Fig 2: Correlation of TMEM density with the proportion of tumor cells expressing stem cell markers in breast cancer tissues from patients.a Clinical and pathological characteristic of the 49 breast cancer cases included in the analysis. b A schematic of tissue collection procedure for the correlation of TMEM doorway score with the proportion of cancer cells expressing stem cell markers. c TMEM correlation with stem cell marker CD44+/24- in 49 invasive ductal carcinomas. Cells were collected by FNA and analyzed by flow cytometry for % cells expressing CD44+/CD24-. TMEM score from the same tumor tissue was correlated with the % CD44+/24- cells by Pearson’s correlation. d Correlation of TMEM with the percentage of breast cancer cells expressing stem cell markers CD44+/24-, ALDH1, and CD133 in 49 invasive ductal carcinomas from patients. Cells were collected by FNA and analyzed for CD133 and ALDH1 by in situ hybridization. CD44+/24- data are plotted from c. TMEM score from the same tumor tissue was correlated with the percentage of cancer cells expressing stem cell markers using Pearson’s correlation. e Correlation of TMEM with the percentage of breast cancer cells expressing stem cell markers ALDH1 and CD133 in 48 invasive ductal carcinomas from patients. Cells were collected by FNA and analyzed for CD133 and ALDH1 by qRT-PCR. TMEM score from the same tumor tissue was correlated with the percentage of cancer cells expressing stem cell markers using Pearson’s correlation. f Sox9 mean fluorescence intensity in FFPE of the above human tissues (n = 13) were correlated with the TMEM score for each human tissue. Pearson’s correlation r = 0.86, p = 0.0002, data normalized to the min and max Sox9 mean fluorescence intensity values. g A composite image of TMEM staining (pink: Mena+ tumor cells, brown: CD68+ macrophages, blue: CD31+ endothelial cells) and Sox9Hi stem cells (green), showing enrichment of Sox9Hi cells close to TMEM areas. Two TMEM doorway ROIs are highlighted with black circles. h Quantification of % Sox9Hi cells in TMEM-rich and TMEM-deficient areas. n = 3 human tissues, data plotted as mean ± SD, paired two-tailed Student’s t-test, *p = 0.0336.

Fig 3: Identification of SALL2 as a CSC-related factor whose translation is dependent on CDKAL1. A) Polysome fractionation was performed in Control-shRNA- (blue line) or CDKAL1-shRNA-expressing (red or green) RD cells. Note that there was no significant change in fractional pattern among the samples. B) Identification of the genes whose transcription (blue dots) or translation (red dots) was altered by CDKAL1 knockdown. The genes whose expression levels in total RNA were altered upon CDKAL1 knockdown are considered transcriptionally changed genes (X-axis) and those in polysome RNA are considered translationally changed genes (Y-axis). Homodirectionally changed genes and oppositely changed genes are shown as purple dots and green dots, respectively. C) GSEA was performed with the genes altered after CDKAL1 knockdown in a transcription-dependent manner in RD cells (upper, blue dots). D) Informatics analysis of lysine codon usage in CDS between the CDKAL1-dependent genes and all other genes (error bars indicate mean ± SD). E) Western blotting analysis of the ER-stress marker BiP in Control-shRNA- or CDKAL1-shRNA-expressing RD cells. Thapsigargin-treated cells were used as a positive control of ER-stress. F) Representative electron microscopy images of Control-shRNA- or CDKAL1-shRNA-expressing RD cells. Scale bars, 500 nm. G) Puromycin incorporation assay was done to monitor the global protein synthesis in Control-shRNA- or CDKAL1-shRNA-expressing RD cells. H) Identification of two de novo motifs, guanine-enriched sequences (GESs) and cytosine-enriched sequences (CESs), in the 5’UTR of the CDKAL1-dependent mRNAs. GO term analysis showing that compared to GESs, CESs correlated highly with the transcription regulator activity. SALL2 was found at the top of gene list of “transcription regulator activity”, whose 5’UTR was correlated with the CESs. I) Immunofluorescent analysis using anti-CD133, anti-SALL2, and anti-CDKAL1 in RD and RH30 cells. Scale bars, 20 µm. J) Effect of CDKAL1 knockdown on SALL2 expression levels in RMS cell lines (n = 2, error bars indicate mean ± SD).

