Article Figures & Data
Figures
- Figure 1. Expression of ALKF1174L in different skin compartments induces skin lesions and cSCC.
(A) Graphical representation of Lgr5-CreERT2;LSL-ALKF1174L genotype and experimental design. Topical application of 4OH-tamoxifen (4-OHT) in ears, back, and tail skin of mice resulted in skin lesion development. Then, mice were euthanized when termination criteria were observed (tumor size and ulceration). (B) Tumor-free survival of Lgr5-CreERT2;LSL-ALKF1174L mice (n = 15, median 47 d) and controls (n = 12). Log-rank (Mantel–Cox) Test P < 0.0001, HR 28.12. All mice developed tumors. (C) From left to right. Topical administration of 4-OHT and after tumor formation per location. (D) Representative picture of in vivo imaging system (IVIS). Analysis of the LSL-ALKF1174L transgene expression shows strong luminescent signal from the tumors on ears and tail. (E) Pie chart representing the number of mice that developed the listed skin lesions out of total skin lesions diagnosed. (F) Representative hematoxylin and eosin (H&E) staining of such lesions from the ears. Scale bars = 100 µm. To note, each mouse developed several tumors. (G, H, I) Top to bottom. Genotypes, representative pictures of the tumors marked by pan-cytokeratin immune-labeling and tumor distribution per location. To note, because of leakage of the Cre expression, tumor formation was observed when no specific topical administration was performed. Scale bars = 250 µm.
- Figure S1.
Overepxression of ALKF1174L induces skin tumorigenesis independently of the cell of origin. (A) Representative pictures of ear and tail tumors of Lgr5-CreERT2;LSL-ALKF1174L mice. (B) Representative pictures of mice that developed tumors on the back skin after wounds (due to fighting) were observed. Because of the interference of wounding on normal skin homeostasis, these mice have been withdrawn from experiments. (C) Western blot for p-ALK and α-tubulin as loading control, in Lgr5-CreERT2;LSL-ALKF1174L ear tumor protein lysate and control ear from sibling. (D, E) Evaluation of the skin lesions of Lgr5-CreERT2;LSL-ALKF1174L mice. (D) Column “Total” represents the percentage of mice that developed the corresponding diagnosed lesion. The P-value has been calculated by Fisher’s exact test in contingency table comparing the proportion of the corresponding skin lesion out of the total lesions per location (tail or ear). (E) Venn diagram representing the distribution of the different skin lesions in the mice. (F) Representative hematoxylin and eosin (H&E) staining of K5-CrePR1;LSL-ALKF1174L and K14-CreERT2;LSL-ALKF1174L skin lesions. Scale bar = 1 mm.
- Figure S2.
Characterization of the expression and mutations in human and mouse ALK genes. (A) Immunohistochemistry of a human skin sample immune-labeled with anti-ALK antibody shows positivity among cells of the basal layer of the epidermis. Scale bar = 25 µm. (B, C) Metanalysis of ALK mutations from different cohorts of cSCC patients. (B) number of cases with mutation in the ALK gene. (C) Graphical representation of the ALK mutations and PolyPhen 2 prediction. Red rectangles point out known gain of function (GOF) mutations. (D) Graphical representation of DMBA/TPA skin carcinogenesis model. Exome 23 of Alk have been sequenced in six tumors and two “normal” skin arising from induced skin areas. No mutations in Alk pF1178 have been observed (mouse residue of the human F1174). (E) Analysis of frequency of co-occurrence of mutations in ALK, TP53, and RAS genes. Co-occurrence or mutual exclusivity has been assessed thanks to the cBioPortal database. P-value is derived from one-sided Fisher’s exact test.
- Figure S3.
