Hich annotations were common to two or more mice and which were unique.KEGG analysisFor each mouse, we fed into KEGG a list of all the transcripts that were significantly changed between AKT inhibitor 2 carcinoma and normal and searched for pathways that were altered in all or many of the mice. We then looked at the gene list in each category and asked which genes appeared in all of the mice.Results Heterogeneity in transcriptional AN 3199 regulation among carcinomas from different miceQuigley et al. [17] induced skin tumors on dorsal back skin of mice from a Mus spretus/Mus musculus 
backcross ([SPRET/ Ei6FVB/N]6FVB/N) using dimethyl benzanthracene (DMBA) and tetradecanoyl-phorbol 
acetate (TPA). This treatment inducedFigure 3. Differential regulation of specific annotations in DAVID. Selected annotations were analyzed to examine the manner in which the specific pathway is regulated. Venn diagrams show the number of mice in which the pathway showed negative (cyan), positive (pink), or both negative and positive regulation (purple). doi:10.1371/journal.pone.0057748.gHeterogeneous Gene Expression in SCC DevelopmentFigure 4. Heatmaps reveal heterogeneity between all 31 mice. Transcripts that were changed at least four-fold according to the average analysis are displayed with their expression level in each mouse separately. Although the mice can be clustered into closely related groups, no two mice show exactly the same expression pattern. doi:10.1371/journal.pone.0057748.gmultiple benign papillomas as well as malignant squamous cell carcinomas (SCC) and spindle cell carcinomas. Quigley et al. deposited Affymetrix Mouse Genome 430 2.0 array data from normal tail, papilloma and carcinoma for 31 mice. To obtain an overall picture of transcription variation among these 31 mice, we first asked how many transcripts were induced or repressed 4-fold in each mouse. A 4-fold (rather than the more usual 2-fold) cutoff was chosen, in order to minimize heterogeneity between mice. Even so, there was a marked variation between the mice in the numbers of transcripts that were up-regulated/down-regulated at each progression step (Figure 1).DAVID annotations vary among miceWe asked whether this mouse-to-mouse variation in transcription reflects the pathways that change during carcinogenesis in each mouse. To this end, significant genes were annotated using DAVID, and we examined which annotations were common to multiple mice and which were unique. Few annotations were significant in all 31 mice: only four annotations were up-regulated at least four-fold in allcarcinomas compared to normal skin (C/N) and two annotations were down-regulated (Figure 2 and Table S1A and S1B). The categories of cell migration and cell motility (as well as the nearsynonymous terms cell localization and cell motion) were upregulated between C/N in all 31 mice. Increased migration and motility are characteristic of epithelial to mesenchymal transition (EMT) and increased invasiveness [18,19]. We note that in some mice the categories of cell migration and cell motility were induced by the papilloma stage (P/N), whereas in others they were significantly induced upon progression from papilloma to carcinoma (C/P). As expected for categories that were over-represented in all 31 mice, these categories were also over-represented in the average analysis. Taken together, the data suggest that carcinoma development in all of the mice was dependent on EMT-like processes. The annotations “fatty acid metabolic process.Hich annotations were common to two or more mice and which were unique.KEGG analysisFor each mouse, we fed into KEGG a list of all the transcripts that were significantly changed between carcinoma and normal and searched for pathways that were altered in all or many of the mice. We then looked at the gene list in each category and asked which genes appeared in all of the mice.Results Heterogeneity in transcriptional regulation among carcinomas from different miceQuigley et al. [17] induced skin tumors on dorsal back skin of mice from a Mus spretus/Mus musculus backcross ([SPRET/ Ei6FVB/N]6FVB/N) using dimethyl benzanthracene (DMBA) and tetradecanoyl-phorbol acetate (TPA). This treatment inducedFigure 3. Differential regulation of specific annotations in DAVID. Selected annotations were analyzed to examine the manner in which the specific pathway is regulated. Venn diagrams show the number of mice in which the pathway showed negative (cyan), positive (pink), or both negative and positive regulation (purple). doi:10.1371/journal.pone.0057748.gHeterogeneous Gene Expression in SCC DevelopmentFigure 4. Heatmaps reveal heterogeneity between all 31 mice. Transcripts that were changed at least four-fold according to the average analysis are displayed with their expression level in each mouse separately. Although the mice can be clustered into closely related groups, no two mice show exactly the same expression pattern. doi:10.1371/journal.pone.0057748.gmultiple benign papillomas as well as malignant squamous cell carcinomas (SCC) and spindle cell carcinomas. Quigley et al. deposited Affymetrix Mouse Genome 430 2.0 array data from normal tail, papilloma and carcinoma for 31 mice. To obtain an overall picture of transcription variation among these 31 mice, we first asked how many transcripts were induced or repressed 4-fold in each mouse. A 4-fold (rather than the more usual 2-fold) cutoff was chosen, in order to minimize heterogeneity between mice. Even so, there was a marked variation between the mice in the numbers of transcripts that were up-regulated/down-regulated at each progression step (Figure 1).DAVID annotations vary among miceWe asked whether this mouse-to-mouse variation in transcription reflects the pathways that change during carcinogenesis in each mouse. To this end, significant genes were annotated using DAVID, and we examined which annotations were common to multiple mice and which were unique. Few annotations were significant in all 31 mice: only four annotations were up-regulated at least four-fold in allcarcinomas compared to normal skin (C/N) and two annotations were down-regulated (Figure 2 and Table S1A and S1B). The categories of cell migration and cell motility (as well as the nearsynonymous terms cell localization and cell motion) were upregulated between C/N in all 31 mice. Increased migration and motility are characteristic of epithelial to mesenchymal transition (EMT) and increased invasiveness [18,19]. We note that in some mice the categories of cell migration and cell motility were induced by the papilloma stage (P/N), whereas in others they were significantly induced upon progression from papilloma to carcinoma (C/P). As expected for categories that were over-represented in all 31 mice, these categories were also over-represented in the average analysis. Taken together, the data suggest that carcinoma development in all of the mice was dependent on EMT-like processes. The annotations “fatty acid metabolic process.