Fig 4: SALL2 is an essential factor for maintaining CSCs in RMS and is controlled by CDKAL1 at the translation initiation phase. A,B) The self-renewal capacity (A, n = 4, error bars indicate mean ± SD) and colony-forming potential (B, n = 3, error bars indicate mean ± SD) of Control-shRNA- or SALL2-shRNA-expressing RMS cells were measured (n = 4, error bars indicate mean ± SD). Western blotting results are shown to confirm SALL2 knockdown (A). C) One million Control-shRNA- or SALL2-shRNA-expressing RMS-YM cells were injected subcutaneously at the backs of BALB/c-nu/nu mice, and tumor growth was measured. Ten injection sites were measured for each group (error bars indicate mean ± SD). D) Immunofluorescent analysis using anti-CD133 and anti-MyHC antibodies in the RMS-YM xenografts obtained in Figure 5C. All scale bars, 50 µm. E) Immunofluorescent analysis of normal and pathological specimens using anti-CDKAL1 and anti-SALL2 antibodies. Note that CDKAL1high tumor cells exhibit SALL2high properties in RMS. Scale bars, 100 µm. F) Luciferase reporter analysis with 5’UTR-SALL2, -GAPDH, and -ACTB in Control-shRNA- or SALL2-shRNA-expressing RD cells (n = 3, error bars indicate mean ± SD). G) A cartoon representing the procedure. H) In situ proximity ligation assay detecting a signal obtained by the reactions from endogenous CDKAL1 and the DIG-labeled sense or antisense RNA probes of the 5’UTR region of SALL2 mRNA in Control-shRNA- or SALL2-shRNA-expressing RD cells. Scale bars, 10 µm. I) RNA immunoprecipitation and reverse transcription PCR assay using RD cells expressing myc-tagged CDKAL1. Primer pairs for 5’UTR of SALL2 mRNA were used to detect the interaction between CDKAL1 and SALL2 mRNA. Reverse transcriptase (-) samples serve as negative controls. Three samples (#1, #2, and #3) were analyzed.

Fig 5: CDKAL1 promotes the maintenance of CSC-related traits in RMS. A) The self-renewal capacity of Control-shRNA- or CDKAL1-shRNA-expressing RMS cell lines was measured (n = 4, error bars indicate mean ± SD). Western blotting results are shown to confirm the CDKAL1 knockdown. B) The clonogenic growth potential of Control-shRNA- or CDKAL1-shRNA-expressing RMS cells (n = 3, error bars indicate mean ± SD). C) The cell proliferation analysis of Control-shRNA- or CDKAL1-shRNA-expressing RD and RH30 cells (n = 4, error bars indicate mean ± SD). D) Flow cytometry analysis showing the effect of CDKAL1 knockdown on CD133 expression in RD cells and RH30 cells. E) Immunofluorescent imaging of myogenic differentiation of RMS cells after CDKAL1 knockdown, as indicated by positive staining for myosin heavy chain (MyHC). F) One million Control-shRNA- or CDKAL1-shRNA-expressing RMS-YM cells were subcutaneously injected at the backs of BALB/c-nu/nu mice, and tumor growth was measured. Ten injection sites were measured for each group (error bars indicate mean ± SD). G) Immunofluorescent analysis using anti-CD133 and anti-MyHC antibodies in the RMS-YM xenografts obtained in Figure 2E. All scale bars, 50 µm. H) Effect of the forced expression of CDKAL1 on the sphere-forming capacity of RD and ICH-ERMS-1 cells (n = 4, error bars indicate mean ± SD).