ALKF1174L cooperates with KrasG12D to drive cSCC tumorigenesis. (A) Graphical representation of Lgr5-CreERT2;LSL-KrasG12D genotype and tumor-free survival (n = 4). (B) Quantification of proliferating cells in back skins from wild-type, Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D mice and representative immunohistochemistry carried out using pan-cytokeratin (Pan-Ck) an Ki67 antibodies. Anagen hair follicles have been excluded. Data showed as mean ± SD (one-way analysis of variance, P = 0.6517). Scale bars = 50 µm. (C, D, E) Evaluation of the skin lesions of Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D mice. (C) The P-value has been calculated by Fisher’s exact test in contingency table comparing the proportion of the corresponding skin lesion out of the total lesions per location (tail or ear). (D) The P-value has been calculated by Fisher’s exact test in contingency table comparing the proportion of Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D mice that developed the corresponding skin lesion out of total mice per group. (E) Venn diagram representing the distribution of the different skin lesions in the mice. (F) Above: table showing the RNA relative expression of significantly mutated keratins between Lgr5-CreERT2;LSL-ALKF1174L versus Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D sorted tumor cells. Below: heat map showing all keratins mutated between sorted ear keratinocytes versus Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D sorted tumor cells. (G) Significantly mutated biological processes and KEGG signaling pathways between Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D tumors. Horizontal columns represent the P-value of the analysis of the down-regulated genes from RNA-seq transcriptional profiling of sorted tumor cells versus normal keratinocytes. Metabolic pathways that is marked by * is significantly changed also in the analysis of transcriptomes of Lgr5-CreERT2;LSL-ALKF1174L versus Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D tumor cells.
- Figure 2. ALKF1174L–driven tumorigenesis relied in bona fide, via PI3K-AKT/focal adhesion–ECM receptor interaction pathways.
(A) Ear tumor from Lgr5-CreERT2;LSL-ALKF1174L mice and ears from control mice were prepared as a single-cell suspension. Tumor cells and keratinocytes were isolated by FACS for EpCAM expression e-negative selected for CD31/CD45/CD140a. RNA was extracted from sorted cells and used for RNA sequencing (RNA-seq). (B) Top 15 biological processes and KEGG signaling pathways of up- and down-regulated genes from RNA-seq transcriptional profiling of sorted tumor cells versus normal keratinocytes. (C) Significantly deregulated keratins. (D) Genes significantly altered that cluster in the PI3K-AKT, Jak-Stat, and focal adhesion–ECM receptor interaction KEGG signaling pathways. (E) Clustering of biological processes using the MSigDB c5.bp.v6.2 gene set. Red nodes represent up-regulated gene sets and blue nodes represent gene sets down-regulated in the tumor cells. Node size shows the size of gene sets. Nodes that clustered together are classes with same or similar function indication. Lines between the nodes represent association of the gene sets within the nodes. (F) Graphical representation of genes/pathways regulated by ALKF1174L, based on RNA-seq data.
- Figure 3. KrasG12D synergizes with ALKF1174L increasing tumorigenicity, epithelial-to-mesenchymal transition properties, and vascularization and proliferation.
(A) Graphical representation of Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D genotype. (B) Distribution of tumor formation per location in Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D mice. (C) Pie chart representing the number of mice that developed the listed skin lesions out of total lesions diagnosed. (D) Number of tumors per mouse in Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D mice. Each dot represents the number of tumors per mouse. Mean ± standard error of the mean (5.917 ± 1.026; n = 12; 10.67 ± 1.280 n = 9; two-tailed t test P = 0.0086). (E) Tumor-free survival of Lgr5-CreERT2;LSL-ALKF1174L (n = 15, median 47 d) and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D (n = 13, median 29 d) mice. Log-rank (Mantel–Cox) test; P = 0.0002, HR = 0.1461. (F) Progression-free survival of Lgr5-CreERT2;LSL-ALKF1174L (n = 15, median 37 d) and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D (n = 13, median 19 d) mice. Log-rank (Mantel–Cox) test; P < 