indicated in three axes. The distance in space between the colored boxes represents the degree of relatedness between the cell lines. B, hierarchical clustering dendrogram was established based on 942 genes identified to have greater than or equal to 2-fold differences between the average values of 10 UM-SCC cell lines and that of 4 primary human keratinocyte (HKC) clones. The heat map was scaled using log2-converted expression ratio of subjects (normal and tumor) to the universal reference for particular gene. Distinct cluster A and B gene expression patterns were denoted.
- Fig. 2.
Validation of microarray data using real-time RT-PCR and in situ hybridization. A, 12 genes were selected from the microarray experiment, of which 8 were from up-regulated gene group and 4 were from down-regulated gene group. The data are presented as the expression ratio in UM-SCC cell lines to human normal keratinocytes using both data from microarray (open square), and real-time RT-PCR (dotted square). B, detection of BAG2, PCNA, CCND1, and CCNB2 genes in HNSCC tissues and normal mucosa by in situ hybridization with the antisense probes specific for each gene (a). Sense probes were used as negative controls (b). Signals of in situ hybridization exhibited as green particles which are indicated by white arrows on the H&E-stained tissues. In each, UM-SCC 11A xenograft tumor section served as the positive control hybridized with antisense probes (a) and negative control with sense probes (b). For BAG2 and PCNA genes (left), HNSCC tumor samples hybridized with antisense probes were in (c) or sense probes in (d); normal mucosa samples hybridized with antisense probes were in (e), or sense probes in (f). For CCND1 and CCNB2 genes (right), the HNSCC tumor samples (c, e), the tissues containing both tumor and normal mucosa (d), and the normal mucosa (f) were hybridized with sense probe. Photomicrographs were taken under 100× magnification.
- Fig. 3.
Identification of transcription factor binding motifs in promoters of the genes in clusters A or B. Differentially expressed genes from clusters A (A) and B (B) are listed, and the putative transcription factor binding motifs for NF-κB, TP53, AP-1, and STAT were predicted in the proximal gene promoter region by Genomatix Suite. One black dot, one or more predicted binding site on the proximal promoter region (see Supplementary Table S3 for detailed analysis). C, the frequency of the predicted binding motifs was calculated in each cluster and analyzed by χ2 statistical method. A significant difference was observed in TP53 and NF-κB p65 binding motifs between the clusters; *, P < 0.05. No statistical significance of AP-1 and STAT predicted frequencies were observed between two clusters of the genes.
- Fig. 4.
Immunohistochemistry of TP53, NF-κB, and related protein expression in HNSCC tissues. A, immunostaining of TP53 (top left in each group), NF-κB p65 (p65, top right), and pan-cytokeratin (CK, bottom left), as well as H&E staining (bottom right) were carried out on tumor and normal mucosa (bottom right) samples using human HNSCC tissue array. All photomicrographs were taken under 100× magnification. B and C, immunohistochemistry of TP53, phospho–NF-κB p65, and its target gene products, CA9, c-IAP1, and YAP were carried on the frozen sections of HNSCC tissues. H&E staining and immunostaining of pan-cytokeratin were used to identify tumor cells from surrounding stroma. The pictures were taken under 400× magnification. D, linear regression analysis showed an inverse correlation with statistical significance between increasing TP53 and decreasing phospho–NF-κB p65 staining.
- Fig. 5.
ChIP assay confirmed the predicted transcription factor binding sites in the promoter regions. IL-8, IL-6, YAP1, and CA9 genes from cluster B were selected with the putative NF-κB binding sites, and ChIP assay was done using rabbit anti-human NF-κB p65 antibody in UM-SCC 11A, 38, and HKC. TNF-α was used as an inducer for NF-κB activity. The same amount of matched isotype antibody was used as the negative control (IgG), as well as no antibody controls (None). The input DNA was used as the loading control for DNA templates in PCR reaction.
- Fig. 6.
Modulation of NF-κB altered gene expression and cell survival. A, UM-SCC cell lines with wt TP53 (UM-SCC 1 and 9) were transiently transfected with p65 RNAi oligos or control oligos and 5× NF-κB luciferase and RSV-LacZ plasmids for 48 h, and the reporter activity was measured by luciferase activities. The relative luciferase unit was calculated by normalizing the luciferase to β-gal activities in the same experimental conditions. The experiments were carried out in triplicates, and statistical significance was examined by Student's t test; *, P < 0.05. B, UM-SCC 1 and 9 cells were transiently transfected with a plasmid containing p65 RNAi or control plasmid for 48 h. RNA was isolated, and real-time RT-PCR analysis was done for IL-6, IL-8, and YAP1 from cluster B gene list. The experiments were carried out in triplicates, and statistical significance was examined by Student's t test; *, P < 0.05. C, UM-SCC 1 and 9 cells were transiently transfected with plasmid containing p65 RNAi or control plasmid for 72 h. Cell morphology was captured under an inverted microscope (100×).
Additional Files
Supplementary Data
Files in this Data Supplement:
Volume 13, Issue 19
