Search

Additional file 1: SupplementaryFigure 1. Copy number variation and expression of EGFR in Kyse30and FaDu cells. Supplementary Figure 2. GSEA of EGF- and EpEX-treated Kyse30and FaDu cells. Supplementary Figure 3. Over-representation analysis of genesof the EGFR-mediated EMT signature. SupplementaryFigure 4. Comparison ofEGFR-mediated EMT, pEMT, and EMT signatures. Supplementary Figure 5. Comparisonof EMT signatures for prognostic purposes.Supplementary Figure 6. ITGB4,ITGA6, LAMA3, LAMB3, and LAMC2 expression in HNSCC. Supplementary Figure 7. ITGB4expression in malignant and non-malignant single cells in different cancerentities. Supplementary Figure 8. ITGA6 expression in malignant andnon-malignant single cells in different cancer entities. Supplementary Figure 9. LAMA3expression in malignant and non-malignant single cells in different cancerentities. Supplementary Figure 10. LAMB3 expression in malignant andnon-malignant single cells in different cancer entities. Supplementary Figure 11. LAMC2expression in malignant and non-malignant single cells in different cancerentities. Supplementary Figure 12. ITGB4 expression in knockdown clonesof Kyse30 and FaDu cells. Supplementary Figure 13. Wound healing capacity of control andITGB4-knockdown cell lines. SupplementaryFigure 14. Tumor buddingintensities in HNSCC.

Additional file 4: Supplementary Table 3. Gene expression correlation with ITGB4 in the HPV-negativeTCGA cohort. Batch correlation analysis identified correlations of geneexpression with integrin beta 4 (ITGB4). Gene ID, Spearman correlation, andp-value are indicated for the top ten positively (co-regulated) and negativelycorrelated genes (counterregulated).

Additional file 3: SI Figure 3. Pathway enrichment analysis using the 50 Hallmark gene sets form MSigDB Fischer Exact Test was applied to test for enrichment of Hallmark gene sets within the 2nd-neighborhood of CLIP2 of the GNA reconstructed from gene set 4. 19 out of 50 gene set exhibited a significant enrichment with an adjusted p-value < 0.05 (red dashed line)

Additional file 1: SI Figure 1. Comparison of the four gene sets used for GNA reconstruction, generated by differential expression analysis (SI Figure 1) A) Upset plot in intersect mode (common genes within the respective gene sets green: present in all four gene sets, black: present in three gene sets, blue: present in two gene sets, orange: number of genes in gene sets 1-4) B) Upset plot in distinct mode (common genes within the respective gene sets, but unique to the respective gene sets / not present in other gene sets) C) Venn diagram D: Visualization of GAN reconstruction statistics derived from GeneNet method applied to gene set 4. Each dot in the plot represents a potential edge between two genes in the network. A total number 237016 edges was calculated from gene set 4, while the application of the edge probability cut-off of 0.5 (black dashed line) results in the final network with 10777 edges and 654 nodes (genes). All direct interactions with CLIP2 (1st neighborhood) are visualized. Negative partial correlation coefficients represent a negative association, while positive partial correlation coefficients indicate a positive association between the genes (nodes).

Additional file 4: SI Figure 4. Validation of CLIP2 and PPIL3 protein co-expression by Immunohistochemical staining in papillary thyroid tumor tissue (10 cases) left column: CLIP2 staining right column: PPIL3 staining

Additional file 2: SI Figure 2. A) number of direct CLIP2 interactors in the GNA reconstructed from gene set 4, dependent on the edge probability cut-off in the GeneNet method. Red dashed line indicates the selected cut-off of 0.5 B) Edge-list of the 1st neighborhood of CLIP2 reconstructed from gene set 4. pcor = partial correlation coefficient, node 1/2 = gene names of nodes, pval = p-value of partial correlation test, qval = q-values (positive false discovery rate), prob = edge probability black line: selected cut-off of 0.5 C) Degree distribution of the GNA generated from gene set 4 and a probability cut-off of 0.5 in the GeneNet reconstruction method. The degree distribution (approximately) follows a power law and thereby represents on attribute of a scale-free network.

Additional file 1 Supplementary simulations. Further plots of the three simulation scenarios, adding “frequency of non-empty signatures”, “mean signature length” and performance scores free from empty signatures (i.e., performance of data usage setting, given a candidate signature was discovered.

Temporal expression and corresponding fold-changes of gene ERV3-1 after irradiation with 8Â Gy. (PDF 531Â kb)

Temporal expression and corresponding fold-changes of gene ERVMER34-1 after irradiation with 8Â Gy. (PDF 535Â kb)

Overview on the cytogenetic aberrations in CAL-33 cells. Clonal structural and numerical chromosomal aberrations detected by SKY and copy number changes detected by array CGH are listed. (DOCX 94Â kb)

Analyses of proliferation rates and cell cycle distribution. Proliferation curves of the CAL-33 parental cell line and the subclones SP and RP are shown in panel (A). Doubling times were determined in the exponential growth phase by semi-log regression. Cell cycle analyses were performed under exponential growth conditions (B). The dynamics of G2-arrest upon irradiation at 4Â Gy are shown in figure panel (C). (PDF 318Â kb)

Technical validation of microarray data. Eight of the differentially expressed genes detected with the microarray technology were arbitrarily chosen for technical validation by qRT-PCR. Correlation analysis results (Spearmanâ s rho coefficient) and fold-changes (microarray and qRT-PCR) are shown. Asterisks (*) indicate fold-change values for genes detected as differentially expressed by microarray analysis. (DOCX 52Â kb)

Cytogenetic characterization of HN2092. (PDF 8434Â kb)

Copy number alterations in HN1957. (PDF 37Â kb)

Pathway enrichment analysis of differentially expressed genes in HN2092 after radiochemotherapy treatment. (PDF 44Â kb)

EGFR and EpCAM surface expression of HN1957, HN2092 and OKF6-hTERT. (PDF 838Â kb)

Pathway enrichment analysis of differentially expressed genes in HN1957 after radiochemotherapy treatment. (PDF 66Â kb)

Copy number alterations in HN2092. (PDF 43Â kb)

Cytogenetic characterization of HN1957. (PDF 8257Â kb)

Pathway enrichment analysis of differentially expressed genes in HN2092 after radiochemotherapy treatment. (PDF 44Â kb)

Significantly deregulated mRNAs in primary HN1957 after in vitro radiochemotherapy treatment. (PDF 146Â kb)

Significantly deregulated mRNAs in primary HN2092 after in vitro radiochemotherapy treatment. (PDF 143Â kb)

Potential target genes in HN2092 for miRNAs responding to therapy in HNSCC patients. (PDF 51Â kb)

Significantly deregulated mRNAs in primary HN1957 after in vitro radiochemotherapy treatment. (PDF 146Â kb)

Potential target genes in HN1957 for miRNAs responding to therapy in HNSCC patients. (PDF 72Â kb)

Primer sequences for TP53 and EGFR sequencing. (PDF 42Â kb)

Pathway enrichment analysis of differentially expressed genes in HN1957 after radiochemotherapy treatment. (PDF 66Â kb)

Significantly deregulated mRNAs in primary HN2092 after in vitro radiochemotherapy treatment. (PDF 143Â kb)