0.0001, HR = 0.07945. (G) Ear tumor from Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D mice were prepared as a single-cell suspension, and tumor cells were isolated by FACS for EpCAM expression e-negative selected for CD31/CD45/CD140a. RNA from sorted cells was used for RNA-seq. (H) Above: Venn diagram representing deregulated genes in tumor cells over controls. Below: significantly altered genes in Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D over Lgr5-CreERT2;LSL-ALKF1174L tumor cells. (I) Significantly altered biological processes and KEGG signaling pathways in Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D tumors. Horizontal columns represent the P-value of the analysis of the up-regulated genes from RNA-seq transcriptional profiling of sorted tumor cells versus normal keratinocytes. Processes and pathways marked by * are significantly changed also in the analysis of transcriptomes of Lgr5-CreERT2;LSL-ALKF1174L versus Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D tumor cells. (J) Genes significantly regulated from RNA-seq analysis, between Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D tumor cells that cluster in the indicated KEGG signaling pathways. (K) Representative immunofluorescences of Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D ear tumors immuno-labelled with DAPI, pan-cytokeratin (Pan-Ck), and vimentin antibodies. (L) Cells that co-expressed Pan-Ck and vimentin were counted as cells in epithelial-to-mesenchymal transition. Every tumor arising in the mice was analyzed and relative quantification is represented by a dot. Analysis of tumors from five Lgr5-CreERT2;LSL-ALKF1174L and five Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D mice showed an increased number of relative vimentin+ cells in Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D tumors (4.907 ± 2.904; 8.547 ± 3.885. Mann–Whitney test P = 0.0117). Scale bars = 50 µm. (M) Representative immunohistochemistry of Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D ear tumors immuno-labelled with Pan-Ck and with Ki67 antibodies, to identify tumors and proliferating cells. (N) Analysis of tumors from nine Lgr5-CreERT2;LSL-ALKF1174L and eight Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D mice showed an increased relative number of proliferating cells in the Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D tumors (5.786 ± 4.958; 7.614 ± 4.828. Mann–Whitney test P = 0.0043). Scale bars = 100 µm. (O) Representative immunohistochemistry of Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D ear tumors immuno-labelled with Pan-Ck and with CD31 antibodies, to identify tumors and vessels. (P) Every tumor arising in the mice was analyzed and relative quantification is represented by a dot. Analysis of tumors from eight Lgr5-CreERT2;LSL-ALKF1174L and eight Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D mice showed an increased relative area occupied by vessels within the Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D tumors (30.75 ± 34.34; 46.10 ± 35.24. Mann–Whitney test P = 0.0035). Scale bars = 100 µm. (J, K, L) Data showed as mean ± SD.
- Figure S4.
Comparison of the signaling pathways in DMBA/TPA-induced cSCC and ALKF1174L-induced cSCC. (A) Common genes between Lgr5-CreERT2;LSL-ALKF1174L tumor epithelial cells and DMBA/TPA genetic signatures. (B) Analysis of Ras-associated pathways in both mouse models reveals that genes from these pathways are significantly altered in DMBA/TPA and not in Lgr5-ALK mouse model. (C) Number of significantly up- or down-regulated genes in both mouse models. Overlapping genes can indicate common genes involved in carcinogenesis of cSCC. (D) Number of genes significantly up-regulated in one model versus down-regulated in the other model. Non-overlapping genes can indicate model-dependent genes involved in the carcinogenesis of cSCC.
- Figure S5.
Characterization of the specific biological process in DMBA/TPA-induced cSCC and ALKF1174L-induced cSCC. (A) Top 10 biological processes associated with different group of genes and pathways associated with common genes significantly up- and down-regulated in both mouse models. (B) Pathways associated with genes significantly up- and down-regulated specifically in DMBA/TPA mouse model. (C) Pathways associated with genes significantly up- and down-regulated specifically in Lgr5-CreERT2;LSL-ALKF1174L mouse model.
- Figure 4. p53 conditional KO increased tumorigenicity driven by ALKF1174L.
(A) Graphical representation of Lgr5-CreERT2;LSL-ALKF1174L;p53fl/fl genotype. (B) Tumor-free survival of Lgr5-CreERT2;LSL-ALKF1174L (n = 15, median 47 d) and Lgr5-CreERT2;LSL-ALKF1174L;p53fl/fl (n = 20, median 49 d) mice. Log-rank (Mantel–Cox) test; P = 0.4998, HR = 0.7834. (C) Progression-free survival of Lgr5-CreERT2;LSL-ALKF1174L (n = 15, median 37 d) and Lgr5-CreERT2;LSL-ALKF1174L;p53fl/fl (n = 20, median 32 d) mice. Log-rank (Mantel–Cox) test; P = 0.4158, HR = 0.7478. (D) Number of tumors per mouse in Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;p53fl/fl mice. Each dot represents the number of tumors in a specific mouse. Mean ± standard error of the mean (5.917 ± 1.026; n = 12; 10.12 ± 1.3 n = 17; Two-tailed t test P = 0.0254). (E) Pie chart representing the number of mice that developed the listed skin lesions out of total lesions diagnosed. (F) Representative hematoxylin and eosin (H&E) staining of acanthopapilloma3 and multiple SCC2 from ear skin of Lgr5-CreERT2;LSL-ALKF1174L;p53fl/fl mice. In the magnification, it is possible to appreciate mesenchymal-like features of tumor cells and pronounced nuclear atypia. Scale bars = 100 µm. (G) Representative immunofluorescences of Lgr5-CreERT2;LSL-ALKF1174L;p53fl/fl ear tumor immuno-labelled with DAPI, pan-cytokeratin (Pan-Ck), and vimentin antibodies. Scale bar = 50 µm. Cells that co-expressed Pan-Ck and vimentin were counted as cells in epithelial-to-mesenchymal transition (EMT). All tumors diagnosed as cSCC type 1 or 2 from different mice were analyzed, and quantification of the relative number of cells in EMT per tumor is represented by a dot. Analysis of tumors showed that no significant changes were observed between SCC1 tumors of three Lgr5-CreERT2;LSL-ALKF1174L or three Lgr5-CreERT2;LSL-ALKF1174L;p53fl/fl mice (4.380 ± 1.425; 7.360 ± 1.996. Mann–Whitney test P = 0.3277). Strong increase in the EMT rate was instead noted in SCC2 tumors when compared with SCC1 tumors of the same cohort of mice (7.360 ± 1.996; 17.10 ± 1.875. Mann–Whitney test P = 0.0056) or to Lgr5-CreERT2;LSL-ALKF1174L mice (4.380 ± 1.425; 17.10 ± 1.875. Mann–Whitney test P = 0.0008). All values are showed as median with 95% confidence interval. (H) Above: the schematic structure of the p53 floxed allele and graphical representation of PCR strategy to determine the recombination efficiency of LoxP sites that drives the removal of exons 2–10 of the p53 gene. White boxes represent exons, arrowheads represent the LoxP sites, and purple lines indicate primers position. Below: PCR analysis of recombination in different tissues. The 612-bp band present in the Lgr5-CreERT2;LSL-ALKF1174L;p53fl/fl tumors confirms that recombination occurred. (I) Representative pictures of in vivo imaging system (IVIS). Analysis performed to detect metastasis in vivo (left) and ex vivo (right). The grey box is used to cover a strong luciferase signal in the gut. The strong luciferase expression observed in the intestinal epithelium of the gut is likely due to the residual amount of 4-OHT entering systemic circulation (mice transferring 4-OHT by scratching their ears and subsequently licking their paws).
- Figure S6.
Histopathological characterization of skin lesions observed in LSL-ALKF1174L p53fl/fl Lgr5-CreERT2. (A, B, C) Evaluation of the skin lesions of Lgr5-CreERT2;LSL-ALKF1174L;p53fl/fl mice. (A) The P-value has been calculated by Fisher’s exact test in contingency table comparing the proportion of Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L; p53fl/fl mice that developed the corresponding skin lesion out of total mice per group. (B) The P-value has been calculated by Fisher’s exact test in contingency table comparing the proportion of the corresponding skin lesion out of the total lesions per location (tail or ear). (C) Venn diagram representing the distribution of the different skin lesions in the mice. (D) Number of tumors per mouse in Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;Stat3fl/fl mice. Each dot represents the number of tumors in a specific mouse. Mean ± standard error of the mean (5.917 ± 1.026 n = 12, 5.727 ± 1.280 n = 11; Two-tailed t test P = 0.9084).
- Figure 5. Stat3 is indispensable for ALKF1174L-driven tumorigenicity.
(A, B) Western blot analysis of phosphorylation status of Stat3 in (A) Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;LSL-KrasG12D tumors and in (B) HEK293T cell line transiently expressing the ALKF1174L transcript or control. (C) Graphical representation of Lgr5-CreERT2;LSL-ALKF1174L;Stat3fl/fl genotype. (D) Tumor-free survival of Lgr5-CreERT2;LSL-ALKF1174L (n = 20, median 51 d) and Lgr5-CreERT2;LSL-ALKF1174L;Stat3fl/fl (n = 11, median 63 d) mice. Log-rank (Mantel–Cox) test; P = 0.0202, HR = 2.540. (E) Progression-free survival of Lgr5-CreERT2;LSL-ALKF1174L (n = 20, median 37 d) and Lgr5-CreERT2;LSL-ALKF1174L;Stat3fl/fl mice (n = 11, median 43 d). Log-rank (Mantel–Cox) test; P = 0.0616, HR = 2.066. (F) Above: structure of the Stat3 floxed allele and graphical representation of PCR strategy to analyze recombination efficiency. Recombined allele will produce a shorter mRNA missing exons 18-19-20. White boxes represent exons, arrowheads represent the LoxP sites, and purple lines represent the primers. Below: RT-PCR analysis of different tissues. Whereas wild-type ear skin shows no detectable expression of Stat3, only 3/6 of the Lgr5-CreERT2;LSL-ALKF1174L;Stat3fl/fl tumors analyzed show a mixed expression of full-length and truncated Stat3. Number of tumors per mouse in Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;Stat3fl/fl mice. Mean ± standard error of the mean (5.917 ± 1.026 n = 12, 5.727 ± 1.280 n = 11; two-tailed t test P = 0.9084). (G) Representative immunofluorescence images of Lgr5-CreERT2;LSL-ALKF1174L, Lgr5-CreERT2;LSL-ALKF1174L;Stat3fl/fl ear tumors immuno-labelled with DAPI, pan-cytokeratin (Pan-Ck), and p-Stat3Y705 antibodies show phosphorylation of Stat3 within the tumors (above) and in the hyperplastic skin and in the hair follicles (HF) adjacent to the tumor masses (below). To note, all Lgr5-CreERT2;LSL-ALKF1174L;Stat3fl/fl tumors analyzed (n = 8) showed p-Stat3Y705 expression. Scale bars = 50 µm. (H) From left to right, wild-type ears HFs, and non-hyperplastic Lgr5-CreERT2;LSL-ALKF1174L and Lgr5-CreERT2;LSL-ALKF1174L;Stat3fl/fl ear HFs adjacent to the tumor masses have been immuno-labelled with DAPI, pan-cytokeratin (Pan-Ck), and p-Stat3Y705 antibodies. Insets display strong p-Stat3Y705 expression within the HFs of Lgr5-CreERT2;LSL-ALKF1174L mice, whereas wild-type skin is devoid of p-Stat3Y705 expression. Scale bars = 25 µm. Quantification of the recombination was measured as relative number of HFs expressing or not expressing p-Stat3Y705. Only non-hyperplastic HFs were being considered for the analysis.
- Figure 6. A schematic illustration summarizing main findings.
(A) Expression of ALKF1174L in Lgr5+ hair follicle stem cells results in formation of different skin lesions as Cysts, acanthopapilloma (AP), keratoacanthoma (KA), or squamous cell carcinoma type 1 (SCC1). Additional KrasG12D expression leads to increased epithelial to mesenchymal transition, proliferation, and vascularization. Loss of p53 primes to formation of a more aggressive tumor: squamous cell carcinoma type 2 (SCC2). Finally, loss of Stat3 prevents the tumor formation ALKF1174L-induced.
